Roger G. Barry - Synoptic and Dynamic Climatology (2001)

11
Synoptic and Dynamic Climatology
111
Synoptic and Dynamic Climatology provides the first comprehensive account of the
dynamical behavior and mechanisms of the global climate system and its components,
together with a modern survey of synoptic-scale weather systems in the tropics and extra-
tropics, and of the methods and applications of synoptic climate classification. It is
unrivalled in the scope and detail of its contents. The work is thoroughly up to date, with
0 extensive reference sections by chapter. It is illustrated with plates and nearly 300 figures.
• Part 1 provides an introduction to the global climate system and the space–time scales
of weather and climate processes, followed by a chapter on climate data and their
analysis.
• Part 2 describes and explains the characteristics of the general circulation of the global
atmosphere, planetary waves and blocking behavior, and the nature and causes of
global teleconnection patterns.
• Part 3 discusses synoptic weather systems in the extratropics and tropics, and satellite-
based climatologies of synoptic features. It also describes the methods and applications
0111 of synoptic climatology and summarizes current climatic research and its directions.
The book is intended for advanced students in climatology and environmental and atmos-
pheric sciences, as well as for professionals in the field of climate dynamics and variability.
It presents both established findings about global climate and unresolved issues. Its
comprehensive reference lists provide an invaluable guide to further study.
Roger G. Barry is Professor of Geography and Director of the National Snow and Ice
Data Center at the University of Colorado and Andrew M. Carleton is Professor of
Geography at Pennsylvania State University.
0
11
1
1
11
Synoptic and Dynamic
Climatology
111 Roger G. Barry and

Andrew M. Carleton
0111
LE
UT D
RO
GE
•
•
0
p
Ta
ou
or
y
r
l
& F r n cis G
a
11 London and New York

1 First published 2001
by Routledge
11 New Fetter Lane, London EC4P 4EE
Simultaneously published in the USA and Canada
by Routledge
29 West 35th Street, New York, NY 10001
Routledge is an imprint of the Taylor & Francis Group
This edition published in the Taylor & Francis e-Library, 2002.
© 2001 Roger G. Barry and Andrew M. Carleton
The right of Roger G. Barry and Andrew M. Carleton to be identified as the
Authors of this Work has been asserted by them in accordance with the
Copyright, Designs and Patents Act 1988
All rights reserved. No part of this book may be reprinted or reproduced or
utilized in any form or by any electronic, mechanical, or other means, now
known or hereafter invented, including photocopying and recording, or in any
information storage or retrieval system, without permission in writing from the
1 publishers.
British Library Cataloguing in Publication Data
A catalogue record for this book is available from the British Library
Library of Congress Cataloging in Publication Data
Barry, Roger Graham.
Synoptic and dynamic climatology / Roger G. Barry and Andrew M. Carleton.
p. cm.
Includes bibliographical references and index.
1. Synoptic climatology. I. Carleton, Andrew M. (Andrew Mark). II. Title.
QC981.7.S8 B36 2001
551.6—dc21
ISBN 0–415–03115–x (hbk)

ISBN 0–415–03116–8 (pbk)
ISBN 0-203-21818-3 Master e-book ISBN
ISBN 0-203-21830-2 (Glassbook Format)
11
Contents
111 List of plates viii

Preface ix
Acknowledgments xi
PART 1
0
The climate system and its study 1
1 Introduction 3
1.1 The global climate system 3
1.2 Time and space scales of weather and climate processes 10
1.3 Dynamic and synoptic climatology 13
1.4 The structure of the book 14
0111 2 Climate data and their analysis 16

2.1 Synoptic meteorological data 16
2.2 Remotely sensed data 17
2.3 Climate variables and their statistical description 31
2.4 Analytical tools for spatial data 40
2.5 Time series 63
2.6 Empirical orthogonal function analysis, clustering, and
classification 78
Appendix 2.1 Eulerian and Lagrangian methods 84
0
PART 2
Dynamic climatology 107
3 Global climate and the general circulation 109

3.1 Planetary controls 109
3.2 Basic controls of the atmospheric circulation and its
maintenance 113
0
3.3 Circulation cells 143
3.4 The Earth’s geography 153
11 3.5 Climate system feedbacks 155
vi Contents
3.6 General circulation models 161
3.7 The global circulation – description 166
3.8 Centers of action 209
3.9 Global climatic features 223
3.10 Air masses 235
Appendix 3.1 Potential vorticity 240
4 Large-scale circulation and climatic characteristics 263

4.1 Time-averaged circulation 263
4.2 Jetstreams 270
4.3 Planetary waves 278
1 4.4 Zonal index 302
4.5 Zonal and blocking flow modes 308
4.6 Blocking mechanisms 313
4.7 Low-frequency circulation variability and persistence 322
4.8 Intraseasonal oscillations 332
Appendix 4.1 Spectral harmonic functions 340
Appendix 4.2 Eliassen–Palm flux 341
Appendix 4.3 Normal modes 342
5 Global teleconnections 358

5.1 Pressure oscillations and teleconnection patterns 358
5.2 The Southern Oscillation and El Niño 361
5.3 ENSO mechanisms 376
5.4 Teleconnections with ENSO 384
1
5.5 Extratropical teleconnection patterns 396
5.6 North Atlantic Oscillation 397
5.7 North Pacific Oscillation 403
5.8 Zonally symmetric oscillations 404
5.9 The southern hemisphere 408
5.10 Tropical–extratropical teleconnections 410
5.11 Teleconnections and synoptic-scale activity 414
5.12 Time-scale aspects of teleconnections 414
5.13 Interannual to interdecadal oscillations 418
Appendix 5.1 Partitioning between equatorially symmetric and
antisymmetric components 424
PART 3
Synoptic climatology 439
6 Synoptic systems 441

6.1 Early studies of extratropical systems 441
6.2 Climatology of cyclones and anticyclone 442
6.3 Development of cyclones 450
Contents vii
11 6.4 Storm tracks 465
6.5 Satellite-based climatologies of synoptic features 476
6.6 Synoptic-scale systems in the tropics 506
Appendix 6.1 The Q-vector formulation 524
7 Synoptic climatology and its applications 547

ROGER G. BARRY AND ALLEN H. PERRY
7.1 Synoptic pattern classification 547
0
7.2 Subjective typing procedures 549
7.3 Objective typing procedures 551
7.4 Principal catalogs and their uses 561
111 7.5 Regional applications 574
7.6 Analogs 578
7.7 Seasonal structure 578
7.8 Climatic trends 587
7.9 Environmental applications 589
0
8 Retrospect and prospect 604
Further reading 606

Index 611
0111
11
Plates
Color
1 The following plates appear between pages 304–5
1 Relief of the surface of the Earth, showing land elevation and ocean
bathmetry, based on the ETOPO 5 database
2 Average extent of snow cover, 1971–95, and sea ice, 1978–95, in the
northern hemisphere for February and August
3 Global images of the normalized difference vegetation index for January,
April, July, and the standard deviation values for July
7 (a) Thermal IR image from the Japanese GMS, rectified to a polar
stereographic format, for 23.30 UTC, November 7 1992. (b)–(f) Color maps
of the DMSP SSM/I retrievals of marine atmospheric variables, for around
22.00 UTC, November 7 1992
Black-and-white
4 Sequence of infrared images illustrating the breakdown and redevelopment
1 of the ITCZ cloud band over the eastern tropical Pacific Ocean between
July 26 and August 12 1988 187
5 GOES-E enhanced thermal infrared images of the North America and
Central America regions on June 25 1988 at 06.00 and 18.00 UTC,
showing convective and frontal high cold cloud tops 479
6 DMSP visible channel mosaic of the Europe/western Russia sector,
February 22 1978, showing a jetstream “shadow band” over the snow
cover 481
8 DMSP infrared image of an instant occlusion about to be initiated on the
frontal cloud band of a North Pacific extratropical cyclone through the
merging of a cold-air mesocyclone at about 36°N, 164°E 489
9 DMSP infrared images of (a) a comma cloud mesocyclone and
(b) a spiraliform mesocyclone in the Labrador Sea during January 1979 491
10 GOES-E enhanced infrared image showing a large MCS over the central
United States on August 13 1982 496
11 GOES-W enhanced infrared image showing an MCC in central Arizona
during the summer “monsoon” season on August 12 1982 497
12 DMSP infrared mosaic showing a tropical–extratropical cloud band
connection in the western Pacific during October 1977 502
13 Infrared mosaic of the southern hemisphere from the NOAA SR
(Scanning Radiometer) for June 17 1975 504
14 GOES-W infrared images showing a moisture plume extending from
the eastern North Pacific into the central United States 505
15 GOES-E image of Hurricane Gilbert on September 12 and 15 1988 518
11
Preface
111 In his book The Earth’s Problem Climates (1961) Professor Glenn Trewartha attempted
to account for apparent regional departures from the “expected” climatic pattern in various
parts of the world. In contrast, this book is principally concerned with understanding the
large-scale regularities of the earth’s climatic patterns. The empirical reconstructions of
climatic patterns on an ideal continent have in fact been shown by numerical model studies
to be unrealistic in several respects. The actual distributions of climatic regions, on which
0 most classification schemes are based, are significantly modified by geographical features
– mountain ranges and ocean currents.
Our approach is to demonstrate, first, the relationships between the dynamical controls
of the global circulation and the major climatic zones, and then to examine the longitu-
dinal variations introduced by the planetary and synoptic wave systems, interacting with
the earth’s surface. The principal time scales of interest are days to years, and we are not
considering the “slow physics” that gives rise to decadal and longer climatic fluctuations,
except in so far as the variability of circulation systems themselves may be involved. The
second part of the book focuses on synoptic climatology. Twenty-seven years have elapsed
since the publication of Synoptic Climatology: Methods and Applications (Barry and Perry,
0111 1973) and the field has advanced considerably during that time (Carleton, 1999).
Computer-based classifications have become the rule, and the ready availability of exten-
sive climatic data bases has permitted diverse analytical studies to be carried out and
evaluated. Results from these studies now allow us to state more confidently the most
useful procedures for the synoptic classification and analysis of regional or local climatic
conditions. The widespread practical application of these studies is illustrated, with partic-
ular emphasis on the linkages between regional and global-scale climate processes. This
is an essential aspect of the problem of interpreting global climate model simulations in
a regional context.
All areas of science have experienced phenomenal growth since the Second World War.
0 The number of scientific papers published between 1960 and 1980 far exceeded the sum
total of previous works, according to Geerts (1960). Six major atmospheric science jour-
nals started in the 1970s, there are now at least five climatology journals, and established
journals have added extra issues and more pages. This information explosion makes a
synthesis of the literature a daunting task and yet such attempts are increasingly neces-
sary, both for educational purposes and for scientists in cognate disciplines. We hope that
this book will provide an up-to-date and comprehensive treatment of dynamic and synoptic
climatology for these audiences.
We begin with an introduction to the global climate system and related time and space
scales. Chapter 2 on climatic data and their analysis may be read in sequence, or used
0 for reference when particular methods are noted in subsequent chapters.
11
x Preface
References
Barry, R.G. and Perry, A.H. 1973. Synoptic Climatology: Methods and Applications, Methuen,
London, 555 pp.
Carleton, A.M. 1999. Methodology in climatology. Prof. Geogr., 51: 713–35.
Greets, B. 1998. Trends in atmospheric science journals: a reader’s perspective. Bull. Amer. Met.
Soc., 80: 639–50.
Trewartha, G.T. 1961. The Earth’s Problem Climates, McGraw-Hill, New York, 334 pp.
1
11
Acknowledgments
111 The preparation of this book has involved many individuals. First and foremost I wish
to thank the following secretarial and clerical staff for their word-processing skills in
preparing the extensive text and many updates: Margaret Strauch, Cindy Brekke-Bauer,
Lyn Ryder, and students Megan Phelan, Shari Fox and Mathew Stones. Pat Hofman and
student Devon Lehman assisted with library materials and Mathew Stones ably scanned
most of the diagrams. A number of colleagues provided valuable comments on chapter
0 drafts: Dr George Kiladis, NOAA, Dr Rol Madden and Dr Gerry Meehl, NCAR, Dr Jeff
Key, Boston University, Professor Atsumu Ohmura, ETH, Zürich, and Anton Seimon,
University of Colorado, but remaining errors and omissions are my own responsibility.
Finally, the editorial staff of Routledge are thanked for their assistance in getting the book
to publication. For my early training in this field I am indebted to F. Kenneth Hare, then
Professor of Geography and head of the Arctic Meteorology Research Group at McGill
University, Montreal. Sabbatical leave from the University of Colorado Graduate School
in spring 1997, when I was hosted by the Geographical Institute of ETH, Zürich, enabled
me to accelerate the completion of the text. The library facilities of NOAA in Boulder
and of the Swiss Meteorological Institute in Zürich provided indispensable literature
0111 resources. The sources of all figures and plates, and associated copyrights, are listed by
chapter. The cooperation of all publishers, professional societies, and individuals is grate-
fully acknowledged.
Roger G. Barry
The authors and publishers would like to thank the following learned societies, editors,
publishers, organizations, and individuals for granting permission to reproduce Plates, Figures,
and Tables in this work. (Please note that the figure number refers to the figure number in
0 this work, but that any page numbers refer to the page of the original publication that the
work was taken from.)
Learned societies
American Geographical Society for Figure 4.6 on p. 535 of “The Stratosphere,” by F.K. Hare,
1962, Geographical Review 52: 525–47.
American Geophysical Union, Washington DC, for Figures 3.30, 4.20, from the Review of
Geophysics and Space Physics; for Figure 3.29 from Climate Processes and Climate
Sensitivity, by J. Hansen et al.; for Figures 3.17, 3.19, 3.33, 4.27, from the Journal of
0 Geophysical Research.
American Meteorological Society for Plate 3 and Figures 2.22, 2.23, 2.24, 3.6, 3.78, 4.36,
5.5, 6.8, from the Bulletin; for Figures 2.15, 2.17, 3.40, 3.65 from the Journal of
11 Applied Meteorology; for Figures 3.66, 3.67, 4.5, 4.7, 4.11, 6.14 from the Meteorological
xii Acknowledgments
Monographs; for Plate 4 and Figures 3.5, 3.22, 3.26, 3.39, 3.43, 3.69, 4.15, 4.21, 4.23, 4.25,
4.26, 4.47, A4.2.1, A4.2.2, A4.2.3, 5.17, 5.35, 5.36, 6.21, 6.22, 6.23, 6.24, 6.25, 6.26, 6.40,
6.42, from the Journal of Atmospheric Sciences; for Figures 3.62, 3.64, 3.81, 4.50, 4.53,
A4.1, 5.1, 5.3, 5.8, 5.9, 5.10, 5.18, 5.23, 5.25, 5.26 a and b, 5.29, 5.31, 5.32, 5.33, 5.34, 6.11,
6.13, 6.16, 6.17, 6.36, 7.14, from the Monthly Weather review; for Figures 2.3, 6.20 from
Weather and Forecasting; for Figure 2.2 from p. 19 of Weather Satellites, by P.K. Rao; for
Figures 3.9, 3.32, 3.44, 3.70, 3.72, 3.77, 3.82, 3.83, 4.1, 4.2, 4.32, 4.33, 5.12, 5.14, 5.15,
5.30, 5.37, 5.38, 5.39, 6.4 from the Journal of Climate; Figure 6.10 from the Eric Palmen
Memorial Volume; Figure 7.4 from the Sixth Conference on Probability and Statistics in
Atmospheric Science.
Association of American Geographers for Figure 4.12 from Synoptic Climatology of the
Westerlies, by J.R. Harman.
Geographical Association, UK, for Figure 7.6 from Geography.
1 Geological Society, London, for Figure 3.27 from the Journal of the Geological Society.
Meteorological Society of Japan, Tokyo, for Figures 3.63, 4.51, 6.32, and 6.38 from the
Journal of the Meteorological Society of Japan; for Figure 4.30 from the Geophysical
Magazine.
Royal Geographical Society/Institute of British Geographers for Figure 7.8 from the
Transactions; for Figure 5.10 from the Geographical Journal.
Royal Meteorological Society for Figures 1.8, 2.4, 2.13, 3.36, 3.42, 3.45, 3.48, 3.49, 4.18, 4.39,
4.40, 4.49, 4.51, A4.1.1, 5.16, 5.20, 5.22, 6.18, 6.31, 6.37, 6.41, A6.1 from the Quarterly
Journal; for Figure 3.14 from p. 63 of The Global Circulation of the Atmosphere, edited by
G.A. Corby, 1970; for Figure 3.80 by E. Augstein from p. 75 of Meteorology over the
Tropical Oceans, edited by D.B. Shaw, 1978; for Figure 6.12 from Meteorology Applied; for
Figure 4.37 from Weather.
Publishers
Academic Press, New York for Figures 1.2, 3.12, 3.16, 4.44, 4.45, and Table 4.1 from Advances
1 in Geophysics; for Figures 3.58 (p. 26), 3.60 (p. 48), 3.61 (p. 54) from Monsoon Meteorology
by C.S. Ramage, 1971.
Academic Press, San Diego, California, for Figure 2.8 from Statistical Methods in the
Atmospheric Sciences, by D.S. Wilks, 1995.
Academic Press, Orlando, Florida, for Figures 4.13 and 4.14 from J.M. Wallace (pp. 27–53),
Figure 4.24 from I.M. Held (p. 132), Figures 4.43 and 5.2 from J.M. Wallace and M.L.
Blackmon (pp. 55–94), all in Large Scale Dynamical Processes in the Atmosphere, edited
by B.J. Hoskins and R.P. Pearce, 1983.
Annual Reviews (www.annualreviews.org) for Figure 3.8 from Annual Review of Fluid
Mechanics.
Birkhaeuser Publishing, Basel, Switzerland, for Figure 3.55 (in “Mechanics effecting the state,
evolution and transition of the planetry scale monsoon,” 1977, by P.J. Webster et al., 15:
1465) and 3.79 from Pure Applied Geophysics.
British Crown Copyright for Figures 4.41 and 4.42 from the Meteorological Office of the UK.
Cambridge University Press, Cambridge, for Figure 1.3 from p. 144 of The Earth as
Transformed by Human Actions, eds B.L. Turner et al.; for Figure 1.4 from p. 142 of The
Global Climate, ed. J.T. Houghton; for Figure 3.54 (p. 63) in Monsoon Dynamics, edited by
Sir J. Lighthill and R.P. Pearce; for Figure 5.6 by K.E. Trenberth, from p. 14 and Figure 5.18
by E.M. Rasmusson from p. 323 of Teleconnections linking Worldwide Climate Anomalies,
edited by M.H. Glantz et al., 1991; for Figure 5.7 by H.F. Diaz and G. Kiladis, from p. 21
of El Nino, edited by H.F. Diaz and V. Markgraf, 1992.
The Controller, Her Majesty’s Stationery Office, for Figure 3.51 from Geophysical Memoir 115
by J. Findlater.
Acknowledgments xiii
11 Elsevier Science for Figure 3.74 from Advances in Space Research.
Kluwer Academic Publishers, for Figure 3.61 (p. 246) of Climate Dynamics of the Tropics, by
S. Hastenrath; for Figure 3.35 from The Climate of Europe, Past, Present and Future, by H.
Flohn, 1984; for Figure 3.5 from Milankovitch and Climate.
Pearson Education Ltd, London, for Figure 2.25 from p. 143 of Multivariate Statistical Analysis
in Geography, by R.J. Johnston, Longman; for Figure 3.38 from Concepts in Climatology,
by P.R. Crowe, Longman.
MeteoSchweiz, Zuerich, Switzerland, for Figure 7.12 by M. Schuepp, from Klimatologie der
Schweiz, Vol.3, 1979.
0 Munksgaard International Publishers Ltd, Copenhagen, Denmark for Plates 8 and 9, and Figures
1.6, 2.14, 3.37, 3.72, 3.85, 4.9, 5.24, 6.30, and 7.10 from Tellus.
National Academy Press, Washington DC, for Figure 4.46 from Understanding Climate Change,
by the National Academy of Sciences, 1975.
111 Oxford University Press Inc., New York, for Figures 3.3 and 3.5 from Paleoclimatology, by T.J.
Crowley and G.R. North, 1991; for Figures 3.20, 3.75 and 4.22 from Global Atmospheric
Circulations, by R. Grotjahn, 1993; for Figure 4.34, by R.M. Dole, pp. 93–9 of Encyclopedia
of Climate and Weather, edited by S.H. Schneider, 1996.
Springer, Berlin & Heidelberg, for Figures 2.21 from p. 455 of Decadal Climate Variability,
edited by D. Anderson and J. Willebrand; for Figures 3.15, 3.23, 3.24, 3.25, 6.1 (D.A. Jones
0 and I. Simmonds, “A Climatology of Southern Hemisphere Anticyclones,” 1994, 10:333–48)
and 6.2 from Climate Dynamics.
Springer, Vienna and New York, for Figure 4.31 from Archiv fur Meteorologie, Geophysik and
Bioklimatologie; for Figures 3.57, 4.52, 5.19, and 5.21 from Meteorology & Atmosphere
Physics; for Figure 4.3 from Theoretical and Applied Climatology; for Figures 3.10 (p. 243),
3.11 (p. 253), 3.13 (p. 159), 3.21 (p. 383), 3.31 (p. 454) from the Physics of Climate, by J.P.
Peixoto and A.H. Oort, American Institute of Physics, New York.
Taylor & Francis, London, for Plate 12 from the International Journal of Remote Sensing
(www.tandf.co.uk/journals) and for Figures 1.8 (p. 90), 2.5 (p. 216), 2.9 (p. 23), 2.10 (p. 26),
2.11 (p. 29), 2.12 (p. 32), 2.13 (p. 33), 2.16 (p. 387), 2.18 (p. 230), 2.19 (p. 232–3), 2.20 (p.
0111 237), 3.50 (p. 447), 3.84 (p. 185), 3.87 (p. 185), 4.31, 6.11 (p. 60), 7.11 (p. 166–7) from
Synoptic Climatology, by R.G. Barry and A.H. Perry, 1973, Methuen; for Figure 2.7 (p. 484)
from Maps and Diagrams, 3rd edition, edited by F.J. Monkhouse and H.R. Wilkinson, 1971,
Methuen; for Figure 3.7, from Climate Past, Present and Future, by H.H. Lamb, 1977,
Methuen; for Figures 3.34 (p. 169), 3.41 (p. 127), 3.59 (p. 259), 3.76 (p. 68), 3.86 (p. 156),
4.4 (p. 121), 5.4 (p. 278), 6.39 (p. 244), 7.15 (p. 195) from Atmosphere, Weather & Climate,
7th edition, by R.G. Barry and R.J. Chorley, 1998, Routledge; for Figures 4.10 and 6.7
from Models in Geography, edited by R.J. Chorley and P. Haggett, 1967, Methuen;
for Figures 4.17 (p. 108), 4.19 (p. 114) from Dynamical Meteorology, by B.W. Atkinson,
1981, Methuen; for Figure 1.5 (p. 70) from The Atmosphere and Ocean, by N. Wells,
0 Taylor & Francis.
University of Chicago Press, for Figure 3.28 from the Journal of Geology.
John Wiley and Sons, Chichester, for Figures 2.26, 4.38, 5.28, 5.29, 6.3, 6.19, 7.2, 7.9, 3.76,
5.27, 6.5, 7.1 and 7.3 from the International Journal of Climatology; for Figure 2.27 from
Computer Applications in Stratigraphic Analysis, 1968, by J.W. Harbaugh and D.P. Merriam;
for Figures 3.52 (p. 5), 3.53 (p. 214), 3.56 (p. 293), 6.33 (p. 240) from Monsoons, edited by
J.S. Fein and P.L. Stephens, 1987; for Plate 15 from Satellite Remote Sensing in Climatology,
by A.M. Carleton, 1991, Belhaven Press.
Organizations
0
Arctic Institute of North America, for Figure 6.6 by E.F. LeDrew and D.G. Barber, 1994,
“The SIMMS Program,” Arctic 47: 256–64.
11 Australian Bureau of Meteorology, Melbourne, for Plate 7a and Figures 6.27 and 6.28.
xiv Acknowledgments
Climate Research Unit, University of East Anglia, UK, for Figure 7.7 by H.H. Lamb, 1994,
“British Isles daily wind and weather patterns,” Climate Monitor 20: 47–71.
Cooperative Institute for Research in Environmental Sciences (CIRES), University of Colorado,
for Figure 4.16.
The Commonwealth of Australia for Figure 3.71 from the Australian Meteorological Magazine.
ETH, Institute for Climate Research, for Figures 1.1, 3.1, and 3.2 from Zurcher Geographische
Schriften (now Zurcher Klimaschriften).
European Centre for Medium Range Weather Forecasts, for Figure 7.5 from the Workshop in
Meteorology, Reading, UK, 1977.
National Center for Atmospheric Research (NCAR), Boulder, Colorado, for Figures 2.1, 3.68,
4.7, 4.28, and 4.29.
National Geophysical Data Center (NGDC), Boulder, Colorado, for Plate 1.
National Research Council Research Press, Canada, for Figure 5.13 from the Canadian Journal
1 of Aquatic Sciences.
National Oceanic and Atmospheric Administration (NOAA), Washington DC, for Plates 1, 5,
10, 11, 13, 14.
National Snow and Ice Data Center (NSIDC), Boulder, Colorado, for Plates 1, 2, 6, 8, 9, and
12.
National Weather Service, USA, for Plates 11 and 14.
Woods Hole Oceanographic Institute, for Figure 3.4 from “Orbital Geometry,” 1986, by N.G.
Pisias and J. Imbrie, Oceanus 29: 46.
World Meteorological Organisation, Geneva for Figures 3.18, 4.19 and 6.34.
Editor
Geocarto International for Figure 6.29.
Individuals
1
Professor B.W. Atkinson, Queen Mary College London, for Figures 1.7 and 1.9.
Professor R.A. Bryson, University of Wisconsin, Madison, for Figures 2.6, 7.16.
Professor A.M. Carleton, Pennsylvania State University, for Plate 15.
Dr H. Diaz, NOAA, for Figure 5.7.
Professor F. Fliri for Figure 7.13.
Professor M. Ghil, University of California, Los Angeles, for Figure 4.48.
Professor S. Grønäs, University of Bergen, Norway for Figures 6.9 and 6.15.
Dr G. Gutman, NOAA, for Plate 3.
Dr A.R. Hansen for Figure 4.45.
Dr E. Harrison for Table 3.4.
Dr I. Held, NOAA, for Figure 4.24.
Professor B.J. Hoskins for Figures 4.13 and 4.14.
Dr L.T. Julian, NOAA, for Figure 4.35.
Dr E.F. LeDrew, University of Waterloo, for Figure 6.6.
Dr L. McMudie and Dr A.M. Carleton for Plates 7b–7f.
Dr P. Niiler for Figure 5.11.
Professor A. Ohmura, Swiss Federal Institute of Technology, Zurich, for Figures 1.1, 3.1, and 3.2.
Professor A. Oort for Figures 3.12 and 3.16.
Professor C.S. Ramage for Figures 3.58, 3.60, 3.61.
Professor B. Saltzman, Yale University, for Figure 1.2.
Dr A. Sutera, Universita Di Camerino, Italy, for Figure 4.44.
Dr K.E. Trenberth, NCAR, for Figure 5.6.
Professor J.M. Wallace, University of Washington, for Figures 4.43 and 5.2.
Acknowledgments xv
11 Professor T. Webb III, Brown University, for Figure 1.3.
Professor D.S. Wilks, Cornell University, for Figure 2.8.
Dr J. Woods for Figure 1.4.
Every effort has been made to contact copyright holders for their permission to reprint mate-
rial in this book. The publishers would be grateful to hear from any copyright holder who is
not here acknowledged and will undertake to rectify any errors or omissions in future editions
of this book.
111
0111
11
1
1
11
Part 1
The climate system
and its study
0
111
0111
11
1
1
11
1 Introduction
1.1 The global climate system

111
Climate can be defined as “the synthesis of weather” considered over a time interval long
enough to determine its essential statistical properties. More broadly, it is the time-aver-
aged state of the physical system that involves the atmosphere, hydrosphere, cryosphere,
lithosphere and biosphere (Bolle, 1985) and their interactions on many different time and
space scales (Figure 1.1). The main domains that comprise the climate system have a
wide range of equilibration times, as shown schematically in Figure 1.2, illustrating the
0
many time scales involved. Internally, the climate system as a whole is a closed system
for exchanges of matter but is subject to external forcing by solar radiation, gravitational
forces, geological processes, and human activity. The whole complex constitutes a
“cascading system,” or chain of subsystems, interconnected by flows of energy, matter,
and momentum.
The spatial dimensions of the components of the climate system are indicated in Figure
1.3. The atmosphere, with a representative thickness of 10 km, is a thin skin of gases
rotating with the earth. Ninety-nine percent of the atmosphere’s mass is in the lowest
30 km. Air is compressible and so its density () decreases nearly exponentially with
0111
11 Figure 1.1 The climate system. (From Hutter et al., 1990)

4 Synoptic and dynamic climatology
Figure 1.2 Schematic summary of the domains of the climate system, showing approximate times
required for equilibrium to be re-established after a perturbation is imposed at the
boundary of each subsystem. (From Saltzman, 1983)
altitude, owing to gravity (g). The essential balance between them is expressed in the
hydrostatic relationship1 describing the decrease of pressure (p) with altitude (z). The
processing of solar radiation by the atmosphere and the earth’s surface as the fundamental
driver of the climate system is treated briefly in Chapter 3. More important for under-
standing the variability of the climate system are the internal variables that undergo
seasonal changes at the earth’s surface. Accordingly, it is useful to outline the spatio-
temporal patterns of these basic components of the climate system.
1.1.1 Land surface

Land areas represent about 30 percent of the earth’s surface, of which about one-quarter
is largely unvegetated. The soil layer plays an important role in the surface energy balance
through its albedo, thermal properties, and moisture content. The lithosphere is essentially
constant other than on geological time scales, with the exception of inputs of dust, volcanic
particles, and gases into the atmosphere, and the weathering of carbonate rocks, which
removes carbon from them. However, large-scale controls on surface–atmosphere inter-
action are exerted by the elevation of the land surface, and the orientation and extent of
Introduction 5
11
111
Figure 1.3 Spatial and temporal scales of variations in weather and climate. Short-term variations
are limited primarily to the atmosphere, but longer-term variations involve progres-
0111 sively more earth systems. Each bubble encloses a group of related types of variations,
in which the shorter-term, smaller-area features are part of the longer-term, larger-area
features. (From McDowell et al., 1990)
mountain ranges and plateaus. Most of the land masses are in the northern hemisphere;
Table 1.1 underlines the dominance of oceans in middle latitudes of the southern hemi-
sphere. Plate 1 depicts the major features of the land surface which have significant effects
on the atmospheric circulation and, in turn, on climatic variables such as cloudiness,
precipitation, and wind. Note that in the Americas, the Cordilleran ranges are essentially
0 perpendicular to the global wind systems of low and middle latitudes.
1.1.2 The hydrosphere

The bulk of the hydrosphere is in the oceans, although the small fraction in the atmos-
phere (as vapor, liquid water droplets, or ice crystals) is a prime determinant of weather
processes through clouds and precipitation. The phase changes involved in evapora-
tion/condensation play an important role in energy transfers. It is also important to note
that terrestrial temperatures are generally close to the triple point of water where the three
phases (vapor, liquid, and solid) may coexist in equilibrium. The six transitions – melting/
0 freezing, evaporation/condensation, and sublimation/deposition (or crystallization) – each
involve substantial cooling (heating) in passing from a lower to higher (higher to lower)
state, respectively. The latent heat parameters involved are, respectively, those of fusion
11 (0.335 J kg1), vaporization (2.5 J kg1) and sublimation (2.835 J kg1 at 0° C). Cloud
Table 1.1 Dimensions of the Earth
Latitude % of total Area of zone Length of % of zone % of zone

area of between two 10° longitude ocean, ocean,
hemisphere 10° circles (km) northern southern
located of latitude hemisphere hemisphere
poleward (106 km2)
90 0 3.9 0 90 0
80 1.5 11.6 193.9 70 27
70 6.1 18.9 381.9 30 91
60 13.5 25.6 558.0 43 99
50 23.5 31.5 717.0 48 97
40 35.9 36.4 854.0 57 89
1 30 50.1 40.2 964.9 63 77
20 65.9 42.8 1,046.5 74 78
10 82.7 44.1 1,096.4 77 76
0 100 1,113.2
Mean (%) 60.7 80.9
Total
(106 km2) 255 154.8a 206.3a
Source: after List (1958).
Note
a
The ocean area totals 361 106 km2 out of a global surface area of 510 106 km2.
reflectance contributes the major part of the planetary albedo and its net effect on the
global energy balance is one of cooling. The terrestrial components rivers, lakes, and
groundwater are key elements in the hydrological cycle, and affect local and regional
climates also.
1 The hydrological reservoirs and the water transfers between them are still poorly known.
Table 1.2 summarizing recent estimates illustrates the predominance of the ocean surfaces
in water exchanges and storage (97 percent of total water). However, land ice represents
the largest reservoir of freshwater, 76 percent of the total. There is a net imbalance over
land (Table 1.2B) as a result of uncertainties in the estimates and possible trends in water
storage in glaciers and groundwater.
The oceans cover 71 percent of the earth, but this extent differs greatly between hemi-
spheres and according to latitude (Table 1.1). As a consequence, the majority of the solar
radiation absorbed by the earth’s surface enters the oceans (Figure 1.4). The direct solar
heating of the oceans averages almost four times that of the land areas. The temperatures
at the ocean surface vary mainly with latitude in the southern hemisphere, but in the
northern hemisphere the major wind-driven current gyres in the North Atlantic and North
Pacific oceans (Figure 1.5) create significant longitudinal contrasts in climate through their
effects on heat and moisture fluxes, on weather systems, and on the large-scale mean
atmospheric circulation.
The oceans are more strongly stratified than the atmosphere, owing to their high density.
Their large heat capacity also creates great thermal inertia so that they serve as a buffer
for changes of temperature, gaseous transfers (such as CO2) etc. The upper ocean, of the
order of 100 m thick, is the most active part, closely interconnected with the atmosphere,
but with a relaxation time of weeks to months. The deep ocean, below the thermocline
(the zone of sharp temperature gradient marking the penetration of annual heating/cooling)
is largely isolated from the surface layers, except on time scales of 102–103 year.
Links between the ocean and atmosphere involve complex feedbacks on a wide range
of time and space scales which are discussed in subsequent chapters. However, their
Introduction 7
11 Table 1.2 Reservoirs and transports involved in the hydrological cycle
A Principal reservoirs Water volume (km3)

Oceans 1,350.0 106
Ice sheets and glaciers 32.4 106
Groundwater (to 4 km depth) 8.2 106
Inland seas 105,000
Lakes and reservoirs 140,000
Atmosphere 130,000
0
B Annual transports (km3) Over oceans Over land
Precipitation 385 111

111 Evaporation 425 71
Runoff
Rivers 27
Groundwater 12
Glaciers 2.5
Sources: after Speidel and Agnew (1982), Baumgartner and Reichel (1975), van der Leeden et al. (1990).
0
0111
Figure 1.4 The zonally averaged distribution of energy incident on the top of the atmosphere in
one year, showing the partition between the continents (black) and the oceans (shaded)
0 in each 5° band. More than half of the energy entering the earth’s climate system is
first absorbed inside the oceans. (From Woods, 1984)
11
1
1
8 Verso running head
Figure 1.5 Major global ocean current systems in February–March. (From Wells, 1986)
Introduction 9
11 fundamental importance can be illustrated by noting that the cooling by 0.1°C of a meter-
thick layer of water is sufficient to raise the temperature of an overlying air layer 30 m
deep by 10°C, through turbulent heat transfer.
1.1.3 The cryosphere

Snow and ice features are collectively referred to as the cryosphere. Ice sheets grow and
decay over 104–105 year intervals, causing lowering/raising of the global sea level by tens
of meters to over 100 m. However, snow cover and sea ice are annually cyclical compo-
0 nents which markedly affect the surface absorption of solar radiation in high and middle
latitudes, owing to their high reflectivity (or albedo). Sea ice also insulates the ocean
surface from the atmosphere, eliminating the flux of moisture by evaporation and greatly
reducing direct heat transfer. Ninety percent of the ice volume is contained in the Antarctic
111 ice sheet but, in terms of areal extent, snow and sea ice cover almost a quarter of the
northern hemisphere in winter, with snow covering half of the land surface in February
(Plate 2).
The principal climatic roles of snow and ice relate to their high reflectivity, low thermal
conductivity, which insulates the underlying ground or ocean, and their thermal inertia
effects. Typical integrated albedo values for the spectral range of solar radiation are 0.8–0.9
0 for fresh snow, 0.6 for bare ice and 0.3–0.4 for melting sea ice with puddles. However,
snow-covered forest areas may have albedos of only 0.25–0.40. Snow cover develops a
cold reserve in winter that stabilizes the lower atmosphere overlying it and depresses
temperatures by 5°–10°C until sufficient energy is absorbed in spring to raise the snow-
pack temperature to 0°C and then to melt it (1.8 106 J kg1).
1.1.4 The biosphere

The biosphere (both terrestrial and oceanic) is important through its effects on exchanges
of energy, moisture, and matter. Land vegetation affects surface albedo, roughness, rain-
0111 fall interception, soil moisture, evaporation, and runoff, for example (Dickinson, 1983).
Photosynthesis and respiration play a major role in the carbon cycle, and the biosphere
is also involved in other significant biogeochemical cycles that affect atmospheric compo-
sition and photochemical processes. From a meteorological perspective, surface properties
(vegetation, agricultural use, soils) are poorly defined. In particular, the open or closed
character of the canopy and its height are of prime importance (Graetz, 1991). Globally,
about 70 percent of canopies are open and 30 percent closed. Maps of global vegetation
types fail to show agricultural land use or the seasonal variation in the biosphere but
specialized archives have been devised for use in modeling (Mathews, 1985; Wilson and
Henderson-Sellers, 1985). Satellite remote sensing using visible and infrared wavelength
0 sensors can now be used to map a “normalized difference vegetation index” (NDVI) on
a regular basis (Justice et al., 1985).
(RNIR Rvis)
NDVI =
(RNIR Rvis)
where R radiance in the visible (vis) channel, 0.55–0.68 m, and near-infrared (NIR)
channel, 0.73–1.1 m.
This is a measure of photosynthetic capacity, although the precise interpretation
remains to be specified. NDVI ranges from <0.2 for deserts to >0.5 for green canopies.
0 The seasonal characteristics of NDVI are illustrated in Plate 3 from five-year averaged
data (April 1985–87 and March 1989–91). For July the standard deviation is included and
this identifies areas of high interannual variability in northeast Brazil, central Siberia
11 and southeast Australia. Other attempts have been made to assemble land use and soil
information from atlases and other maps for use in climate models, but such global data
sets are recognized to be heterogeneous in their information content.
1.2 Time and space scales of weather and climate processes

The time and space scales of atmospheric motion range from seconds to weeks (and
longer), and from essentially random, small-scale turbulence to large-scale horizontal
eddies (weather systems) and the slowly varying global circulation features. Kinetic energy
spectra based on station wind data (e.g. Vinnichenko, 1970) indicate that, apart from
diurnal and annual cycles, the major peak is in the range five to thirty days, associated
with synoptic-scale atmospheric systems. There is a further peak around one minute,
caused by small-scale turbulence in the boundary layer. A clear spectral gap exists between
these peaks (Figure 1.6).
1 Space and time scales of atmospheric motion are closely related. Figure 1.7 illustrates
the full spectrum of motion systems and delimits the scales with which we are concerned.
The principal ones are:
1 Irregular large-scale fluctuations of longer duration and recurrence interval than the
annual cycle (e.g. ENSO events).
2 Seasonal fluctuations of major wind systems (e.g. monsoon systems).
3 Persistent large-scale circulation regimes (e.g. blocking).
4 Planetary waves.
5 Synoptic systems.
6 Subsynoptic cloud clusters and mesoscale features (as they relate to the synoptic and
larger-scale systems).
Figure 1.6 Spectrum of atmospheric kinetic energy. The abscissa shows frequency (log f in day1)
and the ordinate is f s2( f ) in m2 s2, where s2(f) is the explained variance; the area
under the curve is equal to the total variance. (After Vinnichenko, 1970, from Peixoto
and Oort, 1992)
Introduction 11
11 Figure 1.7 The characteristic time and length scales of atmospheric processes. See text. (Courtesy
B.W. Atkinson, 1984)
Descriptive scaling of atmospheric motion is usually made in terms of characteristic hori-

zontal dimensions, or wave number relative to planetary circumference (~40,000 km), and
time periods as illustrated in Figure 1.8. A dynamic basis using three fundamental frequen-
cies is also possible (Atkinson, 1984). These are:
0
1 The inertial frequency due to the earth’s rotation: f 2 sin ~104 s1; latitude,
angular velocity of the earth (see Figure 1.9).
2 The planetary frequency associated with the latitudinal variation of the Coriolis para-
111
meter (the “beta effect”):
P (U )1/2 ~ 106 s1
where
f/
y, f the Coriolis parameter, U horizontal air velocity.
0 3 The Brunt–Väisälä frequency for vertical oscillations:
N= 冢g ∂∂z 冣 ~ 10 2 1
s
where potential temperature, g gravitational acceleration, and z altitude.

N is a measure of the static stability of the atmosphere.
0111
Figure 1.8 Time and space dimensions of atmospheric weather systems and climatic regimes. (After
Mason, 1970, and Barry and Perry, 1973)
Figure 1.9 Important parameters and variables relating to the earth’s atmosphere and its motion.
(Courtesy B.W. Atkinson, 1984)
Introduction 13
11 These frequencies can be used to distinguish four scales of motion:
1 Small scale, where F (the atmospheric frequency) > N.

2 Mesoscale, f < F < N.
3 Synoptic scale, P < F < f.
4 Planetary scale, F < P.
According to classical views of atmospheric motion, kinetic energy cascades from the
largest scale to the smallest in a dissipative process. However, research on the general
0 circulation and its transports of momentum and energy in the 1950s showed this view to
be incorrect. Instead, the eddies in the circulation, especially those in a horizontal plane
(i.e. upper-level waves, surface cyclones and anticyclones), were found to be essential to
the maintenance of the hemispheric zonal flows. Indeed, the eddies transfer momentum
111 and energy to the larger-scale motion, rather than feeding off it. This phenomenon, termed
“negative viscosity” by Starr (1968), is a general characteristic of planetary circulation.
Further details are discussed in Chapter 3. Here it is important to note that this result
underlines the fundamental role played by synoptic systems and planetary waves in the
global circulation and climate. Indeed, it provides the rationale for the focus of this book
on circulation regimes, planetary waves, and synoptic systems.
0
1.3 Dynamic and synoptic climatology
The term “climate” is usually understood to imply the ensemble of events at a location,
or over a homogeneous area, as represented by the mean (or modal) value and appro-
priate statistics of the frequency distributions of weather elements. Bryson (1997) proposes
the axiom: “Climate is the thermodynamic/hydrodynamic status of the global boundary
conditions that determine the concurrent array of weather patterns.” Thus climate theory
should be concerned with the boundary conditions, especially fluxes at the surface, in
Bryson’s view. Surface fluxes are treated in micrometeorology, boundary-layer meteo-
0111 rology, and on a regional to global scale, in physical climatology, as exemplified by
Budyko’s (1956) Atlas of the Heat Budget. However, the direct linkages between
surface fluxes and weather patterns are only part of the story. Climates are also deter-
mined by the combined effects of the global atmospheric and oceanic circulations,
planetary waves, synoptic and smaller-scale weather systems and their interactions. This
book addresses these dynamical elements of climate by treating two major components
of climatology.
Dynamic climatology (now frequently termed “climate dynamics”) is the study of the
global climate system in terms of its origin and maintenance. Hare (1957) defined dynamic
climatology as “the explanatory description of world climates in terms of the circulation
0 or disturbances of the atmosphere.” In the 1930s, however, Hesselberg (1932) stated,
“dynamic climatology must be concerned with the quantitative application of the laws of
hydrodynamics and thermodynamics . . . to investigate the general circulation and state
of the atmosphere, as well as the average state and motion for shorter time intervals.”
Thus it involves the understanding of the dynamic and thermodynamic controls of the
mean features of the large-scale circulation (Marotz, 1987) and their variability on monthly
to interannual time scales. The approaches to these questions are empirical, theoretical,
and model-based (Smagorinsky, 1981; Peixoto and Oort, 1992; Grotjahn, 1993). A history
of the origin of the term “dynamic climatology” proposed by Tor Bergeron in 1930 is
provided by Raynor et al. (1991). However, there has been an explosion of ideas and
0 information since the 1960s and most of this remains unassimilated in climate text-
books.
Synoptic climatology examines the relationship of local and regional climatic condi-
11 tions to the atmospheric circulation. Synoptic meteorological data are used to categorize
selected characteristics of the atmospheric circulation and associated weather phenomena.
The historical evolution of synoptic climatology is treated in detail elsewhere (Barry and
Perry, 1973; see also Harman and Winkler, 1991).
These two aspects of climatology have developed rapidly since the Second World War
through extraordinary developments in computing and modeling, as well as through the
increased availability of upper air data and, since the 1960s and 1970s, of satellite data.
In the last decade or so, many aspects of synoptic and mesoscale meteorology have been
transformed by the use of new measurement technologies, including radar and lidar
profiling of the lower troposphere, mesoscale networks of automatic stations and improved
instrumentation generally. Combined with high-resolution multichannel satellite data,
through new image display and visualization techniques, this wealth of information has
revealed the detailed structure of frontal cyclones, mesoscale convective systems, trop-
ical waves and cyclones, and boundary-layer structure for research purposes, in addition
1 to providing input to “nowcasting” services (Browning and Szejwach, 1994; Conway et
al., 1996). While it is perhaps too soon and, in any case, impracticable to assimilate fully
all of these ideas here, an effort is made to illustrate the additional levels of space–time
data and models now available and to incorporate these findings into the discussion where
appropriate.
1.4 The structure of the book

The book opens with a discussion of data sources for large-scale climatological studies
from conventional weather map and geopotential height data and remote sensing prod-
ucts and data analysis techniques. The global atmospheric circulation controls and the
characteristics of global wind belts and pressure contours are treated in Chapter 3. Then
the planetary waves and circulation modes are examined in depth, followed by a discus-
sion of global teleconnections and their forcings. Chapter 6 treats synoptic-scale systems
in the extratropics and the tropics, and this is followed by a discussion of modern synoptic
climatology and its applications.
1
Note
1 The hydrostatic equation ∂p/∂z g where p pressure, z altitude, g acceleration
due to gravity and air density, expresses the approximate balance in the atmos-
phere between gravitational acceleration (downward) and the vertical pressure gradient
(upward).
References
Atkinson, B.W. 1984. The Mesoscale Atmosphere. Inaugural lecture, Queen Mary College, Univer-
sity of London, 30 pp.
Barry, R.G. 1970. A framework for climatological research with particular reference to scale
concepts. Trans. Inst. Brit. Geog., 49: 61–70.
Barry, R.G. and Perry, A.H. 1973. Synoptic Climatology: Methods and Applications. Methuen,
London, 555 pp.
Baumgartner, A. and Reichel, E. 1975. The World Water Balance. Elsevier, Amsterdam, 179 pp.
Bolle, H.J. 1985. What is climate? In: G. Ohring and H.J. Bolle, eds, Space Observations for
Climate Studies, Adv. Space Research, 5 (6): 5–14.
Browning, K.A. and Szejwach, G. 1994. Developments in operational systems for weather fore-
casting. Met. Applications, 1: 3–22.
Bryson, R.A. 1997. The paradigm of climatology: an essay. Bull. Amer. Met. Soc., 78 (3): 449–56.
Budyko, M.I. 1956. The Heat Balance of the Earth’s Surface (trans. N.I. Stepanova), US Weather
Bureau, Washington DC, 255 pp.
Introduction 15
11 Conway, J., Gerard, L., Labrousse, J., Liljas, E., Senesi, S., Sunde, J., and Zwatz-Meise, V. (eds).
1996. COST 78. Meteorology Nowcasting: a Survey of Current Knowledge, Techniques and
Practice. European Commission, Directorate General XII (EUR 16861 EN), Brussels, 499 pp.
Dickinson, R.E. 1983. Land surface processes and climate–surface albedos and energy balance. Adv.
Geophys., 25: 305–53.
Graetz, R.D. 1991. The nature and significance of the feedback of changes in terrestrial vegetation
on global atmospheric and climatic change. Climatic Change, 18: 147–73.
Grotjahn, R. 1993. Global Atmospheric Circulations: Observations and Theories, Oxford University
Press, Oxford, 430 pp.
Gutman, G., Tarpley, D., Ignatov, A., and Olson, S. 1995. The enhanced NOAA global land data
0 set from the Advanced Very High Resolution Radiometer. Bull. Amer. Met. Soc., 76 (7): 1141–56.
Hare, F.K. 1957. The dynamic aspects of climatology. Geogr. Annal., 39: 87–104.
Harman, J.R. and Winkler, J.A. 1991. Synoptic climatology: themes, applications and prospects.
Phys. Geogr., 12: 220–30.
111 Hesselberg, T. 1932. Arbeitsmethoden einer dynamischen Klimatologie. Beitr. Phy. f. Atmos., 19:
291–305.
Hutter, K., Blatter, H., and Ohmura, A. 1990. Climatic Changes, Ice Sheet Dynamics and Sea Level
Variations. Zürcher Geogr. Schriften, 37, ETH, Zürich, 82 pp.
Justice, C.O., Townshend, J.R.G., Holben, B.N., and Tucker, C.J. 1985. Analysis of the phenology
of global vegetation using meteorological satellite data. Intl. J. Rem. Sensing, 8: 1271–318.
List, R.J. (ed.). 1958. Smithsonian Meteorological Tables. Smithsonian Misc. Collections 114, 6th
0 edition, Smithsonian Institution Press, Washington DC, pp. 483–4.
Marotz, G.A. 1987. Dynamic climatology. In: J.E. Olivier and R.E. Fairbridge, eds, The
Encyclopedia of Climatology, Van Nostrand Reinhold, New York, pp. 395–403.
Mason, B.J. 1970. Future developments in meteorology: an outlook to the year 2000. Quart. J. Roy.
Met. Soc., 96: 349–68.
Mathews, E. 1985. Atlas of Archived Vegetation, Land-use and Seasonal Albedo Data Sets. NASA
Tech. Mem. 86199.
McDowell, P.F., Webb, T. III, and Bartlein, P.J. 1990. Long-term environmental change. In: B.L.
Turner II, W.C. Clark, R.W. Kates, J.F. Richards, J. Mathews and W.B. Meyer, eds, The Earth
as Transformed by Human Actions, Cambridge University Press, Cambridge, pp. 143–62.
Peixoto, J.P. and Oort, A.H. 1992. Physics of Climate, American Institute of Physics, New York,
0111 520 pp.
Raynor, J.N., Hobgood, J.S., and Howarth, D.A. 1991. Dynamic climatology: its history and future.
Phys. Geog., 12: 207–19.
Saltzman, B. 1983. Climatic systems analysis. In: B. Saltzman, ed., Theory of Climate. Adv.
Geophys., 25: 173–233.
Smagorinsky, J. 1981. Epilogue: a perspective on dynamical meteorology. In: B.W. Atkinson, ed.,
Dynamical Meteorology: An Introductory Selection, Methuen, London, pp. 205–19.
Speidel, D.H. and Agnew, A.F. 1982. Water: The Natural Geochemistry of our Environment.
Westview Press, Boulder CO.
Starr, V.P. 1968. Physics of Negative Viscosity Phenomena, McGraw-Hill, New York, 256 pp.
van der Leeden, F., Troise, F.L., and Todd, D.K. 1990. The Water Encyclopedia, 2nd edition. Lewis,
0 Chelsea MI, 808 pp.
Vinnichenko, N.K. 1970. The kinetic energy spectrum in the free atmosphere – 1 second to 5 years.
Tellus, 22: 158–66.
Wells, N. 1986. The Atmosphere and Ocean: A Physical Introduction. Taylor & Francis, London
and Philadelphia, 347 pp.
Wilson, M.F. and Henderson-Sellers, A. 1985. A global archive of land cover and soil data sets
for use in general circulation climate models. J. Climatol., 7: 319–43.
Woods, J.D. 1984. The upper ocean and air–sea interaction in global climate. In: J.T. Houghton,
ed., The Global Climate, Cambridge University Press, Cambridge, pp. 141–87.
11
2 Climate data and their analysis
1
2.1 Synoptic meteorological data
2.1.1 Surface reports
Standard weather observations include instrumental measurements of air temperature,
dewpoint temperature, station pressure (adjusted to mean sea level), wind speed and direc-
tion, pressure change over the last three hours, and pressure tendency; also, visual
observations are made of cloud amount, type, and cloud base height (for low, middle,
and high cloud layers), visibility, and present and past weather. In addition, precipitation
amounts are recorded six-hourly and snow depth once a day. Each element is reported
in the international synoptic code (Stubbs, 1981; World Meteorological Organization,
1995). Observations are made at synoptic weather stations at 00.00, 00.06, 12.00 and
18.00 hours UTC (Universal Coordinated Time based on Greenwich 0° meridian) and
collected at international centers. Under the World Weather Watch program, synoptic
reports are made worldwide at about 4,000 land stations and by 7,000 ships (Figure 2.1).
Ships also report “sea surface” temperature (nowadays usually engine room intake temper-
ature), sea state, and, if present, sea ice conditions.
1 Observations of surface weather with primitive instruments began in various European
countries in the mid-seventeenth century. However, the establishment of networks of
weather stations using standard instruments and procedures largely followed on the heels
of the expansion of telegraphy in the 1850s and 1860s and the organization of national
weather services between the 1840s and the 1880s (Khrgian, 1970, p. 137; Fleming, 1990,
p. 141). Following the first International Meteorological Congress in Vienna in 1873, an
international (“Utrecht”) weather code was adopted in European countries (excluding
Holland, Portugal, Spain, and Turkey) and in Russia, but not the United States, in 1875.
Cloud types, weather, and visibility were not included in international codes, however,
until 1921.
2.1.2 Upper-air reports

Upper-air soundings by instrumental kites and balloons began in the 1890s and high alti-
tude temperature measurements led to the discovery of the isothermal zone or inversion
around 12 km by Teisserenc de Bort, and its confirmation by R. Assmann, in 1902 (Hoinka,
1997). Pilot balloon measurements of winds aloft began in the early 1900s, followed in
the 1920s by instrumented aircraft soundings. Radiosondes were first developed in the
1930s in Russia and the United States, but standardized, calibrated measurements and
radar tracking became widespread only in the late 1940s. Networks in China and India
were established only after 1945. About 900 stations make upper-air soundings of temper-
ature, pressure, humidity, and wind (at 00.00 and 12.00 UTC, or in some countries only
once daily). Errors in temperature measurements have been reduced from 0.7–1.0°C to
0.2–0.5°C in recent years. Radar wind measurements are within 1–2 m s1 and 5–10° for
Climate data and their analysis 17
11
111
0111
Figure 2.1 The distribution of synoptic reports from land stations and ships (above) and upper-air
soundings (below) available over the Global Telecommunications System at the National
Meteorological Center, Washington, DC, for January 1996. There are 7,829 surface
0 and 1,002 upper-air stations shown (Courtesy National Center for Atmospheric
Research, Boulder CO)
direction, and height determinations are within 100–50 m up to 50 mb. However, relative
humidity data still have large departures, especially at the low temperature and humidity
levels characteristic of the upper troposphere (Elliot and Gaffen, 1991; Zaitseva and
Ivanov, 1998). The effects of changes in sensor and corrections applied to the data cause
significant temporal inhomogeneities in station records, especially in the stratosphere
(Gaffen, 1994).
0
2.2 Remotely sensed data
The inhomogeneities of the global surface and upper-air reporting network (section 2.1)
11 are alleviated through the acquisition of satellite remotely sensed data on surface and
atmospheric temperatures, cloud parameters, precipitation occurrence, and estimated
rain rates, surface conditions (e.g. snow cover, sea ice, vegetation, soil moisture, ocean
phytoplankton abundance), and atmospheric gaseous abundances and particulate concen-
trations (especially O3, CO2, water vapor, dust, soot) (Carleton, 1991). At the synoptic
level, the satellite cloud image is the primary data source (Barrett, 1970; Salby et al.,
1991), with supplementary information on temperatures, moisture, and radiation provided
by vertical sounding of the atmosphere at mesoscales (Claud et al., 1991, 1992, 1995;
Carleton et al., 1995). Only certain portions of the electromagnetic spectrum (EMS) are
used routinely in satellite remote sensing of the Earth’s atmosphere and surface. The
wavelengths selected for the retrieval of target information are determined, primarily,
by the spectral, spatial, and temporal attributes of the climatic phenomenon under consid-
eration (cf. clouds, land cover, SSTs) (Smith et al., 1986; Yates et al., 1986). Accordingly,
multispectral remote sensing, involving the use of several narrow wavelength bands, is
1 usually superior to bi-spectral and particularly broad-band uni-spectral methods for
discrimination of targets at the Earth’s surface and in the atmosphere (e.g. clouds). This
is also the basis of the satellite retrieval of vertical temperature and moisture profiles,
such as those from the NOAA/TOVS (TIROS Operational Vertical Sounder) and GOES
VAS (VISSR: Visible-Infrared Spin Scan Radiometer, Atmospheric Sounder). Moreover,
narrow-band remote sensing better permits the removal of “contamination” from sources
at the Earth’s surface or in the atmosphere that is necessary to retrieve accurate infor-
mation (atmospheric water vapor patterns, snow cover discriminated from clouds) (Kidder
and Wu, 1984; Hutchison and Locke, 1997).
Satellite remote sensing for synoptic and dynamic climatology has traditionally involved
the use of bands in the visible: VIS (0.4–0.7 m) and thermal infrared “window”: IR
(8–14 m) regions of the EMS, available from meteorological polar orbiting platforms
since the early 1960s (Bugaev, 1973; Barrett, 1974, 1987). This was supplemented by
sensors in the infrared absorption band at around 6.5 m, used to retrieve the mid-tropos-
pheric water vapor; initially from geosynchronous satellites (Eyre, 1981; Schmetz and
Turpeinen, 1988). More recently, information on precipitation rates, the occurrence of
1 solid versus liquid precipitation in convective situations, the near-surface wind speed and
the column-integrated water vapor and cloud liquid water over ocean areas, has become
available operationally from passive microwave sensors such as the DMSP (Defense
Meteorological Satellite Program) SSM/I (Special Sensor Microwave/Imager) and the
European Remote Sensing (ERS-1) satellite (Goodberlet et al., 1989; Rabin et al., 1991;
Tjemkes et al., 1991; Claud et al., 1992; Bauer and Schluessel, 1993; Liu and Curry,
1993; Rao and MacArthur, 1994; Weng and Grody, 1994; Siefridt et al., 1998). These
operational products build upon the advances made in oceanic remote sensing using active
and passive microwave sensors, by NASA’s Seasat and Nimbus-7 platforms; specifically
the SASS (Seasat-A Satellite Scatterometer) and SMMR (Special Sensor Microwave
Imager) (Alishouse, 1983; Gloersen et al., 1984; Prabhakara et al., 1983; McMurdie
et al., 1987; Katsaros et al., 1989).
2.2.1 History
The era of routine satellite remote sensing of clouds and weather systems began on April
1 1960 when the Television and Infrared Observing Satellite (TIROS-1) was successfully
launched, providing visible and infrared band images over much of the globe. These essen-
tially experimental systems continued through 1965, when they were succeeded by the
TIROS Operational Satellites (TOS), known as ESSA 1–9 (for the Environmental Sciences
Services Administration – a forerunner of NOAA), during 1966–69 (Figure 2.2). These
satellites were all in near-polar orbit, initially at about 750 km altitude and later around
1400 km, making approximately twelve orbits per twenty-four hours. Since 1978 the
TIROS N series satellites have been at about 850 km (Smith, 1985). The later series were
0
0
0
0
11
11
0111
111
Figure 2.2 The history of meteorological satellites since 1960. (updated from Rao et al., 1990)
sun-synchronous, with one satellite having an equatorial crossing time about 09.00 local
solar time and another about 15.00. The data permitted valuable analyses to be made of
snow and ice extent in addition to information used to identify weather systems and map
cloud characteristics – so-called nephanalyses (Barrett, 1974; Matson et al., 1986). A
further major advance was marked in 1966 with the launch of the Geostationary Opera-
tional Environmental Satellites (GOES). Positioned at 60° intervals over the equator at
41,000 km altitude for the First GARP Global Experiment, December 1978–November
1979 (Mason, 1971), they provided views of the globe effectively between about 55°N
and 55°S at twenty-minute intervals.
Daily visible and infrared image mosaics in polar stereographic projections from the
polar orbiting satellite were published in an extensive series of monthly reports between
November 1972 and May 1984 (Environmental Data Service NOAA, 1972) and these
were at a scale suitable for use in many synoptic-climatological analyses. The Meteoro-
1 logical Satellite Center, Japan (1978–96) has published full disk visible and infrared images
centered at 135°E for 13.00 UTC daily since April 1978 as well as regional cloud analysis
charts for East Asia and tabulated cloud vectors. For Europe there is a NOAA-derived
daily image available from 1974 to the present (Institut für Meteorologie, Berlin). NOAA
initiated low-cost Automatic Picture Transmission (APT) to ground stations in 1963 and
High Resolution Picture Transmission (HRPT) in 1972 (Rao et al., 1990).
Technical advances, with improved spatial resolution of terrestrial and atmospheric
features, as well as channels in additional spectral wavelengths, evolved rapidly in the
1970s (Rao et al., 1990). The Polar Orbiting Environmental Satellites (POES) featured
the Improved TIROS Operational System (ITOS) satellites of NOAA during 1970–78,
and the complementary POES of the Defense Meteorological Satellite Program (DMSP)
which supplied broad-band visible and infrared imagery with global coverage at 2.7 km
resolution and local 0.6 km resolution read-out products archived from 1973 to 1990/91
(Scharfen et al., 1995); a digital archive was established by NOAA from 1993 (Kroehl
et al., 1994).
On the NOAA satellites, the Very High Resolution Radiometer (VHRR) replaced the
1 earlier scanning radiometer sensors in 1972 and the data were made available as Global
Area Coverage 4 km resolution products, with limited 1.1 km resolution Local Area
Coverage. Broadly comparable systems were operated on the Meteor series of the Cosmos
satellites of the former Soviet Union from 1969 although the images were not generally
available beyond the State Hydrometeorological Service (Massom, 1991).
The GOES system with the Synchronous Meteorological Satellites (SMS) provided
extensive coverage of the tropics during 1974–89 with satellites of NOAA at longitudes
75°W and 135°W, complemented by the European Meteosat at 0°, the Japanese
Geostationary Meteorological Satellite (GMS) at 135°E; the Indian INSAT was also
geostationary although not formally part of the international system. The TIROS N series
from 1978 to 1989 supplied not only visible and infrared channel data with the Advanced
Very High Resolution Radiometer (AVHRR), but also the TIROS Operational Vertical
Sounders (TOVS) yielding global information on the large-scale vertical temperature and
moisture structure (Susskind, 1993). From 1979 sea surface temperatures were derived
for cloud-free areas from AVHRR with an improved multichannel algorithm employed
from 1981 (see Njoku and Brown, 1993).
Following the success of the single-channel Electrically Scanning Microwave
Radiometer (ESMR) on a NASA Research and Development satellite (Nimbus 5), in
mapping sea ice year-round during 1973–76, the Scanning Multichannel Microwave
Radiometer (SMMR) provided invaluable new information for research on the annual
cycle of sea ice and its regional variations (Gloersen et al., 1992) as well as unique infor-
mation on precipitation over the oceans (Arkin and Ardanuy, 1989) and surface wind
velocity over the oceans (Atlas et al., 1993). SMMR operated during 1978–87 and its
11 data have been continued up to the present via the closely similar Special Sensor
Microwave Imager (SSM/I) on DMSP satellites (Barry, 1991).
The current series of NOAA polar orbiting satellites have continued the tried and tested
systems and added enhancements in spectral coverage with a near-infrared water vapor
channel (3.7 m and/or 1.6 m).
Apart from the many operational uses of satellite data, concerted efforts are finally
being made to generate consistent high-quality research data sets. An example of this is
an archive of five years of International Satellite Cloud Climatology (ISCCP) (Rossow,
1993) and Earth Radiation Budget Experiment (ERBE) data at the National Center for
0 Atmospheric Research, Boulder, Colorado (Hurrell and Campbell, 1992). A nationally
coordinated effort is under way in the United States through the NOAA–NASA Pathfinder
Program; this aims to assemble data sets from operational programs. The products include
an archive of AVHRR, Global Area Coverage (GAC) 5 km data products for 1982–97
111 and TOVS products for 1979–97. Part of this endeavor is focused on a suite of cryos-
pheric products for the polar regions (Barry, 1997). The polar products are all being
generated on a common Equal Area Scalable Earth (EASE) grid so that data from different
sensors and satellites can be directly overlaid. The grid accepts data of differing resolu-
tions in multiples/fractions of the basic 5 km format.
0
2.2.2 Significance
Satellite data are important to synoptic and dynamic climatology in three main areas:
1 They fill in the data-void areas between conventional reporting stations, especially
over the oceans. Figure 2.3 shows the typical coverage of a NOAA polar orbiting
satellite for the 12.00 UTC analysis. It may be compared with Figure 2.1 showing
the heterogeneity of the conventional synoptic network. In this way, pattern recog-
nition of the vortical cloud patterns associated with synoptic and subsynoptic cyclone
systems, as they appear on VIS and infrared data, offers clues to the associated ther-
0111 modynamic and dynamic processes (Evans et al., 1994; Carleton, 1995; Pankiewicz,
1995; Smigielski and Mogil, 1995; Forsythe and Vonder Haar, 1996). This is how
our conceptual models of organized cloud systems, especially frontal and tropical
cyclones, have evolved (Zillman and Price, 1972; Troup and Streten, 1972; Streten
and Kellas, 1973; Dvorak, 1975; Burtt and Junker, 1976; Junker and Haller, 1980;
Jaeger, 1984; Reed and Albright, 1997). Moreover, previously unidentified features,
such as Tropical-Extratropical Cloud Bands (TECBs) linking low and high latitudes
(Kuhnel, 1989, 1990), so-called “instant occlusions,” Mesoscale Convective Systems
(MCSs), cold-air mesocyclones and “polar lows,” and actiniform cloud patterns in
areas of subsidence were also discovered through satellite image analysis (Anderson
0 et al., 1969; Reed, 1979; Rasmussen, 1979, 1981; Maddox, 1980; Carleton, 1985;
Forbes and Lottes, 1985; McGinnigle, 1988, 1990; Augustine and Howard, 1991;
Laing and Fritsch, 1993a, b; Pearson and Stewart, 1994; Carleton, 1996). In most
situations, satellite data give significant improvements to the synoptic analysis and
prediction fields, as well as time-averaged fields, especially for the oceans of the
southern hemisphere (Salstein et al., 1987; Heckley et al., 1990; Anderson et al.,
1991; Keller and Johnson, 1992; Lamberty and Smith, 1993).
2 Satellites provide information to supplement that obtained from conventional meteo-
rological sources in data-rich areas, and permit the associations with synoptic features
to be determined (Zillman et al., 1990). Thus the outgoing long-wave (thermal) radi-
0 ation (OLR) fluxes show strong spatial variations on synoptic time and space scales,
and these are dominated by clouds (Cahalan et al., 1982). By temporal filtering of
the radiant flux variations, or setting threshold values of satellite-retrieved cloud-top
11
temperature (TCT), important information about the radiative and dynamic effects of
clouds, and the likeliest regions of convective precipitation can be obtained (Arkin
and Meisner, 1987; Rossow and Lacis, 1990; Ardanuy et al., 1991; Sohn and Smith,
1992; Thiao and Turpeinen, 1992; Rieland and Stuhlmann, 1993; Gupta et al., 1993).
Moreover, satellite sensors may be used to determine upper-tropospheric thermal
anomalies or to retrieve trace-gas abundances, particularly stratospheric ozone. These
show associations with synoptic features, notably jetstreams, and frontal and tropical
cyclones (Shapiro et al., 1982; Velden 1989, 1992; Bailey et al., 1993).
3 Satellite data can be used both to initialize the GCMs and LFMs used in numerical
weather prediction (NWP), to provide boundary conditions for climate modeling
using GCMs, and also to test the model output against observed data (Stoffelen and
Cats, 1991; Puri and Davidson, 1992). Examples of the satellite-retrieved data used
to provide boundary conditions are SSTs, soil moisture, snow cover and sea ice extent
1 (Ose et al., 1994; van den Hurk et al., 1997; Yamanouchi and Charlock, 1997).
Satellite data comparisons with model-generated fields include those of clouds, precip-
itation rates and OLR (Morcrette, 1991; Raustein et al., 1991; Janowiak, 1992; Mo
and Rasmussen, 1993).
Temporally, satellites are asynoptic (Salby, 1989) because of their nominal capability
of retrieving information over the entire orbit (for polar orbiting satellite) or scan period
(geosynchronous platform) (Hayden et al., 1996). In combination with surface-based radar
data, this makes them suitable for “nowcasting” the weather out to several hours (Menzel
et al., 1998). Satellite data are also a critical component of 4-D data assimilation in NWP.
When used to develop statistical “models” of cloudy circulation systems, only the satel-
lite data within a certain time window either side of the synoptic hour (e.g. ± three hours)
are typically used (Streten and Troup, 1973; Carleton, 1987; Song and Carleton, 1997).
The usable window is dictated by the space and time scales of the variable being consid-
ered. For example, moisture in the mid- and upper troposphere (600–300 mb), acquired
using the infrared absorption band at around 6.7 m, changes less quickly than that at
1 lower levels, which is dominated by the diurnal variations in energy budget and temper-
ature at the Earth’s surface and in the adjacent boundary layer (Chesters et al., 1983,
1987). The sensitivity of climate-scale studies to the satellite asynoptic data is greatest
when using polar orbiting platforms, and for phenomena having a strong diurnal cycle,
Figure 2.3 The typical coverage of a NOAA polar orbiting satellite for 12.00 UTC. (Dey, 1989)
11 such as clouds, surface temperature, and OLR (McGregor and Gorman, 1994; Salby and
Callaghan, 1997). The undersampling that occurs leads to aliasing (see section 2.5) in
time-averaged properties (Zeng and Levy, 1995). This can be reduced either by increasing
the frequency of the satellite observations (e.g. using data from geosynchronous platforms
rather than from one polar orbiter) or by ensuring that the satellite overpasses occur near
times when the diurnal cycle of the phenomenon can be better sampled. For example, the
Heat Capacity Mapping Mission (HCMM) sampled surface temperatures for thermal
inertia studies in the early to mid-afternoon, and again in the early morning. Also, the
Tropical Rainfall Measuring Mission (TRMM) satellite represents an attempt to accom-
0 modate the strong diurnal cycle of convective precipitation in the tropics (Simpson et al.
1988, 1998; Bell and Reid, 1993). For monitoring rapidly changing phenomena at higher
latitudes, particularly the development of “polar lows,” Kidder and Vonder Haar (1990)
advocate the wider application of the Molniya-type orbit of satellites launched by the
111 former Soviet Union. The Molniya orbit is highly eccentric and inclined steeply to the
Earth’s equator, thereby providing the equivalent of a “polar geosynchronous” perspec-
tive for one hemisphere during the approximately eight-hour time period of orbital apogee.
A further contrast between conventional synoptic data and satellite observations lies in
the spatially aggregated nature of the satellite retrievals. Conventional synoptic data are
for point observations, whereas those from satellites are areal averages determined by the
0 sensor spatial resolution, and where the pixel (“picture element”) is the fundamental unit
of the remotely sensed information. The nominal scale represented by the pixel (or instan-
taneous field of view) varies according to the intensity of the radiation that is being sensed
(see below); thus higher spatial resolutions are possible when sensing reflected solar radi-
ation than for emitted thermal infrared, and both are greater than the passive microwave
wavelengths, which have nominal “footprints” on the order of 750 km2. Studies cali-
brating satellite and conventional data on cloud amount, precipitation, surface insolation
receipt (the surface solar irradiance), and SSTs suggest that the two types of observing
technique show the highest correspondence when the conventional observations are aver-
aged over scales of approximately 50 km2 (Barrett and Grant, 1979; Henderson-Sellers
0111 et al., 1981; Njoku, 1985; Griffith, 1987). Thus for example, surface and satellite-retrieved
total cloud amounts show closest correspondence for clear and cloudy skies, but differ
by only about 1 okta when averaged over all sky conditions (Henderson-Sellers et al.,
1987). There is also a “trade-off” between the maximum resolution that can be achieved
and the areal coverage, given by the swath width of a polar orbiter. The swath width
influences how frequently the same location on the Earth’s surface is revisited and, thus
is a component of the temporal undersampling problem (Zeng and Levy, 1995; Salby and
Callaghan, 1997). One may contrast the imaged area acquired by VIS/IR sensors on board
meteorological polar orbiting satellites (approximately 1,000 km altitude) with that
acquired by the higher-resolution sensors on board Earth resources satellites such as
0 Landsat and SPOT (Système probatoire pour l’observation de terre), which are at 185 km
and 60 km altitude, respectively. This means that meteorological polar orbiters, particu-
larly the NOAA AVHRR with its nominal pixel resolution of 1.1 km, revisit a given
location twice per day (twelve hours apart); more frequently at higher latitudes owing to
the reduced Earth surface area. The latter permits the derivation of the cloud-drift winds
(for wind speed and direction estimates) associated with synoptic-scale and subsynoptic
systems in polar regions, over time scales of several hours (Turner and Warren, 1989;
King and Turner, 1997).
0 2.2.3 Principles of satellite remote sensing and applications

The fundamental information obtained by a satellite remote sensing system, in common
with the physical basis for climate, is the radiation interaction in a given wavelength, or
11 group of wavelengths, with targets at the Earth’s surface and in the atmosphere. However,
unlike the “conventional” retrieval of the radiation and energy balance from known infor-
mation on temperature and absorption (due to atmospheric gases), the problem is the
reverse in remote sensing; that is, temperature and moisture characteristics have to be
retrieved from the radiation fields by so-called inversion techniques. Inverting the satel-
lite radiances to yield climatic properties such as surface temperature, precipitation rate,
snow depth, layer-averaged water vapor, or near-surface wind speed, can be handled using
either physically based modeling, or empirical methods derived from regressing the satel-
lite retrievals with conventional data, or a mixture of the two (Petty, 1994a, b).
The utility of satellite remotely sensed data for synoptic and dynamic climatological
studies is predicated on the underlying physical principles, particularly the radiation laws.
There is an inverse relationship between the radiation wavelength (), expressed in
micrometers (m: 1 106 m) and nanometers (1 nm: 1 109 m), and the frequency
(v) which represents the number of wave peaks passing a fixed point per unit time, or:
1
c v (1)
where c the speed of light (2.99792 10 m s ).
8 1
One commonly used measure of v is the gigaHertz (GHz), or 1 109 s1. Whereas the
wavelength (in m) is used typically to denote radiation in the visible and near-infrared
(i.e. solar), and IR regions, the frequency is more commonly used to denote radiation
intensity in the microwave region. There, wavelengths are of the order of centimeters to
meters. Thus from equation 1, wavelength increases as the frequency decreases so that,
for example, 1 cm ⬵ 30 GHz and 0.3 cm ⬵ 100 GHz. The 19.35 GHz channel of the
Nimbus-5 ESMR is at 1.55 cm.
The radiation emission from all bodies having an absolute temperature (Tabs) exceeding
0 K (273.15°C) is expressed by Stefan–Boltzmann’s law:
E T 4 (2)
where E total radiant exitance (W m ), Stefan–Boltzmann constant, or 5.6677
2
108 W m2 K4, and T Tabs (K). In the case of the Sun, this gives a black body curve
1 for an assumed temperature of 6,000 K, as shown in Figure 2.4; the corresponding curve
for an assumed temperature of 255 K (the Earth’s effective temperature) is also shown.
The intensity of the radiation emission, or flux density, differs according to the wave-
lengths considered. This can be seen best by considering the relationship of wavelength
to the energy content of photons (quanta), or “packets” of energy, which is given as:
Q hv (3)
where Q energy of a quantum (J), and h Planck’s constant (6.626 10 34
J s ).1
Thus the energy of a quantum increases with increasing frequency, and is inversely related
to wavelength (from solving equation 1 for v and substituting into equation 3): longer
wavelengths (or lower frequencies) are less intense than shorter wavelengths (or higher
frequencies).
The spectral distribution of radiation intensity across all wavelengths for bodies of
different temperature is given by Planck’s law. Moreover the wavelength of maximum
radiation emission (max), which is based on Planck’s law, is expressed empirically by
Wien’s displacement law:
max 2,897/T (4)
where 2,897 a constant (m K). Thus there is an inverse relationship between the
temperature of a body and its wavelength of maximum emission. Substituting the temper-
ature values for the Sun (6,000 K) and the Earth’s surface (288 K) into the denominator
in equation 4 gives values of max close to 0.5 m and 10 m, respectively. Thus the
Sun has a radiation peak in the wavelengths of visible light, and the Earth emits radia-
tion in the longer IR wavelengths. Moreover, Earth targets of different temperatures have
11
111
0111
Figure 2.4 (a) Black-body radiation (Planck) curves for temperatures of 6,000 K (the Sun) and
255 K (the Earth’s surface) normalized to give equal energies (areas). (b) The percentage
absorption in the total atmospheric column. (c) The percentage absorption from 11 km
0 to the top of the atmosphere. (d) The absorption spectra of individual gases contributing
to (b). (From Harries, 1996)
11
different emission rates and max; compare, for example, high cold clouds, Antarctica, and
the subtropical deserts in summer. This is important for satellite remote sensing because
targets at the Earth’s surface or in the atmosphere (clouds, aerosols, water vapor, O3 and
other gases) can be determined by their effects on the incoming solar radiation, and
by their influence on the long-wave radiation emitted from the Earth’s surface and its
atmosphere.
A fundamental assumption in equations 2 and 4 is that the radiation source (Sun, Earth)
emits at the maximum rate for its temperature (the idealized Planck curves in Figure 2.4).
However, this will occur only when the incident radiation in a given wavelength or
group of wavelengths absorbed by the body is reradiated at maximum efficiency (i.e. there
is no attenuation, or depletion of power due to reflection of radiation away from the
body or by transmission through the target). This is expressed by the principle of energy
conservation, thus:
1
EI () ER() EA() ET() (5)
where EI the incident radiation, ER the reflected radiation, EA the absorbed radiation,
and ET the transmitted radiation. From equation 5 it should be evident that EI ()
EA(), and therefore ER ET 0, only in the case of the Sun, which is a black body.
For the Earth–atmosphere system EA < EI and, thus there is some transmission and also
reflection of the incident radiation. The radiation spectrum of the Earth–atmosphere system,
therefore, features various absorption bands due to the selective effects of atmospheric
gases on the outgoing long-wave radiation, or OLR (see Figure 2.4d). The major gases
involved are water vapor, CO2 , methane, and, to a lesser extent, O3. For most of the 8–14
m region the gases do not hinder appreciably the escape of OLR to space
(or to a satellite sensor). Thus the IR wavelength band is mostly an atmospheric window
region having maximum transmission and minimum absorption of OLR. This is appro-
priate for determining the temperatures of targets at the Earth’s surface (land, ocean) and
also of clouds, for example using the so-called split window method, which involves
comparing the radiances in two non-adjacent bands, such as AVHRR bands 4 (11 m)
1 and 5 (12 m), for which the absorption characteristics are known (Coll and Caselles,
1997; Czajkowski et al., 1998). There are certain wavelength bands of IR radiation in
which the absorption and re-emission by the atmospheric gases (primarily water vapor,
CO2) increases and, correspondingly, the transmission to space decreases. Satellite
remote sensing of the upwelling IR radiation by a particular gas in an absorption band
can be compared with radiances for the wavelengths on either side, yielding information
about the abundance of that gas; for example, using the split window method, whereby
there is greater absorption by water vapor in AVHRR channel 5 compared with channel
4 (Eck and Holben, 1994; Suggs et al., 1998). This principle is also the basis of satel-
lite sounding of the atmosphere in many narrow bands to determine vertical profiles of
the temperature from gases that are well mixed throughout the atmosphere (e.g. oxygen,
CO2), and the relative humidity (from water vapor absorption/re-emission). In the synoptic
context, sensing from geosynchronous orbit in the absorption band near 6.5 m has proven
useful in a variety of ways: for identifying and tracking water vapor “features” (Allison
et al., 1972); showing regions of strong ascent (moistening) and descent (drying) associ-
ated with the equatorward and poleward sides, respectively, of jetstreams and tropopause
breaks (Martin and Salomonson, 1970; Ramond et al., 1981; Mueller and Fuelberg, 1990);
for determining the layer–mean wind vectors in the mid-troposphere (Eigenwillig and
Fischer, 1982); and for depicting time-averaged features of the circulation, particularly in
lower latitudes (Picon and Desbois, 1990; Wu et al., 1993; Soden and Bretherton, 1993,
1996).
The efficiency of absorption and re-emission of the incident radiation, or emissivity
(), as a function of the wavelength and temperature is expressed by Kirchhoff’s law,
or:
11 Eemit
= = f (, T ) (6)
Eabs
where Eemit the intensity of the energy re-emission (W m2) and Eabs the intensity
of the energy absorbed (W m2).
Thus the emissivity is a ratio varying between 0 (white body) and 1.0 (black body).
Because of the change of emissivity according to wavelength and temperature, it is possible
for targets in the Earth–atmosphere system to have very different emissivity values
according to the wavelength bands that are used in remote sensing. For example, most
0 clouds (with the notable exception of cirrus) approach black body status ( ⬵1) in the
IR window (8–14 m) wavelengths. Clouds that have high emissivity are optically thick,
which means that they also tend to be highly reflective in visible wavelengths; hence
cumulonimbus cloud masses are known as highly reflective clouds (HRCs). Their distri-
111 bution in the tropics has been mapped by Garcia (1985). The occurrence of HRCs is a
good proxy for deep convection in the tropics (Grossman and Garcia, 1990; Morrissey
and Greene, 1993). Puri and Davidson (1992) show that HRCs are important sources of
diabatic heating for the troposphere.
The emissivity is also a function of Earth surface properties, such as surface rough-
ness, particle size, water-holding capacity, and thermal inertia. These influence target
0 identification in the microwave region. Most land surfaces have emissivities around 0.9
in the IR, and they remain relatively high in the microwave region. Water targets, such
as the ocean, possess an emissivity very close to unity in the IR (i.e. are radiometrically
warm), but only around 0.4 in the microwave region around 19 GHz (i.e. are radiomet-
rically cold). The latter attribute facilitates the detection of the rather weak atmospheric
signal (humidity, rain) over ocean areas in the passive microwave, which is not possible
over the higher emissivity and spatially more variable land surfaces at these frequencies.
The exception is the detection of microwave radiation at higher frequencies (85–90 GHz)
that is scattered by large ice aggregates (hail, graupel) in the upper parts of convective
cloud systems, and which enables thunderstorms to be detected over land as well as sea
0111 (Negri et al., 1989; Mugnai et al., 1990).
Radiation transfer within the Earth’s atmosphere by absorption or scattering is described
by Schwarzschild’s law (equation 7). This states that the net loss of radiation passing
through the atmosphere (dEI ()) results from its partial absorption at one wavelength (e.g.
solar radiation) and re-emission by the atmosphere at another longer wavelength, or:
dEI () a() EI () a () f (, T) (7)
where a is a coefficient of absorption (or scattering) that is dependent upon the density
of the gas and the thickness of the atmosphere. The attenuation of short-wave radiation
with increasing path length is a function of the zenith angle (from Beer’s law). In equa-
0 tion 7 the absorption is negative, or “warming,” and the emitted intensity is given by
Kirchhoff’s law (equation 6).
The emissivity differences between targets in the Earth–atmosphere system at a given
wavelength, and for the same target at different wavelengths, comprise a fundamental
principle of target differentiation using satellite IR remote sensing by their effects on the
retrieved temperature. In the IR portion of the EMS the temperature sensed by a satel-
lite radiometer (Trad) or sounder is related to the actual, or kinetic, temperature (Tkin), as
follows:
Trad 0.25 Tkin (8)
0 Thus for black bodies, where 1.0, Trad Tkin, and the retrieved target temperature
is the actual measured temperature. For actual emitters, Trad and Tkin differ. Detailed infor-
mation on the emissivity of most land surfaces is lacking and, moreover, these surfaces
11 are typically heterogeneous at satellite pixel resolutions. Hence the retrieved surface
temperature in the infrared window region (or TBB – equivalent black-body temperature)
is only approximately related to the actual surface temperature, and is, in effect, a “skin”
(rather than bulk) temperature that is influenced by solar radiation absorption and evap-
oration effects. An emissivity 1.0 is often assumed in automated cloud retrieval
algorithms, such as that of the US Air Force DMSP 3-D Neph program (McGuffie et al.,
1989; McGuffie, 1993), since most clouds approach black-body status in the infrared.
Thus their TBB is close to the actual TCT measured by radiosondes. Moreover, the detec-
tion of optically thin clouds (i.e. clouds for which there is significant transmission to the
satellite sensor of infrared radiation from lower levels in the atmosphere and from the
ground), and assignment of their altitude, is problematic (Brogniez et al., 1995). When
detected, these clouds appear to be lower and warmer than, in fact, they are. However,
the failure to detect some thin clouds complicates the retrieval of accurate SSTs (Wu,
1984). Similar problems arise when a dust or ash layer is located between the target and
1 the satellite, as occurred with the eruptions of El Chichon in 1982 and Mount Pinatubo
in 1991, which disrupted the satellite mapping of SSTs over large areas of the tropics
during a major short-term shift in Pacific sea surface temperature (see section 5.3).
Inferring rain rates using the so-called “indirect” (or infrared) method (Arkin and
Ardanuy, 1989) relies upon the general negative relationship of precipitation with TCT,
and its positive association with the visible reflectance. This is maximized in the case of
HRCs in the tropics, which are also sites of minima in the OLR field (Hendon and
Woodberry, 1993). While the assumptions underlying the infrared rain-rate estimation
method are generally valid, they are most applicable to tropical and subtropical regions
(Martin et al., 1990; Ebert and LeMarshall, 1995; Todd et al., 1995; Ba and Nicholson,
1998), and can yield reliable approximations to the precipitation rate on climatic, rather
than daily, time scales (Rasmusson and Arkin, 1993). Monthly and seasonal maps of
OLR are used to monitor tropical convection related to large-scale atmospheric circula-
tion changes, particularly the El Niño Southern Oscillation (ENSO) (see Chapter 5).
However, the accuracy of the indirect method is influenced by factors such as the loca-
tion (continental versus oceanic) of the precipitating clouds; the humidity in the sub-cloud
1 layer (arid or semi-arid versus humid climates); season; and the confusion between non-
precipitating thick cirrus clouds and the upper parts of cumulonimbus clouds (Carleton,
1991, chapter 5). The infrared method forms the basis of the operational satellite GPI
(GOES Precipitation Index) (Arkin and Meisner, 1987; Herman et al., 1997), which is a
mapped time-integrated measure of rainfall amounts from geosynchronous VIS and
infrared images.
In the microwave portion of the EMS, where the atmospheric transmittance is at a
maximum, the radiance is proportional to the temperature (or Rayleigh–Jeans approxi-
mation). This brightness temperature (TB) is essentially the corollary of the TBB in the IR.
Thus target differences in TB are directly related to changes in emissivity. In the frequency
range 9–90 GHz there are three atmospheric windows that permit detection of the surface
black-body radiation. These are bounded by absorption lines due to water vapor near 22.2
GHz and oxygen near 60 GHz and 118.8 GHz. Over oceanic regions, and in low frequen-
cies (e.g. 19 GHz, 22 GHz), the infrared method of rain estimation can be compared with
passive microwave estimates (the so-called “direct” method: Arkin and Ardanuy, 1989).
Microwave sensing yields “instantaneous” (approximately twenty-minute average
according to Barrett et al., 1990) rain rates that are derived from the absorption and re-
emission characteristics of raindrops (increasing TB ) occurring over a low-emissivity ocean
background (Alishouse et al., 1990; Petty and Katsaros, 1992). As the frequency of the
microwave radiation considered increases to around 37 GHz (i.e. the wavelength decreases:
equation 1), the signal becomes progressively less emission-based and more due to scat-
tering, with a greater proportion due to hydrometeors occurring higher in the cloud system
(Spencer et al., 1989). Thus at around 19 GHz, mostly liquid precipitation in the lower
parts of the clouds is sensed; at around 37 GHz, rain and ice can potentially be detected,
11 and at 85 GHz and higher frequencies, scattering predominates owing to ice (hail, graupel)
and snow aggregates in the upper parts of clouds. The last property makes it possible to
sense deep convective situations occurring over land as well as sea (e.g. Mohr and Zipser,
1996a, b), and to estimate rainfall over land using algorithms that combine the 37 GHz
and 85 GHz TB (Adler et al., 1993; Ferraro, 1997). Moreover, because the resolution of
the microwave radiometer increases with increasing microwave frequency; from about
30 30 km2 at 19 GHz to around 15 15 km2 at 85 GHz, the problem of beam filling
that results from precipitating convection below sensor resolution (Kummerow, 1998) is
greatest at the lower frequencies. Accordingly, corrections for this effect need to be applied
0 where it cannot be assumed that individual pixels are filled with rain (Shin et al., 1990;
Liu and Curry, 1992; Chiu et al., 1993). Thus oceanic rain rates based on the lower-
frequency microwave data from the Nimbus-5 ESMR underestimated the rain rates
measured using rain gauges.
111 Comparisons of the satellite GPI and passive microwave (SSM/I) estimates of monthly
rain rates over the global oceans for an approximately three-year period (Chiu et al.,
1993) show areas of consistently higher and also lower SSM/I estimates that are differ-
entiated regionally (Ferraro, 1997). These are interpreted as resulting from the merger of
different satellite data sets to develop the GPI, and the interpretation of high and cold
cirrus clouds as raining in the infrared data. Generally, GPI (passive microwave) methods
0 work better in the tropics (extratropics), and in areas where the SSM/I works well a
combination of GPI-passive microwave sensing gives good estimates of monthly precip-
itation over both land and adjacent water (Adler et al., 1993; Negri et al., 1993). When
merged with surface rain gauge observations the satellite remotely sensed data form the
basis of the Global Precipitation Climatology Project (GPCP) Combined Precipitation
Dataset, initially developed for the period July 1987 through December 1995 (Huffman
et al., 1997).
Given the lighter rain rates typical of higher latitude weather systems and the gener-
ally poorer performance of passive microwave methods there, a simpler method of
demarcating the likelihood of rain occurrence (the P37 index) in given pixels has been
0111 suggested by Petty and Katsaros (1992) and Carleton et al. (1995). This is similar to the
polarization-corrected brightness temperature (PCT) developed by Spencer et al. (1989)
for the SSM/I 85 GHz channels. This method has been used to help retrieve precipita-
tion over ocean in higher latitudes (Lachlan-Cope and Turner, 1997), and over
heterogeneous (i.e. variable emissivity) surfaces in a mixture of synoptic conditions (Kidd,
1998). A recent advance involves comparing GOES infrared radiation data with surface-
based radar (i.e. active microwave) for real-time estimates of precipitation rate in the
United States (Vincente et al., 1998).
The column-integrated atmospheric water vapor content and the cloud liquid water
content (in kg m2) can also be detected over low-emissivity ocean surfaces in the
0 microwave region. Retrieval algorithms of these quantities utilize the dual-polarization
measurements of TB for combinations of the SSM/I channels (Petty, 1994a, b). While
these passive microwave measurements have been used to develop large-scale climato-
logies (e.g. Prabhakara et al., 1992; Martin et al., 1993; Ferraro et al., 1996), they also
have proven invaluable for gaining insights into the mesoscale structure of synoptic and
subsynoptic storms over the oceans (Katsaros and Lewis, 1986; Katsaros et al., 1989;
McMurdie and Katsaros, 1985, 1991; Chang et al., 1993; Carleton et al., 1995; Claud
et al., 1995; also Chapter 6). Moreover the emission of passive microwave radiation by
the ocean surface changes in response to the spray and foam generated by near-surface
winds (Goodberlet et al., 1989). Thus the change in TB measured using dual polarization
0 at lower microwave frequencies that occurs as wind speed increases can be inverted to
yield a near-surface (approximately 20 m height) wind speed (in m s1). The spatial vari-
ations of the wind speed associated with synoptic cyclones, as well as with cold-air
11 mesocyclones, have been identified using the SMMR and SSM/I sensors (Claud et al.,
1993; McMurdie et al., 1997; Song and Carleton, 1997). However, the retrievals are erro-
neous in regions of heavy rainfall; typically near cold fronts and in tropical cyclones,
since raindrops falling on the sea surface influence the microwave emission. In these
areas, transects (swaths of approximately 10 km width) of wind speed data provided by
non-imaging active microwave sensors, particularly the Geosat and TOPEX-Poseidon radar
altimeters (Mognard and Katsaros, 1995a, b), supplement the information obtained from
microwave radiometry. Moreover the near-surface wind speed and direction vectors
acquired over wider swaths by the active microwave scatterometer of the ERS-1 satellite
are becoming indispensable to weather analysis and prediction, as well as for depicting
details of the near-surface climatology over the oceans; especially in the middle and higher
latitudes dominated by traveling synoptic and mesoscale storms (Marshall and Turner,
1997; Siefridt et al., 1998). Comparisons between passive and active microwave measure-
ments of ocean surface wind speeds indicate good agreement globally (Boutin and Etcheto,
1 1990), but with regional-scale differences that appear to result mostly from the effects of
atmospheric attenuation (Mognard and Katsaros, 1995b). Given the negative association
between wind speed and isobaric spacing, and the modifying effects of atmospheric
stability on the wind speed in the boundary layer, pressure patterns over data-void ocean
areas can be refined greatly when the scatterometer data are input to a PBL model (Brown,
1986; Levy and Brown, 1991; Brown and Zheng, 1994; Hsu et al., 1997).
The first all-year weather mapping of polar surfaces was made possible by the launch
of the 19.35 GHz passive microwave sensor on Nimbus-5 in December 1972. ESMR
demonstrated the ability of microwave radiometry to detect sea ice in polar regions, from
the strong increase in emissivity and TB that occurs between open ocean and ice-covered
areas, and the relative transparency of most clouds at these latitudes to microwave radi-
ation. Three-day time-lapse movies, and case-study periods, of the ESMR data for the
period 1973–76 clearly revealed the movements of the sea ice edge in response to the
dynamic (wind-field) and thermodynamic (temperature advection) forcings of synoptic-
scale cyclones (Campbell et al., 1980; Crane et al., 1982; Zwally et al., 1983; Parkinson
et al., 1987; Carleton 1984). The unambiguous retrieval of the ice-water concentration
1 and ice type (first year versus multi-year) parameters for the entire seasonal cycle requires
polarization measurements of the microwave emission which the nadir-pointing, single-
channel, horizontally polarized (the plane of polarization of the antenna is parallel to the
earth’s surface) ESMR did not possess. However, the separation of ice concentration
effects on TB from those related to different ice types (age) was possible for times lacking
significant surface melting (i.e. in winter). The SMMR instrument which operated on
Nimbus-7 from October 1978 through August 1987 was dual-polarized at five frequen-
cies: 6.6 GHz, 10.7 GHz, 18 GHz, 21 GHz and 37 GHz. The SSM/I sensors on DMSP
satellites, which were launched in July 1987 and continue through the present, provide
seven channels of data: dual polarized radiances at 19.3 GHz, 37 GHz and 85.5 GHz and
vertically polarized radiances at 22 GHz . The 18 GHz (19.3 GHz) and 37 GHz chan-
nels are the primary ones used in algorithms for the mapping of sea ice extent, type and
concentration (see Carsey, 1992, for detailed descriptions of the techniques and results).
The dual-polarization measurements from SMMR (Gloersen et al., 1984) and SSM/I
(Barry, 1991) made possible the continuous mapping of ice type (age) and concentration
changes in both polar regions (Gloersen et al., 1992; Steffen et al., 1992) and have
provided critical information on recent trends in sea ice cover. There was a decadal
decrease of 2.9 percent in Arctic summer ice extent during 1978–96, for example, with
most of it occurring since the mid-1980s (Bjørgo et al., 1997; Cavalieri et al., 1997).
Low-pressure systems in the Arctic basin appear to have played a significant role in
reducing ice concentrations, especially in the Eurasian shelf seas, in summers 1990, 1993
and 1995 (Maslanik et al., 1995; 1996). The determination of summer ice concentration
from passive microwave data is still problematic owing to the effect of surface snow
11 melt and melt-pond formation (Comiso, 1990). Moreover, atmospheric influences in the
presence of high totals of column water vapor and cloud liquid water are known to cause
overestimates in total ice cover, and to increase (decrease) the apparent concentrations of
first-year (multi-year) ice (Maslanik, 1992; Oelke, 1997). Procedures to reduce weather
effects on the calculation of ice concentrations have been developed (Cavalieri et al.,
1995).
SMMR and SSM/I data have also enabled the identification of surface melt onset in
the Arctic (Anderson, 1987; Crane and Anderson, 1994) and recently maps of Arctic
sea ice melt and freeze-up dates have been constructed (Smith, 1998). However, mapping
0 of the progression of summer albedo values for the Arctic Ocean ice has been based on
DMSP visible band imagery (Robinson et al., 1992) and AVHRR-derived radiances
(Lindsay and Rothrock, 1994; Schweiger et al., 1993). Information on lead (linear ice
fracture) statistics has also depended on these last two sources (Lindsay and Rothrock,
111 1995; Miles and Barry, 1998). Sea ice motion products have been developed for limited
regions with visible band AVHRR data and synthetic aperture radar (SAR) data, indi-
vidually or as blended products including drifting buoy records, and for the polar oceans
using 85 GHz passive microwave data (Emery et al., 1997; Agnew et al., 1997). High
resolution (10–100 m) active microwave (SAR) remotely sensed data can often comple-
ment other sensors in assessing such features as leads and ice type (Onstott, 1992; Kwok
0 et al., 1992), as well as deformation, ridging, and motion. Barber et al. (1991) also discuss
the possibilities of estimating climatic state variables (albedo, latent heat, and atmospheric
drag coefficients) from ice type and cover information. At present, SAR data (ERS-1 and
-2 and Radarsat) are used primarily for operational purposes, although as a result they
now enter into the ice charts prepared by the national ice services.
In the case of snow-covered ground the microwave energy emitted is determined by
the amount of that energy scattered by the snow pack. For dry snow there is a sharp
emissivity decrease in the presence of a snow cover relative to snow-free ground (Maetzler,
1994). Scattering is a function of snow depth and grain size (Armstrong et al., 1993).
Of greater importance on a global scale is the emission from the vegetation canopy or
0111 land cover, especially in the boreal forest zone (Foster et al., 1991; Tait, 1998). Weekly
snow cover extent is mapped by NOAA NESDIS for the northern hemisphere from
AVHRR and the NOAA data product is widely used in climate studies (Robinson et al.,
1993). Operational SSM/I snow products are also now available (Grody 1991; Goodison
and Walker, 1995). Research products based on 18 GHz and 37 GHz passive microwave
data for snow extent and estimated snow water equivalent (SWE) have a long history
(Chang et al., 1987; Rott, 1987). The SMMR/SSM/I data for 1978–97 are used by
Armstrong and Brodzik (1999) to map snow extent on a daily basis with 25 km resolu-
tion using the algorithm of Chang et al. (1987). The trend in the twenty-year record shows
an annual decrease in snow cover of 46,000 km2 compared with a corresponding trend
0 of 64,000 km2 in the NOAA visible band data. Shallow and/or wet snow is not mapped
consistently by the passive microwave (Basist et al., 1996). Thin snow may not be
detectable and snow melt causes a loss of signature but the 37 GHz polarization differ-
ence can serve as a wet-snow indicator (Walker and Goodison, 1993), enabling corrections
to be applied. Algorithms for the determination of snow depth or SWE, however, are still
limited geographically and/or temporally in their applicability (Goodison, 1989; Tait,
1998).
2.3 Climate variables and their statistical description

0 In the words of Durst (1951), “climate is but the synthesis of weather.” There are various
levels of climatological synthesis. Table 2.1 summarizes a conceptual view of these levels
that takes into account the variation in the number of weather elements (e), in time (t)
11
Table 2.1 Levels of climatic synthesis
Order Variables Description Representation

1 ƒ (e0 s0t0) Instantaneous value of one element at one Synoptic observation
instant of one element
1 ƒ (e0 t0s) Spatial distribution of one element at one “Synoptic map” of one
instant element; satellite
cloud photo
1 ƒ (e0 s0t) Time variation of one element at a point Time series
(autographic record)
1 ƒ (s0 t0e) Instantaneous, point value of several elements Station weather report
2 ƒ (e0 st) Changes in the distribution of one element Series of pressure
1 with time maps
2 ƒ (t0es) Spatial covariance of weather elements Synoptic weather map
2 ƒ (s0et) Time covariance of weather elements Climogram
3 ƒ (est) Covariance of weather elements in space Series of synoptic
and time weather maps
Source: modified from Godske (1966).
and in space (s). The zeroth level is the instantaneous value of a single weather element,
such as air temperature, at a given location. It should be noted that the time variation
may refer to any range of time scale (diurnal, intraseasonal, annual, or interannual vari-
ation). In the table, space is regarded as having two dimensions.
2.3.1 Frequency distributions

1
Climatological synthesis is not made completely explicit in Table 2.1. For many purposes,
the description of a climatic variable may involve determination of the frequency of occur-
rence of the complete range of values of the variable, as well as statistics such as the
variance, or standard deviation, and the higher moments of the frequency distribution
(skewness and kurtosis). The appropriate statistics depend heavily on the characteristics
of the frequency distribution for each element and averaging interval.
Each weather element tends to have a distinctive frequency distribution according to
the climatic regime and averaging period. Many distributions are not Gaussian (or normal).
Examples of the frequency distributions for selected weather elements at Bergen, Norway,
are illustrated in Figure 2.5. The distribution of pressure values is close to Gaussian
whereas the temperature plots are positively skewed, with a tail towards higher values.
The distribution of summer visibility values is J-shaped while daily precipitation amounts
show a characteristic reversed J shape. The bimodal distribution of cloud amounts is also
typical of station observations of cloud cover, reflecting the fact that the sky conditions
overhead approximate a binary state.
A multimodal distribution, such as that for Bergen evening temperatures in September
(Figure 2.5), commonly represents a mixed population. The peaks here probably indicate
the occurrence of different air masses or airflow directions (Fiedler, 1965). A numerical
method to decompose such distributions into several partial frequency distributions or
1 collectives was developed by Essenwanger (1955, 1960a), although Bryson (1966) shows
that a close approximation can be obtained by direct graphical analysis (Figure 2.6).
11
111
0111
Figure 2.5 Frequency distributions for selected climatic parameters at Bergen, Norway. P pressure,
T temperature (°C); subscripts or numerals refer to the hour of observation. (After
Godske, 1966, from Barry and Perry, 1973)
0
2.3.2 Exploratory data analysis
Climatic data are often analyzed using standard parametric techniques which assume that
the data distribution is Gaussian. For many practical purposes, non-parametric techniques
of exploratory data analysis (EDA) are actually preferable (Hoaglin et al., 1985). Examples
of climatological applications are provided by Kleiner and Graedel (1980) and Lanzante
(1996). The central tendency of a distribution can usefully be expressed by the median,
or middle value of a distribution. This measure is resistant to outliers (extreme events or
erroneous data). However, its efficiency, representing the effect of sampling variability
on the median, is less than that of the arithmetic mean. The variability can be described by
0 the interquartile range (IQR), which is the difference of the upper quartile minus the lower
quartile (the 75 percent value minus the 25 percent value). For a normal distribution the
IQR 1.349 . Thus a pseudo-standard deviation can be defined as IQR/1.349. Examples
11 of upper-air data are given by Lanzante (1996).
Figure 2.6 Schematic multimodal distribution (heavy line) illustrating two methods of estimating
partial collectives (light lines). Method A: identify ordinate of the median; express the
ordinates one unit above and below the median as a fraction of the median; calculate
the standard deviation and reconstruct the partial collective. Method B: obtain end
collectives by folding the distribution along the median ordinate and subtracting from
the total distribution. (From Bryson, 1966)
2.3.3 Contingency analysis

1 A question that often arises in climatic analysis is the co-frequency of two (or more) vari-
ables. The co-frequencies can be arranged in a two-way contingency table in which each
cell refers to a specified subset of the variables. Examples of synoptic climatological appli-
cations are given by Murray and Lewis (1966), for rainfall categories over England and
Wales against a cyclonicity index, and by Namias (1991), who examines summer temper-
atures over the Great Plains as a function of antecedent spring temperature and
precipitation, using tercile categories in each case. The contingency table can be readily
used for significance tests of association such as the non-parametric 2 statistic (see Wilks,
1995, for example). Carleton (1995) adopts this approach in examining the relationship
between polar low occurrences in relation to surface type over the Southern Ocean and
their cloud cover attributes.
2.2.4 Probability
Probability theory concerns the likelihood of specific chance events happening.
Probabilities (p) are expressed on the scale p 0 to 1.0 (or 0–100 percent). The total
probability in a situation where there are several possible outcomes is always equal to
1.0. Thus if the average frequency of rainy days in April at a station is equal to six, then
the probability of any day in the month having rain is p 0.2 (assuming that rain days
occur at random) and the probability of it being dry is 1 p 0.8. For statistically inde-
pendent variables, joint probabilities are given simply by the product of the individual
probabilities. Suppose, for example, that the wind speed at the same station is unrelated
to precipitation and that the probability of a day with winds below 5 m s1 is 0.05 in
April, then the joint probability of a rainy day with light winds is
11 p 0.2 0.05 0.01
If two variables are not independent, then their conditional probability – the probability
of one occurring given that the other occurred – must be evaluated.
The properties of four statistical distributions that are of particular importance in deter-
mining climatological probabilities are now briefly discussed. They are the normal,
binomial, Poisson and gamma distributions. A useful summary of the properties of prob-
ability distributions and their relationships to one another is given by Rothschild and
Logothetis (1986).
0
Normal distribution
The well known normal or Gaussian distribution, which is symmetrical and bell-shaped
111 (Figure 2.7), is the basis of many parametric statistical methods. Its statistics are the arith-
metic mean value:
n
兺x
1
x= i
n i=1
0 and the second moment, or variance:

n
兺 (x x)
1
s2 = i
2
n i=1
The standard deviation is given by s. The rth moment is:

n
兺 (x x )
1
r = i
r
n i=1
0111 Skewness is usually determined from 3/s3 and the kurtosis from 4/s4. Skewness is a
measure of the symmetry of the distribution; the skewness is positive (negative) when
the distribution has a long tail towards high (low) values of the variate. Kurtosis describes
11 Figure 2.7 The normal (Gaussian) frequency distribution. (From Barry, 1971)
the amplitude of the peak relative to the normal curve. Kaplansky (1945) shows that there
is no reliable relationship between skewness and kurtosis.
The probability density function (pdf) for a normal distribution is written:
1 [(x)2/ 2 2]
f (x) = 1/2 e
(2)
where e the constant 2.71828, is the population mean estimated by the sample mean
x–; is the population standard deviation estimated from s, except that the sample size is
replaced by n1 to represent the number of degrees of freedom (i.e. the number of inde-
pendent observations in a sample minus the number of population parameters that must
be estimated from the sample observations). A sample mean, as an estimate of the popu-
lation mean, has a standard error of s/(n1)0.5. This allows the “true” mean to be estimated
1 within a given range for a specified probability level.
The pdf of the normal distribution expresses the proportion of the distribution under a
specified portion of the normal curve when the total area beneath the curve is equal to 1.
The proportion of the distribution is specified in terms of . For example, 68.26 percent
of the distribution is within ±1 of the mean; 95.46 percent of the distribution is within
±2 of the mean; 99.73 percent of the distribution is within ±3 of the mean. Tables
of areas under the standard normal curve are available in statistical texts and reference
tables. The 3 limit is commonly used to test for likely outliers in observational time
series.
For two-dimensional orientation data, such as wind velocity, a circular normal distrib-
ution can be used (Gumbel, 1954; Curray, 1956; Fisher, 1993; Klink, 1998). The resultant
vector provides a measure of central tendency and magnitude, independent of origin. Its
azimuth is obtained from:
V sin
= arctan
V cos
1 –
where is the azimuth of each wind direction (0–360º), V is the magnitude of the wind
velocity (or the number of observations in the case of grouped data); the numerator
(denominator) represents the u(v) component of the wind. The mean resultant velocity is:
r
V=
V
where r [(V sin )2 (V cos )2]0.5. The standard vector deviation is:
冢VN V 冣
2 0.5
= 2
–
where V2/N is the mean square velocity and V is the mean resultant velocity. The mean
resultant (geostrophic) wind can be inferred from mean pressure maps, but these cannot
discriminate between strong, variable pressure gradients and predominantly weak ones.
Tucker (1960), for example, provides resultant and standard vector deviation statistics
for upper-level winds. It is worth pointing out that the bivariate normal distribution can
be used for cases where there is a pair of related variables, such as wind direction
and pollutant concentration (Essenwanger, 1976). Related discussions on angular corre-
lation for circular data are provided by Johnson and Wehrly (1977) and Fisher and Lee
(1983).
11 Binomial distribution
This is an approximation to the normal curve for data in discrete classes. It is determined
from the expansion (p q)n where q 1 p, when p and q are mutually exclusive
events. The binomial probability law (probability mass function, because discrete values
are involved) is:
ƒ (x) 冢nx冣 p q , for x 0, 1, 2, . . . n

x nx
0 where 冢nx冣 is the number of combinations of x items out of a total of n,

冢nx冣 = x!(nn! x)!
111
x! (x factorial) denotes the expression x(x 1)(x 2) . . . 3, 2, 1.
The mean of a binomial function is np and the variance is npq. In a simple case we
might be interested in the probability of days being wet or dry. If p 0.5 for a wet/dry
day, and assuming no interdependence, the probabilities for two days picked at random
are:
0
State: Two wet One wet One dry Two dry Total
Probability: p2 0.25 0.25 2pq 0.5 q2 0.25 (p q)2 1
For probabilities over three days we use the expansion (p q)3 p3 3p2q 3pq2 q3
and so on. It will be noted that the distribution assumes constant probability, a situation
referred to as Bernoulli trials.
An illustration of the use of the binomial distribution to test for a change in the frequency
of large rainfall events in Wales is provided by Joliffe (1983).
0111
Poisson distribution
For many types of meteorological event the frequency of non-occurrence cannot be spec-
ified. This is true of storms, floods, and droughts, all of which are rather rare events
occurring “at random.” The Poisson distribution, a limiting form of the binomial, is applic-
able to many of these situations. However, it does require large data sets in order to assess
rare events.
The frequency distribution follows an exponential form. If z is the average number of
events during the total time interval, it is assumed that the average remains constant from
trial to trial, so that there is no time trend, and that the probability of an event is unaf-
0 fected by the time elapsed since the preceding one.
∞
z2 z3 zr zx
ez = 1 z ... ...=
2! 3! r! 兺
x=0 x!
This infinite series converges to ez for all values of z. Now, for positive integers of a
random variable x,
zx ez
f (x) =
x!
0
This function satisfies the conditions of a probability density function since
11 ƒ (x) > 0
and
∞ ∞
z x ez
兺
x=0
f (x) = 兺 x=0 x!
= ezez = 1
(i.e. the total area beneath the frequency curve is 1).

A variable satisfying a function of this form is said to be Poisson-distributed. A char-
acteristic of the Poisson distribution is that the mean and variance are both equal to z.
The more extreme an event (e.g. four hurricanes in one season) the much less likely it
is to occur in a given interval. The model is useful for computing the probability that
exactly k events occur in a specific interval, given that the average occurrence in that
interval is z.
1
Gamma distribution
The gamma distribution, like the Poisson, is positively skewed, but it is a continuous
distribution of x between 0 and (Thom, 1958). Here the assumption is that the events
constitute a “renewal process” where the time intervals between these events are distrib-
uted independently and identically (Cox, 1962).
The gamma function is:
() (1)!
where is any positive integer. It can be shown that:
() = 冕0
∞
xa1 ex dz
is a shape parameter estimated from:

1
冦冢
1 4A
冣冧
1/2
1
= 1
4A 3
where:
冢兺ln x冣
n
A = ln x
n
and
x–
2 2
where is a scale parameter.
The probability density function is
x1 ex/
f (x) =
()
To determine cumulative probabilities, for example, of precipitation xi , we use:
t (F) xi / .
Tables of this incomplete gamma function are available (Pearson 1951) for F against
and t(F).
11 The gamma distribution is particularly useful for zero-bounded variables such as
short-period precipitation totals (Suzuki, 1967) and for cloud amounts (Henderson-Sellers,
1978).
Weibull distribution
The Weibull distribution has certain characteristics in common with both the gamma and
the exponential distributions (Olkin et al., 1980; Devore, 1995). The pdf is:
冦ab (x bc) 冤 (x c)a

冥冧
a1
0
f (x) = exp if x c
b
0 if x c
111
As with the gamma distribution, the parameters a and b determine, respectively, the
shape and the scale of the density distribution; c serves as a measure of its location. For
a 1, both the gamma and Weibull distributions correspond to an exponential distribu-
tion. The Weibull distribution is commonly applied in analyses of wind speed data.
0 Beta distribution
The beta distribution is appropriate for variables that vary from 0.0 percent to 1.0 percent
or 0 percent to 100 percent such as cloud amount or relative humidity (Wilks, 1995). The
pdf of the beta distribution is:
f (x) = 冤(a) (b)冥

(a b)
x a1
(1 x)b1
for 0 x 1, a, b > 0. The ranges of a and b determine the typical shape of the
0111 frequency curve (Figure 2.8). Henderson-Sellers (1978) summarizes this as follows:
Shape a b
Single peak > 1 > 1

J-shape < 1 > 1
Reverse J > 1 < 1
U-shape < 1 < 1
Although Henderson-Sellers (1978) used the beta distribution for global cloud frequen-
0 cies, subsequent analysis indicates that the Burger distribution is a more robust model for
this variable (Henderson-Sellers and McGuffie, 1991). Easterling (1989) examines thun-
derstorm rainfall over the United States using both the gamma and beta distributions.
Transformations
Not all meteorological data series can be fitted by the common distribution functions. In
some cases a normal distribution can be approximated by an appropriate transformation
of the original data series. Precipitation data are commonly truncated on the “dry side,”
for example, and a log-normal or cube root transformation is often used to transform the
0 data (Essenwanger, 1960b; Stidd, 1953). The goodness of fit of a raw or transformed data
series to any distribution function can readily be tested by using the 2 function to compare
the observed frequencies with those expected from the assumed model.
11
Figure 2.8 Examples of the beta distribution pdf for different ranges of the parameters (here labeled
p and q). Mirror images of the distributions are obtained by reversing these parameters.
(From Wilks, 1995)
2.4 Analytical tools for spatial data

2.4.1 Synoptic maps
The mean sea level (MSL) pressure field is the most common type of synoptic map.
1 Station pressure observations are adjusted (“reduced”) to the theoretical value at mean
sea level and standard gravity. Since the necessary correction involves the observed
temperature at the station and the assumed lapse rate, “fictitious” MSL pressures may be
reported over montane regions, Greenland and Antarctica (Streten, 1980). The apparent
intensity of the Siberian winter anticyclone is due, in part, to this correlation (Walker,
1967). In synoptic analyses for the hemisphere, MSL isobars are usually drawn at 4 mb
or 5 mb (hPa) intervals and mesoscale features are deliberately smoothed out of the
analysis. Hemispheric or global analyses are made for the main synoptic hours (00.00,
06.00, 12.00, 18.00 UTC), and more frequent analyses may be made for a restricted area,
by national meteorological centers. Daily MSL pressure maps are available for Europe
and most of North America since the 1870s, but the temporal coverage on a hemispheric
or global basis is more limited (see below). Synoptic pressure maps also provide plotted
station weather reports in coded symbols.
Contour charts for constant pressure surfaces have been the international standard for
upper-air analyses since 1945. Earlier, beginning in 1933, pressure maps were drawn at
constant high levels in Great Britain and North America. Standard contour charts are
widely prepared for 1,000 mb, 850 mb, 700 mb, 500 mb, 300 mb, 200 mb, 100 mb levels;
specialized centers may also analyze 50 mb, 30 mb (or 25 mb), 10 mb and 5 mb levels.
Contour heights are expressed in geopotential meters (gpm), where
g
1 gpm = geometric (meter)
9.81
g acceleration due to gravity (m s2).

11 2.4.2 Quality of hemispheric and global analyses
In discussing circulation features it is important to consider the quality and spatial coverage
of the data available for the analyses. Jenne and McKee (1985) and Shea et al. (1996)
provide an overview of sources of atmospheric and oceanographic data. Also, for more
recent products it needs to be recognized that analyses of surface pressure and geopotential
height fields commonly incorporate a prognostic map as the first guess field for a subse-
quent analysis (Trenberth and Olson, 1988). This can lead to good data being rejected in
the analysis! Further inhomogeneities arise as parameterizations of physical processes are
improved, and methods of assimilating satellite or other observations, or of smoothing
0 and interpolating data, are modified in the analysis procedures (Simmons et al., 1989;
Arpe, 1991). The effect of such changes on the data products are now well recognized
and reanalyses are under way at the National Center for Environmental Prediction (NCEP),
Washington, DC (Kalnay et al., 1996, 1998) and the European Centre for Medium Range
111 Forecasts (ECMWF) (Bengtsson and Shukla, 1988; Gibson, 1998), to reanalyze the com-
plete input data streams for the last forty years, or more, to obtain consistent products.
Intercomparison of global data sets prepared by NCEP and ECMWF shows less reliable
analysis in the tropics and southern hemisphere (Trenberth and Olson, 1988). The main
problems so far identified in widely used historical series of hemispheric and global data
sets are summarized in Table 2.2. Problems are particularly severe in the southern hemi-
0 sphere over the ocean areas with few stations (Barnett and Jones, 1992). Jones (1991)
notes that reconstructions of mean sea-level pressure for 15°–60°S back to 1951 are
possible in areas with numerous ocean islands but excluding the eastern equatorial Pacific,
the southeastern and far southern Pacific Ocean and far southern Indian Ocean areas. Even
mean pressure fields for 45°–65°S need to be treated with great caution.
Reconstructions of monthly grid-point sea-level pressures have been obtained for Europe
back to 1780 and for North America back to 1858 (Jones et al., 1987). These are based
on regression equations relating principal components of surface pressure to station data
on monthly mean pressure, temperature, and precipitation. The analyses use a calibration
0111 Table 2.2 Major inconsistencies and problems of data quality in hemispheric and global map
analyses
Source Problems/Changes References

US Weather Bureau: historical Positive biases in Arctic Jones (1987)
weather maps, 1899–1945 pressures, especially pre-1930s
to 1940s
US National Meteorological Changes in analysis techniques Lambert, 1990
Center: northern hemisphere caused major shifts in the data
0 500 mb charts 1946–present. in 1953, 1955, 1962 (especially),
and 1978
Northern hemisphere sea-level The archived analyses were from Trenberth and
pressure grids, 20°N to the US Navy charts up to 1963. In Paolino (1980)
pole. NMC global analyses, 1979 Arctic analyses improved
July 1976–present through the inclusion of drifting
buoy data. A reanalysis is
being prepared
European Center for Medium May 1985, New T106 model; Shaw, et al.
Range Weather Forecasts May 1986, nineteen model levels; (1987), Simmons
0 (ECMWF) global analyses, July 1986, gravity wave drag et al. (1989), and
1980–present incorporated; September 1986, Arpe (1991)
modified analysis scheme; May
1989, new radiation scheme
11
period of 1900–74 for Europe and 1921–80 for North America and have been tested with
independent data for a different period of years.The European reconstuctions for 35°–70°N,
30°W–40°E are updated to 1995 by Jones et al. (1999), using 1936–95 for the calibra-
tion period and 1881–1935 for verification. The application of synoptic weather mapping
to the historical reconstruction of the circulation over Europe for AD 1780–1820 is
presented by Kington (1991). This was subsequently extended to the period AD
1675–1704; mean surface pressure maps over Europe were prepared for the winter and
spring months by J. Kington and H.H. Lamb using the Lamb types for the British Isles.
The maps are presented by Wanner et al. (1994). The pressure fields were interpreted on
the basis of contemporary documentary records of anomalous weather events. However,
Jones et al. (1999) report systematic biases in the reconstructions of Kington and Lamb.
Subsequently, monthly mean grid-point sea-level pressure charts for the eastern North
Atlantic–European region (35°–70°N, 25°W–30°E) have been reconstructed for the entire
1 Late Maunder Minimum period (AD 1675–1715), using canonical correlation analysis
(Pfister et al., 1998). Statistical relationships were established between atmospheric circu-
lation patterns and station measurements of pressure (Paris), air temperature (Kew and
Paris), air temperature indices (Budapest, Lisbon and Zürich), precipitation indices
(Barcelona, Budapest, Kew, Lisbon, Madrid and Zürich), and the western Baltic sea ice
index. A basis for the statistical relationships was first established using station observa-
tions and pressure data for 1901–90 and used for linear prediction of conditions during
AD 1675–1715 assuming climatic stationarity.
2.4.3 Kinematic properties of the wind field

Wind velocity, like any vector quality, may be represented by a directional arrow (showing
the direction from which the wind is blowing) proportionate in length to the magnitude.
For many purposes this type of presentation is cumbersome although it is commonly used
to show the flux of properties such as water vapor, for example.
It is sometimes useful to consider the westerly (u) and southerly (v) components of the
1 horizontal wind velocity (VH). In cartesian coordinates the u component (positive for west
wind) is given by the projection of VH on the west–east (x) axis and the v component
(positive for south wind) by the projection of VH on the south–north (y) axis. Where the
wind direction is the azimuth determined from north (360°) corresponding to the y
axis of a cartesian graph:
u VH sin
v VH cos
In evaluating the sign it is helpful to remember that in the sector
0°–90° sin cos

90°–180° sin cos
180°–270° sin cos
270°–360° sin cos
Component fields are rarely plotted for synoptic analyses, but we shall return to them
again in connection with the global circulation (see section 3.2).
In general the most convenient means of analysis is to examine the scalar quantities
of direction and speed individually. These can be depicted in the form of isogon maps
of constant direction and isotach maps of constant speed (see Figure 2.9). It is more usual,
however, to determine streamlines, or line tangent to the instantaneous motion at each
point. These can be constructed by sketching directly from the velocity vectors. A more
precise method, using an isogon map, is to draw line segments by tangent curves. This
approach is particularly valuable in that wave motions in the flow are readily detected by
11
111
0111
Figure 2.9 An example of isogon and streamline analysis at 850 mb and the corresponding MSL
pressure map. (After Schüepp, 1963, from Barry and Perry, 1973)
isogons. The wind speed may be shown by superimposed isotachs with the spacing of
0 the streamlines independent of the wind speed (Palmer, 1952) or, alternatively, the spacing
of the streamlines is made inversely proportional to the speed (Watts, 1955). Streamlines
may respectively converge into, or diverge from, centers of inflow and outflow known as
11 “singular points.” Inflow may also occur along a singular line or asymptote of conver-
Figure 2.10 The basic patterns of streamline curvature and diffluence (confluence) for the anti-
cyclonic case. (After Jarvis, 1967; from Barry and Perry, 1973)
1
gence. The rate at which flow is converging or diverging with respect to an axis perpen-
dicular to the flow is referred to as confluence or diffluence, respectively.
The basic patterns of streamline curvature and diffluence or confluence are illustrated
for the anticyclonic cases in Figure 2.10. The corresponding cases of cyclonic curvature
give rise to patterns which are the exact reverse of the anticyclonic ones.
Streamline analysis is used most frequently in low-latitude analysis, since pressure
gradients are generally small and consequently the geostrophic wind field is not readily
determined. Illustrations of such streamline maps may be found in Sadler (1965). Because
small variations in the wind vector may be significant in the tropics, and the reliability
and representativeness of observations at a single level is sometimes in doubt, it has
become common to use mean layer winds for 1,000–3,000 m. These averages are routinely
reported in tropical RAWIN-sonde ascents (Zipser and Colon, 1962).
2.4.4 Derived data
Horizontal derivatives
Four primary characteristics, which involve combinations of the horizontal derivatives of
the velocity components, are derived from the wind field (Figure 2.11). They are:
∂u ∂v
= horizontal divergence (H·V in vector notation)
∂x ∂y
11 ∂u ∂v
= “stretching” deformation
∂x ∂y
∂v ∂u
= “shearing” deformation
∂x ∂y
∂v ∂u
= relative vorticity about the vertical axis, HV.
∂x ∂y
0
A word must be said about vector notation (see Appendix 2.1). The gradient operator
(del) is defined as:
111 ∂ ∂ ∂
≡i j k
∂x ∂y ∂z
where i, j and k are unit vectors in the x, y and z directions respectively. In the hori-
zontal case (H) the vertical term
/
z is of course omitted. By definition, . ( ) is the
divergence of, ( ) is the curl (or vorticity) of the term following the operator.
0 The horizontal divergence may be estimated by using a finite difference grid. That is
to say,
u/
x and
v/
y are approximated by finite values u/x and v/y of the velocity
components and lengths. Thus with reference to the coordinates of Figure 2.28:
ux ux vy vy
H · V =
L L
where L a unit length in a rectangular grid (Miller, 1948; Panofsky, 1951), L/2 x
x y y (see Figure 2.28). By definition divergence is positive. Divergence
(convergence) which measures the overall expansion (contraction) of the wind velocity
0111 field must not be confused with confluence (diffluence). Diffluent streamlines may, for
example, be associated with decreased wind speed so that the two effects tend to cancel
out (Figure 2.11a).
In practice, wind data are usually inadequate for obtaining very reliable estimates of
the divergence because of inaccuracy in the basic measurements and because the wind
is nearly geostrophic above the friction layers. The horizontal divergence is zero for
geostrophic flow (when the horizontal pressure force is exactly balanced by the Coriolis
acceleration).
Figure 2.11 Stream function and wind velocity for two-dimensional non-divergent flow. (From
11 Barry and Perry, 1973)
Figure 2.12 Schematic models illustrating (a) divergence/convergence and (b) relative vorticity due
to streamline curvature and lateral shear. The dashed lines are isotachs (nominally
m s1). (From Barry and Perry, 1973)
11 The special case of divergence induced by differential stress at a coastline has been
examined by Bryson and Kuhn (1961). Onshore winds produce convergence through the
slowing down of the flow due to increased friction over land. Winds parallel to the coast,
with high pressure over the land (in the northern hemisphere), also produce convergence,
as a result of the shear set up by the reduced speed over land. The air is thus forced up
the pressure gradient. Bryson and Kuhn determine the divergence by examining the fric-
tional drag across a coastal strip.
The vertical component of relative vorticity, or H V, which is that due to the local
rotation about an axis vertical to the Earth’s surface, is determined by finite differences
0 (see Figure 2.28) from:
vx vx uy uy
H· V =
L L
111
The vertical relative vorticity in plane polar coordinates is made up of two elements –
lateral shear and streamline curvature (Scorer, 1957, 1958). We can write:
∂VS ∂VS
=
∂r r
0
where VS horizontal velocity along a streamline and r radius of curvature of the
streamline. These elements may reinforce one another or tend to cancel out, as illustrated
in Figure 2.12b. By definition, vorticity in the same sense as the Earth’s rotation, cyclonic
in the northern hemisphere, is positive (Figure 2.12b). Relative vorticity is a most impor-
tant synoptic parameter, but its climatological use has been minimal up to the present.
For some purposes it is necessary to consider absolute vorticity. The vertical compo-
nent of absolute vorticity is made up of the sum of the local value of the Coriolis parameter,
f, and the relative vorticity determined by the circulation pattern. The Coriolis para-
meter which is due to the Earth’s rotation has a value of 2 sin , where latitude
0111 angle and the earth’s angular velocity. It increases from zero at the equator to a
maximum of 1.458 104 s1 at the poles. A detailed review of the Coriolis force is
presented by Persson (1998).
The rate of change of absolute vorticity (following the motion) and divergence are
related through the “vorticity equation”:
d( f )
⬵ ( f ) H· V
dt
if we neglect the effects of baroclinicity, tilting of the vortex axis, and friction. This rela-
0 tionship, which shows that horizontal convergence is associated with increased absolute
vorticity, is important in meteorological analysis. Finite difference estimates of are rather
more reliable than those of divergence since they are less commonly close to zero. However,
the neglected terms and the effects of time changes in intensity, and in the relative motion
of a circulation system, all lead to inaccuracy in the estimation of d( f )/dt.
The deformation terms are indicators of zones where frontogenesis is likely to occur.
The two elements of deformation are often considered together, although in modern
synoptic practice new parameters relating to frontal zones are being used (see p. 456).
Finally, we may note that if there is no net divergence, vorticity and deformation of the
motion, then the streamlines are straight and the velocity is unchanged. This distribution
0 is referred to as pure translation.
11
(a)
(b)
Figure 2.13 (a) A stream function field (105 m2 s1) at 200 mb over the western Pacific Ocean,
12.00 GMT, 1 March 1965. (b) The corresponding streamline (solid) and isotach
(dashed, kt) analysis. The streamlines show two anticyclonic centers in each hemi-
sphere, the streamfunction analysis only one. (After Krishnamurti, 1969, from Barry
and Perry, 1973)
11 Stream function
A horizontal wind vector, VH, can be separated into a non-divergent part and an irrota-
tional part, thus:
VH (k )
where a horizontal stream function, a horizontal velocity parameter, k a unit
vertical vector, and a gradient operator. For horizontal, non-divergent flow and VH
parallel to lines of constant as shown in Figure 2.12:
0 ∂ ∂
dx dy = 0
∂x ∂y
so that is defined by:

111
∂ ∂
u= , v=
∂y ∂x
If variations in the Coriolis parameter (f ) are ignored, in an isobaric surface is propor-

tional to gz, i.e. g z /f. Figure 2.13 illustrates a streamline field for the 200 mb level
0 and the equivalent stream functions of the rotational non-divergent wind. In the tropics
at 200 mb this comprises the major part of the total wind (Krishnamurti, 1971).
The velocity potential, , is defined by:
∂ ∂
u= , v=
∂x ∂y
It follows from the above that:
∂2 ∂2
0111 Horizontal divergence = ≡ 2
∂x2 ∂y2
∂2 ∂2
Relative vertical vorticity = 2 ≡ 2
∂x2 ∂y
where 2 the two-dimensional Laplacian operator (see Appendix 2.1).

The point of such procedures is often the elimination of some undesirable character-
istic of the kinematic properties of the wind field. For instance, for purposes such as
smoothing, or in order to obtain conformity with a particular synoptic model in the esti-
0 mation of wind flow from satellite photographs. Unfortunately the solution of the
appropriate equations in and raises serious difficulties, and for this reason Endlich
(1967) proposes an alternative procedure. By iterative methods he produces, for example,
a non-divergent wind field which still retains the original vorticity. Hence the irrotational
wind is the difference between the original and the non-divergent patterns.
2.4.5 Vertical velocity

In terms of its direct significance in the production of weather phenomena, vertical air
motion is undoubtedly the most important single parameter. Unfortunately, it cannot in
0 general be directly measured to provide routine information. Instead it has to be estimated
by one of a variety of techniques, depending on the scale of motion which is of interest
and the applicability of their various assumptions in specific instances. Large-scale vertical
11 motion is normally almost imperceptible (see Table 2.3) whereas the vertical velocities
Table 2.3 Relative magnitudes of vertical motion in different scales of system
System Velocity (cm s1) Time scale

Thunderstorm 103 One hour
Tropical storm; subsynoptic 1025 102 Six hours
(frontal zone)
Intense depression 10 Six to twelve hours
Average depression 5 One to two days
Planetary wave 1 One week
associated with local storm systems may be dramatically evidenced by the build-up of
1 cumulonimbus heads.
The simplest means of determining synoptic or larger-scale vertical velocity is based
on the continuity equation relating the local rate of density change to the mass diver-
gence per unit volume:
∂p
∂t
= ·(pV) =
∂x
冤
∂(u) ∂(v) ∂(w)
∂y

∂z 冥
where density, V wind velocity, and w vertical velocity. From this equation it
can be shown that:

wz = (zH·V)
z
where z an arbitrary height and the bar denotes a vertical average between the surface
and z. For a steady state:
1 wz = zH·V
Also, where density changes are negligible, the continuity equation becomes:
∂u ∂v ∂w
=
∂x ∂y ∂z
In practice, pressure is used as vertical coordinate and the equivalent of vertical velocity
() at an arbitrary pressure level, p, is determined by integration. Thus:
冕
p1
p = p1 H ·V dp
p
where dp/dt. Accordingly, the continuity equation here becomes:

∂u ∂v ∂
=
∂x ∂y ∂p
It is generally assumed that 0 at the surface pressure value. This is satisfactory as

long as the topography has little slope. The integral is approximated by summation over
a number of isobaric layers (Rex, 1958; Vaisanen, 1961).
The horizontal divergence is calculated by the cartesian grid or streamline methods
outlined on p. 46, or by the objective triangle method of Bellamy (1949). The triangle
method is somewhat unsatisfactory in that the value does not refer to any particular point.
11 These kinematic methods are subject to considerable inaccuracies even when the avail-
able winds are numerous (Landers, 1955), and smoothing of any computations is essential
(Palmén and Holopainen, 1962, for example). An advanced method of smoothing based
on spatial autocorrelation has been described by Eddy (1964).
A second method of computing vertical velocity relates to temperature changes in the
free atmosphere. The temperature field is assumed to be altered by horizontal advection
and vertical advection only. Observations of the former and of the local temperature
change allow the vertical motion necessary for balance to be computed:
0 ∂T dT ∂T
w = V·H T
∂z dt ∂t
where
T/
t denotes the local rate of temperature change and dT/dt that following the
111 motion of a particle. V·H T is the advection of the horizontal temperature field. For adia-
batic changes:
dT
= w
dt
0 where the dry adiabatic lapse rate. Thus:
w= 冢(∂T/ ∂t) V· T 冣 = 冢T/t冣

H
where the environmental lapse rate and T/ t temperature change along a horizontal
trajectory. Alternatively, using potential temperature on isobaric surfaces:
[(∂/ ∂t) V (∂/ ∂s)]

w=
0111 ∂/ ∂p
where V horizontal wind speed,

/
t local change of potential temperature on an
isobaric surface, and
/
s variation of potential temperature along a streamline.
This approach is particularly suited to analysis based on records from constant-level
balloons, provided diabatic effects (particularly radiative ones) can be ignored. Alterna-
tively, trajectories can be estimated from geostrophic winds. It should be noted also that
in slant ascent in a changing pressure field the adiabatic lapse rate itself may be overes-
timated by up to 1°C km1 (Staley, 1966). Such errors could seriously affect vertical
velocities estimated by this method.
0 Computations of vertical velocity in conjunction with numerical models have been
based on the vorticity equation (Collins and Kuhn, 1954), the vorticity and thickness
tendency equation (Sawyer, 1949; Bushby, 1952; Knighting, 1960), the “omega equation”
(Pettersen et al., 1962) and the “primitive equations” of motion. It is beyond the scope
of this book to do more than outline the basis of the first two of these methods. They all
involve heavy computational demands and the details are, in any case, subject to continual
improvement in terms of the degree of resolution possible and of the mathematical proce-
dures.
The approximate vorticity equation (see p. 47) is:
0 da ∂a ∂
⬵ V·a a
dt ∂p ∂p
11
where a the vertical component of absolute vorticity if the effects of friction and the
turning of vortex lines are disregarded. The latter is important in frontal zones, however.
Approximate integration of the above equation leads to the following expression:
冢冣 = 冢冣
a p

a p1000
1
2 冤冢(∂ /∂t) V· 冣
a
a
2
a
p1000
冢(∂ /∂t) V· 冣冥
a
a
2
a
where p1000 1,000 mb level and p an arbitrary pressure level. Collins and Kuhn
(1954) computed a from f. Charts at six or twelve-hourly intervals are used to give

a /
t. The vorticity advection V·a can also be calculated by graphical methods.
Knowing these terms, and assuming 0 at the 1,000 mb level, we can determine (/a)
and therefore w at any level p from the hydrostatic assumption, w /pg. The method
provides good estimates of vertical vorticity on a synoptic scale so long as the absolute
1 vorticity is not too small.
Penner (1963) has developed this type of approach in terms of thickness advection and
vorticity advection. The original method related to charts of space–mean vorticity at 500
mb and was modified for use with the 500 mb absolute vorticity analyses subsequently
adopted by the Canadian Weather Service (Harley et al., 1964).
At the level of non-divergence (typically near 600 mb but in practice the 500 mb is
assumed), vertical velocity (6) can be determined from an expression:
6 k1 (Aa )5 k2 (Az)
where (Aa )5 horizontal advection of absolute vorticity at 500 mb and Az horizontal
advection of 1,000–500 mb thickness. k1 and k2 incorporate scale factors including the
Coriolis parameter. 6 is expressed in units of 103 mb s1 (⬵ cm s1). The two terms
of the equation can be determined graphically using a special geostrophic advection scale
(Ferguson, 1961, 1963) and their individual contributions are evaluated in vertical velocity
units before adding.
1 The basis of the “omega equation” approach used by Pettersen et al. (1962) is that the
advection of temperature and vorticity will disturb geostrophic balance unless compen-
sated for by horizontal divergence and therefore the vertical motion field. The relationship
is formulated in terms of divergence and the advection of vorticity by the thermal wind.
Petterssen et al. show that vertical velocity can be separated into a component due to dry
adiabatic motion and one due to diabatic heating by the input of sensible and latent heat.
Patterns of thickness tendency associated with these components were analyzed for typical
stages of development of mid-latitude cyclones. This question has been examined further
by Danard (1964), who showed that the heating term, primarily released latent heat, is
very important with respect to the computed vertical velocity in precipitation areas. Indeed,
vertical velocity is only 25 percent of kinematic estimates if this effect is not incorpo-
rated in the computations. The quasi-geostrophic omega equation is described by Holton
(1992, p. 167); Gordon et al. (1998) provide a brief summary (see Appendix 6.1). The
generalized omega equation which assumes only hydrostatic balance is used by Räisänen
(1995) to examine ageostophic and diabatic effects. He finds that the correlation between
the vertical motion determined by the adiabatic quasi-geostrophic omega equation and the
generalized formulation for the middle and upper troposphere is 0.85 between 60°N and
90°N, 0.7 for 30°–60°N and 0.6 for 15°–30°N.
The most recent generation of numerical prediction models uses the “primitive equa-
tions” of motion (Lorenz, 1967). These are essentially one step removed from the exact
hydrodynamic equations incorporating, for example, hydrostatic equilibrium. Some general
circulation models compute vertical motion on this basis from an equation first formu-
lated by Richardson (1922) involving horizontal divergence and pressure change.
The determination of vertical motion on various scales remains a major problem in
11 synoptic meteorology. However, vertical motion fields are now routinely available for the
NCEP and ECMWF reanalysis products.
Vertical velocity at the lower boundary (in practice the surface) is usually assumed to
be zero in order to simplify the computations. However, the influence of terrain (and also
of friction) on airflow has been stressed by Graystone (1962), Haltiner et al. (1963), and
Jarvis and Agnew (1970). Quantitative assessment of terrain-induced vertical velocity, on
a synoptic scale over a time period of about one day, can be performed as follows:
w H V g · H
0 where Vg horizontal geostrophic wind and H vertical relief, smoothed over a grid
length of the order of 200–300 km. Using this approach, Jarvis and Leonard (1969) have
prepared maps of wH over central and eastern North America for 10 knot winds from
each cardinal point. It should of course be noted that in the case of major topographic
111 barriers the flow may be partially diverted or more or less wholly blocked. Friction effects
are now incorporated in numerical models but, where necessary, a simple approach could
be made along the lines of Bryson and Kuhn’s (1961) estimates of frictional divergence.
2.4.6 Isentropic charts

0 An isentropic chart shows meteorological elements on a surface of constant potential
temperature (an isentropic surface). Potential temperature () is the temperature an air
parcel attains if brought dry adiabatically to a reference pressure, usually 1,000 mb:
冢冣

1,000
=T
p
where (Poisson constant) (cp cv)/ cp 0.288, cp the specific heat at constant
pressure, and cv the specific heat at constant volume (for dry air).
Potential temperature is a valuable diagnostic measure where temperatures are to be
0111 compared at stations with different elevations. This is illustrated in a study of katabatic
winds in Antarctica by Breckenridge et al. (1993). Because air motion tends to be dry
adiabatic, the potential temperature of an air parcel is conserved. It may be noted that
the thermodynamic diagram known as the pseudo-adiabatic chart has p0.288 as ordinate so
that dry adiabats (isentropic surfaces) are straight lines. The tephigram chart is similar in
this respect.
In the atmosphere isentropic surfaces slope upward towards the poles, i.e. towards cold
air, as long as the air is stable. The slope is about 1/500 in air masses and 1/100 in frontal
zones.
The technique of isentropic analysis was first proposed by Sir Napier Shaw (1930,
0 p. 259) and was used extensively in the United States during the 1930s and 1940s by C.-
G. Rossby and his associates (1937), although it was abandoned for operational purposes
at the end of the Second World War owing to the labor involved. The isentropic surfaces
suggested for analysis over the United States are 290–5 K in winter and 310–20 K in
summer (Namias, 1940). Trajectories of air motion computed on such surfaces take
account of dry adiabatic vertical motion, as described below.
2.4.7 Trajectories
A trajectory describes the actual path of an individual air parcel moving in space and
0 time. This represents a Lagrangian view, where the coordinate system moves with the
parcels. It can be contrasted with the more common Eulerian analysis of motion fields,
where there is a fixed reference frame in space through which the air is flowing. Here
11 the instantaneous motion at specified times is depicted by streamlines of flow direction
that are tangent to the velocity vectors everywhere. Streamlines and trajectories coincide
only under stationary conditions.
Some of the earliest work on air trajectories in mid-latitude depressions by Shaw and
Lempfert (1906) was a major stimulus to synoptic meteorology. Modern studies of the
transport of pollutants and aerosols require accurate reconstruction of trajectories
and transport, and dispersion modeling has undergone progressive refinements since the
1950s. To clarify the issues involved, we will first illustrate the simplest approaches.
Suppose we wish to trace the horizontal movement of an air parcel now at point P at
some level above the friction layer during the preceding twenty-four hours, using six-
hourly charts. The simplest, but least accurate, method is as follows:
1 Determine the geostrophic velocity, V0 (km hr1), from the contour spacing at a
selected point, P0 , at the terminal time, t.
1 2 Six hours earlier (t 6) the air parcel was 6 V0 km in the upwind direction, based
on the streamline at P0. Call this point P1.
3 Determine the geostrophic velocity, V1, at t 6 at point P1. Based on the streamline
at the point the air parcel was 6 V1 miles upwind at t 12. Call this point P2,
and so on.
Errors arise in the approximation of the trajectory over some time interval by a stream-
line tangent to isobars, or height contours, owing to the displacement of synoptic systems
over time (Hogben, 1946). The accuracy may be improved if we use the mean speed at
the beginning and end of each time interval, i.e.:
(V0 V1)/2 for the period t to t 6
(V1 V2)/2 for the period t 6 to t 12, etc.
A refinement introduced by Petterssen (1956, p. 27) uses successive approximations based
on the vector mean wind for each time interval. In the boundary layer, allowance has to
be made for the effects of friction. On average, the surface (10 m) wind is backed, or
1
turned counterclockwise, from the geostrophic wind by about 10°–15° over the ocean and
20°–25° over land, owing to the frictional effect on the speed and therefore on the Coriolis
deflection.
The sources of error involved in trajectory calculation are both intrinsic and extrinsic.
The basic trajectory equation involves a position vector, x, and a wind velocity vector,
V:
dx
V[x(t)]
dt
This is normally expanded in a truncated Taylor series:

x (t1) ~ x (t0) t V(t0)
which is known as a zero acceleration solution (Stohl, 1998). The truncation error involved
in the omission of the higher order terms can be constrained by selecting sufficiently short
time steps (Walmsley and Mailhot, 1983; Seibert, 1993). To avoid aliasing, the time step
should be ≥0.5L/U where L is the length of the smallest pattern resolved in the gridded wind
field and U is the typical velocity. A fixed time step of 0.5 hr is recommended. A more accu-
rate approximation is the constant acceleration solution originally formulated by Petterssen:
1
x (t1) ~ x(t0) (t)[V(t0) V(t1)]
2
It needs to be solved by iteration, since V(t1) is not known a priori.
11 The two-dimensional iterative kinematic method developed by Petterssen provides the
simplest possible solution with first-order accuracy. It forms the basis of the NOAA Air
Resources Laboratory Atmospheric Transport and Diffusion (ARL-ATAD) model (Artz
et al., 1985).
In essence, three approaches to trajectory calculation are currently possible: three-
dimensional kinematic analysis, isobaric analysis or isentropic analysis. In complex
topography isoeta, or terrain-following, trajectories may be used. For a three-dimensional
kinematic calculation (Draxler, 1996, for example), the wind at V1 x0 V0 t and
the new vertical position is p2 p0 0.5(0 1). Draxler evaluates three-dimensional
0 kinematic and isentropic trajectories to Nova Scotia in August 1993. The motion is calcu-
lated from two-hourly NMC nested grid (90 km resolution) output of u, v, and (dp/dt)
at ten sigma levels. Isobaric trajectories often show systematic errors as a result of the
warm-air advection associated with rising motion and the clockwise (in the northern hemi-
111 sphere) turning of the wind with height.
Two-dimensional kinematic trajectory analysis can also be performed on isentropic sur-
faces. However, dynamic methods which link information on the mass and velocity fields
are also available (Danielsen, 1961, 1974; Merrill et al., 1986). In so far as the three-
dimensional motion is adiabatic, an isentropic surface can be used to calculate trajectories.
Figure 2.14 illustrates the contrasts between 500 mb isobaric and isentropic trajectories
0 over a five-day period in the Antarctic. The curvatures may be of opposite sign, and the
isentropic trajectory is generally shorter. The ascent of the air in the cases illustrated is
attributable to the air parcels crossing a steep temperature gradient and the fact that
isentropes slope upwards towards cold air.
The method of Peterson and Uccellini (1979) is based on the equation of motion for adi-
abatic flow in isentropic coordinates:
dVh
M f k Vh 0
dt
where Vh is the horizontal wind vector, is the gradient on an isentropic surface, f is
0111 the Coriolis parameter, k is a unit vector in the z direction (see Appendix 2.1) and M is
the Montgomery potential (Montgomery, 1937) or isentropic stream function ()
M cpT gz
where cpT is referred to as the specific enthalpy and gz is the geopotential, with refer-
ence to an isentropic surface. The magnitude of M is given by:
M (1.0046T 9.806z) 103 m2 s2
Trajectory wind vectors are calculated by integrating the equation of motion in isentropic
coordinates using a wind estimate at a starting position.
0 A simple isentropic model used by Pickering et al. (1994) interpolates gridded u and
v components to selected potential temperature surfaces. Parcel velocities at locations on
the isentropic surfaces are obtained by bilinear interpolation of the gridded data. Parcels
are advected (after the first time step) via a leapfrog scheme:
x(t) x(t0 t) 2t{V[x(t0)]}
where x(t) is the new parcel position at time t0 t; x(t0) is the old position and V is
the parcel velocity. Temperature is interpolated on to the parcel path and is used with the
assumption of potential temperature conservation to compute the pressure of a parcel at
its successive positions along the trajectory. Draxler (1996) concludes that 90 percent
0 of all kinematic and isentropic trajectory source pairs are located within ±75 mb of one
another and with a horizontal separation of <10 percent of the travel distance.
Wind data are available either from irregularly spaced stations and soundings or from
11 gridded analyses of observations or model output. Consequently errors arise owing to
Figure 2.14 (a) 500 mb isobaric trajectories and (b) 280 K isentropic trajectories arriving at the
south pole (SPO) on July 16 1989 (upper) and July 13 1989 (lower). The trajectories
are marked at one-day intervals up wind by a numeral. The solid line indicates 00.00
UT and the dashed line 12.00 UT. The plots in the lower right show the heights of
the trajectories (km) on each day. SPO is 2.8 km altitude. (From Harris, 1992)
spatial interpolation (see section below). In a number of related studies Kahl and Samson
(1988) found errors in mean trajectory position of between 400 km and 500 km after sev-
enty-two hours. The effect of interpolation from gridded fields can be examined by artifi-
cially degrading high-resolution gridded values and by varying the temporal sampling.
While linear interpolation is most accurate in time, errors in spatial interpolation can be
reduced by using higher-order interpolations (Walmsley and Mailhot, 1983; Stohl et al.,
1995). The largest errors arise through interpolation of the vertical motion due to its great
11 local variability. Trajectory calculations are also sensitive to the temporal frequency of
wind data (Doty and Perkey, 1993). If diurnal variations in flow are important, a minimum
time step of six hours is required to capture these variations. At high spatial resolutions of
90 km or 180 km, Rolph and Draxler (1990) find that trajectory errors can be reduced most
by shorter time steps. However, at 360 km resolution this is no longer true below about six
hours. Where three-dimensional motion has to be considered, the largest errors generally
arise through interpolation of the vertical component owing to its high variability.
From an analysis of long-range trajectories into the Arctic, Kahl et al. (1989) demon-
strate that the greatest uncertainty in the calculations arises from model sensitivity to input
0 data rather than through the parameterization of vertical motion. Kinematic methods can
give results that are in most cases accurate, if the three-dimensional wind data used in
the calculations are from dynamically consistent analyses (Fuelberg et al. 1996). Stohl
and Seibert (1998) also find that most inaccuracies result from dynamic inconsistencies
111 between meteorological fields. They use potential temperature, potential vorticity and
specific humidity as three-dimensional tracers. They show that three-dimensional isen-
tropic trajectories are optimal but, in the troposphere, kinematic isentropic ones are the
next best option.
Trajectory models developed by Heffter (1980, 1983) have been widely applied in
research on precipitation chemistry (Raynor and Hayes, 1982), pollen transport in eastern
0 Canada (Barry et al., 1981), and flow patterns for South Pole (Harris, 1992). Usually,
five-day or ten-day back trajectories are determined, but forward trajectories may also be
obtained (Small and Sansom, 1983). Harris and Kahl (1990, 1994) provide climatologies
of back trajectories for Mauna Loa, Hawaii, and Barrow, Alaska. For Mauna Loa, Hawaii,
at 700 mb and 500 mb an East Asian source is shown to be frequently present, espe-
cially in winter when the trajectories are longest. Kahl et al. (1997) map ten-day isobaric
back trajectories at 500 mb to Summit, Greenland, for each day of 1946–89. Cluster
analysis was then used to identify seasonal source regions and transport routes. In winter
58 percent of trajectories extend westward into East Asia, while 27 percent follow a
slower zonal path from Canada. In summer East Asia accounts for only 34 percent, the
0111 North Pacific 31 percent, and Canada 26 percent. A climatology of seasonal air mass
trajectories for North America has also been developed by Bryson (1966). Trajectories
have been studied in relation to synoptic types for the North Atlantic (Haagenson and
Sperry, 1989) and northern England (Dorling et al., 1992).
2.4.8 Spatial interpolation

A common problem in climatological analysis is the contouring of irregularly spaced point
data. Various methods are available for such spatial interpolation (Watson, 1992) and for
smoothing and filtering (Essenwanger, 1986). In general, spatial interpolation uses linear
0 combinations of known values at control points to estimate the values at intermediate grid
points. A well established procedure in the meteorological literature is the Cressman (1959)
interpolation scheme. This involves a distance-dependent weighting of data at the control
points within a specified search radius.
For example, an estimate of a climatic variable (E) at a grid point i, j is calculated
from:
E=
兺E W 1 1
0
兺W 1
where W1 is the weight given to observation E1.
11 N2 d2
W=
N 2 d2
where d is the distance between the control point and the i, j grid point, and N is the
search radius at which W → 0. Nuss and Titley (1994) propose an improvement over the
Cressman interpolation scheme, in terms of RMS errors, using a multiquadratic radial
basis function. This has the form:
Q [X Xi]
where the argument represents the vector between an observation point Xi and any other
point in the domain. The spatial field S(X) is described by:
N
S (X) = 兺 Q (X X )
i1
i i
where the coefficients i are weighting factors. For a two-dimensional case the basis
1 function becomes
{| x xi |2 | y yi |2 1.0}1/2
Qi (x, y) =
c2
where c is an arbitrary, typically small, constant.
For mapping on a regional scale, and where there is a dense network of observations,
spatial interpolation can generally ignore the effects of the earth’s curvature. However,
errors grow in data-sparse regions as the discrepancy between true spherical distances and
approximate planar distances (and angles) increases with wider spacing of data points.
For continental- to global-scale analyses, simple interpolation methods within a carto-
graphic projection can introduce significant errors (Willmott et al., 1985). Such errors
may be apparent on global maps where the latitudes of isolines differ at the longitudinal
margins, 180°E and 180°W, for example, or where there are multiple isolines passing
through the poles, as depicted on global-scale rectangular projections (Robeson, 1997).
Techniques are now available enabling a number of spatial interpolation procedures to
1 be applied on a sphere. The main categories are: inverse-distance weighting, kriging,
optimal statistical objective analysis (OSOA), spline methods, tesselation, and spherical
harmonics (Robeson, 1997). An overall consideration is whether the method adopted
uses a subset of the data at each step (a “local” method) or whether all the data are used
together (a “global” method). An inverse-distance weighting scheme is used in the analysis
of both global surface air temperature and continental precipitation by Legates and
Willmott (1990) and Willmott et al. (1994). In addition to weighting control points
according to great circle distances, the angular distribution of control points relative to a
grid point and the spatial gradients within the data are considered. An optimally modified
variant of distance weighting is the OSOA approach pioneered by Gandin (1965). Here
the appropriate form and parameters of the distance-weighting function are determined
from the spatial covariance in the data. For example, the decay of spatial correlation in
contour height fields has been investigated for the standard isobaric levels at stations in
Europe and the North Atlantic by Bertoni and Lund (1963), as illustrated in Figure 2.15.
For tropospheric height, wind and temperature fields the spatial covariance can often be
modeled by a low-order trend surface. Exponential functions are often appropriate,
although sometimes they are directionally dependent, or anisotropic (Thiebaux and Pedder,
1987).
A method that is similar conceptually to OSOA, known as kriging, may also be used.
In meteorology a temporal series of fields (or realizations), is usually available, whereas
for terrestrial science applications there may be only one data set. Kriging, which also
accounts for the distance–decay relationship in the observed variable, uses a measure of
the variation of the observations with increasing separation distance known as the semi-
variogram. Kriging involves a two-step procedure. First, a spatial structure function (the
0
0
0
0
11
11
0111
111
Figure 2.15
Illustration of the
spatial decay of
correlation in sea-
level pressure and
500 mb height fields
over Europe
(Hanover, Germany)
and the North
Atlantic (Ocean
Weather Ship C).
(From Bertoni and
Lund, 1963)
semi-variogram) is calculated from the data and fitted by a model that describes the spatial
continuity of the data. Observations are split into a random part and a deterministic part
that is described by the spatial structure. Then grid values are estimated (kriging) at each
grid node based on data points in their neighborhood and the variogram model. The semi-
variogram describes the transition from good covariation for nearest-neighbor samples to
weaker relationships with increasing sample distance; the form of the variogram may be
linear, Gaussian, or spectral. In kriging the unknown covariances are replaced by the vari-
ogram model values so as to minimize the error variance.
An example is provided for the distribution of January temperature in Scotland by
Hudson and Wackernagel (1994). The spatial variability of some climatic element (x) at
different spacings is assessed by a dissimilarity measure:
(x2 x1)2
* =
1 2
The dissimilarity * depends on the spacing and orientation of a pair of points which is
described by a vector, h ( x2 x1):
*(h) = 0.5[z(x1 h) z(x1)]2
The experimental variogram is obtained by forming the average of the *(h) for all Nh
point pairs that can be linked by a vector h:
Nh
兺 [z(x h) z(x )]
0.5 2
*(h) = k k
Nh k=1
In general, topography modifies the spatial pattern of mean precipitation and tempera-
ture. Information on altitude variations can be included in the estimation procedure
through the method of cokriging, which treats two interrelated variables. Digital eleva-
1 tion data can be incorporated to improve the spatial representation of the rainfall amounts
by determining variogram functions for both precipitation and altitude, and a cross-
variogram of the two variables. Further improvement may be achieved through kriging
with external drift. This model uses a functional relationship between precipitation and
altitude. It is mathematically simpler than cokriging because it only requires a variogram
be calculated for mean precipitation. Examples of Thiessen polygon, kriging, cokriging,
and kriging with external drift (KED) estimates of rainfall in a region of southern Spain
are presented by Pardo-Iguzquiza (1998). He finds that the KED approach gives the most
coherent results, based on a cross-validation assessment. This consists of (1) deleting the
data points from the data set one at a time (2) recalculating the estimates using n1
points, and (3) averaging the estimates over the n deletions of a data point (see Efron
and Gong, 1983).
Spline methods involve local interpolation and are most commonly used for one-dimen-
sional data such as time series. A common interpolation tool is the cubic spline which
locally fits a parabola piecewise to a curve. Constraints can be imposed by ensuring the
existence and continuity of both first and second derivatives of the data series (i.e. smooth-
ness of fit). A thin-plate spline fitting of annually averaged air temperature anomalies is
illustrated by Robeson (1997). Such splines solve a differential equation that describes
the flexing of an infinitely thin plate constrained by point loads at the control points
(Wahba, 1990). They are useful in eliminating small-scale variability or errors in a data
set. Tesselation methods allow a network of control points to be decomposed into a
number of polygons or surface “patches.” This approach was first proposed by Thiessen
(1911), using triangulation between stations to calculate areal average precipitation. The
sides of the triangles are bisected by perpendicular lines which form the sides of poly-
11 gons surrounding each sample point. It is essentially a nearest-neighbor method; all points
inside the polygon are estimated by the nearest sample point. Weights for each station,
or sample point, are proportional to the area ascribed to its polygon. In tesselation an
interpolation surface is fitted to data values at the vertices of the polygon as local control
points. The Thiessen method has been developed subsequently by Diskin (1970)
and others for computer application. A practical problem needs to be noted, that data
for all stations in a network may not always be available and networks change over time,
requiring new weights to be determined. Theoretical questions of the averaging of
meteorological fields, and related practical considerations, are discussed in depth by Kagan
0 (1997).
Spherical harmonic functions (see Appendix 4.1) are widely used in global meteoro-
logical analyses and general circulation models, but they have been little used to date in
the spatial interpolation of global observations.
111
2.4.9 Spatial coherence
Meteorological fields exhibit varying degrees of spatial coherence according to the vari-
able under consideration. The synoptic-scale decay of spatial autocorrelation in
geopotential height fields has been demonstrated by Bertoni and Lund (1963). Figure 2.15
0 shows that there is little change in the scale of the autocorrelation from the surface to
100 mb. Gandin (1965) also examines the correlation structure of MSL pressure, 500 mb
heights and 850 mb dewpoint temperature. Correlation fields for six-day and monthly
precipitation totals are usually elliptical in shape (Cornish et al., 1961). The complex
interaction of moisture sources, precipitation type and topography help shape this
anisotropic structure (Caffey, 1965). For storm rainfall in Illinois the spatial variability
and therefore the correlation decay are largest with air mass showers and least near low-
pressure centers (Huff and Shipp, 1968). The variability also increases as the area
considered is enlarged and decreases exponentially as the network mean precipitation
increases. Based on observations of simultaneous precipitation occurrence in winter, using
0111 the synoptic “present weather” code for stations around Frankfurt, Germany, Wachter
(1968) finds that the frequency distribution is bimodal at around 500 km radial distance
because the area covered by the stations is smaller than the average precipitation area.
Isocorrelate maps of annual precipitation in Great Britain, based on observations at Oxford
and Glenquoich in western Scotland, prepared by Glasspoole (1925), show that at least
two precipitation regions can be distinguished. Similar analyses of monthly and seasonal
precipitation have been carried out for stations north and south of the Alps (Fliri, 1967)
and over Norway (Nordø and Hjortnaes, 1967).
The spatial relationship between climatic elements and atmospheric circulation features
has also been extensively investigated (Stidd, 1954; Klein, 1963 1965; Klein and Kline,
0 1984). Commonly, station values of temperature or precipitation are correlated with
concurrent MSL pressure or geopotential height anomalies. The isocorrelate lines indi-
cate the direction, curvature, and origin of anomalous flow components. Figure 2.16, for
example, illustrates the patterns of correlation between surface temperature and 700 mb
height in winter in the contiguous United States. Dickson (1971) correlates the variance
of daily mean temperatures at eight stations in the United States, separated into three
temporal spectral bands, with seasonal mean 700 mb heights over the North Pacific and
North America. He demonstrates increasing wave amplitude in winter for temperature
oscillations with periods longer than twenty-two days compared with three to ten days.
0
2.4.10 Field intercomparison
The intercomparison of meteorological fields is greatly complicated by the presence of
11 spatial autocorrelation. For static or mean climatological distributions a procedure based
Figure 2.16 Patterns of correlation (negative maximum shaded) between winter temperature and
700 mb height over the contiguous United States. (Klein and Kline, 1984)
on nominal data may be appropriate. Monserud and Leemans (1992) use the kappa ()
statistic (Cohen, 1960) to intercompare global vegetation patterns. A subjective scale of
comparability is used where < 0.4 is a poor match and > 0.85 is an excellent match.
The comparison of a pair of, or more, meteorological fields raises the question of pat-
terns of statistical significance. Correlation fields can be evaluated first by calculating the
Student’s t statistic at all available grid points and evaluating the statistical significance at
the 5 percent level. However, some areas are likely to be correlated by chance and there-
fore it is necessary to assess the field significance rather than just the local significance
11 (Livezey and Chen, 1983). Moreover, because the spatial correlation diminishes the degrees
of freedom the criteria for field significance are more stringent.
The spatial degree of freedom in a climatic field is considered by Wang and Shen
(1999), using four different methods. They find that approaches based on 2 for the distri-
bution of squared differences between realizations of a field, the transformed Z score of
pattern correlation coefficients between two realizations, and the ratio of the variance of
the mean to the mean variance, all tend to give underestimates. This occurs when there
is an insufficient number of realizations of the field, or when the mean and variance vary
spatially. Estimation based on the binomial distribution with Monte Carlo trials (Livezey
0 and Chen, 1983) is preferable. The spatial degrees of freedom estimated for 5° × 5°
gridded air temperature data by Wang and Shen are as follows: for 1,002 available grid
points in the northern hemisphere there are about sixty (ninety) degrees of freedom in
winter (summer). In the southern hemisphere for 658 grid points there are between thirty-
111 five and fifty degrees of freedom, with no clear seasonal cycle.
In the case where the correlation between a temporal index, such as that for the Southern
Oscillation, and global climatic anomalies is examined, two approaches are possible in
assessing the statistical evidence for pattern correlation (Livezey, 1995). A random series
can be generated for the temporal index by random reordering of the data; the local corre-
lations are then recalculated and the number of chance correlations is determined. The
0 resampled series are formed either without replacement (permutation) or with replace-
ment (bootstrap) (Efron and Gong, 1983). The percent of area where the correlations are
significant at some prescribed probability level can then be evaluated. In the case of a
pair of maps, best expressed as anomaly fields with zero mean, the grid-point values
on one of the maps are resampled, giving a distribution from which the probabilities of
statistical significance can be calculated. Graham et al. (1994) discuss the comparison of
time-dependent model simulated fields with observed fields based on some long-term
record (say 1951–90). For example, resampling could be based on twenty-six sets of
fifteen-year contiguous records: say, 1951–65, 1952–66 . . . 1976–90. The pointwise corre-
lation field for the matching simulated and observed data is compared with the
0111 non-matching cases for the different sets of years. Probabilities of obtaining the matching
case value, at each point, are compared with those from the non-matching sets. This
method retains any temporal autocorrelation between the years or seasons, which is lost
when random sets are drawn by permutation.
2.5 Time series

2.5.1 Serial correlation
Almost any set of climatological data is ordered chronologically (see Table 2.1) and so
0 constitutes a time series. Time series of meteorological variables are usually based on
equally spaced observations, although this is not always the case. They exhibit a wide
range of behavior. Typically there are quasi-periodic or irregular fluctuations about some
general mean level. However, for daily and annual series there are more regular periodic
variations. Over several decades a series may also contain some apparently abrupt changes,
marked by shifts in the mean, as well as trends in the mean level. Alternatively the mean
may remain unchanged but the variability, or frequency of extremes, increases or decreases
(Hare, 1979). These changes may be the result of an artifact, such as a change in instru-
mentation, observational procedures, station location or its surroundings, or due to errors
in data recording and transcribing. There is a large literature on appropriate statistical
0 techniques for the analysis of time series in general and meteorological data in particular.
Reference works include Brooks and Carruthers (1953), Chatfield (1989), Box and Jenkins
(1976), Kendall (1976), Essenwanger (1986), and Polyak (1996). Here only a few selected
11 issues will be addressed.
A time series usually consists of a deterministic element and a random or stochastic
element, often referred to as white noise. The deterministic element may be strictly peri-
odic (annual and diurnal cycles, for example) or it may be transient, such as a trend of
indefinite period. White noise can be simulated by a random number generator available
in statistical program packages.
Determination of the characteristics of a time series is commonly hampered by the
availability of only a limited record length, i.e. the series is truncated. The observations
represent a sample of a population that cannot in general be defined because meteoro-
logical phenomena tend not to occur randomly. It is well recognized that most
meteorological time series possess a high degree of persistence or serial correlation (auto-
correlation) (Schumann and Hofmeyr, 1942). Such persistence is also termed red noise.
The time series of a simple first-order autoregressive (AR(1) or Markov) process,
consisting of a trend and stochastic fluctuations, can be written:
1
xt = r1[xt1 ] t
where is the population mean, r1 is the lag-1 serial correlation of xt , t is time, and t
is Gaussian white noise (with zero mean and constant variance). Autoregressive moving
average (ARMA) processes are commonly used to model time series. Katz and Skaggs
(1981) recommend that the data be transformed if they are non-Gaussian (see above)
before fitting the ARMA process. Non-stationarity effects can be suppressed by forming
differenced series (Yt Yt1) and fitting the ARMA process to them. They show that an
AR(1) Markov (or red noise) process fits 90 percent of Palmer drought index data for
climate divisions in the United States. There are similar examples for dry and wet spells
in many other locations.
At the daily time scale, persistence is readily apparent for variables like pressure and
temperature. For surface air temperature in winter, the one-day lag correlation varies
between 0.7 and 0.8 over the western half of the continental United States and is about
0.6 on the east coast (Madden, 1979). For sea-level pressure over the northern hemisphere
in winter, the one-day lag correlation varies from 0.8 over western Europe and the subtrop-
1 ical western North Pacific to <0.5 over most of eastern North America and off Japan
(Schumann and Van Rooy, 1952). For a lag-1 correlation of 0.8 (0.5), the characteristic
time between independent estimates is approximately eight (three) days, respectively,
assuming a first-order autoregressive model (Madden, 1979). These estimates imply that,
for daily data, a stable estimate of the monthly mean can be acquired from pressure obser-
vations on only four days, each eight days apart (or ten days, each three days apart),
according to location, instead of the entire thirty or thirty-one days in the month (Figure
2.17). The time between independent samples of surface pressure also varies seasonally.
At Zurich, Switzerland, it is 4.8 days in summer and autumn, 6.6 days in spring, and
12.3 days in winter (Madden and Sadeh, 1975). For twice-daily winter isobaric height
fields (850 mb, 500 mb and 200 mb) covering most of the northern hemisphere for
1963–67, Lorenz (1973) reports detectable and significant persistence of the patterns out
to twelve days, although at lag-4 the autocorrelation drops below 0.3. The characteristic
time between independent estimates (T0) for different averaging times (T) in an AR(1)
model is depicted in Figure 2.17. Here T0 is determined from the expression
T0 [1 2(1 1/T)r1 2(1 2/T )r12 . . . (2/T)r1T1]
where r1 is the lag-1 autocorrelation. The corresponding variance of the time average of
length T is given by
T0 2
T 2 =
T
where 2 is the variance of daily data (Madden, 1979).
11
111
0 Figure 2.17 The characteristic time between independent estimates (T0) for different averaging times
(T ) of a first-order autoregressive (Markov) process. (From Madden, 1979)
Persistence on long time scales is typical of many geophysical time series. This tendency
in hydrometeorological records was labeled the “Joseph effect” (seven lean years and
seven fat years) by Mandelbrot and Wallis (1968, 1969c). Another characteristic of long
series, their tendency to contain very extreme high and low values, was referred to as the
“Noah effect.” Statistical models to treat such records were developed by Mandelbrot and
Wallis (1969a, b, d). In particular, a statistic proposed by Hurst (1951) for data rescaling
was found to provide a useful analysis tool. The Hurst approach is to remove the
0111 mean from each of n observations and to accumulate the deviations. The Adjusted Range
R(n) (Rmax Rmin) and the Rescaled Range R(n)/s, where s the standard deviation,
varies as nH. The Hurst exponent:
log [R(n)/s]
H=
log n
has an expected value of about 0.5, but is commonly in the range 0.6–0.9. This tendency,
referred to as the Hurst phenomenon, is interpreted to indicate persistence (Mandelbrot
and Wallis, 1969d). However, it may show non-stationarity of the mean (Klemes, 1974),
0 including that caused by artifacts such as a move in station location or a change in instru-
mentation. Outcalt et al. (1997) used the Hurst rescaling to identify and characterize
periods within a time series having different regimes. The method is illustrated for the
sunspot time series, as well as for precipitation, discharge, and soil temperature data.
There are various techniques available for filtering a time series in order to exclude
certain specified frequencies. The simple moving average, or running mean, has equally
weighted filters. These have the effect of reducing to zero the amplitude of oscillations
with a period that is an integer multiple of the length of the moving average, i.e. if the
length of the moving average is t, the response function is 0 for periods of length t, 2t,
3t, etc. Moreover, the response function is negative for periods between t and 2t, 3t and
0 4t, . . . which means that the phase of such periodicities in the smoothed series is inverted
by comparison with the original data. This tendency can be lessened by the use of weighted
moving averages, especially with weights such as 0.25 Xt1, 0.5Xt , 0.25Xt1 (the Hanning
11 filter). It will also be noted that the smoothed series is shorter than the original one, losing
one value at each end for a three-term filter and two values at each end for a five-term
filter. If the number of terms in the filter is odd the new series can be centered, begin-
ning on the second (third) term in the original series for a three (five)-term filter,
respectively. If the filter has an even number of terms, centering requires a second step
in the analysis. Thus for a twelve-month running mean, centering is achieved by deter-
mining a two-month moving average of the values from the first filtering step. The initial
value of the twelve-month centered moving average then refers to July if the original
series began in January, so that six values are lost at each end.
Special filters can be designed to eliminate or suppress unwanted frequencies. Low-
pass filters are used to examine long-term fluctuations (Craddock, 1957) while band-pass
filters select only variations close to a specified period (Landsberg et al., 1963). A detailed
discussion of the design of filters with different polynomial degree and filter width is
given by Polyak (1996).
1 Serial correlation is a serious concern for many standard tests of statistical significance
which require statistically independent data samples (von Storch, 1995; Zwiers and von
Storch, 1995). If the latter condition is not fulfilled, the number of degrees of freedom
has to be adjusted. For example, for an AR(1) process with thirty cases, the effective
sample size is reduced to seventeen for r1 0.3, and to only eight for r1 0.6 (Thiebaux
and Zwiers, 1984). For this reason the standard error of a sample mean has to be expressed
as s/(n1)0.5 where n1 denotes the number of independent samples. The effects of auto-
correlation for a first-order autoregressive series can be filtered to remove the red noise
by “pre-whitening” the series (von Storch, 1995). Here the original time series xt is replaced
by:
yt xt r̂1 xt1
where r̂1 is the estimated lag-1 autocorrelation and t is time.The serial correlation coef-
ficient rk , where k is the number of lags, is given approximately by:
1 rk =
兺(x x )(x x )
i ik
兺(x x )
i
2
Most techniques of time series analysis assume that the series is stationary, meaning that
a shift in the time origin has no effect on the distribution function. In other words, the
mean, the variance, and the higher moments are essentially independent of time. In reality,
climatic time series are commonly non-stationary.
A common problem is identifying shifts or breaks in a time series. A traditional solu-
tion in hydrometeorology is the double mass curve analysis (Kohler, 1949). Cumulative
values of the variable at a station being tested are plotted on the ordinate against the
cumulative mean values for six to ten neighboring stations on the abscissa. If the series
is not homogeneous through time the plot will show a break in the slope of the line. The
amount of change can be estimated and the date when it occurs. However, there is no
objective basis for assessing whether a change has actually occurred. Also, suitable neigh-
boring stations are not always available. There are various other methods in use. Statistical
evaluation of homogeneity can be made using a non-parametric “runs test” with respect
to values above and below the median (Thom, 1966). A procedure for detecting a break
and assessing its “quality” is developed and illustrated by Oerlemans (1978). A break is
defined as the ratio of the amplitude of a change to the corresponding root-mean square
(RMS) difference between the observed change and a “typical” one that is defined a
priori. Oerlemans provides a sample analysis of daily temperature, precipitation and
sunshine duration time series at de Bilt, Netherlands for 1949–74 and finds little corre-
lation between identified breaks. He also presents a spatial application for geopotential
11 height fields and shows a preferred blocking sector over the northeast Atlantic. Howell
(1995) describes a procedure based on adaptive filtering where the record is divided into
blocks and the mean shifts at block boundaries are determined. A bivariate test to detect
a shift in the mean of a given series against a second correlated series was proposed by
Maronna and Yohai (1978). Its application to a precipitation series is described by Potter
(1981), who also demonstrates that it is closely related to double mass analysis. The
test indicates whether a change has occurred and provides maximum likelihood estimates
of both the time (i) and the amount (d) of the change in the mean. Potter shows that the
original formulation can be written such that the linear estimator of a sequence:
0
冢 S 冣 (x X )
Sxy
ŷi = Y i
x
111 where
n
Sxy = 兺 (x X) (y Y )
j =1
j j
– –
is the covariance of the two series. Sx the variance of Xi , X and Y are the series means.
0 A corresponding estimator for Yi where:
i
兺y
1
Yi = j
i j =1
is:
冢 S 冣 (X X ) .
Sxy
Ŷ i = Y i
x
0111 The difference between the regression estimate of Yi and the actual value, with a correc-
tion term (in brackets [ ]) to account for the effects of non-zero d on Sxy , is:
(Ŷ i Yi )
Di =
[(n i) Fi /nSx]
where
Fi = Sx (Xi X )2 冢n ni i冣 , i<n

0
The test statistic is:
T0 = max {Ti}
i <n
where
[i (n i) Di2 Fi]
Ti =
Sx Sy Sxy2
0 When T0 exceeds a critical value depending on n, the likelihood ratio test shows rejec-
tion of the null hypothesis of no shift in the mean. Potter (1981) gives critical values of
T0 as follows: for n 30, T0 8.2 for the 0.05 significance level and 10.7 for the 0.01
11 significance level; for n 100, T0 9.3 and 12.5, respectively. The value of i for which
T0 is a maximum is the maximum likelihood estimator (i0) of the year before the change;
the corresponding estimator of d is Di 0 . The test assumes a normal distribution and inde-
pendent values in the series, which is generally the case for annual precipitation totals.
Another method, developed by Easterling and Peterson (1995), combines regression
and non-parametric statistics. A simple linear regression is first fitted to a difference series
of the candidate station from a homogeneous reference series based on several correlated
stations. The residual sum of squares (RSS) is calculated and then a further two-phase
regression is performed. Segments are tested by adding one year at a time. The point
with the minimum RSS sum from each of the two regressions is noted as a possible
discontinuity. The significance of the two-phase fit is tested by a likelihood ratio statistic
(Solow, 1987).
The detection of trends is more problematic. A frequently used method is the Kendall
statistic, illustrated for monotonic trends by Dettinger and Cayan (1995). A modified
1 form is presented by Hamed and Rao (1998). The number of years of record needed to
detect a trend depends on the magnitude of both the variance and the autocorrelation of
the noise in the series (Weatherhead et al., 1998).
2.5.2 Periodic components

Periodic components of a time series can often be represented by a series of sine and
cosine functions. Figure 2.18 illustrates these basic trigonometric functions for the range
0 to 2 , which can be equated with a diurnal or annual time interval, and the harmonic
curves for n 1, 2, and 4. The series for a finite function f (x) is:
Figure 2.18 (a) Sine and cosine functions between 0 and 2 (0–360°). (b) Simple harmonic curves
for n 1, 2, and 4; the amplitude is shown as being reduced by 50 percent from
n 1 to n 2 to n 4. (From Barry and Perry, 1973)
11 f (x) a0 a1 cos x a2 cos 2x . . . an cos nx
b1 sin x b2 sin 2x . . . bn sin nx . . .
冤兺 a cos kx 兺 b sin kx冥

n n
2
= a0 k k
n k=1 k=1
where a0 the mean, n the total length of the time period. Note that in total there are
n/2 harmonics. The expression can be rewritten:
n/2
0 f (x) = a0 兺 [A cos (kx )]
k=1
k k
where Ak (ak2 bk2)1/2, the amplitude of the harmonic, and:

111
arctan (b/a)
arcsin (b/A)
which avoids the ambiguity of two solutions for the tangent between 0 and 2. repre-
sents the phase difference (time interval) between each harmonic or wave. It follows that:
0 a = A cos and b = [A cos (kx )]
Computer programs for determining the amplitude and phase are widely available. They
incorporate an algorithm known as the fast Fourier transform which significantly shortens
the computation time. Many studies of annual precipitation have used harmonic analysis
to describe the characteristics of the regime and to provide a basis for the delimitation
of regions having a similar regime (Horn and Bryson, 1960; McGee and Hastenrath,
1966). The first harmonic denotes an annual cycle with a maximum and minimum and
the second a semi-annual cycle as illustrated in Figure 2.18. The contribution of each
harmonic to the total variance is calculated from:
0111
Ak2
2s2
(except for k n/2, where the contribution is twice the value) where
n/2
2s2 = 兺A
k=1
2

In cases where the first harmonic accounts for a large proportion of the variance it is worth
0 while mapping this value (the amplitude) and the phase angle. Figure 2.19 shows the phase
angle of the first harmonic over the United States and the ratio of the second to the first
harmonic. The latter indicates the relative strength of the semi-annual component over the
intermontane west. Figure 2.19a indicates that the midwinter maximum (90°) is earlier
going southward in Baja California although the amplitude (not shown) also decreased
sharply, while east of the Rockies the spring maximum (300°–315°) in Wyoming is delayed
until mid-July (270°) in Wisconsin. In Tennessee the 60° line demarcates a February max-
imum. The map indicates some marked changes across New England but, in spite of the
importance of the first harmonic evidenced by the ratio map, it should be noted that the
amplitude of the first harmonic is small. Seasonal contrasts in this region are weak.
0 The results of harmonic analysis need to be interpreted with care. Periodic processes
shorter than semi-annual are unrealistic (Rayner, 1971). Moreover, harmonics of the annual
cycle with frequencies of two, three, four, five, and six cycles a year are superimposed
11 on the computations when monthly data are used.
Figure 2.19 (a) Phase angle of the first harmonic for precipitation over the United States. (b) Ratio
of the second to the first harmonic. (After Horn and Bryson, 1960)
11 Early literature on time series made extensive use of correlograms showing the serial
correlation plotted against the time lag in days or years (Alter, 1933; Wallis and Matalas,
1971). Pronounced peaks in a correlogram represent periodicities at the given lag.
However, the statistical significance of the autocorrelations is not easily determined and
the correlogram cannot be interpreted for long lags because of limited record lengths.
2.5.3 The frequency domain

Because of the severe limitations of analyses in the time domain, time series are now
0 usually investigated through the frequency domain. The variance spectrum (or power spec-
trum) allows readier physical interpretation of a time series and is considerably more
tractable in the evaluation of statistical significance. Essentially, the variance of the quan-
tity examined (Ak2 /2) is plotted against the frequency (cycles per unit time interval).
111 Frequency is the reciprocal of period; thus 0.14 cycle per day corresponds to a period of
approximately one week and 0.33 cycle per day to a period of one month. In practice,
this direct approach is unworkable owing to sampling variations. Instead, the serial corre-
lations are computed and subjected to Fourier analysis. The coefficients, ak , are then
smoothed by a weighted moving average and plotted against frequency. The spectrum is
normalized by the use of the serial correlation coefficients (which have sx2 in the denom-
0 inator), so that there is unit area under the curve. The basic procedures of spectral analysis
are detailed in numerous sources (Blackman and Tukey, 1958; Julian, 1967; Rayner,
1971).
Several basic types of variance spectrum can be identified. Figure 2.20a shows,
schematically, spectra for white noise (random uncorrelated data), quasi-periodicities, and
red noise, where low frequencies (long periods) contribute most of the variance. A pure
periodicity would appear as a vertical line in the frequency domain. In this diagram
frequency is plotted on the abscissa, although it is often preferable to plot the log of
frequency. Mandelbrot and Wallis (1969a) suggest that simple frequency plots may indi-
cate randomness where none exists. An important concern in spectral analysis is the
0111 occurrence of aliasing between different frequencies. The shortest period (highest fre-
quency) that can be identified in a data series is 1/2 t where t is the time interval
between observations. This is known as the Nyquist frequency, F0. Shorter fluctuations
cannot be resolved by the available data. For example, at least two observations per day
are necessary to estimate the diurnal variation. For any frequency, f :
f, 2F0 f, 2F0 f, 4F0 f, 4F0 f …
are aliases such that variance is added to them by unresolved frequencies which exceed
F0. Figure 2.20b illustrates aliasing between frequencies. The low-frequency limit in spec-
0 tral analysis is between one cycle per n/5 to n/10 , where n is the number of observations.
Modern methods of time series analysis can be divided into four categories: Fourier,
maximum entropy, singular spectrum, and wavelet techniques (Yiou et al., 1996). Only
a brief outline of these approaches and their differences can be given here. The basic
technique of Blackman and Tukey (1958) depends on the fact that the lag autocorrela-
tion function (x) of a time series corresponds to a Fourier transform of its variance
spectrum (Vx) and vice versa (Jenkins and Watts, 1968). The variance spectrum is esti-
mated via this relationship using discrete functions:
M
0
Vx( f ) ≈ 兺 f (k) k e
k=0
x
if k
where the first M1 autocorrelation coefficients are {x (k), k 0 . . . M}, and f denotes
11 frequency. A plot of the autocorrelation function x(k) against lag k is called a correlogram.
Figure 2.20 (a) Schematic variance spectra for (i) random data (white noise); (ii) quasi-periodic
oscillations at three frequencies; (iii) an autoregressive series with low frequencies
(persistence or red noise) predominating; (iv) an autoregressive series with high
frequencies predominating. (b) Aliasing of waves (after Blackman and Tukey, 1958).
Sampling causes the sinusoidal wave with a frequency of 4 to appear as a wave of
frequency 1. For time interval 1 the Nyquist frequency is 0.5. For a time interval of
0.2 the Nyquist frequency is 2.5. (From Barry and Perry, 1973)
11 The above approximation smoothes the spectrum, reducing the variance of the spectral
estimates. M is usually chosen to be less than n/5 to avoid spurious results, as noted
above. Estimation of a continuous spectrum by a discrete Fourier transform causes spec-
tral leakage of variance due to the side lobes associated with the discrete step (or boxcar)
function. A window or taper (f(k)) is selected which decreases this leakage. It may be a
modified cosine, cubic or tent function.
The calculations have been transformed by the use of the fast Fourier transform or FFT
(Rayment, 1970). This provides a computationally efficient means of transforming a func-
tion into its Fourier components. The spectral resolution is generally poor when the number
0 of data points, n, and the truncation of M are low. The multi-taper method is a non-para-
metric spectral method. It provides an optimal set of tapers, designed to minimize spectral
leakage, that are based on eigen vectors. The total variance spectrum is obtained by aver-
aging the individual spectra from the independent estimates of each tapered signal (Yiou
111 et al., 1996). There are generally no more than seven tapers. However, the number of
tapers and their bandwidth need to be varied to ensure that the frequency estimates are
stable. The method appears suitable for identifying low-amplitude oscillations in relatively
short time series (Mann et al., 1995).
The maximum entropy method (MEM) (Ulrych and Bishop, 1975) is suitable when the
time series is stationary and is the result of a first-order autoregressive process (Ghil and
0 Yiou, 1996). However, the existence of an AR(1) process is sometimes a contentious
issue, as shown for sunspot numbers by Sneyers (1976). The basis of the method is the
determination of the variance spectrum which corresponds to the most random process
having the same autocorrelation coefficients . In terms of Shannon’s information theory
this random state represents maximal entropy, hence the name. Spectral estimates by MEM
depend heavily on the choice of M. Various heuristic criteria have been suggested but
these do not necessarily prevent errors in the method, according to Penland et al. (1991).
Singular spectrum analysis (SSA) is specifically designed to treat short, noisy time
series resulting from non-linear dynamical processes (Penland et al., 1991; Vautard et al.,
1992). The method involves embedding a time series of observations, {xt}, in a vector
0111 space with dimension D. This dimension D is required to be larger than 2d1, where d
is the effective dimension of the underlying system. Generally, it is appropriate to choose
D < n/5. In decomposing a monthly index of the Southern Oscillation for 1942–90, for
example, Ghil and Yiou (1996) use a vector space with D 60, corresponding to a five-
year window. To embed the data, lagged copies of the series {xt}, with t 1 . . . n, are
used to construct a sequence {xt*} of D-dimensional vectors:
{xt*} {xt , xt 1 . . . xtD1}
with t 1, . . . nD1.
Elementary oscillation patterns are extracted from the phase space of the lagged
0 sequence by decomposing the augmented vectors. The SSA is performed by calculating
the D D covariance matrix and obtaining its eigen elements. These comprise the orthog-
onal eigen vectors (empirical orthogonal functions, or EOFs) representing patterns in the
time series, and the corresponding eigen values give the associated variance. By projecting
the original time series onto each EOF we can obtain the principal components (PCs).
Figure 2.21 illustrates PCs 1–4 of the Southern Oscillation index time series. If these four
PCs are combined, a smoothed record of the index is obtained, confirming that the main
oscillation features have been captured. The phases of pairs 1 and 2, and 3 and 4 of these
PCs and the original EOFs are evidently in quadrature. They are interpreted as the non-
linear analogue of the sine–cosine pairs in standard Fourier decomposition shown in Figure
0 2.22 although red noise can cause spurious pairs. Statistical tests of SSA signals are
described by Yiou et al. (1996).
The most recent technique developed to study the occurrence of frequency variations
11 over time is wavelet analysis (Lau and Weng, 1995). The unique feature of the wavelet
Figure 2.21 The first four principal components of a Southern Oscillation index time series for
1942–90, derived from singular spectrum analysis. (From Ghil and Yiou, 1996)
function is its ability to detect intermittent or transient frequency components (i.e. local
1 waves), whereas Fourier analysis gives oscillations that continue indefinitely. Figure 2.22
shows the different time–frequency windows used in Fourier and wavelet transforms and
the corresponding time series plotted in the time and frequency domains. In contrast to
Fourier analysis, which gives a projection of the signal on to harmonic modes, and
to SSA, which gives a projection on to EOFs, wavelet analysis projects on to either an
orthogonal or a non-orthogonal family of functions (Yiou et al., 1996; Torrence and
Compo, 1998). The wavelet transform uses functions that are flexible in their time
and frequency resolution. The wavelet domain widens to identify low-frequency events
and narrows to a high-pass filter for edge detection. There are a variety of algorithms in
the literature. The Haar wavelet is a boxcar function which is useful to examine step
changes in a series. Among the many others in use are the Morlet, Paul, Derivative of a
Gaussian (DoG), and Mexican Hat wavelets; Figure 2.23 illustrates these last four wavelets
in the time and frequency domains. It may be necessary to use several algorithms in order
to determine whether the results are robust.
Briefly, the wavelet transform decomposes a signal X(t) into elementary functions
b,a (t) that are derived from a “mother wavelet” (t) through translation or position change,
identified by b, and scaling or dilation, a, of the wavelet (Lau and Weng, 1995):
b,a(t) =
a
1
冢冣
0.5
tb
a
The factor a0.5 normalizes the energy of the analyzing wavelets with that of the mother
wavelet. The wavelet transform W(b, a) of the sequence X(t) can be defined as:
0
0
0
0
11
11
0111
111
Figure 2.22 Above The time (t) and frequency () windows used in (a) the Fourier transform, (b) the windowed Fourier trans-
form, and (c) the wavelet transform. Below Their corresponding time series plotted in time space (left) and frequency
space (right). (From Lau and Weng, 1995)
Figure 2.23 Plots of the (a) Morlet, (b) Paul, (c) Mexican Hat, and (d) Derivative of Gaussian
(DoG) wavelets in the time domain (left) and frequency domain (right). In (a) and
(b), left column, the solid line shows the real part and the dotted line the imaginary
part of the function. The wavelet scale s is taken to be ten times the time spacing,
dt. m is the derivative in the wavelet basis function, m 2 in (c) and 6 in (d). (From
Torrence and Compo, 1998)
W (b, a) =
1
冕冢冣
a0.5
*
tb
a
X (t) dt
where * is the complex conjugate1 of defined on the real half-plane (b, a). For the
Morlet wavelet, which comprises a plane wave modulated by a Gaussian (see Figure
2.23a), the Fourier transform is known analytically:
冤
( f ) = (2)0.5 exp
( f f0)
2 冥, f >0
where f0 is the non-dimensional frequency.

11
111
Figure 2.24 Wavelet analysis of (a) sea surface temperatures, 1875–1992, in the Niño 3 region of
0111 the eastern tropical Pacific Ocean and (b) the corresponding all-India rainfall index.
The contours show the normalized variances of 2, 5 and 10 explained at a particular
frequency (1/period) with respect to time. The Paul wavelet gives better time than
frequency localization. The thick black contours denote 95 percent significance; the
hatching in the lower corners identifies the “cone of influence” areas where the vari-
ance in those periods cannot be reliably determined from the data. The lower plot
shows the observations (fine line) and the percent of total variance explained by all
frequencies over time (solid curve) for the all-India rainfall index. (From Torrence
and Compo, 1998; Webster et al., 1998)
The results of a wavelet decomposition are presented on a frequency–time plot. Figure

2.24 is an example of a wavelet analysis of sea surface temperature in the eastern trop-
ical Pacific and Indian rainfall (Webster et al., 1998). There is little inter-annual variability
11 between 1920 and 1950 whereas during 1880–1920 and since the 1950s there are strong
signals in the two-six-year and ten to twenty-five-year ranges. In general, wavelet analysis
is applicable for identifying the modulation of a signal amplitude or its frequency, as well
as for recognizing abrupt changes in a time series or its frequency characteristics.
11
2.6 Empirical orthogonal function analysis, clustering, and
classification
2.6.1 Empirical orthogonal functions and clustering
The purpose of determining empirical orthogonal functions (EOFs) or principal compo-
nent analysis (PCA) is to obtain structures, via a new set of variables, that most closely
and efficiently represent an original data set of N observations on M interconnected vari-
ables; in other words, to simplify the data set for purposes of interpretation and
understanding. The new variables are orthogonal to one another and therefore uncorre-
lated. A summary of the basic concepts is given here. Details and examples may be found
in Gould (1967), Johnston (1978), Joliffe (1986, 1990, 1993), and Jackson (1991).
A matrix of covariances (or correlations) is set up among the M original variables
(M M ). The first new variable (first principal component, or EOF 1) which contains
1
the largest proportion of the total variance in the data set is extracted (by computer
routine). It represents a linear function of the M variables of the form:
zi = a11x1 a12x2 … a1MxM
where a11, a12 … a1M are constants. This set of coefficients (also called loadings or weights)
a11, a12 … a1M in the first principal component comprises the first eigen vector. EOF 1
represents an “average” variable for the (M M ) matrix. The second PC (EOF 2) is
orthogonal to the first and extracts the largest proportion of the remaining variance, and
so on.
The ith principal component (or EOF i) can be expressed as:
zi = ai1x1 ai2x2 … aiMxM
where i 1, 2 … M … .
The correlations between EOF 1 and the M variables are termed the component load-
1 ings, and the square of the loadings is the proportion of associated variance. The sum of
these squared loadings is the “explained” variance by EOF1, known as the eigen value
1. Figure 2.25 illustrates the elliptical hyperspace of a simple bivariate distribution and
the lines representing the paired correlations. Note that the cosine of the angle, , between
these lines is equal to the correlation coefficient between the variables (e.g. 50°;
cos r12 0.64). EOF 1 is the principal axis of the ellipse and bisects the angle
between the two vectors that bracket the correlation. EOF 2 correspondingly bisects the
complements of the angle and is represented by the shortest axis of the ellipse. The
length of the two axes corresponds to the eigen values; 0A 1 , and 0B 2. It should
be noted that EOF analysis requires the ratio of the degrees of freedom to the number of
samples to be not 1/2 in order to obtain a non-biased determination of eigenvalue
estimates in studies with small samples (von Storch and Hannoschock, 1985).
For most real problems there is a large (M M) matrix and a multi-dimensional hyper-
space. The number of EOFs that are usefully interpreted is approximately defined by those
EOFs for which > 1.0 (the variance of the original M variables).
If the M variables are used in their original units of measurement, the analysis is based
on a covariance matrix between the variables. However, this allows the variables with a
large variance to dominate the first few PCs. This arbitrary influence can be avoided by
standardizing the terms x1, x2 , … xM by dividing each by its standard deviation(s). In this
case the matrix between the variables uses correlations between x1, x2, … xM . To normal-
ize the variance of zi, a constraint can be imposed on the coefficients ai1, … aiM :
M
兺a
n=1
2
in = 1.
11
111
0111
Figure 2.25
Upper The elliptical hyperspace of a bivariate
0 distribution; the lines show the paired
correlations. Lower Principal components 1
and 2 as axes of the bivariate scatter plot.
11 (From Johnston, 1978)
In the terminology of Richman (1986) this procedure gives EOFs. Alternatively, some
standard statistical packages (SPSS, BMDP) use the normalization:
M
兺a
n=1
2
in = i
where i is the variance of zi . Richman (1986) uses “principal components” to refer to zi

with this normalization. It has the effect of increasing the loadings in the first few PCs
relative to the subsequent ones. A convenient feature of this normalization is that, where
a correlation matrix is used, aij is the correlation between xj and the ith principal compo-
nent. Note that the basic interpretation of a principal component is unaffected by the
normalization chosen.
Multiplying the eigen vector (ai1 … aiM ) of principal component i by the original (stan-
1 dardized) data (x1/s1 … xM /sM ), and summing over the M variables, gives the projection
of the data on to the component axes; this is called the component score. In the case
where a covariance matrix is used, the component score is the product of the loading and
the original data, summed over the M variables.
To obtain classes from a PCA, the component scores can be grouped via cluster analysis.
A common procedure involves hierarchical clustering. One approach is average linkage
clustering, where a set of groups is obtained by agglomeration; similarity is determined
by the mean distance between all objects in two clusters, weighted by the number of
members (Hawkins et al., 1982; Johnson and Wichern, 1982, p. 543; Kalkstein et al.,
1987). Average linkage tends to generate many clusters with few members that commonly
represent outliers. To overcome this problem, the initial clusters may be rearranged
by iteration of a non-hierarchical procedure such as convergent k-means clustering
(MacQueen, 1967) applied to the mean component loadings of each average linkage
cluster. This two-stage procedure is illustrated by Davis and Kalkstein (1990). Wilson
et al. (1992) examine a daily weather classification in the US Pacific Northwest using
k-means cluster analysis, fuzzy cluster analysis, and principal components.
1 A comprehensive evaluation via Monte Carlo simulations of the reliability of clusters
derived by a suite of procedures (Gong and Richman, 1995) provides a valuable guide
to the options available. The various options include: hierarchical/non-hierarchical
methods; various measures of distance between samples within a cluster, or between clus-
ters; and the use of “hard” (discrete) clusters versus “fuzzy” (overlapping) groups. In
general, they conclude that non-hierarchical methods are better than hierarchical ones,
with rotated PCs giving greatest accuracy. Nuclear agglomeration is the best of the hard
cluster methods; simple linkage in contrast leads to “chaining.” The Euclidean measure
of distance is shown to be slightly better than inverse correlation.
For meteorological patterns, in particular, EOF 1 is an average variable and the subse-
quent components are orthogonal to it. Buell (1975, 1979) showed that the leading EOFs
of meteorological fields over a region display a characteristic set of patterns that is descrip-
tively referred to as “one fried egg,” “two fried eggs,” and so on. These were attributed
to the commonly rectangular boundaries of a limited domain, but Legates (1991) argues
that they are fundamentally related to the underlying spatial correlation structure. Further
discussion is provided by Richman (1993). For this reason these EOFs may not identify
realistic groups of variables. A procedure to achieve this involves rotation of the axes
around the origin, while retaining orthogonality, to obtain an ideal “simple structure.” The
problem of deciding how many modes to rotate is discussed for geopotential fields by
O’Lenic and Livezey (1988). An alternative procedure is an oblique rotation which iden-
tifies factors representing groups of interrelated variables; the axes are no longer orthogonal
and consequently the interpretation requires additional care (Richman, 1986). An appli-
cation of oblique rotation for precipitation regions is provided by Comrie and Glenn
(1998).
11 The variance accounted for by PCs is additive. However, it is necessary to determine
how many components are worth retaining by using one of several tests (Cattell, 1966;
North et al., 1982; Overland and Preisendorfer, 1982). Cattell developed the simple scree
plot in which the eigen value (explained variance) is plotted against the PC number
(1, 2 … n). This is inspected visually to select the point where the variance declines to
a tail. Davis and Kalkstein (1990) suggest some further modifications. In the case of rota-
tion, the Guttman (1954) criterion suggests rotating PCs which contribute more total
variance than the normalized time series, i.e. one unit of total variance. Joliffe (1986,
p. 95) suggests that, when using the correlation matrix, PCs with variance 0.7 are worth
0 retaining. A pragmatic approach is to retain only PCs that can be physically interpreted.
The data matrix discussed above comprised M variables (columns) and the N point
observations (rows). This common analysis in meteorological studies is designated as R
mode; its transpose in which the N observations form the columns and the M variables
111 are the rows is a Q-mode analysis. Figure 2.26 illustrates the possible modes of decom-
position of data matrices (Richman, 1986). The R mode examines correlations between
variables and the Q mode focuses on correlations between spatial observations. In cases
where time is one dimension, the modes are termed O mode, where time points (t) are
the columns, and the M variables are the rows; this examines time-dependent correla-
tions. Its transpose, the P mode, considers correlations between variables over time.
0 Finally, for a single variable, there is the S mode (N point observations form the columns
and t time intervals the rows), which considers spatial correlations of the variable over
time and its transpose the T mode examines correlations between time periods. The S and
T modes are used to study trends of a single variable (e.g. Perry, 1970). S-mode analysis,
where observations are made over time at a network of stations (as variables), provides
time series of component scores. Dyer (1975) delimits climatic regions by identify-
ing geographical areas with similar component loadings. The component scores of a
corresponding T-mode analysis, where time is the variable, yields spatial variations in
temporal eigen values. Green et al. (1993), for example, obtain seasonal divisions from
monthly temperature, precipitation, and wind data based on similarities among the
0111 component loadings.
An analogous tool is singular value decomposition (SVD); this can be used to iden-
tify a hierarchy of paired spatial patterns in two fields (Wallace et al., 1992). By performing
an SVD on a temporal covariance matrix between the ith grid point of one field and the
corresponding jth grid point of the second field, the principal mode accounts for the
maximum possible fraction of the squared covariance between the fields. The singular
vectors for each field are mutually orthogonal. As with EOF analysis, the fields at a spec-
ified observation time can be projected on to the singular vectors to derive expansion
coefficients. This approach was first used in meteorology by Prohaska (1976). Shapiro
and Goldenberg (1998) examine the dominant covarying modes of sea surface tempera-
0 ture gradients and tropospheric vertical wind shear in relation to Atlantic hurricanes, using
the SVD approach. A comparison between methods of identifying coupled patterns in
meteorological fields is provided by Bretherton et al. (1992). Recently, EOF techniques
have been greatly elaborated to treat different frequencies (SSA, discussed in section
2.5.3), as well as patterns that deform or migrate with time. These methods are beyond
the scope of this discussion, but a useful comparison of eight such techniques is given
by Kim and Wu (1999).
2.6.2 Classificatory methods

0 Classification serves the purpose of naming sets of things, and also of grouping things
by resemblance, by relationship, or both. The fundamental objective of all systems of
classification is to obtain the least variability within groups and the maximum differences
11 between groups. Groups (or classes) can be derived by two basic methods – by division
Figure 2.26 The six modes of decomposition (O–T) in principal component analysis (rows) and
the corresponding matrix configurations for the data matrix (variable in the columns
and individual index in the rows), correlation or covariance matrix, component loading
matrix and component score matrix. (From Richman, 1986)
of a data set into subsets, or by agglomeration of similar individuals into groups. Division
generally follows a hierarchical approach where subdivisions are based on the occurrence
or non-occurrence of a single specific attribute. Agglomerative, or clustering, procedures,
in contrast, take all measured attributes into consideration. The groups can be arranged
hierarchically or in a reticulate system. Divisions are quicker to calculate and anomalous
11 data have little effect on the groups, whereas, in the agglomerative process, anomalies
may cause biased groups at the lower levels and these affect subsequent higher order
groups.
Classification involves three separate considerations; the delimitation of groups; the
assignment of new cases to established groups with minimum likelihood of error; and
tests of significance for established groups.
The use of correlation to assess the homogeneity of a region for a single climatic para-
meter has been discussed above (section 2.4). The application of this approach to
teleconnections and to typing synoptic pressure fields is examined in Chapters 5 and 7
0 respectively. Statistical issues in such analyses concern the effects of spatial autocorrela-
tion and of non-linearity on the calculated product–moment correlation coefficients. In
cases where the correlation between two sets of variables (or two eigen vector patterns
determined on two data sets) need to be evaluated, it can be performed through canon-
111 ical correlation analysis, or CCA (Glahn, 1968; Johnston, 1980; von Storch, 1995).
Examples include its use by Crane (1983) in comparing Arctic sea ice conditions and
atmospheric circulation and by Nicholls (1987) in a study of the temporal structure of
teleconnections. Werner and von Storch (1993) analyze anomalies of winter mean sea-
level pressure over the eastern North Atlantic and Europe and air temperatures in central
Europe using a CCA of the data projected on to EOFs.
0 The division of data consisting of a combination of variables into classes can be readily
performed by discriminant analysis. The procedures, first developed by R.A. Fisher, have
been adapted for meteorology by Panofsky and Brier (1958, p. 118) and Miller (1962).
The method is based on multiple regression and so is applicable to normally distributed
and moderately skewed data if the group variances are similar. Suzuki (1969) has extended
it for heterogeneous (discrete and continuous) variables. To illustrate discrimination for
two variables (predictors), x1 and x2, we define a linear discriminant function, D, with D1
greater than a critical value for occurrence, and D2 less than a critical value for non-
occurrence of some specified condition of interest:
D a0 a1x1 a2x2
0111
– –
a1 and a2 are chosen so as to maximize (D1 D2)/sD where the overbar denotes the means
of the respective classes and sD is the standard deviation of the pooled classes. Figure
2.27 shows a simple discriminant function between two groups.
Figure 2.27 A single discriminant function separating two bivariate groups. D2 measures the
11 distance between the group means. (From Harbaugh and Merriam, 1968)
The method has been used to distinguish chinook and non-chinook winds at Calgary
(Brinkmann, 1970) and to identify weather patterns in southern California (McCutchan
and Schroeder, 1973). Discriminant analysis can also be used to assign new observations
to established classes with a minimum of error. This is a valuable feature, since EOFs
have to be recalculated when an observational record is extended.
When a classification has been established, it is important to ensure that the categories
are –
– distinct from one another. If only two groups are involved, the sample means a and
b may be compared using Student’s t statistic:
|a b|
t=
{(s2a /na ) (sb2 /nb )}1/2
with (nanb2) degrees of freedom, for sample sizes of thirty or more (Thiebeaux and
1 Zwiers, 1984). For multiple groups, an analysis of variance within and between groups
may be appropriate. The procedures are discussed in most introductory statistical texts.
Nosek (1967) applies this method to examine the temperature characteristics of weather
types identified at Brno, Czech Republic.
Appendix 2.1 Eulerian and Lagrangian methods

There are two basic analytical approaches to studies of fluid motion. One refers to the
fluid motion past fixed points (Eulerian methods) and the other to the movement of indi-
vidual fluid elements (Lagrangian methods). In the latter system total differentials are
involved, since our reference frame is the trajectory of an individual particle. For example,
the speed components are:
dx dy dz
u= , v= , w=
dt dt dt
1
The various approaches to trajectory analysis and their limitations are presented in
Section 2.4.7
In the Eulerian system we consider the instantaneous change at individual points as the
fluid passes. The acceleration components are now partial derivatives (i.e. functions of
more than one variable):
∂u ∂u ∂u ∂u Du
ax = u v w ≡
∂x ∂y ∂z ∂t Dt
∂v ∂v ∂v ∂u Dv
ay = u v w ≡
∂w ∂w ∂w ∂w Dw
az = u v w ≡
where
D ∂ ∂ ∂ ∂
= u v w
Dt ∂y ∂y ∂z ∂t
(Stokes’s operator).
11
111
Figure 2.28 Grid framework for finite difference evaluation of partial derivatives
0
Finite differences of partial derivatives
∂S (Sx Sx)
≈
∂x L
where S is some field variable (see Figure 2.28).
∂2S ∂ ∂S
=
∂x2 ∂x ∂x 冢冣
(S 2Sx Sx)
= 2x
L2
0111
∂2S ∂2S (Sx Sx Sy Sy 4S0)
2S = ≈
∂x2 ∂y2 L2
Scalars and vectors

A scalar quantity has only magnitude; for example, temperature or pressure. A vector
quantity has magnitude and direction; for example, wind velocity, the gradient of a temper-
ature field. A wind vector (V) can be resolved into components in a cartesian coordinate
system (x, y). The scalars u and v are the lengths of the projections of V on to the x and
0 y axes, respectively. In three dimensions a similar projection for the w component may
be made on the vertical z axis:
V iu jv kw
where u and v horizontal wind components (in x and y directions, respectively), w
vertical wind component (in z direction), i, j and k unit vectors denoting the directions
of u, v, w respectively.
The magnitude of V (u2 v2 w2)1/2. The direction of the vector is measured by
cosines of the angles between the vector V and the three axes x, y, and z. The sum of
0 the two vectors is illustrated in Figure 2.29 (b) and (c) for positive and negative cases.
From the definition of V above it follows that:
V1 ± V2 = i (u1 ± u2) j(v1 ± v2) k(w1 ± w2)
11
Figure 2.29 (a) Vector and components in cartesian coordinates. (b) Vector addition. (c) Vector
subtraction. (d) Vector cross-product
1
The two principal multiplication operations are the dot product and the cross-product of
the two vectors. The definitions are:
V1 ·V2 V1V2 cos
where is the angle 180° between the vectors V1 and V2 . Also V1 · V2 V2 ·V1. The
dot product is a scalar. For unit vectors:
i ·i j· j k· k 1
and
i ·j j·k k· i 0
The cross-product gives a vector of magnitude:
V1 V2 V1V2 sin
and direction normal to both vectors, positive in the sense of a right-hand screw when
turned from direction V1 to V2 through an angle 180° (Figure 2.29d). For unit vectors:
ijk
jki
k i j
11 j i k
iijjkk0
∂S ∂S ∂S
grad S = S = i j k
∂x ∂y ∂z
Vector operators
0
1 Gradient: where i, j, and k are unit vectors in the directions of x, y, and z respec-
tively. S any scalar. The gradient is defined as positive towards lower values
of S.
111
2 Divergence:
∂Sx ∂Sy ∂Sz
div S = ·S =
∂x ∂y ∂z
0 Divergence is the net outflow of fluid from unit volume in unit time.
3 Laplacian operator:
∂2 ∂2 ∂2
2 = · = 2 2
∂x ∂y ∂z2
4 Curl:
0111
curl S = S = i 冢∂S∂y ∂S∂z 冣 j 冢∂S∂z ∂S∂x 冣 k 冢∂S∂x ∂S∂y 冣
z y x z y x
Note that the vertical relative vorticity, , is:
= 冢∂v∂x ∂u∂y冣
corresponding to the last term of the complete three-dimensional curl.
0 Note
1 The complex conjugate of any complex number xiy is xiy.
References
Adler, R.F., Negri, A.J., Keehn, P.R., and Hakkarinen, I.M. 1993. Estimation of monthly rainfall
over Japan and surrounding waters from a combination of low-orbit microwave and geosyn-
chronous IR data. J. Appl. Met., 32: 335–56.
Agnew, T.A., Le, H., and Hirose, T. 1997. Estimation of large-scale sea ice motion from SSM/I
85.5 GHz imagery. Ann. Glaciol., 25: 305–11.
0 Alishouse, J.C. 1983. Total precipitable water and rainfall distributions from SEASAT Scanning
Multichannel Microwave Radiometer. J. Geophys. Res., 88: 1919–35.
Alishouse, J.C., Ferraro, R.R., and Fiore, J.V. 1990. Inference of oceanic rainfall properties from
11 the Nimbus 7 SMMR. J. Appl. Met., 29: 551–60.
Allison, L.J., Steranka, J., Cherrix, G.T., and Hilsenrath, E. 1972. Meteorological applications of
the Nimbus 4 Temperature–Humidity Infrared Radiometer, 6.7 micron channel data. Bull. Amer.
Met. Soc., 53: 526–35.
Alter, D. 1933. Correlation periodogram investigations of English rainfall. Mon. Wea. Rev., 61:
345–52.
Anderson, D., Hollingsworth, A., Uppala, S., and Woiceshyn, P. 1991. A study of the use of scat-
terometer data in the European Center for Medium Range Weather Forecasts operational
analysis–forecast model. 2. Data impact. J. Geophys. Res., 96: 2635–47.
Anderson, M.R. 1987. The onset of spring melt in first-year ice regions of the Arctic as determined
from Scanning Multichannel Microwave Radiometer data for 1979 and 1980. J. Geophys. Res.,
92: 13153–63.
Anderson, R.K., Ashman, J.P., Bittner, F., Farr, G.R., Ferguson, E.W., Oliver, V.J., and Smith,
A.H. 1969. Applications of Meteorological Satellite Data in Analysis and Forecasting, ESSA
Tech. Report NESC 51.
1 Ardanuy, P.E., Stowe, L.L., Gruber, A., and Weiss, M. 1991. Shortwave, longwave, and net cloud-
radiative forcing as determined from Nimbus 7 observations. J. Geophys. Res., 96: 18537–49.
Arkin, P.A. and Ardanuy, P.E. 1989. Estimating climatic-scale precipitation from space: a review.
J. Climate, 2: 1229–38.
Arkin, P.A. and Meisner, B.N. 1987. The relationship between large-scale convective rainfall and
cold cloud over the western hemisphere during 1982–84. Mon. Wea. Rev., 115: 51–74.
Armstrong, R.L. and Brodzik, M.J. 1999. A twenty-year record of global snow cover fluctuations
derived from passive microwave remote sensing data. Preprints, Fifth Conference on Polar
Meteorology, Amer. Met. Soc., Boston MA, 1. 8.1–6.
Armstrong, R.L., Chang, A., Rango, A., and Josberger, E. 1993. Snow depths and grain-size rela-
tionships with relevance for passive microwave studies. Ann. Glaciol., 17: 171–6.
Arpe, K. 1991. The hydrological cycle in the ECMWF short range forecasts. Dynam. Atmos. Oceans,
16: 33–59.
Artz, R., Pielke, R.A., and Calloway, J. 1985. Comparison of the ARL/ATAD constant level
and the NCAR isentropic trajectory analyses for selected case studies. Atmos. Environ., 19:
47–63.
Atlas, R., Hoffman, R.N., and Bloom, S.C. 1993. Surface wind velocity over the oceans. In: R.J.
Gurney, J.L. Foster and C.L. Parkinson, eds, Atlas of Satellite Observations related to Global
1 Change, Cambridge University Press, Cambridge, pp. 129–39.
Augustine, J.A. and Howard, K.W. 1991. Mesoscale convective complexes over the United States
during 1986 and 1987. Mon. Wea. Rev., 119 (7): 1575–89.
Ba, M.B. and Nicholson, S.E. 1998. Analysis of convective activity and its relationship to the rain-
fall over the Rift Valley Lakes of East Africa during 1983–90 using the Meteosat infrared channel.
J. Appl. Met., 37 (10): 1250–64.
Bailey, M.J., O’Neill, A., and Pope, V.D. 1993. Stratospheric analyses produced by the United
Kingdom Meteorological Office. J. Appl. Met., 32: 1472–83.
Barber, D.G., Johnson, D.D., and LeDrew, E.F. 1991. Measuring climatic state variables from SAR
images of sea ice: the SIMS SAR validation site in Lancaster Sound. Arctic, 44 (Suppl. 1):
108–21.
Barnett, T.P. and Jones, P.D. 1992. Intercomparison of two different southern hemisphere sea level
pressure data sets. J. Climate, 5: 93–9.
Barrett, E.C. 1970. Rethinking climatology: an introduction to the uses of weather satellite photo-
graphic data in climatological studies. Progr. Geog., 2: 155–205.
Barrett, E.C. 1974. Climatology from Satellites. Methuen, London and New York, 418 pp.
Barrett, E.C. 1987. Satellite climatology. In: J.E. Oliver and R.W. Fairbridge, eds, The Encyclopedia
of Climatology, Van Nostrand Reinhold, New York, pp. 728–36.
Barrett, E.C. and Grant, C.K. 1979. Relations between frequency distributions of cloud over the
United Kingdom based on conventional observations and imagery from Landsat 2. Weather, 34:
416–23.
Barrett, E.C., Kidd, C., Bailey, J.O., and Collier, C.G. 1990. The Great Storm of 15/16 October
1987: passive microwave evaluations of associated rainfall and marine wind speeds. Met. Mag.,
119: 177–87.
Barry, R.G. 1971. An introduction to numerical and mechanical techniques. In: F.J. Monkhouse
and H.R. Wilkinson, eds, Maps and Diagrams, Methuen, London, pp. 472–517.
11 Barry, R.G. 1991. Cryospheric products from the DMSP SSM/I: status and research applications.
Global and Planetary Change, 4: 231–4.
Barry, R.G. 1997. Satellite-derived data produced for the polar regions. EOS, 78 (5): 52. The Polar
Pathfinders: Data products and science plans. II. The Polar Pathfinder Group. EOS Electronic
Supplement, 96149e.
Barry, R.G., Elliot, D.L., and Crane, R.G. 1981. The palaeoclimatic interpretation of exotic
pollen peaks in Holocene records from the eastern Canadian Arctic. Rev. Paleobot. Palynol., 33:
153–67.
London.
0 Basist, A., Garrett, D., Ferraro, R., Grody, N., and Mitchell, K. 1996. A comparison between snow
cover products derived from visible and microwave satellite observations. J. Appl. Met., 35 (2):
163–77.
Bauer, P. and Schluessel, P. 1993. Rainfall, total water, ice water, and water vapor over sea from
111 polarized microwave simulations and Special Sensor Microwave Imager data. J. Geophys. Res.,
98: 20737–59.
Bell, T.L. and Reid, N. 1993. Detecting the diurnal cycle of rainfall using satellite observations.
J. Appl. Met., 32: 311–22.
Bellamy, J.C. 1949. Objective calculations of divergence, vertical velocity and vorticity. Bull. Amer.
Met. Soc., 30: 45–9.
Bengtsson, L. and Shukla, J. 1988. Integration of space and in situ observations to study global
0 climate change. Bull. Amer. Met. Soc., 69 (10): 1130–43.
Bertoni, E.A. and Lund, I.A. 1963. Space correlations of the height of constant pressure surfaces.
J. Appl. Met., 2: 539–45.
Bjørgo, E., Johanessen, O.M., and Miles, M.W. 1997. Analysis of merged SMMR/SSM/I time series
of Arctic and Antarctic sea ice parameters, 1978–85. Geophys. Res. Lett., 24: 4132–6.
Blackman, R.B. and Tukey, J.W. 1958. The Measurement of Power Spectra from the Point of View
of Communications Engineering. Dover Publications, New York, 190 pp.
Boutin, J. and Etcheto, J. 1990. Seasat scatterometer versus Scanning Multichannel Microwave
Radiometer wind speeds: a comparison on a global scale. J. Geophys. Res., 95 (C12): 22275–88.
Box, G.E.P. and Jenkins, G.M. 1976. Time Series Analysis, Forecasting and Control, Revised
edition. Holden-Day, San Francisco, 575 pp.
0111 Breckenridge, C.J., Radok, U., Stearns, C.R., and Bromwich, D.H. 1993. Katabatic winds along the
Transantarctic mountains. In: D.H. Bromwich and C.R. Stearns, eds, Antarctic Meteorology and
Climatology: Studies based on Automatic Weather Stations, Antarctic Res. Ser. 61, Amer.
Geophys. Union, Washington, DC, pp. 69–92.
Bretherton, C.S., Smith, C., and Wallace, J.M. 1992. An intercomparison of methods for finding
coupled patterns in climate data. J. Climate, 5: 541–60.
Brinkmann, W.A.R. 1970. The chinook at Calgary (Canada). Arch. Met. Geophys. Biokl., B18:
269–98.
Brogniez, G., Burriez, J.C., Giraud, V., Parol, F., and Vanbauce, C. 1995. Determination of effec-
tive emittance and a radiatively equivalent microphysical model of cirrus from ground-based and
satellite observations during the International Cirrus Experiment: the 18 October case study. Mon.
0 Wea. Rev., 123 (4): 1025–36.
Brooks, C.E.P. and Carruthers, N. 1953. Handbook of Statistical Methods in Meteorology, M.O.
538, Air Ministry. HMSO, London, 412 pp.
Brown, R.A. 1986. On satellite scatterometer capabilities in air–sea interaction. J. Geophys. Res.,
91 (C12): 2221–32.
Brown, R.A. and Zeng, L. 1994. Estimating central pressures of oceanic midlatitude cyclones.
J. Appl. Met., 33: 1088–95.
Bryson, R.A. 1966. Air masses, streamlines and the boreal forest. Geogr. Bull. (Ottawa), 8: 228–69.
Bryson, R.A. and Kuhn, P.M. 1961. Stress-differential induced divergence with application to littoral
precipitation. Erdkunde, 15: 287–94.
Buell, C.E. 1975. The topography of the empirical orthogonal functions. Preprints, Fourth
0 Conference on Probability and Statistics in Atmospheric Sciences, Amer. Met. Soc., Boston MA,
pp. 188–93.
Buell, C.E. 1979. On the physical interpretation of empirical orthogonal functions. Preprints, Sixth
11 Conference on Probability and Statistics, Amer. Met. Soc., Boston MA, pp. 112–17.
Bugaev, V.A. 1973. Dynamic climatology in the light of satellite information. Bull. Amer. Met.
Soc., 54 (5): 394–418.
Burtt, T.G. and Junker, N.W. 1976. A typical rapidly developing extratropical cyclone viewed in
SMS-II imagery. Mon. Wea. Rev., 104: 489–90.
Bushby, F.H. 1952. The evaluation of vertical velocity and thickness tendency from Sutcliffe’s
theory. Quart. J. Roy. Met. Soc., 78: 354–61.
Caffey, J.E. 1965. Interstation correlations in annual precipitation and in annual effective precip-
itation. Hydrol. Pap. 6. Colorado State University, Fort Collins, 47 pp.
Cahalan, R.F., Short, D.A., and North, G.R. 1982. Cloud fluctuation statistics. Mon. Wea. Rev.,
110: 26–43.
Campbell, W.J., Ramseier, R.O., Zwally, H.J., and Gloersen, P. 1980. Arctic sea ice variations from
time-lapse passive microwave imagery. Bound.-Layer Met., 18: 99–106.
Carleton, A.M. 1984. Synoptic sea ice–atmosphere interactions in the Chukchi and Beaufort Seas
from Nimbus-5 ESMR data. J. Geophys. Res., 89: 7245–58.
1 Carleton, A.M. 1985. Satellite climatological aspects of the “polar low” and “instant occlusion.”
Tellus, 37A: 433–50.
Carleton, A.M. 1987. Satellite-derived attributes of cloud vortex systems and their application to
climate studies. Rem. Sensing Environ., 22: 271–96.
Carleton, A.M. 1991. Satellite Remote Sensing in Climatology. Belhaven Press, London, 291 pp.
Carleton, A.M. 1995. On the interpretation and classification of mesoscale cyclones from satellite
infrared imagery. Intl. J. Rem. Sensing, 16 (13): 2457–85.
Carleton, A.M. 1996. Satellite climatological aspects of cold air mesocyclones in the Artic and
Antarctic. Global Atmos. Ocean System, 4 (5): 1–42.
Carleton, A.M., McMurdie, L.A., Katsaros, K.B., Zhao, H., Mognard, N.M., and Claud, C. 1995.
Satellite-derived features and associated atmospheric environments of Southern Ocean mesocy-
clone events. Global Atmos. Ocean Sys., 3: 209–48.
Carsey, F.D. (ed.). 1992. Microwave Remote Sensing of Sea Ice. Geophysical Monogr. 68, American
Geophysical Union, Washington, DC, 462 pp.
Cattell, R.B. 1966. The scree test for the number of factors. J. Multivar. Behav. Res., 1: 245–76.
Cavalieri, D.J., St Germain, K., and Swift, C.T. 1995. Reduction of weather effects in the calcu-
lation of sea ice concentration with the DMSP SSM/I. J. Glaciol.,41 (139): 455–64.
Cavalieri, D.J., Gloersen, P., Parkinson, C.L., Comiso, J.C., and Zwally, H.J. 1997. Observed hemi-
1 spheric asymmetry in global sea ice changes. Science, 278: 1104–6.
Chang, A., Foster, J.L., and Hall, D.K. 1987. Nimbus-7 SMMR-derived global snow cover para-
meters. Ann. Glaciol., 9: 39–44.
Chang, S.W., Alliss, R.J., Raman, S., and Shi, J.-J. 1993. SSM/I observations of ERICA IOP4
marine cyclone: a comparison with in situ observations and model simulation. Mon. Wea. Rev.,
121: 2452–64.
Chatfield, C. 1989. The Analysis of Time Series: An Introduction, 4th edn. Chapman and Hall, New
York, 241 pp.
Chesters, D., Robinson, W.D., and Uccellini, L.W. 1987. Optimized retrievals of precipitable water
from the VAS “split window.” J. Clim. Appl. Met., 26: 1059–66.
Chesters, D., Uccellini, L.W., and Robinson, W.D. 1983. Low-level water vapor fields from the
VISSR Atmospheric Sounder (VAS) “split window” channels. J. Clim. Appl. Met., 22: 725–43.
Chiu, L.S., Chang, A.T.C., and Janowiak, J. 1993. Comparison of monthly rain rates determined
from GPI and SSM/I using probability distribution functions. J. Appl. Met., 32: 323–34.
Claud, C., Katsaros, K.B., Mognard, N.M., and Scott, N.A. 1995. Synergetic satellite study of a
rapidly deepening cyclone over the Norwegian Sea, 13–16 February 1989. Global Atmos. Ocean
Sys., 3: 1–34.
Claud, C., Katsaros, K.B., Petty, G.W., Chedin, A., and Scott, N.A. 1992. A cold air outbreak over
the Norwegian Sea observed with the TIROS-N Operational Vertical Sounder (TOVS) and the
Special Sensor Microwave/Imager (SSM/I). Tellus, 44A: 100–18.
Claud, C., Mognard, N.M., Katsaros, K.B., Chedin, A., and Scott, N.A. 1993. Satellite observa-
tions of a polar low over the Norwegian Sea by Special Sensor Microwave Imager, Geosat, and
TIROS-N Operational Vertical Sounder. J. Geophys. Res., 98 (C8): 14487–506.
Claud, C., Scott, N., Chedin, A., and Cascard, J.-C. 1991. Assessment of the accuracy of atmos-
pheric temperature profiles retrieved from TOVS observations by the 3I method in the European
Arctic: application for mesoscale weather analysis. J. Geophys. Res., 96 (D12): 2875–87.
11 Cohen, J. 1960. A coefficient of agreement for nominal scales. Educ. Psychol. Meas., 20: 37–46.
Coll, C. and Caselles, V. 1997. A split-window algorithm for land surface temperature from advanced
very high resolution radiometer data: validation and algorithm comparison. J. Geophys. Res., 102:
16697–713.
Collins, G.O. and Kuhn, P.M. 1954. Computation of precipitation resulting from vertical velocities
deduced from vorticity changes. Mon. Wea. Rev., 82: 173–82.
Comiso, J.C. 1990. Arctic multiyear ice classification and summer ice cover using passive microwave
satellite data. J. Geophys. Res., 95 (C8): 13411–22.
Comrie, A.C. and Glenn, E.C. 1998. Principal components-based regionalization of precipitation
processes across the southwest United States and northern Mexico, with an application to monsoon
0 precipitation variability. Clim. Res., 10 (3): 201–15.
Cornish, E.A., Hill, G.W., and Evans, M.J. 1961. Inter-station correlations of rainfall in southern
Australia. Tech. Pap. 10. Division of Mathematical Statistics, CSIRO, Melbourne, 16 pp.
Cox, D.R. 1962. Renewal Theory. Methuen, London, 142 pp.
111 Craddock, J.M. 1957. An analysis of the slower temperature variations at Kew Observatory by
means of mutually exclusive band pass filters. J. Roy. Stat. Soc., A, 120: 387–97.
Crane, R.G. 1983. Atmosphere–sea ice interactions in the Beaufort/Chukchi Sea and in the European
sector of the Arctic. J. Geophys. Res., 88 (C7): 4505–23.
Crane, R.G. and Anderson, M.R. 1994. Springtime microwave emissivity changes in the southern
Kara Sea. J. Geophys. Res., 99: 14303–9.
Crane, R.G., Barry, R.G., and Zwally, H.J. 1982. Analysis of atmosphere–sea ice interactions in
0 the Arctic Basin using ESMR microwave data. Intl. J. Remote Sens., 3: 259–76.
Cressman, G.P. 1959. An operational objective analysis system. Mon. Wea. Rev., 87: 364–74.
Curray, J.R. 1956. The analysis of two-dimensional orientation data. J. Geol., 64: 117–31.
Czajkowski, K.P., Goward, S.N., and Ouaidrari, H. 1998. Impact of AVHRR filter functions on
surface temperature estimation from the split window approach. Intl. J. Remote Sens., 19 (10):
2007–12.
Danard, M.B. 1964. On the influence of released latent heat on cyclone development. J. Apple.
Met., 3: 27–37.
Danielsen, E.F. 1961. Trajectories: isobaric, isentropic and actual. J. Met., 18: 479–86.
Danielsen, E.F. 1974. A review of trajectory methods. In: F.N. Frenkiel and R.E. Munn, eds,
Turbulent Diffusion in Environmental Pollution, Adv. Geophys., 18B: 73–94.
0111 Davis, R.E. and Kalkstein, L.S. 1990. Development of an automated spatial synoptic climatolog-
ical classification. Intl. J. Climatol., 10: 769–84.
Dettinger, M.D. and Cayan, D.R. 1995. Large-scale atmospheric forcing of recent trends toward
early snow melt runoff in California. J. Climate, 8 (3): 606–23.
Devore, J.L. 1995. Probability and Statistics for Engineering and the Sciences. Duxbury Press,
Belmont CA, 743 pp.
Dey, C.H. 1989. The evolution of objective analysis methodology at the National Meteorological
Center. Wea. Forecast., 4: 297–312.
Dickson, R.R. 1971. On the relationship of variance spectra of temperature to the large-scale atmos-
pheric circulation. J. Appl. Met., 10: 186–93.
Diskin, M.H. 1970. On the complete evaluaton of Thiessen weights. J. Hydrol., 11: 69–78.
0 Dorling, S.R., Davies, T.D., and Pierce, C.E. 1992. Cluster analysis: a technique for estimating the
synoptic meteorological controls on air and precipitation chemistry – method and applications.
Atmos. Environ., 26A: 2575–81.
Doty, K.G. and Perkey, D.J. 1993. Sensitivity of trajectory calculations to the temporal frequency
of wind data. Mon. Wea. Rev., 121: 387–401.
Draxler, R.R. 1996. Boundary layer isentropic and kinematic trajectories during the August 1993
North Atlantic Regional Experiment Intensive. J. Geophys. Res., 101: 29225–68.
Durst, C.S. 1951. Climate: the synthesis of weather. In: T.F. Malone, ed., Compendium of
Meteorology, Amer. Met. Soc., Boston MA, pp. 967–75.
Dvorak, V.F. 1975. Tropical cyclone intensity analysis and forecasting from satellite imagery. Mon.
Wea. Rev., 103 (5): 420–30.
0 Dyer, T.G.J. 1975. The assignment of rainfall stations into homogeneous groups: an application of
principal component analysis. Quart. J. Roy. Met. Soc., 101: 1005–13.
Easterling, D.R. 1989. Regionalization of thunderstorm rainfall in the contiguous United States. Intl.
11 J. Climatol., 9: 567–79.
Easterling, D.R. and Peterson, T.C. 1995. A new method for detecting undocumented discontinu-
ities in climatological time series. Intl. J. Climatol., 15: 369–77.
Ebert, E.E. and LeMarshall, J. 1995. An evaluation of infrared satellite rainfall estimation tech-
niques over Australia. Aust. Met. Mag., 44: 177–90.
Eck, T.F. and Holben, B.N. 1994. AVHRR split window temperature differences and total precip-
itable water over land surfaces. Intl. J. Remote Sens., 15: 567–82.
Eddy, A. 1964. The objective analysis of horizontal wind divergence fields. Quart. J. Roy. Met.
Soc., 90: 424–40.
Efron, B. and Gong, G. 1983. A leisurely look at the Bootstrap, Jack-knife, and cross-validation.
Amer. Statistician 37: 36–48.
Eigenwillig, N. and Fischer, H. 1982. Determination of midtropospheric wind vectors by tracking
pure water vapor structures in METEOSAT water vapor image sequences. Bull. Amer. Met. Soc.,
63: 44–58.
Elliott, W.P. and Gaffen, D.J. 1991. On the utility of radiosonde humidity archives for climate
1 studies. Bull. Amer. Met. Soc., 72 (10): 1507–20
Emery, W.J., Fowler, C.W., and Maslanik, J.A. 1997. Satellite-derived maps of Arctic and Antarctic
sea ice motion, 1988 to 1994. Geophys. Res. Lett., 24 (8): 897–900.
Endlich, R.M. 1967. An iterative method for altering the kinematic properties of wind fields.
J. Appl. Met., 6: 837–44.
Environmental Data Service, NOAA. 1972. Key to Meteorological Records Documentation
54. Environmental Satellite Imagery, US Department of Commerce, Washington, DC (subse-
quently published by the National Environmental Satellite Data and Information Service,
NOAA).
Essenwanger, O. 1955. Zur Realität der Zerlegung von Häufigkeitsverteilungen in Normalkurven.
Arch. Met. Geophys. Biokl. B, 7: 49–50.
Essenwanger, O. 1960a. Frequency distributions of precipitation. In: H. Weickmann ed., Physics of
Precipitation. Geophys. Monogr. 5, Amer. Geophys. Union, Washington, DC, pp. 271–8.
Essenwanger, O. 1960b. Linear and logarithmic scale for frequency distribution of precipitation.
Geof. pura e appl., 45: 199–214.
Essenwanger, O. 1976. Applied Statistics in Atmospheric Sciences. (a). Frequencies and Curve
Fitting. Elsevier, Amsterdam, 412 pp.
Essenwanger, O.M. 1986. Elements of Statistical Analysis. World Survey of Climatology. General
1 Climatology, 1B (ed. in chief E. Landsberg), Elsevier, Amsterdam, 424 pp.
Evans, M.S., Keyser, D., Bosart, L.F., and Lackmann, G.M. 1994. A satellite-derived classification
scheme for rapid maritime cyclogenesis. Mon. Wea. Rev., 122 (7): 1381–416.
Eyre, J.R. 1981. Meteosat water vapor imagery. Met. Mag., 110: 345–53.
Ferguson, H.L. 1961. A Geostrophic Advection Scale for Constant Pressure Surfaces. Met. Branch,
CIR-3516, TEC-363. Department of Transport, Toronto, 9 pp. (and Amendment, CIR-3516, TEC-
411).
Ferguson, H.L. 1963. A Geostrophic Advection Scale for Polar Stereographic Charts. Met. Branch.,
CIR-3957, TEC-473, Department of Transport, Toronto, 5 pp.
Ferraro, R.R. 1997. Special Sensor Microwave Imagery derived global rainfall estimates for climato-
logical applications. J. Geophys. Res., 102: 16715–35.
Ferraro, R.R., Weng, F., Grody, N.C., and Basist, A. 1996. An eight-year (1987–1994) time series
of rainfall, clouds, water vapor, snow cover, and sea ice derived from SSM/I measurements. Bull.
Amer. Met. Soc., 77 (5): 891–905.
Fiedler, V.F. 1965. Woraus bestehen die Haufigkeitsverteilungen der Tagestemperaturen von
Frankfürt. Zeit. Met., 17: 305–10.
Fisher, N.I. 1993. Statistical Analysis of Circular Data. Cambridge University Press, Cambridge,
277 pp.
Fisher, N.I. and Lee, A.J. 1983. A correlation coefficient for circular data. Biometrika, 70:
327–32.
Fleming, J.R. 1990. Meteorology in America, 1800–1870. Johns Hopkins University Press, Baltimore
MD, 264 pp.
Fliri, F. 1967. Beiträge zur Kenntnis der Zeit-Raum-Struktur des Niederschlags in den Alpen. Wetter
u. Leben, 19: 241–68.
Forbes, G.S. and Lottes, W.D. 1985. Classification of mesoscale vortices in polar airstreams and
the influence of the large-scale environment on their evolutions. Tellus, 37A: 132–55.
11 Forsythe, J.M. and Vonder Haar, T.H. 1996. A warm core in a polar low observed with a satellite
microwave sounding unit. Tellus, 48A: 193–208.
Foster, J.L., Chang, A.T.C., Hall, D.K., and Rango, A. 1991. Derivation of snow water equivalent
in boreal forests using microwave radiometry. Arctic, 44 (Suppl. 1): 147–52.
Fuelberg, H.E., Loring, R.O., Jr, Watson, M., Sinha, M., Blake, D.R., and Schoeberl, M.R. 1996.
TRACE: a trajectory intercomparison. 2. Isentropic and kinematic methods. J. Geophys. Res.,
101 (D19): 23927–39.
Gaffen, D.J. 1994. Temporal inhomogeneities in the radiosonde temperature record. J. Geophys.
Res., 99 (D2): 3667–76.
Gandin, L.S. 1965. Objective Analysis of Meteorological Fields (Leningrad, Gidromet. Izdat. 1963).
0 Isreal Prog. Sci. Trans., Jerusalem, Israel, 242 pp.
Garcia, O. 1985. Atlas of Highly Reflective Clouds for the Global Tropics: 1971–1983. NOAA-
ERL, Boulder CO. 365 pp.
Ghil, M. and Yiou, P. 1996. Spectral methods: what they can and cannot do for climatic time series.
111 In D.L.T. Anderson and J. Willebrand, eds, Decadal Climate Variability: Dynamics and
Predictability, Springer-Verlag, Berlin, pp. 445–82.
Gibson, J.K. 1998. ECMWF re-analysis – future plans. In: Proceedings, First WCRP International
Conference on Reanalysis, WCRP-104, WMO/TD. 876, Geneva, pp. 406–10.
Glahn, H.R. 1968. Canonical correlation and its relationship to discriminant analysis and multiple
regression. J. Atmos. Sci., 25: 23–31.
Glasspoole, J. 1925. Relation between annual rainfall over Europe, Oxford and Glenquoich. Brit.
0 Rainfall, 65: 254–69.
Gloersen, P., Campbell, W.J., Cavalieri, D.J., Comiso, J.C., Parkinson, C.L., and Zwally, H.J. 1992.
Arctic and Antarctic Sea Ice, 1978–1987: Satellite passive-microwave observations and analysis,
NASA SP-511, NASA, Washington DC., 290 pp.
Gloersen, P., Cavalieri, D.J., Chang, A.T.C., Wilheit, T.T., Campbell, W.J., Johannessen, O.M.,
Katsaros, K.B., Kunzi, K.F., Ross, D.B., Staelin, D., Windsor, E.P.L., Barath, F.T., Gudmandsen,
P., Langham, E., and Ramseier, O. 1984. A summary of results from the first NIMBUS-7 SMMR
observations. J. Geophys. Res., 89: 5335–44.
Godske, C.L. 1966. A statistical approach to climatology. Archive. Met. Geophys. Biokl., B 14:
269–79.
Gong, X-F. and Richman, M.B. 1995. On the application of cluster analysis to growing season
0111 precipitation data in North America east of the Rockies. J. Climate, 8 (4): 897–931.
Goodberlet, M.A., Swift, C.T., and Wilkerson, J.L. 1989. Remote sensing of ocean surface winds
with the Special Sensor Microwave/Imager. J. Geophys. Res., 94 (C10): 14547–55.
Goodison, B.E. 1989. Determination of areal snow water equivalent on the Canadian prairies using
passive microwave data. IGARSS ’89 (Vancouver), Proceedings, III, pp. 1243–6.
Goodison. B.E. and Walker, A.E. 1995. Canadian development and use of snow cover information
from passive microwave satellite data. In: B.J. Choudhury, Y.H. Kerr, E.G. Njoku and P.
Pampaloni, eds, Passive Microwave Remote Sensing of Land – Atmosphere Interactions, VSP
Publishers, Zeist, Netherlands, pp. 245–62.
Gordon, A., Grace, W., Schwerdtfeger, P., and Byron-Scott, R. 1998. Dynamic Meteorology: A
Basic Course, Arnold, London, 325 pp.
0 Gould, P.R. 1967. On the geographical interpretation of eigen values. Trans. Inst. Brit. Geog., 42:
53–86.
Graham, N.E., Barnett, T.P., Wilde, R., Ponater, M., and Schubert, S. 1994. On the roles of trop-
ical and midlatitude SSTs in forcing interannual to interdecadal variability in winter northern
hemisphere circulation. J. Climate, 7 (9): 1416–41.
Graystone, P. 1962. The introduction of topographic and frictional effects in a baroclimic model.
Quart. J. Roy. Met. Soc., 88: 256–70.
Green, M.C., Flocchini, R.G., and Myrup, L.O. 1993. Use of temporal principal components analysis
to determine seasonal periods. J. Appl. Met., 32: 986–95.
Griffith, C.G. 1987. Comparisons of gauge and satellite rain estimates for the central United States
during August 1979. J. Geophys. Res., 92: 9551–66.
0 Grody, N.C. 1991. Classification of snow cover and precipitation using the Special Sensor
Microwave Imager. J. Geophys. Res., 96 (D4): 7423–35.
Grossman, R.L. and Garcia, O. 1990. The distribution of deep convection over ocean and land
11 during the Asian summer monsoon. J. Climate, 3: 1032–44.
Gumbel, E.J. 1954. Applications of the circular normal distribution. J. Amer. Sta. Assoc., 49:
267–97.
Gupta, S.K., Staylor, W.F., Darnell, W.L., Wilber, A.C., and Ritchey, N.A. 1993. Seasonal varia-
tion of surface and atmospheric cloud radiative forcing over the globe derived from satellite data.
J. Geophys. Res., 98: 20761–78.
Guttman, L. 1954. Some necessary conditions for common-factor analysis. Psychometrika, 19:
149–61.
Haagenson, P.L. and Sperry, P.D. 1989. A relationship of isentropic back trajectories with observed
wind direction and synoptic type in the North Atlantic. J. Appl. Met., 28: 25–42.
Haltiner, G.J., Clarke, L.C., and Lawniczak, G.E., Jr. 1963. Computation of the large-scale vertical
velocity. J. Appl. Met., 2: 242–59.
Hamed, K.H. and Rao, R.A. 1998. A modified Mann–Kendall trend test for autocorrelated data. J.
Hydrol., 204: 182–96.
Harbaugh, J.W. and Merriam, D.P. 1968. Computer Applications in Stratigraphic Analysis. Wiley,
1 New York, 282 pp.
Hare, F.K. 1979. Climatic variation and variability: empirical evidence from meteorological and
other sources. In: Proceedings of the World Climate Conference, WMO Publ. 537, World
Meterological Organization, Geneva, pp. 51–87.
Harley, W.S., Dragert, H., and Rutherford, I.D. 1964. The Determination of Spot Values of Vertical
Velocity and Precipitation. Met. Branch, CIR-4139, TEC-544. Department of Transport, Toronto,
4 pp.
Harries, J.E. 1996. The greenhouse Earth: a view from space. Quart. J. Roy. Met. Soc., 122: 799–818.
Harris, J.M. 1992. An analysis of five-day midtropospheric flow patterns for the South Pole, 1985–89.
Tellus, 44B: 409–21.
Harris, J.M. and Kahl, J.D. 1990. A descriptive atmospheric transport climatology for the Mauna
Loa Observatory using clustered trajectories. J. Geophys. Res., 95 (D9): 13651–7.
Harris, J.M. and Kahl, J.D. 1994. Analysis of ten-day isentropic flow patterns for Barrow, Alaska,
1985–92. J. Geophys. Res., 99 (D): 25845–55.
Hawkins, D.M., Muller, M.W., and Krooden, J.A.T. 1982. Cluster analysis. In: D.M. Hawkins, ed.,
Topics in Multivariate Analysis, Cambridge University Press, Cambridge, pp. 303–53.
Hayden, C.M., Wade, G.S., and Schmit, T.J. 1996. Derived product imagery from GOES-8. J. Appl.
Met., 35: 153–62.
1 Heckley, W.A., Kelly, G., and Tiedtke, M. 1990. On the use of satellite-derived heating rates for
data assimilation within the tropics. Mon. Wea. Rev., 118: 1743–57.
Heffter, J.L. 1980. Air Resources Laboratory atmospheric transport and dispersion model. Tech.
Memo. ERL ARL-81, NOAA, Silver Springs MD, 17 pp.
Heffter, J.L. 1983. Branching Atmospheric Trajectory (BAT) model. Tech. Memo. ERL ARL-121,
NOAA, Silver Springs MD, 19 pp.
Henderson-Sellers, A.J. 1978. Surface type and its effect upon cloud cover: a climatological inves-
tigation. J. Geophys. Res., 83 (C10): 5057–62.
Henderson-Sellers, A.J. and McGuffie K. 1991. An investigation of the Burger distribution to char-
acterize cloudiness. J. Climate, 4: 1181–209.
Henderson-Sellers, A., Hughes, N.A., and Wilson, M. 1981. Cloud cover archiving on a global
scale: a discussion of principles. Bull. Amer. Met. Soc., 62: 1300–7.
Henderson-Sellers, A., Seze, G., Drake, F., and Desbois, M. 1987. Surface-observed and satellite-
retrieved cloudiness compared for the 1983 ISCCP Special Study Area in Europe. J. Geophys.
Res., 92: 4019–33.
Hendon, H.H. and Woodberry, K. 1993. The diurnal cycle of tropical convection. J. Geophys. Res.,
98: 16623–37.
Herman, A., Kumar, V.B., Arkin, P.A., and Kousky, J.V. 1997. Objectively-determined 10-day
African rainfall estimates created for famine early warning systems. Intl. J. Remote Sens., 18:
2147–59.
Hoaglin, D., Mosteller, F., and Tukey, J. 1985. Exploring Data Tables, Trends and Shapes, Wiley,
New York, 527 pp.
Hogben, G.L. 1946. A theoretical note on some errors in estimating the curvature of air trajecto-
ries and streamlines. Quart. J. Roy. Met. Soc., 72: 318–22.
Hoinka, K.P. 1997. The tropopause: discovery, definition and demarcation. Met. Zeit., N.F. 6:
281–303.
11 Holton, J.R. 1992. An Introduction to Dynamic Meteorology 3rd edition, Academic Press, 511 pp.
Horn, L.H. and Bryson R.A. 1960. Harmonic analysis of the march of annual precipitation over
the United States. Ann. Assoc. Amer. Geogr., 50: 157–71.
Howell, J.F. 1995. Identifying sudden changes in data. Mon. Wea. Rev., 123 (4): 1207–12.
Hsu, C.S., Wurtlee, M.G., Cunningham, G.F., and Woiceshyn, P.M. 1997. Construction of marine
surface pressure fields from scatterometer winds alone. J. Appl. Met., 36: 1249–61.
Hudson, G. and Wackernagel, H. 1994. Mapping temperature using kriging with external drift:
theory and an example from Scotland. Intl. J. Climatol., 14 (1): 77–91.
Huff, F.A. and Shipp, W.L. 1968. Mesoscale spatial variability in Midwestern precipitation.
J. Appl. Met., 7: 886–91.
0 Huffman, G.J., Adler, R.F., Arkin, P., Chang, A., Ferraro, R., Gruber, A., Janowiak, J., McNab,
A., Rudolf, B., and Schneider, U. 1997. The Global Precipitation Climatology Project (GPCP)
Combined Precipitation Dataset. Bull. Amer. Met. Soc., 78: 5–20.
Hurrell, J.W. and Campbell, G.G. 1992. Monthly Mean Global Data Sets available in CCM History
111 Tape Format, Tech. Note NCAR/TN 371 STR. NCAR, Boulder CO., 94 pp.
Hurst, H.E. 1951. Long-term storage capacity of reservoirs. Trans. Amer. Soc. Civ. Engr., 116:
770–99.
Hutchison, K.D. and Locke, J.K. 1997. Snow cover identification through cirrus-cloudy atmospheres
using daytime AVHRR imagery. Geophys. Res. Lett., 24: 1791–4.
Jackson, J.E. 1991. A User’s Guide to Principal Components., Wiley, New York, 569 pp.
Jaeger, G. 1984. Satellite indicators of rapid cyclogenesis. Mar. Wea. Log, 28 (1): 1–6.
0 Janowiak, J.E., 1992. Tropical rainfall: a comparison of satellite-derived rainfall estimates
with model precipitation forecasts, climatologies and observations. Mon. Wea. Rev., 120:
448–62.
Jarvis, E.C. and Agnew, T. 1970. A note on the computation of terrain and frictionally induced
vertical velocities. J. Appl. Met., 9: 942–6.
Jarvis. E.C. and Leonard, R. 1969. Vertical Velocities Induced by Smoothed Topography and their
Use in Areal Forecasting. Met. Branch Tech. Mem. 728, Department of Transport, Toronto.
Jenkins, G.M. and Watts. D.G. 1968. Spectral Analysis and its Applications. Holden-Day, San
Francisco, 525 pp.
Jenne, R.L. and McKee, T.B. 1985. Data. In: D.D. Houghton, ed., Handbook of Applied Meteoro-
0111 logy, Wiley, New York, pp. 1175–281.
Johnson, R.A. and Wehrly, T. 1977. Measures and models of angular correlation and angular linear
correlation. J. Roy. Stat. Soc., B 39: 222–9.
Johnson, R.A. and Wichern, D.W. 1982. Applied Multivariate Statistical Analysis. Prentice-Hall,
Englewood Cliffs NJ.
Johnston, R.J. 1980. Multivariate Statistical Analysis in Geography. Longman, London, pp. 127–82
and 183–201.
Joliffe, I.T. 1983. Large falls of rain in Wales: a simple statistical case study. Weather, 38 (4):
103–6.
Joliffe, I.T. 1986. Principal Component Analysis. Springer-Verlag, New York, 271 pp.
Joliffe, I.T. 1990. Principal component analysis: a beginner’s guide. I. Introduction and application.
0 Weather, 45 (10): 375–82.
Joliffe, I.T. 1993. Principal component analysis: a beginner’s guide. II. Pitfalls, myths and exten-
sions. Weather, 48 (8): 246–53.
Jones, P.D. 1987. The twentieth century Arctic high: fact or fiction? Clim. Dynam., 1: 63–75.
Jones, P.D. 1991. Southern hemisphere sea-level pressure data: an analysis and reconstructions back
to 1951 and 1911. Intl. J. Climatol., 11: 585–607.
Jones, P.D., Wigley, T.M.L., and Briffa, K.R. 1987. Monthly Mean Pressure Reconstructions for
Europe (back to 1780) and North America (to 1858). US Department of Energy, TRO 37
(DOE/ER/60397-H1), Washington DC, 99 pp.
Jones, P.D., et al. 1999. Monthly mean pressure reconstuctions for Europe for the 1780–1995 period.
Intl. J. Climatol., 19: 347–64
0 Julian, P.R. 1967. Variance spectrum analysis. Water Resour. Res., 3: 831–45.
Junker, N.W. and Haller, D.J. 1980. Estimation of surface pressures from satellite cloud patterns.
Mar. Wea. Log, 24 (3): 83–7.
11 Kagan, R.L. 1997. Averaging of Meteorological Fields. Kluwer, Dordrecht, 279 pp.
Kahl, J.D. and Samson, P.J. 1988. Uncertainty in estimating boundary-layer transport during highly
convective situations. J. Appl. Met., 27: 1024–35.
Kahl, J.D., Harris, J.M., Herbert, G.A., and Olson, M.P. 1989. Intercomparison of three long-range
trajectory models applied to Arctic haze. Tellus, 41B: 524–36.
Kahl, J.D.W., Martinez, D.A., Kuhns, H., Davidson, C.I., Jaffreze, J-L., and Harris, J.M. 1997. Air
mass trajectories to Summit, Greenland: a 44-year climatology and some episodic events. J.
Geophys. Res., 102 (C12): 26861–75.
Kalkstein, L.S., Tan, G., and Skindlov, J.A. 1987. An evaluation of objective clustering procedures
for use in synoptic climatological classification. J. Clim. Appl. Met. 26: 717–30.
Kalnay E., Kistler, R., and Kananitsu, M. 1998. NCEP/NCAR 40-year reanalysis overview. In:
Proceedings, First WCRP International Conference on Reanalysis, WCRP-104. WMO/TD. No.
876, Geneva, pp.1–7.
Kalnay, E., et al. 1996. The NCEP/NCAR 40-year re-analysis project. Bull. Amer. Met. Soc., 77:
437–71.
1 Kaplansky, J. 1945. A common error concerning kurtosis. J. Amer. Stat. Assoc., 40: 259.
Katsaros, K.B. and Lewis, R.M. 1986. Mesoscale and synoptic-scale features of North Pacific weather
systems observed with the Scanning Multichannel Microwave Radiometer on Nimbus 7. J.
Geophys. Res., 91: 2321–30.
Katsaros, K.B., Bhatti, I., McMurdie, L.A., and Petty, G.W. 1989. Identification of atmospheric fronts
over the ocean with microwave measurements of water vapor and rain. Wea. Forecast., 4: 449–60.
Katz, R.W. and Skaggs, R.H. 1981. On the use of autoregressive-moving average processes to model
meteorological time series. Mon. Wea. Rev., 109 (3): 479–84.
Keller, L.M. and Johnson, D.R. 1992. An atmospheric energy analysis of the impact of satellite lidar
winds and TIROS temperatures in global simulations. Mon. Wea. Rev., 120: 2831–52.
Kendall, M. 1976. Time Series, 2nd edition, Griffin, London, 197 pp.
Khrgian, A.K. 1970. Meteorology: A Historical Survey I, Gidromet. Izdat., Leningrad, 1959; 2nd edi-
tion. Israel Program for Scientific Translations, Jerusalem, 381 pp.
Kidd, C. 1998. On rainfall retrieval using polarization-corrected temperatures. Intl. J. Remote Sens.,
19 (5): 981–96.
Kidder, S.Q. and Vonder Haar, T.H. 1990. On the use of satellites in Molniya orbits for meteoro-
logical observations of middle and high latitudes. J. Atmos. Oceanic Technol., 7: 517–22.
Kidder, S.Q. and Wu, H.-T. 1984. Dramatic contrast between low clouds and snow cover in daytime
1 3.7 m imagery. Mon. Wea. Rev., 112: 2345–6.
Kim, K.-Y. and Wu, Q.-G. 1999. A comparison study of EOF techniques: analysis of nonstationary
data with periodic statistics, J. Climate, 12 (1): 185–99.
King, J.C. and Turner, J. 1997. Antarctic Meteorology and Climatology. Cambridge University Press,
Cambridge, 409 pp.
Kington, J.A. 1991. The application of synoptic weather mapping to historical climatology, with par-
ticular reference to the period 1780–1820. In: R. Glaser and R. Walsh, eds, Historical Climatology
in different Climatic Zones, Wüerrzburg. Geogr. Arbeiten 80, pp. 111–25.
Klein, W.H. 1963. Specification of precipitation from the 700 mb circulation. Mon. Wea. Rev., 91:
527–36.
Klein, W.H. 1965. Five-day precipitation patterns derived from circulation and moisture. In E.J.
Ambdur, ed., Humidity and Moisture. II Applications. Reinhold, New York, pp. 532–49.
Klein, W.H. and Kline, J.M. 1984. The synoptic climatology of monthly mean surface temperature
in the United States during winter relative to the surrounding 700 mb height field. Mon. Wea. Rev.,
112: 443–8.
Kleiner, B. and Graedel, T.E. 1980. Exploratory data analysis in the geophysical sciences. Rev.
Geophys. Space. Phys., 18: 699–717.
Klemes, V. 1974. The Hurst phenomenon: a puzzle? Water. Resour. Res., 10: 675–88.
Klink, K. 1998. Complementary use of scalar, directional and vector statistics with an application to
surface winds. Profess. Geogr., 50: 3–13.
Knighting, E. 1960. Some computations of the variation of vertical velocity with pressure on a syn-
optic scale. Quart. J. Roy. Met. Soc., 86: 318–25.
Kohler, M.A. 1949. Double-mass analysis for testing the consistency of records and making adjust-
ments. Bull. Amer. Met. Soc., 30: 188–9.
Krishnamurti, T.N. 1969. An experiment in numerical prediction in equatorial latitudes. Quart. J.
Roy. Met. Soc., 95: 594–620.
11 Kroehl, H.K., Scharfen, G.R., Arrance, E.S., and Goodman, S.G. 1994. An archive of digital data
from the Defense Meteorological Satellite Program (DMSP). In: Proceedings, Tenth International
Conference on Interactive Information and Processing Systems for Meteorology, Oceano-
graphy, and Hydrology, January 23–28 1994, Nashville TN, Amer. Met. Soc. Boston MA,
pp. 151–3.
Kuhnel, I. 1989. Tropical–extratropical cloudband climatology based on satellite data. Intl. J.
Climatol., 9: 441–63.
Kuhnel, I. 1990. Tropical–extratropical cloudbands in the Australian region. Intl. J. Climatol., 10:
341–64.
Kummerow, C. 1998. Beamfilling errors in passive microwave rainfall retrievals. J. Appl. Met., 37
0 (4): 356–70.
Kwok, R., Cunningham, G., and Holt, B. 1992. An approach to identification of sea ice types from
spaceborne SAR data. In: F.D. Carsey, ed., Microwave Remote Sensing of Sea Ice, Geophys.
Monogr. 68, American Geophysical Union, Washington DC, pp. 355–60.
111 Lachlan-Cope, T.A. and Turner, J. 1997. Passive microwave retrievals of precipitation over the
Southern Ocean. Intl. J. Remote Sens., 18: 1725–42.
Laing, A.G. and Fritsch, J.M. 1993a. Mesoscale convective complexes over the Indian monsoon
region. J. Climate, 6: 911–19.
Laing, A.G. and Fritsch, J.M. 1993b. Mesoscale convective complexes in Africa. Mon. Wea. Rev.,
121: 2254–63.
Lambert, S.J. 1990. Discontinuities in the long-term northern hemisphere 500 mb heights data set.
0 J. Climate, 3: 1479–84.
Lamberty, G.L. and Smith, P.J. 1993. A study of the influence of satellite data on GLA analyses
over the Atlantic Ocean during a period of blocking anticyclone development. Mon. Wea. Rev.,
121: 1881–91.
Landers, H. 1955. A three-dimensional study of the horizontal velocity divergence. J. Met., 12:
415–27.
Landsberg, H.E., Mitchell, J.M., Jr and Crutcher, H.L. 1963. Surface signs of the biennial atmos-
pheric pulse. Mon. Wea. Rev., 91: 549–56.
Lanzante, J.R. 1996. Resistant, robust and non-parametric techniques for the analysis of climate
data: theory and examples, including applications to historical radiosonde stations. Intl. J.
Climatol., 16 (11): 1197–226.
0111 Lau, K.-M. and Weng H.-Y. 1995. Climate signal detection using wavelet transform: how to make
a time series sing. Bull. Amer. Met. Soc., 76 (12): 2391–402.
Legates, D.R. 1991. The effect of domain shape on principal components analyses. Intl. J. Climatol.,
11: 135–46.
Legates, D.R. and Willmott, C.J. 1990. Mean seasonal and spatial variability in gauge-corrected
global precipitation. Intl. J. Climatol., 10 (2): 111–28.
Levy, G. and Brown, R.A. 1991. Southern hemisphere synoptic weather from a satellite scat-
terometer. Mon. Wea. Rev., 119: 2803–13.
Lindsay, R.W. and Rothrock, D.A. 1994. Arctic sea ice albedo from AVHRR. J. Climate, 7 (11):
1737–49.
Lindsay, R.W. and Rothrock, D.A. 1995. Arctic sea ice leads from Advanced Very High Resolution
0 Radiometer Images. J. Geophys. Res., 100: 4533–44.
Liu, G. and Curry, J.A. 1992. Retrieval of precipitation from satellite microwave measurement using
both emission and scattering. J. Geophys. Res., 97: 9959–74.
Liu, G. and Curry, J.A. 1993. Determination of characteristic features of cloud liquid water from
satellite microwave measurements. J. Geophys. Res., 98: 5069–92.
Livezey, R.E. 1995. Field intercomparison. In: H. von Storch and A. Navarra, eds, Analysis of
Climate Variability: Applications of Statistical Techniques, I, Springer, New York, pp. 159–76.
Livezey, R.E. and Chen, W.Y. 1983. Statistical field significance and its determination by Monte
Carlo statistics. Mon. Wea. Rev., 111 (1): 46–59.
Lorenz, E.N. 1967. The Nature and Theory of the General Circulation of the Atmosphere. WMO
218, TP 115, World Meteorological Organization, Geneva, 161 pp.
0 Lorenz, E.N. 1973. On the existence of extended range predictability. J. Appl. Met., 1 (3): 543–6.
MacQueen, J. 1967. Some methods for classification and analysis of multivariate observations. Proc.
Fifth Berkeley Symposium on Mathematical Statistics and Probability, I, University of California
11 Press, Berkeley CA, pp. 281–97.
Madden, R. and Sadeh, W. 1975. Empirical estimates of the standard error of time-averaged climatic
means. J. Appl. Met., 14 (2): 164–9.
Madden, R.A., 1979. A simple approximation for the variance of meteorological time averages.
J. Appl. Met., 18 (5): 703–6.
Maddox, R.A. 1980. Mesoscale convective complexes. Bull. Amer. Met. Soc., 61: 1374–87.
Maetzler, C. 1994. Passive microwave signatures of landscapes in winter. Met. Atmos. Phys. 54:
241–60.
Mandelbrot, B.B. and Wallis, J.R. 1968. Noah, Joseph and operational hydrology. Water Resour.
Res., 4: 909–18.
Mandelbrot, B.B. and Wallis, J.R. 1969a. Computer experiments with fractional Gaussian noises.
1. Averages and variances. Water Resour. Res., 5: 228–41.
Mandelbrot, B.B. and Wallis, J.R. 1969b. Computer experiments with fractional Gaussian noises.
2. Rescaled ranges and spectra. Water Resour. Res., 5: 242–59.
Mandelbrot, B.B. and Wallis, J. R. 1969c. Some long-run properties of geophysical records. Water
1 Resour. Res., 5: 321–40.
Mandelbrot, B.B. and Wallis, J.R. 1969d. Robustness of the rescaled range R/S in the measure-
ment of non-cyclic long-run statistical dependencies. Water Resour. Res., 5: 967–88.
Mann, M.E., Park, J., and Bradley, R.S. 1995. Global interdecadal and century-scale climate oscil-
lations during the past five centuries. Nature, 378: 266–70.
Maronna, R. and Yohai, V.J. 1978. A bivariate test for the detection of a systematic change in
mean. J. Amer. Stat. Assoc., 73: 640–5.
Marshall, G.J. and Turner, J. 1997. Surface wind fields of Antarctic mesocyclones derived from
ERS1 scatterometer data. J. Geophys. Res., 102 (D12): 13907–21.
Martin, D.W., Goodman, B., Schmit, T.J., and Cutrim, E.C. 1990. Estimates of daily rainfall over
the Amazon basin. J. Geophys. Res., 95: 17043–50.
Martin, D.W., Hinton, B.B., and Auvine, B.A. 1993. Three years of rainfall over the Indian Ocean.
Bull. Amer. Met. Soc., 74: 581–90.
Martin, F.L. and Salomonson, V.V. 1970. Statistical characteristics of subtropical jetstream features
in terms of MRIR observations from Nimbus II. J. Appl. Met., 9: 508–20.
Maslanik, J.A. 1992. Effects of weather on the retrieval of sea ice concentration and ice type from
passive microwave data. Intl. J. Rem. Sens., 13: 37–54.
Maslanik, J.A., Fowler, C., Heinrichs, J., Barry, R.G., and Emery, W.J. 1995. Remotely sensed and
1 simulated variability of Arctic sea ice concentrations in response to atmospheric synoptic systems.
Intl. J. Remote Sens., 16: 3325–42.
Maslanik, J.A., Serreze, M.C., and Barry, R.G. 1996. Recent decreases in Arctic summer ice cover
and linkages to atmospheric circulation anomalies. Geophys. Res. Lett., 23 (13): 1677–80.
Mason, B.J. 1971. Global Atmospheric Research Programme. Nature, 333: 382–8.
Massom, R. 1991. Satellite Remote Sensing of Polar Regions. Belhaven Press, London, 307 pp.
Matson, M., Ropelewski, C.F., and Varnadore, M.S. 1986. An Atlas of Satellite-derived Northern
Hemisphere Snow Cover Frequency, NOAA-NESDIS/NWS, NOAA, Washington DC, 75 pp.
McCutchan, M.H. and Schroeder, M.J. 1973. Classification of meteorological patterns in southern
California by discriminant analysis. J. Appl. Met., 7: 1466–75.
McGee, O.S. and Hastenrath, S.L. 1966. Harmonic analysis of the rainfall over South Africa. Notos,
15: 79–90.
McGinnigle, J.B. 1988. The development of instant occlusions in the North Atlantic. Met. Mag.,
117: 325–41.
McGinnigle, J.B. 1990. Numerical weather prediction model performance on instant occlusion devel-
opments. Met. Mag., 119: 149–63.
McGregor, J. and Gorman, A.J. 1994. Some considerations for using AVHRR data in climatolog-
ical studies. I. Orbital characteristics of NOAA satellites. Intl. J. Remote Sens., 15: 537–48.
McGuffie, K. 1993. Australian cloudiness from a high resolution satellite archive. Aust. Met. Mag.,
42 (1): 7–15.
McGuffie, K., Henderson-Sellers, A., and Goodman, A.H. 1989. Regional analysis of 3D
(three-dimensional) Nephanalysis total cloud amounts for July 1983. Intl. J. Remote Sens., 10:
1395–422.
McMurdie, L.A. and Katsaros, K.B. 1985. Atmospheric water distribution in a mid-latitude cyclone
observed by the Seasat Scanning Multichannel Microwave Radiometer. Mon. Wea. Rev., 113:
584–98.
11 McMurdie, L.A. and Katsaros, K.B. 1991. Satellite-derived integrated water vapor distribution in
oceanic midlatitude storms: variation with region and season. Mon. Wea. Rev., 119: 589–605.
McMurdie, L.A., Claud, C., and Atakturk, S. 1997. Satellite-derived atmospheric characteristics of
spiral and comma-shaped southern hemisphere mesocyclones. J. Geophys. Res., 102 (D12):
13889–905.
McMurdie, L.A., Levy, G., and Katsaros, K.B. 1987. On the relationship between scatterometer-
derived convergences and atmospheric moisture. Mon. Wea. Rev., 1281–94.
Menzel, W.P., Holt, F.C., Schmit, T.J., Aune, R.M., Schreiner, A.J., Wade, G.S., and Gray, D.G.
1998. Application of GOES-8/9 soundings to weather forecasting and nowcasting. Bull. Amer.
Met. Soc., 79: 2059–77.
0 Merrill, J.T. 1996. Trajectory results and interpretation for PEM-West A. J. Geophys. Res., 101
(D1): 1679–90.
Merrill, J.T., Bleck, R., and Avila, L. 1986. Techniques of Lagrangian trajectory analysis in isen-
tropic coordinates. Mon. Wea. Rev., 114: 571–81.
111 Meteorological Satellite Center, Japan. 1978–96. Monthly Report of the Meteorological Satellite
Center, Tokyo, Japan (CD-ROMs from January 1996).
Miles, M.W. and Barry, R.G. 1998. A five-year climatology of winter sea ice leads in the western
Arctic. J. Geophys. Res., 103 (C10): 21723–34.
Miller, J.E. 1948. Studies of Large-scale Vertical Motions of the Atmosphere. Met. Papers 1, New
York University, New York, pp. 1–48.
Miller, R.G. 1962. Statistical Prediction by Discriminant Analysis, Met. Monogr. 4 (25), Amer.
0 Met. Soc., Boston MA, 54 pp.
Mo, K. and Rasmussen, E.M. 1993. The 200 mb climatological vorticity budget during 1986–89
as revealed by NMC analyses. J. Climate, 6: 577–616.
Mognard, N.M. and Katsaros, K.B. 1995a. Statistical comparison of the Special Sensor
Microwave/Imager and the Geosat altimeter wind speed measurements over the ocean. Global
Atmos. Ocean Sys., 2: 291–9.
Mognard, N.M. and Katsaros, K.B. 1995b. Weather patterns over the ocean observed with the
Special Sensor Microwave/Imager and the Geosat altimeter. Global Atmos. Ocean Sys., 2: 301–23.
Mohr, K.I. and Zipser, E.J. 1996a. Defining mesoscale convective systems by their 85-GHz ice-
scattering signatures. Bull. Amer. Met. Soc., 77: 1179–89.
Mohr, K.I. and Zipser, E.J. 1996b. Mesoscale convective systems defined by their 85-GHz ice-scat-
0111 tering signature: size and intensity comparison over tropical oceans and continents. Mon. Wea.
Rev., 124: 2417–37.
Monserud, R.A. and Leemans, R. 1992. Comparing global vegetation maps with the kappa statistic.
Ecol. Modell., 62: 275–93.
Montgomery, R.B. 1937. A suggested method for representing gradient flow in isentropic surfaces.
Bull. Amer. Met. Soc., 18: 210–12.
Morcrette, J.-J. 1991. Evaluation of model-generated cloudiness: satellite-observed and model-gener-
ated diurnal variability of brightness temperature. Mon. Wea. Rev., 119: 1205–24.
Morrissey, M.L. and Greene, J.S. 1993. Comparison of two satellite-based rainfall algorithms using
Pacific atoll raingage data. J. Appl. Met., 32: 411–25.
Mueller, B.M. and Fuelberg, H.E. 1990. A simulation and diagnostic study of water vapor image
0 dry bands. Mon. Wea. Rev., 118: 705–22.
Mugnai, A., Cooper, H.J., Smith, E.A., and Tripoli, G.J. 1990. Simulation of microwave brightness
temperatures of an evolving hailstorm at SSM/I frequencies. Bull. Amer. Met. Soc., 71: 2–13.
Murray, R. and Lewis, R.P.W. 1966. Some aspects of the synoptic climatology of the British Isles
as measured by simple indices. Met. Mag., 95: 193–203.
Namias, J. 1940. An Introduction to the Study of Air Mass and Isentropic Analysis, 5th edition.
Amer. Met. Soc., Boston MA, 232 pp.
Namias, J. 1991. Spring and summer 1988 drought over the contiguous United States: causes and
prediction. J. Climate, 4: 54–65.
National Center for Atmospheric Research, Boulder, CO.
Negri, A.J., Adler, R.F., and Kummerow, C.D. 1989. False-color display of Special Sensor
0 Microwave/Imager (SSM/I) data. Bull. Amer. Met. Soc., 70: 146–51.
Negri, A.J., Adler, R.F., Maddox, R.A., Howard, K.W., and Keehn, P.R. 1993. A regional rainfall
climatology over Mexico and the southwest United States derived from passive microwave and
11 geosynchronous infrared data. J. Climate, 6: 2144–61.
Nicholls, N. 1987. The use of canonical correlation to study teleconnections. Mon. Wea. Rev., 115 (2):
393–9.
Njoku, E.G. 1985. Satellite-derived sea surface temperature: workshop comparisons. Bull. Amer. Met.
Soc., 66: 274–81.
Njoku, E.G. and Brown, O.B. 1993. Sea surface temperature. In: R.J. Gurney, J.L. Foster and C.L.
Parkinson, eds, Atlas of Satellite Observations related to Global Change, Cambridge University
Press, Cambridge, pp. 237–49.
Nordø, J. and Hjortnaes, K. 1967. Statistical studies of precipitation on local, national and continental
scales. Geofys. Publik. (Oslo), 26: 46.
North, E.R., Bell, T.L., Cahalan, R.F., and Moeng, F.-J. 1982. Sampling errors in the estimation of
empirical orthogonal functions. Mon. Wea. Rev., 110: 699–706.
Nosek, M. 1967. Varianzanalyse und Signifikanzteste in der dynamischen Klimatologie. Ann. Met., 20:
211–16.
Nuss, W.A. and Titley, D.W. 1994. Use of multiquadratic interpolation for meteorological objective
1 analysis. Mon. Wea. Rev., 122 (7): 1611–31.
Oelke, C. 1997. Atmospheric signatures in sea ice concentration estimates from passive microwaves:
modelled and observed. Intl. J. Rem. Sens., 18: 1113–36.
Oerlemans, J. 1978. An objective approach to breaks in the weather. Mon. Wea. Rev., 106: 1672–9.
O’Lenic, E.A. and Livezey, R.E. 1988. Practical considerations in the use of rotated principal compo-
nents analysis (RPCA) in diagnostic studies of upper air height fields. Mon. Wea. Rev., 116 (8):
1682–9.
Olkin, I., Gleser, L.J., and Derman, C. 1980. Probability Models and Applications. Macmillan, New
York, 576 pp.
Onstott, R.G. 1992. SAR and scatterometer signatures of sea ice. In: F.D. Carsey, ed., Microwave
Remote Sensing of Sea Ice, Geophys. Monogr. 68, American Geophysical Union, Washington, DC,
pp. 73–104.
Ose, T., Mechoso, C.R., and Halpern, D. 1994. A comparison between general circulation model sim-
ulations using two sea surface temperature datasets for January 1979. J. Climate, 7: 498–505.
Outcalt, S.I. , Hinkel, K.M., Meyer, E., and Brazel, A.J. 1997. Application of Hurst rescaling to geo-
physical serial data. Geogr. Anal., 29: 72–87.
Overland, J.E. and Preisendorfer, R.W. 1982. A significance test for principal components applied to a
cyclone climatology. Mon. Wea. Rev., 110: 1–4.
1 Palmén, E. and Holopainen, E.O. 1962. Divergence, vertical velocity and conversion between poten-
tial and kinetic energy in an extratropical disturbance. Geophysica, 8: 89–113.
Palmer, C.E. 1952. Tropical meteorology. Quart. J. Roy. Met. Soc., 78: 126–64.
Pankiewicz, G.S. 1995. Pattern recognition techniques for the identification of cloud and cloud sys-
tems. Met. Appl., 2: 257–71.
Panofsky, H.A. 1951. Large-scale vertical velocity and divergence. In: T.F. Malone, ed., Compendium
of Meteorology, Amer. Met. Soc., Boston MA, pp. 639–46.
Panofsky, H.A. and Brier, G.W. 1958. Some Applications of Statistics to Meteorology, Pennsylvania
State University Press, University Park PA, 224 pp.
Pardo-Iguzquiza, E. 1998. Comparison of geostatistical methods for estimating the areal average cli-
matological rainfall mean using data on precipitation and topography. Intl. J. Climatol., 18 (9):
1031–47.
Parkinson, C.L., Comiso, J.C., Zwally, H.J., Cavalieri, D.J., Gloersen, P., and Campbell, W.J. 1987.
Arctic Sea Ice, 1973–76: Satellite Passive-Microwave Observations. NASA SP-489, National
Aeronautics and Space Administration, Washington DC, 296 pp.
Parrish, D.D., et al. 1992. Indication of photochemical histories of Pacific air masses from measure-
ments of atmospheric trace species at Point Arena, California. J. Geophys. Res., 97 (D14):
15883–901.
Pearson, G.M. and Stewart, R.E. 1994. A diagnostic study of an apparent “instant occlusion” cycloge-
nesis event during ERICA. Atmos.-Ocean, 32: 259–84.
Pearson, K. (ed.) 1951. Tables of the Incomplete Gamma Function. Cambridge University Press,
Cambridge, 164 pp.
Penland C., Ghil, M., and Weickmann, K. 1991. Adaptive filtering and maximum entropy spectra with
application to changes in atmospheric angular momentum. J. Geophys. Res., 90 (D12): 22659–71.
Penner, C.M. 1963. An operational model for determination of vertical velocities. J. Appl. Met., 2:
235–41.
11 Persson, A. 1998. How well do we understand the Coriolis force? Bull. Amer. Met. Soc., 79 (7):
1373–85.
Perry, A.H. 1970. Filtering climatic anomaly fields using principal component analysis. Trans. Inst.
Brit. Geog., 50: 55–72.
Peterson, R.A. and Uccellini, L.W. 1979. The computation of isentropic trajectories using a “discrete
model” formulation. Mon. Wea. Rev., 107: 566–74.
Petterssen, S. 1956. Weather Analysis and Forecasting, I, McGraw-Hill, New York, 428 pp.
Petterssen, S., Bradbury, D.L., and Pedersen, K. 1962. The Norwegian cyclone models in relation
to heat and cold sources. Geofys. Publik. (Oslo) 24: 243–80.
Petty, G.W. 1994a. Physical retrievals of over-ocean rain rate from multichannel microwave imagery.
0 I. Theoretical characteristics of normalized polarization and scattering indices. Met. Atmos. Phys.,
54: 79–100.
Petty, G.W. 1994b. Physical retrievals of over-ocean rain rate from multichannel microwave imagery.
II. Algorithm implementation. Met. Atmos. Phys., 54: 101–22.
111 Petty, G.W. and Katsaros, K.B. 1992. Nimbus-7 SMMR precipitation observations calibrated against
surface radar during TAMEX. J. Appl. Met., 31: 489–505.
Pfister, C., Luterbacher, J., Schwarz-Zanetti, G., and Wegmann, M. 1998. Winter air temperature
variations in western Europe during the Early and High Middle Ages (AD 750–1300). Holocene
8 (5): 535–62.
Pickering, K.E., Thompson, A.M., McNamara, D.P., and Schoeberl, M.R. 1994. An intercompar-
ison of isentropic trajectories over the South Atlantic. Mon. Wea. Rev., 122 (5): 864–79.
0 Picon, L. and Desbois, M. 1990. Relation between METEOSAT water vapor radiance fields and
large-scale tropical circulation features. J. Climate, 3: 865–76.
Polyak, I. 1996. Computational Statistics in Climatology, Oxford University Press, Oxford, 358 pp.
Potter, K.W. 1981. Illustration of a new test for detecting a shift in mean in precipitation series.
Mon. Wea. Rev., 109 (9): 2040–5.
Prabhakara, C., Dalu, G., Liberti, G.L., Nucciarone, J.J., and Suhasini, R. 1992. Rainfall estima-
tion over oceans from SMMR and SSM/I microwave data. J. Appl. Met., 31: 532–52.
Prabhakara, C., Wang, I., Chang, A.T.C., and Gloersen, P. 1983. A statistical examination of
Nimbus-7 SMMR data and remote sensing of sea surface temperature, liquid water content in
the atmosphere and surface wind speed. J. Clim. Appl. Met., 22: 2023–37.
Prohaska, J.T. 1976. A technique for analyzing the linear relationship between two meteorological
0111 fields. Mon. Wea. Rev., 104 : 1345–53.
Puri, K. and Davidson, N.E. 1992. The use of infrared satellite cloud imagery data as proxy data
for moisture and diabatic heating in data assimilation. Mon. Wea. Rev., 120: 2329–41.
Rabin, R.M., McMurdie, L.A., Hayden, C.M., and Wade, G.S. 1991. Monitoring precipitable water
and surface wind over the Gulf of Mexico from microwave and VAS satellite imagery. Wea.
Forecast., 6: 227–43.
Räisänen, J. 1995. Factors affecting synoptic-scale vertical motions: a statistical study using a gener-
alized omega equation. Mon. Wea. Rev., 123 (8): 2447–60.
Ramond, D., Corbin, H., Desbois, M., Szejwach, G., and Waldteufel, P. 1981. The dynamics of
polar jetstreams as depicted by the METEOSAT water vapor channel radiance field. Mon. Wea.
Rev., 109: 2164–76.
0 Rao, G. and MacArthur, P.D. 1994. The SSM/I estimated rainfall amounts of tropical cyclones and
their potential in predicting the cyclone intensity changes. Mon. Wea. Rev., 122: 1568–74.
Rao, P.K., Holmes, S.J., Anderson, R.K., Winston, J.S., and Lehr, P.E., eds, 1990. A history of
civilian weather satellites. In: Weather Satellites: Systems, Data and Environmental Applications.
Amer. Met. Soc., Boston MA, pp. 7–19.
Rasmussen, E. 1979. The polar low as an extratropical CISM disturbance. Quart. J. Roy. Met. Soc.,
105: 531–49.
Rasmussen, E. 1981. An investigation of a polar low with a spiral cloud structure. J. Atmos. Sci.,
38: 1785–92.
Rasmusson, E.M. and Arkin, P.A. 1993. A global view of large-scale precipitation variability.
J. Climate, 6 (12): 1495–522.
0 Raustein, E., Sundqvist, H., and Katsaros, K.B. 1991. Quantitative comparison between simulated
cloudiness and clouds objectively derived from satellite data. Tellus, 43A: 306–20.
Rayment, R. 1970. Introduction to the fast Fourier transform (FFT) in the production of spectra.
11 Met. Mag., 99: 261–70.
Rayner, J.N. 1971. An Introduction to Spectral Analysis. Pion Press, London, 174 pp.
Raynor, G.S. and Hayes, J.V. 1982. Effect of varying air trajectories on spatial and temporal precip-
itation patterns. Water, Air and Soil Pollution, 18: 173–89.
Reed, R.J. 1979. Cyclogenesis in polar air streams. Mon. Wea. Rev., 107: 38–52.
Reed, R.J. and Albright, M.D. 1997. Frontal structure in the interior of an intense mature ocean
cyclone. Wea. Forecast, 12: 866–76.
Rex, D.F. 1958. Vertical atmospheric motion in the equatorial Pacific. Geophysica, 6: 479–500.
Richardson, L.F. 1922. Weather Prediction by Numerical Process. Cambridge University Press,
Cambridge.
Richman, M.B. 1986. Rotation of principal components. J. Climatol., 6: 293–335.
Richman, M.B. 1993. Comments on: “The effect of domain shape on principal components analysis.”
Intl. J. Climatol., 13: 203–18. (A reply to D.R. Legates, ibid., pp. 219–28.)
Rieland, M. and Stuhlmann, R. 1993. Toward the influence of clouds on the shortwave
radiation budget of the Earth–atmosphere system estimated from satellite data. J. Appl. Met., 32:
1 825–43.
Robeson, S.M. 1997. Spherical methods of spatial interpolation: review and evaluation. Cartogr.,
Geogr. Info. Syst., 24: 3–20.
Robinson, D.A., Dewey, K.F., and Heim, R.R., Jr. 1993. Global snow cover monitoring: an update.
Bull. Amer. Met. Soc., 74 (9): 1689–96.
Robinson, D.A., Serreze, M.C., Barry, R.G., Scharfen, G., and Kukla, G. 1992. Large-scale patterns
and variability of snow melt and parameterized surface albedo in the Arctic basin. J. Climate, 5
(10): 1109–19.
Rolph, G.D. and Draxler, R.R. 1990. Sensitivity of three-dimensional trajectories to the spatial and
temporal densities of the wind field. J. Appl. Met., 29: 1043–54.
Rossby, C.-G., et al. 1937. Isentropic analysis. Bull. Amer. Met. Soc., 18: 201–9.
Rossow, W.B. 1993. Clouds. In: R.J. Gurney, J.L. Foster and C.L. Parkinson, eds, Atlas of
Satellite Observations related to Global Change. Cambridge University Press, Cambridge,
pp. 141–63.
Rossow, W.B. and Lacis, A.A. 1990. Global seasonal cloud variations from satellite radiance
measurements. II. Cloud properties and radiative effects. J. Climate, 3: 1204–53.
Rothschild, V. and Logothetis, N. 1986. Probability Distributions, Wiley, New York, 70 pp.
Rott, H. 1987. Remote sensing of snow. In: B.E. Goodison, R.G. Barry and J. Dozier, eds, Large
1 Scale Effects of Seasonal Snow Cover, IAHS Publ. 166, Wallingford, UK, pp. 279–90.
Sadler, J.C. 1965. The feasibility of global tropical analysis. Bull. Amer. Met. Soc., 46: 118–30.
Salby, M.L. 1989. Climate monitoring from space: asynoptic sampling considerations. J. Climate,
2: 1091–105.
Salby, M.L. and Callaghan, P. 1997. Sampling error in climate properties derived from satellite
measurements: consequences of undersampled diurnal variability. J. Climate, 10: 18–36.
Salby, M.L., Hendon, H.H., Woodberry, K., and Tanaka, K. 1991. Analysis of global cloud imagery
from multiple satellites. Bull. Amer. Met. Soc., 72: 467–80.
Salstein, D.A., Rosen, R.D., Baker, W.E., and Kalnay, E. 1987. Impact of satellite-based data on
FGGE general circulation statistics. Quart. J. Roy. Met. Soc., 113: 255–77.
Sawyer, J.S. 1949. Large-scale vertical motion in the atmosphere: a discussion. Quart. J. Roy. Met.
Soc., 75: 185–8.
Scharfen, G.R., Knowles, K.W., Bauer, R.J., and Swick, R.S. 1995. Polar data sets from the Defense
Meteorological Satellite Program (DMSP) digital data archive. In: Proceedings of the Fourth
Conference on Polar Meteorology and Oceanography, January 15–20 1995. Am. Met. Soc.,
Dallas TX, 103–7.
Schmetz, J. and Turpeinen, O.M. 1988. Estimation of the upper tropospheric relative humidity field
from METEOSAT water vapor image data. J. Appl. Met., 27: 889–99.
Schumann, T.E.W. and Hofmeyr, W.L. 1942. The problem of autocorrelation of meteorological
time series. Quart. J. Roy. Met. Soc., 68: 177–88.
Schumann, T.E.W. and Van Rooy, M.P. 1952. The Autocorrelation of Daily Sea-level Pressure
over the Northern Hemisphere. W.B. 17, Weather Bureau, Pretoria, South Africa, 7 pp.
Schweiger, A.J., Serreze, M.C., and Key, J.R. 1993. Arctic sea ice albedo: a comparison of two
satellite-derived data sets. Geophys. Res. Lett., 20: 41–4.
Scorer, R.S. 1957. Vorticity. Weather, 12: 72–83.
Scorer, R.S. 1958. Natural Aerodynamics. Pergamon, Oxford, 312 pp.
11 Seibert, P. 1993. Convergence and accuracy of numerical methods for trajectory calculations.
J. Appl. Met., 32: 558–66.
Shapiro, L.J. and Goldenberg, S.B. 1998. Atlantic sea surface temperatures and tropical cyclone
formation. J. Climate, 11 (4): 578–90.
Shapiro, M.A., Krueger, A.J., and Kennedy, P.J. 1982. Nowcasting the position and intensity of
jetstreams using a satellite-borne total ozone mapping spectrometer. In: K.A. Browning, ed.,
Nowcasting, Academic Press, New York, pp. 137–45.
Shaw, D.B., Lönnberg, P., Hollingsworth, A., and Undén, P. 1987. Data assimilation: the
1984/85 revisions of the ECMWF mass and wind analysis. Quart. J. Roy. Met. Soc., 113:
533–66.
0 Shaw, W.N. 1930. Manual of Meteorology. III. The Physical Processes of Weather. Cambridge
University Press, Cambridge, pp. 259–66.
Shaw W.N. and Lemppfert, R.K.G. 1906. The Life History of Surface Air Currents and a Case
Study of Surface Trajectories of Moving Air. Met. Office 174, London, 107 pp.
111 Shea, D.J., Worley, S.J., Stern, I.R., and Hoar, T.J. 1996. An Introduction to Atmospheric and
Oceanographic Data Sets, NCAR Tech. Note TN-404, National Center for Atmospheric Research,
Boulder CO.
Shin, K.-S., Riba, P.E., and North, G.R. 1990. Estimation of area-averaged rainfall over tropical
oceans from microwave radiometry: a single channel approach. J. Appl. Met., 29: 1031–42.
Siefridt, L., Barnier, B., Legler, D.M., and O’Brien, J.J. 1998. 5-day average wind over
north-west Atlantic from ERS1 using a variational analysis. Global Atmos. Ocean Sys., 5 (4):
0 317–44.
Simmons, A.J., Burridge, D.M., Jarraud, M., Girard, C., and Wergen, W. 1989. The ECMWF
medium-range prediction models: development of the numerical formations and the impact of
increased resolution. Met. Atmos. Phys., 40: 28–60.
Simpson, J., Adler, R.F., and North, G.R. 1988. A proposed Tropical Rainfall Measuring Mission
(TRMM) satellite. Bull. Amer. Met. Soc., 69: 278–95.
Simpson, J., Halverson, J., Pierce, H., Morales, C., and Iguchi, T. 1998. Eyeing the eye: exciting
early stage science results from TRMM. Bull. Amer. Met. Soc., 79: 1711.
Small, M.J. and Sansom, P.J. 1983. Stochastic simulation of atmospheric trajectories. J. Clim. Appl.
Met., 22: 266–77.
Smigielski, F.J. and Mogil, H.M. 1995. A systematic satellite approach for estimating central surface
0111 pressures of mid-latitude cold season oceanic cyclones. Tellus, 47A: 876–91.
Smith, D.M. 1998. Observation of perennial Arctic sea ice melt and freeze-up using passive
microwave data. J. Geophys. Res., 103 (C12): 27753–69.
Smith, W.L. 1985. Satellites. In: D.D. Houghton, ed., Handbook of Applied Meteorology. Wiley,
New York, pp. 380–472.
Smith, W.L., Bishop, W.P., Dvorak, V.F., Hayden, C.M., McElroy, J.H., Mosher, F.R., Oliver, V.J.,
Purdom, J.F., and Wark, D.Q. 1986. The meteorological satellite: overview of 25 years of oper-
ation. Science, 231: 455–62.
Sneyers, R. 1976. Application of least squares to the search for periodicities. J. Appl. Met., 15:
387–93.
Soden, B.J. and Bretherton, F.P. 1993. Upper tropospheric relative humidity from the GOES 6.7
0 m channel: model and climatology for July 1987. J. Geophys. Res., 98: 16669–88.
Soden, B.J. and Bretherton, F.P. 1996. Interpretation of TOVS water vapor radiances in terms of
layer-average relative humidities: method and climatology for the upper, middle, and lower tropos-
phere. J. Geophys. Res., 101: 9333–43.
Sohn, B.-J. and Smith, E.A. 1992. Global energy transports and the influence of clouds on trans-
port requirements: a satellite analysis. J. Climate, 5: 717–34.
Solow, A.R. 1987. Testing for climate change: an application of the two-phase regression model.
J. Clim. Appl. Met., 26 (10): 1406–11.
Song, Y. and Carleton, A.M. 1997. Climatological “models” of cold air mesocyclones derived from
SSM/I data. Geocarto Intl., 12 (1): 79–89.
Spencer, R.W., Goodman, H.M., and Hood, R.E. 1989. Precipitation retrieval over land and ocean
0 with the SSM/I: identification and characteristics of the scattering signal. J. Atmos. Oceanic
Technol., 6 (2): 254–73.
Staley, D.O. 1966. The lapse rate of air temperature following an air parcel. Quart. J. Roy. Met.
11 Soc., 92: 147–50.
Steffen, K., Key, J., Cavalieri, D.J., Comiso, J., Gloersen, P., St Germain, K., and Rubinstein, I.
1992. The estimation of geophysical parameters using passive microwave algorithms. In: F.D.
Carsey, ed., Microwave Remote Sensing of Sea Ice. Geophys. Monogr. 68, American Geophysical
Union, Washington DC, pp. 201–301.
Stidd, C.K. 1953. The cube-root normal precipitation distribution. Trans. Amer. Geophys. Union,
15: 31–4.
Stidd, C.K. 1954. The use of correlation fields in relating precipitation to circulation. J. Met., 11:
202–13.
Stoffelen, A.C.M. and Cats, G.J. 1991. The impact of Seasat-A scatterometer data on high-resolu-
tion analyses and forecasts: the development of the QE II storm. Mon. Wea. Rev., 119: 2794–802.
Stohl, A. 1998. Computation, accuracy and applications of trajectories: a review and bibliography.
Atmos. Environ., 32 (6): 947–66.
Stohl, A. and Seibert, P. 1998. Accuracy of trajectories as determined from the conservation of
meteorological tracers. Quart. J. Roy. Met. Soc., 124: 1465–84.
1 Stohl, A., Wotawa, G., and Kromp-Kolb, H. 1995. Interpolation errors in wind fields as a function
of spatial and temporal resolution and their impact on different kinds of kinematic trajectories.
J. Appl. Met., 34 (10): 2149–65.
Streten, N.A. 1980. Some synoptic indices of the southern hemisphere mean sea level circulation,
1972–77. Mon. Wea. Rev., 108: 18–36.
Streten, N.A. and Kellas, W.R. 1973. Aspects of cloud pattern signatures of depressions in matu-
rity and decay. J. Appl. Met., 12: 23–7.
Streten, N.A. and Troup, A.J. 1973. A synoptic climatology of satellite observed cloud vortices
over the southern hemisphere I. Quart. J. Roy. Met. Soc., 99: 56–72.
Stubbs, M.W. 1981. New code for reporting surface observations: an introduction. Weather, 36
(12): 357–66.
Suggs, R.J., Jedlovec, G.J., and Guillory, A.R. 1998. Retrieval of geophysical parameters from
GOES: evaluation of a split window technique. J. Appl. Met., 37: 1205–27.
Susskind, J. 1993. Water vapor and temperature. In: R.J. Gurney, J.L. Foster and C.L. Parkinson,
eds, Atlas of Satellite Observations related to Global Change, Cambridge University Press,
Cambridge, pp. 89–128.
Suzuki, E. 1967. A statistical and climatological study on the rainfall in Japan. Pap. Met. Geophys.
(Tokyo), 18: 103–82.
1 Suzuki, E. 1969. A discrimination theory based on categorical variables and its application to mete-
orological variables. J. Met. Soc. Japan, ser. 2, 47: 145–58.
Tait, A.B. 1998. Estimation of snow water equivalent using passive microwave radiation data. Rem.
Sensing. Environ., 64: 286–9.
Thiao, W. and Turpeinen, O.M. 1992. Large-scale diurnal variations of tropical cold cloudiness
based on a simple cloud indexing method. J. Climate, 5: 173–80.
Thiebaux, H.J. and Pedder, M.A. 1987. Spatial Objective Analysis, Academic Press, New
York.
Thiebeaux, H.J. and Zwiers, F.W. 1984. The interpretation and estimation of effective sample size.
J. Clim. Appl. Met., 23: 800–11.
Thiessen, A.H. 1911. Precipitation averages for large areas. Mon. Wea. Rev., 39: 1082–4.
Thom, H.C.S. 1958. A note on the gamma distribution. Mon. Wea. Rev., 86: 117–22.
Thom, H.C.S. 1966. Some Methods of Climatological Analysis. Tech. Note 66, World Meteorological
Organization, Geneva, pp. 31–45.
Tjemkes, S.A., Stephens, G.L., and Jackson, D.L. 1991. Spaceborne observations of columnar water
vapor: SSMI observations and algorithm. J. Geophys. Res., 96: 10941–54.
Todd, M.C., Barrett, E.C., Beaumont, M.J., and Green, J.L. 1995. Satellite identification of rain
days over the upper Nile river basin using an optimum infrared rain/no-rain threshold tempera-
ture model. J. Appl. Met., 34: 2600–11.
Torrence, C. and Compo, G. 1998. A practical guide to wavelet analysis. Bull. Amer. Met. Soc.,
79 (1): 61–78.
Trenberth, K.E. and Olson, J.G. 1988. An evaluation and intercomparison of global analyses from
the National Meteorological Center and the European Center for Medium Range Weather
Forecasts. Bull. Amer. Met. Soc., 69: 1047–57.
Trenberth, K.E. and Paolino, D.A., Jr. 1980. The northern hemisphere sea-level pressure data set:
trends, errors and discontinuities. Mon. Wea. Rev., 108: 855–72.
11 Troup, A.J. and Streten, N.A. 1972. Satellite-observed southern hemisphere cloud vortices in relation to
conventional observations. J. Appl. Met., 11: 909–17.
Tucker, G.B. 1960. Upper Winds over the World. 3. Geophys. Mem. 13 (5), London, 101 pp.
Turner, J. and Warren, D.E. 1989. Cloud track winds in the polar regions from sequences of AVHRR
images. Intl. J. Remote Sens., 10: 695–703.
Ulrych, T. and Bishop, T.N. 1975. Maximum entropy spectral analysis and autoregressive decomposi-
tion. Rev. Geophys. Space Phys., 13: 183–200.
Vaisanen, A. 1961. Investigation of the Vertical Air Movement and related Phenomena in selected
Synoptic Situations. Paper 93, Inst. of Met., Helsinki, 72 pp.
Van den Hurk, B.J.J.M., Bastiaanssen, W.G.M., Pelgrum, H., and van Meijgaard, E. 1997. A new
0 methodology for assimilation of initial soil moisture fields in weather prediction models using
Meteosat and NOAA data. J. Appl. Met., 36: 1271–83.
Vautard, R., Yiou, P., and Ghil, M. 1992. Singular spectrum analysis: a toolkit for short noisy chaotic
signals. Physica, D58: 95–126.
111 Velden, C.S. 1989. Observational analyses of North Atlantic tropical cyclones from NOAA polar orbit-
ing satellite microwave data. J. Appl. Met., 28: 59–70.
Velden, C.S. 1992. Satellite-based microwave observations of tropopause-level thermal anomalies:
quantitative applications in extratropical cyclone events. Wea. Forecast., 7 (4): 669–82.
Vincente, G.A., Scofield, R.A., and Menzel, W.P. 1998. The operational GOES infrared rainfall esti-
mation technique. Bull. Amer. Met. Soc., 79: 1883–98.
von Storch, H. 1995a. Misuses of statistical analysis in climate research. In: H. von Storch and
0 A. Navarra, eds, Analysis of Climatic Variability: Applications of Statistical Techniques, Springer-
Verlag, Berlin, pp. 11–26.
von Storch, H. 1995b. Spatial patterns: EOFs and CCA. In: H. von Storch and A. Navarra, eds, Analysis
of Climate Variability: Applications of Statistical Techniques, Springer-Verlag, Berlin, pp. 227–57.
von Storch, H. and Hannoschock, G. 1985. Statistical aspects of estimated principal vectors (EOFs)
based on small sample sizes. J. Clim. Appl. Met., 24: 716–24.
Wachter, H. 1968. Häufigkeitsverteilung klimatologischer Grossen. Ber. Dtsch. Wetterdienst.
(Offenbach) 15 (107) 35 pp.
Wahba, G. 1990. Spline Models for Observational Data, SIAM, Philadelphia PA.
Walker, A.E. and Goodison, B.E. 1993. Discrimination of a wet snow cover using passive microwave
satellite data. Ann. Glaciol, 17: 307–11.
0111 Walker, J.M. 1967. Subterranean isobars. Weather, 22: 296–7.
Wallace, J.M., Smith, C., and Bretherton, C.S. 1992. Singular value decomposition of wintertime sea
surface and 500 mb height anomalies. J. Climate, 5: 561–76.
Wallis, J.R. and Matalas, N.C. 1971. Correlogram analysis revisited. Water Resour. Res., 7 (6): 1448–59.
Walmsley, J.L. and Mailhot, J. 1983. On the numerical accuracy of trajectory models for long-range
transport of atmospheric pollutants. Atmos.-Ocean, 21: 14–39.
Wang, X.-C. and Shen, S.S. 1999. Estimation of spatial degrees of freedom of a climate field.
J. Climate, 12 (5): 1280–91.
Wanner, H., Bradzil, R., Frich, P., Fryendahl, K., Jonsson, T., Kington, J., Pfister, C., Rosenorn, S., and
Wishman, E. 1994. Synoptic interpretation of monthly weather maps for the Late Maunder Minimum
(1675–1715). In: B. Frenzel, C. Pfister and B. Glaeser, eds, Climatic Trends and Anomalies in Europe,
0 1675–1715, Fischer, Stuttgart, pp. 401–25.
Watson, D.F. 1992. Contouring: A Guide to the Analysis and Display of Spatial Data, Pergamon,
Oxford, 340 pp.
Watts, I.E.M. 1955. Equatorial Meteorology, with particular Reference to Southeast Asia. University of
London Press, London, 223 pp.
Weatherhead, E.C., et al. 1998. Factors affecting the detection of trends: statistical considerations and
applications to environmental data. J. Geophys. Res., 103 (D14): 17149–61.
Webster, P.J., Magana, V.O., Palmer, T.N., Shukla, J., Tomas, R.A., Yanai, M., and Yasunari, T. 1998.
Monsoons: processes, predictability, and the prospects for prediction. J. Geophys. Res., 103 (C7):
14451–510.
Weng, F. and Grody, N.C. 1994. Retrieval of cloud liquid water using the special sensor microwave
0 imager (SSM/I). J. Geophys. Res., 99: 25535–51.
Werner, P. and von Storch, H. 1993. Interannual variability of central European mean temperature
in January/February and its relationship to the large-scale circulation. Clim. Res., 3: 195–
11 207.
Wilks, D.S. 1995. Statistical Methods in the Atmospheric Sciences: An Introduction, Academic
Press, San Diego CA, 467 pp.
Willmott, C.J. and Robeson, S.M. 1995. Climatologically aided interpolation (CAI) of terrestrial air
temperature. Intl. J. Climatol., 15: 221–9.
Willmott, C.J., Robeson, S.M., and Fedema, J.J. 1994. Estimating continental and terrestrial precip-
itation averages from rain gauge networks. Intl. J. Climatol., 14: 1403–14.
Willmott, C.J., Rowe, C.M., and Philpot, W.D. 1985. Small-scale climate maps: a sensitivity analysis
of some common assumptions associated with gridpoint interpolation and contouring. Amer.
Cartographer, 12: 5–16.
Wilson, L.L., Lettenmaier, D.P., and Skyllingstad, E. 1992. A hierarchical stochastic model of large-
scale atmospheric circulation patterns and multiple station daily precipitation. J. Geophys. Res.,
97: 2791–809.
WMO. 1995. Manual on Codes, I, 1, WMO 306. World Meteorological Organization, Geneva.
Wu, M.C. 1984. Radiation properties and emissivity parameterization of high level thin clouds.
1 J. Clim. Appl. Met., 23: 1138–47.
Wu, X., Bates, J.J., and Kalsa, S.J.S. 1993. A climatology of the water vapor band brightness
temperatures from NOAA operational satellites. J. Climate, 6: 1282–300.
Yamanouchi, T. and Charlock, T.P. 1997. Effects of clouds, ice sheet, and sea ice on the Earth
radiation budget in the Antarctic. J. Geophys. Res., 102: 6953–70.
Yates, H., Strong, A., McGinnis, D., Jr and Tarpley, D. 1986. Terrestrial observations from NOAA
operational satellites. Science, 231: 463–70.
Yiou, P., Baert, E., and Loutre, M.F. 1996. Spectral analysis of climate data. Rev. Geophys., 17:
619–63.
Zaitseva, N.A. and Ivanov, A.A. 1998. Radiosonding: history and accuracy of the method. In:
Proceedings of the First WCRP International Conference on Reanalysis. WCRP-104, WMO/TD
No. 876: 325–8, World Meterological Organization, Geneva.
Zeng, L. and Levy, G. 1995. Space and time aliasing structure in monthly mean polar-orbiting satel-
lite data. J. Geophys. Res., 100 (D3): 5133–42.
Zillman, J.W. and Price, P.G. 1972. On the thermal structure of mature Southern Ocean cyclones.
Aust. Met. Mag., 20 (1): 34–48.
Zillman, J.W., Griersmith, D., LeMarshall, J., and Gauntlett, D.J. 1990. Remote sensing applica-
tions in the Australian Bureau of Meteorology. Intl. J. Remote Sens., 11: 1979–97.
1 Zipser, E.J. and Colon, J.A. 1962. Mean layer wind charts in tropical analysis. Mon. Wea. Rev.,
90: 465–70.
Zwally, H.J., Comiso, J.C., Parkinson, C.L., Campbell, W.J., Carsey, F.D., and Gloersen, P. 1983.
Antarctic sea ice, 1973–1976: Satellite Passive-Microwave Observations, NASA SP-459, National
Aeronautics and Space Administration, Washington DC, 206 pp.
Zwiers, F.W. and von Storch, H. 1995. Taking serial correlation into account in tests of the mean.
J. Climate, 8 (2): 336–51.
11
Part 2
Dynamic climatology
11
11
3 Global climate and the general
circulation
Early treatments of global climate and its regional anomalies were limited to attempts to
generalize the observed climatic features. The studies of W. Köppen (1931) based on
temperature and precipitation classes sought to identify the hypothetical climate of an
ideal continent, for example, but this approach was limited by the unavoidable inclusion
of the large-scale effects of land–sea distribution and of major mountain barriers on the
atmospheric circulation. In a review of G.T. Trewartha’s (1961) book The Earth’s Problem
0 Climates, Tucker (1962) commented that adequate explanations of “normal” climatic
patterns on the continents are for the most part lacking. This underlying framework is
perhaps more problematic than the so-called anomalies. Climate modeling studies have
clarified considerably the determinants of the background planetary climate, the contrib-
utory role of the Earth’s geography, and the feedbacks internal to the climate system.
These are discussed in turn.
There are a number of ways in which the global circulation of the atmosphere can be
studied. The most obvious approach is empirical, using measurements of upper atmos-
pheric winds, temperature and vapor pressure collected at rawinsonde (radar wind
sounding) stations worldwide. However, as we have seen, the spatial coverage is hetero-
0 geneous and incomplete while the length of the record is barely fifty years. A second
approach is to use analog models, such as the rotating dishpan, to study highly simplified
rotating fluid systems in a controlled situation. A third is the development of increasingly
complex numerical general circulation models (GCMs) to simulate the three-dimensional
behavior of the atmosphere and its time evolution. Modern climates are simulated, as well
as the response of the model either to various imposed forcings (changes of solar input
or atmospheric carbon dioxide concentration, for example), or to changes in surface
boundary conditions (such as sea surface temperature anomalies, and Pleistocene ice
sheets). A fourth possibility is the comparative study of the contrasting circulation features
of planetary atmospheres using theory and limited observations. Elements of each of these
0 approaches are incorporated in the following survey. We begin by considering the plan-
etary setting.
3.1 Planetary controls

The phenomena of planetary atmospheres and their differing characteristics (Lewis and
Prinn, 1983; Chamberlin and Hunter, 1987) are the concern of dynamic meteorology. A
brief review of these questions serves as an introduction to the Earth’s atmosphere.
The basic factors determining planetary climate are its mean solar distance, orbit, mass,
and rotational characteristics, acting through the several laws of conservation. To these
0 must be added its albedo and atmospheric constituents, although their evolution is partly
a response to the astronomical situation (Figure 3.1). Table 3.1 summarizes the differ-
ences in the physical constants (parameters) of the three terrestrial planets and Jupiter.
11 Figure 3.2 illustrates the thermal conditions at the surface of each terrestrial planet and
Figure 3.1 The effective temperatures of the four inner planets in relation to distance from the
Sun. Separate lines are given for the black-body radiation temperature (b.b.r.) and for
different albedos (p). For Venus, Earth, and Mars the radiation temperature is the
lower cross and the mean surface temperature the upper dot. (After Hutter et al., 1990)
Table 3.1 Characteristics of the three terrestrial planets and Jupiter
Planet Distance from Mass (relative Gravitational Sidereal Rotational

sun (A.U.)a to earth = 1) acceleration (m s2) year (days) period (hr)
Venus 0.72 0.82 8.88 244.70 5,820.0b

Earth 1.00 1.00 9.78 365.26 24.0
Mars 1.52 0.11 3.73 687.00 24.7
Jupiter 5.20 318.00 23.20 11.86 (yr) 10.0
Sources: after Pollack and Yung (1980) and Wells (1986).
Notes:
a
Half the major axis of the ellipse, relative to the Earth = 1 Astronomical Unit (A.U.).
b
Clockwise rotation.
the respective state diagrams for water and carbon dioxide. Intercomparisons of planetary
meteorology are now becoming feasible through projects such as the Mars Pathfinder
(Scholefield et al., 1998). The rotation rate determines both day length and the Coriolis
parameter. The orbital parameters determine a planet’s annual cycle; the orbit itself
controls the length of the year, while the inclination of the planet’s axis to the plane of
the ecliptic and the longitude of the vernal equinox determine the length and character
of the seasons. The major gaseous constituents determine the specific heats (cp and cv )
of the atmosphere, whereas its absorbtivity is a result of the proportions of radiatively
active gases, especially H2O, CO2, and O3 (see Table 3.2). Other atmospheric compo-
nents, aerosols and clouds, together with surface properties, determine the radiation budget,
whereas differences in vegetation and hydrological properties affect the exchanges of heat,
moisture, and momentum, as well as of gases and aerosols.
The mean temperature of the Earth is determined by the balance of incoming solar
radiation and outgoing terrestrial radiation. Thus:
S (1 p) r2 = Te4 4r 2

Global climate and the general circulation 111
11
Figure 3.2 Conditions at the surfaces of the terrestrial planets and state diagrams for water and
carbon dioxide. T = the triple point. The range of pressures and temperatures reflects
the latitudinal and topographic ranges on the surface. (After Hutter et al., 1990)
0 Table 3.2 Atmospheric properties of planets
Planet Solar Planetary Surface Surface Dry Major Trace Aerosols

irradiancea albedo pressure temperatureb adiabatic gasesc gases
(W m2) (105 Pa) (K) lapse
(K km1)
Venus 2,630 0.77 90 730 (~230) 10.7 CO2 (0.96), H2O, SO2, H2SO4
N2 CO, HCl
Earth 1,368 0.30 1 287 (254) 9.8 N2 (0.77), H2O, CO2, Water,
O2 (0.21) O3, N2O H2SO4,
0 sea salts,
organics,
dust
Mars 590 0.15 0.007 222 (~215) 4.5 CO2 (0.95), O2, CO, Water ice,
N2 H2O dust, CO2
ice
Jupiter 51 – 100 ~129 H2 (0.89), HD, CH4, Ammonia
He NH3, C2H6 ice, water
Sources: modified after Pollack and Yung (1980) and Wells (1986).
Notes
a
0 Solar radiation flux per unit area at the top of the atmosphere.
b
Effective temperature in parentheses.
c
Numbers in parentheses are volume mixing ratios.
11
and
冤 S (1 p)
冥
1/4
Te =
4
For the appropriate values of solar irradiance (S 1368 W m2) received at the top of
the atmosphere at normal incidence on the Earth’s disc ( r 2) and planetary albedo (p
–
0.30), the equivalent mean Earth temperature (T e) for the planetary surface (4r 2) is about
255 K; this corresponds to a level in the real atmosphere of about 6 km or 500 – mb. A
one percent anomaly in the solar constant will give a 0.6 K anomaly – in (T e). A one
percent planetary albedo anomaly results in a 1.2 K anomaly in (T e). The variation in
solar distance due to the elliptical orbit of the Earth causes an annual variation of ±34
percent in the solar irradiance from 1.034 S at perihelion (currently January 4) to 0.965
S at aphelion (July 3). It is worth pointing out that the above calculation of the Earth’s
“effective” temperature includes the short-wave effect of cloud reflection in the planetary
albedo but neglects the trapping effect of clouds on infrared radiation, which decreases
the net outgoing total (Lindzen, 1994). For a waterless, cloud-free earth with a planetary
–
albedo of about 0.08, T e would be 273 K.
The earth’s orbit round the sun is perturbed by the gravitational effects of the sun and
other planets. There are three principal effects, as illustrated in Figure 3.3. The orbital
ellipticity, e, varies on long time scales (95,000 to 125,000 years and about 400,000 years),
currently being near its minimum (0.0167). Variations of e affect the annual radiation
income for the Earth by about ±0.2 percent. The tilt of the earth’s axis (or obliquity, )
oscillates with a period of about 41,000 years between 21.8° and 24.4°; currently it is
23°27′. A large tilt amplifies simultaneously the seasonal cycle in both polar regions, but
the effect is small in low latitudes. Due to a wobble in the earth’s axis of rotation, with
a 26,000 year period, combined with a precession of the orbital ellipse, the vernal equinox
precesses over a 22,000 year interval (Figure 3.4). However, this is modulated by eccen-
Table 3.3 Orbital forcings and characteristics
Element Index Present Average

range value periodicity (ka)
Obliquity of ecliptic () 22° to 23.4° 41
(tilt of axis of rotation) 24.5°
Effects equal in both hemispheres, effect intensifies
polewards (for caloric seasons)
Low High
Weak seasonality, Strong seasonality, more
steep poleward summer radiation at poles,
radiation gradient weaker radiation gradient
Precession of equinox () 0.05 to 0.0164 19, 23
(wobble of axis of rotation) 0.05
Changing Earth–Sun distance alters seasonal cycle
structure; complex effect, modulated by eccentricity
of orbit
Eccentricity of orbit (e) 0.0050 0.0167 410, 95
to
0.0607
Gives 0.02% variation in incoming radiation;
modifies amplitude of precession cycle changing
seasonal duration and intensity; effects opposite
in each hemisphere; greatest in low latitudes
11
Figure 3.3 The planetary forces on the Earth’s axis and orbit that cause changes in the eccen-
tricity (ellipticity) of the orbit (a), the tilt (obliquity) of the pole of rotation, and the
gyroscopic spin of the planet (or precession). (From Crowley and North, 1991)
0
tricity, splitting the precession frequency into periods of 19,000 and 23,000 years (Figure
3.5). About 10,000 years ago the perihelion occurred during the northern summer; this
enhanced seasonality in the northern hemisphere (Figure 3.4c). The precession effect is
greatest in low latitudes and is opposite between hemispheres. Its effect on the annual
temperature range and on the strength of the Asian monsoon circulation has been demon-
strated by Kutzbach and Otto-Bliesner (1982) and Kutzbach and Gallimore (1988). They
show that solar radiation increases of about 7–8 percent in July 11,000 years ago caused
a rise in summer temperature over the northern continents of 2–4°C; changes over the
0 oceans were small and thus the summer monsoon of South Asia and West Africa was
strengthened. Figure 3.5 illustrates these astronomical periodicities over the past 0.8
million years and their combined effect expressed as a climate index (Berger, 1978, 1979;
Imbrie et al., 1984). Table 3.3 summarizes the characteristics of the orbital forcings.
3.2 Basic controls of the atmospheric circulation and its

maintenance
The general characteristics of the atmospheric circulation are determined first by the
Earth–sun distance and orbital geometry, which account for the terrestrial receipt of solar
0 radiation. The observed latitudinal and vertical gradients of heating rate and temperature
are established by the distributions of solar radiation forcing and planetary factors, partic-
ularly atmospheric composition, cloud cover, and surface properties (albedo, vegetation).
11 The second most important determinants of the atmospheric motion are the planetary
Figure 3.4 The components of precession of the earth’s (a) axial precession (wobble), (b) effect
due to changes in eccentricity of the orbit (c) combined effect on the equinoxes in rela-
tion to the elliptical orbit. (From Pisias and Imbrie, 1986)
rotation rate and the Earth’s dimensions, which together determine the angular momentum
of the Earth–atmosphere system. Table 3.2 summarizes key properties of the Earth and
other terrestrial planets; the significance of some of these differences is discussed below.
3.2.1 Energy balance

The energy balance of the global Earth–atmosphere system involves the balance of
incoming and outgoing solar and terrestrial radiation at the top of the atmosphere and at
the Earth’s surface, as well as the horizontal and vertical transfers of sensible and latent
heat. Details of the radiation laws and radiative transfer are not discussed here, but may
be found in various standard sources, including Houghton (1985) and Peixoto and Oort
(1992, chapter 6), for example.
A summary of the global radiation budget is given in Figure 3.6. The incoming solar
radiation is almost all in the wavelength range 0.15–5.0 m, comprising 9 percent
11
Figure 3.5 Variations in eccentricity, obliquity (tilt) in units of degrees angle, and precession index
(e sin) over the last 800,000 years (after Berger, 1978). The normalized sum of these
0 terms is the curve labeled ETP (units of standard deviation). The calculated variance
spectra and dominant periods (k years) are shown at the right. (After Imbrie et al., 1984)
Figure 3.6 The disposition of solar radiation entering the atmosphere and infrared radiation emitted
11 by the atmosphere and surface(W m–2). (From Kiehl and Trenberth, 1997)
ultraviolet ( < 0.4 m), 49 percent visible (0.4–0.8 m) and 42 percent infrared radia-
tion (> 0.8 m). The peak is around 0.475 m, corresponding to a solar black body
temperature of 6,100 K according to Wien’s law.1 The solar “constant” calculated for
mean solar distance (150 106 km) is approximately 1368 W m2. It refers to the radi-
ation received on a surface normal to the solar beam at the top of the atmosphere (TOA).
Averaged annually over the earth, the TOA amount is 1368 (r2/4r2), or 342 W m2.
The solar radiation entering the atmosphere is absorbed by radiatively active gases (stratos-
pheric ozone <0.3 m and water vapor and aerosols in bands between 1.0 m and 3.0
m), as well as by dust and smoke particle aerosols (sixteen units). There is also some
absorption by cloud droplets and ice crystals; recent studies indicate that this may be
larger than previously thought.
Several authors propose that the modeled absorption of solar radiation by clouds may be
underestimated (Cess et al., 1995; Collins, 1998), while others present evidence that the
calculated absorption in a cloud-free atmosphere may also be too low (Wild et al., 1995;
Arking, 1996; Li et al., 1997). For sites in Germany, the surface solar radiation calculated
by the ECMWF – University of Hamburg (ECHAM)4 general circulation model for cloud-
free conditions agrees closely with high-quality measurements at ground stations (Wild and
Ohmura, 1999). Relative to other radiation schemes, the ECHAM4 shows greater absorp-
tion of solar radiation in the near infrared. Wild and Ohmura suggest that about 21 percent
(72 W m2) of the solar radiation incident at the top of the atmosphere is absorbed in a
cloud-free atmosphere. They also estimate an all-sky absorption of 90 W m2.
Solar radiation is also scattered forward and backward by air molecules of small diam-
eter compared with the wavelengths of the radiation (Rayleigh scattering is proportional
to 4), by larger diameter particles, cloud drops, and ice crystals (Mie scattering approx-
imates 1 dependence). Total backscatter by the atmosphere contributes about four units
to the reflected solar radiation, but reflection of the incident radiation by clouds and the
Earth’s surface is more significant (Figure 3.6). The wavelength-integrated reflectivity, or
albedo, of clouds depends on their fractional coverage, thickness, liquid water content,
and droplet radius, as well as on the solar zenith angle. Thick stratiform or cumuliform
cloud layers have albedos of 0.60–0.70, whereas cirrus or thin stratus clouds have albedos
of 0.20–0.30. Average global cloudiness is approximately 62 percent. Hence cloud
reflection accounts for about twenty units or two-thirds of the mean planetary albedo
(0.30). The earth’s surface has an average albedo of 0.16 (Ohmura and Gilgen, 1993),
relative to the radiation incident at its surface. The ocean surface and terrestrial vegeta-
tion effects are dominant. Consequently, the contribution to the planetary albedo is modest,
only six units. The net solar radiation absorbed by the Earth’s surface is between about
142 W m2 and 170 W m2 (Ohmura and Gilgen, 1993; Li and Leighton, 1993; Kiehl
and Trenberth, 1997) or forty-two to fifty units.
The terrestrial surface has a mean global temperature of about 288 K, giving rise to a
peak emission of infrared radiation near 10 m.1 The emissivity of the surface in these
wavelengths ranges from about 0.92 for soil and rock, to 0.97 for water and 0.98 for
vegetation. The radiation emitted by the Earth’s surface is predominantly absorbed by the
atmospheric greenhouse gases: water vapor (4.5–8.0 m and around 20 m), carbon
dioxide (4 m and 13–17 m) and ozone (9.6 m). This energy is re-radiated back to
the surface and upward to the overlying layers of the atmosphere (Figure 3.6). Hence the
net surface emission is small; Ohmura and Gilgen (1993) estimate a mean annual value
of 40 W m2. However, in the “window regions” of the spectrum, between about 8 m
and 12 m, most of the radiation escapes directly to space (twelve units). A majority of
the outgoing long-wave radiation is emitted by the greenhouse gases, with an additional
contribution from cloud emission. The greenhouse analogy refers to the atmosphere’s
transmittance of most incoming short-wave radiation and absorption of most outgoing
long-wave radiation, although in an actual greenhouse heat is also trapped by the glass
barrier and air movement is restricted. In the Earth–atmosphere system, the Earth’s surface
11 heats the atmosphere by turbulent transfer of sensible heat (enthalpy) (six units) and also
transfers moisture to the atmosphere through evaporation. This provides latent heat of
vaporization when condensation takes place (twenty-four units), and this heat is eventually
lost to space by radiation. The mean surface net radiation for the globe is approximately
100 W m2, or 30 percent of the extraterrestrial radiation.
It is important to note that the atmospheric absorption of solar radiation contributes to
heating of the troposphere by about 0.5 K/day, whereas it is cooled about 1.5–2.0 K/day
by long-wave radiation. Turbulent heat transfer offsets this radiative imbalance. The rela-
tive contributions of sensible and latent heat clearly vary widely from dry desert surfaces
0 to irrigated areas or swamps and the oceans. The ratio of sensible heat to latent heat flux
from a surface is termed the Bowen ratio ( ), which averages about 0.3 globally; it is
less than unity for wet surfaces and ! 1 for dry surfaces. The Bowen ratio is an important
index of the partitioning of the turbulent fluxes of energy from the surface to the atmos-
phere and it varies on a range of time scales from subdaily to annual. Details of these
transfers of radiation and energy for different surfaces may be found in Oke (1987) and
other sources.
The atmosphere is often likened to a heat engine, but it appears to be a remarkably
inefficient one. As we have seen, averaged over the Earth, about 342 W m2 of solar
radiation in the 0.3–3.5 m wavelength range enters the atmosphere. Thirty percent of
0 this incident radiation is reflected to space, mainly by clouds (Figure 3.6) and does not
take further part in atmospheric processes. Twenty percent is absorbed in the atmosphere,
primarily by water vapor, and 50 percent is absorbed by the surface. This absorbed energy
heats the surface and the atmosphere, which then re-emit infrared radiation (3–50 m
wavelength) corresponding to a much lower temperature (255 K for the effective plane-
tary temperature and 288 K for the mean surface temperature as a result of the greenhouse
effect) compared with the sun (6,000 K). Although a large amount of radiation is emitted
by the Earth’s surface, most of it is absorbed by the atmosphere and reradiated, thereby
returning much of it to the surface. The heating of the atmosphere by absorbed solar radi-
ation, reabsorption of infrared radiation emitted by the surface, and by turbulent heat
0 fluxes from the surface, determine its internal and potential energy. This is the order of
1024 J. Only a minute fraction of this (~1020 J) is converted into kinetic energy, which
maintains the circulation of the atmosphere and oceans against friction. Frictional dissi-
pation is of the order of 2 W m2, or 0.6 percent of the absorbed solar radiation. However,
this determination of atmospheric energy efficiency is misleading, because a large fraction
of the total potential energy is not in fact available for conversion. This question is exam-
ined further below (p.140).
It might be anticipated that the Earth’s energy balance would undergo seasonal varia-
tions as a result of changes in orbital geometry and solar declination angle. In actuality
its variations are quite small when integrated over the globe. Table 3.4 shows the average
0 monthly TOA values for February 1985-March 1989 based on the Earth Radiation Budget
Sensor (ERBS) on NOAA-9 and 10. The annual net value of 5 W m2 indicates the
residual uncertainty in the budget components. The annual cycle in solar irradiance is a
result of the timing of perihelion, while that in the long-wave emission suggests the role
of northern summer heating of the continents. The planetary albedo is highest in the boreal
winter associated with the combined influence of ocean cloudiness along the storm tracks
and snow cover on the northern continents. The reason for the minima in the transition
seasons is less obvious, although radiation receipts are then large over low latitudes, while
snow and ice cover in the northern hemisphere is least extensive in September and
Antarctic ice is at its minimum in March. The relative seasonal contributions of cloudi-
0 ness and surface albedo still need to be determined more accurately.
The latitudinal distribution of the net radiation (solar minus infrared radiation) in the
Earth–atmosphere system and its seasonal variations are vital elements in maintaining the
11 atmospheric and oceanic circulations. There is an energy surplus in lower latitudes and
Table 3.4 Radiation balance (W m2) at the top of the atmosphere based on ERBS data for
February 1985–May 1989
J F M A M J J A S O N D Annual
Solar 353 351 346 340 335 332 331 333 338 344 350 353 342
irradiance
Reflected 108 104 100 100 102 99 99 96 96 101 108 110 102
solar
radiation
Long-wave 233 232 233 234 236 238 239 240 238 235 233 232 235
emission
Net radiation 12 14 13 6 3 6 7 3 4 8 8 11 5
Planetary 0.307 0.296 0.290 0.294 0.305 0.299 0.300 0.289 0.285 0.294 0.310 0.311 0.298
albedo
Source: courtesy E.F. Harrison.
Table 3.5 Mean temperatures for each latitude; reduced to sea level, except for south pole
(2,835 m) (°C)
Latitude Air temperature Sea surface

January July Year temperature
90°N 32 1 20 1.7

80° 27.9 1.5 17 1.7
70° 22.6 7.3 9.8 0.7
60° 16.1 14.1 1.1 4.8
50° 7.1 18.1 5.8 7.9
40° 5.0 24.0 14.1 14.1
30° 14.5 27.3 20.4 21.3
20° 21.8 28.0 25.3 25.4
10° 25.8 26.9 26.7 27.2
Equator 26.4 25.6 26.2 27.1
10° 26.3 23.9 25.3 25.8
20° 25.4 20.0 22.9 24.0
30° 21.9 14.7 18.4 19.5
40° 15.6 9.0 11.9 13.3
50° 8.0 4.0 5.0 6.5
60° 2.5 5 2 0.5
70° 2 20 13 1.3
90°S [27] [59] [49]
Source: modified after Lamb (1977).
a deficit in higher latitudes with, for the annual average, a balance at about 35° latitude,
necessitating poleward heat flows in order to maintain the observed latitudinal gradient
of temperature (Table 3.5 and Figure 3.7). General Circulation Model (GCM) experiments
by Hunt (1979) indicate that the equator–pole temperature gradient decreases as the rota-
tion rate increases. Similarly, Kuhn et al. (1988) calculate a 15 K reduction in the gradient
with a rotation rate 1.6 present around 3.5 billion years ago. The reduction is mostly
determined by the higher polar temperatures associated with enhanced sensible heat flux.
Stone (1972) shows that this flux is related to rotation rate as a function of (
/
y)2/f 2,
where
/
y is the local meridional gradient of potential temperature and ƒ is the Coriolis
parameter (2 " sin , where " the Earth’s angular velocity and latitude).
11
Figure 3.7 Mean zonal temperatures adjusted to sea level for January and July. (Based on Lamb,
1977, II, p. 560; Rigor et al., 2000) The inset shows the land fraction of the global
surface. (Based on Lamb, 1972, I, p. 479)
0
Latitudinal temperature gradient
The observed latitudinal gradient of annual mean surface air temperature is 46.7°C for
10°N to 90°N, while the corresponding value for sea surface temperature (SST) is only
29°C (Table 3.5). In January there is an equator–north pole gradient of 58.4°C that
decreases by half in July between 20°N and 90°N. The effects of latitudinal temperature
gradients on global climate have been analysed by GCM experiments. Rind (1998) exam-
ines cases where the latitudinal gradient of zonal mean surface air temperature between
the equator and north pole is varied by ±3.6°C from a control case value of 45.9°C by
0 variously raising and lowering low and high-latitude SSTs. For an increased latitudinal
gradient there is an increase in global average surface winds, as expected, as well as
intensified subtropical jetstreams. The mean meridional Hadley cell in low latitudes is
also strengthened, giving drier conditions in the subtropics. However, increased polar
energy transport by the atmosphere appears to be compensated for by weaker ocean trans-
ports. The tropospheric lapse rate is reduced by about 10 percent for a 7.2°C reduction
in gradient. Global average integrated water vapor is 25 mm for increased gradient, versus
22.9 mm for the control and 21.3 mm for decreased gradient, illustrating a non-linear
asymmetric response to the forcing.
Paleoclimatic reconstructions indicate that the latitudinal distribution of zonally averaged
0 surface temperature varies considerably over geological time. Budyko and Izrael (1991,
tables 8.3–4) estimate that the global mean surface temperature differed from present
(TG) by 3.6ºC for the Pliocene maximum (4–5 Ma) and by 2.0°C for the Eemian
11 Last Interglacial maximum (125 ka). They also suggest that the latitudinal temperature
Figure 3.8 Proposed universal relationship for the variation of global and latitudinal surface temper-
ature change. (Based on Budyko and Izrael, 1991; Lindzen, 1994)
differences (T) expressed as a function of (TG), (i.e. T/TG) display a consistent

relationship with latitude (sin ) (Figure 3.8). The scaling of changes in the latitudinal tem-
perature gradient by the global mean change, implied by Figure 3.8, raises important ques-
tions (Lindzen, 1994):
1 What keeps changes in equatorial temperatures small? A large (small) latitudinal

gradient should increase (reduce) poleward heat transport, implying strong (weak)
thermal forcing in low latitudes.
2 Changes in the zonal temperature gradient and associated changes in poleward heat
flux should produce temperature variations in low/high latitudes that are out of phase.
Observed twentieth-century variations suggest such contrasts, although they are not
apparent on long time scales.
At the present, the required poleward horizontal transport of energy in middle latitude
is approximately 6 1015 W (Trenberth and Solomon, 1994). This takes place in both
the atmosphere (about two-thirds) and the oceans (about one-third) (Figure 3.9). Model
calculations for glacial maximum conditions suggest that there may have been a 70 percent
decrease in poleward transport in the North Atlantic (Miller and Russell, 1989). Energy
balance considerations help provide constraints on model calculations of oceanic and
atmospheric heat transports implied by paleotemperature reconstructions (Horrell, 1990).
Vertical temperature structure

During the eighteenth century there was considerable controversy over the cause of the
observed decrease of temperature with height on mountains (Barry, 1978). More surprising
was the discovery from balloon measurements in the 1890s (Teisserenc de Bort, 1902),
independently confirmed by R. Assmann, that an isothermal layer or inversion around
10–12 km capped the troposphere. Hoinka (1997) recounts the history of this unexpected
finding and the surrounding debate.
The vertical structure of the atmosphere is most clearly demarcated by the vertical
temperature distribution. Four distinct layers are separated by levels where the rate of
11
0
Figure 3.9 Zonal mean distribution of northward energy transport by the oceans and atmosphere
(1015 W). (From Sohn and Smith, 1993)
temperature change with height, or lapse rate, undergoes a rapid shift overlain by a
reversal. The layers are from the Earth’s surface upward, and their altitudes, in mid-
latitudes, are: the troposphere, separated from the stratosphere by the tropopause around
12 km (200 mb); the stratosphere, separated from the mesosphere by the stratopause
around 50 km (1 mb); and the mesosphere, separated from the thermosphere by the
mesopause around 80 km (0.01 mb). The terms troposphere and stratosphere were coined
0 by Teisserenc de Bort in 1908. The word tropopause was introduced in Great Britain
during World War I and its usage was established by Sir Napier Shaw. The stratopause
and mesopause were so named only in 1962 (Sawyer, 1963). It is interesting to note that
neither Mars nor Venus has a warm stratopause due to their gaseous composition (see
Table 3.2).
The troposphere represents 80 percent of the atmospheric mass and contains almost all
of the water vapor and clouds. Weather processes are primarily active in the troposphere.
The mean environmental lapse rate in the troposphere is about 6.0 K km 1. This rate
is determined partly by the radiative equilibrium. This equilibrium involves the absorp-
tion of incoming solar radiation and of upwelling infrared radiation from the surface and
0 underlying atmosphere, on the one hand, and the emission of infrared radiation by succes-
sive overlying atmospheric layers, on the other. The details of this absorption and emission
are controlled by the vertical distribution of the principal radiatively active gases – water
vapor, carbon dioxide and ozone. Water vapor is concentrated below 300 mb, and ozone
is in the stratosphere. Radiative equilibrium is modified, however, by convective over-
turning towards a statically neutral state. A radiative–convective model calculation by
Manabe and Strickler (1964) of the temperature profile produced by the absorption and
emission of these gases in combination, in the absence of clouds, and with a convective
adjustment, reproduces the main features of the temperature structure up to about 40 km.
As noted earlier, the mean global surface temperature is 288 K whereas the mean effec-
0 tive planetary temperature is 255 K. The latter corresponds to the temperature at the level,
where there is approximately one half of the atmospheric mass, or 5.5 km. From these
values, the mean lapse rate is 6.0 K km1. The simple radiative–convective model of
11 vertical temperature structure is deficient in the extratropics, where the lapse rate is
typically 5 K km1. A further possible explanation invokes “baroclinic adjustment,” which
implies that the atmosphere is made baroclinically unstable by large-scale forcing and
that the baroclinic eddies maintain a state close to neutral stability. Like convective adjust-
ment, this represents a dynamical as opposed to a radiative constraint on the tropospheric
lapse rate.
The standard World Meteorological Organisation definition of the tropopause is based
on a lapse rate decreasing to ≤ 2 K km1, and not exceeding 2 K km1 through a
2 km deep layer. Based on such a definition, the tropopause is located at 17–18 km in
low latitudes, 12 km in mid-latitudes, and 9 km in high latitudes. The change in altitude
occurs in steps in association with the upper-level polar and subtropical frontal zones, or
there may exist tropopause folds in these zones (Defant and Taba, 1957; Reed and
Danielsen, 1959). There are also seasonal changes in the tropopause height. Over the
central Arctic Ocean the tropopause is lowest in April (between 8.5 km and 11 km) and
highest in August (9.5–11.5 km), with a secondary minimum in November and a secondary
maximum in December–January, based on north pole drifting station soundings for
1954–91 (Nagurny, 1998).
In addition to the discontinuity in lapse rate at the tropopause, it acts as a lid on vertical
motion in the troposphere. This can be understood by noting that the Brunt–Väisälä
frequency, N S1/2/2, where S is the static stability ( ), increases in magnitude
across the tropopause from 102 rad s1 to ≥ 2 102 rad s1. In addition, only long
Rossby waves, with typical zonal wave numbers ≤ 2, are able to propagate vertically into
the stratosphere. Consequently, vertical motions are small above the tropopause.
In the tropics, the conventional lapse rate tropopause definition is often not relevant
(Highwood and Hopkins, 1998). Other definitions include the top of the layer with con-
vective heating, but this is generally below the conventional tropopause. Another definition
is the altitude of temperature minimum which reaches 80°C near the equator; this is reli-
able only in the deep tropics, where the lower stratosphere is not isothermal. The temper-
ature minimum may otherwise occur above the conventional lapse rate tropopause. Using
European Centre for Medium Range Weather Forecasts (ECMWF) data for nearly four
years, Highwood and Hopkins show that the main level of convective outflow is around
12 km (200 mb), with the top of such outflow at 14 km (145 mb), where 360 K; the
standard tropopause is at 16 km (100 mb) and 375 K; the temperature minimum is at
18 km (80 mb) and 400 K. A stabilization level is identified around 130–40 mb.
The lapse rate definition has disadvantages because it is non-physical and the analyzed
surface often shows erratic displacements associated with synoptic disturbances. A
different approach makes use of a potential vorticity (PV) surface for global analysis. PV
is a conservative property, and a PV surface represents a material surface (a closed surface
that moves with the flow) on a time scale of days. In a frictionless, adiabatic atmosphere,
PV is conserved for 2-D motion on an isentropic surface (Hoskins, 1991). In the tropics,
PV surfaces are almost vertical as a result of the marked equatorward increase of absolute
vorticity. Poleward of about 25° latitude, however, the tropopause can be delineated by
a PV surface corresponding to about two PV units. Hoerling et al. (1991) suggest that
outside the tropics the tropopause may be delineated by potential vorticity values within
the range 1 to 3.5 106 K m2 kg1 s1 (1–3.5 PV units). Above the extratropical
tropopause, PV magnitudes in the lower stratosphere are around four PV units and they
increase rapidly upward and poleward. For this reason, the actual PV value selected to
define the tropopause is arbitrary.
The Ertel potential vorticity on an isentropic () level (see Appendix 3.1) may be calcu-
lated from:
(PV) g ( f ) (
p/
)
where is the vertical component of the relative vorticity on an isentropic surface. The
term:
11 1

g
p
corresponds to the density in coordinates. The absolute vorticity term is:
( f ) =
1
冢冣

M
r 2 cos

where the angular momentum M ua cos "r 2 cos2 , r the earth’s radius,
0 latitude. Differences in tropopause height between the lapse rate definition and a PV defi-
nition can occur, particularly in association with frontal zones, but the large-scale features
of the time-mean tropopause determined by the two procedures show reasonable overall
agreement.
The existence of the tropopause is not well understood. Radiative heating of the Earth’s
surface and the atmosphere produces low static stability near the ground and high values
in the stratosphere, where ozone absorption gives large heating rates. However, the tran-
sition between these is sharp, not gradual. In low latitudes the primary factor appears to
be deep convection, which heats the troposphere by releasing latent heat. The tropos-
pheric lapse rate approximates neutral stability for a saturated atmosphere. The tropopause
0 is close to the level of the high cloud tops, and, because of its height, temperatures are
as low as 215 K.
A composite mapping of the global tropopause, in terms of pressure level, is provided
by Hoinka (1998), using lapse rate and PV criteria. He finds that a consistent spatial
pattern is obtained using a lapse rate criterion equatorward of 19°N, combined lapse rate
and PV criteria for 19°–36°N, and a 3.5 PV units definition poleward of 36°N. The
southern hemisphere tropopause is pre-eminently zonal, whereas in the northern hemi-
sphere there is a planetary two to four wave pattern. In the Arctic the tropopause heights
are circumpolar in summer, but in winter the maximum pressure (around 300 mb) is
displaced to 70°N along 90°W. In the Antarctic the maximum pressures are near 75°S,
0 180° longitude in both summer and winter.
Potential temperature surfaces where is between about 300 K and 340 K are in the
troposphere in low latitudes, but in the lower stratosphere at high latitudes. Potential
vorticity analyses on such isentropic surfaces show that the extratropical tropopause can
develop troughs and ridges and cut-off features through the action of deep baroclinic
disturbances. These convolutions transport high (low) PV air horizontally equatorward
(poleward), respectively. This process strips high PV air from the edge of the stratos-
pheric vortex, sharpening the transition to the lower PV air in the subtropical upper
troposphere (James, 1994). Such “vortex stripping” thereby sharpens the tropopause.
Ambaum (1997) considers that the extratropical tropopause is established by a dynamic
0 equilibrium between diabatic heating and vortex stripping. He also suggests that, in mid-
latitudes, orography plays a role in enhancing meridional PV gradients. In high latitudes,
vertical motion is generally weak in the troposphere and in winter subsidence tends to
balance radiative heat loss. Thus the polar tropopause is usually not well developed.
The sensitivity of the height of the tropopause to external parameters can be examined
using a general circulation model that incorporates the various radiative and dynamical
processes that affect the tropospheric lapse rate. Thuburn and Craig (1997) use the UK
Universities’ Global Atmospheric Modelling Programme (UGAMP) GCM with thirty-
three vertical levels in a January mode for this purpose. They modify surface temperature,
stratospheric ozone and planetary rotation rate. The results show that tropopause height
0 depends strongly on surface temperature acting through changes in the moisture distrib-
ution and its radiative effects. It is less sensitive to ozone distribution and is little affected
by 40 percent changes in the earth’s rotation. Further tests of two types of baroclinic
11 adjustment suggest no apparent relationship with tropopause height.
3.2.2 Angular momentum balance
The total angular momentum of the rotating Earth–atmosphere–ocean system is virtually
constant in time, apart from minor effects due to tidal friction and a gradual secular
decrease caused by the gravitational effects of the moon and planets. For the Earth, the
angular momentum is 5.86 1033 kg m2 s1, compared with about 1 1028 kg m2 s1
for the atmosphere, assumed to be in solid rotation with the Earth (Peixoto and Oort,
1992, chapter 11). The absolute angular momentum of a unit of atmospheric mass about
the Earth’s axis of rotation (M) comprises a component representing the atmosphere in
solid rotation (M") and the angular momentum relative to the rotating earth (MR ) (see
Figure 3.10). For an atmosphere in rotation the absolute angular momentum per unit mass
about the Earth’s axis of rotation is:
M = M" MR
where
M" = " r2 cos2
and
MR = u r cos2
where " the Earth’s angular velocity (7.292 105 rad s1), r the Earth’s radius
(6,371 106 m), = latitude angle, and u zonal wind speed.
Table 3.6 illustrates the theoretical effect of a poleward displacement on the zonal
velocity of an air particle initially at rest for each 10° latitude displacement poleward. In
the upper troposphere, at jetstream levels, a rapid increase of zonal velocity is observed,
on average, poleward to about 30° latitude, implying that absolute angular momentum is
Figure 3.10 Schematic illustration of absolute (M ) and relative angular momentum (Mr ). (Peixoto
and Oort, 1992)
11 Table 3.6 Earth’s rotational velocity and its consequences
Latitude Circumference Rotational Increment of westerly Coriolis

(km) velocity (m s1) motion for stationary air parameterb (f)
transferred 10° poleward (104 s1)
(m s1)a
90° 0 0 81 1.458
80° 6,950 81 77 1.436
70° 13,680 158 74 1.370
0 60° 20,000 232 68 1.263
50° 25,700 300 57 1.117
40° 30,600 357 46 0.937
30° 34,600 403 34 0.0729
20° 37,600 437 21 0.499
10° 39,400 458 7 0.253
0° 40,000 465 0
Source: Lamb (1972, p. 486).
Notes
a
Values refer to intermediate 10° latitude zones.
b
Positive in the northern hemisphere, negative in the southern hemisphere, where f = 2" sin .
0
conserved. Further poleward and at lower levels, however, this is clearly not the case.
Consequently, a poleward gradient of absolute angular momentum exists.
The mean relative angular momentum is of the order of 14 1025 kg m2 s1 on an
annual basis, or 1 percent of M". There is a large annual cycle in M associated with
seasonal variations in wind velocity. In the southern hemisphere there is about a 40 percent
decrease in M from winter to summer, while in the northern hemisphere the summer value
is only 15 percent of that in winter as a result of the weak summer westerlies (Rosen
and Salstein, 1983). The seasonal variations in atmospheric angular momentum, of the
0 order of 5 1025 kg m2 s1, cause the Earth’s rotation rate and angular momentum to
vary also. The rotation rate is greater in July than in January, with a corresponding change
in length of day (LOD). Rosen et al. (1987) indicate that:
LOD (milliseconds) 0.168 Mr (1025 kg m2 s1)
so that the LOD is about 0.8 milliseconds longer in January than in July.
The time rate of change of atmospheric total angular momentum (M) is essentially deter-
mined by pressure and friction torques in the zonal (west–east) direction. For a unit volume,
dM 1
p
= F r cos
0 dt

where F frictional stress in the zonal direction; = longitude.

The pressure and friction torques are considered positive where they increase the west-
erly (eastward) angular momentum. Mountain barriers generate a pressure torque by
east–west pressure differences across them. Calculations suggest that the mean annual
net effect for the major mountain ranges at 40°–50°N is of the order of 2 1018 kg
m2 s2, comprising about 40 percent of the total surface torque due to friction and moun-
tains of 5 1018 kg m2 s2.2
Frictional stress at the Earth’s surface transfers westerly angular momentum from the
0 atmosphere to the Earth in the zones of surface westerlies and, conversely, the atmos-
phere gains relative momentum in the regions of easterlies, where the air rotates more
slowly about the axis of rotation than the Earth. Overall the net torques of westerlies and
11 easterlies must balance in area in order for angular momentum to be conserved (Lorenz,
1967). The observed distribution of wind belts qualitatively supports this concept. More-
over the tropics represent a source of absolute angular momentum in the atmosphere and
the mid-latitudes a sink, with poleward transport across the subtropics.
For the total Earth–atmosphere system, the angular momentum remains constant with
time (the principle of the conservation of absolute angular momentum). Zonal westerly
(easterly) surface winds will add (extract) angular momentum to (from) the earth through
friction. Overall, therefore, the net torques of westerly and easterly winds must balance
as a consequence of the atmospheric general circulation.
The low-level easterlies in low latitudes acquire westerly relative angular momentum,
whereas the mid-latitude westerlies transfer their momentum to the surface. Calculations
suggest that the westerlies would cease in about ten days if their supply of angular
momentum were cut off. How is this supply maintained?
The transport of angular momentum from the tropics to mid-latitudes is accomplished
by the meridional transport of relative angular momentum (Mr). The contribution to the
total angular momentum due to solid rotation with the Earth is very large because
the angular velocity ("r cos ) is 465 m s1 at the equator, 328 m s1 at 45° latitude
and 232 m s1 at 60° latitude, whereas typical wind speeds are an order of magnitude
smaller. However, mass conservation dictates that, in the mean, there is no meridional
transport of the " component of angular momentum.
The specific mechanisms of horizontal transport can be calculated by analyzing the
motion using the concept developed for fluid flow by Osborne Reynolds (1884). The total
zonal flow past a point, u, can be regarded as comprising the time-mean flow, u– , and a
deviation, u′, from the time average:
u u– u′
We can also consider departures with respect to space averages:
u [u] u*
where [ ] denotes an average along a latitude circle and * a spatial deviation from this
average. Thus considering joint time and space averages:
u [u–] u– * [u]′ u*′
where [u–] zonal winds averaged along a latitude circle and over a time interval, u– *
local departure from the longitudinal average, averaged in time, [u]′ time departures
from the longitudinal average, and u*′ instantaneous local departures from the time
and longitudinal averages. It follows that:
u– [u–] u– * and v– [v–] v–*
where v– is the time-mean meridional flow component. By convention, westerly zonal
winds are southerly, meridional winds are positive. By expansion, the mean poleward
transport of relative angular momentum per unit mass as one level across a latitude
circle is:
[uv] = [u ][v ] [u]′[v]′ [u*v*] [u*′v*′] (1)

where the terms on the right refer, respectively, to transports by standing (meridional)
cells, transient cells, standing (zonal) eddies, and transient eddies.
When the analysis is extended to the vertical dimension the terms corresponding to those
in equation 1 represent, respectively, mean and instantaneous cell circulation in the verti-
cal plane along a meridian, and mean and transient circulations in the horizontal plane
across a latitude circle (see Grotjahn, 1993, pp. 386–90; Peixoto and Oort, 1992, pp. 61–5).
In the literature, two different schemes may be encountered. These are compared by
Starr and White (1952):
11 关uv兴 = 关u兴关v兴 [u]′[v]′ 关u*v*兴 (2)
and
关uv兴 = 关u兴关v兴关u*v*兴关 u′v′ 兴 (3)
Note the difference in the order of averaging indicated by the bar over, or within, the
brackets.
The time and space-averaged products are identical:
0 [uv] = 关 uv 兴
since space averaging and time averaging are commutative. Similarly:
[u] [v] = 关u兴关v兴
The other equivalent terms in the two equations are not equal, however. [u]′[v]′ is due to
time variations in the meridional flow, while [u– *v– *] is due to the asymmetry of the mean
–—–
streamlines, i.e. a stationary wave; both terms appear in equation 1 above. [u*v*] is asso-
—
ciated with the asymmetry of instantaneous streamlines and [u′v ′] corresponds to time vari-
0 ations of wind at a given point. These two terms are combined in the transient eddy term
in equation 1.
The horizontal poleward transport of relative (westerly) angular momentum is accom-
plished primarily through the asymmetry of the upper-air streamline patterns. This
mechanism, first proposed by Starr (1948), is illustrated schematically for both hemi-
spheres in Figure 3.11. The annual vertically and zonally averaged poleward momentum
—–
transport [uv] is of the order of 15 m2 s2 near 35°S and 12 m2 s2 at 30°N (corresponding
to angular momentum transports of 26 1018 and 23 1018 kg m2 s2, respectively2 and
there is a larger seasonal contrast in the northern than in the southern hemisphere (Oort
and Peixoto, 1983). Antarctica is also a significant source of equatorward momentum flux.
0 Nearly all of the poleward transport is accomplished by transient eddies [u— ′v ′].. However,
in the northern winter there is a significant flux contribution also from stationary eddies
[u– *v– *] directed poleward at 30°N and equatorward at 60°–70°N (Figure 3.12) (Peixoto
and Oort, 1992). There are also modest contributions from the mean meridional cells of
—–
the Hadley circulations [uv], primarily in the respective winter hemisphere. Vertical cross-
sections of zonally averaged momentum flux show that the transports take place
predominantly in the upper troposphere, around 200 mb at 20°–30°N and about 300 mb
at 30°–35°S, although the mean meridional cells are most significant below 850 mb. To
complete the picture, the vertical transports of momentum (involving frictional stress and
mountain pressure differences) need to be considered. In the annual mean, there are sources
0 in the easterlies of low latitudes, symmetric about the equator, with sinks in the mid-lati-
tude westerlies. This reflects the upward and downward flows of mass in the thermally
direct-meridional Hadley circulation, where warm air rises and cooler air sinks, and a
much weaker (indirect) Ferrel cell in mid-latitudes (discussed further in section 3.3). The
Hadley cells are much stronger in the respective winter hemispheres, with some cross-
equatorward momentum transport from the source regions, southward (northward) in the
boreal (austral) winter (Figure 3.13).
In summary, westerly momentum acquired in the tropical source regions is transferred
upward within the Hadley cells via the w"r2 cos2 term, poleward mainly by large-scale
horizontal eddies in the upper troposphere, and then downward by the sinking arm of the
0 Hadley cells in the subtropics. The high-latitude direct cells associated with the subpolar
easterlies are an additional minor source of westerly momentum for mid-latitudes.
Completion of the cycle of angular momentum requires that a return of momentum
11 must take place from middle to low latitudes either within the oceans or within the litho-
Figure 3.11 Schematic model of the net poleward transport of westerly angular momentum by hori-
zontal eddies in mid-latitudes. The streamlines have a predominant southwest–northeast
(southeast–northwest) tilt in the northern (southern) hemisphere. (After V.P. Starr, from
Peixoto and Oort, 1992)
sphere. Oort (1985, 1989) shows that the oceans largely transport angular momentum
zonally, not meridionally. He suggests that west–east gradients of sea level, associated
with the tropical easterly winds piling up water on the east coasts of Asia and the Americas,
may generate westward torques on the continents analogous to atmospheric pressure
torques across mountains. Correspondingly, the mid-latitude westerlies will create an east-
ward continental torque. The return flow of momentum may involve stress release in the
crust via a preferred tilting in fault displacements.
The total atmospheric mass is 5.1361 1018 kg with a seasonal variation of ±0.0010
(maximum: July, minimum: January) due to the annual cycle of atmospheric moisture
content (specific humidity q ~ 1.0 1015 kg) (Trenberth et al., 1987). Mean surface
pressure (ps ) based on ECMWF data for December 1978–December 1985 is 981.9 mb in
the northern hemisphere (where the mean elevation is 284 m). The effect of orography
alone is estimated to cause surface pressure values 1.1 mb lower than otherwise in the
northern hemisphere and 0.5 mb higher in the southern hemisphere. Water vapor
contributes 2.7 mb of total surface pressure. Seasonal heating and cooling cause inter-
hemispheric mass exchanges. The maximum interhemispheric transfer is from the northern
to the southern hemisphere in boreal spring, with the reverse in austral spring. Water
vapor is transported from the winter to the summer hemisphere, mainly in the monsoon
regions (Chen et al., 1996). The mass deficit in a particular hemisphere is accompanied
by a 5 percent increase in that hemisphere’s angular momentum, implying that the mass
change is a dynamic response rather than a mechanical tendency to mass conservation
(Christy et al., 1989). This may reflect the jetstream strength and the locations of storm
tracks. The primary interhemispheric mode for anomalies of (P cos ) describes exchanges
between northern mid-latitudes and the entire tropics plus southern subtropics with south-
11
Figure 3.12 Zonal-mean annually averaged northward flux of momentum in the atmosphere: (a)
total, (b) transient eddies, (c) stationary eddies, (d) mean meridional cell (m2 s2)
(for units see note2) (From Oort and Peixoto, 1983)
0 ward propagation of anomalies. During the seven-year record (1978–85) Christy et al.
identify eighteen long-lasting, extreme surface pressure anomalies, particularly when the
hemispheric mean is above or below normal. This suggests that regionally persistent anom-
alies are related to global, rather than local, redistributions of mass.
3.2.3 Energy transports

To understand the processes involved in the atmospheric energy cycle we begin by consid-
ering the potential energy generated by diabatic heating (i.e. radiational heating/cooling,
sensible heat transfer from the surface, and latent heat released by condensation during
0 cloud formation). An evaluation has been made of atmospheric heat sources and sinks
between 50°N and 50°S by Yanai and Tomita (1998), using the National Centers for
Environmental Prediction (NCEP)/National Center for Atmospheric Research (NCAR)
11 reanalyses for 1980–94. In the northern winter they identify three major heat sources:
Figure 3.13 Mean meridional circulation cells in the atmosphere illustrated by the mass stream
functions (1010 kg s1) for annual, DJF and JJA conditions. (Peixoto and Oort, 1992)
1 The tropical Indian Ocean – Indonesia and the Southwest Pacific Convergence Zone.
2 The Congo and Amazon basins.
3 The ocean areas off East Asia and eastern North America.
In the northern summer these are replaced by three other centers:
4 The Bay of Bengal.

5 The western tropical Pacific.
6 Central America.
Year-round heat sinks are located over the South Indian Ocean, the eastern North and South
Pacific, and the eastern North and South Atlantic. In addition, the subtropical deserts are a
large source of sensible heat at the surface, but have strong radiational cooling aloft. On
a global scale the major heat sinks are in high latitudes, as implied by Figure 3.9. Nakamura
and Oort (1988) show annual fluxes into the polar caps across 70°N and S of about 95 W m2.
Figure 3.14 shows that, in the mean, the troposphere is radiatively cooled by about
1°–2°C day1. The troposphere warms through sensible heat transfer approximately
0.5°–1°C day1 in the lowest 1–2 km in the tropics and northern middle latitudes. Latent
heat release exceeds 2°C day1 in the tropical middle troposphere and is about 1.5°C
day1 around 2–3 km altitude in mid-latitudes. Net diabatic heating rates for the tropos-
phere in winter and summer are of the order of ± 0.5°–1°C day1. The annual-mean net
planetary radiation budget (at the top of the atmosphere) is about 70 W m2 at the equator
and 100 W m2 at the poles (Kyle et al., 1993). To restore balance, energy is trans-
ferred poleward by the atmosphere and ocean (Figure 3.9). For the oceans, the total energy
transport across a latitude circle is, for practical purposes, that due to heat transport:
冮[ cov T ]dz

11
Figure 3.14 The vertical and latitudinal distribution of (a) total diabatic heating in the atmosphere
11 for December–February by (b) net radiation, (c) latent heat of condensation and (d)
sensible heat in the boundary layer. (From Newell et al., 1970)
where co specific heat of ocean water (4,187 J kg1 K1). The calculation of a zonal
mean involves summing the contributions of each ocean basin, taking account of bathym-
etry. Since direct measurements of ocean currents and temperatures are sparse, the ocean
transports are estimated either as a residual between the required global energy transport
from net radiation data and the calculated atmospheric contribution (Carissimo et al.,
1985), or from regional surface heat balance calculations (Hastenrath, 1982). The ocean
heat flux in Figure 3.9 is based on the former procedure, with the atmospheric compo-
nent constrained by a maximum entropy production principle. It indicates maximum
poleward transports in the ocean of about 2.4 1015 W m2 at 18°N and 1.3 1015
W m2 at 18ºS, compared with atmospheric maxima of 4.5 1015 W m2 at 37°N and
4.7 1015 W m2 at 37°S. The North Pacific Ocean dominates the northward transport,
whereas in the southern hemisphere the Indian Ocean dominates in low latitudes and the
South Pacific Ocean near 55°S. The Atlantic Ocean heat transport is northward in both
hemispheres.
The meridional energy transfer across a latitude circle in an atmospheric column is
given by:
1
g
冕关
v (cp T gz Lq) 兴 dp
where cp T is the sensible heat (enthalpy), gz is the gravitational potential energy, and Lq
is the latent heat. The overbar denotes a time mean and the brackets a zonal mean. Kinetic
energy is negligible in this regard. Figure 3.15 illustrates the total northward energy trans-
ports and the transient component, showing the importance of sensible heat and the
transient terms. The poleward transport of sensible heat is a function of the meridional
gradient of temperature and the v component of wind (van Loon, 1979; Stone and Miller,
1980). It comprises three contributions, as discussed above, for angular momentum:
[(cpvT )] cp[v][T] cp [vT ] cp [v′T ′]

Vertically Mean Mean Mean
integrated meridional stationary transient
zonal mean cell eddies eddies
For sensible heat, the meridional cell term dominates between ± 20º latitude, whereas the
transient eddy term provides the major contribution to poleward transfer in mid-latitudes,
with maxima around 850 mb and 250 mb. However, the meridional cell transports of
potential energy have patterns similar to those of sensible heat but with opposite sign
(because, for adiabatic conditions, g dz cp dT). The net effect of the two terms in low
latitudes is an energy transfer across the equator into the winter hemisphere (Peixoto and
Oort, 1992). Figure 3.16 illustrates the annually averaged northward transport of total
energy due to the three components of circulation. The latitudinal patterns are similar to
those described for sensible heat; the standing eddy transport is prominent only in winter
in the northern higher mid-latitudes, while in the southern hemisphere the ocean trans-
ports fulfill at least part of this role. The transient eddy transports are nearly equivalent
in both seasons, especially in the southern hemisphere.
The poleward transfers of sensible and latent heat are generally positively correlated
in mid and higher latitudes (Figure 3.17). This correlation is strongest along the mid-
latitude storm tracks (see section 6.4). The primary climatic factor is the meridional
temperature gradient (see Table 3.5) and the associated advection of heat and moisture
by the meridional wind component. Raphael (1997) shows that, when the temperature
gradient weakens, the heat and moisture fluxes (at 850 mb) are more dependent on mois-
ture availability and land–sea distribution and the correlation between sensible and latent
heat fluxes is weaker.
0
0
0
0
0
11
11
Figure 3.15 Zonal-mean northward energy transport (PW) for January (upper) and July (lower). The left panels show the total transport,
the right-hand panels show the transient component. The total (—) comprises the dry static energy (DSE), latent energy (LE),
and kinetic energy (not shown). (From Trenberth and Solomon, 1994)
Figure 3.16 Zonal-mean annually averaged northward transport of (a) total energy in the atmos-
phere, and the contributions of: (b) transient eddies, (c) stationary eddies and (d) mean
meridional cell (°C m s1). See note 2. (From Oort and Peixoto, 1983)
The contrasts in land–sea distribution between the northern and southern hemisphere
mid-latitudes result in differences in the amplitude and intensity of the tropospheric
stationary waves. Accordingly, the meridional transports of sensible heat and momentum
carried out by the atmospheric eddies (stationary waves, transients – cyclones and anti-
cyclones), and the role of the upper ocean circulations, vary substantially between the
two hemispheres (Adler, 1975; Kalnay et al., 1986). In the southern hemisphere, much
as in the northern hemisphere, the eddy transports of sensible heat tend to increase over
middle latitudes when the tropospheric temperature gradient is strong in the subtropics,
and vice versa (van Loon, 1979; van Loon and Williams, 1980). Additionally, in the
southern hemisphere the temperature gradients and thermal winds are inversely related
between the subtropical and high latitudes (van Loon and Kidson, 1993). The atmospheric
levels of maximum in the eddy heat fluxes are, primarily, at around 850 mb (Kirk, 1987)
11
Figure 3.17 Correlations (statistically significant) between anomalies of transient eddy fluxes of
sensible and latent heat at 850 mb for (a) January and (b) July. (Raphael, 1997)
0
and, secondarily, at 200 mb; the latter connected with the maintenance of the upper tropos-
pheric temperature structure and the strength of the zonal mean jet just below the
tropopause (Solomon, 1997). This secondary heat flux maximum is dominated by wave
Nos 2–3 in the winter and spring of both northern and southern hemispheres. The flux
of the eddy momentum peaks in the upper troposphere between about 300 mb and 250
mb (Tucker, 1979; Kirk, 1987; Trenberth, 1987; Karoly et al., 1998).
Van Loon (1979) used the Australian and US National Meteorological Center (NMC)
grid-point analyses to compare the tropospheric waves in the southern and northern hemi-
spheres for the extreme seasons. The derivation of accurate spatial patterns of sensitive
0 quantities such as heat fluxes from the Australian analyses is complicated by the undue
influence exerted by the Southern Ocean island stations in the analyses but these effects
can be greatly reduced by zonal averaging (van Loon, 1980). The southern hemisphere
11 quasi-stationary waves are, essentially, barotropic up to at least 70°S, even in winter, in
contrast with the waves in the northern hemisphere. At the latitude of largest amplitude
in the troposphere (55°S, 50°N), the temperature and height waves comprising standing
wave No.1 are separated by between about 34°–55° of longitude in the 700–500 mb layer
in the northern hemisphere, but only by about 15° of longitude in the southern hemi-
sphere. These contrasts in the vertical structure of the stationary waves between the two
hemispheres mean that the stationary eddies (transient eddies) assume relatively greater
importance for transporting heat over most extratropical latitudes of the northern (southern)
hemisphere (van Loon, 1979; Solomon, 1997). Accordingly, the meridional transports of
heat and momentum in the southern hemisphere are effected mostly by the transient eddies
(Kirk, 1987; van Loon and Kidson, 1993), and also by the patterns of upwelling and
downwelling in the upper ocean (Trenberth, 1979). The latter are induced by the wind-
stress patterns in the standing waves (Kalnay et al., 1986): surface divergence and
upwelling to the west (surface convergence and downwelling east) of a trough. Given the
dominant role of the extratropical frontal cyclones in the atmospheric total eddy heat
transport, Tucker (1979) and Carleton and Whalley (1988) used satellite-observed cloud
vortex summaries (by latitude zone), combined with estimates of the heat fluxes, to arrive
at the “flux rate efficiencies” of successive cyclone stages. The stages show a strong lati-
tude dependence in the southern hemisphere. These studies show that the “early
development” vortex (Streten and Troup, 1973, “W” and “B” stages: see section 6.3)
transport heat strongly poleward. In winter the mature (“C” type) and dissipation (“D”
type) stages transports heat equatorward, leading to a flux convergence of the eddy sensible
heat over middle latitudes. This flux convergence is less evident in the intermediate seasons
studied by Tucker (1979), but is much more characteristic of the eddy fluxes of momentum
in those seasons. Regional differences in the eddy fluxes of heat and momentum are also
evident. Kirk (1987) shows, for the Weddell Sea region, that the transient eddy heat flux
decreases south of about 60°S, with the stationary waves increasing in importance close
to Antarctica.
Carleton and Whalley (1988) also show an El Niño Southern Oscillation (ENSO) influ-
ence in the transient eddy heat flux efficiencies by cyclone type at least for the five winters
they studied (1973–77). About 36 percent of the total variance in the meridional eddy
heat transport by the cloud vortices is explained by the Southern Oscillation Index (SOI)
for those winters, with the flux efficiency greatest (lowest) when the SOI of the following
summer is low: El Niño (high: La Niña) (see section 5.2). Similarly, the zonally asym-
metric teleconnection pattern described above exhibits anomalies with the concomitant
variations in cyclonic activity (Karoly, 1990).
Moisture transport
It is appropriate here to examine the meridional transfer of water vapor by the atmos-
phere. This represents the latent heat component of the total energy transport and it also
provides the linkage between the water and energy budgets (see section 1.1). The atmos-
pheric vapor content, averaging about 25 mm of precipitable water, accounts for only
105 of global water (Table 1.2), but its redistribution by the atmosphere is the primary
driver of the hydrological cycle. Thus net precipitation minus evaporation can be deter-
mined from:

W
PE=·Q

t
where # Q is the mean horizontal divergence of moisture in the air column,

W/
t
the change in precipitable water stored in the column. For annual averages this storage
term is near zero.
The meridional transfer of moisture across a latitude circle is given by:
冕关
11
1
qv兴 dp
g
which can be expanded in a form corresponding to that for sensible heat transport. Figure
3.18 illustrates a cross-section of the northward transport of zonal-mean water vapor by
the total circulation and its three components. The Hadley cell transports by the trade
winds, mainly below 850 mb, are directed equatorward. The stationary eddy term is a
major contributor in the northern subtropics in summer, but the transient component is
0 dominant year-round in mid-latitudes.
A different approach identifies most of the vapor flux as occurring in narrow atmos-
pheric filaments, termed “tropospheric rivers” by Zhu and Newell (1998). The separation
of these filaments (QR), from the overall vector field is made by a threshold criterion
based on the latitudinal zonal mean flux [Q], plus an anomaly related to the relative
maximum departure from the mean:
QR ≥ [Q] 0.3 (Qmax – [Q])
Zhu and Newell show that four or five tropospheric rivers account for most of the merid-
ional moisture flux. In middle latitudes northeastward streams are prominent in January
0 in the central eastern North Pacific and the central North Atlantic, as well as southeast-
ward in the central South Atlantic. Southward cross-equatorial flows occur over Indonesia,
0 Figure 3.18 Annual mean latitudinal profile of precipitation and evaporation (103 m3 s1) averaged
over 5° latitude bands (103 m3 s1 = 31.5 109 m3 yr1). Three different estimates
of precipitation are presented showing the wide range of uncertainty in middle latitudes.
11 (After WCRP, 1988)
along the coast of East Africa, and, less continuously, over the Amazon. In July the Indian
Ocean–South Asian monsoon flow is dominant, with strong northward flow also over
East Africa and the northwest Pacific. These “rivers” are analogous to the conveyor belts
in extratropical cyclones (see section 6.3). They also resemble the tropical–extratropical
cloud bands identified in satellite imagery.
Figure 3.19 illustrates the mean latitudinal profile of precipitation and evaporation and
shows the considerable uncertainty remaining in estimates of these basic quantities – due–
to limited network coverage and measurement errors. The indirect estimation of (P E )
from atmospheric wind and specific humidity measurements is similarly constrained by
inadequate upper-air sounding data. Table
– 3.7– summarizes mean annual values of the
precipitation, evaporation and estimated (P E ) from atmospheric data for 10° latitude
zones, showing the overall convergence of moisture except in the tropics and subtropics.
It was pointed out above that the seasonal variation of global-mean atmospheric mass,
and therefore of surface pressure, results from the variation of global-mean vapor pressure.
Reanalysis data for 1985–93 show that the seasonal variation of global, column-integrated
water vapor content is maintained by three pairs of centers of vapor flux divergence–
convergence (Chen et al., 1996). The centers are located on each side of the equator over
Figure 3.19 As Figure 3.16 for water vapor (ms1 g kg1). (From Peixoto and Oort, 1992)
11 Table 3.7 Mean annual estimates of precipitation and evaporation (mm) for 10° latitude belts
(Baumgartner and Reichel, 1975) and (P E) calculated from the zonal mean flux convergence
of water vapor. (From Peixoto and Oort, 1992)
Northern hemisphere Southern hemisphere
Zone a
P E (P E) P E (P E)
80°–90°
70°–80°
46
200
36
126
93
124 冧 (160) b 73
230
12
54
32
98
0 80°–70° 507 276 224 549 229 245
50°–60° 840 447 250 1,003 553 278
40°–50° 874 640 156 1,128 862 150
30°–40° 761 971 23 875 1,181 128
20°–30° 675 1,110 435 777 1,305 312
10°–20° 1,117 1,284 322 1,009 1,507 342
0°–10° 1,885 1,250 478 1,435 1,371 144
Notes:
a
The surface area of each 10° latitude zone is shown in Table 1.1.
b
70°–90°N as calculated by Serreze et al. (1995).
0
the tropical continents, with the ones on the same side of the equator having the same
sign. Contrary to classical views, Van den Dool and Saha (1993) propose a “water mass
forcing mechanism” whereby the atmospheric circulation is driven by sources and sinks
of water vapor. Some support for this argument is provided by Chen et al. (1997). They
show that the seasonal variations of regional surface pressure (ps) and vapor pressure (e)
tend to be spatially out of phase, especially in the subtropics. Surface pressure departures
are maintained by mass divergence from centers of positive ps departure to negative
centers. Conversely, water vapor converges from centers of negative to positive depar-
tures of e, both regionally and hemispherically, through the global divergent circulation3.
0 Divergent circulations are driven by latent heat released by tropical convection in areas
where (E P) < 0 (or water vapor sinks). Note that, without a phase change, water vapor
is a passive constituent transported by low-level divergent flow. The global divergent
circulations transport air mass in response to the large land–ocean pressure differences in
the northern hemisphere, and also transport water vapor to maintain the anomalies in the
sources and sinks. In the southern hemisphere, in contrast, the pressure gradients are
mainly north–south between Antarctica and the subtropical continents.
Atmospheric moisture is depleted by precipitation and restored by evaporation and
moisture flux convergence. Annual mean precipitation for 1979–95 averages 2–4 mm
day1 over large areas of the mid-latitude oceans and is up to 5–10 mm day1 in the
0 intertropical convergence zone, whereas the corresponding evaporation values reach 3–6
mm day1 over the tropical oceans and equatorial land areas, and 2–3 mm day or less
over most middle and higher latitudes (Trenberth, 1998). Temporarily ignoring horizontal
moisture transport, Trenberth shows that the e-folding time constant for the depletion of
atmospheric precipitable water ranges from generally about seven to fifteen days; it is
only five days in the tropical convergence zones, but exceeds one month in desert areas.
The time constants for the globally averaged precipitation and evaporation field are 8.1
days and 8.5 days, respectively. Nevertheless, rainfall is very concentrated in time. It has
long been recognized that in tropical and other rainstorms about 50 percent of the rain-
fall total occurs in 10 percent of the time (Riehl, 1954).
0 At an individual location, precipitation is all derived from atmospheric moisture trans-
port, while for the surface of the earth the source is evaporation. The contributions of
atmospheric moisture transport and local evaporation to precipitation in the same area
11 thus depend on the size of the area being considered. Estimation of the “intensity of the
hydrological cycle” (Drozdov and Grigor’eva, 1965), or the fraction of the moisture trans-
port that is precipitated in a region, can be expressed as:
I PL/F
where P precipitation, F moisture transported through an atmospheric column, and
L a length scale for the region ( area0.5) according to Budyko and Drozdov (1953). It is
assumed that the ratio of precipitation falling from advected moisture (Pa ) versus that from
local evaporation (Pe ) is equal to the ratio the average atmospheric moisture advected
versus that evaporated. The recycling ratio Pe /P can be expressed after Brubaker et al.
(1993) as:
EL
PL 2F
Global mean values, for L 500 (1,000) km, are close to 10 (20) percent, respectively,
according to Trenberth (1998, 1999). Estimates of local precipitation recycling suggest
values of 10–24 percent over a 1,500 km scale for the Mississippi basin and 25–35 percent
in the Amazon basin over the larger 2,500 km scale (Eltahir and Bras, 1996). They suggest
that, where local evaporation dominates, the recycling ratio is independent of L. However,
Burde et al. (1996) consider that the above model, which applies the length scale to an
area, makes an inappropriate substitution. They develop a correction factor which takes
account of the structure of the advective flow (zonal or meandering), although no compar-
isons are made for specific regions.
3.2.4 Energy conversion

The classical idea of momentum and energy conversions in the atmosphere is embodied
in L.F. Richardson’s dictum that:
Big whirls have little whirls that feed on their velocity,

little whirls have lesser whirls, and so on to viscosity.
In fact, this cascade of momentum from larger to smaller scales is often reversed in the
atmosphere – a process termed “negative viscosity” by Starr (1968).
Lorenz (1955) established that the relevant potential energy for the atmosphere is the
available potential energy (APE). This refers to the potential energy above a reference
state in which the atmosphere is rearranged through a dry adiabatic process such that it
is statically stable (lowest potential temperature at the surface), with horizontal isentropic
and isobaric surfaces. It is defined by Haltiner and Williams (1980) as:
冕
––– 2
冢 T 冣 dp
T′
Ps
T
APE = 1/2
o
where the dry adiabatic lapse rate (DALR), actual lapse rate, the overbar denotes
a mean value and the prime is the local departure and
冕
Ps
integration over the entire depth of the atmosphere.
o
The APE is of the order of 0.5 percent of the total potential energy. APE is created
by diabatic and frictional processes. It can be separated into a zonal mean (AZ) and an
eddy component (AE); the former is a result of meridional gradients of temperature/density,
11 i.e. differences between latitude circles, while the latter involves departures from the zonal
mean along a latitude circle. Correspondingly, the kinetic energy
KE =
1
2g
冕o
Ps
V 2 dp
For the atmosphere, KE/APE ~ 0.1. KE can also be separated into zonal (KZ) and eddy
(KE) components. Potential and kinetic energies undergo a variety of conversion processes
in the atmosphere. These include the initial generation of zonal (GZ) and eddy (GE) avail-
0 able potential energy and the eventual dissipation components of kinetic energy (DZ and
DE). Table 3.8 summarizes the atmosphere mechanisms involved in each conversion
process. Intermediate transformations involve:
CA is the conversion of AZ → AE
CE is the conversion of AE → KE
CK is the conversion of KE → KZ
CZ is the conversion of AZ → KZ
In evaluating the terms in the atmospheric energy conversion cycle (Figure 3.20), the kinetic
0 energy (KE) and available potential energy (APE) are partitioned into zonal and eddy
Table 3.8 Energy conversions in the atmosphere
Term Process Mechanism

GZ Generation of AZ Meridionally differentiated diabatic heating; warming of
tropics, cooling of higher latitudes
GE Generation of AE Diabatic heating augments east–west temperature differences
CA Conversion of AZ to AE Horizontal meridional transport by waves and eddies of
warm (cold) air poleward (equatorward) reducing the
0 meridional temperature gradient and therefore AZ
CE Conversion of AE to KE Warm (cold) air rises (sinks) in zonal circulation cells
(Walker circulation)
CZ Conversion of AZ to KZ Warm (cold) air rises (sinks) in low/high latitudes as in the
Hadley cell. The indirect Ferrel cell converts in the
opposite sense by cold (warm) air rising (sinking)
CK Conversion of KE to KZ Eddies transfer angular momentum from regions of high to
low momentum and thus kinetic energy is lost (gained)
by zonally averaged symmetric (asymmetric) motion
DE Dissipation of KE Frictional decay of eddy kinetic energy
DZ Dissipation of KZ Frictional decay of zonal kinetic energy
0
Figure 3.20 Schematic illustration of the energy cycle components in the atmosphere; see text.
11 (From Grotjahn, 1993)
contributions, KZ, KE and AZ, AE respectively. KZ and KE are determined by resolving the
velocity field into zonally averaged motion [u] and eddy motion u* (Lorenz, 1967, p. 108):
1
KZ = { [u] · [u]}
2
1
KE = {u* · u*}
2
Lorenz emphasizes that KZ is not equivalent to the KE of the zonal motion ( u 2/2), or
to zonally averaged KE ( [u # u]/2). The derivation of the equations for KE is presented
in detail by Grotjahn (1993, pp. 96–103).
The approximate equations for the APE components (after Lorenz, 1967) are:
1
冦
AZ = cp ( ⬃
2
⬃ ⬃ 2
)1 T 1[T ]* 冧
1
冦
AE = cp ( ⬃
2
⬃
)1 T 1(T [T])
2
冧
where the DALR, the lapse rate, ~ average over an isobaric surface, and *
deviation from the average. [T] zonal mean temperature along the isobaric surface
and ( ˜) represents the static stability. Thus APE is expressed as a weighted average
of the horizontal (isobaric) variance of temperature, T. APE is produced either by heating
(cooling) of warm (cold) regions, or by heating (cooling) of the lower (upper) troposphere
and so reducing the stability.
In corresponding manner, the energy generation (G), conversion (C) and dissipation
(D) terms are also expressed as covariances of zonal-average or eddy departure quantities.
Lorenz (1967) defines the generation terms:
⬃
GZ = {( ⬃
)1 T 1[Q*][T*]}
⬃
GE = {( ⬃
)1 T 1(Q [Q])(T [T])}
where Q net heat per unit mass. The conversion of total PE into KE, which involves
reversible adiabatic processes, can be represented by:
C g(w
z/
p) g(U ⴢ z).
There are corresponding equations for the conversions CZ, CE, CK and CA.
The dissipation term is expressed by:
D {U ⴢ F}
where F frictional force per unit mass.
Figure 3.20 illustrates these conversions and the associated reservoirs schematically;
the magnitudes of the various components in the global energy cycle are discussed below.
Typical atmospheric processes which act to effect these conversions are summarized in
Table 3.8. The conversion ± CK involves barotropic exchanges whereas ± CZ and ± CE
involve baroclinic processes. The major pathway for the global atmospheric circulation
illustrated in Figure 3.21 is from:
Gz → Az → AE → KE → DE
↓→ KZ → ↑
11
Figure 3.21 The observed annual mean cycle of energy generation, conversion, and dissipation
(W m2) in the atmosphere. The energy reservoirs shown in the boxes are almost
constant (105 J m2). Parentheses indicate terms derived as residual values. P denotes
APE and M a zonal mean. (From Peixoto and Oort, 1992)
In middle latitudes, baroclinic processes dominate in the generation of extratropical

0 cyclones and, from an initial zonal circulation regime, the sequence AZ → AE →KE leads
to an enhancement of meridional circulation resulting in a low index or blocking pattern.
The conversion CK in Figure 3.21 contradicts the classical view of turbulent decay from
larger to smaller scales, as noted earlier.
The spectral distribution of energy across wave numbers has received much attention
in studies of turbulent decay. For three-dimensional quasi-geostrophic flow, an m3 depen-
dence has been determined for the horizontal and vertical wave numbers (m) of the spectral
variance of kinetic energy and temperature (Charney, 1971). This contrasts with the well
known 5/3 law of Kolmogorov (1942) for three-dimensional isotropic turbulence in the
inertial subrange. Kraichnan (1967) first derived the 3 relationship for large-scale flow
0 that was attributed to its essentially two-dimensional characteristics. However, Charney
considered this to be a fortuitous result of the geostrophic constraint on the motion. He
showed that 2-D inviscid flow is governed by vorticity conservation whereas 3-D quasi-
geostrophic flow is governed by the conservation of both potential vorticity and potential
temperature. By combining these latter two adiabatic constants of motion into a pseudo-
potential vorticity parameter Charney demonstrated that the conservation of this property
can forbid an energy cascade. Following from this, the 3 relationship is obtained for
the variance spectra of horizontal kinetic energy and temperature where the wave number
exceeds the excitation wave number.
0
3.3 Circulation cells
The preceding discussion indicates that the atmospheric circulation must satisfy two condi-
11 tions: energy must be transported from the zone of surplus in lower latitudes to the
higher-latitude zone of deficit, and absolute angular momentum must similarly be trans-
ported poleward. The form of circulation that can satisfy these requirements is not
immediately obvious. Early theoreticians postulated circulations in a meridional plane. In
1735 Geoffrey Hadley proposed that rising warm air in low latitudes flowed poleward at
high levels, cooling and sinking in higher latitudes, with a low-level return flow towards
the equator. The upper poleward and lower equatorward motions would be affected by
the Earth’s rotation (although the effect of Coriolis acceleration was not specifically recog-
nized), giving in the northern hemisphere southwesterly components aloft and northeasterly
components (the Trades) at low levels. The mean thermally direct circulation in low lati-
tudes is still described as a Hadley cell (Figure 3.12) to honor this early work. Classical
vertical plane models of the atmospheric circulation depict an intertropical convergence
zone (ITCZ) as the upward limb of the adjacent Hadley cells in the two hemispheres (see
Figure 3.12). However, when the ITCZ moves away from the equator a reversed equa-
torial cell may separate the two Hadley cells, as proposed by Asnani (1968).
By the mid-nineteenth century, wind observations showed that the mid-latitude west-
erlies moved poleward, not equatorward as suggested by Hadley’s direct cell model. Wind
charts compiled by James Coffin (1853) for the northern hemisphere, and later for the
globe (Coffin, 1876), identified three surface wind belts – easterly trades, mid-latitude
westerlies and polar easterlies. William Ferrel (1856, 1859–60) used these maps to illus-
trate the effects of the Earth’s rotation on the atmospheric motion. He applied the principles
of fluid motion and the conservation of absolute angular momentum developed by the
Marquis de Laplace, whose work had been translated into English thirty years earlier, and
the idea of a deflective force (G. Coriolis’s work), to the circulation of the atmosphere
and ocean currents (see Fleming, 1989, p. 138). Ferrel proposed a model with three merid-
ional circulation cells, the subtropical high-pressure belt, a zone of trade wind convergence
and another zone of convergence near the polar circles. The scheme included the indi-
rect circulation (where cold air rises and warm air sinks) in middle latitudes, which is
now called the Ferrel cell in recognition of his contribution.
Key characteristics of the Hadley circulations, based on global tropospheric wind data
for 1964–89, are summarized in Table 3.9. The seasonal cycle is analogous in both hemi-
spheres, with maximum values (~20 1010 kg s1) of the mass stream function $ occur-
ring near 8° latitude in the respective winter seasons.4 The ill-defined summer maxima
located near 26º latitude are an order of magnitude smaller. In the annual mean the Hadley
cells are fairly symmetrical about the equator, with the ascent of warm, moist equatorial
air, associated with the convergence of air in the mean intertropical convergence, and
descending motion in the subtropics. This symmetry is observed in the transition seasons,
particularly October (Oort and Yienger, 1996). In the other seasons, in contrast, there is an
intensified Hadley circulation in the winter hemisphere, strong cross-equatorial flow into
Table 3.9 Measures of the Hadley circulation, 1964–89
Measure January April July October Annual
Max NH $ (1010 kg s1) 22.9 15.4 1.4 8.2 9.2

Latitude $ max (°N) 8.0 10.0 26.0 16.0 12.0
Max SH $ (1010 kg s1) 3.1 7.3 19.3 10.0 7.4
Latitude $ max (°S) 27.0 13.0 7.0 7.0 9.0
v200–v850 (10°N) m s1 5.2 3.2 1.5 0.8 1.8
v200–v850 (10°S) m s1 2.1 1.0 4.0 1.8 1.2
STJ (15°–25°N) m s1 25.7 23.2 4.2 8.3 13.1
STJ (15°–25°S) m s1 6.7 16.7 23.3 21.6 17.1
Source: from Oort and Yienger (1996).
Note: negative signs denote southward transport.
11 the summer hemisphere and the virtual disappearance of the cell in the latter. In the north-
ern winter, meridional velocities in the northern Hadley cell are about twice those in the
southern hemisphere, but in the northern summer the southern cell is five times stronger
than that in the northern hemisphere, according to Mak and Liu (1993). This is attributable
to the cross-equator flow associated with the Asian summer monsoon. However, these pat-
terns refer to values of v obtained by zonal averaging. Using meridional wind components
for July 1957–December 1964, directly observed at 330 stations between 45°N and 45°S,
Schulman (1973) shows that the monsoon circulation dominates the sector 40°–150°E
whereas the remaining two-thirds of the global circumference have a nearly symmetrical
0 Hadley cell pattern. Longitudinally there is considerable spatial variability in the occur-
rence of ascending air. It is concentrated over the tropical continents, in Amazonia, equa-
torial Africa and the “maritime continent” of Indonesia–Malaysia; in the intervening sectors
there is large-scale descent. These longitudinal contrasts give rise to zonal (east–west)
Walker circulations discussed in Section 5. Ascent is highly localized in “hot towers”
(cumulonimbus cells) mainly within synoptic-scale disturbances (Riehl and Malkus, 1958,
1979). GCM experiments by Rind and Rossow (1984) show that the Hadley circulation is
forced by two competing heat sources. These are solar radiation, which at the solstice peaks
at 23° latitude, and latent heat released by near-equatorial precipitation. These are not in
phase at the solstices; in the summer hemisphere they are in opposition, forcing circulation
0 cells of opposite direction between the equator and 23° latitude.
The spatial pattern of atmospheric heat sources and sinks shows strong seasonal contrasts,
as noted above (pp. 129–3). The primary contribution to heating identified in most areas is
the release of latent heat through cumulus convection. Comparing calculated apparent heat
sources and moisture sinks over the western equatorial Pacific warm pool, Yanai and Tomita
(1998) show a vertical pattern of heating between 400–500 mb and drying between 600–700
mb, characteristic of cumulus convection. However, over most of the North Pacific storm
track in winter, initial sensible heat transfer to the atmosphere from the warm ocean surface
is replaced downstream by condensational heating. Also, over much of Asia and North
America in northern spring, sensible heat fluxes are important, especially over the Tibetan
0 Plateau and the western USA. These patterns help account for the principal features of the
Hadley and Walker circulations, as well as some components of the monsoon circulations
(see section 3.7.3).
In addition to the annual shifts, the Hadley cells are affected by considerable variability
on both long and short time scales. Anomalies in Hadley cell strength are inversely corre-
lated with the strength of the low-latitude Walker circulations, and there are also
correlations with the phase of the El Niño Southern Oscillation (Section 5). During El
Niño (La Niña) the winter pattern features two strengthened (weakened) Hadley cells
(Oort and Yienger, 1996) resulting from the intensification (relaxation) of the latitudinal
temperature gradient. Submonthly variability of the Hadley circulation is suggested by
0 the fact that the standard deviation of the zonally averaged meridional wind component
in the upper troposphere (200 mb) over the equator reaches 0.7 m s1 in summer and
winter. A possible mechanism for submonthly fluctuations is suggested below.
The factors controlling the latitudinal extent of the meridional Hadley circulation have
been examined in a number of studies. According to Hunt (1979) the poleward extent,
which is demarcated by the latitude of the subtropical westerly jet core, is determined by
the planetary rotation rate, which affects the momentum balance. Compared with a control
case value of 22° latitude, in a hemispheric simulation with a general circulation model,
the jet core and poleward extremity of the Hadley cell expand to 46° latitude for a five-
fold increase in rotation rate and contract to 10° latitude for a fivefold decrease in rotation
0 rate (Figure 3.22). In each experiment the subtropical jet core, the poleward limit of the
Hadley cell, and the mean surface anticyclone are latitudinally coincident. The descent
of air is necessitated by the requirement for the atmosphere to generate east–west pressure
11 gradients in order to restrict the increase of the atmosphere’s relative angular momentum
Figure 3.22 Changes in the extent

of the Hadley cell in relation to rota-
tion rate. The panels show merid-
ional stream functions (1012 kg s–1)
for fast (top), control case (middle),
and slow (bottom) rotation rates.
(From Hunt, 1979)
11 during a poleward displacement (Lorenz, 1967, p. 74). Local east–west gradients provide
the necessary torque to prevent the zonal wind from attaining excessive values of latitu-
dinal wind shear. Meridional shear values just equatorward of the subtropical jet are
approximately 1.3 105 s1.
In spite of the appealing simplicity of Hunt’s results, other studies imply that the extent
of the Hadley cell also depends on the static stability (Schneider, 1977), the depth of the
circulation, and the magnitude of the differential radiative heating (Held and Hou, 1980).
The work of these authors provides the basis of a theory for a symmetrical Hadley cell.
The width of the cell, the distribution of surface winds, the position of the upper-level jet
0 and the vertically averaged fluxes of heat and momentum (expressed non-dimensionally)
can be specified as universal functions of the external thermal Rossby number:
RE gH H /(r")2
where r radius of the earth (6,000 km), g gravity (9.8 m s2), H 8 km is the
depth of a Boussinesq atmosphere (i.e. where vertical variations in density are negligible),
H a baroclinicity parameter, " earth’s angular velocity.
Schneider’s model requires that the atmosphere adjusts its static stability to balance
radiative cooling with a given heating distribution, which results in apparent dependence
of the width of the Hadley cell on the static stability. In contrast, Held and Hou (1980)
0 adopt a stable and stratified, dry, differentially heated, rotating Boussinesq fluid where
angular momentum is nearly conserved in the poleward flow. They find that here the
width of the Hadley cell is proportional to the square root of both the horizontal temper-
ature gradient and the cell’s depth and inversely proportional to the rotation rate.
The theory is most appropriate at the equinoxes, when the heating is symmetrical
about the equator (Williams, 1988). It predicts that for RE % 1, the Hadley cell extends
approximately to the core of a tropical jet near the latitude (H) where the wind resulting
from angular momentum conservation (uM r " sin2 /cos ) interacts with the west-
erly thermal wind of middle latitudes; this latitude is given by H (5/3RE )1/2 which is
about 30° latitude, as observed. When RE increases, latitude H also increases.
0 The Held and Hou model ignores latent heat release, which will tend to strengthen the
Hadley cell and restrict the zone of ascending air but should have limited effect on the
cell’s width or the strength of the westerly jet. The incorporation in the Held and Hou
model of latent heat release during condensation does in fact reduce the meridional extent
of the ascending air, according to Dodd and James (1997), and in contrast to other studies
it does not weaken the circulation when the heating is away from the equator. In the real
atmosphere, moist convection may raise the tropical tropopause, thereby modifying RE
and the cell’s width, and so circulation-induced changes in cloudiness, and therefore the
heating distribution will also be important. Nevertheless, each of the proposed relation-
ships, as well as laboratory analog experiments with slow rotation, would imply an
0 equator-to-pole cell for the rotation rate observed on Venus. In contrast, Rossow (1985)
notes that observations of Venus’s atmosphere indicate strong zonal flow and fully devel-
oped turbulent motions. Existing models neglect heat and momentum transfer due to
eddies that can modify the mean zonal flow and temperature distribution and the extent
of the Hadley cell. Model experiments by Rind and Rossow (1984) where static stability,
circulation depth, rotation rate, and radiative forcing are held constant still exhibit fluc-
tuations in Hadley cell extent. Thus a balanced mean meridional circulation appears to
involve eddy processes, which must be appropriately parameterized in any model study,
and also surface friction.
The possible range of atmospheric circulations can be explored by GCM experiments
0 where the controlling parameters – the rotation rate, latitudinal gradient of Coriolis, mois-
ture state, surface drag and surface heating – are varied. Williams (1988) uses this approach
to determine the basic controls of the dynamic components of the planetary circulation,
11 namely the mix of meridional cells, jets, and horizontal eddies, and their size and strength.
The rotation rate, "*("/"E), normalized by the terrestrial value ("E), is varied from 0 to
8. The GCM incorporates a zonally symmetric swamp surface, nine vertical levels and
R15 to R42 spectral resolution (i.e. with romboidal truncation at wave Nos 15–42). Some
of Williams’s main findings are as follows.
1 The strength of the Hadley cell, relative to the large-scale horizontal eddies that force
the zonally averaged mean flow (the quasi-geostrophic modes), depends on the mois-
ture content; moist atmospheres are dominated by the Hadley modes, dry ones by
quasi-geostrophic eddies (Figure 3.23). Nevertheless, the circulations resulting from
localized heating in low latitudes for dry atmospheres resemble those in atmospheres
with latent heating, having similar Hadley cells and tropical westerly winds.
Figure 3.23 Schematic summary of the characteristics of the quasi-geostrophic (QG), quasi-Hadley
(QH) and natural Hadley (NH) circulations. The QG(QG ) case represents flow for
nonlinear baroclinic instability (on a midlatitude -plane; ū = mean zonal wind,
—– —–
= stream function, v′T′ = mean meridional eddy heat transport, v′M′ = mean merid-
—––
ional momentum transport, –w′M′ = mean vertical momentum transport, Kz: KE(P:KE
= barotropic (baroclinic) energy conversion. Negative values shaded. (From Williams,
1988)
11 2 The Hadley mode disappears in favour of a quasi-geostrophic barotropic mode if
surface drag is eliminated.
3 For moist atmospheres, low rotation rates ("* 0–0.25) set up a direct cell of almost
hemispheric extent with a near-polar westerly jet (a “Natural Hadley” circulation; see
Figures 3.23–4); low-latitude waves are generated by convective eddies and high-lati-
tude ones by barotropic cascades in the polar jet. The jet moves to mid-latitudes as
"* increases. For "* 0.5–1.0 the direct cell is narrow, with a strong tropical west-
erly jet and mid-latitude Rossby waves (a quasi-Hadley circulation); the Hadley cell
is strong and there is a weak indirect Ferrel cell. For high rotation ("* 2–8), there
0 is a tropical Hadley cell with tropical upper westerlies and several mid-latitude
elements and a polar quasi-geostrophic (QG) element comprising narrow, weak cells
and weak westerly wind maxima (Figure 3.24). Williams shows that the mid-latitude
(QG) elements have poleward jet-traversing momentum fluxes associated with plan-
etary waves that are dispersing asymmetrically and favor equatorward propagation.
The polar QG element has symmetric wave dispersion that produces jet-converging
momentum flux (see Figure 3.25).
The forcing of the average Hadley circulation has been demonstrated via a number of
experiments with primitive equation models for a zonally symmetric regime forced by
0 radiative heating. Analytical investigations of the role of axially symmetric heating about
the equator indicate that an equator-to-pole atmospheric temperature gradient of 40°C (cor-
responding to that observed over the oceans) produces a Hadley circulation much weaker
than the observed annual average condition if the forcing is centered over the equator (Held
and Hou, 1980). Even a gradient of 100°C (corresponding to a state of planetary radiative
equilibrium), is insufficient in these circumstances. Nevertheless, when the thermal forcing
is displaced a few degrees off the equator, the Hadley cell in the winter (summer) hemi-
sphere is greatly amplified (reduced). The mass flux in the low-latitude mean meridional
circulation shows a good degree of symmetry only during equinoctial conditions in April
(Peixoto and Oort, 1992), as shown in Figure 3.13. However, the asymmetry which exists
0 between the hemispheres for most of the year gives rise to an annually averaged Hadley
circulation that is significantly larger than that forced by the annually averaged heating
(Lindzen and Hou, 1988). Moreover the transport of angular momentum by the Hadley cell
is primarily directed into the winter hemisphere, where it excites eddy instability and forms
a strong subtropical westerly jetstream on the poleward margin of the Hadley cell. Lindzen
and Hou point out an apparent discrepancy between their model and observations, namely
the zonally averaged thermal maximum is located slightly north of the equator year-round.
This difficulty appears to be related to the distribution of low-level convergence and latent
heat released by convection that modifies the zone of diabatic heating. In fact, latent heat
release tends to occur south of the equator in the austral summer, especially over the south-
0 west Pacific Ocean. In a further study Hou and Lindzen (1992) show that heating in a nar-
row zone centered symmetrically on the equator strengthens the Hadley circulation up to
fivefold, relative to a non-perturbed case. In the atmosphere the heating concentration in
the equatorial zone is effected by low-level moisture convergence, convection and latent
heat release. For off-equatorial heating, the intensity of the Hadley circulation depends on
the way in which the heating is latitudinally distributed – symmetrically about the heating
maximum or preferentially from the winter hemisphere. There is strong Hadley cell
intensification in the latter case, and only modest heating concentration is required to match
the calculated and observed intensities of the Hadley circulation. The results of Hou and
Lindzen’s analysis are consistent with the observed broad latitudinal distribution of
0 zonally averaged precipitation (Table 3.7), rather than a narrow ITCZ precipitation belt.
For equatorial and off-equatorial cases where the redistribution of heating is restricted to a
zone much narrower than that of the circulation, the circulation strength is not significantly
11 augmented.
Figure 3.24 Meridional distribution of mean zonal wind (m s1) for the moist version of the GCM for "* equal to (a) 4, (b) 1 (= terrestrial value), (c)
1/2, and (d) 1/8. (From Williams, 1988)
0
0
0
0
0
11
11
Figure 3.25 As Figure 3.24 for the mean stream function (1013 g s1). (From Williams, 1988)
Nevertheless, the results imply that a weakening of the easterly wave disturbances (see
section 6.5) and monsoon regimes that currently broaden the zonally averaged precipita-
tion distribution would enhance the Hadley circulation strength and modify the jetstreams
and wave transport in higher latitudes. Using a simplified version of the Goddard
Laboratory for Atmospheric Sciences (GLAS) GCM, Hou (1993) simulates the effect of
a shift of prescribed heating into the summer hemisphere on the extratropical circulation.
In agreement with earlier work he finds an intensified cross-equatorial Hadley circulation
into the winter hemisphere, enhanced upper tropical easterlies and to a lesser degree the
winter subtropical jetstream. The mid to high latitudes of the winter hemisphere are
warmed by increased dynamic heating related to eddy heat fluxes associated with low-
frequency transient waves in middle latitudes and high latitude stationary waves (which
may help organize transient eddies along storm tracks).
In an extension of the steady-state frictionally controlled models of thermal forcing
Hack et al. (1989) examine the evolution of meridional circulations in a frictionless
balanced zonal flow model. The width of the thermal forcing is chosen to be 4° (8°) lati-
tude at latitude 30° (15°), respectively, in accordance with satellite observations of ITCZ
cloudiness. The calculated maximum mass flux from the winter hemisphere, for a narrow
or wide zone of heating, is obtained when the heating is at 12° latitude (Figure 3.26).
The maximum difference between the mass fluxes of the two cells occurs with heating
centered at 13° latitude. The sensitivity of the circulation to the asymmetry of the heating
with respect to the equator is less than that obtained by Lindzen and Hou (1988), but is
in the same sense. The air is drawn across the equator into the ITCZ, increasing the mass
flux in the winter hemisphere Hadley cell, by virtue of the inertial instability of the air
in low latitudes. However, if the heating is displaced more than about 12° latitude, the
transfer of air becomes less effective. Hack et al. (1989) note that this value is consistent
Figure 3.26 Calculated mass fluxes with different thermal forcings. (From Hack et al., 1989)
11 with a Rossby radius of 14° latitude according to the equatorial plane theory of Matsuno
(1966).
In a further study, using the same model formulation with time-dependent radiative
heating, Mak and Liu (1993) confirm the results of the earlier studies and show that the
transitions between the dominance of the respective winter-hemisphere Hadley cells last
only about a month in each case. There is also a strong zonal jetstream at the poleward
margin of the winter cell, as a result of absolute angular momentum conservation in the
poleward upper-level flow. The latitudinal extent of the Hadley cells is controlled by
the thermal gradient needed to sustain baroclinic shear that increases with latitude. Note
0 that the Hadley circulation cools the tropics and warms the extratropical latitudes. The
model simulates bands of disturbances on the poleward side of the winter-hemisphere
Hadley cell that arise from instability in the baroclinic shear flow. Mak and Liu (1993)
propose that these disturbances (causing submonthly fluctuations in v) result from
symmetric instability of slowly evolving background flow.
3.4 The Earth’s geography

The preceding discussion has focused on the two-dimensional meridional circulation com-
ponents. However, the three-dimensional structure involves the zonal Walker circulations
0 (noted above) and monsoon circulations (section 3.7.3). Both play key roles in shaping the
actual global general circulation.
Considerable insight into the role of geography in shaping the Earth’s climate has been
gained from experiments using General Circulation Models to simulate paleoclimatic
conditions (Barron and Washington, 1984). There has not yet been a comprehensive suite
of experiments, using different models, to examine the relative effects of land–sea propor-
tion and distribution and of mountain barriers in a systematic way. Therefore the following
account must be regarded as subject to future revision.
Hay et al. (1990a, b) have used the NCAR Community Climate Model (CCM) to inves-
tigate the large-scale effects of continental location and orientation. In one set of experi-
0 ments they use a north–south continent, 105° longitude wide, extending from pole to pole.
The total land area corresponds to the present value. There is snow cover south of 70°S.
Varying reliefs are used – low (750 m) plateau, high (1,500 m) plateau, and mostly low
plateau but with 3 km mountains on the east or west side. In a related study they contrast
an Earth with two 750 m high polar continents from 45° latitude to the poles, the southern
one with snow cover poleward of 70°S, an Earth having a 750 m high tropical continent
between 17°N and 17°S. All experiments are forced by mean annual solar radiation and a
“swamp” ocean that can form and grow sea ice. Cloud cover and the hydrological cycle in
the experiments are predictive. The land–sea proportions and mean elevations are as now,
except for the elevation of Antarctica.
0 The comparison of polar continents versus a tropical one shows that low-latitude conti-
nents give rise to an earth that is 8°C warmer than present for the mean global surface
temperature (Figure 3.27). Tropical evaporation is much reduced by a low-latitude conti-
nent which decreases tropical precipitation. The Earth with polar continents has a mean
temperature close to present and a strong hydrological cycle, especially in low latitudes.
The continental interiors are dry, but the pole with a snow cover appears to create posi-
tive feedback for precipitation that would maintain an ice cap. By itself a polar continental
location seems insufficient to promote glaciation, suggesting the importance of moisture
as well as temperature effects.
The polar continents give rise to strong temperature gradients near their margins, but
0 the jetstreams are rather weak (25 m s1). The mean sea level (MSL) pressure pattern in
the tropics and subtropics is characteristic of a water-covered earth. The tropical continent
case has a hot interior, and polar temperatures are around 4°C, which contrasts with the
11 results of a seasonal-cycle energy budget model (Crowley et al., 1987) that generated
Figure 3.27 Results of simulations with the NCAR Community Climate Model to illustrate the effect of orography on global temperature (above) and
precipitation (below) for an Earth with pole-to-pole continents for (a) 750 m plateau, (b) 1,500 m plateau, (c) low plateau with 3 km moun-
tains on the east side; (d) as (c) with mountains on the west side. (From Hay et al., 1990a)
11 polar ice. The subpolar lows are deep and the almost symmetrical jetstreams exceed
30 m s1 at 12 km altitude. In both polar regions there are weak circulations.
The results suggest that at present the Earth’s climate is near to one extreme; mean
temperatures in the southern hemisphere now (Table 3.5) are close to those in the snow-
covered polar continent case. More significantly, land–sea distribution is shown to modify
the hydrological cycle’s intensity by limiting moisture availability (Barron et al., 1989).
The experiments for pole-to-pole continents with differing relief (Hay et al., 1990a)
clarify the effects of orography on global climate5. Higher land masses increase the mean
MSL pressure by displacement of mass, and increase the contrast between high and low
0 pressure cells. Increased elevation also reduces temperatures (Figure 3.27). Moisture trans-
port eastward into the continental interior is enhanced by mountain ranges on either coast.
West-coast mountains give rise to an equatorial wet belt, while east-coast mountains cause
considerable continental aridity except for a narrow northern temperature belt. The
north–south coasts eliminate any moisture source for the continent from the ocean to the
east. All four configurations show zonal patterns of precipitation, soil moisture, and runoff,
implying that Köppen’s (1931) idealized climatic pattern, suggesting a southwest–north-
east slope of the arid zone boundary from 20° latitude on the west coast to 40°–60°E
latitude in the east, may be determined primarily by present geography; this is especially
true for the configuration of the Tibetan Plateau and the Cordilleras of North and South
0 America. Modeling studies with varying soil moisture support this suggestion for the
southern continents (Cook, 1994).
The effects of mountain barriers on climate have been extensively studied using several
different GCMs – the GFDL model (Manabe and Terpstra, 1974; Manabe and Broccoli,
1990), the NCAR Community Climate Model (Ruddiman and Kutzbach, 1989). Model
control cases have been run with and without present orography. The simulations show
that mountain ranges indeed account for the dry present-day continental interiors of Asia
and North America (Figure 3.28) (Kutzbach et al., 1993). In the absence of the moun-
tain barriers, these regions experience moist climatic regimes (Manabe and Broccoli,
1990). The role of orography in the planetary wave structure is examined in section 4.2.
0 There is also an extensive literature on paleoclimates for past epochs with radically
different land and ocean distributions and orography (see Crowley and North, 1991) and
postulated changes in orbital geometry (Kutzbach, 1994; see Crowley and North, 1991).
3.5 Climate system feedbacks

The third set of processes that contribute to global and regional climate are internal to
the climate system. They include trace gases and other atmospheric constituents, cloud
cover, air–sea and land surface–atmosphere interactions, including the biosphere. Internal
factors largely operate through feedback effects on the exchanges of energy between the
0 atmosphere and surface. A feedback process occurs when one variable in a system trig-
gers a change in a second variable that in turn affects the system. More generally, we
can say that “information” on the state of the system feeds back to affect the system’s
future state. This may act to amplify and destabilize, or to dampen and stabilize, the initial
perturbation; these are referred to as positive or negative feedback, respectively. For
example, with a temperature increase at the surface, the outgoing infrared radiation
increases, offsetting the temperature rise – a negative feedback. If the soil warmed enough
to release more carbon dioxide, that could enhance the greenhouse effect, thereby
providing positive feedback. The quantitative analysis of feedback in climate model simu-
lations can be performed following alternative definitions (Dickinson, 1985; Schlesinger,
0 1989) and, therefore, some care is needed with the literature on the topic. The relative
importance of its primary contribution to long-term climate change has been estimated
from climate model experiments by Hansen et al. (1984). Figure 3.29 indicates that the
11 approximately 5°C global mean temperature change between the last glacial maximum
Figure 3.28 Simulations with the NCAR Community Climate Model showing the effects of orog-
raphy on temperature and moisture balance in the continental interiors for present-day
orography (M) minus no mountains (NM). (From Kutzbach et al., 1993)
and the present day can be attributed primarily to the water vapor–temperature feedback,
changes in cloud cover (of uncertain magnitude), and snow/ice–albedo feedback. The
initial forcing due to astronomical factors exerts only a minor influence. The feedback
mechanisms operate as follows.
Water vapor–temperature feedback

A rise in air temperature permits an increase in saturation vapor pressure. The relation-
ship, described by the Clausius–Clapeyron equation, is a non-linear, exponential one:
d ln es L
= v2
dT Rv T
where es saturation vapor pressure, Lv latent of vaporization, Rv specific gas constant

for water vapor. The importance of this relationship in the atmosphere is illustrated by
the fact that plots of the atmospheric vapor pressure against air temperature in vertical
air columns at different latitudes are parallel to the esT curve (Figure 3.30). Water vapor
is an infrared-absorbing gas; indeed, it accounts for about two-thirds of the natural terres-
trial “greenhouse” effect (21 K of the 33 K increase in mean surface temperature above
the Te value of 255 K). Hence more of the outgoing radiation is trapped in the atmos-
11
Figure 3.29 Contribution of different feedback processes to the glacial–modern differences in global
mean temperatures. (From Hansen et al., 1984)
phere, causing a further increase in temperature – a positive feedback. Evidence of

increases in evaporation over the tropical oceans (10ºS–14ºN) by 16 percent (20 cm yr2)
0 during 1949–89, and corresponding increases in atmospheric water vapor, suggest an addi-
tional energy input into the global atmosphere of 13 W m2, according to Flohn et al.
(1992).
The relationship between surface temperature and column moisture has been contro-
versial. Lindzen (1990) argued that warming in the tropics might create drying of the
middle–upper troposphere as a result of downdraft areas (decreasing the water vapor
greenhouse). Analysis for 1988–93 shows, however, that surface temperatures and
500–200 mb specific humidity are significantly positively correlated both locally and for
tropical averages, 30°S–30°N (Yang and Tung, 1998). A positive coupling of surface
temperature and mid-tropospheric humidity is also obtained from ERBE data by Inamdar
0 and Ramanathan (1998). They calculate the normalized atmospheric greenhouse effect for
the ice-free parts of the earth (representing 94 percent of the globe). The mean surface
outgoing radiation is 415 W m2, compared with 274 W m2 outgoing at the top of the
atmosphere, leaving to 141 W m2 trapped in the atmosphere – a greenhouse effect of
0.34. Values are only 0.10–0.15 over desert areas, compared with 0.35–0.40 over warm
oceans with deep convection.
Snow and ice – albedo feedback

Snow cover has a spectrally integrated albedo of about 0.87 and bare ice 0.50–0.60,
0 in contrast to tundra or ocean summer values of about 0.15–0.20 in high latitudes (Table
3.10). Rising temperatures reduce the extent of snow and ice, thereby exposing land or
ocean surfaces of much lower albedo. This permits an increase in absorbed solar radia-
11 tion at the surface, which in turn raises air temperature further through turbulent heat
Figure 3.30 Plots of atmospheric vapor pressure (mb) as a function of air temperature (°C) for
mean atmosphere values at the equator (long dashes), 40°N (short dashes) and 75°N
(dotted) at standard pressure levels (marked in hundreds of millibars on each curve),
and the Clausius-Clapeyron es, T curve (C–C). (a) December–January–February, (b)
June–July–August. (From Webster, 1994)
11 Table 3.10 Typical all-wave albedo values of different surfaces during winter and summer.a
Surface Winter (snow-covered surfaces) Summer
Open water 0.07–0.24b 0.05–0.10

Tropical rain forest – 0.07–0.10
Farmland, stony deserts 0.50 0.15–0.17
Rice paddies – 0.12
Light-colored sand deserts – 0.35–0.42
Deciduous/coniferous forests 0.33–0.36 0.14
0 Tundra 0.82 0.17
New snow 0.87 –
Melting old snow – 0.77
Young ice (no snow) 0.21 –
First-year ice (no snow) 0.50–0.60 –
Melting white ice – 0.56–0.68
Old melt pond – 0.15
Sources: modified after Henderson-Sellers and Hughes (1982), Henderson-Sellers and Wilson (1983), Kukla and
Robinson (1980), Hummel and Reck (1979).
Notes
a
Overcast sky conditions typically increase all-wave albedo values by 0.02–0.05.
0 b
The albedo of a water surface is strongly dependent on solar elevation angle and therefore on latitude. Values
are greater than those cited poleward of 50° latitude in winter and 65° latitude in summer.
transfer to the atmosphere from the Earth’s surface. The positive feedback effects of
snow–ice cover and water vapor operate as part of the annual temperature cycle, partic-
ularly in the marginal zones of the seasonal cryosphere.
Model estimates suggest a sea ice–albedo feedback of about 0.18 W m2 K1 (Ingram
et al., 1989), but the results are highly dependent on the parameterization of sea ice
processes, including summer melt ponds, the fraction of open water in leads, and energy
0 storage in brine pockets in the ice, in melt ponds and in the leads (Ebert and Curry, 1993),
as well as on cloud interactions. Cloud cover has a direct effect on clear-sky short-wave
radiation and reflected radiation over snow cover, and an indirect effect through long-
wave radiation feedbacks (Cess et al., 1991).
Cloud feedbacks
Cloud cover affects the surface radiation balance in complex ways that depend on the total
cloud fraction and the type and altitude of the cloud layers. Global cloud cover has a cool-
ing effect through the high albedo of clouds (45 to 50 W m2) and a contrary warm-
0 ing effect through the absorption of terrestrial infrared radiation (30 W m2) (see Table
3.11) (Ramanathan et al., 1989). The contribution of solar radiation, to the net effect is
about 1.7 times that of infrared radiation, according to Hartmann and Doelling (1991).
Overall, the effect of cloud cover relative to clear sky conditions is one of net cooling, aver-
aging about 17 W m2, but there are significant latitudinal and cloud level effects. Table
3.11 summarizes the global radiation budget for clear sky and mean cloud conditions and
the average cloud-forcing components. In general, low and middle cloud layers have a net
cooling effect (short-wave reflectance exceeds the cloud trapping of infrared radiation) and
thin, high cloud a warming effect on the surface (due to the opposite effect). The global
net cloud forcing is between 13 and 21 W m2 during the year. In the tropics, short-
0 wave cloud forcing (60 to 70 W m2) dominates, whereas both short and long-wave
cloud forcing are important in middle latitudes, and poleward of 60° latitude long-wave
cloud forcing (~60 W m2) prevails (Kiehl, 1994). In polar latitudes in winter, cloud cover
11 tends to produce a warming of the surface as a result of the downward infrared radiation
Table 3.11 Components of the Global Radiation Budget for the Earth–atmosphere system
April July October January Annual

Component 1985 1985 1985 1986
Average SW absorbed (W m2) 236.5 234.4 243.0 243.3 239.3

LW emitted (W m2) 234.5 237.5 234.1 231.9 234.5
Net heating (W m2) 2.0 3.1 8.9 11.4 4.8
CLWa (W m2) 31.3 30.1 32.2 30.6 31.1
CSWb (W m2) 45.1 46.7 50.1 51.7 48.4
Net cloud forcing (W m2) 13.8 16.6 17.9 21.1 17.3
Source: from Ramanathan et al. (1989), Harrison et al. (1990).
Notes
a
CLW Cloud long-wave forcing.
b
CSW Cloud short-wave forcing.
increases as cloud amount increases, whereas the opposite is observed for lower surface
albedos. Wendler (1986) refers to this as the “radiation paradox.”
Biosphere feedbacks
A further feedback effect is associated with shifts in vegetation zones and coastlines during
glacial/interglacial cycles or regionally as a result of anthropogenically enhanced deser-
tification and deforestation. Typical surface albedos are shown in Table 3.10. Thus,
lowered glacial sea levels result in less solar radiation absorption at the Earth’s surface.
The reduction in forest cover and its replacement by grassland also increases surface
albedo. Desertification in the tropical savannahs has a similar effect, and the precipita-
tion decrease in the Sahel in the 1970s and 1980s has been attributed to the postulated
biosphere feedback (Charney, 1975). Additional effects involve changes in surface rough-
ness and soil moisture (Garratt, 1993).
The effects of changes in vegetation on global climate have been analyzed mainly
through GCM sensitivity experiments. Most attention has focused on desertification, espe-
cially of the Sahel, and on tropical deforestation, particularly of the Amazon basin. Early
studies concentrated on the responses to a change in albedo (Charney, 1975) but changes
in energy partitioning, especially in the evaporative term, as well as in roughness length
and precipitation interception, also need to be considered (O’Brien, 1996; Zeng and Neelin,
1999). More recently, a biosphere or land surface module has been incorporated into
sophisticated climate models as well as simpler statistical-dynamic models. Various one-
dimensional Surface Vegetation–Atmosphere Transfer Schemes (SVATS) have been used,
including the Biosphere–Atmosphere Transfer Scheme (BATS), the Simple Biosphere
(SiB) model and other derivations. BATS calculates the energy and moisture fluxes of
eighteen different surface types. Canopy–atmosphere and canopy–surface exchanges are
treated. SiB treats the radiative, turbulent heat and momentum fluxes between the vege-
tated surface and the atmosphere via three submodels. A summary of selected results from
these experiments is shown in Table 3.12. Changes in land cover through human activity
are recognized to be of at least regional, and possibly global, significance. The global-
mean radiative forcing due to changes in surface albedo since pre-industrial times is
estimated to be about 0.2 W m2, with a similar contribution from biomass-burning
aerosols (Shine and Forster, 1999), compared to +4 W m–2 for CO2 doubling.
More complex feedbacks with the biosphere involve biogeochemical cycles. Higher
atmospheric concentrations of carbon dioxide may augment photosynthesis, leading to
increases in net primary production and carbon storage on land (a negative feedback).
Methane (CH4), which has contributed about 20 percent (0.5 W m2) of the increase in
radiative forcing due to anthropogenic trace gas increases since pre-industrial times, is
11 Table 3.12 Model experiments of (1) Amazonian deforestation and (2) African desertification
Model Ts (K) Ta (K) P (cm/yr) E (cm/yr) Reference
(1) Deforestation
NCAR CCM – 1 to 3 0 20 Dickinson and
Henderson-Sellers (1988)
COLA – 2 64 50 Nobre et al. (1991)
UKMO – 2.1 30 20 Lean and Rowntree (1993)
SDM – 1.2 18 19 Varejao-Silva et al. (1998)
0 (2) Desertification
COLA – 0.5 55 26 Xue and Shukla (1993)
COLA 0.3 to 1.2 – 11 to 55 – Dirmeyer and Shukla
(1996)
SDM 1.1 0.7 14 16 Varejao-Silva et al. (1998)
Source: from Verajao-Silva et al. (1998).
Notes
NCAR CCM US National Center for Atmospheric Research, Community Climate Model; COLA Center for
Ocean–Land–Atmosphere; UKMO UK Meteorological Office; SDM Statistical dynamical model.
0
released from natural wetlands, bogs, and tundra in addition to releases from rice paddies,
enteric fermentation in ruminants, biomass burning, gas drilling, coal mining, and anaer-
obic decay in landfills (Aber, 1992). Warming in northern high latitudes may potentially
deepen the seasonal active layer overlying permanently frozen ground (permafrost),
thereby enhancing anaerobic methanogenesis in the peatlands of sub-Arctic and Arctic
North America and Eurasia. Thawing of the permafrost itself may also release some of
the methane stored as clathrates. Thus methane concentrations may increase in response
to global warming, providing a positive feedback. It is also possible that drier regimes in
some regions may lower water tables in wetlands, with the opposite result.
0
3.6 General circulation models
Numerical models of the atmospheric circulation, commonly referred to as General
Circulation Models (GCMs), were first developed in the 1960s, building on work by
Phillips (1956) and many others. The essence of an atmospheric GCM (AGCM) is that
the basic equations describing the atmosphere’s large-scale motion and energetics are
solved iteratively over a global grid with numerous vertical layers and a time step of a
few minutes. For numerical weather prediction (NWP), AGCMs typically have high spatial
resolution and are run for time periods of a few days, whereas in climate applications the
0 spatial resolution is lower and the models are integrated over years.
The behavior of the atmosphere is prescribed by the basic laws of hydrodynamics and
thermodynamics which concern the following principles:
1 The conservation of mass and of water substance.

2 The first law of thermodynamics – that internal energy can change only through the
addition/removal of heat or by the system performing work.
3 Newton’s second law of thermodynamics – that the momentum of an air parcel is
changed through the vector sum of the forces acting on it.
0 Using pressure (p) as the vertical coordinate, the momentum equation can be written:

V
V
VⴢV fkV DM
11
t
p
where V is the horizontal velocity, is the vertical motion, f Coriolis parameter, k is
a unit vertical vector, is the gradient operator, gz is the geopotential and DM is
the dissipation of momentum by friction.
The expanded equations for the zonal and meridional components of motion are:
du
dt 冢
fu
tan &
r 冣
v=
1 1
p
r cos &

F
and
dv
dt 冢
fu
tan &
r 冣
u=
1
p

&
F&
where longitude, & latitude, and r the Earth’s radius;

d·
u
v

= w
dt
t r cos &
r
&
z
The mass continuity equation is:

.V =0

p
The equation for continuity of water substance is:

q
q
= V.q E C Dq

t
p
where q specific humidity, E evaporation, C condensation and Dq is the diffu-

sion of moisture.
The thermodynamic equation is:

T

t
V.T
p冢
T
T

p
R
n
cp
冣
LE
cp
DH
where 0.286, cp is the specific heat at constant pressure, and DH the diffusion of
heat. The terms Rn net radiation, LE latent heat of evaporation, C condensation,
and the diffusion components are often referred to as the “model physics.” They are deter-
mined from equations for radiative transfer through the atmosphere, for the processes of
evaporation/condensation and precipitation, and for the turbulent transfers of heat,
momentum, and moisture from the surface.
The above equations contain time derivatives of the independent variables and so are
prognostic. In addition, there is the diagnostic equation for hydrostatic equilibrium:

p
g

z
which, combined with the ideal gas equation, p RT, gives:

p gp
=

z RT
These are known as the “primitive equations” because they filter out sound and gravity
waves that are present if the complete “exact equations” are used. The retention of these
11 Table 3.13 Stages in the development of general circulation models
Stage Characteristics
First generation Grid point. Low spatial and vertical resolution. Idealized or simple
geography. Prescribed SSTs, albedo and cloud. Run in perpetual season
mode. Best results for northern hemisphere winter circulation
Improvements Annual cycle. Hydrological cycle. Cloud calculated at a few levels. Diurnal
cycle
Second generation Spectral representation. Sigma coordinates replace height levels. Improved
0 spatial and vertical resolution. Improved physics – radiation bands, wave
drag. Swamp ocean (energy budget). Time transient experiments
Third generation Coupled atmosphere – ocean (mixed-layer ocean). Improved physics – snow
cover, simple sea ice; cloud physics; trace gases. Coupled AGCM–OGCM
effects in the exact equations permits the rapid growth of errors, and this largely accounts
for the failure of the first attempt at numerical weather prediction by Richardson (1922).
GCMs have evolved steadily over the last thirty years and at least three generations of
successive developments and improvements can be identified. Table 3.13 summarizes the
0 major characteristics of these several stages. Key physical processes treated in atmos-
pheric GCMs include radiative transfers, partitioned into short and long-wave components
and interacting with atmospheric gases, aerosols and clouds; convective and stratiform
cloud development and associated precipitation; convective adjustment and lapse rate
modification; surface and boundary layer fluxes (heat, momentum, and moisture); and
surface hydrology, including snow storage, soil moisture, and roughness. These physical
processes interact with the coupled dynamical and thermodynamic equations, as shown
schematically in Figure 3.31. However, many physical processes operate on scales that
cannot be adequately treated by the finite differences resolved in GCMs. Examples of
subgrid scale processes include cumulus convection, surface heat and moisture fluxes, and
0 terrain-induced vertical motion. Such processes need to be represented by parameteriza-
tions in which the unresolved processes and properties are linked to model variables by
means of simple mathematical relationships (Meehl, 1984; Blackmon, 1986; Washington
and Parkinson, 1986; Cubasch and Cess, 1990; McGuffie and Henderson-Sellers, 1997;
Kiehl, 1992; Trenberth, 1992).
The primitive equations (Lorenz, 1967) are solved for short time steps, either over a
finite difference grid, or by using a spectral harmonic expansion of the scalar fields (see
Appendix 4.1). Where there is a low-order spectral truncation, however, sharp gradients
in a particular field (and in the surface topography) give rise to oscillatory patterns of
over- and underestimated values of the function (Figure 3.32). These are known as Gibbs
0 oscillations and require special functions to reduce their effects (Navarra et al., 1994;
Holzer, 1996; Lindberg and Broccoli,1996), otherwise spurious negative values may occur
in properties that are always positive – such as water vapor content. The horizontal grid
resolution in early GCMs used for climate studies was of the order of 5º–8º latitude and
is currently between 1º and 2.5º latitude, corresponding to spectral truncations of R15–R20
and T63–T106, respectively (see Appendix 4.1). Coarse resolution in models can have
important effects on the simulation of precipitation fields, for example.
The vertical coordinate can be represented in height (z) coordinates, or in terms of
pressure (Hack, 1992). A transformed pressure coordinate ( ) is now common because
of its convenience:
0
p/ps
where ps is surface pressure. The surfaces follow the terrain boundary, thereby avoiding
11 the problems that arise with constant z surfaces which intersect mountains. The surfaces
Figure 3.31 The coupling of physical processes in a GCM. (From Peixoto and Oort, 1992)
are closely spaced in the lower levels and more widely spaced in the troposphere and
stratosphere. The upper boundary may be above 25 mb (~ 25 km) to avoid systematic
errors which result when stratospheric processes are omitted.
The fixed surface boundary conditions are the distribution and height of land areas.
Some models also prescribe surface albedo and roughness length according to land cover
type, as well as climatological sea surface temperatures and sea ice extent. The eventual
goal is to predict all of these properties via prognostic equations. In addition, the vertical
velocity is zero at the highest and lowest surfaces.
Model simulations use initial conditions – an array of wind, pressure, temperature, and
moisture data at all grid points – to begin the forward integration of the prognostic equa-
tions. These initial fields must be balanced to avoid error propagation. The time integration
of the equations cannot use simple time-step measurements with centered finite differ-
ences in space. Computational instability results if this procedure is applied to equations
that represent advection of a given property. Instead, a “leapfrog” scheme is used where
finite differences are centered in time (Gadd, 1981; Hack, 1992). For example:
pn = pn2 2t 冢
p
t 冣
n1
There is a limit on the time step that can be used which depends on the grid spacing and
the maximum speed of advection that is being simulated.
Simulations are performed in a variety of ways. Most early studies carried out a sensi-
tivity analysis to assess the possible effect of changes in an individual factor or a complex
of factors. Such sensitivity experiments have been used to examine the possible role in cli-
11
0
Figure 3.32 Examples of Gibbs Oscillations caused by spectral trunctions. Above An idealistic square
mountain. Below Longitudinal section at 90°E through the Himalayas. Scripps: orographic
data; spectrally truncated orographies at T30 and T42. (From Navarra et al., 1994)
mate change of volcanic dust clouds, nuclear winter (smoke) scenarios, deforestation, deser-
tification, altered solar constant, changes in the extent of sea ice or snow cover, and so on
(Barry, 1975). Recent work investigates the sensitivity to global changes in leaf area index
(Chase et al., 1996). Simulations of the climates of past geological epochs such as the
Cretaceous warm period, the last glacial maximum, and the Holocene thermal maximum
0 are also carried out via sensitivity studies of changed boundary conditions (Schlesinger,
1988; Barry, 1997). True transient climate experiments have become possible only with
coupled ocean–atmosphere models and greatly augmented computer capability (Manabe
and Bryan, 1969; Meehl, 1992; Neelin et al., 1994). Models are now integrated 100 years
or more to assess the evolution of global climate during progressive increases of green-
house gas concentrations. Results are detailed in the assessments of the Intergovernmental
Panel on Climate Change (IPCC) (Gates et al., 1992; Kattenberg et al., 1996). To assess
low-frequency variability, several 1,000 year integrations using coupled atmosphere–
ocean–land surface models have recently been performed (Manabe and Stouffer, 1996).
Models are typically evaluated in terms of their ability to simulate the mean features
0 and variances of global climate variables (pressure, wind, temperature, precipitation, etc.),
their seasonal cycles, and their geographical distributions (Gates et al., 1990). The Köppen
climate classification, which combines temperature and precipitation categories, has proved
11 a useful diagnostic tool in this regard. Lohmann et al. (1993) show how the area of large-
scale climatic regions may alter under warming scenarios. More recently, the synoptic
variability of these elements has been compared with observed synoptic conditions using
empirical orthogonal functions (EOFs) (Crane and Barry, 1988) and catalogs of synoptic
circulation patterns (Hulme et al., 1993, for the British Isles, and McKendry et al., 1996,
for western North America) (see also section 7.4). Model improvements typically occur
in an incremental manner, based on comparative assessments of changes in physical
processes and grid resolutions (Boville, 1991; Marshall et al., 1997). Models are also now
assessed for their ability to simulate intraseasonal variability such as blocking modes and
the interannual variation of ENSO events. There are several major projects designed to
intercompare model capabilities. They include the Atmospheric Model Intercomparison
Program (AMIP), which intercompares various aspects of some twenty-five to thirty GCMs
(Gates, 1992; Phillips, 1994; Lau et al., 1996) and the Program to Intercompare Land
Process Schemes (PILPS) (Henderson-Sellers et al., 1993), as well as research group
studies of individual model elements or regional climate simulation (Hay et al., 1992;
McGinnis and Crane, 1994; Randall, 1996; Kaurola, 1997). Examples of studies of the
effect of model parameterizations include the intercomparison of ice–albedo and cloud
feedbacks (Cess et al., 1990, 1991, 1996). The latest series of intercomparisons is being
performed under the Coupled Model Intercomparison Project. Control runs as well as
integrations for increasing carbon dioxide from many global coupled models are being
intercompared (Meehl et al., 2000a, b).
An important issue in climate assessment and impact studies is the ability of GCMs to
provide useful results for specific regions. Approaches to this problem include the use of
high-resolution mesoscale models nested within a GCM and statistical downscaling of the
GCM results by the use of relationships between large-scale parameters and regional
climate conditions as currently observed (Giorgi and Mearns, 1991; von Storch, 1995;
Hewitson and Crane, 1996). The former has found application in the treatment of
orographic precipitation, see Figure 3.33 for example, while the latter is used in climate
change scenarios.
Operational numerical weather prediction (NWP) has a long history but the model
physics, horizontal and vertical resolution, and other aspects are constantly being upgraded
(Cullen, 1993; McPherson, 1994). This has limited the use of such outputs for climato-
logical analyses. Recognition of this fact led to a plan to reanalyze the archives of synoptic,
upper air, and satellite input data, incorporating subsequent corrections and standardized
procedures for data assimilation, together with a fixed model. Currently, this reanalysis
is being performed in the United States by the National Centers for Environmental
Prediction (NCEP) (Kalnay et al., 1996) and at the Data Assimilation Office, NASA
(Schubert et al., 1993), as well as by the European Centre for Medium Range Weather
Forecasts (ECMWF), European Reanalysis Agency (ERA) in Reading, UK. Studies
utilizing the available results are already under way (WCRP, 1998).
3.7 The global circulation – description

The global circulation in the troposphere displays an essentially tripartite structure from
poles to equator: a polar vortex, subtropical high pressure, and a weak equatorial low-
pressure trough. At sea level the pattern is modified in that the high-latitude minima are
over the subpolar regions, and in the northern hemisphere the land masses disrupt the
zonal regularity (Figure 3.34).
3.7.1 Subtropical high pressure

The most prominent and persistent features of the mean tropospheric circulation in both
hemispheres are the subtropical high-pressure belts. Their locations vary seasonally as a
function of the meridional temperature gradient in the troposphere (Flohn and Korff,
11
Figure 3.33 The simulation of mean precipitation for January over Europe as represented in (a) obser-
11 vations 0.5° 0.5°) from Legates and Wilmott (1990); (b) mesoscale model nested in the
NCAR CCM-1; (c) the CCM-1 (the MM-4 R15 truncation). (From Giorgi et al. 1990)
Figure 3.34 Mean sea-level pressure and winds for January and July. (After Liljequist, 1970)
1969). As the difference in equator–pole temperature in the 700–300 mb layer increases

in the winter months the subtropical anticyclones shift equatorward, and vice versa (Figure
3.35). This figure also illustrates the asymmetry between the hemispheres, associated with
the stronger southern hemisphere circulation, which results in a displacement northward
of the annual average equatorial trough to 6°N. The axes of the subtropical anticyclone
cells slope generally equatorward and westward with height, towards the warm air, as a
consequence of the hydrostatic relationship (p. 162).
The MSL subtropical highs in the northern hemisphere are best developed in July, when
they attain their northernmost positions (Figure 3.34). The highest pressures in July exceed
1,027 mb in both the North Atlantic and the North Pacific oceans. In the southern hemi-
sphere the centers are weaker and show less seasonal variation in amplitude and location.
The South Atlantic cell is stronger in winter (at 30°S) than in summer (at 35°S), whereas
11
0
Figure 3.35 The meridional temperature for the 300–700 mb layer in the previous month plotted
against the latitude of the center of the subtropical high-pressure belt in each hemi-
sphere. A constant vertical lapse rate is assumed. (From Flohn and Korff, 1969)
that in the southeastern Pacific varies little, with maximum mean monthly pressures of
1,022 mb. The seasonal range of latitude is also only 5° for the southeast Pacific cell,
although the ridge axis in the western South Pacific fluctuates over 10° latitude, with an
extreme range of 20°.
The occurrence and strength of the subtropical anticyclones present a number of prob-
0 lems, in particular the fact that they are stronger in summer over the northern hemisphere.
The arid regions of the subtropical anticyclones experience high net radiative heat loss
which is balanced by adiabatic warming in the descending air. There is little descent asso-
ciated with the mean Hadley cell over the northern subtropics in summer, for example,
yet North African precipitation is minimal. Considerable attention has been given to the
ideas first proposed by Charney (1975) concerning enhanced adiabatic warming through
subsidence related to anthropogenically induced biosphere–albedo feedback as a result of
overgrazing and desertification, but this does not address the fundamental question. A
recent hypothesis proposes that the cells over the eastern oceans and adjacent subtropical
deserts are closely linked with the regions of monsoon heating over the continents to their
0 east (Hoskins, 1996; Rodwell and Hoskins, 1996). Calculations of the observed vertical
velocity in the middle troposphere from ECMWF data for June–July–August 1983–88,
and the associated contributions of the diabatic heating and horizontal advection to the
thermodynamic energy balance, show that in the principal areas of descent in the northern
subtropics the horizontal advection contribution is about twice the adiabatic cooling term.
Moreover, the spatial structure of descent regions more closely resembles that of the hori-
zontal advection (Figure 3.36). The terms illustrated in the figure are:
Q /cp = V.p T ( p/po)k

/
p
0 where the term on the left is the diabatic heating and the two on the right are the hori-
zontal advection and the vertical advection, respectively. The result is represented in the
idealized simulation of tropical heating in the Asian summer monsoon by Gill (1980),
11 which indicates weak Rossby wave descent west of the heat source, but the recent work
Figure 3.36 Fields of vertical velocity and thermodynamic budget terms for June–August 1983–88
from ECMWF analyses. (a) Vertical velocity at 477 mb, with contour interval of
0.5 mb hr1. (b) Diabatic heating. (c) Horizontal advection (V.pT). (d) Vertical
advection (( p/po)
/
p). (e) Pressure and horizontal winds on the 325K isen-
tropic surface (40 mb contour interval; H = pressure > 590 mb. The energy terms are
at 477 mb, with contour intervals 0.5 K day1. (From Rodwell and Hoskins, 1996)
provides a more comprehensive explanation. There are four components of the proposed
mechanism:
1 Heating associated with the continental monsoons moves poleward in spring–summer,

generating an enhanced sinking of air on the western and poleward margins. The
descending air is fed by the mid-latitude westerlies, rather than as the sinking arm
of a simple vertical plane (Hadley or Walker) cell.
2 The descending air suppresses convection and permits enhanced radiative cooling,
which acts as a positive feedback on the descent.
3 Strong equatorward flow is set up below the level of maximum descent. Theory shows
that this is accounted for by the vorticity balance: v f w/ z and is equivalent to
anticyclonic vorticity generated in the region being displaced westward as a conse-
quence of the effect.
4 The equatorward flow causes cold upwelling in the coastal waters through Ekman
drift away from the coast, further suppressing any convection.
11 In the non-linear primitive equation model of Rodwell and Hoskins (1996) the pre-
scribed forcing by monsoon heating induces a Rossby wave to the west over the subtropics
which extends poleward into mid-latitudes and warms the atmosphere. The descending
air originates on the southern flank of the westerlies and adiabatic descent is thermo-
dynamically balanced by cold air advection in the mean. The local orography upstream
in northwest Africa (the Atlas–Ahaggar mountains) is considered to be important in the
localization of adiabatic descent over the eastern Sahara–Mediterranean Sea (Semazzi
and Sun, 1997) and a similar relationship is suggested between the Zagros mountains and
descending air over the Kyzlkum desert to the southeast of the Aral Sea. In the eastern
0 Mediterranean the equatorward flow element of the dynamic climatology is also expressed
in the summer occurrence of the northerly Etesian winds over the Aegean Sea.
3.7.2 Low-latitude circulation

The vertical structure of winds in low latitudes displays a variety of patterns (Figure 3.37).
In some sectors and seasons, easterlies are prevalent, although their speed decreases with
height. In other situations, tropical easterlies are overlain by mid-latitude westerlies; these
extend equatorward in the upper troposphere during the cold season. In the summer
monsoon regimes, low-level westerlies are overlain by tropical easterlies. The causes of
0 this diversity are examined below. We begin by describing the tropical easterlies, known
in terms of surface observations as the trade winds.
The trades and their structure

The trade wind systems of the maritime tropics extend over almost 25 percent of the
earth’s surface (Figure 3.38) (Crowe, 1949, 1950, 1971). They were first described by
Figure 3.37 Schematic vertical structure of low-latitude zonal winds. a, b The trades in summer.
11 c The trades in winter overlain by westerlies. d Summer monsoon areas
Figure 3.38 The trade wind systems illustrated by schematic surface streamlines for January and
July. Shaded areas denote relative directional constancy of all winds >3 m s–1 from
the predominant quadrant (percent): 50 (horizontal lines), 70 (stipple), 90 (dark stipple).
(From Crowe, 1950)
Edmond Halley (1686); both he and Hadley (1735) attempted to account for them within
a theory of the global atmospheric circulation. The tropical easterlies are essentially
barotropic, implying an absence of thermal forces, as noted by Rossby (1949, p. 181). They
originate from large-scale subsidence and divergence in the eastern sectors of the subtrop-
ical high-pressure cells. Measurements in the central North Atlantic, 10°–15°N, 30°–40°W,
indicate subsidence of 200 m day1. Brümmer et al. (1974) found that 15 m level winds
are 60–80 percent of the geostrophic value and directed 27º across the isobars towards lower
pressure. Figure 3.38 indicates flow from the northeast (southeast) quadrant in the north-
ern (southern) hemisphere. There is remarkable constancy of direction and steadiness in the
speed; surface winds typically average 7 m s1. Figure 3.39 illustrates annual averages and
standard deviations of the u and v components over the tropical Pacific Ocean based on 5
million ship observations. The u component shows larger variability on the margins of the
“core” regions, in the western equatorial Pacific and in the South Pacific, 20°–30°S. The v
component is most variable about 5°–10°N in the central–eastern equatorial Pacific.
The air subsiding over the eastern parts of the subtropical oceans creates a marked
trade wind inversion that was first identified in 1856, through studies on the Peak of
Teneriffe in the Canary Islands. The inversion base is typically encountered around
1,000–2,000 m altitude, although it is lower over the cold current off Northwest Africa.
Soundings with high vertical resolution made in July 1987 at San Nicolas Island,
California, where the sea surface is similarly cold and there is persistent stratocumulus,
indicate a maximum inversion frequency near 860 mb (1,400 m) with strong inversions
at 915 mb (850 m) (Schubert et al., 1995). However, similar soundings taken during June
1992 at Port Santo in the Madeira Islands (33ºN, 16ºW) show that the most probable
11
Figure 3.39 Annual averages and standard deviations of the u and v components of surface wind
over the tropical Pacific Ocean, 1950–1972. (From Barnett, 1977)
inversion height is near 800 mb (2,000 m) with the stronger ones at 860 mb. Likewise,
soundings during November 1992–February 1993 at Kwajelein Atoll (9ºN, 168ºE)
resemble those at Porto Santo. Aircraft dropwindsonde measurements in the wintertime
Pacific trades south of Hawaii show that, contrary to earlier ideas, the inversion height
0 (about 850 mb) does not increase equatorward (Firestone and Albrecht, 1986). Rather,
subsidence weakens and the frequency of inversions decreases somewhat downwind,
allowing penetrating cumulus towers to transport moisture into the drier air above. Purely
thermodynamic models of the trade wind boundary layer predict that it deepens with
increasing local sea surface temperature and decreasing subsidence, implying a down-
stream slope of about 2,000 m per 1,000 km. In fact, there is typically a gentle slope of
~300 m per 1,000 km until the intertropical convergence zone is reached. Schubert et al.
(1995) propose that the subtropical inversion is extended into the tropics dynamically. A
small poleward movement of low-level equatorial air along isentropic surfaces, which
generates a weak westerly flow (~1 m s1), modifies significantly the static stability. The
0 subtropical inversion is weakened, while a stable inversion layer is extended into the
tropics. Hence the inversion height is determined by horizontally averaged values of sea
surface temperature, divergence, and atmosphere structure above the inversion, and not
11 by local conditions. In the central Pacific north of the equator, deep convective activity
peaks near 7°N, but during January–February 1979 there were still 50 percent of sound-
ings with low-level inversions near the equator. The stability of the inversion is sufficient
to suppress convection, but destabilization can occur within a few hours of vertical
stretching of the inversion layer, not by deepening of the boundary layer (Firestone and
Albrecht, 1986).
The seasonal expansions and contractions of the global trade wind belts are illustrated in
Figure 3.38 (Crowe, 1950, 1965). In general, the margins extend poleward in summer and
contract in winter, in association with shifts of the subtropical anticyclones. The trades also
extend farther westward during the cold season as the high-pressure cells strengthen.
However, the “core regions” are essentially quasi-permanent. Thus the northeast and south-
east trade wind systems are seasonally out of phase; the former reach a maximum in
January–February and a minimum six months later. Wyrtki and Meyers (1976) show that the
southeast trades in the Pacific Ocean are more extensive than the northeast trades (Figure
3.40). Barnett (1977) also finds that variations in the wind field just north (south) of
the northeast (southeast) trades in the Pacific are seasonally in phase with trades in the oppo-
site hemisphere. This demonstrates the close coupling of the wind systems with
the seasonal shifts of the high-pressure cells. Monthly departures from the monthly means of
the u-component winds over the Pacific show little correlation between 5°–19°N and
1°–15°S, although the v components (towards the equator) are quite well correlated (Reiter,
1979). Surges in the meridional components of the North and South Pacific trades are also
well correlated with precipitation anomalies in the equatorial central Pacific (the Line
Islands) (Reiter, 1978). In the tropical Atlantic, however, the v-component anomalies in the
northeast and southeast trades are independent of one another and there are no correlations
in meridional wind components between the Atlantic and Pacific Ocean trade winds (Reiter,
1979). Wyrtki and Meyers (1976) found no relationship (other than the annual cycle between
the strength or area of the southeast and northeast trades in the Pacific) (Figure 3.37), but the
u components of the Pacific trade winds do show a coupling between hemispheres on the
interannual time scales (Barnett, 1977). This is best developed west of the dateline, away
from the core regions of maximum u. In contrast, the largest variations in the v component
are found over the eastern and central Pacific and in the southwest Pacific.
Figure 3.40 The monthly variation in area of the trade winds over the Pacific Ocean. (Wyrtki and
Meyers, 1976)
11 Equatorial westerlies
Westerly winds are present seasonally in the equatorial lower troposphere over much of
the eastern hemisphere. Flohn (1960) describes such climatological equatorial westerlies
across Africa, southern Asia and the Indian Ocean to New Guinea in the northern summer
and primarily in the Australia–Indonesia sector in the southern summer (Figure 3.41).
They originate in response to the poleward displacement of the equatorial trough over the
continents during the summer monsoons, discussed below. Over the Indian Ocean sector
they are sometimes characterized as southeasterly trades that become southwesterly in
response to the Coriolis effect after crossing the equator, yet as Flohn (1960) points out
0 southwesterly surface winds at 2°–3°S were noted by Halley (1686) from sailing ship
reports. Cross-sections of the wind and temperature structure along longitude 90°E show
the seasonal shift between easterly trades and equatorial westerlies.
In the Australian sector, low-level westerlies develop first over the western South Pacific
(0°–10°S, 160°E–170°W), responding to the strengthening of the North Pacific easterly
trades and cross-equatorial northerly flow around 170ºE (Murakami and Sumi, 1982).
Cross-equatorial flow from the northern hemisphere Hadley cell then extends to 10ºS
between 100ºE and 170ºW.
A dynamic explanation of equatorial westerly flow is offered by Tomas and Webster
(1997). In regions where there is a strong pressure gradient from the winter to summer
0 hemisphere, shear in the zonal wind required to balance the cross-equatorial advection
of absolute vorticity ( f ) causes the wind to become westerly (see Figure 3.42). The
westerly winds must also strengthen rapidly north of the zero contour of ( f ); there are
corresponding westerlies in the southern hemisphere summer. The occurrences of wester-
lies in the Indian Ocean in July, south of the zero line of absolute vorticity, is a result of
the additional influence of a west–east pressure gradient along the equator. There are also
synoptic-scale wind burst events in the western equatorial Pacific (Kiladis et al., 1994).
These play an important role in El Niño processes; they are discussed in section 5.3.
Figure 3.41 Equatorial westerlies in any layer below 3 km in January and July. (After Flohn, 1960;
11 from Barry and Chorley, 1998)
Figure 3.42 Schematic illustration of the dynamic balance in the summer hemisphere about the
contour of zero absolute vorticity (' = f) . Left The divergent component of wind,
V. Center The rotational wind component V and low-level westerlies. Right the lati-
tudinal profile of relative vorticity and of zonal wind, u. (Tomas and Webster, 1997)
The near-equatorial trough

Considerable confusion exists in the literature as to the nature of the equatorial pressure
minimum and its associated climatic characteristics. The problem arises in part through
attempts to reconcile several distinct phenomena within a single model. Thus zones of
wind confluence, mass convergence, frontal boundaries, cloud or precipitation maxima,
and pressure minima are treated as if they were synonymous, or mutually consistent.
Moreover, some of these features are more appropriately recognized as climatologically
averaged variables rather than as synoptic entities.
Mariners have long identified a low-latitude zone of light and variable winds – the
Doldrums – which are principally located in the eastern parts of the equatorial Pacific
and Atlantic Oceans. This zone locally separates the trade wind systems of the two hemi-
spheres, but elsewhere the trades may be convergent, or at least confluent. Figure 3.43
shows the average position of the tropical confluence in the Pacific by the zero contour
of the v component dividing northward () and southward () flow. The boundary is
close to the equator near New Guinea (140°E) but reaches 10°N in the eastern Pacific
(Barnett, 1977), closely following the mean location of the isotherms of maximum SST.
Pronounced angular confluence gives rise to the identification of an Intertropical
Convergence Zone (ITCZ), but in other sectors there is only a broad near-Equatorial
Trough of low pressure. In simple models, trade wind confluence within the equatorial
trough is considered to create an ITCZ, with an associated maximum of cloudiness and
precipitation. However, these are often quite distinct features. Moreover, when equatorial
westerlies develop over land sectors, forming the respective summer monsoon flows, the
main convergence is then between trades on the poleward side and westerlies on the equa-
torial side. Sadler (1975) refers to this feature a “monsoon trough.” The axis of maximum
cloudiness is often located within the easterly flow several degrees equatorward of the
line of wind shear. In the central equatorial Pacific during December 1977–January 1978,
Ramage et al. (1981) report that the North Pacific trades were heated and moistened from
below whereas their South Pacific counterparts underwent low-level divergence and slight
cooling in response to this SST gradient. While the strength of the low-level convergence
was related to the trade wind strength, mid and upper-level circulations apparently modu-
lated the distribution of bad weather. Nevertheless, low-level convergence was also
unrelated to upper-level divergence. The convergence zone between Hawaii and Christmas
11
0
Figure 3.43 The tropical wind confluence in the Pacific Ocean, based on the first EOF of the v
components of wind accounting for 67 percent of the variance. (Barnett, 1977)
Island during the experiment was made up of alternating strips of convergence and diver-
gence. Occasional dry zones of subsiding air occurred within the ITCZ, suggesting
mesoscale structures.
Convergence within the near-Equatorial Trough or along the ITCZ thus shows consid-
erable spatial and temporal variation. Low-latitude convergence can be studied indirectly
by identifying large-scale convective systems known as cloud clusters that extend several
0 hundred kilometers horizontally and vertically into the upper troposphere. Indices of such
tropical convective systems, include low values of outgoing long-wave radiation (OLR)
(Gruber and Krueger, 1984), indicative of cold, high cloud-tops associated with cumu-
lonimbus anvils and cirrus cloud shields, and deep highly reflective clouds (HRC) mapped
from visible and infrared satellite images, respectively, to filter out low to middle-level
cloud systems. Garcia (1985) produced a daily HRC data set for 1977–87 identifying
deep “convective systems extending at least 200 km horizontally.” These two indices are
intercompared by Waliser et al. (1993) for 1974–87 and it is concluded that the HRC
maps delimit the deep convection within the tropical convergence zones better than the
OLR data, and compare well with convective cloud areas identified on monthly time
0 scales in the ISCCP data.
A climatology of the ITCZ, based on the HRC data, is shown in Figure 3.44. The ITCZ
in the Atlantic and eastern Pacific remains entirely north of the equator, following a nearly
sinusoidal pattern from near the equator in February–March, when it is weakest, to near
10°N in August–September, when it is strongest. Over Africa it migrates from 10°S in
the austral summer to 8°N in the boreal summer. In the central Pacific there is a clearly
marked double structure, especially evident in April and December, with the summer
hemisphere zone predominating. Over the Indian Ocean there are also two bands, one
between about 2°S and 8°S, and the other associated with the Indian monsoon between
50°N and 20°N. The western Pacific sector also reflects the influence of the Australasian
0 monsoon, although the South Pacific Convergence Zone (SPCZ), discussed below, is a
complex phenomenon. Finally, over South America, the ITCZ is near 10°S in the
austral winter but, after its northward movement to 10°N in July–August, it decays and
11 reappears again south of the equator in the austral spring.
(a)
Figure 3.44 A climatology of the ITCZ: (a) mean monthly frequency of HRC days in January,
April, July, and October, and (b) time–latitude plots for seven regions and a global
average (see p. 179). The white line shows the monthly mean position of the ITCZ.
(From Waliser and Gautier, 1993)
The meridional distributions of OLR, SST and mean sea-level pressure (SLP) across
the equator in the Indian Ocean (55º–85ºE), eastern Pacific (120º–90ºW) and eastern
Atlantic (30º–0ºW) (Figure 3.45) show some striking contrasts. In the eastern Atlantic
and Pacific Oceans, in February, maximum SST, minimum OLR, and a broad minimum
of SLP occur about 50ºN. In July, however, the convection (OLR) in these areas is equa-
torward of the SST maxima and the pressure minima and a similar pattern occurs over
the Indian Ocean in February. There is a closer colocation of the three variables when
11
(b)
Figure 3.44 (cont.)
0 convection is weak (larger OLR) and when there is an insignificant cross-equatorial pres-
sure gradient (Tomas and Webster, 1997). Over the Indian Ocean there are dual SST
maxima. In July one is over the equator, while the OLR minimum is at 5ºN.
In the Atlantic and Pacific Ocean sectors the ITCZ is located north of the equator year-
round. In the southwest Pacific, however, there is a pronounced convergence zone and
cloud band, referred to as the South Pacific Convergence Zone, which extends south-
eastward from New Guinea to about 30°S, 160°W (Figure 3.44). This is primarily a feature
of the austral summer, when a monsoon low-pressure trough is present over northern
Australia and southeasterly trades flow towards the center (Vincent, 1994). The western
section of the SPCZ is located over the western equatorial Pacific “warm pool” where
0 surface temperatures exceed 29°C. Gradients of the sea surface temperature appear to
generate pressure gradients which force the low-level winds, leading to convergence of
moisture and cloud development (Kiladis et al., 1989). The southeastward extension of
11 the SPCZ cloud band is associated with tropical–mid-latitude interactions. Some studies
Figure 3.45 The meridional distributions of convection as indicated by OLR (W m2) (thin solid
line), SST (°C) (heavy solid line), maxima stippled, and mean sea-level pressure, SP
(mb) (heavy dashed line). Upper Indian Ocean (55°–85°E). Middle Eastern Pacific
Ocean (120°–90°W). Lower Eastern Atlantic Ocean (30°W–0°). Left February 1992.
Right July 1992. (From Tomas and Webster, 1997)
identify the role of disturbances propagating toward lower latitudes near New Zealand
and strengthening the convergence zone between themselves and the south-east Pacific
high (Kiladis and Weickmann, 1992). In addition, Kiladis et al. (1989) note that storm
tracks propagate from lower to higher latitudes when the ITCZ interacts with a trough in
the mid-latitude westerlies.
The zonally averaged meridional transports of energy in the Earth–atmosphere system
show a zero value at about 8°S in February and 15°N in August (Figure 3.46). Riehl and
Simpson (1979) point out that this intertropical zone always lies in the winter hemisphere,
so that the net energy source between these latitudes is always exported toward the winter
pole. Hence the equatorial trough (the line of zero meridional energy transport) demar-
cates the export of energy from the tropics to the summer and winter poles, respectively.
11
Figure 3.46 Zonally averaged

meridional transports of energy
in the Earth–atmosphere system,
illustrating the location of the
0 equatorial trough. (From Riehl
and Simpson, 1979)
The seasonal latitudinal shifts of the equatorial trough are only half those of the over-
head sun (between the Tropics of Cancer and Capricorn). Riehl and Simpson propose that
the energy requirement of the high-latitude winter hemisphere exercises a constraint on
the displacement of the trough.
The climatological structure of the equatorial trough in July differs between “oceanic”
and “continental” sectors (Riehl and Simpson, 1979). Over Africa and Asia (30°W east-
0 ward to 150°E) temperature cross-sections show a pronounced low-level warm anomaly
from 8° south to 15° north of the trough compared with mean conditions ±20° latitude
about the trough (Figure 3.47). For the ocean sector (150°E eastward to 30°W), there are
very weak temperature anomalies, but specific humidities are highest in the trough zone.
Highest moisture contents over the continental sector are on the equatorward side of the
trough, which is displaced northward. The vertical structure of the ITCZ from compos-
ites during field experiments in the central Atlantic and central Pacific indicates a warm
core near the surface and at upper levels, with an intermediate cold core (Estoque, 1975).
The cold core at middle levels and warm upper levels is confirmed by dropsondes and
soundings made over the eastern Pacific during the First GARP Global Experiment
0 (FGGE) (Fernandez-Partagas and Estoque, 1985). The structure resembles that of the clas-
sical easterly wave (see section 6.5), but the observational period featured only a vigorous
ITCZ along 11ºN westward from the west coast of Mexico, with no wave disturbances.
11 The upper-level warming is attributed to latent heat release in cumulonimbus towers.
Figure 3.47 Cross-sections of temperature anomaly

(°C) across the equatorial trough zone in July; lati-
tudes are relative to the trough. Temperature
deviations are with respect to a mean sounding
within ± 20° latitude of the trough. Above The
“continental” sector from 30°W eastward to 150°E.
Below The “oceanic” sector from 150°E eastward
to 30°W. (From Riehl and Simpson, 1979)
Explanation of the structure of the ITCZ is not fully resolved, in part because of the
variability of its characteristics in time and space. The asymmetry of its distribution with
respect to the equator is particularly marked in the eastern tropical Atlantic and Pacific
(see Figure 3.44). It is suggested by Philander et al. (1996) that this is a result of two
basic factors. First, is the modification of the basic equatorial symmetry of wind conflu-
ence and the Equatorial Trough by ocean–atmosphere interactions. These interactions are
most effective where the thermocline is shallow. In the eastern tropical Atlantic and Pacific
the northeasterly trades maintain a shallow thermocline by modifying the sea surface
temperature regime, and in turn the cold ocean surface maintains persistent low-level
stratus cloud. Second, the bulges in the continental outlines of West Africa and north-
western South America exert a geographical control on the location of the interactions,
thereby determining where they are most effective. These hypotheses have been tested
with a coupled ocean–atmosphere experiment using the Geophysical Fluid Dynamics
Laboratory (GFDL) model.
There are two basic theoretical viewpoints. One involves an axisymmetric model where,
given an ocean with a zone of maximum sea surface temperature (SST), a single ITCZ
and precipitation maximum are formed over the warmest water. Pike (1971) couples an
axisymmetric atmosphere with an axisymmetric ocean and shows that cold upwelling
along the equator can displace the ITCZ in this situation. Based on moist static energy
11 balance considerations, convergence in the equatorial boundary layer will also localize
moisture flux-convergence and precipitation over the warmest water, according to Neelin
and Held (1987). Experiments with a simple coupled model by Wang and Wang (1999)
further explore the maintenance of the equatorial cold tongue. They first note that east–west
climatic asymmetry is generated primarily by the presence of an eastern boundary to the
ocean. Unstable interactions between the sea surface temperature gradient and the zonal
winds, on the one hand, and counteracting stabilization caused by differential surface
buoyancy fluxes, on the other, set up the equatorial cold tongue. However, they invoke
an important role for solar radiation forcing. Seasonal variations in the solar declination
0 angle cause the solar forcing to be antisymmetric about the equator. Hence it can affect
the cold tongue indirectly through changes in the intensity of the trade winds. The cold
tongue interacts with the ITCZ, creating a mechanism for it to be self-maintaining in one
hemisphere. The cold tongue varies in phase with the annual cycle of SSTs south of the
equator. Consequently the band of highest SSTs and the ITCZ are maintained in the
northern hemisphere, rather than responding to the annual radiation cycle. The meridional
moisture gradient between the cold tongue and the ITCZ favours moisture convergence,
helping to sustain the ITCZ when it is in the summer hemisphere.
The second viewpoint envisages atmospheric processes controlling ITCZ characteris-
tics. Moisture convergence may be driven by boundary-layer pumping, referred to as
0 Conditional Instability of the Second Kind (CISK) by Charney and Eliassen (1964) and
Bates (1973), or by convective activity that is organized as a result of zonally propa-
gating equatorial waves in the tropical easterlies – or wave-CISK (Holton et al., 1971;
Lindzen, 1974). In the latter model, maximum boundary-layer convergence, associated
with four to five-day easterly waves in the tropical eastern Pacific, should occur at 6°
latitude, where the local Coriolis frequency and the wave frequency are equal. In either
case, in an axially symmetric atmosphere, it is feasible to form an ITCZ in each hemi-
sphere parallel to the equator, although both CISK hypotheses have been questioned on
various theoretical grounds. More significantly, observations of deep convection over the
tropical Atlantic and Pacific Oceans indicate that boundary-layer convergence forced by
0 friction or easterly waves plays only a minor role in large-scale forcing of convection,
contrary to both CISK hypotheses. Rather, various combinations of large-scale ITCZ
forcing, cloud cluster feedback, and their diurnal modulations are the principal factors.
Figure 3.48 summarizes the relative magnitudes of these forcings to the 850 mb vertical
motion from composites of extensive data sets (McBride and Gray, 1980). In the eastern
Atlantic, diurnal variations in low-level convergence arise as a result of subsidence from
nocturnal radiative cooling of the zone of stratiform cloud to the north of the ITCZ.
To explore this question, experiments have been carried out with GCMs for the ideal-
ized conditions of a water planet. Such experiments with the NCAR Community Climate
Model (CCM) 1, with solar radiation fixed for March 21, suggest that a primary deter-
0 minant of the location(s) of the ITCZ is the particular formulation used for simulating
atmospheric convection processes (Hess et al., 1993). These experiments use the moist
convective adjustment (MCA) scheme, where a saturated unstable air column is adjusted
locally to the saturated adiabatic lapse rate (SALR), conserving moist static energy; rela-
tive humidities that exceed 100 percent as a result of the convective adjustment are set
equal to 100 percent and excess water vapor instantly rains out. For the CCM 1 simula-
tions using the MCA scheme, an ITCZ forms over the zone of maximum SST even when
the SST gradient is weak. With this MCA scheme, dry air above the boundary layer has
to be moistened by strong vertical advection of moisture in order for convection to occur.
Hence rising air with high moist static energy over the region of warmest water favors
0 convection. Hess et al. (1993) also examine an alternative parameterization due to Kuo
(1965) which assumes that the amount of precipitation is proportional to the moisture
flux-convergence in the air column. Cloud properties are predicted by a steady-state “plume
11 type” of cloud model. The rainfall rate is predicted from this plume model and a rate is
Figure 3.48 The relative magnitude of forcing of 850 mb vertical motion in the ITCZ for three
sectors and diurnal phase. (McBride and Gray, 1980)
also determined for each grid box; the ratio of these rates defines the cloud fraction in a
grid box. With the Kuo scheme, the CCM 1 generates an ITCZ on each side of the equator
at about 7° latitude for a wide variety of SST distributions, including the case where the
maximum SST is located on the equator. This parameterization is very sensitive to weak
convergence in the boundary layer and it appears that this convergence is a result of the
effects of transient low-latitude disturbances. Thus bands of convection can develop away
from the equator. Additionally, the model simulations show that the spectrum of equato-
rial waves is significantly affected by the choice of convective parameterization.
Dynamical arguments to account for the location of areas of heating and convection
away from the equator are provided by two recent studies. Waliser and Somerville (1994)
suggest that low-level convergence is maximized dynamically between 4º and 12º lati-
tude. A heat source near the equator generates a pressure gradient anomaly across the
equator which is balanced by geostrophic motion and friction associated with the low-
level convergence and convection. If the heat source is displaced a small distance
poleward, the frictional effect is enhanced on the equatorward side of the anomaly (where
the Coriolis component is small), but further poleward displacement weakens the conver-
gence as the Coriolis force on the equatorward side increases. Waliser and Somerville
propose a positive feedback between latent heat release in the mid-troposphere and low-
level convergence. An alternative hypothesis (Tomas and Webster, 1997) invokes inertial
instability to generate off-equator convection because, in regions of strong cross-equato-
rial pressure gradients, the ITCZ is displaced poleward of an axis of zero absolute vorticity
( ƒ) (see Figure 3.42). Tomas and Webster (1997) propose three requirements for
near-equatorial convection:
11 1 The advection of absolute vorticity across the equator into the summer hemisphere
by a strong cross-equatorial pressure gradient generates flow that has anticyclonic
(negative) absolute vorticity and is (locally) inertially unstable (e.g. Sawyer, 1949).
Secondary meridional circulations cells make the system stable.
2 The SST is near the maximum tropical value, but need not reach a local or absolute
maximum.
3 Poleward of the zero absolute vorticity axis, the atmosphere locally must be condi-
tionally unstable. Deep convection is then set up, augmented by low-level moist static
energy convergence.
0
In the absence of 1, the convection region is determined by conditions 2 and 3.
Figure 3.49 illustrates monthly mean positions of ( f ) = 0 at 850 mb for 1980–94,
showing its poleward displacements, particularly in the Atlantic and Indian Ocean sectors,
where it is about latitude 5º–10º in the summer hemisphere (Tomas and Webster, 1997).
These sectors experience strong cross-equatorial pressure gradients and low-level diver-
gent winds as a result of either differential land–sea heating, or a meridional SST gradient.
Locally, values of ( f ) are less than zero and the flow is inertially unstable. The pole-
ward displacement of convection is limited as latitude () increases by the weakening of
anticyclonic absolute vorticity advection (with cos ) and the concomitant increase (with
0 sin ) of vortex-tube stretching, which generates cyclonic vorticity. Hence the convec-
tive zone cannot be displaced too far from the equator. Tomas and Webster also note that
inertial instability sets up colocated divergent wind maxima and zero absolute vorticity
away from the equator so that the divergent wind is accelerated (decelerated) on the equa-
torward (poleward) side of the zero ( f ) axis. There is a divergent/convergent doublet
to the south/north of this zero axis in the summer hemisphere. In contrast, they point out
Figure 3.49 Monthly-mean locations of axes of ( f) = 0 at 850 mb, 1980–94, based on ECMWF
11 data. (From Tomas and Webster, 1997)
that the Waliser and Somerville (1994) mechanism, discussed above, would give rise to
maximum velocity divergence on the equator where f 0.
The location of the ITCZ has important effects on the structure and strength of the
mean Hadley cells and on the subtropical jetstreams in the model experiments, whereas
in these respects neither the amount of precipitation in the ITCZ, nor the convective
scheme that is adopted, has any significant influence. Hess et al. (1993) find that the jet
maxima are located near 30° (40°) latitude when the location of the ITCZ(s) is over (off)
the equator. The jet maximum is weaker for a single equatorial ITCZ, and the jet struc-
ture is less barotropic. These differences reflect the additional transport of relative westerly
angular momentum as the air moves farther poleward. In the real atmosphere, it is likely
that in many areas latitudinal gradients of SST determine the ITCZ location through the
mechanism of enhanced boundary layer convergence of moisture. However, geographical
and seasonal differences arise where SST gradients are weak. This is the case over the
Indian Ocean and west Pacific Ocean most of the time and likewise in the central Pacific
during the boreal spring. In such areas the ITCZ may be directly determined by convec-
tive processes in the boundary layer rather than by the SST gradients.
The spatio-temporal variability in the ITCZ over the central equatorial Pacific and
Atlantic Oceans sometimes shows a sequence which involves: (1) undulations developing
in the ITCZ cloud band; (2) disruption of the cloud band and its breakdown into several
equatorial disturbances; (3) the evolution of one or more of these disturbances into trop-
ical storms/cyclones; and (4) the movement of these storms westward and poleward,
allowing the ITCZ cloud band to reform (see Plate 4). Ferreira and Schubert (1997)
suggest that the breakdown is a convectively modified form of barotropic and baroclinic
instability of the mean flow. Simulations of barotropic processes, using the non-linear
shallow water equations on a sphere, where the ITCZ is represented by a prescribed
zonally elongated mass sink near the equator, generate a cyclonic potential vorticity (PV)
anomaly with a reversed meridional gradient of PV on the poleward side. This zone
becomes unstable in the presence of small disturbances and (in the model) it either undu-
lates and breaks down into several cyclones or it changes into a single large cyclone.
Ferreira and Schubert propose that the breakdown of the ITCZ accounts for the tendency
for tropical cyclones to cluster in time and for their development poleward of the ITCZ
and to the east of pre-existing storms (see section 6.5). Moreover the horizontal
morphology of the ITCZ, which is wider north–south to the west of Central America,
may help explain the local maximum of tropical cyclogenesis. It is interesting in this
context to examine the climatology of the ITCZ (Figure 3.44) with its broader extent in
the western and eastern equatorial Pacific and western equatorial Atlantic, in relation to
the major regions of tropical cyclone formation (Waliser and Gauthier, 1993).
Equatorial flow patterns

Airflow in low latitudes is often poorly determined because pressure gradients are typi-
cally weak and the Coriolis parameter is small. However, some important theoretical
considerations merit discussion. Above the boundary layer, in the absence of pressure and
frictional forces, the synoptic conditions in low latitudes commonly satisfy the require-
ments for horizontal flow to display inertial motion. Such motion involves a balance
between centrifugal acceleration (V2/r) and Coriolis acceleration ( f V) i.e.:
V2
fV
r
and thus
V rf
0
0
0
0
0
11
11
Plate 4 Sequence of images illustrating the breakdown and redevelopment of the ITCZ cloud band over the eastern tropical Pacific Ocean, 2°–24°N,
108°–165°W. The infrared images are for 16.46 UTC on (a) July 26, (b) July 28, (c) August 3, and (d) August 12 1988. Hawaii is in the upper left
corner and Baja in the upper right corner of each image. (a) A zonal band of convection around 10°N. (b) Cloud-band undulations (tropospheric east-
erly waves) that evolve into five tropical depressions on July 30. (c) Four of the depressions are now tropical cyclones. (d) The ITCZ has redeveloped.
(From Ferreira, R.N. and Schubert, W.H., 1997, “Barotopic aspects of ITCZ breakdown,” J. Atmos. Sci., 54(2): 262–3)
It follows that, if the latitudinal variation of f is ignored, an air parcel will move hori-
zontally in an inertial circle of radius r. This inertial motion is clockwise (counter-
clockwise) in the northern (southern) hemisphere. A parcel will move around a circle in
2/f, which corresponds to twelve hours at 90°, twenty-four hours at 30°, sixty-nine hours
at 10° and 137 hours at 5° latitude.
The characteristics of inertial flow in low latitudes, taking account of the latitudinal
variation of f, are treated by Wiin-Nielsen (1970; see Asnani, 1993, pp. 419–27). His
solutions show that three types of trajectory may occur according to the initial latitude,
direction, and speed of the air parcel. These trajectories result from a relationship:
V0 >
cos2 0 1/4 0
f02 <
where V0 initial velocity, 0 initial angle with latitude circle (x axis),

f /
y, and
f0 initial value of Coriolis parameter. Where the solution of the equation is positive,
the trajectory repeatedly crosses the equator; where it is negative, the parcel remains in
one hemisphere; and where it is identical to zero, the parcel approaches the equator asymp-
totically. Wiin-Nielsen also shows that, because the path is not actually circular, the time
interval required for a parcel to return to the same latitude (in the equivalent phase of
motion) differs considerably from 2/f depending on 0 and latitude.
Patterns of steady-state flow, identifiable on synoptic and climatological maps over
Africa, have been classified in a simple scheme by Johnson and Mörth (1960) and Johnson
(1965). Figure 3.50 illustrates their six suggested types. In four of them there is a pres-
sure maximum or minimum along the equator. The duct pattern (a) is common over
equatorial East Africa in the transition seasons, while the bridge (b) occurs over West
Africa in July. The drift types (d-f) are observed over Indonesia and East Africa in January.
Divergence computations following trajectories are sensitive to small errors in the initial
winds (Gordon and Taylor, 1970; Ramage, 1971, pp. 85, 130). Near the equator the diver-
gence is determined mainly by the curvature of the pressure gradient, whereas beyond 5°
latitude the Coriolis terms are important. Flohn (1960) points out that in low latitudes
westerly and poleward flows are convergent, easterly and equatorward flows divergent.
The vertical acceleration can be expressed (Haltiner and Martin, 1957, p. 169) as:

w
p
= 2u " cos R g

t
z
where the first term on the right is the vertical component of the Coriolis acceleration
and the second term is the vertical component of the pressure force per unit mass; R
the earth’s radius. Johnson and Mörth (1960) and Flohn (1960) postulate that the vertical
Coriolis component is important near the equator, but even for large values of u it is only
about 0.1 percent of g. The term appears to be sufficient to account for the conver-
gence of equatorial westerlies (Ramage, 1971, pp. 85, 130).
As a consequence of these relationships, air approaching (moving away from) the
equator undergoes horizontal velocity divergence (convergence) and therefore descends
(ascends). The horizontal divergence is given by:
div VH v cot /R
For v 1 m s1, the divergence is 90 106 s1 at 0.1° latitude and 9 106 s1 at
1° latitude. An alternative interpretation from the conservation of potential vorticity is
that, for air approaching the equator, f and therefore sin tend to zero and accordingly
the vertical column shrinks, creating subsidence. Asnani (1993, p. 458) refers to this as
the dynamic valley effect of the equator.Where the isobars cross the equator in the models
of Figure 3.50 they display a cusp or bulge. For the cases of nearly zonal flow the motion
11
Figure 3.50 Idealized equatorial flow patterns. (Johnson, 1965)

0
is quasi-geostrophic, although the pressure or height gradients are generally weak. For
balanced flow in geostrophic westerly winds in the equatorial planetary boundary layer,
Kuo (1973) concludes that the flow is maintained by vertical momentum transport; down-
ward motion is required because there is cross-isobaric motion away from the equator
both to north and to south. For a westerly flow of 5 m s1 there would be subsidence of
0.7 cm s1.
Johnson (1962) examines the temporal distribution of rainfall over East Africa (10°S–
4°N) and finds no evidence of mobile disturbances or zonal bands of precipitation. Rather
0 the rainfall patterns indicate patchiness on a scale of 100–300 km with intervening areas
that are dry or have only scattered precipitation. The rain episodes, with 8–12 mm per
rain day, are superimposed on the broad regimes of wet and dry seasons, and there are
11 no clear relationships between the precipitation areas and the circulation patterns.
A major component of cross-equatorial flow into the northern hemisphere in the boreal
summer is accomplished by a low-level jetstream (LLJ) over East Africa. First identified
and analyzed by Findlater (1966, 1969a, b, 1971, 1972), it has been examined observa-
tionally and modeled by a number of subsequent investigators (Hart et al., 1978; Bannon,
1979). The East African LLJ is an important component of the summer monsoon system
over the Indian Ocean. It develops in May as a narrow band of strong, low-level south-
easterly winds across the East African coast around 10°–5°S. During July–August
streamlines at 1 km show an elliptical flow from about 16°S, 70°E to the equator around
42°E and then northeastward to 16°N over India, as shown in Figure 3.51 (Findlater,
1971). Maximum speeds at 1–1.5 km are about 15 m s1 off northern Madagascar and
15–18 m s1 northeast of the Horn of Africa. There is also a splitting of the jet here,
with one branch directed east-southeastward towards Sri Lanka. The LLJ maintains high
constancy of direction, but varies greatly in intensity. Oscillations in intensity with a four
or five-day period have been detected, according to Asnani (1993). The jet is best devel-
oped at night and in the early morning, suggesting boundary layer forcing. The flow is
forced by low-level velocity divergence in the southeasterly trades emerging from the
South Indian Ocean high and is bounded in the west by the mountains of East Africa
(Anderson, 1976; Bannon 1979). The variation of the term and the diurnal temperature
structure in the lower troposphere also play important roles. The East African LLJ accounts
for about half the cross-equatorial transport of water vapor in the western Indian Ocean
(from 75°E to the East African coast). Moreover, active monsoon regimes over India
appear to be preceded by a strengthening of the LLJ.
3.7.3 Monsoons
The name “monsoon” derives from mausam, the Arabic word for season. The climatic
concept refers both to seasonal rains and to seasonal changes of prevailing winds. The
structure of the wind field in low latitudes and the existence of a southwesterly monsoon
flow in the northern summer over the Atlantic and Indian Oceans was first documented
Figure 3.51 (a) Monthly locations of the mean axis of the low-level jetstream over the western
Indian Ocean and East Africa. (b) Mean velocity (m s1) of the jet at 1 km in July.
(After Findlater, 1971; Barry and Chorley, 1998)
11 by the astronomer Edmond Halley in 1686. Remarkably, his maps from ship observations
correctly depict the southwesterly flow in the Gulf of Guinea and southwesterlies origi-
nating about 2–3°S in the Indian Ocean (see Kutzbach, 1987; Webster, 1987). Clima-
tologists have used various definitions to distinguish the monsoon areas of the world.
Rather comprehensive criteria proposed by Ramage (1971) take account of three factors:
1 A shift of the prevailing wind direction by 120° or more between January and July,
with the prevailing directions having an average frequency greater than 40 percent.
2 A mean resultant wind exceeding 3 m s1 in at least one of the months.
0 3 Fewer than one cyclone/anticyclone alternation occurring every two years in January
or July within a 5° latitude–longitude area.
This definition largely corresponds to the zone where the ITCZ has a large annual oscil-
lation and it encompasses most of southern Asia, northern Australia, West Africa and
eastern Africa. However, this monsoon system interacts dynamically with the circulation
over a far wider area of the tropics and subtropics. In addition, the term “monsoon” has
subsequently been extended to describe summer conditions on the plateaus of the south-
western United States and northwest Mexico (Douglas et al., 1993; Tang and Reiter, 1984;
Adams and Comrie, 1997; Barlow et al., 1998), although these areas do not meet Ramage’s
0 criteria, and to the western Pacific Ocean (Holland, 1995). The extent of seasonal
monsoons shown in Figure 3.52 is based on the criteria of a seasonal wind reversal and
a wet summer/dry winter rainfall regime (Webster, 1987a).
The first theoretical concept of the seasonal wind reversal was offered by Halley (1686),
who proposed a continental-scale sea–land breeze system, due to differential heating of
the land and sea, with upper-level return currents. Fifty years later, George Hadley (1735)
pointed out that the airflows were northeast–southwest (southeast–northwest) in the
northern (southern) hemisphere as a result of the rotational effect of the Earth on the air
motion rather than being caused by the westward movement of diurnal solar heating, as
proposed by Halley. Systematic study of the monsoon in southern Asia, with a view to
0 documenting the nature of its variations and forecasting their behavior, began over a
century ago with the statistical investigations of Blanford (1884), elaborated later by
Walker (1914, 1923). However, significant advances in understanding the dynamics of
the monsoon have occurred only in recent decades through a combination of observa-
tional data, theoretical studies, and modeling. The observational component included
several international field programs; the International Indian Ocean Expedition, 1959–65,
especially 1963–65 (Ramage and Raman, 1972a, b); the Indian–Russian Monsoon ’77
program in the Indian Ocean; the winter and summer Monsoon Experiment (MONEX)
and the West African Monsoon Experiment (WAMEX) in 1978–79 (WMO/ICSU, 1980,
1981; Petrosyants and Belov, 1988); and Chinese programs on the Qinghai-Xizang
0 (Tibetan) Plateau (QXPMEX) in summer 1979 (Xu, 1986). MONEX was carried out over
the sector 40°–180°E, between about 20°S and 32°N, as part of the GARP Global Weather
Experiment, 1978–79.
General characteristics and mechanisms

Monsoon climates are usually characterized in terms of onset/withdrawal dates, rainfall
regime, break and active phases, and synoptic disturbances. These are obviously area-
specific. There are considerable differences between the summer monsoons of the Indian
Ocean, West Africa, and eastern Asia, and in the austral summer, northern Australia.
0 Nevertheless, a common element is the occurrence of equatorial westerlies at low levels
associated with cross-equatorial flow from the easterly trades of the winter hemisphere,
and the development of low pressure over tropical continental areas in the summer hemi-
11 sphere. In essence, the equatorial trough/ITCZ is displaced poleward and becomes a
(a)
(b)
Figure 3.52 Domains of the principal global monsoon systems. (a) Boreal summer. (b) Boreal
winter. The rectangle (dashed) outlines the classical monsoon region based on the
criteria of a seasonal wind reversal and distinct wet summers/dry winters. Arrows indi-
cate predominant surface airflow; areas of maximum seasonal precipitation are outlined.
Land areas with maximum surface temperatures are cross-hatched and the coldest land
surfaces are stippled. NET and SET northeast and southeast trade winds. In (a): EAM
East Africa monsoon, ISWM Indian southwest monsoon, WafM West Africa monsoon,
NAmSM North America summer monsoon. In (b): NEWM Northeast (Asian) winter
monsoon, ANWM Australian northwest monsoon, NAmWM North American winter
monsoon, AfWM African winter monsoon. (From Webster, 1987a)
monsoon trough. Figure 3.53 illustrates conceptually the joint effects of Coriolis deflec-
tion and a heated continent (Young, 1987). If there is a tropical land mass in the summer
hemisphere, strong heating enhances the southwesterlies (Figure 3.53b), and these may
be augmented further by precipitation and latent heat release over the land area (Figure
3.53c). If the continent is in mid-latitudes, however, there is no cross-equatorial flow and
only moderate inflow (Figure 3.53d). An equatorial continent will generate inflow only
towards the heated interior (Figure 3.53f), whereas a land barrier to the west of an equa-
torial ocean will augment the cross-equatorial airflow (Figure 3.53e). In the case of an
equatorial ocean area there will be a weak ITCZ in the summer hemisphere, a few degrees
off the equator (Figure 3.53a).
These ideas are quantified to some degree in experiments by Dirmeyer (1998) with a
coupled land–biosphere version of the COLA GCM with prescribed sea surface temper-
atures. A flat continent with savannah vegetation from 30°W to 30°E is placed in different
11
0
Figure 3.53 Factors influencing monsoon winds. Land areas are shaded. Arrows show winds; strong
winds are bold arrows. The panels illustrate schematically the effects of land–sea distri-
bution, heating, precipitation, and topographic barriers (see text). (From Young, 1987)
latitude locations extending between 0° to 27°N and 27° to 53°N (in 4.4° latitude steps).
For the subtropical continent, 13°–40°N, the width is also varied between 60°, 120° and
360° of longitude. Dirmeyer shows that latitudinal location is critical in establishing the
seasonal precipitation regime. With land in the tropics there is heavy precipitation south
0 of 20°N, with wettest conditions in the eastern half and dry conditions in the northwest,
as observed in India. A low-latitude continent also results in a landlocked precipitation
maximum. A mid-latitude continental location sets up a winter (Mediterranean-type)
precipitation maximum, while a subtropical continent (13°–40°N) has an overall drier
climate. In order to obtain heavy convective precipitation over the land in summer, the
extension of a subtropical continent into the tropics is required. The meridional extent
of the continent affects both the strength of the Hadley circulation and the seasonality of
precipitation over land. In all cases the descending arm of the Hadley cell is locked over
the land area. Increased zonal extent of the continent leads to a decrease in zonally aver-
aged precipitation and an expansion of the low precipitation area in the western part of
0 the continent. The inclusion of a tropical peninsula in the southeastern or southwestern
sector also affects the precipitation distribution. With a peninsula in the southeast, as in
Africa, the summer monsoon is enhanced, although the rains do not penetrate as far north.
11 Nevertheless, in both cases the northwest is dry. Finally, 1,200 m high mountains are
incorporated in the four cardinal directions in turn. Overall, precipitation amounts are
increased (reduced) to the south and east (north and west) of the mountains. The inclu-
sion of a mountain range leads to increased instability and precipitation, through the
elevated heat source effect, and it creates shifts in airflow that produce greater moisture
flux convergence. Mountain uplift is not shown to be an important precipitation mecha-
nism in the model. With the mountains in the south or east, there are enhanced tropical
westerlies and moisture transport. With mountains in the north and west, the westerlies
are kept in higher latitudes. The role of topography in the tropics in observed global
climatic characteristics is reviewed by Meehl (1992).
The basic forcing of the Asian summer monsoon is provided by the annual cycle of
solar radiation interacting with the different heat capacities of the tropical oceans and
land areas (Li and Yanai, 1996), and their respective geographical arrangements. The
monsoon flow, however, represents much more than a giant sea breeze circulation. Strong
heating in spring creates a thermal low over the northern Indian subcontinent, but it is
another month or so before the onset of the monsoon southwesterly flow.
Numerous empirical and modeling studies suggest that the extent of spring Eurasian
snow cover modulates the intensity of the Asian monsoon during the subsequent summer,
as represented by precipitation totals over India (Barnett et al., 1989; Yasunari et al.,
1991). Recent work confirms such an inverse relationship for 1979–92, which is stronger
when El Niño years are excluded (Sankar-Rao et al., 1996). During winters with more
snow over Eurasia the atmosphere is colder (warmer) than normal over the land (ocean).
The troposphere over Asia north of India remains colder in summer following a
winter/spring with more extensive snow cover and the monsoon is delayed and weaker.
The winter (December–January–February) snow depth in European Russia shows a signif-
icant negative correlation with the subsequent Indian monsoon rainfall, according to
Kripalani and Kularni (1999), while there is a positive relation between winter snow depth
in central Siberia and monsoon rainfall; the sign of both these correlations reverses
following the monsoon. They interpret the correlation structure in terms of the long-wave
pattern: there is an anomalous ridge (trough) over Eurasia during the winter preceding a
strong (weak) monsoon. Model simulations by Vernekar et al. (1995) and Dong and
Valdes (1998) suggest that the use of energy for snow melt in spring causes cooler condi-
tions over the Tibetan Plateau, resulting in decreased upward sensible heat flux, a reduced
meridional temperature gradient and thereby a weaker monsoon. Dynamically, the snow
cover favours high pressure over India, a decreased Somali jet, and weakened low-level
southwesterlies and upper-level easterlies. It seems that the snow cover may be largely
an indicator of circulation anomalies that modify surface temperatures, and hence the
land–sea temperature contrast may be primarily responsible for the monsoon strength
(Meehl, 1994).
The patterns of heating are complex because the seasonal variation of ocean tempera-
tures lags both the incoming solar radiation and continental heating; moreover these
variations are strongly dependent on latitude and also show large longitudinal differences.
The large-scale heating in the atmosphere is also a function of altitude, as shown by
vertical variations in the annual temperature oscillations (Verma and Sikka,1981). The
amplitude of the annual oscillation shows two peaks, one in the lower troposphere due
to sensible heating and the other around 300–400 mb as a result of latent heat release
(Figure 3.54). Over southern Asia the lower maximum occurs in June and the upper one
thirty to forty-five days later associated with monsoon precipitation from deep cumulus
cloud over the Himalayas and southeastern Tibet (Flohn, 1968; Yeh and Gao, 1979).
These characteristics indicate the complexity of the diabatic forcing and the additional
role of dynamic factors. The large-scale circulation over the tropics and subtropics in
summer and winter is shown in Figure 3.55 (Webster et al., 1977). In June, July, and
August the 200 mb streamlines demonstrate a dominant anticyclonic outflow from the
Tibetan Plateau, because of the effect of Plateau heating reversing the meridional thermal
11
Figure 3.54 Proposed model of atmospheric heating in the Asian summer monsoon region. (From
Verma and Sikka, 1981)
gradient. Easterlies extend from the western equatorial Pacific, across southern Asia and
Africa into the North Atlantic; these recurve across the equator and form the subtropical
upper westerlies of the southern hemisphere. At 850 mb, equatorial westerlies and south-
0 westerlies in the mean are well developed north of the equator only from central Africa
to the Philippines.
The distribution of sea surface temperatures and the large-scale topography ensure that
the winter and summer monsoon circulations of southern Asia and the Indian Ocean are
far from being mirror images of one another (Webster et al. 1977), as is apparent from
the streamline maps of Figure 3.55. There is a modest anticyclonic outflow aloft over
northwestern Australia, and the equatorial westerlies are confined to the Australasian
sector. Three topographic factors are important in this respect: the relative sizes of Asia
and Australia; the location and form of the Asian mountains and Tibetan Plateau; and the
mountainous islands (particularly Sumatra, Java, Borneo and New Guinea) comprising
0 the “maritime continent” (Ramage, 1968).
The large-scale shift between the winter and summer circulation regimes over southern
Asia clearly involves major adjustments in the atmosphere. Key elements in this transition are:
1 The weakening and northward displacement of the westerly subtropical jetstream

(STJ) and the appearance of the tropical easterly jet (TEJ) over southern India (see
section 4.2).
2 The associated reversal of the thermal gradient in the upper troposphere, which is
directed northward in the cold season, but is directed equatorward over southern Asia
in summer, owing to atmospheric heating over the Tibetan Plateau.
0 3 Strong heating over the Indian subcontinent in the spring transition months estab-
lishing a south–north pressure gradient at low levels.
4 Cross-equatorial motion of the southern hemisphere easterly trade winds in the Indian
11 Ocean and their deflection to a southwesterly direction over the Arabian Sea.
Figure 3.55 Mean streamlines for (a) June–July–August and (b) December–January–February at
850 mb (upper) and 200 mb (lower). (From Webster et al., 1977, after Sadler, 1975,
and Newell et al., 1972)
5 The regional enhancement of this cross-equatorial flow over East Africa in the Somali
low-level jet (LLJ) and its associated transport of moisture in the lower troposphere.
6 The formation of an onset vortex over the Arabian Sea, with westerlies on its equator-
ward side triggering the start of heavy rains over central India (Krishnamurti et al.,
1981). However, this feature does not develop in every year.
11 (a)
0
(b)
Figure 3.56 Schematic illustration of (a) “active” and (b) “break” phases of the Asian summer
monsoon (see text). (From Webster, 1987b)
The monsoon over India is characterized by active phases and breaks with little or no
precipitation. Figure 3.56 illustrates the latitudinal distribution of cloud and precipitation
during these phases. In the active phase there is a strong TEJ and maximum precipita-
tion is over the central Indian peninsula. In the break regime there is subsidence over
peninsular India, with cloud bands and precipitation to the north over the Himalayan
0 foothills as well as to the south around 5°–12°N. Breaks tend to recur over approximately
ten-to-twenty-day and forty-to-fifty-day intervals. Sikka and Gadgil (1980) show that
during 1973–77 bands of maximum cloudiness (and the co-located 700 mb trough) in the
Indian Ocean repeatedly propagated northward from a few degrees north of the equator
to the Himalayan foothills (25°–28°N) in ten to twenty days. The extended forty to fifty
day variations represent modulations of convective activity propagating eastward in low
latitudes (see Madden–Julian Oscillations (MJO), p. 334). Divergent components of the
wind (represented by velocity potential fields5) associated with an MJO enhance (weaken)
the monsoon westerlies as the direction of the divergent wind component has a large
westerly (easterly) component (Webster, 1987b; Webster et al., 1998).
0 A broader picture of break/active phases within the northern summer monsoon is
suggested by intraseasonal oscillations (ISOs) in ECMWF tropospheric wind data and
OLR signatures (Wang and Xu, 1997). These ISOs are found to be phase-locked to the
annual cycle, moving northward from the equator to the northern Philippines during
May–July, and westward along 15°N during August–September to the Bay of Bengal.
Wang and Xu propose that climatological monsoon singularities over this large region
are linked with this sequence and identify four phases:
1 A peak wet phase over the South China Sea and the Philippines in mid-May followed
by a dry phase of the western North Pacific monsoon (see below), the Indian monsoon
0 and the Meiyu/Baiyu rains of East Asia in late May–June.
2 A peak of convection, more or less in phase, in mid-June, marking the onset of the
Indian and western Pacific monsoons and the Meiyu in central China, followed by
11 monsoon breaks and the end of Meiyu over northern China in mid-July.
3 The maximum of the western Pacific monsoon in mid-August, followed by the west-
ward propagation of a dry phase, giving a second western Pacific break, out of phase
with that in India during mid-September.
4 The final active phase of the western Pacific monsoon and withdrawal of the Indian
monsoon.
Despite the temporal association of precipitation events over southern and eastern Asia
implied in 2 above, the Meiyu cloud band variations appear to be an independent mode
linked with the East Asian jetstream (Liang and Wang, 1998; Kang et al., 1999). It seems
that a definitive picture of the southern and East Asian summer monsoons cannot yet be
presented. Wang and Fan (1999) discuss various monsoon indices for South Asia based
on convective activity indicated by OLR data. The monsoon rains over China typically
last thirty to forty days, with totals decreasing northwards, although in northern China
there is also rain from mid-latitude systems (Samel et al., 1999). Precipitation in southern
China is augmented when low-level cold air from the north is overrun by the warm, moist
southwesterly monsoon flow. The Meiyu front is also strengthened when there is high
pressure over the Sea of Okhotsk. Monsoon rainband precipitation in northern China is
greater when there is strong heating over Eurasia and increased southeasterly flow asso-
ciated with a subtropical ridge over the China coast.
Holland (1995) identifies a unique summer monsoon occurring over the western equa-
torial North Pacific. In the lower troposphere there is a monsoon trough or diffuse low-
pressure area, a westerly monsoonal flow extending eastward over the equatorial western
Pacific, and a confluence between this flow and the central Pacific tropical easterlies (see
Figure 3.57) In the upper troposphere there are a diffuse anticyclone and extensive tropi-
cal easterlies. This regime appears to be quite distinct from the Asian summer monsoon.
Key drivers of the system are the lower tropospheric confluence and upper diffluence, which
enhance convection and maintain a feedback cycle with the monsoon circulation.
Precipitation within the Asian summer monsoon is strongly determined by synoptic
systems. The major categories of rain-producing systems are: monsoon depressions,
subtropical cyclones, and tropical cyclones (Hamilton, 1979), with heat lows producing
hot, dry weather. These are categorized schematically in terms of vertical motion, diver-
gence, and weather in Figure 3.58, and are they now discussed in turn.
Figure 3.57 Schematic diagram of the principal lower tropospheric components of the summer
monsoon in the western North Pacific. (From Holland, 1995)
11
Figure 3.58 The circulation systems which occur in monsoon climates, arranged schematically
according to their divergence, vertical motion, and weather. Levels where non-
divergence occurs are marked by heavy lines. (From Ramage, 1971)
Figure 3.59 Monsoon depression over southern Asia, 12.00 GMT, July 4 1957. Above 500 mb
height contours (dam). Below Sea-level pressure (mb) and the Monsoon Trough (dashed
line). Wind arrows (barb indicates 5 m s1) and precipitation areas (oblique shading).
(Based on IGY charts of the Deutsche Wetterdienst, from Barry and Chorley, 1998)
Monsoon depressions. These low-pressure systems move west-north-west across India

from the Bay of Bengal, steered by the upper easterly flow. Records since 1891 show
0 there are two to two and a half per month during June–September (Sikka, 1977). During
1891–1970 about four to five systems each season crossed 85°E, primarily between 18°
and 27°N, although only about one-quarter of these reached 75°E (Mooley, 1973; Rao,
11 1976, p. 107). A few may also occur over the Arabian Sea, usually in June. Monsoon
depressions have a lifetime of three to five days. The circulation is cyclonic for some
500–1,500 km radially, and extends to about 400 mb, with maximum development between
700 mb and 850 mb. There is anticyclonic circulation and divergence above 10 km. The
surface pressure anomaly is typically less than about 12 mb and the winds are between
9 m s1 and 17 m s1 (Sikka, 1977). Their thermal structure is varied; some are cold
cored at lower levels and warm cored above 500 mb, some are mainly warm cored, while
others are neither (Keshvamurty, 1972; Krishnamurti et al., 1975, 1976). In many cases
the axis tilts southwestward with height, with a cold lower troposphere overlain by a
warm pool above the main precipitation area southwest of the surface center, to the left
of its track (Figure 3.59) (Godbole, 1977; Douglas, 1992). Monsoon depressions occur
on only 7 percent of days in July–August at 80°–85°E; consequently they contribute only
8–12 percent of the monsoon rainfall over the Ganges basin (Dhar and Bhattacharya,
1973). Nevertheless, they do modify the spatial distribution and timing of rainfall in indi-
vidual years. Their formation tends to activate the monsoon trough, and, when two systems
occur in close succession along a similar track, floods may result (Sikka, 1977). The
importance of rainfall from these systems is illustrated by the fact that the ratio of mean
precipitation on days with depressions, relative to the mean on other days, is 1.5 at Delhi
(28.8°N, 77°E), but 6.9 at Ahmadabad (23°N, 72.7°E) (Mooley, 1973).
Most of the disturbances originate from lows over the western tropical Pacific, South
China Sea and Southeast Asia. Krishnamurti et al. (1977) propose the following sequence:
1 A tropical storm approaches the coast of Vietnam, and pressures rise around 20°N.
2 In the following week, pressures rise 5–7 mb over Indochina and Burma.
3 A monsoon disturbance forms in the northern Bay of Bengal.
According to Ramage (1971, p. 46), there are three favorable preconditions. The monsoon
trough extends along the Ganges valley from the northern Bay of Bengal. Over the
southern part of the bay, rising air associated with convergent west-southwesterly flow
overlain by divergent east-northeasterly circulation accelerates. The resulting increase in
low-level cyclonic vorticity south of the monsoon trough becomes concentrated into a
vortex in the trough, causing uplift, latent heat release, and circulation intensification.
Strengthening of the monsoon southwesterlies over India and the central Bay of Bengal
plays a major role by increasing cyclonic wind shear farther north, according to Sikka
(1977). Hovmöller longitude–time plots of surface pressure changes show a low–
high–low pattern characteristic of low-latitude downstream amplification. The waves have
6° westward phase velocity (Krishnamurti et al., 1977). The failure of these systems to
intensify into cyclones is attributed by Ramage (1971) to the limited heat source avail-
able over the northern Bay of Bengal and the presence of strong vertical wind shear.
Subtropical cyclones. These systems were first investigated in the eastern North Pacific,
where they originate from cut-off lows in the upper westerlies during winter–spring
(Simpson, 1952; Ramage, 1962). In the Hawaiian Islands, where they tend to be long-
lived, they are known as Kona storms. They are best developed in the middle troposphere,
where convergence between 600 mb and 400 mb is compensated for mainly by ascent
and upper-level divergence. Also, beyond about 500 km radius, there is some low-level
sinking (Figure 3.60). The inner 200 km constitutes a broad eye with only scattered clouds.
A further type of subtropical cyclone is found over the northern Indian Ocean in summer
(Miller and Keshvamurthy, 1968; Ramage, 1971; Hamilton, 1979; Hastenrath, 1991).
These have a similar vertical arrangement of convergence and divergence, with surface
westerlies and upper easterlies. Case studies show that the system’s core is cold at 700
mb and relatively warm at 500 mb. Figure 3.61 illustrates a north–south cross-section
along 72°E and the corresponding kinematic analysis at 600 mb for a system near Bombay,
India, in July 1963. These systems tend to form, intensify, and dissipate during a one-to-
11
Figure 3.60 Schematic radial cross-section of a subtropical cyclone, showing gross features of
vertical motion and clouds. Divergence (convergence) is denoted by () signs.
Regions of dry (moist) adiabatic temperature changes are shown by D (M). (From
Ramage, 1971)
three-week period over the northeastern Arabian Sea. They are an important contributor
to rainfall over western India. Precipitation falls mainly south and west of the cyclone
center, to the left of the cyclone track. Since the development of a subtropical cyclone
0 requires a deep, moist airstream, which is associated with the monsoon southwesterlies,
the rains over western India are typically later than over the east of the country, where
the mean flow steers depressions from the east. According to Ramage, subtropical cyclones
do not develop until a heat low has formed over the Great Indian Desert in northwestern
India. He proposes (1966, 1971, pp. 60–7) that this heat low is the source of mid to upper
tropospheric cyclonic vorticity which is exported towards the Arabian Sea. The devel-
oping subtropical cyclone in turn intensifies the heat low through subsidence. Dissipation
appears to result from the mid-tropospheric advection of dry air into the cyclone.
Three subtropical cyclones studied by Miller and Keshavamurthy (1968) were all
preceded by cyclonic activity over eastern India, and were associated with anticyclonic
0 vorticity over the heat low and enhanced cyclonic vorticity over the Arabian Sea between
18°N and 21°N. Subtropical cyclones are also observed on the equatorward side of heat
low troughs over the Sahara and Australia (Ramage, 1971), over the southern Bay of
Bengal and Burma in April–May, and over Indochina and the South China Sea in summer.
Heat lows. The deserts bordering the summer monsoon regions are dominated in summer
by persistent, quasi-stationary heat lows or troughs. These are found, for example, over
the Sahara, in an arc from the Horn of Africa through Arabia to northwest India, over
Australia, and over the Great Basin of the United States. Their particular significance in
the context of the monsoon systems is their apparent role in exporting energy in the upper
0 troposphere which helps sustain the monsoon system (Rodwell and Hoskins, 1996) and
may also support the genesis of subtropical cyclones, as noted above.
The structure of a heat low, or trough, features low-level converging air. However, little
11 cloud generally develops as a result of the very low relative humidity. Moreover, subsidence
(a)
(b)
Figure 3.61 Composite model of a subtropical

cyclone over the Arabian Sea, showing (a) a
kinematic analysis at 600 mb for July 2–10 1963;
(b) the distribution of clouds, vertical motion and
temperature along 20°N, and (c) along 72°E. In
(a) the isotachs are in knots; each full wind barb
represents 5 m s1. Rainfall exceeded 40 mm day1
along the coast between 21°N and 12°N. In (b) and
(c) the vertical velocity vectors represent 40 cm s–1
(long) and 5 cm s–1 respectively. (From Miller and
Keshavamurty, 1968, and from Ramage, 1971)
(c)
occurs in the middle troposphere, associated with convergence in the upper tropospheric
tropical easterly jet. This subsidence maintains an inversion level in the middle troposphere.
During the transition seasons the heat low may be diurnal, forming by day through surface
heating and weakening or dissipating at night through radiative cooling.
There have been few detailed investigations of heat lows, largely because of the paucity
of stations in desert areas. However, two field campaigns have targeted the Arabian heat
low in the context of the summer monsoon over southwest Asia. Measurements during
11 MONEX in the summer of 1979 identified weak heat lows of about 1,004 mb over the
Arabian Peninsula during 9–12 May (Blake et al., 1983). The motion was anticyclonic
and divergent between 850 mb and 700 mb, with westerlies above 600 mb. During the
day there was ascent up to 1 km and descent above, whereas, prior to sunrise, descent
prevailed at all levels. Long-wave radiative cooling of 2°–5°C day1 throughout the tropos-
phere is counterbalanced around midday by strong shortwave warming which is attributed
to absorption by a dust layer extending to 600 mb. Adiabatic warming also occurs of up
to 3°C day1 around 500 mb, associated with the downward motion. These two heating
sources contribute to stabilizing the mixed layer. Smith (1986) demonstrates that in June
0 1981 the lower and middle layers are heated by radiative absorption and adiabatic descent
(Figure 3.62). The heat low is approximately radiatively neutral or slightly positive at the
top of the atmosphere. This implies that during the monsoon it maintains itself as an
energy source to the surrounding environment. Horizontal advection of the mid-tropos-
pheric excess heat over the northwest Arabian Sea may help in maintaining the low-level
inversion which is generally attributed solely to large-scale subsidence. This inversion
caps the low-level southwesterly jet flow which transports the moisture towards the
monsoon trough over India. Thus the Arabian heat low may play an important role in
the monsoon system.
Measurements of the surface energy budget over the Rub-al Khali Desert during
0 June–July 1981 by Smith (1986) indicate peak daily values of the heating terms near the
center of the heat low of the following order:
Absorbed solar radiation 650 W m2

Net long-wave radiation 250 W m2
Net radiation 400 W m2
Sensible heat 250 W m2
Ground heat storage 175 W m2
11 Figure 3.62 Controls of the structure of the Arabian heat low. (From Smith, 1986)
However, the phases of these peaks are not symmetrical. For daily averages in summer
the net radiation is balanced approximately by sensible heat. The day-night range of skin
temperature is up to 50°C because sand is a poor heat conductor.
Recent work in northwestern Australia suggests that the “classic” heat low, with low-
level cyclonic convergence, a late afternoon pressure minimum, and divergent anticyclonic
circulation aloft, occurs primarily in the absence of large-scale circulations that can lead
to subsidence throughout much of the troposphere (Racz and Smith, 1999). Analysis with
a hydrostatic primitive equation numerical weather prediction model indicates that, in
northwestern Australia, a daytime sea breeze and nocturnal low-level jet play an impor-
tant role in low-level convergence. Convective mixing in the afternoon weakens the
cyclonic relative vorticity, which reaches its maximum in the early morning.
The Asian winter monsoon

In the northern hemisphere the intense polar anticyclones over eastern Siberia generate
recurrent surges of cold, dry northerly airflow in winter over China and southeast Asia.
In contrast, southward flow out of central Asia is largely blocked by the mountain ranges
of the Caucasus, the Elburz and Zagros in Iran, and the Hindu Kush–Karakorum–
Himalayas. Weak northeasterly low-level flow over the Indian subcontinent originates
from air subsiding beneath the subtropical westerly jetstream, aided by the land–sea
pressure gradient that is maintained by the land–sea temperature difference.
High-pressure cells emanating from west of Lake Baikal near 40°N, 90°E tend to move
southeastward across northern China to the Yangtze delta (track 1 in Figure 3.63b) while
others affect northeastern China, the Yellow Sea, the Korean peninsula, and the Sea of
Japan (tracks 3 and 4). Cells that move eastward to the south of 50°N cross Xinxiang
Province and Mongolia and affect eastern China (track 2). During winters 1980–84, 64 per-
cent of high cells moved on track 1, 27 percent on track 2, and 10 percent on track 3 (Ding
and Krishnamurti, 1987). Outbreaks of cold Siberian air are associated with a sharp pres-
sure rise which propagates rapidly southeastward over China. Pressure tendencies at sta-
tions along the path of a cold surge indicate that it can advance at 30 m s1, covering 2,000
km in twenty-four hours; this is up to three times the speed of the winds in the surge, imply-
ing that the mechanism involves the propagation of a gravity wave rather than cold air
advection (Chang and Lau, 1980). Compo et al. (1999) suggest that they may represent a
topographic Rossby (“shelf”) wave related to the eastward-sloping terrain of East Asia.
About twenty such cold events affect some part of China in a given winter (Pan et al.,
1985). Using NCEP/NCAR reanalysis data for 1979–95, Zhang et al. (1997) report an
average of thirteen cold surges during October–April, of which two are typically strong
events. Surges evolve over five to fourteen days and last about nine days. They may travel
as far as Singapore (1°N, 104°E) within a couple of days, although their cooling effect
is negligible in low latitudes. In northwestern China, Mongolia, and eastern China the
temperature reduction is typically 10°–12°C, but drops of 20°C or more are common in
coastal areas of southern China. The frequency of cold surges over the South China Sea
(Hainan Island) is found to be low (high) during El Niño–low SOI (La Niña–high SOI)
conditions (Zhang et al., 1997). However, the interannual variability in cold surges of the
winter monsoon and ENSO is not yet understood, (see sections 5.2 and 5.4).
There are robust correlations between surge events over East Asia and anomalies in
low-level winds and convection over the Bay of Bengal, the eastern Indian Ocean,
Indonesia, and the western Pacific, according to Compo et al. (1999). Also, surges
over the South China Sea are linked with enhanced convection over Indonesia, the
Philippines, and the eastern Indian Ocean. Hence the surges are of more than regional
importance.
11
(a)
(b)
0
Figure 3.63 (a) The frequency of Siberian high-pressure centers calculated for 5° latitude–longi-
tude boxes for October–March, 1955–79. (b) Schematic tracks of high-pressure cells
11 over eastern Asia. (After Pan et al., 1985, and Ding and Krishnamurti, 1987)
The Australian monsoon
The austral summer monsoon in northern Australia is associated with westerly and north-
westerly cross-equatorial airflow in the sector 100°–160°E, from 5° to 12°–15°S. A
monsoon trough or shear line forms between these westerlies and trade wind easterlies
to the south, with heat lows over northwestern and eastern Australia augmenting the
forcing (McBride and Keenan, 1982). In northwestern Australia a regional heat low over
Pilbarra generates onshore westerly flow, identified as a pseudo-monsoon by Gentilli
(1971), while in northern Queensland the shear line/trough is disrupted by the topography
of the Great Dividing Range.
Monsoon onset and withdrawal over northern Australia can be identified in terms of
monsoon rains, lower-tropospheric westerly winds, or changes in upper-level winds
(Suppiah, 1992). For the Indonesian sector, Tanaka (1992, 1994) finds a simple zonal
pattern of monsoon advance and retreat when wind data are used. However, if cloud
images of convective activity are used, the monsoon onset is shown to be fifteen to sixty
days earlier over New Guinea and northern Australia than over the adjacent sea areas.
Figure 3.64 illustrates the monsoon season in 1978–79 defined by the 850 mb zonal wind
component at Darwin. During 1952–82 the mean onset and ending dates were December
24 and March 7 (Holland, 1986), but the seventy-four-day average season has a standard
deviation of twenty-five days. Drözdowsky (1996) finds mean dates to be five or six days
later, using surface 500 mb zonal winds for 1957–92. Much of the interannual variability
is related to ENSO characteristics in the preceding spring (September–October–November)
season. During warm (cold) events the summer monsoon trough shifts equatorward (south-
ward) (Evans and Allan, 1992). During onset there is strong low-level convergence over
northern Australia and the Arafura Sea to the north. Deep convection is often associated
Figure 3.64 Daily mean zonal winds at 850 mb over Darwin during October 1978–September 1979,
showing the summer monsoon season and active and inactive periods. (From Holland,
1986)
11 with an intersection of the northern and southern hemisphere Hadley cells, according
to Davidson et al. (1983). At 200 mb the tropical easterlies are observed to strengthen
over this sector, and a weaker summer subtropical jetstream shifts southward from 25°S
to about 40°S (McBride, 1987; Hendon and Liebmann, 1990). These changes mark part
of the annual cycle of the general circulation in the two hemispheres. In contrast to
southern Asia, however, the upper-level circulation over Australia in summer is only
weakly anticyclonic.
As elsewhere, the Australasian summer monsoon shows characteristic intervals of active
bursts and breaks. Typically there are one or two such cycles, each lasting about forty
0 days during the westerly season, that appear to be linked with the planetary circulation,
the SOI, and probably MJOs. Commonly there is a break regime in late January–mid-
February (Holland, 1986). Active phases are associated either with a broad area of
convection from the equator to 15°S, 100°–140°E, or with monsoon depressions and trop-
ical cyclones along a shear line over northern Australia (McBride, 1986). Monsoon cloud
bands are often located offshore over 29°C water, and the monsoon flow intensifies
following week-long episodes of deep convection which Mapes and Houze (1992) attribute
to MCSs. Their aggregate circulations spin up an active phase of the monsoon. Break
phases feature a shallow monsoon shear line displaced equatorward, although rainfall may
occur, with westward-moving squall lines (cloud lines), as it does in the pre-monsoon
0 period. These mesoscale systems include a northwest–southeast line of convective cells
(sometimes with the well known Morning Glory cloud line signature) that may also
have extensive stratiform cloud layers to the east, or a strong tropical squall line (Holland
et al., 1986).
During the cool season (April–October), 70–90 percent of the precipitation in north-
west Australia comes from cloud bands that originate west of 120°E, while northeastern
Australia has rains from cloud bands originating east of the same meridian (Wright, 1997).
3.7.4 The extratropics

0
The westerlies
Poleward of the subtropical high-pressure cells, the circulation in both hemispheres is
dominated by a circumpolar vortex with associated westerly wind belts, sometimes referred
to as the Ferrel Westerlies (Hare, 1960, 1962). For the zonally averaged u component,
the westerlies at sea level extend from 35° to 65° latitude in both hemispheres in summer
and from 33° to 67° in the northern hemisphere winter, and from 28° to 66° in the southern
hemisphere winter (Figure 3.65). The low values of the zonal wind component at the
surface in the northern hemisphere result from the variability in wind direction caused by
the frequency of traveling storm systems, the land–sea contrasts, and orographic and fric-
0 tional effects on airflow. For example, in the Scilly Isles, off southwest England, 46
percent of the winds are from between southwest and northwest, but 29 percent are from
the opposite quadrant. In contrast, Kerguelen Island (49°S, 70°E) has an 81 percent annual
frequency of winds from between southwest and northwest. The zonally averaged winds
in the southern hemisphere are strongest near 52°S in January and July. However, they
average 12–14 m s1 in both seasons over the South Atlantic and Indian oceans, compared
with 8–10 m s1 over the South Pacific (van Loon, 1972b). The wind maximum over the
South Indian Ocean is closer to the equator than that over the South Pacific in both
seasons. Although the area of the southern westerlies is greatest in the austral winter, they
are stronger at 45°–50°S in some places during summer (van Loon, 1972a; 1991). This
0 is related to the poleward decrease in the annual temperature range over the oceans from
30°S to about 50°S (Figure 3.66) caused by greater cloudiness and heat storage spread
through a deeper mixed layer in mid-latitudes and the location of the southern continents.
11 The greater annual temperature range in the subtropics, and the poleward temperature
Figure 3.65 Profiles of the average west-wind component (m s1) at sea level in the northern and
southern hemispheres during their respective summer (A) and winter (B) seasons.
(After van Loon, 1964)
decrease, result in a steeper latitudinal temperature gradient in summer than winter. From
55°S to 65°S surface winds have maxima in the transition seasons (Figure 3.67).
The velocity of the westerlies increases with altitude in accordance with the poleward
gradients of 1,000–500 mb thickness, as shown by the thermal wind relationship (see p.
266). Figure 3.68 illustrates latitude–height cross-sections for each hemisphere, showing
zonal winds and isotherms. In contrast to the barotropic structure of low latitudes, the
mid-latitude troposphere is weakly baroclinic, with narrow baroclinic zones related to the
polar front separating tropical and polar air masses and a northern hemisphere arctic front
separating polar and arctic air. There are also upper tropospheric baroclinic zones in the
subtropics associated with jetstream confluence. The strength of the mean jetstreams is
in part determined by their persistence in space and time. The location of the polar front
jet, for example, varies over a wide latitudinal range seasonally, in contrast to the subtrop-
ical jetstream. It is also temporally and geographically variable, in association with
traveling waves. The subtropical jetstream in northern winter is strongly constrained by
the orographic influence of the Tibetan Plateau and the Himalaya. The mean locations of
the 200 mb jetstream cores show a close relationship to the standing long-wave troughs
downstream of the Rocky Mountains and the Tibetan Plateau and to the land–sea bound-
aries of eastern Asia and eastern North America (see Figure 4.8). In the southern
hemisphere (Figure 3.69) the subtropical jetstream at 200 mb exceeds 40 m s1 over a
5°–10° latitude range from 90°E to 150°W, but in summer there is only a small core
exceeding 25 m s1 around the dateline (Berberry et al., 1992). In winter the subtropical
jet spirals poleward to merge with the subpolar jet over South America. The jet structure
gives rise to differing seasonal configurations of absolute vorticity. Near New Zealand
the 200 mb absolute vorticity field is nearly flat in winter between the polar and subtrop-
ical jets. The subtropical jet gives regions of maximum latitudinal gradient of absolute
vorticity along the jet core; negative latitudinal gradients exist near 40°S from 105°E east-
ward to 160°W in winter, and there is a separate band around Antarctica in both seasons.
11
Figure 3.66 Amplitude of the annual

0 wave in zonally averaged temperature
at the surface and 500 mb in both
hemispheres. (From van Loon, 1972)
Subpolar easterlies
Poleward of the westerlies at sea level in both hemispheres are weak polar easterlies (see
Figures 3.34 and 3.65). These represent the easterlies on the poleward sides of the Icelandic
and Aleutian lows in the northern hemisphere and the subpolar trough in the southern
hemisphere. There is no permanent high pressure over the Arctic Ocean, although polar
0 anticyclones are common in winter and spring over the Beaufort, Chukchi, and East
Siberian Seas sector (120°W to 140°E) (Serreze et al., 1993). Over Antarctica the topog-
raphy of the ice sheet which rises to 4,000 m in East Antarctica and averages about 2,200
m elevation, makes it impossible to analyze MSL pressures poleward of about 70°S,
except over the Ross and Weddell Seas. Indeed, the influence of the topography of
Antarctica is evident in the zonal asymmetry of the southern hemisphere circulation
(James, 1988) as described in section 4.3.
3.8 Centers of action

0 Mean sea monthly level pressure maps for the globe were first constructed by Alexander
Buchan (1869), who described the zones of prevailing winds. The mean centers of high
and low sea-level pressure were termed centers of action by Teisserenc de Bort (1883)
11 because they represent the principal circulation features that help determine mid-latitude
Figure 3.67 The annual march of

the zonally averaged MSL zonal
geostrophic wind (m s1) (From van
Loon, 1972)
weather and wind systems. Indeed, he described “anticyclones as regulators of weather.”

The centers are clearly defined in the northern hemisphere, whereas in the more oceanic
southern hemisphere (see Table 1.1) the subtropical high and subpolar low-pressure areas
form more or less continuous belts around the hemisphere (see Figure 3.34). There is,
however, a fundamental difference between a mean high and a mean low-pressure center.
The former represents a quasi-stationary system that changes little from day to day; the
latter is an area into which deep cyclones move with high frequency (Petterssen, 1950).
Thus mean pressure charts need to be compared with maps of the frequency and tracks
of cyclone centers, the frequency of cyclogenesis/cyclolysis, and deepening rates for a
proper understanding of their significance (see section 6.4). Nevertheless, the mean
subpolar lows are also stationary disturbances of the zonal flow maintained by the atmos-
phere’s thermal structure, due to land sea contrasts, and by the effects of orography (Spar,
1950; Smagorinsky, 1953).
The four principal centers of action in the northern hemisphere are the Icelandic and
Aleutian lows, and the Azores and North Pacific highs. Their long-term locations and
central pressure averages, based on 5° × 5° gridded mean monthly data derived from the
US Historical Weather Map Series for 1899–1978, are shown in Table 3.14, although
these estimates are relatively crude, owing to limited data coverage in the early record,
approximations caused by the data grids, and uncertainties when two or more locations
had identical pressure values. It is interesting that in the North Pacific the centers are
11 Table 3.14 Long-term mean locations and central pressures of the
northern hemisphere centers of action
Feature Central pressure Location

(mb)
Aleutian low 1,002 56°N, 168°W

Icelandic low 1,001 60°N, 36°W
Azores high 1,024 33°N, 35°W
North Pacific high 1,024 35°N, 143°W
0
Source: from Angell and Korshover (1982).
displaced longitudinally relative to one another, in contrast to the North Atlantic. However,
there are major changes over the annual cycle, as discussed below.
It must be noted that the vertical structure of high and low-pressure centers differs con-
siderably. The Siberian winter anticyclone is a shallow, cold high; the hydrostatic equation
(See Section 1, n. 1) implies that in a cold air column the isobaric surfaces are lowered
with height. The equivalent case of a heat low, in summer over the southwestern United
States for example, has isobaric surfaces that are raised with increased altitude within the
0 warm air column. In contrast, the dynamically induced subtropical highs are warm at all
levels. The axis of the highs in the northern tropics slopes westward and southward with
increasing altitude, corresponding to the locations of warmer air in the upper troposphere.
These highs are maintained by upper-level convergence and subsidence.
The characteristics of each of the main centers of action are briefly described.
3.8.1 The Siberian high

In the winter half-year a shallow semi-permanent cold high extends over most of Asia,
centered about 45°–50°N, 90°–110°E (Perry, 1971; Sahsamanoglou et al., 1991). For
0 1873–1988 mean monthly central pressure values exceed 1,027 mb from October through
March, with highest mean monthly values of 1,037 mb in January. Daily values are often
observed to be 1,050–1,070 mb (Ding, 1994). Walker (1967) notes that the sea-level pres-
sures are problematic because many of the stations are in valley or basin locations. In
winter these experience extremely low temperatures and persistent inversions, making the
adjustment of barometric pressure to sea level dubious, since normal lapse rates are
assumed. Nevertheless, high-pressure centers are observed to move out of the region across
the Arctic coastline. Sahsamanoglou et al. report that 1,000–500 mb layer mean temper-
atures over the anticyclone average only 250–5 K during December–March, consistent
with a cold high structure. Surface pressure values are inversely correlated with the
0 1,000–500 mb thickness, especially for November–February. Moreover the intensity of
the high has weakened slightly since 1980, apparently in response to a warming trend.
In the middle and upper troposphere there is a trough over eastern Asia (Gommel, 1963).
The dynamic structure of the Siberian high features cold highs that move (1) eastward
from southern Europe across central Asia; (2) southeastward from Scandinavia across the
Urals; (3) southeastward from the Kara Sea; and (4) southward from the Taimyr Peninsula
(Figure 3.63) (Ding et al., 1991; Ding, 1994). The Siberian high plays an important role
in the winter climate of East Asia. It gives rise to frequent outbreaks of cold air over
China as high-pressure cells move southeastward and the East Asian trough shifts east-
ward to near 140°E and deepens (Ding and Krishnamurti, 1987). Their analyses show
0 that the development of a surface cold anticyclone typically involves northwesterly flow
in the middle and upper troposphere associated with the western limb of a trough off East
Asia. This northwesterly flow provides significant cold air advection aloft which persists
11 while surface high cells propagate southeastward.
0
0
0
0
0
11
11
(a) (c)
(b)
(d)
Figure 3.68 Latitude–height cross-sections of (a) mean zonal wind, u, and (b) meridional wind, v (m s1), (c) potential temperature, T
(K), and (d) vertical motion, (102 Pa s1), for January 1987–89 (above) and July 1986–88 (below), from ECMWF
analyses. The contour intervals are: 5 m s–1 for u, 0.5 m s–1 for v, 5K for T, and 1 × 10–2 Pa s–1 for . (From Trenberth,
1992)
Figure 3.69 Mean locations of southern hemisphere 200 mb level jetstreams and mean absolute vorticity (contour interval 1 105 s1). (a) Winter. (b)
Summer. (Berberry et al., 1992)
11 The dynamics of the anticyclone can be understood from consideration of the profiles
of relative vorticity and diabatic heating (Ding, 1994). Based on five cases, the inception
of anticyclogenesis is associated with positive vorticity advection in the middle and upper
troposphere. During this inception stage there is convergence in the lower and upper
troposphere, with divergence in between. During the mature stage, relative vorticity is
negative throughout the troposphere as a result of advection of anticyclonic vorticity in
the mid-upper troposphere and low-level divergence. The heating profiles show a heat
sink of about 2ºC day1 in the lower middle troposphere due mainly to radiative cooling,
with a weak heat source in the upper troposphere related to horizontal advection. The
0 adiabatic cooling induces subsidence in the mid-lower troposphere which results in upper
(lower)-level mass convergence (divergence).
In summer, most of Asia is covered by a broad low-pressure system centered over
northwestern India. In early summer this is primarily a surface heat low, but the elevated
terrain over much of the area makes the MSL pressures largely hypothetical.
3.8.2 The North Atlantic subtropical anticyclone

The subtropical high-pressure center in the North Atlantic is a prominent center of action
which, in combination with the Icelandic low, determines the circulation pattern over the
0 North Atlantic (Tucker and Barry, 1984). Only exceptionally is this mean pattern reversed
(Moses et al., 1987). Using sea-level pressure data for 1873–1980, Sahsamanoglou (1990)
shows that the monthly mean central pressure exhibits maxima in July (1,026 mb) and
January (1,024 mb) and minima in October (1,022 mb) and March (1,023 mb). The center
is located mainly between 25°–40°N, 20°–45°W. There is a seasonal clockwise displace-
ment of the center from about 30°N, 25°W (the Azores high) in December, westward to
40°W in July (the Bermuda high) and then northward to 35°N in September and east-
ward to 20°W in November, before returning southward. A correlation of the central
pressure individually with the latitude and longitude of the center (Sahsamanoglou, 1990)
indicates a tendency for the pressure to be higher when the center is farther northward;
0 this also occurs for an eastward displacement during December–February, April and
September–October.
An independent study for 1899–1990 by Davis et al. (1997) provides daily frequen-
cies of pressure values ≥ 1,020 mb over a 5° 5° latitude–longitude grid. This reveals
two principal spatial modes of the anticyclone – a single maximum centered over the
central Atlantic (30°N, 35°–40°W) in summer, and a winter pattern of weaker, dual
maxima with cells over the southeastern United States (32°–35°N, 83°–85°W) and off
northwest Africa (30°N, 22°W in January). The winter mode accounts for 26 percent of
the variance and the summer one for 18 percent. There is, however, substantial variability
in the frequencies of high pressure in the core region and around its fringes. Moreover,
0 Davis et al. (1997) report a long-term decline in the frequency and intensity of the Azores
high, which appears to represent a real net loss of mass over the North Atlantic. A similar
conclusion is reached by Perry (1971) and Sahsamanoglou (1990). Central pressure in the
Azores anticyclone increased from 1873 to 1914 (by 8.8 mb/100 yr), decreased from 1914
to 1963 (5.2 mb/100 yr), and then rose again.
3.8.3 Icelandic low

The subpolar Icelandic low is centered between 55°–70°N, 10°E–60°W, with a monthly
mean central pressure that varies between 994 mb in January and 1,008 mb in May
0 (Sahsamanoglou, 1990). Like the Azores anticyclone, the center of the Icelandic low
shows a seasonal movement, but in an east–west direction. From 60°N, 30°W during
February–April the center moves to 70°W in July–August, returning to 30°W during
11 September. However, in July there is a secondary frequency maximum of the low center
appearing at 18°–30°W, while in March–April this secondary maximum appears at 0°–
10°E. The area from Iceland to the Kara Sea is characterized not only by a broad maximum
of cyclone frequency in winter, but also by maximum deepening rates of 6.8 mb/12 hr
near Iceland. In summer this area is represented by cyclone deepening rates of only
3.3 mb/12 hr and regional contrasts in cyclogenesis and cyclone occurrence are far less
pronounced (Serreze, 1995).
Between 10 percent and 15 percent of cyclone events in the Icelandic low represent
cyclogenesis and 13–18 percent (depending on the month) involve cyclolysis (Serreze
et al., 1997). About half of the cyclones within the Icelandic low develop north of 55°N,
but others can be traced upstream to the Rocky Mountain areas of Alberta or Colorado.
Winter cyclone events are shown to be twice as frequent during the positive mode of the
North Atlantic Oscillation (see section 5.5) as during the negative mode.
Unlike the Azores anticyclone the intensity of the Icelandic low remained almost
constant between 1973 and 1980 (Sahsamanoglou, 1990). However, the central pressure
decreased in winter in the late 1980s (Machel et al., 1998). The mean center drifted west-
ward at 3.7° per year southward at 1.7° per year, with maximum rates during 1873–90
and 1910–30. During winter 1955–70 the center shifted 1°–2° southward of its 1885–1995
mean (59.2°N), moving back northward in the 1980s. Since the 1970s there has also been
a significant eastward displacement of the low in summer.
3.8.4 Aleutian low

The Aleutian low resembles the Icelandic low in intensity. Its characteristics are reported
using data for 1960–85 by Martynova (1990) and for 1891–1990 by Perevedentsev et al.
(1994). It is deepest (approximately 995 mb) in December–January and weakens to about
1,010 mb in June. The center shifts from near 50°N in January to 60°–65°N in August,
it is near the dateline in winter, but moves eastward to about 160°W in the autumn
(Martynova, 1990).
Interannually, the strength of the mean Aleutian low in winter months fluctuates between
a mode when the center is near 50°N, 170°W, with a central pressure below 1,000 mb,
and another where the low is weaker and farther west (near 165°E), with high pressure
over Alaska (Rogers, 1981). These modes correspond respectively to winters with below
(above) normal temperatures in the Aleutians (western Canada), and the opposite pattern.
This North Pacific pressure oscillation is described in section 5.6. Overland et al. (1999)
show that 37 percent of the wintertime interannual variability in the Aleutian low is on
a time scale longer than five years. It was a prominent deep low after 1977, but returned
to near average strength in 1989.
3.8.5 North Pacific subtropical anticyclone

The North Pacific anticyclone, according to Martynova (1990) and the data of Pereve-
dentsev et al. (1994), shows less latitudinal displacement, but more longitudinal
displacement, than its Atlantic counterpart. From about 30°N in January–February it moves
to 40°N in July. During this shift it strengthens from 1,021 mb to 1,026 mb. Its mean
minimum intensity is reached in October (1,020 mb), as in the North Atlantic subtrop-
ical anticyclone. The longitudinal displacements are between about 135°W in November
and 150°W in July (Hastenrath, 1991, p. 137).
3.8.6 The Arctic high

Simple schemes depicting the global circulation usually show high pressure over the Arctic
and subpolar easterly winds north of the subpolar low pressure centers. The concept can
be traced to the theoretical ideas of Helmholtz (1888), who argued on dynamical grounds
11 for a surface high beneath the cold tropospheric polar vortex. The ideas were further
developed in the “glacial anticyclone” theory of Hobbs (1926, 1945), subsequently refuted
for Greenland by Matthes (1946) and Matthes and Belmont (1950). In the virtual absence
of data north of the Arctic Circle, synoptic analysts adopted these climatological concepts
and extrapolated the pressure fields over the Arctic in preparing the US Weather Bureau
historical series of daily northern hemisphere weather maps for 1899–1945. Nevertheless,
arctic expeditions noted the high synoptic variability of the marginal seas and, as early
as 1945, B.L. Dzerdzeevski noted frequent cyclonic activity in the Arctic Basin, espe-
cially in summer. Jones (1987) suggests a positive bias of about 4–6 mb prior to 1930
0 in the US Weather Bureau analyses over the central Arctic.
The establishment of stations in the Canadian High Arctic by the early 1950s and the
deployment of Arctic Ocean drifting buoys since 1979 have greatly improved the data
coverage. More recent surveys using these data (Serreze and Barry, 1988; Colony and
Rigor, 1992; Serreze et al., 1993) indicate that during winter there is more usually a ridge
of high pressure over the Beaufort–Chukchi–East Siberian seas linking centers over Siberia
and the Mackenzie–Yukon, with a trough over the Barents Sea. Serreze and Barry show
highest winter anticyclone frequency (1979–85) over the Beaufort Sea between 75°N and
90°N, 140°W to 160°E. Cyclones from the North Atlantic regularly track into the Arctic
in winter north of Novaya Zemlya (75°N, 20°–80°E) and south of Svalbard (75°N,
0 20°–40°E). These patterns largely persist through spring, although pressures are now
highest (1,022 mb) off the Queen Elizabeth Islands, centered about 80°N, 130°W
(Maxwell, 1980). There are no strong maxima of highs or lows over Greenland. MSL
pressures are, of course, largely fictitious, but upper lows do occasionally cross the ice
sheet at the 700 mb level (Hamilton, 1958). Atmospheric pressure at all levels in the
troposphere and stratosphere are higher (lower) in northern latitudes (south of 40ºN) in
spring than in autumn; this difference between the transition seasons strengthens with alti-
tude and the latitudinal pressure gradient intensifies. Fleming et al. (1987) attribute this
to the fact that tropospheric mean temperatures are lower in spring than in autumn in
middle and higher northern latitudes. In the northern hemisphere the annual temperature
0 cycle in the middle and upper troposphere lags about forty days behind that of incoming
solar radiation.
A major change occurs in summer (Figure 3.70). Anticyclones, with central pressures
of about 1,025 mb, occur in a band about 75°–80°N eastward from near Novaya Zemlya
to the Beaufort Sea. A mean low is usually present over the Canada Basin for at least
one month during the period July–September. For August 1979–1985, the mean low over
the Canada Basin (85°N, 120°W) had a central pressure of about 1,006 mb, comparable
with the mean Icelandic low in summer. Moreover, for individual three-to-four-week
periods the central pressure may average as low as 995 mb, as occurred in mid-
August–mid-September 1980 (Serreze, et al., 1989). This represents a recurrent tendency,
0 in about two years out of three, for a cyclonic system to move into the area, from northern
Eurasia mainly, become stationary, and persist for up to a month (Serreze and Barry,
1988). However, this sector is characterized only by cyclolysis (Serreze, 1995).
3.8.7 Southern hemisphere centers of action

The southern hemisphere is 81 percent ocean surface, and three-fifths of the land area is
concentrated between the equator and 30°S. The mean circulation is much less cellular
than in the northern hemisphere. Nevertheless, important zonal asymmetries in atmos-
pheric and oceanic characteristics are apparent in association with the topography of the
0 Andes cordillera and eastern Antarctica, the relative sizes of the South Atlantic, Indian
and Pacific Oceans (approximately 2, 3 and 5 respectively), and the cold Benguela and
Peru currents and associated upwellings off southwest Africa and western South America
11 (Taljaard, 1967).
Figure 3.70 Mean sea-level pressure over the Arctic

Ocean. (a) Annual. (b) For August 1979–85. The
arrows indicate motion of drifting buoys on the sea
ice. (From Serreze and Barry, 1988; Serreze et al.,
1989)
The most striking element of mean sea-level pressure maps for the southern hemisphere
is the pronounced circumpolar trough of low pressure, located between about 60°S and
65°S (van Loon, 1984; Figure 3.71). This climatological feature is produced by mature
mid-latitude depressions that originate primarily in the subtropical South Pacific and South
Atlantic Oceans and spiral clockwise toward Antarctica with an additional contribution
from high-latitude polar lows and cold air mesocyclones (van Loon, 1972b; Trenberth,
1981a; Carleton, 1992). However, cyclone tracks in the southern hemisphere are less well
defined than their northern hemisphere counterparts (Physick, 1981). The trough expands
(contracts) and simultaneously weakens (intensifies) in a semi-annual oscillation (see
below). Mean zonally averaged pressures are lowest (980 mb) in September (spring); they
average 983 mb in July–August–September and 986 mb in January–February–March
(Hurrell et al., 1998). North of the circumpolar trough there is a zonal belt of subtrop-
ical high pressure. The zonally averaged subtropical ridge (STR) is centered on 28°S in
June–July–August and 33°S in December–January–February. However, the three subtrop-
ical anticyclone centers are located farther south, with highest mean central pressures at
38°S in June–July–August and 46°S in December–January–February (Jones and
Simmonds, 1993). The displacement of the mean STR towards the subtropics suggests
greater persistence, associated perhaps with blocking and/or cold air surge events (Sinclair,
1996). In the winter half-year (May–October) the anticyclones have the following loca-
tions and mean central pressures: South Indian Ocean (40°S, 60°E; 1,023 mb); South
Atlantic Ocean (40°S, 5°W; 1,022 mb) southeast Pacific Ocean (40°S, 90°W; 1,020 mb).
In the summer (November–April) they are 2–3 mb weaker, but occupy almost the same
11
0
Figure 3.71 Elements of the circulation over the Southern Ocean (from Carleton, 1992). Variability
of the mean seasonal axis of the circumpolar trough, 1972–77 (from Streten, 1980).
Relative frequency with longitude of monthly mean low pressure centers in the trough
(1972–77). Tracks (heavy arrowed lines: major; thin arrowed lines: secondary). Mean
location of the continental anticyclone center (1972–77), September/February sea ice
limits (Jacka, 1983)
locations in the South Atlantic and South Pacific, while the Indian Ocean center is now
0 near 85°E. The subpolar lows are also deepest in winter. Central pressures average 976
mb in the Ross Sea, near 70°S, 150°W; 977 mb near 63°S, 100°E, and 980 mb near
65°S, 25°E. In summer the last two centers occupy almost the same locations and are
only slightly weaker, while the South Pacific center is displaced eastward near 82°E, with
a central pressure of 980 mb.
Over Antarctica monthly mean charts for the lower-middle troposphere (800–500 mb)
indicate that an anticyclone is generally present over the high plateau of East Antarctica,
with a cyclone over West Antarctica (Taljaard, 1972; Schwerdtfeger, 1970, 1984;
LeMarshall et al. 1985). This pattern undergoes little seasonal variation. Some cyclones
penetrate into the continental interior from the Ross Sea across Marie Byrd Land. However,
0 daily 500 mb maps show that very few cyclones affect East Antarctica, although the high
may split into two centers (Astapenko, 1964).
11
3.8.8 The semi-annual oscillation in the southern hemisphere
The tropospheric circulation in the southern hemisphere features a well known semi-
annual oscillation (SAO) in the extent and strength of the circumpolar low-pressure trough
(Schwerdtfeger, 1960; van Loon, 1967). The maximum amplitude of the SAO at sea level
is observed at 55°–60°S. The amplitude is ≥ 3 mb in high latitudes, where it accounts
for 50 percent or more of the variance in pressure (Simmonds and Jones, 1998). The
annual wave dominates over the three continents in lower latitudes but the semi-annual
component is evident in middle latitudes. The trough contracts and intensifies between
June and September, and again from December to March, and it expands and weakens
from March to June and from September to December. Pressure difference maps between
these seasons illustrate the changes in the trough and equatorward to 25°S. The June mean
minus that for March shows the expansion of the trough over the Southern Ocean, espe-
cially the South Pacific, contrasting with pressure rises in middle latitudes over the
southern continents (Figure 3.72b) (van Loon and Rogers, 1984b).
This oscillation is also apparent in the troposphere, and van Loon (1967) used the
difference in zonal mean 500 mb temperature between 50°S and 65°S as a suitable index
(Figure 3.73). Fifty-nine percent of the variance in this difference is attributable to the
second harmonic, although the first harmonic is heterogeneous in phase in the zonal means
and so its contribution is reduced in the averaging (van Loon and Rogers, 1984a).
The forcing of the oscillation arises from the differences in annual temperature cycle
in the troposphere over the mid-latitude oceans and Antarctica (Schwerdtfeger,1960; van
Loon, 1967; Mo and van Loon, 1984), and also between continents and adjacent oceans
in the subtropics and lower mid-latitudes (van Loon, 1972b) (Figure 3.66). At 50°S ocean
heat storage delays the seasonal maximum and minimum of SST by two or three months
from the extremes of net surface heat flux. Heat storage also influences the more rapid
fall of SSTs from their maximum in late summer and their slower rise from the late winter
minimum. The decrease in autumn is facilitated by deep vertical mixing, whereas the
spring rise is slowed by the strong winter cooling of the surface layer. These contrasts
in heat capacity and storage influence the longitudinal location and amplitude of the tropos-
pheric waves, especially those between Australia and the Indian and South Pacific Oceans
(van Loon, 1967, 1972b). They are revealed in maps showing the differences in SLP
between mid-season months, whereby pressures rise (fall) over the continents (adjacent
oceans) between March and June, and fall (rise) over the same areas between June and
September (Mo and van Loon, 1984). The resulting changes in the amplitude and longi-
tude position of the waves are also manifest in the meridional wind (v component). A
signal reminiscent of the SAO has been identified in the seasonal changes in latitude of
the speed maximum associated with the Antarctic Circumpolar Current (Large and van
Loon, 1989). The SAO is also believed to play a part in the asymmetric seasonal regimes
(advance, retreat) of the Antarctic sea ice cover occurring via surface wind stress asso-
ciated with the circumpolar trough (Enomoto and Ohmura, 1990).
The seasonal marches of the SLP for islands in southern middle latitudes and the
Antarctic are out of phase (Streten and Zillman, 1984), manifesting the SAO over sub-
antarctic latitudes. The resulting intensification (relaxation) of the gradients of SLP and
zonal wind speed (Figure 3.67) during the equinoctial (solstitial) months produce twice-
yearly changes in the preferred latitude locations and mean central pressures of the
Antarctic circumpolar trough. These comprise two equatorward (in December, June) and
two poleward (March, September) latitude extremes of the trough, when it is also weaker
(stronger) (van Loon, 1967). These excursions are associated with the preferred patterns
of extratropical cyclogenesis in middle latitudes and cyclone tracks into the “graveyards”
(cyclolysis regions) close to Antarctica (Carleton, 1981; Howarth, 1983). Satellite image-
based studies of synoptic-scale cloud vortices in the southern hemisphere (Carleton 1981)
show the strong latitude variations in cyclone occurrence that occur between early winter
11
0 Figure 3.72 Mean sea-level pressure difference for (a) March minus December, (b) June minus
March, (c) September minus June, and (d) December minus September, illustrating the
semi-annual oscillation in the southern hemisphere. (From van Loon and Rogers, 1984)
(June) and late winter (September), in association with the changes in the time-averaged
circumpolar trough associated with the SAO. There is also a signature of the SAO in the
patterns of formation and movement of cold-air mesocyclones over the southern oceans
when considered on weekly to monthly time scales (Carleton and Carpenter, 1990;
Carleton and Song, 1997).
Numerical modeling confirms the primary mechanism of the SAO to be the intensifi-
0 cation and relaxation of the temperature gradient asymmetrically between low and high
latitudes (Meehl, 1991). Anomaly experiments to simulate the impacts on the SAO of
leveling the Antarctic continent and removing the sea ice reveal influences of both but
11 with a stronger sensitivity to the elevation of the ice sheet (Walland and Simmonds, 1998).
Figure 3.73 The difference in zonal mean 500 mb temperature between 50°S and 65°S (dotted
line). Dashed line = first harmonic, solid line = second harmonic. Their variance contri-
bution is shown. (Meehl, 1991)
This is in accord with the apparent influence of East Antarctica in helping to generate
baroclinic instability at high latitudes (Mechoso, 1980; James, 1988). Walland and
Simmonds (1998) suggest that cloud-radiative forcing is the common factor involved in
the results of both experiments, possibly indicating the importance of the higher-latitude
cloudiness regime to the SAO over sub-antarctic latitudes.
There are interannual and also decade-scale changes apparent in the strength of the
SAO that are not explained by differences in data sets between study periods (van Loon
and Rogers, 1984a; Mo and van Loon, 1984). The SAO may be linked to the ENSO: it
is weaker when the westerlies are stronger and temperatures are below-normal in East
Antarctica; as occurs during a warm event (Van Den Broeke, 1998b). The SAO was
strong in the 1970s, when it was the dominant signal in middle and high latitudes of the
southern hemisphere. The SAO weakened in the early 1980s and, in mid-latitudes, in the
early 1990s. On a multi-year time scale, since about 1979 there has been a weakening
and delay into November of the springtime phase of the SAO (Hurrell and van Loon,
1994), especially near Antarctica (Simmonds and Jones, 1998). Simmonds and Jones
(1998) suggest that the variability of the SAO in the high-latitude temperature gradient
could be related to the effects of sea ice extent on the atmospheric dynamics. The change
in the springtime phase may have important implications for the intensity and longevity
of the Antarctic “ozone hole” since that time. Its source is believed to be in the strength-
ened latitude gradient of pressure and temperature that accompanied trends to increased
SSTs over lower latitudes and decreased SLP in the Antarctic circumpolar trough, notably
in the Pacific sector. This change in the temperature gradient is not distributed evenly
throughout the year at each latitude (Meehl et al., 1998), and so the SAO has become
modulated to produce peaks in approximately May and November, rather than in March
and September. Possible “fingerprints” of these changes are an observed intensification
of the poleward-directed fluxes of sensible heat accomplished by the atmospheric eddies
during the 1980s (van Loon and Kidson, 1993), and significant cooling in May and June
at stations in East Antarctica since the mid-1970s (Van Den Broeke, 1998b).
Over the Antarctic continent the regional expression of the SAO is apparent in the
annual temperature cycle which features a “coreless (or kernlose) winter.” This refers to
the lack of a clear midwinter minimum; instead the temperature curve is almost flat during
the winter months after a steep decrease during autumn. Advection of mid-latitude
maritime air and eddy heat flux towards the interior via transient eddies help to delay the
11 annual temperature minimum until early spring (van Loon, 1967; Meehl, 1991). Warm
air with accompanying increases in cloudiness are advected into the region on the eastern
side of a long wave trough (Stearns and Wendler, 1988). These changes affect domi-
nantly wave No. 3 (Van Den Broeke, 1998a). Data from additional manned stations and
also the automatic weather station network show this coreless winter to be widespread in
the Antarctic (Bromwich, 1988; Allison et al., 1993), especially along the west coast of
the peninsula (Van Den Broeke, 1998a). Moreover, it is present in most climate variables,
particularly wind speed and precipitation (e.g. Turner et al., 1997).
0
3.9 Global climatic features
There are some fundamental differences in the characteristics of the atmosphere between
the tropics and extratropics that have profound significance for their weather and climate
(Charney, 1973; Asnani, 1993). These can be summarized as follows:
1 In the tropical easterlies (mid-latitude westerlies), the atmosphere gains (loses) rela-
tive westerly angular momentum, owing to friction at the surface interface. The
low-latitude gain is transferred to middle latitudes, where it is dissipated by friction.
The westerlies would cease in about ten days without this continual resupply from
0 low latitudes.
2 The Earth–atmosphere system in low latitudes has a positive net radiation balance, on
average to about 35° latitude. Poleward there is a net deficit; the tropical surplus is con-
tinuously offset by poleward meridional energy transfer in the atmosphere and oceans.
3 The lower and middle troposphere in the tropics (extratropics) is generally convec-
tively unstable (stable).
4 In the tropics (extratropics) the troposphere is baroclinically stable (unstable) with
respect to dry adiabatic processes. There are hyperbaroclinic (frontal) zones in the
extratropics.
5 There is, respectively, weak (strong) vertical coupling in the tropical (extratropical)
0 troposphere. Strong vertical coupling in the tropics requires moist adiabatic processes
and condensation is necessary for synoptic-scale dynamic instability. Without conden-
sation and latent heat release, vertical motion in the tropics is an order of magnitude
smaller than in mid-latitudes, so that the motion in such cases is quasi-barotropic and
quasi-nondivergent.
6 The Coriolis parameter, f, has a value of approximately 105 s1 in the Tropics and
104 s1 in middle latitudes. Relative vorticity () is of similar magnitude to f in the
tropics but smaller than f in the extratropics.
7 The tropics have quasi-stationary seasonal wind regimes with high zonal wind
constancy. There are superimposed slow oscillations in position and intensity. The
0 extratropics are characterized by irregularly fluctuating regimes of zonal and merid-
ional circulation on submonthly time scales.
8 The extratropics (tropics) are affected by only westerly (both easterly and westerly)
waves.
9 Tropical (extratropical) cyclones are warm (cold) core systems. Tropical systems are
generally smaller in extent but have greater intensity than extratropical cyclones.
10 Average daily totals are of the order of 1 cm (3 cm) for precipitation areas in
the extratropics (tropics). Diurnal precipitation regimes are strongly developed in the
tropics and there is a semidiurnal (twelve-hour) pressure wave.
0
3.9.1 Global cloudiness and precipitation characteristics
The cloudiest regions of the globe are associated with the extratropical storm tracks over
11 the Southern Ocean around 60°S, the mid-high latitude North Pacific and the North Atlantic
(London et al., 1989). In these areas, cloudiness amounts average 80–90 percent in winter
and summer. The summer cloudiness over the Arctic Ocean is similarly high. The least
cloudy region is the Saharan–Arabian desert, where amounts are only 10–20 percent in win-
ter and 20–30 percent in summer. Low amounts are also found over much of Australia and
southwest Africa. There is a second equatorial belt of cloudiness associated with tropical
cloud clusters along the ITCZ and over the tropical Indian Ocean. Figure 3.74 illustrates
the average zonal distributions for summer and winter 1971–81, based on surface obser-
vations. The dominance of the Southern Ocean peak is especially striking.
There is still considerable uncertainty in total global cloudiness. Hughes (1984) showed
the discrepancies existing in publications up to the early 1980s. The statistics assembled
under the International Satellite Cloud Climatology Project (ISCCP) showed more global
cloud than earlier studies, probably as a result of more complete coverage of ocean areas,
particularly the Southern Ocean. Global cloudiness is about 60 percent when mapped on
a 250 km scale, compared with earlier estimates of 50 percent (Rossow and Schiffer,
1991). The initial ISCCP algorithms had difficulty in mapping cloud over snow and ice
surfaces, although adjustments to address this have since been made (Rossow and Garder,
1993).
Statistical analysis suggests that most variation in cloud cover takes place on space
scales of 50–250 km and time scales of a few days (Rossow, 1993). Additionally, mid-
latitude land areas have large diurnal changes in cloud amount, especially in summer.
There is typically more cloud in mid-afternoon, related to the maximum solar heating of
the surface and convective cloud development.
The distribution of cloud types has been investigated using surface observations (Warren
et al., 1986, 1988). A statistical correction is made to minimize the effect on the obser-
vations of the obscuration of higher clouds by lower cloud layers. Figure 3.75 shows the
results for land areas, for 1971–81, and ocean areas, 1952–81. Stratiform cloud is most
common over land and ocean surfaces in higher latitudes during both seasons, while
nimbostratus is predominant in higher middle latitudes. Cumuliform clouds predominate
in the equatorial latitudes, although convective clouds show a secondary maximum over
northern land areas around 55°N in summer. Clouds tops are generally highest in the
tropics, but the high-level cirrus canopies that spread out above the convective towers
bias the areal coverage. In contrast, extratropical cloud systems tend to be large-scale
multilayered systems formed through slantwise (or slope) convection (Ludlam, 1966,
1980). Each system may cover a 106 km2 area and Stewart et al. (1998) consider that
there are ten such systems at any one time over the Earth’s surface. The ISCCP data on
cloud type and total cloud optical thickness are used by Lau and Crane (1995) to develop
a composite classification of cloud top height and thickness. The distribution of eight
cloud top-thickness categories is mapped for October–March and April–September
1983–90, between 60°N and 60°S, and the seasonal climatology is shown to be consis-
tent with the corresponding dynamical characteristics and atmospheric structure in different
areas of the tropics, subtropics, and extratropics.
The mean global distribution of precipitation based on station records for
December–February and June–August is shown in Figure 3.76. The data have been
adjusted for gauge undercatch but probably are still too low where much of the precipi-
tation falls as snow. The Global Precipitation Climatology Center (GPCC) in Offenbach,
Germany, is now producing monthly maps on a routine basis that blend station data and
estimates derived from satellite infrared and passive microwave data (see section 2.2).
They are also endeavouring to incorporate appropriate corrections for the numerous gauge
types and wind-shielding devices employed by different national agencies. A number of
major features are apparent in Figure 3.76:
1 The near-equatorial maximum is displaced into the northern hemisphere over the
Pacific and Atlantic Oceans. In northern summer it is located in the tropics in southern
11
Figure 3.74 The distribution of total cloud amount (percent) for 75°N to 75°S derived from surface
observations for 1971–81 for June–August (above) and December–February (below).
High values are hatched, low values stippled. (From London et al., 1989)
0
11
Figure 3.75 Percentages of zonally averaged cloud amount according to cloud type. (a) Land areas,
December–February 1971–81. (b) Land areas, June–August 1971–81. (c) Ocean areas,
December–February 1952–81. (d) Ocean areas, June–August 1952–81. Ci, Cirriform;
As, Alto-forms; Cu, Cumulus; Cb, Cumulonimbus; St, Stratus and stratocumulus; Ns,
Nimbostratus. (After Warren et al., 1986, 1988; from Grotjahn, 1993)
and eastern Asia. These patterns reflect the ITCZ occurrence and the Asian summer
monsoon regime.
2 There are prominent tropical–subtropical dry zones, extending into the eastern parts
of the major oceans, which are associated with subsidence in the subtropical anti-
cyclones.
3 Mid-latitude west-coast areas have large totals in the respective winter seasons asso-
ciated with the extratropical disturbances in the westerlies and augmented by
orographic effects over the coastal mountains in North and South America. There are
also large totals in the austral summer over the stormy Southern Ocean.
4 Low precipitation in high latitudes and in winter in the interiors of the northern conti-
nents is attributable to low moisture content in the very cold air. Nearly all of this
precipitation falls as snow.
3.9.2 The tropics

Recent advances in ground-based, aircraft, and satellite measurements provide consider-
able information on the quantitative characteristics of tropical cloudiness and precipitation
and help shed light on the regional variations in circulation processes. Over the eastern
parts of the subtropical and tropical oceans, large-scale subsidence of about 40 mb day1
11
0 Figure 3.76 Mean global precipitation (mm day1) for December–February and June–August. (After
Legates, 1995)
occurs in the subtropical anticyclones, and cold coastal currents cause stratiform cloud to
predominate (Figure 3.77; Klein and Hartmann, 1993) and only drizzle or light rain usually
falls. The five regions with such conditions are summarized in Table 3.15. The seasonal
cycle of cloud amount is closely related to that of static stability and inversion strength
rather than to the intensity of the anticyclone or the trade wind divergence (Klein and
Hartmann, 1993); stratus amount increases about 6 percent for each degree Kelvin increase
in static stability, both seasonally and interannually. It also decreases with higher sea
0 surface temperatures, but the relationship is not as strong. In the California region, where
there is extensive summer stratiform cloud, the boundary layer is found to be well mixed.
Observations during the First ISCCP Regional Experiment (FIRE) in 1987 show that the
11 cloud base can be around 200–400 m, with the stratus only a few hundred meters deep.
Figure 3.77 Mean annual stratiform cloud (and fog) from surface observations, 1951–81. (From
Klein and Hartmann, 1993)
Table 3.15 Regional occurrence of subtropical marine stratus
Coastal extent Longitude Season and Value (°C) of Season and Value (°C) of
amount (%) of max. stabilitya amount (%) of max. stability
max. stratus and month min. stratus and month
California 120°–130°W JJA 67 June 22 DJF 45 February 18

20°–30°N
Canary Islands 25°–35°W JJA 35 June 16 SON 17 October 13
15°–25°N
Peru 10°–20°S 80°–90°W SON 72 October 22 DJF 42 February 18
Namibia 10°–20°S 0°–10°E SON 75 September 22 MAM 48 February 17
Western Australia 95°–105°E DJF 45 February 18 JJA 41 June 15
25°–35°S°
Source: from Klein and Hartmann (1993).
Note
a
The difference between 700 mb and surface potential temperature.
The boundary is well mixed, with an almost uniform vertical moisture distribution. The
air temperature is only about 0.5–1.0°C less than that of the sea surface; however, the
convection is readily sustained by cloud-top radiative cooling (Albrecht et al., 1988, 1995).
Off northwest Africa, during the Atlantic Stratocumulus Transition Experiment (ASTEX)
in June 1992, the more broken cloud cover was found to be associated with a decoupled
boundary layer. This develops initially during the day as solar radiation is absorbed by
the clouds, and is assisted by occasional drizzle. Shallow cumulus first develops in the
subcloud layer (Figure 3.78), then, as the air flows over warmer waters, the cloud layer
becomes permanently decoupled from the surface mixed layer. It appears that transitional
processes are occurring beneath the stratiform cloud deck apparent on satellite imagery.
Downstream, in the tropical easterlies, the transition to trade wind cumulus is completed.
The average vertical structure of the lower troposphere in the trade wind zones features
several distinct layers (Figure 3.79). During fair weather there is a shallow surface layer
overlaid by an isentropic mixed layer, to about 700 m altitude, which is maintained by
mechanical mixing and dry convection. This in turn is capped by a cloud layer with trade
11
Figure 3.78 The transition from stratocumulus to trade wind cumulus. (Albrecht et al., 1995)
Figure 3.79 Structure of the lower tropical troposphere. (Sarachik, 1985)
wind cumulus fed by moisture from the surface and mixed layer and buoyant ascent of
0 almost 100 mb day1. Sarachik (1985) proposes a simple theory to explain this structure.
Evaporation and heat transfer from the ocean provide buoyancy for the mixed layer,
whereas buoyancy in the cloud layer is derived from condensation in the clouds. As clouds
form, the cloud layer becomes conditionally unstable, with a lapse rate somewhat greater
than the SALR to the base of the trade wind inversion (Figure 3.80). The inversion is main-
tained by the upward flux of moisture and the evaporation of the trade wind cumulus,
according to Betts (1973). The clouds moisten and cool the dry air descending across the
inversion. The growth of the mixed layer is balanced by subsidence between the cumuli in
the cloud layer and the mixed layer top approximates the lifting condensation level. Fair-
weather cumulus transports moisture upward and heat downward as a result of evaporation,
0 convection, and condensation. This destabilizes the lower layers and serves to deepen the
cloud layer (Betts, 1978). In spite of the heat transport by the cumuli, the re-evaporation
of cloud droplets means that the clouds are not a significant net source of heat for the lay-
11 ers between the cloud tops and the sea surface. As larger cumuli develop, the associated
(a)
(b)
Figure 3.80 (a) Atmospheric structure associated with the trade wind inversion for the ATEX
triangle, February 6–12 1969. (From Augstein et al., 1973; Augstein, 1978.) (b)
Vertical profiles of horizontal velocity divergence, specific humidity (q), air tempera-
ture, and total static energy (J g1) based on ATEX data for the central trades (solid
line), the transition between divergent and convergent flow (dashed), near the origin
of the trade wind trajectory (dash–dot) and for ITCZ conditions during GATE (dotted)
(based on measurements of Brümmer, 1978)
downdraughts serve to dilute the moist subcloud layer and create a more stable stratifica-
tion. Cloud layer growth is balanced by large-scale subsidence above the trade wind inver-
sion. The long-term mean structure of the tropical troposphere overlying the convective
cloud layer is driven by buoyancy and heating generated in deep cumulonimbus clouds.
Figure 3.81 shows soundings illustrating these changes. The stability of the inversion layer
11
Figure 3.81 (a) Undisturbed and (b) disturbed soundings in the trade wind boundary layer of
the central equatorial Pacific, 170°–180°W. Equivalent potential temperature (e) and
0 saturation equivalent potential temperature (es) are plotted. The lifting condensation
level (LCL) is indicated for a parcel originating at 975 mb. (From Firestone and
Albrecht, 1986)
is sufficient to suppress deep convection, except where vertical advection can destabilize
the boundary layer. Cloud downdraughts also modify the velocity field by adiabatic warm-
ing of low-level air (Betts, 1978; Firestone and Albrecht, 1986). Rising and descending air
currents in the cloud layer more or less balance. Mixing, by entrainment of warm dry
air and cloudy moist air, is limited to the top of the inversion layer.
These thermodynamic views of the trade wind inversion imply that it is maintained by
0 a balance between large-scale subsidence and moist convective and radiative processes.
This assumes that the boundary layer depth and structure are determined by local values
of SST and large-scale divergence, but Schubert et al. (1995) argue that horizontal advec-
tion plays a key role. The inversion extends into the tropics from the subtropics as a layer
of high potential vorticity. It is strongly coupled horizontally. Thus the inversion height
is actually controlled by horizontally averaged space values of SST, divergence, and the
atmospheric structure above the inversion.
Research into tropical convection during the Tropical Ocean Global Atmosphere
(TOGA) Coupled Ocean–Atmosphere Response Experiment (COARE) observation periods
in the western Pacific shows that three cloud types are widespread: trade wind cumulus,
0 cumulus congestus, and cumulonimbus. Similar findings have been reported earlier for
other tropical ocean areas. For each category, the cloud tops are near major stable layers
located at altitudes of around 2 km, 5 km (linked with melting below the freezing level)
and 15–16 km (the tropopause), respectively (Johnson et al., 1999). These stable layers
are characterized by stratiform cloud shelves. Moreover, cumulus congestus accounted for
57 percent of the precipitating clouds and 28 percent of the convective rainfall. The cloud
populations fluctuate significantly in association with thirty-to-sixty-day oscillations.
In the eastern tropical Atlantic during the GARP Atlantic Tropical Experiment (GATE)
in 1974 a distinctive convective and precipitation regime was observed (McBride and
Gray, 1980). Moisture flux convergence peaks around 07.00 local time, whereas the
0 maximum rainfall is in the early afternoon. Convection is suppressed by thermal stability
and, because of the vertical structure of the heating, up to eight hours is required for a
squall line to develop. Mesoscale systems are observed to be organized into convective
11 bands about 60 km apart (Frank, 1983). Over the course of a day, cumulonimbus anvils
tend to merge, forming cloud clusters, and these may have a radial dimension of 2°–12°
latitude.
Deep, precipitating convection is buoyancy-driven. Betts (1997) indicates that in this
convective mode about 20 percent of falling precipitation evaporates, forming unsaturated
downdrafts which cause local stabilization. This can help to split the cloud layer vertical
circulation and initiate the build-up of deep cumulonimbus. It is important to note that
the freezing level in the tropics is typically located in mid-troposphere about the 550 mb
level, so that the coalescence process plays the dominant role in precipitation formation.
Convective development is pronounced in the oceanic ITCZ zones. Over the western
tropical Pacific and in the western Atlantic, in contrast to the GATE area, low-level mass
convergence reaches a maximum about 07.00 local time and rainfall peaks at the same
time, demonstrating the rapid coupling of moisture convergence and rain formation
(McBride and Gray, 1980). Over the western equatorial Pacific warm pool, where SSTs
exceed 27°C, deep convection is frequent. However, there is considerable temporal vari-
ability, and, at times, large-scale controls inhibit convection and can give clear skies
(Zhang, 1993). In January the frequency of deep convection generally increases as SSTs
rise from 26°C to around 30°C, but this relationship is not apparent in April or, in general,
along the ITCZ relative to other areas. Indeed, Zhang considers that the relationship
frequently proposed between SSTs and convective activity holds better for their spatial
association than for the temporal variations at an individual location. It is noteworthy that
the seasonal latitudinal displacements of deep convection and low-level convergence in
the central Pacific are in phase with the solar radiation maximum, whereas the develop-
ment of SSTs above 28°C lags the radiation by one to two months (Fu et al., 1994).
Deep convection depends on the convective available potential energy (CAPE), which is
the potential energy available for convection minus that needed to initiate it,6 but also on
the vertical buoyancy profile, particularly boundary layer instability. Thus deep convec-
tion can be forced by either surface wind convergence or high SSTs.
The role of cumulus convection in large-scale tropical circulations presents complex
problems. Convective clouds redistribute heat, moisture, and mass from the subcloud layer
into the upper troposphere. Early studies of the equatorial trough by Riehl and Malkus
(1958) and Riehl and Simpson (1979) proposed the upward transport of energy by undi-
luted cumulus towers penetrating into the upper troposphere. The latent heat release was
presumed to be sufficient to offset radiative cooling and to generate potential energy by
raising high enthalpy air from the subcloud layer to the upper troposphere. Such parcel
ascent requires external forcing by the trade wind flow, generating Ekman pumping. Xu
and Emmanuel (1989) point out that, in much of the tropics, the level of free convection
(LFC) is generally low, so that cumulus convection is widespread. The large amounts of
CAPE required for undilute cloud ascent cannot readily develop. In contrast, over the US
Midwest in spring, capping inversions allow CAPE to accumulate without immediate
release. Arakawa and Schubert (1974) postulate that cumulus clouds are actually dilute
mixtures of mixed-layer and ambient air. Several studies now indicate that the “mixed
layer” is a convectively well adjusted layer and is well mixed only with respect to virtual
potential temperature (v) which measures the buoyancy of unsaturated air (Betts, 1982;
Xu and Emmanuel, 1989).7 The equivalent potential temperature (e) is observed to
decrease with height in the mixed layer, by about 5°C between 1,000 mb and 950 mb,
so the buoyancy of cloud parcels (formed reversibly) depends on where they originate in
the subcloud layer.8 Parcels lifted from the top of the subcloud layer are closer to neutral
stability than those originating nearer the surface, according to Xu and Emmanuel (1989).
They show that the magnitude of the buoyancy over the central and western equatorial
Pacific during July–September 1965–80 depends primarily on the mean surface e. The
buoyancy tends to increase as the height of the LFC decreases (from 875 mb to 975 mb),
and these data show a local maximum buoyancy near 600 mb of about 2 K. Negative
buoyancies of up to 0.5 K are found below LFCs at 925 mb or higher. This analysis
11 excluded conditionally unstable conditions, which accounted for 15 percent, 25 percent,
and 33 percent of all cases at Koror, Truk, and Majoro, respectively. Xu and Emmanuel
find that the troposphere of the area is near neutral to non-precipitating clouds originating
in parcels displaced reversibly from the top of the subcloud layer during 67–85 percent
of the time, confirming earlier observations (Betts, 1982). Larger buoyancies can occur
by lifting air from near the surface. Xu and Emmanuel conclude that moisture conver-
gence is unable to generate kinetic energy in a conditionally neutral atmosphere. Thus
the growth of large-scale disturbances occurs by redistribution of existing kinetic energy
via baroclinic instability or by air–sea feedback mechanisms modifying the subcloud layer.
0 Synoptic cloud patterns in tropical disturbances over the western Pacific, based on data
from the International Satellite Cloud Climatology Project (ISCCP) for 1983–90, are
provided by Lau and Crane (1995). They indicate cloud structures like those observed in
squall lines with low cloud-top altitudes in a zone of subsidence ahead of the leading
edge of the convection, thick cloud extending to 150 mb in a 1° latitude-wide convec-
tive area and thinner high clouds in the trailing edge and outflow region.
Wind profiler measurements made during twelve months of the Tropical Ocean Global
Atmosphere (TOGA) Coupled Ocean Atmosphere Research Experiment (COARE) at 2°S,
147°E provide a valuable climatology of rainfall types over the warm pool (Webster and
Lukas, 1992; Williams et al., 1994). Stratiform, convective, and mixed stratiform–convec-
0 tive types each account for about 30 percent of the total rainfall (160 cm falling in 313
hours), although the rate of stratiform cloud precipitation is half or less than that associ-
ated with the other two types. The remainder of the total is low-level “warm” rainfall.
The ratio of convective to stratiform precipitation appears to influence the thermally direct
zonal circulation cells in the tropics according to Raymond (1994). Calculations of ideal-
ized tropical convection forced by anomalous sea surface temperature gradients (Lindzen
and Nigam, 1987) show that differential heating sets up boundary layer convergence,
giving low-level ascent, and this is further strengthened by upper-level heating gradients
and ascending motions that result from the distribution of stratiform precipitation.
The annual cycle of convection, as determined from harmonic analysis of OLR and
0 HRC data, provides a dynamically based definition of tropical climate regimes (Wang,
1994). Wang identifies a “maritime monsoon” regime, associated with convergence zones
in the western North Pacific, South Pacific and southwest Indian Ocean (Figure 3.82),
that is characterized by a unimodal precipitation distribution with a summer maximum.
In these sectors, surface westerly winds are found equatorward of the monsoon trough.
In the convergence zones located between the trade wind belts over the central North
Atlantic and North Pacific there is a more or less persistent rainy season with a bimodal
(transition seasons) pattern of variation. The Asian–Indonesian–Australian and West
African summer monsoon areas have bimodal rainfall patterns. As noted by Wang, the
0 Figure 3.82 Convection/precipitation regimes in the tropics. Regimes indicated are A arid, SA
semi-arid, PC permanently convective, M summer monsoon and ITCZ trade wind
convergence zone. Permanently convective (arid) regions are hatched (shaded), respec-
11 tively. (Wang, 1994)
peak rainy season shows an apparent eastward progression, almost along the equator, from
60°E in the Indian Ocean to 130°E south of New Guinea from July through February of
the following year, reflecting the monsoon transition from boreal to austral summer (Figure
3.83). In this sector there is also an India–Indonesia–northern Australia displacement in
convection, noted by Meehl (1987), and an analogous feature between Central and South
America, identified in OLR data by Pulwarty et al. (1992). The propagation of convec-
tion eastward during July–October in the tropical North Pacific, shown in Figure 3.83,
may be associated with the seasonal change in the western Pacific monsoon trough and
the autumn peak of tropical cyclone occurrence.
The view that convection drives large-scale circulations, such as the Hadley cell, through
latent heat release in deep cumulonimbus has been challenged by Emmanuel et al. (1994).
In fact, latent heat release is largely compensated for by radiative and adiabatic cooling.
They argue that the equivalent potential temperature (e) of the subcloud layer is deter-
mined through the combined effects of ocean surface temperatures (assumed by Lindzen
and Nigam, 1987, to play the dominant role), low-level winds, large-scale vertical motion
in the convecting layer, turbulent entrainment at the top of the subcloud layer, and down-
draughts of air caused by the evaporation of precipitation. These downdraughts lower e
in the subcloud layer, and this in turn cools the free air. Interactions between disturbances
and convection primarily result in “moist convective damping” which selectively damps
high frequency oscillations. The paradigm advanced by Emmanuel et al. (1994) asserts
that the vertical temperature gradient is controlled not by heating but by convection which
is a response to instability. Their central tenet is that convective systems, in the tropics
at least, are close to a statistical equilibrium with their environment. Nevertheless, depar-
tures from this equilibrium lasting a few hours can play an important role in the dynamics.
For example, where an equatorial Kelvin wave (see section 6.5) moves westward through
an area of convection, enhanced downdraughts associated with the ascending phase of
the wave system lower subcloud e and in turn the temperature of the free air. A short
time lag in the convective response transfers this effect westward behind the zone of
maximum ascent, so that the convection overlaps into the cold phase of the wave. This
creates a negative correlation of heating and temperature leading to wave damping.
Concurrently, however, the baroclinic circulation of the disturbance leads to strengthened
low-level easterlies east of the ascending air. These winds increase the surface heat fluxes
via a wind-induced surface heat exchange (WISHE) process and thereby set up warming,
which is partly in phase with the temperature perturbation of the wave, thus generating
an amplifying effect. Emmanuel et al. also conclude that the magnitude of observed CAPE
values in the tropics undergoes only small variations that are not readily detected (except
Figure 3.83 Phase diagram for the annual mean minimum OLR. Thick solid lines indicate discon-
tinuities of at least two months in the isochrones. Paths 1A, 1B, 2A, 2B, and 2C show
phase propagations. Very dry regions are shown by the stippled areas; areas where
the variance accounted for by the first harmonic is less than 0.2 are shown hatched.
(Wang, 1994)
11 where air parcels are raised through different depths of the atmosphere). Conditional insta-
bility can arise in the upper troposphere, however, through the freezing of cloud
condensate, which alters the lapse rate (Williams and Renno, 1993).
3.9.3 The extratropics

The cloud and precipitation characteristics of middle and high latitudes over the oceans
are strongly shaped by the occurrence of extratropical cyclones as well as mesoscale
cyclones. These relationships are discussed in Chapter 6. Over land areas, in summer,
0 mesoscale convective systems and isolated thunderstorms play the major role and there
are strong diurnal cycles of precipitation. The distribution of mesoscale convective systems
is shown in Figure 6.31. In winter, cyclone systems interact strongly with topography.
It is inappropriate to treat here the wide variety and complexity of regional precipitation
regimes. Details may be found in sources such as Trewartha (1981) or Lockwood (1974),
and the regional volumes of the World Survey of Climatology (Landsberg, 1970–84).
3.10 Air masses

The mean thermal conditions of the atmosphere discussed in section 3.2 obscure important
0 features of the atmospheric structure and significant synoptic variations. One framework
for characterizing this structure is the air mass concept first formulated by Bergeron (1928),
followed by Schinze (1932) and Willett (1933), and subsequently applied in many regional
investigations.
An ideal air mass is a barotropic fluid in which isobaric and isosteric (constant specific
volume) surfaces do not intersect. This implies that the density (or temperature) field is
a unique function of pressure and that the geostrophic wind velocity remains constant
with height. Air mass boundaries are hyper-baroclinic (frontal) zones where isobaric and
isothermal surfaces intersect. The idealized criterion is generally too restrictive, and James
(1969) proposed that an air mass could be considered as spatially homogeneous where
0 standard deviation values for air temperature are less than 1.2°C.
The atmospheric structure is conventionally presented as a function of height (pres-
sure) and latitude in terms of isotherms and isotachs of zonal wind speed (see Figure
4.6). Figure 3.84 shows a schematic cross-section for the northern hemisphere, depicting
quasi-barotropic air masses separated by a baroclinic frontal zone. The frontal zones have
associated arctic and polar front jets, whereas the subtropical jetstream is related to an
upper tropospheric baroclinic zone.
An alternative presentation of atmospheric structure using isentropes was first proposed
by Sir Napier Shaw (1930). A schematic illustration of this type is given in Figure 3.85
(Hoskins, 1991). Shaw distinguished an Overworld where the isentropic surfaces are every-
0 where above the tropopause; a Middleworld, where they cross the tropopause, and an
Underworld, where they intersect the surface. The tropopause intersections imply a direct
route between the troposphere and stratosphere for air parcels moving isentropically.
Hoskins points a number of interesting relationships between PV structure and tropos-
pheric circulation. For a lower boundary with high (low) values, and a tropopause with
low (high) values, there is an associated cyclonic (anticyclonic) circulation and decreased
(increased) static stability in the troposphere. If the anomalies at the tropopause and
lower boundary are not in opposition, the associated circulations tend to cancel one
another. In addition, the mass-weighted PV integrated between two isentropic surfaces 1
and 2 is equivalent to the circulation on the surface (C ) around the mass of air:
0
11
冕1
2
(PV) dm = 冕
1
2
C d
Figure 3.84 Air masses in relation to the structure of the troposphere and the tropopause. J =
jetstream, A = arctic, P = polar, S = subtropical. (After Defant and Taba, 1954)
where dm is the mass element:

1
p

This C circulation is unchanged by heating processes or friction within the region, or by

external forces (Hoskins, 1991).
A climatology of winter season PV for the 315 K surface in the northern hemisphere
(Brunet et al., 1995) shows a clear relationship with the mean stream function distribu-
tion of troughs and ridges, although the polar vortex is bounded by high PV gradients
near 50°N, approximately along the 2 PV units contour. The strongest gradients are over
eastern Asia and eastern North America in association with the mean jetstreams.
On a global scale, the three segments of the tropopause can serve to distinguish between
air with tropical, subtropical, and temperate characteristics (Figure 3.84; Defant and Taba,
1957). This tripartite division is well marked in the middle and upper troposphere but
near the surface, especially in middle and high northern latitudes, there is considerable
spatial and temporal variability as a result of the strong local contrasts in surface condi-
tions. Moreover, analyses of temperature frequency distributions in the free air, over
Budapest for example (Ozorai, 1963; see Barry and Perry, 1973, p. 184), show incon-
sistencies in the number of frequency maxima that are present at 500 m, 1,500 m, 3,000
m, and 5,000 m altitude in both January and July, implying that homogeneity in the
vertical structure of air masses is uncommon.
There are two procedures that can be used in air mass classification. One would iden-
tify conservative atmospheric properties chosen to designate the net effects of the source
area and subsequent dynamic and thermodynamic modications. For surface data, dew
point temperature is a conservative tracer of air movement for dry adiabatic and diabatic
processes, while wet-bulb potential temperature is quasi-conservative for dry and satu-
rated adiabatic processes in the free air. Nevertheless, such an approach requires air
trajectories to be determined regularly. The second procedure, which has been widely
11
Figure 3.85 A schematic vertical cross-section of the isentropic structure of the atmosphere.
Isentropes are shown at 30 K intervals from 270 K to 390 K (thin lines). The tropopause
is shown by the solid line. The arrows indicate transports. The irregular equatorial
boundary is defined by the zero contour of potential vorticity. (From Hoskins, 1991)
followed, takes account of the lapse rate structure despite its spatio-temporal variability.
Primary air mass types are considered to form over extensive areas that have broadly
0 uniform surface conditions, high surface pressures and small pressure gradients with
weakly divergent airflows. The primary division separates arctic (A), polar (P) and trop-
ical (T) air, with a secondary division according to the maritime (m) or continental (c)
nature of the source area. The term “polar” is firmly established in the literature but is
geographically inappropriate; the Russian usage, “temperate,” is a more logical designation
(see Figure 3.84). The six major air mass types are:
Continental Maritime
Arctic cA mA
Polar cP mP
0 Tropical cT mT
The tropical air masses originate in the subtropical anticyclone cells, while cA and cP air
masses originate in polar anticyclone cells and the latter are modified into mA and mP
air masses, respectively, as changes ensue in the vertical structure during passage over
ocean surfaces. The arctic types have been recognized particularly over North America
in winter by Canadian meteorologists in the three front (four air mass) model (Godson
1950; Penner, 1955; Galloway, 1958; McIntyre, 1958).
For climatological purposes the identification of air masses can be based on several dif-
ferent criteria. One involves the decomposition of a frequency distribution of daily values
0 of an appropriate air mass parameter (maximum daily temperature, dew point, etc.) into
several component partial frequency distributions or collectives. A numerical method was
developed by Essenwanger (1954), although Bryson (1966) shows that sufficiently accu-
11 rate results can also be obtained graphically (see Figure 2.6). Alternatively, streamlines of
mean resultant winds can be used to identify source regions according to the areas of
streamline divergence. On this basis the major source regions in each hemisphere and their
annual duration have been determined by Wendland and Bryson (1981) and Wendland and
McDonald (1986) (Figure 3.86). The oceanic subtropical anticyclone sources are dominant
in both hemispheres. In the northern hemisphere they cover at least 25 percent of the surface
and for six months of the year they affect nearly 60 percent of the hemisphere. There is
even greater dominance by the equivalent oceanic anticyclonic source areas in the south-
ern hemisphere, whereas the northern land areas have several regional centers of varying
importance.
The regions of persistent polar anticyclones (Siberia, northwestern Canada, and the
western Arctic Basin) have been traditionally regarded as sources of continental polar
(cP) or arctic (cA) air mass formation (Wexler, 1937, 1951). Using a simple radiative
cooling model, Wexler estimated that cooling of a maritime polar air mass to 3 km would
require about twenty-six days. A revised calculation that incorporated turbulent fluxes,
subsidence, condensation, and ice crystal precipitation suggested that most cP properties
are acquired in four days, but fifteen days are needed for the complete development (Curry,
1983). The question as to whether this interval of time is typically available in arctic air
can be examined in terms of the net mass flux across 70°N. Using Oort’s (1982) data,
Fultz (1986) finds that residence times are only four or five days for the lower tropos-
phere and three days in the vertical mean, in striking contrast to the calculated requirements
for full development. Such low residence times are confirmed by studies of pollution
events at Barrow. Raatz and Shaw (1984) note that pollutants travel from Eurasian source
areas across the Arctic in seven to ten days during quasi-stationary anticyclonic situa-
tions. For the regions of Siberia and Canada (50°–70°N) dominated by polar anticyclones,
Oort’s data for the 1,000–700 mb layer suggest winter residence times of five days and
three days, respectively (Fultz, 1986). The discrepancy lacks a clear explanation. The
omission of processes from Curry’s (1983) model that could significantly speed up the
air mass transformation seems unlikely. The mean residence times and model calcula-
tions could be reconciled if the air mass formation were concentrated in a limited part of
the source area, with stronger exchanges over the remainder. Also, there may be biases
in the observed wind data used to calculate residence times. However, Fultz concludes
that the rapid exchange times imply that air masses must be shallow, low-level structures;
air masses or frontal zones extending to the upper troposphere “must primarily be formed
by dynamics and not by the sort of regional processes originally envisioned.”
There are many studies of air mass modification processes. Transformation of cA or
cP air by diabatic heating and moisture transfer as it flows offshore over relatively warm
ocean surfaces has received much attention, as have the more gradual effects on mT air
flowing poleward over cooler surfaces. Nevertheless, there are other equally important
processes at work in the free air. For example, air moving towards lower latitudes and
conserving its absolute vorticity needs to increase its relative vorticity to offset the decrease
in the Coriolis parameter. If the actual curvature is anticyclonic, the air column shrinks,
causing adiabatic heating. However, if the trajectory is cyclonic a polar outbreak can
maintain its depth into low latitudes (see Figure 3.87).
In view of the importance of atmospheric optical properties for radiative transfer
(Smirnov et al., 1994), renewed attention is being paid to air mass characteristics as a
means of stratifying observed and modeled data. The marine atmosphere over the North
Atlantic, for example, has been characterized in terms of the Canadian three front, four
air mass model by Low and Hudak (1997). Using 850 mb and 700 mb wet-bulb poten-
tial temperature steps of 4°C and a two-day persistence criterion, the frequency of air
mass categories was determined for fourteen subregions of the North Atlantic using a
method analogous to the partial frequency analysis described above. Five main categories
(mT, mP, mA, cA, and a transitional type with shallow arctic air, cmA) are identified, as
well as seven combinations involving up to three adjacent air masses. Two subtypes of
0
0
0
0
0
11
11
(a) (b)
Figure 3.86 Air mass source regions in (a) the northern hemisphere and (b) the southern hemisphere. The number of months per year each
air mass affects the regions delimited is shown. (After Wendland and Bryson, 1981; Wendland and McDonald, 1986)
Figure 3.87 The effect of the air trajectory on the depth of polar air outbreaks in lower latitudes.
(From Barry and Perry, 1973)
mT air for “fair” and “disturbed” conditions are also recognized using discriminant analysis
of 700 mb wind speeds and column water vapor content.
One of the most striking illustrations of the broad significance of air mass character-
istics is the demonstration by Bryson (1966) that the summer limit of arctic air dominance
over North America coincides closely with the southern boundary of tundra vegetation.
In North America the boreal forest zone is bounded by the winter and summer isolines
of 50 percent arctic air occurrence. In Eurasia, Krebs and Barry (1970) also found that
the arctic frontal zone in summer coincides with the boreal forest–tundra ecotone. Pielke
and Vidale (1995) suggest that the air mass contrasts are themselves driven by differ-
ences in energy, moisture and momentum exchanges over the contrasting vegetation
surfaces. However, further investigation of this question appears to be needed.
Appendix 3.1 Potential vorticity

In order to understand and interpret this important dynamic quantity properly, it is neces-
sary to trace the history of its definition and use. Thorpe and Volkert (1997) provide a
useful account which is summarized here.
1. The term potential vorticity was introduced by Rossby (1940) apparently as a concept
analogous to that of potential temperature. Rossby states that potential vorticity “repre-
sents the vorticity the air column would have if it were brought isentropically to a standard
latitude and stretched or shrunk vertically to a standard depth or weight.”
Rossby’s potential vorticity,
p0
PR ( f ) f0
p
where the relative vorticity about an axis perpendicular to an isentropic surface,

f the Coriolis parameter, f0 f for a chosen reference latitude, p the pressure depth
of an air column in which the top and bottom have a potential temperature difference
0, and p0 the reference p selected for the same column.
PR has dimensions of s1 and is conserved for dry adiabatic flow in the absence of
friction; thus it can be used as an air mass tracer. An isentropic relative vorticity can also
be defined by:
p
(P f0) f
p0 R
11 2. An independent hydrodynamical approach to the vorticity of adiabatic motion was
proposed by Ertel (1942) and applied to the study of cyclone dynamics by Kleinschmidt
(1950, 1951; Thorpe, 1994). The Ertel adiabatic vorticity is defined as:
1
PE a #

where a is the absolute vorticity vector. Making the hydrostatic approximation:

0 PE ⯝ –g ( f )

p
PE has dimensions of K m2 s1 kg1.
Comparing the above equation with that of Rossby’s, it can be seen that:
[p0]
PR = PE
[g0]
The term in brackets above can be evaluated using numerical values chosen by Rossby
(1940); for f0 104 s1, p0 100 mb, and 0 4 K, the expression becomes 250
0 kg m2 K1. Using the potential vorticity unit (1 PV unit 106 m2 s1 K kg1), if PE
1 PV unit, then PR 1.5 104 s1. Thorpe and Volkert (1997) point out that, confus-
ingly, the term “potential vorticity” is now commonly applied to Ertel’s expression,
although its dimensions are not in vorticity units.
Notes
1 The radiation flux density (J m2 s1) emitted by a black body is proportional to T 4 (Kelvin).
The emission (F) integrated over the wavelength range for a hemispheric surface is expressed
by the Stefan–Boltzmann law F = oT 4, where (the Stefan–Boltzmann constant) = 5.67
0 108 W m2 K4. For a black body, the wavelength of maximum emission is shown by Wien’s
law to be proportional to the absolute temperature.
max
= T 1
k
where k = 2898 m K.
2 The vertically and zonally averaged momentum transports (m2 s2) are converted to angular
momentum transport units (1018 kg m2 s2) by multiplying the values by 2r2 cos2 (Po /g)
2.56 cos2 where 2r = length of the appropriate latitude circle, r cos = distance to the
axis of rotation, Po /g = mass per unit area (~104 kg m2) (Oort and Peixoto, 1983). To convert
units of ºC m s1 to energy units (PW), multiply the values by 0.4 cos (= 2r cos cp (Po/g)).
0 3 Divergent and rotational wind. The horizontal wind field can be separated into a non-divergent
rotational component and a divergent irrotational component. The non-divergent rotational part,
(Vr ), where the horizontal wind vector VH is parallel to constant streamlines (), can be
expressed:
Vr k
where k a unit vertical vector, a gradient operator, a horizontal stream function
(units m2 s1), is defined as:

u v=

y
x
0
and the relative vorticity as:
2
11
The divergent irrotational part Vd can be expressed in terms of a velocity potential
Vd
where is defined as

u v=

x
y
and the horizontal divergence #VH 2.
4 The (Stokes) mass stream function for the mean meridional circulation is determined from
(Peixoto and Oort, 1992, p. 158):

[v] = g
2 r cos
p
and

[w] = g
2 r 2 cos
p
5 An earlier study of precipitation zones on a pole-to-pole continent, 45º longitude across, and
one extending only equator-to-pole, on either side of the equator, was made by Robinson (1972)
as reported in Lamb (1977, p. 301).
6 Convective available potential energy (CAPE) refers to the buoyancy force in moist air inte-
grated between the level of free convection (LFC) and the equilibrium level (EL). On a
thermodynamic diagram such as the tephigram it is indicated by the area between the environ-
mental temperature conditions and the path curve (a saturated adiabat) connecting the LFC and
the EL (Bluestein, 1993, p. 444). Values of CAPE range from around 103 J kg1 (m2 s2) for
moderate convection to maxima of about 5 103 J kg1. The maximum vertical velocity at
the equilibrium level (Wmax) (assuming there is no entrainment of cooler, drier ambient air) is:
Wmax = (2.CAPE)1/2. Thus Wmax is 50 m s1 for a CAPE value of 1,250 J kg1. For an air parcel
with temperature and humidity mixing ratio corresponding to the mean values of the lowest
500 m, the work needed to lift the parcel to the LFC is termed the convective inhibition.
7 The virtual temperature expressed by:
Tv Tp
q 冥
冤1 1 q/0.622
where q humidity mixing ratio. The difference in virtual temperature of cold air and ambient
air can be used to define buoyancy (Xu and Emmanuel, 1989). For a pseudoadiabatic process,
a parcel is lifted adiabatically without sustaining its liquid water content. Here the virtual temper-
ature:
Tv (A) Tp 冤1 1 q(T
q (T ) 冥
)/0.622
p
s p
where Tp temperature of the parcel lifted adiabatically, and qs saturation mixing ratio.
A parcel lifted adiabatically in a reversible moist adiabatic process retains all of the condensed
water and this increases the effective density of cloud air. Consequently the Tv(R) of cloud air
is lower for a reversible moist adiabatic process:
Tv (R) Tp 冤1 q1 (T Q)/0.622冥

s p
where QT total water content of the parcel. The differences found in the upper troposphere
over the equatorial Pacific between Tv(A) and Tv(R) were 2°–4°C (Xu and Emmanuel, 1989).
8 The equivalent potential temperature is defined as the potential temperature of an air parcel if
all of its water vapor were condensed and the latent heat converted into sensible heat (Houze,
1993, pp. 28–9). This implies that the air is lifted dry-adiabatically to its lifting condensation
level, then moist-adiabatically to a high altitude (with precipitation of condensed water as it
forms), and finally is brought by dry-adiabatic descent to 1,000 mb.
To a first approximation, for saturated air:
11 e ≈ e Lq (T, p)/cpT
For practical purposes:

e ≈ (1 Lq/cpT)
where q humidity mixing ratio (kg of water per kg of air), and L latent heat of vapor-
ization. In the case of saturated air, q is the corresponding saturation value (qs).
A detailed evaluation of errors in the calculations due to various computational approxima-
tions is given by Bolton (1980).
0 References
Aber, J.D. 1992. Terrestrial ecosystems. In: K.E. Trenberth, ed., Climate System Modeling,
Cambridge University Press, Cambridge, pp. 173–200.
Adams D.K.T. and Comrie A.C. 1997. The North American monsoon. Bull. Amer. Met. Soc., 78
(10): 2197–213.
Adler, R.F. 1975. A comparison of the general circulations of the northern and southern hemi-
spheres based on satellite multi-channel radiance data. Mon. Wea. Rev., 103: 52–60.
Albrecht, B., Bretherton, C.S., Johnson, D., Schubert, W.H., and Frisch, A.S. 1995. The Atlantic
stratocumulus transition experiment – ASTEX. Bull. Amer. Met. Soc., 76 (6): 889–904.
Albrecht, B., Randall, D.A., and Nicholls, S. 1988. Observations of marine stratocumulus clouds
0 using FIRE. Bull. Amer. Met. Soc., 69 (6): 618–26.
Allison, I., Wendler, G., and Radok, U. 1993. Climatology of the east Antarctic ice sheet (100°E
to 140°E) derived from automatic weather stations. J. Geophys. Res., (D5): 8815–23.
Ambaum, M. 1997. Isentropic formation of the tropopause. J. Atmos. Sci., 54 (4): 555–68.
Anderson, D.L.T. 1976. The low-level jet as a western boundary current. Mon. Wea. Rev., 104 (7):
907–21
Angell, J.K. and Korshover, J. 1982. Comparison of year-average latitude, longitude and pressure of
the four centers of action with air and sea temperature, 1899–1978. Mon Wea. Rev., 110: 300–3.
Arakawa, A. and Schubert, W.H. 1974. Interaction of a cumulus cloud ensemble with the large-
scale environment. I. J. Atmos. Sci., 31: 674–701.
Arking, A. 1996. Absorption of solar energy in the atmosphere: discrepancy between model and
0 observations. Science, 273: 779–82.
Asnani, G.C. 1968. The equatorial cell in the general circulation. J. Atmos. Sci., 25: 133–4.
Asnani, G.C. 1993. Tropical Meteorology. Indian Institute of Tropical Meteorology, Pune, India.
Vols. 1 and 2, 1,202 pp.
Astapenko, P.D. 1964. Atmospheric Processes in the High Latitudes of the Southern hemisphere,
Israel Program for Scientific Translations, Jerusalem, 286 pp. (Akad. Nauk, SSR, Moscow, 1960).
Augstein, E. 1978. The atmospheric boundary layer over the tropical oceans. In: D.B. Shaw, ed.,
Meteorology over the Tropical Oceans, Roy. Met. Soc., London, pp. 73–105.
Augstein, E., Riehl, H., Ostapoff, E., and Wagner, V. 1973. Mass and energy transport in an undis-
turbed trade wind flow. Mon. Wea. Rev., 101: 101–11.
Bannon, P.R. 1979. On the dynamics of the East African jet. I. Simulation of mean conditions for
0 July: II. Jet transients. J. Atmos. Sci., 36 (11): 2139–52; 2153–68.
Barlow, M., Nigam, S., and Berberry, E.H. 1998. Evolution of the North American monsoon system.
J. Climate, 11 (9): 2238–57.
Barnett, T.P. 1977. The principal time and space scales of the Pacific trade wind fields. J. Atmos.
Sci., 34: 221–36.
Barnett, T.P., Dumenil, L., Schlese, U., Roeckner, E., and Latif, M. 1989. The effect of Eurasian
snow cover on regional and global climate variations. J. Atmos. Sci., 46: 661–85.
Barron, E.J. and Washington, W.M. 1984. The role of geographic variables in explaining paleo-
climates: results from Cretaceous climate model sensitivity studies. J. Geophys. Res., 89: 1267–79.
Barron, E.J., Hay, W.W., and Thompson, S. 1989. The hydrologic cycle: a major variable during
Earth history. Palaeogeogr., Palaeoclim., Palaeoecol. (Global Planet. Change), 75: 157–74.
0 Barry, R.G. 1975. Climate models in palaeoclimatic reconstruction. Palaeogeogr., Palaeoclim.,
Palaeoecol., 17: 123–37.
Barry, R.G. 1978. H. B. de Saussure: the first mountain meteorologist. Bull. Amer. Met. Soc., 59:
11 702–5.
Barry, R.G. 1997. Palaeoclimatology, climate system processes and the geomorphic record. In: D.R.
Stoddart, ed., Process and Form in Geomorphology, Routledge, London, pp. 175–214.
Barry, R.G. and Chorley, R.J. 1998. Atmosphere, Weather and Climate. 7th edition. Routledge,
London, 409 pp.
London, 555 pp.
Bates, J.R. 1973. A generalization of the CISK theory. J. Atmos. Sci., 30 (8): 1509–19.
Baumgartner, A. and Reichel, E. 1975. The World Water Balance. Elsevier, Amsterdam, 179 pp.
Berberry, E.H., Nogués-Paegle, J., and Horel, J.D. 1992. Wavelike southern hemisphere extratrop-
ical teleconnections. J. Atmos. Sci., 49 (2): 155–77.
Berger, A.L. 1978. Long-term variations of daily insolation and Quaternary climate changes. J.
Atmos. Sci., 35 (12): 2362–7.
Berger, A.L. 1979. Spectrum of climatic variations and their causal mechanisms. Geophys. Surveys,
3: 351–402.
Berger, A.L., Imbrie, J., Hays, K.H., Kukla, G.J., and Saltzman, B. (eds) 1984. Milankovitch and
Climate. D. Reidel, Dordrecht, 895 pp.
Bergeron, T. 1928. Über die dreidimensionale verknupfende Wetteranalyse. I. Prinzipelle.
Einführung in das Problem der Luftmassen und Frontenbildung. Geofys. Publ. (Oslo), 5 (6):
1–111.
Betts, A.K. 1973. Non-precipitating cumulus convection and its parameterisation. Quart. J. Roy.
Met. Soc., 99: 178–96.
Betts, A.K. 1978. Convection in the tropics. In: D.B. Shaw, ed., Meteorology over the Tropical
Oceans, Roy. Met. Soc., London, pp. 105–32.
Betts, A.K. 1982. Saturation point analysis of moist convective overturning. J. Atmos. Sci., 39:
1484–505.
Betts, A.K. 1997. Atmospheric convection: Some basic concepts (Abstract). Annal. Geophysicae,
15 (Suppl. II), p. C447.
Blackmon, H.L. 1986. Building, testing and using a general circulation model. In: J. Willebrand
and D.L.T. Anderson, eds, Large-scale Transport Processes in Oceans and Atmosphere, D. Reidel,
Dordrecht, pp. 1–70.
Blake, D.W., Krishnamurti, T.N., Low-Nam, S.V., and Fein, J.S. 1983. Heat low over the Saudi
Arabian desert during May 1979 (Summer MONEX). Mon. Wea. Rev., 111 (9): 1759–75.
Blanford, H.F. 1884. On the connexion of the Himalayan snowfall with dry winds and seasons of
droughts in India. Proc. Roy. Soc., London, 37: 3–22.
Bluestein, H.B. 1993. Synoptic-Dynamic Meteorology in Middle Latitudes, II. Observations and
Theory of Weather Systems. Oxford University Press, New York, 594 pp.
Bolton, D. F. 1980. The computation of equivalent potential temperature. Mon. Wea. Rev., 108:
1046–53.
Boville, B.A. 1991. Sensitivity of simulated climate to model resolution. J. Climate, 4 (5): 469–85.
Bromwich, D.H. 1988. Snowfall in high southern latitudes. Rev. Geophys., 26: 149–68.
Brubaker, K.L., Entekabi, D., and Eagleson, P.S. 1993. Estimation of continental-scale precipita-
tion recycling. J. Climate, 6: 1077–89.
Brümmer, B., Augstein, E., and Riehl, H. 1974. On the low level wind structure in the Atlantic
trade. Quart. J. Roy. Met. Soc., 100: 109–21.
Brunet, G., Vautard, R., Legras, B., and Edouard, S. 1995. Potential vorticity on isentropic surfaces:
climatology and diagnostics. Mon. Wea. Rev., 123 (4): 1037–58.
Bryson, R.A. 1966. Air masses, streamlines and the boreal forest. Geogr. Bull. 8: 228–69.
Buchan, A. 1869. The mean pressure of the atmosphere and the prevailing winds over the globe,
for the month and year. Trans. Roy. Soc. Edinburgh, 25: 575–637.
Budyko, M.I. and Drozdov, O.A. 1953. Zakonomersti vlagoborota v atmosfere (Principles of mois-
ture exchange in the atmosphere). Izvestiya Akad. Nauk SSR, Ser. Geogr. (1953), No. 4: 5–14.
Budyko, M.I. and Izrael, Y.A. 1991. Paleoclimatic evidence of climatic sensitivity to changing
atmospheric composition. In: M.I. Budyko and Y.A. Izrael, eds, Anthropogenic Climate Change,
University of Arizona Press, Tucson AZ, pp. 279–318 (tables 8.3, 8.4).
Burde, G.I., Zangvil, A., and Lamb, P.J. 1996. Estimating the role of local evaporation in precip-
itation for a two-dimensional region. J. Climate 9 (6): 1328–38.
Carissimo, B.C., Oort, A.H., and Vonder Haar, T.H. 1985. Estimating the meridional energy trans-
port in the atmosphere and ocean. J. Phys. Oceanogr., 15: 82–91.
11 Carleton, A.M. 1981. Monthly variability of satellite-derived cyclonic activity for the southern hemi-
sphere winter. J. Climatol., 1: 21–38.
Carleton, A.M. 1992. Synoptic interactions between Antarctica and lower latitudes. Austral. Met.
Mag., 40: 129–47.
Carleton. A.M. and Carpenter, D.A. 1990. Satellite climatology of “polar lows” and broadscale cli-
matic associations for the southern hemisphere. Intl. J. Climatol., 10: 219–46.
Carleton, A.M. and Song, Y. 1997. Synoptic climatology and intrahemispheric associations of cold
air mesocyclones in the Australian sector. J. Geophys. Res., 102 (D12): 13873–87.
Carleton, A.M. and Whalley, D. 1988. Eddy transport of sensible heat and the life history of
synoptic systems: a statistical analysis for the southern hemisphere winter. Met. Atmos. Phys.,
0 38: 140–52.
Cess, R.D. et al. 1991. Interpretation of snow–climate feedback as produced by 17 general circula-
tion models. Science, 273: 888–92.
Cess, R.D. et al. 1995. Absorption of solar radiation by clouds: observation versus models. Science,
267: 496–9.
Cess, R.D. et al. 1996. Cloud feedback in atmospheric general circulation models: an update. J.
Geophys. Res., 101 (D8): 12791–4.
Cess, R.D., Potter, G.L. et al. 1990. Intercomparison and interpretation of climate feedback processes
in 19 atmospheric general circulation models. J. Geophys. Res., 95 (D10): 16601–15.
Chamberlin, J.W. and Hunter, D.M. 1987. Theory of Planetary Atmospheres: An Introduction to their
Physics and Chemistry. 2nd edition. Academic Press, New York, 455 pp.
0 Chang, C.-P. and Lau, K.M. 1980. Northeasterly cold surges and near-equatorial disturbances over
the winter MONEX area during December 1974. II. Planetary-scale aspects. Mon. Wea. Rev., 108:
298–312.
Charney, J.G. 1971. Geostrophic turbulence. J. Atmos. Sci., 28 (6): 1087–95.
Charney, J.G. 1973. Movable CISK. J. Atmos. Sci., 30 (1): 50–2.
Charney, J.G. 1975. Dynamics of deserts and drought in the Sahel. Quart. J. Roy. Met. Soc., 101:
193–202.
Charney, J.G. and Eliassen, A. 1964. On the growth of the hurricane depression. J. Atmos. Sci., 21
(1): 68–75.
Chase, T.N., Pielke, R.A., Kittel, T.G.F., Nemani, R., and Running, S.W. 1996. Sensitivity of gen-
eral circulation model to global changes in leaf area index. J. Geophys. Res., 101: 7393–408.
0 Chen, T.C., Chen, J.-M., Schubert, S., and Takacs, L.L. 1997. Seasonal variation of global surface
pressure and water vapor. Tellus, 49A: 61321.
Chen, T.C., Yen, M.-C., Pfaendtner, J., and Sud, Y.C. 1996. Annual variation of the global precipt-
able water and its maintenance: observation and climate simulation. Tellus, 49A: 1–16.
Christy, J.R., Trenberth, K.E., and Anderson, J.R. 1989. Large-scale redistribution of atmospheric
mass. J. Climate, 2: 137–48.
Coffin, J.H. 1853. On the winds of the northern hemisphere. Smithsonian Contrib. to Knowledge 6,
Washington DC, Article VI.
Coffin, J.H. 1876. The winds of the globe or the laws of the atmosphere’s circulation over the sur-
face of the earth. Smithsonian Contrib. to Knowledge, 268, Washington DC, 756 pp.
Collins, W.D. 1998. A global signature of enhanced shortwave absorption by clouds. J. Geophys.
0 Res., 103 (D24): 31669–79.
Compo, G.P., Kiladis, G.N., and Webster, P.J. 1999. The horizontal and vertical structure of East
Asian winter monsoon pressure surges. Quart. J. Roy. Met. Soc., 125: 29–54.
Cook, K. 1994. Mechanisms by which surface drying perturbs tropical precipitation fields. J. Climate,
7: 400–13.
Corby, G.A. (ed.). 1970. The Global Circulation of the Atmosphere. Roy. Met. Soc., London, pp.
42–90.
Crane, R.G. and Barry, R.G. 1988. Comparison of the MSL synoptic pressure patterns of the Arctic
as observed and simulated by the GISS general circulation model. Met. Atmos. Phys., 39: 169–83.
Crowe, P.R. 1949. The trade wind circulation of the world. Trans. Inst. Brit. Geogr., 15: 39–56.
Crowe, P.R. 1950. The seasonal variation in the strength of the trades. Trans. Inst. Brit. Geogr., 16:
0 25–47.
Crowe, P.R. 1965. The geographer and the atmosphere. Trans. Inst. Brit. Geogr., 36: 1–19.
Crowe, P.R. 1971. Concepts in Climatology. Longman, London, 589 pp.
11 Crowley, T.J. and North, G.R. 1991. Paleoclimatology. Oxford University Press, New York, 339 pp.
Crowley, T.J., Mengel, J.G., and Short, D.A. 1987. Gondwanaland’s seasonal cycle. Nature, 329:
803–7.
Cubasch, U. and Cess, R.D. 1990. Processes and modelling. In: J.T. Houghton, G.J. Jenkins and J.J.
Ephraums, eds, Climate Change. The IPCC Scientific Assessment, Cambridge University Press,
Cullen, M.J.P. 1993. The unified forecast/climate model. Met. Mag., 122: 81–94.
Curry, J. 1983. On the formation of continental polar air. J. Atmos. Sci. 40: 2278–92.
Davidson, N.E., McBride, J.L., and McAvaney, B.J. 1983. The onset of the Australian monsoon
during winter MONEX: synoptic aspects. Mon. Wea. Rev., 111: 496–516.
Davis, C.A. 1996. Potential vorticity. In: S.H. Schneider, ed., Encyclopedia of Weather and Climate,
Oxford University Press, New York, pp. 602–8.
Davis, R.E., Hayden, B.P., Gay, D.A., and Phillips, W.L. 1997. The North Atlantic subtropical anti-
cyclone. J. Climate, 10 (4): 728–44.
Defant, F. and Taba, H. 1957. The threefold structure of the atmosphere and characteristics of the
tropopause. Tellus, 9: 259–74.
Dhar, O.N. and Bhattacharya, B.K. 1973. Contribution of tropical disturbances to the water resources
of the Ganga basin. Vayu Mandal, 3: 76–9.
Dickinson, R.E., 1985. Climate sensitivity. Adv. Geophys., 28A: 99–129.
Dickinson, R.E. and Henderson-Sellers, 1988. Modeling tropical deforestation: a study of GCM
land surface parameterizations. Quart. J. Roy. Met. Soc., 114: 439–62.
Ding, Y.-H. 1994. Monsoons over China. Kluwer, Dordrecht, 432 pp.
Ding, Y.-H. and Krishnamurti, T.N. 1987. Heat budget of the Siberian high and the winter monsoon.
Mon. Wea. Rev., 115 (10): 2428–49.
Ding, Y., Wen, S., and Li, Y. 1991. A study of the dynamic structure of the Siberian high in winter.
Acta Met. Sinica, 49: 428–38.
Dirmeyer, P.A. 1998. Land–sea geometry and its effect on monsoon circulations. J. Geophys. Res.,
103 (D10): 11555–72.
Dirmeyer, P.A. and Shukla, J. 1994. Albedo as a modulator of climate response to tropical defor-
estation. J. Geophys. Res., 99 (D10): 20863–77.
Dirmeyer, P.A. and Shukla, J. 1996. The effect on regional and global climate of expansion of the
world’s deserts. Quart. J. Roy. Met. Soc., 12: 451–82.
Dodd, J.P. and James, I.N. 1997. The impact of latent-heat release on the Hadley circulation. Quart.
J. Roy. Met. Soc., 123: 1763–70.
Dong, B. and Valdes, P.J. 1998. Modelling the Asian summer monsoon rainfall and winter/spring
snow mass. Quart. J. Roy. Met. Soc., 124: 2567–96.
Dorman, C.E. and Bourke, R.H. 1979. Precipitation over the Pacific Ocean, 30ºS to 60ºN. Mon.
Wea. Rev., 107: 896–910.
Douglas, M.W. 1992. Structure and dynamics of two monsoon depressions. I. Observed structure.
II. Vorticity and heat budgets. Mon. Wea. Rev., 120 (8): 1524–47; 1548–64.
Douglas, M.W., Maddock, R.A., Howard, U., and Reyes, S. 1993. The Mexican monsoon. J. Climate,
6 (8): 1665–77.
Drozdov, O.A. and Grigor’eva, A.S. 1965. The Hydrological Cycle in the Atmosphere (Vlagoboorot
v Atmosfere, Gidrometeoizdat, Leningrad, 1963), Israel Program for Scientific Translations,
Jerusalem, pp. 35–40.
Drözdowsky, W. 1996. Variability of the Australian summer monsoon at Darwin, 1957–92. J.
Climate, 9 (1): 85–96.
Dzerdzeevskii, B.L. 1945. Tsirkulatsionnye skhemy v troposfere Tsentralnoi Arktiki (Circulation
schemes for the central Arctic troposphere), Izdatel’stvo Akademii Nauk, 28 pp. (transl. in
Scientific Report No. 3, Contract AF19 (122)-228. Meteorology Department, University of
California, Los Angeles).
Ebert, E.E. and Curry, J.A. 1993. An intermediate one-dimensional thermodynamic sea ice model
for investigating ice–atmosphere interactions. J. Geophys. Res., 98 (C6): 10085–109.
Elliott, W.P. and Reed, R.K. 1984. A climatological estimate of precipitation for the world ocean.
J. App. Met., 23: 434–9.
Eltahir, E.A.B. and Bras, R.L. 1996. Precipitation recycling. Rev. Geophys., 34: 367–78.
Emmanuel, K.A., Neelin, J.D., and Bretherton, C.S. 1994. On large-scale circulations in convecting
atmospheres. Quart. J. Roy Met. Soc., 120: 1111–43.
Enomoto, H. and Ohmura, A. 1990. The influences of atmospheric half-yearly cycle on the sea ice
extent in the Antarctic. J. Geophys. Res., 95 9497–511.
11 Ertel, H. 1942. Ein neuer hydrodynamischer Wirbelsatz. Met. Zeit., 59: 277–81.
Essenwanger, O. 1954. Neue Methoden der Zerlegung von Haufigkeitsverteilungen. Ber. dtsch.
Wetterdienst (US Zone), Bad Kissingen, No. 10, 11 pp.
Estoque, M.A. 1975. Structure of the mid-oceanic intertropical convergence zone. J. Met. Soc. Japan,
53: 317–22.
Evans, J.L. and Allan, R.J. 1992. El Niño/Southern Oscillation modification to the structure of the
monsoon and tropical cyclone activity in the Australasian region. Intl. J. Climatol., 12 (6): 611–23.
Fein, J.S. and Stephens, P.L. (eds). 1987. Monsoons, Wiley, New York, 632 pp.
Fernandez-Partargas, J. and Estoque, M.A. 1985. Characteristics of the ITCZ over the eastern Pacific,
5–8 June 1979. Mon. Wea. Rev., 113 (1): 99–105.
0 Ferreira, R.N. and Schubert, W.H. 1997. Barotropic aspects of ITCZ breakdown. J. Atmos. Sci., 54
(2): 261–85.
Ferrel, W. 1856. An essay on the winds and currents of the ocean. Nashville J. Med. Surgery, 9:
287–301; 375–89.
Ferrel, W. 1859–60. The motions of fluids and solids relative to the earth’s surface. Math. Monthly
(Cambridge MA), 1: 140–8, 210–16, 300–7, 366–73, 397–406, 2: 89–97, 339–46, 374–90.
Findlater, J. 1966. Cross-equatorial jetstreams at low level over Kenya. Met. Mag., 95: 353–64.
Findlater, J. 1969a. A major low-level air current near the Indian Ocean during the northern summer.
Findlater, J. 1969b. Interhemispheric transport of air in the lower troposphere over the western Indian
Ocean. Quart. J. Roy. Met. Soc., 95: 400–3.
0
Findlater, J. 1971. Mean monthly airflow at low levels over the western Indian Ocean. Geophys.
Mem., No. 115, HMSO, London, 53 pp.
Findlater, J. 1972. Aerial exploration of the low-level cross-equatorial current over eastern Africa.
Findlater, J. 1977. Observational aspects of the low-level cross-equatorial jetstream of the western
Indian Ocean. Pure Appl. Geophys., 115: 1251–62.
Firestone, J.K. and Albrecht, B.A. 1986. The structure of the atmospheric boundary layer in the cen-
tral equatorial Pacific during January and February of FGGE. Mon. Wea. Rev., 114 (11): 2219–31.
Fleming, E.L, Lim, G.-H., and Wallace, J.M. 1987. Differences between the spring and autumn cir-
culation of the northern hemisphere. J. Atmos. Sci., 44 (9): 1266–86.
0 Fleming, J.R. 1990. Meteorology in America, 1800–1870. Johns Hopkins University Press, Baltimore
MD, 264 pp.
Flohn, H. 1960. Monsoon winds and the general circulation. In: Monsoons of the World, India
Meteorology Department, Delhi, pp. 65–74.
Flohn, H. 1968. Contributions to a Meteorology of the Tibetan Highlands. Atmos. Sci. Pap. 120,
Colorado State University, Fort Collins CO., 120 pp.
Flohn, H. 1971. Tropical Circulation Patterns. Bonn. Geogr. Abhandl. 15, 55 pp.
Flohn, H. and Korff, H.C. 1969. Zusammenhang zwischen Temperaturgefälle Äquator–Pol und den
planetarischen Luftdruckgürteln. Ann. Met., n.s. 4: 163–4.
Flohn, H., Kapala, A., Knocke, H.R., and Mächel, H. 1992. Water vapor as an amplifier of the green-
house effect: new aspects. Met. Zeitschr, n.s. 1: 122–38.
0 Frank, W.M. 1983. The structure and energetics of the east Atlantic intertropical convergence zone.
J. Atmos. Sci., 40 (8): 1916–29.
Fu, R., Del Genio, A.D., and Rossow, W.R. 1994. Influence of ocean surface conditions on atmos-
pheric vertical structure of deep convection. J. Climate, 7 (7): 1092–108.
Fultz, D. 1986. Residence times and other time-scales associated with Norwegian air mass ideas. In:
J.O. Roads, ed., Namias Symposium, Scripps Institute of Oceanography, Ref. Serv. 86–17, pp. 82–102.
Gadd, A.J. 1981. Numerical modelling of the atmosphere. In: B.W. Atkinson, ed., Dynamical
Meteorology: An Introductory Selection, Methuen, London and the Royal Meteorological Society,
pp. 194–204.
Galloway, J.L. 1958. The three-front model: its philosophy, nature, construction and use. Weather
13: 395–403.
0 Garcia, O. 1985. Atlas of Highly Reflective Clouds for the Global Tropics, 1971–1983. NOAA-ERL,
US Department of Commerce, 365 pp.
Garratt, J.R. 1993. Sensitivity of climate simulations to land–surface and atmospheric boundary layer
11 treatments – a review. J. Climate, 6: 419–49.
Gates, W.L. 1992. AMIP: the Atmospheric Model Intercomparison Project. Bull. Amer. Met. Soc.,
73 (12): 1962–70.
Gates, W.L., Mitchell, J.F.B., Boer, G.J., Cubasch, U., and Meleshko, V.P. 1992. Climate model-
ling, climate prediction and model validation. In: J.T. Houghton, B.A. Callender, and S.K. Varney,
eds, Climate Change 1992: Supplementary Report to the IPCC Scientific Assessment. Cambridge
Gates, W.L., Rowntree, P.R., and Zeng, Q.-C. 1990. Validation of climate models. In: J.T. Houghton,
G.J. Jenkins, and J.J. Ephraums, eds, Climate Change: The IPCC Scientific Assessment, Cambridge
Gentilli, J. 1971. Dynamics of the Australian troposphere. In: J. Gentilli, ed., World Survey of
Climatology, Elsevier, Amsterdam, pp. 53–118.
Gibson, J.K., Kallberg, S., Hernandez, A., Uppala, S., Nomura, A., and Serrano, E. 1997. ECMWF
Re-analysis Project Series 1. ERA Description. ECMWF, Reading, UK, 72 pp.
Gill, A.E. 1980. Some simple solutions for heat-induced tropical circulation. Quart. J. Roy. Met.
Soc., 106: 447–62.
Giorgi, F. and Mearns, L.O. 1991. Approaches to the simulation of regional climate change: a
review. Rev. Geophys., 29: 191–216.
Giorgi, F., Marinucci, M.R., and Visconti, G. 1990. Application of a limited area model nested in
a general circulation model to regional climate simulation over Europe. J. Geophys. Res., 95
(D11): 18413–31.
Godbole, R.V. 1977. The composite structure of the monsoon depression. Tellus, 29: 25–40.
Godson, W.L. 1950. The structure of North American weather systems. Centen. Proc. Roy. Met.
Soc., London, pp. 89–106.
Gommel, W.R. 1963. Mean distribution of 500 mb topography and sea-level pressure in middle
and high latitudes of the northern hemisphere during the 1950–9 decade, January and July. J.
Appl. Met., 2: 105–13.
Gordon, A.H. and Taylor, R.C. 1970. Numerical Steady-state Friction Layer Trajectories over the
Oceanic Tropics related to Weather. International Indian Ocean Expedition Meteorological
Monograph 7, East–West Center, Honolulu HI.
Grotjahn, R. 1993. Global Atmospheric Circulations: Observations and Theory. Oxford University
Press, New York, 430 pp.
Gruber, A. and Krueger, A.F. 1984. The status of the NOAA outgoing long-wave radiation data
set. Bull. Amer. Met. Soc., 65 (9): 958–62.
Hack, J.J. 1992. Climate system simulation: basic numerical and computational concepts. In:
K.E. Trenberth, ed., Climate System Modeling, Cambridge University Press, Cambridge, pp.
283–318.
Hack, J.J., Shubert, W.H., Stevens, D.E., and Kuo, H.C. 1989. Response of the Hadley circulation
to convective forcing in ITCZ. J. Atmos. Sci., 46 (19): 957–73.
Hadley, G. 1735. Concerning the cause of the general trade winds. Phil. Trans. Roy. Soc., London,
29: 58–62.
Halley, E. 1686. An historical account of the trade-winds and the monsoons, observable in the seas
between and near the tropics with an attempt to assign the physical cause of the said winds. Phil.
Trans. Roy. Soc., London, 16: 153–68.
Haltiner, G.J. and Martin, F.L. 1957. Dynamical and Physical Meteorology. McGraw-Hill, New
York, 470 pp.
Haltiner, G.J. and Williams, R.T. 1980. Numerical Prediction and Dynamic Meteorology. 2nd
edition. J. Wiley, New York, 477 pp.
Hamilton, M.G. 1979. The South Asian Summer Monsoon. E. Arnold, Port Melbourne, Australia,
72 pp.
Hamilton, R.A. 1958. The meteorology of north Greenland during the midwinter period. Quart. J.
Roy. Met. Soc., 84: 355–74.
Hansen, J., Lacis, A.A., Rind, D., Russell, G., Stone, P., Fung, I., Ruedy, R., and Lerner, J. 1984.
Climate sensitivity analysis of feedback mechanisms. In: J.E. Hansen and T. Takahaski, eds,
Climate Processes and Climate Sensitivity, Geophys. Monogr. 29, Amer. Geophys. Union,
Washington DC, pp. 130–63.
Hansen, J.E., Russell, G., Rind, D., Stone, P., Lacis, A.A., Lebedoff, S., Ruedy, R., and Travis, L.
1983. Efficient three-dimensional global models for climate studies. Models I and II. Mon. Wea.
Rev., 111 (4): 609–62.
11 Hare, F.K. 1960. The westerlies. Geog. Rev., 50: 345–67.
Hare, F.K. 1962. The stratosphere. Geog. Rev., 52: 525–47.
Harrison, E.P., Minnis, P., Barkstrom, B.R., Ramanathan, V., Cess, R.D., and Gibson, G.G. 1990.
Seasonal variation of cloud radiative forcing derived from the Earth Radiation Budget Experiment.
J. Geophys. Res., 95 (D11): 18687–703.
Hart, J.E., Rao, G.V., van de Boogard, H., Young, J.A., and Findlater, J. 1978. Aerial observations
of the East African low-level jetstream. Mon. Wea. Rev., 106 (12): 1714–24.
Hartmann, D.L. and Doelling, D. 1991. On the net radiative effectiveness of clouds. J. Geophys. Res.,
96 (D1): 869–92.
Hastenrath, S. 1982. On the meridional heat transports in the world ocean. J. Phys. Oceanogr., 12:
0 922–7.
Hastenrath, S. 1991. Climate Dynamics of the Tropics. Kluwer, Dordrecht, 488 pp.
Hay, L.E., McCabe, G.J., Jr, Wolock, D.M., and Ayers, M.A. 1992. Use of weather types to disag-
gregate general circulation model predictions. J. Geophys. Res., 97 (D3): 2781–90.
Hay, W.W., Barron, E.J., and Thompson, S.L. 1990a. Results of global atmospheric circulation exper-
iments on an Earth with a meridional pole-to-pole continent. J. Geol. Soc., London, 147: 385–92.
Hay, W.W., Barron, E.J., and Thompson, S.L. 1990b. Global atmospheric circulation experiments on
an Earth with polar and tropical continents. J. Geol. Soc., London, 147: 749–57.
Held, I.M. and Hou, A.Y. 1980. Non-linear axisymmetric circulations in a nearly inviscid atmos-
phere. J. Atmos. Sci., 37 (3): 515–33.
Helmholtz, H. von. 1888. Über atmosphärische Bewegungen. Met. Zeit., 5: 329–40.
0 Henderson-Sellers, A. and Hughes, N.A. 1982. Albedo and its importance in climate theory. Progr.
Phys. Geog., 6: 1–44.
Henderson-Sellers, A. and Wilson, M.F. 1983. Surface albedo data for climatic modeling. Rev.
Geophys. Space Phys., 21: 1743–78.
Henderson-Sellers, A., Yang, Z.-L. and Dickinson, R.E. 1993. The Project for Intercomparison of
Land–Surface Paramerization Schemes. Bull. Amer. Met. Soc., 74 (7): 1335–49.
Hendon, H. and Liebmann, B. 1990. A composite study of onset of the Australian summer.
J. Atmos. Sci, 47: 2227–40.
Hess, P.G., Battisti, D.S., and Rasch, P.J. 1993. Maintenance of the intertropical convergence zones
and the large-scale tropical circulation on a water-covered Earth. J. Atmos. Sci., 50 (5): 691–713.
Hewitson, B.C. and Crane, R.G. 1996. Climate downscaling: techniques and applications. Climate
0 Res., 7: 85–95.
Highwood, E.J. and Hopkins, B.J. 1998. The tropical tropopause. Quart. J. Roy. Met. Soc., 124:
1579–604.
Hobbs, W.H. 1926. The Global Anticyclones: The Poles of the Atmospheric Circulation. Basingstoke,
Macmillan, 198 pp.
Hobbs, W.H. 1945. The Greenland glacial anticyclone. J. Met., 2: 143–53.
Hoerling, M.P., Schaak, T.K., and Lenzen, A.J. 1991. Global objective tropopause analysis. Mon.
Wea. Rev., 119 (8): 1816–31.
Hoinka, K.P. 1997. The tropopause: discovery, definition and demarcation. Met. Zeit., 6: 281–303.
Hoinka, K.P. 1998. Statistics of the global tropopause pressure. Mon. Wea. Rev., 1265 (12): 3303–25.
Holland, G.J. 1986. Interannual variability of the Australian summer monsoon at Darwin, 1952–82.
0 Mon. Wea. Rev., 114 (3): 594–604.
Holland, G.J. 1995. Scale interaction in the western Pacific monsoon. Met. Atmos. Phys. 56: 57–79.
Holland, G.J., McBride, J.L., Smith, R., Jasper, D., and Keenan, T.D. 1986. The BMRC Australian
monsoon experiment: AMEX. Bull. Amer. Met. Soc., 67 (12): 1466–72.
Holton, J.R., Wallace, J.M., and Young, J.A. 1971. On boundary layer dynamics and the ITCZ.
J. Atmos. Sci., 28 (2): 275–80.
Holzer, M. 1996. Optimal spectral topography and its effect on model climate. J. Climate, 9 (10):
244–63.
Horrell, M.A. 1990. Energy balance constraints on 18O based paleo-sea surface temperature estimates.
Paleoceanog., 5: 339–48.
Hoskins, B.J. 1991. Towards a PV- view of the general circulation. Tellus, 43AB: 27–35.
0 Hoskins, B.J. 1996. On the existence and strength of the summer subtropical anticyclones. Bull. Amer.
Met. Soc., 77 (6): 1287–92.
Hou, A.Y. 1993. The influence of tropical heating displacements on the extratropical climate.
11 J. Atmos. Sci., 50 (21): 3553–70.
Hou, A.Y. and Lindzen, R.S. 1992. Intensification of the Hadley circulation due to concentrated
heating. J. Atmos. Sci., 49 (14): 1233–41.
Houghton, H.G. 1985. Physical Meteorology. MIT Press, Cambridge MA, 442 pp.
Howarth, D.A. 1983. An analysis of the variability of cyclones around Antarctica and their relationship
to sea ice extent. Ann. Assoc. Amer. Geogr., 73: 519–37.
Hughes, N.A. 1984. Global cloud climatologies: a historical review. J. Clim. Appl. Met., 23:
724–51.
Hulme, M., Briffa, K.R., Jones, P.D., and Senior, C.A. 1993. Validation of GCM control simulations
using indices of daily airflow types over the British Isles. Climate Dynam., 9: 95–105.
Hummell, J.R. and Reck, R.A. 1979. A global surface albedo model. J. Appl. Met., 18: 239–53.
Hunt, B.G. 1979. The influence of the earth’s rotation rate on the general circulation of the atmosphere.
J. Atmos. Sci., 36 (8): 1392–408.
Hurrell, J.W. and van Loon, H. 1994. A modulation of the atmospheric annual cycle in the southern
hemisphere. Tellus, 46A: 325–38.
Hurrell, J.W., van Loon, H., and Shea, D.J. 1998. The mean state of the troposphere. In: D.J. Karoly
and D.G. Vincent, eds, Meteorology of the southern hemisphere, Met. Monogr. 27 (49), Amer. Met.
Soc., Boston MA, pp. 1–46.
Hutter, K., Blatter, H., and Ohmura, A. 1990. Climatic changes, ice sheet dynamics and sea level.
Zürcher Geogr. Schriften, Geographisches Inst. ETH, Zürich, 37, 82 pp.
Imbrie, J., Hays, J.D., Martinson, D.G., McIntyre, A., Mix, A.C., Morley, J.J., Pisias, N.G., Prell, W.L.,
and Shackleton, N.J. 1984. The orbital theory of Pleistocene climate: support from a revised chronol-
ogy of the marine 18O record. In: A.L. Berger, J. Imbrie, K.H. Hays, G.J. Kukla and
B. Saltzman, eds, Milankovitch and Climate, D. Reidel, Dordrecht, pp. 269–305.
Inamdar, A.K. and Ramanathan, V. 1998. Tropical and global-scale interactions among water vapor,
atmospheric greenhouse effect and surface temperature. J. Geophys. Res., 103 (D24): 32, 177–94.
Ingram, W.P., Wilson, C.A., and Mitchell, J.F.B. 1989. Modeling climate change: an assessment of sea
ice and surface albedo feedbacks. J. Geophys. Res., 94 (D6): 8609–22.
Jacka, T.H. 1983. Computer Database for Antarctic Sea Ice Extent. Austral. Nat. Res. Exped.
(ANARE), Res. Notes 13, 54 pp.
James, I.N. 1988. On the forcing of planetary-scale Rossby waves by Antarctica. Quart. J. Roy. Met.
Soc., 114: 619–37.
James, I.N. 1994. Introduction to Circulating Atmospheres. Cambridge University Press, Cambridge,
422 pp.
James, R.W. 1969. Elementary air mass analysis. Met. Rdsch., 22: 75–9.
Johnson, D.H. 1962. Rain in East Africa. Quart. J. Roy. Met. Soc., 88: 1–19.
Johnson, D.H. 1965. African synoptic meteorology. In: Meteorology and the Desert Locus. WMO
Tech. Note 69, T.P.85, Geneva, pp. 48–90.
Johnson, D.H. and Mörth, H.T. 1960. Forecasting research in East Africa. In: D.J. Bargman, ed.,
Tropical Meteorology in Africa, Munitalp Foundation, Nairobi, pp. 56–137.
Johnson, R.H., Rickenbach, T.M., Rutledge, S.A., Ciesielski, P.E., and Schubert, W.H. 1999. Trimodal
characteristics of tropical convection. J. Climate 12 (8): 2397–418.
Jones, D.A. and Simmons, I. 1993. Time and space spectral analysis of southern hemisphere sea level
pressure variability. Mon. Wea. Rev., 121: 661–72.
Jones, D.A. and Simmonds, I. 1994. A climatology of southern hemisphere anticyclones. Climate
Dynam., 10: 131–45.
Jones, P.D. 1987. The twentieth-century Arctic high – fact or fiction? Climat. Dynam., 1: 63–75.
Kalnay, E., et al. 1996. The NCEP/NCAR 40-year reanalysis project. Bull. Amer. Met. Soc., 77 (3):
437–71.
Kalnay, E., Mo, K.C., and Paegle, J. 1986. Large-amplitude short-scale stationary Rossby waves in the
southern hemisphere: observations and mechanistic experiments to determine their origin. J. Atmos.
Sci., 43 (3): 252–75.
Kang, I.-S., Ho, C.-H., Lim, Y.-W., and Lau, K.-M. 1999. Principal modes of climatological seasonal
and intraseasonal variations of the Asian summer monsoon. Mon. Wea. Rev., 127 (3): 322–40.
Karoly, D.J. 1990. The role of transient eddies in low-frequency zonal variations of the southern hemi-
sphere circulation. Tellus, 42A: 41–50.
Karoly, D.J., Vincent, D.G., and Schrage, J.M. 1998. General circulation. In: D.J. Karoly and
D.G. Vincent, eds, Meteorology of the Southern Hemisphere. Met. Monogr. 27 (49), pp. 47–85.
Kattenberg, A., Giorgi, F., Grassl, H., Meehl, G.A., Mitchell, J.F.B., Stouffer, R.J., Tokioka, T.,
11 Weaver, A.J., and Wigley, T.N.L. 1996. Climate models – projections of future climate. In:
J.T. Houghton, L.G. Meira Filho, B.A. Callander, N. Harris, A. Kattenberg and K. Maskell, eds,
Climate Change, 1995: The Science of Climate Change. Cambridge University Press, Cambridge,
pp. 285–357.
Kaurola, J. 1997. Some diagnostics of the northern wintertime climate simulated by the ECHAM
3 model. J. Climate 10 (2): 201–22.
Keshvamurty, R.N. 1972. On the vertical tilt of monsoon disturbances. J. Atmos. Sci., 29: 993–9.
Kiehl, J.T. 1992. Atmospheric general circulation modeling. In: K.E. Trenberth, ed., Climate System
Modeling, Cambridge University Press, Cambridge, pp. 319–69.
Kiehl, J.T. 1994. Clouds and their effects on the climate system. Physics Today, 47 (11): 36–41.
0 Kiehl, J.T. and Trenberth, K.E. 1997. Earth’s annual global mean energy budget. Bull. Amer. Met.
Soc., 78 (2): 197–208.
Kiladis, G.N. and Weickmann, K.M. 1992. Circulation anomalies associated with tropical convec-
tion during the northern winter. Mon. Wea. Rev., 120 (9): 1900–23.
Kiladis, G.N., Meehl, G.A., and Weickmann, K.M. 1994. Large-scale circulation anomalies asso-
ciated with westerly wind bursts and deep convection over the western equatorial Pacific.
J. Geophys. Res., 99 (D9): 18527–44.
Kiladis, G.N., von Storch, H., and van Loon, H. 1989. Origin of the South Pacific convergence
zone. J. Climate, 2: 1185–95.
Kirk, A. 1987. Southern hemisphere energy fluxes from analyses with special respect to the Weddell
Sea region (Antarctica). Met. Atmos. Phys., 37: 171–82.
0 Klein, S.A. and Hartmann, D.L. 1993. The seasonal cycle of low stratiform clouds. J. Climate, 6
(8): 1587–606.
Kleinschmidt, E. 1950. Über Aufbau und Enstehung von Zyklonen. I and II. Met Rundschau, 3:
1–16; 54–61.
Kleinschmidt, E. 1951. Über Aufbau und Enstehung von Zyklonen. III. Met Rundschau, 4: 89–96.
Kolmogorov, A.N. 1942. The equations of turbulent motion in an incompressible fluid (in Russian).
Izy. Akad. Nauk SSSR, Ser. Fizika, 6 (2): p. 56; (2), p. 58.
Köppen, W. 1931. Grundriss der Klimatkunde, de Gruyter, Berlin, 388 pp.
Kraichnan, R.H. 1967. Inertial range in two-dimensional turbulence. Phys. Fluids, 10: 1417–23.
Krebs, S.J. and Barry, R.G. 1970. The arctic front and the tundra–taiga boundary in Eurasia. Geogr.
Rev., 60: 548–54.
0 Kripalani, R.H. and Kularni, A. 1999. Climatology and variability of historical Soviet snow depth data:
some new perspectives in snow–Indian monsoon teleconnections. Climate Dynam., 15 (6): 475–89.
Krishnamurti, T.N. (ed.). 1977. Monsoon dynamics. Pure Appl. Geophys., 115: 1087–529.
Krishnamurti, T.N. 1979. Tropical Meteorology. WMO Publ. 364, Compendium of Meteorology,
Part 4. WMO, Geneva, 428 pp.
Krishnamurti, T.N., Aradanuy, P., Ramanathan, Y., and Pasch, R. 1981. On the onset vortex of the
summer monsoon. Mon. Wea. Rev., 109 (2): 344–63.
Krishnamurti, T.N., Kanamitsu, M., Godbole, R., Chang, C.B., Carr, F., and Chow, J. 1975. Study
of a monsoon depression. I. Synoptic structure. J. Met. Soc., Japan 53: 227–40.
Krishnamurti, T.N., Kanamitsu, M., Godbole, R., Chang, C.B., Carr, F., and Chow, J. 1976. Study
of a monsoon depression. II. Dynamic structure. J. Met. Soc. Japan, 54: 208–22.
0 Krishnamurti, T.N., Molinari, J., Pan, H.-L., and Wong, V. 1977. Downstream amplification and
formation of monsoon disturbances. Mon. Wea. Rev., 105 (10): 1281–97.
Kuhn, W.R., Walker, J.C.G., and Marshall, H.G. 1988. The effect on Earth’s surface temperature
from variations in rotation rate, continent formation, solar luminosity and carbon dioxide.
J. Geophys. Res., 94 (D8): 11129–36.
Kukla, G. and Robinson, D. 1980. Annual cycle of surface albedo. Mon. Wea. Rev., 108: 56–67.
Kuo, H.L. 1965. On the formation and intensification of tropical cyclones through latent heat release
by cumulus convection. J. Atmos. Sci., 22: 40–63.
Kuo, H.L. 1973. On the planetary boundary layer at the equator. J. Atmos. Sci., 30 (1): 153–4.
Kutzbach, G. 1987. Concepts of monsoon physics in historical perspective: the Indian monsoon
(seventeenth to early twentieth century). In: J.S. Fein and P.L. Stephens, eds, Monsoons, Wiley,
0 New York, pp. 269–330.
Kutzbach, J.E., 1994. Idealized Pangean climates: sensitivity to orbital change. In: G.D. Klein, ed.,
Pangea: Paleoclimate, Tectonics, and Sedimentation during Accretion, Zenith, and Breakup of a
11 Supercontinent, Special Paper 288, Geological Society of America, Boulder CO, 55 pp.
Kutzbach, J.E. and Gallimore, R.G. 1988. Sensitivity of a coupled atmosphere/mixed layer ocean
model to changes in orbital forcing at 9,000 years B.P. J. Geophys. Res., 93: 803–21.
Kutzbach, J.E. and Otto-Bliesner, B.L. 1982. The sensitivity of the African–Asian monsoonal climate
to orbital parameter changes for 9,000 years B.P. in a low-resolution general circulation model. J.
Atmos. Sci., 39: 1177–88.
Kutzbach, J.E., Prell, W.L., and Ruddiman, W.F. 1993. Sensitivity of Eurasian climate to surface
uplift of the Tibetan Plateau. J. Geol., 101: 177–90.
Kyle, H.L., et al. 1993. The Nimbus Earth Radiation Budget (ERB) experiment, 1975 to 1992. Bull.
Amer. Met. Soc., 74: 815–30.
Lamb, H.H. 1972. Climate Present, Past and Future. I. Fundamentals and Climate Now. Methuen,
London, 613 pp.
Lamb, H.H. 1977. Climate Present, Past and Future. II. Climate History and the Future. Methuen,
London, 560 pp.
Landsberg, H.E. (ed.-in-chief). 1970–84. World Survey of Climatology. Volumes V–XV, Elsevier,
Amsterdam.
Large, W.G. and van Loon, H. 1989. Large-scale low frequency variability of the 1979 FGGE sur-
face buoy drifts and winds over the southern hemisphere. J. Phys. Oceanogr., 19: 216–32.
Lau, K.M., Kim, J.H., and Sud, Y. 1996. Intercomparison of hydrologic processes in AMIP GCMs.
Bull. Amer. Met. Soc., 77 (10): 2209–27.
Lau, N.-C. and Crane, M.W. 1995. A satellite view of the synoptic-scale organization of cloud prop-
erties in midlatitude and tropical circulation systems. Mon. Wea. Rev., 123 (7): 1984–2006.
Lean, J. and Rowntree, P.R. 1993. A GCM simulation of the impact of Amazonian deforestation on
climate using an improved canopy representation. Quart. J. Roy. Met. Soc., 119: 509–30.
Legates, D.R. 1995. Global and terrestrial precipitation: a comparative assessment of existing clima-
tologies. Intl. J. Climatol., 15: 237–58.
Legates, D.R. and Wilmott, C.S. 1990. Mean seasonal and spatial variability in gauge-corrected,
global precipitation. Intl. J. Climatol., 10 (2): 111–18.
LeMarshall, J.F., Kelly, G.A.M., and Karoly, D.J. 1985. An atmospheric climatology of the southern
hemisphere based on ten years of daily numerical analyses (1972–1982). I. Overview. Austral. Met.
Mag. 33: 65–86.
Lewis, J.S. and Prinn, R.G. 1983. Planets and their Atmospheres. Academic Press, New York, 470
pp.
Li, A.-Q. and Leighton, H.G. 1993. Global climatologies of solar radiation budgets at the surface and
in the atmosphere from 5 years of ERBE data. J. Geophys. Res., 98 (D3): 4919–30.
Li, C. and Yanai, M., 1996. The onset and interannual variability of the Asian summer monsoon in
relation to land–sea thermal contrast. J. Climate, 9, 358–75.
Li, Z., Moreau, L., and Arking, A. 1997. On solar energy disposition. Bull. Amer. Met. Soc., 78:
53–70.
Liang, X.-Z. and Wang, W.-C. 1998. Association between China monsoon rainfall and tropospheric
jets. Quart. J. Roy. Met. Soc., 124: 2597–623.
Liljequist, G.H. 1970. Klimatologi. Generalstubens Litografiska Anstalt, Stockholm.
Lindberg, C. and Broccoli, A.J. 1996. Representation of topography in spectral climate models and
its effect on simulated precipitation. J. Climate, 9 (11): 2641–59.
Lindzen, R.S. 1974. Wave-CISK in the tropics. J. Atmos. Sci., 31 (1): 156–79.
Lindzen, R.S. 1990. Some coolness concerning global warming. Bull. Amer. Met. Soc., 71: 288–99.
Lindzen, R.S. 1994. Climate dynamics and global change. Annu. Rev. Fluid Mech., 26: 353–78.
Lindzen, R.S. and Hou, A.Y. 1988. Hadley circulations for zonally averaged heating centered off the
equator. J. Atmos. Sci., 45 (17): 2416–27.
Lindzen, R.S. and Nigam, S. 1987. On the role of sea-surface temperature gradients in forcing low
level winds and convergence in the tropics. J. Atmos. Sci., 44 (17): 2418–36.
Lockwood, J.G. 1974. World Climatology: An Environmental Approach. E. Arnold, London, 330 pp.
Lohmann, U., Sausen, J., Bengtsson, L., Cubasch, U., Perliwitz, J., and Roeckner, E. 1993. The
Köppen climate classification as a diagnostic tool for general circulation models. Climate Res., 3:
177–93.
London, J., Warren, S.G., and Hahn, C.J. 1989. The global distribution of observed cloudiness – a
contribution to the ISCCP. Adv. Space. Res., 9: 161–5.
Lorenz, E.N. 1955. Available potential energy and the maintenance of the general circulation. Tellus,
7: 271–81.
11 Lorenz, E.N. 1967. The Nature and Theory of the General Circulation of the Atmosphere. WMO
Publ. 218 (TP 115), WMO, Geneva, 161 pp.
Low, T.B. and Hudak, D.A. 1997. Development of air mass climatology analysis for the determi-
nation of characteristic marine atmospheres. I. North Atlantic. Theor. Appl. Climatol., 57: 133–53.
Ludlam, F.H. 1966. The cyclone problem: a history of models of the cyclonic storm. Inaugural
lecture, Imperial College of Science and Technology, London, pp. 19–49.
Ludlam, F.H. 1980. Clouds and Storm: The Behavior and Effect of Water in the Atmosphere.
Pennsylvania State University Press, University Park PA, 405 pp.
Machel, H., Kapala, A., and Flohn. H. 1998. Behavior of the centres of action above the Atlantic
since 1881. I. Characteristics of seasonal and interannual variability. Intl. J. Climatol., 18 (1):
0 23–36.
Mak, M. and Liu, Z.-G. 1993. Hadley circulation and climate variability. In: D.-Z. Ye et al., eds,
Climate Variability, China Meteorological Press, Beijing, pp. 201–8.
Manabe, S. and Broccoli, A.J. 1990. Mountains and arid climates of middle latitudes. Science, 247:
192–5.
Manabe, S. and Bryan, K. 1969. Climate calculations with a combined ocean–atmosphere model.
J. Atmos. Sci., 26 (4): 786–9.
Manabe, S. and Stouffer, R.J. 1996. Low-frequency variability of surface air temperature in a
1,000-year integration of a coupled atmosphere–ocean–land surface model. J. Climate 9 (2):
376–93.
Manabe, S. and Strickler, R.F. 1964. Thermal equilibrium of the atmosphere with a convective
0 adjustment. J. Atmos. Sci., 21: 361–85.
Manabe, S. and Terpstra, T.B. 1974. The effects of mountains on the general circulation of the
atmosphere as identified by numerical experiments. J. Atmos. Sci., 31: 3–42.
Mapes, B. and Houze, R.A., Jr. 1992. An integrated view of the 1987 Australian monsoon and its
mesoscale convective systems. I. Horizontal structure. Quart. J. Roy. Met. Soc., 118: 927–64.
Marshall, S., Roads, J.O., and Oglesby, R.J. 1997. Effects of resolution and physics on precipita-
tion in the NCAR Community Climate Model. J. Geophys. Res., 102 (D16): 19529–41.
Martynova, T.V. 1990. O kolebanniyakh polozheniya i intensivnosti tsentrov deistviya atmosfery
(Fluctuations in position and intensity of the atmospheric centers of action). Meteorologiya i
Gidrologiya, 1990 (4): 50–7.
Mathes, F.E. 1946. The glacial anticyclone theory examined in light of recent meteorological data
0 from Greenland. 1. Trans. Amer. Geophys. Union, 27: 324–41.
Matthes, F.E. and Belmont, F.E. 1950. The glacial anticyclone theory examined in light of recent
meteorological data from Greenland. 2. Trans. Amer. Geophys. Union, 31: 174–82.
Matsuno, T. 1966. Quasi-geostrophic motions in the equatorial area. J. Met. Soc. Japan, 44: 25–43.
Maxwell, J.B. 1980. The Climate of the Canadian Arctic Islands and Adjacent Waters. I. Climatol.
Studies 30, Atmospheric Environment Service, Environment Canada, 532 pp.
McBride, J.L. 1986. Tropical cyclones in the southern hemisphere summer monsoon. Proc. Second
International Conference on Southern Hemisphere Meteorology, Wellington, New Zealand,
pp. 358–64.
McBride, J.L. 1987. The Australian summer monsoon. In: C.P. Chang and T.N. Krishnamurti, eds,
Monsoon Meteorology, Oxford University Press, Oxford, pp. 203–31.
0 McBride, J.L. and Gray, W.M. 1980. Mass divergence in tropical weather systems. I. Diurnal vari-
ation. II. Large-scale controls on convection. Quart. J. Roy. Met. Soc., 106: 501–16, 517–38.
McBride, J.L. and Keenan, T.D. 1982. Climatology of tropical cyclone genesis in the Australian
region. J. Climatol., 2: 13–33.
McCabe, G.J. Jr and Legates, D.R. 1992. General circulation model simulations of winter and
summer sea-level pressures over North America. Intl. J. Climatol., 12: 815–27.
McGinnis, D.L. and Crane, R.G. 1994. A multivariate analysis of arctic climate in GCMs. J. Climate,
7 (8): 1240–50.
McGuffie, K. and Henderson-Sellers, A. 1997. Climate Modelling Primer. 2nd edition. Wiley, New
York, 253 pp. and CD-ROM.
McIntyre, D.P. 1958. The Canadian three-front, three-jetstream model. Geophysica (Helsinki), 6:
0 309–24.
McKendry, I.G., Steyn, D.G., and McBean, G. 1996. Validation of synoptic circulation patterns
simulated by the Canadian Climate Center General Circulation Model for western North America.
11 Atmos–Ocean, 33: 809–25.
McPherson, R.D. 1994. The National Center for Environmental Prediction: operational climate,
ocean and weather prediction for the 21st century. Bull. Amer. Met. Soc., 75 (3): 363–73.
Mechoso, C.R. 1980. The atmospheric circulation around Antarctica: linear stability and finite-
amplitude interactions with migrating cyclones. J. Atmos. Sci., 37: 2209–23.
Meehl, G.A. 1984. Modeling the earth’s climate. Climate Change, 6: 259–86.
Meehl, G.A. 1987. The annual cycle and interannual variability in the tropical Pacific and Indian
Ocean regions. Mon. Wea. Rev., 115 (1): 27–50.
Meehl, G.A. 1991. A re-examination of the mechanism of the semiannual oscillation in the southern
hemisphere. J. Climate 4: 911–26.
Meehl, G.A. 1992a. Global coupled models: atmosphere, ocean, sea ice. In: K.E. Trenberth, ed.,
Climate System Modeling, Cambridge University Press, Cambridge, pp. 555–81.
Meehl, G.A. 1992b. Effect of tropical topography on global climate. Ann. Rev. Earth Planet Sci.,
20: 85–112.
Meehl, G.A. 1994. Coupled land–ocean–atmosphere processes and south Asian monsoon variability.
Science, 266: 263–7.
Meehl, G.A., Boer, G.J., Covey, C., Latif, M., and Stouffer, R.J. 2000a. The Coupled Model
Intercomparison Project (CMIP). Bull. Amer. Met. Soc. 81 (2): 313–18.
Meehl, G.A., Collins, W., Boville, B., Kiehl, J.T., Wigley, T.M.L., and Arblaster, J.M.
2000b. Response of the NCAR Climate System Model to increased CO2 and the role of phys-
ical process. J. Climate 13 (11): 1879–98.
Meehl, G.A., Hurrell, J.W., and van Loon. H. 1998. A modulation of the mechanism of the semi-
annual oscillation in the southern hemisphere. Tellus, 50: 442–50.
Miller, F.R. and Keshvamurty, R.N. 1968. Structure of an Arabian Sea Summer Monsoon System.
International Indian Ocean Expedition, Met. Monogr. (East–West Center), University Press of
Hawaii, Honolulu HI, 94 pp.
Miller, J. and Russell, G. 1989. Ocean heat transport during the Last Glacial Maximum.
Paleoceanography, 4: 161–55.
Mo, K.C. and van Loon, H. 1984. Some aspects of the interannual variation of mean monthly sea
level pressure on the southern hemisphere. J. Geophys. Res., 89: 9541–6.
Mooley, D.A. 1973. Some aspects of Indian monsoon depressions and the associated rainfall. Mon.
Wea. Rev., 101: 271–80.
Moses, T., Kiladis, G.N., Diaz, H.F., and Barry, R.G. 1987. Characteristics and frequency of rever-
sals in mean sea level pressure in the North Atlantic sector and their relationship to long-term
temperature trends. J. Climatol., 7: 13–30.
Murakami, T. and Sumi, A. 1982. Southern hemisphere summer monsoon circulation during the
1978–79 WMONEX. I. Monthly mean wind fields. J. Met. Soc. Japan, 60: 638–48.
Nagurny, A. 1998. Climatic characteristics of the tropopause over the Arctic Ocean. Ann.
Geophysicae 16: 110–15.
Nakamura, N. and Oort, A.H. 1988. Atmospheric heat budgets of the polar regions. J. Geophys.
Res., 93: 9510–24.
Navarra, A., Stern, W.F., and Miyakoda, K. 1994. Reduction of the Gibbs Oscillation in spectral
model simulations. J. Climate, 7 (8): 1169–83.
Neelin, J.D. and Held, I.M. 1987. Modeling tropical convergence based on the moist static energy
budget. Mon. Wea. Rev., 115 (1): 3–12.
Neelin, J.D., Latif, M., and Jin, F.-F. 1994. Dynamics of coupled ocean–atmosphere models: the
tropical problem. Annu. Rev. Fluid. Mech., 26: 617–59.
Newell, R.E., Kidson, J.W., Vincent, D.G., and Boer, G.J. 1972. The General Circulation of the
Tropical Atmosphere. II. MIT Press, Cambridge MA, 258 pp.
Newell, R.E., Kidson, J.W., Vincent, D.G., and Boer, G.J. 1974. The General Circulation of the
Tropical Atmosphere, II. MIT Press, Cambridge MA, 371 pp.
Newell, R.E., Vincent, D.G., Dopplick, T.G., Ferruzza, D., and Kidson, J.W. 1970. The energy
balance of the global atmosphere. In: G.A. Corby, ed., The Global Circulation of the Atmosphere,
Roy. Met. Soc., London, pp. 42–90.
Nobre, C.A., Sellers, P.J., and Shukla, J. 1991. Amazonian deforestation and regional climate change.
J. Climate, 4 (10): 957–87.
O’Brien, K.L. 1996. Tropical deforestation and climate change. Progr. Phys. Geogr., 20:
311–35.
Ohmura, A. and Gilgen, H. 1993. Re-evaluation of the global energy balance. In: Interactions
11 between Global Climate Subsystems: the Legacy of Hann. Geophys. Monogr. 75, Amer. Geophys.
Union, Washington DC, pp. 93–110.
Oke, T.R. 1987. Boundary Layer Climates. 2nd edition. Routledge, London, 435 pp.
Oort, A.H. 1982. Global Atmospheric Circulation Statistics, 1958–73, NOAA Prof. Paper 14, U.S.
Dept. of Commerce, 180 pp.
Oort. A.H. 1985. Balance conditions in the earth’s climate system. Adv. Geophys., A28:
75–98.
Oort, A.H. 1989. Angular momentum cycle in the atmosphere–ocean–solid earth system. Bull. Amer.
Met. Soc., 70: 1231–42.
Oort, A.H. and Peixoto, J.P. 1983. Global angular momentum and energy balance requirements
0 from observations. Adv. Geophys., 25: 355–490.
Oort, A.H. and Yienger, J.J. 1996. Observed interannual variability in the Hadley circulation and
its connection to ENSO. J. Climate, 9 (11): 2751–67.
Overland, J.E., Adams, J.M., and Bond, N.A. 1999. Decadal variability of the Aleutian low and its
relation to high-latitude circulation. J. Climate 12 (5, Pt. 2): 1542–8.
Ozorai, Z. 1963. An assessment of ideas relative to air masses. Idojaras, 67: 193–203.
Pan, H.L., Jia, X., and Yang, X.-Z. 1985. The climatological features of outbreaks of cold air in
China. Beijing Met. Center, State Meteorological Agency, China, pp. 120–31 (in Chinese; cited
by Ding, 1994).
Peixoto, J.P. and Oort, A.H. 1992. Physics of Climate. American Institute of Physics, New York,
Chs. 6, 11, 13 and 14.
0 Penner, C. M. 1955. A three-front model for synoptic analysis. Quart. J. Roy. Met. Soc., 81:
89–91.
Perevedentsev, Yu P., Ismagilov, N.V., and Shantalinskii, K.M. 1994. Tsentry deistviya atmosfery
i ikh vzaimosvyaz’ s makrotsirkulyatsionnymi protsessami (Atmospheric centers of action and
their relations with macrocirculation processes). Meteorologiya i Gidrologiya, 1994 (3): 43–51.
Perry, A.H. 1971. Changes in position and intensity of major northern hemisphere centers of action.
Weather, 26 (6): 268–70.
Petrosyants, M.A. and Belov, P.N. (eds). 1988. Tropicheskie Mussony (Tropical Monsoons) (in
Russian), Gidrometeoizdat, Leningrad, 336 pp.
Petterssen, S. 1950. Some aspects of the general circulation of the atmosphere. Centen. Proc. Roy.
Met. Soc., London, pp. 120–55.
0 Philander, S.G.H., Gu, D., Halpern, D., Lambert, G., Lau, N.C., Li, T., and Pacanowski, R.C. 1996.
Why the ITCZ is mostly north of the equator. J. Climate, 9 (12): 2958–72.
Phillips, N.A. 1956. The general circulation of the atmosphere: a numerical experiment. Quart. J.
Roy. Met. Soc., 82: 123–64.
Phillips, T.J. 1994. A Summary Documentation of the AMIP Models, PCDMI Report 18, UCRL-ID-
116384, University of California, Lawrence L. Livermore National Laboratory, Livermore, CA.
Physick, W.L. 1981. Winter depression tracks and climatological jetstreams in the southern hemi-
sphere during the FGGE year. Quart. J. Roy. Met. Soc., 107: 383–98.
Pielke, R.A. and Vidale, P.L. 1995. The boreal forest and the polar front. J. Geophys. Res., 100
(D12): 25, 755–8.
Pike, A.C. 1971. Intertropical Convergence Zone studied with an interacting atmosphere and ocean
0 model. Mon. Wea. Rev., 99 (6): 469–77.
Pisias, N.G. and Imbrie, J. 1986. Orbital geometry, CO2 and Pleistocene climate. Oceans, 29: 43–9.
Pollack, J.B. and Yung, Y.L. 1980. Origin and evolution of planetary atmospheres. Ann. Rev. Earth
Planet. Sci., 8: 425–88.
Pulwarty, R.S., Barry, R.G., and Riehl, H. 1992. Annual and seasonal patterns of rainfall variability
over Venezuela. Erdkunde, 51: 273–89.
Raatz, W.E. and Shaw, G.E. 1984. Long-range tropospheric transport of pollution aerosols into the
Alaskan Arctic. J. Climate Appl. Met., 23: 1052–64.
Racz, Z. and Smith, R.K. 1999. The dynamics of heat lows. Quart. J. Roy. Met. Soc., 125: 225–52.
Ramage, C.S. 1962. The subtropical cyclone. J. Geophys. Res., 67: 1401–11.
Ramage, C.S. 1966. The summer atmospheric circulation over the Arabian Sea. J. Atmos. Sci., 23:
0 144–50.
Ramage, C.S. 1968. Role of a tropical “Maritime Continent” in the atmospheric circulation. Mon.
Wea. Rev., 96: 365–70.
11 Ramage, C.S. 1971. Monsoon Meteorology. Academic Press, New York, 296 pp.
Ramage, C.S. and Raman, C.R.V. 1972. Meteorological Atlas of the International Indian Ocean
Expedition. II. Upper Air. National Science Foundation, Washington DC.
Ramage, C.S., Khalsa, J.J.S., and Meisner, B.N. 1981. The central Pacific near-equatorial conver-
gence zone. J. Geophys. Res., 86 (C7): 6580–98.
Ramage, C.S., Miller, F.R., and Jeffries, C. 1972. Meteorological Atlas of the International Indian
Ocean Expedition. I. The Surface Climates of 1963 and 1964. National Science Foundation,
Washington DC.
Ramanathan, V., Cess, R.D., Harrison, E.F., Minnis, P., Barkstrom, B., Amhad, E., and Hartmann,
D. 1989. Cloud-radiative forcing and climate: results from the Earth Radiation Budget Experiment.
Science, 243: 57–63.
Randall, D. 1996. A university perspective on global climate modeling. Bull. Amer. Met. Soc., 77
(11): 2685–700.
Rao, Y.P. 1976. The Southwest Monsoon. Monograph, Synoptic Meteorology 1, Indian Meteor-
ological Department, New Delhi, 367 pp.
Raphael, M. 1997. The relationship between the transient, meridional eddy sensible and latent heat
flux. J. Geophys. Res., 102 (D12): 13487–94.
Raymond, D.J. 1994. Convective processes and tropical atmospheric circulation. Quart. J. Roy. Met.
Soc., 120: 1431–55.
Reed, R.J. and Danielsen, E.F. 1959. Fronts in the vicinity of the tropopause. Arch. Met. Geophys.
Biokl., 11: 1–17.
Reiter, E.R. 1978. Long-term wind variability in the tropical Pacific: its possible causes and effects.
Mon. Wea. Rev., 101 (6): 324–30.
Reiter, E.R. 1979. Trade-wind variability, Southern Oscillation and Quasi-biennial Oscillation. Arch.
Met. Geophys. Biokl., A28: 113–26.
Reynolds, O. 1884. An experimental investigation of the circumstances which determine whether
the motion of water shall be direct or sinuous, and of the law of resistance in parallel channels.
Phil. Trans. Roy. Soc. (London) A, 174: 935–82.
Richardson, L.F. 1922. Weather Prediction by Numerical Process. Cambridge University Press,
London (reprinted by Dover Publications, New York, 1965), 236 pp.
Riehl, H. 1950. On the role of the tropics in the general circulation of the atmosphere. Tellus, 2:
1–17.
Riehl, H. 1954. Tropical Meteorology. McGraw-Hill, New York, 392 pp.
Riehl, H. and Malkus, J.S. 1958. On the heat balance of the equatorial trough zone. Geophysica
(Helsinki), 6: 503–38.
Riehl, H. and (Malkus) Simpson, J.S. 1979a. The heat balance of the equatorial trough zone revis-
ited. Contrib. Atmos. Phys., 52: 287–305.
Riehl, H. and Simpson, J. (Malkus) 1979b. The heat balance of the equatorial trough revisited.
Beitr. Phys. Atmos., 52: 287–305.
Rigor, I.G., Colony, R.L., and Martin, S. 2000. Variations in surface air temperature observations
in the Arctic, 1979–97. J. Climate, 13 (5): 896–914.
Rind, D. 1998. Latitudinal temperature gradients and climate change. J. Geophys. Res., 103 (D6):
5943–71.
Rind, D. and Rossow, W.B. 1984. The effects of physical processes on the Hadley circulation.
J. Atmos. Sci., 41: 479–507.
Robinson, P., 1972. Palaeoclimatology and Continental Drift. (Unpublished paper, Geology section,
British Association, Leicester.)
Rodwell, M.J. and Hoskins, B.J. 1996. Monsoons and the dynamics of deserts. Quart. J. Roy. Met.
Soc., 122: 1385–404.
Rogers, J.C. 1981. The North Pacific oscillation. J. Climatol., 1: 39–57.
Rosen, R.D. and Salstein, D.A. 1983. Variations in atmospheric angular momentum on global and
regional scales and the length of day. J. Geophys. Res., 88: 5451–70.
Rosen, R.D., Salstein, D.A., Miller, A.J., and Arpe, K. 1987. Accuracy of atmospheric angular
momentum estimates from operational analyses. Mon. Wea. Rev., 115: 1627–29.
Rossby, C.-G. 1940. Planetary flow patterns in the atmosphere. Quart. J. Roy. Met. Soc., 66 (supple-
ment): 68–87.
Rossby, C.-G. 1949. On the maintenance of the general circulation of the lower atmosphere. In:
G.P. Kuiper, ed., The Atmosphere of the Earth and Planets. University of Chicago Press, Chicago,
pp. 16–48.
11 Rossow, W.B. 1985. Atmospheric circulation of Venus. In: S. Manabe, ed., Adv. Geophys., 28A
(Issues in Atmospheric and Oceanic Modeling, Part A. Climate Dynamics), pp. 347–79.
Rossow, W.B. 1993. Clouds. In: R.J. Gurney, J.L. Foster, and C.L. Parkinson, eds, Atlas of Satellite
Observatioms related to Global Change, Cambridge University Press, Cambridge, pp. 141–63.
Rossow, W.B. and Garder, L.C. 1993. Validation of ISCCP cloud detections. J. Climate, 6 (12):
2370–93.
Rossow, W.B. and Schiffer, R.A. 1991. ISCCP cloud data products. Bull. Amer. Met. Soc., 72(1): 2–20.
Ruddiman, W.F. and Kutzbach, J.E. 1989. Forcing of late Cenozoic northern hemisphere climate by
plateau uplift in southern Asia and the American west. J. Geophys. Res., 94: 18409–27.
Sadler, J.C. 1975. The monsoon circulation and cloudiness over the GATE area. Mon. Wea. Rev., 122
0 (9): 369–87.
Saha, K.R. and Bavadekar, S.N. 1973. Water vapor budget and precipitation over the Arabian Sea dur-
ing the northern summer. Quart. J. Roy. Met. Soc., 99: 273–8.
Sahsamanoglou, H.S. 1990. A contribution to the study of action centers in the North Atlantic. Intl. J.
Climatol., 10: 247–61.
Sahsamanoglou, H.S., Makrogiannis, T.J., and Kallimopolous, P.P. 1991. Some aspects of the basic
characteristics of the Siberian anticyclone. Intl. J. Climatol., 11: 827–39.
Samel, A.N., Wang, W.-C., and Liang, X.-Z. 1999. The monsoon rain band over China and relation-
ships with the Eurasian circulation. J. Climate. 12 (1) : 155–81.
Sankar-Rao, M., Lau, K.M., and Yang, S. 1996. On the relationship between Eurasian snow cover and
the Asian summer monsoon. Intl. J. Climatol., 16 (6): 605–16.
0 Sarachik, E.S. 1985. A simple theory for the vertical structure of the tropical atmosphere. Pure Appl.
Geophys., 123: 261–71.
Sawyer, J.S. 1949. The significance of dynamic instability in atmospheric motions. Quart. J. Roy. Met.
Soc., 75: 364–74.
Sawyer, J.S. 1963. Notes on terminology and conventions for the high atmosphere. Quart. J. Roy. Met.
Soc., 89: 156.
Schinze, G. 1932. Troposphärische Luftmassen und vertikaler Temperaturgradient. Beitr. Phys. fr.
Atmos., 19: 79–90.
Schlesinger, M.E. (ed.). 1988. Physically based Modelling and Simulation of Climate and Climate
Change, I. Reidel, Dordrecht, 624 pp.
Schlesinger, M.E. 1989. Quantitative analysis of feedbacks in climate model simulations. In: A.
0 Berger, R.E. Dickinson and J.W. Kidson, eds, Understanding Climate Change, Geophys. Monogr.
52, Amer. Geophys. Union, Washington DC, pp. 177–87.
Schneider, E.K. 1977. Axially symmetric steady-state models of the basic state for instability and cli-
mate studies. I. Linearized calculations. J. Atmos. Sci., 34 (2): 263–79. II. Non-linear calculations.
J. Atmos. Sci., 34 (2): 280–96.
Scholefield, J.T., et al. 1998. The Mars Pathfinder Atmospheric Structure Investigation/Meteorology
(ASI/MET) Experiment. Science, 278 (5344): 1752–8.
Schubert, S.D., Rood, R.B., and Pfaendtner, J. 1993. An assimulated data set for earth science appli-
cations. Bull. Amer. Met. Soc., 74 (12): 2331–42.
Schubert, W.H., Cieselski, P.E., Lu, C.G., and Johnson, R.H. 1995. Dynamical adjustments of the trade
0 wind inversion layer. J. Atmos. Sci., 52 (16): 2941–52.
Schulman, L.L. 1973. On the summer hemisphere Hadley cell. Quart. J. Roy. Met. Soc., 99: 197–201.
Schwerdtfeger, W. 1960. The seasonal variation of the strength of the southern circumpolar vortex.
Mon. Wea. Rev., 88: 203–8.
Schwerdtfeger, W. 1970. The Climate of the Antarctic, World Survey of Climatology 14. Elsevier,
Amsterdam, pp. 253–355.
Schwerdtfeger, W. 1984. Weather and Climate of the Antarctic. Elsevier, Amsterdam, 261 pp.
Semazzi, F.H.M. and Sun, L.-Q. 1997. The role of orography in determining the Sahelian climate. Intl.
J. Climatol., 17: 581–96.
Serreze, M.C. 1995. Climatological aspects of cyclone development and decay in the Arctic. Atmos.-
Ocean, 33: 1–23.
0 Serreze, M.C. and Barry, R.G. 1988. Synoptic activity in the Arctic Basin, 1979–85. J. Climate, 1:
1276–95.
Serreze, M.C., Barry, R.G., and McLaren, A.S. 1989. Seasonal variations in sea ice motion and effects
11 on sea ice concentration in the Canada Basin. J. Geophys. Res., 94 (8): 10955–70.
Serreze, M.C., Barry, R.G., Rehder, M.C., and Walsh, J.E. 1995. Variability in atmospheric circulation
and moisture flux over the Arctic. Phil. Trans. Roy. Soc. (London), A352: 215–25.
Serreze, M.C., Box, J.E., Barry, R.G., and Walsh, J.E. 1993. Characteristics of Arctic synoptic activ-
ity, 1952–89. Met. Atmos. Phys., 51: 147–64.
Serreze, M.C., Carse, F., Barry, R.G., and Rogers, J.C. 1997. Icelandic low cyclone activity: clima-
tological features, linkages with the NAO and relationships with recent changes in the northern
Shaw, Sir Napier. 1930. Manual of Meteorology. III. The Physical Processes of Weather. Cambridge
University Press, Cambridge.
Shine, K.P. and Forster, P.M. de F. 1999. The effect of human activity on radiative forcing of
climate change: a review of recent developments. Global Planet. Change, 20: 205–25.
Sikka, D.R. 1977. Some aspects of the life, history, structure and movement of monsoon depres-
sions. Pageoph., 115: 1501–24.
Sikka, D.R. and Gadgil, S. 1980. On the maximum cloud zone and the ITCZ over Indian longi-
tudes during the south-west monsoon. Mon. Wea. Rev., 108: 1840–53.
Simmonds, I. and Jones, D.A. 1998. The mean structure and temporal variability in the semi-annual
oscillation in the southern extratropics. Intl. J. Climatol., 18: 473–504.
Simpson, R.H. 1952. Evolution of the kona storm: a subtropical cyclone. J. Met., 9: 24–35.
Sinclair, M.R. 1996. A climatology of anticyclones and blocking for the southern hemisphere. Mon.
Wea. Rev., 124 (2): 245–63.
Smagorinsky, J. 1953. The dynamic influence of large-scale heat sources and sinks on the quasi-
stationary mean motions of the atmosphere. Quart. J. Roy. Met Soc., 79: 342–66.
Smirnov, A., Royer, A., O’Neill, N.T., and Tarussov, A. 1994. A study of the link between synoptic
air mass type and atmospheric optical parameters. J. Geophys. Res., 99 (D10): 20967–82.
Smith, E.A. 1986. The structure of the Arabian heat low. I. Surface energy budget. II. Bulk tropo-
spheric heating budget and implications. Mon. Wea. Rev., 114 (6): 1067–83; 1084–102.
Sohn, B.J. and Smith, E.A. 1993. Energy transports by ocean and atmosphere based on an entropy
extremum principle. I. Zonally averaged transports. J. Climate, 6 (5): 886–99.
Solomon, A.B. 1997. An observational study of the spatial and temporal scales of transient eddy
sensible heat fluxes. J. Climate, 10: 508–20.
Spar, J. 1950. On the theory of annual pressure variations. J. Met., 7: 167–80.
Starr, V.P. 1948. An essay on the general circulation of the Earth’s atmosphere. J. Met., 5: 39–43.
Starr, V.P. 1968. The Physics of Negative Viscosity Phenomena. McGraw-Hill, New York, 256 pp.
Starr, V.P. and White, R.M. 1952. Schemes for the study of hemispheric exchange processes. Quart.
J. Roy. Met. Soc., 78: 407–10.
Stearns, C.R. and Wendler, G. 1988. Research results from Antarctic automatic weather stations.
Revs. Geophys., 26: 45–61.
Stewart, R.E., Szeto, K.K., Reinking, R.F., Clough, S.A., and Ballard, S.P. 1998. Midlatitude
cyclonic cloud systems and their features affecting large scales and climate. Rev. Geophys., 36:
245–73.
Stone, P.H. 1972. A simplified radiative-dynamical model for the static stability of rotating atmos-
pheres. J. Atmos. Sci., 29: 405–18.
Stone, P.H. and Miller, D.A. 1980. Empirical relations between seasonal changes in meridional
temperature gradients and meridional fluxes of heat. J. Atmos. Sci., 37: 1708–21.
Streten, N.A. 1968. A note on multiple image photo-mosaics for the southern hemisphere. Austral.
Met. Mag., 16: 127–36.
Streten, N.A. 1980. Some synoptic indices of the southern hemisphere mean sea level circulation,
1972–77. Mon. Wea. Rev. 108: 18–36.
Streten, N.A. and Troup, A.J. 1973. A synoptic climatology of satellite-observed cloud vortices in
the southern hemisphere. Quart. J. Roy. Met. Soc., 99: 56–72.
Streten, N.A. and Zillman, J.W. 1984. Climates of the South Pacific. In: H. van Loon, ed., Climates
of the Oceans, World Survey of Climatology 15, Elsevier, Amsterdam, pp. 263–429.
Suppiah, R. 1992. The Australian summer monsoon: a review. Progr. Phys. Geogr., 16:
263–318.
Taljaard, J.J. 1967. Development, distribution and movement of cyclones and anticyclones in the
southern hemisphere during the IGY. J. Appl. Met., 6: 973–87.
Taljaard, J.J. 1972. Synoptic meteorology of the southern hemisphere. In: C.W. Newton, ed.,
Meteorology of the Southern Hemisphere. Meteorol. Monogr., 13 (35), pp. 139–213.
11 Tanaka, M. 1992. Intraseasonal oscillation and the onset and retreat dates of the summer monsoon
over East, Southeast Asia and the western Pacific region using GMS high cloud amount data.
J. Met. Soc. Japan., 70: 613–29.
Tanaka, M. 1994. The onset and retreat dates of the austral summer monsoon over Indonesia,
Australia and New Guinea. J. Met. Soc. Japan., 72: 255–67.
Tang, M.-C. and Reiter, E.R. 1987. Plateau monsoons of the northern hemisphere: a comparison
between North America and Tibet. Mon. Wea. Rev., 112 (4): 617–37.
Teisserenc de Bort, L. 1883. Étude sur l’hiver de 1879–80 et recherches sur la position des centres
d’action de l’atmosphère dans les hivers anormaux. Ann. Bureau Central Met. de la France 1881,
Pt 4: 17–62.
0 Teisserenc de Bort, L. 1902. Variations de la temperature de l’aire libre dans la zone comprise
entre 8 km et 13 km d’altitude. Comptes Rendus, Seances Acad. Sci. (Paris), 134: 987–9.
Thorpe, A.J. 1994. The contribution of Ernst Kleinschmidt to cyclone research. In: S. Grønås and
M.A. Shapiro, eds, The Life Cycles of Extratropical Cyclones, I. Alma Mater Forlag, Bergen,
pp. 59–69.
Thrope, A.J. and Volkert, H. 1997. Potential vorticity: a short history of its definitions and uses.
Met. Zeit., N.F. 6: 275–80.
Thuburn, J. and Craig, G.C. 1997. GCM tests of theories for the height of the tropopause. J. Atmos.
Sci., 54: 869–82.
Tomas, R.A. and Webster, P.J. 1997. The role of inertial instability in determining the location and
strength of near-equatorial convection. Quart. J. Roy. Met. Soc., 123: 1445–82.
0 Trenberth, K.E. 1979. Mean annual poleward energy transports by the oceans in the southern hemi-
sphere. Dyn. Atmos. Oceans, 4: 57–64.
Trenberth, K.E. 1981a. Interannual variability of the southern hemisphere 500 mb flow: regional
characteristics. Mon. Wea. Rev., 109: 127–36.
Trenberth, K.E. 1981b. Observed southern hemisphere eddy statistics at 500 mb: frequency and
spatial dependence. J. Atmos. Sci., 38 (12): 2585–605.
Trenberth, K.E. 1992a. Global analyses from ECMWF and atlas of 1000 to 10 mb circulation
statistics. NCAR Tech. Note TN-373STR, National Center for Atmospheric Research, Boulder
CO.
Trenberth, K.E. (ed.). 1992b. Climate System Modeling. Cambridge University Press, Cambridge,
788 pp.
0 Trenberth, K.E. 1998. Atmospheric moisture residence times and cycling: implications for rainfall
rates and climate change. Climate Change 39: 667–94.
Trenberth, K.E. 1999. Atmospheric moisture recycling: role of advection and local evaporation.
J. Climate 12 (5, Pt 2): 1368–81.
Trenberth, K.E. and Solomon, A. 1994. The global heat balance: heat transports in the atmosphere
and ocean. Climate Dynam., 10: 107–34.
Trenberth, K.E., Christy, J.R., and Olson, J.G. 1987. Global atmospheric mass, surface pressure and
water vapor variations. J. Geophys. Res., 92 (D2): 14815–26.
Trewartha, G.T. 1961. The Earth’s Problem Climates. University of Wisconsin Press, Madison WI,
334 pp.
0 Trewartha, G.T. 1981. The Earth’s Problem Climates. 2nd edition. University of Wisconsin Press,
371 pp.
Tucker, G.B. 1962. Review of The Earth’s Problem Climates by G.T. Trewartha. Quart. J. Roy.
Met. Soc., 87: 460–1.
Tucker, G.B. 1979. Transient synoptic systems as mechanisms for meridional transport: an obser-
vational study in the southern hemisphere. Quart. J. Royal Met. Soc., 105: 657–72.
Tucker, G.B. and Barry, R.G. 1984. The climate of the North Atlantic Ocean. In: H. van Loon,
ed., Climates of the Oceans, World Survey of Climatology 15 (H. Landsberg, ed.-in-chief),
Elsevier, Amsterdam, pp. 193–262.
Turner, J., Colwell, S.R., and Harangozo, S. 1997. Variability of precipitation over the coastal
western Antarctic Peninsula from synoptic observations. J. Geophys. Res., 102: 13999–14007.
0 Van Den Broeke, M.R. 1998a. The semi-annual oscillation and Antarctic climate. 1. Influence on
near surface temperatures, 1957–79. Ant. Sci., 10: 175–83.
Van Den Broeke, M.R. 1998b. The semi-annual oscillation and Antarctic climate. 2. Recent changes.
11 Ant. Sci., 10: 184–91.
van den Dool, H. and Saha, S. 1993. Seasonal redistribution and conservation of atmospheric mass
in a general circulation model. J. Climate, 6: 22–30.
van Loon, H. 1956. Blocking action in the southern hemisphere. 1. Notos, 5: 171–5.
van Loon, H. 1964. Mid-season average zonal winds at sea level and at 500 mb south of 25°S and
a brief comparison with the northern hemisphere. J. Appl. Met., 3: 553–63.
van Loon, H. 1967. The half-yearly oscillations in the middle and high southern latitudes and the core-
less winter. J. Atmos. Sci., 24: 472–86.
van Loon, H. 1972a. Temperature in the southern hemisphere. In: C.W. Newton, ed., Meteorology of
the Southern Hemisphere. Met. Monogr. 13 (35), Amer. Met. Soc., Boston MA, pp. 25–58.
van Loon, H. 1972b. Wind in the southern hemisphere. In: C.W. Newton, ed., Meteorology of the
Southern Hemisphere. Met. Monogr. 13 (35), Amer. Met. Soc., Boston MA, pp. 87–100.
van Loon, H. 1979. The association between latitudinal temperature gradient and eddy transport. I.
Transport of sensible heat in winter. Mon. Wea. Rev., 107 (1): 2–34.
van Loon, H. 1980. Transfer of sensible heat by transient eddies in the atmosphere on the southern
hemisphere: an appraisal of the data before and during FGGE. Mon. Wea. Rev., 108: 1774–81.
van Loon, H. (ed.), 1984. Climates of the Oceans. World Survey of Climatology 15 (H.E. Landsberg,
ed.-in-chief), Elsevier, Amsterdam, 716 pp.
van Loon, H. 1991. A review of the surface climate of the southern hemisphere and some comparisons
with the northern hemisphere. J. Mar. Systems, 2: 171–94.
van Loon, H. and Kidson, J.W. 1993. The association between latitudinal temperature gradient and
eddy transport. III. The southern hemisphere. Aust. Met. Mag., 42: 31–7.
van Loon, H. and Rogers, J.C. 1984a. Interannual variations in the half-yearly cycle of pressure gradi-
ents and zonal wind at sea level on the southern hemisphere. Tellus, 36A: 76–86.
van Loon, H. and Rogers, J.C. 1984b. The yearly wave in pressure and zonal geostrophic wind at sea
level on the southern hemisphere and its interannual variability. Tellus, 36A: 348–54.
van Loon, H. and Williams, J. 1980. The association between latitudinal temperature gradient and eddy
transport. II. Relationships between sensible heat transport by stationary waves and wind, pressure
and temperature in winter. Mon. Wea. Rev., 108: 604–14.
von Storch, H. 1995. Inconsistencies at the interface of climate impact studies and global climate
research. Met. Zeit., n.s., 4: 72–80.
Varejao-Silva, M.A., Franchito, S.H., and Rao, V.B. 1998. A coupled biosphere–atmosphere climate
model suitable for studies of climatic change due to land surface alterations. J. Climate, 11 (8):
1749–67.
Verma, R.K. and Sikka, D.R. 1981. The annual oscillation of the tropospheric temperature in the north-
ern hemisphere. In: Sir J. Lighthill and R.P. Pearce, eds, Monsoon Dynamics, Cambridge University
Press, Cambridge, pp. 49–64.
Vernekar, A.D., Zhou, J., and Shukla, J. 1995. The effect of Eurasian snow cover on the Indian mon-
soon. J. Climate 8 (2): 248–66.
Vincent, D.G. 1994. The South Pacific convergence zone (SPCZ): a review. Mon. Wea. Rev., 122 (9):
1949–70.
Waliser, D.E. and Gautier, C. 1993. A satellite-derived climatology of the ITCZ. J. Climate, 6 (11):
2162–74.
Waliser, D.E. and Somerville, R.J.C. 1994. Preferred latitudes of the intertropical convergence zone.
J. Atmos. Sci., 51 (12): 1619–39.
Waliser, D.E., Graham, N.E., and Gautier, C. 1993. Comparison of the highly reflective cloud and out-
going long-wave radiation datasets for use in estimating tropical deep convection. J. Climate 6 (2):
331–53.
Walker, G.T. 1914. Further study of relationships with Indian monsoon rainfall. Mem. Indian Met.
Dept., 21 (8): 1–12.
Walker, G.T. 1923. Correlations in the seasonal variations of weather. VIII. A preliminary study of
world weather. Mem. Indian Met. Dept., 24 (4): 75–131.
Walker, J.M. 1967. Subterranean isobars. Weather, 22: 296–7.
Walland, D.J. and Simmonds, I. 1998. Sensitivity of the southern hemisphere semi-annual oscillation
in surface pressure to changes in surface boundary conditions. Tellus, 50A: 424–41.
Wang, B. 1994. Climate regimes of tropical convection and wind. J. Climate, 7 (7): 1109–18.
Wang, B. and Fan, Z., 1999. Choice of South Asian summer monsoon indices. Bull. Amer. Met. Soc.,
804: 629–38.
Wang, B. and Wang, Y.-Q. 1999. Dynamics of the ITCZ–Equatorial Cold Tongue complex and causes
of the latitudinal climate asymmetry. J. Climate 12 (6): 1830–47.
11 Wang, B. and Xu, X.-H. 1997. Northern hemisphere monsoon singularities and climatological intrasea-
sonal oscillations. J. Climate, 10 (5): 1071–85.
Warren, S.G., Hahn, C.J., London, J., Chervin, R.M., and Jenne, R.L. 1986. Global Distributions of
Total Cloud Cover and Cloud Type Amounts over Land. NCAR Tech. Note TN-273STR, NCAR,
Boulder CO., 25 pp + 199 maps.
Warren, S.G., Hahn, C.J., London, J., Chervin, R.M., and Jenne, R.L. 1988. Global Distribution of
Total Cloud Cover and Cloud Type Amounts over the Ocean. NCAR Tech. Note NCAR/TN-
317STR, NCAR, Boulder CO, 42 pp. + 170 maps.
Washington, W.M. and Parkinson, C.L. 1986. An Introduction to Three-dimensional Climate
Modeling. University Science Books, Mill Valley CA., 422 pp.
0 Webster, P.J. 1987a, The elementary monsoon. In: J.S. Fein and P.L. Stephens, eds, Monsoons,
Wiley, New York, pp. 3–320.
Webster, P.J. 1987b. The variable and interactive monsoon. In: J.S. Fein and P.L. Stephens, eds,
Monsoons, Wiley, New York, pp. 269–330.
Webster, P.J. 1994. The role of hydrological processes in ocean–atmosphere interactions. Rev.
Geophys., 32: 427–76.
Webster, P.J. and Lukas, R. 1992. TOGA COARE: the Coupled Ocean–Atmosphere Response
Experiment. Bull. Amer. Met. Soc., 73 (9): 1377–416.
Webster, P.J., Chou, L., and Lau, K.M. 1977. Mechanisms of affecting the state, evolution and
transition of the planetary-scale monsoon. Pure Appl. Geophys., 115 (5–6): 1463–92.
Webster, P.J., Magana, V.O., Palmer, T.N., Shukla, J., Tomas, R.A., Yanai, M., and Yasunari, T.
0 1998. Monsoons: processes, predictability, and the prospects for prediction. J. Geophys. Res.,
103: 14451–510.
Wells, N. 1986. The Atmosphere and Ocean: A Physical Introduction. Taylor & Francis, London,
pp. 5 and 43.
Wendland, W.M. and Bryson, R.A. 1981. Northern hemisphere airstream regions. Mon. Wea. Rev.,
109: 255–70.
Wendland, W.M. and McDonald, N.S. 1986. Southern hemisphere airstream climatology. Mon. Wea.
Rev., 114: 88–94.
Wendler, G. 1986. The “radiation paradox” on the slopes of the Antarctic continent. Polarforschung,
56: 33–42.
Wexler, H. 1937. Formation of polar anticyclones. Mon. Wea. Rev., 65: 229–36.
0 Wexler, H. 1951. Anticyclones. In: T.F. Malone, ed., Compendium of Meteorology, Amer. Met.
Wiin-Nielsen, A. 1970. On Inertial Flow, Report No. 1, Institute of Theoretical Meteorology,
University of Copenhagen, 25 pp.
Wild, M. and Ohmura, A. 1999. The role of clouds and the cloud-free atmosphere in the problem
of undetermined absorption of solar radiation in GCM atmospheres. Phys. Chem. Earth, B24 (3):
261–8.
Wild, M., Ohmura, A., Gilgen, H., and Roeckner, E. 1995. Validation of GCM simulated radiative
fluxes using surface observations. J. Climate 8: 1309–24.
Willett, H.C. 1933. American Air Mass Properties. Pap. Phys. Oceanog. Met. 2 (2), WHOI/MIT,
Cambridge MA, 116 pp.
0 Williams, C.R., Gage, K.S., and Eckland, W.L. 1994. Application of 915 MHz wind profiles to the
classification of tropical precipitating cloud systems observed during TOGA COARE. Sixth
Conference on Climate Variations, Preprints, Amer. Met. Soc., Boston MA., pp. J68–J77.
Williams, E. and Renno, N. 1993. An analysis of the conditional instability of the tropical atmos-
phere. Mon. Wea. Rev., 121 (1): 21–36.
Williams, G.D. 1988. The dynamical range of global circulations I. Climate Dynam., 2: 205–60.
WMO/ICSU. 1980. Winter MONEX Field Phase Report, FGGE Operations Report Ser. 7, WMO,
Geneva (n.p.).
WMO/ICSU. 1981. Summer MONEX Field Phase Report, FGGE Operations Report Ser. 7, WMO,
Geneva (n.p.).
World Climate Research Programme. 1998. Proceedings of the First WCRP International
0 Conference on Reanalysis, WCRP-104, WMO/TD No. 876, World Meteorological Organization,
Geneva, 461 pp.
World Climate Research Programme. 1988. Concept of the Global Energy and Water Cycle
11 Experiment. WCRP-5, WMO/TD No. 215, World Meteorological Organization, Geneva, 70 pp.
Wright, W.J. 1997. Tropical–extratropical cloudbands and Australian rainfall. I. Climatology. Intl.
J. Climatol., 17 (8): 807–29.
Wyrtki, K. and Meyers, G. 1976. The trade wind field over the Pacific Ocean. J. Appl. Met, 15:
698–704.
Xu, K.-M. and Emmanuel, K.A. 1989. Is the tropical atmosphere conditionally unstable? Mon. Wea.
Rev., 117(7): 1471–9.
Xu, Y.-G. (ed.) 1986. Proceedings of International Symposium on the Qinghai-Xijang Plateau and
Mountain Meteorology. Science Press, Beijing., and American Meteorological Society, Boston
MA, 1,036 pp.
Xue, X. and Shukla, J. 1993. The influence of land surface properties on Sahel climate. I.
Desertification. J. Climate, 6 (12): 2232–45.
Yanai, M. and Tomita, T. 1998. Seasonal and interannual variability of atmospheric heat sources
and moisture sinks as determined from NCEP–NCAR reanalysis. J. Climate, 11 (3): 463–82.
Yang, H. and Tung, K.K. 1998. Water vapor, surface temperature and the greenhouse effect – a
statistical analysis of tropical-mean data. J. Climate, 11 (10): 2686–97.
Yasunari, T., Kitoh, A., and Tokioka, T. 1991. Local and remote responses to excessive snow mass
over Eurasia appearing in the northern spring and summer climate: a study with the MRI–GCM.
J. Met. Soc. Japan, 69: 473–87.
Yeh, T.C. and Gao, Y.X. 1979. Meteorology of the Tibetan Plateau (in Chinese), Scientific
Publication Agency, Beijing, 278 pp.
Young, J.A. 1987. Physics of monsoons: the current view. In: J.S. Fein and P.L. Stephens, eds,
Monsoons, Wiley, New York, pp. 211–43.
Zeng, N. and Neelin, J.D. 1999. A land–atmosphere interaction theory for the tropical deforesta-
tion problem. J. Climate 12 (3): 857–72.
Zhang, C.D. 1993. Large-scale variability of atmospheric deep convection in relation to sea-surface
temperatures in the tropics. J. Climate, 6 (10): 1898–913.
Zhang, Y., Sperber, K.R., and Boyle, J.S. 1997. Climatology and interannual variation of the East
Asian winter monsoon: results from the 1979–95 NCEP–NCAR reanalysis. Mon. Wea. Rev., 125
(10): 2005–19.
Zhu, Y. and Newell, R.E. 1998 A proposed algorithm for moisture fluxes from atmospheric rivers.
Mon Wea. Rev., 126 (3): 725–35.
11
4 Large-scale circulation and
climatic characteristics
Understanding the characteristics of the large-scale circulation and climate requires both
the analysis of observational data and interpretation of the physical and dynamical
processes involved through theoretical analysis and modeling. In the following sections
the spatio-temporal characteristics of the circulation in extratropical and tropical latitudes
are discussed, together with consideration of their interactions. In section 3.2 the focus
of general circulation transport mechanisms was on zonal mean flow, meridional standing
0 cells, and time and space eddies. The alternative paradigm, implicit in the classical work
of Rossby, Namias, Palmén, and Riehl, among others, is that the high and low-frequency
transient components can be separated from the time mean (zonal mean and stationary
wave) components. Wallace (1987) points out that low-frequency dynamics, including
important longitudinally dependent interactions between the time-mean flow and tran-
sients, began to receive attention in the late 1970s. This chapter examines the outcomes
of the extensive work that has been performed. The discussion begins with an account
of the westerly vortex and jetstreams, followed by consideration of planetary waves, zonal
and blocked flows, and low-frequency variability and persistence.
0
4.1 Time-averaged circulation
The extratropical circulation is dominated by the tropospheric westerlies. In the free atmos-
phere these constitute a broad flow in each hemisphere around the respective polar
low-pressure centers. Climatically, the extent to which a more or less symmetrical polar
vortex exists depends on the season and level in the atmosphere (Figures 4.1 and 4.2).
In general, the symmetry is greater in the southern than in the northern hemisphere, it is
more developed at high levels (200–100 mb) and, for the northern hemisphere, it is more
pronounced in summer.
In the middle and upper troposphere of the southern hemisphere the polar vortex extends
0 to about 30°S in summer (January) and 10° nearer the equator in winter (July) (van Loon,
1972b). In middle latitudes there is strong baroclinicity, but conditions are almost
barotropic poleward of 60°S. At 500 mb the vortex center is displaced slightly towards
the Ross Sea (180° longitude) in both summer and winter. Geopotential heights are lower
over the colder south pole than the north pole, with a 400 gpm height difference at
500 mb. It must also be emphasized that in spite of the strong zonality of the mean circu-
lation in the southern hemisphere, the daily variability of the height field is closely similar
to that in the northern hemisphere (van Loon, 1972b).
The extent of the northern circumpolar vortex undergoes important temporal variations.
Its size has been estimated by planimetry along a fixed contour for mean monthly maps
0 at 500 mb (Markham, 1985; Davis and Benkovic, 1992) and 300 mb (Angell and
Korshover, 1978). Using the 546 dam contour, which is within the mid-latitude baroclinic
zone in January, Davis and Benkovic (1994) obtain an alternate day record for January
11 1947–90. The mean latitudinal extent of the vortex, so defined, is shown in Figure 4.3.
Figure 4.1 Mean geopotential heights of the 500 mb surfaces (a) January 15 and (b) July 15
1982–86 for the northern hemisphere. The contour interval is 30 m. (From Epstein,
1988)
Large-scale circulation and climate 265
11
0
Figure 4.2 As Figure 4.1 for the southern hemisphere, (a) January 15 and (b) July 15 1982–86.
11 The contour interval is 30 m. (From Epstein, 1988)
Figure 4.3 Mean latitudinal extent of the northern circumpolar vortex, defined by the 546 dam
contour for January 1947–90. (From Davis and Benkovic, 1992)
The principal trough is at 140°E off eastern Asia with another at 80°W and ridges at
10°W and 130°W. The mean position of the zonally averaged 546 dam shifted 1.2° lati-
tude equatorward between 1947–65 and 1966–90, associated with an expansion of the
vortex and amplified troughs over the central North Pacific and eastern North America.
There was also some contraction over western North America, implying a pattern of wave
amplification. Burnett (1993) suggests that these changes reflect a more frequent positive
mode of the Pacific/North American teleconnection pattern (see section 5.6) compared
with an expanded vortex in the early 1960s and 1976. At 300 mb, however, the vortex
contracted sharply in the 1980s (Angell, 1992).
The structure and variability of the polar vortex can also be described by so-called
elliptical diagnostics (Waugh, 1997). An ellipse is fitted to the contour of a quasi-conser-
vative tracer like potential temperature or potential vorticity. Parameters such as the
equivalent latitude of a zonal circle enclosing the same area, the latitude and longitude
of the ellipse center, and the aspect ratio and orientation of the ellipse can be determined.
Waugh and Randel (1999) illustrate the technique for the Arctic and Antarctic stratos-
pheric vortices, showing differences in their structure and seasonal evolution.
It has proved hard to draw a distinction between atmospheric conditions in the tropos-
phere over middle latitudes and those within the polar vortex. Climatological average values
and variance spectra of daily upper-air sounding data (temperature, u and v wind between
850 mb and 100 mb) at Churchill, Manitoba (59°N, 94°W), Oimyakon, eastern Siberia
(63°N, 143°E) and Wakkanai, Japan (45°N, 142°E), provide little evidence to separate the
two zones (Müller et al., 1979). However, conditions within the polar vortex in winter are
most strongly modulated by ultralong period circulation patterns (more than ten days),
related to lower tropospheric cold air outbreaks, with high persistence in the fifteen to
twenty-day range. In summer, changes within the vortex are mainly associated with the
Arctic Front over both short (less than five days) and long (five to ten days) periods.
The strength of the westerlies is determined by the temperature gradient between the
tropics and high latitudes, as a consequence of the thermal wind relationship. As shown
by combining the hydrostatic relationship and the equation of state:
11 ∂p pg
=
∂z RTv
where R the gas constant for dry air (287 J kg1 K1) and Tv virtual temperature;
Tv ≈ T r/6, where r the humidity mixing ratio (g kg1). Thus a warm air column
expands vertically, compared with a cold air column, having the same pressure at sea
level. Hence in this case, at any level there is a horizontal pressure gradient from the
warm air to the cold air. In analogy with the geostrophic wind relationship, air does not
move directly along this pressure gradient but is turned 90° to the right (left) in the
0 northern (southern) hemisphere as a result of the Coriolis force. The balanced state is
referred to as the thermal wind, which blows at right angles to the mean thermal gradient
(with a speed inversely proportional to the isotherm spacing), with cold air to the left
(right) in the northern (southern) hemisphere, viewed downwind. This thermal wind is a
hypothetical component of the actual wind velocity at upper levels. It can be analyzed
directly from charts of the geopotential height difference, or thickness, between two
standard pressure surfaces (e.g. 1,000–500 mb, 700–300 mb, etc.) as illustrated schemat-
ically in Figure 4.4. The thickness value is proportional to the mean temperature of the
air column, as indicated by the following relationship:
冕
0 R p1
dp
(z2 z1) = Tv
g p2 p
For example, the 1,000–500 mb thickness is:

RTv
g
ln 冢
1,000
500 冣
= 20.3Tv
The 1,000–500 mb thickness is 5,685 m for Tv 280 K (a tropical value) and 4,770 m
for Tv 235 K (a polar value). Note that the 500 mb level is near the mid-point of the
0
atmospheric mass and corresponds approximately to the steering level of baroclinic waves.
Figure 4.4 Schematic illustration of the pattern of 1,000–500 mb thickness and thermal wind.
11 (From Barry and Chorley, 1998)
Table 4.1 Mean temperatures at standard pressure levels 1963–73 (°C)
p (mb) 80°S 65° 50° 35° 20° Eq. 20° 35° 50° 65° 80°N
December–January–February
100 40.8 -43.8 52.9 66.1 76.0 79.9 75.4 63.0 55.2 57.5 62.7
200 44.9 47.0 52.1 54.4 53.2 53.3 53.9 55.8 55.6 57.1 59.5
300 52.4 50.6 45.0 37.7 32.0 31.2 34.1 44.8 52.2 55.6 58.1
500 33.1 29.3 20.8 11.3 6.3 5.5 8.1 19.5 30.0 35.7 39.8
700 19.1 15.1 5.8 4.5 9.1 9.6 7.5 3.8 15.0 21.5 26.3
850 – 17.6 1.1 11.7 17.5 17.7 14.7 3.2 8.9 16.5 22.0
1,000 – 0.1 7.9 19.0 25.2 26.8 23.6 13.9 4.3 11.6 23.7
June–July–August
100 78.6 69.9 58.3 60.6 72.5 76.2 73.7 66.1 53.5 46.7 42.7
200 71.1 66.4 59.2 55.1 54.2 53.9 52.7 52.4 52.1 48.1 43.2
300 63.2 59.2 52.4 44.0 34.5 31.8 31.0 33.7 41.1 45.1 46.6
500 41.4 36.2 27.7 18.7 8.6 5.9 5.7 7.9 14.7 19.9 23.8
700 27.3 21.2 11.2 2.4 6.3 8.9 10.6 8.4 0.8 4.6 8.9
850 – 15.3 3.6 4.9 12.9 17.2 19.7 16.6 9.4 3.7 2.3
1,000 – 2.6 5.2 4.4 22.0 26.4 26.6 22.6 11.8 7.9 0.2
Source: from Oort and Peixoto (1983).
In the northern hemisphere the tropics-to-pole temperature gradient increases consider-

ably from summer to winter. Table 4.1 gives mean seasonal temperatures for selected levels
and latitudes. Peixoto and Oort (1992, p. 143) show that the meridional gradient at low lev-
els in northern mid-latitudes is 4°C/1,000 km in summer, but in excess of 8°C/1,000
km in winter. As a result, the upper westerlies, expressed as a zonal average around the
hemisphere, strengthen twofold from summer to winter. In the southern hemisphere, in con-
trast, there is a much more complex pattern of seasonal change in the meridional temper-
ature gradient (Figure 4.5). This is a result of the latitudinal distribution of the land masses
and its effects on the heating of the atmosphere by the surface. Consequently there is no
simple summer to winter strengthening of the zonal circulation. In contrast to the limited
seasonal variation, in the extent of the circumpolar vortex, in the southern hemisphere there
is considerable variation interannually (Trenberth, 1979; 1981a).
Easterly winds dominate the circulation in low latitudes (Figure 4.6). The zonal compo-
nents are only of the order 2–3 m s1, except over the equator in June–August, where
they exceed 5 m s1 above 400 mb. The strongest tropical easterlies are found at 100 mb
and above, between 5°N and 20°N. The occurrence of zonal mean easterlies in the trop-
ical upper troposphere is generally attributed to deceleration of the zonal westerly winds
on the equatorward margin of the subtropical jetstream. According to Feldstein and Held
(1989), the absorption of wave disturbances near their critical latitude (see section 4.3,
p. 302) implies momentum flux divergence from midlatitude-generated eddies. Recent
work challenges this interpretation. As shown in section 3.2 (p. 126), the zonal mean
momentum flux is normally decomposed into three components, owing to the mean
meridional circulation, stationary eddies, and transient eddies. Lee (1999) points out that
the transient eddy term [u′v′] represents the sum of the transient eddy momentum flux
[u′*v′*] and the transient meridional momentum flux [u′][v′]. The contribution of the tran-
sient eddies to the zonal mean momentum flux has been assumed to be small. Lee’s
analysis of 200 mb wind data for 1980–95 shows that in the subtropics the transient
eddies serve to decelerate the zonal mean zonal winds. However, for the deep tropics
(approximately 10°N–10°S) the deceleration is attributable to the horizontal momentum
flux divergence in the transient meridional circulation. This circulation arises from the
seasonal north–south shifts of the Hadley cells. Interestingly, Lindzen and Hou (1988)
find that the upper-tropospheric equatorial easterlies are strongest for a non-equinoctial
11
Figure 4.5 Seasonal change of the zonally

averaged temperature gradient per 5° latitude
at (a) the surface, (b) 500 mb in the southern
hemisphere, and at (c) 500 mb in the northern
0 hemisphere. (From van Loon, 1972a)
11
Figure 4.6 Meridional cross-section along 120°E for December 4 1959, observing temperature,
zonal wind component (knots), and frontal zone. The polar front and subtropical
jetstreams are evident. Jet maxima are shown. (From Hare, 1962)
circulation. This implies that a critical role in their maintenance is played by the earth’s
obliquity (Table 3.3). The transient eddy momentum flux actually accelerates the zonal
mean zonal winds in the deep tropics. The acceleration is supported on the intraseasonal
(thirty to seventy days) time scale by the Madden–Julian Oscillation (see p. 334). There
are also intra- and interannual transient eddy fluxes that feature eastward-propagating
disturbances of zonal wave No. 2 (see section 4.3.1) and a two-to-four-year period of
ENSO-like characteristics. Nevertheless, the deceleration by the transient meridional circu-
lation far outweighs the accelerations by the transient eddy momentum fluxes.
4.2 Jetstreams
North–south cross-sections of temperature reveal that in each hemisphere the atmosphere
is partitioned into two major barotropic zones separated by a narrow baroclinic or frontal
zone (Figure 4.6); that is, there are so-called tropical and polar air masses on either side of
11 the polar front. Associated with the horizontal temperature gradient in the upper tropos-
phere is a maximum of the zonal wind, or jetstream (Riehl et al., 1954, 1962; Reiter, 1996).
In a barotropic atmosphere there are no horizontal temperature gradients, and height
contours of the pressure surfaces have corresponding patterns in the vertical dimension.
Hence there is no vertical shear of the geostrophic wind. Where there are horizontal
temperature gradients, but the isotherms are parallel to the height contours (an equivalent
barotropic atmosphere), the geostrophic wind direction is independent of height but the
speed varies in the vertical. Such an atmosphere may feature warm highs/cold lows where
the pressure anomalies and geostrophic wind speeds increase upward. In the case of cold
0 highs/warm lows the opposite is observed.
Following from the thermal wind relationship discussed above:
VG2 VG1 =
R ∂T
f ∂n 冢冣
p
ln 1 ⬵
p2
constant ∂T
f ∂n
–
where T mean layer temperature between pressure levels 1 and 2, VG the geostrophic
wind velocity, and n is normal to the temperature gradient, implies that the difference in
geostrophic wind speed between levels 1 and 2, or the vertical wind shear, is proportional
to the horizontal gradient of mean layer temperature. This is the basis of the horizontal
0 temperature gradient–jetstream association illustrated in Figure 4.6.
A jetstream is a strong narrow current, typically a few hundred kilometers wide and
several kilometers in depth, extending horizontally over distances of up to several thou-
sand kilometers. The wind speeds are at least 30 m s1 (an arbitrary limit) and may exceed
100 m s1 near the tropopause. They were “discovered” during US air raids over Japan
in 1944, according to Bryson (1994), and were named by C.-G. Rossby, who had previ-
ously studied jets in water. However, Seilkopf (1939) introduced the German term
Strahlströme for upper tropospheric wind maxima, according to Reiter (1963), and Kington
(1999) reports that in the 1870s Clement Ley (in England) assembled cirrus cloud motion
reports to infer very strong upper winds. Early balloon measurements confirmed such
0 inferences, but there was no systematic study of jetstreams until the late 1940s. The
contrast in the horizontal and vertical dimensions of jetstreams – the former exceed the
latter by nearly two orders of magnitude – results in much larger vertical than horizontal
gradients of temperature and wind speed (Reiter, 1963). The horizontal wind shears in a
mid-latitude jetstream are of the order of 104 s1, compared with vertical shears of about
5 103 s1. The horizontal temperature gradient is ~ 6 105 K km1 compared with
a vertical gradient of ~6 K km1. Nevertheless, it is the former which determines the
vertical wind shear, as discussed above.
In winter in each hemisphere the upper troposphere zonally averaged geostrophic wind
maximum is about 40 m s1 and located in the subtropics near 30° latitude at 200 mb
0 (see Figures 3.68 and 4.7) (Trenberth, 1992; Hurrell et al., 1998). However, there are
major differences in higher latitudes (van Loon, 1972c). In the southern hemisphere the
west wind maximum extends upward into the stratosphere in higher latitudes, intensifying
with height to above 10 mb. This has no northern counterpart, although there are weaker
“polar night” stratospheric westerlies. In the summer cases the zonal maxima are weaker
in both hemispheres; they average 20 m s1 at 42°N in July, but are 50 percent stronger
in January at 45°S, where they attain 32 m s1 (Trenberth, 1987). There are also strong
upper tropospheric easterlies in the northern tropics in July.
In summer in the northern hemisphere the average (1979–93) 200 mb zonal wind forms
a virtually continuous spiral jet maximum (Figure 4.8). The jet cores are located just
0 north of the Tibetan Plateau (40°N, 80°–100°E), with a 32 m s1 average, and over the
St Lawrence estuary and Newfoundland (50°N, 60°W) where the maximum averages
34 m s1. The winter jets are 10°–15° nearer the equator and are about twice as strong,
11 although the maximum off East Asia peaks at over 70 m s1 (Figure 4.8).
Figure 4.7 Zonally averaged mean zonal wind (m s1) for JFM (upper) and JAS (lower) from
ECMWF data for 1979–93. Values over 5 m s1 are stippled and those under 5 m s1
are hatched. (From Hurrell et al., 1998)
Although the mean temperature gradient in the troposphere is directed from the equator
to the pole, giving rise to a strong westerly thermal wind component, in summer this
gradient is reversed in the tropical upper troposphere, particularly in the northern hemi-
sphere. Hence the thermal wind relationship (section 4.1) dictates an easterly flow in low
latitudes and as a result of the small magnitude of the Coriolis parameter there is substan-
tial vertical wind shear. The existence of a tropical easterly jetstream at 200–150 mb over
India and northern Africa was first identified in the early 1950s and it was subsequently
analysed by Koteswaram (1958); Figure 4.9 provides an illustration for July 25 1955. The
mean speed in the core over southern India (9–10°N) is about 30 m s1 at 14–15 km and
the direction, but not the speed, shows high constancy. The summer reversal of the thermal
gradient in the upper troposphere is associated with the occurrence of minimum temper-
atures in the equatorial upper troposphere, as well as with diabatic heating of the 500–200
mb layer over the Tibetan Plateau and Himalayan–Karakorum mountains. Strong radia-
tive heating of the surface following the disappearance of the thin and discontinuous
winter snow cover provides a source of sensible heat flux to the atmosphere, and this is
strengthened by latent heat release in cumulonimbus cloud development along the southern
11
Figure 4.8 Mean 200 mb zonal winds (m s1) for JFM (upper) and JAS (lower) from ECMWF
data for 1979–93. Dashed lines show negative values. (From Hurrell et al., 1998)
0
flanks of the mountain ranges and high plateau in midsummer (Flohn, 1968). Energy
budget measurements in the area give a range of estimates for this heating. Table 4.2
provides estimates from station data during 1961–70. There is a large sensible heat flux
of about 220 W m2 over the arid western plateau in June; over the eastern plateau the
sensible flux is about half as large, but in July in this area it is supplemented by a larger
contribution from latent heat of condensation. Subsequent calculations suggest lower
amounts of heat transfer over western Tibet (Ding, 1994, chapter 5). For May 26–July 4
1979 Luo and Yanai (1984) determined a mean daily heating rate of about 3 K day1 for
the 500–200 mb layer over eastern Tibet, comparable with the rate in the monsoon trough
0 over Assam–Bengal to the south. However, Smith and Shi (1996) estimate net warming
in summer of only 3 K day1 in the plateau boundary layer and net cooling in the upper
troposphere through radiative divergence. They find modest heating only over the western
11 part of the plateau.
Figure 4.9 The tropical easterly jetstreams at 200 mb on July 25 1955. Streamlines (solid) and
isotachs (dashed lines). Wind speeds are in knots, easterly negative, westerly positive;
2 kt ≈ 1 m s1. (From Koteswaram, 1958)
Table 4.2 Estimates of net diabatic heating (Q) and Earth–atmosphere system energy budget
(Qsys) over the Tibetan Plateau (daily values, W m2)
F M A M J J A S O N D Year
Q 42 25 60 94 109 101 74 44 10 54 77 21

Qsys 43 21 58 88 100 98 75 49 4 47 74 21
Source: after Yeh et al. (1979).
Over northern Africa a midtropospheric jetstream termed the African Easterly Jetstream
(AEJ) is present at about 650 mb with winds of 12–15 m s1 from the Sudan to the West
African coast near Dakar. It is primarily a response to low-level baroclinicity induced by
the steep surface temperature gradient (about 1°C deg1 latitude) between the equatorial
forests of the Congo and coastal West Africa and the semi-arid Sahel zone to the north.
In addition, this temperature gradient is strengthened by the parallel meridional gradient
in soil moisture (Cook, 1999). Newell and Kidson (1984) found that the AEJ is indeed
stronger (weaker) and moves southward (northward) in dry (wet) years in the Sahel. Model
results suggest that the AEJ is maintained by two diabatically forced meridonial circula-
tions (Thorncroft and Blackburn, 1999). One is associated with surface heat fluxes and
dry convection in the Saharan heat low, which extends up to 700 mb, and the other is
linked with deep convection in the ITCZ. The AEJ is both barotropically and baroclini-
cally unstable, so that easterly waves grow at the expense of the AEJ.
The thermal structure of the atmosphere discussed in section 4.1 does not “explain”
the jetstreams. Their existence is due primarily to the meridional pressure gradient asso-
ciated with that in the thermal field and to transports of angular momentum. An early
attempt to account for jetstream maxima was put forward by Namias and Clapp (1949).
They related jet maxima to wind confluence in a baroclinic zone. The zonally averaged
wind maxima represent a composite of the polar front and subtropical jetstreams that are
apparent on individual vertical cross-sections, especially at particular longitudes. Wind
confluence provides a qualitative explanation for a jet maximum, but confluence itself is
caused by some perturbation in the flow that gives rise to the superimposition of
ageostrophic motions on the geostrophic flow. Figure 4.6 illustrates a jetstream associated
with the polar front and the subtropical jet associated with an upper-level baroclinic zone
located approximately over the subtropical anticyclone belt. These jetstream features are
generally not distinct on mean sections, since the polar front jet is highly variable in
intensity in time and space (it shifts north and south in association with waves in the
11 upper westerlies), whereas the subtropical jets are more stable and persistent. Commonly,
the strongest winds occur when the trough of a mid-latitude wave penetrates into the
subtropics and approaches the crest of a subtropical ridge (Krishnamurti, 1961). In the
southern hemisphere, meridional vertical cross-sections in the Australia–New Zealand
sector show both a subtropical and a mid-latitude wind maximum in January (summer)
that is absent elsewhere and in winter.
The extratropical jetstreams of the winter hemisphere are located poleward of regions
of maximum convective activity (as shown by minima of tropical OLR values) and there-
fore upper tropospheric heating in the summer hemisphere. Thus, for example, in the
0 boreal winter, the extratropical jetstreams over North America, Europe, and Asia are
located northward of the zones of rainfall and of heating maxima over the Amazon, Central
Africa, and Indonesia (Krishnamurti, 1979, p. 325); correspondingly, in the austral winter,
the jets over South America, South Africa, and Australia are southward of heating areas
over North America, North Africa, and the Asian monsoon (Yang and Webster, 1990).
The heating field in the summer hemisphere exerts a strong influence on both the loca-
tion and the intensity of the subtropical and polar front jetstreams in the winter hemisphere.
Yang and Webster show that the annual changes of intensity of the jetstreams are in phase
with the maximum heating. The strength of the heating is also positively correlated with
the interannual variability of jetstream intensity. On annual and interannual time scales,
0 the linkage between the summer hemisphere heating and the winter hemisphere westerly
jetstreams involves the upper-level meridional wind component. The outflow from the
East Asian monsoon towards the Australian jet is stronger, for example, than the reverse
coupling. In addition to these overall associations between the hemispheres, Yang and
Webster (1990) demonstrate relationships between the extratropical jetstreams and the El
Niño Southern Oscillation mode (see section 5.2). The Australian jet undergoes down-
stream strengthening (weakening) in El Niño (La Niña) years, whereas upstream, over
western Australia, the westerlies are weaker (stronger), respectively. During El Niño years
the summer hemisphere heating is shifted eastward. There is also an analogous effect on
the Asian winter jetstream, with downstream strengthening (weakening) during El Niño
0 (La Niña) years. However, in this case the upstream effects are lacking.
Theoretical limits to the latitudinal profile of geostrophic zonal wind are provided by
constant angular momentum on the equatorward side of the velocity distribution and by
constant absolute vorticity (CAV) on the poleward side. The limiting value of constant
angular momentum is determined by the latitudinal change of the term cos2 2/cos2 1,
since u/CE cos where latitude and CE the equatorial speed of the earth’s rota-
tion; thus cos2 is a measure of angular momentum. Air displaced from 30°N to 40°N
would increase its velocity by 99 m s1, for example. The CAV limit on the zonal wind
profile poleward side is determined by
0 u (1 sin )
= cos
CE (1 sin )
(Reiter, 1963).
Figure 4.10 illustrates these theoretical limits and some observed upper tropospheric
profiles of geostrophic zonal wind. For example, air at rest at 20° latitude would, if it
were displaced 10° poleward and conserved its absolute angular momentum, accelerate
eastward with a speed of 34 m s1 (see Table 3.3). Rossby (1947) proposed that lateral
mixing of vorticity provides the redistribution of absolute angular momentum poleward
of the polar front jets, with, theoretically, a sharply defined equatorward border. On the
0 anticyclonic side of a zonal jetstream a limiting condition is imposed by hydrodynamic
stability, i.e. the Coriolis parameter, f, represents a critical value of lateral anticyclonic
shear (
u/
y) for hydrodynamic instability at a given latitude. If the absolute vorticity
11 ( f
u/
y) becomes negative, the zonal flow breaks and small disturbances can amplify
Figure 4.10 Theoretical limits to latitudinal profiles of upper tropospheric geostrophic zonal winds
(see text) and some observed wind profiles. Zonal wind speed is expressed as a rate
of equatorial speed, CE . (After University of Chicago, 1947; from Barry 1967)
rapidly. However, the observed latitudinal wind profile in individual cases may depart
from the theoretical one associated with geostrophic wind components. Rossby used this
argument, expressed in terms of the conservation of absolute vorticity (CAV) (see section
4.3) to account for the observed mean velocity profile on the poleward side of the jet
maximum (Figure 4.10).
Jetstreams characteristically have regional wind maxima both synoptically and in a
climatological sense; these maxima may be quasi-stationary or progress slowly in asso-
ciation with extratropical cyclone systems. The entrance and exit zones of these jet cores
are important in determining the distribution of upper-tropospheric divergence and rela-
tive vorticity. Figure 4.11 shows that, with respect to a straight westerly or easterly
jetstream in the northern hemisphere, there are distinct patterns of convergence/diver-
gence, relative vorticity (), and its time derivative (
/
t). The major terms of the vorticity
equation in general are:
1 d(f )
·VH = 0
(f ) dt
For straight flow, is determined by the lateral wind shear, which is strongest on the
cyclonic side. In the right entrance and left exit regions of the westerly jet maximum
there is divergence, as a result of ageostrophic flow components and associated ascending
air in the troposphere, while in the other two quadrants there is convergence and sinking.
Hence there are also transverse wind components with respect to the jet axis (Figure
4.11b). In the easterly jet the right entrance and left exit are also the quadrants where
upper-tropospheric divergence is located. For corresponding westerly and easterly
jetstreams in the southern hemisphere, divergence and ascent are located in the left entrance
and right exit zones of the jet maxima. The vorticity maximum (minimum) located to the
left (right) of the jet maximum in Figure 4.11c, implying positive (negative) vorticity
11
0
(a)
(c)
(b)
(d)
Figure 4.11 (a, b) The transverse ageostrophic circulations and (c) patterns of divergence/conver-
gence associated with the entrance and exit regions of a straight jetstream maximum.
0 (a, b) Vertical cross-sections along lines A–A′ and B–B′ showing the vertical motion
and direct (indirect) transverse circulations in the jet entrance (exit) regions. (d) Patterns
of relative vorticity and NVA/PVA corresponding to the straight jet streak in (c) (From
Kocin and Uccelini, 1990)
advection in the left exit, right entrance (left entrance, right exit) sectors. Where the
jetstream is associated with a wave trough or ridge, there are additional curvature contri-
butions to the vorticity. For a jet maximum located in a wave trough there is a single
vorticity maximum, giving a two-quadrant model with positive (negative) vorticity advec-
tion everywhere in the exit (entrance) region of the jet core.
0 The maintenance of the climatological jetstreams is explained by accelerations in the
jet entrance that are provided by the time-mean meridional ageostrophic flow. The ther-
mally direct circulation in the entrance region, with ascent on the equatorward side (Figure
4.11b), supplies kinetic energy to the jet. In the jet core and exit regions, however, fluxes
from transient eddies are involved according to Blackmon et al. (1977) and Hoskins et
al. (1983). Blackburn (1985) suggests, from scale analysis, that the meridional plane
vertical circulations inferred by Blackmon et al. (1977) may be invalid. The interpreta-
tion of ageostrophic flow depends on whether the local value of the Coriolis parameter
is used in the geostrophic balance relationship or whether a constant f plane approach is
adopted. Blackburn shows that ageostrophic flow based on the local Coriolis arises in
0 part from rotational flow, where air parcels move up and down isobaric surfaces in closed
circulations, converting potential energy to kinetic energy. The rotational flow gives rise
to the wave retrogression in a non-divergent Rossby mode. He concludes that vertical
11 circulations associated with the climatological jetstreams are mainly forced geostrophi-
cally. They are secondary features maintaining thermal balance with the horizontal advec-
tion by the time-mean flow.
4.3 Planetary waves

4.3.1 General characteristics
The basic zonal circulation (wave number zero) in the mid-latitude troposphere is perturbed
by a spectrum of superimposed waves. Contour maps of mean monthly geopotential height
in the middle troposphere for the northern hemisphere show a well developed wave struc-
ture (Epstein, 1988; Harman, 1991). In January there are three mean waves over eastern
North America, Kamchatka, and eastern Europe, with a prominent high-latitude ridge over
Alaska (Figure 4.1a). The April map suggests four waves and a much weaker circulation.
In July the subtropical highs have moved poleward and the circulation has weakened
further; the principal trough is located over eastern North America (Figure 4.1b). The
October map shows a return toward the winter intensity, but with four waves in high lati-
tudes and five in low latitudes.
An alternative, but less common, representation is a superposed plot of the trough and
ridge axes for individual months (Stark, 1965). The mean locations and outliers are readily
apparent. These maps can be summarized by a longitudinal plot, for a specific latitude,
of the frequencies of mean monthly trough and ridge axes. Figure 4.12 illustrates this for
45°N (Harman, 1991). It confirms some of the features noted above, but the July pattern
shows a complexity not readily apparent in Figure 4.1. These distribution plots identify
the month-to-month variability in the ridge and trough positions by the size of the respec-
tive peaks and the degree of symmetry about the indicated modes.
The wave structure can be separated into its harmonic components by a Fourier analysis
of the heights along a given latitude circle (see Appendix 4.1 for details). The wave
Figure 4.12 Longitudinal plots of the frequencies of mean monthly trough and ridge axes for 45°N
in January, April, July, and October 1946–87. (From Harman, 1991)
11 pattern along a latitude circle is specified by its scale (zonal wave number, m indicates
the number of waves along the latitude circle; m 2R cos /L where R earth radius,
latitude angle and L wavelength), the phase angle, indicating the longitudinal
occurrence, and the amplitude or intensity, in terms of the height departure from the lati-
tude mean value. Wave number m 1 has a ridge and one trough, m 2 has two ridges
180° apart and two troughs, and so on. The nature of these different waves varies consid-
erably over the wavelength spectrum. Wave No. 1 represents an eccentricity of the polar
vortex; the circulation pole in the northern hemisphere is commonly displaced 5°–10°
latitude between longitudes 150° and 160°W (La Seur, 1954). The eccentricity with respect
0 to the geographic north pole increases, according to Barrett (1958), as the amplitude of
wave No. 1 increases.
Analysis of the wave structure for a series of years shows that the longitudinal patterns
of the major wave components (m 1–4) recur regularly in monthly or seasonal aver-
ages. This reflects the fact that there are quasi-stationary planetary waves (Trenberth,
1980) related to forcing by orography and/or land–sea contrasts in diabatic heating. They
have wavelengths above 5,000 km and are dynamically different from traveling synoptic
waves in that the advection of earth’s vorticity greatly exceeds that of geostrophic vorticity
(Bluestein, 1992, p. 62). Wave Nos 5–8 are free planetary waves in the zonal basic current
and waves m 9–12, approximately, are associated with traveling synoptic disturbances.
0 Eady (1950, 1953) first suggested that the waves act continually to regenerate the basic
zonal current. Calculations of kinetic energy transfers between wave numbers show that
the cyclone waves largely maintain both the long waves (m 1–5) and the basic current,
as well as supplying energy to shorter waves (m 11–15) (Saltzman and Fleisher, 1960).
In simplest terms, the stationary planetary waves are represented by the deviations from
zonal symmetry in the time-averaged circulation fields described above. These departures
from zonal symmetry represent two rather different phenomena. One involves stationary
waves, i.e. real waves with zero phase speed, and the other concerns climatological (statis-
tical) wave-like structures that are shown on monthly mean contour maps (Ashe, 1978).
Monthly mean height fields in fact represent a quasi-stationary wave. Rossby et al. (1939)
0 pointed out that the theoretical determination of two or three stationary waves at 60°N
and 3–6 at 30°N was in fair agreement with the observed dimensions of the subpolar
lows and the subtropical high-pressure cells. The very long waves (m 1–3) tend to
dominate in high latitudes, and m 6–7 in low latitudes, but it is important to note that
the actual wavelength of m 7 at 30° latitude is approximately equal to that of m 3
at 60° latitude. The mean amplitude of these three waves and locations of the ridges at
500 mb for average winter conditions, 1964–77, at 50°N are shown in Table 4.3. It will
be noticed that the ridge of m 3 at 60°W is near that for m 2. For January 1976
the component wave numbers are shown in Table 4.4. Wave No. 1 dominates at 75°N
and 30°N, whereas in middle latitudes wave No. 3 is predominant. The circulation in
0 January 1976 was characterized by fast mid-latitude westerly flow and a pronounced three-
wave pattern at 700 mb with troughs located over eastern North America, eastern Europe,
and the western Pacific (Wagner, 1976).
In summer there are two distinct regimes of standing waves (White, 1982). There is a
subtropical regime consisting of lower troposphere oceanic highs overlain by 200 mb mid-
Table 4.3 The amplitude of zonal wave numbers m = 1–3 for winters 1964–77 at 50°N
Wave no. Mean amplitude (m) Longitude of ridges(s)
1 102 ± 17 5°E ± 14°

0 2 93 ± 20 81°W ± 27° (99°E)
3 81 ± 22 60°E ± 42° (180°, 60°W)
11 Source: from Fogarasi and Strome (1978).

Table 4.4 Component wave numbers of northern hemisphere circulation at 500 mb in January
1976
North latitude Zonal wave no. (% contribution)
1 2 3 4 5 6
74° 81.4 10.6 7.3 0.2 0.2 0.2

60° 8.7 2.8 72.6 11.1 3.8 0.6
45° 36.9 8.7 48.8 4.6 0.6 0.2
30° 65.3 2.2 13.4 9.2 2.6 4.3
Source: Fogarasi and Strome (1978).
Figure 4.13 Longitude–height cross-sections of geopotential height for the stationary waves. (a)
60°N, (b) 45°N. (c) 25°N. The contour interval is 50 m. (From Wallace, 1983, after
Lau, 1979)
oceanic trough and continental heat lows overlain by upper-level anticyclones which repre-
sent a monsoonal response to thermal forcing. In higher latitudes there are weaker,
smaller-scale waves than in winter, showing a barotropic structure.
The structure of the stationary waves is most clearly revealed by longitude plots of
geopotential height ( –z ) (Figure 4.13). In the northern hemisphere during winter the high
latitudes show greater wave amplitude near–the tropopause and also a westward tilt with
height. Corresponding mean temperature (T ) sections show a small longitudinal phase
11
0
Figure 4.14 Longitude–height cross-sections of temperature corresponding to Figure 4.13. The
isotherm interval is 2 K. (From Wallace, 1983)
difference from the geopotential section (Figure 4.14). Theoretically, a westward tilt with
height implies vertically propagating waves and upward eddy energy flux, whereas the
presence of maximum
– wave amplitudes in the upper troposphere and minimal difference
between ( –z ) and (T ) imply an equivalent barotropic structure with vertical trapping of
waves (Shutts, 1987). In contrast to the northern hemisphere, the southern hemisphere
0 wave structure is equivalent barotropic except in high latitudes (Karoly, 1985). This
analysis updates earlier studies by van Loon and Jenne (1972) and van Loon et al. (1973).
Southern hemisphere waves have amplitudes about half those of their northern hemi-
sphere counterparts and tend to be of longer wavelength (Trenberth, 1979; LeMarshall et
al., 1985). This is attributed to the predominantly zonal configuration of land and ocean
in middle and high latitudes and the greater mobility of the waves (Adler, 1975; Trenberth,
1981b). Upper tropospheric mean height fields, or mean stream functions, show a domi-
nant wave No. 1 (Trenberth, 1980; Lejenäs, 1984). This has a mean longitude of the ridge
about 140°W and a maximum amplitude throughout the year around 60°S (Quintanar and
Mechoso, 1995). Nevertheless, the zonal wind is enhanced over the South Atlantic–South
0 Indian Oceans and the jet splits near 40°S, 130°E, south of Australia, with a zone of
blocking and weak zonal winds over New Zealand (Figure 4.8b). This structure is closely
linked with the topographic asymmetry of Antarctica, since altitudes exceed 4,000 m in
11 East Antarctica at 82°S, 75°E. Using a non-linear barotropic model, James (1988) shows
that the orographic forcing due to Antarctica is mainly in wave No. 1 south of 40°S. The
asymmetric distribution of sea-surface temperatures and the Antarctic Ocean Convergence
in the Southern Ocean, as well as of the extent of Antarctic sea ice, have been linked
with the pattern of wave No. 1 (Anderssen, 1965; Raynor and Howarth, 1979), but it
seems more likely that they are all forced by the topography of the East Antarctic ice
sheet (Watterson and James, 1992). However, in spite of the contributory role of orog-
raphy, Quintanar and Mechoso (1995) argue the primary forcing of quasi-stationary wave
No. 1 is from low latitudes, especially the Indian Ocean in June and October, based on
Eliassen–Palm flux vectors (see Appendix 4.2). They show poleward propagation of wave
No. 1 from the subtropical Indian Ocean in October. At 50°S the annual mean 500 mb
wave No. 1 accounts for 72 percent of the height variance, with an amplitude of 54 gpm;
m 2 and m 3 both have amplitudes of only 23 gpm and each accounts for 13 percent
of the variance (Trenberth, 1979). Karoly (1985), using data for 1972–82, finds higher
contributions to the variance from m 2 and m 3 at the 300 mb level and 55°S. Wave
No. 1 has an amplitude of about 85 m in winter and summer, m 2 is between 35 m
and 40 m, and m 3 between 40 m and 50 m amplitude.
Wave No. 3 is especially represented on ten-to-fifty day time scales in middle and
higher latitudes (Shiotani, 1990; Kidson, 1991). It is also prominent in the monthly aver-
aged frequencies of cyclonic cloud vortices identified on satellite imagery (Carleton, 1979,
1981; Rogers and van Loon, 1982); year-round troughs are located over the three ocean
basins. A wave train (m 5) set up by the Andes equatorward of 40°S is confined to
subtropical latitudes, and South Africa has a similar effect. However, neither source affects
the circulation in high southern latitudes. The Andes is a high but narrow range, whereas
South Africa has less relief (Trenberth, 1979).
Analysis of daily 500 mb height data at 50°N for ten winters and three summers in
terms of wave number frequency spectra indicates three variance maxima (Böttger and
Fraedrich, 1980). There are ultralong stationary waves (m 1–4) with a greater than
twelve-day period, eastward-propagating waves (m 5–6) with a ten-day period, and
shorter synoptic waves (m 7–8 with periods of four to six days). A similar analysis at
50°S for May–July and November–January 1964–70 finds only two variance maxima
(Fraedrich and Kietzig, 1983). They are associated with wave Nos 4–5, with periods in
the six to twelve-day range, and wave Nos 6–7 traveling eastward with periods shorter
than six days. Figure 4.15 illustrates the time–longitude progression of 500 mb traveling
waves at 50°N for winter 1976–77 on a Hovmöller diagram.
Space–time spectra of 250 mb geopotential heights at 60°E latitude in both hemispheres
show maxima of variance in the thirteen to thirty-two-day range that are westward-prop-
agating and of zonal wave No. 1 (Speth and Madden, 1983; Speth et al., 1992). The
disturbances at 60°N have slightly longer periods (sixteen to twenty days) and are more
prevalent, up to 38 percent of the time in December–February, than their southern hemi-
sphere counterparts (thirteen to seventeen days and 28 percent occurrence in June–August).
In all seasons there is considerable vertical coherence in the waves near 55°–60°S from
the surface up to about 100 mb (Speth et al., 1992) The westward-propagating wave No.
1 episodes are seldom coincident in the two hemispheres. However, it seems that, although
the disturbances are global, they are forced in one hemisphere (Lau, 1979; Held, 1983;
Wallace, 1983).
4.3.2 Properties of waves

It is worth pausing here to review briefly the essential features of atmospheric waves.
The “instantaneous” wavelike motion of the atmospheric circulation is readily illustrated
by a plot superimposing the path of a single contour of the 500 mb height field on succes-
sive days (Figure 4.16). The configuration of a typical sinusoidal wave is described
mathematically by its length (), amplitude (a), and velocity (c). The geographical location
11
Figure 4.15 Hovmöller diagram of 500 mb geopotential during winter 1976/77 at 50°N showing long
0 (short) traveling waves by continuous (dashed) lines. (From Fraedrich and Böttger, 1978)
(longitude) at a particular time, t, is represented by a phase parameter. Following Atkinson

(1981b, p. 104) a one-dimensional wave of constant profile moving eastward (positive
direction along the x axis) with constant velocity, c, can be expressed in terms of the
meridional component of motion (v) as:
v f (x ct)
For pure sine and cosine waves:
0 v a sin b (x ct)
and
11 v a cos b (x ct)
Figure 4.16 The superimposition of single contours of the 500 mb height field on successive days,
June 3–11, 1998, illustrating zonal and blocking patterns. The ensemble mean
(7.0–11.0) is for June 10–14. (Climate Diagnostics Center, CIRES)
where a is the amplitude; b can be expressed as 2/ for a periodic sine or cosine wave.
(The latter is identical, but shifted by /2.) Hence:
2
v = A sin (x ct)

and
2
v = B cos (x ct)

Because the maximum amplitude of the wave need not occur at x 0, t 0, this phase
shift can be incorporated by describing the peak at x0 when t 0:
2
v = A cos (x ct x0) .

These relationships are generally transformed in terms of exponentials involving complex

numbers, i.e.:
11 ei ei ei ei
sin = ; cos =
2i 2i
where i √

 1.
The wave velocity (in the x direction) may also involve a complex form:
c cr ici
The real part cr and the imaginary part ci must both be real numbers. The wave velocity
0 c is real if ci 0; it is imaginary if cr 0 but ci ≠ 0. Where ci 0 the wave ampli-
tudes remain constant with time, so that the wave is neutral or stable. If ci ≠ 0 the wave
can in certain conditions become unstable and amplify. cr measures the wave motion in
the x direction for all cases. For unstable waves ci specifies the rate of growth.
In the atmosphere there are several pure types of wave motion. Sound waves caused by
density variations are longitudinal (compression) waves where the trajectories of air parti-
cles are parallel to the direction of wave propagation. Gravity waves (e.g. mountain lee
waves), which are driven by gravity restoring an external perturbation to the current, involve
particle motion vertically and transversely (back and forward) as the waves propagate hori-
zontally in the x direction (Figure 4.17). The planetary (Rossby) waves are horizontal- trans-
0 verse, so that particles move meridionally (north–south) as the waves propagate horizontally
in the x direction. These large-scale waves conserve absolute vorticity ( f ) by changes in
the vertical component of relative vorticity of air motion () in response to latitudinal varia-
tions in the Coriolis parameter, f (the vertical component of the earth’s vorticity), i.e.:
d
( f ) = 0
dt
Figure 4.17 The instantaneous distribution of velocity, pressure and buoyancy perturbations in an
internal gravity wave, viewed in the x–z plane. The phase of the wave is constant
along all the slanting lines. Velocity and pressure perturbations have extrema along
the solid lines and buoyancy perturbations have extrema along the dashed lines where
velocity and pressure perturbations are zero. Small arrows show the pertubated veloc-
0 ities. Air parcels move parallel to the wave fronts (lines of constant phase) and over
time the wave fronts move perpendicular to the parcel trajectories. The group velocity
(showing direction of energy propagation) is upward, parallel to the air parcel trajec-
11 tories. (From Durran, 1990)
It must be stressed that the dynamics of large-scale waves, which involve motions in
a very shallow atmospheric layer (10 km deep but several thousand kilometers in hori-
zontal extent), are strongly influenced by the earth’s rotation. A useful index of this effect
is provided by the Rossby number:
U
R0 =
2"L
where U horizontal flow velocity, " the earth’s angular velocity (7.3 105 s1), and
L the characteristic length scale (the distance from ridge crest to wave trough). If R0 1
(i.e. if a fluid element travels the distance L in a time less than the period of the earth’s
rotation), so that L/U (1/"), the fluid will be scarcely influenced by the earth’s rotation on
this time scale (Pedlovsky, 1987). Large-scale waves have values of R0 % 1. Note that the
Rossby number as defined above is not appropriate in low latitudes because the formulation
is in terms of the component of absolute vorticity perpendicular to the earth’s surface.
A Rossby wave, assuming horizontal and non-divergent flow (approximated at the
500 mb level), can be represented by:
2
v = A cos (x ct)

It has a wave speed
冢冣
2
c=U
2
where U is the speed of the zonal current on which the wave is superimposed and
f /
y
(the latitudinal variation of the Coriolis parameter). Note that the waves propagate slowly
upstream (westward) with no change of shape; their speed of motion depends on the wave-
length (Platzman, 1968).
The Rossby wave has an angular speed along a latitude circle of:
2"
m(m 1)
where " the earth’s angular velocity.

Another source of changes in wave structure is resonance. This occurs, for example,
if the wave number is close to the barotropic stationary Rossby wave number (Austin,
1980). Its presence in individual zonal wave number plots is shown by 180° phase rever-
sals; also, low-frequency anomalies are of global scale with resonant waves, but are
localized when the waves are dissipative or dispersive (Held, 1983, p. 133). Resonant
forcing of stationary planetary waves by topography is usually interpreted via the concep-
tual model of changes in relative vorticity caused by flow over a mountain barrier,
assuming conservation of potential vorticity and adiabatic motion:
( f )
= constant
p
where p depth of an air column in pressure units. The denominator now implies diver-
gent motion.
Upstream of the mountain, relative vorticity () is zero in the zonal westerly flow
(Figure 4.18). As an air column approaches the barrier it first undergoes vortex stretching
and lateral contraction, giving convergence and cyclonic rotation. As the air rises over
11
Figure 4.18 Schematic illustration of airflow over an extensive mountain range, assuming conser-
0 vation of absolute vorticity. This model ignores upstream influences (see text).The
solid line indicates a trajectory; A, B, and C illustrate the deformation of vortex tubes.
(Buzzi and Tibaldi, 1977)
the barrier and shrinks, causing divergence, the relative vorticity becomes negative and
the air curves anticyclonically southward. Downstream of the mountains, as the air column
resumes its original depth, the air is at a lower latitude (smaller f value) and so the rela-
0 tive vorticity increases and the air moves poleward. However, this model fails for an
easterly airstream approaching a barrier. The negative vorticity and increasing f, as the
air moves up the mountains, would cause the air column to recurve eastward and it would
be unable to cross the mountain range (Holton, 1993). The true explanation of airflow
over topography with a large meridional extent involves the influence of the barrier in
producing upstream lifting and volumetric expansion which produces anticyclonic vorticity
(Smith, 1979a, b). Holton notes, however, that lifting extends only about a radius of defor-
mation (NH/f ) upstream where H a standard atmospheric scale height and N the
Brunt–Väisäla frequency. The latter refers to the natural frequency of vertical oscillations
that are superimposed on large-scale flow patterns:
0
冢g ∂∂z 冣
1/2
N=
where = potential temperature. It is of the order of 102 rad s1 in the troposphere,
corresponding to an oscillation period (2p/N) of ten minutes.
Stationary waves can exist for long wavelengths; the stationary wavelength, assuming
Cartesian coordinates on a flat earth, is:
冢冣
1/2
U
s = 2
0
At latitude 45°, s is about 5,000 km for U 10 m s1, corresponding to m 6 and

11 7,500 km for U 20 m s1, corresponding to m 4 (see Table 4.5a).
Table 4.5a The wavelength (km) of stationary Rossby waves for selected latitudes and mean
zonal wind speeds
Mean zonal wind speed (m s1)
Latitude 5 10 20
60° 4,140 5,850 8,300

45° 3,490 4,940 6,980
30° 3,190 4,460 6,310
Source: after Haurwitz (1941).
Table 4.5b The relationship between zonal wind speed, U (m s1), and stationary zonal wave
number, ms, with latitude
Wave No. (ms )
Latitude Earth circumference (km) 2 3 4 6 8
60° 20,000 29.0 12.9 7.3 3.2 1.8

45° 28,380 90.1 40.1 22.9 10.0 5.6
30° 37,600 150.7 67.0 37.7 16.7 9.4
Where the equations of motion are expressed in spherical instead of Cartesian coordi-
nates, the relationship between wave number (m) and zonal wind speed, U, is:
m(m 1) = 2 1 冢 "R
U sin 冣
where R the earth’s radius, latitude angle, and " the angular velocity of the
earth (Haurwitz, 1941).
At latitude 45° from this equation U 16 m s1 for m 6 and 36 m s1 for m 4;
these values are substantially greater than those in Table 4.5b.
Rossby’s formulation assumes that the zonal current is constant in the vertical and merid-
ionally. The extratropics are, however, baroclinic so that wind speeds increase upward.
Mathematical analysis shows that the stability of a baroclinic wave depends on its
wavelength and the vertical wind shear. Figure 4.19 indicates that waves longer than about
3,500 km are unstable for vertical shear over 1 m s1 per km; such values of shear are
common. The amplitude of unstable waves will double or triple within about two days.
4.3.3 Global wave modes

Atmospheric pressure oscillations were first examined around AD 1800 by P.S. Laplace.
Extending his treatment of ocean tides to those in the atmosphere, he calculated lunar
and solar gravitational forcing for a hypothetical isothermal atmosphere of uniform density
(incompressible) having an “equivalent depth” of about 8 km. However, despite a twofold
greater gravitational potential for lunar forcing, that effect is negligible. Laplace expected
that solar thermal forcing would generate a twenty-four-hour pressure oscillation, whereas
late nineteenty-century studies by Lord Kelvin (1882) and Hann (1889) showed the domi-
nance of a twelve-hour pressure wave with an amplitude of about ±1.5 mb in the tropics.
Also, the twelve-hourly wave was found to occur at a fixed time near the poles, like a
standing oscillation, whereas the low-latitude component traveled with the sun having a
11
0 Figure 4.19 The occurrence of stable and unstable waves in relation to wavelength and vertical
wind shear. (From Atkinson 1981, after Wiin-Nielsen, 1973)
maximum (minimum) near 10.00 (22.00) LST. A global traveling pressure wave was
indeed observed as a result of the eruption of Krakatoa in August 1883; it had a velocity
(c) of 319 m s1 and, assuming c (gH)1/2 for a gravitational wave, the equivalent depth
would be 10.4 km (Asnani, 1993). However, nuclear bomb tests in the 1950s gave a
range of 285–310 m s1 for such global waves. Kelvin proposed that a free oscillation
associated with the solar gravitational tidal potential could be enhanced by a factor of
0 two through resonance in the atmosphere. Haurwitz and Möller (1955) rejected solar tidal
forcing but suggested thermal forcing was enhanced by resonance. The role of resonance
has subsequently been discounted and a mechanism to account for thermal forcing adopted.
Chapman and Lindzen (1970) consider thermal forcings from radiative absorption by water
vapor and ozone, as functions of latitude and height only, for twenty-four-hour and twelve-
hour periods. The diurnal oscillation which is driven mainly by the absorption of solar
radiation by tropospheric water vapor is small because the energy mainly goes into trapped
modes which do not reach the ground (Lindzen, 1967). The model accounts quite well
for the observed semi-diurnal pressure amplitude, but not its phase, nor the polar standing
oscillation. The phase difference is attributed to additional heating with maxima around
0 03.00 and 15.00 LST, associated with the observed semi-diurnal oscillations in precipi-
tation and latent heat release in low latitudes by Lindzen (1978). However, van den Dool
et al. (1997) argue that the moisture variation is an order of magnitude too small to
11 explain the semi-diurnal oscillation. In NCEP six-hour reanalysis fields, they find that m
(a)
(b)
Figure 4.20 (a) Associated Legendre functions (Pnm sin ) versus latitude for P21, P31, and P41. (b)
The corresponding patterns of geopotential height. The amplitude is arbitrary. (From
Madden, 1979)
11
Figure 4.21 Schematic global fields of geopotential and velocity for modes (m, nm) = (1, 1) and
(1, 3) of the Laplace tidal equations. L and H denote low and high-pressure centers,
0 respectively. (From Ahlquist, 1982)
2 accounts for 30–50 percent of the variance of diurnal heights. For surface pressure
it gives an equatorial amplitude of 1.47 mb, compared with 0.67 mb for m 1; these
values are probably overestimated in the reanalysis by comparison with the estimates of
Haurwitz (1965) of 1.16 mb and 0.6 mb, respectively. m 1 is largest at 18.00 and 00.00
hours, m 4 at 06.00 and 18.00 hours.
Large-scale waves in the atmosphere exhibit a variety of spatial structures and frequen-
cies. For the normal modes (Appendix 4.3), or free waves representing resonant states of
the atmosphere, these depend on the earth’s rate of rotation and the scale height of the
0 atmosphere (the equivalent depth of an isothermal atmosphere – approximately 10 km).
In the real atmosphere the zonal wind modifies this basic wave structure (Ahlquist, 1982;
Madden and Speth, 1989; Speth and Madden, 1983). Figure 4.20(a) illustrates the asso-
ciated Legendre functions Pnm versus latitude for P21, P31 and P41 (see Appendix 4.1), while
Figure 4.20(b) shows the corresponding patterns of geopotential height or pressure
(Madden, 1979). The superscript m denotes the zonal (longitudinal) wave number, and n
the meridional index where (nm) is the number of latitudes between the poles at which
the stream function of the wave is zero. Modes where (nm) is even (odd) are antisym-
metric (symmetric) about the equator. The global geopotential and velocity fields for
modes with (m, nm) (1, 1) and (1, 3) are shown in Figure 4.21 (Ahlquist, 1982).
0 The well known planetary waves first described by Rossby (1938) are -plane (
f/
y
0) and/or non-divergent approximations of the global waves. In the absence of zonal
winds (i.e. U 0), these waves propagate westward with a local period of days n
11 (n1)/2m. Haurwitz (1940) demonstrated that for U 0, with a constant longitudinal
Table 4.6 Normal mode characteristics in the troposphere
Meridional wave No. (n m)
Zonal wave No. (m) 0 1 2 3 4 5
1 Period (d) – 4.6 7.5–20 12–30

Amplitude (mb)a – 0.5 0.6 2.0
Percent varianceb – 38.5 24.2 38.0
2 Period – 3.5–5 6–13 10–30 12–30 17–50
Amplitude – 0.3 0.6 1.0 2.0 2.0
Percent variance – 33.8 15.3 28.0 – –
3 Period 2–2.5 4–5.5 6–14 17 approx.
Amplitude – 0.4 0.6 –
Percent variance – 35.0 21.2 25.5
4 Period 3–3.5 5–7.5 6–14
Amplitude – 0.3 ?
Percent variance – 28.2
Sources: from Ahlquist (1982); Venne (1989).
Notes
a
Surface pressure amplitude at the latitude where the wave is a maximum.
b
Percent of variance explained by the normal modes in the filter passband for the 850–200 mb levels.
scale (i.e. m constant), westward propagation slows as the latitude scale decreases (i.e.
(nm) increases). Westward propagation also slows as longitudinal scale decreases
(i.e. m increases) with (nm) constant. Zonal wind effects are small for modes where the
wave numbers are small.
In the atmosphere, planetary waves m 1 and m 2 nearly always move westward,
m 4 waves usually move eastward, while m 3 waves may move in either direction
(Madden, 1979). The periods and amplitudes of the ideal modes calculated by Ahlquist
(1982) are shown in Table 4.6. The periods are shorter (longer) in lower (higher) lati-
tudes. It is apparent that while many normal mode waves have periods of two to thirty
days, they are strongly excited. The amplitudes or surface pressure given in the table are
for ideal modes and may therefore differ from those of the actual free modes in the atmos-
phere. At the 500 mb level, waves m 1–4 have annual mean amplitudes of about
5 gpm for the gravest symmetric meridional mode (nm 1), and 10 gpm and 20 gpm,
respectively, for the next two meridional modes (nm 2 and 3) (Ahlquist, 1985). The
amplitudes vary seasonally within about ±25 percent of the annual mean. Ahlquist finds
significant coherence and correlation between mode (1, 1) at 60°N and 60°S, but not for
mode (1, 3) although it exists in each hemisphere. In another study the forced modes as
represented in monthly mean 500 mb height fields, smoothed to retain only the annual,
semi-annual, the third, and half of the fourth cycle per year, are projected mathematically
on to the free modes (Tribbia and Madden, 1988). The height projections show that the
amplitudes are generally about 10–15 m, typical of transient long Rossby waves. Mode
(1, 0) is large (20–5 m) in boreal winter and spring at latitude 37°, with a phase near
90°E all year; mode (1, 1) is about 20 m in January and April; mode (1, 2) has an ampli-
tude of 40 m in DJF when it is near 45°W and mode (1, 3) has a maximum of 20 m in
November, when the phase is about 60°W; modes (2,0) and (2, 1) have small amplitudes
relative to wave No. 1, whereas the (2, 2) mode is 10–15 m in boreal winter. The largest
wave No. 3 mode (3, 2) is about 15 m in winter, when the phase is near 90°E.
Analyses of global geopotential data for December–February indicate that the west-
ward-moving (1, 1) five-day wave is in phase vertically in mid-latitudes and the tropics
and has its largest signal-to-noise ratio in the tropics. It represents almost 40 percent of
the passband filtered variance (Venne, 1989). The subsequent first symmetric modes (2, 1),
11 (3, 1), and (4, 1) are also in the four to seven-day time range and each account for about
30 percent of the filtered variance in tropospheric heights. The second antisymmetric
modes (1, 2), (2, 2), and (3, 2) which have longer periods are more weakly represented in
the explained variance (Table 4.6). The second symmetric mode (1, 3) sixteen-day wave
has amplitude peaks at about 70°N and 25°N and 50°S and 25°S, with those in the
subtropics nearly 180° out of phase with those in higher middle latitudes (Venne, 1989).
The (2, 3) and (3, 3) modes have a similar frequency and a similar structure; all of these
second symmetric modes contribute substantially to the variance.
The global characteristics of planetary waves are best characterized by the use of spher-
0 ical harmonic analysis (Volland, 1988) (Appendix 4.1). The global wave modes and their
theoretical and observed periods in the atmosphere are shown in Table 4.7. There is good
agreement for the gravest Rossby normal mode 1, 1 and also reasonable similarity for the
next antisymmetric mode about the equator (1, 3). The Rossby waves of largest longitu-
dinal and latitude scales move most rapidly westward (Madden, 1979). Thus wave mode
1,1 travels around the earth in about five days, whereas that with a smaller latitudinal
scale (1, 3) takes up to three weeks (Table 4.7). Zonal waves m 3 may propagate both
eastward and westward and m 4 almost always eastward. Note that mode 1,0 is an
eastward-propagating Kelvin (gravity) wave with a period of about thirty-two hours (Salby,
1984). The equatorially symmetric modes 2,0 and 3,0 have theoretical periods of 1.8 days
0 and 2.1–2.4 days, respectively; the latter corresponds well with observations.
An analysis of 500 mb height data for the International Geophysical Year (1957–58)
provides a useful starting point to characterize the contribution of different ranges of wave
number to the total variance of kinetic energy of the geostrophic wind (Eliassen and
Machenhauer, 1969). Globally, at the 500 mb level, 39 percent of the total variance resides
in the zonal mean state (m 0), zonal wave Nos m 1–4 account for 37 percent and
m 5–12 for 22 percent. There is a large contrast between the hemispheres, as seen in
Table 4.8. Firstly the zonal mean state predominates in the southern hemisphere, with a
similar contribution in both seasons, whereas there is a considerable summer–winter
contrast in the northern hemisphere. Secondly, there is a disparity between the contribu-
0 tion of the lower and higher-frequency waves only in the northern hemisphere winter. In
the northern summer, and in the southern hemisphere in both seasons, the energy in the
two spectral bands is almost equal.
The relative contributions of the stationary and transient waves to eddy kinetic energy
in the northern hemisphere (25°E–75°EN, 925 to 100 mb) has also been analyzed by
Salby (1984). Table 4.9 shows that the transient contribution in the first three zonal wave
numbers dominates those of the stationary waves, especially in summer.
0
Table 4.7 Periods (days) of the planetary wave modes in the presence of westerly zonal winds
in mid-latitudes and easterlies at the equator
Meridional wave No. (n – m)
Zonal wave No. (m) 1 2 3 4
1 5 [5] ~ 10 [8.3–10.6] 7–21 [11.1–20] 28–32

2 4–6 [3.8–4.5] 7–8 14–15; 60 21
3 4 7–8 12–14 15.5
0 4 4–5 8 11–14 12.5
Sources: from Madden (1979); Salby (1984); Williams and Avery (1992).
Note
11 Square brackets indicate theoretical periods.
Table 4.8 The contribution of zonal wave numbers for (n m) = 0 to 7 to the total variance of
the kinetic energy of the 500 mb geostrophic wind (103 m2)
Hemisphere January July

Northern
m=0 469 112
1–4 307 61
5–8 139 72
Southern
m=0 577 665
1–4 131 166
5–8 123 142
Source: from Eliassen and Machenhauer (1969).
Table 4.9 Relative contributions of zonal wave numbers to eddy kinetic energy, normalized to
one, between 25°–75°N and 925–100 mb
Winter Summer
Wave No. Stationary Transient Stationary Transient
1 0.42 0.58 0.58 0.77

2 0.31 0.69 0.21 0.79
3 0.30 0.64 0.15 0.85
Source: from Salby (1984).
4.3.4 Orographic and thermal forcing

There has been a long controversy over the relative roles of diabatic heating and surface
topography in forcing responses by the atmospheric circulation. Barotropic model studies
(Charney and Eliassen, 1949) initially indicated orographic forcing as predominating, while
considerations of the similarity between 1,000–500 mb thickness patterns and 500 mb
height contours pointed to the influence of thermal forcing (Sutcliffe, 1951; Bolin, 1952).
Fundamentally, we should expect orographic effects to be more apparent in the northern
hemisphere, related to the presence of both the extensive Rocky Mountains and the Tibetan
Plateau, relative to the southern hemisphere, where only the Andes are present. Also,
orographic effects should be evident in all seasons, whereas thermal forcing should vary
its phase seasonally as a result of differences in land–sea heating. A model calculation
of thermal forcing by fixed diabatic heat sources and sinks for January and July was
performed by Smagorinsky (1953), showing that both factors contribute to the stationary
waves. An updated version of this analysis is shown in Figure 4.22. We should note that
effects due to orography are not solely mechanical, but involve diabatic heating as a result
of the release of latent heat in cloud systems. Moreover, diabatic heating itself is depen-
dent on the circulation and is not geographically and seasonally fixed.
Topographic effects on airflow that can generate planetary waves include: (1) compres-
sion and expansion of air columns, leading to vortex stretching; this may be balanced by
vorticity advection; (2) adiabatic heating and cooling due to rising and sinking air motions;
this may be balanced by temperature advection; (3) secondary effects such as orograph-
ically induced precipitation releasing latent heat, and momentum transport in smaller-scale
gravity waves. Orographically forced vertical motion (Wb ) is often approximated as:
∂h
Wb = u
∂x
11
Figure 4.22 Thermal forcing and orographic forcing of stationary waves in winter (above) and
0 summer (below). Solid lines show 500 mb heights and dashed lines sea-level pressure
averaged for 40°–50°N. The topography is shown schematically (shaded). The
upward/downward arrows show the integrated diabatic heating and cooling
(surface–500 mb); the triangle size indicates the relative magnitude of the heating.
(From Hoskins et al., 1989)
where u– is the low-level mean zonal wind and h the height of the terrain. However, this
linearized expression neglects the contribution of eddy winds by introducing a prescribed
field of vertically averaged divergence (Ashe, 1978; Dickinson, 1980). A more complete
expression is given by:
0
∂h ∂h
Wb = u v
∂x ∂y
Barotropic theory, for a plane where

f/
y is constant with latitude, implies that west-
erly flow over the upslope of a north–south mountain barrier undergoes shrinking of the
air column and anticyclonic vorticity generation, with corresponding downslope genera-
tion of cyclonic vorticity by stretching (Smith, 1979a, b; Wallace, 1993). However, the
response of westerly flow to an orographic barrier has been shown to be wavelength-
dependent by Hoskins and Karoly (1981). For wavelengths shorter than the stationary
0 zonal wavelength, the dominant term in the barotropic vorticity equation (on a sphere) is
zonal vorticity advection whereby shrinking generates an anticyclone over the mountains
(see Figure 4.23). For long wavelengths, the dominant term sets up a cyclone in the
11 same location. In the upper troposphere the stationary wavelength is ~ 7,000 km, however,
Figure 4.23 Schematic vertical sections illustrating the response of the atmosphere to westerly flow
over a mountain range. (a) and (b) are barotropic atmospheres; clockwise (counter-
clockwise) arrows indicate the generation of anticyclonic (cyclonic) vorticity. H, high
pressure ridge; L, trough. (c) and (d) are baroclinic atmospheres; the circled cross (dot)
indicates poleward (equatorward) flow. W, warmest air; c, coldest air. H and L are the
mean sea-level pressure ridge and trough; the sloping lines show their vertical tilts. k is
the wave number and ks the stationary wave number. (From Hoskins and Karoly, 1981)
so that the first case is more likely. For a baroclinic model there is adiabatic cooling
(warming) on the windward (leeward) slopes. For high mountains (> 2 km), upslope
cooling tends to be balanced by a poleward flow of warmer air, and vice versa on the
downslope. This forcing also generates anticyclonic conditions over the mountains. Hence
the two theoretical models, “fortuitously,” both generate anticyclonic circulations
(Dickinson, 1978).
The solution of the Charney–Eliassen barotropic model on a plane (i.e. no latitudinal
variation of
f/
y), with uniform westerly zonal wind flowing over a schematic topog-
raphy along 45°N, is illustrated in Figure 4.24. The response has been split into the part
due to the topography of the western hemisphere and the part due to that of the eastern
hemisphere. Each wave train shows rapid eastward decay and limited interference with
the other (Held, 1983).
The interacting effects of different major topographic barriers is illustrated by perturba-
tion experiment with a linearized baroclinic model of Hoskins and Karoly (1981). Using
smoothed earth orography and mean zonal flow for the northern hemisphere winter, they
simulated wave trains at 300 mb from the Himalayas and Rocky Mountains, and to a small
extent from Greenland (Figure 4.25). Examining the effects of each mountain area sepa-
rately, it appears that the Himalayas generate almost all of the downstream perturbations
at 120°E, 160°E and 130°W, and about half of a ridge at 75°E. The Rocky Mountains
generate a ridge at 40°W, a trough at 20°E and half of the trough at 75°E. However, the
upstream ridge and downstream trough set up by the Rocky Mountains are modified by
the effects of the Himalayas to produce a ridge at 100°W, 40°N, and a trough at 65°W.
The results of model assessments of orographic forcing must be interpreted with care
in view of the simplifications introduced by the representation of the topography by spher-
ical harmonic functions (Hoskins, 1980) (see Appendix 4.1). For a T42 truncation, for
11
Figure 4.24 Uniform westerly flow of 17 m s1 over schematic topography along 45°N from
the Charney–Eliassen barotropic model with dissipation over five days, showing the
response of the total height (m) (continuous line), which closely resembles the 500 mb
January waves, and the responses due to the topography of the eastern and western hemi-
spheres separately. (From Held, 1983)
0 Figure 4.25 Simulated wave trains at 300 mb, using a linearized baroclinic model, smoothed Earth
orography and winter mean zonal flow. (From Hoskins and Karoly, 1981)
11
example, the Rockies have a maximum altitude of 2,120 m compared with 3,100 m in a
1° grid input grid. The Himalayas–Tibet reach 5,250 m in T42 compared with 6,100 m
in the input data. The usual effect of harmonic representations is to lower the altitude of
a mountain range and increase the width, especially with narrow ranges. The importance
of fine model resolution in the horizontal and vertical domains is stressed by Jacqmin
and Lindzen (1985), who consider that deficiencies in these respects give misleading esti-
mates of the sensitivity of model responses to forcings. Calculations with a high resolution
spectral primitive equation model, linearized about observed zonal wind and temperature
fields for the northern winter, show that the response to topographic forcing in mid-lati-
tudes exceeds that from thermal forcing. Wave Nos 1 and 2 are influenced by the
Himalayas and Rocky Mountains at 30°–45°N and wave No. 3 by the northern Rockies.
In the mid-troposphere, the topographic response is insensitive to changes in the zonal
wind. However, thermal forcing plays a larger role in the interannual variability of the
waves, which amounts to less than 40 gpm in amplitude at 500 mb for combined topo-
graphic and thermal factors.
The vertical structure of planetary waves provides important information on their ability
to transmit energy to high altitudes. The vertical propagation of stationary Rossby waves
depends on the zonal wind exceeding a critical zonal wind speed Uc (Charney and Drazin,
1961; Shutts, 1978). If 0 < U(z) < Uc for all heights, z, then a local heat source gives
rise to waves that tilt upstream with height and show a rapid phase change with respect
to the heat source. If U(z) > Uc , or U(z) < 0 for all z, then vertical wave propagation is
prohibited and there is a sharp 180° phase change in the lower troposphere. For either
case, low-level anticyclones become cold lows aloft, as over eastern Siberia in winter. If
0 < U < Uc for z < zc and U > Uc for z > zc , then planetary waves are trapped below zc
(Shutts, 1987).
The vertical amplitude and phase of the low-frequency waves can be analyzed in terms
of harmonic (Fourier) components (Eliassen and Machenhauer, 1965; van Loon et al.,
1973; Dickinson, 1980; Wallace, 1983). These studies show that zonal wave Nos 1 and
2 tilt westward with height in winter, and the westerly winds in the troposphere
and stratosphere enable energy to be transmitted up to 30–60 km. In July, in contrast,
wave Nos 1 and 2 tilt eastward and wave No. 3 is essentially vertical. The easterly strato-
spheric winds above 20 km cause rapid attenuation of the waves. Linearized quasi-
geostrophic models of large-scale forcing indicate that thermal forcing plays a major role,
relative to the (untrapped) orographic modes (Shutts, 1987b). Thermally generated waves
also make the primary contribution to poleward energy transport (see section 3.7).
However, thermal forcing involves both stationary heat sources and the effects of atmos-
pheric flow over different surfaces in modifying the heating. Shutts (1987b) proposes that
thorough interpretation of the forcings requires non-linear effects to be incorporated where
free-mode structures are taken into account. Non-linear free modes, including isolated free
modes such as the Modon (McWilliams, 1980), are likely to dominate the instantaneous
circulation patterns where thermal and/or vorticity advection is strong. In some sense, the
atmosphere tends to be close to stationary free mode behavior.
The forcing of planetary waves by diabatic heating is not well determined. It involves
inputs of sensible heat, latent heat release through condensation, and solar and infrared
radiative heating. In winter the largest sensible heat fluxes to the atmosphere in the northern
hemisphere are found off the east coasts of Asia and North America, with cooling over
the continents. The vertically integrated sensible heat contributed to each mid-latitude
wave (m 1, 2, 3) is about 10 W m2, corresponding to 20 percent for each wave of
the excess heating over the oceans compared with the land areas (50 W m2), according
to Dickinson (1980). He notes that there is also three to five times more precipitation
over mid-latitude oceans than over the continents in winter. For each of waves 1, 2, and
3 the Fourier amplitudes of precipitation in winter are about 0.8 mm day1, corresponding
to an integrated lower tropospheric heating of 20 W m2.
11
Figure 4.26 Mean heating of the troposphere below 500 mb in the northern hemisphere due to the
asymmetric portions of sensible and latent heat (the mean zonal component removed)
in January and July. The maps are low-pass filtered. Units are 106 K s1 (艑5 W
m2). (From Ashe, 1979)
0
The maximum excess of heating over ocean versus land areas is approximately 100
W m2. Figure 4.26 illustrates the patterns of heating due to the turbulent heat fluxes in
January and July. In July the large longitudinal contrasts in the tropics are associated with
variations in rainfall amount and the land–sea phase has reversed. The Sahara is not a
heating source in either season.
Solar radiation absorbed by the atmosphere is estimated to be of the order of 30
W m2 over land, compared with 50 W m2 over the oceans, giving a difference of only
0 ~20 W m2 (Dickinson, 1980). The net effect of infrared fluxes overall is probably close
to zero. However, the heating/cooling of the lower troposphere from surface infrared emis-
sion may give land–sea differences comparable with those due to sensible and latent
11 heating.
Thermal forcing involves not only heating by the surface primarily from sensible heat
transfer, but also heating in the atmosphere by latent heat release. However, both may be
balanced by horizontal advection or vertical motion. In mid-latitudes, with large baro-
clinic gradients, heating at any level is generally balanced by horizontal temperature
advection (Smagorinsky, 1953; Hoskins and Karoly, 1981). A mid-tropospheric heat
source, for example, is usually balanced by advection of polar air and the trough is then
located to the east of the heat source. Low-level heating may be balanced in part by zonal
advection where the temperature gradient is in the same direction. In this case, also, the
trough is east of the heating. In contrast, Hoskins and Karoly point out that in the tropics
low-level heating is balanced by vertical motion. The vorticity generation for waves of
long wavelength causes stretching that is balanced by poleward motion (the term).
Hence the surface trough is now located west of the heat source. It should also be noted
that boundary layer mixing causes damping of thermal forcings, while at higher levels
damping may result from radiative processes or the dissipative effects of transient eddies.
Experiments using general circulation models with and without mountains confirm the
strong control of the mountain and high plateau of eastern Asia and western North America
on the northern hemisphere wave pattern in winter (Manabe and Terpstra, 1974).
Comparing the circulation in January with present-day mountains, no mountains and an
intermediate “half-mountain” case, using the NCAR Community Climate Model, Kutzbach
et al. (1989) show that mountains cause the planetary waves to increase in amplitude,
with the low-level flow being progressively blocked or diverted around the barriers (Figure
4.27). In July monsoon-like circulations develop near the Tibetan and Colorado plateaus
in response to higher barriers.
4.3.5 Propagation of wave trains

The response of Rossby waves (Platzman, 1968) to topography is influenced by both
meridional dispersion and vertical dispersion. Meridional effects arise from the earth’s
spherical geometry and the latitudinal variation of mean flow. Barotropic flow models on
a sphere suggest that the Tibetan Plateau and Rocky Mountains set up wave trains prop-
agating equatorward into the tropics. Such patterns can be analyzed by ray tracing, which
is based on kinematic wave theory and geometrical optics (Hoskins and Karoly, 1981;
Held, 1983; James, 1994). A ray path indicates the direction of energy propagation away
from a source region. The propagation takes place with a speed equal to the magnitude
of the group velocity. Zonal wave number (m) and wave frequency are conserved along
a ray path, whereas the meridional wave number (k) evolves (Karoly, 1978). According
to Hoskins and Karoly (1981):
k (K s2 m2)1/2
where the total steady wave number:
冢 [u]冣
1/2
M
Ks =
M the meridional gradient of potential vorticity (on a Mercator map), [u] the longi-
tudinally averaged zonal wind.
Ks is analogous to a refractive index in optics; rays bend towards (away from) a
maximum (minimum) of Ks (James, 1988). A source localized in latitude produces two
rays for each zonal wave number (m) less than the stationary value (ms) that propagate
away from the source, while zonal wave numbers m > ms are meridionally trapped near
the source (ms ( /U)1/2). For a source localized in latitude and longitude, there are two
rays for each m < ms that correspond to the two possible meridional wave numbers k
± (ms2 – m2)1/2. For m positive (negative) there is poleward (equatorward) propagation.
11
(a)
0
(b)
(c)
Figure 4.27 The 250 mb circulation in January simulated by the NCAR Community Climate Model
0 for (a) no mountains (NM) (b) “half-mountains” (HM) and (c) present-day mountains
(M). (From Kutzbach et al., 1989)
11
Ray paths are refracted towards larger (ms m2)1/2 and so towards larger ms . Held (1983)
observes that at the latitude where k ms , k → 0, and cy (the meridional group velocity)
→ 0; an incident poleward-propagating wave train is reflected and continues to propa-
gate eastward. At a critical latitude where the mean zonal wind is zero, k → , but cy
→ 0, and rays are refracted into this region. For zonal flows with uniform angular rota-
tion the rays are great circles. For Rossby waves on a sphere, ray paths are curved as a
result of the meridional variation of . Figure 4.28 illustrates calculations by Karoly (1978)
for ray paths of Rossby waves forced at 30°N in a zonal flow of 15 m s1 westerly at
the equator with constant angular velocity (termed “super-rotation”); the ray paths are
great circles between the source and the antipodean point. Note that the speed of energy
propagation is a maximum for the nearly zonal propagation of m 6. For a realistic
jetstream wind profile, Figure 4.29 shows ray paths for a 10°N source in the 300 mb
zonal flow. The two wave number regimes in low and middle high latitudes create
contrasting ray paths for high and low zonal wave numbers, respectively (Hoskins and
Karoly, 1981).
4.4 Zonal index

The state of the westerlies is commonly described by the difference in pressure, or geopo-
tential height, between two latitude circles (usually 35° and 55°) expressed as a geostrophic
wind speed (Vg).
1 ∂p
Vg = ·
f ∂x
Figure 4.28 Rays paths of steady waves forced at 30°N in a zonal westerly flow of 15 m s1 at
the equator with constant angular velocity. (From Karoly 1978)
11
0
Figure 4.29 Ray paths of steady waves forced at 10°N, in 300 mb winter jetstream flow, peaking
at 30°N. The rays are labeled according to wave number. The total wave number
(K = n2 m2) is large equatorward of the jet maximum and small on the poleward
side. Rays propagating equatorward are not shown, as there is a critical line near the
equator. (From Karoly, 1978)
where 1.2 kg m3, f 1.11 104 s1, and

p/
x zonal pressure gradient; 1 mb
102 Pa ( 102 kg m1 s2). The geostrophic westerly wind measured in this way is
an average around the hemisphere or some sector of it (Forsdyke, 1951). This zonal index
0 was first investigated by Rossby et al. (1939). It was recognized that the character of the
large-scale circulation varies in association with the strength of the westerlies. Rossby
and Willett (1948) identified fluctuations in the zonal index with a four to six-week time
scale during which the westerlies intensify and the circumpolar vortex expands, followed
by a decrease in westerly circulation and its breakdown into cellular patterns. High index,
or zonal, circulations typically display elongated subtropical highs, deep subpolar lows
and rapid west–east movement of cyclones, whereas low index circulations have cellular
blocking highs (Willett, 1948). Figure 4.30 illustrates composite maps of surface pressure
for fifteen winter months of each mode and highlights the pressure anomaly patterns.
However, this view of two extreme modes was recognized to be overly simplistic. Riehl
0 et al. (1952) demonstrated four distinct low index states with different synoptic patterns
at the surface and 700 mb.
The westerlies of the northern hemisphere have a strong annual cycle of intensity
(Winston, 1954). A statistical analysis of the hemispheric zonal index (35°–65°N)
of surface pressure since 1899 shows that the annual cycle peaks in October at 7.5 mb
(1.7 m s1) and has its minimum of 2.5 mb (0.6 m s1) in May. The index has an absolute
range for mean monthly values varying from 16.5 mb (3.7 m s1 geostrophic westerly
wind) in January 1978 to 11.5 mb (2.6 m s1) in February 1947 (Kozuchowski, 1993).
The seasonal trend in the strength of the westerlies is often removed from the zonal index.
There are also the effects of the longitudinal asymmetry of the planetary waves. A
0 moderate index may represent a combination of strong zonal flow in one sector and wave
or cellular patterns elsewhere. Indices for limited sectors (Allen et al., 1940) can over-
come this problem, but the movement of individual pressure systems across the boundaries
11 can introduce spurious oscillations in a sector index. The degree of departure from circular
Figure 4.30 Composite of mean monthly surface pressure (mb) for fifteen high-index winter months
(upper) and fifteen low-index winter months (lower) during 1950–76. Pressure depar-
tures (dashed lines) are at 2 mb intervals. (Arai, 1981)
11 zonal flow (waviness) can be described by a meander index (M) representing the ratio of
the circumference of a selected geopotential height contour to the area it encloses (Mörth,
1977; Jones and Mörth, 1978). For each 10° meridian, the intersection of a chosen height
contour is determined both from the pole equatorward and from 25° latitude poleward,
in order to include closed contours or slanting trough/ridge features that are inside or
outside the circumference of the main vortex contour.
M (ai ) (bi )
where ai distance equatorward from the pole along meridian i to the grid point with a
0 smaller height value than the chosen contour, b the corresponding distance poleward from
25° latitude; bi (yn ai) where yn the number of grid points along meridian i exclud-
ing the pole, and standard deviation. The effect of eccentricity of the vortex is incor-
porated by summing the index values for each meridian 180° apart around the hemisphere.
The circumpolar vortex, especially in winter, is often eccentric with respect to the pole, as
reflected in the dominance of wave No. 1; the circulation pole may be displaced 10° lati-
tude from the geographical in the sector 160°–170°W (La Seur, 1954), which also affects
a fixed-latitude hemispheric index. For this reason Riehl et al. (1952) made use of zonal
profiles by plotting geostrophic westerly wind for 5° latitude belts versus time. This is useful
for displaying northward/southward displacements of relative wind maxima.
0 The recurrent fluctuations in the westerlies between high and low index, and subse-
quent recovery, termed an index cycle, were first investigated systematically by Namias
(1950). He showed that such fluctuations of thirty to fifty days’ duration are especially
characteristic of late winter and suggested that they are caused by the accumulation of a
polar reservoir of cold air. This subsides as a series of extensive outbreaks of cold air
occurs, disrupting the zonal westerlies and creating a low index with strong meridional
flow components. The meridional thickness gradient over land areas does reach a peak
in February, according to Miles (1977). The main cold air outbreaks leading to a transi-
tion from high to low index occur over northeast Asia, northwestern North America, and
Greenland–Iceland, according to Defant (1954). The transition of a low to high index
0 pattern features a rise in surface pressure and tropospheric warming over the subtropical
western North Atlantic, in winter especially, and also over the western Pacific. Eventually
a more zonal jetstream develops in higher latitudes and the polar vortex contracts.
Namias’s theory implies a negative correlation between the index and eddy activity,
although this was not found by Wallace and Hsu (1985). The regularity of changes in
hemispheric circulation was analyzed by Kletter (1962) using time–latitude profiles of
850 mb zonal wind for 1955–59. Figure 4.31 illustrates schematically the most frequently
Figure 4.31 The most frequently occurring sequences of 850 mb circulation regimes, 1955–58, in
11 the northern hemisphere. (From Kletter, 1962; Barry and Perry, 1973)
occurring sequences of circulation regime and it is clear that the evolution is consider-
ably more complex than pictured by the simple index cycle. The “characteristic cycle” is
comparatively rare. There have been many suggestions of periodicities with three to four
weeks’ duration in the index (Panofsky and Wolff, 1957; Webster and Keller, 1975), but
Julian (1966) found no evidence of strong covariance in the three to eight-week range
between polar, temperate and subtropical indices at 700 mb. Nevertheless, in the southern
hemisphere, Kidson (1988) reports index cycle components of twelve and a half to fifty
days’ duration at 500 mb that are significant at 60°S in winter. These represent transi-
tions from a single to a double jetstream structure.
During the 1979 Global Weather Experiment the transition from zonal flow to a blocking
mode over the Atlantic sector was associated with an increase in the poleward flux of
heat in mid-latitudes, according to Kidson (1985). Consequently the cycle, which lasts
about twenty-five days, may help to regulate the latitudinal temperature gradient between
50° and 70°N. Earlier, Lorenz (1952) and Mintz and Kao (1952) investigated the predict-
ability of the zonal index and Lorenz found lag correlations between the poleward transport
of relative angular momentum averaged about a latitude circle, [uv], and the mean zonal
wind, indicating a relationship between meridional and zonal motions. Zonal and merid-
ional indices are not significantly correlated on time scales of days to weeks, although
Wahl (1972) notes a strong positive correlation on an annual scale, when the meridional
index lags the zonal ones by twenty-four days.
Laboratory experiments with a rotating annulus (Fultz and Kaylor, 1959; Hide, 1970)
have shown that rotating fluids commonly exhibit regular periodic time variations in the
flow pattern. This behavior, referred to as vacillation, resembles index cycles of westerly
circulation in the atmosphere. Vacillation is defined by Hide as comprising pulsations in
amplitude which may include changes in wave number, the progression of a large distor-
tion around the wave pattern, or a wavering of the shape of the pattern. Because they are
spatially and temporally periodic, these flows may be predictable. However, in different
experiments, the same external conditions can give rise to different regular flows,
suggesting that regular flows are intransitive (Hide, 1985) (see Figure 4.46). The criteria
necessary for steady or vacillating waves are still unknown, according to Hide.
The question arises as to whether the circulation in middle latitudes exhibits “random”
fluctuations or whether there are preferred modes. The first alternative involves “transient
variability of linear wavelike perturbations about a climatological mean basic state”
whereas the latter may be represented by “discrete transitions of the polar vortex,” i.e.
an index cycle, according to Wallace et al. (1991). These alternatives were already recog-
nized by Berggren et al. (1949), but a definitive resolution of the problem is still awaited.
The question is important in terms of the mechanisms involved and because it could be
of value in extended-range forecasting.
Using a simple numerical geostrophic model, Lorenz (1962) simulated index cycles
that had a chaotic, rather than a periodic, nature. That is, the time series contain a degree
of regularity (Lorenz, 1986). This seems to accord with the suggestion of Wallace and
Hsu (1985) that index fluctuations are perhaps a “statistical residue of dynamically unre-
lated events at different longitudes.” For 1899 to 1990, monthly index values show spectral
peaks of two to three, six and twelve months, as well as a quasi-biennial (~ 2.2 yr) and
approximately thirteen to fifteen-year periodicity, according to Kozuchowski (1993),
although the last two are not stable. Over the period of record the index has been higher
than average between 1899 and 1938, and from 1972 to 1990, and lower during 1939 to
1971. Figure 4.32 illustrates the fluctuations of the 500 mb index (55°–35°N) for winters
1947–94. It is apparent that sequences of easterly anomalies show a tendency to recur
(1948–52, 1972–76, 1989–94) and there is a statistically significant one-year lag correla-
tion of 0.34. There are also abrupt interannual reversals in 1976 and 1988. The reason
for interannual persistence in zonal wind could be related to persistence in anomalies of
sea surface temperature in the North Pacific (Ting et al., 1996).
11
0 Figure 4.32 The 500 mb zonal geostrophic wind index (55°–35°N) for winter 1947–94 (m s1),
from NCEP analyses. The indicated year refers to January of each winter season.
(From Ting et al., 1996)
Robinson (1991), using a global primitive equation model, also found high and low index
states. Forcing of momentum fluxes by synoptic eddies is shown to be strongly correlated
with the index and leads it slightly, i.e. the synoptic eddies sustain the index against dissi-
pation. In contrast, forcing by low-frequency eddies is less well correlated with the index
and these eddies tend to erode the index. Robinson shows that blocking is not a critical fea-
0 ture. Blocks may occur with extreme high and low indices, which is contrary to the earlier
view of Namias (1950). Robinson’s model does not reproduce a split jet, as reported by
Kidson (1988) for the southern hemisphere. Further examination of the role of eddy forc-
ing through wave–zonal mean flow interaction1 has been based on the NCEP reanalysis
(with R30 truncation) for 1979–95 (Feldstein and Lee, 1998). They find that although unfil-
tered eddy forcing lacks significant feedback, high-frequency eddies (less than ten days) do
prolong anomalies of the zonal index against dissipation by low-frequency and “cross-fre-
quency” (high low frequency) eddy forcing and surface drag.
The spatial structure of the zonal wind relative to the mean zonal index at 500 mb is
illustrated in Figure 4.33. The regression is for geostrophic zonal wind (u– ) and westerly
0 (positive) zonal index. There is considerable coherence in u– values across the subtropics
with the mean westerly component at 35°N, whereas large negative regression coefficients
are found around 60°N, associated with the winter jetstreams over eastern North America,
the North Atlantic, and the Aleutian Islands (Ting et al., 1996). Thus the anomalies of
zonal wind at 35° and 55°N are out of phase. Interestingly, although the Atlantic and
Pacific climatological jets apparently fluctuate “in phase,” the winter-mean fluctuations of
the two jets for winters 1967–94 are uncorrelated.
The relationships between regional climatic anomalies and zonal indices have been the
subject of many investigations, but recent studies can draw on lengthy records, enabling
identification of multi-decadal fluctuations in circulation characteristics. Sector indices for
0 the eastern North Atlantic and northern Europe, for example, show few episodes of per-
sistently low or high index during the period 1899 to 1992 (Jonsson and Börring, 1994).
There was an interval with low index during winters 1955–70 over the North Atlantic
11 (45°–65°N, 5°–40°W), followed by stronger westerly circulation after 1970 in winter and
Figure 4.33 Regression of the December–February geostrophic zonal wind against the zonal wind
index. Anomaly amplitudes are obtained by multiplying the regression coefficients by
the standard deviation of 2.7 m s1. The contour interval is 1 m s1; negative contours
are dashed. The right-hand panel shows the corresponding zonal mean (m s1). (Ting
et al., 1996)
spring. The zonal indices for these two sectors are shown to be positively (negatively) cor-
related with winter (summer) temperatures in southern Scandinavia. Increases in zonal
index generally correspond with higher temperatures over Europe, except in Scandinavia
in summer. The strongest correlations between zonal index and temperature are found over
the British Isles in winter and over northern Germany, Denmark, and Poland in spring. In
the southern hemisphere, several circulation indices have been calculated using station data.
A Trans-polar Index (TPI) based on monthly pressure values at Hobart, Tasmania, and
Stanley, Falkland Islands, was tabulated for 1931–60 by Pittock (1980) and has been
extended to cover seasonal values for 1895–1997 by Jones et al. (1999). It is a measure of
wave No. 1, or the eccentricity of the polar tropospheric vortex. Carleton (1989) shows
links between the TPI and Antarctic sea ice extent. During austral summer there are neg-
ative correlations (0.3 to 0.5) between the TPI and temperatures in southern South
America (Jones et al., 1999). Pittock also calculates an index of the tropical South Atlantic
zonal circulation for 1941–60 using sea-level pressure departures from normal at Ascension
Island and St Helena minus those at two stations in coastal Brazil. Zonal and meridional
indices for the New Zealand sector were introduced by Trenberth (1976), based on pres-
sure differences between Auckland and Dunedin (ZNZ) and Hobart and Chatham Island
(MNZ). Seasonal values for 1853–1997 (ZNZ) and for 1878–1997 (MNZ) are presented
by Jones et al., as well as corresponding zonal and meridional indices for 1903–97 (ZSAAP)
and 1895–1997 (MSAAP) over southern South America and the Antarctic Peninsula, based
on Stanley–Orcadas and Punta Arenas–Stanley, respectively. They show that surface
temperatures in southern South America and New Zealand depend primarily on the local
meridional circulation, with correlations of 0.4 to 0.7 for MNZ. However, temperature
trends in New Zealand are unrelated to meridional flow intensity or its frequency. It appears
that the temperature rise since the mid-1970s is caused by warming of the Southern Ocean,
rather than by decadal variations in atmospheric circulation.
4.5 Zonal and blocking flow modes

4.5.1 Synoptic characteristics
The mid-latitude westerlies of both hemispheres exhibit a wide variety of flow behavior.
Over sectors spanning about 90° of longitude there can be strongly zonal (west–east) flow,
well developed waves, or meridional (north–south) flows associated with a ridge of high
pressure. Interruption of the westerly circulation is termed blocking. There are numerous
11 definitions of blocking, and many statistical analyses have been made of its temporal and
spatial characteristics (Rex, 1950; Geb, 1966; Dole and Gordon, 1983; Wallace and
Blackmon, 1983; Knox and Hay, 1985; Dole, 1996).
Synoptic descriptions of blocking identify a stationary warm anticyclone, persistent for
five days or more, in the westerly wind belt. The blocking high may develop from a ridge
extending poleward from the subtropical anticyclone, or it may form in high latitudes, as
over Scandinavia, with limited effect on the zonal index, since the westerlies are enhanced
upstream (Berggren et al., 1949; Rex, 1950). The blocking high or ridge has an equiva-
lent barotropic structure, forming a closed anticyclone at lower levels and a ridge in the
0 upper troposphere. In the “dipole pattern” (Rex blocking), a “cut-off” cold low usually
develops about 30°– 40° latitude equatorward of a high-latitude anticyclone. The jetstream
splits into a northern branch poleward of the high, and a branch equatorward of the cut-
off low with weak winds in between (Figure 4.34). Diffluence in the westerlies upstream
of the block is matched by confluence downstream. Disturbances in the mid-latitude west-
erlies are steered around the block, often on the poleward side. Another blocking pattern
has a high-latitude anticyclone flanked at lower latitudes by cyclones to east and west.
This pattern is termed an omega block from its resemblance to the Greek letter ".
The identification of blocking patterns has been performed by manual analysis of
synoptic charts at the surface and upper levels (Elliott and Smith, 1949; Rex, 1950; Geb,
0 1966; Treidl et al., 1981) and by the use of objective criteria (Dole, 1978; Illari et al.,
1981; Lejenäs and Økland, 1983; Shukla and Mo, 1983; Knox and Hay, 1985; Schilling,
1986; Lupo and Smith, 1995a). For example, Dole (1978) identifies blocking as a persis-
tent, large positive anomaly of 500 mb height. Illari et al. (1981) examine the weekly
variation of a selected 500 mb height contour between 60°W and 30°E (a meandering
index). Lejenäs and Økland arbitrarily define blocking as occurring locally and instanta-
neously when the 500 mb height difference between 40°N and 60°N is negative and a
high-pressure cell is poleward of a cut-off low. Tibaldi and Molteni (1990) modify this
criterion in order to exclude cases of cut-off lows that are anomalously displaced pole-
ward. Shukla and Mo recognize blocking as an anomaly, of 500 mb geopotential height
0 ≥200 gpm in winter or ≥100 gpm in summer, that persists at a grid point for seven days
or less. Knox and Hay (1985) examined a selected sample of 1,200 positive anomalies
of five-day mean 500 mb heights ( k) stratified seasonally. A blocking signature is defined
according to the magnitude of the anomaly, weighted according to latitude. They identi-
fied a blocking signature as occurring when “the absolute value of the distance between
k [the height anomaly] and k1 is less than 1905 km”; this corresponds to a threshold
speed of 4.4 m s1 at 60° latitude.
Having examined the nature of zonal and blocking flows, we consider next their
geographical occurrence. Knox and Hay (1985) analyzed anomaly centers at 500 mb over
the northern hemisphere for 1946–78, using the criteria described above. The sectors with
0 maximum frequency of blocking signatures are:
1 10°W in the North Atlantic.

2 75°W over northeastern Canada.
3 60°E near the Ural mountains of Eurasia.
4 A broad sector from 140°W to 180°W over Alaska and the northeastern Pacific Ocean.
These results generally match earlier findings. Rex (1950) identified the major regions in
the northeast Atlantic and eastern North Pacific sectors, using analyses for 1932 to 1950,
and these were confirmed by Treidl et al. (1981) for 1945–77. These two sectors coin-
0 cide with regions where the annual temperature range is minimal. Johansen (1958) points
out that the mean temperature of the 1,000–700 mb layer varies by only 6°–10°C. There
are larger variations in the 700–500 mb layer, implying that oceanic effects damp the
11 temperature range in the lower troposphere, strengthening anticyclonic tendencies. Dole
Figure 4.34 Schematic illustrations of 500 mb contours for (a) February 15 1983, the dipole (Rex),
and (b) the omega patterns of blocking. (From Dole, 1996)
11 and Gordon (1983) found persistent positive and negative anomalies of 500 mb heights
in winter corresponding to sectors 1, 3, and 4 above. Knox and Hay show that the frequen-
cies are highest in spring, especially over eastern Canada and the North Atlantic, and
lowest in autumn, except at 20°W and 60°–70°E. They also identify a high Arctic
maximum, clockwise from 90°W to 40°E, which is prominent in spring and summer, and
is associated with warm anticyclones that migrate poleward. The blocking maximum over
northeastern Canada that peaks in spring is surprising in light of the mean upper air trough
at 70°W. It is explained by variations in the intensity and location of the mean trough.
Tibaldi et al. (1994) find, from 500 mb ECMWF analyses for December 1980–November
0 1987, that European blocking centered around 10°–20°E has a pronounced spring
maximum, whereas the eastern North Pacific maximum is in winter, with an autumn
minimum. Their definition is based on a modified version of that by Lejenäs and Økland
(1983) with a requirement of five consecutive blocked days in a given sector.
Results based on Tarleton’s (1986, 1987) definition for 500 mb height anomalies are
shown in Figure 4.35; blocking is common in three preferred sectors in the northern hemi-
sphere: the eastern North Pacific–western North America, the northeastern North
Atlantic–western Europe, and northern Eurasia. However, only the first two locations are
identified using absolute heights. Anomaly-based results emphasize the oceanic situations
and understate orographically related blocking areas such as western north America
0 and western Europe. In the southern hemisphere both definitions show the New Zealand
sector as being the primary location for blocking. The northern hemisphere pattern also
shows seasonal differences between winter–spring on the one hand and summer–
autumn on the other (Tibaldi et al., 1994). Cold-season blocking anticyclones are more
intense, persistent, and larger than their summer counterparts, according to Lupo and
Smith (1994). Other definitions produce similar, but not identical, results. For example,
Dole (1983) finds positive and negative height anomalies co-located over ocean areas.
Precise reasons for such definition-related differences are not yet clear.
For the southern hemisphere, several studies find that the New Zealand sector (160°–
180°E) is a primary area of blocking, often with a split jet (van Loon, 1956; Lejenäs,
0 1984; Trenberth and Mo, 1985). Other areas are in the South Atlantic east of South
America (50°–70°W), and in the southern Indian Ocean east of South Africa (40°E). The
overall frequency is about half that in the northern hemisphere, and 55 percent of the
cases identified by Lejenäs (1984) lasted only one day. However, the method used is
apparently sensitive to the accuracy of the southern hemisphere charts. Comparison of
this procedure with the use of positive height anomalies for 500 mb fields during December
1978–November 1979 suggests that more frequent blocking ridges, without a dipole cut-
off low, occur near South Africa and South America (Lejenäs, 1987). In the analysis of
Tibaldi et al. (1994) the number of blocking days in the Australian sector is comparable
in the austral summer and autumn with that in the European sector in the corresponding
0 boreal seasons.
Blocking occurrences in the southern hemisphere differ in several ways from those in
the northern hemisphere. Coughlan (1983) notes that there are three main preferred regions
in the southern hemisphere, rather than two, as found by Rex (1950) and Treidl et al.
(1981) for the northern hemisphere, and the mean latitude is about 45°S, compared with
56°N. However, the majority of events, and particularly those lasting a week or more, all
occur in the southwest Pacific (Lejenäs, 1984). There is a late winter–early spring
maximum in both hemispheres, but in the southern hemisphere there is weaker season-
ality and also a secondary autumn maximum in the Indian Ocean region. The three
dominant blocking regions in the southern hemisphere are positioned just downstream of
0 the land masses, but they do not appear to be a result of orographic forcing. Coughlan
notes that they correspond instead with sectors where the sea surface is warmest and a
west–east positive gradient of temperature is observed upstream of the warm anomalies
11 (with respect to a zonal average).
Figure 4.35 Frequency of blocking in each hemisphere based on 500 mb height anomalies. (From
Tarleton, 1987)
11 The blocking indices so far discussed incorporate, directly or indirectly, some measure of
the zonal mean circulation. For this reason they are unable to address the question as to
whether blocking frequency is influenced by zonally symmetric flow. Accordingly, Kaas and
Branstator (1993) devised an index for this purpose which is based only on 500 mb eddy
fields; they use opposite anomalies of the meridional wind component (>10 m s1 or
<10 m s1) at locations within ±1,000 km of each grid point (Pi ) selected along a latitude
circle (positive anomalies west of Pi, negative anomalies to the east). Coherent space–time
anomaly clusters are identified via a Euclidean distance metric. The index delimits the famil-
iar maxima of blocking (10 percent of days for winter months) centered over the west coast
0 of North America and from northwest of Ireland to northern Scotland. In both regions, strong
winter blocks are accompanied by weak zonal winds at 50°–60°N and strong ones near 30°N.
They suggest that localized blocking is more likely when the mean zonal wind is in an anom-
alous state, resembling the structure of the leading EOF of zonally averaged winds.
4.5.2 The climatology of zonal and blocking flow patterns

The locations of the centers of action and the intermediate tracks of traveling cyclones
differ considerably between high index (zonal) and low index circulations. Figure 4.36
illustrates that, with high index conditions in winter, cyclones are more frequent in the
0 central eastern North Atlantic and northern North Pacific, whereas with low index they
are more frequent in the Mediterranean and mid-latitudes of the North Pacific.
Anticyclones tend to shift from ocean to mid-continent areas, and from middle to higher
latitudes, as the zonal index goes from high to low values (Bradbury, 1958). Corresponding
differences in cyclonic and anticyclonic activity are less pronounced in summer, when
the overall flow in the northern hemisphere is weaker.
The best indication of the contrast between zonal and blocking patterns is provided by
studies of monthly or seasonal climatic anomalies. Typically, in western Europe, zonal
circulation gives mild, stormy winters and rather cool, changeable or wet summers.
Blocking conditions, in contrast, typically give rise to warm, sunny summer weather, as
0 shown in Figure 4.37 for July 1976. A persistent ridge from the Azores anticyclone gave
record high temperatures in southern England (Green, 1977). Depending on the location
of the anticyclone, northerly and easterly airflows give cold winters in western Europe,
often with recurrent snowfalls and long-lasting snow cover.
An extreme type of low index regime occurs occasionally in winter in the North Atlantic
sector. It involves a reversal of the normal pressure pattern, with an anticyclone over
Iceland and low pressure over the Azores. Based on monthly sea-level pressures at
Stykkisholmur and Ponta Delgado between 1867 and 1980, there is a 6 percent frequency
of this “reversal pattern,” with two thirds of these cases occurring in the cold season,
October through March (Moses et al., 1987). For positive and negative pressure depar-
0 tures of at least one standard deviation from normal, there were forty cases (3 percent of
the total). This pattern may be part of a hemispheric anomaly, with subpolar high pres-
sure and a deep low-pressure belt across the mid-latitude North Pacific and Atlantic. Shutts
(1987a) refers to this as a “severe winter pattern” (for western Europe). The winters of
1947–48 and 1962–63 were of this type (Figure 4.38).
4.6 Blocking mechanisms

Theories on the origins of blocking can be distinguished according to those concerned
with the initial development and those that relate to the maintenance of an established
0 block. Blocking onset has been attributed to:
1 The non-linear interaction of a traveling wave and a topographically forced wave

11 (Egger, 1978).
Figure 4.36 Distribution of anticyclones and cyclones during high/low index patterns. (From
Bradbury, 1953)
2 Rossby wave dispersion (Hoskins et al., 1977; Webster, 1982).

3 The amplification of planetary waves through resonance effects of topography or
heating (Tung and Lindzen, 1979).
4 Constructive interference between stationary planetary waves (Austin, 1980), espe-
cially downstream of orographic barriers.
5 Instability of a three-dimensionally varying basic flow (Frederiksen, 1982).
6 Transport of eddy vorticity in a region of jetstream splitting (Shutts, 1983; Neilley
and Dole, 1991).
11
Figure 4.37 Summer weather conditions over western Europe associated with blocking. Mean
500 mb contour chart (16 dam spacing) for July 1976, showing jetstream locations
and above-normal rainfall areas (shaded). (Green, 1977)
0
Figure 4.38 Sea-level measure anomalies associated with the North Atlantic pressure reversal of
January 1963 which gave rise to severe winter conditions in Great Britain. (Moses
11 et al., 1987)
In diagnostic terms, the onset of blocking coincides with a minimum of eddy kinetic
energy and a relatively low value of mean kinetic energy (Smith, 1973). Comparison of
blocking and transient ridges over the North Atlantic and North Pacific in winter indi-
cates that blocks show a zonal scale of 60° longitude in the vorticity pattern; there is a
40 percent reduction in the local 500 mb zonal wind and a 20 percent decrease in the
zonal mean, according to Hartmann and Ghan (1980). The major difference between
blocks and transient ridges over the Pacific is a reduced eastward advection of vorticity
in the blocking ridges; these results point to barotropic mechanisms. In contrast, in the
Atlantic sector, blocking ridges are more baroclinic than transient ones and indicate
the need for a large conversion of potential to kinetic energy.
The amplification of planetary waves that occurs during the initiation of blocking implies
an increase in wave energy. Tanaka (1991) notes that in principle this may result from
either a downscale cascade of energy from the zonal flow, or an upscale energy transfer
from synoptic or shorter-scale waves. The index cycle of Namias (1950) illustrates energy
transfer downscale from the zonal flow into large-scale waves, although Tanaka points
out that the growth rates in themselves are small. Instabilities of the zonal mean state
may involve baroclinic instability whereby zonally available potential energy is trans-
ferred into planetary wave energy. Also, barotropic instability in a zonal jet transfers zonal
kinetic energy into planetary waves (Schilling, 1986). The contrary upscale transfer of
energy may occur owing to non-linear wave–wave interactions with barotropic energy
conversions. Such transfers represent a major energy source for planetary wave amplifi-
cation (Hansen, 1988; Kung et al., 1989). The zonal and wave modes of circulation
identified by Sutera (1986) (see p. 325), have been examined in terms of the energy
sources involved in the transitions from one mode to the other by Hansen (1988). Based
on thirteen winter cases in four years, the low-amplitude (zonal) circulation (mode 1) and
the large-amplitude waves, m 2–4 (mode 2) each persist ten to eleven days and the
transitions between them last typically two to four days. The main source of the approx-
imately 50 percent increase (1.7 105 J m2) in both kinetic energy and available
potential energy during the mode 1 → 2 transition involves non-linear barotropic inter-
actions between intermediate scale waves. The energy calculations are for 20°–80°N,
1,000–10 mb. During the mode 2 → 1 transition, kinetic energy is dissipated primarily
by upscale transfer from waves 2–4 to wave No. 1.
The apparently conflicting results of numerous studies suggest that different energy
sources may set up blocking modes having a similar structure and behaving in a similar
manner (Tanaka and Kung, 1989). Thus a low-frequency free eigen mode can be excited
by various energy sources. That the eigen vector has a dipolar nature implies that positive
and negative anomaly patterns should have similar structures, as confirmed by Dole (1986).
The theory of Rossby wave dispersion, generalized to a sphere, proposes that a train
of waves initiated by an anomaly source in low latitudes propagates into middle latitudes
of the winter hemisphere. This is because the upper westerlies are closest to the equator
there. However, Frederiksen and Webster (1988) point out that the theory implies that
wave trains will shift longitudinally in response to displacements of tropical heating anom-
alies. This is not in accord with observations or modeling results. Moreover, the theory
assumes a zonally averaged basic state, whereas Arkin and Webster (1985) show impor-
tant variations with longitude in the time-averaged zonal flow and in the transient eddies
relative to that basic state. The observed vertical structures of mature anomalies are also
found to be different from those generated by local diabatic heating anomalies (Dole,
1986). Dole remarks additionally that the anomalous circulations “grow and decay while
the external forcing remains nearly fixed.”
The idea that blocking occurs through simple interference between stationary planetary
waves has been explored by Austin (1980). She concluded from observational data and
a simple model of interaction between planetary and baroclinic waves that blocking anti-
cyclones are initiated by constructive interference between stationary planetary waves
11
0
Figure 4.39 Constructive interference between stationary planetary waves, as shown by the normal
phases of the waves at 500 mb, compared with the longitudes of initial jet splitting.
(From Austin, 1980, based on Rex, 1950, and Eliassen, 1958)
(m 1–3) with normal phase but very large amplitudes (Figure 4.39). As discussed in
section 4.3, stationary planetary waves are forced by land–sea thermal contrasts and large-
scale orography. Wave amplification may occur as a result of surface anomalies (sea
surface temperatures, snow cover) or through a change in propagation conditions in the
stratosphere. Austin proposes that blocking in the North Atlantic sector involves inter-
0 ference between large-amplitude wave Nos 1 and 2 at 500 mb, whereas Pacific blocking
requires large amplitude wave Nos 2 and 3, with wave No. 1 small. She notes that in
summer the 500 mb mean waves are displaced 15° eastward of the January position,
causing more continental blocking, in line with observations. Contrary to these results,
Quiroz (1987) found that traveling, mainly retrogressive, waves were significant in twenty
out of twenty-four blocking cases. The resonance theory, elaborated by Tung and Lindzen
(1979), indicates that amplifying waves are resonant with topographic forcing and differ-
ential land–sea heating. If the energy is constrained horizontally and vertically, as in a
wave guide, the wave can become resonant. They show that the necessary conditions are
readily satisfied for m 4 and much less so for m 1 or 2.
0 The role of synoptic/planetary-scale interactions during blocking onset involves four
key elements, according to Tsou and Smith (1990). They are: (1) a quasi-stationary plan-
etary-scale ridge at 500 mb; (2) a developing precursor cyclone at the surface located
40°–60° longitude upstream of the block, corresponding to about half a planetary wave-
length; (3) an associated amplifying short-wave ridge at 500 mb; (4) a strong jet maximum
on the upstream flank of the developing short-wave ridge. Interestingly, a rare occurrence
of continental blocking over Saskatchewan, Canada, in April 1980, when a ridge over
western North America amplified into a blocking anticyclone, also contained these key
features (Lupo and Bosart, 1999). In this event, the time-mean thermal wind provided
lower tropospheric thermal advection and anticyclonic vorticity advection, leading to subsi-
0 dence which created abnormal warmth, but synoptic-scale cyclones played an important
role in forming and maintaining the block.
The instability theory advanced by Frederiksen (1982), Frederiksen and Webster (1988),
11 and Frederiksen and Bell (1990) takes account of the three-dimensional instability of a
three-dimensionally varying basic flow. Both climatological mean and instantaneous syn-
optic flows have been studied. Using a five-level, quasi-geostrophic model, with spherical
geometry, applied to the development of a blocking event in January 1979, Frederiksen and
Bell (1990) examine the growth rates and structures of the fastest-growing modes.2 Large-
scale barotropic dipole or multipole instability modes are found to be quasi-stationary and
undergo rapid growth. Blocking onset in the central North Atlantic occurred, with rapid
cyclogenesis off eastern North America and large-scale wave-train modes propagating east-
ward. This stage involves both barotropic and baroclinic instability, whereas mature block-
ing is characterized by barotropic instability. The breakdown of the westerlies through the
growth of localized disturbances is dependent on the wavelength of the Rossby wave trav-
eling in the westerly flow. For a short wave in strong zonal flow the disturbance moves
downstream, whereas a large-amplitude wave in weak zonal flow can grow in situ over a
few days (Illari et al., 1981). Illari et al. (1981) refer to fast-moving unstable disturbances
that increase in space rather than time as displaying “convective instability” and ones that
grow in time as having “absolute instability.” Topographic instability, proposed by Charney
and DeVore (1979) and Tung and Lindzen (1979), appears to play a small role in any of
these stages. The fastest-growing monopole cyclogenesis-instability modes are initially
focused in the climatological storm tracks over the North Pacific, North Atlantic, and
Siberia, but subsequently the cyclogenesis modes in the Atlantic split into wave trains to
north and south of the block.
Given the initiation of blocking, the next consideration is: how is it maintained?
Frederiksen and Webster (1988) suggest two possible mechanisms. One involves the
concept of multiple flow equilibria (Charney and DeVore, 1979) although, as has been
pointed out, this proposition may be only hypothetical. A second mechanism is based on
the “modon” theory of a solitary Rossby wave. The role of a modon, or vortex pair, in
blocking has been examined by Haines and Marshall (1987), based on non-linear solu-
tions of the equivalent barotropic vorticity equation.3 Vortex pairs can be excited by an
appropriate vorticity forcing function; for example, synoptic systems that propagate in a
diffluent jet, as described above (Figure 4.40). However, the results of Frederiksen and
Figure 4.40 Schematic illustration showing how transient eddies reinforce the dipole pattern of
blocking. (Shutts, 1983)
11 Bell (1990) suggest that the modon mechanism may be unrealistic, because blocks are
constantly changing and being regenerated by dynamic processes; they are not stationary.
Numerical model studies demonstrate that blocking can develop in experiments with
no mountains, although the duration of blocks in such cases is shorter than observed
(Mullen, 1987). Kikuchi (1971) notes that the dynamic effect of mountains appears to
cause blocking occurrences in the sector clockwise from 150°E to 150°W and from 30°E
to 30°W. In the sectors 30°–150°E and 30°–150°W, however, blocks seem related to
dynamic and thermodynamic forcings. Scorer (1979) drew attention to Rossby’s relation
for vorticity conservation along a streamline:
0
y constant
where = relative vertical vorticity,
f /
y, and f the Coriolis parameter. This
implies that blocking anticyclones can form downstream of an orographic barrier, or any
other extensive flow perturbation. For example, during the winter 1978–79 blocking in
the North Atlantic and North Pacific developed in two instances downstream of exten-
sive regions of intense cyclogenesis (Hansen and Chen, 1982).
The role of transient eddies in producing blocking through the transport of vorticity
was first proposed by Green (1977), based on a study of blocking during the European
drought of summer 1976. Shutts (1983) notes that continual replenishment of mass in a
0 blocking anticyclone is required in order to overcome frictional dissipation or mass
removal by the mean flow. The location where a jetstream split forms is a region of defor-
mation within which synoptic eddies transfer vorticity to the block, with anticyclonic
(cyclonic) eddies becoming absorbed into the poleward (equatorward) parts of the pattern.
Shutts suggests that this eddy propagation sets up a region of high eddy enstrophy (one-
half the square of the relative vorticity) with net downgradient energy flux. Figure 4.40
illustrates schematically how transient eddies reinforce the dipole pattern of vorticity.
Shutts (1983, 1986) uses the non-linear barotropic vorticity equation to show that blocking
modes can be generated by introducing a “wave maker” to excite small-amplitude waves
superimposed on simple flow fields.
0 The split jet can be enhanced by the westward motion of planetary-scale waves 1 and
2. Lejenäs and Madden (1992) show that some 20–40 percent of 500 mb blocks in the
North Atlantic and North Pacific during 1950–79 were associated with a westward-prop-
agating wave No. 1 which has a sixteen to twenty-day period and maximum amplitude
at 60°N. A westward-propagating wave No. 2, which has maximum amplitude at 40°N,
occurred within 20° longitude of the blocked flows for an additional 10–20 percent of
the blocks. The results lend support to the eddy straining mechanism proposed by Shutts.
These ideas are extended to the inception of blocking events by Neilley and Dole
(1991). Using composite analyses of eighteen wintertime blocks in the eastern North
Atlantic, they show that the development of large-scale flow anomalies is preceded (six
0 days earlier) by a general increase in synoptic eddy activity over a large part of the hemi-
sphere. As wave activity over the North Pacific and western North America propagates
eastward, eddies approach the upstream margin of the developing block and the large-
scale flow anomalies rapidly intensify. This suggests that feedbacks between the eddies
and the large-scale flow serve as a self-sustaining mechanism for the flow anomalies. The
net forcing of the transient eddies on the time-mean zonal flow over the central North
Atlantic (calculated from the divergence of the zonal flow E vector; see Appendix 4.2)
indicates that the eddies tend to weaken the westerlies. The eddies also become merid-
ionally elongated as they move into the area of diffluent large-scale flow over the
west-central Atlantic in accordance with Shutts’s model (1983, 1986, 1987a; see Figure
0 4. 41). Subsequently the storm track in the Atlantic shifts northward towards Iceland and
eddy activity weakens over the east-central Atlantic as the block develops. The climax
of the blocking event during February 12–16 1983 is a pronounced dipolar block over
11 western Europe and the British Isles (see Figure 4.34a). The Ertel potential vorticity (Q)
Figure 4.41 Meridional elongation of eddies moving into an area of diffluent large-scale flow (“eddy
straining”) over the west-central Atlantic associated with a split jetstream and a dipolar
block north of Scotland (see Figure 4.40), February 15 1983 (Shutts, 1986). The arrows
show the Eliassen–Palm flux vectors which point in the direction of westerly
momentum transport, superimposed on the mean 300 mb stream function field centered
at 30°W (British Crown copyright)
distribution for February 15 (Figure 4.42) shows that a narrow tongue of low values
extends from the Canary Islands to Greenland and then eastward into the anticyclone.
High Q values are present over 40°–50°N, 0°–10°W, associated with an upper cut-off
low (Figure 4.34a), and a tongue of high values over the western Atlantic pushes south-
eastward to form a cut-off low on the next day. Another case study of a North Atlantic
blocking cycle confirms the role of anticyclonic vorticity advection at 500 mb in the
formation and maintenance of the blocking anticyclone (Lupo and Smith, 1995b).
A climatological analysis of blocking onset over northern Europe and over the Gulf of
Alaska during 1955–84 shows that baroclinic wave activity in the two and a half to six-
day range is enhanced along the upstream storm track about five days before the block
is fully established (Nakamura and Wallace, 1990). This wave activity then separates into
a northeastward branch along the western margin of the blocking anticyclone and a south-
eastward one around an anomaly of low frequency (over six days) cyclonic circulation
developing south of the block. Decay conditions are associated with a tendency to
suppressed upstream wave activity. It seems that baroclinic wave activity helps to maintain
the climatological 500 mb ridges located over the eastern oceans of the northern hemi-
sphere. It now appears that the onset of blocking is linked with antecedent explosive
11
Figure 4.42 Transport of Ertel potential vorticity (plotted as Q1/4 to show the gradients in mid-lati-
0 tudes and the subtropics) for the blocking pattern shown in Figure 4.41. The contour
interval is 1 103. The map illustrates the northward transfer of subtropical air of
low potential vorticity into a blocking high over western Europe. (From Shutts, 1986;
British Crown copyright)
cyclogenesis over the western North Pacific and western North Atlantic. These cycloge-
netic events occur about once every five to seven days. However, statistical analyses fail
to identify a clear relationship (Colucci and Alberta, 1996). The most frequent precursor
of blocking onset within five days (76 percent of 222 cases of explosive cyclogenesis
during seven winters) is anomalously strong planetary-scale southerly flow and weaker
0 westerly flow at 500 mb over the antecedent cyclone. However, the dynamical mecha-
nisms involved in blocking onset remain to be resolved. Colucci and Alberta (1996)
suggest either interaction between synoptic-scale waves and “preconditioned” anomalously
amplified planetary waves or anomalous planetary-scale advection of synoptic-scale
vorticity associated with the upstream cyclone.
A diagnostic analysis of synchronous blocking in the North Atlantic and North Pacific
sheds some light on scale interactions (Lupo, 1997). Such simultaneous blocks are compar-
atively rare – about 7 percent of all the blocking days identified by Lejenäs and Økland
(1983). By partitioning the planetary- and synoptic-scale components of 500 mb height
tendency, non-interaction and corresponding scale interactions, Lupo (1997) demonstrates
0 no dynamic connection between the paired blocks. Hence “local” mechanisms appear to
be responsible for the synchronous developments. In particular, upstream cyclogenetic
events contribute to the formation of blocks, and whereas vorticity advection is impor-
11 tant for low-amplitude development, temperature advection is the primary factor in
high-amplitude wave blocking events. As shown by Hartman and Ghan (1980), there are
important differences between blocks forming over the North Pacific (56°N, 160°W) and
Europe (54°N, 10°E). In a study by Nakamura et al. (1997) composites of 250 mb Ertel
potential vorticity (see Appendix 3.1) for the thirty strongest events during 1965–92 are
derived and the synoptic eddies are removed by filtering. In the North Atlantic, upstream
of European blocks, there is a quasi-stationary wave train, which has no counterpart over
the western Pacific. Nakamura et al. suggest that, for European blocks, wave activity flux
is derived from the upstream Rossby wave train. Vorticity flux convergence from tran-
sient synoptic eddies supports 75 percent of the amplification of Pacific blocks but only
45 percent of the European blocks.
The barotropic model of Shutts (1986) cannot explain the relative warmth of blocking
highs. Green (1977) proposes that subsidence is forced by differential eddy vorticity in
the vertical, whereas Shutts (1983) suggests that tropical air (with low potential vorticity)
is advected in southerly flow upstream of the block as each cyclone approaches from the
west. Mullen (1987) examines the net forcing of blocking by synoptic eddies. Temporal
filtering is used to separate transient eddy transports of heat and vorticity from longer
time-scale effects during observed and modeled blocking events. Anticyclonic eddy forcing
is in a quadrature relationship (one-quarter wavelength upstream) with the blocking high,
as found by Austin (1980). Mullen shows that the thermal anomalies of the block are
– —–
maintained by advection of time-mean temperature by the time-mean wind (V # T );
heat transports by the eddies act in the opposite direction to decrease zonal asymmetry
in the thermal field. The eddy heat transports thus perform a similar role for climatolog-
ical-mean and blocking flows. In contrast, the vorticity transports by the synoptic eddies
are dissimilar for blocking and climatological flows. For the latter, the eddies tend to shift
the jetstream axis poleward. Upstream of a block the eddy vorticity transport has the same
tendency, but at the block itself the eddy forcing acts to displace the blocking pattern
westward. Trenberth (1986) found the same result in an independent analysis of southern
hemisphere blocks.
Less attention has been paid to the decay of blocking anticyclones. As noted above,
Nakamura and Wallace (1990) find climatologically that baroclinic wave activity upstream
of a decaying block is suppressed; anomalies are of opposite polarity to those during the
onset phase, and presumably this applies also to the eddy transports of vorticity and heat.
4.7 Low-frequency circulation variability and persistence

The tendency of the planetary circulation to display low-frequency variability, with time
scales of several weeks to several years, and with distinct geographical distributions, is
an important problem in climate dynamics and long-range prediction. Theoretical aspects
of this problem are now attracting the attention of dynamic meteorologists, but we begin
with an overview of the empirical nature of the variability.
Beyond the life span of a family of mid-latitude cyclones, about ten to fifteen days,
detailed predictability of weather conditions is lost. The theoretical limit to such
predictability marks the threshold of what is called low-frequency variability of the atmos-
pheric circulation. Wallace and Blackmon (1983) analyzed northern hemisphere daily
500 mb height data for winters 1962–63 to 1979–80 to study its spatiotemporal charac-
teristics. Three time scales were isolated by filtering the data, using (1) a band-pass filter
to distinguish baroclinic waves, (2) a low-pass filter to emphasize fluctuations longer than
ten days, and (3) a thirty-day average. The spatial variance characteristics of each field
were compared with those of the original unfiltered data (Figure 4.43). Category (1) depicts
a zone resembling that of storm tracks over the oceans, whereas (2) and (3) show high
variance in regions of the North Atlantic, North Pacific, and Siberian Arctic that have
been identified as centers of blocking activity by Knox and Hay (1985). The similarity
of (2) and (3) suggests that the geographical pattern of variance is reasonably constant
11
Figure 4.43 Variance of 500 mb height data for winters 1962/63–1979/80; 10 m contour intervals.
(a) Unfiltered twice daily data. (b) Band-pass filtered to show two and a half to six-
day baroclinic waves. (c) Low-pass filtered for periods over ten days. (d) Thirty-day
0 mean value. The contribution of the mean annual cycle to the December–February
values has been removed. (From Wallace and Blackmon, 1983)
at time scales beyond ten days. The baroclinic waves were also found to contribute less
to the total variance than the low-frequency components. The great majority of the atmos-
pheric variability in the northern hemisphere 500 mb heights is attributable to the 10–90
time scale (Blackmon et al., 1977). The low-frequency fields are determined principally
by the planetary waves (0 ≤ m ≤ 6) at high latitudes and synoptic-scale waves (7 ≤ m
≤ 12) in middle latitudes. In contrast, in the southern hemisphere the transient eddies with
periods less than a week play a much larger role in mid-latitudes. Jones and Simmonds
0 (1993) indicate that the contributions from medium and short-wave (7 ≤ m ≤ 12 and
13 ≤ m ≤ 18, respectively) to MSL pressure variations peak around 50°–60°S in the hemi-
spheric storm track. The bandpass filtered (two and a half to six-day) component of the
11 daily sea-level pressure data for 1972–91 contributes about 25 percent of the variance,
and locally 40 percent between 30° and 60°S. The low pass (over ten-day) frequencies
contribute to the larger share, with maximum values equatorward of 30°S and in high
latitudes, where the contribution of planetary waves (0 m 6) becomes increasingly
important.
The processes involved in establishing the major modes of atmospheric behavior have
been identified by the analysis of a multilevel, quasi-geostrophic model (Frederiksen and
Bell, 1987). Four classes of motion system are distinguished as the fastest-growing modes
on a climatological basic state for January 1978:
1 Rapidly propagating cyclogenesis lasting two to six days, corresponding to shallow

disturbances with maximum amplitudes in the storm track zones over the North
Atlantic and North Pacific.
2 A dipole mode in the meridional direction, with maximum amplitude at 300 mb,
corresponding to onset-of-blocking patterns of larger scale and lasting seven to eleven
days.
3 Barotropic modes of intermediate scale, lasting eleven to seventeen days and resem-
bling mature anomaly patterns over the Atlantic and Pacific.
4 Large scale, long-lasting (more than seventeen days) equivalent barotropic modes
resembling the major northern hemisphere teleconnection patterns.
These four classes closely match the variance patterns found by Wallace and Blackmon
(1983), shown in Figure 4.43. Analysis of a blocking event in the North Atlantic in
January 1979 identified the periods of corresponding instability modes growing on the
instantaneous synoptic flows as being in the ranges one to five, five to ten, ten to twenty,
and over twenty days, respectively.
Before considering further the geographical features of low-frequency variability, we
need to examine the behavior of particular modes of circulation in the time domain. There
are three primary characteristics of circulation modes or “weather regimes” – recurrence,
quasi-stationarity, and persistence (Michelangeli et al., 1995). Recurrence can be identi-
fied by locating atmospheric states that have the largest probability of occurrence.
Examples include the work of Molteni et al. (1990) and Kimoto and Ghil (1993). Cluster
analysis affords a further practical approach (Mo and Ghil, 1988; Cheng and Wallace,
1993). Michelangeli et al. (1995) use a dynamic clustering procedure with iterative
partioning in order to obtain stable clusters. The data are agglomerated around randomly
chosen seeds and the partition is found which minimizes the sum of the variances within
each cluster with respect to their centroids; the Euclidean distance metric is used as a
similarity measure. A classifiability index is developed, based on the correlation between
the cluster centroids obtained from different seeds.
Quasi-stationarity of regimes occurs when the large-scale motion is stationary in a
statistical sense. The large-scale patterns are maintained despite feedback from transient
eddies; this is termed non-linear equilibration. Michelangeli et al. (1995) compare the
recurrent and quasi-stationary patterns in 1949–92 winter 700 mb maps for the northern
hemisphere, sampled on alternate days. Scale separation is achieved by an EOF analysis
for both the Pacific and the North Atlantic–European sectors. Only three clusters for the
Pacific sector and four for the Atlantic can be distinguished from a first-order Markov
process having the same covariance as the data at lag 0 and lag 1 (two days). Identification
of quasi-stationarity of regimes follows the procedure of Vautard (1990). A state vector
containing the large-scale variables of flow is defined as the vector containing the projec-
tion coefficients on to the first few EOFs. The composite time tendency of any large-scale
pattern is calculated as the mean of the instantaneous tendencies associated with adjacent
patterns within a specified radius. This mean is weighted inversely with distance. About
100 convergent solutions were identified for 2,974 iterations, but only 10–20 percent of
them were retained as significant, based on reproducibility. The patterns obtained by the
11 two approaches differ in their temporal characteristics. The recurrent patterns show a slow,
systematic evolution, or drift, in many cases. This may involve amplification of an
anomaly, for example, but in other cases such evolution may significantly change the
pattern over a few days.
Persistence of regimes can be defined from patterns of anomalies (above/below a spec-
ified threshold) that persist for a given time interval, as illustrated by Dole and Gordon
(1983). There are a large number of investigations of persistence. For example, Horel
(1985a) examined daily northern hemisphere 500 mb maps for winters 1965–66 to
1981–82, using time series of pattern correlations. This neglects their amplitude and spatial
0 configuration. Based on correlations between pairs of days exceeding 0.5, there were fifty-
eight quasi-stationary regimes lasting seven days or more. These runs accounted for 25
percent of all days, and a third of them began in December. The most common regime
features a wave No. 3 pattern in mid-latitudes superposed on zonally symmetric anom-
alies of opposing signs between middle and polar latitudes. The circulation is also more
persistent than red noise over two to seven-day intervals on 25 percent of winter days,
although this persistence varied considerably from year to year. In comparison with indi-
vidual time series of station weather (Oerlemans, 1978), there are few “breaks” in the
hemispheric patterns of height anomalies, i.e. transitional situations are more common
than a switch in mode. Horel also concludes that persistence originates through plane-
0 tary-scale, not regional, processes. A further analysis of 500 mb height fields, by
application of Oerlemans’s technique for defining breaks in a time series, their transition
speed, and amplitude, is described by Yang and Reinhold (1991). Grid-point values of
500 mb heights for 1946–84 are filtered to retain scales longer than ten days after removal
of the seasonal cycle. The “quality” of a break is defined by the ratio of the amplitude
of the shift relative to the root mean square of the variance. The quality is 1.0 when the
amplitude is equal to the fluctuations (“noise”) about an ideal break function. The
frequency of breaks with a quality of 3.5 and a height change of 200 gpm at 45°N, 170°W
over the thirty-year record has a peak at 5.4 days for the transition time. This is confirmed
at other locations in mid-latitudes, with distinct longitudinal maxima around 10°–20°W
0 and 160°–180°W, corresponding to the maxima in frequency of persistent anomalies
reported by Dole and Gordon (1983) (see section 4.5). Breaks are relatively rare events,
such as occur following explosive cyclogenesis. The frequency of six-day transitions
suggests the possibility of baroclinic instability as a mechanism, although breaks occur
over a wide range of time intervals (or speeds).
There is need for caution in interpreting analyses of single variables; 500 mb height
data spanning 15°–75°N for four winters (detrended for variability longer than ninety
days) are used by Sutera (1986) to examine the frequency characteristics of both zonal
geostrophic wind and the amplitude of planetary waves 2, 3, and 4. The probability density
distribution of height variance associated with these waves is found to be bimodal in each
0 winter, and of the four winters together, whereas the corresponding distributions of zonal
wind are unimodal but slightly skewed, suggesting interannual variability (Figure 4.44).
Mean 500 mb height maps for the days representing each of the two modes of height
variance resemble a high-index pattern and a large-amplitude wave pattern. The differ-
ence maps between modes 1 and 2 in each winter are analogous to patterns obtained by
Charney et al. (1981) using a different objective criterion for blocking. In an extension
of Sutera’s study, Hansen (1986) used daily 500 mb height data for March 1980–May
1984 to compute wave amplitude for Fourier wave Nos 2–4. The results were filtered to
remove high-frequency (of five days or less periods) and low-frequency (annual and inter-
annual) variability. A frequency histogram again reveals bimodality. The 500 mb field is
0 strongly zonal in Mode 1 and has prominent troughs over the western North Pacific and
eastern North America in Mode 2 (Figure 4.45). Their relative frequencies are 54 percent
and 46 percent, respectively. It is noteworthy that days subjectively classified as showing
11 “Rex blocks” (20 percent of the total) do not generally occur in Mode 2 events; the latter
Figure 4.44 Probability density distribution of (above) 500 mb height variances for four winters
and (below) the corresponding variances of zonal wind. (From Sutera, 1986)
11
0
Figure 4.45 Mean 500 mb maps (gpm) for (a) Mode 1 days representing a high-index pattern and
(b) Mode 2 days representing a large amplitude wave pattern. (From Hansen, 1986)
11
are amplified wave patterns. Rex blocks can occur, however, on days designated as Mode
1 with overall zonal flow at 500 mb.
In spite of the apparently considerable evidence for bimodality, there are other results
that cast doubt on the idea. Using 500 mb northern hemisphere height data for thirty-nine
winters (with ten-day low-pass filtering), Wallace et al. (1991) analyze the spatial covari-
ance, the spatial anomaly correlation, and the statistical distance (RMS difference) between
each map and all others in phase space. There is no evidence of bimodality, although the
frequency distributions show that large positive spatial covariances and positive anomaly
correlations are much more frequent that negative ones. In other words, there are more
positive anomaly analogs than antilogs (or “mirror images”). Van den Dool (1991),
however, finds that antilogs are almost as common as analogs for 500 mb flow patterns
over eastern North America, except for very deep lows which lack antilogs. While these
asymmetries may suggest regime-like behaviors, they may alternatively represent subtle
non-linearities in the dynamical behavior of the atmosphere at low frequencies.
The relative persistence of zonal and blocking patterns is a subject of considerable debate.
Using a measure of day-to-day persistence in 500 mb flow over western Europe, Madden
and Lejenäs (1989) found the degree of persistence to be inversely proportional to the
strength of the zonal wind at 55°N for 20°W–40°E. However, contrary to these results and
traditional ideas, hemispheric zonal flows have been found to be more persistent than block-
ing regimes over the central Pacific (Horel, 1985b; Legras and Ghil, 1985). Strong 500 mb
zonal flows are more persistent than either random or red noise processes, according to
Horel (1985a). He also identified a North Atlantic “seesaw mode” (m 1) (see section 5.6)
and an enhanced polar vortex pattern (m 3) among the diverse patterns that occur in
extended runs during winter months. These conflicting findings may depend on the size of
domain and the particular regions selected. The persistence properties of circulation anom-
alies do differ regionally. Blocked patterns in the central North Pacific tend to occur with
zonal flow across the west coast of North America, while in the North Atlantic an omega
pattern of blocking is common. In contrast, flow regimes are less persistent over East Asia
and eastern North America, according to Dole and Gordon (1983).
Positive and negative anomaly patterns of 500 mb heights in the northern hemisphere
during 1963–64 to 1976–77 are found to have a nearly constant probability of persisting
for an additional day beyond five days, i.e. persistence is independent of the duration of
a run (Dole and Gordon, 1983). For moderate magnitude and duration events occurring
south of the Aleutians, west of the British Isles, and near Novaya Zemlya, the frequency
of positive and negative anomalies is about equal. Positive anomalies are rather more
frequent for long duration, large-magnitude events. An important finding by Dole and
Gordon is that these persistence patterns can be described by a simple non-linear autore-
gressive model and, contrary to Charney and DeVore’s (1979) theoretical ideas (described
below), there is no evidence of multiple quasi-equilibria.
Under the assumption that the statistics of westerly circulation are contained in a
normally distributed continuum from intense zonal flow to blocking flow, considerations
of symmetry would imply that both extremes should exhibit persistence tendencies. This
idea has been tested by Tarleton (1986, 1987). The geostrophic flow is determined for a
channel defined by a pair of 500 mb height contours at intervals of 10° longitude.
Following a rank ordering of the flow intensity (from high to low), the statistics are
divided into halves. Tarleton calculates the frequency of high and low index flows for
thirty years of daily 500 mb data at two-day intervals, using actual heights and height
anomalies in both hemispheres. The autocorrelation of these categories of flow index at
lags of two days to six days shows significantly greater persistence for blocking than for
zonal flows. This implies that strong blocking and zonal flows are not symmetrical with
respect to their temporal characteristics.
There is theoretical and empirical evidence that intense zonal flow becomes unstable.
Theoretical considerations show that on the equatorward side of a westerly jetstream the
11 flow becomes unstable where anticyclonic shear exceeds the Coriolis parameter (see
section 4.2). When the critical value of dynamic instability is reached, meridional motion
will tend to develop. Moreover, under these conditions a small initial perturbation rapidly
amplifies and a low-index, blocking regime may ensue.
Lindzen (1986) concluded from theoretical and statistical considerations that blocking
regimes should not be regarded as more persistent than other atmospheric circulation
anomalies of corresponding geographical scale. This proposition continues to be re-exam-
ined. For example, the persistence of blocking events as defined by the subjective criteria
of Rex (1950) and by the definition of Dole (1983) for positive height anomalies is
0 compared by Liu (1994). Positive height anomalies at 45°N represent a ridge extending
northward from the subtropical anticyclone, whereas positive anomalies at 60°N typically
feature a block centered near 50°N with negative height departures in lower latitudes. Liu
finds that day-to-day changes in geopotential heights are less than average over the block,
and greater than average to the north of it, as a result of the jetstream and eddies being
displaced poleward. Blocking and strong zonal flow modes are found to have similar
height anomaly patterns, of opposite sign, as well as similar average persistence, resem-
bling a first-order Markov processing in each instance.
The existence of distinct regimes of blocked and zonal flows is important for predic-
tion purposes. Rather than purely statistical behavior, it implies some underlying orderly
0 dynamical process that Ghil (1987a) characterizes as “order in chaos.” The idea that there
are several possible circulation modes (i.e. multiple flow equilibria) for a unique set of
boundary conditions has intrigued meteorologists. The concept of transitive and intransi-
tive flow behavior was introduced to general circulation studies by Lorenz (1969a); a
transitive or ergodic system is one which all initial states lead to the same solution, in
terms of its statistical properties. An intransitive system has two or more sets of statis-
tics, depending on the initial states. An almost intransitive system has two or more sets
of statistics during finite periods of its evolution from different initial states (Figure 4.46).
It is unclear which category describes the climate system. Using a set of simple equa-
tions for thermal convection, Lorenz (1963) demonstrated that the system shifted randomly
0 between a pair of steady states. The behavior of a system of three (or more) variables
0
Figure 4.46 Schematic illustration of a transitive, intransitive, and almost intransitive system in a
climate system with time-independent forcing according to E.N. Lorenz. (From US
11 National Research Council, 1975)
Figure 4.47 The behavior of three variables in a model of thermal convection over a long time
interval and the directions of the corresponding empirical orthogonal functions (EOFs
1, 2, and 3). (a) The time variations of one variable, Y. (b) and (c) Projections of the
trajectory in phase space on to the Y, Z plane and X, Y plane, respectively. Numerals
in (b) and (c) denote successive epochs 1,000 time units apart. (From Lorenz, 1963)
over a long time interval can be displayed as the trajectory of one variable, Y, in phase
space on to the X, Y- or Y, Z-planes (Figure 4.47). The path of a trajectory converges,
as a result of dissipation, to a subset with an infinite number of layers of dimension two
(or more). All points on this “attractor” are stable in the direction(s) perpendicular to the
layer it is in, but unstable in directions tangential to the attractor. Trajectories may approach
one another closely, but will still diverge. The long-term behavior of the trajectory for a
system with three (or more) variables has been termed a strange attractor (Ruelle, 1980).
Daily climatic data have been used to illustrate the possible existence of strange attrac-
tors (Essex et al., 1987). Numerical experiments by Charney and DeVore (1979) with a
simple equivalent barotropic model showed two stable stationary solutions (zonal and
blocked flow) rather than the single solution that had been expected. More elaborate,
but still simple, versions of this approach indicate that, when the forcing of
the flow is allowed to dissipate over realistic time scales, stable solutions break down
and periodic oscillations or aperiodic solutions may develop (Ghil, 1987a). Figure 4.48a
and c shows the time behavior of variable x, with respect to the long-term mean x–, for
solutions to forced dissipative models with non-linear dynamics. A periodic solution in
time, when displayed in three-dimensional phase space (variables x, y, z), can approach
a limit cycle (Figure 4.48d), whereas a convergent solution appears as a fixed point in
11
Figure 4.48 Long-term behavior of solutions to forced dissipative models of non-linear dynamics.
0 (a) and (c) How variable x behaves over time, t; x denotes time-mean. (b) and (d)
Phase–space maps of the attractor set where (b) is a fixed point convergence, (d) is a
limit cycle, and arrows show the directions of stability and motion. (From Ghil, 1987)
phase space (Figure 4.48b) (Ghil, 1987a). Aperiodic solutions are more complex than
those in Figure 4.48.
Ghil (in Radok, 1987, p. 141) summarized these achievements in understanding the
structure of atmospheric circulation modes and their temporal behavior by the following
statement:
0
this modern mathematical and computational work validates a lot of ideas which have
been around the long-range forecasting community for a long time. In other words,
regimes are just Grosswetterlagen [see section 7.4], and some of these persistence
times are the synoptic periods of Multanovski.
Low-frequency modes of circulation variability have been attributed to at least three

different dynamical processes: multiple equilibria, or stationary solutions, of a barotropic
or baroclinic model (Charney and DeVore, 1979); Rossby wave propagation from low
latitudes in response to thermal forcing (Hoskins and Karoly, 1981); and barotropic insta-
0 bility of a zonally asymmetric circulation (Simmons et al., 1983). Charney and DeVore’s
theory predicts bimodality of the mean zonal wind and the planetary wave amplitude; as
noted above (p. 226), observations confirm the latter but not the former (Hansen, 1986;
Sutera, 1986; Hansen and Sutera, 1987).
The frequency distribution of wave amplitude can be compared with hemispheric circula-
tion regimes identified by clustering techniques. For example, Mo and Ghil (1988) analyze
a non-linear barotropic model with orographic forcing for the northern extratropics. In the
phase space generated by the first two EOFs of the model fields they show a multimodal
probability density function. For 500 mb northern hemisphere height field data, six clusters,
obtained using pattern correlations, represent 62 percent of the sample. Two stationary
0 clusters (a zonal and a blocked pattern) have persistences greater than ten days; another
persistent mode resembles the Pacific–North American pattern of Wallace and Gutzler
(1981) (see section 5.10). Transient clusters (of wave trains) occur in the two and a half to
11 six-day band. The analysis supports the existence of bimodality of wave amplitudes.
A different approach to defining circulation regimes involves identifying clusters of
atmospheric states in a five-dimensional phase space generated by the leading EOFs of
eddy geopotential fields (Molteni et al., 1990). A cluster is defined about each local density
maximum of points in the phase space, applying a Euclidean distance measure. The study
uses five-day mean eddy fields of 500 mb height for thirty-two winters (December–
February), 1952/53–1983/84 for 20°E–90°EN. The EOFs were computed on a larger
sample of October–March for the thirty-two years so that the first EOF includes the
seasonal cycle (Molteni et al., 1988). Six clusters, illustrated by the average height fields
of their centroids (Figure 4.49), are identified by Molteni et al. (1990). The largest, C1,
representing 45 percent of the 576 pentads, approximates the winter climatology. The
other five clusters (17 percent of the fields) represent anomalous flow regimes: C3 corres-
ponds to zonal flow with below-average wave amplitude; clusters C1 and C2 contain
features characteristic of positive Pacific–North American anomalies (see section 5.10)
and C5b has a negative-Pacific–North American signature; C4 suggests blocking over
Europe whereas C5 has strong ridges over Europe and the eastern Pacific. Molteni et al.
remove the redundancy in these patterns by determining three rotated EOFs from a linear
combination of the five original EOFs. A new five-dimensional density–cluster analysis,
performed on the 576 three-dimensional vectors defined by the rotated EOF coefficients,
generated eight local density maxima; the corresponding eight clusters contain 92 percent
of the 576 fields (40 percent in the first one). The new clusters (three of which are
subgroups that can be combined to give a total of six) closely resemble the original ones
but include more of the fields sampled. Probability density estimates for the total wave
amplitude computed for the first sixteen and last sixteen winters both show a clear
bimodality (near 70 gpm and 90 gpm). From the rotated EOF coefficients, this variability
appears to be associated with wave No. 3. New clusters 1 and 3 represent the low-ampli-
tude class and 2, 4, 5, and 6 the high-amplitude class. (The new clusters are numbered
according to their similarity with the first set.) Transition frequencies calculated between
the clusters show about 50 percent persistence for clusters 1 and 3 (low-amplitude waves),
40 percent for clusters 4 and 5, and only 30 percent for cluster 2 (high-amplitude). Among
275 transition cases, 74 percent are to or from cluster 1. Transitions from cluster 1 to
high-amplitude modes occur more frequently toward a regime of large, negative
Pacific–North American index than its positive counterpart. Two main asymmetrical tran-
sition cycles were noted: from cluster 1 to 5 (wave amplification), then either to cluster
3 (low amplitude but negative Pacific–North American index) and back to cluster 1; or
to cluster 4 and then to cluster 2 (positive Pacific–North American index, high ampli-
tude) before returning to cluster 1. This type of analysis can be carried out on GCM fields
(Molteni and Tibaldi, 1990) in order to diagnose their biases and particularly their ability
to predict regime transitions.
4.8 Intraseasonal oscillations

The preceding discussion of low-frequency variability focused on the tendency for the
circulation to exhibit large-scale episodic spatial structure (zonal/blocking modes) and
considered the persistence characteristics of these regimes. In this section attention
turns to the oscillatory behavior of the atmosphere on the intraseasonal time scale, i.e.
approximately ten to 100 days. While there are connections between the episodic
and oscillatory components (Ghil, 1987a, for example), it is convenient here to treat
them separately. Intraseasonal oscillations (ISOs) are of considerable importance because
of the significant percentages of the variance which they account for in tropospheric
winds and cloudiness, relative to synoptic-scale disturbances (under ten days), especially
in the tropics. The discussion covers the tropics and extratropics, although studies of
tropical oscillations have received much more attention and now span the last three
decades.
11
0
Figure 4.49 Anomaly patterns for the centroids of six clusters of mean eddy fields at 500 mb from
rotated EOF analysis of data for winters 1952/3–1983/4. (From Molteni and Tibaldi,
11 1990)
4.8.1 Tropical intraseasonal oscillations
The primary oscillation identified in dynamical fields in the tropics is the forty to fifty
(or thirty to sixty) day Madden–Julian Oscillation (MJO). The signal was first identified
through spectral analysis of lower stratospheric winds at Kanton (3°S, 172°W), where a
spectral peak of forty-one to fifty-three days was reported (Madden and Julian, 1971).
Similar characteristics were also found in zonal wind, pressure, and temperature data from
distant stations in the tropics, showing cross-correlation peaks and a high degree of coher-
ence. Surface pressure oscillations suggest an eastward-propagating wave, with the
characteristics of a standing oscillation in the vertical plane that originates in the Indian
Ocean. At 150 mb, the u component of wind at tropical stations shows that the oscilla-
tion is strongest in December–February and is most pronounced over the Indian Ocean
and from the Date Line east to South America. Indeed, the oscillation is now established
as a global-scale disturbance, concentrated in zonal wave No. 1, that propagates eastward
with periods in the thirty to sixty-day range (Madden and Julian, 1972, 1994; von Storch
and Xu, 1990). Figure 4.50 illustrates a conceptual zonal plane model of the disturbance
over the eastern hemisphere; the letters denote temporal progression of low (A) to high
(E) pressure at Kanton; negative sea-level pressure anomalies at Kanton, shown at the
bottom of the figure, are shaded.
The oscillation is represented in various climatic variables such as winds, cloud amount
and OLR (Weickmann et al., 1985), as well as in global relative angular momentum.
Spectral analysis of fifteen years of OLR data, for example, shows a dominant peak at
forty to fifty days, with additional peaks about twenty to thirty and seventeen days
(Anyamba and Weare, 1995). These peaks are distinct away from the equator in the
western hemisphere. The OLR data confirm the equatorial dipole at 80°E and 160°E. A
study using extended EOFs of the time–height modes of u and v components and convec-
tive heating and drying from rawinsonde data for 0°–10°S confirms the importance of a
forty-one-day oscillation in u, accounting for 48 percent of the variance (Fraedrich et al.,
1997). There are westerly (easterly) anomalies in the upper (lower) troposphere, indica-
tive of an internal Kelvin wave. The heating–drying index has a corresponding forty-day
oscillation, accounting for 41 percent of the variance, with a mid-tropospheric maximum.
The second pair of EOFs represents a twenty-four-day upper tropospheric zonal wind
variation (14 percent of the variance), while the second pair for convective heating–drying
represents a thirteen-day oscillation with a structure similar to the forty-day one.
Associated with the MJO is a spectrally broader convection signal that seems to be concen-
trated in the eastern hemisphere, where the climatological mean pattern indicates
convection. It propagates eastward at about 5 m s1 with periods between thirty-five and
ninety-five days (Hendon and Salby, 1994). The convective signal, identified in daily OLR
data for 1979–89, shows a concentration in zonal wave Nos 1–3 over the Indian
Ocean–western Pacific Ocean (Salby and Hendon, 1994). In the Indian Ocean–western
tropical Pacific the eastward movement of cloud complexes, for example, shows a pattern
closely similar to the propagation of the oscillation (Figure 4.49).
The zonal wind oscillations in the tropical upper troposphere resemble a thermally
forced Kelvin wave4 trapped within the tropical easterlies by the change in sign of the
Coriolis parameter at the equator (Webster, 1972, 1973; Holton, 1973). The low-latitude
forcing is attributed to the release of latent heat by precipitation and the effects of subtrop-
ical steady forced motions. The longitudinal heating distribution in low latitudes appears
to give rise to a slowly varying Kelvin wave response with a time scale of about one
month, but several aspects of the atmospheric response to local heating and convection
remain to be explored (Madden and Julian, 1994). Salby and Hendon (1994) propose that
in the eastern hemisphere there is a Kelvin structure in the equatorial convective anomaly,
with flanking subtropical Rossby gyres, which are all propagating eastward as a forced
response at about 5 m s1. In the western hemisphere, there is mainly a Kelvin compo-
nent response radiating from the convective anomaly at about 10 m s1. The convection
11
Figure 4.50 Schematic diagram of the mean spatiotemporal variations in the forty to fifty-day MJO
in the zonal plane. The letters at the left indicate time evolution of the station pres-
0 sure at Kanton (2°S, 172°W). A corresponds to low pressure and E to high pressure
at Kanton, with intermediate times denoted by the other letters. The mean pressure
disturbance is shown at the foot of the figure; negative anomalies are shaded. The
11 cumulus and cumulonimbus clouds indicate regions of enhanced convection. (From
Madden and Julian, 1972a)
anomaly is positively correlated with a vertically integrated tropospheric temperature
anomaly and with surface convergence when the MJO is amplifying, implying the gener-
ation of eddy available potential energy; this is converted into eddy kinetic energy in the
upper troposphere (Hendon and Salby, 1994). When the MJO is decaying, however, the
temperature anomaly and convection are out of phase with the surface convergence. It is
striking that anomalous 850 mb convergence and 200 mb divergence are in phase with
the anomalous convection throughout the MJO life cycle. In contrast, the spatial associ-
ation of surface convergence and the tropospheric temperature anomaly, indicating
convective latent heat release, undergo a progressive evolution. This evolution suggests
that moisture convergence in the boundary layer is essential in organizing the convec-
tion. The occurrence of boundary-layer frictional convergence some 40°–50°E of the zonal
convergence at 850 mb, when convection is amplifying, supports the hypothesis of a fric-
tional wave–CISK mechanism, according to Hendon and Salby (1994).
New details concerning extratropical–tropical interactions are continuing to emerge.
Matthews and Kiladis (1999) find that when the Asian jet and upper tropospheric east-
erlies over the western Pacific warm pool are shifted westward, during northern winters,
high-frequency transient waves can propagate equatorward to enhance the MJO. The waves
enhance convective variability in the central Pacific ITCZ, and this can project onto ISO
time scales.
Satellite OLR data near the equator also indicate eastward-propagating cloudy areas
with a scale of zonal wave Nos 1–2. They comprise several eastward-moving super cloud
clusters (SCC) of about 103 km dimension that are made up of smaller clusters moving
westward with a one to two-day lifetime as illustrated in Figure 4.51. During active ISO
phases, large persistent cloud systems occur in the late afternoon (14.00–19.00 LST) and
develop maximum extent of cold cloud tops between 00.00 and 06.00 LST (Chen
and Houze, 1997). These do not decay until the second day, and so a two-day period-
icity (“diurnal dancing”) is set up. The westward-propagating elements are the envelopes
of many convective systems, which may be a combination of a two-day inertio-gravity
wave with the superposed one-day convective regime, rather than merely representing the
footprint of westward-moving clusters (Nakazawa, 1988; Chen and Houze, 1997).
Analogous tropical intraseasonal convection anomalies (TICA) in the equatorial Indian
and Pacific Oceans, with a minimum scale of 30° longitude and a minimum lifetime of
twenty days, as identified from OLR data, show a variety of behaviors (Wang and Rui,
1990). Over a ten-year period, out of 122 such TICAs originating in the west-central
Indian Ocean or off equatorial West Africa, seventy-seven (63 percent) moved eastward,
while twenty-seven moved northward during June–August, and eighteen (weak systems)
moved westward. The seventy-seven eastward-moving systems also show varying behavior
(Figure 4.52); thirty-two remained within ±15° of the equator, twenty-five moved east-
ward initially but near 100°E they moved northeast or southeast; and the other twenty
moved eastward and northward over India and/or the western Pacific. The second group
that move southeastward near 100°E occur during November–April, apparently related to
the Australian monsoon. The corresponding ones moving northeast are primarily associ-
ated with the Asian summer monsoon, while the third group, which combines eastward
and northward movements, occurs predominantly in May and October.
A case study for December 1978–February 1979 over the western tropical Pacific (Sui and
Lau, 1992) identified two intraseasonal oscillations between 0°S and 10°S, propagating east-
ward from the Indian Ocean. One was associated with monsoon onset and surface westerlies
in northern Australia. The oscillations are enhanced as they become quasi-stationary over the
western Pacific warm pool, where they develop a strong lower tropospheric westerly jet.
Westerly wind bursts with a spatial scale of 500–400 km and lasting two to ten days, during
November–April especially, were recognized earlier by Keen (1987).
Cloud areas moving northward at about 1° latitude per day and eastward at 5° longi-
tude per day are characteristic of “active” phases of the Indian monsoon, as has been
0
0
0
0
0
11
11
Figure 4.51 Large-scale eastward-propagating cloud systems in the tropics. Supercloud clusters (SCC) propagate within larger clusters or intraseasonal
variability (left). The SCC contains smaller westward-moving clusters (right).
Figure 4.52 Contour plot of the total number of occurrences of strictly eastward-moving cloud
complexes in 2° by 2° grid boxes for 1975–77, 1979–85. The heavy line and arrows
indicate the central paths; the contour interval is 0.8 per 10 years in (a) and (c), and
0.6 per ten years in (b). (a) Strictly eastward motion. (b) Splitting over the eastern
Indian Ocean. (c) Moving eastward and northward over southern Asia. (From Wang
and Rui, 1990)
11 confirmed by several studies. Madden and Julian (1994) show that maximum (minimum)
cloudiness and precipitation over India coincide with maximum (minimum) zonal wind
at 150 mb over Chuuk (Truk) (7°N, 152°E), consistent with a stronger (weaker) circula-
tion in the equatorial plane. Spectral peaks of about forty days in summer rainfall affect
southern India (Hartmann and Michelsen, 1989). Active periods are associated with the
eastward movement of an equatorial low and a trough progressing northward.
Corresponding thirty to fifty-day modulations of monsoonal westerlies occur in Australia
and are associated with monsoon onset events as well as active periods (Hendon and
Liebmann, 1990).
0 There are also some mid-latitude manifestations of these tropical oscillations in the
upper troposphere. Analyses at 250 mb show eastward propagation of OLR anomalies,
following areas of upper-level divergence (Weickmann et al., 1985). They are strongest
over the warmer waters of the Indian and western Pacific Oceans, particularly in the
summer hemisphere, and negligible over the cooler eastern Pacific and Atlantic. When
cloudiness is present over Indonesia (100°–140°E) the upper tropospheric circulation
consists of paired tropical anticyclones located near the convective area, with twin cyclones
to the east, as illustrated in Figure 4.53. The pattern is more or less reversed when cloudi-
ness extends eastward to the dateline. The circumpolar vortex is expanded in the regions
of equatorial cloudiness and subtropical anticyclones and contracted in the less cloudy
0 cyclone regions.
In addition to the MJO spectral band in the tropics, a twenty-four to twenty-eight-day
mode in OLR variability is equally strong during 1974–86 in the sector from 170°W to
90°W, according to Ghil and Mo (1991). Analysis of area-averaged rainfall over the
tropical Asia–Australia monsoon sector 60°E–120°W and the extra-monsoon sector
120°W–60°E shows a twelve to twenty-four-day mode during 1979–80 (Chen et al., 1995).
At 200 mb this latter mode has a zonal wave No. 1 structure in circulation parameters
and propagates eastward at 25 m s1, especially over the sector 120°W–60°E. This mode
and the MJO cause a seesaw oscillation in intraseasonal variation between these two
sectors of the tropics.
0
0
Figure 4.53 Schematic relationship between OLR areas of cloud and clear skies and 250 mb circu-
lation during stage of maximum cloudiness over Indonesia (corresponding to stage H
11 in Figure 4.50). (From Weickmann et al., 1985)
4.8.2 Intraseasonal oscillations in the extratropics
Intraseasonal oscillations outside the tropics have only recently been examined, but the
results are intriguing. Ghil and Mo (1991) study the ten to 120 day band, using singular
spectral analysis (section 2.5) on northern hemisphere 700 mb daily data for December
1949–December 1986 and southern hemisphere 500 mb data for June 1972–September
1984. The height fields were first analyzed for the dominant spatial patterns of variability,
using EOFs and rotated EOFs (section 2.4); then the leading principal components were
examined in the time domain via singular spectrum analysis (SSA) (see p. 73). Singular
spectrum analysis is appropriate for identifying non-linear oscillations in time series that
are noisy and of limited length. The traveling and standing components of the oscilla-
tions are distinguished by using Hovmöller time–longitude plots of geopotential height
anomalies. In the northern hemisphere there are important oscillations near forty-eight
days and twenty-three days. The former is primarily represented in zonal wave No. 2 and
has both traveling and standing components. The twenty-three-day mode resembles that
identified in the winter season over the western hemisphere by Branstator (1987) and the
fifteen to thirty-day retrograding wave disturbances that create high/low index modes over
the North Pacific (Kushnir, 1987). The twenty-three-day mode is westward-propagating,
has equivalent barotropic structure in the troposphere, and its amplitude increases upward.
For November–March it can account for 25 percent of the variance of tropospheric geo-
potential heights, based on twenty-one years of daily data analyzed by complex EOFs.
Further analysis of northern hemisphere 500 mb height data for 1946–88 finds sixty to
seventy-day as well as thirty to forty-day oscillations (Zhang et al., 1997). While the
latter show a PNA-like pattern and account for only about 13 percent of the local vari-
ance, the sixty to seventy day mode (about 20 percent of the variance) involves an
alternation between anomalies of opposite sign over the polar cap and mid-latitudes of
Europe to North America and the Pacific–East Asia. It is best developed in winter and
spring. The anomalies propagate northward over Greenland and the Ural mountains and
southwestward over Europe and the North Atlantic to North America.
Appendix 4.1 Spectral harmonic functions

The longitudinal departures of a height contour from the zonally averaged value for a
given latitude circle are described by harmonic or Fourier functions (see section 2.5).
The spatial structure of large-scale waves at standard pressure levels in the atmosphere
is represented most conveniently by spherical harmonic functions. The stream function
$ associated with a given n wave has a shape which is expressed by:
$(t, , ) = Am,n sin mr冤冢2t 冣冥 P (sin &)

m
n
where Am,n is an arbitrary constant, longitude, t time, period, and Pnm (sin &)
is a Legendre polynomial of order m and degree n. A global field of geopotential, , can
be expressed in spherical harmonics:
兺兺 P
m
= m
n (sin ) e im
n
m n
where Pnm (sin ) is an associated Legendre function, normalized to unity, of order m and
degree n, longitude, and latitude. These Legendre functions represent solutions
to the non-divergent vorticity equation on a sphere (Haurwitz, 1940). The associated
Legendre functions for m 1, 2, 3, and 4 are illustrated in Figure 4.20a, and the corre-
sponding latitudinal structure of the geopotential height fields is shown in Figure 4.20b.
11 Note that the associated Legendre functions themselves are (antisymmetric) symmetric
about the equator if (n m) is odd (even).
With |m | n, m is the zonal (west–east) wave number, (n – m) the meridional wave
number, i.e. the number of nodal zero values such that $ vanishes along a meridian
between (but not including) the north and south poles (Madden, 1979). If m 1 and n
1, no north–south wave exists. The gravest mode is that with the largest latitudinal
scale. As m increases, the number of zonal waves increases, and the number of merid-
ional waves decreases.
The spherical harmonic series is generally truncated; the truncation may be triangular or
0 rhomboidal (Figure 4.54). In a triangular truncation, every position and direction on the
sphere is treated identically. Spherical functions with m 0 are called zonal harmonics and
express the mean zonal flow. Here the meridional coordinate is geographical latitude, &.
Appendix 4.2 Eliassen–Palm flux

A vector quantity, E, based on Eliassen and Palm (1961), serves as a diagnostic measure
of small-amplitude waves superimposed on a mean zonal flow. The vector E displayed
on cross-sections in a meridional plane characterizes the northward eddy fluxes of angular
momentum and heat (Edmon et al., 1980). For a plane with pressure, p, as a vertical
0 coordinate, the quasi-geostrophic approximation for E, the Eliassen–Palm flux, is:
冦
E = u′v′ ,
f v′′
p 冧
where u and v are zonal and meridional flow components, f the Coriolis parameter
( f (y) where y northward distance), potential temperature, p is a measure of
static stability, and is negative for a stable atmosphere; the overbars denote a zonal mean
and the primes are departures from the means.
0 Figure 4.54 Illustration of rhomboidal and

triangular wavenumber truncations of a
spherical harmonic series (for corresponding
11 degrees of freedom). (From Hoskins, 1980)
The direction of E shows the relative importance of the eddy heat and momentum
fluxes. The sign of the vertical component of E is positive when the arrow points toward
high pressure (downward). The divergence of E:
∂(u′v′) f ∂((′)
·E =
∂y p ∂p
is zero for steady, conservative finite-amplitude wave disturbances of the zonal wind. For
atmospheric cross-sections, the terms in both equations incorporate r cos, where r
the Earth’s radius and latitude to allow for spherical geometry.
Edmon et al. (1980) show that # E represents the magnitude of transient eddy processes
at each latitude and vertical level; it is a direct measure of the total eddy forcing of the
zonal-mean state. The divergence of E is also shown to be proportional to the northward
flux of quasi-geostrophic potential vorticity, which involves only horizontal advection.
Cross-sections illustrating the contribution of transient waves in winter and summer
(1966–77) are shown in Figure 4.55. The upward E flux, especially in winter, has a pattern
resembling that for simulated cases of non-linear baroclinic wave instability. The equiv-
alent cross-sections for stationary eddies (not shown) have E fluxes less than half those
of the transient ones.
Extensions of the basic approach have been used to study local interactions between
transient eddies and the time-mean zonal flow (Hoskins et al., 1983). Trenberth (1986)
defines appropriate horizontal components of E by:
1 2
(v ′ u ′2) i u ′v ′ j
2
Arrows of E directed from weak to strong values of the zonal velocity indicate positive
net growth of eddy energy. The divergence ( # E) and curl (k # E) describe the eddy-
induced accelerations of the zonal and meridional wind components due to barotropic
processes. For the barotropic case, the group velocity of the transient eddies relative to
the local time-mean flow is parallel to E; the shape and horizontal tilt of the eddy can
be inferred from the major or minor axis of the anisotropic part of the eddy stress tensor
which subtends with the zonal direction an angle equal to twice the corresponding angle
for the E vector (Hoskins et al., 1983).
The E flux is a zonally averaged quantity, but an extension of this diagnostic tool
enables determination of quasi-geostrophic stationary waves on zonal flow (Plumb, 1985).
To illustrate this, Figure 4.56 shows the wave activity flux, Fs , for the northern hemi-
sphere stationary wave field at 500 mb averaged over ten winters. There are two distinct
wave trains, one from eastern Asia across the North Pacific and a weaker one eastward
from eastern North America, with a smaller feature emanating from western North
America. The vertical component, shown by contours in Figure 4.56, is directed upward
(Figure 4.57). It is noteworthy that there is no evidence for the propagation of stationary
waves out of the tropics. Moreover the apparent origins of the wave trains are not closely
related to orography, at least in the case of the Rocky Mountains. Plumb (1985) suggests
that areas of strong diabatic heating gradients are important sources of the wave trains,
supplemented by interactions with transient eddies within the major hemispheric storm
tracks.
Appendix 4.3 Normal modes

Internal and surface wave motions in a fluid made up of two or more layers of different
density can be described by a set of equations known as the shallow water equations.
These are applicable to fluids where the horizontal scales are large compared with the
11
Figure 4.55 The contribution of transient waves to vertical and meridional fluxes for (a) winter
0 and (b) summer 1966–77. The Eliassen–Palm vectors (arrows) are in units of (a)
1.25 1015 m3 and (b) 5 1015 m3 for the arrow scale in the horizontal direction; a
vertical arrow of the same length in the vertical component is in m3 k Pa 80.4. The
units of m3 are equivalent to dimensions of energy/pressure. (From Edmon et al., 1980)
11
Figure 4.56 Total wave flux by the 500 mb stationary wave field in winter (1965–75); horizontal
component shown by arrows (see scale, lower right); vertical component shown by con-
tours plotted at values (n 1/2), where is the contour interval 4.12 102 m2 s2;
contours solid for n = 0. (From Plumb, 1985)
Figure 4.57 Longitude–height cross-section of wave flux along 45°N. (Scales are in the inset.)
Schematic topography is shown. (From Plumb, 1985)
11 vertical scale. Thus the hydrostatic equation is valid (Gill, 1982). In a closed basin system
with two layers of different density there are two modes of oscillation, corresponding to
the number of degrees of freedom. These modes are termed the normal modes of oscil-
lation; they behave independently of one another. Gill (1982, pp. 119–21) shows that, for
a fluid system with two layers of different densities that do not mix, the structure of the
two modes is determined by a quadratic equation:
ce4 gHce gg*H1 H2 0 (1)
where the total fluid depth H H1 H2, the subscripts 1 and 2 referring to the upper
0
and lower layers, respectively; g* is the reduced gravity g(2 1)/ 1; and ce is a
wave speed. For each of the two solutions (eigen values) ce2 of this equation there is a
normal mode structure expressed by the equation:
h (x, y, t) z*(x, y, t) (2)
where h is the upward displacement of the density interface, z* is the perturbed position
of the free surface, and is an eigen vector which is independent of x, y, and t and has
an appropriate value.
For ocean waves, the small vertical density differences permit approximations to be
0 made; g*/g ≈ 0.03. This leads to two distinct roots, ce2 in equation 1. The larger root:
冢
c12 ≈ gH 1
g*H1H2
gH2… 冣
and
z H1
≈ ;
h H2
0
the ratio of the horizontal velocities of the fluid layers:
u2 g*H1
=1
u1 gH…
As g*/g → 0 in the limit, the solution corresponds to the surface gravity wave for a fluid
of uniform density. This is conventionally termed the barotropic mode.
The smaller root (for small g*/g) is:
0 c12 = 冢g*HH H 冣冢1 g*H

1 2
gH … 冣
H
2
1 2
and
z* gH u H
≈ 2; 2 ≈ 1
h gH u1 H2
This is termed the baroclinic mode. For the ocean, c1 is about 2–3 m s1, corresponding
to an equilibrium depth He ce2/g of 0.5–1 m. In the atmospheric two-layer model c1 is
0 about 10–20 m s1 and H ≈ 10–50 m. If the lower layer is thick relative to the upper
layer (H2 ! H1), the equation for the smaller root becomes c1 ≈ g*H. Internal waves
travel more slowly than surface waves because: g* % g
11
Notes
1 The E vector technique gives information on wave structures, propagation, and net forcing of
time-mean flows by transient eddies. It is an extension of Eliassen–Palm flux vectors for zonal-
mean flows (see Appendix 4.2). Following Trenberth (1986), the components of the zonal flow
E vector (in pressure coordinates) are given by:
冤(v′ 2 u′ ) u′v′, f v′′
S 冥
2 2
Eu =
where u and v are the flow components, primes denote departures from time-mean values, =
potential temperature, and S is a measure of the horizontally and temporally averaged static
stability. The x component of the Eu vector is related to the horizontal asymmetry of the eddies,
the y component to horizontal phase variations. The divergence of the E vector ·Eu > 0 implies
a positive net (eastward) forcing of the mean flow (i.e. augmentation of the westerlies).
2 “For time-dependent flows the perturbation equations for instability literally describe the linear
error growth of disturbances superimposed on the basic flow. However, . . . the rapid growth
of particular instability modes was a precursor to subsequent dynamical development”
(Frederiksen and Bell, 1990).
3 Quasi-geostrophic flow is equivalent barotropic if there is no vertical phase tilt and zero thermal
advection (Hoskins and Pearce, 1983, p. 389).
4 Equatorial Kelvin waves are limited to low latitudes, are symmetrical about the equator, and affect
only the u component of motion (Figure 5.27). They propagate eastward and downward. For east-
erly basic flow, the Kelvin wave has a phase speed slow enough for it to appear quasi-stationary
(Webster, 1983). The wave is a solution of the planetary wave equations on a plane (where
there is no latitudinal variation of the Coriolis parameter (f ), i.e. (
f /
y = = 0), assuming con-
stant basic zonal flow. The amplitude of the wave is proportional to exp ( 0 y2/2c) where 0 is
the rate of change of f at the equator and c = the zonal phase speed relative to the mean flow.
Temperature oscillations lead oscillations of u by a quarter of a cycle, i.e. they are in quadrature
(Lindzen, 1967; Holton and Lindzen, 1968; Parker, 1973).
References
Adler, R.F. 1975. A comparison of the general circulations of the northern and southern hemi-
spheres based on satellite, multi-channel radiance data. Mon. Wea. Rev., 103: 52–60.
Ahlquist, J.E. 1982. Normal-mode Rossby waves: theory and observations. J. Atmos. Sci., 39 (2):
193–202.
Ahlquist, J.E. 1985. Climatology of normal mode Rossby waves. J. Atmos. Sci., 42 (19): 2059–68.
Allen, R.A., Fletcher, R., Holmboe, J., Namias, J. and Willett, H.C. 1940. Report on an Experiment
in Five-day Weather Forecasting. Pap. Phys. Met. 8 (3) (MIT and WHOI), 94 pp.
Anderssen, E.C. 1965. A study of atmospheric long waves in the Southern hemisphere, Notos, 14:
57–65.
Angell, J.K. 1992. Relation between the 300 mb north polar vortex and equatorial SST, QBO, and
sunspot number and the record contraction of the vortex in 1988–89. J. Climate, 5: 22–9.
Angell, J.K. and Korshover, J. 1978. The expanded north circumpolar vortex of 1976 and winter
of 1976–77, and attendant vortex displacement. Mon. Wea. Rev., 106: 137–42.
Anyamba, E.K. and Weare, B.C. 1995. Temporal variability of the 40–50 day oscillation in trop-
ical convection. Intl. J. Climatol., 15: 379–402.
Arai, Y. 1981. An observational study of transient and quasi-stationary waves in the lower half of
the troposphere. Geophys. Mag., 39 (4): 173–247.
Arkin, P.A. and Webster, P.J. 1985. Annual and interannual variability of tropical–extratropical
interaction: an empirical study. Mon. Wea. Rev., 113: 1510–22.
Ashe, S.M. 1978. The seasonal-mean planetary-scale waves: an introduction. In: General
Circulation: Theory, Modeling and Observations, NCAR/CQ-6 1978-ASP, National Center for
Atmospheric Research, Boulder CO, pp. 83–97.
Ashe, S.M. 1979. A non-linear model of the axially asymmetric flow induced by topography and
diabatic heating. J. Atmos. Sci., 36: 109–26.
Asnani, G.C. 1993. Tropical Meteorology. Indian Institute of Tropical Meteorology, Pune,
chapter 1.
11 Atkinson, B.W. (ed.) 1981a. Dynamical Meteorology. An Introductory Selection. Methuen, London,
228 pp.
Atkinson, B.W. 1981b. Atmospheric waves. In: B.W. Atkinson, ed., Dynamical Meteorology: An
Introductory Selection, Methuen, London, pp. 100–15.
Austin, J.F. 1980. The blocking of middle latitude westerly winds by planetary waves. Quart.
J. Roy. Met. Soc., 106: 327–50.
Barrett, E.E. 1958. Eccentric circumpolar vortices in a barotropic atmosphere. Tellus, 10: 395–400.
Barry, R.G. 1967. Models in meteorology and climatology. In: R.J. Chorley and P. Haggett, eds,
Models in Geography, Methuen, London, pp. 97–144.
Barry, R.G. and Chorley, R.J. 1998. Atmosphere, Weather and Climate, 7th edn. Routledge, London.
0 Barry, R.G. and Perry, A.H. 1973. Synoptic Climatology: Methods and Applications. Methuen,
London, 555 pp.
Berggren, R., Bolin, B., and Rossby, C.-G. 1949. An aerological study of zonal motion, its pertur-
bations and breakdown. Tellus, 1: 14–37.
Blackburn, M. 1985. Interpretation of ageostrophic winds and implications for jetstream mainte-
nance. J. Atmos. Sci., 42 (23): 2604–20.
Blackmon, M.L. 1976. A climatological spectral study of the 500 mb geopotential height of the
northern hemisphere. J. Atmos. Sci., 33 (8): 1607–23.
Blackmon, M.L., Wallace, J.M., Lau, N.C., and Mullen, S.L. 1977. An observational study of the
Northern hemisphere wintertime circulation. J. Atmos. Sci., 34 (7): 1040–53.
Bluestein, H.B. 1992. Synoptic–Dynamic Meteorology in Midlatitudes. II. Observations and Weather
0 Systems. Oxford University Press, Oxford, 594 pp.
Bolin, B. 1952. Studies of the general circulation of the atmosphere. Adv. Geophys., 1: 87–118.
Böttger, H. and Fraedrich, K. 1980. Disturbances in the wavenumber–frequency domain observed
along 50°N. Contrib. Atmos. Phys., 53: 90–105.
Bradbury, D.L. 1958. On the behavior patterns of cyclones and anticyclones as related to zonal
index. Bull. Amer. Met. Soc., 39: 149–51.
Branstator, G. 1987. A striking example of the atmosphere’s leading traveling pattern. J. Atmos.
Sci., 44 (16): 2310–23.
Bryson, R.A. 1994. The discovery of the jetstream. Wisc. Acad. Rev., summer: 15–17.
Burnett, A.W. 1993. Size variations and long-wave circulation within the January northern hemi-
sphere circumpolar vortex, 1946–89. J. Climate, 6: 1914–20.
0 Buzzi, A. and Tibaldi, S. 1977. Inertial and frictional effects on rotating and stratified flow over
topography. Quart. J. Roy. Met. Soc., 103: 135–50
Carleton, A.M. 1979. A synoptic climatology of satellite-observed extratropical cyclone activity for
the southern hemisphere winter. Arch. Met. Geophys. Bioklim., B27: 265–79.
Carleton, A.M. 1981. Monthly variability of satellite-derived cyclonic activity for the southern hemi-
sphere winter. J. Climatol., 1: 21–38.
Carleton, A.M. 1989. Antarctic sea ice relationships with indices of the atmospheric circulation of
the southern hemisphere. Clim. Dynam., 2: 207–20.
Chapman, S. and Lindzen, R.S. 1970. Atmospheric Tides. Reidel, Dordrecht, 200 pp.
Charney, J.G. and DeVore, J.G. 1979. Multiple flow equilibria in the atmosphere and blocking.
J. Atmos. Sci., 36: 1205–16.
0 Charney, J.G. and Drazin, P.G. 1961. Propagation of planetary-scale disturbances from the lower
to the upper atmosphere. J. Geophys. Res., 66: 83–109.
Charney, J.G. and Eliassen, A. 1949. A numerical method for predicting the perturbations of the
middle latitude westerlies. Tellus, 1: 38–54.
Charney, J.G., Shukla, J., and Mo, K.C. 1981. Comparison of barotropic blocking theory and obser-
vation. J. Atmos. Sci., 38: 762–79.
Chen, S.S. and Houze, R.A., Jr. 1997. Diurnal variation and life cycle of deep convective systems
over the tropical Pacific warm pool. Quart. J. Roy. Met. Soc., 123: 359–88.
Chen, T.-C., Chen, J.M., Pfaendtner, J., and Susskind, J. 1995. The 12–24 day mode of global
precipitation. Mon. Wea. Rev., 123 (1): 140–52.
Cheng, X. and Wallace, J.M. 1993. Cluster analysis of the northern hemisphere wintertime 500 hPa
0 height field: spatial patterns. J. Atmos. Sci., 50: 2674–96.
Climate Diagnostics Center, CIRES, University of Colorado.
Colucci, S.J. and Alberta, T.L. 1996. Planetary-scale climatology of explosive cyclogenesis and
11 blocking. Mon. Wea. Rev., 124 (11): 2509–20.
Cook, K.H. 1999. Generation of the African Easterly Jet and its role in determining West African
preciptation. J. Climate, 12 (5): 1165–84.
Coughlan, M. 1983. Comparative climatology of blocking action in the two hemispheres. Austral.
Met. Mag., 31: 3–14.
Davis, R.E. and Benkovic, S.R. 1992. Climatological variations in the northern hemisphere circum-
polar vortex in January. Theor. Appl. Climatol., 46: 63–73.
Davis, R.E. and Benkovic, S.R. 1994. Spatial and temporal variations of the January circumpolar
vortex over the northern hemisphere. Intl. J. Climatol., 14: 415–28.
Defant, F. 1954. Über den Mechanismus der unperiodischen Schwankungen der allgemeinen
Zirkulation der Nordhalbkügel. Archiv. Met. Geophys. Biokl., A6: 253–79.
Dickinson, R.E. 1978. On planetary waves. In: The General Circulation: Theory, Modeling and
Observations. National Center for Atmospheric Research, Boulder CO, pp. 59–82.
Dickinson, R.E. 1980. Planetary waves: theory and observations. In: R. Hide and P.W. White, eds,
Orographic Effects in Planetary Flows, GARP Publ. Series 23, pp. 1–49, World Meteorological
Organization, Geneva.
Ding, Y.-H. 1994. Monsoons over China. Kluwer, Dordrecht, 432 pp.
Dole, R.M. 1978. The objective representation of blocking patterns. In: The General Circulation:
Theory, Modeling, and Observations. Notes from a Colloquium, Summer 1978, NCAR/CQ-
61978-ASP, NCAR, Boulder CO, pp. 406–26.
Dole, R.M. 1983. Persistent anomalies of the extratropical northern hemisphere wintertime circu-
lation. In: B.J. Hoskins and R.P. Pearce, eds, Large-scale Dynamical Processes in the Atmosphere,
Academic Press, New York, pp. 95–109.
Dole, R.M. 1986. The life cycle of persistent anomalies and blocking over the North Pacific. Adv.
Geophys., 29: 31–69.
Dole, R.M. 1987. Persistent large-scale flow anomalies. Part I. Characteristics of development. Part
II. Relationships to variations in synoptic-scale eddy activity and cyclogenesis. In: Proceedings
of 1987 ECMWF Workshop on the Nature and Prediction of Extratropical Weather Systems, II,
pp. 27–72, 73–122. European Center for Medium Range Weather Forecasts, Reading, England.
Dole, R.M. 1996. Blocking. In: S.H. Schneider, ed., Encyclopedia of Climate and Weather, Oxford
University Press, New York, pp. 93–9.
Dole, R.M. and Gordon, N.D. 1983. Persistent anomalies of the extratropical northern hemisphere
wintertime circulation: geographical distribution and regional persistence characteristics. Mon.
Wea. Rev., 111: 1567–86.
Durran, D.R. 1990. Mountain waves and downslope winds. In: W. Blumen, ed., Atmospheric Pro-
cesses over Complex Terrain, Meteorol. Monogr. 23 (45), Amer. Met. Soc., Boston MA, pp. 59–81.
Eady, E.T. 1950. The cause of the general circulation of the atmosphere. In: Centenary Proceedings
of the Royal Meteorological Society, London, pp. 156–72.
Eady, E.T. 1953. The maintenance of the mean zonal surface currents. In: Proceedings, Toronto
Meteorological Conference, pp. 124–28.
Edmon, H.J., Jr, Hoskins, B.J., and McIntyre, M.E. 1980. Eliassen–Palm cross-sections for the
troposphere. J. Atmos. Sci., 37: 2600–16 (also corrigendum, ibid., 38: 1115).
Egger, J. 1978. Dynamics of blocking highs. J. Atmos. Sci., 35: 1788–801.
Eliassen, E. 1958. A study of long atmospheric waves on the basis of zonal harmonic analysis.
Tellus, 10: 206–15.
Eliassen, E. and Machenhauer, B. 1965. A study of the fluctuations of the atmospheric planetary
flow patterns represented by spherical harmonics. Tellus, 17: 220–38.
Eliassen, E. and Machenhauer, B. 1969. On the observed large-scale atmospheric wave motion.
Tellus, 21: 149–66.
Eliassen, A. and Palm, E. 1961. On the transfer of energy in stationary mountain waves. Geophys.
Publ., 22 (3): 1–23.
Elliott, R.D. and Smith, T.B. 1949. A study of the effect of large blocking highs on the general
circulation in the northern hemisphere westerlies. J. Met., 6: 67–85.
Epstein, E.S. 1988. A spectral climatology. J. Climate, 1 (1): 88–107.
Essex, C., Lookman, T., and Nerenberg, M.A.H. 1987. The climate attractor over short time scales.
Nature, 326: 64–6.
Feldstein, S. and Lee, S.Y. 1998. Is the atmospheric zonal index driven by an eddy feedback?
J. Atmos. Sci., 55 (19): 3077–86.
11 Feldstein, S.B. and Held, I.M. 1989. Barotropic decay of baroclinic waves in a two-layer beta plane
model. J. Atmos. Sci., 46: 3416–30.
Flohn, H. 1968. Contributions to a Meteorology of the Tibetan Highlands. Atmos. Sci. Pap. 130,
Colorado State University, Fort Collins CO, 120 pp.
Fogarasi, S. and Strome, M. 1978. Computer Program for Calculating Atmospheric Planetary
Waves. Inland Waters Directorate, Environmental Canada, 8 pp.
Forsdyke, A.G. 1951. Zonal and other indices. Met. Mag., 80: 151–61.
Fraedrich, K. and Böttger, H. 1978. A wavenumber–frequency analysis of the 500 mb geopotential
at 50°N. J. Atmos. Sci., 35 (4): 745–50.
Fraedrich, K. and Kietzig, E. 1983. Statistical analysis and wavenumber–frequency spectra of the
0 500 mb geopotential along 50°N. J. Atmos. Sci., 40 (4): 1037–45.
Fraedrich, K., McBride, J.L., Frank, W.M., and Wang, R.S.L. 1997. Extended EOF analysis of trop-
ical distubances: TOGA COARE. J. Atmos. Sci., 54 (19): 2363–72.
Frederiksen, J.S. 1982. A unified three-dimensional instability theory of the onset of blocking and
cyclogenesis. J. Atmos. Sci., 39: 969–87.
Frederiksen, J.S. and Bell, R.C. 1987. Teleconnection patterns and the roles of baroclinic, barotropic
and topographic instability. J. Atmos. Sci., 44: 2200–18.
Frederiksen, J.S. and Bell, R.C. 1990. North Atlantic blocking during January 1989: linear theory.
Frederiksen, J.S. and Webster, P.J. 1988. Alternative theories of atmospheric teleconnections and
low-frequency fluctuations. Rev. Geophys., 26: 459–94.
0 Fultz, D. and Kaylor, R. 1959. The propagation of frequency in experimental baroclinic waves in
a rotating annular ring. In: B. Bolin, ed., The Atmosphere and the Sea in Motion (Rossby memo-
rial volume), New York, pp. 359–71.
Geb, M. 1966. Synoptische-statistische Untersuchungen zur Einleitung blockierender Hoch-
drucklagen über dem Nordatlantik und Europa. Met. Abhandl., 69 (1): 94 pp.
Ghil, M. 1987a. Predictability of planetary flow regimes: dynamics and statistics. In: U. Radok,
ed., Toward Understanding Climate Change, Westview Press, Boulder Co, pp. 91–147.
Ghil, M., 1987b: Dynamics, statistics and predictability of planetary flow regimes, In C. Nicolis
and G. Nicolis, eds, Irreversible Phenomena and Dynamical Systems Analysis in the Geosciences,
Reidel, Dordrecht, pp. 241–83.
Ghil, M. and Mo, K.-C. 1991. Intraseasonal oscillations in the global atmosphere. I. Northern hemi-
0 sphere and tropics. II. Southern hemisphere. J. Atmos. Sci., 48 (5): 752–79, 780–90.
Gill, A.E. 1982. Atmosphere–Ocean Dynamics. Academic Press, San Diego CA, 662 pp.
Green, J.S.A. 1977. The weather during July 1976: some dynamical considerations of the drought.
Weather, 32: 120–6.
Grotjahn, R. 1993. Global Atmospheric Circulations: Observations and Theory. Oxford University
Press, New York, 430 pp.
Haines, K. and Marshall, J. 1987. Eddy-forced coherent structures as a prototype of atmospheric
blocking. Quart. J. Roy. Met. Soc., 113: 681–704.
Hann, J. 1889. Untersuchungen über die tägliche Oscillation des Barometers. Denkschr. Akad. Wiss.,
Wien, Math.-Nat. Kl. 55: 49–121.
Hansen, A.R. 1986. Observational characteristics of atmospheric planetary waves with bimodal
0 amplitude distributions. Adv. Geophys., 29: 101–33.
Hansen, A.R. 1988. Further observational characteristics of bimodal planetary waves: mean struc-
ture and transitions. Mon. Wea. Rev., 116: 386–400.
Hansen, A.R. and Chen, T.-C. 1982. A spectral energetics analysis of atmospheric blocking. Mon.
Wea. Rev., 110: 1146–65.
Hansen, A.R. and Sutera, A. 1987. The probability density distribution of the speed and horizontal
and vertical shear of the zonal-mean flow. J. Atmos. Sci., 44: 1525–33.
Hare, F.K. 1962. The stratosphere. Geogr. Rev., 52: 525–47.
Harman, J.R. 1991. Synoptic Climatology of the Westerlies: Process and Patterns. Assoc. Amer.
Geographers, Washington DC, 80 pp.
Hartmann, D.L. and Ghan, S.J. 1980. A statistical study of the dynamics of blocking. Mon. Wea.
0 Rev., 108: 1144–59.
Hartmann, D.L. and Michelsen, M.L. 1989. Intraseasonal periodicities in Indian rainfall. J. Atmos.
Sci., 46: 2838–62.
11
Haurwitz, B. 1940. The motion of atmospheric disturbances on the spherical earth. J. Mar. Res.,
3: 254–67.
Haurwitz, B. 1941. Dynamic Meteorology. McGraw-Hill, New York, 365 pp.
Haurwitz, B. 1965. The diurnal surface pressure oscillation. Archiv. Met. Geophys. Biokl., A14: 361–79.
Haurwitz, B. and Möller, F. 1955. The semi-diurnal and temperature variation and the solar air
tide. Archiv. Met. Geophys. Biokl., A8: 332–56.
Held, I.M. 1983. Stationary and quasi-stationary eddies in the extratropical troposphere: theory. In:
B.J. Hoskins and R.P. Pearce, eds, Large-scale Dynamical Processes in the Atmosphere, Academic
Press, London, pp. 127–88.
Hendon, H.H. and Liebmann, B. 1990. The intraseasonal (30–50 day) oscillation of the Australian
summer monsoon. J. Atmos. Sci., 47: 2909–23.
Hendon, H.H. and Salby, M.L. 1994. The life cycle of the Madden–Julian oscillation. J. Atmos.
Sci., 51 (15): 2225–37.
Hide, R. 1970. Some laboratory experiments on free thermal convection in a rotating fluid subject
to a horizontal temperature gradient and their relation to the theory of the global atmospheric
circulation. In: G.A. Corby, ed., The Global Circulation of the Atmosphere, Royal Meteorol. Soc.,
London, pp. 196–221.
Hide, R. 1985. Dynamics of rotating fluids and planetary atmospheres. In: Recent Advances in
Meteorology and Physical Oceanography, Royal Meteorol. Soc., London, pp. 37–45.
Holton, J.R. 1973. On the frequency distribution of atmospheric Kelvin waves. J. Atmos. Sci., 30
(3): 499–501.
Holton, J.R. 1993. The Second Haurwitz Memorial Lecture: Stationary Planetary Waves. Bull. Amer.
Met. Soc., 74: 1735–42.
Holton, J. and Lindzen, R. 1968. A note on Kelvin waves in the atmosphere. Mon. Wea. Rev., 96: 385–6.
Horel, J.D. 1985a. Persistence of the 500 mb height field during northern hemisphere winter. Mon.
Wea. Rev., 113 (11): 2030–42.
Horel, J.D. 1985b. Persistence of wintertime 500 mb height anomalies over the central Pacific. Mon.
Wea. Rev., 113 (11): 2043–8.
Hoskins, B.J. 1980. Representation of the earth’s topography using spherical harmonics. Mon. Wea.
Rev., 108: 111–15.
Hoskins, B.J. and Karoly, D.J. 1981. The steady linear response of a spherical atmosphere to thermal
and orographic forcing. J. Atmos. Sci., 38 (6): 1179–96.
Hoskins, B.J. and Pearce, R.P. (eds). 1983. Large-scale Dynamical Processes in the Atmosphere.
Academic Press, London, 397 pp.
Hoskins, B.J., Hsu, H.H., James, I.N., Masutani, M., Sardeshmukh, P.D., and White, G.H. 1989.
Diagnostics of the Global Atmospheric Circulation based on ECMWF Analyses, 1979–1989.
WCRP-27, WMO/TD No. 326, World Meteorological Organization, Geneva.
Hoskins, B.J., James, I.N., and White, G.H. 1983. The shape, propagation and mean flow interac-
tion of large-scale weather systems. J. Atmos. Sci., 40: 1595–612.
Hoskins, B.J., Simmons, A.J., and Andrews, D.G. 1977. Energy dispersion in a barotropic atmos-
phere. Quart. J. Roy. Met. Soc., 103: 553–67.
Hurrell, J.W., van Loon, H., and Shea, D.J. 1998. The mean state of the troposphere. In: D.J. Karoly
and D.G. Vincent, eds, Meteorology of the Southern Hemisphere, Meteorol. Monogr. 27 (49),
Amer. Met. Soc., Boston MA, pp. 1–46.
Ilari, L., Malguzzi, P., and Speranza, A. 1981. On breakdowns of the westerlies. Geophys. Astrophys.
Fluid Dynam., 17: 27–49.
Jacqmin, D. and Lindzen, R.S. 1985. The causation of sensitivity of the northern hemisphere plan-
etary waves. J. Atmos. Sci., 42: 724–45.
James, I.N. 1988. On the forcing of planetary-scale Rossby waves by Antarctica. Quart. J. Roy.
Met. Soc., 114: 619–38.
James, I.N. 1994. Introduction to Circulating Atmospheres, Cambridge University Press, Cambridge,
pp. 171–84.
Johansen, H. 1958. On continental and oceanic influences on the atmosphere. Met. Ann. (Oslo), 4
(8): 143–58.
Jones, D.A. and Simmonds, T. 1993. Time and space spectral analysis of southern hemisphere sea
level pressure variability. Mon. Wea. Rev., 121 (3): 661–72.
Jones, P.D. and Mörth, H.T. 1978. A new approach to indexing the circumpolar wind circulation.
2. Climate Monitor, 7: 54–63.
11 Jones, P.D., Salinger, M.J., and Mulland, A.B. 1999. Extratropical circulation indices in the southern
hemisphere based on station data. Intl. J. Climatol., 19 (12): 1302–17.
Jonsson, P. and Börring, L. 1994. Zonal index variations, 1899–1992: links to air temperature in
southern Scandinavia. Geogr. Annal., A76: 207–20.
Julian, P.R. 1966. Index cycle: a cross-spectral analysis of zonal index data. Mon. Wea. Rev., 94:
283–93.
Kaas, E. and Branstator, G. 1993. The relationship between a zonal index and blocking activity. J.
Atmos. Sci., 50 (18): 3061–77.
Karoly, D.J. 1978. Rossby wave ray paths and horizontal wave propagation. In: General Circulation:
Theory, Modeling and Observations, NCAR/CQ-6 1978-ASP, National Center for Atmospheric
0 Research, Boulder CO, pp. 474–84.
Karoly, D.J. 1985. An atmospheric climatology of the southern hemisphere based on ten years of
daily numerical analyses, 1972–82. II. Standing wave climatology. Austral. Met. Mag., 33:
105–16.
Keen, R.A. 1987. Equatorial westerlies and Southern Oscillation. In: R. Lukas and D.J. Webster,
eds, Proc. of the US–TOGA Western Pacific Air–Sea Interaction Workshop, Honolulu HI,
pp. 121–40.
Kelvin, Lord (Thomson, W.) 1882. On the thermodynamic acceleration of the earth’s rotation. Proc.
Roy. Soc. Edinburgh, 11: 396–405.
Kidson, J.W. 1985. Index cycles in the northern hemisphere during the Global Weather Experiment.
Mon. Wea. Rev., 113: 607–23.
0 Kidson, J.W. 1988. Indices of the southern hemisphere zonal wind. J. Climate, 1: 183–94.
Kidson, J.W. 1991. Intraseasonal variations in the southern hemisphere circulation. J. Climate, 4:
939–53.
Kikuchi, Y. 1971. Influence of mountains and land–sea distribution on blocking action. J. Met. Soc.
Japan, Ser. II, 49 (special issue): 564–72.
Kimoto, M. and Ghil, M. 1993. Multiple flow regimes in the northern hemisphere winter. I.
Methodology and hemispheric regimes. J. Atmos. Sci., 50: 2625–43.
Kington, J.A. 1999. W. Clement Ley: nineteenth-century cloud study and the European jetstream.
Bull. Amer. Met. Soc. 80 (5): 901–3.
Kletter, L. 1962. Die Aufeinanderfolge charakteristische Zirkulationstypen in mittleren Breiten der
nordlichen Hemisphäre. Arch. Met. Geophys. Biokl., A13: 1–33.
0 Knox, J.L. and Hay, J.E. 1985. Blocking signatures in the northern hemisphere: frequency distrib-
ution and interpretation. J. Climate, 5: 1–16.
Kocin, P.J. and Uccellini, L.W. 1990. Snowstorms along the northeastern coast of the United States,
Met. Monogr., No. 44, Amer. Met. Soc., Boston, MA.
Koteswaram, P. 1958. Easterly jetstreams in the tropics. Tellus, 10: 43–57.
Kozuchowski, K.M. 1993. Variations of hemispheric zonal index since 1899 and its relationships
with air temperature. Intl. J. Climatol., 13: 853–64.
Krishnamurti, T.N. 1961. The subtropical jetstream of winter. J. Met., 18 (2): 172–91.
Krishnamurti, T.N. 1979. Compendium of Meteorology II, Part 4, Tropical Meteorology, WMO
Report No. 364, WMO, Geneva.
Kung, E.C., Tanaka, H.L., and Baker, W.E. 1989. Energetics examination of winter blocking simu-
0 lations in the northern hemisphere. Mon. Wea. Rev., 19: 2019–40.
Kushnir, Y. 1987. Retrograding wintertime low-frequency disturbances over the North Pacific Ocean.
J. Atmos. Sci., 44 (19): 2727–42.
Kutzbach, J.E., Guetter, P.J., Ruddiman, W.F., and Prell, W.L. 1989. Sensitivity of climate to late
Cenozoic uplift in southern Asia and the American west: numerical experiments. J. Geophys.
Res., 94: 18393–407.
La Seur, N.E. 1954. On the asymmetry of the middle-latitude circumpolar current. J. Met., 11:
43–57.
Laplace, P.S. 1799–1825. Mécaniques Célestes, 5 volumes, Paris.
Lau, N.-C. 1979. The observed structure of tropospheric stationary waves and local balances of
vorticity and heat. J. Atmos. Sci., 36: 996–1016.
0 Lee, S.-Y. 1999. Why are the climatological zonal winds easterly in the equatorial upper tropo-
sphere? J. Atmos. Sci., 56 (10): 1353–63.
Legras, B. and Ghil, M. 1985. Persistent anomalies, blocking and variations in atmospheric vari-
11 ability. J. Atmos. Sci., 42: 433–71.
Lejenäs, H. 1984. Characteristics of southern hemisphere blocking as determined from a time series
of observational data. Quart. J. Roy. Met. Soc., 44: 2575–625.
Lejenäs, H. 1987. A comparative study of southern hemisphere blocking during the Global Weather
Experiment. Quart. J. Roy. Met. Soc., 113: 181–8.
Lejenäs, H. and Madden, R.A. 1992. Traveling planetary scale waves and blocking. Mon. Wea.
Rev., 120 (12): 2821–30.
Lejenäs, H. and Økland, H. 1983. Characteristics of northern hemisphere blocking as determined
from a long time series of observational data. Tellus, 35A: 350–62.
LeMarshall, J.F., Kelly, G.A.M., and Karoly, D.J. 1985. An atmospheric climatology of the southern
hemisphere based on ten years of daily numerical analyses, 1972–82. I. Overview. Austral. Met.
Mag., 33: 65–85.
Lindzen, R.S. 1967a. Thermally driven diurnal tide in the atmosphere. Quart. J. Roy. Met. Soc.,
93: 18–42.
Lindzen, R. 1967b. Planetary waves on beta-planes. Mon. Wea. Rev., 95: 441–51.
Lindzen, R.S. 1978. Effect of daily variation of cumulonimbus activity on the atmospheric semi-
diurnal tide. Mon. Wea. Rev., 106 (4): 526–32.
Lindzen, R.S. 1986. Stationary planetary waves, blocking, and interannual variability. In: R. Benzi
and A.C. Winn-Nielsen, eds, Anomalous Atmospheric Flows and Blocking, Adv. Geophys., 29:
251–73.
Lindzen, R.S. and Hou, A.Y. 1988. Hadley circulations for zonally averaged heating centered off
the equator. J. Atmos. Sci., 45: 2416–27.
Liu, Q. 1994. On the definition and persistence of blocking. Tellus, 46A: 286–98
Lorenz, E.N. 1952. Flow of angular momentum as a predictor for the zonal westerlies. J. Met., 9:
152–7.
Lorenz, E.N. 1962. The statistical predictions of solutions of dynamic equations. In: Proc. Internat.
Sympo. Numerical Weather Prediction, Japan Met. Soc., Tokyo, pp. 629–35.
Lorenz, E.N. 1963. Deterministic nonperiodic flow. J. Atmos. Sci., 20: 130–41.
Lorenz, E.N. 1969a. The predictability of a flow which possesses many scales of motion. Tellus,
21: 89–307.
Lorenz, E.N. 1986. The index cycle is alive and well. In: J.O. Roads, ed., Namias Symposium,
Scripps Inst. Oceanog., Ref. Series 86–17, University of California, San Diego CA, pp. 188–96.
Luo, H. and Yanai, M. 1984. The large-scale circulation and heat sources over the Tibetan Plateau
and surrounding areas during the early summer of 1979. II. Heat and moisture budgets. Mon.
Wea. Rev., 112 (5): 966–89.
Lupo, A.R. 1997. A diagnosis of two blocking events that occurred simultaneously in the midlat-
itude northern hemisphere. Mon. Wea. Rev., 125 (8): 1801–23.
Lupo, A.R. and Bosart, L.F. 1999. An analysis of a rare case of continental blocking. Quart.
J. Roy. Met. Soc., 125: 107–38.
Lupo, A.R. and Smith, P.J. 1994. An investigation of observed mid-latitude blocking characteris-
tics in the northern hemisphere. Sixth Conference on Climatic Variations. Preprints. Amer. Met.
Lupo, A.R. and Smith, P.J. 1995a. Climatological features of blocking anticyclones in the northern
Lupo, A.R. and Smith, P.J. 1995b. Planetary and synoptic scale disturbances during the life cycle
of a mid-latitude blocking anticyclone over the North Atlantic. Tellus, 47A: 575–96.
Madden, R.A. 1979. Observations of large-scale traveling Rossby waves. Rev. Geophys., 17: 1935–49.
Madden, R.A. and Julian, P.R. 1971. Description of a 40–50 day oscillation in the zonal winds in
the tropical Pacific. J. Atmos. Sci., 28 (5): 702–8.
Madden, R.A. and Julian, P.R. 1972. Description of global-scale circulation cells in the tropics with
a 40–50 day period. J. Atmos. Sci., 29 (6): 1109–23.
Madden, R.A. and Julian, P.R. 1994. Observations of the 40–50 day tropical oscillation – a review.
Mon. Wea. Rev., 122 (5): 814–37.
Madden, R.A. and Lejenäs, H. 1989. Flow at 500 mb associated with a measure of persistence over
western Europe. Mon. Wea. Rev., 117: 2843–54.
Madden, R.A. and Speth, P. 1989. The average behavior of large-scale westward traveling distur-
bances evident in the northern hemisphere geopotential heights. J. Atmos. Sci., 46 (21): 3225–39.
Mahringer, G. and Zwatz-Meise, V. 1993. A semi-operational diagnosis method. Contrib. Atmos.
Phys., 66 (1–2): 89–106.
11 Manabe, S. and Terpstra, T.B. 1974. The effects of mountains on the general circulation of the
atmosphere as identified by numerical experiments. J. Atmos. Sci., 31 (1): 3–42.
Markham, C.G. 1985. A quick and direct method for estimating mean monthly global temperatures
from 500 mb data. Prof. Geogr., 37: 72–4.
Matthews, A.J. and Kiladis, G.N. 1999. The tropical–extratropical interaction between high-
frequency transients and the Madden–Julian Oscillation. Mon. Wea. Rev., 127 (5): 661–77.
McWilliams, J.C. 1980. An application of equivalent modons to atmospheric blocking. Dyn. Atmos.
Ocean., 5: 43–66.
Michelangeli, P.-A., Vautard, R., and Legras, B. 1995. Weather regimes: recurrence and quasi-
stationarity. J. Atmos. Sci., 52 (8): 1237–56.
0 Miles, M.K. 1977. The annual course of some indices of the zonal and meridional circulation in
middle latitudes of the northern hemisphere. Met. Mag., 106: 52–66.
Minz, Y. and Kao, S.-K. 1952. A zonal index tendency equation and its application to forecasts of
the zonal index. J. Met., 9: 87–92.
Mo, K.C. and Ghil, M. 1988. Cluster analysis of multiple planetary flow regimes. J. Geophys. Res.,
93 (D): 10927–52.
Molteni, F. and Tibaldi, S. 1990. Regimes in the wintertime circulation over northern extratropics.
II. Consequences for dynamical predictability. Quart. J. Roy. Met. Soc., 116: 1263–88.
Molteni, F., Sutera, A., and Tronci, N. 1988. EOFs of the geopotential eddies at 500 mb in winter
and their probability density distribution. J. Atmos. Sci., 45: 3063–80.
Molteni, F., Tibaldi, S., and Palmer, T.N. 1990. Regimes in the wintertime circulation over northern
0 extratropics. I. Observational evidence. Quart. J. Roy. Met. Soc., 116: 31–67.
Mörth, H.T. 1977. A new approach to indexing the circumpolar upper wind circulation. I. Clim.
Monitor, 6 (5): 158–62.
Mullen, S.L. 1987. Transient eddy forcing of blocking flows. J. Atmos. Sci., 44: 3–22.
Müller, K., Buchwald, K., and Fraedrich, K. 1979. Further studies on single station climatology.
1. The summer confluence of subtropic and polar front jet. 2. The two northern Cold Poles. Beitr.
Phys. frei. Atmos., 52: 362–73.
Nakamura, H. and Wallace, J.M. 1990. Observed changes in baroclinic wave activity during the
0 life cycles of low-frequency circulation anomalies. J. Atmos. Sci., 47: 1100–16.
Nakamura, H., Nakamura, M., and Anderson, J.L. 1997. The role of high- and low-frequency
dynamics in blocking formation. Mon. Wea. Rev., 125 (9): 2074–93.
Nakazawa, T. 1988. Tropical superclusters within intraseasonal variations over the western Pacific.
J. Met. Soc. Japan, 66: 823–39.
Namias, J. 1950. The index cycle and its role in the general circulation. J. Met., 7: 130–9.
Namias, J. and Clapp, P.F. 1949. Confluence theory of the high-tropospheric jetstream. J. Met., 6:
330–6.
National Research Council 1975. Understanding Climate Change: A Program for Action, National
Academy of Sciences, Washington DC, 239 pp.
Neilley, P.P. and Dole, R.M. 1991. Interactions between synoptic-scale eddies and the large-scale
0 flow during the development of blocking over the North Atlantic Ocean. In: Proc. International
Symposium on Winter Storms. Preprint Volume. Amer. Met. Soc., Boston MA, pp. 50–5.
Newell, R.E. and Kidson, J.W. 1984. African mean wind changes between Sahelian wet and dry
periods. J. Climatol., 4: 27–33.
Oerlemans, J. 1978. An objective approach to breaks in the weather. Mon. Wea. Rev., 106: 1672–9.
Oort, A.H. and Peixoto, J.P. 1983. Global angular momentum and energy balance requirements
from observations. Adv. Geophys., 25: 355–490.
Panofsky, H.A. and Wolff, P. 1957. Spectrum and cross spectrum analysis of hemispheric westerly
index. Tellus, 9: 195–200.
Parker, D. 1973. Equatorial Kelvin waves at 100 millibars. Quart. J. Roy. Met. Soc., 99: 116–28.
Pedlovsky, J. 1987. Geophysical Fluid Dynamics. Springer-Verlag, New York, 710 pp.
0 Peixoto, J.P. and Oort, A.H. 1992. Physics of Climate. American Institute of Physics, New York,
520 pp.
Pittock, A.B. 1980. Patterns of climatic variation in Argentina and Chile. I. Precipitation. Mon.
11 Wea. Rev., 108: 1347–61.
Platzman, G. 1968. The Rossby wave. Quart. J. Roy. Met. Soc., 94: 225–48.
Plumb, R.A. 1985. On the three-dimensional propagation of stationary waves. J. Atmos. Sci., 42
(3): 217–29.
Quintanar, A.I. and Mechoso, C.R. 1995. Quasi-stationary waves in the southern hemisphere. I.
Observational data. II. Generation mechanisms. J. Climate, 8 (11): 2659–72; 2673–90.
Quiroz, R.S. 1987. Traveling waves and regional transitions in blocking activity in the northern
hemisphere. Mon. Wea. Rev., 115: 919–35.
Radok, U. (ed.). 1987. Toward Understanding Climate Change. Westview Press, Boulder CO, 200 pp.
Raynor, J.N. and Howarth, D.A. 1979. Antarctic sea ice, 1972–75. Geog. Rev., 69: 202–23.
Reiter, E.R. 1963. Jetstream Meteorology, University of Chicago Press, Chicago. 515 pp.
Reiter, E.R. 1996. Jetstream. In: S.H. Schneider, ed., Encyclopedia of Climate and Weather, Oxford
University Press, New York, pp. 455–9.
Rex, D.F. 1950. Blocking action in the middle troposphere and its effects upon regional climate.
Tellus, 2: 196–211; 275–301.
Riehl, H. 1962. Jetstreams of the Atmosphere. Tech. Rep. 32, Dept of Atmos. Sci., Colorado State
University, Fort Collins CO, 117 pp.
Riehl, H., Alaka, M.A., Jordan, C.L., and Renard, R.J. 1954. The Jet Stream. Met. Monogr., 2 (7),
100 pp.
Riehl, H., La Seur, N.E., et al. 1952. Forecasting in Middle Latitudes, Met. Monogr., 1 (5) Amer.
Met. Soc., Boston MA, 80 pp.
Robinson, W.A. 1991. The dynamics of the zonal index in a simple model of the atmosphere.
Tellus, 43A: 295–305.
Rogers, J.C. and Van Loon, H. 1982. Spatial variability of sea level pressure and 500 mb height
anomalies over the southern hemisphere. Mon. Wea. Rev., 110: 1375–92.
Rossby, C.-G. 1938. On the mutual adjustment of pressure and velocity distributions in certain
simple current systems. II. J. Mar. Res., 1: 239–63.
Rossby, C.-G. 1947. On the distribution of angular velocity in gaseous envelopes under the influ-
ence of large scale horizontal mixing processes. Bull. Amer. Met. Soc., 28: 53–68.
Rossby, C.-G., et al. 1939. Relation between variations in the intensity of the zonal circulation of the
atmosphere and the displacement of the semi-permanent centers of action. J. Marine Res., 2: 38–54.
Rossby, C.-G. and Willett, H.C. 1948. The circulation of the upper troposphere and lower stratos-
phere. Science, 108: 643–52.
Ruelle, D. 1980. Strange attractors. Math. Intelligencer, 3: 126–237.
Salby, M.L. 1984. Survey of planetary-scale traveling waves: the state of theory and observations.
Rev. Geophys. Space Phys., 22: 209–36.
Salby, M.L. and Hendon, H.H. 1994. Intraseasonal behavior of clouds, temperature, and motion in
the tropics. J. Atmos. Sci., 51 (15): 2207–24.
Saltzman, B. and Fleisher, A. 1960. The exchange of kinetic energy between larger scales of atmos-
pheric motion. Tellus, 12: 374–7.
Schilling, H.-D. 1986. On atmospheric blocking types and blocking numbers. Adv. Geophys., 29:
71–99.
Scorer, R.S. 1979. One further remark on blocking. Weather, 34: 361–3.
Seilkopf, H. 1939. Maritime Meteorologie. Handbuch der Fliegerwetterkunde. II. Radetzki Verlag,
Berlin, 150 pp.
Shiotani, M. 1990. Low-frequency variations of the zonal mean state of the southern hemisphere
troposphere. J. Met. Soc. Japan, 68: 461–71.
Shukla, J. and Mo., K.C. 1983. Seasonal and geographical variation of blocking. Mon. Wea. Rev.,
111: 388–402.
Shutts, G.J. 1978. Quasi-geostrophic planetary wave forcing. Quart. J. Roy. Met. Soc., 104: 331–50.
Shutts, G.J. 1983. The propagation of eddies in diffluent jetstreams: eddy vorticity forcing of
“blocking flow fields.” Quart. J. Roy. Met. Soc., 109: 737–62.
Shutts, G.J. 1986. A case study of eddy forcing during an Atlantic blocking episode. Adv. Geophys.,
29: 135–62.
Shutts, G.J. 1987a. Persistent anomalous circulation and blocking. Met. Mag., 116: 116–24.
Shutts, G.J. 1987b. Some comments on the concept of thermal forcing. Quart. J. Roy. Met. Soc.,
113: 1387–94.
Simmons, A.J., Wallace, J.M., and Branstator, G.W. 1983. Barotropic wave propagation and insta-
bility, and atmospheric teleconnection patterns. J. Atmos. Sci., 40: 1363–91.
11 Sinclair, M.R. 1997. Objective identification of cyclones and their circulation intensity and clima-
tology. Wea. Forecasting, 12: 595–612.
Smagorinsky, J. 1953. The dynamical influence of large-scale heat sources and sinks in the quasi-
stationary mean motions of the atmosphere. Quart. J. Roy. Met. Soc., 79: 342–66.
Smith, E.A. and Shi, L. 1996. Reducing discrepancies in atmospheric heat budget of Tibetan Plateau
by satellite-based estimates of radiative cooling and cloud–radiation feedback. Met. Atmos. Phys.,
56: 229–60.
Smith, R.B. 1979a. The influence of mountains on the atmosphere. Adv. Geophys., 21: 87–230.
Smith, R.B. 1979b. Some aspects of the quasi-geostrophic flow over mountains. J. Atmos. Sci., 36
(12): 2385–93.
0 Smith, R.F.T. 1973. A note on the relationship between large-scale energy functions and charac-
teristics of climate. Quart. J. Roy. Met. Soc., 99: 693–703.
Speth, P. and Madden, R.A. 1983. Space–time spectral analysis of northern hemisphere geopotential
heights. J. Atmos. Sci., 40 (5): 1086–100.
Speth, P., May, W., and Madden, R.A. 1992. The average behavior of large-scale westward-traveling
disturbances evident in the southern hemisphere geopotential heights. J. Atmos. Sci., 49 (2): 178–85.
Stark, L. 1965. Positions of the monthly mean troughs and ridges in the northern hemisphere,
1949–63. Mon. Wea. Rev., 93: 705–20.
Sui, C.-H. and Lau, K.-M. 1992. Multiscale phenomena in the tropical atmosphere over the western
Pacific. Mon. Wea. Rev., 120: 407–30.
Sutcliffe, R.C. 1951. Mean upper contour patterns of the northern hemisphere: the thermal synoptic
0 viewpoint. Quart. J. Roy. Met. Soc., 77: 435–40.
Sutera, A. 1986. Probability density distribution of large-scale atmospheric flow. Adv. Geophys.,
29: 227–49.
Tanaka, H.L. 1991. A numerical simulation of amplification of low-frequency planetary waves and
blocking formations by the upscale energy cascade. Mon. Wea. Rev., 119: 2919–35.
Tanaka, H.L. and Kung, E.C. 1989. A study of low-frequency unstable planetary waves in realistic
zonal and zonally-varying basic states. Tellus, 41A: 179–99.
Tarleton, L.F. 1986. Some characteristics of 500 mb blocking and zonal flows in the southern hemi-
sphere. Second International Conference on Southern Hemisphere Meteorology (Wellington, New
Zealand). Extended Abstracts, Amer. Met. Soc., Boston MA, pp. 274–7.
Tarleton, L.F. 1987. “Persistence Characteristics of 500 mb Blocking and Zonal Flows in the Middle
0 Latitudes of the Northern and Southern Hemispheres.” Ph.D. Dissertation, University of Colorado,
Boulder CO.
Thompson, P.D. 1961. Numerical Weather Analysis and Prediction, Macmillan, London, 170 pp.
Thorncroft, C.D. and Blackburn, M. 1999. Maintenance of the African Easterly Jet. Quart. J. Roy.
Met. Soc., 125: 763–86.
Tibaldi, S. and Molteni, F. 1990. On the operational predictability of blocking. Tellus, 42A: 343–65.
Tibaldi, S., Tosi, E., Navarra, A., and Pedulli, L. 1994. Northern and southern hemisphere vari-
ability of blocking frequency and predictability. Mon. Wea. Rev., 122 (9): 1971–2003.
Ting, M.-F., Hoerling, M.P., Xu, T.-Y. and Kumar, A. 1996. Northern hemisphere teleconnection
patterns during extreme phases of the zonal-mean circulation. J. Climate, 9 (10): 2614–33.
Treidl, R.A., Birch, E.C., and Sajecki, P. 1981. Blocking action in the northern hemisphere: a clima-
0 tological study. Atmos.-Ocean, 19: 1–23.
Trenberth, K.E. 1976. Fluctuations and trends in indices of the southern hemisphere circulation.
Trenberth, K.E. 1979. Interannual variability of the 500 mb zonal mean flow in the southern hemi-
sphere. Mon. Wea. Rev., 107 (11): 1515–24.
Trenberth, K.E. 1980. Planetary waves at 500 mb in the southern hemisphere. Mon. Wea. Rev.,
108: 1378–89.
Trenberth, K.E. 1981a. Interannual variability of the southern hemisphere 500 mb flow: regional
characteristics. Mon. Wea. Rev., 109: 127–36.
Trenberth, K.E. 1981b. Observed southern hemisphere eddy statistics at 500 mb: frequency and
spatial dependence. J. Atmos. Sci., 38 (12): 2585–605.
0 Trenberth, K.E. 1986. An assessment of the impact of transient eddies on the zonal flow during a
blocking episode using localized Eliassen–Palm flux diagnostics. J. Atmos. Sci., 43 (19): 2070–88.
Trenberth, K.E. 1987 The zonal mean westerlies over the southern hemisphere. Mon. Wea. Rev.,
11 115: 1528–33.
Trenberth, K.E. 1992. Global Analyses from ECMWF and Atlas of 1000 to 10 mb Circulation
Statistics, NCAR Tech. Note TN-373STR, National Center for Atmospheric Research, Boulder
CO.
Trenberth, K.E. and Mo, K. 1985. Blocking in the southern hemisphere. Mon. Wea. Rev., 113 (1):
3–21.
Tribbia, J.J. and Madden, R.A. 1988. Projection of time-mean geopotential heights on to normal,
Hough modes. Met. Atmos. Phys., 38: 9–21.
Tsou, C.H. and Smith, P.J. 1990. The role of synoptic/planetary-scale interactions during the devel-
opment of a blocking anticyclone. Tellus, 42A: 174–93.
Tung, K.K. and Lindzen, R.S. 1979. A theory of stationary long waves. 1. A simple theory of
blocking. 2. Resonant Rossby waves in the presence of realistic vertical shears. Mon. Wea. Rev.,
107: 714–34; 735–50.
University of Chicago, staff members, Meteorology Department. 1947. On the general circulation
of the atmosphere in middle latitudes. Bull. Amer. Met. Soc., 28: 255–80.
van den Dool, H.M. 1991. Mirror images of atmospheric flow. Mon. Wea. Rev., 119 (9): 2095–106.
van den Dool, H.M., Saha, S., Schemm, J., and Huang, J. 1997. A temporal interpolation method
to obtain hourly atmosphere surface pressure tides in Reanalysis 1979–1995. J. Geophys. Res.,
102 (D18): 22013–24.
van Loon, H. 1956. Blocking action in the southern hemisphere. 1. Notos, 5: 171–8.
van Loon, H. 1972a. Temperature in the southern hemisphere. In: C.W. Newton, ed., Meteorology
of the Southern Hemisphere, Met. Monogr., 13 (35), Amer. Met. Soc., Boston MA, pp. 25–58.
van Loon, H. 1972b. Pressure in the southern hemisphere. In: C.W. Newton, ed., Meteorology of
the Southern Hemisphere, Met. Monogr., 13 (35), Amer. Met. Soc., Boston MA, pp. 59–86.
van Loon, H. 1972c. Winds in the southern hemisphere. In: C.W. Newton, ed., Meteorology of the
Southern Hemisphere, Met. Monogr., 13 (35), Amer. Met. Soc., Boston MA, pp. 87–100.
van Loon, H. and Jenne, R. 1972. The zonal harmonic standing waves in the southern hemisphere.
J. Geophys. Res., 77: 992–1003.
van Loon, H., Jenne, R., and Labitzke, K. 1973. Zonal harmonic standing waves. J. Geophys. Res.,
78: 4463–71.
Vautard, R. 1990. Multiple weather regimes over the North Atlantic: analysis of precursors and
successors. Mon. Wea. Rev., 118 (10): 2056–81.
Venne, D.E. 1989. Normal-mode Rossby waves observed in the wave number 1–5 geopotential
fields of the stratosphere and troposphere. J. Atmos. Sci., 46: 1042–56.
Volland, H. 1988. Atmospheric Tidal and Planetary Waves. Kluwer, Dordrecht, 348 pp.
von Storch, H. and Xu, J.-S. 1990. Principal oscillation pattern analysis of the 30 to 60-day oscil-
lation in the tropical troposphere. I. Definition of an index and its prediction. Clim. Dynam., 4
(3): 175–90.
Wagner, A.J. 1976. Weather and circulation of January 1976. Mon. Wea. Rev., 104: 491–8.
Wahl, E. 1972. Climatological studies of the large-scale circulation in the northern hemisphere. 1.
Zonal and meridional indices at the 700 mb level. Mon. Wea. Rev., 100: 553–64.
Wallace, J.M. 1983. The climatological mean stationary waves: observational evidence. In: B.J.
Hoskins and R.P. Pearce, eds, Large-scale Dynamical Processes in the Atmosphere, Academic
Press, London, pp. 27–53.
Wallace, J.M. 1987. Low-frequency dynamics – observations. In: Dynamics of Low Frequency
Phenomena in the Atmosphere. 1. Observations. Notes from an NCAR Summer Colloquium,
National Center for Atmospheric Research, Boulder CO, pp. 1–75.
Wallace, J.M. 1993. The Second Haurwitz Memorial Lecture: Stationary Planetary Waves. Bull.
Amer. Met. Soc., 74: 1735–47.
Wallace, J.M. and Blackmon, M.L. 1983. Observations of low-frequency atmospheric variability.
In: B.J. Hoskins and R.P. Pearce, eds, Large-scale Dynamical Processes in the Atmosphere,
Academic Press, London, pp. 55–94.
Wallace, J.M. and Gutzler, D.S. 1981. Teleconnections in the geopotential height field during the
northern hemisphere winter. Mon. Wea. Rev., 109: 784–812.
Wallace, J.M. and Hsu, H.-H. 1985. Another look at the index cycle. Tellus, 37A: 478–86.
Wallace, J.M., Cheng, X.-H., and Sun, D.Z. 1991. Does low-frequency atmospheric variability
exhibit regime like behavior? Tellus, 43A: 16–26.
Wang, B. and Rui, H. 1990. Synoptic climatology of transient tropical intraseasonal convective
anomalies. Met. Atmos. Phys., 44: 43–61.
11 Watterson, I.G. and James, I.N. 1992. Baroclinic waves propagating from a high-latitude source.
Quart. J. Royal Met. Soc., 118: 23–50.
Waugh, D.W. 1997. Elliptical diagnostics of stratospheric polar vortices. Quart. J. Roy. Met. Soc.,
116: 913–27.
Waugh, D.W. and Randel, W.J. 1999. Climatology of Arctic and Antarctic polar vortices using
elliptical diagnostics. J. Atmos. Sci., 56 (11): 1594–613.
Webster, P.J. 1972. Response of the tropical atmosphere to local steady forcing. Mon. Wea. Rev.,
100 (7): 518–40.
Webster, P.J. 1973. Temporal variation of low-latitude zonal circulation. Mon. Wea. Rev., 101 (1):
803–16.
0 Webster, P.J. 1982. Seasonality in the local and remote atmospheric response to sea surface anom-
alies. J. Atmos. Sci., 38: 554–71.
Webster, P.J. 1983. Large-scale structure of the tropical atmosphere. In: B.J. Hoskins and R.P.
Pearce, eds, Large-scale Dynamical Processes in the Atmosphere, Academic Press, London,
pp. 235–75.
Webster, P.J. and Keller, J.L. 1975. Atmospheric variations: vacillations and index cycles. J. Atmos.
Sci., 32: 1283–300.
Weickmann, K.M., Lussky, G.R. and Kutzbach, J.E. 1985. Intraseasonal (30–60 day) fluctuations
of outgoing long-wave radiation and 250 mb stream function during northern winter. Mon. Wea.
Rev., 113: 941–61.
White, G.H. 1982. An observational study of the northern hemisphere extratropical summertime
0 general circulation. J. Atmos. Sci., 39: 24–40.
Wiin-Nielsen, A. (ed.). 1973. Compendium of Meteorology, 1, World Meteorological Organization,
Geneva.
Willett, H.C. 1948. Patterns of world weather changes. Trans. Amer. Geophys. Union, 29: 803–9.
Williams, C.R. and Avery, S.K. 1992. Analysis of the long-period waves using the mesosphere–
stratosphere–troposphere radar at Poker Flats, Alaska. J. Geophys. Res., 97 (D18): 20856–61.
Winston, J.S. 1954. The annual course of zonal wind at 700 mb. Bull. Amer. Met. Soc., 35: 468–71.
Yang, S. and Webster, P.J. 1990. The effect of tropical heating on the location and intensity of the
extratropical westerly jetstreams. J. Geophys. Res., 95 (D11): 18705–21.
Yang, S.-T. and Reinhold, B. 1991. How does the low frequency vary? Mon. Wea. Rev., 1119 (1):
119–27.
0 Yeh, D.Z., Gao, Y.X., et al. 1979. The Meteorology of Qinghai-Xizang (Tibet) Plateau (in Chinese).
Science Press, Beijing.
Zhang, X., Corte-Real, J. and Wang, X.L. 1997. Low-frequency oscillations in the northern hemi-
sphere. Theor. Appl. Climatol., 57: 125–33.
11
5 Global teleconnections
5.1 Pressure oscillations and teleconnection patterns

The study of hemispheric and global-scale oscillations in sea-level pressure has a century-
long history. Inverse pressure variations over southeastern Australia and southern South
America were first noted by Hildebransson (1897) in his studies of centers of action.
Low-frequency pressure seesaws were confirmed by Lockyer (1906) and, undoubtedly,
these results provided the basis for the extensive investigations of Sir Gilbert Walker
between 1909 and the 1930s. Several large-scale pressure patterns were distinguished by
Walker in an attempt to isolate predictors useful in long-range forecasting. Through studies
of the temporal correlation of monthly mean sea-level pressure at various locations around
the world he discovered three large-scale oscillations of pressure and associated temper-
ature and precipitation anomalies. The circulation modes are identified according to the
strongest simultaneous negative correlations with a given location at some remote distance,
3,000–6,000 km away. The three patterns identified by Walker (1924) and Walker and
Bliss (1932) were:
1 The North Atlantic Oscillation (NAO), involving the Icelandic low and Azores high.
2 The North Pacific Oscillation (NPO), involving the Aleutian low and North Pacific
high.
3 The Southern Oscillation (SO) between the southeast Pacific high and the equatorial
trough in the Indian Ocean–Indonesian region (Figure 5.1).
These pressure oscillations imply changes in the strength (or anomalous components) of
surface wind.
The term teleconnection was introduced by Ångström (1935) in the context of patterns
of climatic fluctuations; Bjerknes (1969) later used it to describe patterns of atmospheric
response to a remote surface forcing. The 1990s saw renewed interest in teleconnection
patterns between the pressure oscillations described by Walker and more distant global
anomalies. Their global characteristics and causes have been explored using various spatial
analysis techniques (Horel, 1981; Wallace and Gutzler, 1981; Barnston and Livezey, 1987;
Mo and Ghil, 1987, 1988; Kushnir and Wallace, 1989; Rogers, 1990; Cheng and Wallace,
1992) and atlases of teleconnections have been published (e.g. O’Connor, 1969; Namias,
1981; Kousky and Bell, 1992). Two aspects of teleconnection patterns need to be consid-
ered. First, the nature of the patterns that are identified by various classification techniques.
Second, mechanisms that may be responsible for their occurrence. The principal statis-
tical approaches used to identify the modes of atmospheric circulation are (1) correlation
analysis of teleconnections and (2) principal component analysis (PCA), or empirical
orthogonal function (EOF) analysis, often combined with cluster analysis. These methods
are first briefly described.
Global teleconnections 359
11
11
Figure 5.1 Simultaneous correlations (×10) of annual mean sea-level pressure with that at Darwin,
Australia, based on a composite assessment of several sources. The figure shows the
Southern Oscillation pattern. (From Trenberth and Shea, 1987)
5.1.1 Correlation analysis

The correlation field is a statistical construct representing the “net” result of various under-
lying constituent patterns. Thus the outlying features of a teleconnection pattern may blur
0 other response modes. Barnston and Livezey (1987) note that the reference location for
a teleconnection pattern has an artificially high correlation compared with the remote
centers and that the teleconnections tend to be selected according to the strength of the
negative correlation with the subjectively selected reference location, while neglecting
the spatial extent of the pattern. Nevertheless, simple composite analyses of the positive
and negative modes of Walker’s three primary teleconnection patterns have proved very
informative (van Loon and Rogers, 1978; Rogers, 1981b).
The statistical description of concurrent or time-lagged weather relationships between
different parts of the globe by the use of correlation coefficients and multiple regression
equations was pioneered by G.T. Walker between 1910 and the 1930s. Walker, as a math-
0 ematician, was aware of many of the limitations of linear correlation methods and their
application to data that possess autocorrelation in time and space, although his empirical
studies were subsequently criticized (Montgomery, 1940, for example). The use of corre-
lation patterns involves two principal problems: the multiplicity of correlations and the
existence of autocorrelation in bivariate time series (Brown and Katz, 1991). The first
problem concerns the bias caused by the limitation in the selection of factors to those
with the highest correlations. Walker recognized the need to adjust the probability levels
in determining the true significance of many correlations; otherwise there is an increased
likelihood of assuming a relationship exists when the correlation is actually negligible
(a Type I error). Brown and Katz demonstrate the need to adopt an appropriately conser-
0 vative threshold of significance. The so-called Bonferroni approach assumes that to achieve
the same probability level for multiple tests (o) as for an individual test (), tests of the
individual correlations should employ the criterion 0 /k ; for example, for k ten
11 pairs of correlations, an overall test level of 0 0.05 would require that the individual
test level be 0.005. The modern “bootstrap” procedure of cross-validation by Monte
Carlo replication is a more sophisticated alternative (Ephron and Gong, 1983).
The second problem of autocorrelation in time series can generate apparent lead and lag
relationships where none exists; smoothing of a time series has a similar effect. Brown and
Katz suggest that an autocorrelated time series be transformed into an uncorrelated time
series (“pre-whitening”) by the use of a first-order autoregressive model. A similar method
is to take the “first differences” of values in a time series (xi xi1) that is known to be highly
temporally correlated, such as monthly values of sea ice extent (e.g. Carleton, 1989). These
first differences may then be tested for serial autocorrelation to confirm that the series is now
comprised of independent observations. Alternatively, an adjustment can be incorporated
to account for the effective number of independent samples through a so-called variance
inflation factor (Katz, 1988). The recommended approach that takes into account multiplicity
and autocorrelation for a desired 0 0.05 is 1 (0.95)1/k (Brown and Katz, 1991).
Teleconnections are established by constructing one-point correlation maps for all grid
points. Wallace and Gutzler (1981) propose their summarization by a teleconnectivity field
which selects the strongest negative correlations in these one-point maps. The telecon-
nectivity at grid point i is defined:
Ti = | (rij) minimum for all j |
where rij is the correlation of point i with all other j grid points. Large values of Ti are
usually part of a standing oscillation involving one or more remote areas.
5.1.2 EOF and clustering methods

The use of empirical orthogonal functions (EOFs, or principal components) to obtain the
most efficient possible representation of a data set has become commonplace in meteo-
rology since the mid-1960s. An outline of the basic aspects of EOF analysis and references
to detailed descriptions are given in section 2.4. Only the salient features of this approach
need to be noted here. In the analysis of a time series of pressure (or height) fields, a
matrix of grid-point pressure values is converted into a matrix of covariance or correla-
tion coefficients between the pressure fields. The principal component analysis (PCAs)
yields: a set of principal components (PCs), orthogonal to one another, representing
normalized time series; eigen values which describe the normalized variance attributable
to each principal component; and eigen vectors (or loading vectors) describing the spatial
patterns associated with each principal component (Horel, 1981). Each eigen vector can
be scaled by the square root of the eigen value to obtain coefficients (loadings) relating
the PC to the original time series. The PCs can be linearly transformed (rotated) so that
the variance of the squared correlation coefficients between each rotated PC (RPC) and
each of the original time series is maximized – the varimax solution. This solution is also
independent of the spatial domain of the analysis (Kaiser, 1958). This important feature
overcomes the domain-dependent sequence of ordinary orthogonal PC patterns identified
by Buell (1975; see also Richman, 1993).
The component scores obtained by PCA, representing the projection of the original
time series data on to the PC axes, can be clustered to obtain a classification. A summary
of the procedures used to group similar objects together is contained in section 2.4.
Several independent studies, using correlation pattern analysis (Wallace and Gutzler,
1981) and rotated principal component analysis (Horel, 1981; Barnston and Livezey, 1987;
Kushnir and Wallace, 1989; Rogers, 1990), have identified the principal northern hemi-
sphere tropospheric teleconnection patterns. From monthly mean 700 mb height fields,
Barnston and Livezey show two north–south dipole patterns located in the eastern Pacific
and the western Pacific, and two uncorrelated patterns each with three centers – the
Pacific–North America and Northern Hemisphere Tropics pattern, and a northern Asia
11 pattern. Wallace and Gutzler’s analysis of 500 mb fields for winter months depicts five
main patterns (Figure 5.2) over the west Atlantic, east Atlantic, Eurasia, west Pacific, and
Pacific/North America (PNA). Rotated PCA by Horel (1981) of 500 mb heights for ninety
winter months (December–February 1950/51–1979/80) and a separate analysis for ninety
summer months (June–August 1959–79) gives ten patterns for each season (Figure 5.3).
For winters, patterns associated with RPCs 1, 2, 3, and 5 resemble the PNA, east Atlantic,
west Pacific and west Atlantic teleconnections, although pattern 2 in the subtropics is dis-
placed eastward. They account for 26 percent of the variance. RPCs 4, 6, 7, 8, and 9 are
not apparently teleconnection patterns. In summer a number of RPCs appear to be related
0 to the grid periphery. However, RPCs 4, 5, 8, and 9 are regional fluctuations analogous to,
but displaced northward of, the winter ones and accounting for only 13 percent of the vari-
ance. When the same RPC procedure is performed for five-day mean wintertime sea-level
pressures, five modes are identified by Hsu and Wallace (1985), including the North
11 Atlantic Oscillation and the Pacific/North American. These two spatial patterns were iden-
tified by Kushnir and Wallace (1989) as the dominant modes of interannual variability at
500 mb during 1946/47 to 1984/85. Hsu and Wallace (1985) indicate that these two patterns
show a barotropic structure whereas the other three – one over northern Asia, the North
Pacific Oscillation and one over Tibet–China – differ considerably at 500 mb from the sea-
level patterns in their shape and polarity. Before describing these teleconnection patterns,
0 however, we first examine the pressure oscillations themselves.
5.2 The Southern Oscillation and El Niño

The Southern Oscillation (SO) comprises a standing atmospheric wave that involves a
west–east vertical circulation (“Walker”) cell between an area centered on Indonesia and
the eastern Pacific Ocean. The concept of a thermally driven mass circulation, with rising
air over the “maritime continent” of Indonesia–Malaysia, divergent westerly upper tropo-
spheric flow across the Pacific, descent over the eastern Pacific and low-level easterly
flow along the equator, was developed by Schell (1956), Troup (1965), Bjerknes (1969),
0 and Julian and Chervin (1978), among others, and later confirmed observationally (Streten
and Zillman, 1984). An EOF analysis of pressure data shows the Southern Oscillation to
be a preferred mode of tropical circulation (Kidson, 1975).
The Southern Oscillation is most simply measured by the sea-level pressure difference
between Tahiti (Papeete) and Darwin, Australia (Troup, 1965; Parker, 1983) or Djakarta,
Indonesia (Berlage, 1957), although the term introduced by Walker was not meant to
imply that the pattern is confined to the southern hemisphere. The immense spatial scale
of the Southern Oscillation, as represented by simultaneous correlations of MSL pressure
with that at Darwin, is illustrated in Figure 5.1. It is the major contributor to variance in
climatic fields globally on interannual time scales. A positive Southern Oscillation Index
0 (SOI) represents a strong southeastern Pacific high with anomalously low pressure centered
over Indonesia–northern Australia, well developed low-level easterlies and strong convec-
tion over Indonesia; during low (negative) SOI the region of ascending air and convection
shifts eastward to the central Pacific and this sets up low-level westerlies over the western
Pacific (Figure 5.4). During negative SOI regimes, warm surface waters (temperatures
28°C) extend eastward from their normal location between 10°N and 10°S in the western
Pacific as the trade winds relax, and the normal cold upwelling waters off Peru and
Ecuador (Figure 5.4c and d) are replaced by a well developed current of warm equatorial
water flowing southward along the coast (Rasmussen and Carpenter, 1982). The associated
atmosphere–ocean links are discussed below.
0 Various indices are used to describe the SO and to identify its modes. Selected atmos-
pheric and oceanic variables exhibit a characteristic pattern of variations in time, as
illustrated in Figure 5.5. The principal indices used to identify the different modes of the
11 SO are: sea-level pressure anomalies at Darwin (12.4°S, 130.9°E) and Tahiti (17.5°S,
(a)
(b)
Figure 5.2 Simultaneous correlation patterns in winter monthly mean 500 mb height data,
1962/63–1976/77. (a) Centers of the five strongest patterns: EU Eurasian, WP West
Pacific, PNA Pacific North American, WA West Atlantic, EA East Atlantic. The + and
– signs denote the sign of the correlation within each pattern. The light lines show the
wintertime mean 500 mb height contours. (b) Arrows and shaded areas denote strong
negative correlations between distant locations; heavy shading > 0.75 and light
shading > 0.6. Correlations are plotted ×100 and arrows point the direction(s) of the
correlations. (After Wallace and Gutzler, 1981; from Wallace and Blackmon, 1983)
11
11
(a)
(b)
0
Figure 5.3 Summary map of the loading vectors associated with the first ten RPCs, derived from
(a) ninety winter months and (b) ninety summer months, for 500 mb height fields. The
respective seasonal mean 500 mb height contours are indicated. (From Horel, 1981)
11
Figure 5.4 Schematic models of (a) non-ENSO and (b) ENSO mode vertical circulation cells in
equatorial latitudes. (c) Sea-surface temperature anomalies corresponding to (a), (d) and
(e). Changes in sea level and thermocline depth corresponding to (a) and (b). (Wyrtki,
1985; from Barry and Chorley, 1998)
149.6°W), usually expressed as the anomaly at Tahiti minus that at Darwin (the SOI1);
sea surface temperatures in the equatorial Pacific east of 180° longitude; rainfall in the
equatorial central Pacific; and zonal wind anomalies also in the equatorial central Pacific.
Figure 5.5 shows a close similarity between these different variables. An enhancement of
the climatological-mean high pressure over Tahiti and low values over Darwin are highly
correlated with a cooler sea surface in the central eastern equatorial Pacific, reduced rain-
fall in the equatorial central Pacific and stronger easterlies – referred to as the High/Dry
mode of the SO and its converse as the Low/Wet phase. Modern usage refers to the
related “warm” (El Niño) or “cold” (La Niña) events described below.
11
11
Figure 5.5 Monthly values of seven indices of the Southern Oscillation. (a) SST, mean anomaly
6°–2°N, 170°–90°W; 2°N–6°S, 180°–90°W; 6°–10°S, 150°–110°W. (b) Rainfall, mean
0 percentage of cube root at up to six stations in central equatorial Pacific. (c) Pressure
anomaly at Darwin. (d) Pressure anomaly at Darwin minus that at Tahiti. (e) Mean
zonal wind anomaly 5°N–7°S, 150°E–150°W. (f) As (c) but smoothed using filter (0.25,
0.5, 0.25). (g) As (e) but smoothed as in (f). (From Wright, 1985)
The basic indices are sometimes contaminated by local scale or transient disturbances.
In order to extract the SO signal from pressure data, Trenberth (1984) calculated a ratio of
the SOI, defined as the difference between the monthly MSL pressure values at Tahiti and
Darwin, each normalized by the respective standard deviations, divided by a noise index,
defined as the sum of the normalized pressures at Tahiti and Darwin. The signal-to-noise
0 ratio can be further amplified by applying a low-pass eleven-point filter to the series to
remove fluctuations with a period less than about a year. Figure 5.6 illustrates the clear
inverse relationship between sea-level pressure anomalies at Darwin and Tahiti filtered in
this manner. An alternative to averaging sea surface temperatures over a grid box is to
analyze the longitude of the 28.5°C (warm pool) isotherm between 4°S and 4°N, as illus-
trated in Figure 5.7. The extreme nature of the 1982–83 warm event (Caviedes, 1984) is
readily apparent. The 1997–98 warm event gave rise to extreme seasonal anomalies of tem-
perature and precipitation (the upper quintiles of the distribution) over about 80 percent of
the contiguous United States. One-fifth of the country experienced fifty-year record winter
temperature anomalies. Even so, Harrison and Larkin (1998b) emphasize that these per-
0 centages are not exceptional compared with other non-El Niño years.
Figure 5.5 shows also that the various SO indices exhibit significant persistence; for
1946–81 the autocorrelation of seasonal values of the Darwin–Tahiti pressure index is
11 0.63 for a one-season lag and 0.40 for a two-season lag, while sea surface temperatures
Figure 5.6 Sea-level pressure anomalies at Darwin and Tahiti, 1950–99. Monthly values are
smoothed with an eleven-point filter (Courtesy Dr K.E. Trenberth)
Figure 5.7 Variation of the longitude of the 28.5°C isotherm of sea surface temperature for
4ES–4EN for January 1951–June 1991. Values are three-month running means. The
forty-year mean longitude is 175.6°W. (From Diaz and Kiladis, 1992)
for the equatorial Pacific from the Galapagos to longitude 180° show corresponding auto-
correlations of 0.76 and 0.41, respectively (Wright, 1985). There are smaller, but
statistically significant, negative autocorrelations for a six-season lag in these same indices.
This persistence is high between July and February, and least in April, which creates a
barrier to seasonal predictability (see section 5.4). The correlations between indices are
11 also greatest from September through February. In effect, the SO pattern has a tendency
to become established around April, although this is by no means a fixed recurrence date.
The SO is closely linked with the occurrence of sea-surface temperature anomalies in
the eastern equatorial Pacific, particularly off Ecuador–Peru. During the austral summer,
warm equatorial water moves southward along the west coast of South America as the
southeasterly trades slacken. Typically this weak warm current reaches 5°–6°S by late
December or early January, hence the name El Niño, or Christ child. Every few years,
however, this water is anomalously warm (2°–4°C above normal) and penetrates along
the coast of Peru, replacing the usual cold, nutrient-rich Humboldt Current and thereby
0 leading to disastrous effects on marine life. Heavy precipitation at this time also causes
severe flooding and mudslides in the normally arid coastal zone. The phenomenon was
first studied in its local and regional context (Bjerknes, 1966, for example). In the nine-
teenth century, reductions in guano production by island bird colonies during El Niños
11 had major impacts on the Peruvian economy by cutting exports of guano fertilizer.
Anchoveta fishmeal, used as animal feed, became economically important in Peru during
the 1950s–60s until overfishing, exacerbated by the 1972–73 El Niño, virtually eliminated
the industry.
Various criteria are used to define ENSO events. Wright et al. (1988) use the MSL
pressure at Darwin only, while van Loon and Madden (1981) emphasise MSL pressure
0 in the zone from Darwin to Cocos Island (12°S, 97°E) and rainfall at stations in the equa-
torial South Pacific. Rainfall indices are commonly determined from the cube root of
monthly amounts, which serves approximately to normalize the frequency distribution
(Wright, 1984). Changes in SST patterns and the distribution of deep cloud masses, as
an indicator of precipitation occurrence, can be readily monitored from satellite infrared
data (see Chapter 2). Rasmusson and Carpenter (1982) define ENSO events in terms
of sea surface temperatures off the Peruvian coast. Kiladis and van Loon (1988) use the
SOI combined with an SST anomaly index for the eastern equatorial Pacific. A multi-
variate ENSO Index (MEI) has been developed by Wolter (1987; Wolter and Timlin,
1993). It combines information on six observed variables over the tropical Pacific – sea-
0 level pressure, zonal and meridional components of surface wind, sea surface and air
temperature, and total fractional cloudiness. The spatial fields of these variable are simpli-
fied by clustering and the MEI is determined by the first unrotated principal component
of the six fields combined. Time series of the index are published on the Web by the
NOAA Climate Diagnostics Center (http://www.cdc.noaa.gov).
The usage of the term “El Niño” has evolved over time. The original regional defini-
tion has generally been superseded through the recognition that such events are commonly
part of a Pacific-wide warming pattern extending from the coasts of Ecuador and Peru to
the date line. Trenberth (1997) indicates that an appropriate quantitative definition for El
Niño is provided by the index introduced by NOAA in April 1996. This is based on a
0 five-month running mean of anomalies of sea surface temperature (SST) of ±0.4°C or
more lasting for at least six months in the “Niño 3.4” region (5°N–5°S, 120°–170°W).
The annual mean SST for the Niño 3.4 box is 26.8°C, with a standard deviation of 0.77°C;
monthly means range from 26.4°C in NDJ to 27.6°C, in April. The positive mode, with
a negative SOI, and its large-scale manifestations, is referred to as an El Niño–Southern
Oscillation (ENSO) (warm event); the other non-ENSO pattern (cold event) is now termed
La Niña (the girl) because the atmospheric and oceanic conditions are essentially oppo-
site to those associated with El Niño (Aceituno, 1992; Enfield, 1989; Philander, 1990).
However, episodes of low SOI values and El Niño events have both occurred separately
(Deser and Wallace, 1987). For the period of January 1950–March 1997 Trenberth (1997)
0 identifies 31 percent of months as El Niño (fifteen events), and 23 percent as La Niña
(ten events), with neither being present for 45 percent of the time.
The relationship between El Niño events and the SO is apparently not fixed (Wright
11 et al., 1988). Diaz and Pulwarty (1992) suggest that during 1882–1938 negative SOI
events may have reached maximum intensity late in year 0 of El Niño onset, continuing
into year 1, whereas during 1939–88 the SOI was strongest earlier in year 0. The large-
scale SO was absent during the interval 1925–35, according to Trenberth and Shea (1987),
although El Niño events occurred in 1925, 1930 and 1932. Nearly half of the last 112
years can be characterized as cold or warm events; Diaz and Pulwarty list twenty-seven
El Niño (warm) years and twenty-one cold ones between 1877 and 1988. Because ENSO
events differ in strength, listings provided by various authors commonly differ according
to the criteria adopted. Quinn et al. (1978) distinguish strong El Niño events as having
monthly sea-surface temperature anomalies > 3°C range and weak ones in the 1.0°–2.5°C
range. Weak events may also appear relatively late in the year.
El Niño events have major climatic impacts regionally and worldwide, through ENSO-
related teleconnections (Glantz et al., 1991; Allan et al., 1996). Based on cases between
1950 and 1975, the anatomy of these warm episodes is illustrated by a composite ENSO
developed by Rasmusson and Carpenter (1982) (Figure 5.8).
1 Strongly positive SST anomalies develop in the eastern equatorial Pacific and along
the coast of South America from the equator to about 12°S. The anomalies tend to
reach a maximum between April and June, with a second peak around the following
December.
2 Weaker ocean surface warming extends across the equatorial Pacific westward to the
dateline, reaching a peak around December; mean SST anomalies for six events
exceeded 1°C eastward of 150°E.
3 The normal east–west gradient of MSL pressure in the tropical South Pacific collapses,
leading to negative SOI values which peak around September.
4 The rainfall belt, normally located over Indonesia, is displaced eastward towards the
usually dry central Pacific. This begins in April–May, culminating around December.
5 Westerly wind anomalies develop at the surface in the western equatorial Pacific, on
the western edge of the region of anomalous rainfall over the Pacific.
A more fully documented composite based on the ten warm events between 1950 and
1991 is now available (Harrison and Larkin, 1996, 1998). Harrison and Larkin distinguish
six phases associated with the following statistically robust features of sea level pressure,
wind and sea surface temperatures:
1 Pre-ENSO, March (1) to November (1): strong westerly anomalies over the equa-
torial Indian Ocean; weak surface warming in the tropical and subtropical western
and central South Pacific. There is a negative pressure anomaly over southeast
Australia–New Zealand.
2 Ante-ENSO, December (1) to February (0): initial warming of the equatorial Pacific
surface from the dateline to the South American coast.
3 Onset, March (0) to June (0): significant warming of the eastern and central equato-
rial Pacific; sea-level pressure anomalies are negative in the eastern equatorial Pacific
and the subtropical southeast Pacific, and positive over northern Australia and the
Arabian Sea.
4 Peak, July (0) to December (0): maximum equatorial Pacific warmth and low-level
convergence over the central equatorial Pacific; westerlies on the equator from 150°E
to 130°W, easterlies over Indonesia and in the ITCZ over the eastern equatorial
Pacific; strong westerlies and negative SST anomalies in the North Pacific near 170°E,
40°N. The pressure anomalies of the onset phase continue with the Arabian high
weakening and a new positive anomaly appearing in the subtropical western North
Pacific.
5 Decay, January (1) to April (1): the warmth in the equatorial Pacific decays near
South America but persists in the central sector; after January (1) the North Pacific
11
11
Figure 5.8 Composite SST anomalies illustrating the development of the canonical El Niño. (a)
March–May of the ENSO year, the peak phase off South America. (b) August–October,
with anomalies developing in the central Pacific. (c) December–February, the mature
phase. (From Rasmusson and Carpenter, 1982)
11
westerlies weaken; the easterlies persist over Indonesia. The eastern equatorial nega-
tive pressure anomaly is the only robust feature.
6 Post-ENSO, May (1) to August (1): no significant equatorial SST anomalies; east-
erlies persist over Indonesia.
A further composite description of warm and cold events in the Pacific based on the
Comprehensive Ocean Atmosphere Data Set (COADS) for 1946–85 and outgoing long-
wave radiation (OLR) data for 1974–89 provides some additional information (Deser and
Wallace, 1990). There is an equatorward expansion and intensification of the ITCZ over
the eastern Pacific during warm events. There is a net increase of deep convection over
the tropical Pacific, and enhanced convection in the ITCZ is accompanied by a net increase
in surface wind convergence. However, moisture convergence extends through a deeper
layer in the western equatorial Pacific than in the ITCZ over the eastern Pacific.
Corresponding to these associations, high correlations are found between MSL pressure
at Darwin, SSTs and precipitation in the central equatorial Pacific, and zonal wind in the
western equatorial Pacific, particularly between July and November (Wright et al., 1988).
A composite analysis of eighteen warm events (1902–76) shows a precipitation anomaly
averaged at three stations in the “maritime continent” of 2.2 mm day1 compared with
3.2 mm day1 at three equatorial island stations in the central Pacific; the mean anomaly
of this dipole is 2.7 mm day1 (Lau and Sheu, 1988). The corresponding precipitation
anomalies during sixteen cold events are 2.8 mm day1 and 1.9 mm day1, with a
mean dipole anomaly of 2.3 mm day1.
Both the warm and cold ENSO events appear to be “phase-locked” with the annual
cycle, showing a tendency to develop during March–June and lasting at least a year. The
extreme modes have a tendency to follow one another. Composite anomalies of global
temperature and precipitation show opposite signs in seasonal anomalies between the year
prior to the event (year 1) and year 0 for both cold and warm events, although this
relationship is not invariable (Deser and Wallace, 1987; Diaz and Kiladis, 1992).
Moreover, since the late 1970s, by comparison with earlier decades, the paradigm of a
quasi-periodic annually phase-locked ENSO cycle has failed to provide an adequate repre-
sentation of ENSO complexities (Wallace et al., 1998).
Anomalies of March–April wind and pressure over the Atlantic Ocean for ten El Niño
years (and its opposite phase) show inverse pressure tendencies in the subtropical anti-
cyclones over the South Pacific and South Atlantic Oceans and associated wind fields
(Covey and Hastenrath, 1978). During the positive SO phase in March–April, when eastern
equatorial Pacific waters are cold, there are cold surface waters associated with north-
easterly trades in the tropical North Atlantic, but positive anomalies in the tropical South
Atlantic. Hastenrath et al. (1987) show that the near-equatorial low-pressure trough is
displaced southward at this season, but well northward in July–August, associated with
enhanced meridional displacements in positive SO phases. These northward shifts favour
precipitation in the Sahel and the Caribbean–Central America. During negative SO
(El Niño) years the excursions of the equatorial trough tend to be suppressed.
The tropical Atlantic Ocean resembles the tropical Pacific in having an equatorial low-
pressure trough, trade wind systems, and a convergence zone, as well as the cold Benguela
Current off southwest Africa analogous to the Peru current. Anomalies similar to those
during El Niño in the Pacific may occur, as in 1984, for example (Horel et al., 1986).
In this phase, heavy rains can affect the usually arid coast of Namibia, southwest Africa.
In the tropical Atlantic, however, the anomalies have no east–west negative correlation
patterns; rather there are meridional displacements of the ITCZ. There is also no east-
ward movement of a convergence center, such as occurs over the western Pacific. The
upper ocean in the equatorial Atlantic responds in phase with the seasonal variation of
the tropical easterlies, not interannually as in the Pacific. Philander (1990, p. 88) notes
that the east-to-west slope of the thermocline in the equatorial Atlantic is steepest from
11 July through December, when the winds are strongest. These ocean basin contrasts are
related in part to the smaller east–west extent of the tropical Atlantic (one-third of the
Pacific) and to the land–sea arrangement and its effects on currents.
The Southern Oscillation displays distinct evidence of leads and lags in its spatial evolu-
tion, that must be seen as a variant of the normal annual cycle of pressure over the South
Pacific Ocean. From March to June mean pressure falls east of New Zealand, reflecting
the half-yearly pressure wave in southern temperate latitudes (van Loon, 1972), and rises
over Australia; in the subtropics, southerlies strengthen and the trades weaken. The
seasonal pressure trend from June to September reverses, with pressure falling over
0 Australia and rising east of New Zealand. The subtropical ridge strengthens over the
Pacific, with faster trades and northerly wind components over the western South Pacific
(van Loon, 1986; van Loon and Shea, 1987). In June–August of the year preceding a
warm event, a weaker than normal trough east of New Zealand is often a precursor (van
11 Loon and Shea, 1985). The northerly anomalous geostrophic winds in the southwest Pacific
form positive anomalies of sea-surface temperature which persist into the austral spring.
Heating by this warm surface intensifies the South Pacific Convergence Zone as it extends
southeastward and, as a result, below-normal pressure develops over the Tasman Sea.
During the succeeding austral summer (January of the El Niño year) the anomalous
wind component becomes westerly in the trade-wind belt of the Pacific Ocean south of
0 10°S, as illustrated by the sequence in Figure 5.9. During the warm event year there is
an amplification of the normal annual cycle of trade winds between 15°S and 30°S, and
of the westerly trough in the South Pacific Ocean, compared with the preceding year (van
Loon, 1984).
The concept of a simple standing oscillation between Indonesia and the eastern equa-
torial Pacific has given way to the view that ENSO is a propagating system (Barnett,
1985, 1988; Xu and von Storch, 1990). Barnett (1984) characterized the SO as “one part
of a larger propagating wavelike phenomenon in the sea level pressure field that appears
first over the Indian Ocean.” Correlation analysis suggests that the pattern of pressure
anomalies reflects the motion of a wave train that begins in the Pacific, crosses South
0 America and ends in the southern Indian Ocean, with a phase that reverses from the year
preceding the warm event to the El Niño year itself (van Loon, 1986). Composite analysis
of eight warm events between 1951 and 1985 shows that anomalies of a given sign
develop sequentially over a roughly two-year interval in a counterclockwise trajectory
from southern Asia to the tropical Indian–Pacific ocean, to the eastern Pacific and North
Pacific, before returning to Asia (Barnett, 1988).
Further information is available from a Principal Oscillation Pattern (POP) analysis of
monthly mean sea-level pressures for 15°–40°S during 1951–58 and 1972–83 by Xu and
von Storch (1990). This diagnostic technique extracts standing or oscillating spatial
patterns from a dynamically complex multicomponent system. The time series, filtered in
0 time and space to retain the specific scales of interest (x(t)), is assumed to be generated
by a first-order autoregressive process:
x (t 1) Ax(t) forcing
where A is a matrix, whose real (or complex) eigen vectors are called POPs; real valued
POPs represent standing patterns, and pairs of complex-valued POPs represent oscillatory
patterns that often migrate in space. POPs describe the normal modes of a linear stochastic
process. They do not comprise an orthogonal set of patterns and are not necessarily inde-
pendent. There is also no information from the analysis about the variance (von Storch
et al., 1990). Xu and von Storch (1990) find an oscillatory POP pair with an oscillation
0 period of about thirty-seven months. It accounts for 48 percent of the variance in the SOI,
or 73 percent for the 1972–88 data. A sequence of patterns is identified:
11 …→ P1 → P → P1 → P2 → P1 →
Figure 5.9 Anomalies of sea-level pressure and anomalous wind components in the South Pacific
Ocean for the year preceding (1) and the year of (0) a warm ENSO event for (a)
May–June–July (1), (b) November–December (1), January (0), (c) May–June–July
(0) and (d) November–December (0), January (1). (From van Loon and Shea, 1987)
P1 represents a positive SOI mode which is replaced after seven to nine months by P2,
associated with westerly wind anomalies in the western tropical Pacific. About sixteen
months later, there are easterly anomalies over the eastern tropical Pacific and southerly
anomalies near the dateline (P1). By twenty-seven months there are easterly wind anom-
alies over the western tropical Pacific associated with positive pressure anomalies north
of New Zealand (P2) with a return to P1 by thirty-six to thirty-seven months. These
results support earlier studies of southern hemisphere pressure fields using composites
based on the state of the SO and the annual cycle (van Loon and Shea, 1985, 1987).
The warm and cold events represent features superimposed on an annual cycle of
convective activity and the west–east circulation cell linking the Indian and southwest
Pacific sectors of the tropics (Meehl, 1987a). During a cold event the Indian summer
monsoon is usually strong, and there is also lower pressure, increased rainfall, and higher
11
11
0
Figure 5.10 A schematic model of zonal and meridional circulation cells in the Indian and Pacific
ocean sectors. (From Meehl, 1987)
SSTs to the west of the South Pacific Convergence Zone (SPCZ). To the east of the SPCZ
the descending arm of an intensified zonal Walker circulation sets up higher surface pres-
sure and less rainfall (Figure 5.10). During the following October–January–April, the
convective maximum moves southeastward across southeast Asia to northern Australia,
maintaining its strength, partly owing to the warmer ocean surface, which also induces a
southwestward displacement of the SPCZ. The South Pacific high now weakens, and
0 lower pressure in the SPCZ is associated with anomalous cyclonic flow nearer the equator.
Westerly wind anomalies at the surface are typical of the transition to a weak annual
cycle (Figure 5.8), which in the extreme is a warm event. During years with a weak cycle,
11 conditions across the region are the opposite of those described. During a warm event
the ITCZ and SPCZ are both displaced equatorward, becoming merged near the dateline
(Kiladis and van Loon, 1988).
Links between the Walker and Hadley circulations in the equatorial western Pacific are
apparent from a comparison of indices of the zonal and meridional wind components in
a sector approximately 10°S to 10°N, 150°E to 140°W (Barnett, 1984). Changes in anom-
alous surface winds in the Walker cell lead those in the Hadley cell by two to six months.
Thus, strengthening in the Walker cell between January and April is followed by weak-
ening of the Hadley circulation in September–October (and vice versa). The mechanism
appears to involve forcing exerted by anomalous convergence and precipitation within
the shifting mode of the Walker cell. The relationship is not apparent, however, in northern
winter.
Time series analyses of various SO indices reveal that the variance is concentrated
around 2.3 years (a quasi-biennial oscillation, or QBO, in surface climate variables) and
between three and six years (Berlage, 1957; Trenberth, 1976) (see Table 5.1). When
singular spectrum analysis2 is used to filter out the two to three-year variability in the
SO, a four to six-year ENSO cycle is isolated (Keppenne and Ghil, 1992). Further study
by Barnett (1991) using filters to distinguish quasi-biennial and three to seven-year periods
Table 5.1 Twentieth-century warm and cold

events in the Pacific
Warm events Cold events

1902 1903
1904 1906
1911 1908
1913 1916
1918 1920
1923 1924
1925 1928
1930 1931
1932 1938
1939 1942
1951 1949
1953 1954
1957 1964
1963 1970
1965 1973
1969 1975
1972 1988
1976 1995
1982 1998
1986
1991
1997
Note: only the zero year is tabulated when the SOI
changes sign and central-eastern equatorial Pacific SST
anomalies became strongly positive (warm event) or
negative (cold event). The events are identified on three
criteria up to 1976: SSTs in the equatorial Pacific, sea-
level pressure at Darwin (and Tahiti from 1935), and
rainfall anomalies in the equatorial Pacific dry zone. The
events in 1977 and subsequently are based on the first
two of these criteria (Kiladis and Diaz, 1989).
Source: updated after van Loon (1984); Diaz and Kiladis
(1992).
11 in near-global SST and SLP data indicates that ENSO time scales involve annual and
QBO periodicities of twenty to thirty and thirty-six to eighty months. Barnett suggests
that both data sets indicate that the ENSO phenomenon is attributable to a non-linear
interaction of the two longer time scales. There is also a traditional SO in the SLP records
and El Niños in the SST data, in both frequency bands. The quasi-biennial mode of ENSO
variability is phase-locked with the annual cycle of surface winds in the eastern equatorial
Indian Ocean (Rasmusson et al., 1990). This component provides some regularity to ENSO
occurrences, although its amplitude is variable and some phase changes were recorded
during 1950–87. Warm and cold events apparently reflect opposite phases of the biennial
0 component, modulated by the low-frequency (four to five-year) mode. Records of 115 El
Niño events from Peru during AD 1524–1987 compiled by Quinn et al. (1987; Quinn
and Neal, 1992) show an average periodicity of almost four years, with several episodes
(45 percent of cases) recurring at about nine-year intervals. However, the recurrence short-
11 ened to 3.1 year during 1824–1941 (Diaz and Pulwarty, 1992). Tree ring records from
the southern Great Plains and Sierra Madre Occidental, which correlate well with the SOI,
support this. Southern Oscillation extremes of either sign have an average interval of 6.04
year for 1699–1850, but only 3.71 year for 1850–1971 (Stahle and Cleveland, 1993).
The intensity of the SO signal also seems to vary over longer time scales. This is illus-
trated by the correlation of annual mean pressure at Darwin and Tahiti; it was 0.75
0 during 1940–80, but only 0.38 for 1883–1940 (Elliott and Angell, 1988). Older data
may be less reliable, but it seems more probable that the southeast Pacific pole of the
Southern Oscillation is subject to changes in location and/or intensity. Moreover, while
the SO consists of a dominant east–west standing wave, it also has other components.
Historical reports since the early 1600s in Peru and Chile indicate that El Niño activity
was pronounced in the periods 1607–24, 1701–28, 1812–32, 1864–91, and 1925–32, but
weak in the intervening years. The 1982–83 event, like those of 1925–26 and 1891, is
categorized as having been very strong; other strong events occurred in 1940–41, 1957–58,
and 1972–73. Very strong events have occurred only eight times in five centuries, with
none of them recurring within twenty years (Enfield, 1992). Low-frequency, especially
0 decadal, ENSO modes are identified by a Fourier transform analysis of the records for
1876–1995. Brassington (1996) shows that from the 1870s to the 1930s very severe ENSOs
recurred with significant seven to eight and thirty-five year periods in addition to quasi-
biennial and quasi-quadrennial (four to five and five to six-year) periodicities. Spectral
power is also identified at about thirty-five to forty-five years in records of 18O from ice
cores in the Quelccaya ice cap, Peru, and in El Niño, but with low coherence between
ENSO and 18O (Diaz and Pulwarty, 1994).
The use of historical records of Nile floods at Cairo as a proxy for SOI-related climatic
activity since AD 622 is proposed by Quinn (1992) and receives some statistical support
from Diaz and Pulwarty (1992). The annual maximum flow of the Nile registers the
0 summer monsoon rainfall over the Ethiopian Highlands, with below-normal flood levels
occurring during low (negative) SOI phases. Spectral analysis of the composite record
suggests cycles of around ninety, forty to fifty, and twenty-two to twenty-four years in
SOI-related activity (Anderson, 1992; Diaz and Pulwarty, 1992), suggestive of low-
frequency solar forcing. There is also evidence of fewer and weaker El Niño events during
episodes of strong solar activity. Using the data of Quinn et al. (1987) for AD 1525–1988,
spectral analyses by Diaz and Pulwarty (1992) show that the period of ENSO events is
three to four years when solar activity and the Quasi-Biennial Oscillation (QBO) are weak,
but lengthens to four or five years during high solar activity and a well developed QBO.
Enfield (1992) confirms this and by confining the analysis to strong (and very strong)
0 events, finds corresponding modal intervals of 8.5 years for low solar activity and 12.8
years for high solar activity.
Analyses of several ENSO indices indicate a possible regime shift in 1976/77 (Zhang
11 et al., 1997). SSTs in the tropical Pacific and off the west coast of the Americas are
higher during 1977–93 than during 1950–76. However, it is also feasible that a cooling
trend occurred during 1943–55, followed by gradual warming, rather than a regime shift.
Nevertheless, the frequency and duration of ENSO increased over the last 20 years of the
century, with 1990–95 having the longest sustained high pressures recorded at Darwin
during 1892–1995 (Rajagopalan et al., 1997). Rajagopalan et al. suggest an average return
period for this five-year event of 350 years (based on non-parametric statistics and a
Markov model), and suggest that the event could occur with 0.28 probability in the
1882–1995 record. However, a linear autoregressive moving average (ARM) analysis by
Trenberth and Hoar (1996, 1997) which is a more appropriate statistical model, focuses
on the size of the SOI anomalies and shows that both the post-1976 anomalies and the
1990–95 event have less than a 0.05 percent chance of occurring in the 1882–1995 record.
It is still debatable whether the event could be greenhouse effect-induced, as proposed
by Trenberth and Hoar. Latif et al. (1997) argue against this view on the grounds that
the decadal mode of SST variability is dominant in the 1990s. This mode has SST anom-
alies of the same sign in all three tropical oceans. There is a “horseshoe” anomaly pattern
centered in the western equatorial Pacific, with branches extending northeast and south-
east into the subtropics. This pattern is unlike that associated with ENSO anomalies in
the eastern Pacific (see Figure 5.8).
5.3 ENSO mechanisms

The climatological state of ocean temperatures in the equatorial Pacific is determined by
three dynamical effects (Bjerknes, 1969; Cane, 1992). These involve:
1 Horizontal advection of cold water westward from the west coast of South America
by the easterly trades.
2 Upwelling along the equator resulting from the divergence of the westward surface
current (Figure 5.11). Waters move poleward in each hemisphere under the influence
of the Coriolis deflection (to the right/left in the northern/southern hemisphere).
3 The upward displacement of the thermocline off the west coast of South America as
the upper warmer-water layers are forced westward. This is accompanied by cold
upwelling and a reduction in ocean heat storage (Wyrtki, 1985).
The relative importance of these three processes is still uncertain.
Figure 5.11 Schematic model of the processes contributing to the maintenance of a tongue of cold
water in the eastern equatorial Pacific. (From Niiler, l982)
11 A model to account for El Niño conditions was proposed by Wyrtki (1975). During
periods with strong trade winds (high SOI) the ocean level is raised (~20 cm) in the
western Pacific by the accumulation of water against the coasts of East Asia and New
Guinea–Australia. Coastal wind-induced divergence off Ecuador and Peru and the shallow
westward-tilting thermocline facilitate the normal upwelling (Figure 5.4e). As the trade
winds relax, warm water from the western Pacific returns eastward, and the Peruvian
upwelling is suppressed as the thermocline becomes deeper. Wyrtki (1985) observes that
positive height anomalies (h) are associated with a depression of the thermocline and
an increase in mixed layer temperature, as represented by a selected isotherm depth (D):
0
h = D 冢QQ冣
11 where Q/Q is the relative density difference between two layers (~ 5 to 6 103). The
slow build-up of water level in the western ocean during easterly years is rapidly reversed
during an ENSO cycle. For example, the 1982–83 ENSO was associated with an east-
ward flux of warm water (about 40 Sv) in the upper layer of the equatorial Pacific,
15°N–15°S, according to Wyrtki’s hypothesis. He proposed that the relaxation of the trade
winds prompts the eastward displacement of water by triggering a Kelvin surge. This
0 warm water is deflected towards the subtropics by the west coast of the Americas and so
out of the tropics.
Attractive though this simple model may be, it fails to explain the history of the 1986–87
El Niño, which was not preceded by a build-up of warm tropical water (Miller and Cheney,
1990). Moreover, during the extreme 1982–83 ENSO event, heat content was enhanced
only between 5°N and 5°S, not 15°N–15°S. Most strikingly, in 1982–83 warm waters
appeared not along the South American coast but along the equator in the austral winter,
from where they spread eastward. The warm and cold ENSO modes show a persistent
tendency to alternate from one to the other. Bjerknes (1969) remarked, “Just how the
turnabout between trends takes place is not yet quite clear.” The mechanism appears to
0 involve adjustments in the subsurface ocean thermal structure associated with internal
ocean waves (Neelin et al., 1998). Nevertheless, the record El Niño of 1997–98 was
terminated primarily by the rapid and unexplained onset in May-June 1998 of strong trade
winds driving cold upwelling (McPhadden, 1999).
The pattern of sea surface temperature change in the equatorial Pacific shows a general
tendency for the warmest area (28.5°C) to migrate eastward from the west-central Pacific
(about 150°E) during the development of ENSO events and to retreat westward during
the decaying phase. However, there are differences between events in the pattern of
warming. Three SST patterns can be defined on the basis of temperatures in the zone
4°N–4°S, 120°E–80°W (Fu et al., 1986). Pattern A has much warmer surface water in
0 the east-central Pacific, below normal west of 160°E (1957, 1965, 1972, and 1982); pattern
B is warm mainly in the eastern Pacific, below normal in the west (1976); and pattern C
is warm almost everywhere (1963, 1969). The equatorial westerlies extend farther east
during strong events of pattern A than during weaker ones like 1976. This may reflect a
positive feedback between the westerlies and warmer surface water. The warmest areas
also appear to be associated with strong convection.
The warming of the ocean surface layer is determined primarily by the advection of heat
by wind-driven ocean currents (Barnett, 1984). Vertical heat exchange between the ocean
and atmosphere plays only a minor role in the heat budget of the central equatorial Pacific
(Ramage and Hori, 1981). Indicative also of this oceanic heat advection is the high corre-
0 lation found by Barnett between anomalies of surface wind SSTs near the equator and in
the southwest Pacific (0.74 in January–February and 0.86 in August–September).
An important related aspect of ocean–atmosphere interaction concerns the postulated
11 association between higher sea surface temperatures and enhanced convective activity and
precipitation proposed by Bjerknes (1969). Observations over the tropical Indian and
Pacific Oceans show that convection is closely controlled by SSTs, being much more
active for SST 27.5°C (Graham and Barnett, 1987). However, Graham and Barnett’s
suggestion of a decrease in the activity beyond about 28°C has not withstood further
analysis. The frequency and intensity of deep convection increases with SSTs between
about 26°C and 30°C (especially in January), but this relationship is less apparent within
the ITCZ and at times can be overridden by other factors over the warm pool in the
western Pacific (Zhang, 1993).
Large-scale vertical motion at 850 mb across the tropical Pacific matches the patterns
of HRC and there is also a strong correlation on a seasonal basis between maximum
upward motion and maximum sea surface temperatures according to Zimmerman et al.
(1988). They find that warming of the eastern Pacific surface strengthens the Hadley circu-
lation, as postulated by Bjerknes (1966), but there is no evidence of changes in the zonal
gradient of equatorial sea surface temperatures affecting east–west circulation cells.
Observed sensible and latent heat fluxes turn out not to be correlated with SSTs and
precipitation in the manner hypothesized by Ramage (1977). A possible resolution of this
contradiction, involving a simple model linking low-level convergence and the moist static
energy, has been offered by Neelin and Held (1987). Precipitation in the model depends
on moisture convergence, not heat fluxes. High SSTs do cause high rainfall areas, but no
change in heat fluxes into the atmosphere is required. Variations in precipitation are caused
almost entirely by moisture convergence, which is dependent on SST, through the effects
of SST on large-scale moist stability. Within a shallow equatorial boundary layer, surface
wind convergence, and therefore precipitation, is proportional to the Laplacian of sea
surface temperature (2 SST), according to Barnett et al. (1991). Figure 5.12 shows how
condensation heating can set up a secondary circulation that feeds back on the structure
of the planetary boundary layer (PBL).
Figure 5.12 A schematic model of coupling between an SST gradient which sets up convergence
in the planetary boundary layer (PBL) and to first order dictates the position of the
precipitation zone. Condensation heating sets up a secondary circulation cell that feeds
back positivity on the PBL structure. (From Barnett et al., 1991)
11 Recent oceanographic research highlights the important role of internal waves in the
upper layer of the equatorial Pacific for ENSO mode transitions (Cane, 1992).
Atmospherically forced Kelvin waves (Figure 5.13a), propagating rapidly eastward, are
trapped along the equator because the eastward motion of water on either side of the
equator is deflected equatorward by Coriolis forces in each hemisphere. Figure 5.13a illus-
trates a downwelling equatorial Kelvin wave, moving eastward in the upper layer and
westward below the thermocline. Hence the wave-induced currents are counter to the
mean westward-moving South Equatorial current and the underlying Equatorial under-
current flowing eastward (Mysak, 1986). Kelvin waves at the thermocline travel at about
0 1–3 m s1 and have a Rossby radius of deformation3 of the order of 125–250 km. They
cross the Pacific Ocean in about three months. The ocean Kelvin waves are forced by
variations in zonal wind stress anomalies over the western equatorial Pacific. Two poten-
tial mechanisms have been identified. The first involves anomalous westerly winds in the
11 western/central Pacific (Harrison and Schopf, 1984). Observations of post-1950 El Niño
events show that, in nine out of ten cases, a burst of west winds of a few weeks’ duration
preceded coastal warming off South America by one to four months (Luther et al., 1983).
A climatological analysis for 1958–87 indicates that equatorial westerlies in the western
Pacific have a maximum frequency of almost 20 percent in November–December, but
low steadiness, typically lasting five to seven days (Chu et al., 1991). These episodes
0 may result from a pair of tropical storms north and south of the equator (Keen, 1982).
Spells of fourteen days or more peak in March–April, and these are more likely to generate
a Kelvin wave. Verbickas (1998) reports seven WWB events in 1982 and eight during
January–September 1997. Westerly windburst (WWB) episodes in the western tropical
Pacific have a varied character. Harrison and Giese (1991) identify 155 episodes of west-
erly wind anomalies >5 m s1 (relative to the mean monthly value) affecting two or more
western Pacific island stations. The composite characteristics for events centered between
0°S and 7°S is that they have maximum westerly winds of 8 m s1, a meridional extent
(half amplitude) of 3° and a duration of ten days if centered near the equator, or up to
5° extent and sixteen-day duration if centered near 5°–7°S. About 40 percent of the cases
0 relate to named cyclones (Keen, 1982) and half of all cases are associated with ENSO
events. A further study of WWBs for 1980–89, using ECMWF 1,000 mb winds, applies
Harrison and Giese’s classification and finds that almost half (47 percent) of the patterns
have maximum westerlies centered between 5°S and 10°S, and 37 percent between 3°N
and 7°N (Hartten, 1996). They are most common in December–February. Also, WWBs
tend to occur around 10°N in July and move southward over a twelve-month period, with
no complementary northward progression. There is large interannual variability, with a
tendency for more/less WWBs to occur during low/high SOI.
The second mechanism for generating Kelvin waves invokes near-resonant forcing
by Madden–Julian Oscillations (see section 4.8). The dominant Kelvin wave period is
0 about seventy days and the zonal wavelength is 13,000–14,000 km (the width of the
Pacific Ocean), while the MJO has low frequency components around sixty days, according
to Hendon et al. (1998). West of the dateline the two have similar eastward phase speeds
below 5 m s1. Using a simple Kelvin wave model, forced by the observed intraseasonal
variations in zonal stress, Hendon et al. simulate 80 percent of the observed intra-
seasonal variance in the depth of the 20°C isotherm – indicating vertical motion associ-
ated with Kelvin waves. The predominant forcing takes place in the western equatorial
Pacific and the Kelvin waves travel eastward, initially at less than 5 m s1, increasing to
>10 m s1 east of the dateline. Westerly windbursts still play a significant role in forcing
oceanic Kelvin waves, individually and collectively, within the MJO spectrum. Their
0 amplitude is about 0.02 N m2 with a longitudinal extent of 2,000 km, whereas the MJO
has an extent of 7,000 km but an amplitude of only about 0.015 N m2 (Hendon et al,
1998). Equatorial westerly winds in the western Pacific tend to elevate the ocean thermo-
11 cline north and south of the equator in the west, owing to latitudinal shear both northward
Figure 5.13 (a) Side and (b) rear views of a downwelling equatorial Kelvin wave pulse in a two-layer ocean generated by relaxation of the easterly trade
winds. The rear view shows a Gaussian-shaped trapping region. (c) A westward-traveling baroclinic Rossby wave in a two-layer ocean showing
the velocity pattern and thermocline structure. The wave is transverse, with current fluctuations perpendicular to the direction of phase propaga-
tion. (d) The transformation of a first baroclinic mode Kelvin wave into poleward-traveling coastal Kelvin waves and reflected equatorial Rossby
waves. The wave guides are approximately delimited by dashed lines; re = equatorial internal Rossby radius, and r = mid-latitude value (~ 35
km at ± 25° latitude, Emery et al., 1984). (From Mysak, 1986)
11 and southward, creating cyclonic vorticity. In the eastern Pacific, maximum thermocline
displacements are along the equator (Chao and Philander, 1993). However, westerly wind-
bursts are neither necessary nor sufficient conditions for ENSO events, according to Luther
et al. (1983). Moreover, based on a study for 1980–85 over the Indian and Pacific Oceans,
the frequency of westerly bursts increases after ENSO onset and such bursts are not
restricted to the western equatorial Pacific (Murakami and Sumathipala, 1989).
When Kelvin waves encounter the eastern boundary some of the wave energy is trans-
mitted polewards via coastally trapped waves, extending the equatorial waveguide to
higher latitudes, in each hemisphere, and some is reflected westward as a more slowly
0 moving Rossby wave front (Figure 5.13c). It is worth noting that coastal upwelling does
not shut off; it continues in the 50–100 m subsurface layer, but these upwelling waters
are not abnormally warm. The coastal Kelvin waves disperse (“radiate”) energy into baro-
clinic Rossby waves that propagate slowly westward along the subtropical thermocline
11 between 15° and 30° latitude (Figure 5.13b). The Rossby waves travel westward, in
response to the geostrophic balance, much more slowly than the Kelvin wave motion,
taking two years to cross the Pacific at 15° latitude and up to nine years at 30°N or 30°S
(McCreary, 1983). These time scales may help determine the irregularity of ENSO events.
Nevertheless, the faster Rossby waves at lower latitudes transport energy and mass, deliv-
ered by the Kelvin waves, in the reverse direction. At the western boundary the mass
0 flux transported by the Rossby waves is moved equatorward in the boundary currents and
then returned eastward via the Kelvin waves. Kessler (1991) demonstrates that westward-
traveling Rossby waves need to be equatorward of 8° latitude in order for reflected Kelvin
waves to have significant amplitude. Satellite altimetry data clearly identified such equa-
torial Kelvin waves prior to and during the 1986–87 and 1991–95 El Niños (Miller
et al., 1988; Kessler and McPhadden, 1995; McPhadden et al., 1998). In 1997, however,
reflected Rossby waves were not initially present, and eastward advection of warm water
was forced by MJO-induced WWBs (McPhadden, 1999).
The western coastal boundary of the tropical Pacific is discontinuous, with seven major
island complexes between 10°N and 10°S extending from about 120°E to 160°E. However,
0 Clarke (1991) and du Penhoat and Cane (1991) show that the Asia–Borneo–Indonesia
and New Guinea–Australia land masses reflect most of the Kelvin wave energy that would
be reflected eastward by a continuous meridional boundary; for example, half of the inci-
dent energy of a mode m 1 Rossby wave (symmetric about the equator) would be
reflected eastward as a Kelvin wave by a solid north–south wall, compared with 34 percent
for the island boundary. The amplitude of the reflected Kelvin wave is 85 percent of that
for a solid boundary (for m 1 incoming) and correspondingly is 74 percent for Rossby
wave mode m 3 incoming.
The sea-level amplitude, as determined by satellite altimetry data for 1986–87, shows
that mode 2 (Rossby waves) peaking at 5°N and 5°S account for about 75 percent of the
0 mode 0 Kelvin wave amplitude (White and Tai, 1992). The circumstances that prevailed
in the Pacific when the strong, protracted El Niño of 1939–41 and 1942–44 succeeded a
seven-year interval without an event are described by Biggs and Inoue (1992). There were
two episodes of wave reflection at the eastern and western boundaries. Waves arrived
first at the coasts of South America and eastern Australia and later at the west coast of
Mexico and in the northwestern Pacific. It seems that interaction between upper ocean
conditions and surface winds is critical for the transition between modes.
Coupled with these features of equatorial ocean dynamics, the warm ENSO mode
requires the availability of an equatorial reservoir of warm water (Cane, 1992). This water
is transferred eastward by Kelvin waves to initiate a warm event. The anomalous warm
0 water in the east then sets up westerly wind anomalies, forcing Kelvin waves that depress
the thermocline further in the east, thereby requiring a compensatory shallow thermocline
elsewhere. Westerly wind forcing, as described above, and westward-propagating Rossby
11 wave packets accomplish this. The thermocline thus becomes shallower and SSTs are
lowered. “Cold” Kelvin waves are now reflected from the western boundary, and several
years are necessary for the warm water pool to be recharged. Biggs and Inoue (1992)
propose that an extensive deep warm pool in the western equatorial Pacific in the late
1930s was the energy source of the El Niños of the early 1940s. Nevertheless, the 29°C
warm pool as measured in January–February 1986 between 8°S and 5°N, near 150°E,
had an average mixing layer depth of only 30 m and the thermocline at 60 m (Lukas and
Lindstrom, 1991). Low salinities due to net excess of precipitation minus evaporation
determine the density structure. Westerly winds may cool the mixed layer by increasing
evaporation and through entrainment of cooler water from below and thereby cause the
warm pool to move eastward.
The role of atmospheric pressure anomalies in the evolution of the ENSO cycle is
suggested by the characteristics of the propagating oscillatory patterns across the South
Pacific described earlier. An hypothesis linking the SO and the South Pacific Convergence
Zone (SPCZ) is outlined in Figure 5.14, based on van Loon and Shea (1985, 1987) and
Kiladis and van Loon (1988). For example, northerly wind anomalies near the dateline,
that develop with the strengthened subtropical high of a cold event, set up positive SST
anomalies within the SPCZ. This intensifies convective activity in the SPCZ, leading to
negative pressure anomalies (von Storch et al., 1988). Westerly wind anomalies then
develop over the western equatorial Pacific and may trigger El Niño onset, as outlined
above. Figure 5.14 also shows the sequence initiated by a warm event. Xu and von Storch
(1990) illustrate the applicability of this hypothesis in a hindcast analysis for 1974–88.
During a warm event, when the South Pacific high is weak, the SPCZ shifts north and
eastward, merging with the ITCZ over the central Pacific and enhancing convection
(Trenberth, 1976; Yarnal and Kiladis, 1988). Nevertheless, other prediction schemes
consider the tropical wind field and the role of equatorial waves (Graham et al., 1987).
This suggests that there is a large, complex interacting system involved in ENSO
phenomena.
Figure 5.14 Schematic relationships between the SO and SPCZ. (From Xu and von Storch, 1990)
11 The current paradigm for the warm/cold transition is characterized as a “delayed oscil-
lator mechanism,” with the oscillation memory provided by the westward propagation of
Rossby waves through wind stress, and the reflection of a Kelvin wave from the western
boundary to the eastern Pacific. This mechanism provides the basis for coupled atmos-
phere–ocean models of SST anomalies in the eastern equatorial Pacific (Graham and
White, 1988; Zebiak and Cane, 1987; Battisti and Hirst, 1989; Schopf and Suarez, 1988,
1990). ENSO behaves as a delayed oscillator because the slow dynamic adjustments of
the ocean never catch up with changes in the wind field. The Zebiak–Cane model simu-
lates the characteristics of ENSO SST anomalies when forced by composite ENSO wind
0 anomalies. When initiated with an imposed 2 m s1 westerly wind anomaly of four months’
duration, it reproduces realistic irregular three to four-year SST anomalies, with peaks of
varying amplitude, that tend to be phase-locked to the annual cycle and reach maximum
amplitude at the end of the calendar year. The model has greater forecast skill than clima-
11 tology out to fifteen months’ lead time, and beyond six months is substantially better than
persistence (Cane, 1992). In an enhanced version of the Zebiak–Cane model, Jin et al.
(1994) find that seasonal cycles generate quasi-periodic motion or irregular oscillations
in ENSO structure. ENSO can also be entrained by non-linearities into synchrony with
the annual cycle, leading to longer-period oscillations. They also note that the inherent
frequency of the system may lock onto a sequence of annual subharmonics of the external
0 frequency and this leads to a transition to chaos through the overlapping of the non-linear
resonances – a behavior referred to as “the devil’s staircase.” A detailed review of coupled
model experiments of tropical air–sea interaction is presented by Neelin et al. (1992,
1998). However, these models overlook the oceanic complications associated with a
continuing oscillation of several years’ duration (Chao and Philander, 1993). Zonal wind
fluctuations must excite a spectrum of superimposed ocean waves, making it impossible
to identify explicitly individual Kelvin and Rossby waves. Newer ocean models have been
developed (Jin, 1997; Wang and Fang, 1996), but there is as yet no single consensus
model (Neelin et al., 1998). The role of free Rossby/Kelvin waves is downplayed by
Zhang and Levitus (1997). They emphasize the importance of changes in the depth of
0 the thermocline for SST anomalies in the eastern and central Pacific. Based on compos-
ites of ocean temperatures and surface wind fields for five warm and five cold events
during 1966–99, they argue that the interannual variations in the coupled system evolve
through the slow propagation of subsurface thermal anomalies around the tropical Pacific
basin. These subsurface anomalies take two years to cross the Pacific.
The above models postulate that ENSO is triggered in the eastern Pacific, but some
recent modeling studies focus on forcing in the western and central Pacific. Moore and
Kleeman (1997) show that perturbations in a coupled ENSO model grow most rapidly in
the western and central equatorial Pacific. Error growth in the ocean can originate in asso-
ciation with different combinations of upwelling (U), thermocline (T) displacement, and
0 ocean heat advection (A), but the resulting SST patterns appear to have a preferred
response to these various processes acting in the mixed layer. In the atmosphere, error
growth is initiated by SST anomalies, but requires SSTs > 28°C, deep penetrative convec-
tion, and latent heat release through moisture flux convergence. Recall that the Clausius–
Clapeyron relationship is more effective at the higher temperatures in the western
compared with the eastern equatorial Pacific. Including only the observed seasonal cycle,
the perturbation energy growth associated with the three terms together reaches a
maximum in the model during JAS. However, the effect of T is largest in AMJ, U in
JAS, and A in OND. The ENSO cycle can modify perturbation growth. During the
different phases of an El Niño (or La Niña) event the mechanisms creating perturbation
0 growth vary geographically in their relative contributions to growth, thereby forcing longi-
tudinal differences in the evolution of ENSO. Moore and Kleeman also show that the
patterns of wind perturbation over the tropical western Pacific Ocean in their model
11 resemble observed tropical cyclone pairs on opposite sides of the equator. Such cyclone
pairs are associated with westerly windbursts, and they propose that westerly and east-
erly windbursts are a preferred response to stochastic variability in the coupled system.
The termination of an El Niño warm anomaly is generally preceded by the return of the ther-
mocline to shallow depths in the central and eastern equatorial Pacific (Harrison and Vecchi,
1999). The “delayed oscillator” model depends on this process for the removal of positive SST
anomalies. The reason for the rise of the thermocline is thought to be either a direct result of
wind forcing or equatorial Kelvin wave forcing in the ocean. The southward shift of westerly
wind anomalies, from a distribution that is symmetrical about the equator in October–
November to one that is centered south of the equator in December–January, is simulated in
an oceanic GCM by Harrison and Vecchi. This southward shift is a response to the normal sea-
sonal displacement of waters warmer than 28°C south of the equator by December–January.
The southward movement of warm water, which is present in both “normal” and El Niño
years, seems to be responsible for the thermocline returning to shallow depths.
Analyses of air–sea interaction in the equatorial Atlantic (Zebiak, 1993) show the exis-
tence of a coupled mode resembling ENSO. It is farther west in the Atlantic basin than
its Pacific counterpart and less robust. However, the oscillation periods are similar; the
difference in basin size is apparently offset by the strength of the air–sea coupling and
differences in zonal structure.
The tendency for decadal scale modulations of ENSO to occur was noted in section
5.2. Mechanisms to explain these modulations are still not yet understood. Two processes
involving linkages from mid-latitudes into the tropics have been suggested. One invokes
the generation of wind anomalies in mid-latitudes that extend into low latitudes and force
the ocean circulation (Barnett et al., 1999). The other proposes the formation of SST
anomalies, by latent heat flux anomalies in mid-latitudes, and their advection southward
within the subsurface branch of a shallow meridional ocean cell in the North Pacific (Gu
and Philander, 1997). A different ocean teleconnection is suggested by Kleeman et al.
(1999), in which wind speed anomalies in the northeastern subtropical Pacific create LE
flux anomalies. These provide a positive feedback for the decadal oscillation, while a
delayed negative feedback results from anomalous horizontal advection and overturning.
Observational data are so far inadequate to resolve these arguments.
Analysis of irregular ENSO oscillations via stochastic forcing of a non-linear dynam-
ical coupled model suggests that ENSO can respond to different mechanisms in various
stages of its evolution, depending on the basic state and the non-linear dynamics (Wang
et al., 1999). ENSO oscillations may be (1) irregular, when stochastic forcings act on a
stable basic state; (2) unstable and non-linear (a stable limit cycle), when stochastic forcing
perturbs the limit cycle; or (3) bi-stable, when stochastic forcing causes multi-equilibrium
states to oscillate irregularly between warm and cold stable states. White and red noise
are shown to be more effective in stochastic resonance than band-limited white noise in
modifying the non-linear oscillations. Strong white noise forcing can, however, destroy
resonant or non-linear oscillations.
5.4 Teleconnections with ENSO

Meteorological and oceanographic variables in many areas of the world show high corre-
lations with the SO in its core regions of the Indian Ocean–Pacific Ocean. These
teleconnections both lead and lag events in the Pacific sector. For example, Wright (1986)
reports that warm water in the southeast Pacific, accompanied by low pressure, low cloudi-
ness, and weak southeast trades in December–February is associated with a low pressure
value at Darwin in the following twelve-month April to March period. In contrast, warm
waters in the equatorial Atlantic in December–February are associated with high twelve-
month pressure at Darwin in the following April to March.
The effects of an ENSO warm event on the circulation over the North Pacific were first
outlined by Bjerknes (1966, 1969). According to his hypothesis, the warm equatorial
11 ocean surface produces enhanced convective activity, and the release of latent heat in the
cloud systems augments the poleward fluxes of heat and momentum in the Hadley circu-
lation, which, in turn, lead to an intensification of the extratropical jetstreams and an
enhanced Aleutian low. Symmetrical effects are observed over the western tropical Pacific,
where the intensity of the Australian subtropical jet is also increased (Nogués-Paegle and
Mo, 1988).
An important association with the western North Atlantic–Caribbean Sea is shown by
the reduction in frequency of tropical cyclones during El Niño years (Gray and Sheaffer,
1991). For the Atlantic cyclone seasons of 1900–88, the average number of tropical storms
0 and hurricanes for El Niño years is only 5.4, compared with 9.1 in non-El Niño years;
moreover, the corresponding ratio for major hurricanes is 1:3. The mechanism respon-
sible for the decreased activity during El Niño events involves the occurrence of an
anomalous upper tropospheric westerly flow over central America and the Caribbean. This
11 is a result of enhanced deep convection in the eastern equatorial Pacific. The anomalous
flow creates unfavorable conditions of vertical wind shear for hurricane formation in the
western Atlantic and Caribbean.
Climatic anomalies associated with extremes of pressure in the Indian Ocean sector
were recognized in the late nineteenth century by meteorologists working in India and
Australia (Diaz and Kiladis, 1992). Walker (1923) subsequently demonstrated that
0 droughts in India and Australia accompany a negative SOI, as well as cool, rainy winters
in the southeastern United States. The variability in annual precipitation is one-third to
one-half greater in areas immediately affected by ENSO events and those linked by consis-
tent teleconnections than in other parts of the world, according to Nicholls (1988).
Subsequently there have been many investigations of global and regional patterns of
precipitation anomalies associated with ENSO (Stoeckenius, 1981; Behrend, 1987;
Ropelewski and Halpert, 1987; Lau and Sheu, 1988; Kiladis and Diaz, 1989). The evolu-
tion of global anomalies of precipitation during warm events (June–August of year 0 to
December–February of year 1) is illustrated in Figure 5.15. The maps show the composite
anomaly for warm minus cold events so that W wetter and D drier than normal
0 during warm events, with the opposite signs for cold events. Summer droughts are indi-
cated for India, particularly over the peninsula (Kiladis and Sinha, 1991), and the Ethiopian
Highlands (Quinn, 1992), and over Australia in the following seasons. Figure 5.15d
demonstrates the widespread extent of positive anomalies in the tropics during December–
February of year 1 following warm events. Over the Americas the strongest signals are
above-average winter and spring precipitation over southern South America in year 0 of
a warm event. In North America cool, wet winters over the south and southeastern United
States follow the warm ENSO mode. A study of precipitation anomalies associated with
SO phase shows that in fifteen out of nineteen regions around the globe there are opposite
anomalies between high and low SO and that the relations were consistent in thirteen out
0 of nineteen years, with high SO mode between 1885 and 1983 (Ropelewski and Halpert,
1989). The precipitation–SO relationship also occurs in the same seasons for both high
and low SO in thirteen of the fifteen regions.
Long-term, global-scale associations of air temperature with “warm” and “cold” ENSO
events, and the seasons of maximum association, have been documented by Halpert and
Ropelewski (1992). They find that the temperature anomalies associated with ENSO tend
to be of opposite sign for the two phases, especially in the tropics, where they have the
same sign as the local anomaly of SST. Outside the tropics, the SO–temperature associ-
ation may only be evident regionally for one phase. Japan and western Europe/North
Africa exhibit an SO–temperature relationship only during cold events, while the south-
0 eastern United States has an association only during warm events, despite having drier
than normal conditions in cold events.
Ten ENSO events between 1950 and 1988 have been differentiated according to their
11 duration. Using the persistence of positive SST anomalies, Tomita and Yasunari (1993)
Figure 5.15 Composite differences for warm minus cold ENSO events for precipitation (a–h) and temperature (i–p) in the eastern hemisphere and the Americas
for the seasons SON 1, DJF 0, SON 0 and DJF 1, where the event occurs in year 0. W wetter, D drier than normal; A above and B below normal
temperature. Solid symbols represent significant differences at the 1 percent level, open symbols at the 5 percent level. (From Kiladis and Diaz, 1989)
identify one group (1951, 1953, 1963, 1965, 1972, and 1982) where the event lasts about
a year and includes one boreal winter, and a second group (1957, 1968, 1976, and 1986)
which includes two boreal winters and persists for more than two years. Another cate-
gorization, by Wang (1995), distinguishes changes in the onset phase and SST anomaly
patterns between the events of 1957, 1965, and 1972, on the one hand, and 1982, 1986/87,
and 1991, on the other, which moved eastward to set up anomalous westerlies in the
western equatorial Pacific. In the earlier group, the onset phase (in November–December
of year 1) featured a large anomalous anticyclone over eastern Australia. These cases
also showed warming off South America for three seasons prior to warming in the central
Pacific. In the later group there was an anomalous low over the Philippines during the
onset phase which generated westerlies. Coastal warming occurred only after that in
the central Pacific. These differences appear to reflect the control of the background SSTs
in the Pacific. A sharp interdecadal warming in the equatorial Pacific waters took place
in the late 1970s, with cooling in the northern and southern extratropical latitudes.
Associated changes occurred in the onset cyclone, the western Pacific westerly anomalies,
and the southeast Pacific trades which resulted in different modes of warming in coastal
South America and the central Pacific (Wang, 1995).
Composite analysis of El Niño and La Niña events for northern winters 1950–96 indi-
cates non-linearity in their anomaly patterns. There is a 35° longitude phase shift in eddy
500 mb height anomalies, respectively westward/eastward, during warm and cold events
(Hoerling, et al., 1997). The wave trains appear to originate in different parts of the tropics
since the positive anomalies of tropical rainfall occur east (west) of the dateline during
warm (cold) events. However, it is also noted that the composite SST anomalies are not
exact inverse counterparts of one another.
A fundamental question concerning these worldwide teleconnections with the ENSO
signal is: what mechanisms link the regional-scale forcing of sea surface temperature
anomalies in the equatorial Pacific to circulation anomalies that extend into the extrat-
ropics and vertically throughout the troposphere (Tribbia, 1991)? Simple models of the
circulation response to an equatorial heat source reproduce the main elements of the zonal
Walker cell, the north–south Hadley cell, and low-level cyclonic flow anomalies to the
west of the local heating source (Heckley and Gill, 1984) (Figure 5.16). Observational
wind data for September–October 1972 and September 1982 match this model quite well
near the equator (Barnett, 1984). The anomalous equatorial heat source generates low-
level convergence and upper-level divergence (a thermally direct circulation). This upper-
level divergence serves as a source of vorticity for a wave train which becomes a stationary
pattern. Horizontal propagation of the anomalies from a vorticity source region to mid-
latitudes has been analyzed by extending the wave equation of C.-G. Rossby (p. 286) to
a sphere, where the latitudinal variation of the Coriolis parameter causes the latitudinal
changes in the direction of energy propagation and the wavelength. Hoskins and Karoly
(1981) use wave ray paths to trace the propagation of wave energy, initially poleward
and then recurving to cross the equator. A critical limit is reached where the zonal phase
speed of a Rossby is equal to the zonal flow; waves cannot propagate through this region.
Figure 5.17 illustrates that waves with smaller wave numbers (1–3) can penetrate further
poleward before they are reflected (Lau and Lim, 1984). It takes between seven and forty-
two days for wave energy to propagate from an equatorial source to its critical latitude
for wave Nos 6 and 1, respectively, for a zonal flow of 10 m s1 and phase speed of
120 m s1. For zonally uniform westerly background flow, wave trains of dispersing
stationary Rossby waves move eastward following great-circle paths. Tribbia (1991) points
out that the rays emitted from a stationary wave source are absorbed in regions of mean
easterly flow. In the northern hemisphere summer the easterlies extend into the subtropics.
Thus larger extratropical responses may be expected during the respective winter seasons.
Moreover, cross-equatorial Rossby wave propagation is possible where upper-level extra-
tropical westerlies extend across the equator, as in the eastern equatorial Pacific and
11
11
Figure 5.16 Schematic illustration of equatorial low-level flow and vertical motion associated with
asymmetrical heating north of the equator over a region of about 40° of longitude.
For a steady state, after at least ten days, the circulations in the two hemispheres are
almost independent. In the model the wave front propagates westward at about 8°
longitude per day. (From Heckley and Gill, 1984)
0 Atlantic sectors during the austral summer (Kiladis and Mo, 1998). These regions are
termed “westerly ducts” by Webster and Holton (1982).
Subsequent work indicates limitations in these interpretations. The mid-latitude circu-
lation anomalies are predicted by the theories of Gill (1980) and Hoskins and Karoly
(1981) to shift in location, in response to the longitude of the equatorial SST anomaly,
but a lack of such sensitivity is apparent (Simmons et al., 1983; Geisler et al., 1985).
This arises because the disturbances produced by a vorticity source are able to extract
energy from the mean upper-level flow and thereby amplify. Simmons et al. (1983) show
that in the northern hemisphere winter, when there are large-amplitude stationary long
waves, the circulation structure that does this most efficiently closely resembles typical
0 teleconnection patterns. Moreover the extraction of energy is greatest off the east coasts
of the continents in subtropical jet exit regions. Tribbia (1991) points out that although
vortex stretching is the principal vorticity source for the anomaly patterns, this contribu-
tion is small off the east coasts, remote from the SST anomaly. The locally strong vorticity
gradients off the east coasts dictate that convergence of the transport of mean vorticity
by the divergent wind, induced by the SST anomaly, becomes an important term
(Sardeshmukh and Hoskins, 1985). Hence there is no inconsistency in the fact that the
anomalies associated with ENSO events are in the central eastern Pacific. In summary,
the teleconnection mechanism according to Tribbia (1991) involves the following steps:
0 1 A locally warm ocean surface sets up anomalous low-level convergence (see Figure
5.16), enhancing precipitation.
2 Enhanced precipitation increases mid-tropospheric release of latent heat, generating
11 anomalous upper-level divergent flow and, through non-linear relationships, absolute
a.
b.
Figure 5.17 Ray paths on a sphere for a zonal basic flow of 10 m s1 for wave Nos 1–8 from an
equatorial heat source. (a) Non-divergent flow. (b) Divergent motions with a phase
speed of 120 m s1. (From Lau and Lim, 1984)
vorticity. This divergent component of flow is illustrated by the mean velocity poten-
tial at 200 mb (Figure 5.18) (see Chapter 3, note 3).
3 Vorticity transport created by the anomalous upper-level outflow excites instability
of the barotropic flow, extracting energy from the climatological mean flow off the
east coasts of Asia and North America. Recent work suggests that vertical wind shear
is required to complete the link between upper-level divergent flow, which creates
vorticity by vortex stretching, and the energy conversion to the vertically averaged
flow.
4 A train of dispersing, almost stationary, Rossby waves emanates from this region of
energy extraction.
Thus tropical forcing sets up a geographically fixed circulation pattern. For an initial cold
anomaly, the subsequent anomalous upper-level inflow in step 3 above also excites circu-
lation instability.
Additional variations in the forcing are introduced by the north–south seasonal migra-
tion of the regions of upper tropospheric outflow over South America and central Africa
(Rasmusson, 1991). Tropical convection over the “maritime” continent of Indonesia–
Malaysia, which promotes divergent outflow into the Walker and Hadley circulations, also
oscillates seasonally (Meehl, 1987a). These seasonal fluctuations cause variations in the
intensity and location of the vorticity sources and the ensuing large-scale circulation
patterns. As noted earlier, there is an eastward progression of tropical convective activity
from India (in July) to northern Australia and the southwest Pacific for the austral summer
11
11
0
Figure 5.18 Divergent component of flow at 200 mb as illustrated by the mean velocity potential
for December–February 1986/87–1988/89 (above) and June–August 1986–88 (below);
contour interval is 1 × 106 m2 s1. (From Rasmusson, 1991)
monsoon (Meehl, 1987a). The SPCZ in turn may link the tropical anomalies with higher
southern latitudes.
The linkage between the South Asian summer monsoon and ENSO is still being
explored. However, it was recognized over forty years ago that variations in Indian
monsoon activity appear to lead those in the Southern Oscillation rather than the converse
0 (Normand, 1953). Monsoon anomalies lead tropical SSTs by about six months (Yasunari,
1991). Monsoon precursors of ENSO are very epoch-specific (Kumar et al., 1999). Such
correlations occurred in 1951–90, but not during 1911–50, nor after 1991. From GCM
experiments, the Asian monsoon appears to be coupled to the Southern Oscillation via
an upper-level circulation couplet which acts as a radiating node for teleconnection signals
to be transmitted from the monsoon area into the extratropics (Lau and Buc, 1998). Weak
Asian monsoons are associated with a warm eastern equatorial Pacific Ocean and a nega-
tive SO, with low pressure over Tahiti and the reverse for a strong monsoon. There are
two wave trains associated with the monsoon fluctuations. One extends over northeast
Asia via the Aleutians to North America and the other extends from northwest Europe
0 via Siberia to northern India. Webster and Yang (1992) point out that the summer monsoon
is developing rapidly in spring at a time when zonal equatorial pressure gradients are
minimal and lagged correlations with SOI are decreasing rapidly. The center of convec-
11 tive activity (as depicted by minimum outgoing long-wave radiation) shifts northwestward
Figure 5.19 Gradients in latent heat (LE) and radiative flux convergence (Rn) between the main
branches of the South Asian summer monsoon and the near-equatorial circulations.
The zonal (“transverse monsoon”) has gradients that are one-third larger than merid-
ional (“lateral monsoon”) or the Walker circulation. (From Webster, 1995)
from the western equatorial Pacific “warm pool” region in winter to East and South
Asia in June–July while the areas of strong radiational cooling (maximum OLR) over
North Africa and the Middle East remain stationary (Webster, 1995). Figure 5.19 illus-
trates the boreal summer pattern of zonal (“transverse monsoon”) and meridional (“lateral
monsoon”) heating gradients due to latent heat (LE) and radiative flux convergence
(Rn) between the main branches of the South Asian monsoon and the near-equatorial
circulations. The link between the Asian monsoon and north African–Arabian deserts
is apparent.
Empirical analyses and coupled model studies all indicate a “predictability barrier” in
the northern hemisphere spring for the tropical ocean–atmosphere system; that is to say,
forecasts for time periods straddling this season have no skill. An explanation of this
decrease in forecast skill, and the observed corresponding decline in lagged SOI correla-
tions (Figure 5.20), has been sought in two hypotheses. The first (Figure 5.21a) suggests
that there is an inherent predictability limit of the coupled ocean–atmosphere system which
is set by the fragility of the near-equatorial (ENSO) circulation in the boreal spring and
its susceptibility to random perturbations. This hypothesis is examined by Torrence
and Webster (1998), using historical SST and pressure data. They show that the strength
of the predictability barrier is controlled by the degree of phase locking of ENSO to the
annual cycle. Anomalies of the SST and the SOI show a slow decrease in autocorrelation
within each series over several months. However, defining persistence in terms of the
fixed-phase correlation between pairs of months, there is a sharp decrease in persistence
in April for the SOI and in June for Niño 3 SST. Statistical modeling suggests that the
persistence barrier occurs because boreal spring represents a time of transition from one
state of the atmosphere to another with a low signal/noise ratio. There are significant vari-
ations in the degree of persistence on interdecadal scales. For 1871–1920 and 1960–90,
ENSO variance is large and the persistence barrier is strong. In contrast, ENSO variance
is low during 1921–50 and the barrier is weak. Such fluctuations in persistence may
simply result from stochastic variability in the wind forcing on SST persistence, according
to Flügel and Chang (1999). Weiss and Weiss (1999) examine the statistics of persis-
tence in ENSO, particularly its annual phase-dependence. A persistence barrier may arise
where the frequency of a sine wave function is a biennial cycle, or one of its harmonics.
Their analysis indicates that the barrier in the SOI series and Niño 3 region SSTs is statis-
tically distinguishable from that of a red-noise process. They also conclude that the barrier
was weak from about 1915 to 1945, strong in the 1960s and early 1970s, and weakened
in the late 1970s.
11
11
Figure 5.20 Lagged correlations of the mean monthly SOI show the rapid decline during boreal
spring (boxed area). (From Webster and Yang, 1992)
The second hypothesis (Fig 5.21b) invokes the role of external influences, particularly
an inverse coupling in the strength of the South Asian monsoon and the Pacific trades.
0 Another factor could be the extent of Eurasian snow cover in spring. Anomalously
strong/weak monsoons are associated with stronger/weaker summer trade winds over the
Pacific Ocean. Moreover, annual cycle composites of outgoing long-wave radiation for
strong/weaker monsoons show coherent patterns over southern Asia and the tropical Indian
Ocean during the previous winter–spring seasons. The Asian monsoon and Walker circu-
lation appear to have a six-month phase difference and to be selectively interactive,
according to Webster and Yang (1992). In spring the developing monsoon dominates the
near-equatorial Walker circulation. In the boreal autumn–winter the monsoon is weakest,
with convection near the equator. The Walker circulation is now strongest and may domi-
nate the winter circulation. Figure 5.22 illustrates the composite variation of the monthly
0 SOI over twenty months during strong, normal and weak monsoons between 1871 and
1992 (sixteen in each category) (Lau and Yang, 1996). These several studies are comple-
mentary in suggesting that a combination of influences are responsible for the springtime
predictability barrier (Figure 5.21c). Moreover a weak Asian monsoon is associated
with strong tropical predictability from May to August but a sharp decline for predictions
from September to April, whereas a strong monsoon is associated with a recovery of
predictability from September to April (Lau and Yang, 1996.) Linkage of the two
hypotheses is apparent, since Torrence and Webster (1998) find from wavelet analysis
that interdecadal variations in ENSO are correlated with changes in the strength of the
Indian monsoon.
0 According to Frederiksen and Webster (1988), remote linkages to the extratropical
circulation associated with anomalous tropical heating represent one type of teleconnec-
tion mechanism. Visual evidence of the extratropical teleconnection with convective
11 activity in the central and eastern equatorial Pacific is given by the appearance of elongated
(c)
(d)
(a) (b)
Figure 5.21 Schematic representations of hypotheses to account for the decrease in boreal spring tropical correlations and the associated spring predictability
barrier. (a) Hypothesis 1, showing oscillations of ENSO between two regimes (upper). The vertical amplitude of the annual cycle (upper)
determines the “robustness” of the system. There is a weak restraint in boreal spring (middle) but a strong restraint in boreal summer (lower).
(b) and (c) Hypothesis 2 invokes coupled variability in monsoon and Pacific Ocean trade wind strengths. The annual cycle and ENSO are
regarded as distinct systems. (d) A unified hypothesis combining internal and external influences. (Webster, 1995)
11
11
Figure 5.22 Composite variation of the monthly SOI for sixteen cases each of strong, normal, and
weak monsoon. The normalized Indian monsoon rainfall index values (Parthasarathy
et al., 1991) for each category are shown in parentheses. (From Lau and Yang, 1996)
0
plumes of moisture and cloud (“moisture bursts”) in satellite imagery (Iskenderian, 1995).
These bands connect with the subtropical jet on their poleward edge. A study of four
cold seasons suggests that these features comprise a transient Hadley circulation (McGuirk
et al., 1987). However, the cloud bands in southwesterly flow occur ahead of upper-level
troughs that propagate into the tropics over the eastern North Pacific in winter from the
exit region of the East Asian jet (Kiladis and Weickmann, 1992a, b; Kiladis, 1998).
Extratropical teleconnections are discussed more fully below (section 5.8). The occurence
of positive SST anomalies in remote ocean areas (the South China Sea and the Indian
Ocean) three to six months after maximum positive anomalies in the tropical Pacific are
0 linked with the effects of changes in atmospheric circulation and energy fluxes by Klein
et al. (1999). They suggest that the circulation changes lead to reductions in cloudiness
or evaporation, which increases the absorption of solar radiation and warms the surface.
In the tropical North Atlantic the SST increase is attributed to weaker trade winds reducing
evaporation. Such linkages provide evidence for a so-called “atmospheric bridge.”
A second type of teleconnection mechanism involves extratropical effects on low lati-
tudes. Observations and theory show that mid-latitude disturbances can propagate into
low latitudes in regions of equatorial westerlies (Webster, 1983). During the northern
winter, westerlies in the upper troposphere of the eastern Pacific are sufficiently strong
to permit propagation across the equator into the southern extratropics (Kiladis and
0 Weickmann, 1992b, 1997). Interactions in the lower troposphere are particularly impor-
tant in the western Pacific during northern winter, when surges of cold air penetrate into
the tropics (Arkin and Webster, 1985). Their antecedents include the intensification of the
East Asian jetstream (Lau and Lau, 1984) and the development of organized wave trains
over the extratropical North Atlantic some six or seven days earlier (Joung and Hitchman,
1982; Compo et al., 1999). If the cold surge is strong, the enhanced subsidence may
suppress convection farther east in the tropics. However, in the case of weak or moderate
surges, moisture and energy may be advected eastward by upper westerlies and, ulti-
mately, towards extratropical latitudes in longitudes of the central and eastern Pacific (Lau
et al., 1983). Two cases of cold advection into the tropical Pacific during northern winter
0 that forced convection within the eastern Pacific ITCZ and the SPCZ are illustrated by
Kiladis and Weickmann (1992).
A third type of linkage – acting within the tropics – is suggested by Fredericksen and
11 Webster (1988) on theoretical grounds. Satellite observations of OLR reveal intraseasonal
variations of tropical convection, particularly in the Indian Ocean, on time scales of forty
to fifty days (the forty- to fifty-day oscillations; see section 4.9). Strong changes in the
eastward propagation of the convection patterns into the Pacific, between ENSO warm
event and non-ENSO years, may imply that the forty- to fifty-day oscillations help trigger
this lower-frequency ENSO (Lau and Chan, 1986). This hypothesis has been modified by
the result of modeling (Zebiak, 1989), which suggests that intraseasonal variations may
be important at certain times but, on average, do little to alter the evolution of ENSO as
predetermined by the initial conditions. The biennial signals that are apparent in the
ocean–atmosphere system of the tropical Indian and Pacific Oceans, and involve modu-
lations of the annual convective cycle (Meehl, 1993, 1997), may be another example of
intratropical linkage.
A biennial signal has been identified in a Tropic-wide Oscillation Index (TOI) defined
by the first EOF of normalized rainfall anomalies over 20°S–40°N (Navarra et al., 1999;
Miyakoda et al.,1999). Teleconnections within the tropics are examined using observa-
tions (gridded precipitation for 1961–94, SSTs, ECMWF 500 mb heights, MSL pressure,
and vertical velocity) and corresponding ECHAM-4 model runs. The TOI best charac-
terizes the ENSO/Asian monsoon oscillation in July–September for years with large SST
anomalies and is an effective precursor of SST and vertical motion in the eastern equa-
torial Pacific, representing El Niño intensification in the following November–December–
January. Whereas the teleconnection pattern of pressure is a dipole across the Pacific
Ocean to Indonesia, the patterns of teleconnection for SST, precipitation, and vertical
motion show a horseshoe pattern from the SPCZ, through Indonesia, and northeastward
into the western North Pacific. Omitting eight out of thirty-four years with near-normal
SSTs and Walker circulation over the equatorial Indo-Pacific Ocean (45°E–180°–82°W),
the TOI for July–September shows a biennial period, apart from 1971. Moreover, in line
with earlier work, wet/dry Indian monsoons can be projected fifteen months previously
(May the previous year) with good success, based on SST anomalies in the eastern equa-
torial Pacific Ocean. It should be noted that the two states of coupling/decoupling of
ENSO and the monsoon defined by the Walker circulation (SST) index in these studies
are distinct from the two types of ENSO proposed by Tomita and Yasunari (1993).
5.5 Extratropical teleconnection patterns

The primary teleconnection patterns are now routinely identified by the Climate Analysis
Center, NOAA, using a rotated principal component analysis as proposed by Barnston
and Livezey (1987). Indices are calculated for each calendar month based on 700 mb
monthly height anomalies for January 1964–July 1994; they are determined from the
height anomaly fields for the three-month period centered on a given month. Time series
are displayed as standardized amplitude (mean zero, variance 1.0) determined simul-
taneously for each calendar month such that the combined sum of their products with the
corresponding pattern eigen vector explains the maximum spatial structure of the observed
monthly height anomaly field.
The major patterns and their seasonal occurrence in the northern hemisphere are as
follows:
North Atlantic Oscillation (NAO) All months

East Atlantic (EA) pattern September to April
East Atlantic Jet (EA-Jet) pattern April to August
West Pacific (WP) pattern All months
East Pacific (EP) pattern All months except
August–September
Pacific/North America (PNA) pattern All months except June–July
North Pacific (NP) pattern March to July
11 Pacific Transition pattern May to August
East Atlantic–Western Russia (Eurasia, EU) pattern September to May
Scandinavia (SCAND) pattern All months except June–July
Polar/Eurasian pattern December to February
Tropical/Northern Hemisphere (TNH) pattern November to January
Asian Summer pattern June to August
The EA and SCAND patterns, as well as two others – over southern Europe and the
northern Atlantic (SENA), and over the Bering Sea–central North Pacific Ocean (BER)
0 – were first identified in an RPCA of sea-level pressure data for 1899–1986 by Rogers
(1990). Monthly NAO and Eurasian (EU) indices have been reconstructed back to AD
1675 by canonical correlation analysis based on observed station pressure, temperature
and precipitation values, and multiple proxy records (Luterbacher et al., 1999). The predic-
11 tive skill is highest in autumn and winter, but the indices are considered reliable except
for summer before about AD 1750. Figure 5.23 illustrates the modes for five primary
patterns in the northern hemisphere, based on ten-day mean 700 mb heights (Barnston
and Livezey, 1987). Several of these patterns are now discussed in more detail.
0 5.6 North Atlantic Oscillation

The North Atlantic Oscillation (NAO) involves a negative correlation in winter months
between sea-level pressures in the subtropical Atlantic high and the Icelandic low. For
example, a correlation of 0.76 was obtained for the winters of 1963–77 between the
pressures at 65°N, 20°W and those in the region of the Azores, with analogous negative
correlations at 500 mb (Wallace and Gutzler, 1981). The gradient of sea-level pressure
between the Azores high and the Icelandic low ranges from 9 mb/1,000 km in
December–January to 5.5 mb/1,000 km in May–August (Mächel et al., 1998). The NAO
is strongest in winter and weakest in autumn, but it is unambiguously evident each month
of the year (Rogers, 1990). An NAO index can be defined, following Walker (1924), as
0 the normalized mean sea-level pressure anomaly for Ponta Delgada, Azores, minus that
for Akureyri, Iceland (Rogers, 1984). A time series of this index for winters since 1879
suggests quasi-decadal oscillations and a trend toward negative values from the early
twentieth century to the 1960s, followed by more recent positive peaks and a general
positive mode since 1980 (see Figure 5.24) (Koslowski and Loewe, 1994). The NAO
index has recently been extended back to 1820, using pressure data from Gibraltar and
a composite south-west Iceland series (Jones et al., 1997). Their record identifies inter-
vals of lower correlations between southwest Iceland and the Azores, Lisbon, or Gibraltar
during 1821–60 and 1951–95, especially during spring, summer, and autumn, implying
variability or shifts in the pressure centers. Wavelet analysis of the 315 year reconstructed
0 NAO and EU indices by Luterbacher et al. (1999) indicates low-frequency periodicities.
There are significant periods in the NAO (Azores–Iceland) index around ninteen to twenty-
three years and fifty to sixty-eight years in spring, and fifty-four to eighty-eight years in
summer. The annual (April–March) mean NAO has a fifty-four to sixty-eight-year period.
In the EU (Great Britain–Black Sea) index, the periods are sixteen to twenty-two years
in spring, and six or seven years and twenty-two to twenty-eight years in summer. There
is a general positive correlation between the NAO and EU indices in autumn and winter:
strong (weak) Atlantic westerlies occurring with strong (weak) northerly flow over central
Europe, except during the mid-nineteenth century.
From an EOF analysis of hemispheric MSL pressure anomaly maps for January, it
0 appears that the NAO is essentially represented by the first eigen vector pattern (Kutzbach,
1970). Thus it has hemispheric significance. Comparisons between composite anomalies
for the Southern Oscillation and the NAO for 1900–83 suggest a weak tendency for strong
11 Atlantic westerlies to occur simultaneously with high SOI and dry conditions in the equa-
Figure 5.23 The primary northern hemisphere teleconnection patterns identified by rotated PCA,
using ten-day mean 700 mb height patterns. The modes are: (a) PNA in January. (b)
TNH in January. (c) EU in January. (d) NAO in April. (e) Subtropical zonal in July
(see text). (Barnston and Livezey, 1987)
11
11
Figure 5.24 Variations of the normalized NAO index for winters 1874–2000, based on tabulations
0 provided by J.C. Rogers. The indices are based on the monthly sea-level pressure
difference between Ponta Delgada, Azores, and Akureyri, Iceland. The pressure data
are normalized by calculating the monthly anomalies from the 1874–1999 mean divided
by the mean monthly standard deviation for the same period. The standardized value
for iceland is then subtracted from that for the Azores, so that positive indices repre-
sent strong westerlies. The figure shows the individual winter values and an unweighted
nine-year moving average.
torial Pacific and for Atlantic blocking during the low SOI/wet (ENSO) mode (Rogers,
1984). However, the two oscillations are more frequently unassociated. Rogers notes that
0 spectral estimates for the NAO show peak energy at 7.3 years, compared with about six
years for Darwin pressures. The percentage of surface area north of 20°N over which
pressure differences are statistically significant between extremes of the NAO and SO is
37 percent for the NAO and 29 percent for the SO, but the significant area in common
is only about 10 percent during 1900–42 and 1943–80, and only 3 percent of the surface
area is involved for the two oscillations for both time periods. Cross-spectral analysis of
NAO and ENSO indicates that the coherence between them is dependent on both the
frequency band and the year (Huang et al., 1998). There is significant coherence between
the NAO and SSTs in the Niño region (see p. 367) in nineteen out of twenty-seven El
Niño years during 1900–95. Nine of these cases were associated with five to six-year
0 period events and ten of them with a two to four-year period. During these events the
initial winter had a dominant positive Pacific North American teleconnection pattern (see
section 5.9). In the eight cases with little or no NAO–ENSO coherence, SST anomalies
in the Niño 3 region were weak. These winters had a characteristic strong negative NAO
teleconnection pattern.
The NAO pressure anomalies are apparent not only at the surface. There are corre-
sponding height anomalies of the same sign at the 500 mb level, as illustrated in Figure
5.25 showing differences in heights between winters with strong and weak NAO indices
(Rogers, 1984). As noted by Walker and Bliss (1932), a deep Icelandic low gives rise to
strong advection of cold air over Baffin Bay–west Greenland and mild southwesterly flows
0 over northwestern Europe (Figure 5.26). The NAO creates a characteristic seesaw pattern
of winter temperature anomalies in western Greenland and northern Europe that was first
recognized in Greenland and Denmark in the eighteenth century. Loewe (1937) and van
11 Loon and Rogers (1978) have detailed the seesaw modes and their relationship to circu-
Figure 5.25 The difference in height (m) at 500 mb between winters when the NAO index is above
and below normal, based on the period 1947–83. Eleven winters were above the normal
NAO index, and fourteen below. (From Rogers, 1984)
lation anomalies (Figure 5.26). When Greenland is warmer than normal (“Greenland
Above”), pressure anomalies average 9 mb over Denmark Strait, 3 mb over southern
Europe and 6 mb over the North Pacific Ocean. Almost the reverse pattern is observed
when Greenland is colder than normal (“Greenland Below”); the pressure anomalies are
6 mb north of Iceland and 5 mb in the North Pacific. Strong (weak) NAO winters
correspond to the seesaw modes Greenland Below (Greenland Above), respectively,
although the match is imperfect as a result of differing definitions. Figure 5.27 illustrates
the anomaly patterns of pressure, winds, temperatures, and relative humidity for strong
and weak NAO winters (Kapala et al., 1988). Storm tracks are strongly concentrated from
eastern Canada northeastward to Iceland and the Norwegian Sea for positive NAO,
whereas for the negative mode they are highly variable: many move into the Labrador
Sea, others into northwest Europe, and some through the Mediterranean and into eastern
Europe and Russia.
An association between the NAO and ocean circulation in the western North Atlantic
has recently been suggested. Taylor and Stephens (1998) find that almost 50 percent of
the variance in annual mean latitude of the Gulf Stream between 65°W and 79°W is
predictable from NAO intensity. A positive NAO mode with stronger westerly and trade
winds gives rise to a more northerly path of the Gulf Stream two years later. Persistence
explains a further 10 percent of the variance. The cause of the time lag is uncertain but
may represent the time needed for the ocean gyre at 35°N to adjust. Some of the unex-
plained variance seems to be attributable to the Southern Oscillation (Taylor et al., 1998).
Two years after ENSO events the mean latitude of the Gulf Stream is 0.2° farther north-
ward than in years without an event. A slight southward shift following La Niña years
is not statistically significant.
An analysis of the composite patterns of eddy sensible heat associated with NAO events
for the time period 1948/49–1979/80 finds the anti-phase behavior of the Icelandic and
11
11
(a)
Figure 5.26 (a) Pressure anomalies for the

0 (b) “Greenland Above” mode (warmer than average
in West Greenland, colder than average in
Norway). (b) Corresponding anomalies for the
11 “Greenland Below” mode. (From van Loon and
Figure 5.27 Departures from “normal” for (a) strong and (b) weak NAO winters. Shown are sea-
level pressures (isolines, dashed: negative departures), wind vectors (arrows), sea
surface and air temperatures (over 0.2°C dark gray, under 0.2°C light gray), C
cool, W warm, w wet, d dry for relative humidity anomalies. (c) Sea-level pressure
(mb) and wind velocity for “normal” NAO winters, 1950–89. (From Kapala et al.,
1998)
Aleutian lows to be less pronounced, with the largest variations between GA and GB
centered on longitudes of the Icelandic low (Carleton, 1988b). In that region there is a
marked change in the heat transport by the quasi-stationary waves and by the transients
(traveling lows and highs). Transport by the quasi-stationary waves (transients) is consid-
erably smaller (greater) in GA compared with GB modes of NAO. The circulation and
temperature anomalies also have important effects on sea ice conditions. Ice extent in the
Baltic Sea (Koslowski and Loewe, 1994) is inversely related to that in Davis Strait and
to iceberg frequency off Newfoundland in the seesaw years (Rogers and van Loon, 1979).
The seesaw modes have occurred in half of all winter seasons since 1840, based on
observations at Oslo and Jakobshavn, when the temperature anomalies differed by at least
4°C. The “Greenland Below” mode was most frequent before 1924 and during the early
1970s and 1980s. The “Greenland Above” mode was most frequent in the 1870s, the
1940s, and during 1976–86. In addition, van Loon and Rogers note that other months
occur when both Greenland and northern Europe have above/below average temperatures.
The “both above” mode was most frequent between 1926 and 1939 and contributed to
the 1920s warming over the northern North Atlantic (Rogers, 1985). These “both
above/below” patterns each have frequencies of about 10 percent during the winter months,
leaving 30 percent of months falling into none of these four categories. The GA and GB
11 modes of the NAO also favor different storm track locations (Rogers, 1999). With the
GB mode they are concentrated west of Portugal, whereas with the GA mode they are
more often, but not uniquely, concentrated in the northeast North Atlantic.
An extreme version of the GA mode of the NAO occurs when a high is centered over
the Iceland area and low pressure is found near the Azores. These reversals in the normal
gradient occurred in 6 percent of months during 1873–1980, with two-thirds of the eighty-
two cases in the cold season (Moses et al., 1987). They are most common in February–
March. The highs are commonly between 70° and 80°N, 20° and 40°W, and the lows
about 45°–50°N, 20°–40°W. For twenty-two cases where the pressure departures exceeded
0 one standard deviation at Stykkisholmur, Iceland, and Ponta Delgada, the pressure depar-
tures ranged from 8.6 to 23.5 mb at the former and 6.7 to 18.1 mb at the latter station.
In January 1963, for example, the maximum departures were 24 mb and 10 mb (see
Figure 4.38). This pattern closely resembles a composite map of pressure anomalies for
11 the coldest Januaries in northwest Europe. The patterns are associated with minimum
values of the zonal index over the North Atlantic (Makrogiannis et al., 1982). These
reversals were most frequent in the late nineteenth century and in the 1960s (Moses et
al., 1967); none occurred in the 1900s, 1930s, or 1970s, paralleling the frequency of GA
winters identified by van Loon and Rogers (1978). Studies by Hurrell (1995) and Jones
et al. (1997) show that NAO has had strongly positive values since 1980, especially in
0 winters 1993, 1989, and 1995, when the index had the highest values on record. Northern
Europe experienced wetter and warmer than normal conditions, whereas the opposite
occurred in southern Europe and the Mediterranean (Hurrell and van Loon, 1997). Winters
1995/96 and 1996/97 interrupted the run of mild winters in northeast Europe, however.
5.7 North Pacific Oscillation

The North Pacific Oscillation (NPO) involves a subtropical–subpolar pressure oscillation
analogous to the NAO. The related winter temperature seesaw between western Canada
Figure 5.28 Composite pressure difference map for “Aleutians below” minus “Aleutians above”
11 Januaries. (From Rogers, 1981)
(Edmonton) and western Alaska (Dutch Harbor, St Paul), featuring an Aleutians Above-
normal (AA) temperature mode and an Aleutians Below-normal (AB) mode has been
analyzed by Rogers (1981a). The AB mode is characterized by an eastward elongated
Aleutian low with strong westerlies in the central North Pacific and strong polar easter-
lies over Alaska and the Beaufort–Chukchi–Bering seas, as well as a deep Icelandic low.
The AA mode features a weaker Aleutian low displaced 25° to the west off Kamchatka,
and a high over northwest Canada –Alaska; the Icelandic low is now weaker. Thus warmer
winters over western Canada are associated with low pressure in the Gulf of Alaska. This
AB pattern also has more extensive ice in the Bering Sea. The pressure difference map
for AB Januaries minus AA Januaries (Figure 5.28) shows main centers of 12 mb over
the Gulf of Alaska and 8 mb over northwestern Europe. Positive differences (also statis-
tically significant) are located over the northwestern United States and Kamchatka.
In contrast to the NAO, the NPO is essentially a regional-scale oscillation and matches
Kutzbach’s (1970) second eigen vector pattern for winter MSL pressure anomalies. The
AA and AB modes of the NPO are also identified by the rotated PCA of five-day mean
sea-level pressure data for twenty-nine winters (Hsu and Wallace, 1985). However, the
corresponding 500 mb fields are considerably different. The AA mode features a strong
blocking ridge over the Aleutians, whereas the AB mode has a strongly zonal circulation
over the North Pacific. There are no apparent trends in frequency of the two NPO modes.
There were thirteen AB winters and eleven AA winters during 1906–76; the NPO occurs
in 43 percent of winter months.
From an RPCA of gridded northern hemisphere pressure data from 160°E eastward to
40°E for 1899–1986, Rogers (1990) identifies three major patterns of variability in the
North Pacific. They are: the NPO, a north-central Pacific pattern and a Bering Sea pattern.
The reproducibility of the NPO is quite low between 1899–1945 and 1946–86 in January
and February, although it is strong in December and March. The other two patterns are
more consistent except for the Bering center in January. It is possible that some of the
patterns are not unique modes of variability or that they are subject to spatial and temporal
variations in the distribution of anomalies.
It is suggested by Gershunov and Barnett (1998) that the NPO modulates ENSO signals
over North America. Their results are based on a two-way classification of high and low
NPO with El Niño and La Niña events. During high (low) NPO phases, El Niño (La
Niña) signals are strong and stable, whereas they are weak, spatially incoherent and
unstable during El Niño–low NPO and La Niña–high NPO combinations. They propose
that a deeper Aleutian low displaces Pacific storms southward while the El Niño-enhanced
moisture contributes to the storms from the eastern tropical Pacific. In contrast, with a
weaker Aleutian low during La Niña storms are steered northward, increasing precipita-
tion in the Pacific Northwest, and giving fewer storms and drier conditions in the US
southwest.
5.8 Zonally symmetric oscillations

The north–south MSL pressure oscillations in the North Atlantic and North Pacific appear
to be coupled to one another such that the intensity of the Icelandic and Aleutian lows in
winter is inversely related. This zonally symmetric pattern was noted by Lorenz (1951),
Kutzbach (1970), van Loon and Rogers (1978), Trenberth and Paolino (1981), and Wallace
and Gutzler (1981), using different approaches and data sets. Rogers (1990) notes that the
North Atlantic and North Pacific patterns each have a mode with a zonal cyclone track in
mid-latitudes. Thompson and Wallace (1998) show that the leading EOF of wintertime
geopotential heights in the northern hemisphere comprises an Arctic Oscillation (AO)
between the Arctic Ocean and a zonal ring in middle latitudes, particularly over the oceans;
land–sea contrasts weaken the zonal symmetry. Figure 5.29a illustrates the AO signature
at 1,000 mb, based on the leading principal component of November-April monthly mean
11 height anomalies for 1947–97. Here it accounts for 22 percent of the variance at sea level.
The other panels show regression maps on this pattern for tropopause pressure, 50 and 500
mb heights, 1,000–500 mb thickness and surface air temperature anomalies. The pattern is
robust throughout almost a century of sea-level pressure data and dominates both intra-
seasonal and interannual variability, provided that interdecadal trends are removed from the
data. The AO is also identified, especially in winter–spring, in phase alternations of
500 mb anomalies over the northern polar cap versus the mid-latitudes of Europe–North
America and the North Pacific–east Asia. Zhang et al. (1997) find oscillations of sixty to
seventy and thirty to forty days’ duration, using multichannel singular spectrum analysis
0 of five-day mean 500 mb height fields for 1946–88. The anomalies are shown to propagate
poleward along the Ural mountains and over Greenland and southwestard over Europe and
the North Atlantic toward North America. Nevertheless, these patterns account for only 20
percent of the local variance. Importantly, the variance of winter surface air temperatures
11 accounted for by the AO index (39 percent) is over twice that due to the NAO index (17
percent). The correlation of November–April temperatures over Eurasia is 0.55 with the
AO, compared with only 0.23 with the NAO. Thompson and Wallace demonstrate a deep
coupling within the polar vortex through the troposphere and lower stratosphere. Figure
5.29 indicates the presence of both a deep, zonally symmetric, barotropic signature and a
more wavelike baroclinic signature in the troposphere, as shown by the thickness and
0 500 mb patterns. It is noteworthy that LeDrew et al. (1991; LeDrew and Barber, 1994) pro-
posed a coupling between the stratospheric polar vortex and late summer cyclones over the
Beaufort Sea (see Figure 6.6), but he did not explore this conceptual model. Baldwin et al.
(1994) and Perlewitz and Graf (1995) also demonstrate links between the stratospheric polar
night jet, the tropospheric circulation and surface temperature.
In the southern hemisphere there are generally out-of-phase relationships at 500 mb
between high and low latitudes, and also for mid-latitudes versus low and high latitudes
(Mo and White, 1985). The anomalies are essentially barotropic (Rogers and van Loon,
1982). The sea-level pressure field in middle and high latitudes of the southern hemi-
sphere also shows a strong seesaw tendency, which was referred to as the Antarctic
0 Oscillation by Gong and Wang (1998). Subsequently they define an Antarctic Oscillation
Index (AAOI) as the difference in normalized zonal-mean sea level pressure between lati-
tudes 40°S and 60°S (Gong and Wang, 1999). This zonally symmetric oscillation is
characterized by the first EOF of monthly sea-level pressure. It is a year-round feature,
with EOF 1 accounting for between 17 percent (March) and 33 percent (December) of
the variance. Thompson and Wallace (2000) make a comprehensive study of the southern
hemisphere counterpart of the Arctic Oscillation using monthly circulation data poleward
of 20°S for 1957–97. They show that the structure of monthly mean fields in the southern
hemisphere is an annular mode represented by the leading principal component (Figure
5.30). The percentage of variance in all calendar months explained by PC1 is 27 percent
0 for the zonally varying 850 mb heights, 45 percent for the 1,000–50 mb zonal-mean zonal
wind and 47 percent for the zonal-mean 1,000–50 mb geopotential height. The corres-
ponding values for the northern hemisphere are: 20 percent, 35 percent and 45 percent.
The meridional scale of the annular modes in the southern hemisphere is almost identical
throughout the troposphere and lower stratosphere. As Thompson and Wallace point out,
the similarity and robustness of the annular modes in both hemispheres are notable,
given the contrasts in land–sea distribution and in their planetary wave structure (see
section 4.3). Coupling of the respective tropospheric modes with the stratospheric circu-
lation is evident only in late austral spring (boreal winter) in the southern (northern)
hemisphere, when the patterns amplify upward into the stratosphere. This occurs during
0 times of the year when strong zonal flow in the stratospheric polar vortex favours inter-
actions between the planetary and wave-mean flow. In the southern hemisphere this
represents the time of breakdown of the vortex, whereas in the northern hemisphere it
11 occurs when the vortex is strongest.
Figure 5.29 Regression maps for monthly anomalies of tropopause pressure, 50 mb geopotential
height (z50 ), 1,000–500 mb thickness (z500–z1000 ), 500 mb geopotential height (z500 )
surface air temperature, and with the first EOF of sea-level pressure (z1000 ) for
November–April 1947–97 (the AO index). Contour intervals (negative values dashed)
expressed in units per standard deviation of the AO index are: 12.5 m for z1000, 20 m
for z500 and 10 m for z500–z1000; as labeled for z50 surface air temperature; and 5 mb
for tropopause pressure. Extreme values are labeled in the corresponding units. (From
Thompson and Wallace, 1998)
11
11
(a) (b)
0
(c) (d)
Figure 5.30 Southern hemisphere (left, a and c) and northern hemisphere (right, b and d) struc-
tures of the hemispheric circulations associated with their respective annular modes
(the AAO and the AO). Zonal-mean geostrophic wind (m s1) (a and b) and lower-
tropospheric geopotential height (in meters per standard deviation of the respective
0 time series) (c and d) are regressed on the standardized indices of the AAO and AO.
The contour interval is 10 gpm for height and 0.5 m s1 for zonal wind. (From
Thompson and Wallace, 2000)
The AAOI was higher in the 1980s than in the 1960s to early 1970s, indicating a
strengthening of the zonal circulation (Gong and Wang, 1999). This is confirmed by
Thompson et al. (1999), who show a linear trend during 1968–97 in the leading PC of
the 850 mb height field in all months of the year except June.
A different zonal structure, with a mode near the equator and anti-phase relationships
between the hemispheres, has been identified from 250 mb global stream functions
0 analysed for winters 1978/79–1988/89 (Hsu and Lin, 1992). Figure 5.31 shows seven
waveguides associated with eddy activity. After removing this zonal structure, inter-
hemispheric teleconnections were found in the low-frequency (over thirty days) domain,
11 with dipoles straddling the equator near 90°W and 20°W and others between the tropics
Figure 5.31 Seven wave guides at 250 mb in winter inferred from lag correlation maps for base
points located in areas of maximum teleconnectivity. These relate to ten to thirty-day
eddy activity. (From Hsu and Lin, 1992)
and mid-latitudes across the exit zones of the mid-latitude jetstreams in the Pacific (Figure
5.31). These exit zones represent regions where perturbations are likely to grow.
5.9 The southern hemisphere

EOF analysis of SLP and tropospheric height has helped to determine the dominant modes
of low-frequency variability of the atmospheric circulation in the southern hemisphere.
These teleconnections also reveal associations with the tropical ENSO, to varying degrees,
depending on the eigen vectors studied and seasons considered.
The zonally varying pattern Several authors (Rogers and Van Loon, 1982; Mo and White,
1985; Kidson, 1988b, 1991; Shiotani, 1990) have described a dominant teleconnection
pattern characterized by the anomalies of SLP, 500 mb height and zonal winds which are
out of phase between low and middle and middle and high latitudes. These zonally asym-
metric anomalies comprise a barotropic “seesaw” pattern centered on 60oS and upon which
zonal wave No. 3 anomalies are superimposed (Mo and White, 1985; Shiotani, 1990).
The extreme modes of the seesaw, representing out-of-phase variations of geopotential
height between middle and high latitudes in one case, and between the tropics and middle
latitudes in the other (Karoly, 1990), are also characterized by concomitant variations in
temperature gradients, cyclonic activity, and the eddy transports (Rogers, 1983; Shiotani,
1990). These patterns of zonally varying anomalies also exhibit some association with
the ENSO, and are such that the trade winds are stronger in the cold phase, or La Niña
(weaker: El Niño), and the westerlies north (south) of about 45oS are correspondingly
stronger (weaker) (Trenberth, 1981; Rogers and van Loon, 1982).
The Trans-polar Index (TPI) and wave No. 1 Eigen vector analysis of southern hemi-
sphere SLP and 500 mb height data reveals a prominent teleconnection pattern separate
from that associated with ENSO, involving an out-of-phase relationship of the pres-
sure/height anomalies between the Australasian and South America sectors (Rogers and
van Loon, 1982). The variability in the eccentricity of planetary wave No. 1 that is implied
can be depicted by a Trans-polar Index (TPI), or pressure anomaly difference for Hobart,
11 Australia, 147.3°E, 43°S, minus Stanley, Falkland Islands, 58°W, 51.6°S (Pittock, 1980a,
b). A negative TPI implies a shift of the polar vortex towards Australia, which is more
likely during June–September. Rogers and van Loon (1982) confirmed the presence of
the TPI as EOF 2, secondary to the zonally asymmetric pattern and Connolley (1997)
finds an analogous result in a coupled GCM. The Trans-polar Index has been correlated
significantly with annual rainfall and temperature on the southern continents, and with
variations in the extent of ice in the Scotia Sea, Antarctica (Pittock, 1980a, b, 1984;
Rogers and van Loon, 1982). Positive (negative) values of the index are correlated with
the variations in sea ice conditions in the Scotia Sea: mild (severe) ice years being asso-
0 ciated with the trough in the Australia (South America) sector. Other workers (Streten,
1983; Carleton, 1989) have shown statistical associations between TPI and sea ice extent
and concentration in the Antarctic south of Australia, and in the Ross Sea. These are
linked physically by the effects of changes in zonal wind speeds associated with the TPI
11 on the sea ice cover. Confirmation of these results, and development of a plausible phys-
ical model to explain them, awaits the application of a longer-term and more extensive
data set.
The TPI and southern westerlies are positively correlated with the SOI when TPI leads
by up to one year (Pittock, 1984), although there is some dependence in the strength of
this association on the choice of time period studied (Carleton, 1989; Villalba et al., 1997).
0 This may imply a long-term variation of the “poles” of the oscillation. The TPI associa-
tion with ENSO probably also relates to the appearance of significant circulation anomalies
in the storm track area near New Zealand in the period leading up to a “warm event”
(Trenberth and Shea, 1987).
The Pacific–South America pattern Mo and Ghil (1987) applied EOF analysis to the
Australian daily hemispheric 500 mb height analyses for the period June 1972 to July
1983. The first EOF confirms a previously established tendency for southern hemisphere
heights to be out of phase between lower and higher latitudes (Rogers and van Loon,
1982; Mo and White, 1985); the second EOF resembles a quasi-stationary three-wave
0 pattern that is related to blocking events (Trenberth and Mo, 1985; Sinclair et al., 1997).
In particular, Mo and Ghil (1987) also found a strong wave No. 3 component in the third
EOF of the winter data that they called the Pacific–South America (PSA) pattern, because
of its resemblance to the PNA of the northern hemisphere. However, the two telecon-
nections (PSA, PNA) do not necessarily occur at the same time (Mo and Ghil, 1987).
The PSA comprises a wave train of alternating anomalies that extends southeastwards
from the subtropical Pacific to the Antarctic Peninsula. A subsequent EOF analysis by
Farrara et al. (1989) retained all wave numbers in the anomalies, including those of wave
No. 5 and greater that were discarded by Mo and Ghil (1987). This resulted in the PSA
comprising EOF 2, rather than EOF 3, of the winter 500 mb height anomalies, with EOF
0 1 remaining as the zonally symmetric anomalies identified in Mo and Ghil (1987). Kidson
(1988a) and Berberry et al. (1992) have also identified a PSA-like pattern in southern
hemisphere anomaly fields that have been filtered to retain time scales of between ten
and fifty days.
Southern extratropical teleconnections with the El Niño–Southern Oscillation The trop-

ical ENSO phenomenon links directly with the southern extratropics in the Pacific Ocean
sector (Zillman and Johnson, 1985; van Loon and Shea, 1985). In the lead-up (approxi-
mately twelve to eighteen-month lead) to a warm ENSO event (El Niño) the seasonal
cycle of the trough in the Tasman Sea is strongly enhanced relative to its intensity in a
0 non-ENSO year (van Loon, 1984). As a result, extratropical cyclone activity in the New
Zealand region is typically increased (Trenberth and Shea, 1987). This is associated with
a greater frequency of blocking near New Zealand and enhanced troughing downstream
11 in longitudes of the Ross Sea during El Niño. Accordingly, there are more outbreaks of
cold air towards lower latitudes of the southwest Pacific. There are also large changes in
cyclone frequency elsewhere in the southern extratropics associated with ENSO (Sinclair
et al., 1997). These include increases (decreases) in the number of wintertime cyclones
during an El Niño over the Indian Ocean and the Amundsen Sea (near Wilkes Land, and
the subtropical eastern Pacific). Interestingly, the patterns of greatest cyclonic activity
during the La Niña phase of ENSO are virtually the opposite to those in the El Niño,
suggesting a more or less linear response of the southern hemisphere cyclone eddies
to ENSO.
The influence of the Southern Oscillation, particularly the El Niño phase, is evident in
the pressure and temperature anomaly fields, and also in the sea ice conditions of the
Antarctic and sub-antarctic (Carleton, 1988a, 1989; Smith and Stearns, 1993; Gloersen,
1995; Simmonds and Jacka, 1995). In particular, synoptic studies of atmospheric circu-
lation variations in the southeast South Pacific and Antarctic Peninsula region link the
PSA pattern with the ENSO (Marshall and King, 1998; Carleton et al., 1998). The center
of action comprising the Amundsen Sea low exhibits a strong out-of-phase relationship
with ENSO, such that the “mean” low is stronger (weaker) during La Niña (El Niño)
events, resulting in positive (negative) temperature anomalies in winter on the Antarctic
Peninsula (Marshall and King, 1998). Hemispheric-scale maps of the 500 mb height anom-
alies composited for sets of “warm” and “cold” winters on the Antarctic Peninsula bear
a strong resemblance to the extreme phases of the PSA pattern. Similarly, for two winters
characterized by strong anomalies in the intensity of the Amundsen Sea low (1988: weak;
1989: strong), Carleton et al. (1999) find that the frequencies of occurrence of cold-air
mesocyclones in the Bellingshausen and Amundsen seas were greatly increased in 1989
relative to 1988. The mesocyclone strong increases in 1989 accompanied an area of wide-
spread cooling of the sea surface and greater sea ice extent in the May through September
period that were, apparently, tied to the greater intensity of the “mean” Amundsen Sea
low and the more frequent cold-air outbreaks associated with this feature during that
winter. These contrasted with weaker seasonal changes in the SST and also with less
extensive sea ice in the Bellingshausen/Amundsen seas, for the 1988 winter.
The variations in longitude position of the Amundsen Sea low that are related to its
intensity variations (i.e. a displacement westward towards the Ross Sea when weaker;
displaced eastward toward the Antarctic Peninsula when stronger, e.g. Carleton and Fitch,
1993), dominate the interannual variations of moisture fluxes into West Antarctica
(Bromwich et al., 1995). Values of the moisture convergence into this region increase
(decrease) when the low is displaced to the west (east) of its long-term mean position.
The mechanism by which the intensity variations of the Amundsen Sea low are linked
with ENSO likely involves the eddy momentum flux convergence patterns associated with
the subtropical jet (STJ) and polar front jet (PTJ). The two jetstreams vary out of phase
with each other according to the phase of ENSO, whereby the STJ is weaker (stronger)
and the PFJ stronger (weaker) in cold La Niña (warm: El Niño) events (Chen et al., 1996;
Cullather et al., 1996).
5.10 Tropical–extratropical teleconnections

As noted earlier (section 5.1), in addition to the dipole teleconnections in the North Atlantic
and North Pacific sectors, there are independent patterns each with three centers – in
Eurasia, the northern tropics, and the Pacific/North Atlantic.
Tropical–extratropical teleconnections in 500 and 200 mb height fields during northern
winter, identified by simultaneous and lagged correlation statistics for 1963–81, are the
Pacific–North American (PNA), a Tropical/Northern Hemisphere (TNH) pattern, and a
mixed one of these two labeled West Pacific Ocean (WPO) (Mo and Livezey, 1986). The
three cumulatively account for about one-third of the variance of the mean northern hemi-
sphere winter, and a larger fraction over the North American sector. The PNA pattern
11 features in both low-pass (seasonal) and high-pass (monthly) filtered data, but the seasonal
connection with the tropics is limited to ENSO years. Further analysis of tropical heating,
inferred from SST and OLR data, indicates that positive projections are likely for all three
circulation modes (at 700 mb) during strong ENSO winters (Livezey and Mo, 1987). The
absolute strength of the TNH is shown to be directly related to SST anomalies in the
central Pacific, for example.
The PNA pattern refers to the relative amplitudes of the ridge over western North
America and the troughs over the central North Pacific and eastern North America
(Leathers et al., 1991). A strong (weak) ridge–trough pattern is designated as a positive
0 (negative) PNA regime (Figure 5.32). An index of PNA strength proposed by Wallace
and Gutzler (1981) uses a linear combination of standardized 700 mb height anomalies
at grid points nearest the four mean centers of the anomaly field; Leathers et al. use a
different variant for three centers. During the positive mode of the PNA (Figure 5.33a)
11 with a strong Aleutian low and a strong ridge over western Canada, there is a well devel-
oped storm track from Asia extending eastward into the central Pacific between 40°N and
50°N and then northeastward to the Gulf of Alaska. For the negative phase (Figure 5.33b),
cyclones track northeastward along the East Asian coast to the Bering Sea, with a second
area near the west coast of Canada (Lau, 1998; Klasa et al., 1992; Ueno, 1993). The
PNA is generally in the positive mode during a warm ENSO event (see Horel and Wallace,
0 1981) and in the negative mode during the cold phase of ENSO. In autumn, winter, and
spring the PNA is a principal mode of circulation variability in the middle troposphere.
It has strong effects on winter temperatures in the United States, with warm anomalies
in the northwest and cold outbreaks in the southeast under positive PNA patterns (Leathers
et al., 1991).
The frequency of the two modes of the PNA pattern is not symmetrical. Using EOFs
of winter 500 mb height data to identify the modes, Dole (1986) found more occurrences
of low or moderate amplitude positive PNA than negative PNA, but for larger amplitude
anomalies the opposite is true. Predictability experiments suggest that the atmosphere is
more barotropically stable during episodes of positive PNA (Palmer, 1988). This is
0 reflected in the tendency for a composite of negative PNA mode pentad-mean fields of
500 mb height over the northern hemisphere during winters 1952–84 to have lower vari-
ability than a similar composite of positive mode events.
Figure 5.32 Schematic plot of dominant low-frequency (thirty-day) teleconnectivities at 250 mb in

11 winter. (After Hsu and Lin, 1992)
Figure 5.33 The composite sea-level pressure fields for (a) positive and (b) negative modes of the
Pacific/North American pattern. (From Wallace and Gutzler, 1981)
11 This teleconnection depends on the location and strength of the jetstream over East
Asia which extracts energy from the mean flow (Nakamura et al., 1987) (see section 5.4).
In winter the PNA is significantly associated with the East Asian jet, Asian land temper-
atures, and tropical SSTs in the central Pacific (Leathers and Palecki, 1992). In contrast,
only mid-latitude variables are involved in spring: SSTs in the western North Pacific and
Asian land temperatures. In summer and autumn there is little PNA variance associated
with any of these variables.
The role of synoptic eddies in forcing the tropospheric PNA pattern has been exam-
ined by Klasa et al. (1992) for twenty-five winters. By analysis of the conversion of eddy
0 kinetic energy from barotropic eddies into the mean flow (through Eliassen–Palm flux
vectors), and also the eddy vorticity forcing for positive and negative PNA modes, they
show that the strongest anomaly of eddy forcing is collocated with large amplitude PNA
centers in the Pacific. The eddy forcing has a six to ten-day time scale. During well devel-
11 oped PNA patterns, the conversion of synoptic-scale eddy kinetic energy into the mean
flow is most important within jet exit zones. For PNA positive, the maximum eddy-mean
flow conversion is over the eastern Pacific, while for PNA negative the maximum is
displaced westward into the central Pacific.
The primary source of kinetic energy for the major teleconnection patterns is the conver-
sion of the KE of the basic flow to the KE of the response, according to Li and Ji (1997),
0 rather than that supplied directly by the external forcing. Using the barotropic vorticity
equation, linearized about the 300 mb level zonally varying climatological flow for the
northern hemisphere winter, they analyze the location of “efficient” forcing modes – which
lead to the growth of anomalies within about five days – and the preferred response
patterns. The sources of efficient forcing are localized in the subtropics, south of the major
jetstream maxima, and over the Arctic. The subtropical loci give responses resembling
various observed teleconnection patterns: forcing over Central America and the western
Atlantic gives a North Atlantic pattern (Hsu and Lin, 1992); forcing over North Africa
sets up a south Eurasia pattern; forcing over South Asia and the western Pacific gives an
East Asia–Pacific pattern (Nitta, 1987), while central Pacific forcing leads to the
0 Pacific–North America pattern. In addition to these, the Arctic forcing gives propagating
responses from northern Canada to the western Atlantic and from northern Asia to the
western Pacific. Hoskins and Ambrizzi (1993) documented a zonal wave train from forcing
in North Africa to response centers over southern Asia. However, Li and Ji (1997) note
that the PNA pattern does not correspond to a wave guide and they also point out that
the possible path of Rossby wave rays (Hoskins and Karoly, 1981) may vary according
to the wave number involved. Teleconnection patterns do not appear to be determined
solely by the dispersion of wave energy.
It must be emphasized that the tropical SST forcing of interannual climate variability
in the extratropics accounts for less than the total variability of wintertime mean surface
0 air temperature. Horel and Wallace (1981) show that the correlation between the SOI and
extratropical height fields in the northern hemisphere is stronger at 700 mb and 300 mb
than at the surface and is stronger for North America in winter. Studies show that extra-
tropical seasonal climate anomalies are a result of the combined effects of ENSO states
and zonal index anomalies on teleconnection patterns and that anomalies of u are essen-
tially independent of tropical SST variations (Hoerling et al., 1995; Ting et al., 1996).
ENSO explains an important part of the seasonal variance over the North Pacific and
central Canada. In contrast, anomalies of the stationary waves associated with u anomalies
produce “centers of action” over the Pacific–North America and North Atlantic–Eurasia
regions, where they account locally for 30–40 percent of the interannual variability (IAV)
0 of 500 mb heights. Moreover, Ting et al. (1996) find that the IAV over many parts of
the northern hemisphere extratropics is largely independent of both the variations of ENSO
and the zonal index.
11
5.11 Teleconnections and synoptic-scale activity
The principal teleconnection patterns identified by Wallace and Gutzler (1981) and others
are associated with characteristic spatial distributions and levels of activity of the 500 mb
synoptic-scale storm tracks as indicated by RMS statistics of two and a half to six-day
band-pass filtered heights (Lau, 1988). For example, the dipole modes of the western
Pacific and western Atlantic teleconnection patterns in the northern hemisphere are accom-
panied by changes in wintertime storm track intensity and mean zonal wind over the
western oceans. There are similar variations in storm track intensity over Siberia associ-
ated with the northern Asian teleconnection pattern noted by Esbensen (1984). Eddy
activity is strengthened over western Siberia (50°–60°N) when 500 mb heights are above
normal over Mongolia. In contrast, the modes of the Eastern Atlantic and Pacific/North
American patterns, which display more of a wave-like character, are associated with lati-
tudinal displacements of the storm tracks and jetstreams over the eastern oceans, rather
than with intensity changes, Lau (1988) also finds that enhanced (weakened) 500 mb eddy
activity over the North Pacific and North Atlantic is associated with strengthened surface
westerly (easterly) winds to the south of negative (positive) sea-level pressure anomalies,
respectively. Patterns of composite sea-level pressure differences for several of the storm
track modes resemble North Pacific and North Atlantic oscillation patterns.
Northern hemisphere circulation regimes, as characterized by the pattern classification
of Dzerdzeevski (see section 7.3.3), are found to differ considerably during extreme (warm
and cold) modes of the ENSO sequence (Fraedrich et al., 1992). There is enhanced
(reduced) zonality in winters following the warm (cold) event. A warm event forces the
North Pacific storm track into a more zonal arrangement, especially in the eastern part.
Overall, warm event winters have a reduction in the eight to ten-day “residence time” of
meridional circulation types.
An extension of this analysis for nine warm and nine cold events shows that the latter
have more variance in the 500 mb heights along 50°N owing to transient eddies relative
to stationary eddies (Fraedrich and Müller, 1994). Thus, for height variance in periods of
fifteen days or more, cold event winters have most variance in zonal wave Nos. 3 and 4
whereas warm event winters show no peak in quasi-stationary wave No. 3.
5.12 Time-scale aspects of teleconnections

The analysis of fields of simultaneous pressure values provides no information about their
temporal evolution, although lag correlations and band-pass filtering are useful in this
regard. The time scale dependence of 500 mb circulation patterns examined by lag corre-
lations suggests that ten to thirty-day features represent wave trains. They display zonally
oriented curved ray paths that migrate with the reference point, whereas patterns that last
longer than a month are mainly fixed north–south dipoles situated over the oceans
(Blackmon et al., 1984a). The zonally oriented wave-like patterns do not actually prop-
agate zonally; rather, downstream centers develop and intensify as the ones upstream
decay, providing an eastward dispersion of energy (Blackmon et al., 1984b).
The spatial structure of low-frequency variability differs considerably according to the
time scale. An analysis of monthly 700 mb height anomalies for winters 1949/50 through
1976/77, where the intermonthly and interannual signals are obtained by filtering, finds
important differences in the teleconnection patterns on these two time scales (Esbensen,
1984). The intermonthly fields resemble the map of Wallace and Gutzler (1981) with
prominent western Pacific, western Atlantic and Pacific–North American (PNA) centers
and a further dipole over northern Asia.4 The interannual field, in contrast, displays patterns
that are considerably more global in extent; they are the PNA, the Eurasian, the Zonally
Symmetric Seesaw, and the North Pacific dipole (Figure 5.34). These differences in pattern
of teleconnectivity suggest that the associated dynamical and physical mechanism may
11
11
(a)
0
(b)
0 Figure 5.34 Teleconnection fields in winter for (a) the intermonthly signal and (b) the interannual sig-
nal. Hatched regions indicate correlations greater than 0.6 and dark shaded regions greater
than 0.7. Solid arrows denote patterns supported by teleconnectivity and one-point corre-
11 lation maps, open arrows are based primarily on the latter. (From Esbensen, 1984)
also differ. Analysis of connections between tropical and mid-latitude circulations at 250
mb confirms that the mechanisms do indeed differ according to time scale (Mo and
Kousky, 1993). On the intraseasonal scale there is a substantial zonally symmetric connec-
tion in both summer and winter, comprising a forty-eight-day oscillation. Anomalies of
OLR show this to be related to convection over the tropical Pacific. In contrast, Mo and
Kousky (1993) identify a PNA wave-train mode in the boreal winter extending north-
eastward from the Pacific into western North America, and then southeastward to the
subtropical Atlantic.
The mid-tropospheric winter circulation in the northern hemisphere has been analyzed
with daily 500 mb height data for November–April 1946–84, filtered to distinguish periods
of ten to sixty, sixty to eighty, and over 180 days, by Kushnir and Wallace (1989). The
interannual (over 180 days) variability displays modal structure – the NAO and PNA
patterns – which account for much of the total height variance in these sectors (Figure
5.35c). In the sixty to 180 day intraseasonal band, however, only the western Atlantic
Figure 5.35 Teleconnectivity of wintertime 500 mb height fluctuations for (a) ten to sixty-day
periods, (b) sixty to eighty days, (c) longer than 180 days. Extrema show correlation
values with decimal point omitted; bold contours indicate r = 0.4 in (a) and 0.5 in (b)
and (c). The contour interval is 0.1. (From Kushnir and Wallace, 1989)
11 dipole pattern identified by Wallace and Gutzler (1981) rises above the background
continuum. In the ten to sixty-day interval no geographically fixed patterns are apparent
in the spatial distributions of variance, or teleconnectivity, or the coefficient of anisotropy
(measuring the shape and orientation of transients). There are zonally oriented wave trains,
particularly over the continents, and weak north–south dipoles over the oceans, as illus-
trated in Figure 5.35a for teleconnectivity. The dipole patterns occur downstream of the
climatological jetstream maxima.
A general issue that needs to be recognized is the tendency of mid-latitude SST anom-
alies to recur in successive winters without being present in the intervening summers
0 (Namias and Born, 1970). A re-emergence of anomalies from beneath the summer mixed
layer has been shown to take place via entrainment in the following autumn–winter as
the mixed layer deepens (Alexander and Deser, 1995). Confirmation that SST anomalies
of given sign in the central Pacific, and of opposite sign off the west coast of North
11 America, are sequestered in the summer thermocline and re-emerge in autumn is provided
by Alexander et al. (1999). The shallower mixed depth in the eastern Pacific allows an
earlier return than in the western Pacific. Re-examination of this process (Zhang et al.,
1998) shows that the primary reason for the SST signal is the fact that the dominant SST
anomaly mode has a similar spatial structure all year, with maximum amplitude in summer.
The pattern is more persistent from one summer to the next than from one winter to the
0 next. However, the persistence from summer to winter in the anomaly pattern is greater
than that of the local SST anomalies at their primary centers. It is unclear why the pattern
does persist. West of 140°W at 40°N there is the same polarity of anomaly all year, but
a winter-to-summer reversal to the east of this longitude. Zhang et al. suggest that nega-
tive SST anomalies east of 140°W in winter favor a positive PNA and this circulation
reinforces negative SSTs to the west. Southerly flow in the eastern Pacific tends to raise
SSTs there.
Meehl et al. (1998) using a global coupled ocean–atmosphere–sea ice model, simulate
variability in the nine to twenty-year range and show that the thirteen to fifteen-year time
scale in the model is set approximately by the average circuit of the ocean gyre circula-
0 tions. Mehta (1998) finds twelve to thirteen year periods in SSTs for 1882–1991 in the
tropical South Atlantic but not in its northern counterpart. There are three modes of vari-
ability in tropical Atlantic SSTs. In the South Atlantic decadal mode, anomalies form in
situ and travel from the subtropical South Atlantic along the eastern boundary into the
tropics. They reside there several years and they tend to travel southward along the western
boundary into the subtropical South Atlantic. There is also an energetic North Atlantic
mode with anomalies from the extratropics making a clockwise rotation around the ocean
boundary. Analysis of SST and sea-level pressures (SLPs) by White and Cayan (1998)
distinguishes periods in the three to seven, nine to thirteen, and eighteen to twenty-three-
year range. During 1900–89 the largest SST and SLP anomalies occur in the extratropics
0 and near the eastern boundaries of the oceans. Peaks of a tropical warm phase are noted
in 1900, 1920, 1940, 1960, and 1980. During 1955–94, they identify global reflection
symmetries about the equator and global translation symmetries between ocean basins.
Sea-surface temperatures in the tropical and eastern ocean areas are above (below) average
in association with stronger (weaker) extratropical westerlies. Positive (negative) SST
departures in the west-central subarctic and subantarctic frontal zones covary with stronger
(weaker) subtropical and subantarctic gyres in the North Pacific and North Atlantic and
with basin and global SSTs that are about 0.1 K above (below) normal.
Changes in the frequency of major hemispheric circulation regimes are recognized to
be implicated in climatic change. Corti et al. (1999) examine the structure of the first two
0 EOFs of monthly-mean 500 mb height fields for November–April 1949–94. The proba-
bility density function of the atmospheric state vector, as described by the loadings of
the monthly-mean heights onto the first and second EOFs plotted in two-dimensional
11 phase space, identifies four maxima. The spatial structures of these clusters resemble
several well known circulation modes. One represents the 500 mb height field associated
with the cold ocean, warm land (COWL) pattern in 1,000–500 mb thickness anomalies
discussed by Wallace et al. (1995, 1996). The second and third clusters are related to the
positive mode of the NAO and the negative mode of the PNA, while the fourth is well
correlated with the negative phase of the Arctic Oscillation. These clusters are still evident
when ENSO years are removed from the data. Corti et al. conclude that changes in the
frequency or duration of the principal circulation regimes in the northern hemisphere over
the second half of the twentieth century have contributed substantially to hemispheric
temperature trends. These circulation changes may, nevertheless, be a response to anthro-
pogenic forcing.
In the southern hemisphere, monthly mean patterns as well as low-pass filters suggest
the existence of a zonally symmetric pattern that features a change in the sign of the
anomalies near 60°S and also of a zonal wave (n 3) in middle to high latitudes (Mo
and White, 1985; Mo and Ghil, 1987). For the intraseasonal (ten to sixty-day) time domain,
extratropical teleconnectivity maps at 200 mb indicate short-wave (3,000 – 4,000 km)
negative correlations with slow-moving zonal wave patterns in winter (Berberry et al.,
1992). A wave train that appears to originate in the South Indian Ocean (45°S) splits into
two in the vicinity of Australia, with the subpolar and subtropical jetstreams serving as
wave guides (Figure 5.36). The two wave trains merge near South America, with equa-
torial propagation towards northeast Brazil and the subtropical South Atlantic. In summer
the wave patterns are more meridional (equatorward) and geographically fixed.
5.13 Interannual to interdecadal oscillations

Several recent observational and coupled GCM studies suggest the occurrence of a range
of multi-year coupled ocean–atmosphere oscillations. Most of these are still incompletely
documented, owing to the short available record length, but it seems worth while at least
to point out their major characteristics and suggested mechanisms involved. The occur-
rence of variability on interannual to interdecadal time scales has been examined by
various spectrum analysis techniques (see section 2.6) for global fields of air temperature
(Mann and Park, 1994), and sea-surface temperature (Moron et al., 1998), as well as for
the SOI (Allen and Smith, 1994) and other variables. The global analysis of sea-surface
temperatures since 1901 by Moron et al. (1998) identifies a number of oscillations: (1)
a thirteen- to fifteen-year seesaw oscillation between the Gulf Stream region (Bermuda
to Cape Hatteras) and the North Atlantic Current sector south of the Denmark Strait; (2)
a quasi-decadal oscillation over the North Atlantic (also Tourre et al., 1999) as well as
over the South Atlantic and Indian Oceans; (3) a seven to eight-year oscillation involving
the subtropical and subpolar gyres in the North Atlantic; and (4) interannual oscillations
of sixty to sixty-five, forty-five, and twenty-four to thirty months, particularly in the trop-
ical Pacific Ocean. In the Pacific Ocean from 20°S to 58°N Zhang et al. (1998) find a
quadriennial (fifty-one-month) oscillation in SSTs accounting for about 20 percent of the
variance. It represents a standing wave in the tropics but there is propagation northeast-
ward from the Philippines Sea and then eastward along 40°N. There are also weaker
interdecadal and QBO signals.
5.13.1 Quasi-biennial oscillations

The existence of a twenty-five and a half to twenty-four-month (quasi-biennial) oscilla-
tion was first noted in North American temperature data (Clayton, 1885) and subsequently
in many climatic records. These include snow cover (Voeikov, 1895), sea-surface temper-
atures off Norway (Helland-Hansen and Nansen, 1920), surface pressure in mid-latitudes
(Shapiro, 1964), blocking (Boehme, 1967), and other surface and tropospheric climate
parameters (Landsberg et al., 1963). The discovery in the 1950s of a twenty-six-month
11
11
0 Figure 5.36 Teleconnection patterns in the southern hemisphere for (a) winter and (b) summer.
The correlation values are ×100; values below 50 percent are shaded. Points of
maximum negative teleconnection are connected. (From Berberry et al., 1992)
11
oscillation in easterly and westerly phases of the equatorial stratospheric winds – the
stratospheric QBO (Veryard and Ebdon, 1961; Naujokat, 1986) – focused much interest
on the question of stratospheric–tropospheric interactions. The stratospheric oscillations
have a maximum amplitude about 25–30 km and propagate downward at ~1 km/month.
Lindzen and Holton (1968) proposed a two-way interaction between the mean flow and
momentum fluxes from waves in the troposphere. Waves on all scales from gravity waves
to planetary waves appear to play a role (Dunkerton, 1997). Momentum is transferred
upward from the equatorial tropopause to a critical layer below 40 km where the phase
of the oscillation in the zonal winds is triggered. The amplitude of the momentum fluxes
appears to play a primary role in determining the oscillation frequency, but this can vary
with changes in the thermal relaxation time and the horizontal diffusivity. The momentum
transfer to the mean flow causes the critical layer to move downward. The response to
momentum eddy flux convergence is a flow acceleration which has a limited latitudinal
extent. Haynes (1998) suggests that the internal dynamics of longitudinally symmetric
motion in a rotating, stratified atmosphere, with thermal relaxation, may be a sufficient
cause of the equatorially confined QBO.
A possible influence of the equatorial stratospheric QBO phase on the Southern
Oscillation and the northern hemisphere winter 700 mb height field was suggested by van
Loon and Labitzke (1987, 1988; Labitzke and van Loon, 1989). The data record is rather
short, but a statistical association exists between the eleven-year cycle in 10.7 cm solar
flux and northern hemisphere tropospheric conditions for the westerly phase of the QBO:
high solar flux, westerly QBO and high SOI for 1951–88. The relationship with northern
hemisphere winter climate failed in January–February 1989, apparently because the Pacific
cold event of 1988–89 superseded solar–QBO forcing (Barnston and Livezey, 1991).
Barnston et al. (1991) find a preference for the TNH and WPO patterns during the east-
erly QBO phase at 45 mb for 1951–89 in response to anomalies in the Southern Oscillation.
During the westerly phase there is a preference for the PNA pattern. Changes in the height
of the tropopause with QBO phase, that affect the potential for convergence in the tropos-
phere, may provide the forcing for the responses in mid-latitudes. However, a physical
mechanism for the QBO to mediate the effects of the solar cycle, or the extratropical
effects of the Southern Oscillation, remains to be established.
Biennial oscillations in the tropical troposphere appear to be linked with variability in
the Southern Oscillation and ocean–atmosphere coupling (Brier, 1978; Barnett, 1991;
Yasunari, 1989; Meehl, 1997). The South Asian monsoon plays an active role in the
tropospheric biennial oscillation (TBO), according to Meehl (1997). He proposes mech-
anisms for both ocean–atmosphere (OA) and land–atmosphere (LA) couplings on a
biennial scale. The OA coupling involves a biennial alternation in the intensity of local
convection during the season of local convective maximum, arising from SST anomalies.
The LA coupling involves a corresponding alternation in monsoon strength associated
with the relative land–sea (meridional) temperature contrast. Linkages between the two
mechanisms, operating over the tropical Indian and Pacific sectors, require that anomalies
of land surface temperatures in southern Asia and SST anomalies in the tropical Indian
Ocean and eastern equatorial Pacific vary roughly in phase over the annual cycle.
Observations indicate that the air–sea coupling tends to be best developed during the
season when the seasonal convective cycle is also strong at some location in the
Indian–Pacific sector, as postulated by Meehl (1993). Moreover the upper ocean heat
content is known to have sufficient memory to maintain SST anomalies for about a year.
Figure 5.37 illustrates schematically the evolution of the TBO over a two-year period.
Beginning in the boreal winter (DJF) of year 0, anomalies of SST and convection, set up
previously, are associated with an anomalous ridge over central Asia, giving warm, dry
conditions. By boreal spring (MAM0), the weak Australian monsoon has left a warm sea
surface north of Australia and a relatively cool surface in the tropical Pacific. As the
ITCZ moves northward, convection is weak over the Pacific and stronger over Indonesia,
11
11
Figure 5.37 The idealized evolution of the tropospheric biennial oscillation over a two-year cycle.
Year 0 (1) refers to years with a strong (weak) Asian monsoon. The cycle begins in
DFJ0, counter-clockwise to SON1. (From Meehl, 1997)
maintaining the pattern from DJF0. In the following boreal summer the anomalously warm
0 land surface of South Asia enhances the land–sea (meridional) temperature contrast,
creating a strong Asian monsoon. The maximum of convection shifts southeastward during
SON0, leaving a relatively cool moist land surface in South Asia. Enhanced evaporation
persists there through DJF1 and there is a strong Australian monsoon, associated with
the warm sea surface. The east–west Walker circulation suppresses convection in the
central tropical Pacific and in East Africa. Over Asia a mid-latitude trough advects cold
air and causes increased snow cover. This pattern is maintained into the next boreal spring
(MAM1). Now SST anomalies become positive in the tropical Pacific, and negative over
Indonesia, with corresponding convective anomalies. The cool South Asian land surface
weakens the land–sea (meridional) temperature gradient and the summer monsoon (JJA1).
0 By autumn (SON1) South Asia is relatively dry and warm, setting the stage for the TBO
to continue.
5.13.2 Antarctic Circumpolar Wave

A set of atmospheric variables (sea-level pressure and meridional wind stress) and ocean
variables (sea surface temperature and sea ice concentration) for the Southern Ocean all
show evidence of anomalies propagating eastward in the circumpolar southern westerlies
at about 6 km/day. White and Peterson (1996) term this feature the Antarctic Circumpolar
Wave (ACW). The overall wave No. 2 pattern has a period of four to five years and
0 propagates around Antarctica in eight to ten years. It is best developed in the South Pacific
sector. The anomalies at 56°S have ranges of up to 8 mb in MSL pressure, 0.03N m2
in meridional wind stress, 1.6°C in sea surface temperature, and 350 km in the sea ice
11 margin. Lagged cross-correlations indicate that positive (negative) anomalies of sea surface
temperature (SST) follow high (low) pressure anomalies by about one year (90° phase)
and are about 180° out of phase with equatorward (poleward) anomalies in meridional
wind stress and the sea ice margin. The SST anomalies seem to originate in the western
subtropical South Pacific and propagate southeastward before moving eastward in the
Southern Ocean. It should be noted that the observational record in this region spans little
more than a decade and that its overall quality is uncertain in the case of atmospheric
analyses. A case study for 1982–94 identifies the major source of the ACW in the western
subtropical South Pacific (Peterson and White, 1998). Here anomalies develop in SST
and moisture content and these move, together with SLP anomalies, into the Southern
Ocean, where they migrate eastward in the Antarctic Circumpolar Current. Parts of the
interannual SST anomalies branch northward into the South Alantic and South Indian
oceans. These return to the tropics some six to eight years after appearing in the low-
latitude Pacific. Remarkably, an ACW has been simulated in an extended integration with
the Max Planck Institute coupled model (Christoph et al., 1998). Christolph et al. suggest
that, while the SST and sea ice anomalies are advected eastward in the Antarctic
Circumpolar Current, the pressure and meridional wind stress anomalies appear to be
standing waves that are amplified or weakened as the SST and associated surface heat
flux anomalies move in and out of phase with the waves.
White and Peterson postulate that the ACW may originate through a teleconnection
between ENSO-related precipitation (and resultant latent heat) anomalies in the central
and western tropical Pacific and the atmosphere over the Southern Ocean. However,
Christoph et al. find that ENSO forcing explains only 30 percent of the variance at most.
They propose that a standing wave pattern associated with the Pacific–South American
(PSA) oscillation (see section 5.9) generates surface heat flux anomalies. These fluxes
Figure 5.38 Schematic relationship between the coupled components in a specific phase of the
ACW. The contours show SST anomalies (negative dashed), H and L denote high and
low-pressure centers, Q and Q represent upward and downward heat fluxes, and the
open arrows marked show meridional wind stress maxima. The bold arrows indi-
cate the eastward progression of SST anomalies; the other components undergo a
standing oscillation. (From Christoph et al., 1998)
11 warm (cool) the eastern (western) margins of the pressure anomalies, as illustrated
schematically in Figure 5.38. Then the SST and wind stress anomalies drive the fluctua-
tions in the sea ice margin. The oceanic components encircle the Southern Ocean in about
twelve to sixteen years, according to the model. The winter sea-level pressure and 500
mb heights in high southern latitudes characteristically feature a zonal wave No. 3
barotropic pattern (Mo and White, 1985), with a geographic distribution that favours the
ACW regime. The wave No. 3 pattern coupled with the circumglobal ocean wave causes
locally reappearing energy peaks at four or five-year intervals.
0
5.13.3 Quasi-decadal oscillations
A number of studies of climate records in the northern hemisphere point to the presence
of decadal and multidecadal fluctuations. Meehl et al. (1998), using a global coupled
11 ocean–atmosphere–sea ice model, simulate variability in the nine to twenty-year range
and show that the thirteen to fifteen-year time scale in the model is set approximately by
the average circuit of the ocean gyre circulations. Mehta (1998) finds twelve to thirteen-
year periods in SSTs for 1882–1991 in the tropical South Atlantic but not in its northern
counterpart. There are three modes of variability in tropical Atlantic SSTs. In the South
Atlantic decadal mode, anomalies form in situ and travel from the subtropical South
0 Atlantic along the eastern boundary into the tropics. They reside there several years and
tend to travel southward along the western boundary into the subtropical South Atlantic.
There is also an energetic North Atlantic mode, with anomalies from the extratropics
making a clockwise rotation around the ocean boundary. Analysis of SST and sea level
pressures (SLPs) by White and Cayan (1998) distinguishes periods in the three to seven,
nine to thirteen, and eighteen to twenty-three-year range. During 1900–89 the largest SST
and SLP anomalies occur in the extratropics and near the eastern boundaries of the oceans.
Peaks of a tropical warm phase are noted in 1900, 1920, 1940, 1960, and 1980. During
1955–94 they identify global reflection symmetries about the equator and global transla-
tion symmetries between ocean basins. Sea surface temperatures in the tropical and eastern
0 ocean areas are above (below) average in association with stronger (weaker) extratrop-
ical westerlies. Positive (negative) SST departures in the west-central subarctic and
subantarctic frontal zones covary, with stronger (weaker) subtropical and subantarctic
gyres in the North Pacific and North Atlantic and with basin and global SSTs that are
about 0.1 K above (below) normal. In the North Atlantic region these seem to be related
to several factors: the ocean thermohaline circulation (THC), the formation of North
Atlantic deep water (NADW), Arctic ice export through the Fram Strait, and associated
fresh-water anomalies in the northern North Atlantic, like the Great Salinity Anomaly
(GSA) of the late 1960s (Mysak et al., 1990; Mysak and Venegas, 1998). Bjerknes (1963,
1964) first developed a model of variations in the surface heat balance of the North
0 Atlantic on short and long time scales. He proposed that, on the two to five-year scale,
strong zonal flow at 35°N is associated with large turbulent heat fluxes and, since oceanic
heat transport responds slowly, there is cooling of the surface layers. Over fifty-year
periods, wind speeds and surface temperature in the Icelandic low area are also nega-
tively correlated. On this time scale, cooling due to upwelling is weak when the Icelandic
low is weak. Warm water transported in the Irminger Current branch of the North Atlantic
Current lags the Icelandic low strength by a few years, so that these opposing influences
are out of phase. The mechanisms responsible for such long-term trends are still under
debate. Kushnir (1994) notes that decadal-scale variations in North Atlantic SSTs for
1900–87 show a distinct dipole pattern centered in the Iceland–Labrador Sea area and
0 east of Bermuda. This pattern, he suggests, provides evidence for the role of the THC in
the northwestern Atlantic–Greenland Sea. A coupled model study of the North Atlantic,
however, indicates that interaction of the extratropical atmosphere and the wind-driven
11 subtropical ocean gyre can account for the dipole pattern in quasi-decadal fluctuations
observed in winter temperatures east of Newfoundland and off the southeastern United
States (Deser and Blackmon, 1993; Grötzner et al., 1998). A relationship analogous to
that reported for the North Atlantic by Bjerknes is also observed in the North Pacific
(Latif and Barnett, 1996). It involves interdecadal variations in sea surface temperatures
in the subtropical gyre of the North Pacific, heat transport by the Kuroshio Current, and
the strength of the Aleutian low.
A pan-Atlantic decadal oscillation (PADO) has been proposed, based on a zonally aver-
aged linear dynamic ocean–atmosphere model of the tropics (Xie and Tanimoto, 1998).
Sea surface temperature anomalies are related to evaporation and the surface advection
of heat. There are zonal bands of SST and wind anomalies with alternating polarities
from the subtropical South Atlantic to Greenland. In the model, extratropical wind forcing
sets up the observed decadal oscillations in SST. Low-frequency SST anomalies form an
anti-symmetric dipole about the equator between 20°N and 20°S, whereas poleward of
20°N there is an equatorward-propagating perturbation. Rajagopalan et al. (1998) find a
strong coherence between the NAO and the SST difference between the northern and
southern tropical Atlantic. Since both these tropical ocean regions are separately corre-
lated with the NAO index over this time band, it is posssible that ocean–atmosphere
interaction affects climate variability in the North Atlantic.
Using the ECMWF–Hamburg (ECHAM-3) atmospheric model coupled with the large-
scale geostrophic (LSG) ocean model in a 700 year control integration, Timmerman et
al. (1998) find evidence of an approximately thirty-five-year period of oceanic meridional
overturning and matching anomalies of North Atlantic SSTs in the region 20°–70°W,
30°–50°N. A possible model of the interactions involved, based on a lag regression
analysis, is shown in Figure 5.39. A similar thirty-five-year oscillation is identified in the
North Pacific sea surface temperatures for 120°E–160°W, 20°–40°N despite the absence
of GSA-type events in the Pacific.
Appendix 5.1 Partitioning between equatorially symmetric

and antisymmetric components
The spatial symmetry of climatological anomaly fields about the equator is often a useful
diagnostic tool for examining manifestations of the annual cycle and its variations. For
Figure 5.39 Schematic model of interactions that

may produce an interdecadal cycle in the North
Atlantic. The negative feedback loop begins with a
negative SST anomaly, leading to a weakened NAO.
This causes reduced fresh water transport, which
creates positive surface salinity (SSS) anomalies in
the northwest Atlantic, enhancing deep convection.
This subsequently strengthens the THC and pole-
ward heat transport, giving positive SST anomalies,
completing the cycle. (Timmermann et al., 1998)
11 example, the thermal field over a tropical ocean in response to solar radiation forcing
should be symmetrical about the equator, with a semi-annual regime in response to the
corresponding second harmonic in solar forcing; the amplitude of response would be a
maximum at the equator. There would also be an annual response in the thermal (and
pressure) field, antisymmetric about the equator, with a minimum amplitude at the equator,
increasing toward the tropics. The latter regime is characteristic of the eastern Pacific
Ocean; the former resembles the annual variation over continental Africa, the tropical
Indian Ocean, and the western Pacific (Wang, 1993).
For any monthly mean anomaly field (x, y) in a given month, the symmetric (s) and
0 antisymmetric (a) components are:
1
Fs(x, y) = [ f (x, y) F(x, y)]
2
11
1
Fa (x, y) = [ f (x, y) F(x, y)]
2
where x, y are respectively longitude and latitude distance. Wang (1993) notes that
partioning of the continuity equation (in pressure coordinations) yields:
0
∂us ∂va ∂a
=0
∂x ∂y ∂p
and
∂ua ∂vs ∂a
=0
∂x ∂y ∂p
This suggests a requirement of mass conservation for a linkage between the a (or s) part
0 of the meridional wind with the s (or a) part of the zonal wind.
Notes
1 Three principal versions of the SOI are in use (Climate Analysis Center, 1986):
Troup (1965): sea-level pressure anomalies at Tahiti minus Darwin are divided by the standard
deviation for that month of the difference time series. This index is generally multiplied
by 10.
Trenberth (1984): the Tahiti and Darwin anomalies are normalized by the standard deviations
of the respective anomaly time series. Twelve-month mean standard deviations are 0.931
0 for Tahiti and 1.003 for Darwin. The (Tahiti–Darwin) difference of the standardized anomaly
is then computed. CAC (1986) follows the Trenberth procedure and then further normal-
izes the difference by the standard deviation of the TN minus DN difference time series.
(TN and DN are the standardized anomalies for Tahiti and Darwin, respectively, based on
the 1951–80 base period.) This gives a time series with a zero mean and unit variance.
CAC (1986): tabulates monthly values for 1935–85. The Troup and CAC values are closely
similar. CAC is now the Climate Prediction Center at NCEP
2 Singular spectrum analysis is an application of PCA to a univariate time series, i.e. time lags
replace the spatial direction. Unlike spectral analysis, the functions are determined empirically,
not a priori. Oscillatory modes have pairs of nearly equal eigen values; their temporal EOFs
0 and PCs have the same time scale of oscillation and are approximately 90° out of phase. The
root of the eigen value represents the singular value of the corresponding temporal PC (Keppenne
and Ghil, 1992). The method is equivalent to the extended EOFs described by Weare and
11 Nasstrom (1982).
3 The Rossby radius of deformation, LR = c/f, where c = wave speed, f = the Coriolis parameter,
defines the distance where the amplitude becomes negligible. For a stratified fluid, the Rossby
radius of deformation, LR, is
冢
冣 )
1/2
LR = gh f

where h the vertical extent of the disturbance and / the degree of density stratifica-
tion (Hasse and Dobson, 1986, p. 71). In the ocean LR ~ 50 km and in the atmosphere LR
~ 800 km.
4 Time series of five main teleconnection indices defined by Wallace and Gutzler (1981) are now
routinely updated so that low-frequency fluctuations in these teleconnections can be examined
(Bell and Halpert, 1993). (ftp://ftp.ncep.noaa.gov:/pub/cpc/wd52dg/data.indices/tele_index.nh)
References
Aceituno, D. 1992. El Niño, the Southern Oscillation, and ENSO: confusing names for a complex
ocean–atmosphere interaction. Bull. Amer. Met. Soc., 73: 483–85.
Alexander, M.A. and Deser, C. 1995. A mechanism for the recurrence of wintertime midlatitude
SST anomalies. J. Phys. Oceanogr., 25: 122–37.
Alexander, M.A., Deser, C., and Timlin, C.S. 1999. The reemergence of SST anomalies in the
North Pacific Ocean. J. Climate, 12 (8, Pt 1): 2419–33.
Allan, R., Lindesay, J., and Parker, D. 1996. El Niño, Southern Oscillation and Climate Variability.
CSIRO, Collingwood, Victoria, 405 pp.
Allen, M.R. and Smith, L.A. 1994. Investigating the origins and significance of low-frequency
modes of climatic variability. Geophys. Res. Lett., 21: 883–6.
Anderson, R.Y. 1992. Long-term changes in the frequency of El Niño events. In: H.F. Diaz and
V. Markgraf, eds, El Niño: Historical and Paleoclimatic Aspects of the Southern Oscillation,
Ångstrom, A. 1935. Teleconnections of climate changes in present time. Geogr. Annal., 17: 242–58.
Arkin, P.A. and Webster, P.J. 1985. Annual and interannual variability of tropical–extratropical
interaction: an empirical study. Mon. Wea. Rev., 113: 1510–22.
Baldwin, M.P., Cheng, X., and Dunkerton, T.J. 1994. Observed correlations between winter-mean
tropospheric and stratospheric circulation anomalies. Geophys. Res. Lett., 21: 1141–4.
Barnett, T.P. 1984. Interaction of the monsoon and the Pacific tradewind system at interannual time
scales. III. A partial anatomy of the Southern Oscillation. Mon. Wea. Rev., 112 (12): 2388–400.
Barnett, T.P. 1985. Variations in near-global sea level pressure. J. Atmos. Sci., 42: 478–501.
Barnett, T.P. 1988. Variations in near-global sea level pressure: another view. J. Climate, 1: 225–30.
Barnett, T.P. 1991. The interaction of multiple time scales in the tropical climate system. J. Climate,
4: 269–85.
Barnett, T.P., Latif, M., Kirk, E., and Roeckner, E. 1991. On ENSO physics. J. Climate, 4: 487–515.
Barnett, T.P., Pierce, D.W., Latif, M., Dommenget, D., and Saravanan, R. 1999. Interdecadal inter-
actions between tropical and midlatitudes in the Pacific basin. Geophys. Rev. Lett., 26: 615–18.
Barnston, A.G. and Livezey, R.E. 1987. Classification, seasonality and persistence of low-frequency
atmospheric circulation patterns. Mon. Wea. Rev., 115: 1083–126.
Barnston, A.G. and Livezey, R.E. 1991. Statistical prediction of January–February mean northern
hemisphere lower tropospheric climate from the 11-year solar cycle and the Southern Oscillation
for west and east QBO phases. J. Climate, 4 (2): 249–62.
Barnston, A.G., Livezey, R.E., and Halpert, M.S. 1991. Modulation of Southern Oscillation–northern
hemisphere midwinter climate relationships by the QBO. J. Climate 4 (2): 203–17.
Barry, R.G. and Chorley, R.J. 1998. Atmosphere, Weather and Climate, 7th edition. Routledge,
London, 409 pp.
Battisti, D.S. and Hirst, A.C. 1989. Interannual variability in the tropical atmosphere/ocean system:
influence of the basic state, ocean geometry and non-linearity. J. Atmos. Sci., 46: 1687–712.
Behrend, H. 1987. Teleconnections of rainfall anomalies and of the Southern Oscillation over the
entire tropics and their seasonal dependence. Tellus, 39A: 138–51.
Bell, G.D. and Halpert, M.S. 1993. The global climate of March–May 1991: anomalous low-
11 frequency fluctuations dominate the mid-latitudes; ENSO becomes established in the tropics. J.
Climate, 6: 1413–33.
Berberry, E.H., Nogués-Paegle, J., and Horel, J.D. 1992. Wavelike southern hemisphere extratrop-
ical teleconnections. J. Atmos. Sci., 49 (2): 155–72.
Berlage, H.P. 1957. Fluctuations of the General Atmospheric Circulation of more than one Year:
their Nature and Prognostic Value. Meded. Verhandel. Kon. Nederland. Met. Inst. (de Bilt) 69,
152 pp.
Biggs, G.R. and Inoue, M. 1992. Rossby waves and El Niño during 1935–46. Quart. J. Roy. Met.
Soc., 118: 125–52.
Bjerknes, J. 1963. Climatic change as an ocean–atmosphere problem. In: Changes of Climate, Arid
0 Zone Research 20, UNESCO, Paris, pp. 297–321.
Bjerknes, J. 1964. Atlantic air–sea interaction. Adv. Geophys., 10: 1–82.
Bjerknes, J. 1966. Survey of El Niño 1957–58 in its relation to tropical Pacific meteorology. Inter-
American Tropical Tuna Commission Bull., 12: 1–62.
11 Bjerknes, J. 1969. Atmospheric teleconnections from the equatorial Pacific. Mon. Wea. Rev., 97:
165–72.
Bjerknes, J. 1996. A possible response of the atmospheric Hadley circulation to equatorial anom-
alies of ocean temperature. Tellus, 18: 820–9.
Blackmon, M.L., Lee, Y.H., and Wallace, J.M. 1984a. Horizontal structure of 500 mb height fluc-
tuations with long, intermediate and short time scales. J. Atmos. Sci., 41: 961–79.
Blackmon, M.L., Lee, Y.H., Wallace, J.M., and Hsu, H.H. 1984b. Time variation of 500 mb height
0 fluctuations with long, intermediate and short time scales as deduced from lag-correlation statis-
tics. J. Atmos. Sci., 41: 981–91.
Böehme, W. 1967. Eine 26-monatige Schwankung der Haüfigkeit meridionaler Zirkulationsformen
über Europa. Zeit. Met., 19, 113–15.
Brassington, G.B. 1996. The modal evolution of the Southern Oscillation. J. Clim., 10 (5): 1021–34.
Brier, G.W. 1978. The quasi-biennial oscillation and feedback processes in the atmosphere–
ocean–earth system. Mon. Wea. Rev., 106: 938–46.
Bromwich, D.H., Robasky, F.M., Cullather, R.I., and Van Woert, M.L. 1995. The atmospheric
hydrologic cycle over the Southern Ocean and Antarctica from operational numerical analyses.
Mon. Weath. Rev., 123: 3518–38.
Brown, B.G. and Katz, R.W. 1991. Use of statistical methods in the search for teleconnections:
0 past, present and future. In: M.H. Glantz, R.W. Katz, and N. Nicholls, eds, Teleconnections
Linking Worldwide Climate Anomalies, Cambridge University Press, Cambridge, pp. 371–400.
Buell, C.E. 1975. The topography of the empirical orthogonal functions. Preprints, Fourth
Conference on Probability and Statistics in Atmospheric Sciences, Amer. Met. Soc., Boston MA,
pp. 188–93.
Cane, M.A. 1992. Tropical Pacific ENSO models: ENSO as a mode of the coupled system. In: K.E.
Trenberth, ed., Climate System Modeling, Cambridge University Press, Cambridge, pp. 583–614.
Carleton, A.M. 1988a. Sea-ice atmosphere signal of the Southern Oscillation in the Weddell Sea,
Antarctica. J. Climate, 1, 379–88.
Carleton, A.M. 1988b. Meridional transport of eddy sensible heat in winters marked by extremes
of the North Atlantic Oscillation, 1948/49–1979/80. J. Climate, 1: 212–23.
0 Carleton, A.M. 1989. Antarctic sea-ice relationships with indices of the atmospheric circulation of
the southern hemisphere. Climate Dynamics, 3: 207–20.
Carleton, A.M. and Fitch, M. 1993. Synoptic aspects of Antarctic mesocyclones. J. Geophys. Res.,
98: 12997–13018.
Carleton, A.M., John, G., and Welsch, R. 1999. Interannual variations and regionality of Antarctic
sea ice – temperature associations. Ann. Glaciol., 27: 403–8.
Caviedes, C.N. 1984. El Niño, 1982–83. Geogr. Rev., 74: 267–90.
Chao, Yi and Philander, S.G.H. 1993. On the structure of the Southern Oscillation. J. Climate, 6:
450–69.
Chen, B., Smith, S.R., and Bromwich, D.H. 1996. Evolution of the tropospheric split jet over the
South Pacific Ocean during the 1986–89 ENSO cycle. Mon. Wea. Rev., 124: 1711–31.
0 Cheng, X. and Wallace, J.M. 1992. Cluster analysis of the northern hemisphere wintertime 500-
hPa height field: spatial patterns. J. Atmos. Sci., 49: 2674–96.
Christoph, M., Barnett, T.P., and Roeckner, E. 1998. The Antarctic circumpolar wave in a coupled
11 ocean–atmosphere GCM. J. Climate 11 (7): 1659–72.
Chu, P.S., Frederick, J., and Nash, A.J. 1991. Exploratory analysis of surface winds in the equa-
torial western Pacific and El Niño. J. Climate, 4: 1087–102.
Clarke, A.J. 1991. On the reflection and transmission of low-frequency energy at the irregular
western Pacific Ocean boundary. J. Geophys. Res., 96, supplement, pp. 3221–37.
Clayton, H.H. 1885. A lately discovered meteorological cycle. Amer. Met. J., 1: 130, 528.
Clayton, H.H. 1936. Long-range weather changes and methods of forecasting. Mon. Wea. Rev., 64:
359–76.
Climate Analysis Center 1986. Climate Diagnostics Bulletin, March 1986. Global analysis and
indices. NOAA/National Weather Service, Washington DC, pp. 9–10; tables A1–3.
Compo, G.P., Kiladis, G.N., and Webster, P.J. 1999. The horizontal and vertical structure of East
Asian winter monsoon pressure surges. Quart. J. Roy. Met. Soc., 125: 29–54.
Connolley, W.M. 1997. Variability in annual mean circulation in southern high latitudes. Clim.
Dynam., 13 (10): 745–56.
Corti, S., Molteni, F., and Palmer, T.N. 1999. Signature of recent climate change in frequencies of
natural and atmospheric circulation regimes. Nature, 398: 799–802.
Covey, D.L. and Hastenrath, M. 1978. The Pacific El Niño phenomenon and the Atlantic circula-
tion. Mon. Wea. Rev., 106: 280–7.
Cullather, R.I., Bromwich, D.H., and Van Woert, M.L. 1996. Interannual variations in Antarctic
precipitation related to El Niño–Southern Oscillation. J. Geophys. Res., 101: 19109–18.
Deser, C. and Blackmon, M.L. 1993. Surface climate variations over the North Atlantic Ocean
during winter, 1900–89. J. Climate, 6 (9): 1743–53.
Deser, C. and Wallace, J.M. 1987. El Niño events and their relation to the Southern Oscillation,
1925–86. J. Geophys. Res., 92 (C13): 14189–96.
Deser, C. and Wallace, J.M. 1990. Large-scale atmospheric circulation features of warm and cold
episodes in the tropical Pacific. J. Climate 3: 1254–81.
Diaz, H.F. and Kiladis, G.N. 1992. Atmospheric teleconnections associated with the extreme phases
of the Southern Oscillation. In: H.F. Diaz and V. Markgraf, eds, El Niño: Historical and Paleo-
climatic Aspects of the Southern Oscillation, Cambridge University Press, Cambridge, pp. 7–28.
Diaz, H.F. and Markgraf, V. (eds). 1992. El Niño: Historical and Paleoclimatic Aspects of the
Southern Oscillation, Cambridge University Press, Cambridge, 476 pp.
Diaz, H.F. and Pulwarty, R.S. 1992. A comparison of Southern Oscillation and El Niño signals in
the tropics. In: H.F. Diaz and V. Markgraf, eds, El Niño. Historical and Paleoclimatic Aspects
of the Southern Oscillation, Cambridge University Press, Cambridge, pp. 175–92.
Diaz, H.F. and Pulwarty, R.S. 1994. An analysis of the time-scales of variability in centuries-long
ENSO-sensitive records in the last 1000 years. Clim. Change, 26: 317–34.
Dole, R.M. 1986. Persistent anomalies of the extratropical northern hemisphere wintertime circu-
lation structure. Mon. Wea. Rev., 114: 178–207.
du Penhoat, Y. and Cane, M.A. 1991. Effect of low-latitude western boundary gaps on the reflec-
tion of equatorial motion. Geophys. Res., 96, supplement, pp. 3307–22.
Dunkerton, T.J. 1997. The role of gravity waves in the quasi-biennial oscillation. J. Geophys. Res.,
102: 26053–76.
Elliott, W.P. and Angell, J.K. 1988. Evidence for changes in Southern Oscillation relationships
during the last 100 years. J. Climate, 1: 729–737.
Emery, W.J., Lee, W.G., and Magaard, L. 1984. Geographic and seasonal distribution of
Brunt–Väisälä frequency and Rossby radii in the North Pacific and North Atlantic. J. Phys.
Oceanogr., 14: 294–317.
Enfield, D.B. 1989. El Niño, past and present. Rev. Geophys, 27: 159–87.
Enfield, D.B. 1992. Historical and prehistorical overview of El Niño/Southern Oscillation. In:
H.F. Diaz and V. Markgraf, eds, El Niño: Historical and Paleoclimatic Aspects of the Southern
Oscillation, Cambridge University Press, Cambridge, pp. 95–117.
Ephron, B. and Gong, G. 1983. A leisurely look at the bootstrap, the jackknife and cross-validation.
Amer. Statistician, 37: 36–48.
Esbensen, S.K. 1984. A comparison of intermonthly and interannual teleconnection in the 700 mb
geopotential height field during the northern hemisphere winter. Mon. Wea. Rev., 112: 2016–32.
Farrara, J.D., Ghil, M., Mechoso, C.R., and Mo, K.C. 1989. Empirical orthogonal functions and
multiple flow regimes in the southern hemisphere winter. J. Atmos. Sci., 46: 3219–23.
Flügel, M. and Chang, P. 1999. Stochastically induced climate shift of El Niño–Southern Oscillation.
Geophys. Res. Lett., 26 (16): 2473–6.
11 Fraedrich, K. and Müller, K. 1994. Climatology of wavenumber-frequency spectra of the 500 mb
height along 50 EN during the El Niño/Southern Oscillation extremes. Met. Zeit., NF, 3: 80–4.
Fraedrich, K., Müller, K., and Kuglin, R. 1992. Northern hemisphere circulation regimes during
the extremes of the El Niño/Southern Oscillation. Tellus, 44A: 33–40.
Frederiksen, J.S. and Webster, P.J. 1988. Alternative theories of atmospheric teleconnections and
low-frequency fluctuations. Rev. Geophys., 26: 459–94.
Fu, C.-B., Fletcher, J.O., and Diaz, H.F. 1986. Characteristics of the response of sea surface temper-
atures in the central Pacific associated with warm episodes of the Southern Oscillation. Mon.
Wea. Rev., 114: 1716–38.
Geisler, J., Blackmon, M.L., Bates, G.T., and Muñoz, S. 1985. Sensitivity of January climate
0 response to the magnitude and position of equatorial Pacific sea surface temperature anomalies.
J. Atmos. Sci., 42: 1037–49.
Gershunov, A. and Barnett, T.P. 1998. Interdecadal modulation of ENSO teleconnections. Bull.
Amer. Met. Soc., 79 (12): 2715–25.
11 Gill, A.E. 1980. Some simple solutions for heat induced tropical circulation. Quart. J. Roy. Met.
Soc., 106: 447–62.
Glantz, M.H., Katz, R.W., and Nicholls, N. (eds). 1991. Teleconnections linking Worldwide Climate
Anomalies. Cambridge University Press, Cambridge, 535 pp.
Gloersen, P. 1995. Modulation of hemispheric sea-ice cover by ENSO events. Nature, 373: 503–6.
Gong, D.-Y. and Wang, S.-W. 1998. Antarctic Oscillation: concept and applications. Chinese Sci.
Bull., 43: 734–8.
0 Gong, D.-Y. and Wang, S.-W. 1999. Definition of an Antarctic Oscillation Index. Geophys. Res.
Lett., 26 (4): 459–62.
Goswani, B.N. 1995. A multiscale interaction model for the origin of the tropospheric QBO. J.
Climate, 8: 524–34.
Graham, N.E. and Barnett, T.P. 1987. Sea surface temperature, surface wind divergence, and convec-
tion over tropical oceans. Science, 238: 657–9.
Graham, N.E. and White, W.B. 1988. The El Niño cycle: a natural oscillator of the Pacific
Ocean–atmosphere system. Science, 240: 1293–302.
Graham, N.E., Michaelson, J., and Barnett, T.P. 1987. An investigation of the El Niño–Southern
Oscillation cycle with statistical models. 2. Model results. J. Geophys. Res., 92: 14271–89.
0 Gray, W.M. and Sheaffer, J.D. 1991. El Niño and QBO influences on tropical cyclone activity. In:
M.H. Glantz et al., eds, Teleconnections Linking Worldwide Climate Anomalies. Cambridge
Grötzner, A., Latif, M. and Barnett, T.P. 1998. A decadal climate cycle in the North Atlantic Ocean
as simulated by the ECHO coupled GCM. J. Climate, 11 (5): 831–47.
Gu, D. and Philander, S.G.H. 1997. Interdecadal climate fluctuations that depend on exchanges
between the tropics and the extratropics. Science, 275: 805–7.
Halpert, M.S. and Ropelewski, C.F. 1992. Surface temperature patterns associated with the Southern
Oscillation. J. Climate, 5: 577–93.
Harrison, D.A. and Vecchi, G.A. 1999. On the termination of El Niño. Geophys. Res. Lett., 26
(11): 1593–6.
0 Harrison, D.E. and Giese, B.S. 1991. Episodes of surface westerly winds as observed from islands
in the western tropical Pacific. Geophys. Res., 96, supplement, 3221–37.
Harrison, D.E. and Larkin, N.K. 1996. The COADS sea level pressure signal: a near-global El Niño
composite and time series view. J. Climate, 9: 3025–55.
Harrison, D.E. and Larkin, N.K. 1998a. El Niño–Southern Oscillation sea surface temperature and
wind anomalies. Rev. Geophys., 36: 353–99.
Harrison, D.E. and Larkin, N.K. 1998b. Seasonal US temperature and precipitation anomalies asso-
ciated with El Niño: historical results and comparison with 1997–98. Geophys. Res. Lett., 25
(21): 3959–62.
Harrison, D.E. and Schopf, P.E. 1984. Kelvin wave-induced anomalous advection and the onset of
surface warming in El Niño events. Mon. Wea. Rev., 112: 923–33.
0 Hartten, L.M. 1996. Synoptic settings of westerly wind bursts. J. Geophys. Res., 101 (D12): 16997–
17,019.
Hasse, L. and Dobson, F. 1986. Introductory Physics of the Atmosphere and Ocean. Reidel,
11 Dordrecht, 126 pp.
Hastenrath, S., de Castro, L.-C., and Aceituno, P. 1987. The Southern Oscillation in the tropical
Atlantic sector. Beiträge Phys. Atmos., 60: 447–63.
Haynes, P.H. 1998. The latitudinal structure of the quasi-biennial oscillation. Quart. J. Roy. Met.
Soc., 124: 2645–70.
Heckley, W.A. and Gill, A.E. 1984. Some simple analytical solutions to the problem of forced
equatorial long waves. Quart. J. Roy. Met. Soc., 110: 203–17.
Helland-Hansen, B. and Nansen, F. 1920. Temperature Variations in the North Atlantic Ocean and
in the Atmosphere. Smithsonian Misc. Coll. 70 (4), Smithsonian Institution, Washington DC,
408 pp.
Hendon, H.H., Liebmann, B., and Glick, J.D. 1998. Oceanic Kelvin waves and the Madden–Julian
Oscillation. J. Atmos. Sci., 55 (1): 88–101.
Hildebransson, H.H. 1897. Quelques recherches sur les centers d’action de l’atmosphère: écarts
et moyennes barométriques mensuelles. Kongl. Svenska Vetenscaps-Akad. Handlinger 29,
Stockholm, 36 pp.
Hoerling, M.P., Kumar, A., and Zhang, M. 1997. El Niño, La Niña and the non-linearity of their
teleconnections. J. Climate, 10 (8): 1769–86.
Horel, J.D. 1981. A rotated principal component analysis of the interannual variability of the northern
hemisphere 500 mb height field. Mon. Wea. Rev., 109 (10): 2080–92.
Horel, J.D. and Wallace, J.M. 1981. Planetary-scale atmospheric phenomena associated with the
Southern Oscillation. Mon. Wea. Rev., 109: 813–29.
Horel, J.D., Kousky, V.E., and Kagano, M.T. 1986. Atmospheric conditions in the Atlantic sector
during 1983 and 1984. Nature, 322: 248–51.
Hoskins, B.J. and Ambrizzi, T. 1993. Rossby wave propagation on a realistic longitudinally varying
flow. J. Atmos. Sci., 50 (12): 1661–76.
Hoskins, B.J. and Karoly, D. 1981. The steady linear response of a spherical atmosphere to thermal
and orographic forcing. J. Atmos. Sci., 38: 1179–96.
Hsu, H.H. and Lin, S.H. 1992. Global teleconnections in the 250-mb stream function field during
the northern hemisphere winter. Mon. Wea. Rev., 120 (7): 1169–90.
Hsu, H.H. and Wallace, J.M. 1985. Vertical structures of wintertime teleconnection patterns.
J. Atmos. Sci., 42: 1693–710.
Huang, J.P., Higuchi, K., and Shabbar, A. 1998. The relationship between the North Atlantic
Oscillation and the El Niño–Southern Oscillation. Geophys. Res. Lett., 25 (14): 2707–10.
Hurrell, J.W. 1995. Decadal trends in the North Atlantic Oscillation: regional temperatures and
precipitation. Science, 269 (5224): 676–9.
Hurrell, J.W. and van Loon, H. 1997. Decadal variations in climate associated with the North
Atlantic Oscillation. Clim. Change, 36: 301–26.
Iskenderian, H. 1995. A 10-year climatology of northern hemisphere tropical cloud plumes and
their composite flow patterns. J. Climate, 8: 1630–7.
Jin, F.-F. 1997. An equatorial recharge paradigm for ENSO: conceptual model. J. Atmos. Sci., 54:
830–45.
Jin, F.-F., Neelin, J.D., and Gill, M. 1994. El Niño on the Devil’s Staircase: annual subharmonic
steps to chaos. Science, 264 (5155): 70–2.
Jones, P.D., Jonssen, T., and Wheeler, D. 1997. Extension to the North Atlantic Oscillation using
early instrumental pressure observations from Gibraltar and south-west Iceland. Intl. J. Climatol.,
17 (13): 1433–50.
Joung, C.H. and Hitchman, M.H. 1982. On the role of successive downstream development in East
Asian polar air outbreaks. Mon. Wea. Rev., 110: 1224–37.
Julian, P.R. and Chervin, R.M. 1978. A study of the Southern Oscillation and Walker Circulation
phenomenon. Mon. Wea. Rev., 106: 1433–51.
Kaiser, H.F. 1958. The varimax criterion for analytic rotation in factor analysis. Psychometrika, 23:
187–200.
Kapala, A., Mächel, H., and Flohn, H. 1998. Behavior of the centers of action above the Atlantic
since 1881. II. Association with regional climate anomalies. Intl. J. Climatol., 18 (1): 23–6.
Karoly, D.J. 1990. The role of transient eddies in low-frequency zonal variations of the southern
hemisphere circulation. Tellus, 42A: 41–50.
Katz, R.W. 1988. Use of cross-correlations in the search for teleconnections. J. Climatol., 8: 241–53.
Keen, R.A. 1982. The role of cross-equatorial tropical cyclone pairs in the Southern Oscillation.
Mon. Wea. Rev., 110: 1405–16.
11 Keppenne, C.L. and Ghil, M. 1992. Adaptive filtering and prediction of the Southern Oscillation
Index. J. Geophys. Res., 97 (D18): 20449–54.
Kessler, W.S. 1991. Can reflected extra-equatorial Rossby waves drive ENSO? J. Phys. Oceanogr.,
21: 444–52.
Kessler, W.S. and McPhadden, M.J. 1995. Oceanic equatorial waves and the 1991–93 El Niño.
J. Climate, 8 (7): 1757–74.
Kidson, J.W. 1975. Tropical eigenvector analysis and the Southern Oscillation. Mon. Wea. Rev.,
103: 187–96.
Kidson, J.W. 1988a. Interannual variations in the southern hemisphere circulation. J. Climate, 1:
1177–98.
0 Kidson, J.W. 1988b. Indices of the southern hemisphere zonal wind. J. Climate, 1: 183–94.
Kidson, J.W. 1991. Intraseasonal variations in the southern hemisphere circulation. J. Climate, 4:
939–53.
Kiladis, G.N. 1998. Observations of Rossby waves linked to convection over the eastern tropical
11 Pacific. J. Atmos. Sci., 55: 321–9.
Kiladis, G.N. and Diaz, H.F. 1989. Global climatic anomalies associated with extremes of the
Southern Oscillation. J. Climate, 2 (9): 1069–90.
Kiladis, G.N. and Mo, K.C. 1998. Interannual and intraseasonal variability in the southern hemi-
sphere. In: D.J. Karoly and D.G. Vincent, eds, Meteorology of the Southern Hemisphere, Met.
Monogr., 27 (49), pp. 307–36.
Kiladis, G.N. and Sinha, S.K. 1991. ENSO, monsoon and drought in India. In: M.H. Glantz, R.W.
0 Katz, and N. Nicholls, eds, Teleconnections Linking Worldwide Climate Anomalies, Cambridge
Kiladis, G.N. and van Loon, H. 1988. The Southern Oscillation. VII. Meteorological anomalies
over the Indian and Pacific sectors associated with the extremes of the oscillation. Mon. Wea.
Rev., 116: 120–36.
Kiladis, G.N. and Weickmann, K.M. 1992a. Circulation anomalies with tropical convection during
northern winter. Mon. Wea. Rev., 120: 1900–23.
Kiladis, G.N. and Weickmann, K.M. 1992b. Extratropical forcing of tropic Pacific convection during
northern winter. Mon. Wea. Rev., 120: 1924–38.
Kiladis, G.N. and Weickmann, K.M. 1997. Horizontal structure and seasonality of large-scale circu-
lations associated with tropical convection. Mon. Wea. Rev., 125 (9): 1997–2013.
0 Klasa, M., Jerome, D., and Sheng, J. 1992. On the interaction between the synoptic-scale eddies
and the PNA teleconnection pattern. Contrib. Atmos. Phys., 65: 211–22.
Kleeman, R., McCreary, J.P., Jr, and Klinger, B.A. 1999. A mechanism for generating ENSO
decadal variability. Geophys. Res. Lett., (26) 11: 1743–6.
Klein, S.A., Soden, B.J., and Lau, N.-C. 1999. Remote sea surface temperature variations during
ENSO: evidence for a tropical atmospheric bridge. J. Climate, 12 (4): 917–32.
Koslowski, G. and Loewe, P. 1994. The western Baltic Sea ice season in terms of a mass-related
severity index, 1879–1992. I. Temporal variability and association with the North Atlantic
Oscillation. Tellus, 46A: 66–74.
Kousky, V.E. and Bell, G.D. 1992. Atlas of Southern Hemisphere 500-mb Teleconnection Patterns
derived from National Meteorological Center Analyses. NOAA Atlas No. 9, Department of
0 Commerce, Washington DC, 90 pp.
Kumar, K.K., Kleeman, A.K., Cane, M.A., and Rajagopalan, B. 1999. Epochal changes in Indian
monsoon–ENSO precursors. Geophys. Res. Lett., 26 (1): 75–8.
Kushnir, Y. 1994. Interdecadal variations in North Atlantic sea surface temperature and associated
atmospheric conditions. J. Climate, 7 (1): 141–57.
Kushnir, Y. and Wallace, G.M. 1989. Low-frequency variability in the northern hemisphere winter:
geographical distribution, structure, and time scale dependence. J. Atmos. Sci., 46 (20): 3122–42.
Kutzbach, J.E. 1970. Large-scale features of the monthly mean northern hemisphere anomaly maps
of sea level pressure. Mon. Wea. Rev., 107: 708–16.
Labitzke, K. and van Loon, H. 1989. Association between the 11-year solar cycle, the QBO, and
the atmosphere. J. Climate, 2: 254–65.
0 Landsberg, H.E., Mitchell., J.M., Jr, Crutcher, H.L., and Quinlan, F.T. 1963. Surface signs of the
biennial atmospheric pulse. Mon. Wea. Rev., 91: 549–56.
Latif, M. and Barnett, T.P. 1996. Decadal climate variability over the North Pacific and North
11 America: dynamics and predictability. J. Climate, 9 (10): 2407–23.
Latif, M., Kleeman, R., and Eckert, C. 1997. Greenhouse warming, decadal variability, or El Niño:
An attempt to understand the anomalous 1990s. J. Climate, 10 (9): 2221–39.
Lau, K.-M. and Buc, W. 1998. Mechanisms of monsoon–Southern Oscillation coupling. Climate
Dynam., 14 (11): 759–80.
Lau, K.-M. and Chan, P.H. 1986. The 40–50 day oscillation and the El Niño/Southern Oscillation:
a new perspective. Bull. Amer. Met. Soc., 67: 533–4.
Lau, K.-M. and Lim, H. 1984. On the dynamics of equatorial forcing of climate teleconnections.
J. Atmos. Sci., 41: 161–76.
Lau, K.-M. and Yang, S. 1996. The Asian monsoon and predictability of the tropical ocean–atmos-
phere system. Quart. J. Roy. Met. Soc., 122: 945–57.
Lau, K.-M., Chang, C.P., and Chan, P.H. 1983. Short-term planetary-scale interactions over the
tropics and midlatitudes. II. Winter-MONEX period. Mon. Wea. Rev., 111: 1372–88.
Lau, K.-M. and Sheu, P.J. 1988. Annual cycle, quasi-biennial oscillation and Southern Oscillation
in global precipitation. J. Geophys. Res., 93 (D9): 10975–88.
Lau, N.C. 1998. Variability of the observed midlatitude storm tracks in relation to low-frequency
changes in the circulation pattern. J. Atmos. Sci., 45 (19): 2718–43.
Lau, N.C. and Lau, K.M. 1984. The structure and energetics of midlatitude disturbances accompa-
nying cold-air outbreaks over East Asia. Mon. Wea. Rev., 112: 1309–27.
Leathers, D.J. and Palecki, M.A. 1992. The Pacific/North American teleconnection pattern and
United States climate. II. Temporal characteristics and index specification. J. Climate, 5 (7):
707–16.
Leathers, D.J., Yarnal, B., and Palecki, M.A. 1991. The Pacific/North American teleconnection
pattern and United States climate. I. Regional temperatures and precipitation associations.
J. Climate, 4 (5): 517–22.
LeDrew, E.F. and Barber, D.G. 1994. The SIMMS programme: a study of change and variability
within the marine cryosphere. Arctic, 47: 256–64.
LeDrew, E.F., Johnson, D., and Maslanik, J.A. 1991. An examination of the atmospheric mecha-
nisms that may be responsible for the annual reversal of the Beaufort Sea ice field. Intl. J.
Climatol., 11: 841–59.
Li, Z.J. and Ji, L.R. 1997. Efficient forcing and atmospheric teleconnections. Quart. J. Roy. Met
Soc., 123: 2401–23.
Lindzen, R.S. and Holton, J.R. 1968. A theory of the quasi-biennial oscillation. J. Atmos. Sci., 25:
1095–107.
Livezey, R.E. and Mo, K.C. 1987. Tropical–extratropical teleconnections during the northern hemi-
sphere winter. II. Relationships between monthly mean northern hemisphere circulation patterns
and proxies for tropical convection. Mon. Wea. Rev., 115: 3115–32.
Lockyer, W.J.S. 1906. Barometric variations of long duration over large areas. Proc. Roy. Soc., A
78: 43–60.
Loewe, F. 1937. A period of warm winters in western Greenland and the temperature seesaw
between western Greenland and Europe. Quart. J. Roy. Met. Soc., 63: 365–71.
Lorenz, E.N. 1951. Seasonal and irregular variations of the northern hemisphere sea-level pressure
profile. J. Met., 8: 52–9.
Lukas, R. and Lindstrom, E. 1991. The mixed layer of the western equatorial Pacific Ocean. J.
Geophys. Res., 96, supplement: 3343–57.
Luterbacher, J., Schmutz, C., Gyalistras, D., Xoplaku, E., and Wanner, H. 1999. Reconstruction of
monthly NAO and EU indices back to AD 1675. Geophys. Res. Lett., 26 (17): 2745–8.
Luther, D.S., Harrison, D.E., and Knox, R.A. 1983. Zonal winds in the central equatorial Pacific
and El Niño. Science, 222: 327–30.
Mächel, H., Kapala, A., and Flohn, H. 1998. Behavior of the centers of action above the Atlantic
since 1881. I. Characteristics of seasonal and interannnual variability. Intl. J. Climatol., 18 (1):
1–22.
Makrogiannis, T.J., Bloutsos, A.A., and Gile, B.D. 1982. Zonal index and circulation change in the
North Atlantic area, 1873–1972. Intl. J. Climatol., 2 (2): 159–70.
Mann, M.E. and Park, J. 1994. Global-scale modes of surface temperature variability on interan-
nual to century timescales. J. Geophys. Res., 99: 25819–33.
Marshall, G.J. and King, J.C. 1998. Southern hemisphere circulation anomalies associated with
extreme Antarctic Peninsula winter temperatures. Geophys. Res. Lett., 25: 2437–40.
11 McCreary, J.P., Jr. 1983. A model of tropical ocean–atmosphere interaction. Mon. Wea Rev., 111
(2): 370–87.
McGuirk, J.P., Thompson, A.H., and Smith, N.R. 1987. Moisture bursts over the tropical Pacific
Ocean. Mon. Wea. Rev., 115: 787–98.
McPhadden, M.J. 1999. Genesis and evolution of the 1997–98 El Niño. Science, 283: 950–4.
McPhadden, M.J., et al. 1998. The Tropical Ocean–Global Atmosphere observing systems: a decade
of progress. J. Geophys. Res., 103 (C7): 14169–240.
Meehl, G.A. 1997. The South Asian monsoon and the tropospheric biennial oscillation. J. Climate,
10 (7): 1921–43.
Meehl, G.A. 1987a. The annual cycle and interannual variability in the tropical Pacific and Indian
0 Ocean regions. Mon. Wea. Rev., 115 (1): 27–50.
Meehl, G.A. 1987b. The tropics and their role in the global climate system. Geogr. J., 153: 21–36.
Meehl, G.A. 1993. A coupled air–sea biennial mechanism in the tropical Indian and Pacific regions:
role of the ocean. J. Climate, 6: 31–41.
11 Meehl, G.A. 1997. The South Asian monsoon and the tropospheric biennial oscillation. J. Climate,
10 (7): 1921–43.
Meehl, G.A., Arblaster, J.M., and Strand, W.G., Jr. 1998. Global decadal climate variability.
Geophys. Res. Lett., 25 (21): 3983–6.
Mehta, V.K. 1998. Variability of the tropical ocean surface temperatures at decadal–multidecadal
timescales. 1. The Atlantic Ocean. J. Climate, 11 (9): 2351–75.
Miller, L. and Cheney, R.E. 1990. Large-scale meridional transport in the tropical Pacific Ocean
0 during the 1986–87 El Niño. J. Geophys. Res., 95 (C10) 17: 905–20.
Miller, L., Cheney, R.E., and Douglas, B.C. 1988. Geostat altimeter observations of Kelvin waves
and the 1986–87 El Niño. Science, 239: 52–4.
Miyakoda, K., Navarra, A., and Ward, M.N. 1999. Tropical-wide teleconnection and oscillation. II.
The ENSO–monsoon system. Quart. J. Roy. Met. Soc., 125: 2937–63.
Mo, K. and Ghil, M. 1987. Statistics and dynamics of persistent anomalies. J. Atmos. Sci., 44: 877–901.
Mo, K. and Ghil, M. 1988. Cluster analysis of multiple planetary flow regimes. J. Geophys. Res.,
93: 10927–52.
Mo, K. and Kousky, V.E. 1993. Further analysis of the relationship between circulation anomaly
patterns and tropical convection. J. Geophys. Res., 98 (D3): 5103–13.
Mo, K. and Livezey, R.E. 1986. Tropical–extratropical geopotential height teleconnections during
0 the northern hemisphere winter. Mon. Wea. Rev., 114: 2488–515.
Mo, K.C. and White, G.H. 1985. Teleconnections in the southern hemisphere. Mon. Wea. Rev.,
113: 22–37.
Montgomery, R.B. 1940. Report on the work of G.T. Walker. Mon. Wea. Rev., Supplement 39: 1–22.
Moore, A.M. and Kleeman, R. 1997. The singular vectors of a coupled-atmosphere model of ENSO.
I. Thermodynamics, energetics and error growth. II. Sensitivity studies and dynamical interpre-
tation. Quart. J. Roy. Met. Soc., 123: 953–81, 983–1006.
Moron, V., Vautard, R., and Ghil, M. 1998. Trends, interdecadal and interannual oscillations in
global sea-surface temperatures. Clim. Dynam., 14: 545–69.
0 temperature trends. J. Climatol., 7: 13–30.
Murakami, T. and Sumathipala, W.L. 1989. Westerly bursts during the 1982/83 ENSO. J. Climate,
2: 71–90.
Mysak, L.A. 1986. El Niño, interannual variability and fisheries in the northeast Pacific Ocean.
Can. J. Aquat. Sci., 43: 464–97.
Mysak, L.A. and Venegas, S.A. 1998. Decadal climate oscillations in the Arctic: a new feedback
loop for atmosphere–ocean–ice interactions, Geophys. Res. Lett., 25 (A): 3607–10.
Mysak, L.A., Manak, S.K., and Marsden, R.F. 1990. Sea ice anomalies in the Greenland and Labrador
seas during 1901–84 and their relation to an interdecadal Arctic climate cycle. Clim. Dynam., 8:
103–16.
Nakamura, H., Tanaka, M., and Wallace, J.M. 1987. Horizontal structure and energetics of northern
0 hemisphere wintertime teleconnection patterns. J. Atmos. Sci., 44: 3377–91.
Namias, J. 1981. Teleconnections of 700 mb height anomalies for the northern hemisphere. Calcofi
Atlas No. 29 (California Cooperative Oceanic Fisheries Investigations), Scripps Institution of
11 Oceanography, La Jolla CA, 265 pp.
Namias, J. and Born, R.M. 1970. Temporal coherence in North Pacific sea surface temperatures.
J. Geophys. Res., 75: 5952–5.
Navarra, A., Ward, M.N., and Miyakoda, K. 1999. Tropical-wide teleconnection and oscillation. I.
Teleconnection indices and type1/type 2 states. Quart. J. Roy. Met. Soc., 125: 2909–35.
Neelin, J.D. and Held, I.M. 1987. Modeling tropical convergence based on the moist static energy
budget. Mon. Wea. Rev., 115 (1): 3–12.
Neelin, J.D., et al. 1992. Tropical air–sea interaction in general circulation models. Clim. Dynam.
7: 73–104.
Neelin, J.D., Battisti, D.S., Hirst, A.C., Jin, F.-F., Wakata, Y., Yamagata, T., and Zebiak, S.E. 1998.
ENSO theory. J. Geophys. Res., 103 (7): 14261–90.
Nicholls, N. 1988. El Niño–Southern Oscillation and rainfall variability. J. Climate, 1: 418–21.
Niiler, P. (ed.). 1982. Tropic Heat. A Study of the Tropical Pacific Upper Ocean Heat, Mass and
Momentum Budgets. (The CORE Research Program), Corvallis OR, 30 pp.
Nitta, T. 1987. Convective activity in the tropical western Pacific and their impact on the northern
hemisphere summer circulation. J. Met. Soc. Japan, 65: 373–90.
Nogués-Paegle, J. and Mo, K.C. 1988. Transient response of the southern hemisphere subtropical
jet to tropical forcing. J. Atmos. Sci., 45: 1493–508.
Normand, C. 1953. Monsoon seasonal forecasting. Quart. J. Roy. Met. Soc., 79: 463–73.
O’Connor, J.F. 1969. Hemispheric Teleconnections of Mean Circulation Anomalies at 700 mb. US
Weather Bureau Tech. Rep. 10, ESSA, Washington DC, 103 pp.
Palmer, T.N. 1988. Medium and extended range predictability and stability of the Pacific/North
American mode. Quart. J. Roy. Met. Soc., 114: 691–713.
Parker, D.E. 1983. Documentation of a Southern Oscillation index. Met. Mag., 112: 114–18.
Perlewitz, J. and Graf, H.-F. 1995. The statistical connection between tropospheric and stratospheric
circulation of the northern hemisphere in winter. J. Climate, 8: 2281–95.
Peterson, R.G. and White, W.B. 1998. Slow oceanic teleconnections linking the Antarctic
Circumpolar Wave with the tropical El Niño–Southern Oscillation. J. Geophys. Res., 103 (C11):
24573–83.
Philander, S.G.H. 1990. El Niño, La Niña, and the Southern Oscillation. Academic Press, San Diego
CA, 293 pp.
Pittock, A.B. 1980a. Patterns of climatic variation in Argentina and Chile. I. Precipitation, 1931–60.
Mon. Wea. Rev., 108: 1347–61.
Pittock, A.B. 1980b. Patterns of climatic variation in Argentina and Chile. II. Temperature, 1931–60.
Mon. Wea. Rev., 108: 1362–9.
Pittock, A.B. 1984. On the reality, stability, and usefulness of southern hemisphere teleconnections.
Austral. Met. Mag., 32: 75–82.
Quinn, W.H. 1992. A study of Southern Oscillation-related climatic activity for AD 622–1900 incor-
porating Nile River flood data. In: H.F. Diaz and V. Markgraf, eds, El Niño: Historical and
Paleoclimatic Aspects of the Southern Oscillation, Cambridge University Press, Cambridge, pp.
119–49.
Quinn, W.H. and Neal, V.T. 1992. The historical record of El Niño events. In: R.S. Bradley and
P.D. Jones, eds, Climate since AD 1500, Routledge, London, pp. 623–48.
Quinn, W.H., Neal, V.T., and Antunez de Mayolo, S.A. 1987. El Niño occurrences over the past
four and a half centuries. J. Geophys. Res., 92 (C13): 14449–61.
Quinn, W.H., Zopf, D.O., Short, K.S., and Kuo Yang, R.T.W. 1978. Historical trends and statis-
tics of the Southern Oscillation, El Niño, and Indonesian droughts. Fish. Bull., 76: 663–78.
Rajagopalan, B., Kushnir, Y., and Tourre, Y. 1998. Observed decadal midlatitude and tropical
Atlantic climate variability. Geophys. Res. Let., 25 (21): 3967–70.
Rajagopalan, B., Lill, K., and Cane, M.A. 1997. Anomalous ENSO occurrence: an alternate view.
J. Climate, 10 (9): 2351–7.
Ramage, C.S. 1977. Sea surface temperature and local weather. Mon. Wea. Rev., 105: 540–4.
Ramage, C.S. and Hori, A.M. 1981. Meteorological aspects of the El Niño. Mon. Wea. Rev., 109:
1827–35.
Rasmusson, E.M. 1991. Observational aspects of ENSO cycle teleconnections. In: M.H. Glantz et
al., eds, Teleconnections Linking Worldwide Climate Anomalies, Cambridge University Press,
Rasmusson, E.M. and Carpenter, T.H. 1982. Variations in tropical sea surface temperature and surface
wind fields associated with the Southern Oscillation/El Niño. Mon. Wea. Rev., 110: 354–84.
11 Rasmusson, E.M., Wang, X.-L., and Ropelewski, C.F. 1990. The biennial component of ENSO
variability. J. Mar. Systems, 1: 71–96.
Richman, M.B. 1993. Comments on “The effect of domain shape on principal components analyses.”
Intl. J. Climatol., 13: 203–18.
Rogers, J.C. 1981a. The North Pacific oscillation. J. Climatol., 1: 39–85.
Rogers, J.C. 1981b. Spatial variability of seasonal sea level pressure and 500 mb height anomalies.
Mon. Wea. Rev., 109 (10): 2093–106.
Rogers, J.C. 1983. Spatial variability of Antarctic temperature anomalies and their association with
southern hemisphere atmospheric circulation. Ann. Assoc. Amer. Geogr., 73: 502–18.
Rogers, J.C. 1984. The association between the North Atlantic Oscillation and the Southern
0 Oscillation in the northern hemisphere. Mon. Wea. Rev., 112: 1999–2015.
Rogers, J.C. 1985. Atmospheric circulation changes associated with the warming over the northern
North Atlantic in the 1920s. J. Clim. Appl. Met., 24: 1303–10.
Rogers, J.C. 1990. Patterns of low-frequency monthly sea level pressure variability (1899–1986)
11 and associated wave cyclone frequencies. J. Climate, 3 (12): 1364–79.
Rogers, J.C. 1997. North Atlantic storm track variability and its association to the North Atlantic
Oscillation and climate variability of northern Europe. J. Climate, 10 (7): 1635–47.
Rogers, J.C. and Mosley-Thompson, E. 1999. Atlantic Arctic cyclones and the mild Siberian winters
of the 1980s. Geophys. Res. Lett., 22 (7): 799–802.
Rogers, J.C. and van Loon, H. 1979. The seesaw in winter temperatures between Greenland and
northern Europe. II. Some oceanic and atmospheric effects in middle and high latitudes. Mon.
0 Wea. Rev., 107: 509–19.
Rogers, J.C. and van Loon, H. 1982. Spatial variability of sea level pressure and 500 mb height
anomalies over the southern hemisphere. Mon. Wea. Rev., 110: 1375–92.
Ropelewski, C.F. and Halpert, M.S. 1987. Global and regional scale precipitation patterns associ-
ated with El Niño/Southern Oscillation. Mon. Wea. Rev., 114: 2352–62.
Ropelewski, C.F. and Halpert, M.S. 1989. Precipitation patterns associated with the high index
phase of the Southern Oscillation. J. Climate, 2: 268–84.
Sardeshmukh, P.D. and Hoskins, B.J. 1985. Vorticity balances in the tropics during the 1982–83
El Niño–Southern Oscillation event. Quart. J. Roy. Met. Soc., 111: 261–78.
Schell, I.I. 1956. On the nature and origin of the Southern Oscillation. J. Met., 13: 592–8.
0 Schopf, P.S. and Suarez, M.J. 1988. Vacillations in a coupled ocean–atmosphere model. J. Atmos.
Sci., 45: 549–66.
Schopf, P.S. and Suarez, M.J. 1990. Ocean wave dynamics and the time scale of El Niño. J. Phys.
Oceanogr., 20: 629–45.
Shapiro, R. 1964. A mid-latitude biennial oscillation in the variation of the surface pressure distri-
bution. Quart. J. Roy. Met. Soc., 90: 328–31.
Shiotani, M. 1990. Low-frequency variations of the zonal mean state of the southern hemisphere
troposphere. J. Met. Soc. Japan, 68: 461–71.
Simmonds, I. and Jacka, T.H. 1995. Relationship between the interannual variability of Antarctic
sea ice and the Southern Oscillation. J. Climate, 8: 637–47.
Simmons, A., Wallace, J.M., and Branstator, G.W. 1983. Barotropic wave propagation and insta-
0 bilities and atmospheric teleconnection patterns. J. Atmos. Sci., 40: 1363–92.
Sinclair, M.R., Renwick, J.A., and Kidson, J.W. 1997. Low-frequency variability of southern hemi-
sphere sea level pressure and weather system activity. Mon. Wea. Rev., 125: 2531–43.
Smith, S.R. and Stearns, C.R. 1993. Antarctic pressure and temperature anomalies surrounding the
minimum in the Southern Oscillation index. J. Geophys. Res., 98: 13071–83.
Stahle, D.W. and Cleveland, M.K. 1993. Southern Oscillation extremes reconstructed from tree
rings of the Sierra Madre Occidental and southern Great Plains. J. Climate, 6: 129–40.
Stoeckenius, T. 1981. Interannual variations of tropical precipitation patterns. Mon. Wea. Rev., 109:
1233–47.
Streten, N.A. 1968. A note on multiple image photo-mosaics for the southern hemisphere. Austral.
Met. Mag., 16: 127–36.
0 Streten, N.A. 1983. Circulation contrasts in the southern hemisphere winters of 1972 and 1973.
Austral. Met. Mag., 31: 161–70.
Streten, N.A. and Zillman, J.W. 1984. Climate of the South Pacific Ocean. In: H. van Loon, ed.,
11 Climates of the Oceans, World Survey of Climatology, 15, Elsevier, Amsterdam, pp. 263–429.
Taylor, A.H. and Stephens, J.A. 1998. The North Atlantic Oscillation and the latitude of the Gulf
Stream. Tellus, 50A: 134–42.
Taylor, A.H., Jordan, M.B., and Stephens, J.A. 1998. Gulf Stream shifts following ENSO events.
Nature, 393 (6686): 638.
Thompson, D.W.J. and Wallace, J.M. 1998. The Arctic Oscillation signature in the wintertime
geopotential height and temperature fields. Geophys. Res. Lett., 25 (9): 1297–300.
Thompson, D.W.J. and Wallace, J.M. 2000. Annual modes in the extratropical circulation. I. Month-
to-month variability. J. Climate 13 (5): 1000–16.
Thompson, D.W.J., Wallace, M.J., and Hegerl, G.C. 2000. Annular modes in the extratropical circu-
lation. II. Trends. J. Climate 13 (5): 1018–36.
Timmerman, A., Latif, M., Voss, R., and Groetzner, A. 1998. Northern hemisphere interdecadal
variability: a coupled air–sea mode. J. Climate, 11 (8): 1906–31.
Ting, M.F., Hoerling, M.P., Xu, T.Y., and Kumar, A. 1996. Northern hemisphere teleconnection
patterns during extreme phases of the zonal mean circulation. J. Climate, 9 (10): 2614–33.
Tomita, T. and Yasunari, T. 1993. On the two types of ENSO. J. Met. Soc. Japan, 71 (2): 273–84.
Torrence, C. and Webster, P.J. 1998. The annual cycle of persistence in the El Niño/Southern
Oscillation. Quart. J. Roy. Met. Soc., 124: 1985–2004.
Tourre, Y.M., Rajagopalan, B., and Kushnir, Y. 1999. Dominant patterns of climate variability in
the Atlantic Ocean during the past 136 years. J. Climate, 12 (8, Pt 1): 2285–99.
Trenberth, K.E. 1976. Spatial and temporal variations of the Southern Oscillation. Quart. J. Roy.
Met. Soc., 102: 639–53.
Trenberth, K.E. 1984. Signal versus noise in the Southern Oscillation. Mon. Wea. Rev., 112: 326–32.
Trenberth, K.E. 1991a. Storm tracks in the southern hemisphere. J. Atmos. Sci., 48: 2159–78.
Trenberth, K.E. 1991b. General characteristics of El Niño–Southern Oscillation. In: M.H. Glantz,
R.W. Katz, and N. Nicholls, eds, Teleconnections Linking Worldwide Climate Anomalies,
Trenberth, K.E. 1997. The definition of El Niño. Bull. Amer. Met. Soc., 78 (12): 2771–7.
Trenberth, K.E. and Hoar, T.J. 1996. The 1990–95 El Niño–Southern Oscillation: longest on record.
Geophys. Res. Lett., 23 (1): 57–60.
Trenberth, K.E. and Hoar, T.J. 1997. El Niño and climate change. Geophys. Res. Letter., 24 (23):
3057–60.
Trenberth, K.E. and Paolino, D.A. 1981. Characteristic patterns of variability of sea level pressure
in the northern hemisphere. Mon. Wea. Rev., 109: 1169–89.
Trenberth, K.E. and Shea, D.J. 1987. On the evolution of the Southern Oscillation. Mon. Wea. Rev.,
115 (12): 3078–96.
Trenberth, K.E. and Mo, K.C. 1985. Blocking in the southern hemisphere. Mon. Wea. Rev., 113: 3–21.
Tribbia, J.J. 1991. The rudimentary theory of atmospheric teleconnections associated with ENSO.
In: M.H. Glantz, R.W. Katz, and N. Nicholls, eds, Teleconnections Linking Worldwide Climate
Anomalies, Cambridge University Press, Cambridge, pp. 285–308.
Troup, A.J. 1965. The Southern Oscillation. Quart. J. Roy. Met. Soc., 91: 490–506.
Ueno, K. 1993. Inter-annual variability of surface cyclone tracks, atmospheric circulation patterns
and precipitation patterns in winter. J. Met. Soc. Japan, 71 (6): 655–71.
van Loon, H. 1972. Pressure in the southern hemisphere. In: C.W. Newton, ed., Meteorology of
the Southern Hemisphere, Met. Monogr. 13 (35), Amer. Met. Soc., Boston MA, pp. 59–86.
van Loon, H. 1984. The Southern Oscillation. III. Associations with the trades and with the trough
in the westerlies of the South Pacific Ocean. Mon. Wea. Rev., 112 (5): 947–52.
van Loon, H. 1986. The characteristics of sea level pressure and sea surface temperature during
the development of a warm event in the Southern Oscillation. In: J.O. Roads, ed., Namias
Symposium, Scripps Inst. Oceanog. Ref. Ser. 86–17, La Jolla CA, pp. 160–73.
van Loon, H. and Madden, R.A. 1981. The Southern Oscillation. I. Global associations with pres-
sure and temperature in northern winter. Mon. Wea. Rev., 109 (6): 1150–62.
van Loon, H. and Rogers, J.C. 1978. The seesaw in winter temperatures between Greenland and
northern Europe. 1. General description. Mon. Wea. Rev., 106: 296–310.
van Loon, H. and Shea, D.J. 1985. The Southern Oscillation. IV. The precursors south of 15°S to
the extremes of the oscillation. Mon. Wea. Rev., 113 (12): 2063–74.
van Loon, H. and Shea, D.J. 1987. The Southern Oscillation. VI. Anomalies of sea level pressure
in the southern hemisphere and of Pacific sea surface temperature during the development of a
warm event. Mon. Wea. Rev., 115 (2): 370–9.
11 van Loon, H. and Labitzke, K. 1987. The Southern Oscillation. V. The anomalies in the lower
atmosphere of the northern hemisphere in winter and a comparison with the quasi-bienniel oscil-
lation. Mon. Wea. Rev., 115: 357–69.
van Loon, H. and Labitzke, K. 1989. Association between the 11-year solar cycle, the QBO, and
the atmosphere. II. Surface and 700 mb in the northern hemisphere in winter. J. Climate, 1:
905–20.
Verbickas, S. 1998. Westerly wind bursts in the tropical Pacific. Weather, 53 (9): 282–7.
Veryard, R.G. and Ebdon, R.A. 1961. Fluctuations in tropical stratospheric winds. Met. Mag., 90:
125–43.
Villalba, R., Cook, E.R., D’Arrigo, R.D., Jacoby, G.C., Jones, P.D., Salinger, M.J., and Palmer,
0 J. 1997. Sea-level pressure variability around Antarctica since AD 1750 inferred from subantarctic
tree-ring records. Clim. Dyn., 13: 375–90.
Voeikov, A. 1895. Die Schneedecke in “paaren” and “unpaaren” Wintern. Met. Zeit., 12: 77.
von Storch, H., van Loon, H., and Kiladis, G.N. 1988. The Southern Oscillation. VIII. Model sensi-
11 tivity to SST anomalies in the tropical and subtropical regions of the South Pacific Convergence
Zone, J. Climate, 1 (3): 325–31.
von Storch, H., Weese, U., and Xu, J.-S. 1990. Simultaneous analysis of space–time variability:
principal oscillation patterns and principal interaction patterns with applications to the Southern
Oscillation. Zeit. Met., 40: 99–103.
Walker, G.T. 1909. Correlations in seasonal variations of climate. Mem. Indian Met. Dept, 20 (6):
117–24.
0 Walker, G.T. 1923. Correlation in seasonal variations of weather. VIII. A preliminary study of
world weather. Mem. Indian Met. Dept, 24: 75–131.
Walker, G.T. 1924. Correlation in seasonal variations of weather. IX. A further study of world
weather. Mem. Indian Met. Dept, 24 (9): 275–332.
Walker, G.T. and Bliss, E.W. 1932. World weather. V. Mem. Roy. Met. Soc. (London), 4: 53–84.
Wallace, J.M. and Blackmon, M.L. 1983. Observations of low-frequency atmospheric variability.
In: B.J. Hoskins and R.P. Pearce, eds, Large-Scale Dynamical Processes in the Atmosphere,
Academic Press, London, pp. 55–94.
Wallace, J.M. and Gutzler, D.S. 1981. Teleconnections in the geopotential height field during the
northern hemisphere winter. Mon. Wea. Rev., 109 (4): 784–812.
Wallace, J.M., Rasmusson, E.M., Mitchell, T.P., Kousky, V.E., Sarachik, E.S., and von Storch, H.
0 1998. On the structure and evolution of ENSO-related climate variability in the tropical Pacific:
lessons from TOGA. J. Geophys. Res., 103 (C7): 14241–59.
Wallace, J.M., Zhang, Y., and Bajuk, L. 1996. Interpretation of interdecadal trends in northern
hemispheric surface air temperature. J. Climate, 9 (2) 249–59.
Wallace, J.M., Zhang, Y. and Renwick, J.A. 1995. Dynamic contribution to hemispheric mean
temperature trends. Science, 270: 780–3.
Wang, B. 1993. On the annual cycle in the equatorial Pacific cold tongue. In: D.-Zh. Ye et al.,
eds, Climate Variability, China Meteorological Press, Beijing, pp. 114–36.
Wang, B. 1995. Interdecadal change of El Niño onset in the last four decades. J. Climate, 8 (2): 267–85.
Wang, B. and Fang, Y. 1996. Chaotic oscillations of tropical climate: a dynamic system theory for
ENSO, J. Atmos. Sci., 53: 2768–802.
0 Wang, B., Barcilon, A., and Fang, Z. 1999. Stochastic dynamics of El Niño–Southern Oscillation.
J. Atmos. Sci., 56: 5–23.
Weare, B.C. and Nasstrom, J.S. 1982. Examples of extended empirical orthogonal function analysis.
Mon. Wea. Rev., 110 (6): 481–5.
Webster, P.J. 1983. Large-scale structure of the tropical atmosphere. In: B.J. Hoskins and
R.P. Pearce, eds, Large-scale Dynamical Processes in the Atmosphere, Academic Press, London,
pp. 235–75.
Webster, P.J. 1995. The annual cycle and the predictability of the tropical coupled ocean–
atmosphere system. Met. Atmos. Phys., 56: 33–56.
Webster, P.J. and Holton, J.R. 1982. Cross-equatorial response to middle-latitude forcing with a
latitudinally and zonally nonuniform basic state. J. Atmos. Sci., 41: 1187–201.
0 Webster, P.J. and Yang, S. 1992. Monsoon and ENSO: selectively interactive systems. Quart. J.
Roy. Met. Soc., 118: 877–926.
Weiss, J.P. and Weiss, J.B. 1999. Quantifying persistence in El Niño. J. Atmos. Sci., 56 (16):
11 2737–60.
White, W.B. and Cayan, D.R. 1998. Quasi-periodicity and global symmetries in interdecadal upper
ocean temperature variability. J. Geophys. Res., 103 (C10): 22335–54.
White, W.B. and Peterson, R.G. 1996. An Antarctic circumpolar wave in surface pressure, wind,
temperature and sea ice extent. Nature, 380 (6576): 699–702.
White, W.B. and Tai, C.-K. 1992. Reflection of interannual Rossby waves at the maritime western
boundary of the tropic Pacific. J. Geophys. Res., 97 (C9): 14305–22.
Wolter, K. and Timlin, M.S. 1993. Monitoring ENSO in COADS with a seasonally adjusted prin-
cipal component index. Proc. Seventeenth Climate Diagnostics Workshop, Norman OK, CIMMS
and the School of Meteorology, University of Oklahoma, pp. 52–7.
Wolter, K. 1987. The Southern Oscillation in surface circulation and climate over the tropical
Atlantic, eastern Pacific, and Indian oceans as captured by cluster analysis. J. Clim. Appl. Met.,
26: 540–58.
Wright, P.B. 1984. Relationships between the indices of the Southern Oscillation. Mon. Wea. Rev.,
112 (9): 1913–19.
Wright, P.B. 1985. The Southern Oscillation: an ocean–atmosphere feedback system? Bull. Amer.
Met. Soc., 66: 398–412.
Wright, P.B. 1986. Precursors of the Southern Oscillation. J. Climate, 6 (1): 17–30.
Wright, P.B., Wallace, J.M., Mitchell, T.P., and Deser, C. 1988. Correlation structure of the El
Niño/Southern Oscillation phenomenon. J. Climate, 1 (6): 609–25.
Wyrtki, K. 1975. El Niño – the dynamic response of the equatorial Pacific to atmospheric forcing.
J. Phys. Oceanog., 5: 572–84.
Wyrtki, K. 1985. Water displacements in the Pacific and the genesis of El Niño cycles. J. Geophys.
Res., 90: 7129–32.
Xie, S.-P. and Tanimoto, Y. 1998. A pan-Atlantic decadal climate oscillation. Geophys. Res. Lett.,
25 (12): 2185–8.
Xu, J.-S. and von Storch, H. 1990. Predicting the state of Southern Oscillation using Principal
Oscillation Pattern analysis. J. Climate, 3 (12): 1316–29.
Yarnal, B. and Kiladis, G. 1988. Tropical teleconnections associated with ENSO events. Prog. Phys.
Geog., 9: 524–58.
Yasunari, T. 1989. A possible link of the QBOs between the stratosphere, troposphere and the
surface temperature in the tropics. J. Met. Soc. Japan, 67 (3): 483–93.
Yasunari, T. 1991. The monsoon year – a new concept of the climate year in the tropics. Bull.
Amer. Met. Soc., 72 (9): 1331–8.
Zebiak, S.E. 1989. On the 30–60 day oscillation and the prediction of El Niño. J. Climate, 2: 1381–7.
Zebiak, S.E. 1993. Air–sea interaction in the equatorial Atlantic zone. J. Climate, 6: 1567–86.
Zebiak, S.E. and Cane, M.A. 1987. A model El Niño–Southern Oscillation. Mon. Wea. Rev., 115:
2262–78.
Zhang, C.-D. 1993. Large-scale variability of atmospheric deep convection in relation to sea surface
temperatures in the tropics. J. Climate, 6 (10): 1898–913.
Zhang, R.-H. and Levitus, S. 1997. Interannual variability of the coupled tropical Pacific Ocean–
atmosphere system associated with the El Niño–Southern Oscillation. J. Climate, 10 (6): 1312–30.
Zhang, X., Corte-Real, J., and Wang, X.L. 1997. Low-frequency oscillations in the northern hemi-
sphere. Theor. Appl. Climatol., 57: 125–33.
Zhang, X.-B., Sheng, S., and Shabbar, A. 1998. Modes of interannual and interdecadal variability
of Pacific SST. J. Climate, 11 (10): 2556–69.
Zhang, Y., Norris, J.R., and Wallace, J.M. 1998. Seasonality of large-scale atmosphere–ocean inter-
action over the North Pacific. J. Climate, 11 (10): 2473–90.
Zhang, Y., Wallace, J.M., and Battisti, D.S. 1997. ENSO-like interdecadal variability: 1900–93.
J. Climate, 10 (5): 1004–20.
Zillman, J.W. and Johnson, D.R. 1985. Thermally forced mean mass circulations in the southern
Zimmerman, P.H., Selkirk, H.B., and Newall, R.E. 1988. The relationship between large-scale
vertical motion, highly reflective cloud, and sea surface temperature in the tropical Pacific region.
J. Geophys. Res., 93 (D9): 11205–15.
11
Part 3
Synoptic climatology
11
11
1
11
6 Synoptic systems
11 Synoptic systems have a horizontal scale of about 1,000–2,000 km and a lifetime of five
to seven days. In mid-latitudes a weak low (high) pressure system typically has a central
MSL pressure of less (greater) than about 1,008 (1,016) mb. In low latitudes, wave distur-
bances are more usually identified in streamline patterns because of the weak pressure
gradients and the semi-diurnal pressure oscillation, except in the case of tropical cyclones.
In view of their importance to mariners, identification and tracking of storm systems over
0 the tropical oceans began in the mid-nineteenth century. W.C. Redfield (1831), for
example, traced the parabolic paths of hurricanes over the western North Atlantic–Gulf
of Mexico and, about the same time, Henry Piddington (1842) documented the motion
of tropical storms in the Bay of Bengal–Arabian Sea. He also introduced the term cyclone
(from the Greek kyklon, meaning revolving). Its counterpart, anticyclone, was first used
by Sir Francis Galton in 1863 (Khrgian, 1970).
6.1 Early studies of extratropical systems

The nature and causes of storms became a major scientific controversy in the 1830s–1840s
0 in North America and Great Britain, prompting efforts to compile meteorological reports
and to systematize observations. A crucial step was the development of telegraph networks
in the 1840s. Reviews and assessments of these early observations and theories are
provided by Khrgian (1970), Kutzbach (1979), and Fleming (1990). Monmonier (1999)
presents a fascinating history of the early meteorological charts prepared for Europe for
the year 1783 by H.W. Brandes (1820) and of storm events over the eastern United States
in the 1830s–1840s analyzed by Espy (1841), Redfield (1843), and Loomis (1846).
Identification of the role of air currents with differing properties in mid-latitude weather
systems became possible when synoptic weather maps were routinely prepared in the
1850s–1860s. Sir Francis Galton (1863) analyzed daily streamlines and rain areas, as well
0 as the distributions of temperature and pressure over Europe for the month of December
1861, but his unique pioneering effort was not followed up. The “meteorograms” produced
by seven British observatories for 1869–80, showing daily time traces of dry and wet
bulb temperatures, vapor pressure, air pressure, wind speed and direction, and rainfall in
orthogonal coordinates on a single 5-day chart (see Bergeron, 1980, figure 2, for example)
were also badly underutilized, in part owing to the lack of understanding of their value.
In the view of Bergeron (1980), emphasis on the study of the relation of weather to
surface isobaric patterns by R. Abercromby (1878), Julius von Hann (1901), and others
delayed the analysis of air trajectories and the recognition of fronts for up to fifty years.
The characteristics and life cycle of frontal cyclones were not described until 1919 by
0 the ‘Bergen school’ of meteorologists in Norway (Bjerknes, 1919; Friedman, 1989, pp.
150–201). Their model, discussed below, remained little modified until the 1950s (Namias,
1983). The availability of upper air balloon soundings, aircraft observations, and radar
11 measurements, followed later by data from satellites, Doppler radar, and boundary-layer
1 profilers has greatly extended our three-dimensional view of mid-latitude cyclones.
Anticyclones have received much less attention, primarily because they present fewer
forecasting problems.
6.2 Climatology of cyclones and anticyclones

There is a long history of studies on the characteristics of synoptic systems, beginning
with classical work on mid-latitude cyclones (Mohn, 1870; Loomis, 1874) and anticy-
clones (Loomis, 1887; Russell, 1893; Rawson, 1908, 1909). A comprehensive view was
presented for the northern hemisphere by Petterssen (1950), using manual analyses of the
frequencies of cyclone/anticyclone centers, as well as of cyclogenesis/anticyclogenesis and
rates of alternation between high and low-pressure values. Such information is an essen-
tial adjunct to the interpretation of mean pressure or height fields (Klein, 1958). It is well
known that climatological low-pressure systems tend to represent areas through which
deep lows frequently move whereas high pressure cells tend to be quasi-stationary or
slow-moving within the corresponding climatological anticyclone centers. The first detailed
investigations of cyclone/anticyclone centers were based on the 1899–1939 Historical
Weather Maps of the US Weather Bureau, either in their entirety (Petterssen, 1950) or
selectively for 1909–14 and 1924–37 by Klein (1957). However, these charts are known
to be unreliable in high latitudes until the 1950s, owing to the paucity of stations, and
there are gaps in the station network in other areas. As longer time series and more
reliable pressure and geopotential height fields have become available in the form of
gridded values, the earlier analyses have been updated and improved upon. Algorithms
to identify cyclones and anticyclones based on pressure or height data are described by
Murray and Simmonds (1991) and Serreze et al. (1993), for example. The algorithms
permit searches to be made for blocks of grid points to detect local pressure
minima/maxima. Jones and Simmonds (1993, 1994) also introduce tests of curvature to
avoid the inclusion of weak systems. The delimitation of anticyclone centers is commonly
ambiguous because of slack pressure gradients and the tendency for weak maxima, that
may shift irregularly over time, located within the highest closed isobar. System centers
are tracked in space and time in order to obtain data on genesis/lysis and rates of cyclone
deepening/filling. An assessment of the relative performance of three automated proce-
dures for identifying cyclones and determining their tracks shows that the scheme of
Murray and Simmonds (1991) identifies the largest number of systems and of tracks
(Leonard et al., 1999). An alternative procedure, proposed by Sinclair (1994, 1996, 1997)
is to calculate geostrophic relative vorticity, although this is better suited to cyclones than
anticyclones, as the latter have light winds and a wide separation between the loci of
pressure maxima and relative anticyclonic vorticity maxima.
For the southern hemisphere, analyses of fifteen years of pressure data provide exten-
sive statistics of the climatology of synoptic systems (Jones and Simmonds, 1993, 1994),
updating the early work of Taljaard (1967). Figure 6.1 compares the zonally averaged
behavior of mean sea-level pressure, anticyclone system density, and anticyclone
mean central pressure on a seasonal basis. Interestingly, the anticyclone mean central
pressure maximum at 38°S in JJA and 44°S in DJF is located 8°–10° south of the mean
subtropical ridge (STR) of high pressure and 5°–8° south of the maximum zone of
system density. The strongest anticyclones occur poleward of both the ridge and the
maximum zone of system density. The poleward side of the STR is affected by frequent
and intense lows, whereas the equatorward side generally has undisturbed flow. Anti-
cyclones are most numerous over the eastern subtropical oceans, with fewer over the
southern land areas (excluding Antarctica). Genesis occurs over the southwestern Atlantic
and Indian Oceans and over the Australian Bight and the Tasman Sea. Systems gener-
ally move eastward and somewhat equatorward, decaying near the oceanic centers of the
time-mean anticyclones. There is a bifurcation in the distribution pattern of system density
Synoptic systems 443
11
11
Figure 6.1 The zonally averaged distribution with latitude of highest mean sea-level pressure
0 (MSLP), anticyclone system density (SD), and anticyclone mean central pressure (MCP)
in the southern hemisphere. (Jones and Simmonds, 1994)
in winter from east of Tasmania to 150°W; in this sector there are maxima around 30°S
and 45°S.
Southern hemisphere cyclones as analyzed from weather maps for 1975–89 are char-
acterized by a year-round frequency maximum in the circumpolar trough between about
60°S and 70°S (Jones and Simmonds, 1993). During winter and the transition seasons
this maximum is fed by two branches spiralling towards it; one originates in the Tasman
0 Sea and the other in the South American sector (Figure 6.2). However, when centers
are computed by identifying local minima from 1,000 mb-level geostrophic vorticity, a
different picture emerges. Sinclair (1994) uses ECMWF data for 1980–86 for this purpose,
avoiding the bias towards slower/deeper systems in the traditional approach. This analysis
thereby takes account of mobile vorticity centers in middle latitudes. These are fairly
uniformly distributed and give rise to a primary eastward track between 45°S and
55°S, which also includes heat lows and lee troughs over or near the three land masses.
A further maximum lies off East Antarctica, where there are well known katabatic out-
flows. There is a secondary maximum in winter–spring associated with the subtropical
jetstream near 40°S east of New Zealand. Intense cyclones occur near New Zealand, east
0 of South America and in the southern Indian Ocean in winter.
Extension of the vorticity analysis to cyclogenesis shows that cyclones typically form
in preferred areas in middle latitudes – near the jetstream baroclinic zones and to the east
of the southern Andes year-round, as well as off the east coasts of Australia and South
America in winter. Systems forming over the oceans intensify over strong gradients of
sea surface temperature. Rapid cyclogenesis is particularly concentrated east of South
America, southeast of South Africa, south of Australia, and near New Zealand.
For the northern hemisphere, a twenty-year (1958–77) climatology of cyclones based
on more consistent data than the earlier studies is available. Whittaker and Horn (1984)
performed manual frequency counts over 5° latitude–longitude boxes for mid-season
0 months using surface pressure charts. Figure 6.3a, b presents their maps of cyclone
frequency and cyclogenesis. Whereas Klein (1957) makes counts at specific times,
Whittaker and Horn tabulate a system only once in a given box. The principal findings
11 of the analysis are as follows:
1
Figure 6.2 (a) Cyclone system density, 1975–89 in the southern hemisphere. (b) Density of cyclo-
genesis for winter. (From Jones and Simmonds, 1993)
11
11
0 Figure 6.3 (a) Cyclone frequency for 5° latitude and longitude boxes and (b) frequency of cyclo-
genesis in the northern hemisphere (corrected for latitudinal-scale change) for January
and July (1958–77). (From Whittaker and Horn, 1984)
11
1 1 In January the primary maxima are in the western North Atlantic, with an extension
westward towards the Great Lakes, a broad zone in the western and central North
Pacific, with a peak about 45°–50°N extending into the Gulf of Alaska, where there
is a secondary peak, and a subsidiary maximum over the north-central Mediterranean.
2 In April the pattern is similar but with a decrease in the frequency of centers and a
northward shift over the North Pacific, where the activity now begins to the west
over China along 50°N.
3 In July the frequencies are further reduced and the hemispheric maximum is over
eastern Canada at 55°N. The other main focus is from the western Pacific to the
western Aleutian Islands, with a subsidiary area over China.
4 The October pattern resembles winter, except that the Atlantic maximum is off south-
east Greenland, the main Pacific center is in the Gulf of Alaska, and there is little
activity in the Mediterranean.
Trends in cyclone frequencies in the northern hemisphere for 1958–97 are examined
at the 1,000 and 500 mb levels by Key and Chan (1999), using the NCEP reanalyses.
They show that, for 60°–90°N, closed lows increased in frequency at 1,000 mb in all
seasons, but they decreased in frequency at 500 mb except in winter. In mid-latitudes,
the frequency of lows decreases at 1,000 mb, with increases at 500 mb, except in winter.
For 0°–30°N, lows became more frequent at both levels in winter and spring and at 500
mb only in summer and autumn. Agee (1991) used three previous analyses of cyclone
and anticyclone frequency to examine trends in relation to intervals of warming and
cooling in the northern hemisphere. Based on the works of Parker et al. (1989) on annual
500 mb cyclone and anticyclone frequency over the western hemisphere for 1950–85,
Zishka and Smith (1980) on surface cyclone and anticyclone frequency over North
America and adjacent oceans for January and July 1950–77, and Hosler and Gamage
(1956) on surface cyclones in the United States for 1905–54, Agee suggested that warming
(cooling) periods are accompanied by increases (decreases) in frequencies of both cyclones
and anticyclones. The findings of Key and Chan (1999) indicate greater complexity. They
also found no significant difference in cyclone frequencies between El Niño and La Niña
years. In both North America and Europe cyclone frequencies are poorly correlated with
the NAO.
Prior to the availability of pressure data from the Arctic drifting buoy program, which
began in 1979, information on pressure systems over the Arctic Ocean was limited to the
one or two manned drifting stations. These operated continuously from 1950 to 1991
under the North Pole Drifting Station Program of the Soviet Union, supplemented by a
small number of US stations mainly in the 1950s and 1960s. Consequently the statistics
based on pre-1979 data must be treated with caution (Jones, 1987; Serreze and Barry,
1988). The spatial distribution of systems identified by Serreze et al. (1993; Serreze, 1995)
during 1973–92, shows that in winter months the cyclone maximum near Iceland extends
northeastward into the Norwegian–Barents Sea (Figure 6.4). In the summer half-year this
tendency is almost absent. In winter the rate of cyclone deepening and the frequency of
deepening events peak in the area of the Icelandic low, southwest of Iceland, with a sepa-
rate maximum in the Norwegian Sea (Serreze et al., 1997). Cyclogenesis is common in
these areas, as well as in northern Baffin Bay. Deepening rates are up to 6.8 mb (12
h1) for the Greenland Sea–North Atlantic sector. The combined effects of ice-edge baro-
clinicity, orographic forcing, and rapid boundary layer modification in off-ice airflows are
probably involved. Additionally, these same locations show high frequencies of systems
filling and cyclolysis, implying that this sector of the Arctic is a dynamically active one
with alternating regimes.
In summer, high-latitude cyclones are considerably weaker. There is a frequency
maximum over the central Arctic Ocean but it is characterized by cyclolysis, as are
Baffin Bay and Davis Strait, the Icelandic Low area and the Norwegian Sea. In summer,
0
0
0
0
0
11
11
11
Figure 6.4 Cyclone events over the northern hemisphere north of 30°N for 1966–93. Systems lasting twenty-four hours are counted once in a 1,200 km
(1,250 km north of 60°N) search area and counts are adjusted to a 60° reference latitude. Contour interval is 100: areas with over 300 are
stippled. (Serreze et al., 1997)
1
Figure 6.5 Estimates from a Q-vector analysis of the contributions to 850 mb vertical motion of
differential vorticity and thickness advection and diabatic heating in a persistent low-
pressure system over the Arctic Ocean during August–September 1990. (From Le Drew,
et al., 1991)
deepening rates in the Greenland Sea are only 3.3 mb (12 h1), although this area
remains the leading sector. Cyclogenesis in summer is common over central and eastern
Eurasia and northwest Canada. The low over the Arctic in summer is a deep, cold-cored
barotropic structure that is continually regenerated by thickness and vorticity advection
associated with mature systems moving from northern Siberia via the Kara and Laptev
seas. LeDrew (1988, 1989) uses the Q-vector formulation of the omega equation (Hoskins
et al., 1978) to examine the contributions of differential advection (of vorticity and thick-
ness combined) and diabatic heating to vertical velocity in five depression systems within
the Polar Basin in June, July, and September 1979. A Q vector represents the rate of
change of potential temperature gradient in the direction of low-level ageostrophic flow
and towards ascending air, assuming frictionless adiabatic motion (see Appendix 6.1).
Q-vector convergence denotes cyclonic vorticity generation. Advection of cold (warm)
air with the Q vectors implies frontogenesis (frontolysis). LeDrew finds much weaker
vertical circulation at 850 mb within the Polar Basin than in North Atlantic cyclones.
Advective processes contributed 38–53 percent of total 850 mb vertical circulation, latent
heat 12–59 percent, sensible heat 28 percent to 51 percent and friction 4 percent to
42 percent. The ratio of diabatic heating to advective effects averaged 0.8 for trajectories
over the central basin, 1.2 over the Chukchi–Beaufort seas and 1.4 over the Barents–Laptev
seas. Further analysis by LeDrew et al. (1991) of the persistent low over the Canada
Basin from August 15 to September 11 1988 (Figure 6.5) shows a mean vertical velocity
of 0.8 cm s1 at 850 mb (based on ten days through the twenty-seven-day interval).
Diabatic heating contributed 62 percent, advection 29 percent and friction 9 percent of
the total vertical velocity. Surface heat fluxes associated with areas of more open ice
cover – forced in part by the cyclonic activity (Serreze et al., 1990) – may provide a
11
11
Figure 6.6 A schematic model of atmosphere–surface coupling with cold low development over
the Beaufort Sea in late summer. (From LeDrew and Barber, 1994)
0
feedback to help maintain the cyclone. However, LeDrew et al. note that the calculations
do not permit a firm conclusion to be reached. Ageostrophic cold air advection also is
important. The cyclone initiation in mid-August may be related to cooling in the strato-
sphere and the circulation transition to a wintertime barotropic westerly vortex (Figure
6.6).
Anticyclogenesis has received less attention except for that occurring in polar air (Curry,
1987) and blocking events. A study for the northern hemisphere identified 1,250 events
during 1984 (Colucci and Davenport, 1987). Cases were defined by a twenty-four-hour
0 MSL pressure change averaging 2.7 mb with a closed isobar (8 mb spacing) appearing
on two successive 12.00 UTC charts. For the western hemisphere, anticyclogenesis is
concentrated over Alaska–western Canada, associated with cold air outbreaks, and over
southeastern Canada, where cold anticyclones that have moved southeastward over snow-
covered areas re-intensify. Zishka and Smith (1980), however, identify an area of winter
anticyclogenesis over western Texas, Oklahoma, and Kansas in response to cold air advec-
tion in the rear of Colorado lee cyclones. These highs are usually shallow mobile systems.
In summer a similar process operates farther north, over southern Alberta.
Boyle and Bosart (1983) examine the transformation of a polar anticyclone over Alaska
into a warm dynamic system off the east coast of the United States over a seven-day
0 period in November 1969. The system is initially confined to the layer below 850 mb,
although vertical motion associated with the anticyclogenesis extends through the tropo-
sphere. The system first moves southward towards the Gulf coast and then recurves
11 northeastward. In the first stages, upper-level vorticity advection and cold air advection
1 lead to subsidence downstream over the anticyclone. The anticyclone moves southward
towards the region of maximum descent, forced by the cold advection. It is supported to
the west by a jet streak maintaining a thermal gradient. In the later stages, warm advec-
tion in the lower middle troposphere west of the anticyclone forces an upper-level ridge.
6.3 Development of cyclones

6.3.1 Historical background
Three ideas on cyclone development were current in the nineteenth century (Kutzbach,
1979; Bergeron, 1980; Namias, 1983). The American meteorologist J.P. Espy proposed
that low-pressure centers were driven by rising warm air associated with latent heat release
and sustained by lateral outflow aloft, a situation which resembles tropical storms. Julius
von Hann objected to this “conventional hypothesis” on the grounds that mountain
observatories showed the air to be warmer above anticyclones than above cyclones, as
was later confirmed by W.H. Dines and others. Von Hann suggested that cyclones
and anticyclones derive their energy as eddies feeding off the zonal westerly current (c.f.
L.F. Richardson’s dictum, p. 140). A third group, represented by H.W. Dove, R. FitzRoy,
W. Blasius, and H. von Helmholtz, recognized the necessity for the juxtaposition of
contrasting air masses in the formation of weather systems. Subsequently Napier Shaw
and his associates identified the complementary spiral motions of air trajectories in
cyclones and anticyclones. These last two concepts were incorporated in the frontal cyclone
model of the Bergen school (Bjerknes, 1920; Bjerknes and Solberg, 1922), building on
the dynamic concepts of Vilhelm Bjerknes and the three-dimensional synoptic analyses
performed at Leipzig, 1913–17 (Friedman, 1989, p. 88; Eliassen, 1994). Using a dense
station network in Norway and inferred upper air conditions (based on cloud motion and
hydrometeors), the Bergen meteorologists showed that cloud and rain bands were asso-
ciated with lines of wind convergence and also that temperature (or density) differences
between air masses are concentrated at frontal discontinuities.
The polar front theory of the life cycle of extratropical cyclones formulated by the
Bergen school dominated synoptic meteorology from 1920 until the 1950s, although its
formal adoption in the United States was slow (Namias, 1983; Newton and Newton, 1994).
The principal idea of the theory is that cyclones form, mature, and decay along the polar
front. The frontal boundary develops a small wave and incipient cyclonic circulation.
Because the leading edge of the cold air in the rear of the system moves faster than the
air in the warm sector, the cold front catches up with the warm front. The air in the warm
sector is then lifted off the surface, forming an occlusion – a process identified in 1919
by Tor Bergeron (1980; Friedman, 1989, pp. 212–23). The dynamical mechanism involves
the conversion of the potential energy stored in the atmospheric thermal gradients into
kinetic energy. When occlusion occurs, the cyclone weakens as the low-pressure center
becomes detached from the warmest air. Thus cyclonic shear across an initially quasi-
stationary front leads to unstable growth of the incipient pressure perturbation, with the
familiar frontal wave sequence evolving over a period of three to five days.
As upper air observations became more numerous the important role of divergence in
the upper troposphere on cyclogenesis was recognized by Scherhag (1934), Sutcliffe
(1939), and Bjerknes and Holmboe (1944). The fall of pressure in a deepening surface
low is made possible because air ascends in the low and diverges aloft. Such upper-
level divergence is typically located on the eastern limb of an upper trough in the west-
erlies, and so cyclogenesis is favored in that sector. The spatial pattern of convergence/
divergence in the upper troposphere is also strongly coupled with the distribution and
structure of jetstreams (see section 4.3).
Major theoretical advances occurred in the 1940s along two different lines of thought.
We note first the concept of baroclinic instability, formulated by Charney (1947) and
11
11
Figure 6.7 Baroclinic instability model. (From Barry, 1967)
0
Eady (1949). It envisages that cyclones may form as a result of the breakdown of a zonal
jetstream when the horizontal temperature gradient or the vertical wind shear reaches
certain critical limits. Small perturbations can grow at the expense of potential energy in
the mean motion when warm (cold) air undergoes a small upward (downward) displace-
ment in air moving poleward (equatorward) ahead of (behind) an upper trough in the
westerlies. Figure 6.7 illustrates this concept for (a) short and (b) medium-wavelength
disturbances. In (a) the slope of the air motion exceeds that of the isentropic () surfaces
and, since the rising air has lower than the subsiding air, the potential energy is increased
and there is no wave growth. In medium wavelength systems (b), the potential energy is
0 reduced by air of lower (higher) sinking (rising) and moving equatorward (poleward).
The north–south slope of the air motion, as determined by Fleagle (1957, 1960), is:
( f /2HS2)
=
1 ( f 2 /2H 2S2)
where the slope of the isentropic () surfaces, (
f/
y), H the depth of the
atmosphere (10 km), S static stability, and the wave number ( 2/L where L
wavelength).
The maximum conversion of potential to kinetic energy takes place where /2
and L 2H(2S/f )1/2. The maximum isentropic slopes are in mid-latitudes and, since
0 /f increases equatorward, causing → , maximum baroclinic instability tends to occur
climatologically in higher mid-latitudes. Theoretically there is a critical latitude, with baro-
clinic stability tending to prevail equatorward of 35°. Despite the simplicity of the Charney
and Eady models, they demonstrated that synoptic disturbances with a horizontal scale
of a few thousand kilometers can amplify with a doubling time of about twenty-four hours
and propagate eastwards.
The second line of advance in the 1940s was provided by the analysis of development
in cyclones and anticyclones by Sutcliffe (1939). He proposed that surface development
be diagnosed in terms of the difference in divergence/convergence within an air column,
and he showed that development (D) occurs with ascending air and low-level stretching
0 creating cyclonic vorticity (Hoskins, 1994):
∂2
D = ·VU ·VL = p
11 ∂p2
1 where VU and VL refer to the upper and lower-level wind velocities, and is the vertical
motion in pressure units. Subsequently, Sutcliffe linked the horizontal divergence with
vorticity considerations, following the ideas of Rossby et al. (1939). In particular, vertical
motion is linked with the advection of vorticity by the thermal wind. Sutcliffe (1947)
related development to the imbalance between changes in horizontal thermal gradients
and the vertical difference in vorticity caused by differential advection with height.
Thermal advection was analyzed via thickness charts (Sutcliffe and Forsdyke, 1950):

∂
∂t
1
冤冢
z zL
[u L] = 2 u u
f x
v u冣冢
z zL
y 冣冥
where zu zL denotes the difference in geopotential between an upper and lower isobaric
level, or thickness; u– and v– are layer-average wind components and 2 is the Laplacian
operator.
The development theory (Sutcliffe, 1947) can be related to an approximate form of the
omega equation for vertical motion, expressed in terms of vorticity and thermal advec-
tion (Hoskins et al., 1978), while Sutcliffe’s earlier work (1939) is consistent with the
Q-vector formulation (Hoskins, 1994).
6.3.2 Modern views

The traditional view of the frontal cyclone has been gradually modified. Ideas on the
nature of occlusions have also undergone substantial change. Cold fronts propagate faster
than warm ones, suggesting that one front overruns the other, but few studies of this have
been made. Mass (1991) points out that there appears to be “a lack of consistent and
well-defined procedures for defining fronts and for analysing surface synoptic charts.”
Figure 6.8 illustrates the variety of analyses by specialist participants in a workshop on
surface analysis (Uccellini et al., 1992). Analysts disagree on the type, location and even
the existence of fronts. Moreover, analyzed fronts are not necessarily located where there
are strong temperature gradients. Problems occur in the charting of shallow zones of
temperature contrast that are related to topography (cold air damming and lee troughs)
or to discontinuities in the surface type (coastal fronts; snow and ice boundaries on land
or in the ocean) and difficulties also arise from the modification of fronts by mountain
barriers (Williams et al., 1992).
The Bjerknes model was developed for North Atlantic systems. However, Petterssen
and Smeybe (1971) showed that not all cyclones develop as frontal waves within a baro-
clinic zone. These Type A storms are prevalent over the North Atlantic Ocean and they
evolve towards a classical occlusion (Petterssen et al., 1962). A second group (Type B)
develops over North America, east of the Rocky Mountains, when a pre-existing upper
trough advances over a zone of low-level warm advection that is weakly baroclinic.
Initially there is strong upper-level vorticity advection on the forward side of the trough;
this decreases as the system develops, while thermal advection increases when the storm
transports cold air southward, enhancing the temperature gradient (Petterssen and Smeybe,
1971). Work by Locatelli et al. (1995) and Hobbs et al. (1996) provides a schematic
model of cold-season frontogenesis east of the Rocky Mountains. The scheme involves
an outflow of arctic air east of the mountains forming an arctic front and a dry lee trough,
associated with dry air with low equivalent potential temperature (e ) crossing the moun-
tains and overriding warm, moist, high e air from the Gulf of Mexico. Instability in the
trough generates a rain band analogous to a warm front. The arctic air moves southward
west of the trough or low center, causing lifting of warmer air but little precipitation. The
flow of low e air may represent an upper cold front advancing eastward over the trough
and setting up a rain band along its leading edge, ahead of the trough. Type B systems
are now thought to be widespread in their occurrence. Some polar lows resemble the
Type B cyclones.
11
11
Figure 6.8 Surface data analysis over North America for 21.00 UTC, February 13 1991, illus-
trating the variability of subjective analyses. The solid lines are fronts analyzed by
participants in a professional workshop; troughs and squall lines are dashed. Heavy
lines are the fronts or troughs from the National Meteorological Center analysis. Light
0
lines are objectively analyzed isobars (4 mb). (From Uccellini et al., 1992)
Modifications of the Bergen model of frontal cyclone analysis since 1945 have followed
two primary directions. The first approach considers the synoptic analysis of frontal char-
acteristics and the second treats the dynamical relationships involved in frontogenesis and
frontolysis. For example, three-dimensional frontal analysis, involving the plotting of
contours of frontal location at different isobaric levels (frontal contour analysis), by
Canadian synopticians (Crocker, 1949; see also Palmén, 1951) and the associated three-
front (four air-mass) model (Penner, 1955; Galloway, 1958) took account of the strong
0 latitudinal temperature gradient in North America, especially in winter. This gradient
commonly gives rise to cyclones developing simultaneously along the arctic and polar
frontal zones with an associated double jetstream structure (McIntyre, 1958). A valuable
aspect of this scheme was the recognition of the trough of warm air aloft (trowal) (Morris,
1972). However, the potential drawbacks of a rather rigid framework for analysis were
also recognized (Longley, 1959). The complexity of frontal weather patterns, attributable
in part to the vertical motion of air relative to a front, was identified in the 1930s by
Bergeron (1937), but the terminology of anafronts (katafronts) with air ascending
(descending) over the frontal surface proposed by Bergeron came into widespread
use only in the 1950s and 1960s with the application of Doppler radar to detect such
0 motions. This technology allowed conveyor belts (see Figure 6.9) to be mapped and incor-
porated into frontal models (Browning, 1990, 1994; Carlson, 1991, chapter 12). They are
typically a few hundred kilometers wide and about a kilometer deep. The associated “slant-
11 wise” ascent/descent gives rise to characteristic distributions of cloud and precipitation
1
Figure 6.9 Conveyor belts and precipitation in a developing frontal cyclone. The following features
are represented. (a) Sea level pressure and fronts. The unshaded triangles indicate a
cold front aloft. Between the two well marked cold fronts is a diffuse surface cold
front indicated by a single triangle. The low is moving northeastward (arrow). (b) Cloud
(stippled), showing a cloud head in the west and the polar front cloud band in the east.
Precipitation is shown as follows. Solid lines in the cloud head, cross-hatched where
convective (moderate–heavy); broken lines show warm conveyor belt (light) precipita-
tion; broken cross-hatching: patchy, moderate mid-level convective precipitation; solid
shading: narrow cold front bands of heavy rain. (c) Conveyor belts: W1, main warm
conveyor belt (WCB). W2, lower part of WCB separates and rises in the upper cloud
head. CCB, cold conveyor belt; diffluent flow giving cloud head precipitation. J, upper-
level jet core; W1 and CCB westward flow is an ageostrophic circulation at the jet
exit. (d) Dry intrusion. Dry air descends from the upper troposphere upwind and rises
towards the cyclone center, overrunning a shallow moist zone (SMZ in (c)), associated
with W2, and gives rise to the dry slot (panel (b)). (From Browning, 1994)
11
11
0
Figure 6.10 A schematic view of the life cycle of a marine frontal cyclone. (I) Incipient system.
(II) Cold front fracture. (III) Bent-back warm front frontal T bone. (IV) Warm core
seclusion. Upper sequence shows sea-level isobars, fronts, and satellite cloud signa-
ture (stippled). Lower sequence depicts warm (solid) and cold (dashed) air currents.
(From Shapiro and Keyser, 1990)
areas in frontal cyclones and also redistributes heat and moisture over long distances.
More recently the combination of satellite and aircraft data has led to the identification
0 of split cold fronts south of the cyclone center and the back-bent warm front/T-bone
pattern (Shapiro and Keyser, 1990; Shapiro and Grell, 1994). Figure 6.10 illustrates
schematically a revised view of the life cycle of a marine frontal cyclone depicting these
features. The process whereby relatively warm air is left behind in the center of circula-
tion is termed “seclusion” by Shapiro.
Fronts are associated with gradients of temperature, dew point, wind velocity, baro-
metric tendency, and vertical motion. For example, a study of five cold fronts over Brittany,
France, during the winter 1987–88, using rawinsonde and acoustic sound detection and
ranging (SODAR) data, found that across the front there was a mean temperature drop
of 2°C, a reduction of surface wind speed of about 6 m s1 and a 60° rotation in the
0 direction of the wind (Lefloch and Amory-Mazaudier, 1998). A low-level jet averaging
27 m s1 between 600 and 1,000 m was also present over Brittany. Studies of frontal
dynamics in the 1960s analyzed the field properties of such parameters to develop objec-
11 tive, physically meaningful criteria for frontal definition. Potential temperature at 850 mb
1 was selected as a thermal parameter for frontal identification by Renard and Clarke (1965).
The divergence of the gradient of potential temperature is:
2 | | # n | | #n
where n is a unit vector in the direction of . The first term on the right-hand side is
the directional derivative of the magnitude of the gradient of potential temperature along
its gradient. It is proportional to the horizontal shear of the thermal wind. The second
term represents the tangential curvature of the potential isotherms. Renard and Clarke use
the axis of maximum ||, which is coincident with a zero value of the first term, as the
locus of the baroclinic zone. Figure 6.11 illustrates, for a one-dimensional case, the
meaning of the variation of the terms: X, | X| , and | X|#nx , respectively, with nx .
Note that a zero value of the term | X|#nx corresponds to both maximum and minimum
|X| (see Figure 6.11b and c). Maximum |X|, which may be of the order of 1 K km1,
is an index of frontal intensity: the term |X|#nx is of the order of ±0.2 K (100 km)2.
It must be noted that the unit vector in the direction of the gradient is ill defined when
| | is small or changes direction abruptly, as occurs in the conveyor belt zone ahead
of a cold front (Hewson, 1998).
Alternative approaches, based on other thermal characteristics, include that of Huber-
Pock and Kress (1989), who use 850–500 mb thickness instead of 850 mb temperature.
Their technique has been applied by Hoinka and Volkart (1992) and Zwatz-Meisinger
and Mahringer (1990), but the fronts are still manually drawn. Steinacker (1992) uses
850 mb equivalent potential temperature and identifies where |
2/
s2| is maximized, where
s is a tangent to a streamline. Sanders (1999) proposes the use of potential temperature
analysis in areas of complex terrain, in lieu of a surface temperature analysis. This allows
non-frontal baroclinic zones to be distinguished. A front is identified where a wind shift
Figure 6.11 The relationship between

(a) X , (b) |X | and (c) |X |·nx for the
one-dimensional case where parameters vary
with nx. (From Renard and Clarke, 1966)
11
11
Figure 6.12 Above Plan view of idealized contours of a thermodynamic variable in the vicinity of
a straight cold front with no along-front thermal gradient. Arrows represent vector
differentials of , | , and ||||. Below Graphical plot showing the variation of
|
/
x| and its scalar differentials in the cross-front direction. Points on the x axis
denote where
3 /
x3 0. Lines A and B delineate the baroclinic zone, and line B
0 (and the ringed dot) shows the front. The term (m) is a positive fractional number
(m) of grid lengths (). (From Hewson, 1998)
coincides with the warm edge of a baroclinic zone. Sanders also notes that most cold
fronts lack a pronounced temperature contrast and advocates their mapping as baroclinic
troughs.
Hewson (1998) developed a procedure to plot fronts objectively based on the concepts
discussed above. The following elements are used: a line adjacent to a baroclinic zone
across which the magnitude of the thermal gradient changes most abruptly; the rate of
0 change of the thermal gradient across a front exceeds a specified threshold; and the thermal
gradient in the adjacent baroclinic zone also exceeds a threshold value. Figure 6.12 illus-
trates schematically the way in which the scalar differential values of a thermodynamic
11 variable vary across a simple cold front where there is no along-front thermal gradient.
1 The baroclinic zone is defined by the vertical lines A and B in the lower figure and line
B demarcates the front. The Laplacian of , or
2/
x2, has a minimum value at the front
and here also the gradient of the Laplacian, or the function
3/
x3 0. Hewson also
develops additional diagnostics for cases of straight fronts with an along-front thermal
gradient, and for curved fronts which may or may not have an along-front gradient.
A frontal zone can also be identified from the vertical rate of change of geostrophic
relative vorticity (g), which is proportional to the Laplacian (2T) (Kirk, 1966b). For
quasi-geostrophic vorticity:
∂g R
f = 2T
∂p p
where R the gas constant for dry air, f the Coriolis parameter, and
∂2T ∂2T
2T =
∂x2 ∂y2
This Laplacian term is determined by the spacing and curvature of the isobars (Kirk,
1970). A corresponding approach based on the divergence of the gradient of potential
temperature was developed by Clarke and Renard (1966). This term (2) can be broken
down into components that are proportional to the horizontal shear of the thermal wind
and the tangential curvature of the potential isotherms. Creswick (1967) extended their
work using wet-bulb potential temperature. Much recent attention has focused on the use
of potential vorticity analysis as a diagnostic tool (Hoskins et al., 1985); Thorpe (1990)
illustrates its application to frontal cyclone study. Other workers addressed the nature of
front–jet coupling (Bleeker, 1958), and a modern example is given in Figure 6.13 showing
Figure 6.13 Conceptual model showing the ageostrophic circulations associated with low- and
upper-level jets in the vicinity of a cold front. (Sortais et al., 1993)
11 the transverse, indirect, ageostrophic circulation associated with upper and low-level jets
at a cold front (Sortais et al., 1993). Yet a further approach to the dynamics of baroclinic
systems makes use of semi-geostrophic theory (Hoskins, 1971; Hoskins and West, 1979).
6.3.3 Cyclogenesis
The key aspects of cyclogenesis involve: (1) horizontal divergence of air in the upper
troposphere, in excess of low-level convergence, thereby permitting surface pressures to
decrease; (2) the existence of a wavelength of maximum amplification for typical frontal
0 perturbations; (3) a frontal zone of strong baroclinicity and vertical wind shear.
The role of upper-level divergence in cyclonic development was first formulated by M.
Margules and later extended by J. Bjerknes and by R.C. Sutcliffe (Bjerknes and Holmboe,
1944; Palmén and Newton, 1969, p. 134 ff.). Bjerknes’s “pressure-tendency” equation
11 (from the hydrostatic and continuity equations) states that the pressure change with time
can be expressed:
冢冣
∂p
∂t
= g 冕0
∞
H· V dz g 冕
0
∞
V ·H dz gw
0 where the terms on the right-hand side are: first, the integrated horizontal divergence of
the wind velocity (V), second, the advection of density (), and, third, the vertical motion.
In practice, the first term is about an order of magnitude larger than the second one, and
the third term is of similar magnitude to the first but opposite in sign. A surface pres-
sure tendency of ±1 mb hr1 corresponds to an average divergence of ±0.3 106 s1
Figure 6.14 Schematic illustration of the effects of streamline curvature in upper tropospheric flow
on cyclone development in the lower troposphere. (a) Ageostrophic wind components
along the streamline at upper levels cause convergence (divergence) as a result of
deceleration (acceleration) from supergeostrophic flow in the ridges to subgeostrophic
0 flow in the trough. (b) Corresponding maximum (minimum) centers of relative cyclonic
(anticyclonic) vorticity and associated regions of negative, or anticyclonic, vorticity
(NVA) and positive, or cyclonic, vorticity advection (PVA) for an upper level wave.
11 (From Shapiro and Kennedy, 1981; Kocin and Uccellini, 1990)
1 whereas measured values of >105 s1 are typical for individual levels. The calculation
of surface pressure changes from observed winds is generally unreliable, owing to the
usual compensation of convergence/divergence between the upper and lower troposphere
and “noise” in wind soundings due to small-scale eddies. Pedder (1981) illustrates proce-
dures to adjust these kinematic estimates.
In spite of these practical problems of estimating divergence, the basic concept is useful
in understanding that surface cyclogenesis is favored below the eastern limb of an upper
level trough (Bjerknes and Holmboe, 1944). The upper flow accelerates in this region as
the air passes from cyclonic curvature (where the gradient wind is subgeostrophic) towards
anticyclonic curvature (where the gradient wind is supergeostrophic). This acceleration
causes upper-level divergence and hence rising motion and a surface pressure fall (Figure
6.14). In this way we see that cyclone development is related to the long wave structure
and, hence to Rossby’s treatment of absolute vorticity conservation.
The vorticity equation can be written:
d( f )
= ( f ) (u·V)
dt
where relative vorticity (positive cyclonic) and H#V is the horizontal divergence of
velocity. The role of patterns of vorticity advection and thermal advection in the upper
troposphere in cyclogenesis was formulated by Sutcliffe (1947; see Hoskins, 1994) and
Petterssen (1955; Petterssen et al., 1962), in particular.
In synoptic terms, the upper-level flow locally accelerates towards a jetstream maximum
(core) and decelerates ahead of it (Figure 6.14). The entrance region of a jet streak is
associated with a transverse ageostrophic component directed toward the cyclonic-shear
side as shown in Figure 4.11, with the converse in the exit region. Case studies and
composite average patterns of jet maxima confirm the tendency for rising (sinking) motion
on the anticyclonic (cyclonic) side of the jet entrance and the converse in the exit region.
Consequently, this has an important influence on the associated patterns of vertical motion,
cloud, and precipitation distribution. These relationships are apparent even in a mean
sense, as shown by studies of Koteswaram (1958) and others for the Tropical Easterly
Jetstream over southern Asia and West Africa. A perspective view of the life cycles of
extratropical frontal cyclones in the context of planetary and synoptic waves is shown in
Figure 6.15. A 200 mb subtropical wave (m~3) and jetstream and a 300 mb polar wave
(m ~ 6) and jetstream each have suspended potential vorticity anomalies. The western
system with a trailing cold front and indeterminate warm front is within the meridional
anticyclonic shear south of the polar jet; the T-bone bent-back warm-front occlusion is
below the vertically aligned jets and potential vorticity anomalies and is characterized by
non-shear; the eastern system is a classical Norwegian warm-front occlusion evolving
north of the subtropical jetstream within meridional cyclonic shear.
The curvature of the upper contours modifies the distribution of divergence/conver-
gence in the jet entrance and exit zones, as illustrated in Figure 4.11. Rapid cyclogenesis
events have received considerable attention over the last decade or so, particularly owing
to events like the President’s Day storm, February 18–19 1979, and the QE II storm,
September 10–11 1978 (Bosart, 1981; Gyakum, 1983). Sanders and Gyakum (1980) coined
the term “bomb” to refer to storms that deepen explosively with pressure falls of at least
24 mb/24 hr (1 Bergeron) and up to 60 mb/24 hr. It is worth noting that, in contrast,
anticyclogenesis is considered rapid if pressure rises by 5 mb/24 hr and maximum rates
are only about 13 mb/24 hr (Colucci and Davenport, 1987). Leaving aside the mesoscale
polar lows, rapid deepening occurs primarily over warm ocean currents in the North
Atlantic and North Pacific (Sanders, 1986; Uccellini, 1990). Figure 6.16 illustrates this
distribution for 1976–82. However, rapid cyclogenesis is not limited to the oceans or
coastal areas, such as the east coasts of the United States and Australia; it can occur
11
11
Figure 6.15 Frontal cyclone life cycles in relation to planetary and synoptic-scale waves. Upper
plane (light shading): 200 mb subtropical wave No. ~3 and jetstream (open arrow)
with associated, suspended PV anomalies. Middle plane (heavy shading): 300 mb polar
wave No. ~6 and polar jetstream (open arrow) and associated, suspended PV anom-
alies. Lower plane: three characteristic surface-level frontal cyclones; open frontal
0 symbols indicate occlusion aloft. Left cyclone: a cause of anticyclonic shear south of
the polar front jetstream with weak trailing cold front. Middle cyclone: non-shear case
beneath vertically aligned polar and subtropical jetstreams featuring T-bone polar
occlusion and bent-back warm frontal seclusion. Right cyclone: a case of cyclonic
shear north of the subtropical jetstream representing the Norwegian model with back-
bent polar warm frontal occlusion. (From Shapiro and Grell, 1994)
in the cold seasons over the east-central United States, for example. A key element is a
“preconditioned” lower troposphere which features baroclinic development, cyclonic
vorticity, and reduced static stability (Bosart, 1994).
0 The development of these systems involves a rather short interval (twenty-four to thirty-
six hours) of extremely rapid deepening which tends to begin when the exit region of an
upper tropospheric jet moves over an area of diffluent flow downstream of a trough axis
(Uccellini, 1990). The contributions of thermal advection, diabatic heating, particularly
from latent heat release, and the reduction in static stability due to low-level air motion
over a warm ocean surface appear to vary in different cases (Kocin and Uccellini, 1990).
However, there are certainly non-linear synergistic interactions between them, according
to Uccellini. One of the most characteristic elements of rapidly developing cyclones is
an asymmetric pattern of clouds and precipitation situated on the poleward side of the
surface low. This feature is attributable to a rather cold and moist easterly flow moving
0 through the storm, with an ascending motion, and to a warm southerly flow that rises
over the warm front. These “conveyor belts” are illustrated in Figure 6.9. A related, explic-
itly Lagrangian, analysis has been carried out by Wernli and Davies (1997); they trace
11 the movement of potential vorticity anomalies, as well as changes in specific humidity
1
Figure 6.16 Distribution of maximum deepening positions of explosive cyclones, 1976–82. (From
Roebber, 1984)
and potential temperature in relation to ensembles of air parcel trajectories. Typically,

banded mesoscale cloud and precipitation structures are embedded within the conveyor
belts. The origin of such structures remains under discussion; suggestions include
boundary-layer friction effects, gravity waves, and conditional symmetric (or inertial)
instability (CSI), resulting from excess buoyancy acquired when air parcels are displaced
along surfaces of constant absolute momentum (Bennetts and Hoskins, 1979). Small-scale
processes associated with precipitation as well as the evaporation/sublimation and melting
of raindrops and snowflakes can also modify frontal cloud and rain systems and may help
to initiate and maintain multilayer cloud structures (Stewart et al., 1998).
Recent studies of cyclogenesis in the western North Atlantic indicate cases that depart
significantly from the Norwegian model. Neiman and Shapiro (1993) analyse a winter
storm that deepened 60 mb in twenty-four hours over 20°C Gulf Stream waters where
the turbulent fluxes approach 3,000 W m2 compared with mean January values of up to
400–500 W m2. They report a relative westward development of the warm front in the
polar air, with a bent-back occlusion form, and a warm core frontal seclusion within the
post-cold front cold airstream (see Figure 6.10). Different types of rapid maritime cyclo-
genesis can be identified from satellite imagery. This approach is especially valuable in
forecasting over oceanic data sparse areas (Bader et al., 1995). Building on the informa-
tion from the Experiment on Rapidly Intensifying Cyclones over the Atlantic (ERICA),
Evans et al. (1994) examine imagery for fifty rapid cyclogenisis events over the western
North Atlantic during the 1970s and 1980s. Four types of cyclogenesis are distinguished
and these are shown schematically in Figure 6.17. The “emerging cloud head” forms on
the poleward side of a cirriform cloud band along the polar front. The development takes
place in the left exit zone of the jetstream, downstream of the upper trough. As Figure
6.17a shows, there are two jet streaks and the cloud head forms in association with stream-
line diffluence and the phase adjustment of the wave trough and equatorward jetstream.
0
0
0
0
0
11
11
11
Figure 6.17 Schematic diagram of cloud signatures for four types of rapid maritime cyclogenesis, showing their evolution. (a) Emerging cloud head.
(b) Comma cloud. (c) Left exit (of jetstream). (d) Instant occlusion. The pairs of diagrams indicate early (left) and twelve to twenty-four
hours later (right) stages of evolution. (From Evans et al. 1994)
1 The cloud head signals rapid surface deepening, with the surface low situated near the
equatorward edge of the cloud head. The comma cloud (Figure 6.17b) has no interaction
with a polar-front cloud band. Development involves the expansion of a cloud cluster
into a “baroclinic leaf,” typically in the left exit region of a jet streak associated with a
diffluent upper-level trough. The longitudinal axis of the leaf tends to rotate cyclonically
as cyclogenesis occurs and the cloud system assumes a comma shape. A surface low
forms on the southwest edge of the comma cloud. In a variant of left-exit cyclogenesis,
a baroclinic leaf of relatively warm stratiform cloud forms poleward of an existing polar
front cloud band with colder, higher cloud tops (Figure 6.17c). The leaf, showing cyclonic
rotation, merges with the main cloud band and a surface low develops below the inter-
section of the leaf and the cloud band. The instant occlusion category (Figure 6.17d)
involves the merging of a “cold-air cloud cluster,” characterized by open cellular convec-
tion, with a polar front cloud band in a region of confluent upper flow. Rapid deepening
of the surface low takes place on the equatorward edge of the cold-air cloud cluster.
Evans et al. find that in the instant occlusion the longitudinal axis of the cloud head is
typically parallel to the upper tropospheric flow, whereas in the left-exit category it is
oriented perpendicular to the upper flow direction. A study of twelve frontal cyclones in
the North Pacific shows that they doubled in size over a four-day period (Grotjahn et al.,
1999). The lows were tracked by a wavelet analysis. This finding appears to contradict
the assumption of some theories of cyclone development. It remains to be determined
whether this result is a broadly representative one and whether it holds true in the upper
troposphere.
Secondary cyclogenesis within a primary frontal cyclone is a common phenomenon.
During the field campaign FRONTS 92, March–May 1992, one wave per day was observed
over the North Atlantic, of which half showed deepening. Secondary cyclones may form
in various locations within a primary system, as illustrated in Figure 6.18, although the
processes involved in their formation are not well understood. Parker (1998) notes that
during FRONTS 92 cold front waves were the most common type, whereas col waves
were most likely to undergo development. Secondary lows are commonly shallow systems,
implying that boundary layer processes play an important role.
The mechanisms involved in secondary cyclogenesis seem to differ qualitatively from
those in the primary systems. Proposed theoretical models invoke either a dry or a moist
frontal instability mechanism. Dry instability can occur within a warm band ahead of a
cold front or in a narrow, low-level zone of positive potential vorticity (PV). The width
of the zone determines the wavelength and growth rate – narrower zones giving shorter,
faster-growing waves (Parker, 1998). The Charney–Stern instability criterion (Charney
and Stern, 1962) requires that the basic state PV gradient and the equivalent PV gradi-
ents, representing the boundary gradients, have regions of opposite sign for normal-mode
dry instability of quasi-geostrophic systems. Wave growth first involves barotropic conver-
sions, with energy acquired from the kinetic energy of the basic state (the wind shear
across a front, for example); then, if the perturbation has enough vertical depth, baro-
clinic growth occurs. Laboratory studies with a rotating two-layer tank of water suggest
that, given a shallow depth of buoyant fluid layer relative to the total fluid depth, waves
grow mainly through barotropic instability for a large Richardson number (Ri), which is
a measure of the ratio of the PE of the basic state to its kinetic energy1 (Griffiths and
Linden, 1981). For small Ri and a large depth of buoyant fluid, baroclinic processes are
dominant. The advection of an upper-level PV anomaly over a surface front can also
trigger secondary cyclogenesis (Thorncroft and Hoskins, 1990). Moist instability may
develop when latent heat release generates an unstable PV zone. Latent heat release by
deep convection is of considerable importance in small-scale systems. The intensity of
the October 15–16 1987 storm in southern England may have been a result of such heating,
according to Hoskins and Berrisford (1988). However, many details remain to be resolved.
It is not certain whether frontal waves are equally common at all scales, or whether there
11
11
Figure 6.18 Schematic illustration of four locations within a parent (primary) frontal cyclone where
secondary cyclogenesis may occur. (From Parker, 1998)
0
is any scale dependence of the dynamical processes and wave structure. For an idealized
front separating air masses of different density, Orlanski (1968) showed that unstable
modes are possible for all values of the along-front wave number.
6.4 Storm tracks

6.4.1 Climatology
0 The conventional Lagrangian approach to defining storm tracks involves tracing the
movement of low pressure centers. Such manual investigations of cyclone tracks in mid-
latitudes began when synoptic weather maps were systematically prepared in the 1850s
and 1860s. Early studies were performed by E. Loomis (1874) for North America, H.
Mohn (1870) for Norway, V. Köppen (1880) and J. van Bebber (1891) for Europe, and
N.A. Rykachev (1896) for Europe, including European Russia. Monthly maps of cyclone
paths over the United States began to be published in Monthly Weather Review in 1875
and for the northern hemisphere four years later. Loomis (1885) was the first to assemble
information on cyclone paths over the northern hemisphere, but a comprehensive analysis
was possible only in the mid-twentieth century (Petterssen, 1950; Klein, 1957). Petterssen
0 drew attention to the importance of the zones where there is a high rate of alternation
between high and low-pressure centers, which he termed pressure ducts. Recently, tracking
algorithms have been developed for digital pressure data (Murray and Simmonds, 1991;
Jones and Simmonds, 1993; Serreze et al., 1993). The main results of these studies are
now summarized. For winter in the northern hemisphere (Figure 6.19):
1 Major tracks extend from the east coasts of the northern hemisphere continents north-
eastward across the oceans: in the North Atlantic, systems either turn northward into
Baffin Bay or, more frequently, continue northeastward to Iceland and the
Norwegian–Barents Sea; in the North Pacific, systems move from eastern Asia towards
0 the Gulf of Alaska.
2 Cyclones form or redevelop east of the Rocky Mountains in Alberta and Colorado
and move eastward towards the Great Lakes and Newfoundland before turning north-
11 ward towards Greenland and Iceland.
1
Figure 6.19 Cyclone tracks in the northern hemisphere (a) January, (b) July, 1958–77. Solid
(dashed) lines indicate primary (secondary) tracks; main cyclogenesis areas are
stippled. (From Whittaker and Horn, 1984)
11 A Lagrangian climatology of North Atlantic storm tracks illustrates a further novel
methodology (Blender et al., 1997). Cyclonic minima, defined from ECMWF 1,000 mb
height data at T106 resolution (~ 1.1° 1.1°) over 3 3 grid points, were tracked at
six-hour intervals for winters 1990–94. Cluster analysis of the track data defined three
groups of storms that are persistent for at least three days – quasi-stationary (representing
56 percent of the total), northeastward-moving (27 percent), and eastward-moving
(16 percent). Those moving northeastward from the east coast of North America about
40°–50°N toward the Norwegian Sea have a clear life cycle in terms of height anomaly
and height gradient, whereas the zonal group has only a weak cycle, and the stationary
0 systems none. Analysis of the zonal (x), meridional (y), and total displacement over time
(t) for each group demonstrates that mean-squared displacements of the cyclones follow
a power law scaling:
11 dx2(t) dy2(t) ≈ t
where is about 1.6 for the traveling systems, in line with other scaling analyses of
geophysical flows, and 1.0 for stationary systems, indicative of random walk type of
behavior. The persistence of a northeastward storm track regime averages about five days
(three to eight-day range), while the zonal regime has a slightly shorter duration.
0 In the southern hemisphere, in contrast to the northern, storm tracks are virtually circum-
global, with little seasonal variability (Figure 6.20). The track density in winter is a
maximum between 50°S and 60°S in the South Atlantic and South Indian oceans and
south of 60° in the South Pacific, with a secondary maximum near 40°S across the Pacific,
according to Sinclair (1997). This analysis for winters 1980–94 shows the frequency of
centers per 5° latitude circle per month and translation vectors. The maxima shown are
in higher latitudes than in earlier studies by the same author, where the grid spacing
favored detection at lower latitudes. Cyclogenesis is most frequent downstream of the
0
Figure 6.20 Cyclone track density in the southern hemisphere for winters (April–September)
1980–94 (contours show one center per 5° latitude circle per month) and average
11 translation vector (From Sinclair, 1997)
1 east coast of South America and southeast of South Africa, extending south of Australia
in a band into the South Pacific around 55°S.
A weakness of such analyses is the fact that changes in the intensity of the system and
its rate of movement have to be taken into account independently. An alternative frame-
work for the diagnostic analysis of the atmospheric circulation uses the analysis of the
variance of the geopotential height field. By high-pass filtering of the data to extract vari-
ance in the two to six-day range, the behavior of positive/negative height anomalies can
be examined. These are observed to propagate along zonally oriented wave guides. There
are close overall similarities between the two sets of patterns. Differences between them
are caused by zonal variations in the climatological mean flow, which may displace the
cyclone (anticyclone) relative to the corresponding anomaly center (Wallace et al., 1988).
For example, as an area of negative (positive) height anomaly moves eastward from east
Asia it enters a region where the mean 1,000 mb height gradient becomes more strongly
southward directed. Hence the cyclone (anticyclone) center is displaced northward (south-
ward) of the baroclinic wave guide. Figure 6.21a shows the divergent movement of cyclone
(a)
(b)
Figure 6.21 Storm tracks determined by traditional methods (a) from surface weather maps and
(b) from high-pass filtered 1,000 mb height data for two selected locations (40°N,
70°W and 35°N, 135°E) on the major storm tracks in the northern hemisphere. (a)
shows paths of cyclones (dashed arrows) and anticyclones (light solid arrows), also
contours of the mean winter 1,000 mb heights (30 m intervals). (b) depicts paths of
negative (dashed arrows) and positive (light solid arrows) height anomalies, also
contours of the mean winter 700 mb height field (60 m intervals). Both diagrams show
the movement of high-pass filtered anomalies inferred from lag-correlation maps
(heavy arrows). (Wallace et al., 1988)
11
11
Figure 6.22 Schematic relationships between the southern hemisphere mean jetstream (heavy solid
arrows), storm track (broad stippled arrow), and high-frequency eddy statistics.
Maxima in the various quantities are shown against the latitude scale. The double
0 broken lines denote the trough/ridge axes, the temperature perturbations (cold/warm)
are indicated by the bold dashed line, and the hatched zone demarcates cloud cover.
(Trenberth, 1991)
and anticyclone centers, whereas the positive/negative height anomaly tracks are almost
identical.
In the southern hemisphere, high-pass filtering (two to eight-day range) of 300 mb
height fields for 1979–89 shows a highly zonal storm track from 45°W eastwards to
150°W, centered about 50°S in January (austral summer), with a maximum concentration
over the South Indian Ocean. In July there is a more asymmetric pattern with a primary
0 track from the South Atlantic through the Indian Ocean around 45°–50°S, curving
poleward to 65°S at 160°W. There is no pronounced equatorward displacement of the
storm track in winter, as occurs in the northern hemisphere, and the occurrence of
maximum mid-latitude meridional temperature gradients during austral summer determines
the degree of storm track activity and the tendency to zonal symmetry. The observed
location of the primary storm track just downstream and poleward of the polar jetstream
maxima (Trenberth, 1991) is accounted for by linear baroclinic theory for the observed
basic state of the atmosphere in the southern hemisphere, according to Frederiksen
(1985). However, James and Anderson (1984) emphasize the role of moisture entrain-
ment into the low-level westerlies over the mid-latitudes of the South Atlantic, downstream
0 of the source in the Amazon basin, as responsible for the large increase there of transient
eddy activity.
The relation between the mean jetstream, the storm track, and the associated high-
frequency eddy statistics for the zonally symmetric circulation in the southern hemisphere
is illustrated schematically in Figure 6.22. Maximum height variance (z′2), indicating a
high rate of alternation, is located along the storm track, whereas perturbations of the
vorticity (′2) are greatest just equatorward of the track as a result of the variation of the
Coriolis parameter and consistent with the geostrophic relationship. Accordingly, pertur-
bations of the meridional wind (v′) are displaced correspondingly, but zonal wind
perturbations (u′) have maxima north and south of the storm track. The perturbations are
0 elongated meridionally, thusv′2 <u′2, and have a characteristic eastward-bowed shape
of trough and ridge axes. This generates momentum convergence from eddies into the
storm track (v′′ > 0). The perturbations of temperature, T ′, are greatest at low levels and
11 are highly correlated with the east–west variations in v′. Also, maximum perturbations
1 of moisture, q′ and vertical velocity, ′, are closely related and are located in lower lati-
tudes in association with the patterns of v′ and T ′.
It is noteworthy that the positive/negative height anomalies identified by high-pass
filtering of 1,000 mb height anomaly data in the northern hemisphere display the char-
acteristics of finite-amplitude baroclinic waves: over the oceans the waves are elongated
in the east–west direction, have a mean wavelength of 4,000 km, tilt westward with height,
and show a mean eastward phase propagation of 12–15 m s1 (Wallace et al., 1988).
However, there are significant differences east of the Rocky Mountains and the Tibetan
Plateau. The height anomalies move eastward following distinct wave guides which are
associated with maximum variance of the height fields. In contrast, cyclone (anticyclone)
paths in the northern hemisphere tend to be orientated southwest–northeast (north-
west–southeast), as illustrated in Figure 6.21. This is attributable to the fact that as
baroclinic waves move eastward they encounter changes in the mean basic state. It is
well known that, on a day-to-day basis, cyclones tend to be steered by the mean tropos-
pheric flow pattern in which they are embedded. Nevertheless, a cyclone may itself modify
the steering pattern through thickness and vorticity advection as it approaches maturity
(Palmén and Newton, 1969, chapter 11). On the time scale of weeks, steering patterns
are greatly influenced by the thermodynamic effects of heat sources (sinks) which act to
generate cyclonic (anticyclonic) vorticity in preferred locations. The long wave structure
may be reinforced or damped by such heating and it is this pattern that determines the
overall steering mode for intervals of perhaps two to four weeks.
Wave-guides themselves can be defined in several different ways (Wallace et al., 1988).
First, as described above, they are identified as regions of strong variability in the high-
frequency height anomaly fields (Blackmon, 1976). If the standard deviation values of
geopotential height are converted into geopotential stream function, &, by using an f 1
weighting of the geopotential height, the maximum variability is shifted about 5° equa-
torward in closer agreement with results obtained using vorticity and kinetic energy.
Second, bands of strong teleconnectivity in the high-pass filtered data, determined from
composite maps, show wave-like fluctuations. Third, using lag correlation maps, the phase
propagation of the high-frequency fluctuations shows vectors parallel to the wave guides.
Figure 6.23 presents an idealized representation of 1,000 and 500 mb wave guides for
winter conditions in the northern hemisphere, showing consistency between the three
methods.
The traditional mapping of cyclone/anticyclone tracks yields information relevant to
determinations of the wind field and the sequence of weather conditions over given loca-
tions which is appropriate for many synoptic climatological purposes. However, it is the
combined effect of the height anomalies in all frequency bands that make up the observed
circulation pattern at a given time. As pointed out by Trenberth (1991), cyclonic vorticity
advection and associated bad weather arise from the advent of a negative height anomaly
or the departure of a positive height anomaly.
6.4.2 Processes
The forcing and maintenance of storm tracks clearly merits explanation. There are two
basic hypotheses concerning the development of storm tracks. One considers that the
meridional temperature gradients formed by land–sea contrasts induce planetary wave
structures through heating and orographic effects that are modified by transient influences.
The second idea involves a self-organizing mechanism whereby eddies feed back onto
the time-mean flow. In this view, the storm track ends downstream through destruction
of the eddy energy. Recently it has been demonstrated that the statistics of extra-
tropical synoptic eddies can be derived from the assumption that they are stochastic-
ally forced disturbances evolving on a baroclinically stable background flow (Farrell
and Ioannou, 1993). Further, Whitaker and Sardeshmukh (1998) deduce the observed
0
0
0
0
0
11
11
11
Figure 6.23 Idealized representation of northern hemisphere wave guides in winter, based on high-pass filtered height fields. (a) 1,000 mb. The axes of
high teleconnectivity having wave-like high-frequency fluctuations are denoted by heavy arrows (dashed where less pronounced); the contours
are the 50 m (inner) and 40 m (outer) standard deviations and the arrows (scale at lower right) are 1,000 mb phase propagation vectors.
(b) Corresponding 500 mb patterns. The contours are 60 m (inner) and 50 m (outer) standard deviations. (Wallace et al., 1988)
1 wintertime statistics of the zonally varying synoptic eddies that are associated with the
observed zonally varying background flows. They use a two-level hemispheric quasi-
geostrophic model, linearized about the observed mean flow (for 400 and 800 mb winds,
1982–95), and forced by Gaussian white noise. The synoptic eddy covariance is linked
with the spatial structure of the background flow and with the covariance of the stochastic
forcing by a fluctuation–dissipation relationship; this relationship implies that the tendency
of eddies to decay is balanced by forcing. The model reproduces most major features of
the climatological winter storm tracks over the North Pacific and North Atlantic as well
as some aspects of their seasonal cycle and interannual variability. Using a similar
modeling approach, Whitaker and Dole (1995) examine the sensitivity of storm track
organization to zonally varying large-scale flow. They identify two competing processes
that are associated with the locations of a local baroclinicity maximum and a horizontal
deformation minimum. If the equilibrium state comprises a zonally symmetric tempera-
ture field and a barotropic stationary wave, the storm track is just downstream of a
minimum in horizontal deformation in the upper jet entrance zone. However, as zonal
variations in baroclinicity increase, the storm track is displaced to the jet exit region just
downstream of a baroclinicity maximum. With flows intermediate between these two
cases, there are storm track maxima in both the jet entrance and exit zones.
Midlatitude cyclones, at least off the east coasts of Asia and North America, develop
and intensify primarily through baroclinic instability associated with diabatic heating.
Hoskins and Valdes (1990) show that the major North Atlantic and North Pacific tracks,
identified using high-pass filtered 250 mb height data for winters 1979–84, are centered
somewhat eastward and poleward of the regions of maximum column-averaged diabatic
heating. Figure 6.24 shows column-averaged mean values of 50 W m2 or more over the
western oceans. This heating is mainly attributable to sensible heat fluxes in these loca-
tions during outflows of cold continental air, supplemented by latent heat released in
frontal cloud systems. Hoskins and Valdes show that the storm tracks are characterized
by a baroclinic instability parameter due to Eady (1949); the maximum growth rate is
BI (day1) = 0.31 f | |
V
z
N 1
where the Brunt–Väisäla frequency: N (g )1/2/( z) is the static stability parameter
(N 102 s1). Maxima of BI exceeding 0.6 day1 at the 780 mb level are found over
the western North Pacific and Atlantic oceans, implying amplification of systems by factors
of 2–3 day1. Figure 6.24 also shows poleward and vertical heat fluxes at 700 mb located
at the upstream ends of the storm tracks, with baroclinicity minima at the eastern ends
of the tracks. The horizontal eddy transports of heat in mid-latitude storms act to reduce
baroclinicity and therefore storm tracks might be expected to shift in time and space as
systems move through an area, yet this is not observed because vorticity fluxes help to
offset the effect. Rather, the storm tracks tend to be self-maintaining as a result of the
diabatic heating patterns primarily caused by the storm tracks. The vectors of the E flux

( v′2u′2 ,

u′v′ ) shown in Figure 6.24 diverge from the storm tracks, indicating a tendency
for cyclonic (anticyclonic) circulation on the poleward (equatorward) flanks, which serves
to force the mean westerly flow by counteracting the destructive effects of the eddy heat
fluxes on the baroclinicity. Orlanski (1998) confirms the increase in the barotropic compo-
nent of the zonal jet due to the second term of the E flux (Figure 6.25a). He also notes
that the first term displays a quadruple pattern of vorticity forcing, with cyclonic (anti-
cyclonic) forcing located to the northwest and southeast (southwest and northeast) of the
maximum in the  ( v′2 u′2) pattern. The quadruple pattern is also more or less in phase
with the trough–ridge system in the stationary flow (Figure 6.25b). The combined effect
of the two terms of E is to tilt the storm track axis into a southwest–northeast orientation
(Figure 6.25c). Hoskins and Valdes use a linear stationary wave model, with representa-
11
11
Figure 6.24 Winter storm tracks in the northern hemisphere for December–February 1979–84,
based on the two to six-day variance of the 250 mb height fields (thin contours at 15
0 m intervals) and the Eliassen–Palm flux vectors shown by arrows (see text). Also
shown are the 700 mb horizontal temperature flux   ′) (thick dashed contour, 10 K
(v′T
m s1), the 700 mb vertical temperature flux  ′T ′) (thick dotted contour, 0.2 K pa
(
s1), and the column mean diabatic heating (thick solid contour, 50 W m2). (Hoskins
and Valdes, 1990)
tive forcing in the storm track regions, to show that the mean diabatic heating off the
east coasts of the northern continents in winter provides the necessary environment for
storm track maintenance, overriding the thermal effects of the eddies. Nevertheless, the
low-level winds that arise as a result of cyclone passages set up wind stresses that help
0 to strengthen the Gulf Stream and Kuroshio currents, thereby in turn providing the initial
diabatic heating and baroclinicity for the atmosphere.
There are some unusual features in the seasonal occurrence of baroclinic wave activity
over the northern hemisphere. Nakamura (1992) finds a strong midwinter peak in the
frequency of 250 mb height variability under six days over the North Atlantic (70°–30°W),
whereas over the North Pacific (160°E–160°W) there are maxima in November and
March–April (Figure 6.26). However, the double peak is only weakly evident in the middle
troposphere and barely identifiable at the surface. There is also a larger seasonal shift in
the latitude of storm activity in the Pacific than in the Atlantic. Nakamura reports a posi-
tive correlation between jetstream strength and baroclinic wave activity for winds up to
0 about 45 m s1. At higher speeds, which occur in midwinter over the western North
Pacific, the correlation reverses. It seems that barotropic feedback from the baroclinic
waves is weakened when the upper tropospheric westerlies are very strong (> 50 m s1).
11 Stronger winds cause the phase speed of the waves to increase, but the steering level
1
Figure 6.25 Schematic illustration of the correlation velocity patterns, the induced circulation and
vorticity tendency. (a) The meridional flux of momentum, showing a dipole circula-
tion. (b) The variance of the v velocity minus the variance of the u velocity and the
induced quadropole circulation. (c) The total response of the eddy forcing. (From
Orlanski, 1998)
lowers from about 3 km for a 250 mb wind of 40 m s 1 to 2 km for a 65 m s1 wind.

This implies trapping of the waves near the surface. Wave activity may also be suppressed
because strong advection rapidly transfers developing waves downwind of a baroclinic
region, thereby limiting their ability to amplify. It is shown by Nakamura and others that
maximum baroclinic wave amplitude at 250 mb is located downstream of its surface-level
counterpart over the North Atlantic and Pacific. Nevertheless, other processes may be
involved, because the wave amplitude in the Pacific is not always pronounced even when
the jetstream wind speed is optimal. Nakamura proposes that atmospheric moisture, which
is higher during the transition seasons, may be a contributory factor.
Climatological storm tracks are commonly identified by maxima in the variance of
geopotential height. We have seen that large-amplitude, high-frequency eddies occur
preferentially downstream of the major stationary wave troughs at 500 mb, giving rise to
stationary storm tracks (Blackmon et al., 1977, for example). However, the planetary-
scale waves oscillate in position. Therefore it is important to understand how traveling
storm tracks may move in association with these planetary-scale waves. Low-frequency
(seven to ninety days) and high-frequency (less than seven days) components of geopo-
tential height can be separated by taking Fourier components of gridded height values in
the frequency domains, for example. Cai and van den Dool (1991) apply this separation
to twice-daily 500 mb heights for winters 1967–68 to 1976–77, 22°–90°N. At 50°N
the time-averaged amplitude of the stationary waves in the 500 mb heights is mainly
11
11
0 Figure 6.26 Latitude time section of the annual march of baroclinic wave amplitude in 250 mb
heights averaged (a) for the western North Pacific, 160 E°–160° W, and (b) for the
western North Atlantic, 70°–30° W. The plot was derived from a thirty-one-day
running mean envelope of six-day high-pass filtered daily height data. The data are
plotted at five-day intervals, with a contour interval of 10 m. (From Nakamura, 1992)
concentrated in zonal wave Nos 1 and 2, the amplitudes of the low-frequency waves are
similar for wave Nos 1 to 4 and then decrease slowly, while the smaller contribution of
high-frequency waves is largest for wave Nos 5 to 8. The low-frequency variability is
about twice that of the high-frequency component and represents regions of recurring
0 high-amplitude anomalies in the central North Atlantic, Gulf of Alaska, western Siberia,
and northern Hudson Bay. The high-frequency component has a background value of
about 40 gpm and maxima of 70–80 gpm in elongated zones resembling the storm tracks
of the North Atlantic and North Pacific. The high-frequency transient eddies reinforce the
barotropic component of the stationary waves, i.e. they lose energy barotropically to the
stationary waves, whereas the low-frequency eddies gain energy from the stationary waves.
According to Cai and van den Dool (1991), the low-frequency waves act to organize the
high-frequency wave troughs. By analyzing the RMS heights of the high-frequency eddies
within moving coordinates that are related to the low-frequency waves (wave Nos 1–4)
along 50°N, they show for 1,800 cases that a traveling storm track exists statistically
0 downstream of each trough of a planetary-scale traveling wave. For zonal wave Nos 1
and 2 there are two storm tracks, one centered about 45°N and the other about 55°N.
The vorticity flux of the high-frequency eddies in the traveling storm tracks serves to
11 reinforce the low-frequency waves and retards their propagation.
1 Model studies support the idea that storm track anomalies are driven by, and through,
feedback effects and may also modify large-scale, low-frequency circulation anomalies.
Using a series of GCM integrations, Branstator (1995) shows that the distribution of
storms can be altered by the barotropic component of the low-frequency perturbations
through the steering of synoptic systems by the mean winds. Alternatively, storm tracks
can be reorganized by changes in the location or intensity of baroclinic zones. However,
because the climatological distribution of storms is not random but has distinctive spatial
structure, large-scale circulation pattern anomalies can redistribute storm tracks such
that anomalous momentum fluxes may feed back positively on to the large-scale anom-
alies for some, but not all, of the primary circulation modes observed in the northern
hemisphere.
6.5 Satellite-based climatologies of synoptic features

The cloud fields organized in association with synoptic-scale circulation features are the
most obvious characteristics of satellite images (section 2.2; Conway and the Maryland
Space Grant Consortium, 1997). The early TIROS platforms confirmed ground-based
observations (Kuettner, 1959) that the atmosphere is organized primarily into banded
structures, such as fronts, jetstreams, and long waves (Whitney, 1966; Viezee et al., 1967;
Anderson et al., 1969; Kornfield et al., 1967; Kletter, 1972; Streten, 1968a, 1973, 1978;
Erickson and Winston, 1972; Downey et al., 1981). In addition, these images showed the
prominence of vortical (rotational) features associated with cyclones in tropical and extra-
tropical latitudes and, in some cases, anticyclones (Oliver, 1969; Chang and Sherr, 1969;
Troup and Streten, 1972; Streten and Troup, 1973; Dvorak, 1975). Cyclone centers are
identified by the presence of a cloud vortex, the configuration and tightness of the central
cloud relating to successive stages in the Norwegian model of frontal cyclone develop-
ment. Over ocean areas of the subtropics, high-pressure systems are characterized by
regions of only low-level cloudiness or, in visible channel imagery, by “sunglint.” Sunglint
is direct solar radiation reflected from a smooth sea surface back to the satellite, and
indicates the very light winds or calm conditions found within high-pressure areas.
While cloud patterns in the tropics are often organized linearly, comprising the ITCZ
(Gadgil and Guruprasad, 1990; Waliser and Gautier, 1993; Waliser et al., 1993), they
also exhibit an embedded globular, or clustered, appearance arising from the convection
associated with highly reflective clouds (HRCs) (Garcia, 1985; Grossman and Garcia,
1990; Hastenrath, 1990; Evans and Shemo, 1996). Thus cloud fields show structure on a
range of spatio-temporal scales, detected largely as a function of satellite sensor spatial
resolution (Hopkins, 1967; Houghton and Suomi, 1978; Sui and Lau, 1992). These scaling
characteristics imply important scale interactions of cloud physical processes, particularly
the moisture budget and cloud radiative forcing (Miller and Katsaros, 1992; Weaver and
Ramanathan, 1996; Stewart et al., 1997).
The earliest TIROS images confirmed the general synoptic model of the mid-latitude
cyclone, proposed some forty years previously by the Bergen school (cf. Bjerknes and
Solberg, 1922; Boucher and Newcomb, 1962; Boucher et al., 1963; Leese, 1962; Minina,
1964; Widger, 1964; Sherr and Rogers, 1965; Barr et al., 1966; Brodrick, 1969; Kondrat’ev
et al., 1970). Only relatively minor modifications and refinements occurred to that basic
model in the subsequent period (Katsaros and Brown, 1991), when satellite VIS and
infrared images were acquired at increasing spatial and temporal resolutions, in combina-
tion with intensive field programs to study extratropical cyclones at mesoscales, such as
CYCLES (CYCLonic Extratropical Storms) and ERICA (Experiment on Rapidly
Intensifying Cyclones over the Atlantic). These modifications include the so-called
“T-bone” structure exhibited by some intense maritime cyclones (section 6.3). However,
operational meteorological satellite imagery has also shown the existence of previously
unknown, or little known, features at synoptic and subsynoptic scales (Nagle and Clark,
11 1968; Chang and Sherr, 1969; Troup and Streten, 1972; Streten and Kellas, 1973); most
notably, the cold air mesoscale cyclone (“polar low”), the “instant occlusion” cyclone,
the Mesoscale Convective Complex (MCC), and the very long cloud bands linking the
tropics and higher latitudes and known generally as TECBs: tropical–extratropical cloud
bands (Kuhnel, 1989, 1990; Wright, 1997) or as “moisture bursts” in the eastern North
Pacific (McGuirk et al., 1987, 1988; McGuirk and Ulsh, 1990; Iskenderian, 1995).
While the spatial resolutions of satellite data have improved greatly since 1960 (section
2.2), the role of human analysts to interpret the cloud fields organized at synoptic and
subsynoptic scales in VIS/IR data, and to infer their basic physical associations with
0 tropospheric humidity, stability, vertical motion, and horizontal wind speed, has changed
little. However, quantitative evaluations became possible in the 1980s and 1990s with the
use of operational atmospheric sounders; notably the TOVS for retrieval of vertical
profiles of layer temperature and humidity at mesoscales (see section 2.2), and the Nimbus-
11 7 Total Ozone Mapping Spectrometer (TOMS) used to derive stratospheric ozone
concentrations. These data permit inferences about the important physical processes asso-
ciated with synoptic system development and, accordingly, help improve their prediction
(Velden et al., 1991; Velden, 1992; McGuirk, 1993; Barsby and Diab, 1995). Moreover,
microwave radiometry in dual polarization now permits more direct examination of
processes within and below the clouds over ice-free ocean surfaces (section 2.2). These
0 include precipitation rates, near-surface wind speeds, and the column-integrated cloud
liquid water and water vapor.
The following summarizes the development of satellite-based analyses of synoptic cloud
systems in the tropics and extratropics. The stages are not necessarily mutually exclusive
in time; for example, stages 2 and 3 occurred more or less concurrently during the 1970s.
1 The recognition and basic interpretation of synoptic cloud features for the enhance-
ment of meteorological analyses in data-sparse areas (Boucher et al., 1963; Alvarez
and Thompson, 1965; van Loon and Thompson, 1966; Blackmer et al., 1968; Zillman
and Martin, 1968).
0 2 The development of generic and genetic classifications of cloud signature evolution
to help identify associations with the conventional tropospheric fields for case
systems (Leese, 1962; Nagle and Serebreny, 1962; Brodrick, 1964; Brooks and Shenk,
1965; Chang and Sherr, 1969; Brodrick and McClain, 1969; Oliver, 1969; Kondrat’ev
et al., 1970; Sekioka, 1970; Dvorak, 1975; Evans et al., 1994).
3 Composite (multi-case) studies of the meteorological fields associated with specific
synoptic cloud features, useful for “bogusing” in numerical analyses (Elliot and
Thompson, 1965; Shenk and Brooks, 1965; Barr et al., 1966; Rogers and Sherr, 1966,
1967; Brodrick, 1969; Nagle and Hayden, 1971; Troup and Streten, 1972; Streten
and Kellas, 1973; Kelly, 1978; Sovetova and Grigorov, 1978; Junker and Haller,
0 1980; Carleton, 1987; Smigielski and Mogil, 1995).
4 Determination of the characteristic regimes and variability (i.e. “climatology”) of
specific cloud features on synoptic and subsynoptic scales (fronts, cyclones, TECBs,
ITCZ, mesoscale cyclones and polar lows, the MCC) based upon the identification
of their areas of formation and decay, and tracking system movements (Streten and
Troup, 1973; Carleton, 1979, 1981a, b, 1985a, 1995, 1996; Carrasco and Bromwich,
1996; Carrasco et al., 1997a, b; Fitch and Carleton, 1992; Forbes and Lottes, 1985;
Heinemann, 1990; Reed, 1979; Turner and Thomas, 1994; Turner et al., 1996; Waliser
and Gautier, 1993).
5 Associations of the synoptic climatologies developed in 4 above with climate system
0 features; for example, SST anomaly distributions, teleconnections with ENSO and
NAO, polar sea ice and snow cover extent and their variations, and the transport of
eddy sensible heat (Carleton, 1981c, 1983, 1985b, 1987, 1988a, b, 1996; Carleton
11 and Whalley, 1988; Kuhnel, 1989; Iskenderian, 1995).
1 6 Attempts to automate the classification of cloud systems and their intensity, to reduce
the subjectivity of manual methods (stages 2 and 3 above) (Burfeind et al., 1987;
Arnaud et al., 1992; Pankiewicz, 1995; Evans and Shemo, 1996; Velden et al., 1998).
7 Application of mesoscale information from newer satellite sensors to help determine
the structure of, and physical processes associated with, the cloud systems appearing
in VIS/IR images; particularly the use of passive microwave and sounder data on
precipitation, humidity, geopotential thickness, winds, stratospheric ozone (Velden,
1989, 1992; Alliss et al., 1993; Carleton et al., 1995; Carleton and Song, 1997; Claud
et al., 1992, 1995; Heinemann et al., 1995; Katsaros et al., 1989; McMurdie and
Katsaros, 1991; Rao and MacArthur, 1994; McGaughey et al., 1996; McMurdie et
al., 1997; Petty and Miller, 1995; Song and Carleton, 1997; Miller and Petty, 1998;
Carleton and Song, 2000).
8 Assimilation of the satellite synoptic and mesoscale information in stage 7 into real-
time numerical analysis and prediction models (Zillman et al., 1990; Velden et al.,
1992; Hewson et al., 1995; Marshall and Turner, 1997).
6.5.1 Physical processes of satellite-viewed organized cloud fields

Clouds are tracers of atmospheric energy, moisture, and stability. Most often, they become
evident in satellite VIS/IR images when air rises through a given depth of the tropos-
phere, and the adiabatic expansion of air parcels results in cooling to the dewpoint
temperature (TD). As this occurs, the outgoing long-wave radiation (OLR) in the infrared
window region decreases and the albedo (visible reflectance) increases. The thickness of
the cloud and its form (either cumuliform or stratiform) are dependent on the depth of
the moist air and the atmospheric stability (Kruspe and Bakan, 1990). Over ocean areas
moisture is not a limiting factor, so the presence and form of the clouds are determined
mostly by the atmospheric stability and larger-scale vertical motion patterns (see Zillman
and Price, 1972). However, over land the characteristics and heterogeneity of the surface
help to determine the availability of moisture and the partitioning of the net radiation into
the latent and sensible heat fluxes, which are important in cloud development – at least
for boundary-layer clouds – and the ease with which air parcels can be moved above the
condensation level by either forced or free convection (Weston, 1980; Pielke and Zeng,
1989; Gibson and Vonder Haar, 1990; Wetzel, 1990; Rabin et al., 1990; Raymond et al.,
1994; Cutrim et al., 1995; Betts et al., 1996; Doran and Zhong, 1995; Lyons et al., 1993;
Rabin and Martin, 1996; Brown and Arnold, 1998).
Plate 5 shows enhanced infrared black-body GOES images of much of the region
centered on North and Central America on a day in June 1988. The images are taken
twelve hours apart and show phenomena on a range of spatial scales. The following
features are particularly evident:
1 A larger diurnal range of the surface temperature over land compared with the ocean
(compare the gray tones between the two images).
2 Nocturnal thunderstorm activity (see image at 06.00 UTC) over the southwest and
upper Midwest United States (the enhanced cloud tops).
3 Shower and thunderstorm activity in the ITCZ, located in northern South America
and the eastern tropical Pacific.
4 Areas that are clear or have only low stratiform clouds over the eastern portions of
the subtropical oceans, associated with strong subsidence.
The associations between atmospheric stability and the satellite-observed cloud form
were demonstrated by Tang et al. (1964) and Shenk and Brooks (1965). These authors
showed that, over the ocean outside the tropics, air moving equatorward (poleward) is
potentially colder (warmer) than the surface over which it is traveling. Thus cloud fields
11
11
Plate 5 GOES-E enhanced thermal infrared images of the North America and Central America
11 regions on June 25 1988 at (a) 06.00Z, and (b) 18.00Z. The high cold cloud tops associ-
ated with convective and frontal systems are particularly evident (see text). (NOAA)
1
Figure 6.27 The Guymer-type synoptic cloud model of an extratropical cyclone (southern hemi-
sphere perspective). (From Guymer, 1978)
tend to be dominantly cumuliform (stratiform) in response to the different atmospheric

stability regimes that result. When cold air moves equatorward sensible heat and latent
heat are directed strongly upward; however, when warmer air moves poleward, the sensible
heat is directed towards the surface. These differences in stability and heat flux were
shown for the different satellite-viewed sectors of mature extratropical cyclones over the
southern oceans by Zillman and Price (1972). Their satellite synoptic “model” (Figure
6.27) was subsequently elaborated by Kelly (1978) for incorporation into the Australian
Bureau of Meteorology’s numerical analysis and prediction scheme. The heavy reliance
on satellite VIS/IR imagery for the synoptic analysis of the southern oceans by the Bureau
of Meteorology has a long history (Zillman and Martin, 1968; Troup and Streten, 1972;
Streten and Kellas, 1973; Streten and Downey, 1977), and is still important today (Zillman
et al., 1990; Seaman et al., 1993; Leighton, 1994).
Given sufficient moisture, clouds will develop whenever lower-level convergence of
air occurs. Excluding the more local effects of surface features (topography, contrasting
land covers) on cloud development, widespread upward motion and surface pressure falls
are generated when divergence in the upper troposphere exceeds convergence in the lower
troposphere. On average, the magnitude of the vertical motion decreases as the speed
divergence (or convergence) of the horizontal wind component increases away from the
level of non-divergence located at about 600 mb (see sections 2.4.4–5). Dines compen-
sation ensures that the divergence and convergence have opposite sign above and below
the level of non-divergence. On the forward (back) side of a mid-tropospheric wave, warm
11 (cool) air rises and cools (sinks and warms), providing the energy for continued ampli-
fication of the wave (slantwise convection). However, in both sectors of a wave the sign
of the vertical motion induced by adiabatic processes is insufficient to counteract the
temperature change due to the combination of advective and also diabatic (radiation
absorption and loss, latent heating) processes. The last is positive, or warming, for conden-
sation and when the resulting thick clouds absorb solar radiation and impede the infrared
emission to space. The diabatic component is negative, or cooling, when clouds evapo-
rate or when infrared losses from the surface increase.
As illustrated in Figure 6.14, divergence is favored in the forward, or exit, region of
0 a jetstream maximum, and on the eastern side of an upper trough because of the nega-
tive relationship of the divergence with the vorticity in the horizontal plane. Thus upper
divergence (convergence) and ascent (subsidence) are favored on the forward (back) side
of an upper trough by the advection of vorticity. Accordingly, satellite-viewed cloud fields
11 are characteristically deeper (shallower) on the forward (back) side of the trough. On
synoptic time and space scales, where traveling medium-wavelength, or baroclinic, waves
predominate (wave Nos approximately 7–10), the quasi-horizontal motion of air parcels
cuts across surfaces of constant potential temperature, or isentropes. Potentially warmer
Plate 6 DMSP visible

channel mosaic
(5.4 km resolution) of the
Europe/western Russia
sector, February 22 1978.
A “shadow band”
0 associated with a jetstream
is clearly evident overlying
the snow-covered land.
11 (NISDC)
(b)
1
(a)
Figure 6.28 Atmospheric circulation over the Australasian sector at 00Z on November 8 1992. (a) SLP. (b) 500 mb height. (From the Commonwealth
Bureau of Meteorology, Melbourne)
11 (colder) air rises and cools (sinks and warms) ahead of (behind) a trough, enhancing the
amplification of the wave. This process accompanies synoptic “development,” which in
middle latitudes is evident on satellite VIS/IR imagery by the following two features
(Burtt and Junker, 1976): (1) a brightening of the clouds, indicating rising cloud tops
(increased albedo, lowering cloud top temperature), and (2) increasing anticyclonic curva-
ture to a cloud band (indicating divergence aloft). A satellite-observed feature associated
with (1), especially over the ocean, is enhanced convection, which may precede the forma-
tion of some “polar lows” and “instant occlusions” (below). Typically associated with (2)
is the formation of the “open wave” stage of an extratropical cyclone. This satellite-
0 viewed stage, termed a “baroclinic leaf” (Evans et al., 1994; Smigielski and Mogil, 1995),
is particularly prominent in cases of rapid cyclogenesis (Jaeger, 1984), including the forma-
tion of synoptic cyclone “bombs” (Bottger et al., 1975).
The changes in sign of the relative vorticity and divergence between the equatorward
11 and poleward sides of an upper-tropospheric jetstream result in opposing patterns of
vertical motion: from ascent on the equatorward side to subsidence on the poleward side.
Thus the jetstream axis is often evident in satellite VIS images by a sharp boundary, or
“shadow band” (Whitney, 1966; Anderson et al., 1969), that demarcates the higher cloud
on the warm side of the jet from the cold side of the jet that has lower clouds or is clear
(Plate 6), or by the cirrus streamers on the warm side of the jet that terminate at the jet
0 axis. In infrared images a similar feature to the shadow band is a marked discontinuity
in the TBB across the jet axis (Martin and Salomonson, 1970), which can be used to iden-
tify the jetstream in the absence of conventional data (Togstad and Horn, 1974).
In synoptic weather analysis, and increasingly also for synoptic climatological studies,
much use is made of satellite remote sensing in more than one spectral band, and from
more than one platform (Carleton et al., 1995; Claud et al., 1995). This is illustrated here
for the Australian sector of the Southern Ocean on a day in November 1992. The passive
microwave images from the SSM/I illustrate the wealth of additional new mesoscale infor-
mation that can be retrieved over data-void ocean areas. Figure 6.28 a and b shows,
respectively, the broad-scale fields of SLP and 500 mb height at this time. A high-ampli-
0 tude trough extended well into central Australia from a cyclonic system near Adelie Land,
Antarctica. A secondary low had formed on the equatorward end of the front in the Great
Australian Bight. Given the high amplitude of the trough, and the presence of the subtrop-
ical high just west of Western Australia, strong cold air advection was occurring west of
the trough. To the east of the trough, moisture of subtropical origin was moving pole-
ward and ascending, resulting in a broad meridionally oriented cloud band ahead of the
main frontal system.
The SLP map is derived from observations from surface land stations, ships, and ocean
drifting buoys. Additional, so-called “bogus,” information is also provided by VIS/IR
cloud imagery from polar orbiting and geostationary satellites (Plate 7a). The latter include
0 lower-level winds over the ocean (which help improve the isobaric analysis via applica-
tion of the geostrophic wind rule), derived from automated tracking of cumuliform clouds
in successive satellite images at high temporal resolution (Wylie et al., 1981; Nieman et
al., 1993, 1997; Ottenbacher et al., 1997).
The constant pressure analyses (e.g. Figure 6.28b) are model-generated fields of geopo-
tential height, or thickness, and isotachs. They also incorporate considerable amounts of
satellite information, particularly from the NOAA TOVS. The TOVS provide informa-
tion on layer-mean temperatures (directly related to the thickness) and moisture for a
range of cloud cover conditions, and may even be extended over ice-covered surfaces in
polar regions. However, the TOVS retrievals are most reliable for the middle to upper
0 troposphere (Kopken et al., 1995). Automated tracking of cirrus-level clouds on succes-
sive geosynchronous images (lower latitudes) or images from polar orbiters (higher
latitudes) helps yield supplementary information on jetstream winds (e.g. the small jet
11 maximum west of the upper trough along longitude 120oE and centered at about 45oS in
1 Figure 6.28b. Thus the 500 mb analysis shown in Figure 6.28b shows strong vertical
consistency with the manually constructed SLP chart (Figure 6.28a).
Processes occurring within and below the clouds have to be inferred from visible and
infrared images. To expand upon the synoptic situation for November 7–8 1992, the SSM/I
retrievals of column-integrated water vapor (IWV), integrated cloud liquid water, solid
(ice) precipitation and liquid precipitation (Petty, 1994), and near-surface wind speeds,
are presented in Plate 7b–f. Successive orbits leave a data-void area in and south of the
Australian Bight, which unfortunately excludes part of the secondary wave cyclone in
that region. However, the broad-scale fields are consistent with those depicted in Figure
6.28a, b. For example, the IWV (Plate 7b) confirms the considerable latitudinal extent of
the cold, dry air to the west of the high amplitude trough in the Australian region, and
the warm and moist air to the east. (The very high values around Antarctica in every
SSM/I image are spurious; they are due to high microwave emission from sea ice.)
Highest values of water vapor are confirmed as occurring in the meridionally oriented
cloud band to the east of the trough, which is also the region of greatest cloud liquid
water content. Areas of liquid (rain) and solid (snow, graupel) precipitation are deter-
mined using the P37 and S85 indices (Plate 7d, e).2 Since P37 values < 0.8 indicate
increasing probability of rain (section 2.2), these occur in the frontal cloud band east of
the trough and also in the upstream frontal system located well south of Western Australia.
Scattering of the microwave signal by large ice particles (Plate 7e) occurs in the deeper,
or higher and colder, parts of cloud systems. While there is almost certainly precipitation
also occurring in the cold air cumulus over the Southern Ocean west of the meridional
trough, the coarse resolution of the SSM/I and the phenomenon of “beam filling,” whereby
individual convective elements go unresolved, do not allow these showers to be detected
(section 2.2).
Finally, Plate 7f shows the near-surface (19 m height) wind speed retrieved from
backscattering of the microwave signal by the ocean surface. Comparisons of SSM/I winds
with those derived from buoys and other remotely sensed wind speed measurements over
the ice-free oceans, such as Geosat (Mognard and Katsaros, 1995a, b) show high relia-
bility; the major exception is in areas of large cloud liquid water or heavy precipitation.
Thus one should ignore the wind speed values located between about 45°–55°S in the
eastern cloud band (Plate 7f), as well as those contaminated by sea ice. Accordingly, the
SSM/I data yield highest wind speeds (16–22 m s1) just west of the secondary low in
the Australian Bight. Lowest wind speeds of about 2–6 m s1 occur on either side of the
long-wave trough at higher latitudes.
Satellite-based climatologies of synoptic and sub-synoptic cloud features in the extra-
tropics consist of three main types. These are:
1 Composite (average) statistical “models” of the cloud system features (e.g. jetstream
maxima, cyclonic vortices, cloud bands) on a movable grid (Cox, 1969; Businger,
1990), based on the analysis of many cases having broadly similar satellite-observed
characteristics, and in which either conventional atmospheric data or other satellite
information (e.g. SSM/I) are averaged (Martin and Salomonson, 1970; Troup and
Streten, 1972; Streten and Kellas, 1973; Carleton, 1987; Katsaros et al., 1989; Rao
and MacArthur, 1994; Rodgers et al., 1994; Petty and Miller, 1995; Song and
Carleton, 1997; Miller and Petty, 1998).
2 The time-averaged regimes of circulation features, particularly synoptic and subsynoptic-
scale cyclonic vortices. This involves mapping the formation, maturity and dissipation
areas of many systems over weekly, monthly, or seasonal time scales, and also the tracks
of those systems (Streten, 1968b, 1974; Streten and Troup, 1973; Carleton, 1979, 1981a,
b, 1996; Carleton and Carpenter, 1990; Carleton and Fitch, 1993; Carleton and Song,
1997; Carrasco et al., 1997a, b; Forbes and Lottes, 1985; Turner and Thomas, 1994;
Turner et al., 1996; Yarnal and Henderson, 1989a, b).
11 3 For subsynoptic-scale systems, particularly MCCs and cold-air mesocyclones, the
composite synoptic anomaly fields (e.g. 500 mb height, SLP, vorticity, surface–air
temperature difference) associated with those features (Maddox, 1983; Businger, 1985,
1987; McAnelly and Cotton, 1989; Sinclair and Cong, 1992; Fitch and Carleton,
1992; Carleton and Fitch, 1993; Carleton and Song, 1997).
The above three categories comprise the basic framework for examining synoptic–
dynamic climatologies of satellite-observed cloud systems in the extratropics, on a range
of spatial scales (below).
0
6.5.2 Synoptic-scale extratropical cloud vortices
Statistical models of extratropical synoptic cyclones have been developed, based on the
11 composite fields of SLP, 500 mb height, wind speed, sea–air temperature difference, and
vorticity, in different stages of cloud system development (Troup and Streten, 1972; Streten
and Kellas, 1973; Zillman and Price, 1972; Kelly, 1978; Junker and Haller, 1980; Carleton,
1987; Seaman et al., 1993; Lau and Crane, 1997). These subjective (manual) classifica-
tion schemes of cloud vortices rely on assigning discrete stages of development in frontal
cyclogenesis to a process that occurs over a time continuum. Fortunately the satellite-
0 viewed cloud signatures associated with each stage (incipient → developing → maturity
→ dissipation → decay) are relatively distinct, especially for vortices occurring over the
ocean. The successive stages of cloud vortex evolution also tend to occur at successively
higher latitudes in the extratropics, on average. The incipient, or open-wave stage (W in
the Troup and Streten, 1972, scheme), is marked by the appearance of a bulge, thick-
ening, or anticyclonic turning of an existing frontal cloud band; the developing stage (B
stage) is signalled by a “dry slot” which develops to the rear of the cloud band. Composites
of the SLP and tropospheric height anomalies (i.e. synoptic climatology Type 1, above)
for these early stages of cyclonic development (Troup and Streten, 1972; Carleton, 1987)
show strong baroclinicity: the minima in these fields are displaced towards the colder air
0 with increasing height in the system.
Cyclone maturity (C stage) is heralded by the dry slot taking on a distinct cyclonic
curvature, either hook-shaped or spiral in shape (Troup and Streten, 1972; Carleton, 1987).
It is at this stage that the center of cyclonic rotation is easiest to discern, and central
surface pressures are usually at their lowest (Sovetova and Grigorov, 1978; Junker and
Haller, 1980; Carleton, 1987; Smigielski and Mogil, 1995). Moreover, microwave radiom-
etry shows that the most rapidly deepening cyclones typically have heaviest rain rates in
their northeastern quadrant, associated with the highest latent heat release (Petty and
Miller, 1995; Miller and Petty, 1998). Disorganization of the central cloud vortex struc-
tures – for example, a “loosening” of the spiral cloud marking the C stage, or filling-in
0 of the central cloud vortex – marks the onset of the dissipation (D stage). The D stage
is usually more protracted than the preceding stages of development of extratropical cloud
vortices, and can endure for up to several days (Troup and Streten, 1973; Carleton, 1987).
There are also variations in the morphology of the D-stage that indicate strong differ-
ences in the associated SLP and tropospheric height departures from “climatology.” Streten
and Kellas (1973) showed that D vortices can exhibit either a symmetrical or an asym-
metrical frontal cloud band – the latter being the more intense system, and sometimes
associated with “reintensification.” Ultimately the D-stage vortices may be observed to
enter a protracted “decay” (E stage), marked by the loss of a single frontal band and the
appearance of a swirl of generally lower-level clouds. At this stage the vortex is cold-
0 cored throughout. However, when located over ice-free ocean areas in higher latitudes,
such as the south-east South Pacific and northern North Atlantic, the weak static stability
means that cyclogenesis on a mesoscale is favored within these pre-existing centers of
11 cyclonic circulation when PVA maxima also occur (Zick, 1983). Accordingly, multiple
1 mesoscale vortices that evolve and rotate within E-stage vortices at higher latitudes
comprise the so-called “merry-go-round” configuration of mesocyclones (below).
Satellite-based climatologies of synoptic-scale extratropical cloud vortices (synoptic
climatology Type 2, above) have been developed almost exclusively for the southern
hemisphere (Streten and Troup, 1973; Carleton, 1979, 1981a, b, 1983). The basic features
which they reveal have been confirmed and extended by recent studies using long-period
grid-point analyses (Jones and Simmonds, 1993; Sinclair, 1994, 1995). The few studies
undertaken for the northern hemisphere are typically of shorter duration and identify asso-
ciations between the different stages of cyclonic development and climate teleconnections,
such as the North Atlantic Oscillation (Carleton, 1985a, 1988b, 1996), or cryosphere vari-
ations (Carleton, 1987). In the southern hemisphere, frontal wave cyclogenesis (W and B
stages in the Troup and Streten classification scheme) predominates in the western parts
of ocean basins in all seasons (Streten and Troup, 1973, Carleton, 1979). This empha-
sises a three-wave pattern in the time-averaged synoptic fields. Cyclonic maturity (C
stage), dissipation and decay (D and E stages) occur to the southeastward, at progres-
sively higher latitudes. The Antarctic circumpolar trough comprises mostly C and D/E
vortices, although a considerable amount of mesoscale cyclogenesis also occurs at higher
southern latitudes in all seasons (Carleton, 1992; Turner et al., 1993). The average frequen-
cies of cloud vortices are greatest for the winter season. Highest densities of “late stage”
(C, D, E) cloud vortices occur near Wilkes Land, in the Bellingshausen/Amundsen
seas, and northeast of the Weddell Sea. These coincide with the areas of highest snow
accumulations (>300 kg m2 a1, or 300 mm a1) in coastal Antarctica (Carleton, 1992),
and emphasize the importance of synoptic cyclones for transporting moist air from lower
latitudes on to the ice sheet (Bromwich, 1988). The satellite-observed successive stages
of frontal cyclone development are also linked with the efficiency of poleward transport
of the eddy sensible heat, occurring via their predominance in particular latitude zones.
Carleton and Whalley (1988) show that the “early” (W, B) stages of cloud vortices trans-
port the most heat polewards, on average, with strong eddy convergence of sensible heat
into the circumpolar trough in the short-lived C stage. Moreover, there is an ENSO signal
in the statistics of the eddy heat transport that is effected by cyclones, at least for the
winters (June through September) of 1973–77. For that period, about 36 percent of the
meridional heat transport by the vortices was explained by the SOI, with greater (lesser)
fluxes effected by the cyclones when the SOI is low: El Niño (high: La Niña).
Interannual variations in the spatial patterns of extratropical cyclones of the southern
hemisphere are large, and show dependence on the El Niño Southern Oscillation, partic-
ularly in the South Pacific sector (Streten, 1975; Berlin, 1991). Sinclair et al. (1997),
using grid-point analyses, show a more or less linear response of synoptic cyclones to
ENSO. In the winter of an El Niño warm event these are reduced (increased) over the
Indian Ocean, Australasia, and the Amundsen Sea (near Wilkes Land and the subtropical
eastern Pacific), and vice versa for the La Niña cold event. The change in intensity of
the Amundsen Sea low associated with ENSO is an important component of the PSA
(Pacific–South America) teleconnection pattern, which is discussed in the context of low-
frequency circulation variations in section 5.9.
6.5.3 Tropical cloud clusters and vortices

The classification systems designed to identify and describe the time evolution of the
cloud vortices associated with rotating storm systems originating in the tropics are based
upon pattern recognition in VIS/IR imagery like those for extratropical cyclones
(Oliver, 1969; Dvorak, 1975; Lander, 1990). The classification systems of Oliver (1969)
and Dvorak (1975) describe the intensification (decreasing surface pressure, increasing
surface wind speeds) associated with the progression from highly reflective cloud clus-
ters of tropical depressions that exhibit little organization through the increasing circularity
11 and size associated with tropical storms, and the appearance and structure of a relatively
cloud-free eye (lower albedo, higher black-body temperatures) in full-blown hurricanes,
typhoons, and tropical cyclones. Moreover the TBB data from geosynchronous plat-
forms show that the cloud tops associated with tropical vortices, as well as smaller cloud
clusters, undergo strong intradiurnal variations associated primarily with cloud-top
radiational processes (Muramatsu, 1983; Steranka et al., 1984; Mapes and Houze, 1993;
Machado et al., 1993). These are confirmed from passive microwave data (Rodgers and
Pierce, 1995).
Because of their development primarily over ocean areas, and their potential for destruc-
0 tion along populated coastal areas, tropical cloud vortices appearing on VIS/IR imagery
have always been highlighted in the synoptic analysis. Thus, unlike the situation with
extratropical cloud vortices, the synoptic chart is usually a reliable indicator of the loca-
tion of tropical cyclones – at least, in their more organized stages of development. The
11 locations of tropical cyclogenesis, tracks of movement, and approximate central surface
pressures (i.e. synoptic climatologies) are readily compiled from synoptic charts that
include the interpretation of satellite images – as in the Mariners Weather Log monthly
series – rather than from the imagery alone (Evans and Shemo, 1996).
The emphasis in satellite remote sensing of tropical cloud systems has been mainly in
the following areas:
0
1 Identifying the initial stages of development of tropical waves in VIS/IR imagery.
For example (Velasco and Fritsch, 1987; Laing and Fritsch, 1993a, b) many tropical
vortices originate from Mesoscale Convective Systems forming over or near elevated
terrain, such as the highlands of Mexico and West Africa. These MCSs may become
more organized upon moving westward over ocean areas where SSTs exceed about
27oC and atmospheric conditions (especially low-level convergence and upper diver-
gence) are favorable (Gray, 1968; Lau and Crane, 1997).
2 Identifying and tracking the development of eastward-moving “superclusters” asso-
ciated with the thirty to sixty-day oscillation, as revealed in Hovmöller analyses of
0 outgoing long-wave radiation (OLR) (Nakazawa, 1986; Lau and Chan, 1988). The
basic characteristics of superclusters, such as cloud-top temperatures, cloud system
size and propagation speed, have relied upon the analysis of VIS/IR images from
geosynchronous platforms (Nakazawa, 1988; Mapes and Houze, 1993). Associations
between the extent of supercluster activity in the western tropical Pacific, and the
ENSO in the central and eastern Pacific, have been suggested (Lau and Chan, 1988;
Nakazawa, 1988).
3 Passive microwave observations (e.g. SMMR, SSM/I) of rain rate, cloud water, and
ice scatter, and radar altimetry (e.g. Geosat, TOPEX-Poseidon) of surface wind speeds
help to depict and predict the intensification and movement of tropical cyclones
0 (Porter, 1990; Fishman, 1991; Alliss et al., 1993; Rodgers and Pierce, 1995; Rodgers
et al., 1991; Velden et al., 1992; Peng and Chang, 1996). These data reveal that the
most intense systems tend to have greater rain rates and more ice associated with
increased latent heat release in the eye-wall clouds (Rao and MacArthur, 1994;
Rodgers et al., 1994). In addition, the mesoscale wind vectors retrieved by scat-
terometers (Seasat, ERS-1) can be used in PBL models to generate detailed SLP fields
(Hsu and Liu, 1996).
4 Using TOVS-derived temperatures to determine tropical cyclone intensity from the
strength of the warm pool in the upper troposphere (e.g. 250 mb) that results primarily
from latent heat release in the eye wall (Velden and Smith, 1983; Velden, 1989).
0 Moreover, the application of ozone observations from the Nimbus-7 TOMS adds
information on the stratosphere–troposphere exchange of air and its relation to the
dynamics of tropical cyclones in different regions (Rodgers et al., 1990; Stout and
11 Rodgers, 1992).
1 6.5.4 Mesoscale cyclones (mesocyclones, “polar lows”)
Visible and infrared imagery shows the relatively frequent occurrence of subsynoptic-
scale cyclonic cloud vortices over the ocean in the cold airstreams to the rear of major
synoptic cyclones (Anderson et al., 1969; Whitney and Herman, 1968; Martin, 1968;
Chang and Sherr, 1969; Zillman and Price, 1972). These systems became known as
“comma clouds” because of their characteristic signature, and are observed to develop
out of enhanced convection in response to the positive vorticity advection (PVA) associ-
ated with short waves or jetstream maxima (Harrold and Browning, 1969; Ninomiya,
1989; Craig, 1993; Evans et al., 1994). Comma clouds are the satellite-observed signa-
ture of some “polar lows” that generate much of the organized post-cold frontal shower
activity and pressure/wind discontinuities noted in surface weather observations but which
the basic Norwegian model of the middle-latitude cyclone could not adequately explain.
Before their confirmation as discrete subsynoptic systems in satellite imagery, “polar lows”
were often identified on synoptic charts as “back-bent occlusions” or as polar “trofs”
(Reed, 1979) that were associated with the main frontal cyclone. Broadly similar systems
occur, only with less frequency, over the Great Lakes and interior continental areas during
winter (Mullen, 1982; Forbes and Merritt, 1984; Mills and Walsh, 1988).
The interpretation of satellite imagery revealed that a comma cloud may either remain
separate from the cold frontal cloud band that precedes it (typically where a large hori-
zontal distance separates the two features) or it may “catch up” with the cloud band and
initiate frontal cyclogenesis as an “instant occlusion” (Plate 8). The term “instant occlu-
sion,” or “instant frontogenesis” (Anderson et al., 1969), was coined for this phenomenon
because of the large temporal gap between successive visible band images from the early
polar orbiters, which gave the appearance of sudden and rapid development of the mature
stage of a frontal cyclone. Because frontogenesis occurs simultaneously with cyclogen-
esis in instant occlusions, these systems were believed to manifest the Type B cyclogenesis
mode advanced by Petterssen and Smebye (1971) (Streten and Troup, 1973). In this mode
the strong low-level temperature advection typically associated with a developing frontal
wave (or Type A development) is absent or weak, and cyclogenesis proceeds mainly from
the import of baroclinicity (i.e. PVA) in the mid to upper troposphere. Consequently,
thermal advection is weak to begin with, but intensifies as the system evolves (i.e. fron-
togenesis and cyclogenesis occur concurrently in the Type B model). Case and composite
analyses of the instant occlusion (Browning and Hill, 1985; Carleton, 1985a; Businger
and Walter, 1988; Pearson and Stewart, 1994) confirm that this cyclogenesis can occur
quite rapidly, and often leads to a deep extratropical cyclone.
Studies of cold-air cyclone systems conducted for north-west Europe and the west coast
of North America (Harley, 1960; Lyall, 1972; Mansfield, 1974; Monteverdi, 1976; Mullen,
1979, 1983) confirmed the generally strong baroclinicity of comma clouds throughout the
middle and upper troposphere. This was particularly apparent in the composite statistical
“models” of southern hemisphere extratropical cloud vortices developed by Troup and
Streten (1972), and in which the comma cloud (their Type A vortex) was the most strongly
represented subsynoptic-scale vortex type. This system is associated with negative depar-
tures of SLP and tropospheric height, implying a cold-cored vortex.
The acquisition of higher resolution imagery with successive generations of polar
orbiters, such as the DMSP, showed that mesoscale cloud vortices occurred over a wide
range of length scales, down to about 100 km diameter. The characteristic sizes of meso-
cyclones and synoptic-scale cyclones over the southern oceans, identified using DMSP
infrared imagery, show a statistically significant difference (Carleton, 1995). The modal
class of mesocyclone cloud vortex has a diameter of around 240–460 km, and 680–880
km for frontal cyclones; there is a lack of vortices in the diameter range 460–680 km.
This is suggestive of a possible “spectral gap” in cloud vortex size for the southern oceans.
However, a similar satellite-based analysis of extratropical cloud vortices over the northern
0
0
0
0
0
11
11
11
Plate 8 DMSP infrared image of an instant occlusion about to be initiated on a frontal cloud band associated with a North Pacific extratropical cyclone
(year and date unknown). The area of enhanced convection associated with a cold-air mesocyclone is merging with the band at approximately 36°N,
164°E. (From Carleton, A.M., 1985, “Satellite climatological aspects of the ‘polar low’ and ‘instant occlusion’,” Tellus, 37A (5): 436–7, 438–40)
1 hemisphere oceans suggests the absence of such a spectral gap: frontal cyclones appear
to be smaller and mesocyclones larger than their southern hemisphere counterparts
(Carleton, 1996).
The higher resolution satellite data also confirm that comma clouds are one of several
classes of oceanic mesoscale cyclones, albeit often the largest. Cold-air mesocyclones
located deep within the cold air far from the jetstream, often develop just equatorward
of the sea ice edge. These mesocyclones are known under a variety of names, including
“arctic lows,” “arctic hurricanes,” or “true polar lows” (Emanuel and Rotunno, 1989;
Businger, 1991; Rasmussen, 1981; Rasmussen et al., 1996). They have a spiraliform rather
than comma cloud signature (Rabbe, 1987; Nordeng and Rasmussen, 1992), sometimes
with a clear “eye” reminiscent of a tropical cyclone (Rasmussen, 1989). The signature
differences between comma cloud and spiraliform mesocyclones suggest the importance
of different physical processes, or at least different relative contributions of the same
processes, in their formation (Carleton, 1996). Diagnostic case studies of Arctic lows have
mostly been undertaken for the East Greenland and Norwegian seas (Shapiro et al., 1987;
Shapiro and Fedor, 1989; Rasmussen et al., 1992; Douglas et al., 1995). The spiraliform
signature type seems to occur more frequently there than in the Bering Sea of the western
Arctic, the Sea of Japan, or in the Subantarctic – apparently because of the strong SST
gradients located close to the sea ice edge, which enhance the surface–atmosphere fluxes
of latent and sensible heat when cold air moves out across the ice edge in winter. Case
studies show a lack of deep baroclinicity associated with these spiraliform systems (e.g.
Rasmussen, 1979, 1981). Rather, the baroclinicity tends to be confined to a shallow surface
layer that has moved out over warmer water to form a “boundary layer front” (Fett, 1989;
Shapiro and Fedor, 1989). This shallow layer is surmounted by a barotropic cold core
low in the upper troposphere in which air is subsiding and warming adiabatically (Bresch
et al., 1997). The latter process effectively limits the mesocyclone to the lower to mid-
troposphere in most cases. Convection occurring along the boundary layer front is
enhanced by the addition of heat and moisture from the upper ocean by the process of
ASII (Air–Sea Interaction Instability) (Emanuel and Rotunno, 1989). ASII is the high-
latitude variant of CISK (Conditional Instability of the Second Kind), which predominates
in the tropics (see p. 183). CISK involves the organization of simple cumulus into a
cyclonic center that becomes self-sustaining through latent heat release in the clouds, and
divergence aloft, with subsidence surrounding the growing vortex. The diabatic processes
encourage frictional convergence of the wind at lower levels, or the “spin up” of a cyclonic
vortex such as a tropical cyclone. In ASII the presence of cold and stable, rather than
warm and conditionally unstable, air prevents the deep convection typified by CISK
(Bratseth, 1985). Moreover the boundary layer front near the ice edge is clearly also
different from the barotropic conditions associated with tropical system development.
The latent heat release within the convective clouds associated with spiraliform “polar
lows” is believed to be responsible for the “warm pool” often observed in the 1,000–700
mb thicknesses (Rasmussen, 1981). However, a competing theory is that this feature results
more from a seclusion type of process, similar to that observed in some synoptic cyclones
(Montgomery and Farrell, 1992). Both the “warm pool” and the multiple spiral cloud
bands associated with polar air vortices are reminiscent of tropical cyclones, although the
latter are larger systems because of the weaker Coriolis force in lower latitudes.
The recognition of the two broad classes of cold air mesocyclone (comma cloud, spiral-
iform) on the basis of satellite image interpretation (Plate 9) is pivotal to the cloud vortex
classification systems developed subsequently for use in climatological studies (e.g. Forbes
and Lottes, 1985; Carleton, 1985a; Carleton and Carpenter, 1989, 1990; Yarnal and
Henderson, 1989a, b). Composite models of these systems in the northern hemisphere,
developed through the averaging of SLP and tropospheric height anomaly fields for
multiple cases (Carleton, 1987), confirmed the more baroclinic (barotropic) character of
the comma clouds (spiraliform systems). However, for mesocyclones in subantarctic lati-
11
11
(a)
0 (b)
Plate 9 DMSP infrared images of (a) a spiraliform mesocyclone and (b) a comma cloud meso-
11 cyclone, in the Labrador Sea during January 1979 (exact dates unknown). (From Carleton,
A.M., 1985, “Satellite climatological aspects of the ‘polar low’ and ‘instant occlusion’,”
Tellus, 37A (5): 436–7, 438–40)
1
(a)
(b)
Figure 6.29 Composite SSM/I retrievals of (a) integrated water vapor (kg m2 ) and (b) near-
surface wind speed (m s1) for twelve comma-cloud mesocyclones in the mature stage
over the southern oceans during 1992. The infrared-observed position of the meso-
cyclone is at the center of each grid. Left mean values. Right standard deviation. (From
Song and Carleton, 1997)
tudes during the cold season, passive microwave data, and also the TOVS-derived geopo-
tential heights, suggest that some degree of baroclinicity may be present in both signature
types (Carleton et al., 1995). Antarctic mesocyclones observed in the summer are more
barotropic (Turner et al., 1993) because they either occur over cold ocean or ice surfaces
where the air is very stable, or where air moving out from the Antarctic ice sheet under-
goes column stretching and an increase in its cyclonic vorticity (Carrasco and Bromwich,
1997a; Heinemann, 1990). These systems are important snowfall producers for some
coastal areas, such as the western Ross Sea, in all seasons (Rocky and Braaten, 1995).
Song and Carleton (1997) determined the prominently recurring patterns of SSM/I-
derived water vapor and surface wind speeds for different development stages of southern
ocean “comma cloud” mesocyclones, using the “storm-following” technique of Businger
(1990). The mean patterns for the mature stage of mesocyclones (Figure 6.29) show a
latitude gradient of water vapor with evidence of the ingestion of warm air into the storm
center, and strong cyclonic shear of wind speed from the northwest to southeast quad-
rants. The standard deviation fields of the retrieved fields show remarkably small variation,
suggesting that some degree of confidence can be placed in the mean mesocyclone micro-
wave patterns. However, the sample of cases used in the composites should be increased
to confirm their representativeness more generally.
The climatological spatial distributions of cold air mesocyclones (comma cloud polar
lows) were known first for the southern hemisphere, given the heavy reliance on the inter-
pretation of satellite cloud data for synoptic analysis (Streten and Troup, 1973; Carleton,
1979). Subsynoptic-scale cyclones occur more frequently with increasing latitude over the
11
11
Figure 6.30 Monthly frequency distribution of “polar lows” in the Norwegian and Barents seas,
based on synoptic data for the period 1978–82. (From Wilhelmsen, 1985)
0 southern oceans, possibly comprising up to 50 percent of all cyclogenesis events in the

winter season. This has helped to revise our impression of the subantarctic trough as being
a predominantly cyclolytic region, to one in which cyclogenesis also frequently occurs
(Carleton, 1992). However, southern hemisphere cold-air mesocyclones are most abun-
dant around the equinoctial months (March, September/October) with a secondary peak
in midwinter (July) (Lyons, 1983; Carleton, 1995, 1996; Carleton and Song, 1997). This
seasonal frequency pattern resembles the semi-annual oscillation (SAO) in tropospheric
temperatures, pressure/heights and wind speed over subantarctic latitudes (van Loon, 1967;
van Loon and Rogers, 1984). The SAO gives rise to extrema in the latitude locations
of the circumpolar trough, such that it is closer to (further from) Antarctica in the equi-
0 noctial (solstitial) months. This contrasts strongly with the simpler annual pattern of polar
low development in the North Pacific and North Atlantic, in which maximum frequen-
cies occur during the winter season, although there is a secondary minimum evident in
February (Wilhelmsen, 1985; Businger, 1987) (Figure 6.30). The minimum occurs around
the time of maximum latitude displacement of the westerly circumpolar vortex, and its
baroclinic zones, away from the regions favorable to polar low formation.
Carleton and Carpenter (1990) and Turner et al. (1996) indicate that spiraliform meso-
cyclones are more frequent than comma clouds in the Subantarctic, but estimates
differ as to their relative abundance. Favored areas for the formation of polar meso-
cyclones in the southern hemisphere are in longitudes of the Ross Sea, the Weddell Sea,
0 and the Amundsen–Bellingshausen seas (Bromwich, 1991; Carrasco and Bromwich, 1994,
1996; Carrasco et al., 1997a, b; Heinemann, 1990, 1996; Turner and Thomas, 1994; Turner
et al., 1996). In all these areas cold air frequently sweeps out from the Antarctic ice sheet
or sea ice zone, sometimes via the high-speed katabatic winds in the boundary layer,
which are evident in satellite infrared images as warmer “plumes” of air overlying the
colder ice surfaces (Bromwich, 1989, 1991; Bromwich et al., 1993). These are also favored
areas for the stagnation of old cold-cored synoptic lows within which mesocyclones
may develop and move cyclonically as a so-called “merry-go-round” formation type (e.g.
Zick, 1983; Forbes and Lottes, 1985). Composites of the SLP and tropospheric height
fields associated with “outbreak” times (i.e. synoptic climatology Type 3, above) confirm
0 these synoptic-scale associations of mesocyclones, although there are regional differences
(Fitch and Carleton, 1992; Carleton and Fitch, 1993). For example, mesocyclones in the
Ross Sea tend to develop in the enhanced thickness gradient (i.e. baroclinic southerly
11 thermal wind) of a cold-cored low located to the northeast, whereas those in the
1 Bellingshausen–Amundsen seas predominantly occur well within a cold-cored low (i.e.
barotropic atmosphere).
The interannual variability of cold-air mesocyclone frequency is large, in accordance
with the variations undergone by the larger synoptic circulation. For example, the southern
winters of 1988 (1989) were characterized by large numbers of mesocyclones in the Ross
Sea (Amundsen–Bellingshausen seas). These major changes in mesocyclone frequencies
between the two years accompanied strong shifts in the planetary waves and, hence, in
the preferred longitudes of cold-air outbreaks and the greatest extent of the antarctic sea
ice (Carleton and Fitch, 1993). In 1988 (1989) an enhanced trough, given by negative
departures of the monthly-averaged SLP and upper tropospheric heights, was located in
the Ross Sea (Amundsen) sectors. There may also be a “signal” of the El Niño Southern
Oscillation (ENSO) in these interannual mesocyclone variations over middle and higher
southern latitudes. The 1988 (1989) winter followed an El Niño (La Niña) phase of ENSO.
The intensity of the Amundsen Sea “mean” low varies according to the phase of ENSO
(Chen et al., 1996; Cullather et al., 1996), and this influences the longitudes of frequent
cold-air outbreaks and surface air temperature anomalies over the Antarctic Peninsula
(Marshall and King, 1997). The low is weaker (stronger) during El Niño (La Niña), and
accompanies an out-of-phase association in the intensity of the STJ and PFJ of the South
Pacific region: the STJ (PFJ) is stronger (weaker) in the El Niño, and vice versa for La
Niña. Moreover, Carleton and Carpenter analyzed southern hemisphere DMSP infrared
mosaics for mesocyclones during the southern winters (June through September) of
1977–83. In the winter leading up to the development of the major ENSO warm event
of 1982–83, large numbers of mesocyclones occurred south-east of Australia and around
New Zealand, consistent with the amplification in the mean of the seasonal cycle of the
Tasman Sea trough during ENSO events (van Loon and Shea, 1985). Conversely, in the
winter before the warm event (winter 1981), mesocyclone activity was reduced markedly
in the New Zealand region, while far greater frequencies occurred about 90o of longitude
to the west. While this pattern is consistent with the reduced annual cycle of the Tasman
trough in year 1 of warm events (van Loon and Shea, 1985), satellite studies of meso-
cyclone activity during additional ENSO events (both “warm” and “cold” extremes) need
to be undertaken to confirm these synoptic and teleconnective associations.
The only climatological study of the “instant occlusion” mode of satellite-observed
extratropical cyclogenesis is for the southern oceans and covers five winters (Carleton,
1981b). Highest frequencies of this phenomenon occur in the latitude zone 45o–50oS,
which is the zone of maximum overlap of the frontal wave (Troup and Streten, 1993, W
category) systems over lower middle latitudes and the comma cloud (A type) vortices
over higher-middle latitudes, in this season. The analysis of higher resolution DMSP
imagery for the Ross Sea sector (Carleton, 2001) confirms the occurrence of instant occlu-
sions north of 60oS. Spatially, most instant occlusions occur to the southwest and south
of Australia, which is west of the area in the southern hemisphere characterized by the
highest frequency of blocking (the Australia–New Zealand sector). Frontal waves
approaching the block are slowed, increasing the opportunity for the interaction with cold-
air comma clouds approaching from the west.
Synoptic climatologies of cold-air mesocyclones using satellite data number fewer for
the northern hemisphere and are more recent than those in the southern hemisphere. This
reflects, at least in part, the better coverage by conventional synoptic data. Reed (1979)
developed the first satellite-based synoptic climatology of comma clouds for the eastern
North Pacific using two years of infrared mosaics from the NOAA Scanning Radiometer.
Yarnal and Henderson (1989a, b) analyzed DMSP infrared images for six cold seasons
(November through March) to document the synoptic climatology of mesocyclones over
the entire North Pacific. Their study confirms the strong seasonality of mesocyclone occur-
rences in the northern hemisphere documented in other studies (Wilhelmsen, 1985; Lystad,
1986; Businger, 1987; Ninomiya, 1989). Despite the strong intraseasonal and interannual
11 variations in frequency and patterns of mesocyclogenesis, the majority of systems tend
to develop off the Asian land mass in the zone of strong baroclinicity associated with the
jetstream, then track southeastwards to the mid-Pacific and subsequently northeastward
into the Aleutian mean low cyclolytic region. Interestingly, the winter preceding the major
El Niño event of 1982–83 recorded the smallest number of cold-air mesocyclones for the
North Pacific basin as a whole. Whether this climatological feature of mesocyclones is
typical of ENSO warm events, or whether the unusual strength of the 1982–83 event
atypically influenced the development patterns, remains to be determined.
The synoptic climatological context of cold air mesocyclones in the North Atlantic
0 region (Forbes and Lottes, 1985; Carleton, 1985a) reflects a tendency for these systems
to develop in groups as “outbreaks” associated with anomalous atmospheric circulation –
typically highly meridional wave patterns. Accordingly, there also appears to be a link
with the dominant atmospheric teleconnection pattern of this region, the North Atlantic
11 Oscillation (NAO), on seasonal and interannual time scales (Carleton, 1985a). There were
marked changes in the regions of mesocyclone formation, tracks, and dissipation between
two winters of contrasting sign in the index used to characterize the NAO (1974/75: posi-
tive; 1976/77: negative). These changes were similar to the composite shifts in the
planetary waves associated with this teleconnection pattern described for the middle and
higher-latitude regions of the North Atlantic sector by Rogers and van Loon (1979).
0 Carleton (1988b) subsequently showed that the satellite-observed changes in frequency
and locations of both the synoptic and mesoscale cyclones between these two years are
consistent with the composite mean patterns of sensible heat transport for the respective
extremes of the NAO. That is, there were considerably more (fewer) cloud vortices in
the positive (negative) mode of NAO, and the peak frequency was shifted poleward (equa-
torward) in those winters. Clearly, again, the data for more years should be studied to
confirm these possible NAO–mesocyclone associations.
The synoptic climatology of cold-air mesocyclone outbreaks in the North Atlantic and
North Pacific regions has also been described where the satellite data were not the primary
data source indicating the occurrence of these systems (Businger, 1985, 1987; Ese et al.,
0 1988). These studies confirm that the conditions necessary for cold-air mesocyclogenesis
involve an area of negative heights in the mid to upper troposphere that moves equator-
ward prior to the development of mesocyclones, a high-amplitude tropospheric wave
pattern, and the rapid advection equatorwards of cold air just to the west of the anomalous
trough. Moreover, these conditions are similar to those necessary for meso-cyclogenesis
over some higher southern latitude regions (Fitch and Carleton, 1992; Carleton and
Fitch, 1993; Turner and Thomas, 1994; Carleton and Song, 1997), particularly the
Bellingshausen–Amundsen Sea sector and the Weddell Sea.
6.5.5 The Mesoscale Convective Complex

0
A corollary to the cold air mesocyclone that predominates over land in the warm season is
the Mesoscale Convective Complex (MCC) (Maddox, 1980). The MCC is a class of multi-
celled thunderstorm organized as a cluster, and contained within the broader phenomenon
of Mesoscale Convective Systems (MCSs) (Plate 10). MCSs include MCCs and also
thunderstorms organized linearly as squall lines, in both the tropics and the middle latitudes
(Zipser, 1982; Lau and Crane, 1995). In the tropics they make up much of the ITCZ, which
is particularly prominent over land and in the summer hemisphere (Machado and Rossow,
1993; Machado, et al., 1993). The tropical squall line or squall cluster has a line of
convective cells extending laterally some hundred(s) of kilometers. It is accompanied
0 by a strong wind squall and heavy rainfall, which is followed by a wide band of steady
precipitation from upper-level stratiform cloud (Cotton and Anthes, 1989, p. 595).
MCCs were first identified as discrete systems for the US Great Plains (Fritsch and
11 Maddox, 1981). They deliver very heavy rainfall and also hail, cause flooding and strong
1
Plate 10 GOES-E enhanced infrared image showing a large MCS over the central United States
(Arkansas–Missouri–Illinois region) on August 13 1982. (NOAA)
winds, including derechos, and even spawn tornadoes (Maddox, 1980). It is estimated
(Fritsch et al., 1986) that MCSs (i.e. MCCs and other convective weather systems) are
responsible for about 30–70 percent of the precipitation falling on the Great Plains in the
April through September period, and even more during the summer (June–August). While
the convective cells contribute intense rain and hail showers, precipitation also falls from
the deep upper stratiform layer formed by the cloud anvils.
Similar systems to those in the Great Plains have now been cataloged for other mid-
latitude and also tropical and subtropical regions (Velasco and Fritsch, 1987; Miller and
Fritsch, 1991; Laing and Fritsch, 1993a, b). For example, it is now believed that much
of the precipitation falling in the southwest United States summer rainfall singularity (or
“monsoon”) of July and August is associated with organized thunderstorm complexes
having much in common with the MCCs of the central United States (Perry, 1990;
McCollum et al., 1995). Moreover, MCCs occurring in the low latitudes may provide an
important “seed” circulation for some tropical cyclones (Velasco and Fritsch, 1987; Laing
and Fritsch, 1993a, b).
Like cold-air mesocyclones, MCCs were first characterized using satellite images –
specifically, those obtained in the thermal infrared (IR) wavelengths, which have been
enhanced to accentuate the lowest temperatures associated with the highest cloud tops.
An example of this type of image is shown in Plate 5, which is a view from GOES-E
for the early morning of June 25 1988. The highest and coldest clouds are identified by
the enhancement which repeats the gray scale in cloudy areas. Notice how these very
11
11
Plate 11 GOES-W enhanced infrared image showing an MCC in central Arizona during the summer
“monsoon” season on August 12 1982. The system developed over the elevated topo-
graphy of the Mogollon Rim, and both enlarged and moved to the northwest in the ensuing
0 several hours. (NOAA/National Weather Service)
cold cloud tops occur in both the tropics (associated with the ITCZ) and over land (e.g.
over northern Arizona and the upper Midwest) at this time.
Mesoscale convective systems tend to peak in intensity (i.e. they exhibit the greatest
horizontal extent of the cold cloud shield and lowest cloud top temperature) at night,
when the vertical development of the system is enhanced by cooling aloft through long-
wave radiation loss from the thunderstorm tops, and by an intensified low-level jet (LLJ)
advecting moist air poleward in the planetary boundary layer (Anderson and Arritt, 1998).
0 Accordingly, the cloud-radiative forcing due to MCSs may comprise a crucial determi-
nant of the Earth–atmosphere energy balance over a wide range of latitudes in the warm
season. Mesoscale convective systems interact dynamically and thermodynamically with
their surroundings. They result from the larger-scale atmospheric environment, and also
help to modify it via strong and widespread vertical transports of heat and moisture (cf.
Maddox, 1983; Read and Maddox, 1983). Like cold-air mesocyclones, MCSs may be
concentrated temporally and spatially as “outbreaks” due to the persistence of synoptic
conditions favorable to their development (Wetzel et al., 1983; Leary and Rappaport,
1987). Mesoscale convective complexes typically endure for around ten to twelve hours,
although a small proportion of systems may persist for up to two or even three days
0 (Wetzel et al., 1983). The individual thunderstorms comprising an MCC have lifetimes
of just an hour or two.
Maddox (1980) developed a set of criteria for identifying MCCs from satellite-enhanced
11 infrared images, and for characterizing their successive stages of development (Table 6.1).
1 Table 6.1 Criteria for identification of a Mesoscale Convective Complex based upon analyses of
enhanced infrared satellite imagery
Criterion Physical characteristics

Size Cloud shield with continuously low infrared temperature ≤ 52°C
must have an area > 50,000 km2
Initiation Size definition first satisfied
Duration Size definition must be met for a period of six hours or more
Maximum extent Contiguous cold cloud shield (infrared temperature ≤ –52°C)
reaches maximum size
Shape Eccentricity (minor axis/major axis) ≥ 0.7 at time of
maximum extent
Temination Size definition is no longer satisfied
Source: Maddox (1980); modified by Anderson and Arritt (1998).
Like the frontal cyclone and cold-air mesocyclone, MCCs exhibit a well defined temporal
evolution in the imagery that can be used to classify them into genesis, mature, and dissi-
pation stages. The synoptic fields within which an MCC is embedded also change between
these different stages (Maddox, 1983). The Maddox criteria emphasize the meso-alpha
(i.e. larger subsynoptic) scale of the MCC, and include the areal extent of the coldest
cloud tops, the duration of the system, and its shape. In the enhanced infrared imagery
the coldest cloud tops are located in the central part of the system, as shown in the
example for the southwest United States (Plate 11). At maturity the area covered by the
coldest cloud tops (often <52°C) reaches its largest extent and the MCC is at its most
elliptical (around 0.7: Table 6.1). Augustine and Howard (1988) modified Maddox’s orig-
inal criteria by excluding weaker systems, whose cloud-top temperatures may be lower
than 32°C but not reach the “critical” threshold temperature of 52°C established by
Maddox. Other authors (McAnelly and Cotton, 1989; Anderson and Arritt, 1998) have
made more minor modifications to the basic satellite-based criteria defining MCCs.
The longer horizontal axis of an MCC is parallel to the direction of movement of the
system, which is to the right (left) of the upper steering wind in the northern (southern)
hemisphere. Accordingly the MCC exemplifies a class of right-moving storm in the
northern hemisphere. For middle-latitude systems this means that the MCC typically moves
from north of west to south of east, but always into a region of warm, moist, and unstable
air. This net motion is the vector difference of the direction of movement of new thun-
derstorm cells forming on the warm side of the complex (the so-called propagation
component), and the advective component, which is the mean motion of cells comprising
the system (Corfidi et al., 1996). In effect the MCC moves with the “thermal wind.”
The availability of satellite-enhanced infrared images from geosynchronous platforms
has meant that MCCs have been identified, and synoptic climatologies of their occurrence
compiled, for large areas of the tropics and middle latitudes. As with the study of cold-
air mesocyclones, these regional and larger-scale climatologies of MCCs have typically
taken three forms:
1 Composite analyses of standard meteorological fields, including precipitation (Kane

et al., 1987; McAnelly and Cotton, 1989), stratified according to the characteristic
temporal stages of development of MCCs denoted on the infrared images (Maddox,
1983).
11 2 Mapped distributions of MCCs for a particular season or multi-year composite of
seasons, such as those obtained by tracking the movements of the “centroid locations”
(central point of the coldest cloud-top temperatures) of many such systems (Rodgers
et al., 1983, 1985; Augustine and Howard, 1988, 1991; Anderson and Arritt, 1998).
3 Composites of the larger-scale synoptic fields within which are embedded MCCs for
time periods of their occurrence (Augustine and Howard, 1991).
The system-composite studies of MCCs (category 1 above) permit the temporal evolu-
tion of these storms to be analyzed. For example, the composite precipitation patterns of
0 MCCs over the US Great Plains generated by McAnelly and Cotton (1989) reveal that
the heaviest precipitation occurs about 50–100 km equatorward of the cloud-shield
centroid. Moreover, these types of studies help reveal the dominant forcing mechanisms
of MCCs. The MCCs that form in middle latitudes in summer show some baroclinic
11 attributes, although they are largely thermally driven (Fritsch and Maddox, 1981). Thus
many MCCs exhibit a warm core in the mid-troposphere resulting from latent heat release
within the clouds, and a cold upper troposphere due to radiation loss from the cloud tops
(Fritsch and Maddox, 1981).
The second group of satellite-based climatological studies of MCCs emphasizes the
locations, tracks, and dissipation areas of MCCs for different regions; initially North
0 America (Rodgers et al., 1983, 1985; McAnelly and Cotton, 1989; Augustine and Howard,
1991). Over the US Great Plains there is an intraseasonal variation in the locations of
MCC development such that systems are generally displaced to the southward (north-
ward) in the late spring and early summer (mid to late summer). This pattern is connected
with the heating of the land mass and retreat of the upper westerlies of the circumpolar
vortex as the warm season progresses. Interannual variations in MCC development are
apparently a response to larger-scale atmospheric circulation patterns (Rodgers et al.,
1985) and, possibly also, teleconnections with ENSO.
Velasco and Fritsch (1987) extended the use of GOES-E data to the study of MCCs
in North and South America, as well as the tropical zone between. These authors found
0 broadly similar features associated with MCCs in the middle-latitude areas of both conti-
nents, except that the cloud shield is about 60 percent larger in the Argentinian systems.
The frequency of MCCs showed strong variations between the two years that they studied
(May 1981–May 1983), and these appear connected with the major ENSO event of
1982–83. In middle southern latitudes more than twice the number of MCCs formed in
the El Niño year (1982–83) compared with the non-El Niño year (1981–82). There were
also considerably more MCCs that formed over the anomalously warm water off the
Peruvian coast associated with the El Niño event. In common with the Brazilian rain-
forest region of South America (Velasco and Fritsch, 1987), relatively few MCCs develop
over the equatorial rain-forest of central Africa despite the large amount of deep convec-
0 tion occurring there and the presence of the ITCZ. This observation emphasizes the
differences between MCSs and MCCs (cf. Machado et al., 1992).
Satellite-based climatologies of MCCs have now been undertaken for almost all regions
of the globe. Comparisons of the infrared-derived signatures of MCCs in these different
regions with those occurring in the Americas suggest strongly that they are all the same
class of phenomenon. A global summary developed by Laing and Fritsch (1997) considers
MCCs occurring over South and Central America (Velasco and Fritsch, 1987), the United
States (Augustine and Howard, 1991), Africa (Laing and Fritsch, 1993a), India (Laing
and Fritsch, 1993b), and the western Pacific Ocean and adjacent areas (Miller and Fritsch,
1991); they also extend the analysis to Europe. Approximately 400 MCCs occur annu-
0 ally around the globe, with two-thirds in the northern and one-third in the southern
hemisphere, related to the land–sea fractions. The great majority (91.6 percent) form over
land areas. The systems are predominantly nocturnal, with the cold cloud shield attaining
11 its maximum extent of about 3 × 105 km2 in the early morning hours. They persist for
1
Figure 6.31 The global occurrence of MCCs and regions of widespread, frequent deep convec-
tion, indicated be outgoing long-wave radiation (ORL). The shaded areas indicate OLR
minima. The OLR data (W m2) are from the ERBE for (a) July (north of the equator)
and January (to the south) 1985–86 and (b) June–August (north of the equator) and
December–February (to the south) 1974–78. (From Laing and Fritsch, 1997)
about ten hours, but the oceanic systems tend to end later in the morning than those over
land. In summer there is a positive correlation (0.57) in mid-latitudes between cloud area
and duration, illustrating the fact that systems with large cold-cloud shields tend to be
more persistent.
The geographical distribution of MCCs (Figure 6.31) shows two preferred zones of
occurrence: (1) on the peripheries of minima in outgoing long-wave radiation and (2) in
the lee of mountain ranges and high plateaus, relative to the prevailing mid-tropospheric
airflow. Seasonally, the region of activity over the continents shifts between 35°S in the
austral summer and north of 50°N in July, and this tendency is matched by the occur-
rence of midnight lightning, reported by Goodman and Christian (1993) and Barry et al.
(1994).
Higher frequency passive microwave sensing, particularly the 85 GHz dual-polarized
channels of the SSM/I (section 2.2), readily lends itself to the study of MCS-type features
(Adler et al., 1991; Mohr and Zipser, 1996a), because these channels are sensitive to the
scattering of microwave radiation by large ice particles and snow aggregates in the upper
parts of precipitating cumulonimbus clouds (McGaughey et al., 1996). Mohr and Zipser
(1996b) used the 85 GHz channels to determine the basic distribution and characteristics
11 of MCSs (i.e. all classes of mesoscale convective system) occurring over land and sea in
the tropical zone. They found the greatest frequency of MCSs over tropical South America,
tropical Africa, and the western Pacific “warm pool,” The smallest (largest) MCSs occur
over equatorial land regions, including tropical Africa (subtropical oceans). MCSs over
land (ocean) tend to have the coldest (warmest) brightness temperatures. Comparisons of
the sunset and sunrise orbital data of the SSM/I revealed diurnal variations of MCS
frequency and size that show dependence on their occurrence over land or sea domains
and, accordingly, differences in the extent of ice scattering (McGaughey et al., 1996).
The third group of MCC studies comprises those in which the synoptic environments
0 within which the MCC is embedded can be characterized. From those studies, the domi-
nant forcing mechanisms of MCCs are determined to be as follows:
1 Weak static stability due to their presence in, and movement towards, warm and moist
11 air.
2 Occurrence on the lee side of major mountain ranges, whereby air columns moving
into the plains leading away from the mountains are forced to undergo vertical
stretching to conserve potential vorticity, with convergence at lower levels and diver-
gence aloft. Additionally, the slope of the plains can aid the development of LLJs
and also enhance the ascent of moist air at low levels, both of which are important
0 in MCC development (e.g. Miller and Fritsch, 1991; Leary and Rappaport, 1987).
3 An upper short-wave trough that is evident at the surface as a weak cold front or
stationary front located to the westward or northwestward (Maddox, 1980).
4 A subtropical jetstream in the warm air that curves anticyclonically as the MCC
develops (Maddox, 1983). This advects subtropical moisture into the region, and has
associated divergence aloft, with convergence at lower atmospheric levels.
5 An LLJ concentrated between about 950 mb and 850 mb that advects heat and mois-
ture within the boundary layer into the storms on the warm side of the complex. The
LLJ is typically diurnally varying, being strongest just below the top of the boundary
layer when the surface-based inversion is strongest (i.e. at night and in the early
0 morning). Accordingly the LLJ helps to initiate new thunderstorms on the warm side
of the MCC, in association with the weak stability generated by the outflow bound-
aries either from existing or from recently deceased thunderstorms within the complex.
It also helps provide a clockwise turning of the wind with height (northern hemi-
sphere), which denotes warm advection into the MCC. Both of these signify a
baroclinic atmosphere. Modeling work for the US Great Plains (McCorcle, 1988;
Chang and Wetzel, 1991; Zhong et al., 1996; Wu and Raman, 1997) suggests that
horizontal contrasts in land surface conditions, particularly soil moisture and vegeta-
tion, may increase the instability and thereby enhance the development and intensity
of the LLJ and, potentially, also, the development of MCCs, under favorable broader-
0 scale circulation conditions.
6.5.6 Tropical-extratropical cloud band connections

The very long bands of deep cloud that develop on synoptic time scales, and which
connect convection in the lowest latitudes with cyclonic circulations over higher latitudes
(Plate 12), were also first identified in satellite imagery (Oliver and Anderson, 1969). The
lower-latitude branch of a tropical–extratropical cloud band connection (Erickson and
Winston, 1972) originates either in clusters of convection or in a tropical cyclone that
has begun to dissipate (Sekioka, 1970; Gray and Clapp, 1978; Davis, 1981); however, it
0 seems that the movement equatorwards of a middle-latitude trough provides the initial
impetus for development of such a connection (McGuirk et al., 1987; McGuirk and Ulsh,
1990). Once initiated, the cloud-band feature transports moisture and energy rapidly pole-
11 wards and eastwards via the STJ (Thepenier and Cruette, 1981), the PFJ over middle
1
Plate 12 DMSP infrared mosaic (5.4 space km resolution) showing a tropical–extratropical cloud
band connection in the western North Pacific during October 1977 (exact date unknown).
The cloud band extends from a typhoon which is dissipating, through several extratrop-
ical cloud vortices, and terminates in the Bering Strait at 60°N. (From Carleton, A.M.,
1985, “Synoptic cryosphere–atmosphere interactions in the northern hemisphere from
DMSP image analysis,” Intl. J. Remote Sens. 6 (1): 245)
latitudes (Zwatz-Meise and Hailzl, 1980), and, sometimes, with a feature resulting from
the merger of the two (Reiter and Whitney, 1969). The poleward terminus of a trop-
ical–extratropical cloud band is usually a cyclonic cloud vortex; however, it is not
uncommon for a series of cyclonic vortices in different stages of development to extend
back along the band in middle latitudes (Plate 12). These tropical–extratropical cloud-
band connections manifest a preferred mode of the atmospheric cross-latitude transport
of energy and moisture, and confirm that it does not take place equally across all longi-
tudes. Rather, this transport occurs rapidly and along relatively narrow zones that can be
as short-lived as a couple of days (Gray and Clapp, 1978; Thepenier and Cruette, 1981;
Davis, 1981).
Tropical cloud-band connections are known by a variety of names. While Kuhnel (1989)
referred to them more generally as Tropical–Extratropical Cloud Bands (TECBs), they
are also differentiated on a regional basis. Several authors (McGuirk et al., 1987, 1988;
11 McGuirk and Ulsh, 1990; Iskenderian, 1995) term the cloud bands in the central and
eastern North Pacific “moisture bursts” and “tropical cloud plumes,” while those origi-
nating in the Arafura Sea northwest of Australia are known as “northwest Australia cloud
bands” (Downey et al., 1981; Tapp and Barrell, 1984; Bell, 1986). The cloud band in the
South Pacific, or SPCB, is the most frequently occurring and persistent TECB, especially
in the November through May period (Kuhnel, 1989). This feature was prominent in the
early analyses of multi-day composite visible images of the southern hemisphere used for
developing the first satellite cloud climatologies and for determining their associations
with the atmospheric circulation (Kornfield et al., 1967; Streten, 1968a, 1970, 1973; Miller
0 and Feddes, 1971). The SPCB links the area of enhanced convection over the maritime
continent with frontal systems in the southeast Pacific Ocean (Plate 13), and is involved
in the SST variations associated with the ENSO over tropical and subtropical latitudes
(Trenberth, 1976). The role of the SPCB in transporting energy, momentum, and mois-
11 ture poleward has been studied extensively for case periods (Huang and Vincent, 1985).
In Kuhnel’s (1989) study it was present on the infrared imagery on 778 days in five years,
but showed substantial interannual variability associated with the Southern Oscillation
Index (SOI) consistent with other non-satellite studies (Trenberth, 1976). Streten (1978)
found a quasi-periodicity in SPCB location of around twenty to twenty-five days, similar
to the time scales of zonal indices and kinetic energy in the southern hemisphere. The
0 association of the SPCB variations with ENSO was confirmed and extended by Berlin
(1991), who studied the SPCB using daily NOAA hemispheric infrared mosaics for the
period June 1 1980 through May 31 1984. This includes the major warm event of 1982–83.
She confirmed the spring and summer frequency maximum of this band, and also found
that the seasonal-scale shifts were more evident than a forty to fifty-day oscillation, at
least for the tropical reach of the cloud band. Berlin (1991) also found the frequencies
of extratropical cyclones in the South Pacific region to vary inversely with the frequency
of cloud-band days, suggesting a net balance in the contribution to the total poleward
fluxes of heat and momentum by the standing and transient eddies, represented by the
cloud-band and cyclonic vortices, respectively. Interannually, an eastward movement in
0 the cloud band axis commenced in the southern spring of 1982 and reached its most east-
ward location in the summer of 1982–83, coincident with the ENSO warm event. This
pattern is similar to that found for the 1972–73 El Niño event by Streten (1975). Also
occurring during the 1982–83 event was an increase in the frequency of satellite-observed
extratropical cyclone activity over the South Pacific (Berlin, 1991).
There are fourteen separate TECBs for the globe, seven in each hemisphere, sometimes
having different seasons of occurrence (Kuhnel, 1989). While TECBs are typically a cool-
season phenomenon, some have a late autumn maximum, and still others, like the SPCB
and that over southern Africa, may occur relatively frequently in summer (Erickson and
Winston, 1972; Harrison, 1984). This suggests that TECBs fulfill slightly different func-
0 tions with respect to the atmospheric general circulation. Erickson and Winston (1972)
studied the occurrence of TECBs in the western North Pacific, and showed that the
frequency of cloud bands increases in the late summer and early fall seasons, coincident
with the build-up of the westerly circulation in middle latitudes. Since many tropical
storms and typhoons are observed to decay upon interacting with these cloud bands at
these times, the implication was that TECBs facilitate the rapid poleward transport of
energy and moisture out of the tropics and into the extratropics of the northern hemi-
sphere.
The “moisture bursts” of the central North Pacific that develop near the dateline, prop-
agate across North America (McGuirk et al., 1987, 1988; McGuirk and Ulsh, 1990), and
0 often extend into the northern North Atlantic (Thepenier and Cruette, 1981), are even
more transient features. Half the bursts develop and dissipate over a period of between
two and four days. Depending on the latitude at which moisture bursts enter North
11 America, they may bring heavy precipitation into either the US west coast or across Texas
1
Plate 13 Infrared mosaic of the southern hemisphere from the NOAA SR (Scanning Radiometer),
8 km resolution, for June 17 1975. The image shows a well developed SPCB extending
from the area near northeast Australia south and eastwards into the Bellingshausen/
Amundsen seas. (NOAA)
and the southern Gulf states (e.g. Dickson, 1973). When penetrating the interior regions
in midwinter, dominated by air of Arctic origin, moisture bursts can be associated with
severe ice storms (Plate 14). McGuirk et al. (1987) studied moisture bursts from GOES-
W imagery for the cool seasons (November–April) of 1975–76, 1977–78, 1981–82, and
1982–83. They found a substantial interannual variation in the frequency and longitude
locations of moisture bursts that suggested an association with the ENSO teleconnection.
The hypothesis of McGuirk et al. (1987) was tested in a ten-year climatology developed
for the October through May periods of 1974–84 by Iskenderian (1995). That period
recorded 1,062 “tropical plume” events. Composite analysis of streamline data showed
that cloud plumes tend to occur during the preferred times of westerly wind ducts in low-
latitudes, which develop as a low latitude trough extends towards the region of plume
origin, and a trough from the southern hemisphere also extends northwards. Tropical
plumes also exhibit substantial intraseasonal variability, from a maximum in the October
and May months but a short minimum in February and March. Moreover, there is a
tendency for plume frequency to vary out of phase between the eastern Pacific and central
Atlantic on a seasonal basis. While Iskenderian (1995) confirmed the strong decrease in
plume events during 1982–83 found by McGuirk et al. (1987), no similar decrease was
found for the 1986–87 ENSO warm event. In fact, plume frequencies were above average
in the eastern Pacific for that season. Thus the true association of moisture bursts and
ENSO remains to be clarified.
11
11
Plate 14 GOES-E infrared images showing a moisture plume extending from the eastern North
Pacific into the central United States. In moving above arctic air near the Earth’s surface
0 the moisture plume helped to produce a heavy ice storm over southern Illinois and Indiana
on February 14 1990. (NOAA/National Weather Service)
The northwest Australia cloud band is a potent rainfall producer for the inland areas
of central and southeastern Australia in winter (Kininmonth, 1983; Wright, 1988), unlike
the rainfall typically associated with cold frontal passages in this region. Diagnostic case
studies of the northwest Australia cloud band (Bell, 1986) indicate it to be comprised
mostly of middle and high-level clouds generated ahead of an elongated trough in the
westerlies. Air parcels moving poleward on the forward, or eastern, side of the trough
0 ascend more or less moist isentropically, leading to deepening clouds which give exten-
sive pre-frontal precipitation over subtropical and lower middle latitudes of Australia. The
probability of precipitation is increased poleward along the cloud band owing to increasing
11 cyclonic vorticity and also slantwise convection.
1 6.6 Synoptic-scale systems in the tropics
Progress in understanding synoptic behavior in low latitudes spans the last fifty years,
and a number of significant milestones of this period can be identified. The first land-
mark contributions describing synoptic-scale waves in the tropical easterlies were based
on time series analysis of single station sounding data and the use of streamline maps as
a key diagnostic tool (Riehl, 1945, 1954; Palmer, 1950, 1952). It was recognized that the
three-dimensional distribution of mass divergence and the vertical moisture profile are
important determinants of the occurrence of cloud cover and precipitation. A second major
step came in the 1960s with the greatly expanded spatial view provided by satellite
imagery, which brought recognition of the wide variety of tropical disturbances. Merritt
(1964) identified five satellite cloud patterns, only one of which resembled the easterly
wave (Frank, 1969); the others had vortical cloud distributions and included upper lows.
The value of satellite remote sensing was further enhanced with the advent of operational
geostationary satellites over the equator in 1974 (see section 2.2), enabling analyses to
be made of diurnal cloud and convection regimes. A third phase, beginning in the late
1960s, involved the use of spectral analyses of meteorological data to identify wave
periods, propagation, and wavelength characteristics. The fourth phase, which began
almost concurrently and is still continuing, is represented by the mounting of interna-
tional field programs combining measurements from ships, aircraft, and satellites.
These have included the Line Islands Experiment (LIE) in the central equatorial Pacific
in March–April 1967; the Atlantic Tropical Experiment (ATEX) in the central equatorial
Atlantic in February 1969; the Barbados Oceanographic and Meteorological Experiment
(BOMEX) in the western tropical Atlantic during May–July 1969; and the Venezuela
International Meteorological and Hydrologic Experiment (VIHMEX) during June–October
1969 and May–September 1972. Subsequently there were more intensive campaigns orga-
nized as contributions to the Global Atmospheric Research Program (GARP), coordinated
by the World Meteorological Organization and International Council of Scientific Unions
(ICSU). Programs in the tropics included the GARP Atlantic Tropical Experiment (GATE)
off West Africa in June–September 1974, the Monsoon Experiment (MONEX) over
Southeast Asia–Indonesia–northern Australia in January–February 1979, the Arabian
Sea–western Indian Ocean in May–June 1979 (WMO-ICSU JOC, 1976) and the linked
West African Monsoon Experiment (WAMEX) in summer 1979. Subsequently there has
been the Tropical Ocean Global Atmosphere (TOGA) Coordinated Ocean Atmosphere
Response Experiment (COARE) over the West Pacific warm pool in November 1992–
February 1993 (Lukas et al., 1995). A crucial component of GARP is the assembly of
comprehensive and consistent atmospheric fields through the Global Data Assimilation
System (GDAS) operated at the primary international centers for operational weather
prediction. This development has laid the foundations for intensive analyses of global
circulation characteristics and the role of disturbances in combination, inter alia, with
satellite-derived information on clouds, deep convection, and precipitation (Garcia, 1985;
Rasmusson and Arkin, 1993; Mohr and Zipser, 1996a).
In parallel with the enhanced observational capabilities and data sets, advances in statis-
tical techniques and in theoretical work and modeling studies also played essential roles.
The spatiotemporal characteristics of low-latitude circulation and weather have been more
clearly defined by the statistical analysis of time series, as well as by intensive diagnostic
analysis of case studies. Nevertheless, it needs to be pointed out that the period and wave-
length of waves identified in spectral analyses of the variance of wind, temperature, cloud
cover, and so on, do not always coincide with the corresponding values determined from
synoptic information. Burpee (1974), for example, reports a period of four and a half
days and a wavelength of 3,800 km from spectral analysis of meridional wind and pres-
sure data, but three and a half days and 3,100 km, respectively, based on analysis of
forty-three waves over West Africa. He also found stronger signals by composite analysis
than using spectral analysis.
11 6.6.1 Global waves in the tropics
Global waves that have been identified in the tropics on synoptic time scales include a
five-day traveling wave (Deland, 1964; Wallace and Chang, 1969; Misra, 1972; Madden
and Julian, 1972; Madden and Stokes, 1975; Madden, 1978) as well as a five-day standing
oscillation; the latter was first noted over India in the late nineteenth century by J. Eliot
(Asnani, 1993) and subsequently reported by Frolow (1941) from central Africa to Central
America across the Atlantic Ocean, and also over the equatorial Pacific Ocean (Palmer
and Ohmstede, 1956). For the 1957–58 International Geophysical Year, Misra (1972)
analyzed sea-level pressure data at seventy-six stations between 20°N and 20°S. The data
0 show four to five-day oscillations from a westward-moving zonal wave No. 1, in accord-
ance with earlier work by Deland (1964). Madden and Stokes (1975) support this finding
from a cross-spectral analysis of a seventy-two-year summer series of surface pressure
data at 25°N, 90°W, 25°N, 150°W, and 25°N, 30°E. The wave is not evident in the winter
11 series although this could be the result of a lower signal/noise ratio. Wallace and Chang
(1969) indicate a phase speed of 100 m s1 in the surface pressure wave and an ampli-
tude of about 1 mb at the equator. The significance of these features for direct modulation
of weather and climate is evidently modest, but they may have important dynamical inter-
actions in the tropics and extratropics.
Several waves having long wavelengths and periods have been identified in low lati-
0
tudes. They appear to play an important role in the stratospheric quasi-biennial oscillation,
the Walker circulations and thirty-to-sixty-day oscillations in the troposphere. Analytical
studies of these waves by Matsuno (1966) and Longuet-Higgins (1968) are summarized
by Webster (1983) and Asnani (1993).
The pressure and velocity distribution of a theoretical westward-propagating inertio-
gravity wave of wave No. 1 is shown in Figure 6.32 (Matsuno, 1966). An important
feature of the equatorial region (equatorward of about 15° latitude) is that inertio-gravity
waves are reflected as they try to propagate poleward by virtue of the increase of f pole-
ward, deflecting the flow towards the equator. They have a frequency:
0 [f 2 c02 m2]1/2
where c0 wave speed, m zonal wave number, and thus the minimum frequency is
equal to f, the Coriolis parameter. The reflected inertio-gravity waves are trapped in an
equatorial duct and tend to propagate zonally; they have a maximum period of about two
days (Young, 1987). The duct has a width proportional to the Rossby radius of defor-
mation (c0/ )1/2.
Figure 6.32 The pressure and velocity distribution for a theoretical westward-propagating gravity
11 wave, meridional wave No. 1. (From Matsuno, 1966)
1
Figure 6.33 Schematic diagram of equatorial Rossby wave (left), Kelvin wave (center), and mixed
Rossby–gravity wave (right). Lower level shows isobar (H, L centers) and winds;
mid-level shows warm (W) and cold (C) centers and vertical motion. Typical phase
and energy propagation vectors are indicated. (From Young, 1987)
Other important equatorial waves are the Rossby wave and the mixed Rossby–gravity
wave. The Rossby wave is symmetric about the equator, with strongest zonal winds along
the equator, and the circulation is almost geostrophic (see Figure 6.33). Note that the
phase and energy propagate slowly westward. The second type resembles a hybrid between
the Rossby and the inertia-gravity wave. It is asymmetric with respect to the equator and
features strong ageostrophic and vertical motions. The phase propagation is westward at
about 23 m s1, but energy propagates eastward (see Figure 6.33). The figure illustrates
the cross-isobaric flow components near the equator. The mixed Rossby–gravity wave has
a horizontal wavelength of about 10,000 km and a period of four to five days in the lower
equatorial stratosphere (Wallace, 1973).
A further important wave in the equatorial lower stratosphere is the Kelvin wave (Figure
6.33). It features almost geostrophic zonal motion on both sides of the equator and
strong cross-isobaric flow. It has a period of about fifteen days, propagating eastward at
25 m s1, with a wavelength around 30,000 km. The amplitude of the u component is
about 8 m s1 (Wallace, 1973). The Kelvin wave transports energy and westerly momen-
tum upward, whereas the mixed Rossby–gravity wave transports easterly momentum
upward. These act together through wave-mean flow interactions to produce the Quasi-
biennial Oscillation in the equatorial lower stratosphere.
6.6.2 Characteristics of tropical waves

The literature on tropical waves and depressions has helped to create the impression that
there exists a wide variety of synoptic disturbances in low latitudes. Among the models
described are: the easterly wave over the Caribbean (Riehl, 1945), the tropical Atlantic
(Carlson, 1969) and North Africa (Burpee, 1972, 1974); monsoon depressions over India
(Duggupaty and Sikka, 1977); the subtropical cyclone (Ramage, 1962); and wave distur-
bances over the equatorial central Pacific (Palmer, 1952), western Pacific (Reed and
Recker, 1971), and eastern equatorial Atlantic (Thompson et al., 1979). Nevertheless, as
Riehl (1979, p. 321) has emphasized, it is unlikely that there are as many physical mech-
anisms for these storms as there are for superficial differences. The essential element,
except along the subtropical margins, is the conversion of latent heat into potential energy
and then into kinetic energy. This conversion is concentrated in mesoscale convective
systems (MCSs) which show a wide variety of organizational patterns – as squall lines
(Aspliden et al., 1976; Houze et al., 1981; Zipser, 1970), cloud clusters (Mohr and Zipser,
11
11
Figure 6.34 The basic types of zonal wind structure observed in the tropics. The profiles are for:
76°W (Caribbean easterly wave), 12°W (African monsoon cyclone), 75°E (Indian
0 monsoon depression), 135°E (western Pacific ITCZ wave), 170°E (central Pacific ITCZ
wave). (From Krishnamurti, 1979)
1996b), and vortical cloud systems (Merritt, 1964) – and generally occupy areas of only
about 2°–5° latitude radius. Within a given MCS there are individual convective cells
covering about 2,000 km2 (Henry, 1974). The apparent variety of tropical disturbances
seems to be a result of the variations in external controls; the vertical wind structure
differs seasonally and interannually (Riehl, 1973) as well as geographically. Figure 6.34
illustrates the different basic structures that have been reported to occur in different sectors
over the belt 5°–25°N during northern summer. The contrasting profiles of vertical wind
0 shear are readily apparent.
The classical easterly wave structure (Figure 6.35) pictures the trough tilted eastward
with height and identifies maximum cloud cover, convection, and precipitation ahead of
the moving wave trough (Riehl, 1954). This relationship is interpreted in terms of the
distribution of low-level convergence; if the easterlies are moving faster than the wave,
then conservation of potential vorticity requires the air ahead of the wave to gain cyclonic
vorticity to offset the decrease in f, thus the air converges and ascends, i.e., for adiabatic
motion, potential vorticity conservation requires:
∂( f )
0 =K
∂p
An increase of cyclonic relative vorticity following the motion implies increased conver-
gence, since:
∂( f )
= ( f ) ·VH
∂t
At higher levels, where the easterlies are weaker, the air moves eastward relative to the
wave, and anticyclonic vorticity and divergence result ahead of the trough. The structure
0 supports ascending air and convection in association with a deeper moist layer. In the
rear of the trough the opposite structure and descending air are found. However, obser-
vational analysis of many disturbances in the tropical Atlantic shows a different picture.
11 For July 1975 traveling wave disturbances surveyed by Shapiro (1986) show a westward
1
Figure 6.35 The structure of the classical easterly wave in the Caribbean. Vertical time section at
San Juan, Puerto Rico, July 11–13 1944 and corresponding 5,000 ft (1,640 m) winds,
surface weather, and twenty-four-hour pressure change, July 12 1944 (from Riehl,
1954)
11 tilt of the trough with height, with a 90° westward phase shift of meridional velocity and
vorticity at 200 mb, relative to the lower troposphere, equivalent to a one-day lead. The
pattern resembles that in developing tropical cyclones. For summer–autumn 1968 a survey
of thirty-three Atlantic disturbances between 11°N and 15°N shows increased cloud
cover on the wave axis (Carlson, 1969), while a more extensive analysis for summers
1960–64 indicates no preferred location with respect to the wave (Burpee, 1972). However,
in a subsequent study of summers 1968 and 1969, Burpee (1974) reports maxima of
precipitation and thunderstorm frequency west of the trough (succeeding ridge) for systems
to the south (north) of 12.5°N. Analyses of GATE observations show the maximum
0 convection and squall lines to be located just ahead of the trough axis (Aspliden et al.,
1976; Reed et al., 1977) and this is confirmed by Meteosat infrared and water vapor
channel data and ECMWF analyses for June–September 1983–85 over the eastern trop-
ical Atlantic and West Africa (Duvel, 1990). In contrast to the classical interpretation,
11 Duvel attributes the vertical distribution of convergence and divergence to feedback
processes from the vertical motion and convection induced by diabatic heating.
Significantly, he finds that over 70 percent of the intraseasonal variance in circulation,
cloud, and water vapor parameters in the eastern tropical Atlantic is concentrated in the
one to eight day frequency band.
Wave development in low latitudes requires deep easterlies throughout the troposphere.
0 When upper-level westerlies are present, tropical wave activity is substantially reduced.
Deep easterlies are prevalent over West Africa, for example, whereas upper westerlies are
not uncommon over the Caribbean and northern Venezuela (Riehl, 1973), as shown in
Figure 6.34. As pointed out by Frank (1969), the Caribbean is not an ideal region in which
to identify a “pure” easterly wave. The structure of wave disturbances in relation to the ver-
tical shear of the zonal wind has been examined in several modeling studies. For example,
Holton (1971) concluded that westerly vertical wind shear (a trade-wind regime), produces
the classical easterly wave structure, tilting eastward with height, whereas the trough axis
tilts westward with easterly wind shear. The diabatic heat source for the waves in Holton’s
model is specified and there is no feedback from the wave to the heating. Shapiro et al.
0 (1988) use a linear primitive equation model which incorporates a basic state zonal wind
in gradient thermal wind balances, a forced westward-propagating disturbance, and diabatic
heating centered at 400 mb. The tilt of the trough is found to depend on interactions between
the wave and the environment, involving not only the vertical wind structure but also the
latitude of the disturbance. For westerly shear in the lower middle troposphere and maxi-
mum heating at 19°N, westward tilt with height above 700 mb is determined by a down-
ward flux of energy below 200 mb towards the surface. An eastward tilt can occur below
400 mb with heating at 9.4°N; here the westward phase shift is focused just above the max-
imum heating level.
A major source area of tropical waves for the North Atlantic is the sector from 0°W
0 to 100°W in northern Africa. Disturbances in the tropical easterlies over the eastern North
Atlantic were recognized in the 1930s by Piersig (1944). They appear to have their origins
on either side of the mid-tropospheric African Easterly Jetstream (AEJ) located at about
15°N. They tend to develop either around 5°–15°N, 20°E downstream of the Ethiopian
Highlands, or around 20°–25°N, 5°W–5°E, downstream of the Hoggar mountains (Reed
et al., 1988), although it is not thought that topography plays any major role in wave
initiation, in contrast with the generation of diurnal disturbance lines to the south (Eldridge,
1957; Aspliden et al., 1976). Figure 6.36 illustrates wave occurrence in 850 mb stream-
line and vorticity analyses during summer 1985. Systems formed in these regions may
subsequently merge downstream. African waves have their greatest amplitude near the
0 coast of West Africa, between 600 and 700 mb, in association with the 12–15 m s1 AEJ
around 15°N. There is a secondary maximum in wave amplitude at 200 mb. For
June–September 1983–85, 40–50 percent of the spectral amplitude of integrated water
11 vapor, and over 60 percent near the West African coast at 15°N, occurs in the 2.8–5.1
1
Figure 6.36 Wave structure over West Africa; 850 mb streamlines and relative vorticity (dotted
lines) (105 s1), from ECMWF analysis for 12.00 GMT, August 29 1985. H/L denotes
vorticity maxima/minima; a full barb on the wind arrow denotes 5 m s1. Wave K
formed in the proximity of the Hoggar mountains. (From Reed et al., 1988)
day band (Duvel, 1990). The 850 mb v wind in this frequency interval also has maximum
wave amplitude near the coast at 20°N; this amplitude is stronger in August–September
than in June–July. The amplitude of waves over West Africa in summers 1968 and 1969
was about 1.5 m s1 for the surface meridional wind at 19°N and 1 mb for sea-level pres-
sure at 18°N, determined spectrally. Synoptic composites of forty-three waves gave a
stronger amplitude reaching 3 m s1 for v and 1°C for temperature at 700 mb. In GATE
cases (Reed et al., 1977), ascent averaged 5 mb/hr at 700 mb ahead of the trough. However,
the associated cloud amount was only about 40 percent and the precipitation averaged
10–20 mm/day. The waves move westward at about 8 m s1 along zonal paths. The waves
studied during GATE were cold-core systems below 650 mb and warm-core above 250
mb. Particularly on the equatorward side of the AEJ, the waves show a southwest–north-
east tilt which is consistent with growth through barotropic energy conversion (Norquist
et al., 1977).
Thorncroft and Hoskins (1994) find in a primitive equation simulation that the most
unstable wave No. 10 grows by conversion of zonal to eddy kinetic energy five to six times
11
11
Figure 6.37 Schematic illustration of the possible zones of convective activity in relation to a posi-
tive vorticity anomaly (PVA) on the African Easterly Jet (AEJ). (a) Longitude height
section, looking northward, of an idealized PVA and vertical motions (short arrows)
and meridional wind (· and x). (b) Possible distribution over dry surfaces at 20°N. (c)
Possible distribution over moist surfaces at 10°N. (Thorncroft and Hoskins, 1994)
0
more effectively than by converting eddy available potential energy. Linear instability at
the jetstream level is dominated by the north–south contrasts between positive (negative)
gradients of potential vorticity on the flanks (in the core) of the jet. However, the nega-
tive gradient of zonal mean quasi-geostrophic potential vorticity in the jet core also causes
the instability to be destroyed by the energy conversion. The distribution of convection
appears to be closely dependent on surface moisture conditions as well as on vorticity
anomalies along the AEJ. Figure 6.37 illustrates this schematically for a meridional-height
cross-section through a positive vorticity anomaly on the AEJ. At 20°N above a dry
0 boundary layer, convection is possible only in southerly airflows east of the vorticity
anomaly above the jet; at 20°N above a moist boundary layer, however, it can occur ahead
of the wave trough, despite northerly airflow, as well as in the rear of the trough in the
11 southerlies. These patterns may explain the observational evidence of convective activity
1 west of the easterly wave trough around 10°N but east of it around 20°N (Burpee, 1974;
Duvel, 1990). Inclusion of latent heating effects with wave–CISK type convection in the
Thorncroft and Hoskins analysis has little effect on the growth rate, but does increase
the phase velocity of the waves and change the normal mode structures by augmenting
baroclinic instability. It is also found that when the jetstream is farther north the baroclinic
instability increases relative to the barotropic, although the growth rate and wave frequency
are not affected. Thorncroft and Hoskins propose that waves grow linearly for about six
days, then baroclinic instability and non-linear growth takes over. Downward propagation
of Rossby waves is strongly damped by boundary layer effects, so that the waves have
maximum amplitude in the mid-troposphere.
During May–November there are on average (1967–89) about 90–100 tropical distur-
bances over the North Atlantic, and tropical systems originating from northern Africa
typically account for about 60 percent of the total number of depressions and also the
same percentage of the approximately ten systems per year that subsequently intensify
into tropical storms (Avila, 1990).
Disturbances in the tropical western Pacific show a period of about six days and a
wavelength of 2,500 km at 850 mb, based on an analysis of the variance in relative
vorticity for 1980–87 (Lau and Lau, 1990). They propagate west-northwestward and are
elongated southwest–northeast, implying the conversion of barotropic energy from the
time-mean circulation to the transient eddies. The maximum activity is equatorward of
the mean trade winds and occurs in association with a “monsoon trough” orientated west-
northwest from about 5°N, 165°E into the South China Sea. The disturbances dissipate
over eastern Asia. An earlier comprehensive study finds that waves forming in the eastern
tropical Pacific, west of Panama, also travel west-northwestward with a period of 5.7 days
and a wavelength of about 2,700 km (Nitta et al., 1985a, b); their structure is similar to
that of systems in the western Pacific. Using infrared imagery and ECMWF analyses for
June–July–August 1980–83 and 1985–89, Takayabu and Nitta (1993) identify tropical
depressions west of 160°E, particularly in El Niño years, propagating westward with
periods of three to five days. Wave convergence and convection are tightly coupled, and
the structure corresponds to the easterly waves described by Reed and Recker (1971).
However, in La Niña years, differences in the large-scale environment (zonal wind, vertical
wind shear, and SST distribution) lead to mixed Rossby–gravity waves developing along
the equator. Near the dateline these have a large amplitude over a sector spanning about
4,000 km. Liebmann and Hendon (1990; Hendon and Liebmann, 1991) find four-to-five-
day-period Rossby–gravity waves only in October–November and within about 30°
longitude of the dateline. Their peak amplitude is near 7.5° latitude. They have wave-
lengths of 7,000–10,000 km and propagate westward at 15–20 m s1. Their occurrence
is attributed to the warm ocean surface (≥28°C) and the double ITCZ structure (see Figure
Figure 6.38 Schematic illustration of “tropical depression” (TD) and mixed Rossby–gravity (MRG)
wave disturbances. The lower tropospheric circulation, high/low pressure anomaly
centers and areas of convection cloud are indicated. (From Takayuba and Nitta, 1993)
11 3.44). Although the meridional wind components are small (~1 m s 1 at 850 mb), the
waves account for over 50 percent of the variance of OLR. These waves appear to be
convectively coupled, although convection is only loosely linked with the wave conver-
gence (see Figure 6.38).
6.6.3 Tropical cyclones
Distribution
0 Tropical cyclones are the most severe category of large-scale weather system, bringing
winds of up to 33–50 m s1 or more, torrential rain, and coastal storm surges. In the
North Atlantic they are referred to as hurricanes and in the western North Pacific as
typhoons. They develop over warm tropical oceans, evolving over four to seven days
11 from weak depressions into named tropical storms with highest sustained winds (aver-
aged over one minute or more) of at least 18 m s1 and becoming a tropical cyclone if
the winds intensify to 33 m s1 or more and an eye forms. Figure 6.39 shows their global
distribution, relationship to ocean surface temperatures of at least 27°C, and typical paths.
Only exceptionally do they occur over land, or cold ocean waters, or within about 5° of
the equator (Gray, 1968, 1979). The seasonality is related to both oceanic and atmos-
0 pheric conditions. Tropical storms in the North Atlantic and in the eastern and western
North Pacific are most frequent in late summer, whereas those in the southwest Pacific
and southwest Indian Ocean peak in midsummer (January). Those in the southeast Indian
Ocean peak in March and January. In contrast, the storms in the Bay of Bengal and (more
rarely) the Arabian Sea occur in spring and autumn. Vertical wind shear is large over
these two areas in northern summer, as it is over the central Pacific, and this appears to
be a powerful constraint on tropical cyclogenesis (Gray, 1968). Annually, there are on
average about fifty-five tropical storms and cyclones in the northern hemisphere and
twenty-seven in the southern hemisphere (Lander and Guard, 1998). The northwest Pacific
sector alone experiences about one-third of the global total (twenty-seven systems) (see
0 Table 6.2). The global total has varied between seventy-five in 1986 and 103 in 1971.
An analysis of Atlantic systems for 1967–93 indicates that of fifty-nine tropical waves,
on average, per season, twenty become tropical depressions but only nine (four) inten-
sify to the tropical storm (cyclone) stage (Avila and Pasch, 1995). Systems of African
origin represent about 60 percent of each intensity category, the remainder developing
over the tropical Atlantic or Caribbean. African waves are also the main contributor to
storms in the eastern North Pacific, according to Avila and Clark (1989), although they
are not a necessary precursor of tropical cyclogenesis in the eastern Pacific (Molinari
et al., 1997). The proportion of waves of African origin that develop in the Atlantic varies
0
Figure 6.39 Frequency of hurricane genesis for a twenty-year period (isopleths) and primary tracks
(arrowed). Areas with sea surface temperature exceeding 27°C in the warmest month
11 are shaded. (After Gray, 1979, from Barry and Chorley, 1998)
1 Table 6.2 Mean annual frequency of tropical storms and cyclones, 1966–95
Measure Western Eastern N. Atlantic N. Indian S. hemi- Global

N. Pacific N. Pacific sphere
Mean 27.1 15.8 9.8 4.8a 27.2a 84.9a
Standard
deviation 4.5 3.8 3.4 2.4 3.8 9.8
Source: from Lander and Guard (1998).
Note
a 1969–95.
greatly from year to year (Frank and Clark, 1980). Avila and Pasch distinguish “African”
years if the ratio of African waves to total waves in the Atlantic is 0.7 or more, and “non-
African” years when the ratio is 0.5 or less; otherwise years are termed average. During
1967–95 twelve years are categorized as African, eight as non-African, and nine as
average. Hurricanes of category 3–5 intensity on the Hurricane Disaster Potential (now
the Saffir/Simpson) Scale (Simpson, 1974) are twice as likely to occur in African years
as in non-African years. In this context “Out of Africa” takes on a significant new meaning!
Non-African years, such as 1972, 1977, 1983, and 1986 coincide with El Niño years,
implying an unfavorable large-scale environment for cyclogenesis over the tropical
Atlantic (Avila and Clark, 1989). Tropical cyclone activity in the western North Atlantic
varies almost independently of that in the other ocean basins, whereas there are weak but
significant rank correlations (0.3–0.4) between the annual frequency of cyclones in the
two North Pacific regions and between each of them and the southern hemisphere. More
important, cyclone frequency in the Atlantic is correlated (0.52) with the average ENSO
index for boreal spring to autumn, and with the stratospheric QBO for June to September
(0.44) (Lander and Guard, 1998). More cyclones are observed in the Atlantic during the
west phase of the QBO, as was observed in 1995, for example (Landsea et al., 1998).
For the other ocean basins such correlations are lacking.
The 1979 season exemplifies an African year: there were eighty-five tropical waves,
characterized by a trough or cyclonic curvature in the easterly trades, with maximum
amplitude in the lower troposphere (Avila and Pasch, 1992). Of these eighty-five systems,
fifty-two originated over Africa and twenty-seven became closed depressions. The African
disturbances accounted for seven of the eight named storms. Forty-three waves of African
origin also moved into the eastern North Pacific, where seven intensified into named
storms. In addition, one Pacific storm developed out of nine ITC disturbances originating
in the Caribbean, one formed from an Atlantic wave, and another formed from an eastern
Pacific wave. In the non-African season of 1992 there were sixty-nine Atlantic waves.
Nine tropical depressions formed, four of which originated from Africa; only two out of
six storms, and one out of four hurricanes, had an African origin.
Structure
The structure of a mature tropical cyclone has several characteristic features (Miller, 1967;
Palmén and Newton, 1969; Frank, 1977, 1982; Anthes, 1982; Pielke, 1990). The mean
radius, as defined by the outer closed surface isobar, is 350 km in the western North Atlantic
and 500 km in the western North Pacific (Merrill, 1984), which is much smaller than a
typical extratropical cyclone. However, the radius ranges from less than 100 km up to 1,100
km in extreme cases. Surface winds spiral towards the center, with a 10–20 km-wide
annulus of maximum winds (in near-cyclostrophic balance) surrounding the central eye,
which may have a diameter of between 15 km and 100 km (Figure 6.40). The wind
11
11
Figure 6.40 Schematic plan view and cross-section of a mature hurricane. The low-level wind
streamlines and cloud bands based on radar echo patterns are shown above. (From
Willoughby et al., 1984.) The lower figure shows the vertical motion and cloud struc-
ture. (From Musk, 1988)
0 maximum coincides with a cloud wall and the most intense convection (Ramage, 1995,
p. 242). Winds are weak within the eye and there is often a break in the middle–high
cloud decks. The storm’s circulation may extend throughout the troposphere to 14–15 km
altitude. Inflow is pronounced in the lower levels and, according to Gray (1979), may
extend to 7 km altitude. In the mid-troposphere there is more or less tangential flow, while
outflow occurs above 8 km, with a maximum typically around 12 km. The upper circula-
tion appears to be cyclonic in some cases, with outflow beyond about 300–500 km radius,
especially on the poleward side of the storm center. Aircraft and satellite data show
the storm comprises spiral cloud and rain bands, an annular ring of subsidence around the
cirrus shield capping the storm, and an outer convective band at about 800 km from the
0 center (Plate 15). In contrast to subtropical cyclones and tropical depressions, the hurricane
is warm-cored as a result of the concentrated moist adiabatic ascent of air and the latent
heat released in the eye wall and rain bands. The warmth is amplified within the eye by
11 subsiding air, giving rise to temperature anomalies of 10° to 20°C.
1
Plate 15 GOES-E image of Hurricane Gilbert on (above) September 12 (visible) and (below)
September 15 (IR) 1988. The eye and spiral cloud bands are clearly visible. (From
Carleton, A.M., 1991, Satellite Remote Sensing in Climatology, Belhaven Press, London,
p. 147)
11 This subsidence creates an inversion in the mid-lower troposphere (850–500 mb),
according to Willoughby (1998). The dry air above this inversion may be resident in the
eye following its formation and it subsides only a few kilometers; the lower air is moist,
owing to frictional inflow below the eye wall, sea surface evaporation, and moist down-
drafts. Willoughby notes that the central pressure in an intense system may be 50–100
mb below that outside the vortex, but only 10–30 mb of this drop occurs between the
eye wall and the center. This represents the contribution of warming due to subsidence.
The storm acquires most of its energy from the inflow of latent heat, supplemented by
sensible and latent heat transfer from the warm ocean surface. Sensible heat counteracts
0 adiabatic cooling caused by the rapid pressure decrease of 50–100 mb as the air flows
towards the eye. For example, 27°C air at 1,000 mb pressure moving to a storm center
of 900 mb would theoretically cool 9°C without heat transfer from the ocean surface
(Pielke, 1990). Central pressure and wind intensities are closely related. A category 3
11 cyclone on the Saffir/Simpson scale has winds of 49–58 m s1 and a central pressure of
945–64 mb; a category 5 cyclone (such as Hurricane Camille in 1969 and Hurricane
Gilbert in 1989), which can cause catastrophic damage, has winds exceeding 69 m s1
and a central pressure below 920 mb. Hurricane Andrew, which struck southern Florida
in 1992, was rated as a category 4 cyclone, although it was the most costly weather
disaster in the United States.
0 Holland and Merrill (1984) propose using three related but weakly correlated parameters
to describe tropical cyclones: size, intensity, and strength. “Size” is measured by the radius
of the outermost closed isobar or by the axisymmetric radius of gale-force winds.
“Intensity” is defined by the maximum winds or central pressure. “Strength” is deter-
mined from the average relative angular momentum of the low-level circulation within
300 km radius. Systems of similar intensity may have very different sizes and strengths
(Merrill, 1984). Large cyclones develop most commonly in October in the North Atlantic
and northwest Pacific near latitude 30°N.
0 Cyclogenesis
Several factors regulating the development of tropical cyclones are illustrated by Figure
6.41, as noted above. Warm tropical waters are typically at least 50–60 m deep, so that
wind mixing does not bring cool water to the surface (Gray, 1979). However, the rela-
tionship with sea surface temperature (SST) is not straightforward, since two Atlantic
hurricanes formed in 1980 over waters where the SSTs were only 23° and 20°C (Ramage,
1995, p. 260). Small horizontal gradients of SST are also important. Storms tend to develop
from existing waves or depressions where cyclonic vorticity is locally concentrated. This
condition may exist within a monsoon trough, in the ITCZ, in waves in the tropical east-
erlies, or at the trailing edge of a cold front that has penetrated to low latitudes. The
0 disturbance needs to be at least 4°–5° from the equator so that the Coriolis parameter is
large enough to support flow curvature. A moist, warm atmosphere that is close to satu-
ration and able to generate towering cumulus and cumulonimbus clouds is also a
prerequisite. Deep convection can be enhanced by low-level convergence (Ekman pumping
in the boundary layer) and by upper-level mass divergence.
Important features of the upper circulation in summer are the tropical upper-tropos-
pheric troughs (TUTTs) orientated westsouthwest–eastnortheast in the North Atlantic and
North Pacific oceans and westnorthwest–eastsoutheast in the South Atlantic and South
Pacific. At 200 mb over the South Pacific in January the TUTT extends from 30°S, 105°W
to the equator at 175°W; in the South Atlantic the TUTT stretches from 30°S, 15°W
0 across South America to the equator at 75°W. In the North Atlantic in July the TUTT
spans 35°N, 30°W to 22°N, 95°W and in the North Pacific from 35°N, 145°W to 22°N,
130°E (Sadler and Wann, 1984; Ramage, 1995, pp. 57–61). The western ends of the
11 TUTTs appear to provide upper-level support for tropical cyclogenesis. However, Ramage
1
Figure 6.41 Schematic illustration of secondary circulations resulting from convective heating in
the inner core and momentum forcing in the outer region, and their potential effects
on cyclone intensity, size, and strength. (From Holland and Merrill, 1984)
(1995, p. 263) suggests that 85–90 percent of western North Pacific tropical cyclones
form in the surface monsoon trough and only 10–15 percent develop where the TUTT
overlies the trade winds. Synoptically, the TUTT consists of a series of upper cold lows,
but it is most readily identified in upper-level vector wind analyses (Sadler, 1975). Such
upper lows occasionally trigger tropical cyclogenesis, according to Sadler (1976).
Divergence east of an upper cold low induces a low-level trough and, as convection builds
deep cumulus, the heat released by condensation forms a ridge east of the cold low, which
begins to shrink. A low-level depression now forms, with convective cloud concentrated
to the east of the vortex. The upper tropospheric flow becomes distorted and the low-
level depression strengthens into a tropical storm overlain by a high-pressure cell in the
upper troposphere. The dynamic process involves intense convection setting up low-level
convergence, with the inflow generating cyclonic spin-up (Hastenrath, 1991, p. 226). This
follows from the vorticity equation (see p. 47).
For a tropical cyclone to develop, a pressure drop of 25–30 mb is needed (Ramage,
1971). The difference between the central pressure in a mature storm and values in the
undisturbed surroundings is in the range 50–100 mb, i.e. central pressures of 960–910
mb; the pressure gradient is typically 1–2 mb km1. The extreme pressures are made
possible by the latent heat release in the cloud bands, especially the eye wall cloud, and
by adiabatic warming of subsiding air in the eye. The inward spiraling of air, with conser-
vation of absolute angular momentum about the cyclone axis at distance r:
f r2
Mr = vT r
2
where vT tangential velocity and r is the radial distance, would imply infinite wind
speeds at the center. Thus a limiting tangential velocity is achieved at some radial distance,
and here the air is forced upward and outward.
Composite vertical cross-sections of radial winds in intensifying storms and hurricanes
(Holland and Merrill, 1984) show two low-level inflow maxima in the inner and outer
11 sectors of the storm, a secondary inflow near 400 mb, and an extensive outflow region
between 300 and 100 mb. The inflow at low levels within 6° radius imports angular
momentum, strengthening the cyclone and offsetting frictional effects; it also promotes
moisture flux convergence. The inflow from the outer environment supplies the angular
momentum needed to increase cyclone size. The secondary upper tropospheric inflow
provides the vertical wind shear from cyclonic to anticyclonic winds. The high-level
outflow removes air with high potential temperatures to the outer part of the storm and
can thereby effect changes of cyclone intensity. Figure 6.40 illustrates schematically the
effects of the various forcings on cyclone structure. The upper circulation is often asym-
0 metric, in contrast to the lower levels. There is typically a major poleward outflow jet,
ahead of a trough in the subtropical westerlies, and a subsidiary equatorward outflow, as
demonstrated in the northwest Pacific (Sadler, 1976) and the southwest Pacific (Holland
and Merrill, 1984). The import of angular momentum needed for intensification of a trop-
11 ical cyclone is very minor relative to that required for its initial growth and subsequent
maintenance. Merrill (1984) calculates that the requirements are as follows:
Angular momentum import across

8° radius (1017 kg m2 s1)
0 Cyclone growth 2.86

Poleward motion 1.94
Surface stress 0.90
Intensification 0.08
Gray (1998) argues that lower-tropospheric wind surges play the critical role in trig-
gering cyclone development from an initial convective disturbance. Zehr (1992) and Gray,
in particular, identify a two-stage process. A first antisymmetric wind surge lasting only
six to twelve hours strengthens the convective activity within a large active cloud cluster,
establishing a mid-level circulation and convective vortex. During the subsequent one to
0 three days the broad system characterized by only weak to moderate convection under-
goes little change. Then a second wind surge triggers a blow-up of deep convection and
transforms the lower levels of the disturbance from a cold core to a warm core. This
occurs because the inertial stability of the mid-level circulation, set up by the first surge,
concentrates the penetration of the second surge into the disturbance within the lower
troposphere. Mechanically forced convergence and ascent moisten the upper levels in the
inner core. Buoyancy sustained by cumulonimbus clouds near the storm center then begins
to concentrate convection around the eye wall, and this is maintained by unstable, self-
sustaining convergence. The storm becomes isolated from its external environment, and
internal physical processes are dominant until the system encounters a cooler sea surface,
0 strong vertical wind shear, or undergoes landfall. The wind surges originate in a variety
of ways: as surges in the trade wind or monsoon flow; as a cross-equatorial surge related
to a winter-hemisphere baroclinic cyclone; from convergence induced by an easterly wave
or by an upper-level trough (in the subtropics); or from convergence resulting from the
low-latitude penetration of a cold front.
The mean structures of western Pacific and western North Atlantic tropical cyclones
are compared with those of summer season cloud clusters in Figure 6.42 (McBride, 1981).
These composites are each based on about eighty disturbances. The main difference over
a 4° latitude radius between the cloud clusters and cyclones is in the relative magnitudes
of the upper warm core, moisture anomalies, and tangential wind velocities. Using a
0 “seasonal genesis parameter” (SGP) proposed by Gray (1979), which can be represented
as the product of a dynamic potential and a thermal potential, McBride finds that the
large differences in SGP between the cloud clusters and the cyclones are determined by
11 the dynamic potential. The latter term is the product of a vorticity parameter, the Coriolis
1
Figure 6.42 Mean structure of eighty-seven cloud clusters in the western Pacific and forty-six in the western Atlantic (left) and of a composite western Pacific
typhoon (147 cases) and Atlantic hurricane (seventy-three cases) (right). The clusters are all embedded in deep easterly flow and moving west-
ward at about 6 m s1. The typhoons have a mean location of 13°N, 136°E and the hurricanes 23°N, 73°W . (a) Difference between the mean
temperature within 3° radius of the center of the cluster and 3°–7° radius to east and west of the cluster. (b) Height anomalies (m) of the 200 mb,
500 mb and 900 mb surfaces versus radius from the center. (c) Differences of specific humidity (g kg1 ) determined as in (a). (d) Radial wind
(m s1 0 at 4° radius. (e) Vertical velocity (mb day1) for 0–4° radius. (f) Tangential wind speed at 4° radius. (From McBride, 1981)
11 parameter, and a vertical wind shear parameter. The thermal potential for deep convec-
tion appears to play only a climatological role. Comparison of non-developing versus
developing cloud clusters shows that the differences between them are rather subtle
(McBride and Zehr, 1981). Both categories are warm-cored in the upper troposphere and
the convective instability and moisture anomalies are similar in each case. In the devel-
oping cases the warm core has a larger horizontal extent but the critical differences are
in the dynamic parameters. Developing clusters are in areas where the low-level relative
vorticity is about twice that in non-developing cases; the difference between 900 mb and
200 mb relative vorticity over a 6° radius is 2.7 versus 0.7 105 s1, respectively. The
0 conclusion is that cyclogenesis is dependent on a cloud cluster being located in a favor-
able large-scale environment, particularly one where upper-level anticyclonic vorticity
overlies lower-level cyclonic vorticity. In a further study of developing and non-devel-
oping cloud clusters in the western Pacific, Lee (1989) confirms the importance of
11 large-scale low-level convergence, a stronger middle and low-level cyclonic circulation,
and mid-level moisture in the developing cases. The formation of tropical cyclones in
subtropical latitudes and in the off season often involves a hybrid baroclinic process
(Bosart and Bartlo, 1991). This is common in the southern hemisphere, where the upper
westerlies penetrate to low latitudes. Upper-level troughs play a role by supporting outflow
aloft and ascending motion.
0
Scale interactions
The concept of a cooperative interaction between a cluster of deep cumulus and an incip-
ient tropospheric vortex was put forward independently by Ooyama (1964) and Charney
and Eliassen (1964). The central idea is that a cumulus cluster located in a region of
vortical flow generates buoyancy which serves as a driver for the vortex to develop.
Ascent of air in the cumulus towers causes vortex stretching in the lower troposphere and
therefore is a source of relative vorticity. The Charney and Eliassen theory considers the
latent heat release by the cumulus convection to be proportional to the moisture flux
0 convergence in the boundary layer. Intensification of the vortex enhances this moisture
convergence and the release of latent heat, thereby generating positive feedback. This
organization of convection leading to the growth of instability in a moist, conditionally
unstable atmosphere is termed Conditional Instability of the Second Kind (CISK) to distin-
guish it from the simple instability responsible for the initiation of cumulus clouds. The
linear model of Charney and Eliassen contains unstable modes, but the growth rates are
relatively uniform over a range of horizontal scales. Moreover the fact that mid-tropos-
pheric heating would eventually stabilize the atmosphere is neglected (Smith, 1997).
The necessity for non-linear scale interactions is stressed by many investigators,
although the precise mechanisms seem elusive. The concept of non-linear cooperative
0 interactions between the cumulus and cyclone scales is proposed by Ooyama (1982). A
depression organizes the clouds into a consistent pattern. Cloud entrainment forces a deep
lower tropospheric inflow and shallow upper-level outflow. The formulation of coopera-
tive intensification by Ooyama (1964, 1982) differs in several fundamental ways from the
“classical” CISK model, according to Smith. Deep cumulus clouds are able to entrain
mid-tropospheric air and transfer it to upper levels. The heating rate is proportional to
the boundary layer convergence and ascent, but it also depends on the degree of convec-
tive instability (i.e. on the convective available potential energy, or CAPE; see Chapter
3, n. 5) through a variable entrainment parameter. Also, the radial entrainment in the
middle layer leads to convergence in the mid-troposphere which is required for the spin-
0 up of the vortex (Ooyama, 1982). Ooyama’s non-linear model allows the local Rossby
radius of deformation (see Chapter 5, n. 3) to shrink as the inertial stability of the vortex
increases in its inner core, thereby reducing the scale separation between deep cumulus
11 and the tangential circulation in the vortex. It is important to note that Ooyama’s concept
1 addresses the transformation of a vortex to hurricane intensity and scale, given the prior
organization of cumulus convection by vortical flow resembling a tropical depression. His
results also indicate that the transfer of latent and sensible heat from a warm ocean is
critical for vortex intensification, although energy budget calculations suggest these terms
represent only 9–16 percent of the total heat source (Palmén and Newton, 1969, table
15.4).
A related theory of hurricane intensification by Emmanuel (1986) also involves inter-
actions between a vortex and a warm ocean. Emmanuel views the energy cycle of a
mature tropical cyclone as a Carnot cycle that converts heat energy acquired from the
ocean into mechanical energy. The thermodynamic efficiency, (Ts To)/Ts , where
Ts sea surface temperature and To is the mean temperature of the upper-level outflow.
Emmanuel states that is typically one-third. Finite-amplitude instability is attributed to
wind-induced surface heat exchange (or WISHE). Latent heat flux increases markedly
with increasing surface wind speeds, and this seems to be critical in hurricane intensifi-
cation. The actual sea–air temperature difference in the central area is small, but according
to Emmanuel (1991) the near-surface air is undersaturated. The surface heat transfers
increase the large-scale radial gradient of equivalent potential temperature in the boundary
layer, and this gradient is communicated to the troposphere by convection. The concur-
rent increase in the large-scale gradient of buoyancy leads to convergence above the
boundary layer, which enables the vortex to spin up. Thus cumulus convection redistrib-
utes heat acquired from the ocean surface upward and outward to the upper troposphere,
where it is exported, or lost by radiation. Smith (1997) suggests that the only difference
from Ooyama’s model is that the cumulus convection pattern is organized by the boundary
layer entropy gradients rather than directly by boundary layer convergence. Rotunno and
Emmanuel (1987) simulate realistic hurricanes, starting from a conditionally neutral
ambient atmosphere. However, the simulated storm intensities regularly attain the theo-
retical minimum sustainable central pressures whereas only a few observed storms attain
such intensities. The arguments of Gray (1998) concerning the dominant role of mechan-
ical rather than thermodynamic factors in tropical cyclone spin-up are in disagreement
with the theories of air–sea interaction.
Appendix 6.1 The Q-vector formulation

The Q-vector is a form of the omega equation for large-scale vertical motion, developed
so as to combine the generally opposing effects of vorticity advection and thermal advec-
tion (Hoskins et al., 1978). The omega equation expresses the synoptic-scale vertical
motion (represented by the three-dimensional Laplacian of omega – term A) as the sum
of the advection of absolute geostrophic vorticity by the geostrophic wind (term B) and
the Laplacian of temperature (or thickness) advection (term C):
冢 2 f o2
∂p冣
∂2
2 = fo
2 ∂
∂p冤冥冤
Vg·(g f ) 2 Vg·
∂
∂p 冢冣冥 (1)
A B C
where fo the (spatially averaged) Coriolis parameter, vertical motion in pressure
coordinates, 2 the Laplacian operator, = geopotential and static stability. This
equation assumes no effects from diabatic heating, surface friction, or orography. Term
A is proportional to , the upward motion (see Bluestein, 1992). Billingsley (1997)
shows that this relationship can readily be appreciated by considering the Laplacian of
omega in only the zonal direction:
∂(∂)
x2() =
∂x (∂x)
11 (
)/(
x) is the rate of change of with x (or the gradient of the omega curve with
increasing x) and the second derivative is the Laplacian of (i.e. its curvature). Where
is a simple sine wave, the first derivative is a cosine wave and the second derivative
(the Laplacian) corresponds to a negative sine wave (). The relationship between omega
and its Laplacian is generally more complex. Nevertheless, for synoptic-scale systems,
well defined and dominant areas of forcing largely determine the vertical motion field.
Term B in the omega equation represents the vertical variation of vorticity advection
by the geostrophic wind. Where cyclonic (anticyclonic) vorticity advection increases with
altitude (p) there is upward (downward) motion, in the northern hemisphere case. The
0 static stability parameter ( ) implies an inverse relationship between the static stability
and the magnitude of the forcing. The absolute vorticity advection is sometimes evalu-
ated at 500 mb where vorticity tends to be nearly conserved. However, calculation of the
differential advection in a layer such as 700–300 mb is preferable.
11 The Laplacian of the thickness advection (term C) implies that warm (cold) forces
upward (downward) motion. There is again an inverse relationship with static stability.
Terms B and C are generally of opposing sign and the net result is a rather small residual.
For this reason, alternative approaches have been proposed.
Trenberth (1978) rearranges terms A, B, and C to combine the thermal and vorticity
effects. Term A can be expressed as:
0
∂ug
A ≈ 2fo 冢∂x∂ ∂p ∂∂y ∂v∂p ∂v∂p 冣
g g g g
(2)
by ignoring the deformation terms, which are mainly important near frontal boundaries
or jetstreams (Carlson, 1991, p. 186; Martin, 1998).
The preceding term can also be written as an approximation of Trenberth’s formula-
tion in the form:
2fo
V · ( f )
0 p500 T p g
where the thermal wind VT represents the advection in the layer 1,000–500 mb and the
vorticity advection by the thermal wind is at a mean level of 700 mb, for example p500
500 mb. This equation enables the sign and approximate magnitude of the vertical
motion at 700 mb to be estimated. The Trenberth version of the omega resembles the
formulation of the development equation by Sutcliffe (1947). Practical applications are
described by Carroll (1995), although Billingsley (1997) suggests that the method should
be abandoned because of its neglect of deformation terms that are important near fronts
and jets which can now be readily computed.
0 A second approach takes account of transverse (ageostrophic) circulations normal to a
frontal zone or jet streak and the vertical advection of geostrophic momentum. The differ-
ential vorticity advection term (B) and the thickness advection term (C) can be replaced
by a combined Q-vector expression, which contains no explicit advection terms, plus a
diabatic heating contribution (Hoskins et al., 1978):
R 2
i.e. A = 2·Q H (3)
Cp
Equation 3 demonstrates that when the field of Q-vectors is convergent (divergent), there
0 is upward (downward) motion, . On constant pressure surfaces:
11
Q= 冤 ∂Vg
∂x 冢
·
∂
∂p
, 冣
∂Vg
∂y
·
∂
冢
∂p 冣冥
1
Figure 6.43 A schematic illustration of two-dimensional cross-front circulation, showing (a) fron-
togenetic and (b) frontolytic situations. The Q vectors point in the direction of the
low-level ageostrophic motion and towards ascent. In (a) there is an enhancement of
the gradient and development; in (b) the converse is true. (From Hoskins and Pedder,
1980)
A procedure to estimate Q-vectors is discussed by Sanders and Hoskins (1990); they also
illustrate typical configurations of Q-vectors and vertical motion for schematic patterns
of surface highs and lows, upper-level ridges and troughs, and frontogenetic and fron-
tolytic situations.
The Q-vector defines the rate of change of the potential temperature gradient moving
with the geostrophic wind. Q-vectors are directed with the low-level ageostrophic flow
(assuming frictionless adiabatic motion) and indicate a frontogenetic tendency. Advection
of cold (warm) air in the same direction as the Q-vectors implies an enhanced (weak-
ened) potential temperature gradient indicative of frontogenesis (frontolysis), respectively
(Hoskins and Pedder, 1980; Billingsley, 1998). Thus there is frontogenesis when Q is
oriented within 90° of p , giving a thermally direct circulation, and frontolysis when
Q is within 90° to 180° of p , giving a thermally indirect circulation (see Figure 6.43).
p is the potential temperature gradient vector following geostrophic motion. LeDrew
(1988), for example, uses the Q-vector technique to evaluate the relative importance of
the dynamic properties of the advected airflow compared with surface effects for five low-
pressure systems in the Arctic.
Notes
1 The Richardson number (Ri) is a measure of the importance of buoyancy forces (represented
by the static stability, or the square of the Brunt–Väisälä frequency) to inertial accelerations.
Thus:
Ri =
(g ∂ ln )
∂z ) (∂ |V |)2
∂z
Flow becomes turbulent for small Ri. In the context of frontal waves, Ri can also be regarded
as the ratio of the potential energy of the basic state to its kinetic energy.
gH (1 2)
Ri =
(U2 U1)2
11 where U and are respectively the velocities and densities of the two layers, H is the total
channel depth, and – is a m–ean density (Orlanski, 1968.)
2 Passive microwave-derived precipitation indices. P37 and S85 are indices of the likelihood of
rain occurrence and cold-cloud precipitation, respectively, based on passive microwave radi-
ances retrieved by SMMR and/or SMM/I. P37 is obtained from the ratio of the difference
between the brightness temperatures observed at 37 GHz for vertical and horizontal polariza-
tions, divided by the expected clear-sky difference. The rain–no rain threshold is approximately
0.8, with rain probability increasing as P37 decreases (Petty and Katsaros, 1992). S85 is an
analogous ratio of polarized brightness temperature differences at 85 GHz (Claud et al., 1992).
It is a scattering-based index of cold-cloud precipitation (graupel, hail, snow aggregates) applic-
0 able to convective situations, that can be used over land or ocean. Values of S85 of about
30–60 are good indicators of ice hydrometeors associated with convection (McMurdie et al.,
1997).
11 References
Abercromby, R. 1878. On the general character and principal sources of variation in the weather
at any part of a cyclone or anticyclone. Quart. J. Met. Soc., 4: 1–13.
Adler, R.F., Yeh, H.-Y.M., Prasad, N., Tao, W.-K., and Simpson, J. 1991. Microwave simulations
of a tropical rainfall system with a three-dimensional cloud model. J. Appl. Met., 30: 924–53.
Agee, E.M. 1991. Trends in cyclone and anticyclone frequency and comparison with periods of
0 warming and cooling over the northern hemisphere. J. Climate, 4 (2): 263–7.
Alliss, R.J., Sandlin, G.D., Chang, S.W., and Raman, S. 1993. Applications of SSM/I data in the
analysis of Hurricane Florence (1988). J. Appl. Met., 32: 1581–91.
Alvarez, J.A. and Thompson, A.H. 1965. Improvement of weather analysis in isolated areas of the
southern hemisphere by use of meteorological satellite information: a case study. Notos, 14:
33–42.
Anderson, C.J. and Arritt, R.W. 1998. Mesoscale convective complexes and elongated convective
systems over the United States during 1992 and 1993. Mon. Wea. Rev., 126: 578–99.
Anderson, R.K., Ashman, J.P., Bittner, F., Farr, G.R., Ferguson, E.W., Oliver, V.J., and Smith,
A.H. 1969. Applications of Meteorological Satellite Data in Analysis and Forecasting, ESSA
Tech. Rept. NESC 51.
0 Anthes, R.A. 1982. Tropical Cyclones: Their Evolution, Structure and Effects. Met. Monogr. 19
(41), Amer. Met. Soc., Boston MA, 208 pp.
Arnaud, Y., Desbois, M., and Maizi, J. 1992. Automatic tracking and characterization of African
convective systems on Meteosat pictures. J. Appl. Met., 31: 443–53.
Asnani, G.C. 1993. Tropical Meteorology, I, Indian Inst. Trop. Meteorology, Pune, chapter 5.
Aspliden, C.I., Tourre, Y., and Sabine, J.B. 1976. Some climatological aspects of West African
disturbance lines during GATE. Mon. Wea. Rev., 104 (8): 1029–35.
during 1985. Mon. Wea. Rev., 116: 685–701.
during 1986 and 1987. Mon. Wea. Rev., 119: 1575–89.
0 Avila, L.A. 1990. Atlantic tropical systems of 1989. Mon. Wea. Rev., 118 (5): 1178–85.
Avila, L.A. and Clark, G.B. 1989. Atlantic tropical systems of 1988. Mon. Wea. Rev., 117 (10):
2260–5.
Avila, L.A. and Pasch, R.J. 1992. Atlantic tropical systems of 1991. Mon. Wea. Rev., 120 (11):
2688–96.
Avila, L.A. and Pasch, R.J. 1995. Atlantic tropical systems of 1993. Mon. Wea. Rev., 123 (3):
887–96.
Bader, M.J., Forbes, G.S., Grant, J.R., Lilley, R.B.E., and Waters, A.J. 1995. Images in Weather
Forecasting, Cambridge University Press, Cambridge, 499 pp.
Barnett, T.P. 1984. Interaction of the monsoon and Pacific trade wind system at interannual time
scales. 3. Partial anatomy of the Southern Oscillation. Mon. Wea. Rev., 112 (12): 2388–400.
0 Barnston, A.G. and Livezey, R.E. 1987. Classification, seasonality, and persistence of low-frequency
atmospheric circulation patterns. Mon. Wea. Rev., 115: 1083–126.
Barr, S., Lawrence, M.B., and Sanders, F. 1966. TIROS vortices and large-scale vertical motion.
11 Mon. Wea. Rev., 94: 675–96.
1 Barry, R.G. 1967. Models in meteorology and climatology. In: R.J. Chorley and P. Haggett, eds,
Models in Geography, Methuen, London, pp. 97–144.
Barry, R.G. and Chorley, R.J. 1998. Atmosphere, Weather and Climate, 7th edn. Routlege, London.
Barry, R.G., Scharfen, G.R., Knowles, K.W., and Goodman, S.J. 1994. Global distribution of light-
ning mapped from night-time visible band DMSP satellite data. Revue gén. d’électricité (Paris),
6: 13–16.
Barsby, J. and Diab, R.D. 1995. Total ozone and synoptic weather relationships over southern Africa
and surrounding oceans. J. Geophys. Res., 100: 3023–32.
Bell, I.D. 1986. The northwest Australian cloud band, In: Preprints, Second International Conference
on Southern Hemisphere Meteorology, Wellington, New Zealand, December 1986. Amer. Met.
Bennetts, D.A. and Hoskins, B.J. 1979. Conditional symmetric instability – a possible explanation
for frontal rainbands. Quart. J. Roy. Met. Soc., 105: 945–62.
Bergeron, T. 1937. On the physics of fronts. Bull. Amer. Met. Soc., 18: 265–75.
Bergeron, T. 1980. Synoptic meteorology: an historical review. Pageoph., 119: 443–73.
Berlin, C.J. 1991. “A Satellite Climatology of the South Pacific Cloud Band and Related Synoptic
Extratropical Cyclone Activity.” Unpubl. Master’s thesis, Department of Geography, Indiana
University, Bloomington IN, 99 pp.
Betts, A.K., Hall, J.H., Beljaars, A.C.M., Miller, M.J., and Viterbo, P.A. 1996. The land
surface–atmosphere interaction: a review based on observational and global modeling perspec-
tives. J. Geophys. Res., 101: 7209–25.
Billingsley, D. 1997. Review of QG theory. Part II. The omega equation. Nat. Wea. Digest, 21: 43–51.
Billingsley, D. 1998. Review of QG theory. Part III. A different approach. Nat. Wea. Digest, 22
(3): 3–10.
Bjerknes, J. 1919. On the structure of moving cyclones. Geophys. Publ., 5 (6): 7–111.
Bjerknes, V. 1920. The structure of the atmosphere when rain is falling. Quart. J. Roy Met. Soc.,
46: 119–40.
Bjerknes, J. and Holmboe, J. 1944. On the theory of cyclones. J. Met., 1: 1–22.
Bjerknes, J. and Solberg, H. 1922. Life Cycle of Cyclones and the Polar Front Theory of Atmospheric
Circulation. Geophys. Publikasjoner, 3 (1), Norske Videnskaps-Akad. Oslo.
Blackmer, R.H., Davis, P.A., and Serebreny, S.M. 1968. Satellite-viewed Cloud Cover as a Des-
criptor of Atmospheric Properties. Final Rept., Contract E-126–67(N), Stanford Research Inst.,
Menlo Park CA, 47 pp.
Blackmon, M.L. 1976. A climatological spectral study of the northern hemisphere winter time circu-
lation. J. Atmos. Sci., 33: 1607–23.
Blackmon, M.L., Wallace, J.M., Lau, N.-C., and Mullen, S.L. 1977. An observational study of the
northern hemisphere winter time circulation. J. Atmos. Sci., 34: 1040–53.
Bleeker, W. 1958. Fronts and the jet stream: the front as a circulation system. Met. Abhandl., 9
(1): 85–3.
Blender, R., Fraedrich, K., and Lunkeit, F. 1997. Identification of cyclone-track regimes in the
North Atlantic. Quart. J. Roy. Met. Soc., 123: 727–41.
Bluestein, H.B. 1993. Synoptic–Dynamic Meteorology in Midlatitudes. II. Observations and Theory
of Weather Systems. Oxford University Press, Oxford, 594 pp.
Bosart, L.F. 1981. The Presidents’ Day snowstorm of 18–19 February 1979: a subsynoptic scale
event. Mon. Wea. Rev., 109 (7): 1542–66.
Bosart, L.F. 1994. Observed cyclone life cycles. In: S. Grønäs and M.A. Shapiro, eds, The Life
Cycles of Extratropical Cyclones, 1, Alma Mater Forlag, Bergen, pp. 111–48.
Bosart, L.F. and Bartlo, J.A. 1991. Tropical storm formation in a baroclinic environment. Mon.
Wea. Rev., 119: 104–18.
Bottger, H., Eckardt, M., and Katergiannakis, U. 1975. Forecasting extratropical storms with hurri-
cane intensity using satellite information. J. Appl. Met., 14: 1259–65.
Boucher, R.J. and Newcomb, R.J. 1962. Synoptic interpretation of some TIROS vortex patterns: a
preliminary cyclone model. J. Appl. Met., 1: 127–36.
Boucher, R.J., Bowley, C.J., Merritt, E.S., Rogers, C.W.C., Sherr, P.E., and Widger, W.K., Jr. 1963.
Synoptic Interpretations of Cloud Vortex Patterns as Observed by Meteorological Satellites. Final
Rept, Contract No. Cwb-10630, ARACON Geophysics, Concord MA, 211 pp.
Boyle, J.S. and Bosart, L.F. 1983. A cyclone/anticyclone couplet over North America: an example
of anticyclone evolution. Mon. Wea. Rev., 111 (5): 1025–45
11 Brandes, H.W. 1820. Beiträge zur Witterungskunde. Barth, Leipzig.
Branstator, G. 1995. Organization of storm track anomalies by recurring low-frequency circulation
anomalies. J. Atmos. Sci., 52 (2): 207–26.
Bratseth, A.M. 1985. A note on CISK in polar air masses. Tellus, 37A: 403–6.
Bresch, J.F., Reed, R.J., and Albright, M.D. 1997. A polar-low development over the Bering Sea:
analysis, numerical simulation, and sensitivity experiments. Mon. Wea. Rev., 125: 3109–30.
Brodrick, H.J., Jr. 1964. Tiros Cloud Pattern Morphology of some Mid-latitude Weather Systems.
Met. Satellite Lab. Rept No. 24, US Department of Commerce, Weather Bureau, Washington
DC, 33 pp.
Brodrick, H.J., Jr. 1969. Some Aspects of the Vorticity Structure Associated with Extratropical Cloud
0 Systems. ESSA Tech. Memo. NESCTM 15. US Department of Commerce, Washington DC,
8 pp.
Brodrick, H.J. and McClain, E.P. 1969. Synoptic/dynamic Diagnosis of a Developing Low-level
Cyclone and its Satellite-viewed Cloud Patterns. ESSA Tech. Rept NESC 49. Washington DC,
11 26 pp.
Bromwich, D.H. 1988. Snowfall in high southern latitudes. Revs. Geophys., 26: 149–68.
Bromwich, D.H. 1989. Satellite analyses of Antarctic katabatic wind behavior. Bull. Amer. Met.
Soc., 70: 738–49.
Bromwich, D.H. 1991. Mesoscale cyclogenesis over the south-western Ross Sea linked to strong
katabatic winds. Mon. Wea. Rev., 119: 1736–52.
Bromwich, D.H., Carrasco, J.F., Liu, Z., and Tzeng, R.-Y. 1993. Hemispheric atmospheric varia-
0 tions and oceanographic impacts associated with katabatic surges over the Ross Ice Shelf,
Antarctica. J. Geophys. Res., 98: 13045–62.
Brooks, E.M. and Shenk, W.E. 1965. Synoptic Studies of Vortex Cloud Patterns. Final Rept, Contract
Cwb-10817, GCA Tech. Rept No. 65–5–G. Prep. for US Weather Bureau, Washington DC, by
GCA Corp., Bedford MA, 163 pp.
Brown, M.E. and Arnold, D.L. 1998. Land surface–atmosphere interactions associated with deep
convection in Illinois. Intl. J. Climatol., 18: 1637–53.
Browning, K.A. 1990. Organization of clouds and precipitation in the extratropical cyclones. In:
C.W. Newton and E.O. Holopainen, eds, Extratropical Cyclones: The Erik Palmén Memorial
Volume. Amer. Met. Soc., Boston MA, pp. 129–53.
Browning, K.A. 1994. Airflow and structure of precipitation systems in extratropical cyclones. In:
0 S. Grønäs and M.A. Shapiro, eds, The Life Cycles of Extratropical Cyclones, 1, Alma Mater
Forlag, Bergen, pp. 210–19.
Browning, K.A. and Hill, F.F. 1985. Mesoscale analysis of a polar trough interacting with a polar
front. Quart. J. Royal Met. Soc., 111: 445–62.
Browning, K.A. and Roberts, N.M. 1994. Structure of a frontal cyclone. Quart. J. Roy. Met. Soc.,
120: 1535–57.
Brysom R.A. 1966. Air masses, streamlines, and the boreal forest. Geogr. Bull. (Ottawa), 8: 228–69.
Burfeind, C.R., Weinman, J.A., and Barkstrom, B.R. 1987. A preliminary computer pattern analysis
of satellite images of mature extratropical cyclones. Mon. Wea. Rev., 115: 556–63.
Burpee, R. 1972. The origin and structure of easterly waves in the lower troposphere of North
Africa. J. Atmos. Sci., 29 (1): 77–90.
0 Burpee, R. 1974. Characteristics of North African easterly waves during the summers of 1968 and
1969. J. Atmos. Sci., 31 (6): 156–70.
Burtt, T.G. and Junker, N.W. 1976. A typical rapidly developing extratropical cyclone as viewed
in SMS-II imagery. Mon. Wea. Rev., 104: 489–90.
Businger, S. 1985. The synoptic climatology of polar low outbreaks. Tellus, 37A: 419–32.
Businger, S. 1987. The synoptic climatology of polar-low outbreaks over the Gulf of Alaska and
the Bering Sea. Tellus, 39A: 307–25.
Businger, S. 1990. Storm following climatology of precipitation associated with winter cyclones
originating over the Gulf of Mexico. Wea. Forecast., 5: 378–99.
Businger, S. 1991. Arctic hurricanes. Amer. Scient., 79: 18–33.
Businger, S. and Walter, B. 1988. Comma cloud development and associated rapid cyclogenesis
0 over the Gulf of Alaska: a case study using aircraft and operational data. Mon. Wea. Rev., 116:
1103–23.
Cai, M. and van den Dool, H.M. 1991. Low-frequency waves and traveling storm tracks. I.
11 Barotropic component. J. Atmos. Sci., 48: 1420–36.
1 Carleton, A.M. 1979. A synoptic climatology of satellite-observed extratropical cyclone activity for
the southern hemisphere winter. Arch. Met. Geophys. Bioklim., B27: 265–79.
Carleton, A.M. 1981a. Monthly variability of satellite-derived cyclonic activity for the southern
hemisphere winter. J. Climatol., 1: 21–38.
Carleton, A.M. 1981b. Climatology of the “instant occlusion” phenomenon for the southern hemi-
sphere winter. Mon. Wea. Rev., 109: 177–81.
Carleton, A.M. 1981c. Ice–ocean–atmosphere interactions at high southern latitudes in winter from
satellite observation. Aust. Met. Mag., 29: 183–95.
Carleton, A.M. 1983. Variations in Antarctic sea ice conditions and relationships with southern
hemisphere cyclonic activity, winters 1973–77. Arch. Met. Geophys. Biokl., B32: 1–22.
Carleton, A.M. 1985a. Satellite climatological aspects of the “polar low” and “instant occlusion.”
Tellus, 37A: 432–50.
Carleton, A.M. 1985b. Synoptic cryosphere–atmosphere interactions in the northern hemisphere
from DMSP image analysis. Intl. J. Remote Sens., 6: 239–61.
Carleton, A.M. 1987. Satellite-derived attributes of cloud vortex patterns and their application to
climate studies. Remote Sens. Environ., 22: 271–96.
Carleton, A.M. 1988a. Sea ice-atmosphere signal of the Southern Oscillation in the Weddell Sea,
Antarctica. J. Climate, 1: 379–88.
Carleton, A.M. 1988b. Meridional transport of eddy sensible heat in winters marked by extremes
of the North Atlantic Oscillation, 1948/49–1979/80. J. Climate, 1: 212–23.
Carleton, A.M. 1992. Synoptic interactions between Antarctica and lower latitudes. Aust. Met. Mag.,
40: 129–147.
Carleton, A.M. 1995. On the interpretation and classification of mesoscale cyclones from satellite
infrared imagery. Intl. J. Remote Sens., 16: 2457–85.
Carleton, A.M. 1996. Satellite climatological aspects of cold air mesocyclones in the Arctic and
Antarctic. Global Atmos. Ocean Sys., 5 (1): 1–42.
Carleton, A.M. 2001. The Antarctic. In: E. Rasmussen and J. Turner, eds, Mesoscale Circulation
Systems in the Arctic and Antarctic. Cambridge University Press, Cambridge (forthcoming).
Carleton, A.M. and Carpenter, D.A. 1989. Intermediate-scale sea ice–atmosphere interactions over
high southern latitudes in winter. Geojournal, 18: 87–101.
Carleton, A.M. and Carpenter, D.A. 1990. Satellite climatology of “polar lows” and broadscale
climatic associations for the southern hemisphere. Intl. J. Climatol., 10 (3): 219–46.
Carleton, A.M. and Fitch, M. 1993. Synoptic aspects of Antarctic mesocyclones. J. Geophys. Res.,
98 (D7): 12997–13018.
Carleton, A.M. and Song, Y. 1997. Synoptic climatology, and intrahemispheric associations, of cold
air mesocyclones in the Australasian sector. J. Geophys. Res., 102 (D12): 13873–87.
Carleton, A.M. and Song, Y. 2000. Satellite passive sensing of the marine atmosphere associated
with cold-air mesoscale cyclones. Prof. Geogr., 52 (2): 289–306.
Carleton, A.M. and Whalley, D. 1988. Eddy transport of sensible heat and the life history of synoptic
systems: a statistical analysis for the southern hemisphere winter. Meteorol. Atmos. Phys., 38:
140–52.
Carleton, A.M., McMurdie, L.A., Katsaros, K.B., Zhao, H., Mognard, N.M., and Claud, C. 1995.
Satellite-derived features and associated atmospheric environments of Southern Ocean meso-
cyclone events. Global Atmos. Ocean Sys., 3: 209–48.
Carlson, T.N. 1969. Some remarks on African disturbances and their progress over the tropical
Atlantic. Mon. Wea. Rev., 97 (10): 716–26.
Carlson, T.N., 1991. Mid-latitude Weather Systems. HarperCollins, London, 507 pp.
Carrasco, J.F. and Bromwich, D.H. 1994. Climatological aspects of mesoscale cyclogenesis over
the Ross Sea and Ross Ice Shelf regions of Antarctica. Mon. Wea. Rev., 122: 2405–25.
Carrasco, J.F. and Bromwich, D.H. 1996. Mesoscale cyclone activity near Terra Nova Bay and
Byrd Glacier, Antarctica, during 1991. Global Atmos. Ocean Sys., 5: 43–72.
Carrasco, J.F., Bromwich, D.H., and Liu, Z. 1997a. Mesoscale cyclone activity over Antarctica
during 1991. 1. Marie Byrd Land. J. Geophys. Res., 102 (D12): 13923–37.
Carrasco, J.F., Bromwich, D.H, and Liu, Z. 1997b. Mesoscale cyclone activity over Antarctica
during 1991. 2. Near the Antarctic Peninsula. J. Geophys. Res., 102 (D12): 13939–54.
Carroll, E.B. 1995. Practical subjective application of the omega equation and Sutcliffe develop-
ment theory. Met. Appl., 2: 71–81.
11 Chang, D.T. and Sherr, P.E. 1969. Cloud Pattern Models for Extratropical Cyclogenesis. Final
Rept, ESSA Contract E-203–68, Allied Research Associates, Concord MA, 146 pp.
Chang, J.-T. and Wetzel, P.J. 1991. Effects of spatial variations of soil moisture and vegetation
on the evolution of a prestorm environment: a numerical case study. Mon. Wea. Rev., 119: 1368–
90.
Charney, J.G. 1947. The dynamics of long waves in a baroclinic current. J. Met., 4: 135–62.
Charney, J.G. and Eliassen, A. 1964. On the growth of the hurricane depression. J. Atmos. Sci.,
21: 68–75.
Charney, J.G. and Stern, M.E. 1962. On the stability of internal baroclinic jets in a rotating atmos-
phere. J. Atmos. Sci., 19: 159–72.
0 Chen, B., Smith, S.R., and Bromwich, D.H. 1996. Evolution of the tropospheric split jet over the
South Pacific Ocean during the 1986–89 ENSO cycle. Mon. Wea. Rev., 124: 1711–31.
Clarke, L.C. and Renard, R.J. 1966. The U S Navy numerical frontal analysis scheme: further devel-
opments and a limited evaluation. J. Appl. Met., 5: 764–77.
11 Claud, C., Scott, N.A., and Chedin, A. 1992. Use of TOVS observations for the study of polar and
arctic lows. Intl. J. Remote Sens., 13: 129–39.
Claud, C., Katsaros, K.B., Mognard, N.M., and Scott, N.A. 1995. Synergetic satellite study of a
rapidly deepening cyclone over the Norwegian Sea: 13–16 February 1989. Global Atmos. Ocean
Sys., 3: 1–34.
Colucci, S.J. and Davenport, J.C. 1987. Rapid surface anticyclogenesis: synoptic climatology and
attendant large-scale circulation changes. Mon. Wea. Rev., 115 (4): 822–36.
0
Commonwealth Bureau of Meteorology, Melbourne.
Conway, E.D. and the Maryland Space Grant Consortium. 1997. An Introduction to Satellite Image
Interpretation. Johns Hopkins University Press, Baltimore MD and London. 242 pp CD-Rom.
Corfidi, S.F., Merritt, J.H., and Fritsch, J.M. 1996. Predicting the movement of mesoscale convec-
tive complexes. Wea. Forecast., 11: 41–6.
Cotton, W.R. and Anthes, R.A. 1989. Storm and Cloud Dynamics, Academic Press, San Diego CA,
883 pp.
Cox, S.K. 1969. Radiation models of midlatitude synoptic features. Mon. Wea. Rev., 97 (9): 637–51.
Craig, G.C. 1993. Polar lows over the North Atlantic. Met. Mag., 122: 278–9.
Creswick, W.S. 1967. Experiments in objective frontal analysis. J. Appl. Met., 6: 774–81.
0 Crocker, A.M. 1949. Synoptic applications of the frontal contour chart. Quart. J. Roy. Met. Soc.,
75: 57–71.
Cullather, R.I., Bromwich, D.H., and Van Woert, M.L. 1996. Interannual variations in Antarctic
precipitation related to El Niño–Southern Oscillation. J. Geophys. Res., 101: 19109–18.
Curry, J. 1987. Contribution of radiative cooling to the formation of cold-core anticyclones. J.
Atmos. Sci., 44 (18): 2575–92.
Cutrim, E., Martin, D.W., and Rabin, R. 1995. Enhancement of cumulus clouds over deforested
lands in Amazonia. Bull. Amer. Met. Soc., 76: 1801–5.
Davis, N.E. 1981. Meteosat looks at the general circulation. III. Tropical–extratropical interaction.
Weather, 36: 168–73.
Deland, R.J. 1964. Traveling planetary-scale waves. Tellus, 16: 271–3.
0 Dickson, R.R. 1973. Weather and circulation of February 1973: an active low-latitude storm track
across the United States. Mon. Wea. Rev., 101: 461–6.
Doran, J.C. and Zhong, S. 1995. Variations in mixed-layer depths arising from inhomogeneous
surface conditions. J. Climate, 8: 1965–73.
Douglas, M.W., Shapiro, M.A., Fedor, L.S., and Saukkonen, L. 1995. Research aircraft observa-
tions of a polar low at the East Greenland ice edge. Mon. Wea. Rev., 123: 5–15.
Downey, W.K., Tsuchiya, T., and Schreiner, A.J. 1981. Some aspects of a northwestern Australian
cloudband. Aust. Met. Mag., 29: 99–113.
Duggupaty, S.M. and Sikka, D.R. 1977. On the vorticity budget and vertical velocity distribution
associated with the life-cycle of a monsoon depression J. Atmos. Sci., 34 (5): 773–92.
Duvel, J.P. 1990. Convection over tropical Africa and the Atlantic Ocean during northern summer.
0 II. Modulation by easterly waves. Mon. Wea. Rev., 118 (9): 1855–68.
Dvorak, V.F. 1975. Tropical cyclone intensity analysis and forecasting from satellite imagery. Mon.
Wea. Rev., 103: 420–30.
11 Eady, E.T. 1949. Long waves and cyclone waves. Tellus, 1: 33–52.
1 Eldridge, R.H. 1957. A synoptic study of West African disturbance lines. Quart. J. Roy. Met. Soc.,
83: 303–14.
Eliassen, A. 1994. Vilhelm Bjernes’s early studies of atmospheric motions and their connections
with the cyclone model of the Bergen school. In: S. Grønäs and M.A. Shapiro, eds, The Life
Elliott, R.D. and Thompson, J.R. 1965. Relationships between TIROS Cloud Patterns and Air Mass
(Wind and Thermal) Structure. Final Rept, Contract N189 (188)-58870A to US Navy Weather
Research Facility, Norfolk VA, by Aerometric Research, Goleta CA, 54 pp.
Emmanuel, K.A., 1986. An air–sea interaction theory for tropical cyclones. I. Steady-state mainte-
nance. J. Atmos. Sci., 43: 585–604.
Emmanuel, K.A., 1991. The theory of hurricanes. Ann. Rev. Fluid Mech., 23: 179–96.
Emmanuel, K.A. and Rotunno, R. 1989. Polar lows as arctic hurricanes. Tellus, 41A: 1–17.
Erickson, C.O. and Winston, J.S. 1972. Tropical storm, mid-latitude cloud-band connections and
the autumnal buildup of the planetary circulation. J. Appl. Met., 11: 23–36.
Ese, T., Kanestrom, I., and Pedersen, K. 1988. Climatology of polar lows over the Norwegian and
Barents Seas. Tellus, 40A: 248–55.
Espy, J.P. 1841. The Philosophy of Storms. Little and Brown, Boston MA, 552 pp.
Evans, M.S., Keyser, D., Bosart, L.F. and Lackmann, G.M. 1994. A satellite-derived classification
scheme for rapid maritime cyclogenesis. Mon. Wea. Rev., 122 (7): 1381–416.
Evans, J.L. and Shemo, R.E. 1996. A procedure for automated satellite-based identification and
climatology development of various classes of organized convection. J. Appl. Met., 35: 638–52.
Farrell, B.F. and Iannou, P.J. 1993. Stochastic dynamics of baroclinic waves. J. Atmos. Sci., 50:
4044–57.
Fein, J.S. and Stephens, P.L. (eds). Monsoons. Wiley, New York, 632 pp.
Fett, R.W. 1989. Polar low developments associated with boundary layer fronts in the Greenland,
Norwegian, and Barents seas, In: P.F. Twitchell, E.A. Rasmussen, and K.L. Davidson, eds, Polar
and Arctic Lows, Deepak, Hampton VA, pp. 313–22.
Fishman, J. 1991. Comment on “Tropical cyclone–upper-atmospheric interaction as inferred from
satellite total ozone observations.” J. Appl. Met., 30: 1047–8.
Fitch, M. and Carleton, A.M. 1992. Antarctic mesocyclone regimes from satellite and conventional
data. Tellus, 44A: 180–96.
Fleagle, R.G. 1957. On the dynamics of the general circulation. Quart. J. Roy. Met. Soc., 83: 1–20.
Fleagle, R.G. 1960. The general circulation. Science Progress, 48: 72–81.
Fleming, J.R. 1990. Meteorology in America, 1800–1870, Johns Hopkins University Press, Baltimore
MD, 264 pp.
Forbes, G.S. and Lottes, W.D. 1985. Classification of mesoscale vortices in polar airstreams and
the influence of the large-scale environment on their evolutions. Tellus, 37A: 132–55.
Forbes, G.S. and Merritt, J.H. 1984. Mesoscale vortices over the Great Lakes in wintertime. Mon.
Wea. Rev., 112: 377–81.
Frank, N.L. 1969. The “inverted V” cloud pattern – An easterly wave? Mon. Wea. Rev., 97 (1): 130–40.
Frank, N.L. and Clark, G. 1980. Atlantic tropical systems of 1979. Mon. Wea. Rev., 108 (7): 966–72.
Frank, W.M. 1977. The structure and energetics of the tropical cyclone. I. Storm structure. Mon.
Wea. Rev., 105: 1119–35.
Frank, W.M. 1982. Large-scale characteristics of tropical cyclones. Mon. Wea. Rev., 110 (6): 572–86.
Frederiksen, J.S. 1985. The geographic locations of southern hemisphere storm tracks: linear theory.
J. Atmos. Sci., 42 (7): 710–23.
Friedman, R.M. 1989. Appropriating the Weather: Vilhelm Bjerknes and the Construction of a
Modern Meteorology. Cornell University Press, Ithaca NY, 251 pp.
Fritsch, J.M. and Maddox, R.A. 1981. Convectively driven mesoscale weather systems aloft. I.
Observations. J. Appl. Met., 20: 9–19.
Fritsch, J.M., Kane, R.J., and Chelius, C.R. 1986. The contribution of mesoscale convective weather
systems to the warm-season precipitation in the United States. J. Clim. Appl. Met., 25: 1333–45.
Frolow, S. 1941. Sur des variations simultanées de la pression dans la région tropicale. Bull. Amer.
Met. Soc., 23: 239–54 (Engl. transl. 1942).
Gadgil, S. and Guruprasad, A. 1990. An objective method for the identification of the Intertropical
Convergence Zone. J. Climate, 3: 558–67.
Galloway, J.L. 1958. The three-front model: its philosophy, nature, construction and use. Weather,
13: 3–10.
11 Galton, Sir F. 1863. Meteorographica or Methods of Mapping the Weather. Macmillan, Lon-
don.
Garcia, O. 1985. Atlas of Highly Reflective Clouds for the Global Tropics, 1971–83. NOAA-ERL,
Boulder CO, 365 pp.
Gibson, H.M. and Vonder Haar, T.H. 1990. Cloud and convection frequencies over the southeast
United States as related to small-scale geographic features. Mon. Wea. Rev., 118: 2215–27.
Goodman, S.J. and Christian, H.J. 1993. Global observations of lightning. In: R.J. Gurney, J.L.
Foster and C.L. Parkinson, eds, Atlas of Satellite Observations related to Global Change,
Gray, W.M. 1968. Global view of the origin of tropical disturbances and storms. Mon. Wea. Rev.,
0 96 (10): 669–700.
Gray, W.M. 1979. Hurricanes: their formation, structure and likely role in the tropical circulation.
In: D.B. Shaw, ed., Meteorology over the Tropical Oceans. Roy. Met. Soc., Bracknell, UK, pp.
155–218.
11 Gray, W.M. 1998. The formation of tropical cyclones. Met. Atmos. Phys., 67: 37–69.
Gray, Jr., T.I. and Clapp, P.F. 1978. An interaction between low- and high-latitude cloud bands as
recorded on GOES-1 imagery. Bull. Amer. Met. Soc., 59: 808–9.
Griffiths, R.W. and Linden, P.E. 1981. The stability of vortices in a rotating, stratified fluid. J.
Fluid Mech., 105: 283–316.
Grossman, R.L. and Garcia, O. 1990. The distribution of deep convection over ocean and land
during the Asian summer monsoon. J. Climate, 3: 1032–44.
0
Grotjahn, R., Hodyss, D., and Castello, C. 1999. Do frontal cyclones change size? Observed widths
of North Pacific lows. Mon. Wea. Rev., 127 (6): 1089–95.
Guymer, L.B. 1978. Operational Application of Satellite Imagery to Synoptic Analysis in the
Southern Hemisphere. Commonweath Bureau of Meteorology, Tech. Rep. No. 29, Dept. of
Science, Melbourne, 87 pp.
Gyakum, J.R. 1983. On the evolution of the QE II storm. I. Synoptic aspects. Mon. Wea. Rev., 111
(6): 1137–55.
Harley, D.G. 1960. Frontal contour analysis of a “polar” low. Met. Mag., 89: 146–7.
Harrison, M.S.J. 1984. A generalized classification of South African summer rain-bearing synoptic
systems. J. Climatol., 4: 547–60.
0 Harrold, T.W. and Browning, K.A. 1969. The polar low as a baroclinic disturbance. Quart. J. Roy.
Met. Soc., 95: 710–23.
Hastenrath, S. 1990. The relationship of Highly Reflective Clouds (HRCs) to tropical climate anom-
alies. J. Climate, 3: 353–65.
Hastenrath, S. 1991. Climate Dynamics of the Tropics. Kluwer, Dordrecht, 488 pp.
Heinemann, G. 1990. Mesoscale vortices in the Weddell Sea region (Antarctica). Mon. Wea. Rev.,
118: 779–93.
Heinemann, G. 1995. TOVS retrievals obtained from the 3I algorithm: a study of a mesoscale
cyclone over the Barents Sea. Tellus, 47A: 324–30.
Heinemann, G. 1996. A wintertime polar low over the eastern Weddell Sea (Antarctica): a study
with AVHRR, TOVS, SSM/I and conventional data. Met. Atmos. Phys., 58: 83–102.
0 Heinemann, G., Noel, S., Chedin, S.A., Scott, N.A., and Claud, C. 1995. Sensitivity studies of
TOVS retrievals with 3I and ITPP retrieval algorithms: application to the resolution of mesoscale
phenomena in the Antarctic. Met. Atmos. Phys., 55: 87–100.
Henderson-Sellers, A. and McGuffie, K. 1990. Are cloud amounts estimated from satellite sensor
and conventional surface-based observations related? Intl. J. Remote Sens., 11: 543–50.
Hendon, H.H. and Liebmann, B. 1991. The structure and annual variation of antisymmetric fluc-
tuations of tropical convection and their association with Rossby-gravity waves. J. Atmos. Sci.,
48 (19): 2127–40.
Henry, W.K. 1974. The tropical rainstorm. Mon. Wea. Rev., 102 (10): 717–25.
Hewson, T.D. 1998. Objective fronts. Met. Appl., 5: 37–65.
Hewson, T.D., Craig, G.C., and Claud, C. 1995. A Polar Low Outbreak: Evolution and Mesoscale
0 Structures. University of Reading Joint Center for Mesoscale Meteorology Internal Rept 45,
Meteorological Office, London, 26 pp.
Hobbs, P.V., Locatelli, J.D., and Martin, J. 1996. A new conceptual model for cyclones generated
11 in the lee of the Rocky Mountains. Bull. Amer. Met. Soc., 77 (6): 1169–78.
1 Holland, G.J. and Merrill, R.T. 1984. On the dynamics of tropical cyclone structural changes. Quart.
J. Roy. Met. Soc., 110: 723–46.
Holton, J. 1971. A diagnostic model for equatorial wave disturbances: the role of vertical shear of
the mean zonal wind. J. Atmos. Sci., 28 (1): 55–64.
Hopkins, M.M. 1967. An approach to the classification of meteorological satellite data. J. Appl.
Met., 6: 164–78.
Hoskins, B.J. 1971. Atmospheric frontogenesis models: some solutions. Quart. J. Roy. Met. Soc.,
97: 139–53.
Hoskins, B.J. 1994. Sutcliffe and his development theory. In: S. Grønäs and M.A. Shapiro, eds,
The Life Cycles of Extratropical Cyclones, 1, Alma Mater Forlag, Bergen, pp. 52–8.
Hoskins, B.J. and Berrisford, P. 1988. A potential vorticity perspective of the storm of 15–16
October 1987. Weather, 43: 122–9.
Hoskins, B.J. and Pedder, M.A. 1980. The diagnosis of middle latitude synoptic development. Quart.
J. Roy. Met. Soc., 106: 707–19.
Hoskins, B.J. and Valdes, P.J. 1990. On the existence of storm tracks. J. Atmos. Sci., 47 (15): 1854–64.
Hoskins, B.J. and West, N.V. 1979. Baroclinic waves and frontogenesis. II. Uniform potential
vorticity jet flows – cold and warm fronts. J. Atmos. Sci., 36: 1663–80.
Hoskins, B.J., Draghici, I., and Davies, H.C. 1978. A new look at the omega equation. Quart. J.
Roy. Met. Soc., 104: 31–8.
Hoskins, B.J., McIntyre, M.E., and Robertson, A.W. 1985. On the use and significance of isen-
tropic potential vorticity maps. Quart. J. Roy. Met. Soc., 111: 877–946.
Hosler, C.L. and Gamage, L.A. 1956. Cyclone frequencies in the United States for the period 1905
to 1954. Mon. Wea. Rev., 84: 388–90.
Houghton, D.D. and Suomi, V.E. 1978. Information content of satellite images. Bull. Amer. Met.
Soc., 59: 1614–17.
Houze, R.A., Jr, Geotis, S.G., Marks, F.D., Jr., and West, A.K. 1981. Winter monsoon convection
in the vicinity of North Borneo. I. Structure and time variation of clouds and precipitation. Mon.
Wea. Rev., 109 (8): 1595–64.
Houze, R.A., Jr., 1993. Cloud Dynamics, Academic Press, San Diego CA, 573 pp.
Hsu, C.S. and Liu, W.T. 1996. Wind and pressure fields near tropical cyclone Oliver derived from
scatterometer observations. J. Geophys. Res., 101: 17021–7.
Huang, H.-J. and Vincent, D.G. 1985. Significance of the South Pacific Convergence Zone in energy
conversions of the southern hemisphere during FGGE, 10–27 January 1979. Mon. Wea. Rev.,
113: 1359–71.
Iskenderian, H. 1995. A 10-year climatology of northern hemisphere tropical cloud plumes and
their composite flow patterns. J. Climate, 8: 1630–7.
Jaeger, G. 1984. Satellite indicators of rapid cyclogenesis. Mar. Wea. Log, 28 (1): 1–6.
James, I.N. and Anderson, D.L.T. 1984. The seasonal mean flow and distribution of large-scale
weather systems in the southern hemisphere. Quart. J. Roy. Met. Soc., 110: 943–66.
Jones, D.A. and Simmonds, I. 1993. A climatology of southern hemisphere extratropical cyclones.
Clim. Dynam., 9: 131–45.
Jones, D.A. and Simmonds, I. 1994. A climatology of southern hemisphere anticyclones. Clim.
Dynam., 10: 333–48.
Jones, P.D. 1987. The twentieth-century Arctic high – fact or fiction? Clim. Dynam., 1: 63–75.
Junker, N.W. and Haller, D.J. 1980. Estimation of surface pressures from satellite cloud patterns.
Mar. Wea. Log, 24 (3): 83–7.
Kane, R.J., Jr, Chelius, C.R., and Fritsch, J.M. 1987. Precipitation characteristics of mesoscale
convective weather systems. J. Clim. Appl. Met., 26: 1345–57.
Katsaros, K.B. and Brown, R.A. 1991. Legacy of the Seasat mission for studies of the atmosphere
and air–sea–ice interactions. Bull. Amer. Met. Soc., 72 (7): 967–81.
Katsaros, K.B., Bhatti, I., McMurdie, L.A., and Petty, G.W. 1989. Identification of atmospheric
fronts over the ocean with microwave measurements of water vapor and rain. Wea. Forecast.,
4: 449–60.
Kelly, G.A.M. 1978. Interpretation of satellite cloud mosaics for southern hemisphere analysis and
reference level specification. Mon. Wea. Rev., 106: 870–89.
Key, J.R. and Chan, A.C.K. 1999. Multidecadal global and regional trends in 1000 mb and 500
mb cyclone frequencies. Geophys. Res. Lett., 26 (14): 2053–6.
11 Khrgian, A.K. 1970. Meteorology: A Historical survey, 1, 2nd edition, ed. K.P. Pogosyan. Israel
Program for Scientific Translations, Jerusalem (published in Russian, Gidromet. Izdat., Leningrad,
1959) (NTIS TT.69–55106, Springfield VA), 387 pp.
Kidson, J.W. and Sinclair, M.R. 1995. The influence of persistent anomalies on southern hemi-
sphere storm tracks. J. Climate, 8 (8): 1938–50.
Kininmonth, W.R. 1983. Variability of rainfall over northern Australia, In: A. Street-Perrott, M.
Beran, and R. Ratcliffe, eds, Variations in the Global Water Budget. Reidel, Dordrecht,
pp. 265–72.
Kirk, T.H. 1966a. Some aspects of the theory of fronts and frontal analysis. Quart. J. Roy. Met.
Soc., 92: 374–81.
0 Kirk, T.H. 1966b. A parameter for the objective location of frontal zones. Met. Mag., 94: 351–3.
Kirk, T.H. 1970. The Laplacian and its relevance for analysis. Met. Mag., 99: 151–2.
Klein, W.H. 1957. Principal Tracks and Mean Frequencies of Cyclones and Anticyclones in the
Northern Hemisphere. US Weather Bureau, Res. Paper 40, Washington DC, 60 pp.
11 Klein, W.H. 1958. The frequency of cyclones and anticyclones in relation to the mean circulation.
J. Met., 15: 98–102.
Kletter, L. 1972. Globale Beobachtung signifikanter Wirbelstrukturen. Arch. Met. Geophys. Bioklim.,
A21: 353–72.
Kocin, P.J. and Uccellini, L.W. 1990. Snowstorms along the Northeastern Coast of the United
States, 1955 to 1985. Met. Monogr. 22 (44), 280 pp.
Kondrat’ev, K.Y., Borisenkov, E.P., and Morozkin, A.A. 1970. Interpretation of Observation Data
0 from Meteorological Satellites. Israel Program for Scientific Translations, Jerusalem, 370 pp.
Kopken, C., Heinemann, G., Chedin, A., Claud, C., and Scott, N.A. 1995. Assessment of the quality
of TOVS retrievals obtained with the 3I algorithm for Antarctic conditions. J. Geophys. Res.,
100: 5143–58.
Köppen, W. 1880. Die Zugstrassen der barometrischen Minima in Europa und auf dem nordatlantischen
Ozean und ihr Einfluss auf Wind und Wetter bei uns. Mitt. Geogr. Ges. Hamburg 1880–81, pp. 76–97.
Kornfield, J., Hasler, A.F., Hanson, K.J., and Suomi, V.E. 1967. Photographic cloud climatology
from ESSA III and V computer produced mosaics. Bull. Amer. Met. Soc., 48: 878–83.
Koteswaram, P. 1958. The easterly jetstream in the tropics. Tellus, 10: 43–57.
Krishnamurti, T.N. 1979. Tropical Meteorology: Compendium of Meteorology, Part 4. WMO Publ.
364, Geneva, 428 pp.
0 Kruspe, G. and Bakan, S. 1990. The atmospheric structure during episodes of open cellular convec-
tion observed in KonTur, 1981. J. Geophys. Res., 95 (D2): 1973–84.
Kuettner, J. 1959. The band structure of the atmosphere. Tellus, 11: 267–94.
Kuhnel, I. 1989. Tropical–extratropical cloudband climatology based on satellite data. Intl. J.
Climatol., 9: 441–63.
Kuhnel, I. 1990. Tropical–extratropical cloudbands in the Australian region. Intl. J. Climatol., 10:
341–64.
Kutzbach, G. 1979. The Thermal Theory of Cyclones: A History of Meteorological Thought in the
Nineteenth Century, American Met. Soc., Boston MA, 255 pp.
Laing, A.G. and Fritsch, J.M. 1993a. Mesoscale convective complexes over the Indian monsoon
region. J. Climate, 6: 911–19.
0 Laing, A.G. and Fritsch, J.M. 1993b. Mesoscale convective complexes in Africa. Mon. Wea. Rev.,
121: 2254–63.
Laing, A.G. and Fritsch, J.M. 1997. The global population of mesoscale convective complexes.
Lander, M.A. 1990. Evolution of the cloud pattern during the formation of tropical cyclone twins
symmetrical with respect to the equator. Mon. Wea. Rev., 118: 1194–202.
Lander, M.A. and Guard, C.P. 1998. A look at tropical cyclone activity during 1995: contrasting
high Atlantic activity with low activity in other basins. Mon. Wea. Rev., 126 (5): 1163–73.
Landsea, C.W. 1993. A climatology of intense (or major) Atlantic hurricanes. Mon. Wea. Rev., 121
(6): 1703–13.
Landsea, C.W., Bell, G.D., Gray, W.M., and Goldenburg, S.B. 1998. The extremely active 1995
0 Atlantic hurricane season: environmental conditions and verification of seasonal forecasts. Mon.
Wea. Rev., 126 (5): 1174–93.
Lau, A.H.-K. and Lau, N.-C. 1990. Observed structure and propagation characteristics of summer-
11 time synoptic-scale disturbances over the tropical western Pacific. In: P. Sham and C.P. Chang,
1 eds, East Asia and Western Pacific Meteorology and Climate, World Scientific Publ., Singapore,
pp. 48–57.
Lau, K.M. and Chan, P.H. 1988. Intraseasonal and interannual variations of tropical convection:
a possible link between the 40–50 day oscillation and ENSO? J. Atmos. Sci., 45: 506–21.
Lau, N.-C. and Crane, M.W. 1997. Comparing satellite and surface observations of cloud patterns
in synoptic-scale circulation systems. Mon. Wea. Rev., 125: 3172–89.
Leary, C.A. and Rappaport, E.N. 1987. The life cycle and internal structure of a mesoscale convec-
tive complex. Mon. Wea. Rev., 115: 1503–27.
LeDrew, E.F. 1988. Development processes for five depressions systems within the polar basin.
J. Climatol., 8: 125–53.
LeDrew, E.F. 1989. Modes of synoptic development within the Polar Basin. Geojournal, 18 (1):
79–95.
LeDrew, E.F. and Barber, D.G. 1994. The SIMMS program: a study of change and variability
within the marine cryosphere. Arctic, 47: 256–64.
LeDrew, E.F., Johnson, D., and Maslanik, J.A. 1991. An examination of the atmospheric mecha-
nisms that may be responsible for the annual reversal of the Beaufort Sea ice field. Intl. J.
Climatol., 11 (8): 841–59.
Lee, C.S. 1989. Observational analysis of tropical cyclogenesis in the western North Pacific. I.
Structural evolution of cloud clusters. J. Atmos. Sci., 46 (16): 2580–98.
Leese, J.A. 1962. The role of advection in the formation of vortex cloud patterns. Tellus, 14 (4):
409–21.
Lefloch, C. and Amory-Mazaudier, C. 1998. Boundary-layer mean dynamic characteristics of cold
fronts over Brittany. Quart. J. Roy. Met. Soc., 124: 857–71.
Leighton, R.M. 1994. Monthly anticyclonicity and cyclonicity in the southern hemisphere: aver-
ages for January, April, July, and October. Intl. J. Climatol., 14: 33–45.
Leonard, S., Turner, J., and Van der Wal, A. 1999. An assessment of three automatic depression
tracking schemes. Met. Applic., 6: 173–83.
Liebmann, B. and Hendon, H.H. 1990. Synoptic-scale disturbances near the equator. J. Atmos. Sci.,
47 (12): 1463–79.
Locatelli, J.D., Hobbs, P.V., and Werth, J.A. 1982. Mesoscale structures of vortices in polar air
streams. Mon. Wea. Rev., 110: 1417–33.
Locatelli, J.D., Martin, J.E., Castle, J.A., and Hobbs, P.V. 1995. Structure and evolution of winter
cyclones in the central United States and their effects on the distribution of precipitation. III. The
development of a squall line associated with weak, cold frontogenesis aloft. Mon. Wea. Rev.,
123: 2641–62.
Longley, R.W. 1959. The Three-front Model – a Critical Analysis. Met. Branch, CIR 3245, TEC
309, Department of Transport, Toronto, Ontario, 14 pp.
Loomis, E. 1846. On two storms which were experienced throughout the United States, in the month
of February 1842. Trans. Amer. Philosoph. Soc., 9: 161–84.
Loomis, E. 1874. Results derived from examinations of the United States weather maps for 1872
and 1873. Amer. J. Sci. (3rd ser.), 8: 1–15.
Loomis, E. 1885. Areas of low pressure, their form, magnitude, direction and velocity of move-
ment. In: Contributions to Meteorology, Tuttle Moorehouse & Taylor, New Haven CT, 67 pp.
Lukas, R., Webster, P.J., Ji, M., and Leetman, A. 1995. The large-scale context for the TOGA
coupled Ocean–Atmosphere Response Experiment. Met. Atmos. Phys., 56: 3–16.
Lyall, I.T. 1972. The polar low over Britain. Weather, 27: 378–90.
Lyons, S.W. 1983. Characteristics of intense Antarctic depressions near the Drake Passage. In:
Preprint vol., First International Conference on Southern Hemisphere Meteorol., San Jose dos
Campos, Brazil, July 31–August 6, 1983. Amer. Met. Soc., Boston MA, pp. 238–40.
Lyons, T.J., Schwerdtfeger, P., Hacker, J.M., Foster, I.J., Smith, R.C.G., and Xinmei, H. 1993.
Land–atmosphere interaction in a semiarid region: the bunny fence experiment. Bull. Amer. Met.
Soc., 74: 1327–34.
Lystad, M. 1986. Polar Lows in the Norwegian, Greenland, and Barents Seas. Final Rept of Polar
Lows Project, Norwegian Meteorological Institute, Oslo, 196 pp.
Machado, L.A.T. and Rossow, W.B. 1993. Structural characteristics and radiative properties of trop-
ical cloud clusters. Mon. Wea. Rev., 121: 3234–60.
Machado, L.A.T., Desbois, M., and Duvel, J.-P. 1992. Structural characteristics of deep convective
systems over tropical Africa and the Atlantic Ocean. Mon. Wea. Rev., 120: 392–406.
11 Machado, L.A.T., Duvel, J.-P., and Desbois, M. 1993. Diurnal variations and modulation by east-
erly waves of the size distribution of convective cloud clusters over West Africa and the Atlantic
Madden, R.A. 1978. Further evidence of traveling planetary waves. J. Atmos. Sci., 41 (8): 1320–35.
Madden, R.A. and Julian, P.R. 1972. Further evidence of global-scale five-day pressure waves. J.
Atmos. Sci., 29 (8): 1464–9.
Madden, R.A. and Stokes, J. 1975. Evidence of global-scale five-day waves in a 73-year pressure
record. J. Atmos. Sci., 32 (4): 831–6.
Maddox, R.A. 1980. Mesoscale convective complexes. Bull. Amer. Met. Soc., 61: 1374–87.
Maddox, R.A. 1983. Large-scale meteorological conditions associated with midlatitude, mesoscale
0 convective complexes. Mon. Wea. Rev., 111: 1475–93.
Mansfield, D.A. 1974. Polar lows: the development of baroclinic disturbances in cold air outbreaks.
Mapes, B.E. and Houze, R.A., Jr, 1993. Cloud clusters and superclusters over the oceanic warm
11 pool. Mon. Wea. Rev., 121: 1398–415.
Marshall, G.J. and King, J.C. 1997. Southern hemisphere circulation anomalies associated with
extreme Antarctic Peninsula winter temperatures. Geophys. Res. Lett., 25: 2437–40.
Marshall, G.J. and Turner, J. 1997. Surface wind fields of Antarctic mesocyclones derived from
ERS1 scatterometer data. J. Geophys. Res., 102: 13907–21.
Martin, D.W. 1968. Satellite Studies of Cyclonic Developments over the Southern Ocean, Tech.
Rept No. 9, International Antarctic Meteorological Research Center, Bureau of Meteorology,
0 Melbourne, Australia, 72 pp.
Martin, F.L. and Salomonson, V.V. 1970. Statistical characteristics of subtropical jet-stream features
in terms of MRIR observations from Nimbus II. J. Appl. Met., 9 (3): 508–20.
Martin, J.E. 1998. On the deformation terms in the quasigeostrophic omega equation. Mon. Wea.
Rev., 126 (7): 2000–7.
Mass, C.F. 1991. Synoptic frontal analysis: time for a reassessment? Bull. Amer. Met. Soc., 72 (3):
348–63.
McAnelly, R.L. and Cotton, W.R. 1989. The precipitation life cycle of mesoscale convective
complexes over the central United States. Mon. Wea. Rev., 117: 784–808.
McBride, J.L. 1981. Observational analysis of tropical cyclone formation. I. Basic descriptions of
0 data sets. J. Atmos. Sci., 38 (6): 1117–31.
McBride, J.L. and Zehr, R. 1981. Observational analysis of tropical cyclone formation. II.
Comparison of non-developing versus developing systems. J. Atmos. Sci., 38 (6): 1132–51.
McCollum, D.M., Maddox, R.A., and Howard, K.W. 1995. Case study of a severe mesoscale convec-
tive system in central Arizona. Wea. Forecast., 10: 643–65.
McCorcle, M.D. 1988. Simulation of surface-moisture effects on the Great Plains low-level jet.
Mon. Wea. Rev., 116: 1705–20.
McGaughey, G., Zipser, E.J., Spencer, R.W., and Hood, R.E. 1996. High-resolution passive
microwave observations of convective systems over the tropical Pacific Ocean. J. Appl. Met., 35:
1921–47.
McGinnigle, J.B. 1988. The development of instant occlusions in the North Atlantic. Met. Mag.,
0 117: 325–41.
McGinnigle, J.B. 1990. Numerical weather prediction model performance on instant occlusion devel-
opments. Met. Mag., 119: 149–63.
McGuirk, J.P. 1993. Impact of increased TOVS signal on the NMC global spectral model: a trop-
ical-plume case study. Mon. Wea. Rev., 121: 695–712.
McGuirk, J.P. and Ulsh, D.J. 1990. Evolution of tropical plumes in VAS water vapor imagery.
Mon. Wea. Rev., 118: 1758–66.
McGuirk, J.P., Thompson, A.H., and Schaefer, J.R. 1988. An eastern Pacific tropical plume. Mon.
Wea. Rev., 116: 2505–21.
McGuirk, J.P, Thompson, A.H., and Smith, N.R. 1987. Moisture bursts over the tropical Pacific
0 McIntyre, D.F. 1958. The Canadian three-front, three-jetstream model. Geophysica (Helsinki), 6:
309–24.
McMurdie, L.A. and Katsaros, K.B. 1991. Satellite-derived integrated water-vapor distribution in
11 oceanic midlatitude storms: variation with region and season. Mon. Wea. Rev., 119: 589–605.
1 McMurdie, L.A., Claud, C., and Atakturk, S. 1997. Satellite-derived atmospheric characteristics of
spiral and comma-shaped southern hemisphere mesocyclones. J. Geophys. Res., 102: 13889–905.
Merrill, R.T. 1984. A comparison of large and small tropical cyclones. Mon. Wea. Rev., 112 (7):
1408–18.
Merritt, E.S. 1964. Easterly waves and perturbations: a reappraisal. J. Appl. Met., 3 (4): 367–82.
Miller, B.I. 1967. Characteristics of hurricanes. Science, 157: 1389–99.
Miller, D.B. and Feddes, R.G. 1971. Global Atlas of Relative Cloud Cover, 1969–70, Based on
Photographic Signals from Meteorological Satellites. US Department of Commerce and US Air
Force, Washington DC, 237 pp.
Miller, D. and Fritsch, J.M. 1991. Mesoscale convective complexes in the western Pacific region.
Mon. Wea. Rev., 119: 2978–92.
Miller, D.K. and Katsaros, K.B. 1992. Satellite-derived surface latent heat fluxes in a rapidly inten-
sifying marine cyclone. Mon. Wea. Rev., 120: 1093–107.
Miller, D.K. and Petty, G.W. 1998. Moisture patterns in deepening maritime extratropical cyclones.
I. Correlation between precipitation and intensification. Mon. Wea. Rev., 126: 2352–68.
Mills, J.M. and Walsh, J.E. 1988. A winter mesocyclone over the midwestern United States. Wea.
Forecast., 3: 230–42.
Minina, L.S. 1964. The vortex structure of cloud cover according to weather satellite data.
Meteorologija i Gidrologija, 1: 22. (English trans.: NASA TTF 206, Washington DC).
Misra, B.M. 1972. Planetary pressure wave of four to five day period in the Tropics. Mon. Wea.
Rev., 100 (4): 313–16.
Mognard, N.M. and Katsaros, K.B. 1995a. Weather patterns over the ocean observed with the
Special Sensor Microwave/Imager and the Geosat altimeter. Global Atmos. Ocean Sys., 2: 301–23.
Mognard, N.M. and Katsaros, K.B. 1995b. Statistical comparison of the Special Sensor
Microwave/Imager and the Geosat altimeter wind speed measurements over the ocean. Global
Atmos. Ocean Sys., 2: 291–9.
Mohn, H. 1870. Det Norske Meteorologiske Instituts Storm-Atlas, Bengtzen, Christiania, 26 pp.
Mohr, K.I. and Zipser, E.J. 1996a. Defining mesoscale convection systems by their 85 GHz ice-
scattering signatures. Bull. Amer. Met. Soc., 77 (6): 1179–89.
Mohr, K.I. and Zipser, E.J. 1996b. Mesoscale convective systems defined by their 85 GHz ice scat-
tering signature: size and intensity comparison over tropical oceans and continents. Mon. Wea.
Rev., 124: 2417–37.
Molinari, J., Knight, D., Dickinson, M., Vollaro, D., and Skubis, S. 1997. Potential vorticity, east-
erly waves, and eastern Pacific tropical cyclogenesis. Mon. Wea. Rev., 125 (10): 2699–708.
Monmonier, M. 1999. Air Apparent: How Meteorologists Learned to Map, Predict and Dramatize
Weather. University of Chicago Press, Chicago, 309 pp.
Monteverdi, J.P. 1976. The single air mass disturbance and precipitation characteristics at San
Francisco. Mon. Wea. Rev., 104: 1289–96.
Montgomery, M.T. and Farrell, B.F. 1992. Polar low dynamics. J. Atmos. Sci., 49: 2484–505.
Morris, R.M. 1972. The trowal, an important feature of frontal analysis. Met. Mag., 101: 150–3.
Mullen, S.L. 1979. An investigation of small synoptic-scale cyclones in polar air streams. Mon.
Wea. Rev., 107: 1636–47.
Mullen, S.L. 1982. Cyclone development in polar airstreams over the wintertime continent. Mon.
Wea. Rev., 110: 1664–76.
Mullen, S.L. 1983. Explosive cyclogenesis associated with cyclones in polar air streams. Mon. Wea.
Rev., 111: 1537–53.
Muramatsu, T. 1983. Diurnal variations of satellite-measured TBB area distribution and eye diam-
eter of mature typhoons. J. Met. Soc. Japan, 61: 77–90.
Murray, R.J. and Simmonds, I. 1991. A numerical scheme for tracking cyclone centres from digital
data. Austral. Met. Mag., 39: 155–66, 167–80.
Nagle, R.E. and Clark, J.C. 1968. An Approach to the SINAP Problem: a Quasi-objective Method
of Incorporating Meteorological Satellite Information in Numerical Weather Analysis. Final Rept,
Contract No. E-93-67(N), Meteorology International, Montgomery CA, 155 pp.
Nagle, R.E. and Hayden, C.M. 1971. The Use of Satellite-observed Cloud Patterns in Northern
Hemisphere 500-mb Numerical Analysis. NOAA Tech. Rept NESS 55. Washington DC, 36 pp.
Nagle, R.E. and Serebreny, S.M. 1962. Radar precipitation echo and satellite cloud observations of
a maritime cyclone. J. Appl. Met., 1: 279–95.
11 Nakamura, H. 1992. Midwinter suppression of baroclinic wave activity in the Pacific. J. Atmos.
Sci., 49 (17): 1629–42.
Nakazawa, T. 1986. Main features of 30–60 day variations as inferred from 8-year OLR data. J.
Met. Soc. Japan, 64: 777–86.
Nakazawa, T. 1988. Tropical superclusters within intraseasonal variations over the western Pacific.
J. Met. Soc. Japan, 66: 823–39.
Namias, J. 1983. The history of polar front and air mass concepts in the United States – an eyewit-
ness account. Bull. Amer. Met. Soc., 64: 734–55.
Neiman, P.J. and Shapiro, M.A. 1993. The life cycle of an extratropical marine cycle. I. Frontal-
cyclone evolution and thermodynamic air–sea interaction. Mon. Wea. Rev., 121: 2153–76.
0 Newton, C.W. and Holopainen, E.O. (eds). 1990. Extratropical Cyclones: The Erik Palmén
Memorial Volume. Amer. Met. Soc., Boston MA, 262 pp.
Newton, C.W. and Newton, H.R. 1994. The Bergen School concepts come to America. In: S. Grønäs
and M.A. Shapiro, eds, The Life Cycles of Extratropical Cyclones, 1, Alma Mater Forlag, Bergen,
11 pp. 22–31.
Nieman, S.J., Menzel, W.P., Hayden, C.M., Gray, D., Wanzong, S.T., Velden, C.S., and Daniels,
J. 1997. Fully automated cloud-drift winds in NESDIS operations. Bull. Amer. Met. Soc., 78:
1121–33.
Nieman, S.J., Schmetz, J., and Menzel, W.P. 1993. A comparison of several techniques to assign
heights to cloud tracers. J. Appl. Met., 32: 1559–68.
Ninomiya, K. 1989. Polar/comma-cloud lows over the Japan Sea and the northwestern Pacific in
0 winter. J. Met. Soc. Japan, 67: 83–97.
Nitta, T., Nagakomi, Y., Suzuki, Y., Hasegawa, N., and Kadokura, A. 1985a. Global analysis of
the lower tropospheric disturbances in the tropics during the northern summer of the FGGE year.
I. Global features of the disturbances. J. Met. Soc. Japan, 63: 1–19.
Nitta, T., Nagakomi, Y., Suzuki, Y., Hasegawa, N., Kadokura, A., and Takayuba, Y. 1985b. Global
analysis of the lower tropospheric disturbances in the tropics during the northern summer of the
FGGE year. II. Regional characteristics of the disturbances. Pure Appl. Geophys., 123: 272–92.
Nordeng, T.E. and Rasmussen, E.A. 1992. A most beautiful polar low: a case study of a polar low
development in the Bear Island region. Tellus, 44A: 81–99.
Norquist, D.C., Reckes, E.E. and Reed, R.J. 1977. The energetics of African wave disturbances as
observed during phase III of GATE. Mon. Wea. Rev., 105 (3): 334–42.
0 Økland, H. 1987. Heating by organized convection as a source of polar low intensification. Tellus,
39A: 397–407.
Oliver, V.J. 1969. Tropical storm classification system. In: Satellite Meteorology, Australian Bureau
of Meteorology, Melbourne, pp. 27–9.
Oliver, V.J. and Anderson, R.K. 1969. Circulation in the tropics as revealed by satellite data. Bull.
Amer. Met. Soc., 50: 702–7.
Ooyama, K. 1964. A dynamical model for the study of tropical cyclone development. Geofis. Int.,
4: 187–98.
Ooyama, K. 1982. Conceptual evolution of the theory and modelling of the tropical cyclone. J.
Met. Soc. Japan, 60: 369–80.
Orlanski, I. 1968. Instability of frontal waves. J. Atmos. Sci., 25: 178–200.
0 Orlanski, I. 1998. Poleward deflection of storm tracks. J. Atmos. Sci., 55 (16): 2577–602.
Ottenbacher, A., Tomassini, M., Holmlund, K., and Schmetz, J. 1997. Low-level cloud motion
winds from Meteosat high-resolution visible imagery. Wea. Forecast., 12: 175–84.
Palmén, E. 1951. The aerology of extratropical cyclones. In: T.F. Malone, ed., Compendium of
Meteorology, Amer. Met. Soc., Boston, MA, pp. 599–620.
Palmén, E. and Newton, C.W. 1969. Atmospheric Circulation Systems: Their Structure and Physical
Interpretation. Academic Press, New York, 603 pp.
Palmer, C.E. 1950. Tropical meteorology. In: T.F. Malone, ed., Compendium of Meteorology. Amer.
Met. Soc., Boston MA, pp. 859–80.
Palmer, C.E. 1952. Reviews of modern meteorology. 5. Tropical meteorology. Quart. J. Roy. Met.
Soc., 78: 126–64.
0 Palmer, C.E. and Ohmstede, W.D. 1956. The simultaneous oscillation of barometers along or near
the equator. Tellus, 8: 495–507.
Pankiewicz, G.S. 1995. Pattern recognition techniques for the identification of cloud and cloud
11 systems. Met. Appl., 2: 257–71.
1 Parker, D.J. 1998. Secondary frontal waves in the North Atlantic region: a dynamic perspective of
current ideas. Quart. J. Roy. Met. Soc., 124: 829–56.
Parker, S.S., Hawes, J.T., Colucci, S.J., and Hayden, B.P. 1989. Climatology of 500 mb cyclones
and anticyclones, 1950–85. Mon. Wea. Rev., 117: 558–70.
Pearson, G.M. and Stewart, S.J. 1994. A diagnostic study of an apparent “instant occlusion” cyclo-
genesis event during ERICA. Atmos.-Ocean, 32: 259–84.
Pedder, M.A. 1981. Practical analysis of dynamical and kinematic structure: some applications and
a case study. In: B.W. Atkinson, ed., Atmospheric Circulation Systems: Their Structure and
Physical Interpretation, Academic Press, New York, pp. 169–86.
Peng, M.S. and Chang, S.W. 1996. Impacts of SSM/I retrieved rainfall rates on numerical predic-
tion of a tropical cyclone. Mon. Wea. Rev., 124: 1181–98.
Penner, C.M. 1955. A three-front model for synoptic analysis. Quart. J. Roy. Met. Soc., 81: 89–91.
Perry, M.D. 1990. “A Satellite Climatology of Mesoscale Convective Systems for the Southwestern
United States.” Unpubl. Masters thesis, Department of Geography, Indiana University,
Bloomington IN.
Petterssen, S. 1950. Some aspects of the general circulation of the atmosphere. Centen. Proc. Roy.
Met. Soc. (London), pp. 120–55.
Petterssen, S. 1955. A general survey of factors influencing development at sea level. J. Met., 12:
36–42.
Petterssen, S. and Smeybe, S.J. 1971. On the development of extratropical cyclones. Quart. J. Roy.
Met. Soc., 97: 457–82.
Petterssen, S., Bradbury, D.L., and Pedersen, K. 1962. The Norwegian cyclone model in relation
to heat and cold sources. Geofys. Publ. 24: 243–80.
Petty, G.W. 1994. Physical retrievals of over-ocean rain rate from multichannel microwave imagery.
Met. Atmos. Phys., 54: 79–100, 101–22.
Petty, G.W. and Katsaros, K.B. 1992. Nimbus-7 SMMR precipitation observations calibrated against
surface radar during TAMEX. J. Appl. Met., 31: 489–505.
Petty, G.W. and Miller, D.K. 1995. Satellite microwave observations of precipitation correlated
with intensification rate in extratropical oceanic cyclones. Mon. Wea. Rev., 123: 1904–15.
Piddington, H. 1842. The Sailor’s Hornbook for the Law of Storms. Wiley, New York and London.
Pielke, R.A. 1990. The Hurricane. Routledge, London, 228 pp.
Pielke, R.A. and Zeng, X. 1989. Influence on severe storm development of irrigated land. Nat. Wea.
Digest, 14 (2): 16–17.
Piersig, W. 1944. The cyclonic disturbances of the subtropical eastern Atlantic. Bull. Amer. Met.
Soc., 25: 2–17 (translation of Parts 2 and 3 of paper in Arch. Dtsch. Seewarte 54 (6), 1936).
Porter, D.L. 1990. Geosat observations of the tropical Pacific cyclone pair of May 1986. J. Geophys.
Res., 95: 3705–20.
Rabbe, A. 1987. A polar low over the Norwegian Sea, 29 February–1 March 1984. Tellus, 39A:
326–33.
Rabin, R.M. and Martin, D.W. 1996. Satellite observations of shallow cumulus coverage over the
central United States: an exploration of land use impact on cloud cover. J. Geophys. Res., 101:
7149–55.
Rabin, R.M., Stadler, S.J., Wetzel, P.J., Stensrud, D.J., and Gregory, M. 1990. Observed effects of
landscape variability on convective clouds. Bull. Amer. Met. Soc., 71: 272–80.
Ramage, C.S. 1962. The subtropical cyclone. J. Geophys. Res., 67: 1401–11.
Ramage, C.S. 1971 Monsoon Meteorology. Academic Press, New York, 296 pp.
Ramage, C.S. 1995. Forecaster’s Guide to Tropical Meteorology, Air Weather Service AWSR/TR-
95/001, Scott Air Force Base IL, 392 pp.
Rao, G.V. and MacArthur, P.D. 1994. The SSM/I estimated rainfall amounts of tropical cyclones
and their potential in predicting the cyclone intensity changes. Mon. Wea. Rev., 122: 1568–74.
Rasmussen, E. 1979. The polar low as an extratropical CISK disturbance. Quart. J. Roy. Met. Soc.,
105: 531–49.
Rasmussen, E. 1981. An investigation of a polar low with a spiral cloud structure. J. Atmos. Sci.,
38: 1785–92.
Rasmussen, E. 1989. A comparative study of tropical cyclones and polar lows, In: P.F. Twitchell,
E.A. Rasmussen, and K.L. Davidson, eds, Polar and Arctic Lows, Deepak, Hampton VA, pp.
47–80.
11 Rasmussen, E.A., Claud, C., and Purdom, J.F. 1996. Labrador Sea polar lows. Global Atmos.-Ocean
Sys., 4 (2–4): 275–333.
Rasmussen, E., Pedersen, T.S., Pedersen, L.T., and Turner, J. 1992. Polar lows and arctic insta-
bility lows in the Bear Island region. Tellus, 44A: 133–54.
Rasmusson, E.M. and Arkin, P.A. 1993. A global view of large-scale precipitation variability.
J. Climate, 6 (8): 1595–22.
Rawson, H.E. 1980. The anticyclonic belt of the southern hemisphere. Quart. J. Roy. Met. Soc.,
34: 165–88.
Rawson, H.E. 1909. The anticyclonic belt of the northern hemosphere. Quart. J. Roy. Met. Soc.,
35: 233–48.
0 Raymond, W.H., Rabin, R.M., and Wade, G.S. 1994. Evidence of an agricultural heat island in the
lower Mississippi river floodplain. Bull. Amer. Met. Soc., 75: 1019–25.
Read, W.L. and Maddox, R.A. 1983. Picture of the month: apparent modification of synoptic-scale
features by widespread convection. Mon. Wea. Rev., 111: 2123–8.
11 Redfield, W.C. 1831. Remarks on the prevailing storms at the Atlantic coast of the North American
states. Amer. J. Sci. and Arts, 20: 17–51.
Redfield, W.C. 1843. Observations on the storm of December 15, 1839. Trans. Amer. Philosophy
Soc., 8: 77–80.
Reed, R.J. 1979. Cyclogenesis in polar air streams. Mon. Wea. Rev., 107: 38–52.
Reed, R.J. and Recker, E.E. 1971. Structure and properties of synoptic-scale wave disturbances in
the equatorial western Pacific. J. Atmos. Sci., 28 (7): 1117–33.
0
Reed, R.J., Norquist, D.C., and Recker, E.E. 1977. The structure and properties of African wave
disturbances as observed during Phase III of GATE. Mon. Wea. Rev., 105 (3): 317–33.
Reed, R.J., Hollingsworth, A., Heckley, W.A., and Delsol, F. 1988. An evaluation of the perform-
ance of the ECMWF operational system in analyzing and forecasting easterly wave disturbances
over Africa and the tropical Atlantic. Mon. Wea. Rev., 116 (4): 824–65.
Reiter, E.R. and Whitney, L.F. 1969. Interactions between subtropical and polar-front jet stream.
Mon. Wea. Rev., 97: 432–8.
Renard, R.J. and Clarke, L.C. 1965. Experiments in numerical objective analysis. Mon. Wea. Rev.,
93: 547–56.
Richardson, L.F. 1922. Weather Prediction by Numerical Process, Cambridge University Press,
0 Cambridge, 236 pp. (reprinted 1965 by Dover Publications, New York).
Riehl, H. 1945. Waves in the Easterlies and the Polar Front in the Tropics. Miscellaneous Report
17, Department of Meteorology, University of Chicago, 79 pp.
Riehl, H. 1954. Tropical Meteorology. McGraw-Hill, New York, chapters 9, 10.
Riehl, H. 1973. Controls of the Venezuela rainy seasons. Bull. Amer. Met. Soc., 54 (1): 9–12.
Riehl, H. 1979. Climate and Weather in the Tropics. Academic Press, London, chapter 8.
Riehl, H., Alaka, M.A., Jordan, C.L., and Renard, R.J. 1954. The Jet Stream. Met. Monogr., 2 (7):
100 pp.
Rocky, C.C. and Braaten, D.A. 1995. Characterization of polar cyclonic activity and relationship
to observed snowfall events at McMurdo Station, Antarctica. In: Preprint vol., Fourth Conference
on Polar Meteorology and Oceanography, Dallas, Texas, January 15–20, 1995. Amer. Met. Soc.,
0 Boston, MA, pp. 244–5.
Rodgers, E.B. and Pierce, H.F. 1995. A satellite observational study of precipitation characteristics
in western North Pacific tropical cyclones. J. Appl. Met., 34: 2587–99.
Rodgers, E.B., Chang, S.W., and Pierce, H.F. 1994. A satellite observational and numerical study
of precipitation characteristics in western North Atlantic tropical cyclones. J. Appl. Met., 33:
129–39.
Rodgers, E.B., Chang, S.W., Stout, J., Steranka, J., and Shi, J.-J. 1991. Satellite observations of
variations in tropical cyclone convection caused by upper-tropospheric troughs. J. Appl. Met., 30:
1163–84.
Rodgers, E.B., Stout, J., Steranka, J., and Chang, S. 1990. Tropical cyclone–upper atmospheric
interactions as inferred from satellite total ozone observations. J. Appl. Met., 29: 934–54.
0 Rodgers, D.M., Howard, K.W., and Johnston, E.C. 1983. Annual summary: mesoscale convective
complexes over the United States during 1982. Mon. Wea. Rev., 111: 2363–9.
Rodgers, D.M., Magnano, M.J., and Arns, J.H. 1985. Annual summary: mesoscale convective
11 complexes over the United States during 1983. Mon. Wea. Rev., 113: 888–901.
1 Roebber, P.J. 1984. Statistical analysis and updated climatology of explosive cyclones. Mon. Wea.
Rev., 112 (8): 1577–89.
Rogers, C.W.C. and Sherr, P.E. 1966. Toward the Dynamical Interpretation of Satellite-observed
Extratropical Vortical Cloud Patterns. Final Rept, Contract Cwb-11123, ARACON Geophysics
Div., Concord MA. Prepared for Environmental Satellite Services Administration, National
Environment Satellite Center, May 1966. 135 pp.
Rogers, J.C. and van Loon, H. 1979. The seesaw in winter temperatures between Greenland and
northern Europe. II. Some oceanic and atmospheric effects in middle and high latitudes. Mon.
Wea. Rev., 107: 509–19.
Rossby, C.-G. et al. 1939. Relation between variations in the intensity of the zonal circulation of the
atmosphere and the displacement of the semi-permanent centers of action. J. Marine Res., 2: 38–55.
Rotunno, R. and Emmanuel, K.A. 1987. An air–sea interaction theory for tropical cyclones. II.
Evolutionary study using a non-hydrostatic axisymmetric numerical model. J. Atmos. Sci., 44:
542–61.
Russell, H.C. 1893. Moving anticyclones of the southern hemisphere. Quart. J. Roy. Met. Soc., 19:
23–34.
Rykachev, M.A. 1896. Tipy putei tsiklonov v Evrope po nabludeniyam 1872–1887 gg (Types of
cyclone paths in Europe for 1872–1887). Zapisky Akad. Nauk, Ser. 8 Fiz.-Mat. Otdelenie (St
Petersburg), 3 (3), 71 pp.
Sadler, J.C. 1975. The Upper Tropospheric Circulation over the Global Tropics. Department of
Meteorology over the Global Tropics, University of Hawaii, Honolulu (UHMET 75–05), 35 pp.
Sadler, J.C. 1976. A role of the upper-tropospheric trough in early season typhoon development.
Mon. Wea. Rev., 104: 1137–52.
Sadler, J.C. and Wann, T.C. 1984. Mean upper tropospheric flow over the global tropics. Air
Weather Service, Tech. Rep. 83/002, Vol. II (Charts).
Sanders, F. 1986. Explosive cyclogenesis in the west-central North Atlantic Ocean. I. Composite
structure and mean behavior. Mon. Wea. Rev., 114: 1781–94.
Sanders, F. 1999. A proposed method of surface map analysis. Mon. Wea. Rev., 127 (6): 945–55.
Sanders, F. and Gyakum, J.R. 1980. Synoptic climatology of the “bomb.” Mon. Wea. Rev., 108
(10): 1589–606.
Sanders, F. and Hoskins, B.J. 1990. An easy method for estimation of Q-vectors from weather
maps. Wea. Forecasting, 5: 346–53.
Scherhag, R. 1934. Zur Theorie des Hoch- und Tiefdruckgebiete. Die Bedeutung der Divergenz in
Druckfeldern. Met. Zeitschr., 51: 129–38.
Schwartz, M.D. 1991. An integrated approach to air mass classification in the north-central United
States. Prof. Geogr., 43: 77–91.
Seaman, R., Steinle, P., Bourke, W., and Hart, T. 1993. The impact of manually derived southern
hemisphere sea level pressure data upon forecasts from a global model. Wea. Forecast., 8: 363–8.
Sekioka, M. 1970. On the behavior of cloud patterns as seen on satellite photographs in the trans-
formation of a typhoon into an extratropical cyclone. J. Met. Soc. Japan, 48: 224–32.
Serreze, M.C. 1995. Climatological aspects of cyclone development and decay in the Arctic. Atmos.-
Ocean, 33: 1–23.
Serreze, M.C. and Barry, R.G. 1988. Synoptic activity in the Arctic Basin, 1979–85. J. Climate,
1 (12): 1276–95.
Serreze, M.C., Box, J.E., Barry, R.G., and Walsh, J.E. 1993. Characteristics of Arctic synoptic
activity, 1952–89. Met. Atmos. Phys., 51: 147–64.
Serreze, M.C., Carse, F., Barry, R.G., and Rogers, J.C. 1997. Icelandic low cyclone activity: clima-
tological features, linkages with the NAO, and relationships with recent changes in the northern
Serreze, M.C., Maslanik, J.A., Preller, R., and Barry, R.G. 1990. Sea ice concentrations in the
Canada Basin during 1988: comparisons with other years and evidence of multiple forcing mech-
anisms. J. Geophys. Res., 95 (C12): 22253–67.
Shapiro, L.J. 1986. The three-dimensional structure of synoptic-scale disturbances over the tropical
Atlantic. Mon. Wea. Rev., 114 (10): 1876–91.
Shapiro, M.A. and Fedor, L.S. 1989. A case study of an ice-edge boundary layer front and polar
low development over the Norwegian and Barents seas. In: P.F. Twitchell, E.A. Rasmussen, and
K.L. Davidson, eds, Polar and Arctic Lows, Deepak, Hampton VA, pp. 257–77.
11 Shapiro, M.A. and Grell, E.D. 1994. In search of synoptic/dynamic conceptualizations of the life
cycles of fronts, jet streams, and the tropopause. In: S. Grønäs and M.A. Shapiro, eds, The Life
Shapiro, M.A. and Keyser, D. 1990. Fronts, jet streams and the tropopause. In: C.W. Newton and
E.O. Holopainen, eds, Extratropical Cyclones: The Eric Palmén Memorial Volume, Amer. Met.
Shapiro, M.A., Fedor, L.S., and Hampel, T. 1987. Research aircraft measurements of a polar low
over the Norwegian Sea. Tellus, 39A: 272–306.
Shapiro, L.J., Stevens, D.E., and Cieselski, D.E. 1988. A comparison of observed and model-derived
structures of Caribbean easterly waves. Mon. Wea. Rev., 116 (4): 921–38.
0 Shenk, W.E. and Brooks, E.M. 1965. Thermal and wind structures related to major cloud bands of
TIROS-photographed extratropical vortices. J. Appl. Met., 4: 676–92.
Sherr, P.E. and Rogers, C.W.C. 1965. The Identification and Interpretation of Cloud Vortices using
TIROS Infra-red Observations. Final Rept, Contract No. Cwb-18812, ARACON Geophysics Co.,
11 74 pp.
Simpson, R.H. 1974. The hurricane disaster potential scale. Weatherwise, 27: 169, 186.
Sinclair, M.R. 1994. An objective cyclone climatology for the southern hemisphere. Mon. Wea.
Rev., 122: 2239–56.
Sinclair, M.R. 1995. A climatology of cyclogenesis for the southern hemisphere. Mon. Wea. Rev.,
123: 1601–19.
Sinclair, M.R. 1996. A climatology of anticyclones and blocking for the southern hemisphere. Mon.
0
Wea. Rev., 124: 245–63.
Sinclair, M.R. 1997. Objective identification of cyclones and their circulation intensity and clima-
tology. Wea. Forecasting 12: 595–612.
Sinclair, M.R. and Cong, X. 1992. Polar airstream cyclogenesis in the Australasian region: a
composite study using ECMWF analyses. Mon. Wea. Rev., 120: 1950–72.
Sinclair, M.R., Renwick, J.A., and Kidson, J.W. 1997. Low-frequency variability of southern hemi-
sphere sea-level pressure and weather system activity. Mon. Wea. Rev., 125: 2531–43.
Smigielski, F.J. and Mogil, M.H. 1995. A systematic approach for estimating central surface pres-
sures of mid-latitude cold season oceanic cyclones. Tellus, 47A: 876–91.
Smith, R.K. 1997. On the theory of CISK. Quart. J. Roy. Met. Soc., 123: 407–18.
0 Song, Y. and Carleton, A.M. 1997. Climatological “models” of cold air mesocyclones derived from
SSM/I data. Geocarto Int., 12: 79–89.
Sortais, J.L., Cammas, P.-L., Yu, X.D., Richard, E., and Rosset, R. 1993. A case-study of coupling
between low- and upper-level jet-front systems: investigation of dynamical and diabatic processes.
Mon. Wea. Rev., 121 (8): 2339–53.
Sovetova, V.D. and Grigorov, S.I. 1978. Utilization of satellite information on cloud vortexes to
determine pressure characteristics. Soviet Met. and Hydrol., 2: 15–22.
Steranka, J., Rodgers, E.B., and Gentry, R.C. 1984. The diurnal variation of Atlantic Ocean trop-
ical cyclone cloud distribution inferred from geostationary satellite infrared measurements. Mon.
Wea Rev., 112: 2338–44.
Stewart, R.E., Shaw, R.W., and Isaac, G.A. 1997. Canadian Atlantic Storms Program: the meteo-
0 rological field project. Bull. Amer. Met. Soc., 68 (4): 338–45.
Stewart, R.E., Szeto, K.K., Reinking, R.F., Clough, S.A., and Ballard, S.P. 1998. Midlatitude cyclonic
cloud systems and their features affecting large scales and climate. Rev. Geophys., 36 (2): 245–73.
Stout, J. and Rodgers, E.B. 1992. Nimbus-7 total ozone observations of western North Pacific trop-
ical cyclones. J. Appl. Met., 31: 758–83.
Streten, N.A. 1968a. A note on multiple image photo-mosaics for the southern hemisphere. Aust.
Met. Mag., 16: 127–36.
Streten, N.A. 1968b. Some aspects of high latitude southern hemisphere summer circulation as
viewed by ESSA 3. J. Appl. Met., 7: 324–32.
Streten, N.A. 1970. A note on the climatology of the satellite observed zone of high cloudiness in
the central South Pacific. Aust. Met. Mag., 18: 31–8.
0 Streten, N.A. 1973. Some characteristics of satellite-observed bands of persistent cloudiness over
the southern hemisphere. Mon. Wea. Rev., 101: 486–95.
Streten, N.A. 1974. A satellite view of weather systems over the North American Arctic. Weather,
11 29: 369–80.
1 Streten, N.A. 1975. Satellite-derived inferences to some characteristics of the South Pacific atmos-
pheric circulation associated with the Niño event of 1972–73. Mon. Wea. Rev., 103: 989–95.
Streten, N.A. 1978. A quasi-periodicity in the motion of the South Pacific cloud band. Mon. Wea.
Rev., 106: 1211–14.
Streten, N.A. and Downey, W.K. 1977. Defence Meteorological Satellite Program (DMSP) imagery:
a research tool for the Australian region. Aust. Met. Mag., 25: 25–36.
Streten, N.A. and Kellas, W.R. 1973. Aspects of cloud pattern signatures of depressions in matu-
rity and decay. J. Appl. Met., 12: 23–7.
Streten, N.A. and Troup, A.J. 1973. A synoptic climatology of satellite observed cloud vortices
over the southern hemisphere. Quart. J. Roy. Met. Soc., 99: 56–72.
Sui, C.-H. and Lau, K.-M. 1992. Multiscale phenomena in the tropical atmosphere over the western
Pacific. Mon. Wea. Rev., 120: 407–30.
Sutcliffe, R.C. 1939. Cyclonic and anticyclonic development. Quart. J. Roy. Met. Soc., 65: 519–24.
Sutcliffe, R.C. 1947. A contribution to the problem of development. Quart. J. Roy. Met. Soc., 73:
370–83.
Sutcliffe, R.C. and Forsdyke, A.G. 1950. The theory and use of upper air thickness patterns in fore-
casting. Quart. J. Roy. Met. Soc., 76: 189–217.
Takayabu, Y.N. and Nitta, T. 1993. Three to five-day period disturbances coupled with convection
over the tropical Pacific Ocean. J. Met. Soc. Japan, 71: 221–46.
Taljaard, J. 1967. Development, distribution, and movement of cyclones and anticyclones in the
southern hemisphere during the IGY. J. Appl. Met., 6: 973–87.
Tang, W., Brooks, E.M., and Watson, B.F. 1964. Theoretical and Observational Studies of Vortex
Cloud Patterns. GCA Tech. Rept No. 64–2–G, Final Rept, Contract Cwb-10626, Geophysics
Corporation of America, Bedford MA. Prepared for US Weather Bureau, Washington DC, 133
pp.
Tapp, R.G. and Barrell, S.L. 1984. The north-west Australian cloud band: climatology, character-
istics and factors associated with development. J. Climatol., 4: 411–24.
Thepenier, R.-M. and Cruette, D. 1981. Formation of cloud bands associated with the American
subtropical jet stream and their interaction with midlatitude synoptic disturbances reaching Europe.
Mon. Wea. Rev., 109: 2209–20.
Thompson, R.M., Jr., Payne, S.W., Recker, E.E., and Reed, R.J. 1979. Structure and properties of
synoptic-scale wave disturbances in the intertropical convergence zone of the eastern Atlantic.
J. Atmos. Sci., 36 (1): 53–72.
Thorncroft, C.D. and Hoskins, B.J. 1990. Frontal cyclogenesis. J. Atmos. Sci., 47: 2317–36.
Thorncroft, C.D. and Hoskins, B.J. 1994. An idealised study of African easterly waves. I. A linear
view. II. A non-linear view. Quart. J. Roy Met. Soc., 120: 953–82, 983–1015.
Thorpe, A.J. 1990. Frontogenesis at the boundary between air-masses of different potential vorticity.
Togstad, W.E. and Horn, L.H. 1974. An application of the satellite indirect sounding technique in
describing the hyperbaroclinic zone of a jet streak. J. Appl. Met., 13: 264–76.
Trenberth, K.E. 1976. Spatial and temporal variations of the Southern Oscillation. Quart. J. Roy.
Met. Soc., 102: 639–53.
Trenberth, K.E. 1978. On the interpretation of the diagnostic quasi-geostrophic omega equation.
Mon. Wea. Rev., 106: 131–7.
Trenberth, K.E. 1991. Storm tracks in the southern hemisphere. J. Atmos. Sci., 48 (19): 2159–78.
Troup, A.J. and Streten, N.A. 1972. Satellite-observed southern hemisphere cloud vortices in rela-
tion to conventional observations. J. Appl. Met., 11: 909–17.
Turner, J. and Thomas, J.P. 1994. Summer-season mesoscale cyclones in the Bellingshausen–
Weddell region of Antarctica and links with the synoptic-scale environment. Intl. J. Climatol.,
14: 871–94.
Turner, J., Corcoran, G., Cummins, S., Lachlan-Cope, T., and Leonard, S. 1996. Seasonal vari-
ability of mesocyclone activity in the Bellingshausen/Weddell region of Antarctica. Global
Atmos.-Ocean Sys., 5 (1): 73–97.
Turner, J., Lachlan-Cope, T.A., and Thomas, J.P. 1993. A comparison of Arctic and Antarctic
mesoscale vortices. J. Geophys. Res., 98: 13019–34.
Uccellini, L.W. 1990. Processes contributing to the rapid development of extratropical cyclones.
In: C.W. Newton and E.O. Holopainen, eds, Extratropical Cyclones: The Erik Palmén Memorial
Volume. Amer. Met. Soc., Boston MA, pp. 81–105.
11 Uccellini, L.W., Corfidi, S.F., Junker, N.W., Kocin, P.J., and Olson, D.A. 1992. Report on the
Surface Analysis Workshop held at the National Meteorological Center, March 25–28 1991. Bull.
Amer. Met. Soc., 73 (4): 459–72.
van Bebber, J. 1891. Die Zugstrassen der barometrischen Minima nach den Bahnenkarten der
Deutschen Seewarte für den Zeitraum 1875–1890. Met. Zeit., 8: 361–6.
Van Loon, H. 1967. The half-yearly oscillations in middle and high southern latitudes and the core-
less winter. J. Atmos. Sci., 24: 472–86.
Van Loon, H. and Rogers, J.C. 1984. Interannual variations in the half-yearly cycle of pressure
gradients and zonal wind at sea level on the southern hemisphere. Tellus, 36A: 76–86.
Van Loon, H. and Shea, D.J. 1985. The Southern Oscillation. IV. The precursors south of 15oS to
0 the extremes of the oscillation. Mon. Wea. Rev., 113: 2063–74.
Van Loon, H. and Thompson, A.H. 1966. A note on southern hemisphere analysis incorporating
satellite information. Notos, 15: 91–7.
Velasco, I. and Fritsch, J.M. 1987. Mesoscale convective complexes in the Americas. J. Geophys.
11 Res., 92: 9591–613.
Velden, C.S. 1989. Observational analyses of North Atlantic tropical cyclones from NOAA polar
orbiting satellite microwave data. J. Appl. Met., 28: 59–70.
Velden, C.S. 1992. Satellite-based microwave observations of tropopause-level thermal anomalies:
qualitative applications in extratropical cyclone events. Wea. Forecast., 7: 669–82.
Velden, C.S. and Smith, W.L. 1983. Monitoring tropical cyclone evolution with NOAA-satellite
microwave observations. J. Clim. Appl. Met., 22: 714–24.
0 Velden, C.S, Goodman, B.M., and Merrill, R.T. 1991. Western North Pacific tropical cyclone inten-
sity estimation from NOAA polar-orbiting satellite microwave data. Mon. Wea. Rev., 119: 159–68.
Velden, C.S., Hayden, C.M., Menzel, W.P., Franklin, J.L., and Lynch, J.S. 1992. The impact of
satellite-derived winds on numerical hurricane track forecasting. Wea. Forecast., 7: 107–18.
Velden, C.S., Olander, T.L., and Zehr, R.M. 1998. Development of an objective scheme to esti-
mate tropical cyclone intensity from digital geostationary satellite infrared imagery. Wea.
Forecast., 13: 172–86.
Viezee, W., Endlich, R.M., and Serebreny, S.M. 1967. Satellite-viewed jetstream clouds in relation
to the observed wind field. J. Appl. Met., 6: 929–35.
von Hann, J. 1901. Lehrbuch der Meteorologie, 1st edition, Leipzig.
0 Waliser, D.E. and Gautier, C. 1993. A satellite-derived climatology of the ITCZ. J. Climate, 6 (11):
2162–74.
Waliser, D.E., Graham, N.E., and Gautier, G. 1993. Comparison of the Highly Reflective Cloud
and Outgoing Long-wave Radiation datasets for use in estimating tropical deep convection. J.
Climate, 6: 331–53.
Wallace, J.M. 1973. General circulation of the tropical lower stratosphere. Rev. Geophys. Space
Phys., 11: 191–222.
Wallace, J.M. and Chang, C.P. 1969. Spectrum analysis of large-scale wave disturbances in the
tropical lower troposphere. J. Atmos. Sci., 26 (5): 1010–25.
Wallace, J.M., Lim, G.-H., and Blackmon, M.L. 1988. On the relationship between cyclone tracks,
anticyclone tracks and baroclinic wave guides. J. Atmos. Sci., 45 (3): 439–62.
0 Weaver, C.P. and Ramanathan, V. 1996. The link between summertime cloud radiative forcing and
extratropical cyclones in the North Pacific. J. Climate, 9: 2093–109.
Webster, P.J. 1983. Large-scale structure of the tropical atmosphere. In: B.J. Hoskins and R.P. Pearce,
eds, Large-scale Dynamical Processes in the Atmosphere, Academic Press, London, pp. 235–75.
Wernli, H. and Davies, H.C. 1997. A Lagrangian-based analysis of extratropical cyclones. I. The
method and some applications. Quart. J. Roy. Met. Soc., 123: 467–89.
Weston, K.J. 1980. An observational study of convective cloud streets. Tellus, 32: 433–8.
Wetzel, P.J. 1990. A simple parcel method for prediction of cumulus onset and area-averaged cloud
amount for heterogeneous land surfaces. J. Appl. Met., 29: 516–23.
Wetzel, P.J., Cotton, W.R., and McAnelly, R.L. 1983. A long-lived mesoscale convective complex.
II. Evolution and structure of the mature complex. Mon. Wea. Rev., 111: 1919–37.
0 Whitaker, L.M. and Horn, L.H. 1984. Northern hemisphere extratropical cyclone activity for four
mid-season months. J. Climatol., 4 (3): 297–310.
Whitaker, J.S. and Dole, R.M. 1995. Organization of storm tracks in zonally varying flows.
11 J. Atmos. Sci., 52 (8): 1178–91.
1 Whitaker, J.S. and Sardeshmukh, P.D. 1998. A linear theory of extratropical synoptic eddy statistics.
J. Amos. Sci., 55 (2): 237–58.
Whitney, L.F. Jr. 1966. On locating jet streams from TIROS photographs. Mon. Wea. Rev., 94:
129–38.
Whitney, L.F., Jr. and Herman, L.D. 1968. The Nature of Intermediate-scale Cloud Spirals. ESSA
Tech. Rept NESC-45, Washington DC, 69 pp.
Widger, W.K., Jr. 1964. A synthesis of interpretations of extratropical vortex patterns as seen by
TIROS. Mon. Wea. Rev., 92: 263–82.
Wilhelmsen, K. 1985. Climatological study of gale-producing polar lows near Norway. Tellus, 37A:
451–9.
Williams, R.T., Peng, M.S., and Zankofski, D.A. 1992. Effects of topography on fronts. J. Atmos.
Sci., 49 (4): 287–305.
Willoughby, H.E. 1998. Tropical cyclone eye thermodynamics. Mon. Wea. Rev., 126 (12): 3053–67.
Willoughby, H.E., Marks, F.D., Jr, and Feinberg, F.J. 1984. Stationary and propagating convective
bands in asymmetric hurricanes. J. Atmos. Sci., 41: 3189–277.
Wright, W.J. 1988. The low latitude influence on winter rainfall in Victoria, south-eastern Australia.
I. Climatological aspects. J. Climatol., 8: 437–62.
Wright, W.J. 1997. Tropical–extratropical cloudbands and Australian rainfall. I. Climatology. Intl.
J. Climatol., 17: 807–29.
Wu, Y. and Raman, S. 1997. Effect of land-use pattern on the development of low-level jets.
J. Appl. Met., 36: 573–90.
Wylie, D.P., Hinton, B.B., and Millett, K.M. 1981. A comparison of three satellite-based methods
for estimating surface winds over oceans. J. Appl. Met., 20: 439–49.
Yarnal, B. and Henderson, K.G. 1989a. A satellite-derived climatology of polar-low evolution in
the North Pacific. Intl. J. Climatol., 9: 551–66.
Yarnal, B. and Henderson, K.G. 1989b. A climatology of polar low cyclogenetic regions over the
North Pacific Ocean. J. Climate, 2: 1476–91.
Young, J.A. 1987. Physics of monsoons: the current view. In: J.S. Fein and P.L. Stephens, eds,
Monsoons. Wiley, New York, pp. 211–43.
Zehr, R. 1992. Tropical Cyclogenesis in the Western North Pacific. NOAA Technical Report,
NESDIS 16, Washington DC, 181 pp.
Zhong, S., Fast, J.D., and Bian, X. 1996. A case study of the Great Plains low-level jet using wind
profiler network data and a high-resolution mesoscale model. Mon. Wea. Rev., 124: 785–806.
Zick, C. 1983. Method and results of an analysis of comma cloud development by means of vorticity
fields from upper tropospheric satellite wind data. Met. Rundsch., 36: 69–84.
Zillman, J.W. and Martin, D.W. 1968. A sharp cold frontal passage at Macquarie Island in the
Southern Ocean. J. Appl. Met., 7: 708–12.
Zillman, J.W. and Price, P.G. 1972. On the thermal structure of mature Southern Ocean cyclones.
Aust. Met. Mag., 20: 34–48.
Zillman, J.W., Griersmith, D., LeMarshall, J., and Gauntlett, D.J. 1990. Remote sensing applica-
tions in the Australian Bureau of Meteorology. Intl. J. Remote Sens., 11: 1979–97.
Zipser, E.J. 1970. The Line Islands Experiment: its place in tropical meteorology and the rise of
the fourth school of thought. Bull. Amer. Met. Soc., 51 (12): 1136–45.
Zipser, E.J. 1982. Use of a conceptual model of the life-cycle of mesoscale convective systems to
improve very short-range forecasts. In: K.A. Browning, ed., Nowcasting, Academic Press, London
and New York, pp. 191–204.
Zishka, K.M. and Smith, P.J. 1980. The climatology of cyclones and anticyclones over North
America and surrounding ocean environs for January and July 1950–77. Mon. Wea. Rev., 108:
387–401.
Zwatz-Meise, V. and Hailzl, G. 1980. Interpretation of so-called shear bands in satellite images.
Arch. Met. Geophys. Bioklim., B28: 299–315.
11
7 Synoptic climatology and its
applications
Roger G. Barry and Allen H. Perry
11
7.1 Synoptic pattern classification
There is a long history of interest in classifying the pressure patterns on synoptic weather
maps into a small number of categories for the purpose of studying local climatic
conditions stratified on a meaningful basis. Arbitrary monthly averages conceal much
information as to the nature of the controlling weather processes, while three or six-hourly
0 synoptic observations usually give too much data for convenient summarization.
The determination of categories of atmospheric circulation types prevailing in a
particular area or region is the first stage in a synoptic climatology analysis. In addition,
a synoptic typing scheme offers a descriptive tool for summarizing typical modes of the
atmospheric circulation with the potential for reducing data volume (Key and Crane,
1986). The assumption that is implicit in the categorization procedure is that the weather
conditions associated with a type are more or less homogeneous, and that the differences
among types are relatively marked. Identification of discrete categories of pressure patterns
or circulation types, as they are commonly called, presents several problems:
0 1 Atmospheric modes are continuous, so that the delimitation of any boundary between
classes is extremely difficult. Although patterns may switch abruptly, on other occa-
sions a gradual evolution takes place.
2 Pressure systems can be of variable intensities, and decisions have to be made as to
when to include or neglect minor features such as a small weak low or transitory
ridge.
3 There may be seasonal variation in type characteristics. Generally weather systems
are of greatest intensity in winter, when equator–pole heat differences are greatest.
Considerable effort has been expended on deriving appropriate classifications of weather

0 systems, and Smithson (1988) notes that some researchers regard this topic as the prin-
cipal purpose of synoptic climatology.
The variables most commonly analyzed in synoptic climatological studies are the MSL
pressure field and 700 mb or 500 mb contour fields. Daily pressure maps covering
most of the northern hemisphere are available from 1899 and for the North Atlantic
sector since 1873. Height data (500 mb) essentially begin from 1946 for the northern
hemisphere. However, Lambert (1990) identifies discontinuities in the US National
Meteorological Center (NMC) 500 mb fields due to analysis changes in 1953, 1955, 1962
and 1978; those in 1962 are particularly serious. A reanalysis project for the entire record
from 1957 to 1996 is in progress at the National Center for Environmental Prediction
0 (Kalnay et al., 1996). This uses a consistent scheme, T62 (210 km) model resolution
and an augmented data base. A similar reanalysis project has been carried out by the
European Centre for Medium Range Weather Forecasts. Daily surface pressure and upper
11 air charts for the southern hemisphere are available from 1957 (the International
1 Geophysical Year), with different series in the 1960s and 1970s from South Africa and
Australia. Even in the 1980s, global analyses were often in error in the southern hemi-
sphere and of low quality south of 30°S (Trenberth and Olsen, 1988). Other variables
that can be analyzed in a similar manner include satellite cloud patterns and diagnostic
products of numerical analyses such as the fields of vertical velocity, horizontal diver-
gence, and relative vorticity.
The spatial scale of synoptic climatological studies tends to be rather arbitrarily deter-
mined. Some classifications refer to a small region – the British Isles, the European Alps;
others to a large sector – western Europe and the eastern North Atlantic is the area covered
by the well known Grosswetterlage scheme (Baur, 1951; Hess and Brezowsky, 1977).
This has subsequently been tabulated for 1899–1992 (Gerstengarbe and Werner, 1993).
Hemispheric patterns have been categorized independently by Dzerdzeevski (1968) and
by Vangengeim (1935), and later modified by Girs (1974). Savina (1987) illustrates the
patterns recognized in the Dzerdzeevski scheme and tabulates daily types for the northern
hemisphere over the period 1899–1985. The Vangengeim–Girs catalogue for 1891–1970
is detailed by Girs (1974).
When the objective is to investigate conditions in specific areas, however, there is no
simple way of determining a priori what sector is most critical. In a study of the response
of glacier mass balances in British Columbia and Alberta to the atmospheric circulation,
Yarnal (1984b) finds that the best scale for analysis of the 500 mb circulation pattern
may change with season and differ between the locations being studied. Brinkmann
(1999a) shows that, contrary to the expectations that, weather conditions are primarily
determined upstream in the westerlies, a grid “window” centered east of her study area
– the Lake Superior basin – gave superior classification results. The number of map types,
identified using a correlation coefficient measure of similarity for the overall grid and for
latitudinal and longitudinal sectors, was larger with the center of the grid window moved
7.5° longitude eastward, than the same distance westward. With the window extending
farther east, the classification contained more distinct well defined types that provided a
better description of the climatic elements.
Another consideration is the number of different synoptic categories recognized. This
is necessarily an arbitrary choice, although since pressure patterns display a particular
range of behavior (from zonal to meridional and blocked patterns), there is not an infi-
nite variety of possible configurations of the pressure field over a specific geographical
area. However, within this constraint, there remains a large number of overlapping possi-
bilities. Two practical concerns help determine the solution to this question. First, if the
time period under study is limited, we may specify that there be an adequate number of
cases (say more than ten) representing each group. Alternatively, or as a further criterion,
we might permit groups to be designated only if they constitute more than a certain
percentage (say 5 percent) of all cases. Such subjective decisions do in fact have an
outcome on the final classification scheme, even if the rest of the procedure is objective
(Key and Crane, 1986; Yarnal and White, 1987; Yarnal et al., 1988).
A different approach to this question was taken by Fliri (1965). For the European Alps
he was able to compare the results of classifications by F. Lauscher (nineteen types), W.
Gressel (twenty-three types), and M. Schüepp (thirty-three and 121 types) in terms of the
standard deviations of temperature and cloudiness for each classification, compared with
their climatological standard deviation values. Fliri shows that, as expected, the standard
deviations diminish as the number of types increases, but there is no appreciable reduc-
tion beyond about thirty types.
An historical overview of the development of synoptic typing methods was given by
Barry and Perry (1973) and updated in short progress reports by Barry (1980) and Perry
(1983). El-Kadi and Smithson (1992) review methods of classifying pressure patterns used
by climatologists. Yarnal (1993) has provided a comprehensive analysis of atmospheric
circulation classification procedures and hence we shall only briefly review manual, subjec-
Synoptic climatology and its applications 549
11 Table 7.1 Classification of synoptic climatology
1 Global scale
1.1 Subjective approach
Description of seasonal changes of pressure and flow fields
Characteristics of typical circulation patterns (zonal–meridional–blocked)
2 Continental scale
0 Classification of surface pressure and flow fields and important action enters
(Grosswetterlagen).
Delimitation of zonal and meridional circulation patterns
Fixing preferred positions of high and low-pressure enters
11 Classification of air masses and fronts
2.2 Objective approach
Classification of the surface characteristics and parameters of pressure and flow fields
(pressure gradients, relative vorticity flow durations)
Classification with the help of mathematical and statistical methods (e.g. principal component
analysis, orthogonal polynomials)
0
3 Regional scale
Grouping of similar pressure and flow fields (e.g. airflow, pressure field climatology)
Classification of air masses and fronts
Classification of the surface characteristics parameters of pressure and flow fields
(pressure gradients, relative vorticity flow durations)
Classification with the help of mathematical and statistical methods (e.g. principal component
analysis, orthogonal polynomials)
0
4 Local scale
The definition of surface weather types by means of measured weather parameters
(complex climatology)
Correlations between diverse local weather parameters and statistical summaries of local
weather types
Source: partly after Wanner (1980).
0
tive approaches and automated objective classifications. In addition to considering methods
of classification we shall include examples of different scale studies, illustrating many of
the categories proposed by Wanner (1980) shown in Table 7.1. Synoptic classifications
can be defined in terms of the climatological phenomena addressed as well as their spatial
applicability (Kalkstein et al., 1996), as illustrated in Figure 7.1. These various approaches
are explained below.
7.2 Subjective typing procedures

0 Until the 1960s synoptic classifications were developed through subjective manual proce-
dures. Indeed, these studies began almost as soon as weather maps were first produced
in the late nineteenth century, by meteorologists of the day like Abercromby (1883) and
11 Van Bebber and Köppen (1895). By studying long series of daily weather map sequences
1
Figure 7.1 The categorization of synoptic approaches according to spatial scale and climatological
phenomena. (Modified from Kalkstein et al., 1996)
the researcher gains familiarity with the more commonly recurring patterns of synoptic
weather map features. By noting differences, similarities, and preferred modes of atmos-
pheric circulation it is possible to abstract the most salient and frequently occurring patterns
and designate a limited number of map pattern types. One major contentious problem has
always been how many types to categorize. The recognition of numerous types will reduce
the variability of weather conditions within each category, but unless a very long record
is considered, most type categories will be represented by only a few cases, which makes
their characterization difficult. Conversely if few types are distinguished then each cate-
gory inevitably contains a wide variability of weather conditions.
Although subjective methods are still being used for particular applications (e.g.
Sturman et al., 1984; Sumner and Bonnell, 1986; Carleton, 1987; Mock et al., 1998;
Jacobheit et al., 1999), for a long period of record such analyses tend to be very time-
consuming. Also, as Key and Crane (1986) have pointed out, the subjectivity involved
makes replication of the results difficult, because the identification of types by any two
analysts may well differ. Even a single analyst may be inconsistent in the categorization
of transitional situations and weak patterns, leading to “internal drift” in a type catalogue
spanning a long time period. Tests by Frakes and Yarnal (1997b) indicate a replication
rate approaching 90 percent in winter, when patterns tend to be well defined, compared
with only 60 percent in summer, when pressure gradients are weak. However, in his orig-
inal work Lamb (1950) selected the years from 1899 to 1947 in random order to minimize
the risk of any gradual changes in the subjective manual typing.
We can broadly classify the approaches used in identifying synoptic types under three
headings:
1 The static weather map, where the location of selected features of the pressure field
is emphasized.
2 Kinematic classification, where the large-scale movement of pressure systems or
airflow is considered.
11 3 Weather element classifications. These originated in the Soviet Union in the 1920s
(Federov, 1927; Lydolph, 1959), and such studies are generally described as “complex
climatology,” i.e. dealing with weather complexes. The classical approach to air
masses can also be viewed in this same context.
Of these three, kinematic classifications, where the synoptic weather map is viewed in
terms of airflow and the movement of pressure systems, have yielded some of the best-
known and most widely used classifications, and several of these will be considered in
some detail in section 7.4, since they have been widely employed in research and
0 copied in other parts of the world. One new variant for kinematic analysis is the use of
trajectory clustering (Dorling et al., 1992). Stohl and Scheifinger (1994) calculate forty-
eight-hour back trajectories from Sonnblick Observatory, Austria, at 850 mb and then
develop a weather pattern classification by distinguishing nine directional clusters. The
11 geometric distances between the trajectory positions are minimized and the inter-group
distances maximized. More weight is assigned to the last part of each trajectory.
Previously, trajectories had been examined only for already specified airflow types,
whereas here they are used to define the types. Attempts have since been made to combine
regional classifications of weather elements with the kinematic descriptions. These involve
complex statistical procedures, as described below.
0
7.3 Objective typing procedures
With the advent of high-speed computers and digital data sets of sea-level pressure and
geopotential height fields there has been a transformation in procedures for preparing cata-
logs of synoptic types in the 1980s and 1990s (Yarnal, 1993). Initially numerical
techniques were applied rather uncritically but recent experimentation with various options
available to researchers has provided useful practical guidelines of so-called “objective”
typing methods. There are basically two approaches that can be adopted in developing a
classification:
0
1 A determination of pattern similarity based on correlation methods.
2 The use of one of a range of statistical techniques to extract components of the fields,
perhaps combined with a clustering approach to obtain pattern types.
Blasing (1975) considers that neither technique seems universally superior to the other.
“Map pattern typing appears to have distinct advantages over principal component analysis
at least when used for descriptive as opposed to predictive purposes.” However, “lack of
orthogonality between patterns can be a disadvantage if further statistical analyses are to
be performed.”
0
7.3.1 Correlation-based methods
Correlation methods were developed by Lund (1963). Examination of the similarity of
pressure patterns over an area is achieved by correlating grid-point pressure values for
each possible part of maps. The pattern (known as the key day) which has the largest
number of maps correlated with it, using an arbitrary threshold of r > 0.8, is selected as
Type A. After abstraction of these days the date with the next highest number of related
maps is designated Type B, and so on. On completion of the process, each case is
rechecked to see that it is assigned to the key-day group with which it has the highest
0 correlation. The Lund correlation method typically classifies 60–80 percent of the maps
analyzed. Although Petzold (1982) has developed a technique to improve significantly the
percentage of maps classified by this method, the fact that not every pattern can be clas-
11 sified is considered by many climatologists to be a serious drawback of Lund’s methods,
1 and Willmott (1987) suggests that correlation in any form is not generally a satisfactory
measure of the similarity between weather maps.
The map pattern correlation approach has been quite widely used, for example with
700 mb and 500 mb height data to classify tropospheric flow patterns in the western
United States (Barry, 1973; Paegle and Kierulff, 1974), and to derive synoptic patterns
over southeastern Australia by Jasper and Stern (1983). On a larger scale, Blasing and
Lofgren (1980) have used a pattern recognition algorithm based on the use of the corre-
lation coefficient to identify recurrent types of seasonal sea-level pressure anomalies over
the North Pacific–western North American area, and this work was later extended (Blasing,
1981) to analyses of characteristic type patterns in summer for the whole northern hemi-
sphere. Winter circulation anomalies over the North Pacific Ocean were classified, using
a threshold correlation coefficient of 0.5, to define quasi-stationary regimes (Horel and
Mechoso, 1988).
An alternative procedure using sums of squares, rather than inter-map correlations, was
proposed by Kirchhofer (1974). The Kirchhofer metric is the score, S:
N
S= 兺 (Z
i=1
ai Zbi)2
where Zai is the normalized grid-point value at i and day a, Zbi the corresponding value
for day b, N is the number of grid points, and:
Zi = 冤x s x冥
i
the normalized grid point value where xi is the value of pressure (or height) at grid
point i,
N
兺x
1
x= i
N i=1
and s is the standard deviation of x over the grid. Toth (1991a) finds that root-mean
square difference is significantly better, as a measure of circulation similarity, than
correlation for selecting analogs from daily 700 mb height fields. Nevertheless, this pro-
cedure is essentially equivalent to the correlation method developed by Lund (1963): S /N
2(1 r), where 0 ≤ S/N ≤ 4 (Willmott, 1987). Thus use of the Kirchhofer score with
a threshold of 1.0 N for similarity gives identical results to the correlation method using
a coefficient of 0.5. El-Kadi and Smithson (1996) find that this threshold for the total
worked well for a study of the British Isles. Blair (1998) points out that the original
Kirchhofer formulation contains an error which may bias the results. The grid-point values
are normalized over the whole array, whereas the row and column values should be
normalized separately. Less serious is the fact that the equivalence relationship with the
correlation coefficient should use the sample standard deviation rather than the popula-
tion value, i.e. N should be replaced by N 1. In an evaluation of the consequence of
these deficiencies, for 2,000 sample grids of five rows and seven columns, Blair finds
that the corrected algorithm generates more key days and that more “corrected types” are
needed to describe 80 percent of all the grids. Accordingly, published results based on
the original algorithm may contain significant biases if the separate row and column scores,
as well as the total score, were used in the classification. In some studies only the total
score is actually used. It is also worth noting that a distance function that measures the
difference between the gradient of pressure, or height, on a pair of maps appears to give
better results than either root-mean square difference or correlation (Toth, 1991a). This
function is defined as:
11 N
兺关 a
1
xai, l )2 (yai, j yai, l )兴
1/2
D= x i, j
N i=1
where xai,j the zonal gradient, and yai,l the meridional gradient. However, this
metric does not yet appear to have been adopted in any classifications.
Because the results of correlation-based classifications can be substantially influenced
by decisions about the grid and sample sizes, and the threshold adopted for similarity, it
is important to observe several practical points in order to obtain a valid synoptic clas-
sification. Based on studies by Key and Crane (1986) and Yarnal and White (1987), the
0 following recommendations are made:
1 The types should be identified using a sample of at least 1,000 individual grids in
order to obtain stable results. For example, Barry et al. (1981) used sixty months of
11 daily grids for the western United States; for each calendar month, two months were
selected with zonal circulations, two with strong meridional circulations, and one with
conditions similar to the long-term mean for the region, as indicated by mean monthly
fields.
2 The threshold for similarity must be sufficiently stringent to avoid too much internal
diversity within the types. For correlation coefficients, a threshold of >0.7 is gener-
0 ally satisfactory for MSL pressure fields and >0.9 for 700 mb maps (Hartranft et al.,
1970). For the Kirchhofer S score, the use of row and column thresholds of between
1.0 n and 1.4 n where n is the number of grid points in each row (or column) of the
grid array seems to produce satisfactory results in terms of the percentage of the days
classified in the population under study (Yarnal and White, 1987). However, this may
also produce an inconveniently large number of types (~50).
3 The “optimum” number of types used to describe the synoptic characteristics of an
area depends on several considerations. One is the effectiveness of the categories in
describing the area climatology. However, experience suggests that twenty to thirty
types are an unmanageable number for interpretation and study. Rather than reducing
0 the categories by lowering the threshold for similarity, it appears preferable to group
the initial categories by objective clustering or by a subjective grouping appropriate
to the problem under study (Yarnal, 1985).
A Monte Carlo technique for assessing the statistical significance of a Kirchhofer classi-
fication is illustrated by Kaufmann et al. (1999). A three-dimensional surface is generated
from 5,796 observations of daily maximum temperature at sixteen grid points over the
central United States for June–August 1931–93. One hundred experimental data sets are
obtained by random drawings from a normal distribution (with zero mean and appropriate
standard deviation) based on data at one grid point for Julian Day 200. Each 100 sets are
0 analyzed by the Kirchhofer technique, using the same criteria as for the observational
data, and their statistical significance is then determined. For example, with a threshold
for S of 0.50 and a minimum group size of sixty, fifteen groups are distinguished from
the observational data. However, fifteen groups are at position 45 out of 100 values gener-
ated by the experimental data for the same criteria. This means that fifteen or more groups
will be identified 45 percent of the time from data where no meaningful patterns exist.
It is apparent that further research is needed to assess the reliability of the critical values
used in the Kirchhofer technique.
Kaufmann et al. point out that the use of spatial averages and deviations in the
Kirchhofer formula assumes spatial stationarity. They normalize the maximum tempera-
0 ture data as Zij scores with respect to a temporal mean and deviation. This avoids obtaining
a simple south–north temperature gradient as pattern 1. Also, this normalization of loca-
tion value addresses the problem identified by Blair (1998) with row and column scores.
11 Kirchhofer’s approach has been widely adopted, especially in the United States. Daily
1
Figure 7.2 Six key MSL circulation patterns (mb) over western North America, obtained using
the Kirchhofer typing procedure. (Barry et al., 1998)
11
11
0
Figure 7.3 The basic cyclone model and the associated synoptic types. (From Yarnal and Frakes,
1997)
catalogues of sea-level pressure types have been produced for the Canadian Arctic (Bradley
and England, 1979), the Arctic Ocean (Barry et al., 1987) Alaska (Moritz, 1979),
and western North America (Barry et al., 1981); Yarnal (1984a, b, c, 1985) has used the
available 500 mb data to classify upper air patterns in this last area. An illustration of
the circulation patterns over western North America is shown in Figure 7.2. The issue
0 of grid resolution for the Kirchhofer classification applied to western Canada is consid-
ered by Saunders and Byrne (1999). More realistic precipitation patterns are simulated
with a grid spanning western North America than with a regional grid. A recent appli-
11 cation to the British Isles is discussed below. The Kirchhofer scheme has also been applied
1 to validate the representation of synoptic circulations patterns in GCM control runs with
observational data (Crane and Barry, 1988; McKendry et al., 1995).
A procedure combining some of the strengths of the manual and correlation-based
approaches has been proposed by Frakes and Yarnal (1997b). They first use a manual
procedure to develop a classification of daily surface pressure maps for a twelve-year
period. This training set has ten types based on the Bergen cyclone model (see Figure
7.3) applied to daily weather maps. The authors also examine the preceding and following
days in order to select the pattern that best represents each day’s weather. Composite
mean pressure maps for each type are then used as “key” days in a correlation analysis
of gridded pressure data for the region. An optimum threshold is determined by succes-
sively varying the threshold value by 0.01 and checking its effect on the percentage of
days unclassified and on the standard deviation fields of the composites. There appeared
to be no advantage in selecting r 0.60 with 80 percent of days classified over r 0.30
with 99 percent of days classified. Analysis of the manual and the hybrid daily classifi-
cation results for the twelve years showed only fair agreement with a 42 percent
correspondence, although the total frequencies of the ten types agree better overall. A
different procedure to modify a correlation-based classification is illustrated by Brinkmann
(1999a). She notes that within-type variability caused by small-scale circulation features
can be reduced by incorporating subtypes, based on 700 mb geostrophic relative
vorticity, into the classification. Increasing the correlation threshold failed to provide any
improvement.
7.3.2 Classifications based on data reduction methods

The earliest studies in objective specification of isobaric or contour patterns made use of
Fischer–Tschebyschev orthogonal polynomial equations (e.g. Hare, 1958), but by the
1960s the advantages of the more flexible principal component, or eigen vector, analysis
was generally recognized. The principal components are the optimal set of mathemati-
cally determined functions that provide the most efficient representation of variance in
the data set. Each function is mutually uncorrelated (orthogonal) in space and the coef-
ficients of the functions are orthogonal in time. Because each component extracted is
orthogonal with respect to all others the variances are additive. A researcher may there-
fore extract only the number of components that explain a significant portion of the total
variance of the original system (Kalkstein et al., 1987). The result is a considerable reduc-
tion in the volume of the data without loss of valuable information. The first few principal
components of pressure fields usually describe simple zonal and cellular patterns, as can
be seen in Figure 7.4, which illustrates the first three components of a daily 500 mb level
analysis for western Europe (Kruizinga, 1979). An individual pressure or height map is
represented by some combination of these components and a classification based on prin-
cipal components is constructed by defining arbitrary ranges of the amplitude of each
component. It must be emphasized that each component does not necessarily represent
an actual circulation type. They are abstractions from reality by virtue of their orthogo-
nality. It has been suggested by Richman (1981, 1986) that rotating some component
scores with respect to others by using canonical variates can improve results. However,
Joliffe (1987) and Richman (1987) have concluded that there are advantages and disad-
vantages with both rotated and unrotated solutions, and further work is needed to clarify
which is more appropriate for particular applications.
Vargas and Compagnucci (1983), analyzing pressure fields over southern South
America, have drawn a useful distinction between two ways of resolving the input data
matrix. These are: (1) correlating pairs of fixed points over time – the S mode, (2) corre-
lating pairs of time occurrences over fixed points – the T mode. The S-mode (multiple
station over time) correlation matrix is commonly used to develop a regionalization of
climatic variables (White et al., 1991; Sumner et al., 1993; Comrie and Glenn, 1998).
11
11
0
Figure 7.4 The first three components of daily 500 mb height data over western Europe. They
account for 25 percent, 29 percent, and 13 percent of the variance respectively.
11 (Kruizinga, 1979)
1 The T-mode (day-by-day) correlation matrix, followed by a cluster analysis, is particu-
larly suitable for defining circulation patterns (Romero et al., 1999a). S-mode analyses
have been used by Crane and Barry (1988), Richman (1981), Cohen (1983), Bonnell
and Sumner (1992), and Maheras and Kutiel (1999). However, Huth (1993) finds that
S-mode analysis of 500 mb heights failed to detect the main circulation patterns over
Europe, whereas T-mode (with oblique rotation of components) was successful. T-mode
analysis is used by Romero et al. (1999b) to classify circulation types for days with signif-
icant rainfall in the Spanish Mediterranean area. Wilson et al. (1992) compared four
methods of classification – k-means clustering, fuzzy clustering, principal components,
and principal components coupled with k means – and found that in terms of discrimi-
nating the circulation patterns responsible for precipitation events all methods performed
approximately equally well. In a later paper (Huth 1996a) no unique solution is found
to the circulation classification problem, and no method was found to be best in all
aspects. This result is substantiated by the work of Brinkmann (1999b) for eastern North
America. Twenty-two 700 mb circulation types were obtained by using an S-mode PCA
of both covariance and correlation matrices, with rotation of the first five components,
followed by k-means clustering of the PC scores. Four classifications were derived using
the two matrices each with unrotated and rotated PCs. The within-type variability of
temperature for each PC-based classification was found to be comparable to, or in the
case of the rotated covariance matrix worse than, that obtained through a correlation-
based classification (Brinkmann, 1998). Even when warm and cold circulation subtypes
were identified, the best regression results for monthly temperature anomalies estimated
from monthly circulation type frequencies were still inferior to those with the correlation-
based approach.
On the scale of a whole hemisphere, eigen vectors have been used to classify 500 mb
fields in the northern hemisphere by Craddock and Flood (1969) and although the moti-
vation behind this and subsequent work (Craddock and Flintoff, 1970) was primarily in
connection with long-range forecasting, it throws into focus the primary patterns, ten of
which contribute almost 80 percent of the total variance.
In conclusion it should be noted that some disagreement remains as to whether objec-
tive approaches are always qualitatively an improvement on subjective approaches (Ladd
and Driscoll, 1980). Many climatologists believe that the classification of weather patterns
should be as objective as possible but the subjective selection of both data and method
preclude complete objectivity. In their review of the future challenge for climatic studies
Yarnal et al. (1987) go so far as to say that “what is now being observed is a renewed
confidence in the more subjective pressure-pattern typing schemes.” Jones et al. (1993)
find little difference between an objective scheme for allocating daily weather types on
the British Isles and the Lamb classification, which is described next.
7.3.3 Weather types

Multivariate statistical techniques have been applied to the categorization of weather types
(“complex climatology”) since the 1970s, building on the pioneering study of Christensen
and Bryson (1966). They used principal components analysis to identify ensembles of
weather conditions from surface weather observations. Fifteen weather variables recorded
twice daily over five years of observation at Madison, Wisconsin, were reduced by prin-
cipal component analysis to nine new components accounting for about 80 percent of the
original variance. They were then able to show that these weather types were synopti-
cally reasonable, in this spatial organization.
A different illustration is provided by Fechner (1977), who shows how empirical orthog-
onal functions can be used to provide a classification of the vertical profile for weather
situations above a certain geographical point. Data on geopotential height, temperature,
humidity, and wind measured by radiosonde ascents over the period 1948–65 at Ocean
11
11
Figure 7.5 The first EOF of radiosonde data, including the annual range 1948–65 at Ocean Weather
Ship C (52.75°N–35.50°W), representing a cold air mass situation. (From Fechner,
1977)
Weather Ship C provided the input data. The first (most important) eigen vector, which
represents a cold air mass situation, is shown in Figure 7.5.
0 The multivariate technique of stepwise discriminant analysis of eight meteorological
variables has been used to distinguish weather situations in southern California
(McCutchan and Schroeder, 1973; McCutchan, 1978). The primary purpose of the tech-
nique is to set up the best combination of variables to differentiate classes. Cluster analysis
to group cases automatically on the basis of minimum squared error was then employed.
A variety of distance measures and measures of similarity can be used in weather typing,
including string and tree-type representations (Tsui and Hay, 1979; Kalkstein et al., 1987).
The recent approach to identifying synoptic weather complexes follows a series of steps
(El-Kadi and Smithson, 1992; Yarnal, 1993):
0 1 Principal component (eigen vector) analysis is used to reduce the input weather vari-
ables to a set of orthogonal components.
2 Each day is described in terms of these components via a multiple regression analysis
of the data.
3 Using a clustering procedure, an objective grouping of the regression coefficients is
obtained.
4 A threshold criterion is selected to derive the weather types.
An example of the clustering approach is provided by Kidson (1994a) for New Zealand.
He identifies thirteen circulation types by clustering daily sea-level pressure data for
0 1980–90. Standard weather elements (temperatures, precipitation, sunshine, and wind
run) are further examined for eighty-two stations by ranking daily values into quintiles
to remove seasonal effects. (Kidson, 1994b). The mean departures for the different classes
11 are typically only in the second to the fourth quintiles, but they show dynamical
1 consistency. Kidson also performs an EOF analysis of the thirteen types and finds that
the first three rotated modes represent 60 percent of the variance in frequency. These
EOF’s describe the strength of the westerlies, the north–south displacement of high-
pressure centers, and opposing variations in the frequency of northwesterly troughs and
highs to the south. Further cluster analysis, based on the three-component scores for all
months, led to the definition of eight sets of synoptic regime, for which weather anom-
alies are also calculated.
A standardized set of eigenvector-based procedures has been developed, building on
the single station analysis of twenty-eight weather variables at Greater Wilmington,
Delaware, by Kalkstein and Corrigan (1986). Daily sea-level pressure and 500 mb charts
are chosen to represent the synoptic circulation associated with each cluster. This proce-
dure provides a “temporal synoptic index,” or TSI (Kalkstein et al., 1987). The approach
was extended by Davis and Kalkstein (1990), to incorporate weather data for a network
of US stations. However, the associated synoptic types had to be grouped subjectively.
Moreover a fundamental and unresolved problem concerns the subjective selection of
weather variables (Stone, 1989). Conceptually these should represent the radiatively,
thermodynamically, and advectively determined components of local weather in a parsi-
monious manner. A methodology leading to a reduction of the minimum sample size that
should be considered in order to obtain a reliable synoptic weather type classification has
been described by Lana and Fernandez Mills (1994). Most of the research studies following
the “Kalkstein TSI approach” have been directed towards local environmental problems
connected with air quality (Davis and Kalkstein, 1990), acid rain (Ezcurra et al., 1988),
mortality statistics (Kalkstein, 1991), urban energy budgets (Todhunter, 1989), and climate
change analysis (Kalkstein et al., 1990). Up to now, no major catalogs of circulation
patterns developed through these procedures and spanning extended time intervals have
been published. However, for the western United States, Davis and Walker (1992) devel-
oped a Spatial Synoptic Index to classify daily types of synoptic situation for 1979–88.
Based on radiosonde data from 800 mb to 250 mb from twenty-one stations, thermal,
moisture, and flow parameters were used in a principal components analysis and clustering
procedure to identify thirteen types.
A new procedure to study weather types on a continental scale is proposed by Kalkstein
et al. (1996). The first step is to designate air mass types based on specified ranges of
weather elements (afternoon air temperature, dew point, dew point depression, wind speed
and direction, cloud cover, and diurnal temperature range) at a number of locations. A
seed day group is chosen for each of six air mass types (polar, temperate, and tropical,
each with dry and moist variants). Linear discriminant analysis is then used to obtain a
daily categorization of air masses, and the process is repeated, with seed days for tran-
sitional synoptic events to identify transitional days. Maps of air mass frequency and their
principal characteristics are prepared for the eastern United States for 1961–90. The goal
of this “spatial synoptic classification” is to facilitate continental-scale climatic impact
analysis.
7.3.4 Artificial neural networks and self-organizing maps

Weather systems display considerable non-linearity and an interconnectedness across
scales. These characteristics are not well captured in correlation-based synoptic climato-
logical methods, including PCA. Another limitation of these methods is their inability to
learn in an iterative fashion, by converging on the “best” solution (transfer function) to
the relationship between atmospheric predictor variables (e.g. 500 mb geopotential height,
850 mb moisture) and the local to regional-scale climate (dependent variables) at Earth’s
surface (e.g. temperature, precipitation). These limitations are largely overcome with the
use of artificial neural networks (ANNs) and self-organizing maps (SOMs). Artificial
neural networks attempt to replicate the primary processes involved in human learning
11 (e.g. pattern recognition, “trial and error”). The use of ANNs in synoptic climatology,
and also satellite image classification, exploits the increases in desktop and network
computer power over the 1990s by enabling a large number of simulations to be under-
taken efficiently (Key et al., 1989; Pankiewicz, 1995; McGinnis, 1997). The ANN “learns”
the predictor–target relationship by successive iterations to produce a solution containing
the smallest error (Hewitson and Crane, 1992a). Artificial neural networks are also proving
to be a valuable tool in empirical “downscaling” (translation across scales) of the coarse-
grid resolution output from GCMs to small (local) scales, and on time scales ranging
from daily to centennial (Hewitson and Crane, 1992b, 1994; Crane and Hewitson, 1998;
0 Frakes, 1998). Thus they are a valid alternative to more computer-intensive dynamical
methods of downscaling which use regional GCMs or the “nesting” of grid models
(Hewitson and Crane, 1996).
An ANN consists of layers of interconnected nodes that each contain an activation
11 function (Cavazos, 1997). In the feed-forward, or back-propagating, type of ANN, the
“best” relationship describing the target variable (e.g. precipitation) and the input layer
(e.g. 850 mb humidity) of the training data set, is achieved by successive iterations through
one or more hidden layers. The output error is initially large, but enables adjustments to
be made to the input layer through a series of weights (back-propagation). This leads
to a reduction in the output–target error over successive iterations. A point is eventually
0 reached where no further improvement in the output of the net occurs when the test data
are used as input. Using an ANN to predict local-scale precipitation, for example, typically
results in a high degree of predictability of the phase (timing) of events but a tendency
to underestimate (overestimate) extremely high (low) events (Cavazos, 1999). These may
be particularly expected during extremes in teleconnection modes, such as those associ-
ated with El Niño and La Niña. The use of ANN in ENSO prediction is discussed by
Hsieh and Tang (1998).
Self-organizing maps are another form of ANN that can be applied to synoptic clima-
tological research, although such work is only just beginning. Self-organizing maps
classify complex matrices having data elements that are related to each other non-linearly
0 (Cavazos, 1998, chapter 3). They extract map patterns without being “trained” (i.e. they
are unsupervised classifiers), relying upon the emergence of clustered structures from the
bottom up (self-organization) (Kohenen, 1995). By reducing multidimensional processes
to a two-dimensional problem, SOMs also permit the study of persistence of weather map
patterns, as well as their day-to-day evolution (Cavazos, 1998, chapter 3, 1999). Self-
organizing maps may provide an alternative to PCA (Cavazos, 1999). The map patterns
so derived are meaningful physically, and readily permit the identification of circulation
features such as “split flow” and blocking regimes.
0 7.4 Principal catalogs and their uses

The principal catalogues that have been developed for different areas of the globe are
listed in Table 7.2.
7.4.1 The Lamb classification

Lamb (1950, 1972), building in part on earlier work by Levick (1949), classified the
airflow over the British Isles and the immediate surroundings for each day from 1881 to
the present on the basis of eight directional types (Figure 7.6), each of which is sub-
divided into anticyclonic, cyclonic, and unspecified categories (Table 7.3). The Lamb
0 weather type catalog constitutes the longest daily history of airflow patterns for any part
of the world and provides an unrivalled perspective on the changing behavior of the
atmospheric circulation around the British Isles (Perry and Mayes, 1998). A complete
11 listing of the daily classification from 1972 to 1995 can be found in Hulme and Barrow
1
Figure 7.6 The basic Lamb circulation types for the British Isles. (From O’Hare and Sweeney, 1993)
11 Table 7.2 Principal synoptic catalogs
Area Catalog sources Time period

covered
Northern hemisphere
Vangengeim (1960)
Girs (1974) 1891–present
Dzerdzeevskii (1968)
Savina (1987) 1899–1985
0
Europe
Central Hess and Brezowsky (1977)
Gerstengarbe and Werner (1993) 1881–1992
Eastern Alps Lauscher (1985) 1946–83
11 Hungary Peczely (1961) 1901–57
Switzerland Schüepp (1968) 1955–present
Switzerland Perret (1987) 1955–85
Italy Urbani (1961) 1945–60
Spain Goodess and Palutikof (1998) 1956–89
British Isles Lamb (1972)
0 Hulme and Barrow (1997) 1861–present
Regions of Britain Mayes (1991) 1950–present
Asia
East Yoshino (1968)
Yoshino and Kai (1974) 1941–80
North America
Contiguous US Barchet and Davis (1984) 1969–78
Eastern US Comrie and Yarnal (1992) 1978–87
Western US Barry et al. (1982) 1899–1980
Western US Davis and Walker (1992) 1979–88
0 Alaska Putnins (1966)
Moritz (1979) 1946–84
Canadian Arctic Bradley and England (1979) 1946–74
Gulf Coast Muller and Wax (1977) 1971–74
Australasia
New Zealand –
South Island Sturman et al. (1984) 1961–80
0 (1997), and updates are posted on the Climatic Research Unit website (http://www.
cru.ac.uk). In addition three non-directional types are recognized, anticyclonic where high
pressure dominates, cyclonic where a depression stagnates over or crosses the British
Isles, and an unclassified type where the pattern is weak or chaotic. Figure 7.6 illustrates
the basic types identified by Lamb. Among the advantages of the classification is the rela-
tive ease with which individual days can be classified, and this allows updating to be
carried out very readily. Less satisfactory is the fact that the size of area being classified
often exceeds the size of the prevailing circulation features (O’Hare and Sweeney, 1993).
Thus a day that is classified as cyclonic may have an easterly flow in the north, over
Scotland, and a westerly flow over southern England. Mayes (1991) has addressed this
0 problem with a simple airflow classification for four regions of the British Isles. In fact
the original airflow analysis scheme of Levick (1949, 1975) distinguished different types
in five regions of the British Isles to take account of such regionally different patterns.
11 Mayes and Wheeler (1997) illustrate the variable conditions over the British Isles during
1 Table 7.3 Mean seasonal and annual frequencies (days) of the Lamb “weather types” for the
British Isles
Type Winter Spring Summer Autumn Annual

Anticyclone 14.3 17.4 15.9 17.0 65.5
Directional AC (except AW) 8.1 11.1 9.8 8.8 37.7
AW 4.3 3.4 5.2 4.3 17.2
W 20.9 12.6 16.7 18.2 68.5
CW 4.0 2.8 4.3 4.9 15.0
NE, E, SE 5.5 8.8 3.3 4.8 22.4
S, SW 8.3 5.7 4.2 6.8 25.0
NW, N 6.7 8.4 8.7 7.4 31.1
C 9.5 11.6 14.7 10.8 46.5
Directional Cyclone (except CW) 5.3 6.5 4.8 5.6 22.4
Unclassified 3.3 3.8 3.4 3.5 14.1
90 92 92 91 365
Source: from Briffa et al. (1990).
individual months with a predominant airflow pattern regime. Notwithstanding these diffi-
culties, the Lamb classification has been extensively used to produce regional synoptic
temperature and precipitation climatologies both for large areas, e.g. England and Wales
(Lawrence, 1971; Sweeney and O’Hare, 1992), Ireland (Sweeney, 1985) and for more
local studies, e.g. southern England (Barry, 1963; Stone, 1983) and south Wales (Faulkner
and Perry, 1974). Beaumont and Hawksworth (1997) calculate the areal mean daily precip-
itation associated with the Lamb types over Wales for 1982–91. The Cyclonic type, which
is wettest, with 6.4 mm per day, occurs on forty-six days per year in the long term; the
Anticyclonic type, which is driest (0.46 mm per day) occurs on sixty-five days annually.
Precipitation trends during 1861–1995 are largely determined by these types; the Westerly
type itself is not well correlated with precipitation in Wales, although strong zonal circu-
lation and precipitation may occur with the Cyclonic pattern or hybrid types such as CW,
CSW, CS, and CSE. Heavy rainfall events also occur predominantly with the Cyclonic
type in many parts of the British Isles, excluding southwestern England and Ireland, where
the Southerly type brings most such events (Perry and Mayes, 1998). The usefulness of
the type categories as a “predictor” of local and regional annual precipitation amounts
can be substantially improved, however, if the frequency of warm, cold, and occluded
fronts is incorporated within the type categories (Wilby et al., 1995).
With more than a century of daily data available, the Lamb classification provides a
useful tool for investigating the temporal context of seasonal weather (Figure 7.7). Similar
regional airflow classifications based on the general trajectory of the flow, subjectively
assessed, have been developed for areas as geographically diverse as Labrador-Ungava
(Barry, 1959, 1960), Poland (Litynski, 1970), southern South America (Sturman, 1979),
and the South Island of New Zealand (Sturman et al., 1984).
Various indices have been formulated in order to condense the vast amount of data in
the Lamb catalog. Murray and Lewis (1966) devised a set of four indices to summarize
the circulation over Britain for a given period of time, and these are known as P
(Progression) – a measure of the frequency of westerly circulation types, S (Southerliness),
C (Cyclonicity), and M (Meridionality). These indices have been related to monthly mean
temperatures and summer rainfall totals in Britain (Murray, 1972; Hughes, 1980). They
are also applied by Mayes (1994) to identify trends in airflow.
The information content of the Lamb (1972) catalog of 27 daily “weather types” can
be represented adequately by eight principal components, based on an analysis of annual
type frequencies for 1861–1980 by Jones and Kelly (1982). Four combinations of the
11
11
0
Figure 7.7 Annual totals and ten-year moving averages of the frequency (days) of the main Lamb
types, 1861–1991. (From Lamb, 1994)
major types are identified that account for about 70 percent of the total variance. Thus
PC1 contrasts Westerly airflow and blocking Anticyclonic conditions; PC2 represents
Anticyclonic and/or Cyclonic types versus Northwesterly and/or Northerly airflow; and
PC4 contrasts Northerly and/or Northwesterly airflow with Anticyclonic Westerly,
Southerly and/or Southwesterly flows. Using the same approach, Briffa et al. (1990) show
0 that the principal components can be used to fashion four generic indices of climatic vari-
ation over the British Isles. Jones et al. (1993) take a different approach by relating the
Lamb types to an objective classification of sea-level airflow and vorticity developed by
11 Jenkinson and Collison (1977). The latter use grid-point pressure values over an array
1 from 45°N to 65°N, 10°E to 20°W to define westerly flow (W), southerly flow (S), resul-
tant flow F (S2 W2)1/2, westerly shear vorticity (ZW), southerly shear vorticity (ZS),
and total shear vorticity (Z ZW ZS). Rules are devised to define flow directions from
45° sectors (e.g. W is between 247.5° and 292.5°); if | Z| < F, the flow is straight; if | Z |
> 2F the flow is strongly cyclonic (Z > 0) or anticyclonic (Z < 0); if F < |Z | < 2F a
hybrid direction/curvature, type is indicated; and if F < 6 and |Z | < 6 the light, indeter-
minate flow is unclassified. The method was tested for daily maps during December
1880–December 1989 against the Lamb catalog for seasonal and annual totals of Lamb’s
seven basic types. Seasonal correlations are about 0.9 for anticyclonic and cyclonic types
and 0.70–0.85 for directional types. Jones et al. attribute the differences primarily to
Lamb’s attention to large-scale steering of weather systems rather than instantaneous
surface winds. The good overall agreement suggests that an objective scheme is suitable
for many applications, particularly the use of GCM outputs (Briffa, 1995).
A number of studies suggest that low-level airflow may be a better predictor of precip-
itation than the Lamb types. Sumner (1996) reaches this conclusion from a cluster analysis
of standardized patterns for more than 1,000 daily cases of significant rainfall events in
Wales. Mayes (1996) uses the W, S, and C indices determined from the Lamb types to
examine trends in monthly precipitation over the British Isles. Similarly, Wilby (1997)
argues that airflow indices of shear vorticity, cyclonicity, angular flow direction, and flow
strength are preferable to discrete types in GCM downscaling because the indices are
continuous. He also points out that weather type–precipitation relationships used in down-
scaling cannot be assumed to be stationary. The options are to ignore any non-stationarity
or to incorporate it empirically or stochastically.
The Kirchhofer method has recently been used to generate an MSL pressure pattern
catalog for the British Isles, comparable with Lamb’s scheme (Figure 7.8). El-Kadi and
Smithson (1996) identify fifteen types for 1977–86. Seven types (westerly, northerly,
southerly, easterly northwesterly, Netherlands high and British high) account for 78.5
percent of cases, and only 0.5 percent of days remained unclassified. Hybrid types are
eliminated, and the smaller number of categories simplified applications of the catalog.
Another virtue of the automated approach is the differentiation of three anticyclonic and
three cyclonic types, according to the location of the system centers.
7.4.2 The Grosswetter classification

The concept of Grosswetter (literally, large-scale weather) has been developed in Germany
and applied in a daily classification by Hess and Brezowsky (1977). Baur (1947, 1951),
the originator of the concept, defined a Grosswetter as the mean pressure distribution (at
sea level) over a time interval during which the position and the tracks of the major
depressions and anticyclones remain essentially unchanged. Later, account was taken of
the 500 mb mid-tropospheric circulation pattern also (Baur, 1963). The thirty Grosswetter-
lagen are described in Table 7.4 and the relative frequencies of each during the 1881–1970
period are shown.
Median durations, frequencies of occurrence, and the probabilities of transition from one
type to another have been calculated by van Dijk and Jonker (1985). The daily classifica-
tion extends back to 1881 and has been updated in the monthly publication Die Grosswetter-
lagen Mitteleuropas, published since 1947 by the Deutscher Wetterdienst (1994). Type
averages and frequency values have been determined for various climatic parameters for
regions and individual German cities (Bürger, 1958), as well as for other central European
countries, e.g. Switzerland (Grutter, 1966), Hungary (Peczely, 1961). The frequencies of
the main Grosswetter in relation to phases of the El Niño/Southern Oscillation have been
studied by Fraedrich (1990). Schiesser et al. (1997) analyze the occurrence of severe winter
storms in lowland Switzerland during 1864/65–1993/94 in relation to the Grosswetter; half
the cases had cyclones located over the British Isles–North Sea.
11
11
Figure 7.8 Objectively derived pressure patterns for the British Isles. (From El Kadi and Smithson,
11 1996)
1 Table 7.4 Description of the GWL and relative frequencies of the total number of Witterungen
(Nt) that occurred during the periods 1881–1970 and 1949–1970 (%)
No. GWL Description 1881–1970 1949–70

1 Ws Zonal circulation type, displaced southward 2.55 2.34
2 Wa Zonal circulation type, displaced northward 5.62 5.10
3 Wz Zonal circulation type (most frequently occurring
GWL) 11.58 10.11
4 BM Ridge of high pressure over Central Europe 5.90 7.12
5 HM Anticyclone over Central Europe 9.89 8.50
6 SWa Southwesterly flow, anticyclonic conditions 2.26 3.03
7 SWz Southwesterly flow, cyclonic conditions 1.79 4.14
8 NWa Northwesterly flow, anticyclonic conditions 4.87 1.79
9 NWz Northwesterly flow, cyclonic conditions 4.45 5.74
10 HNa Anticyclonic over Norwegian Sea, anticyclonic
conditions 3.69 2.21
11 HNz Anticyclonic over Norwegian Sea, cyclonic conditions 1.60 2.02
12 HB Anticyclone over the British Isles 3.17 3.13
13 Na Northerly flow, anticyclonic conditions 1.41 0.97
14 Nz Northerly flow, cyclonic conditions 3.19 2.71
15 TrM Trough over Central Europe 4.52 5.79
16 TM Cyclone over Central Europe 2.79 2.44
17 TB Cyclone over the British Isles 2.17 2.53
18 TrW Trough over Western Europe 3.30 5.28
19 Sa Southerly flow, anticyclonic conditions 2.06 1.75
20 Sz Southerly flow, cyclonic conditions 0.94 1.10
21 SEa Southeasterly flow, anticyclonic conditions 2.17 1.98
22 SEz Southeasterly flow, cyclonic conditions 1.48 1.06
23 HFa Anticyclonic over Scandinavia and/or Finland,
anticyclonic conditions 3.59 2.67
24 HFz Anticyclonic over Scandinavia and/or Finland, cyclonic
conditions 1.06 1.61
25 HNFa Anticyclonic over the Norwegian Sea and Scandinavia,
anticyclonic conditions 1.16 1.15
26 HNFz Anticyclonic over the Norwegian Sea and Scandinavia,
cyclonic conditions 1.54 2.16
27 NEa Northeasterly flow, anticyclonic conditions 3.04 2.44
28 Ww Zonal circulation type, in Eastern Europe a southerly
flow, blocking anticyclone over Russia 2.86 3.54
29 U Not defined, transitional situation 2.21 2.80
30 NEz Northeasterly flow, cyclonic conditions 3.12 2.80
Nt 8,554 2,176
Source: from van Dijk and Jonker (1985).
Although the classification takes account of the circulation over a wide area of central
and western Europe (30°W–45°E, 24°–70°N), it has been supplemented by modified
schemes in some areas, particularly the Alps (Wanner et al., 1997, 1998), partly because
it takes little account of Mediterranean influences, which become increasingly important
towards the southern fringes of central Europe. Lauscher (1985), for example, has given
details of a daily classification for the period 1946–83 for the eastern Alps. The subtypes
had a mean duration of four to eight days, with an overall mean of 6.4 days. Kahlig
(1989) suggests that the categorization of circulation types can be assisted by the use of
artificial intelligence techniques. He notes that an expert system has been applied to the
classification of Grosswetterlagen for the eastern Alps.
11
11
0
Figure 7.9 Position of troughs and ridges in the mid-troposphere for the W, E and C types of
hemispheric circulation pattern proposed by Vangengeim and Girs. (a) Winter. (b)
11 Summer. (From Kozuchowski and Marciniak, 1988)
1 7.4.3 Hemispheric classifications
Studies of hemispheric circulation patterns began in the Hydrometeorological Service of
the Soviet Union in the 1930s and 1940s. Two “schools” developed under the leader-
ship of B.L. Dzerdzeevski (1945) in the Institute of Geography, Moscow, and G. Y.
Vangengeim (1935, 1946) in the Arctic and Antarctic Research Institute, St Petersburg.
Interestingly, neither group references the other and there does not appear to have been
detailed intercomparison of the two approaches.
Elementary synoptic processes (Vangengeim–Girs)

Vangengeim (1935, 1946) distinguishes three basic types of circulation in the zone
35°–80°N – westerly (W), easterly (E), and meridional (C). Each is characterized by a
particular distribution of depressions and anticyclones at the surface and by an organiza-
tion of the major long-wave pattern. The ridge and trough positions differ seasonally
(Figure 7.9). W type refers to patterns with essentially zonal movement of small-ampli-
tude waves; nine subtypes are distinguished on the basis of the latitude of the subtropical
anticyclone cells. With C type (seven subtypes) there are large-amplitude, stationary
waves. The subpolar lows are shallow, there is a well developed high, and the subtrop-
ical anticyclone cells are split and displaced northward. Pressure over Europe and western
North America is low. E type (ten subtypes) is comparable with C, but the troughs are
in different locations. The subpolar lows are well developed, the Siberian high is weaker
and farther west than with C, the Azores and Pacific anticyclone cells are also west of
their normal position, and there are stationary highs over Europe and western North
America. Subsequently Girs (1948, 1960, 1981) showed that in the North America–Pacific
Ocean sector two meridional patterns (M1 and M2) and one zonal type (Z) are the most
significant, and he indicated that these may combine with W, E, and C to give nine basic
types. Table 7.5 indicates these and the annual frequencies for 1900–57. An algorithm
based on a Euclidean distance measure was subsequently developed, and used, for twelve
of the Vangengeim circulation types (Reitenbakh and Kozulin, 1982) to obtain a catalog
for 1880–1973.
The Vangengeim classification has been widely used by Russian scientists in assessing
seasonal departures from normal of temperature and precipitation for the former Soviet
Union, or regions of the territory. Vorobieva (1967), for example, provides maps of
seasonal departures of precipitation for the W, E, and C types. Correlations between the
three major type categories and temperature and precipitation patterns over Europe are
presented by Kozuchowski and Marciniak (1988).
Elementary circulation mechanisms (Dzerdzeevski)

Dzerdzeevski (1945, 1968) proposed that Elementary Circulation Mechanisms (ECMs)
operate over a short time interval but govern the circulation pattern over an entire hemi-
Table 7.5 Frequency of Vangengeim’s types, 1900–57 (%)
Annual frequency
Type Wz WM1 WM2 Ez EM1 EM2 Cz CM1 CM2

Subtypes 9.9 7.5 9.1 14.3 11.1 9.0 10.1 8.9 10.1
Main types ← 26.5 → ← 44.4 → ← 29.1 →
Wz Zonal in the North Pacific sector.
M1 Surface anticyclone near the Aleutian Islands, lows to the north.
M2 Ridge from the Pacific high extending to western North America.
11 sphere. He recognized that the identification of circulation types for a limited sector is
hindered by the different synoptic histories of the airflows entering the region. The
“influence field” of the atmosphere for a twenty-four-hour forecast period at a location
in mid-latitudes extends over about a quarter of the earth’s circumference and from the
tropics to the polar circle. Discrete categories of atmospheric process can be distinguished
because:
1 The hemispheric circulation is determined by a finite number of characteristic circu-

lation mechanisms. The number of those mechanisms is small over short time periods
0 when the incoming solar radiation and the properties of the earth’s surface are
constant, but their characteristics differ greatly with season.
2 The features of each circulation mechanism (including its spatial organization) persist
longer than the time scale of synoptic processes. Thus the hemispheric circulation is
11 a real macroprocess, not a chance combination of independent synoptic processes.
Individual disturbances and fronts are regarded as “noise”.
The approach takes account of the degree of organization of the hemispheric flow
(Dzerdzeevski et al., 1946). Cyclone and anticyclone tracks at the 700 mb or 500 mb
level are used as an indicator of the main mid-tropospheric steering currents (Dzerdzeevski
0 and Monin, 1954). It is emphasized that charts averaged over several days provide the
best view of the various types. Special attention is given to polar intrusions and associ-
ated blocking in the westerlies in six hemispheric sectors of 50°–60°. The four major
patterns are shown in Figure 7.10. They are:
1 A zonal ring of cyclone tracks in high latitudes; two or three breakthroughs of mid-
latitude cyclones (two types, five subtypes).
2 Interruption of zonality with a single polar intrusion; one to three breakthroughs of
mid-latitude cyclones (five types according to the sector of the intrusion, thirteen
subtypes).
0 3 Northerly meridional motion with two, three, or four polar intrusions (five types
according to their location, twenty-one subtypes).
4 Southerly meridional; no polar intrusions and poleward movement into the Arctic in
two to four locations (one type, two subtypes).
The patterns for the forty-one subtypes are illustrated in Dzerdzeevski (1970), and their
seasonal tendencies are summarized by Savina (1987), together with a catalog of their
daily occurrence, 1899–1985. She also uses them to characterize six “natural seasons”
and tabulates the durations for each year.
A similar approach is evident in Russian studies of the southern hemisphere (Davidova,
0 1967). The term “synoptic process” is used to refer to the movement of pressure systems
over a two to three-day period. The classification refers to the three southern oceans from
20°S to the coast of Antarctica. The types, between five and seven for each ocean, refer
primarily to the zonal or meridional character of the circulation at MSL, based on six
years of data. Davidova shows that meridional patterns of circulation are dominant in all
three oceans; in the South Pacific meridional forms have an 81 percent annual frequency,
with 94.8 percent in the winter months. Figure 7.11 shows the most common patterns in
the south Indian Ocean.
The earliest hemispheric-scale categorizations of circulation regime in Western literature
were based on the concept of high and low index described in section 4.3. A classification
0 for the northern hemisphere 500 mb circulation was developed by Wada and Kitahara
(1971), based on 500 mb zonal index anomalies in four quadrants (0°–80°E, 90°–170°E,
100°–180°W, and 10°–90°W). Based on the degree of zonality and meridionality, ten basic
11 types and six subtypes were distinguished for five-day mean and monthly circulations for
1
Figure 7.10 The four basic types of elementary circulation mechanism of Dzerdzeevski (1962).
Solid arrows: cyclone tracks; open arrows: anticyclone tracks.
1946–70. This approach was extended by Kletter (1962), based on the identification of pro-
gressive, stationary, and retrogressive planetary wave motion, using the Rossby wave for-
mula (section 4.3.2) applied to time–latitude profiles of the 850 mb zonal wind component
in the northern hemisphere. For 1955–58 the type frequencies shown in Table 7.6 were
observed. The subtypes had a mean duration of four to eight days, with an overall mean of
6.4 days. The typical sequence of these types shows that a zonal pattern generally evolves
progressively towards the omega or cellular pattern through amplification of the waves.
However, the cellular and blocked patterns may revert directly to zonal flow (see Figure
4.31). The approach is inadequate for regional applications, however, because the geo-
graphical aspects of the patterns are insufficiently specified.
A hemispheric classification for monthly sea-level pressure patterns has been devel-
oped by Bartzokas and Metaxas (1996). Monthly pressure grids for January, February,
July, and August 1890–1989 and 1,000–500 mb thickness values for 1945–89 were used
poleward of 20°N. Patterns were obtained by T-mode eigen vector analysis with varimax
11
11
Figure 7.11 The most frequent large-scale circulation patterns over the south Indian Ocean. (a) Type
1. Zonal, with mid-latitude traveling depressions, 42 percent frequency. Above (1) MSL
isobars, (2) cyclone tracks. Below (1) Streamlines, (2) isotachs (m s1). (b) Type 2.
Meridional, with break-up of the subtropical high southeast of Madagascar, 26 percent
0 frequency. (Davidova, 1967)
Table 7.6 Hemispheric circulation types
Circulation type Frequency

(%)
1 Zonal
a Zonal flow 40°–65°N; main low enters north of 60°N 2
0 b Zonal flow (subdivided according to latitude of maximum westerlies
and degree of uniformity over different sectors) 27
c North or southward trend of the wind maximum over the North
American or European–North Atlantic sectors 20
2 Planetary waves (at least three waves) 23
3 Cellular
a Blocking anticyclone with split westerlies 11
b Omega block linked to subtropical high pressure 9
c Weak cellular circulations 8
0 Source: from Kletter (1962).
11
1 rotation. The February and August analyses are used to check the likely validity of the
patterns obtained in January and July, respectively. Pattern 1 in January (pattern 2 in
February) account for 30.6 percent (27.4 percent) of the variance and show a strong exten-
sive negative pressure anomaly over the north Siberian coast, with a positive anomaly
southwest of the British Isles, giving a zonal circulation over the northeastern Atlantic.
Pattern 2 in January (28 percent) and pattern 1 in February (21.3 percent) feature a strong
positive anomaly in the Denmark Strait and negative anomalies over the eastern subtrop-
ical Atlantic and the central North Pacific (and for February also over north-central
Siberia). In July (August) pattern 1 (pattern 2) accounts for 45.3 percent (41.6 percent)
of the variance. The closely similar patterns show a strong negative anomaly over the
Arctic Ocean and a weak positive anomaly over Europe. The July pattern 2 (August
pattern 1), with 43.5 percent (42 percent) explained variance, is the opposite of the previous
patterns, with a strong positive anomaly over the Arctic Ocean. Since about 1900 the
positive (negative) pressure anomaly over the Arctic Ocean in summer has been decreasing
(increasing) in frequency.
7.5 Regional applications

The original purpose of most synoptic climatological studies was extended, or long-range,
weather forecasting. This is apparent in the description of the World War II developments
by Jacobs (1947); airflow patterns and related weather conditions over Japan were analyzed
as a basis for prediction. Modern developments along these lines are discussed below.
More generally, it was assumed that airflow patterns would discriminate between the
weather conditions in a particular region. Hence the use of the term “weather type” as
applied by Lamb (1972), among others. Many regional studies sought to describe the
typical conditions experienced during each airflow, or circulation type, according to season.
This has been done for the Alps (Lauscher, 1985; Schüepp, 1968; Kerschner, 1989;
Kirchhofer, 1976), Germany (Flohn and Huttary, 1950; Bürger, 1958), the British Isles
(Barry, 1963; Sowden and Parker, 1981; Storey, 1982; Stone, 1983), Alaska (Moritz,
1979), the Canadian High Arctic (Bradley and England, 1979), New Zealand (Kidson,
1994b) and many other individual regions.
Few synoptic classifications, apart from the Hess–Brezowsky Grosswetter, have
addressed the problem of linking surface and mid-tropospheric circulation characteristics,
although several schemes consider surface and upper-air patterns independently (Mosino,
1964, for Mexico; Schüepp, 1957, for the Alps). In a study of precipitation in Italy,
Gazzola (1969) identified twenty-two patterns separately at the surface and at the 500 mb
level. Contingency tables of their joint occurrence on a seasonal basis showed that, out
of 484 possible combinations, 44 percent never occurred and a further 19 percent occurred
three times or less over an eleven-year period. Sixty-nine percent of days were acccounted
for by 13.4 percent (or sixty-five) of the possible types. Thus the concern that a multi-
plicity of types may result when the vertical structure of the circulation is considered may
not represent an insuperable problem. Synoptic systems and upper-level waves possess
considerable vertical coherence, as discussed in Chapter 6.
The use of a unified system in categorizing surface and upper air fields is illustrated
by the second classification developed for Switzerland by Schüepp (1959). He identifies
thirty-three weather situations that may last several days (Witterungslagen), although the
number is increased to forty patterns in a subsequent analysis of related station weather
conditions (Schüepp, 1979). The scheme is based on principles set out by Lauscher (1958,
1972) for low-level airflow and upper-level circulation patterns over the eastern Alps but
modified to take account of airflow curvature and vertical motion. The six basic classes
and the convective, advective, and mixed types are summarized in Table 7.7. Convective
situations are 46 percent of the total, advective situations 46 percent, and mixed patterns
8 percent of days during 1951–70. The types for the four cardinal directions and low
11 Table 7.7 Witterungslagen scheme for the Alps
Partial collective 500 mb pattern Subtypes

I Convective patterns 1 High pressure Each has five subtypes: weak
2 Weak (average) pressure surface winds with W, N, E, S,
3 Low pressure or weak winds at upper levels
II 4 Westerly flow Each has five subtypes: upper

5 Northerly flow jet, upper/lower flows almost
0 6 Easterly flow parallel with above or below-
7 Southerly flow average pressure aloft, wind
turns height with above or
below average pressure aloft
11 III Mixed patterns 8Z Wave (strong surface winds) B and C have two subtypes:
8B Upper jetstream pressure aloft and above or
8C Surface flow (weak aloft) below average
Source: Schüepp (1979).
Figure 7.12 Climate of Zürich in

winter and summer for the Schüepp
(1979) Witterungs-lagen classifica-
tion. (Above) mean 13.00 tempera-
ture (circle) and ± range (vertical
0 lines); (center) sunshine duration
(columns, hr), + (dots), and days
< 2 hr d–1 shown by open circles
plotted downward (left scale, per-
cent); (below) mean daily precipita-
tion plotted downward (solid
columns, mm), + (vertical lines)
and percent days 1 mm (solid cir-
cles), for each of the forty types.
The horizontal lines show annual
mean values of the three elements.
0 H anticyclonic, F average pressure,
L low pressure; W, N, E, and S are
the advective types; X mixed types.
11 (From Schüepp, 1979)
1 pressure are illustrated for a representative day in the climate atlas of Switzerland
(Kirchhofer, 1995) by a satellite image, fields of 500 mb height and sea-level pressure,
and a map of significant weather over Switzerland; a synoptic climatological summary at
twelve Swiss stations for each of the forty types is provided by Schüepp (1979). Figure
7.12 shows the winter and summer conditions associated with the types at Zürich. A
further classification for Switzerland by Perret (1987) adopts elements of the Grosswetter-
lagen over Europe and the northeastern Atlantic. Perret recognizes directional patterns of
the upper circulation and associated steering of synoptic systems, upper anticyclonic
patterns, and upper cyclonic or trough patterns, giving nine basic types with subtypes
(thirty-one patterns in all). The Swiss Meteorological Institute (1985) updates thirty-four
synoptic parameters on a daily basis, including the Schüepp Wetterlagen type, three para-
meters from Perret (1987), and two from Fliri and Schüepp (1984), and maintains a
database of this information.
For the Alpine region spanning Switzerland, the Tirol and northern Italy, a compre-
hensive precipitation synoptic climatology based on the Schüepp classification has
been developed by Fliri and Schüepp (1984). For each type in each season a map of
mean sea-level pressure and 500 mb contours is provided, together with maps of the
mean daily precipitation and the probability of daily amounts of at least 1 mm. There
are also a description of the weather conditions over the region and tables for selected
stations of temperature averages, Föhn and thunderstorm frequency with each type. Figure
7.13 illustrates the conditions associated with Northerly and Southerly types in winter
1946–79.
Stochastic models of daily precipitation have recently been developed using the
conditional probability of specified precipitation characteristics for selected atmospheric
circulation patterns. In a study of daily precipitation in Nebraska, for example, Bogardi
et al. (1993) use a first-order autoregressive model of precipitation days with nine
types of 500 mb circulation pattern, obtained by PCA and k-means clustering for the
area 25°–60°N, 80°–125°W. For three stations in Washington state Wilson et al. (1992)
examine wet/dry days and precipitation amount on rain days in relation to four circula-
tion classes, defined by tropospheric height patterns and wind components over western
North America, 35°–65°N, 100°–145°W. Precipitation conditions at the three stations
are linked through a hierarchical precipitation event model. In these approaches, geo-
graphical factors that may influence precipitation are not considered. In contrast, Kilsby
et al. (1998) use regression models to estimate points precipitation statistics (mean,
proportion of dry days) at any station in England and Wales, taking account of
MSL airflow indices and factors of geographical location (distance from sea, latitude, and
longitude).
Recently the value of synoptic catalogues for validating the simulations of modern
climate via general circulation models (GCMs) has been recognized (Hay et al., 1992;
McHendry et al., 1995). Most GCM validation studies use mean fields and their standard
deviation, although diagnostic parameters, such as eddy kinetic energy as a measure of
synoptic eddies, have also been examined. Schubert (1994) has noted that empirical
(Wigley et al., 1990), semi-empirical (Giorgi and Mearns, 1991), and dynamic method-
ologies have been developed as downscaling techniques. Hulme et al. (1993) employ an
objectively derived version of Lamb’s classification for the British Isles to analyze the
annual course of the principal airflow pattern types in versions of the UK Meteorological
Office and Max Planck Institute GCMs, in comparison with the observed values. Jones
et al. (1993) show how objectively defined weather types derived from surface pressure
maps can achieve strong correlation with the original Lamb types. This means that objec-
tively defined Lamb weather types could be derived from model-derived pressure patterns,
either to express future scenarios in a meaningful form or to validate control runs of
general circulation models against the actual Lamb types. For Canada, McHendry et al.
(1995) analyze the Canadian Climate Model, using types derived by the Kirchhofer
0
0
0
0
0
11
11
11
(a) (b)
Figure 7.13 The circulation pattern and mean daily precipitation over the Alpine region during (a) Northerly and (b) Southerly type of the Schüepp clas-
sification. Upper Sea-level pressure (mb) (solid lines) and 500 mb contours (Dm) (dashed). Lower Precipitation (mm). n number of days.
(From Fliri and Schüepp, 1984)
1 method. For the New Zealand region Kidson (1995) also uses the Kirchhofer approach
with the CSIRO model daily sea-level pressure fields for 1980–88 and examines the
changes in type frequency with the results of a doubled carbon dioxide simulation.
7.6 Analogs
The concept that atmospheric flow patterns tend to recur, i.e. that analogs of any pattern
exist, has a long history. This argument served as one approach to long-range forecasting
beginning in the 1920s with the work of F. Baur (1951, 1963) in Germany. More recently
it has been utilized in seasonal forecasts by Bergen and Harnack (1982), but with only
marginal skill.
Good analogs on a hemispheric scale appear to be uncommon (Lorenz, 1969). A study
of 500 mb circulation patterns over the northern hemisphere north of 20°N, using twice-
daily analyses for 1956–79, shows that only about 3 million out of 135 million pairs (2
percent) are relatively good analogs (Ruosteenoja, 1988). These analogs are most common
in late winter and least frequent in summer and autumn. Good analogs are more likely,
and their lifetime is longer, when the seasonal difference between the patterns is short
(three weeks). However, only 842 good analogs were found, and most of these were not
independent (separated by at least ten days). Additional difficulties arise because the
surface and upper-level fields may not match equally well. Even in rare cases where they
do match well the patterns both two days earlier and two days later differ considerably
from one another, as illustrated in Figure 7.14. By using EOFs to describe the hemi-
spheric 500 mb field Ruosteenoja was able to determine the contributions of planetary,
large-synoptic (3,000–4,500 km) and small-scale synoptic (1,000–3,000 km) waves to the
analog index.1 Analog similarity is determined mostly by the planetary waves; the simi-
larity of short-wave patterns at 500 mb, for example, rapidly diminishes (Gutzler and
Shukla, 1984).
At the hemispheric scale, Toth (1991b) concludes that analogs of daily circulation and
time-derivative analogs are no more useful than traditional analogs in forecasting beyond
a few days. However, they yield useful information on the gross structure and phase space
of circulation patterns. Toth (1991c) finds that the best analog to a circulation pattern is
likely to be closer to the climatic mean than to the base case and that analog predictability
does not depend on the distance, in phase space, of the initial flow from the mean. Another
interesting suggestion from his analysis is that persistence of flow increases with increasing
proximity to the climatic mean.
The use of limited regions in which to define analogs is re-examined by van den Dool
(1989). For fifteen years of data, the first five analogs, weighted according to their quality,
were found to provide a sufficient basis for twelve-hour forecasting within a limited area
(a circle of 900 km radius), which avoids rapid advection of changes. A further test was
made for a 2,000 2,000 km area in the eastern United States using 500 mb height fore-
cast maps. The overall conclusion is that analog forecasting, in these situations, has more
skill than previous work suggests. The possibility of using other parameters to define
analogs for different locations and situations is also proposed. For example, in a further
regional study of eastern North America van den Dool (1991) finds that negative or anti-
analogs are almost as common as analogs, except for cases of deep low-pressure systems.
Such “antilogs” show a comparable skill level to analogs for twelve-hour forecasts. This
is a result of the linear component of the absolute vorticity advection (represented in the
linearized barotropic model).
7.7 Seasonal structure

Synoptic climatological methods have found wide application in descriptions of the
temporal characteristics of climate. These include both the structure of the annual climatic
0
0
0
0
0
11
11
11
Figure 7.14 Illustration of the temporal evolution of an exceedingly good 500 mb analog pattern (b) for March 12 1956 and March 6 1976, and the corre-
sponding patterns (a) two days earlier and (c) two days later. (Ruosteenoja, 1988)
1 cycle, in terms of “natural seasons,” weather spells and singularities, and the contribution
of synoptic variability to long-term climatic trends (discussed in the subsequent section).
The idea that certain types of weather tend to recur around a specific calendar date has
a long history in traditional weather lore (Pilgram, 1788; Zimmer, 1941; Inwards, 1950;
Yarham, 1966). Nineteenth-century European meteorologists were particularly interested
in cold spells occurring in winter and spring (Brandes, 1820, 1826; Dove, 1857; Talman,
1919). Dove wrote a memoir on the frosty weather spell, referred to as the “Ice Saints”
after the corresponding saints’ days, that was thought to be frequent in central Europe
during May 11–15. He concluded that associated outbreaks of cold polar air were too
irregular in their occurrence to be connected reliably with the specified dates.
The recurrence tendency of certain weather characteristics around a specific time of
year is referred to as a singularity. Schmauss (1928) used the term in the mathematical
sense of a single point on a time plot of some weather element. However, three approaches
have been used to identify singularities:
1 Examination of mean daily values of a climatic parameter over a long period to iden-
tify irregularities in the seasonal trends. Harmonic analysis is one tool used to filter
an annual time series for this purpose (Craddock, 1956).
2 Analysis of synoptic catalogs to identify time intervals when particular types are
unusually prevalent (Brooks, 1946).
3 Study of the synoptic and spatial characteristics of selected singularities (Bryson and
Lowry, 1955; Duquet, 1963; Carleton, 1986; Carleton et al., 1990).
Many of the weather spells that were recognized in early studies were not substanti-
ated when tests of statistical significance were applied (Marvin, 1919). Also, it was found
that false singularities could be generated in a random series by the introduction of persis-
tence (Bartels, 1948; Baur, 1948). In a statistical analysis of the mean daily central England
temperature series Battye (1980) found no evidence of consistent year-to-year variations
of one to ten-day duration that cannot be ascribed to chance or seasonal trends. Flohn
and Hess (1949) note that the Ice Saints occurred with 77 percent frequency from 1881
to 1910 but that this decreased to 58 percent during 1911–47. Bissolli and Schoenwiese
(1990) show that the period May 6–18 had become a warm anomaly for 1949–85 in six
regions of Germany. Likewise, the Old Wives’ Summer (September 27–October 1)
occurred on September 17–20 during 1916–50 and September 25–6 during 1951–85
compared with 1881–1915. Such differences probably reflect changes in the large-scale
circulation.
Table 7.8 Singularities in Central Europe, 1881–1947
Period Circulation type Characteristics Frequency

(%)
January 15–26 Continental anticylcones Dry, night frosts 78

May 22–June 2 Northern and Central Dry
European highs 80
June 9–18 Northwesterly Summer monsoon,
thundery rains, cool 89
July 21–30 Westerly Summer monsoon 89
August 1–20 Westerly Thundery rains 84
September 3–12 Central European highs Dry 79
September 21–October 2 Central and South-eastern “Old wives’ summer,”
European highs dry 76
December1–10 Westerly Mild 81
Source: from Flohn and Hess (1949).
11 Table 7.9 Singularities in the British Isles, 1873–1961
Period Circulation Characteristics Type Period

type frequency
(%) and
significance
level
January 20–3 AC, S, and E Generally dry and sunny in 50 D
together central and southern England.
Year’s lowest frequency of C
0 type (10–12%) January 24–6
March 12–23 AC, N, and E Notable rainfall minimum in 35 D
together central and southern England. (1% level)
March 12–14 peak of AC
11 May 12–18 N type Annual maximum about these 30 A, B, C
dates; May 14–20 is sunniest
week of the year in Ireland
May 21–June AC type Annual maximum frequency, 5% level A, B, C
10 40% or more on some days
during most of this period;
driest weeks of the year in
0 Scotland, Ireland; more year-
to-year variations in southern
half of England
June 18–22 W, NW, and Generally dry and sunny in 70 D
AC together southern England; cloudy and
wet in Scotland and Ireland
W type frequency 52% on 1% level D
20 June
July 31– C type Sharp peak (replaced by twin 35% B, A, C
August 4 maxima around July 20 and (5% level)
mid-August)
0
August 17– W and NW Wet in most areas 70 D
September 2 together
C type Peaks August 19 and 28 30 (5% level) D
September AC, N, and Dry, especially east and 80 A, B, C
6–19 NW together central England
C type frequency, <20% 5% level A, B, C
between September 6 and 12
October 5–7 AC type Slight check to seasonal 40 (5% level) D
cooling
October 24–31 C, E, and N Great decline to year’s 1% level A, B, C
0 types minimum frequency of AC
type (10%) about October
28–31
Stormy, wet weather 5% level
November 17–20 AC type Dry, foggy period in central 30 (1% level) A, B, C
and southern England
December W and NW Wet and stormy in most areas 70 A, B, C
3–11 together with December 3–9 generally
wettest week of year on
average
December 17–21 AC type Generally dry, foggy weather 25 A, B, C
0
Source: from Lamb (1964).
Note
11 Period A 1873–98, B 1898–1937, C 1938–61, D 1890–1950 ± about ten years.
1 The first calendar of singularities was developed by Schmauss (1938, 1941), based on
investigations of temperature, pressure, precipitation frequency, and wind data for
Germany. Later, research in Germany, Britain, the United States and Japan focused on
circulation characteristics. Using the Hess–Brezowsky Grosswetter catalog for 1881–1947,
Flohn and Hess (1949) identify periods when a particular Grosswetterlage occurred on
three or more consecutive days in at least two-thirds of the sixty-seven-year record. The
definition allows a variation of ± 5 days about the middle date for each singularity. Table
7.8 summarizes periods having a frequency of 75 percent or more. The high frequency
of the summer events is particularly striking. Singularities in the winter half-year, that
are not shown, have frequencies of around 70 percent and are mainly anticyclonic in char-
acter. For the British Isles, Lamb (1964) provides a similar calendar covering 1873–1961.
Events with at least a 25 percent frequency are shown in Table 7.9. It is apparent that
several of the intervals correspond closely to those in central Europe.
Synoptic studies of well known singularities or weather spells began in the early twen-
tieth century. Lehmann (1911) investigated the autumnal warm spells in Europe known
as the Altweibersommer (“Old Wives’ Summer”). In the 1950s such regional phenomena
began to be explored in a global context. Wahl (1953) differentiated between primary
singularities affecting the general circulation and the secondary regional manifestations
of the former. Bayer (1959) defines a primary singularity as a period with more inten-
sive meridional air mass exchange in particular sectors of the hemisphere. It is associated
with enhanced long waves in the 500 mb circulation. A secondary singularity represents
the increased tendency for particular Wetterlage to occur in a specified period and region.
Examples of regional events linked with the larger-scale circulation are the January thaw
in New England, associated with Gulf Coast cyclones (Wahl, 1952; Guttman, 1991) and
the “midsummer high jump” in the southwestern United States (Bryson and Lowry, 1955).
The latter involves the northward shift of the subtropical high over the eastern North
Pacific and the incursion of southwesterly airflow advecting moisture into Arizona. From
a harmonic analysis of the 700 mb heights for 30°–90°N, 160°E to 0° longitude, Lanzante
(1983) identifies some temporal anomalies of regional significance. In Alaska height rises
in late December to early January are terminated by a January thaw. A rapid height rise
in the eastern Pacific in February may account for a minimum in Hawaiian rainfall totals.
An Indian Summer pattern, associated with low heights in high latitudes and weakly posi-
tive anomalies over most of North America, occurs generally in late September–early
October. The second harmonic of the heights suggests that a semi-annual east–west oscil-
lation in the North Pacific may explain these singularities.
In addition to pronounced peaks or troughs in the frequency of types of weather regime
about a particular date, it is apparent that there may be times of the year when there is
a tendency to the abrupt onset or decline of a certain regime. These short-period events
are superimposed on broad maxima and minima of individual circulation regimes during
the course of the year. Synoptic climatologists recognized that the overall frequency of
circulation types could be used to characterize seasons, and that abrupt transitions often
mark their beginning or termination. The idea that calendar intervals defined by partic-
ular types of weather regime can be distinguished was first developed by Lamb (1950).
Based on the frequency of time intervals with twenty-five or more days of a particular
weather pattern (a weather spell), he proposed a calendar of natural seasons for the British
Isles. The frequency and timing of such long spells, disregarding brief interruptions up
to three days in length, are illustrated in Figure 7.15. During 1898–1947 158 long spells
occurred, seventy-nine of which involved Westerly type and fifty-six involved Anti-
cyclonic and a hybrid type. Long spells are most frequent in October and midsummer,
with a minimum in April–June. A similar seasonal calendar was prepared for central
Europe by Baur (1958), and the results are compared in Table 7.10.
For Japan the duration of six seasons – winter monsoon, spring, the Bai-u rains,
midsummer, the Shurin rains and late autumn – has been determined from curves of
11
11
Figure 7.15 Percentage frequency of spells of more than twenty-five days of Lamb weather types
and five natural seasons, 1898–1947. (After Lamb, 1950)
Table 7.10 Natural seasons in the British Isles and Central Europe
0
Season Calendar period in Calendar period in
Central Europe the British Isles
High winter January 1–February 14 January 20–March 31
Fore-spring February 15–March 31 January 20–March 31
Spring April 1–May 16 April 1–June 17
Fore or early summer May 17–June 30 April 1–June 17
High summer July 1–August 15 June 18–September 9
Late summer or early autumn August 16–September 30
Autumn October 1–November 15 September 10–November 19
0 Fore-winter November 16–December 31 November 20–January 19
Source: after Lamb (1950) and Baur (1958).
five-day and daily means of sunshine duration, cloudiness, and precipitation; temperature
and vapor pressure are secondary criteria for the winter monsoon season (Maejima, 1967).
For East Asia, Yoshino (1968) uses the pentad frequency of the seven major types of
pressure pattern for 1946–65 to develop a seasonal calendar. The concept of a “monsoon
year” has been revived by Yasunari (1991), based on anomalies of precipitation and SSTs
0 in the eastern and western Pacific. He emphasizes the high autocorrelation of the vari-
ables during May–December and the low persistence in January–April.
For the Pacific Northwest of the United States a more conventional single variable
11 approach is adopted by Alsop (1989). He applies cluster analysis to weekly maximum,
1 minimum, and mean temperatures to define four seasons. Wos (1980) also uses a taxonomic
clustering method to group geometric distances between every pair of pentads in a year at
thirty-nine stations in northwest Poland for 1951–65. Forty weather states are defined, based
on mean temperature (grouped according to maximum and minimum values above or below
freezing), cloudiness (above/below 50 percent), and above/below-average precipitation.
Nine types of seasonal structure, with four to six seasons, are found within northwest
Poland. Temporal principal component analysis of temperature, vector wind, and precipi-
tation is used to determine monthly seasons in southern California by Green et al. (1993).
For the Atlantic and Gulf coasts of the United States, Pielke et al. (1987) suggest using an
air mass–frontal approach, defining seasons according to the frequency with which the polar
front is south of a given location. They distinguish five synoptic categories:
1 mT air in the warm sector of an extratropical cyclone.

2 mT/cP or mP/cA air ahead of a warm front in a region of cyclonic curvature.
3 cP or cA air behind a cold front in a region of cyclonic curvature.
4 cP or cA air in a polar anticyclone.
5 mT air near a subtropical ridge.
Winter (summer) is defined as having the highest frequency of category 2, 3, and 4 (1

and 5) air occurrences, with spring and autumn being transitional. Quantitative criteria,
based on cumulative changes in categories 2, 3, and 4, were developed in order to provide
threshold values for the seasonal transitions. They show that, over most of these south
and east coastal regions of the United States, winter lasts from late October or early
November to late March or early April and summer from late May or early June to late
August or late September. Moreover the frontal movement is characterized by rapid shifts
rather than steady displacement. They suggest that this approach can have hemisphere-
wide application in middle latitudes and could be refined using such parameters as 700
mb thermal advection and 500 mb vorticity advection in relation to location within a
frontal cyclone.
Shifts in the mid-latitude long-wave structure have been investigated as a cause of
abrupt seasonal changes. For example, the 500 mb trough over eastern Europe is farther
east in summer than in winter, and a subsidiary trough may develop upstream (De la
Mothe, 1968). Since the wave number is an integer, such changes in the tropospheric
long waves can greatly modify the regional circulation. In an attempt to define primary
singularities and the seasonal structure of the year, Bryson and Lahey (1958) assembled
a synthesis of various hemispheric circulation indices and regional criteria for North
America (Figure 7.16). Considerable coherence is apparent in late March, mid-August,
and the beginning of October, although this last date is not identified as a seasonal break.
Kalnicky (1987) used EOFs of the daily Dzerdzeevski types for 1899–1969 to examine
seasonal transitions in terms of the frequencies of hemispheric patterns. The first four
eigen vectors of thirteen major types accounted for 74 percent of the variance in the
frequency of daily circulation patterns. Twenty-five days of maximum discontinuity were
identified in annual plots of the day-to-day variation of the coefficients of these four eigen
vectors. The four seasons he defined were winter: October 23–March 8; spring: March
9–June 17; summer: June 18–August 31; and autumn: September 1–October 22. The
summer–autumn transition had the largest discontinuity and occurred within ± four days
of the specified date in 51 percent of years. However, the length and character of the
seasons were found to be variable. The period 1899–1919 had the longest winters, while
1920–52 had the longest summers and the most frequent Indian Summer episodes. During
the 1953–69 cool episode the summers tended to have circulation features typical of spring
and autumn. As a result, plots of frequency distributions of the start dates of northern
hemisphere seasons for 1899–1965 show a variation about the mean date of around plus-
or-minus twelve days (Bradka, 1966).
0
0
0
0
0
11
11
11
Figure 7.16 A summary of the criteria used to establish a natural calendar for the northern hemisphere. (From Bryson and Lahey, 1958).
Abbreviations: Di2 40° – Storminess index. MSL meridional circulation at 40°N. Anticyc. – Difference between mid-latitude west-
erlies and subtropical easterlies. Baro. – Difference between MSL and 700 mb zonal indices; SP = subpolar; ML = mid-latitude; ST
= subtropical. Index var. – Standard deviation of circulation indices. Ortho Poly. – Orthogonal polynomials for the western hemi-
–
sphere at 500 mb. G.A. and M. – Occurrence of Aleutian low in Gulf of Alaska and Manchuria. 45/155P5 – Normal five-day mean
–
pressure at 45°N, 155°W. Ariz. P5 – Same as above over Arizona. N.W. U.S. High – Anticyclones in northwestern USA; migr. =
migratory; basin = Great Basin. Alberta – Dominant synoptic pattern in southern Alberta. GWT. – Grosswetter type; = 3 change,
x = 2 change. Medit. Low–High – Cyclonic/anticyclonic regime in the Mediterranean area.
1 Conceptual efforts have been made to relate natural seasons to gross features of the
energy budget and its latitudinal gradient. Bradka (1966) argued that the latitudinal energy
budget gradient determines the zonal circulation in middle latitudes, while the differen-
tial effects of solar radiation over land and ocean foster meridional components.
Additionally, radiative cooling in high latitudes intensifies the meridional temperature
gradient and favors blocking situations. James (1970) applied the (essentially linear) quasi-
barotropic relationship between air mass temperature and mean pressure:
∂T
T ≈ T0 (p p0)
∂p
where T0 the air mass temperature at a conventional pressure level p0 (say, 1,012 mb).
Plots of mean daily maximum temperature against mean pressure for each month are used
as seasonal indicator diagrams for Australian stations to show when the surface air
performs work against the ambient environment (decreasing p and increasing T) or vice
versa.
A number of factors are involved in the seasonal transitions and what Lamb (1964)
termed master seasonal trends. Undoubtedly, too, there are complex interactions between
them. Following Lamb and Bradka (1966), seven factors can be recognized as playing a
role in the seasonal development of the large-scale circulation in the northern hemisphere:
1 The midwinter maximum in the latitudinal gradient of the energy balance and the
associated maximum of the zonal index.
2 The late winter maximum extent of snow cover and sea ice, linked with the expan-
sion of the tropospheric polar vortex and the southernmost position of extratropical
cyclone tracks about February–March. Meridional circulation patterns are common,
with enhanced north–south interchange of air masses.
3 Persistent negative net radiation values over the Beaufort Sea–Canadian Arctic
Archipelago and the related development of the major polar anticyclone in this region
in March–April.
4 The minimum meridional energy balance gradient and related minimum intensity of
the circulation in the Atlantic sector in May; the circulation over the North Pacific
remains predominantly zonal.
5 The summer contrast in heating between the snow-free land in northern Eurasia and
North America and the cold Arctic Ocean and its sea ice cover. This probably controls
the northward displacement of cyclone tracks over western North America in July
and August. The “cold pole” shifts towards the Atlantic–European sector and this is
accompanied by a southward displacement of cyclone tracks.
6 Early autumn cooling, assisted by the cold seas off eastern Canada (Barry, 1993),
helps to strengthen the upper trough over eastern North America in September. The
Icelandic low deepens and the Atlantic circulation intensifies. Depressions now move
northeastward into the Barents Sea.
7 High-latitude cooling and the expansion of sea ice and snow cover lead to expan-
sion and intensification of the circumpolar vortex and the southward displacement of
cyclone tracks, particularly from October onward (Lamb, 1955).
It is apparent that the jumps which often accentuate the seasonal transitions can in
some cases be attributed to specific factors. These include the rapid change in surface
albedo, and the associated relative magnitudes of the surface energy budget terms, with
the disappearance of snow cover. In the Arctic tundra this transition typically lasts only
ten days (Weller and Holmgren, 1974). A delay is imposed on the seasonal poleward
retreat of the subtropical jetstream by the influence of the Tibetan Plateau–Himalayan
ranges on the circulation; the eventual disappearance of snow cover over the Tibetan
11 Plateau in June contributes to the circulation shift over South Asia in June. Nevertheless,
changes in the external forcing may not be required. Chapter 4 discussed the role of
synoptic eddies in transitions between zonal and blocking modes, and the ordered chaos
of the “almost intransitive” atmospheric circulation has also been noted.
7.8 Climatic trends

The existence of a number of long catalogs of airflow types and circulation patterns (see
Table 7.2) has stimulated studies of the role of changes in regional or hemispheric circu-
0 lation in climatic fluctuations and trends. Dzerdzeevski (1963, p. 293) stated, “all changes
in temperature and precipitation are connected with changes in frequency and duration of
large-scale circulation patterns.” Many studies in the 1950s and 1960s focused on the
synoptic climatology of extreme conditions (cold summers, snowy winters) that might
11 favor a glacial climatic regime. For example, Rex (1950) concluded that reduced precip-
itation over Europe associated with Scandinavian blocking implied that such patterns
would inhibit glacial onset. Namias (1957) showed that extreme continentality, with warm
summers and interdependent cold winters in Sweden, occurs during weakened zonal circu-
lation which would therefore interdict a glacial regime. Regional investigations of airflow
patterns and associated weather conditions in Labrador-Ungava (Barry, 1960, 1966) and
0 Keewatin, Canada (Brinkmann and Barry, 1971), and in Tasmania (Derbyshire, 1971)
attempted to identify patterns favoring winter snowfall and cool summers that would be
important for glacierization. Synoptic weather typing is now finding increasing applica-
tion in historical climate reconstruction for Europe from the seventeenth to the nineteenth
centuries (Glaser and Walsh, 1991). A recent attempt to characterize circulation patterns
during the late Maunder Minimum cool episode (1675–1704) uses standardized monthly
temperature and precipitation indices for winter and spring in Iceland, England, and
Switzerland, together with documentary information on wind direction and weather events,
to infer air masses and monthly mean surface pressure maps over Europe. Each month
of 1675–1704 was also classified according to the Lamb scheme (Wanner et al., 1994)
0 and this was repeated for comparison for 1961–90. Finally, the P and S indices (see
section 7.4.1) were determined. The results show a clear difference between the two thirty-
year periods. The late Maunder Minimum experienced frequent blocking and northerly
flow, with outbreaks of cold continental air, especially in the 1690s. The extreme NAO
(“pressure reversal”) mode of January 1963 (Moses et al., 1987) is proposed as a candi-
date analog.
Changes in the frequency of circulation types during the twentieth century have received
increased study as a result of the accumulating evidence for global warming. Lamb (1965)
drew attention to the long-term fluctuations in the frequency of the Westerly circulation
type over the British Isles (Figure 7.7); the available record shows an average frequency
0 of eighty-five days per year in the 1880s and 1890s, a peak of about 110 days per year
around 1920, followed by a decline to the early 1940s and, after a brief recovery,
a decrease to a minimum of only about sixty days per year in the early 1980s. Recent
years show an irregular recovery. Trends in circulation have similarly been identified and
related to climate changes over Europe, using the Grosswetter catalogue for 1881–1989
(Bardossy and Caspary, 1991; Klaus, 1993). In winter, zonal types prevailed from 1890
to 1920, followed by more meridional conditions until about 1965, when zonal patterns
again increased considerably. This is thought to have contributed to the winter warming,
especially after 1980. In summer, zonal flows peaked around 1900, 1920, and 1945, with
meridional patterns increasing in the 1960s; the frequency and persistence of southerly
0 flows increased significantly, for example, while northerly flows decreased. Earlier, van
Dijk and Jonker (1985) analyzed the duration of Grosswetterlagen (GWL) and Gross-
wettertypen (GWT) for 1899–1970. The durations are found to be log-normally distributed.
11 For the GWL, the median duration is about three and a half days, except for Wa and
1 Table 7.11 Circulation epochs based on the
Vangengeim–Girs large-scale patterns of
circulation
1891–99 WC
1900–28 W
1929–39 E
1940–48 C
1949–71 EC
1972–93 E
Source: Dmitriev (1994).
Wz, where it is four days. The median duration of the GWT is about half a day larger
in each case. For 1949–70 the frequency of occurrence of GWL and GWT differed signif-
icantly from the overall values. However, the type durations for 1949–79 indicated that
no changes in circulation could be demonstrated statistically for this period. For the British
Isles, Perry (1970) concluded that a decrease in frequency of Westerly types in winter
between 1910–30 and 1948–68 was accompanied by a pronounced reduction in the mean
duration of a Westerly spell, whereas Easterly types increased in frequency and run length.
For the northern hemisphere the Russian catalogues of Vangengeim–Girs and
Dzerdzeevski have been used by several workers to identify “circulation epochs.” Table
7.11 shows those recognized by Dmitriev (1994) with the former scheme; however, this
differs in some details for the 1930s from one put forward earlier by Kozuchowski and
Marciniak (1988) for 1891–1983. The annual ratio of zonal to meridional patterns in the
Dzerdzeevski catalogue is low in the early twentieth century (around 0.5 in 1915), rises
to 1.1 from the 1930s to the 1940s, declines to 0.75 around 1960 and recovers to 0.9–1.0
in the 1970s (Savina, 1987). The major trends seem to be broadly consistent with changes
in other indices, although more thorough intercomparisons are needed.
On a regional scale, the frequencies of the three general classes of synoptic type for
the Alps, according to Schüepp (1979) – namely convective, advective, and mixed – show
pronounced trends during 1945–94 (Stefanicki et al., 1998). The convective types show
a linear increase of thirteen days per decade, while advective types decrease by eleven
days per decade (both significant at the 0.1 percent level) and the mixed types decline
slightly. Most of the increase in convective types is due to high-pressure situations in
winter and autumn which are shown to be strongly correlated with the positive NAO
mode. Wanner et al. (1997) also show an increase in Anticyclonic GWL, especially in
July, and of West–maritime and SW GWL patterns in autumn during the 1970s and 1980s.
The possibility that changes may also occur in the weather (or air mass) characteris-
tics of the individual circulation types was pointed out by Barry and Perry (1969, 1970).
For example, the change in mean daily temperature for a given month can be regarded
as comprising a component due to changes in the frequencies of the circulation types and
another component due to changes in the characteristics of individual types. The total
change in mean temperature between two time periods can be expressed (Perry and Barry
1973):
兺冤冥
K
fi (Ti Ti ) fi Ti
T =
i=1 n n
where fi frequency of type i during the first time period, Ti mean temperature of
type i during the first time period, n total number of days in the first time period,
f i f frequency of type i during the second period, and Ti Ti mean temper-
ature of type i during the second time period. The last term on the right-hand side of the
11 equation, which is independent of any change in frequency, represents the component due
to within-type temperature changes. The first term on the right side represents the effect
on T of a change in type frequency when a change occurs in the temperature of type i
in the second period. For some purposes it may be of interest to separate the two elements
making up the first term on the right-hand side. For four stations in the British Isles for
1925–35 and 1957–67, the changes in mean daily maximum and minimum temperatures
were associated with changes in frequency of the Lamb types in January; in April, July,
and October, however, there were significant “within-type” temperature changes affecting
Anticyclonic, Westerly, Northerly, and Easterly types. Although it is unlikely that the
0 types were inhomogeneous between the two periods, owing to Lamb’s procedure of clas-
sifying the years in random order, there could be subtle differences in wind speed or
trajectory within a single type grouping between the two periods (Lawrence, 1970). For
large samples, however, it would be reasonable to assume a comparable range of wind
11 speeds and trajectories within each type. This technique has also been applied to temper-
ature and precipitation associated with circulation patterns over the western United States
(Barry et al., 1981). The most common type, a zonal pattern which is most frequent in
winter, is warmer than average in the mean in Colorado but becomes colder than normal
in cold years. Type 2 with a high over the northern Rocky Mountains tends to remain
cold even in months with a warm anomaly, and changes in its frequency exert strong
0 control on temperatures in all seasons in Colorado.
Other studies of climate changes employing a synoptic climatological approach for
stations in Yukon–Alaska (Kalkstein et al., 1990) and the contiguous United States
(Kalkstein et al., 1998) examine changes in air mass frequencies and types. In Alaska
the frequency of warm (cold) air masses has increased (decreased) since about 1948–88.
Additionally, the coldest air masses have warmed by 1°– 4°C. However, they do not quan-
tify the separate effects of between and within-type changes. The second study finds a
winter decline in the frequency of air mass transitions, related to more persistent merid-
ional circulation patterns. There are increased frequencies of dry polar and dry moderate
air masses during winter 1948–93. In summer there are increases in the frequency of
0 moist tropical air. This gave rise to higher summer temperatures and increased cloud
cover.
7.9 Environmental applications

Applications of synoptic climatology use information about airflow movement close to
the earth’s surface to evaluate a variety of environmental problems. Muller (1977) was
among the first to recognize that a climatology of synoptic weather types could form an
important baseline for studying the link between atmospheric conditions and environ-
mental phenomena. Synoptic approaches have unique appeal because they permit
0 evaluation of the synergistic impacts of an entire suite of weather elements, and, as a
consequence, other physical scientists are incorporating them within their research design.
Synoptic climatology provides a convenient framework for interpreting variations in
atmospheric chemistry because changes in transport, diffusion, and stability accompany
different patterns of airflow and weather type. Hence there have been numerous synoptic
studies of spatiotemporal variations in the concentrations of atmospheric pollutants. For
example, Kalkstein and Corrigan (1986) analyzed sulphur dioxide concentrations in
Wilmington, Pennsylvania, using air mass categories. The typical synoptic circulations
associated with each category were also illustrated. The Lamb weather types have been
used to evaluate ozone pollution in the United Kingdom and Ireland (O’Hare and Wilby,
0 1995), establishing the dominant role of stable Anticyclonic and Easterly types in
producing peak ozone episodes. The Lamb types are also used to study synoptic influ-
ences on air pollution in Birmingham, UK (McGregor and Bamzelis, 1995). In the arid
11 southwestern United States air quality is shown to be related to synoptic-scale variations
1 in middle and upper tropospheric wind direction, temperature, and humidity. Categories
are specified in a four-dimensional classification of circulation types developed through
a two-stage hierarchical clustering approach (Davis and Kalkstein, 1990; Davis and Gay,
1993). Similar work has been carried out by Comrie and Yarnal (1992) in developing a
synoptic-sequence climatology of surface ozone concentrations for Pittsburgh. Comrie
(1992) demonstrates a procedure to distinguish within-type (non-synoptic) components
from between-type (synoptic) components in a time series of visibility data at Pittsburgh.
This allows removal of the synoptic climate signal from environmental data.
Precipitation composition has been interpreted in relation to synoptic types in the United
Kingdom (Davies et al., 1990, 1991). Cyclonic, Westerly, and Anticyclonic types play
the dominant roles in affecting precipitation chemistry. Dorling et al. (1992, 1995) combine
the pattern recognition capabilities of cluster analysis with isobaric trajectory data to study
source effects on precipitation ion concentration. They show that, for the United Kingdom,
trajectory clusters, combined with precipitation amount, explain much of the daily vari-
ability in precipitation chemistry. In Norway, however, topographic effects confound the
other signals.
In an interesting application to wind-energy potential, Palutikof et al. (1987) investi-
gated the relationship between Lamb’s types and wind speed. Todhunter (1989) uses
weather type clusters to stratify the energy budgets of urban surfaces for Boston,
Massachusetts.
Since the pioneering work by Hoinkes (1968), relating glacier variations in Switzerland
to frequencies of particular circulation types, there have been a number of similar studies.
Glacier mass balance in British Columbia was related to synoptic-scale atmospheric circu-
lation by Yarnal (1984a). Trends in summer energy balance on the West Gulkana glacier,
Alaska, are analyzed in terms of a Temporal Synoptic Index (TSI) and changes in type
frequency by Brazel et al. (1992). In an extension of Hoinkes’s work, Fitzharris (1998)
has looked at four glaciers in the European Alps that have direct mass balance measure-
ments extending back more than thirty-five years and analyzed the frequency of
Grossswetterlagen during the accumulation and ablation seasons for high and low mass
balance glacier years. High mass balance years have higher frequencies of westerlies,
troughs, and depressions in summer and winter than low mass balance years. The asso-
ciations between weather types and circulation pattern can be quantified into formulae
that allow estimates of Alpine glacier mass balance from a weather index based on
Grosswetterlagen frequency.
There have been several synoptic climatology studies of snow melt and snow accu-
mulation. Recent examples are the analysis of synoptic controls of melt conditions on the
Greenland ice sheet (Mote, 1998) and of snowpack melt conditions in the Southern Alps
of New Zealand (Hay and Fitzharris, 1988; Neale and Fitzharris, 1997). Snowfall vari-
ability in relation to synoptic type has been examined for New England and the Great
Lakes (Ellis and Leathers, 1996). Fitzharris and Bakkehøi (1986) relate winters with a
high risk of avalanches in Norway to synoptic type. In addition to weather much colder
than normal, northerly and easterly airflow patterns tend to predominate in major avalanche
winters.
Many investigations have linked atmospheric circulation and basin hydrology, and espe-
cially runoff over seasons (Keables, 1988), basin-scale discharge and water quality (Yarnal
and Draves, 1993; Yarnal, 1997), and flood events (Maddox et al., 1980). Hay et al.
(1991) used six conceptual daily weather types to simulate average and extreme precip-
itation in the Delaware River basin. The variability in type frequency and precipitation
characteristics among types was determined for a GCM control case and one with doubled
carbon dioxide concentration. Hughes and Guttrop (1994) devised a statistical model to
relate circulation fields, rainfall, and hydrologic processes in Washington state (1993,
1997), while Wilby (1993) used the Lamb weather types to look at river water quantity
and quality in the United Kingdom.
11 Weather types have been used as a concise surrogate measure of weather conditions
in a number of ecological studies. Masterman et al. (1996) have examined the autumn
migration of the bird cherry aphid, while Boyd and Bell (personal communication) are
investigating the role of wind direction in affecting the migration of geese between Iceland
and Scotland. The literature in agricultural meteorology using synoptic techniques is
surprisingly sparse, although Dilley (1992) studies corn (maize) yields in southwestern
Pennsylvania, making use of composite maps of surface pressure. Pepin (1995), investi-
gating the potential for ecological change in the northern Pennines of England, shows
that scenarios of warming for high-altitude sites depend critically on the frequency of
0 airflow types. The beneficial effects of any future regional warming would be reduced
for this area if the frequency of westerly winds were to increase, since the lapse rates in
such airstreams are generally much steeper than those of continental airstreams reaching
northern England.
11 In bioclimatology, attention is turning to the identification of weather types associated
with human health and mortality. Kalkstein (1991) found that summer days with the
highest mortality rate in St Louis, Missouri, were associated with cases of tropical air
that were oppressively hot at night, although such cases were infrequent. Unlike many
other cities, high pollution levels did not appear to have a significant effect on mortality
in St. Louis. Using the air mass-based Temporal Synoptic Index (TSI), Jamason et al.
0 (1997) find that air mass-related pollution levels in spring and summer strongly influence
New York city hospital admissions of asthma sufferers. The TSI has also been used with
twenty-four and forty-eight-hour local forecasts for Philadelphia, Pennsylvania, to develop
a hot weather health watch/warning system (Kalkstein et al., 1996). The relationship
between air pollution and mortality has been studied in Utah, using the TSI (Pope and
Kalkstein, 1996). More recently the continent-wide Spatial Synoptic Classification (SSC)
of air masses has been used to determine high-risk summer and winter air masses for
both total and elderly mortality rates in forty-four US cities (Kalkstein and Greene, 1997).
In the eastern and Midwest United States two air masses are associated with sharp increases
in mortality. A very warm maritime air mass is important in the eastern United States
0 and a hot dry air mass is important in many cities. Cities in the south and southwest have
a weaker climate–mortality relationship and, in winter, air mass-related mortality increases
are small. Work on extreme mortality levels has been initiated in the United Kingdom
for Birmingham by McGregor (1998), and wider application of synoptic climatological
techniques in environmental health can be anticipated.
The number and diversity of applied synoptic climatological studies continues to prolif-
erate, which in itself is a tribute to the usefulness and robustness of the method. As Yarnal
(1993) asserts, “relating the atmospheric circulation to the complex surface environment
adds another major dimension to the demands on the synoptic climatologist.”
0
Note
1 The analog index (Y) used by Ruosteenoja (1988) is the mean-square difference between two
500 mb fields:
100
Y= 兺 [C (t ) C (t )]
n=2
n a n b
2
where C(t) = the time-dependent coefficient of the respective EOF, n = the EOF component,
n = 2 to 100, and ta , tb = the two dates. The first EOF is omitted, since it largely represents
the annual variation of the 500 mb height pattern. The Y value is further normalized, to remove
the greater variability of winter 500 mb heights, using the function:
0
N() = 1 0.5 cos
where is an angle describing the time of year; N varies sinusoidally from 1.5 on January 31
11 to 0.5 on July 31. The normalized index is:
1 Y
YN =
[N(a) · N(b)]1/2
where a and b are the seasonal angles of the two dates. The expected value of YN ~ 20,000 m2.
A relatively good analog has YN < 8,000 and a good one YN < 6,000.
References
Abercromby, R. 1883. On certain types of British weather. Quart. J. Roy. Met. Soc., 9: 1–25.
Alsop, T.J. 1989. The natural seasons of western Oregon and Washington. J. Climate, 2 (8): 888–96.
Barchet, W.R. and Davis, W.E. 1984. A Weather Pattern Climatology of the United States. PNL-
4889, Battelle Pacific Northwest Laboratory, Richland WA., 31 pp.
Bardossy, A. and Caspary, H.J. 1991. Detection of climate change in Europe by analyzing European
circulation patterns from 1881 to 1989. Theoret. Appl. Climatol., 42: 155–67.
Bardossy, A., Duckstein, L., and Bogardi, I. 1994. Fuzzy rule-based classification of atmospheric
circulation patterns from 1881 to 1989. Theoret. Appl. Climatol., 42 : 155–67.
Barnston, A.G. and Livezey, R.E. 1987. Classification, seasonality and persistence of low-frequency
atmospheric circulation patterns. Mon. Wea. Rev., 115 (6): 1083–126.
Barry, R.G. 1959. A synoptic climatology for Labrador-Ungava. Arctic Met. Research Group,
Publication in Meteorology 17. McGill University, Montreal, 159 pp.
Barry, R.G. 1960. A note on the synoptic climatology of Labrador-Ungava. Quart. J. Roy. Met.
Soc., 86: 557–65.
Barry, R.G. 1963. Aspects of the synoptic climatology of central south England. Met. Mag., 92:
300–8.
Barry, R.G. 1966. Meteorological aspects of the glacial history of Labrador-Ungava with special
reference to atmospheric vapor transport. Geogr. Bull., 8, 319–40.
Barry, R.G. 1973. A climatological transect on the east slope of the Front Range, Colorado. Arct.
Alp. Res., 5: 89–110.
Barry, R.G. 1980. Synoptic and dynamic climatology. Progr. Phys. Geogr., 4: 88–96.
Barry, R.G. 1993. Canada’s cold seas. In: H.M. French and O. Slaymaker, eds, Canada’s Cold
Environments, McGill-Queen’s University Press, Montreal, pp. 29–61.
Barry, R.G. and Perry, A.H. 1969. “Weather type” frequencies and the recent temperature fluctua-
tion. Nature, 222: 463–4.
Barry, R.G. and Perry, A.H. 1970. “Weather type” frequencies and temperature fluctuation: a reply.
Nature, 226: 634.
London, 555 pp.
Barry, R.G., Bradley, R.S., and Tarleton, L.F. 1982. A synoptic type classification and catalog for
the western United States. In: R.S. Bradley, R.G. Barry, and G. Kiladis, eds, Climatic Fluctuations
of the Western United States during the Period of Instrumental Records, Contrib. No. 42. Depart-
ment of Geology and Geography, University of Massachusetts, Amherst MA, pp. 96–150.
Barry, R.G., Crane, R.G., Schweiger, A., and Newell, J. 1987. Arctic cloudiness in spring from
satellite imagery. Intl. J. Climatol., 7: 423–51.
Barry, R.G., Kiladis, G.N., and Bradley, R.S. 1981. Synoptic climatology of the western United
States in relation to climatic fluctuations during the twentieth century. J. Climatol., 1: 97–113.
Bartels, J. 1948. Anschauliches über den statistischen Hintergrund der sogenannten Singularität im
Jahresgang der Witterung. Ann. Met., 1: 106–27.
Bartzokas, A. and Metaxas, D.A. 1996. Northern hemisphere gross circulation types: climatic change
and temperature distribution, Met. Zeit., 5: 99–109.
Battye, D.G.H. 1980. A note on singularities. Met. Mag., 109: 358–62.
Baur, F. 1947. Musterbeispiele europäischer Grosswetterlagen. Deiterich, Wiesbaden, 36 pp.
Baur, F. 1948. Zur Frage der Echtheit der sogenannten Singularitäten im Jahresgang der Witterung.
Ann. Met. 1: 372–8.
Baur, F. 1951. Extended-range weather forecasting. In: T.F. Malone, ed., Compendium of
Meteorology, Amer. Met. Soc., Boston MA, pp. 814–33.
Baur, F. 1958. Physikalische-statistische Regeln als Grundlagen für Wetter und Witterungsvorher-
sagen, 2, Akad. Verlag, Frankfurt-am-Main, 152 pp.
11 Baur, F. 1963. Grosswetterkunde und Langfristige Witterungsvorhersagei. Akad. Verlag, Frankfurt-
am-Main, 91 pp.
Bayer, K. 1959. Witterungssingularitäten und allegmeine Zirkulation der Erdatmosphere. Geofys.
Sbornik (Prague), 125: 521–634.
Beaumont, P. and Hawksworth, K. 1997. A calibration of the Lamb airflow classification model to
predict past precipitation in Wales. Intl. J. Climatol., 17 (13): 1397–420.
Bergen, R.E. and Harnack, R.P. 1982. Long-range temperature prediction using a simple analogue
approach. Mon. Wea. Rev., 110: 1083–99.
Bijvoet, H.C. and Schmidt, F. 1958. Dutch Weather Type Classification. Koninklijk Ned. Met. Inst.,
Wet. Rapt, 58–4.
0 Bissolli, P. and Schoenwiese, C.-D. 1990. Spatial and temporal variations of singularities in the
Federal Republic of Germany, 1881–1986. Met. Rdsch., 42: 33–42.
Blair, D. 1998. The Kirchhofer technique of synoptic typing revisited. Intl. J. Climatol., 18 (14):
1625–35.
11 Blasing, T.J. 1975. A comparison of map-pattern correlation and principal component eigenvector
methods for analyzing climatic anomaly patterns. Preprint vol., Fourth Conference on Probability
and Statistics in Atmospheric Sciences, Amer. Met. Soc., Boston MA, pp. 96–101.
Blasing, T.J. 1981. Characteristic anomaly patterns of summer sea-level pressure for the northern
hemisphere. Tellus, 33: 428–37.
Blasing, T.J. and Lofgren, G.R. 1980. Seasonal climatic anomaly types for the North Pacific sector
and western North American. Mon. Wea. Rev., 108: 700–19.
0 Bogardi, I., Matyasovszky, I., Bardossy, A., and Duckstein, L. 1993. Application of a space–time
stochastic model for daily precipitation using atmospheric circulation patterns. J. Geophys. Res.,
98 (D9): 16653–67.
Bonnell, M. and Sumner, G. 1992. Autumn and winter daily precipitation areas in Wales, 1982–83
to 1986–87. Intl. J. Climatol., 12 (1): 77–102.
Bradka, J. 1966. Natural seasons in the northern hemisphere. Geofys. Sbornik (Prague), 14: 597–648.
Bradley, R.S. and England, J. 1979. Synoptic climatology of the Canadian High Arctic. Geogr.
Annal., 61A: 187–201.
Brandes, H.W. 1820. Untersuchungen über den mittleren Gange der warme Änderungen durchs
ganze Jahre. Beiträge zur Witterungkunde. Leipzig.
0 Brandes, H.W. 1826. Bemerkungen über die Zeitpunkte grösser Kälte nach der Mitte des Winters.
Unterhaltungen f. Freunde d. Physik u. Astron., 2: 148–59.
Brazel. A.J., Chambers, F.B., and Kalkstein, L.S. 1992. Summer energy balance on West Gulkana
glacier, Alaska, and linkages to a temporal synoptic index. Zeit. Geomorph., N.F. suppl. 86:
15–34.
Briffa, K.R. 1995. The simulation of weather types in ECMs: a regional approach to control-run
validation. In: H. von Storch and A. Navarra, eds, Analyses of Climate Variability: Applications
of Statistical Techniques. Springer-Verlag, Berlin pp. 121–38.
Briffa, K.R., Jones, P.D., and Kelly, P.M. 1990. Principal component analysis of the Lamb cata-
logue of daily weather types. 2. Seasonal frequencies and update to 1987. Intl. J. Climatol., 10:
549–64.
0 Brinkmann, W.A.R. and Barry, R.G. 1971. Palaeoclimatological aspects of the synoptic climatology
of Keewatin, Northwest Territories, Canada. Palaeogeogr., Palaeoclimatol, Palaeoecol., 11:
87–91.
Brinkmann, W.A.R. 1999a. Within-type variability of 700 hPa winter circulation patterns over the
Lake Superior basin. Intl. J. Climatol., 19: 41–58.
Brinkmann, W.A.R. 1999b. Application of non-hierarchically clustered circulation components to
surface weather conditions: Lake Superior basin winter temperatures. Theor. Appl. Climatol., 63:
41–56.
Brooks, C.E.P. 1946. Annual recurrences of weather: singularities. Weather, 1: 107–13, 130–4.
Bryson, R.A. 1966. Air masses, streamlines, and the boreal forest. Geogr. Bull. (Ottowa), 8 (3):
228–69.
0 Bryson, R.A. and Lahey, J.F. 1958. The March of the Seasons. Final Rep., Contract AF 19-(604)-
992. Madison WI: Meteorological Department, University of Wisconsin, 41 pp.
Bryson, R.A. and Lowry, W.P. 1955. Synoptic climatology of the Arizona summer precipitation
11 singularity. Bull. Amer. Met. Soc., 36: 329–39.
1 Bürger, H. 1958. Zur Klimatologie der Grosswetterlagen. Ein witterungsklimatologischer Beitrag.
Berichte dtsch. Wetterdienstes, 6 (45), Offenbach-am-Main, 79 pp.
Carleton, A.M. 1986. Synoptic-dynamic character of bursts and breaks in the southwest US summer
precipitation singularity. J. Climatol., 6: 605–23.
Carleton, A.M. 1987. Summer circulation climate of the American southwest, 1945–84. Annals
Assoc. Amer. Geogr., 77: 619–34.
Carleton, A.M., Carpenter, D.A., and Weser, P.J. 1990. Mechanisms of interannual variability of
the southwest United States summer rainfall maximum. J. Climate, 3 (9): 999–1015.
Cavazos, T. 1997. Downscaling large-scale circulation to local winter rainfall in northeastern Mexico.
Cavazos, T. 1998. “Downscaling Large-scale Circulation to Local Winter Climate using Neural
Network Techniques.” Unpubl. Ph.D. dissertation, Department of Geography, Pennsylvania State
University, University Park PA, 127 pp.
Cavazos, T. 1999. Large-scale circulation anomalies conducive to extreme precipitation events and
derivation of daily rainfall in northeastern Mexico and southeastern Texas. J. Climate, 12 (5, Pt.
2): 1506–23.
Christensen, W. and Bryson, R.A. 1966. An investigation of the potential of opponent analysis for
weather classification. Mon. Wea. Rev., 94: 697–709.
Cohen, S.J. 1983. Classification of 500 mb height anomalies using obliquely rotated principal compo-
nents. J. Clim. Appl. Met., 22: 1975–88.
Comrie, A.C. 1992. An enhanced synoptic climatology of ozone using sequencing techniques. Phys.
Geog., 13: 53–65.
Comrie, A.C. and Glen, E.K. 1998. Principal components-based regionalization of precipitation
regimes across the southwest United States and northern Mexico, with an application to monsoon
precipitation variability. Clim. Res., 10: 201–15
Comrie, A.C. and Yarnal, B. 1992. Relationships between synoptic-scale atmospheric circulation
and ozone concentration in Pittsburgh, Pennsylvania. Atmos. Environ., 26B: 301–12.
Craddock, J.M. 1956. The representation of the annual temperature variation over central and
northern Europe by a two-term harmonic form. Quart. J. Roy. Met. Soc., 82: 275–88.
Craddock, J.M. and Flintoff, S. 1970. Eigenvector representations of northern hemispheric fields.
Craddock, J.M. and Flood, C.R. 1969. Eigenvectors for representing the 500 mb geopotential surface
over the northern hemisphere. Quart. J. Roy. Met. Soc., 95: 576–93.
Crane, R.G. and Barry, R.G. 1988. Comparison of the MSL synoptic pressure patterns of the Arctic as
observed and simulated by the GISS general circulation model. Meteorol. Atmos. Phys., 39: 169–83.
Crane, R.G. and Hewitson, B.C. 1998. Doubled CO2 precipitation changes for the Susquehanna
basin: down-scaling from the Genesis general circulation model. Intl. J. Climatol., 18: 65–76.
Davidova, N.G. 1967. Types of synoptic process and associated wind fields in oceanic regions of
the southern hemisphere. In: Polar Meteorology. Tech. Note No. 87 (WMO, no. 211, TP 111),
World Meteorological Organization, Geneva, pp. 263–91.
Davies, T.D., Dorling, S.R., Pierce, C.E., Barthelmie, R.J., and Farmer, G. 1991. The meteoro-
logical control on the anthropogenic ion content of precipitation at three sites in the UK: the
utility of Lamb types. Intl. J. Climatol., 11 (7): 795–807.
Davies, T.D., Farmer, G., and Barthelmie, R.J. 1990. Use of simple daily atmospheric circulation
types for the interpretation of precipitation composition at a site (Eskdalemuir) in Scotland,
1978–84. Atmos. Environ., 24: 63–72.
Davis, R.E. and Gay, D.A. 1993. An assessment of air quality variations in the southwest USA
using an upper air synoptic climatology, Intl. J. Climatol., 13: 755–82.
Davis, R.E. and Kalkstein, L.S. 1990. Development of an automated spatial synoptic climatolog-
ical classification. Intl. J. Climatol., 10: 769–94.
Davis, R.E. and Walker, D.R. 1992. An upper-air synoptic climatology of the western United States.
J. Climate, 5 (12): 1449–67.
De la Mothe, P.D. 1968. Middle latitude wavelength variation at 500 mb. Met. Mag., 97: 333–9.
Derbyshire, E. 1971. A synoptic approach to the atmospheric circulation of the Last Glacial
Maximum in southeastern Australia. Palaeogeogr., Palaeoclim., Palaeoecol., 10: 103–24.
Deutscher Wetterdienst. 1994. Grosswetterlagen Europas. Amtsblatt des deutschen Wetterdienstes
47, Offenbach-am-Main.
Dilley, F.B. 1992. The statistical relationship between weather-type frequencies and corn (maize)
yields in southwestern Pennsylvania. Agric. Forest Met., 59 (3–4): 149–64.
11 Dmitriev, A.A. 1994. Izmenchivost’ atmosfernykh protsessov Arktiki i yeyo v dolgosrochnykh prog-
nozakh. (Variability of atmospheric processes in the Arctic and their calculation for long-range
forecasts), Gidrometeoizdat, St Petersburg, 207 pp.
Dorling, S.R. and Davies, T.D. 1995. Extending cluster analysis–synoptic meteorology links to char-
acterise chemical climates at six northwest European monitoring stations. Atmos. Environ., 29
(2): 145–67.
Dorling, S.R., Davies, T.D., and Pierce, C.E. 1992. Cluster analysis: a technique for estimating the
synoptic meteorological controls on air and precipitation chemistry – method and applications.
Atmos. Environ., 26A (14): 2575–82.
Dove, H.W. 1857. Über die Rückfälle der Kälte in Mai. Abh. Königl. Preuss. Akad. Wissenchaften
0 (Berlin), 121–92.
Duquet, R.T. 1963. The January warm spell and associated large-scale circulation changes. Mon.
Wea. Rev., 91: 47–60.
Dzerdzeevski, B.L. 1945. Typification of atmospheric processes over the northern hemisphere as a
11 method of characterizing seasons. Dokl. Gos. Okean. Inst. (Moscow), 42: 24 (in Russian).
Dzerdzeevski, B.L. 1962. Fluctuations of climate and of general circulation of the atmosphere in
extratropical latitudes of the northern hemisphere and some problems of dynamic climatology.
Tellus, 14: 328–36.
Dzerdzeevski, B.L. 1963. Fluctuations of general circulation of the atmosphere and climate in the
twentieth century. In: Changes of Climate, Arid Zone Research 20, UNESCO, Paris, pp. 285–95.
Dzerdzeevski, B.L. 1968. Tsirkulatsionnye mekhanizmy v atmosfere Severnogo Polusharii v XX
0 veke (Circulation mechanisms of the atmosphere in the northern hemisphere in the twentieth
century). Results of Meteorological Investigations. IGY Committee. Inst. of Geography, Akad.
Nauk SSSR, Moscow, 220 pp. (in Russian).
Dzerdzeevski, B.L. 1970. Calendar of changes of ECM for 1967, 1968, 1969. In: A.S. Caplygina,
ed., The comparison of the characteristics of the atmospheric circulation over the northern hemi-
sphere with the analogous characteristics for its sectors: circulation of the atmosphere. Results
of Meteorological Investigations, IGY Committee. Inst. of Geography, Akad. Nauk SSSR,
Moscow, 19–22 pp. (in Russian).
Dzerdzeevski, B.L. and Monin, A.S. 1954. Typical schemes of the general circulation of the atmos-
phere in the northern hemisphere and circulation indices. Izv. Akad. Nauk. SSSR, Geofiz. Ser., 6:
562–74 (in Russian).
0 Dzerdzeevski, B.L., Kurgansakaia, V.M. and Vitvitskaia, Z.M., 1946. Classification of circulation
mechanisms over the northern hemisphere and characteristics of synoptic seasons. Trudy Nauk.
Issled. Uchrezden (Glav. Uprav. Gidromet. Sluzhby, Moscow) ser. 2 (21), 80 pp. (in Russian).
El-Kadi, A.K.A. and Smithson, P.A. 1992. Atmospheric classifications and synoptic climatology.
Progr. Phys. Geog., 16: 432–55.
El-Kadi, A.K.A. and Smithson, P.A. 1996. An automated classification of pressure patterns over
the British Isles, Trans. Inst. Brit. Geog., n.s., 21, pp. 141–561.
Ellis, A.W. and Leathers, D.J. 1996. A synoptic climatic approach to the analysis of lake-effect
snowfall: potential forecasting applications. Weather and Forecasting, 11: 216–29.
Escourrou, G. 1978. “Climate et types de temps en Normandie.” Ph.D. thesis, University of Paris,
Librairie Champion, Paris, 3 volumes, 1,730 pp.
0 Ezcurra, A., Casado, H., Lacause, J.P., and Garcia, C. 1988. Relationships between meteorological
situations and acid rain in the Spanish Basque country. Atmos. Env., 22: 2779–86.
Faulkner, R. and Perry, A.H. 1974. A synoptic precipitation climatology of South Wales. Cambria,
1–2: 127–38.
Fechner, H. 1977. Representation of weather situations above a certain geographical point and for
the whole hemisphere by EOFs. European Centre for Medium Range Weather Forecasts.
Workshop in Meteorology. Reading, UK, pp. 50–83.
Federov, E.E. 1927. Climate as totality of weather (trans. E.S. Nichols). Mon. Wea. Rev., 55: 401–3.
Fitzharris, B. 1998. A synoptic climatology of glacier mass balance for the European Alps.
Proceedings of the Conference on Climate and Environmental Change (Evora, Portugal), IGU
Commission on Climatology, pp. 141–2.
0 Fitzharris, B.B. and Bakkehøi, S. 1986. A synoptic climatology of major avalanche winters in
Norway. J. Climatol., 6 (4): 431–45.
Fliri, F. 1965. Über Signifikanzen synoptisch-klimatologischer Mittelwerte in verschiedenen Alpinen
11 Wetterlagensystemen. Carinthia II (Vienna), 24: 36–48.
1 Fliri, F. and Schüepp, M. 1984. Synoptische Klimatographie der Alpen zwischen Mont Blanc und
Hohen Tauern (Schweiz–Tirol–Oberitalien). Wiss. Alpenvereinshefte 29, Innsbruck, 686 pp.
Flohn, H. and Hess, P. 1949. Grosswettersingularitäten in jährlichen Witterungsverlauf
Mitteleuropas. Met. Rund., 2: 258–63.
Flohn, H. and Huttary, J. 1950. Über die Bedeutung des V-b Lagen für das Niederschlagsregime
Mitteleuropas. Met. Rund., 3: 167–70.
Fraedrich, K. 1990. European Grosswetter during warm and cold extremes of the El Niño/Southern
Oscillation. Intl. J. Climatol., 10: 21–31.
Frakes, B. 1998. “A Synoptic Climatology of Short- and Long-term Relationships between the
Atmospheric Circulation and River Flow in the West Branch of the Susquehanna River Basin,
Pennsylvania.” Ph.D. dissertation, Department of Geography, Pennsylvania State University,
University Park PA, 235 pp.
Frakes, B. and Yarnal, B. 1997a. Using synoptic climatology to define representative discharge
events. Intl. J. Climatol., 17: 323–41.
Frakes, B. and Yarnal, B. 1997b. A procedure for blending manual and correlation-based synoptic
classifications. Intl. J. Climatol., 17 (13): 1381–96.
Gazzola, A. 1969. First results of an investigation of precipitation in Italy in relation to the mete-
orological situation. Riv. Met. Aeronaut. (Rome) 29, Suppl. to No. 4: 84–114 (in Italian).
Gerstengarbe, F.-W. and Werner, P.C. 1993. Katalog der Grosswetterlagen Europas nach Paul
Hess und Helmuth Brezowski 1881–1992. Ber. Dtsch. Wetterdienstes, no.113, Offenbach-am-
Main, 249 pp.
Giorgi, F. and Mearns, L.O. 1991. Approaches to the simulation of regional climate change: a
review. Rev. Geophys, 29: 191–216.
Girs, A.A. 1948. On the problem of investigation of the main types of atmospheric circulation. Met.
i Gidrol. (Moscow) No. 3 (in Russian).
Girs, A.A. 1960. Principles of Long-range Weather Forecasting. Gidrometeoizdat, Leningrad, 560 pp.
Girs, A.A. 1974. Makrotsirkulatsionnie metod dolgosrochnikh meteorologicheskikh prognoz (Large-
scale circulation methods of long-range weather forecasting). Gidrometeoizdat. Leningrad, 488 pp.
Girs, A.A. 1981. K voprosy o formakh atmosfernoi tsirkulyatsii i ikh prognosticheskom ispol’zo-
vanie. (On the question of the forms of atmospheric circulation and their use in prediction). Trudy
Arkt. Antarkt. Nauch.-Issled. Inst., 373: 4–13.
Glaser, R. and Walsh, R. (eds) 1991. Historische Klimatologie in verschiedenen Klimazonen,
Würzburger Geogr. Arbeiten 80, 251 pp.
Goodess, C.M. and Palutikof, J.P. 1998. Development of daily rainfall scenarios for southeast Spain
using a circulation-type approach to downscaling. Intl. J. Climatol., 10: 1051–83.
Green, M.C., Flocchini, R.G. and Myrup, L.O. 1993. Use of temporal principal components analysis
to determine seasonal periods. J. Appl. Met., 32 (5): 986–95.
Grutter, M. 1966. Die bemerkenswerten Niederschlage der Jahre 1948–64 in der Schweiz. Veröff.
Schweiz. Meteorol. Zentr. Anstalt 3, Zurich, 20 pp.
Guttman, N.B. 1991. January singularity in the Northeast from a statistical viewpoint. J. Appl. Met.,
30 (3): 358–67.
Gutzler, D.S. and Shukla, J. 1984. Analogs in the wintertime 500 mb height field. J. Atmos. Sci.,
41: 177–89.
Hare, F.K. 1958. The quantitative representation of the north polar pressure field. In: R.C. Sutcliffe,
ed., The Polar Atmosphere Symposium, 1. Oxford, Pergamon Press, pp. 137–50.
Hartley, S. and Keable, M.J. 1998. Synoptic associations of winter climate and snowfall variability
in New England 1950–92. Intl. J. Climatol., 18: 281–98.
Hartranft, F.R., Restivo, J.S., and Sabin, R.C. 1970. Computerized Map Typing Procedures and
their Application in the Development of Forecasting Aids. Tech. Paper 70–2, HQ 14th Weather
Wing, Ent Air Force Base CO, 57 pp.
Hay, L.E. and Fitzharris, B.B. 1988. The synoptic climatology of ablation on a New Zealand glacier.
J. Climatol., 8: 201–15.
Hay, L.E., McCabe, G.J. Jr, Wolock, D.M., and Ayers, M.A. 1992. Use of weather types to dis-
aggregate general circulation model predictions. J. Geophys. Res., 97 (D3): 2781–90.
Hess, P. and Brezowsky, H. 1977. Katalog der Grosswetterlagen Europas (1981–1976). Ber. Dtsch.
Wetterdienstes, 15 (113), Offenbach-am-Main, 249 pp.
Hewitson, B.C. and Crane, R.G. 1992a. Large-scale atmospheric controls on local precipitation in
tropical Mexico. Geophys. Res. Lett., 19 (18): 1835–8.
11 Hewitson, B.C. and Crane, R.G. 1992b. Regional-scale climate prediction from the GISS GCM.
Global Planet. Change, 97: 249–67.
Hewitson, B.C. and Crane, R.G. (eds). 1994. Neural Nets: Applications in Geography. Kluwer,
Norwell MA, 208 pp.
Hewitson, B.C. and Crane, R.G. 1996. Climate downscaling: techniques and application. Climate
Res., 7: 85–95.
Hoinkes, H.C. 1968. Glacier variations and weather. J. Glac., 7: 3–19.
Holawe, F. and Nagl, H. 1997. Wetterlagen als Basis klimageographischer Forschung. Wetter u.
Leben, 49: 177–201.
Horel, D.J. and Mechoso, C.R. 1988. Observed and simulated interseasonal variability of winter-
0 time planetary circulation. J. Climate, 1: 582–99.
Hsieh, W.W. and Tang, B. 1998. Applying neural network models to prediction and data analysis
in meteorology and oceanography. Bull. Amer. Met. Soc., 79 (9): 1855–70.
Hughes, G.M. 1980. Summer rainfall totals related to the P.S.C.M. indices. Weather, 35: 230–3.
11 Hughes, J.P. and Guttrop, P. 1994. A class of stochastic models for relating synoptic atmospheric
patterns to regional hydrologic phenomena. Water. Resour. Res., 30: 1535–46.
Hulme, M. and Barrow, E. 1997. Climates of the British Isles. Routledge, London, 454 pp.
Hulme, M., Briffa, K.R., Jones, P.D. and Senior, C.A. 1993. Validation of GCM control simula-
tions using indices of daily airflow types over the British Isles. Climate Dynam., 9: 95–105.
Huth, R. 1993. An example of using obliquely rotated principal components to detect circulation
types over Europe. Met. Zeitschr., NF2: 285–93.
0
Huth, R. 1996a. Intercomparison of computer-related circulation classification methods. Intl. J.
Climatol., 16: 893–922.
Huth, R. 1996b. Properties of the circulation classification scheme based on the rotated principal
component analysis. Meterol. Atmos. Phys., 59: 217–33.
Inwards, R.L. 1950. Weather Lore. 4th edition. Rider, London 251 pp.
Jacobheit, J., Wanner, H., Koslowski, G., and Gudd, M. 1999. European surface pressure patterns
for months with outstanding climatic anomalies during the sixteenth century. Climatic Change,
43: 210–21.
Jacobs, W.C. 1947. Wartime Developments in Applied Climatology. Met. Monogr. Amer. Met. Soc.,
Boston MA, 1 (1), 52 pp.
0 Jamason, P.F., Kalkstein, L.S., and Gergen, P.J. 1997. A synoptic evaluation of asthma hospital
admissions in New York city. Amer. J. Respir. Crit. Care Med., 156: 1781–8.
James, R.W. 1970. Air mass climatology. Met. Rund. 23: 65–70.
Jasper, J.D. and Stern, H. 1983. Objective classification of synoptic pressure patterns over S.E.
Australia. In: Second International Conference on Statistical Climatology, Nat. Inst. Met.
Geophys., Lisbon, Portugal, pp. 31–5.
Jenkinson, A.F. and Collison, F.P. 1977. An Initial Climatology of Gales over the North Sea. Syn.
Climatology. Branch Mem. 62, Meteorological Office, Bracknell, UK.
Joliffe, I.T. 1987. Rotation of principal components: some comments. J. Climate, 7 (5): 507–10.
Jones, P.D. and Kelly, P.M. 1982. Principal component analysis of the Lamb catalogue of daily
weather types. 1. Annual frequencies. J. Climatol., 2: 147–57.
0 Jones, P.D., Hulme, M., and Briffa, K.R. 1993. A comparison of Lamb circulation types with an
objective classification scheme. Intl. J. Climatol., 13 (6): 655–63.
Kahlig, P. 1989. Grosswetterlagen und Artificial Intelligence. Wetter u. Leben, 41: 109–16.
Kalkstein, L.J., Jamason, P.F., Greens, J.S., Libby, J., and Robinson, L. 1996. The Philadelphia hot
weather – health watch/warning system. Bull. Amer. Met. Soc. 77 (7): 1519–28.
Kalkstein, L.S. 1991. A new approach to evaluate the impact of climate on human mortality. Environ.
Health Perspectives, 96: 145–50.
Kalkstein, L.S. and Corrigan, P. 1986. A synoptic climatological approach for geographical analysis:
assessment of sulfur dioxide concentrations. Ann. Assoc. Amer. Geogr., 76: 381–95.
Kalkstein, L.S. and Greene, J.S. 1997. An evaluation of climate/mortality relationships in large U.S.
cities and the possible impacts of climate change. Environ. Health Perspectives, 105: 84–93.
0 Kalkstein, L.S., Nichols, M.C., Barthel, C.D., and Greene, J.S. 1996. A new spatial synoptic clas-
sification: application to air-mass analysis. Intl. J. Climatol., 16 (9): 983–1004.
Kalkstein, L.S., Sheridan, S.C., and Graybeal, D.Y. 1998. A determination of character and frequency
11 changes in air masses using a spatial synoptic classification. Intl. J. Climatol., 18 (11): 1223–36.
1 Kalkstein, L.S., Tan, G., and Skindlov, J.A. 1987. An evaluation of three clustering procedures for
use in synoptic climatological classification. J. Clim. Appl. Met., 25: 717–30.
Kalnay, E. et al., 1996. The NCEP/NCAR 40-year reanalysis project. Bull. Amer. Met. Soc., 77
(3): 437–71.
Kalnicky, R.A. 1987. Seasons, singularities and climatic changes over the midlatitudes of the
Northern hemisphere during 1899–1969. J. Clim. Appl. Met., 26 (11): 1496–1510.
Kaufmann, R.K., Snell, S.E., Gopal, S. and Dezzani, R. 1999. The significance of synoptic patterns
identified by the Kirchhofer technique: a Monte Carlo approach. Intl. J. Climatol., 19: 619–26.
Keables, M.J. 1988. Spatial associations of mid-tropospheric circulation and upper Mississippi River
basin hydrology, Ann. Assoc. Am. Geogr., 78: 74–92.
Kelly, P.M., Jones, P., and Briffa, K. 1997. Classifying the winds and weather. In M. Hulme and
E. Barrow, eds, Climates of the British Isles: Present, Past and Future. Routledge, London and
New York, pp.153–72.
Kerschner, H. 1989. Beiträge zür synoptischen Klimatologie der Alpen zwischen Innsbruck und dem
Alpenstrand. Innsbrucker Geogr. Studien, 17, 253 pp.
Key, J. and Crane, R.G. 1986. A comparison of synoptic classification schemes based on “objec-
tive” procedures. J. Climatol., 6: 375–88.
Key, J., Maslanik, J.A. and Schweiger, A.J. 1989. Classification of merged AVHRR and SMMR
Arctic data with neural networks. Photog. Eng. Remote Sens., 55: 1331–8.
Kidson, J.W. 1994a. An automated procedure for the identification of synoptic types applied to the
New Zealand region. Intl. J. Climatol., 14: 711–21.
Kidson, J.W. 1994b. Relationships of New Zealand daily and monthly weather patterns to synoptic
types. Intl. J. Climatol., 14: 723–37.
Kidson, J.W. 1995. A synoptic climatological evaluation of the changes in the CSIRO nine-level
model with doubled CO2 in the New Zealand region. Intl. J. Climatol., 15: 1179–94.
Kilsby, C.G., Cowperthwait, P.S.P., O’Connell, P.E., and Jones, P.D. 1998. Predicting rainfall statis-
tics in England and Wales using atmospheric circulation variables. Intl. J. Climatol., 18: 523–39.
Kirchhofer, W. 1974. Classification of European 500 mb Patterns. Schweiz. Met. Zentralanstalt,
Arbeitsbericht 43, Zürich, 16 pp.
Kirchhofer, W. 1976. Stationsbezogene Wetterlagen Klassifikation. Veröff. Schweiz. Met.
Zentralanstalt, 34, Zürich, 50 pp.
Kirchhofer, W. 1995. Klimaatlas der Schweiz (Part 5). Bundesamt für Landestographie, Waber-
Bern; map sheets, 2.19–2.14.
Klaus, D. 1993. Zirkulations- und Persistenzänderungen der Europaischen Wettergeschehens in
Spiegel der Grosswetterlagenstatistik. Erdkunde, 47: 85–104.
Kletter, L. 1962. Die Aufeinanderfolge charakteristische Zirkulationstypen in mittleren Breiten der
nordlichen Hemisphäre. Arch. Met. Geophys. Biokl., AB: 1–33.
Kohenen, T. 1995. Self-Organizing Maps. Springer-Verlag, Berlin,. 362 pp.
Kozuchowski, K. and Marciniak, K. 1988. Variability of mean monthly temperatures and semi-
annual precipitation totals in Europe in relation to hemispheric circulation patterns. J. Climatol.,
8: 191–9.
Kruizinga, S. 1979. Objective classification of daily 500 mb patterns. Sixth Conference on
Probability and Statistics in Atmospheric Science, Amer. Met. Soc., Boston MA: 126–9.
Ladd, J.W. and Driscoll, D.M. 1980. A comparison of objective and subjective means of weather
typing – an example from west Texas. J. Appl. Met., 19: 691–704.
Lamb, H.H. 1950. Types and spells of weather around the year in the British Isles: annual trends,
seasonal structure of the year, singularities. Quart. J. Roy. Met. Soc., 76: 393–429.
Lamb, H.H. 1955. Two-way relationships between the snow or ice-limit and 1000–500 mb thick-
ness in the overlying atmosphere. Quart. J. Roy. Met. Soc., 81: 172–89.
Lamb, H.H. 1964. The English Climate. 2nd edition. English Universities Press, London, 212 pp.
Lamb, H.H. 1965. Frequency of weather types. Weather, 20: 9–12.
Lamb, H.H. 1972. British Isles Weather Types and a Register of the Daily Sequence of Circulation
Patterns, 1861–1971. Geophys. Mem. 116, HMSO, London, 85 pp.
Lamb, H.H. 1994. British Isles daily wind and weather patterns 1588, 1781–86, 1972–91 and shorter
early sequences. Climate Monitor, 20: 47–71.
Lambert, S.J. 1990. Discontinuities in the long-term northern hemisphere 500–millibar heights data
set. J. Climate, 3: 1479–84.
Lana, X. and Fernandez-Mills, G. 1994. Minimum sample size for synoptic weather type classifi-
11 cation: applications to winter period data recorded on the Catalan coast (N.E. Spain). Intl. J.
Climatol., 14: 1051–60.
Lanzante, J.R. 1983. Some singularities and irregularities in the seasonal progression of the 700
mb height field. J. Clim. Appl. Met., 22 (6): 967–81.
Lauscher, F. 1958. Studien zur Wetterlagen-Klimatologie der Ostalpenländer. Wetter u. Leben, 10:
79–83.
Lauscher, F. 1972. 25 Jahre mit täglicher Klassifikation der Wetterlage in den Ostalpenländern.
Wetter u. Leben, 24: 185–9.
Lauscher, F. 1985. Klimatologische Synoptik Österreichs mittels der Ostalpinen Wetterlagen-
klassifikation. Arbeiten, Zentralanstalt f. Meteorologie u. Geodynamik. 302, Vienna, 65 pp.
0 Lawrence, E.N. 1970. Variation in weather-type temperature averages. Nature, 226: 633–4.
Lawrence, E.N. 1971. Synoptic-type rainfall averages over England and Wales. Met Mag., 100:
333–9.
Lehmann, A. 1911. Altweibersommer: die Wärmerückfälle des Herbstes in Mitteleuropa. Landwirt-
11 schaftliche Jährbücher, Zeit. f. wiss. Landwirtschaft (Berlin) 41: 57–129.
Levick, R.B.M. 1949. Fifty years of English weather. Weather, 4: 206–11.
Levick, R.B.M. 1975. British Isles – weather types. Weather, 30: 342–6.
Litynski, J.K. 1970. Classification numérique des types de circulation et des types de temps en
Pologne. Cah. Géogr. Québec, 33: 329–38.
Lorenz, E.N. 1969. Atmospheric predictability as revealed by naturally occurring analogues.
J. Atmos. Sci., 26: 636–46.
0 Lund, I.A. 1963. Map pattern classification by statistical methods. J. Appl. Met., 2: 56–65.
Lydolph, P.E. 1959. Federov’s complex method in climatology. Ann. Assoc. Amer. Geogr., 49:
120–44.
Maddox, R.A., Canova, F., and Hoxit, R.L. 1980. Meteorological characteristics of flash flood events
over the western United States. Mon. Wea. Rev., 108 (11): 1866–77.
Maejima, I. 1967. Natural seasons and weather singularities in Japan. Geogr. Rep., 2 (Tokyo:
Metropolitan Univ.), 77–103.
Maheras, P. and Kutiel, H. 1999. Spatial and temporal variations in the temperature regime in the
Mediterranean and their relationship with circulation during the last century. Intl. J. Climatol.,
19 (7): 745–64.
Marvin, C.F. 1919. Normal temperatures (daily): are irregularities in the annual march of temper-
0 ature persistent? Mon. Wea. Rev., 47: 544–55.
Masterman, A.J., Foster, G.N., Holmes, S.J., and Harrington, R. 1996. The use of the Lamb daily
weather types and indices of progressiveness, southerliness and cyclonicity to investigate the
autumn migration of Rhopalosiphum padi. J. Appl. Ecol., 33: 23–30.
Mayes, J.C. 1991. Regional airflow patterns in the British Isles. Intl. J. Climatol., 11: 473–91.
Mayes, J.C. 1994. Recent changes in the monthly distribution of regional weather types in the British
Isles. Weather, 49: 156–62. (Also discussion, by J. Kington and J.C. Mayes, ibid., pp. 494–5.)
Mayes, J. 1996. Spatial and temporal fluctuations in monthly rainfall in the British Isles and vari-
ations in mid-latitude westerly circulation. Intl. J. Climatol., 16 (5): 585–96.
Mayes, J. and Wheeler, D. 1997. The anatomy of regional climates in the British Isles. In: D.
Wheeler and J. Mayes, eds, Regional Climates of the British Isles, Routledge, London, pp. 9–44.
0 McCutchan, M.H. 1978. A model for predicting synoptic weather types based on model output
statistics. J. Appl. Met., 7: 1466–75.
McCutchan, M. and Schroeder, M.J. 1973. Classification of meteorological patterns in southern
California by discriminant analysis. J. Appl. Met., 12: 571–2.
McGinnis, D.L. 1997. Estimating climate-change impacts on Colorado Plateau snowpack using
downscaling methods. Prof. Geogr., 49: 117–25.
McGregor, G. 1998. Killer weather types in Birmingham. Proceedings of the Conference on Climate
and Environmental Change, Evora, Portugal, IGU Commission on Climatology, pp. 173–4.
McGregor, G.R. and Bamzelis, D. 1995. Synoptic typing and its application to the investigation of
weather air pollution relationships in Birmingham, U.K. Theor. Appl. Climatol, 51: 223–36.
McHendry, J.G., Steyn, D., and McBean, G. 1995. Validation of synoptic circulation patterns simu-
0 lated by the Canadian Climate Center general circulation model for western North America.
Atmos.-Ocean, 33 (4): 809–25.
Mock, C.M., Bartlein, P., and Anderson, P.M. 1998. Atmospheric circulation patterns and spatial
11 climatic variations in Beringia. Intl. J. Climatol., 18 (10): 1085–104.
1 Moses, T., Kiladis, G.N., Diaz, H.F., and Barry, R.G. 1987. Characteristics and frequency of rever-
Mosino, P.A. 1964. Surface weather and upper-air flow patterns in Mexico. Geofis. Internac., 4:
117–68.
Mote, T.L. 1998. Mid-tropospheric circulation and surface melt on the Greenland ice sheet. 2.
Synoptic climatology. Intl. J. Climatol, 18: 133–45.
Muller, R.A. 1977. A synoptic climatology for environmental baseline analysis: New Orleans.
J. Appl. Met., 16: 20–33.
Muller, R.A. and Wax, C.L. 1977. A comparative synoptic climate baseline for coastal Louisana.
Geosci. and Man, 18: 121–9.
Murray, R. 1972. Monthly mean temperatures related to synoptic types over Britain specified by
PSCM indices. Met. Mag., 101: 305–11.
Murray, R. and Lewis, R.P.W. 1966. Some aspects of the synoptic climatology of the British Isles
as measured by simple indices. Met. Mag., 95: 193–203.
Namias, J. 1957. Characteristics of cold winters and warm summers over Scandinavia in relation
to the general circulation. J. Met. 14, 235–50.
Neale, S.M. and Fitzharris, B.B. 1997. Energy balance and synoptic climatology of a melting snow-
pack in the southeast Alps, New Zealand. Intl. J. Climatol., 17 (14): 1595–609.
O’Hare, G. and Sweeney, J. 1993. Lamb’s circulation types and British weather: an evaluation.
Geography, 78: 43–60.
O’Hare, G.P. and Wilby, R. 1995. A review of ozone pollution in the U.K. and Ireland with an
analysis using Lamb weather types. Geog. J., 161: 1–20.
Paegle, J.N. and Kierulff, L.P. 1974. Synoptic climatology of 500–mb winter flow types. J. Appl.
Met., 13: 205–12.
Palutikof, J., Holly, P.M., Davies, T.D., and Halliday, J.A. 1987. Impacts of spatial and temporal
windspeed variability on wind energy output. J. Clim. Appl. Meteorol., 26 (9): 1124–32.
Pankiewicz, G.S. 1995. Pattern recognition techniques for the identification of cloud and cloud
systems. Met. Appl., 2: 257–71.
Peczely, C. 1961. Die klimatologische Charakterisierung der makrosynoptischen Lagen Ungarns.
As Orszagos Meteorol. Intezet, Budapest, 128 pp. (In Hungarian, German abstract pp. 36–49).
Pepin, N. 1995. The use of GCM scenario output to model effects of future climate change on the
thermal climate of marginal maritime uplands. Geogr. Annal., 77A: 167–84.
Perret, R. 1987. Une classification des situations méteorologiques à l’usage de la prévision. Veröff.
Schweiz. Met. Anst. 46, Zürich, 127 pp.
Perry, A.H. 1970. Changes in duration and frequency of synoptic types over the British Isles.
Weather, 25: 123–6.
Perry, A.H. 1983. Growth points in synoptic climatology. Progr. Phys. Geog., 7: 90–6.
Perry, A.H. and Barry, R.G. 1973. Recent temperature changes due to changes in the frequency
and average temperature of weather types over the British Isles. Met. Mag, 102: 73–82.
Perry, A.H. and Mayes, J. 1998. The Lamb weather type catalogue. Weather, 53 (7): 222–9.
Petzold, D.E. 1982. The Summer Weather Types of Quebec-Labrador. McGill Subarctic Res. Pap.
34, McGill University, Montreal, 160 pp.
Pielke, R.A., Garstang, M., Lindsey, C., and Gusdorf, J. 1987. Use of a synoptic classification
scheme to define seasons. Theoret. Appl. Climatol., 38: 57–68.
Pilgram, A. 1788. Untersuchungen über das Wahrscheinliche der Wetterkunde durch vieljährige
Beobachtungen. J. Edlen von Kurzbeck, Vienna.
Pope, C.A., III, and Kalkstein, L.S. 1996. Synoptic weather modeling and estimates of the expo-
sure – response relationship between daily mortality and particulate air pollution. Environ. Health
Perspectives, 104: 414–20.
Putnins, P. 1966. The Sequence of Baric Pressure Patterns over Alaska, Studies on the Meteorology
of Alaska, First Interim Report, Environmental Data Service, ESSA, Washington DC, 81 pp.
Reitenbakh, R.G. and Kozulin, K.N. 1982. Objective classification of ground-level pressure fields
and distinction of climatic seasons in the northern hemisphere. Soviet Met. Hydrol., 1982, no. 8:
13–20.
Rex, D.F. 1950. Blocking action in the middle troposphere and its effects upon regional climate.
Tellus, 2: 275–301.
11 Richman, M.B. 1981. Obliquely rotated principal components: an improved meteorological map
typing technique. J. Appl. Met., 20: 1145–59.
Richman, M.B. 1986. Rotation of principal components. Intl. J. Climatol., 6: 293–335.
Richman, M.B. 1987. Rotation of principal components: a reply. Intl. J. Climatol., 7 (5): 511–20.
Romero, R., Ramis, C., and Guijarro, J.A. 1999a. Daily rainfall patterns in the Spanish Mediterranean
area: an objective classification. Intl. J. Climatol., 19: 95–112.
Romero, R., Sumner, G., Ramis, C., and Genoves, A. 1999b. A classification of the atmospheric
circulation patterns producing significant daily rainfall in the Spanish Mediterranean area. Intl.
J. Climatol., 19 (7): 765–85.
Ruosteenoja, K. 1988. Factors affecting the occurrence and lifetime of 500 mb height analogues:
0 a study based on a large amount of data. Mon. Wea. Rev., 116: 368–76.
Saunders, I.R. and Byrne, J.M. 1999. Using synoptic surface and geopotential height fields for
generating grid-scale precipitation. Intl. J. Climatol., 19 (11): 1165–76.
Schiesser, H.H., Pfister, C., and Bader, J. 1997. Winter storms in Switzerland north of the Alps
11 1864/1865–1993/1994. Theor. Appl. Climatol., 58: 1–19.
Schmauss, A. 1928. Singularitäten in jährlichen Witterungsverlauf von München. Dtsch. Met.
Jahrbuch (Bavaria) 50, 22 pp.
Schmauss, A. 1938. Synoptische Singularitäten. Met. Zeit., 55: 384–403.
Schmauss, A. 1941. Kalendermässige Bindungen des Wetters (Singularitäten). Z. angew. Met., 58:
237–44, 373–6.
Schmutz, C. and Wanner, H. 1998. Low frequency variability of atmospheric variability over Europe
0
between 1785 and 1994. Erdkunde, 52: 81–94.
Schubert, S. 1994. A weather generator based on European Grosswetterlagen. Climate Res., 4:
191–202.
Schüepp, M. 1957. Klassifikationschema. Beispiele und Probleme der Alpenwetterstatistik.
Meteorologie, ser. 4, 45–6: 291–9.
Schüepp, M. 1959. Die Klassifikation der Witterungslagen. Geofis. Pura e Appl., 44: 242–8.
Schüepp, M. 1968. Kalender der Wetter-und Witterungslagen von 1955 bis 1967 im zentralen
Alpengebiet. Veröffentlichungen 11, Met. Zentralanstalt, Zürich, 43 pp.
Schüepp, M. 1979. Witterungsklimatologie. In: Klimatologie der Schweiz, 3, Meteorologischen
Anstalt, Zürich, 94 pp.
0 Schwartz, M.D. 1991. An integrated approach to air mass classification in the north central United
States. Prof. Geogr., 43: 77–91.
Smithson, P.A. 1988. Synoptic and dynamic climatology. Progr. Phys. Geog., 11: 121–32.
Sowden, I.P. and Parker, D.E. 1981. A study of variability of daily central England temperatures
in relation to the Lamb synoptic types. J. Climatol. 1: 3–10.
Stefanicki, G., Talkner, P., and Weber, R.O. 1998. Frequency changes of weather types in the
Alpine region since 1945. Theor. Appl. Climatol., 60: 47–61.
Stohl, A. and Scheifinger, H. 1994. A weather pattern classification by trajectory clustering. Met.
Zeit., NF 6: 333–6.
Stone, J. 1983. Circulation types and the spatial distribution of precipitation over central, eastern
and southern England. Weather, 39: 173–7, 200–4.
0 Stone, R.G. 1989. Weather types at Brisbane, Queensland: an example of the use of principal
components and cluster analysis. Intl. J. Climatol., 9: 3–32.
Storey, A.M. 1982. A study of the relationship between isobaric patterns over the UK and central
England temperatures and England and Wales rainfall. Weather, 37: 2–11, 46, 88, 122, 151, 170,
208, 244, 260, 294, 327, and 360.
Sturman, A.P. 1979. Aspects of the synoptic climatology of southern South America and the
Antarctic peninsula. Weather, 34: 210–23.
Sturman, A.P., Trewinnard, A.C., and Gorman, P.A. 1984. A study of atmospheric circulation over
the South Island of New Zealand, 1961–1980. Weather and Climate, 4: 53–62.
Sumner, G. 1996. Daily precipitation patterns over Wales: towards a detailed precipitation clima-
tology. Trans. Inst. Brit. Geogr., n.s. 21: 157–76.
0 Sumner, G. and Bonnell, M. 1986. Circulation and daily rainfall in the north Queensland wet seasons
1979–1982. Intl. J. Climatol., 6: 531–49.
Sumner, G., Ramis, C. and Guijarro, J.A. 1993. The spatial organization of daily rainfall over
11 Mallorca. Intl. J. Climatol., 13 (1): 89–109.
1 Sweeney, J.C. 1985. The changing synoptic origins of Irish precipitation. Trans. Inst. Brit. Geog.,
n.s. 10: 467–80.
Sweeney, J.C. and O’Hare, G.P. 1992. Geographical variations in precipitation yields and circula-
tion types in Britain and Ireland. Trans. Inst. Brit. Geogr., n.s. 17: 448–63.
Swiss Meteorological Institute. 1985. Alpenwetterstatisk Witterungskalender, Internal Report,
Schweiz. Met. Anstalt, Zürich, 26 pp.
Talman, C.F. 1919. Literature concerning supposed recurrent irregularities in the annual march of
temperature. Mon. Wea. Rev., 47: 555–65.
Todhunter, P.E., 1989. An approach to the variability of surface energy budgets under stratified
synoptic weather types. Intl. J. Climatol., 9: 191–201.
Toth, Z. 1991a. Intercomparison of circulation similarity measures. Mon. Wea. Rev., 199: 55–64.
Toth, Z. 1991b. Estimation of atmospheric predictability by circulation analogs. Mon. Wea. Rev.,
119: 65–72.
Toth, Z. 1991c. Circulation patterns in phase space: a multinormal distribution? Mon. Wea. Rev.,
119: 1501–11.
Trenberth, K.E. and Olsen, J.G. 1988. An evaluation and intercomparison of global analyses from
the National Meteorological Center and the European Centre for Medium Range Weather
Forecasts. Bull. Amer. Met. Soc., 69: 1047–57.
Tsui, H.T. and Hay, K.P. 1979. On the problems of classification of weather contour maps. Proceed-
ings, Fourth International Joint Conference on Pattern Recognition, IEEE, New York, pp. 635–7.
Urbani, M. 1961. Una classificazione di tipi di tempo sull Europa e sul Mediterraneo. Dynamic
Climatology 1, Min. di Feesa Aeronaut. Publi., 75 pp.
Van Bebber, W.J. and Köppen, W. 1895. Die Isobarentypen des Nordatlantischen Ozeans und
Westeuropas, ihre Beziehung zur Lage und Bewegung der barometrischen Maxima und Minima.
Arch. Dtsch. Seewarte, 18 (4), Hamburg, 27 pp.
van den Dool, H.M. 1989. A new look at weather forecasting through analogs. Mon. Wea. Rev.,
117: 2230–47.
van den Dool, H.M. 1991. Mirror images of atmospheric flows. Mon. Wea. Rev., 119: 2095–106.
van Dijk, W. and Jonker, J.P. 1985. Statistical remarks on European weather types. Arch. Met.
Geophs. Bioklim, B 35: 277–306.
Vangengeim, G. Ya. 1960. O stepeni odnorodnosti atmosfernoi tsirkulatsii razlichnykh chastei
Severnogo Polusharii pri osnovnykh formakh E, W i C (On the degree of uniformity of the
atmospheric circulation in different regions of the northern hemisphere during the basic E, W
and C types of circulation). Trudy Arkt. Antarkt. Inst., 240: 4–23.
Vangengeim, G.I. 1935. Opyit Primeneniya Sinopticheskikh Metodov u Izucheniyu i Kharakteristike
Klimata (Application of Synoptic Methods to the Study and Characterization of Climate),
Gidrometeoizdat, Moscow, 112 pp. (in Russian).
Vangengeim, G.I. 1946. On variations in the atmospheric circulation over the northern hemisphere.
Izv. Akad. Nauk SSSR, Ser. Geogr. i. Geofiz. 5 (in Russian).
Vargas, W.M. and Compagnucci, R.H. 1983. Methodological aspects of principal component
analysis in meteorological fields. Preprints, Second International Conference on Statistical
Climatology, Nat. Inst. Met. Geophys., Lisbon, Portugal, 5.3.1.
von Dijk, W. and Jonker, P.J. 1985. Statistical remarks on European weather types. Arch. Met.
Geophys. Biokl., B35: 277–306.
Vorobieva, E.V. 1967. Characteristics of precipitation for the basic forms of atmospheric circula-
tion. Trud Glav. Geofiz. Obs. (Leningrad), 211: 81–93 (in Russian).
Wada, H. and Kitahra, E. 1971. A proposal for a classification of 500 mb patterns over the northern
hemisphere. J. Met. Soc. Japan, 49: 790–7.
Wahl, E.W. 1952. The January thaw in New England: an example of a weather singularity. Bull.
Amer. Met. Soc., 33, 380–6.
Wahl, E.W. 1953. Singularities and the general circulation. J. Met., 10: 42–5.
Wanner, H. 1980. Grundzüge der Zirkulation der mittleren Breiten und ihre Bedeutung für der
Wetterunganalyse im Alpenraum. In: Das Klima, Springer Verlag, Heidelberg.
Wanner, H., Brazdil, R., Frich, P., Fryendahl, K., Jonsson, T., Kington, J., Pfister, C., Rosenørn,
S., and Wishman, E. 1994. Synoptic interpretation of monthly weather maps for the late Maunder
Minimum (1675–1704). In: B. Frenzel, C. Pfister, and B. Glaeser, eds, Climatic Trends and
Anomalies in Europe 1675–1715, Fischer, Stuttgart, pp. 401–25.
11 Wanner, H., Pfister, C., Brazdil, R., Frich, P., Fryendahl, K., Jonsson, T., Kington, J., Lamb, H.H.,
Rosenørn, S., and Wishman, E. 1995. Wintertime European circulation patterns during the late
Maunder Minimum cooling period (1675–1704). Theor. Appl. Climatol., 51: 167–75.
Wanner, H., Rickli, R., Salvisberg, E., Schmutz, C., and Schüepp, M. 1997. Global climate change
and variability and its influence on Alpine climate – concepts and observations. Theor. Appl.
Climatol., 58: 221–43.
Wanner, H., Salvisberg, E., Rickli, R., and Schüepp, M. 1998. Fifty years of Alpine Weather
Statistics (AWS). Met. Zeit., NF, 7: 99–111.
Weller, G. and Holmgren, B. 1974. The microclimates of the arctic tundra. J. Appl. Met., 13 (8):
854–62.
0 White, D., Richman, M., and Yarnal, B. 1991. Climate regionalization and rotation of principal
components. Intl. J. Climatol., 11: 1–25.
Wigley, J.M.L., Jones, P.D., Briffa, K.R., and Smith, G. 1990. Obtaining sub-grid scale informa-
tion from coarse-resolution general circulation model output. J. Geophys. Res., 95 (D2): 1943–53.
11 Wilby, R.L. 1993. The influence of variable weather patterns on river water quantity and quality.
Wilby, R. L. 1997. Non-stationarity in daily precipitation series: implications for GCM downscaling
using atmospheric indices. Intl. J. Climatol., 17 (4): 439–54.
Wilby, R.L., Barnsley, N., and O’Hare, G. 1995. Rainfall variability associated with Lamb weather
types: the case for incorporating weather fronts. Intl. J. Climatol., 15 (11): 1241–52.
Willmott, C.J. 1987. Synoptic weather-map classification: correlation versus sum-of-square. Prof.
0 Geogr., 30: 205–7.
Wilson, L., Lettenmaier, D.P., and Skyllingstad, E. 1992. A hierarchical stochastic model of large-
scale atmospheric circulation patterns and multiple station daily precipitation. J. Geophys. Res.,
97 (D3): 2791–809.
Wos, A. 1980. An outline of a method of distinguishing the climatic seasons. Geogr. Polonica, 43:
49–59.
Yarham, E.R. 1966. Weather lore of the twelve months. Weather, 21: 433–9.
Yarnal, B. 1984a. Relation between synoptic-scale atmospheric circulation and glacier mass balance
in SW Canada during the IHD. J. Glaciol., 30: 188–98.
Yarnal, B. 1984b. The effects of weather map scale on the results of synoptic climatology.
J. Climatol., 4: 481–93.
0 Yarnal, B. 1984c. Synoptic-scale atmospheric circulation over British Columbia in relation to the
mass balance of Sentinel Glacier. Ann. Assoc. Amer. Geogr., 74: 375–92.
Yarnal, B. 1985. A 500–mb synoptic climatology of Pacific Northwest coast winters in relation to
climatic variability, 1948–1949 to 1977–1978. J. Climatol., 5: 237–52.
Yarnal, B. 1993. Synoptic Climatology in Environmental Analysis: A Primer. Belhaven Press,
London, 195 pp.
Yarnal, B. and Draves, J.D. 1993. A synoptic climatology of stream flow and acidity. Climate Res.,
2: 193–203.
Yarnal, B. and Frakes, B. 1997. Using synoptic climatology to define representative discharge events.
Yarnal, B., and White, D.A. 1987. Subjectivity in a computer-assisted synoptic climatology. I.
0 Classification results. J. Climatol., 7: 119–28.
Yarnal, B., White, D.A., and Leathers, D.J. 1988. Subjectivity in a computer-assessed synoptic
climatology. II. Relationships to surface climate. J. Climatol., 8: 227–39.
Yarnal, B., Crane, R.G., Carleton A.M., and Kalkstein, L.S. 1987. A new challenge for climate
studies in geography. Prof. Geogr., 39: 465–73.
Yasunari, T. 1991. The monsoon year – a new concept of the climatic year in the tropics. Bull.
Amer. Met. Soc., 72 (9): 1331–8.
Yoshino, M.M. 1968. Pressure pattern calendar of east Asia. Met. Rund., 21: 162–9.
Yoshino, M.M. and Kai, K. 1974. Pressure Pattern Calendar of East Asia. 1941–1970, and its
Climatological Summary, Climatol. Notes 16, Inst. of Geoscience, University of Tsukuba, Japan,
71 pp.
0 Zhang, X.-B., Wang, X.-L.L., and Corte-Real, J. 1997. On the relationship between daily circula-
tion patterns and precipitation in Portugal. J. Geophys. Res., 102 (D12): 13495–507.
Zimmer, F. 1941. Der Wert der Baurenregel über den jährlichen Temperaturgang und die
11 Witterungsabschnitte des Jahres. Met. Zeit., 58: 210–19, 464.
1
8 Retrospect and prospect
The foundations of dynamic and synoptic climatology were laid in the nineteenth century
with the advent of weather maps, the study of cyclones, and theories of the general circu-
lation. The great strides that have been made in these areas over the last thirty years can
be attributed to several factors. First, the importance of the increased availability of gridded
hemispheric and global meteorological fields and satellite data cannot be overemphasized.
Until at least the 1970s the absence of such readily accessible data was hampering clima-
tological research, as noted by Barry and Perry (1973, p. 448). The recent advent of the
global reanalysis products and the Pathfinder satellite products is providing a continuing
stimulus. Second is the increasing use of GCM simulations to diagnose large-scale
processes and to treat their statistics as a surrogate for observational data. Third, there
has been recognition of the importance of climatic processes on intraseasonal, interan-
nual, and decadal-centennial time scales, fostered by the World Climate Research Program
(WCRP). The current programmatic elements of the WCRP are: the Global Water and
Energy Experiment (GEWEX), the Arctic Climate System (ACSYS), Climate Variability
(CLIVAR), and Stratospheric Processes and their Role in Climate (SPARC). There is also
a new project for Climate and Cryosphere (CliC). Information about these programs can
be obtained via http://www.wmo.ch. Fourth, the establishment of several new climate
journals in the 1970s and 1980s also contributed to the renaissance of dynamic and
synoptic climatology by creating a wider visibility for publications about climate and for
the related scientific community.
Observational records made possible the detailed investigation of mesoscale and
synoptic weather systems worldwide, and of large-scale recurring circulation anomalies
and teleconnections. Crucial to these advances were the existence of more than a century
of basic meteorological observations and surface weather maps for at least a sector of the
northern hemisphere, near-global surface and upper data since 1945, and global satellite
data available digitally from the 1980s. Longer time series are needed, however, to analyze
the characteristics and mechanisms of multidecadal phenomena. Until such records do
become available, coupled global model simulations of these features, using integrations
performed for several model centuries, are likely to receive increased attention. Areas
where progress may be more limited include the southern hemisphere, especially the
Southern Ocean, owing to the unavoidable sparsity of the surface-based observational
network, and the Arctic Ocean since the discontinuance of the Soviet North Pole Drifting
Stations in 1991. The coverage available for the period 1950–90 in both the Arctic and
across the former Soviet Union, in terms of hydrometeorological data, is unlikely to be
repeated in the foreseeable future. Degradation of surface hydrometeorological networks
is also occurring in other areas, including Canada and parts of Africa. Expanded satellite
capabilities will enhance the data assimilated into operational products and future
reanalysis products, although the continuance of some important long time series will
necessitate data blending and careful homogenization of records collected by different
sensors.
Retrospect and prospect 605
11
The development of new models of mesoscale and synoptic systems has been facili-
tated by special field campaigns involving aircraft, ocean buoys, and augmented surface
networks, as described in Chapter 6. Additionally, the renewed interest in analyses of
isentropic surfaces using potential vorticity, as well as new diagnostic tools such as Q
vectors and Eliassen-Palm flux, has served to sharpen the insight into dynamical processes.
In studies of planetary to global-scale systems, many recent advances have relied on
the use of advanced statistical techniques to identify patterns that, it is assumed, may
have an underlying physical basis. EOF analyses and their variants are particularly
common. The separation of stationary and transient circulation features has been another
0
important topic, and applications of the new technique of wavelet analysis are becoming
common. Empirical analyses frequently go hand in hand with diagnostic analysis of model
output. The four-dimensional assimilation procedures incorporated into operational models
make their fields, in principle, self-consistent and therefore of especial value in data-sparse
11
regions. Nevertheless, considerable care is needed due to the effects of the truncation of
the spectral resolution on smaller-scale features, as a result of Gibbs oscillations, and the
particular errors that arise in areas of mountainous terrain. Coupled GCMs are beginning
to be widely used and there are continuing investigations of the effects of the parame-
terizations used in the various subcomponent models, especially for the land surface.
The field of synoptic climatology has witnessed a renaissance in the 1980s, with wide
0
application of its methods to environmental problems, including human health and the
analysis of climatic change, as well as to the diagnosis of GCM performance. Classification
procedures, which evolved from manual to automated methods based on correlation or
EOFs and clustering routines, are now being further modified to combine the strengths
of both approaches.
The pace of research in dynamic and synoptic climatology shows no sign of slack-
ening, so that any text can inevitably provide only a snapshot of the state of knowledge.
Nevertheless, it is hoped that the present work represents the main strands of our under-
standing of the global climate and its main components at the turn of the century and the
millennium.
0
11
1
Further reading
Inevitably, important new literature has appeared since the main text was completed. Here
we provide references to some of these works and their context in the book.
Chapter 2 Climate data and their analysis

2.1 Synoptic meteorological data
Strangeways, I. 2000. Measuring the Natural Environment. Cambridge University Press, Cambridge,
365 pp.
2.3 Climate variables and their statistical description

von Storch, H. and Zwiers, F.W. 1999. Statistical Analysis in Climate Research. Cambridge
University Press, Cambridge, 484 pp.
2.4 Analytical tools for spatial data

Persson, A. 2000. Back to basics: Coriolis. 1. What is the Coriolis force? 2. The Coriolis force
according to Coriolis. 3. The Coriolis force on the physical earth. Weather, 55: 155–70, 182–7,
234–9.
Phillips, N. 2000. An explication of the Coriolis effect. Bull. Amer. Met. Soc., 81 (2): 299–303.
2.5 Time series

Mann, M.E. and Park, J. 1999. Oscillatory spatiotemporal signal detection in climate studies: a
multiple-taper spectral domain approach. Adv. Geophys., 41: 1–131.
Chapter 3 Global climate and the general circulation

3.1 Planetary controls
Gittelman, A.I., Risbey, S.J., Kass, R.E., and Rosen, R.D. 1999. Sensitivity of a meridional temper-
ature gradient index to latitudinal domain. J. Geophys. Res., 104 (D14): 16709–17.
Hoinka, K.P. 1999. Temperature, humidity, and wind at the global tropopause. Mon. Wea. Rev.,
127 (10): 2248–65.
3.2 Basic controls of the atmospheric circulation

Barry, R.G. and Serreze, M.C. 2000. Atmospheric components of the Arctic Ocean freshwater
balance and their interannual variability. In: E.L. Lewis, ed., The Freshwater Budget of the Arctic
Ocean, Kluwer, Dordrecht, pp. 45–56.
Further reading 607
11
3.3 Circulation cells
Wunsch, C. 2000. Moon, tides and climate, Nature, 405 (5788): 743–4.
3.6 General circulation models

Mote, P. and O’Neill, A. (eds). 2000. Numerical Modeling of the Global Atmosphere in the Climate
System. Kluwer, Dordrecht, 517 pp.
0 3.7.2 Low-latitude circulation

Hastenrath, S. 1999. Dynamics of the equatorial Pacific dry zone. Met. Atmos. Phys., 71: 243–54.
Hastenrath, S. 2000. Zonal circulations over the equatorial Indian Ocean. J. Climate, 13 (15):
2746–56.
11 Robertson, A.W. and Mechoso, C.R. 2000. Interannual and interdecadal variability of the South
Atlantic Convergence Zone. Mon. Wea. Rev., 128 (8): 2947–57.
3.7.3 Monsoons
Lau, K.M., Kim, K.M., and Yang, S. 2000. Dynamical and boundary forcing characteristics of
0 regional components of the Asian summer monsoon. J. Climate, 13 (14): 2461–82.
Qian, W. and Lee, D.K. 2000. Seasonal march of Asian summer monsoon. Intl. J. Climatol., 20
(11): 1371–86.
3.8 Centers of action

Burnett, A.W. and McNicoll, A.R. 2000. Interannual variations in the southern hemisphere winter
circumpolar vortex: relationships with the semi-annual oscillation. J. Climate, 13 (5): 991–9.
Cohen, J., Saito, K., and Entekhabi, D. 2001. The role of the Siberian High in northern hemisphere
climate variability. Geophys. Res. Lett., 28 (2): 299–302.
0
3.9 Global climatic features
Raymond, D.J. 2000. Thermodynamic control of tropical rainfall. Quart. J. Roy. Met. Soc., 126:
889–98.
Yang, D.-Q. 2001. Compatibility evaluation of national precipitation gage measurements. J.
Geophys. Res., 106 (D2): 425–22.
Chapter 4 Large-scale circulation and climatic characteristics

4.2 Jet streams
0
Inatsu, M., Mukougawa, H., and Xie, S.-P. 2000. Formation of a subtropical westerly jet core in
an idealized GCM without mountains. Geophys. Res. Lett., 27 (4): 529–32.
4.3 Planetary waves

DeWeaver, E. and Nigam, S. 2000. Do stationary planetary waves drive the zonal-mean jet anomalies
of the northern winter. J. Climate, 13 (13): 2160–76.
4.4 Zonal index

0 Feldstein, S.B. 2000. Is interannual zonal–mean flow variability simply climate noise? J. Climate,
13 (13): 2356–62.
Robinson, W.A. 2000. A baroclinic mechanism for the eddy feedback on the zonal index. J. Atmos.
11 Sci., 57 (3): 415–22.
608 Further reading
1 4.7 Low-frequency circulation variability and persistence
Bongioannini Cerlini, P., Corti, S., and Tibaldi, S. 1999. An intercomparison between low-frequency
variability indices. Tellus, 51 (5): 773–89.
Chapter 5 Global teleconnections

5.1 Pressure oscillations and teleconnection patterns
Robertson, A.W. and Ghil, M. 2000. Large-scale weather regimes and local climate over the western
United States. J. Climate, 12 (6): 1796–813.
5.2 The Southern Oscillation and El Niño

An, S.I. and Wang, B. 2000. Interdecadal change of the structure of the ENSO mode and its impact
on the ENSO frequency. J. Climate, 13 (11): 2044–55.
Burgers, G. and Stephenson, D.B. 1999. The “normality” of El Niño. Geophys. Res. Lett., 26 (8):
1027–30.
Federov, A.V. and Philander, G.S. 2000. Is El Niño changing? Science, 288 (5473): 1997–2002.
Trenberth, K.E. 1996. El Niño–Southern Oscillation. In: T.W. Giambelluca and A. Henderson-
Sellers, eds, Climate Change: Developing Southern Hemisphere Perspectives, Wiley, New York,
pp. 145–73.
White, W.B. and Cayan, D.R. 2000. A global El Niño–Southern Oscillation wave in surface temper-
ature and pressure and its interdecadal modulation from 1900 to 1997. J. Geophys. Res., 105
(C5): 11223–42.
5.3 ENSO mechanisms

Chan, J.C.L. and Xu, J. 2000. Physical mechanisms responsible for the transition from a warm to
a cold state of the El Niño–Southern Oscillation. J. Climate, 13 (12): 2056–71.
Matthews, A.J. 2001. Propagation mechanisms for the Madden–Julian Oscillation. Quart. J. Roy.
Met. Soc., 126: 2637–51.
Neelin, J.D., Jin, F.-F., and Syu, H.-H. 2000. Variations in ENSO phase locking. J. Climate, 13
(14): 2570–90.
Saravanan, R. and Chang, P. 2000. Interaction between tropical Atlantic variability and El Niño–
Southern Oscillation. J. Climate, 13 (13): 2177–94.
Vecchi, G.A. and Harrison, D.E. 2000. Tropical Pacific sea surface temperature anomalies, El Niño,
and equatorial westerly wind events. J. Climate, 13 (11): 1814–30.
Webster, P.J. and Fasullo, J. 2000. Atmospheric and surface variations during westerly windbursts
in the tropical western Pacific. Quart. J. Roy. Met. Soc., 126: 899–924.
5.4 Teleconnections with ENSO

Hoskins, B.J. and Young, G.Y. 2000. The equatorial response to higher-altitude forcing. J. Atmos.
Sci., 57 (9): 1197–1213.
Kirtman, B.P. and Shukla, J. 2000. Influence of the Indian summer monsoon on ENSO. Quart. J.
Roy. Met. Soc., 126: 213–39.
Krishnamurty, V. and Goswani, B.N. 2000. Indian monsoon–ENSO relationship on interdecadal
timescale. J. Climate, 13 (3): 579–95.
Slingo, J.M. and Annamalai, H. 2000. The El Niño of the century and the response of the Indian
monsoon. Mon. Wea. Rev., 128 (6): 1778–96.
Torrence, C. and Webster, P.J. 2000. Comments on “The connection between the boreal spring
Southern Oscillation persistence barrier and biennial variability.” J. Climate, 3 (3): 665–7. (A.J.
Clarke and S. Van Gordes, Reply. Ibid., pp. 668–71.)
5.6 North Atlantic Oscillation

Dickson, R.R. et al. 2000. The Arctic Ocean response to the North Atlantic Oscillation. J. Climate,
13 (15): 2671–96.
Further reading 609
11 Garcia, R., Ribera, P., Gimenoo, L., and Hernandez, E. 2000. Are the NAO and SO related in any
time scale? Annal. Geophyicae, 18: 247–51.
Joyce, T.M., Deser, C., and Spall, M.A. 2000. The relationship between decadal variability of
subtropical mode water and the North Atlantic Oscillation. J. Climate, 13 (14): 2250–69.
Kodera, K., Koide, H., and Yoshimura, H. 1999. Northern hemisphere winter circulation associ-
ated with the North Atlantic Oscillation and stratospheric polar night jet. Geophys. Res. Lett., 26
(4): 443–6.
Kwok, R. 2000. Recent changes in Arctic Ocean sea ice motion associated with the North Atlantic
Oscillation. Geophys. Res. Lett., 27 (6): 775–8.
Morison, J., Aagard, K., and Steele, M. 2000. Recent environmental changes in the Arctic: a review.
0 Artic, 53 (4): 359–71.
5.7 North Pacific Oscillation

11 Sheng, J. 1999. Correlation between the Pacific/North America pattern and the eastward propaga-
tion of sea surface temperature anomalies in the North Pacific. J. Geophys. Res., 104 (D24):
30885–95.
5.8 Zonally symmetric oscillations

0 Baldwin, M.P. and Dunkerton, T.J. 1999. Propagation of the Arctic Oscillation from the strato-
sphere to the troposphere. J. Geophys. Res., 104 (D24): 30937–46.
Deser, C. 2000. On the teleconnectivity of the “Arctic Oscillation.” Geophys. Res. Lett., 27 (6):
779–82.
Perlewitz, J., Graf, H.F., and Voss, R. 2000. The leading variability mode of the coupled tropo-
sphere–stratosphere winter circulation in different climatic regimes. J. Geophys. Res., 105 (D5):
6915–26.
Wallace, J.M. 2000. North Atlantic Oscillation/annular mode: two paradigms for the same pheno-
menon. Quart. J. Roy. Met. Soc., 126: 791–805.
0 5.10 Tropical–extratropical teleconnections

Mo, R.P., Fyfe, J., and Derome, J. 1998. Phase-locked and asymmetric correlations of the winter-
time atmospheric patterns with the ENSO. Atmos.-Ocean, 36: 213–40.
5.12 Time-scale aspects of teleconnections

Pozo-Vasquez, D., Esteban-Parra, M.J., Rodrigo, F.S., and Castro-Diez, Y. 2000. An analysis of
the variability of the North Atlantic Oscillation in the time and frequency domains. Intl. J.
Climatol., 20 (14): 1675–92.
0 5.13 Interannual to interdecadal oscillations

Dunkerton, T.J. 2000. Inferences about QBO dynamics from the atmospheric “tape recorder” effect.
J. Atmos. Sci., 57 (2): 230–46.
Salby, M. and Callaghan, P. 2000. Connection between the solar cycle and the QBO: the missing
link. J. Climate, 13 (14): 2652–62.
Haarsma, R.J., Selten, F.M., and Opsteegh, J.D. 2000. On the mechanism of the Antarctic Circum-
polar Wave. J. Climate, 13 (5): 1461–80.
Chapter 6 Synoptic systems

0
6.1 Early studies of extratropical systems
Grahame, N. 2000. The development of meteorology over the last 150 years as illustrated by
11 historical weather charts. Weather, 55 (4): 108–17.
610 Further reading
1 6.2 Climatology of cyclones and anticyclones
Sickmoller, M.A., Blender, R., and Fraedrich, K. 2000. Observed winter cyclone tracks in the
northern hemisphere in reanalyzed ECMWF data. Quart. J. Roy. Met. Soc., 126: 591–620.
Simmonds, I. and Keay, K. 2000. Variability of southern hemisphere extratropical cyclone behavior,
1958–97. J. Climate, 13 (3): 550–61.
6.3 Development of cyclones

Joly, A. et al. 1999. Overview of the field phase of the Fronts and Atlantic Storm Track Experiment
(FASTEX) project. Quart. J. Roy. Met. Soc., 125: 3131–63.
Parker, D.J. 2000. Frontal theory. Weather 55 (4): 120–34.
6.5 Satellite-based climatologies of synoptic features

Hewson, T.D., Craig, G.C., and Claud, C. 2000. Evolution and mesoscale structure of a polar low
outbreak. Quart. J. Roy. Met. Soc., 126: 1031–63.
Laing, A.G. and Fritsch, J.M. 2000. The large-scale environment of the global populations of meso-
scale convective complexes. Mon. Wea. Rev., 128 (8): 2756–76.
6.6 Synoptic-scale systems in the tropics

Elsner, J.B., Liu, K.-B., and Kocher, B. 2000. Spatial variation in major US hurricane activity:
statistics and a physical mechanism. J. Climate, 13 (13): 2293–305.
Emmanuel, K. 2000. A statistical analysis of tropical cyclone intensity. Mon. Wea. Rev., 128 (4):
1139–51.
Chapter 7 Synoptic climatology and its applications

7.3 Objective typing procedures
Brinkmann, W.A.R. 2000. Modification of a correlation-based circulation pattern classification to
reduce within-type variability of temperature and precipitation. Intl. J. Climatol., 20 (8): 839–52.
Cavazos, T. 2000. Using self-organizing maps to investigate extreme climate events: an applica-
tion to wintertime precipitation in the Balkans. J. Climate, 13 (10): 1718–32.
El-Kadi, A.K.A. and Smithson, P.A. 2000. Probability assessment of pressure patterns over the
British Isles. Intl. J. Climatol., 20 (11): 1351–69.
7.5 Regional applications

Kidson, J.W. 2000. An analysis of New Zealand synoptic types and their use in defining weather
regimes. Intl. J. Climatol., 20 (3): 299–316.
7.7 Seasonal structure

Keller, L.M. and Houghton, D.D. 2000. Characteristics of the large-scale circulation changes during
the sudden onset of the fall transition season. Intl. J. Climatol., 20 (4): 397–415.
7.8 Climatic trends

Werner, P.C., Gerstengrabe, F., Fraedrich, K., and Oesterle, H. 2000. Recent climate change in the
North Atlantic/European sector. Intl. J. Climatol., 20 (5): 463–71.
11
Index
absorption spectra 25 61, 372, 399, 400–1; sea surface temperature

11 adiabatic method of computing vertical velocity 367, 368–70, 376, 383–4, 388–9, 395, 417, 420
51 422–3; temperature 385–8, 406; wind 382, 389
adiabatic lapse rate 111, 140, 242 Antarctic Circumpolar Current 230
adiabatic motion 55 Antarctic Circumpolar Trough 486
advection: thickness 525; vorticity 185, 276–7, Antarctic Circumpolar Wave 421–3
320, 459, 525 Antarctic ice-sheet 9
aerosols 111, 116, 160 Antarctic Peninsula 223, 308, 410, 494
0 Africa 499, 501, 507, 514, 605; East 188–90; Antarctica 40, 209, 219, 281–2, 409, 484, 493,
West 274, 511–12 571
African waves 512–16 anticyclogenesis 215, 449–50
airflow: regional types 563–6 anticyclone 441; blocking 309–11, 313, 317,
air mass 235, 550, 559, 584; Arctic 237-8, 504, 319–20, 22, 573; frequency 442–3; polar 209,
characteristics 237–40, 588, 589, 591; 238, 449–50, 586; Siberian 204–5, 238, 570;
classification 237–8, 560, 584; definition 235; subtropical 166–71, 215, 216, 218, 238, 370;
depth changes 238, 240; modification 238–40; tracks, 469, 571; warm 309, 311
polar 237–8; relation to general circulation aphelion 112
238–9; temperature 586; tracers 236–7; Arctic 57, 238, 526, 555, 574, 604; front 208,
trajectories 57, 238–40; tropical 237–8 240; high 216–17; low 490
0 air pollution 238, 589, 591 Arctic Ocean 209, 224, 446, 448–9, 555, 574,
air–sea interaction (see also atmosphere–ocean 586, 604
interaction) 490 ARMA (autoregressive moving averages) 64, 376
Alaska 57, 555, 574, 582, 589 Asia: East 57, 205, 271, 583
albedo: cloud 116, 159, 478; feedback 157–9; Asian monsoon 113, 145, 169–71, 192–7, 204–5,
planetary 110–1; sea ice 9; snow 9, 157; 275, 236, 391–2, 421
surface 159 ASII (air–sea interaction instability) 490
aliasing 23, 71 assimilation 22, 166, 478, 605
Aleutian low 211, 216, 385, 404, 411, 424, 495 Assmann, R. 16
Alps 548, 568, 574, 576–7, 588, 590 astronomical periodicities 112–13, 115
alternation index of cyclones and anticyclones asynoptic (satellite) data 22
0 442 Atlantic Ocean 370–1
altimetry 381 ATEX (Atlantic Tropical Experiment) 230,
Amazon Basin 160 506
AMIP (Atmospheric Model Intercomparison atmosphere: barotropic 235, 271, 281;
Program) 166 Boussinesque 147; gases 25, 26, 110–11; mass
Amundsen Sea low 486 128; tides in 288–91; vertical structure 120–3,
analog 578–9 230, 235–6
Andes 282, 294 atmosphere–ocean interaction 6–9, 231–3, 364,
angular momentum 124–6, 223, 275–6, 520–1; 377–8, 524
transport 125–9, 135, 149, 241, 306 atmospheric bridge 395
angular velocity 124, 126, 286 atmospheric circulation 113–14, 139, 166–209;
0 ANN (artificial neural network) 560–1 and hydrology 590
anomaly: field 424–5; height 61, 311, 399, 406, atmospheric energy 114–18, 129–43
468–9, 470; pattern 388; potential vorticity atmospheric GCMs 161–3 (see also general
11 461, 464, 513; precipitation 385–8; pressure circulation models (GCMs)
612 Index
1 AMIP (Atmospheric Model Intercomparison Brunt–Väisälä frequency 11, 122, 472
Program) 166 buoyancy 229, 232–3, 242
atmospheric moisture 139, 474; transport 136–40
atmospheric motion 11–13; laws of horizontal calendar: synoptic 580–7, 583, 584–5
motion 161–2; vertical 49–53, 212–13, 223 California 559, 584
atmospheric window 26, 116 Canada 57, 555, 576, 586, 587, 604
Australia 275, 494, 502, 505, 552; summer canonical correlation 42, 183
monsoon 206–7, 336, 421 CAPE (convective available potential energy)
Australian Bureau of Meteorology 480 232, 234, 242, 523
autocorrelation 64, 66, 83, 360, 365 carbon dioxide 116, 160
avalanches 590 Caribbean 508, 510, 511
averages 126–7 Carnot cycle 524
AVHRR (Advanced Very High Resolution catalog of weather types (see synoptic)
Radiometer) 20, 21, 23, 26, 31 CAV (constant absolute vorticity) 275–6
axial path wobble 112 center of action 209–20
Azores 313, 397; high pressure 215, 313 China: summer monsoon 198; winter cold waves
204–5, 211
Baiyu rains 197, 582 circulation: atmospheric 113–14, 139; cells
band-pass filter 322–3 143–53, 373; epochs 588; hemispheric 414,
baroclinic adjustment 122, 123 569–74, 578, 587; types 547–66, 576–7, 580–1
baroclinic instability 148, 186, 450–1, 472, 514 circumpolar trough (Antarctica) 218–19, 443
baroclinic leaf 463–4, 483 circumpolar vortex 263–6, 305–6, 339, 586;
baroclinic wave 470, 473–5, 481 eccentricity 279
baroclinic zone 208, 235, 270, 452, 456–9, 476 CISK (Conditional Instability of the Second
baroclinicity 490 Kind) 183, 336, 490, 523
barotropic: atmosphere 271; fluid 235, 346; classification: air mass 237–8, 560; circulation
instability 331 572–8; cloud 477–8; evaluation of synoptic
barotropic mode 324 classifications 549; Grosswetterlagen 566–8,
Bay of Bengal 199, 201, 515 576; objective 558; pressure field 549, 551–8,
Bergen school 441, 450, 476 566–8; weather map 547–51; weather types
Bernoulli probability 37 550, 558–60, 574
beta distribution 39–40 classificatory methods 81–4
beta effect 11, 188 Clausius–Clapeyron relationship 156
biennial cycle 392 climate: definitions 13, 31; global 223–35;
biennial oscillation 420–1 (see also Quasi- regimes 233–4; scales of variation 5, 10–13;
biennial oscillation) system 1–2; variables 31–2
bimodality 325–8, 331, 332 climate change: causes of 112–15, 119–20, 587;
binomial distribution 37 glacial period 157, 587; historical period 587;
bioclimatology 591 recent 587–9
biosphere 9–10; feedback 160–1, 169 climate models 153–6, 161
black body radiation 25, 27–8, 241 climatic forcing 3–5
blocking 305, 307, 309–22, 324; anticyclone climatic regime 81, 233, 575
309–11, 313, 316–17, 319–20, 322, 573; in climatic regions 81, 223–8, 238–9
southern hemisphere 311–12, 409, 494; cloud band: 484, 485, 517, 520; classification
mechanisms 313–22; pattern 284, 308–13, 477–8; cluster 177, 336–8, 486–7, 508, 521–2,
328–9, 571, 587 523; fields 478–85;forcing 159–60; system
“Bomb” (cyclone) 460, 483 224, 484, 505; top temperature 21–2, 498,
BOMEX (Barbados Oceanographic and 499–500; types 224, 226, 231
Meteorological Experiment) 506 cloud cover 23, 27, 116, 223–26; feedback
boreal forest 240 159–60
boundary conditions 164 cloud vortices 485–6, 502; tropical 486–7, 506,
boundary layer 227–31, 462, 501, 513 508–9
Bowen ratio 117 clustering methods 80–2, 360–1, 558, 559, 584
break: in monsoon 197, 207; in time series 66–7, clusters: of atmospheric states 332–3; of cyclone
325 tracks 467; of trajectories 551, 590
brightness temperature 28, 527 coherence: spatial 61–3
British Isles 552, 555, 561–7, 574, 581, 582, 587, cold fronts 452, 455–8, 501, 521
589; airflow types 561–6 cold lows 449, 493–4
Index 613
11 cold surge 204–5, 395 deformation 45
cold tongue: equatorial 183, 376 degrees of freedom 63, 66, 78
comma cloud 463–4, 488–92 delayed oscillator 383–4
complex climatology 551, 558 depression tracks (see cyclone tracks)
composite differences 386–7 depressions: monsoon 198–200; tropical 186–87
composite pressure fields 412 desertification 160–1, 169
conditional instability 462 deserts 201–3, 392
confluence 44–5, 176–7; zone 274 development 525; of cyclones 451–2, 459–65,
conservation of angular momentum 126, 275–6 483; historical ideas on 450–2
continental location: effect on climate 153–5, Devil’s staircase 383
0 192–3 diabatic heating 130–1, 275, 294, 298
contingency analysis 34 diffluence 44
continuity equation 50, 162, 425 Dines compensation 480
contour chart 40, 547 discriminant analysis 83–4, 559–60
11 convection 233–4, 378, 390, 396, 478, 500, 501, dissipation 142–3
513, 515, 520, 521, 524; slantwise 224, 453, distance measures 80, 552–3, 570
481, 505 distribution (see frequency distributions)
convective systems 514–15; mesoscale 21, 207, diurnal dancing 336
235, 487, 595–7, 508 diurnal pressure wave
convergence 451, 480–1, 501 divergence 44–6, 49, 50, 87, 188, 238, 450, 451,
convergence zone (see ITCZ and SPCZ) 459, 480–1, 501; coastal 47
0 conveyor belt 453–4; oceanic DMSP (Defense Meteorological Satellite
Coriolis force 47 Program) 20, 29, 31, 488–9, 494, 502
Coriolis, G. 144 doldrums 176,
Coriolis parameter 11, 47, 186, 188, 223, 286, downscaling 166, 561, 566, 576
507, 519 drift pattern 188–9
correlation: canonical 42, 183; decay 58–9; field dry adiabatic lapse rate (DALR) 111, 140
61, 359–60; lag-correlation maps 470; matrix duration of circulation types 587–8
78, 82; patterns 359–60, 362, 414–16; serial dynamic climatology 13–14, 604–5
63–8; spatial 58–9, 101–3 dynamic instablility 275, 329
correlation velocity patterns 474 Dzerdzeevski classification 414, 548, 570–1, 584,
correlogram 71 588
0 coupled models 163–4, 166, 192–3, 383, 417,
422, 424 Earth dimensions 6, 110
COWL (cold ocean, warm land) 418 Earth–atmosphere system 223
Cressman interpolation 57 easterly tropical jet stream 195
cryosphere 9 easterlies: subpolar 209; tropical 171–4, 228, 268,
cumulonimbus 145, 177, 181, 226, 230–2, 521 509–12
cumulus convection 173, 228–33, 500, 523, 524 easterly wave 508–12, 514
curl 48, 87 eccentricity of Earth’s orbit 112, 115
cut-off low 309, 320 ECM (Elementary Circulation Mechanism) 570–1
cycle: index 305–6; solar 420 (see also ECMWF (European Centre for Medium Range
periodicity) Weather Forecasts) 41, 166, 511, 547
0 CYCLES (cyclonic Extratropical Storms) project ecological studies using weather types 591
476 eddies 126–7, 413, 470; horizontal 128;
cyclogenesis 216, 318, 321, 324, 442–3, 459–65, stationary 126, 132; transient 126, 132, 268,
486, 493; secondary 464–5; tropical 515, 277, 319, 323, 414, 469
519–23; types 452–5, 460–4, 488 eddy transport 126–8, 132–5
cyclone 441; development 450–9; frequency 410, eddy vorticity 314
442–7; models 441–2, 452–5, 485–6, 555–6; eigen vectors (see empirical orthogonal
stage 136, 485–6; tracks 217, 218, 219, 442, functions (EOFs))
465–76 Ekman pumping 232
cyclonic vortices: classification 484–6 El Niño 136, 145, 275, 364, 367, 377, 382,
cyclosis 216 385, 486
0 Eliassen-Palm flux 320, 341–4, 346, 472–3,
data: quality 41–2; sources 41–2 606
De Bort, Teisserenc 16, 120–21, 209 emissivity 27–8
11 deforestation 160–61 EMS (electromagnetic spectrum) 18
614 Index
1 energy 7; balance 114–17; conservation 27; forecasting: long range 574, 578; numerical 22,
conversion 140–3, 512; kinetic 10, 13, 117, 161, 166; seasonal 578
141, 143, 277, 293–4, 413, 451, 464, 512, 526; Fourier analysis 71–3, 278, 325; time series 71–3
latent 117; potential 117, 129, 526; static 133; frequency distributions 32–3; beta 39–40;
total 130, 132–4; transport 129–36; vertical binomial 37; gamma 38–9; normal 35–6;
transfer 114–15 Poisson 37–8
energy budgets 274, 590; atmospheric 114–18; of frequency domain 71–7, 474
heat low 203 friction 125
energy transfers: in the atmosphere 120, 121, front: anafront 453; Arctic 208, 240; cross-front
129–36, 180–1; in the ocean 130, 132 circulation 526; katafront 453; Polar 584
enhanced convection 483 frontal analysis 453–9; objective 456–8; three-
ENSO (El Niño Southern Oscillation) 136, 145, front model 453
204, 364, 367–76, 384–96, 404, 409–10, 414, frontal contours 453
422, 496, 494, 502, 505, 516, 561; El Niño frontal cyclone; life cycle 455, 460–1
136, 145, 275, 364, 367, 377, 382, 385, 404, frontal zones 270
410, 486, 494, 495, 502, 514, 561, 566; events frontogenesis 448, 452, 526; instant (see
367–70, 374–6, 377, 385–8, 495; indices 367; occlusion, instant)
mechanisms 376–84; Southern Oscillation frontolysis 526
358–9, 361–75, 391; teleconnections 384–96, FRONTS 92 experiment 464
409–10
ENSO/PNA circulation modes 411 gamma distribution 38–9
enstrophy 319 GATE (GARP Atlantic Tropical Experiment)
enthalpy 117, 132 230–2, 506, 511–12
EOFs (empirical orthogonal functions) 78–81, general circulation 109
330, 332–3, 334, 340, 360–1, 417, 556, 558, general circulation models (GCMs) 109, 116,
559–60, 572–3, 578, 586, 605 118, 119, 148–51, 160–6, 183–4, 300–1, 561,
equatorial duct 188–9, 507 576, 590, 605; AMIP (Atmospheric Model
equatorial trough 176–82 Intercomparison Program) 166
equatorial westerlies 175, 195 geopotential 40; height fields 263–5
equivalent temperature Geosat 30, 484, 487
ERICA experiment 462, 476 geostrophic wind; resultant 238
Ertel potential vorticity 122, 241, 319, 321–2 Germany 566, 574, 580, 582
ESMR (Electrically Scanning Microwave Gibbs oscillation 163, 165, 605
Radiometer) 20, 29 glacial anticyclone 217
Etesian winds 171 glacial period 155, 157, 587
Ethiopian Highlands 511 glacier variations 590
Eulerian method 53, 84–7 global analyses 41–2
Eurasia: teleconnection pattern 397–8 GOES (Geostationary Operational Environmental
Europe 403, 449, 556–7, 566, 568, 570, 580, Satellites) 18, 20, 478–9, 496, 497, 504, 505,
587 518
evaporation 137, 139, 462 GPCP (Global Precipitation Climatology Project)
exploratory data analysis 33 29, 224
eye 517–18 GPI (GOES Precipitation Index) 28, 29
gradient operator 45, 87
factor analysis (see empirical orthogonal gravity wave 285, 420
functions) Great Salinity Anomaly 423, 424
fast Fourier transform 69 greenhouse gases 116, 156
feedback: mechanisms 155–61 Greenland 40, 57, 217, 296, 400–1, 590
feedback studies (see air–sea interaction) Grosswetter 566, 574, 587
Ferrel cell 127, 141, 144 Grosswetterlagen 331, 548, 566–8, 576, 582,
Ferrel, W. 144 587–8, 590
Ferrel westerlies 207 GTS (Global Telecommunications System) 17
field: intercomparison 61–3; significance 62–3 Gulf Stream 400, 418, 462, 473
field of view 23
filtering: time series 65–6, 74, 323–4, 468, 469 Hadley cell 119, 127, 137, 141, 144, 148–53,
finite differences 85 169, 186, 234, 268
flow patterns 566; equatorial 188–9 Hadley circulation 144–5, 148–9, 374, 378, 385,
föhn 576 391, 395
Index 615
11 Hadley, G. 144, 191 isotach 42
Halley, E. 191 Italy 574
harmonic analysis 68–70; spherical 296, 298, ITCZ (Intertropical Convergence Zone) 152,
340–1 176–80, 186–7, 191, 224, 232, 370, 478, 495,
Hawaii 57, 200 497, 514, 519
HCMM (Heat Capacity Mapping Mission) 23
heat; latent 117, 132, 145, 299–300, 508, 519; January thaw 582
sensible (enthalpy) 117, 132, 145, 298, 472, jet streak 459–60, 525
486, 519; sources 129–30, 470; transport 130, jet streams 270–8, 408, 413, 469, 472, 481, 483,
132–6, 486; turbulent fluxes 117, 299 513; African easterly 274, 511, 513; Australian
0 heat low 201–4, 215, 274 275, 385; discovery 271; low-level 190, 196,
hemispheric circulation 414, 569–74, 578; Fourier 455, 458–9, 497, 501; polar front 208, 270,
analysis 278–9; indices of 587; types 569–74 410, 494; polar night 405; related to
Himalayas 272, 296, 586 precipitation 26; split 311, 319; subtropical
11 hot towers 145 145, 147, 186, 195, 204, 207–8, 270, 274–5,
Hovmöller diagram 200, 282–3, 487 385, 416, 494, 501, 586; tropical easterly 195,
HRCs (highly reflective clouds) 27, 177–9, 233, 272–4, 460; velocity maxima 271, 276–7
378, 476 Joseph effect 65
human health and weather types 591
hurricanes: development 515; intensity 516, 522, Kelvin wave 234, 293, 334, 346, 379–84, 508
524 Kernlose winter 222–5
0 Hurst phenomenon 65 kinematic analysis 42–5, 55, 551
hydrological cycle 136 kinetic energy 10, 13, 117, 141, 143, 277, 293–4,
hydrosphere 5–9 413, 451, 464, 512, 526; spectrum 10
hydrostatic equation 14 kinetic temperature 27
hydrostatic equilibrium 162 Kirchhofer method 552–6, 566, 576, 578
Kirchhoff’s Law 26–7
ice (see sea ice) Kona storm 200
Ice Saints 580 Köppen climate classification 155, 165
ice sheets 7, 9 kriging 58–60
ice storm 504, 505 Kuroshio current 424, 473
Iceland 397, 587 kurtosis 35–6
0 Icelandic low 211, 215–16, 399, 404, 423, 446
IGY (International Geophysical Year) 507, 547–8 La Niña 136, 145, 275, 364, 404, 410, 486, 494,
index cycle 305–6 514, 561
India 77, 499, 507; summer monsoon 195–8; Labrador-Ungava 564
relationship to ENSO 391–6 Lagrangian method 53–7, 84–7, 465, 467
Indian Ocean 385, 390–6, 571, 573 Lamb catalog 561–6, 587, 589
Indonesia 339 Lamb types 42, 562, 564–5, 576, 589, 590
inertial circle 188 Landsat 23
influence field 571 Laplace tidal functions 288, 291
infrared radiation 24, 26–7, 112, 115, 121 Laplacian operator 49, 87, 458, 524
INSAT (Indian National Satellite System) 20 lapse rate 119, 121
0 instability: baroclinic 148, 186, 451; Charney- latent heat 5, 117, 508, 519
Stern criterion 45; frontal 464; hydrodynamic laws of thermodynamics 161
275–6 Legendre functions 290–1, 340–1
intransitive system 329 LIE (Line Islands Experiment) 506
inverse distance weighting 58 lightning 500
inversion 501, 519; trade wind 172–4, 182–4, limit cycle 330–1
229–31; of satellite sounding radiances 24 linkage 80
ISCCP (International Satellite Cloud Climatology long waves (see planetary waves)
Project) 224, 233 low frequency variablility 322–32, 411, 414–16
isentropic analysis 53–5 low-latitude circulation 171–90
isentropic surface 451, 605
0 ISO (intra-seasonal oscillation) 332–40 maritime continent 145, 198, 370, 391
isobaric analysis 55 Markov chain (process) 64
isocorrelate 61 mass flux 152
11 isogon 42 master seasonal trends 586
616 Index
1 Maunder Minimum 587 natural season 571, 580, 582–6
maximum entropy method 73 NCEP (National Centers For Environmental
MCC (Mesoscale Convective Complex) 477, Prediction) 41, 166, 446, 547
495–501; criteria 498 NDVI (Normalized Difference Vegetation Index)
MCS (Mesoscale Convective System) 21, 207, 9
235, 487, 495–6, 497, 508 negative viscosity 13, 140
mean sea level pressure 40; reconstruction 41–2 nephanalysis 20
meander index 305, 309 neural network 560–1
median 23 New Zealand 308, 371, 409, 443, 494, 559, 564,
Mediterranean 558 574
Meiyu 197–8 Noah effect 65
meridional circulation 130, 144–53 non-linear dynamics 330–1, 384, 523
merry-go-round formation 493 normal (Gaussian) distribution 35–6
mesocyclone (see mesoscale cyclone) normal modes 291–2, 342, 345, 514
mesoscale cyclone 4, 10, 477, 486, 488–505 North America 385–7, 479, 499, 502, 554–5, 558,
mesosphere 121 576, 578, 586
methane 160–1 North Atlantic 215–16, 417, 423, 452, 472, 495,
Mexico 574 502, 511, 515, 519; seesaw mode 328, 398,
microwave radiation 27, 28–31 400–2; subtropical anticyclone 215
MJO (Madden – Julian Oscillation) 197, 270, North Atlantic Current 418, 423
334–6, 339 North Pacific 216, 417, 472, 494–5, 502, 515,
models: air–sea interaction; baroclinic 296; 519–20, 552; oscillation 358, 361, 403–4
barotropic 296–7, 300, 322, 331; cloud and Norwegian Sea 490, 493
circulation 454–5, 462–4; 480–3; coupled 63–4, nowcasting 22
166, 192–3, 383, 417, 422, 424, 605; cyclone numerical weather prediction 22, 161, 166
441–2, 450–5, 462–4, 476, 480–3, 485–6; Nyquist frequency 71–2
GCMs 161–6, 561, 576, 590; hemispheric
circulation 144; numerical prediction 22, 161, objective analysis 58
166; parameterization 163, 183–4; primitive objective typing 551–61, 565–7; artificial neural
equation 511–12; probability 34–5; regional networks 560–1; correlation-based 551–6; data
circulation 166; resolution 163; stochastic 576; reduction 556–60; multivariate techniques
synoptic 441–2, 455–60, 476, 480, 555–6 558–60
modon 298, 318–19 obliquity 112–13
moist convective adjustment 183 observing stations: surface 16–17, 604; upper air
moisture: recycling ratio 140; transport 136–40, 16–17
410 occlusion 452, 460; bent-back 455, 460; instant
moisture burst 395, 503–4 21, 463–4
Molniya satellite 23 ocean-atmosphere interaction 182, 383, 392,
momentum: angular 124–5; equation 161–2; flux 420–1
268, 270, 474 ocean currents 8
MONEX (Monsoon Experiment) 191, 203 oceans 6–8; heat transport 130, 132; thermocline
monsoon trough 176, 192, 200 514, 519 182, 376, 377, 379, 381, 383–4, 417
monsoons 190–8; Asian 113, 145, 169–71, Old Wives’ Summer 580–1
204–5, 275, 336, 391, 421; Australian 206–7, OLR (outgoing longwave radiation) 26, 28, 180,
336, 421; depression 198–200, 508; East Asian 233–4, 334, 336, 339, 392, 395
winter 204–5, 583; Indian 195–8, 336; North omega equation 51, 52, 448
American 496, 497; West African 113; operators 87
Western Pacific 198 orbital forcing 112–13
Monte Carlo simulation 63, 80, 553 orographic effects: on climate 155, 171, 193–5,
Montgomery potential 55 226, 282, 294–8, 300–1
Morning Glory 207 orthogonal polynomial surfaces 556
mountain barriers 501; airflow over 287, 295–7 oscillation: Antarctic 405–8; Arctic 404–7, 418;
moving average 65 interdecadal 418; interannual 418;
multiple flow equilibria 318, 330–1 Madden–Julian 197, 270, 334–6, 339; North
Atlantic 358, 361, 396–403, 413, 416, 418,
NAO (North Atlantic Oscillation) 358, 361, 424, 495, 587, 588; Pacific-South American
396–403, 413, 416, 418, 424, 495, 587, 588; 409, 422; Pan-Atlantic 424; quasi-biennial
index 397, 399–400 374–5, 507; quasi-decadal 423–4; semi-annual
Index 617
11 220–3; semi-diurnal 289, 291; Southern 358–9, pressure: mean surface 42, 128, 168
361–75, 391; Tropic-wide 396; zonally pressure fieldanomaly 61, 129, 372, 399,400–1;
symmetric 404–8 classification 549, 551–8, 566–8; decay of
ozone 116, 121, 123, 589, 590; hole 222 correlation in 58–9; mean 168, 566; sea level
41–2, 168, 218, 483, 547
Pacific Ocean 172–4, 176–80 pressure pattern types (see circulation types)
palaeoclimate 119, 153–5, 157, 165 pressure tendency equation 459
parameterization 163, 183–4, 605 pressure torque 125
partial collectives 32, 34, 575 primitive equations 162–3, 307
partial derivatives 85 principal component 78–80, 82, 360–1, 391, 398,
0 partial frequency 237 404; analysis 551, 556, 558–9, 565
partitioning 424–5 probability: distribution 326; models 34–5
passive microwave radiation 29, 527; imagery PSA (Pacific–South American) pattern 409, 422
483 PSCM index 564, 566
11 Pathfinder Program 21, 604 pseudo-adiabatic chart 53
PBL (planetary boundary layer) 378
perihelion 112, 113 Q vector 524–6, 605
periodicity 68–9, 112–13, 115 QBO (quasi-biennial oscillation) 374–5, 418–21,
persistence 64–6, 72, 325–32, 392, 417, 561, 578 516
Peru 367, 375, 377 quasi-barotropic relationship 586
phase changes 5 quasi-geostrophic flow 346
0 photosynthesis 9, 160
PILPS (Program to Intercompare Land Process radar 31
Schemes) 166 radar altimetry 30
pixel 23 radiation: balance 118; budget 114–15, 159–60;
Planck’s Law 24 net 117–18
planetary climate 109–13 radiation laws: Beer 27; Kirchoff 26–7; Planck
planetary waves 278–302, 325, 572–3, 578; 24; Schwarzchild 27; Stefan-Boltzmann 24,
stationary 241; Wien 24, 116, 241
PNA (Pacific–North American) pattern 266, radiative convective model 121
331–2, 360, 361,396, 398, 410–14, 416, 418, radiosonde 16; data 558–9, 560
420 rain band 198, 452, 454
0 Poisson distribution 38 rain rates 28–9
Poland 564 rainfall 77, 139, 564; in East Africa 189; indices
polar front 208, 274, 462, 584; jet stream 208, 367; related to ENSO events; types 233–4
270; theory 450 rawinsonde 413–14
polar low 452, 483, 490, 493 ray path 300–3, 388–90, 413–14
polar orbiter 18, 20, 23 reanalysis 41, 166, 446, 547
polar vortex 217, 263–6, 268, 305–6, 339, 405, recent climatic fluctuations 578–9
409 recurrence 324
pollution 238, 589, 591 red noise 66, 72
POP (Principal Oscillation Pattern) 371–2 regimes: of circulation 322–32; of climate 233–4;
potential energy 117, 129, 277, 526; available of precipitation 333–4; quasi-stationary 324–5
0 140–2, 513 regional climate 561–9, 574–8
potential temperature 123, 212–13, 455–6; remote sensing 18–21; data 17–23; history 18–21;
equivalent 232, 234, 242–3, 452, 456; wet bulb infrared 18, 20 21, 23, 26, 27; multispectral 18;
237, 458 passive microwave 20–21, 23, 26, 27; radar 31;
potential vorticity 122, 143, 186, 188, 235–6, visible 18–20, 21, 23
240–1, 458, 464, 606; conservation of 286–7, resonance 286, 289, 314
509; Ertel 122, 241, 319, 321–2 resultant vector 36
precession of the equinox 112–13 Reynolds, Osborne 126
precipitable water 136–9 Richardson, L.F. 140
precipitation 61, 137, 139, 224–7, 231, 235, 298, Richardson number 464, 526
527, 576, 577; anomaly 370; recycling 140; Rocky Mountains 294, 296, 452
0 seasonal regime 223–4 (see also rainfall) Ross Sea 492, 493, 494
precipitation estimation: from satellite data Rossby, C.-G. 271
28–9 Rossby number 147, 286
11 predictability barrier 392, 394 Rossby radius of deformation 379, 426, 507
618 Index
1 Rossby wave 122, 171, 285–6, 288, 291, 314, South America 234, 308, 377, 385, 388, 499,
388, 390, 508; dispersion 316; equation 286, 501, 519, 556, 564
572; in ocean 379–81, 383 South Atlantic Ocean 417, 423
rotating annulus 306 South China Sea 197, 200, 204, 514
rotation rate 110, 118, 125, 145, 147 South Pacific Ocean 371–2, 519, 502, 571, 604
rotational velocity 125 southern hemisphere 208, 210, 214, 217–22, 281,
running mean 65–6 268, 281, 371–2, 407–10, 418–19, 462–4,
467–9, 492–4, 502, 547–8, 571, 604
Sahara 171 Southern Ocean 219, 223, 224, 282, 308, 407–10,
SAR (synthetic aperture radar) 31 418–19, 421–2, 442–44, 484, 493–4, 604
SASS (Seasat-A Satellite Scatterometer) 18 Southern Oscillation 358–9, 361–75, 397, 420,
satellite imagery 18–21; cloud classification 21–2, 502; index 73, 74, 136, 361, 364–6, 392, 395,
28 397, 399, 425, 486, 502
satellites 18–23 Soviet Union 446, 551, 570, 604
scalar 85 spatial coherence 61
scale interactions 523–4 spatial data 40–63
scale of weather systems 11–12 spatial interpolation 56, 57–61
Scandinavia: teleconnection pattern 397 SPCB (South Pacific Cloud Band) 502
scanning radiometer 20 SPCZ (South Pacific Convergence Zone) 179–80,
scattering 116 371, 373–4, 382
scatterometer 16, 30 spectral gap 10, 488
Schwarzschild’s Law 27 spectrum analysis 71–3, 506; singular 73, 340
sea ice 9, 30–1, 402, 409, 586; Arctic Ocean 31, spell: statistical properties 64
83; albedo feedback 159; motion 31; Southern spells of weather 580, 582–3
Ocean 219, 220, 221, 282, 308, 410 spherical harmonic analysis 296, 298, 340–1
season 578–80, 582–4; natural (see natural spline methods 60
season) split window method 26
seasonal trends 586 SPOT (Système probatoire pour l’observation de
seclusion 455, 461 terre) satellite 23
secondary cyclones 464–5 squall lines 495, 508
seesaw pattern 398, 400–2, 418 SSM/I (Special Sensor Microwave/Imager) 21,
self-organizing map 560–1 29, 31, 483–4, 487, 492, 500, 527
semi-annual oscillation 220–3 SST (sea surface temperature) 16, 18, 20, 220,
sequences of circulation 305, 572 368–70, 377–8, 399, 417, 418; anomalies 367,
shadow band 481, 483 368–70, 376, 383–4, 388, 389, 395, 417, 420,
Shurin rains 583 422, 423
Siberian high pressure 40, 204–5, 211–15, 570 standard deviation 33, 35, 548
significance tests 61–3, 66–7, 68, 84, 359–60 standing wave 136
similarity measures 551–3 state diagram 110–11
simulation 153–6, 164–5 static energy 133
Singular Spectrum Analysis 73, 340, 374, 425 static stability 142, 147, 526
singularities 580–2; primary 582 stationarity 66
skewness 35 stationary wave 127, 287–88, 294–96, 314, 474–5
SMMR (Scanning Multichannel Microwave statistical tests 61, 63, 84; for shifts of mean
Radiometer) 20, 31, 527 66–7; of trend 68
snow 226, 590 steering 267, 566, 571
snow cover 31, 194, 393, 418, 586 storm tracks 322, 400, 403, 414, 465–76
SOI (Southern Oscillation index) 73, 74, 136, strange attractor 230
361, 364–66, 392, 395, 397, 399, 425, 486, stratiform cloud 227–8, 477–8, 495, 496
502 stratocumulus 177, 228–9
soil moisture 9, 155 stratopause 121
solar constant 116 stratosphere 121, 405, 409; quasi-biennial
solar influences on climate 375, 420 oscillation 420
solar irradiance 111–12, 118; at surface stream function 45, 48, 49, 130, 241, 470;
solar radiation 114–17, 299 isentropic 55
Somali jet stream 274 streamlines 42–4, 196, 459, 504, 512; analysis
sounder (see TOVS, VAS) 48, 53–4
South Africa 502 Student’s t 84
Index 619
11 subpolar easterlies 209 thermocline 182, 376, 377, 379, 381, 383–4, 417
subpolar lows 210 thermodynamic equation 162
subsidence 160–70, 215, 226–7, 322, 478, 519 thermohaline circulation 423
subtropical cyclone 200–2, 508, 517 thickness 208, 267–8, 456; advection 52, 525
subtropical high pressure 166–71, 218, 442 Thiessen method 60–1
subtropical jet stream 145, 147, 186, 195, 208, three-cell circulation model 144
270, 274–5 three-front model 238
summer monsoon 233–4; Asian 169–71, 194–97; thunderstorm 478, 495, 497, 501, 576
Australian 206–7; East Asian 197–8; Indian Tibetan Plateau 155, 194, 208, 272–4, 294, 300,
195–8, 372 586
0 sunglint 476 TICA (tropical intraseasonal convention
supercluster 487 anomalies) 336
surface/upper-air relationships: and the formation time series 63–7, 506; deterministic elements 64;
of depressions 459–64 Fourier analysis 71–3; periodic components
11 SVATS (Surface Vegetation–Atmosphere 68–71; persistence 64–5; random elements
Schemes) 160 64
SVD (singular value decomposition) 81 TIROS (Television and Infrared Orbiting
Switzerland 64, 566, 574, 576, 587 (see also Satellite) 18–20, 476
Alps) TOGA (Tropical Ocean Global Atmosphere) 231,
symmetry about the equator 424–5 233, 506
synoptic: catalogs 561–6; data 16–17; indices TOMS (Total Ozone Mapping Spectrometer)
0 560, 564, 565–6, 578, 591–2; models 450–5, 477
476; systems 441–2, 476–87, 506–24; weather TOPEX-Poseidon 30, 487
map 40–1, 549–50, 604 topographic effects 155, 171, 193–5, 287, 300–1,
synoptic classification: objective 551–61; 452, 501, 590
subjective 549–51 tornado 496
synoptic climatology 487, 494, 497, 547–91, TOVS (TIROS Operational Vertical Sounder) 18,
604–5; analysis 547–9; definitions 13–14; 20, 21, 477, 483, 487, 492
environmental applications 560, 589–91; trade wind cumulus 229–31
historical development of 548–9; regional trade winds 171–4
applications 574–8; scales 549–50 trajectory 53–7, 188; air mass 57, 238–40;
synoptic weather maps 40–1, 549–50 isentropic 55–6; isobaric 55–6, 590
0 Trans-Polar Index 308, 408–9
Tasmania 587 transformation of data 39
TECB (Tropical-Extratropical Cloud Bands) 21, transition of circulation types 325, 332, 566 (see
477, 501–5 also sequences of circulation)
teleconnection 83, 358, 384–96, 415–16; forcing transport mechanisms in the general circulation
413; indices 329, 426; patterns 358, 425; 126–39, 143–53
tropical-extra tropical 410–13 trends: in cyclone frequency 446
temperature; annual cycle 220; annual wave 209; tropical cyclones 186–87, 385, 487, 496, 515–24;
anomaly 385–8, 406; change components frequency 516; intensity 519–20, 52; structure
588–9; correlation with 700 mb height 61–2; 516–19
effective 111–12; frequency distribution 33, tropical easterly jet 195, 272–4, 460
0 236; latitude–height cross-sections 212–13, tropical/northern hemisphere teleconnection
281; latitudinal gradient 119–20, 149, 269; pattern 397–8, 410–11, 420
mean for Earth 110, 112, 118, 119; mean at tropical plume 503–4
standard pressure levels 269; potential 123, tropical waves 506–16
212–13; range 207, 478; sea surface 16, 20, tropopause 120–3, 235–7, 406; definition 122;
118–19, 220, 368–70, 377–8, 417, 418, 422, discovery 120–1
423, 487, 519; vertical structure 120–23, troposphere 121, 130, 237
194–95; virtual 242; wet bulb potential 458 tropospheric rivers 137–8
tephigram 53 trowal 453
terrain-induced vertical velocity 53, 294–5 TRMM (Tropical Rainfall Measuring Mission) 23
tesselation methods 60 truncation 341, 605
0 thermal advection 452, 460 turbulence 143
thermal forcing of planetary waves 294, 298–300 TUTT (tropical upper-tropospheric trough)
thermal Rossby number 147 519–20
11 thermal wind 266–7, 271, 456, 458, 498, 525 typhoons 515, 522
620 Index
1 United States 61, 62, 64, 70, 365, 385, 411, water vapor flux 136–8
495–6, 499, 501, 502, 505, 519, 552–3, 560, Wave-CISK 336, 514
582–4, 589, 591 wave disturbance 441, 508–15
upper air: charts 41; cyclones 200–2, 446; data wave guide 407–8, 413
16–17 wave train 391, 413, 414, 417, 418; propagation
upwelling 217, 377, 381, 383 300–2
US Weather Bureau analyses 41, 217 waves: African 512–16; easterly 508–11; global
modes 288–93; gravity 285, 420; inertio-
vacillation 306 gravity 507–8; Kelvin 234, 293, 334, 346,
Vangengeim-Girs catalogue 548, 569–70, 588 379–84, 508; numbers 143, 279, 282, 288, 300;
vapor flux (see moisture transport) properties 282–8; planetary 278–302, 314;
vapor pressure 156, 158 Rossby 285–6, 288, 291, 298, 508; Rossby-
variance 35, 64; analysis of 84; spectrum analysis gravity 508, 514; unstable 288–9; tropical
71–3 506–15
variogram 60 wavelet analysis 73–7, 393, 397, 605
VAS 18 WCRP (World Climate Research Programme)
vector notation 45 604
vegetation 9, 62, 160–1 weather: elements 551; satellites 18–23; types
velocity potential 69, 197, 390–1 550, 558–60, 574, 591
Venezuela 511 Weibull distribution 39
Venus 110–11, 147 wet-bulb potential temperature 458
vertical velocity 49–53, 212–13, 227, 389, 448, West Africa 274; monsoon 113
459, 480; computation methods 50–3, 524–5; westerly ducts 389
relative magnitudes 50; terrain-induced 53, westerly winds 207–9, 263, 303; bursts 336,
294–5 379–81; equatorial 175, 195, 379; polar night
VIHMEX (Venezuela International Meteorological 271
and Hydrologic Experiment) 506 white noise 64, 472
virtual temperature 242 Wien’s Law 24, 116, 241
VIS/IR (visible/infrared) imagery 476, 478, 480, wind: belts 126, 144, 207; components 42;
483, 486, 487 constancy 223; field 42; geostrophic 267, 271,
vortex stripping 123 302; irrotational 241–2; katabatic 493; non-
vorticity: absolute 47, 52, 123, 175–6, 184–5, divergent 49, 241–2; resultant 238; shear
208, 275–6; advection 52, 276–7, 459–60, 470, (vertical) 271; streamline 42–4, 196, 459, 504,
525; anomaly 513; conservation 319, 460; 512; thermal 266–7, 271, 456, 458, 498, 525;
cyclonic 451, 459; equation 47, 51, 276, 460; velocity 42, 267; zonal 210, 212–13, 272,
flux 474–5; geostrophic 458, 556; potential 275–6, 288, 326, 328
122, 143, 186, 188, 513; relative 45–7, 49, 87, WISHE (wind-induced surface heat exchange)
286–7, 442, 458, 459–60, 512, 556; shear 566; 234, 524
tangential 520–2 Witterungslagen 574–5
World Weather Watch 16
Walker circulation 145, 361, 374, 391–3, 396,
507 Zebiak–Cane model 383
Walker, Sir Gilbert 358, 359 zonal flow 303, 308–9, 568, 573, 587
WAMEX (West African Monsoon Experiment) zonal index 302–8, 403, 571, 586; cycle 305–6;
191, 506 high 303–4, 313, 327, 328, 571; low 303–4,
warm front 455, 460, 462 307, 313–14, 328, 571
warm pool: in mesocyclones/polar lows 490; in zonal pattern 284, 313–14, 328, 568, 570,
western Pacific 145, 179, 232, 233, 365–6, 573
392, 501, 506 zonal wind (see wind)
water vapor 26, 29, 116, 121, 136, 138, 139, 484, zonally-varying teleconnection pattern 284,
492; temperature feedback 156–7 313–14, 328, 408

Roger G. Barry - Synoptic and Dynamic Climatology (2001)

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Roger G. Barry - Synoptic and Dynamic Climatology (2001)

Uploaded by

Copyright:

Available Formats

11

Synoptic and Dynamic Climatology

111 Roger G. Barry and

11 London and New York

ISBN 0–415–03115–x (hbk)

111 List of plates viii

0111 2 Climate data and their analysis 16

3 Global climate and the general circulation 109

4 Large-scale circulation and climatic characteristics 263

5 Global teleconnections 358

6 Synoptic systems 441

7 Synoptic climatology and its applications 547

Further reading 606

Geocarto International for Figure 6.29.

1.1 The global climate system

11 Figure 1.1 The climate system. (From Hutter et al., 1990)

1.1.1 Land surface

1.1.2 The hydrosphere

Table 1.1 Dimensions of the Earth

Latitude % of total Area of zone Length of % of zone % of zone

A Principal reservoirs Water volume (km3)

Precipitation 385 111

1.1.3 The cryosphere

1.1.4 The biosphere

1.2 Time and space scales of weather and climate processes

Descriptive scaling of atmospheric motion is usually made in terms of characteristic hori-

0 3 The Brunt–Väisälä frequency for vertical oscillations:

where potential temperature, g gravitational acceleration, and z altitude.

1 Small scale, where F (the atmospheric frequency) > N.

1.4 The structure of the book

2.1.2 Upper-air reports

0 2.2.3 Principles of satellite remote sensing and applications

2.3 Climate variables and their statistical description

Order Variables Description Representation

2.3.1 Frequency distributions

2.3.3 Contingency analysis

0 and the second moment, or variance:

The standard deviation is given by s. The rth moment is:

ƒ (x) 冢nx冣 p q , for x 0, 1, 2, . . . n

0 where 冢nx冣 is the number of combinations of x items out of a total of n,

(i.e. the total area beneath the frequency curve is 1).

 is a shape parameter estimated from:

冦ab (x bc) 冤 (x c)a

f (x) = 冤(a) (b)冥

Single peak > 1 > 1

2.4 Analytical tools for spatial data

g acceleration due to gravity (m s2).

Source Problems/Changes References

2.4.3 Kinematic properties of the wind ﬁeld

0°–90° sin  cos 

2.4.4 Derived data

so that  is deﬁned by:

If variations in the Coriolis parameter (f ) are ignored,  in an isobaric surface is propor-

It follows from the above that:

where 2 the two-dimensional Laplacian operator (see Appendix 2.1).

2.4.5 Vertical velocity

System Velocity (cm s1) Time scale

where  dp/dt. Accordingly, the continuity equation here becomes:

It is generally assumed that  0 at the surface pressure value. This is satisfactory as

w= 冢(∂T/ ∂t)  V· T 冣 = 冢T/t冣

[(∂ / ∂t)  V (∂ / ∂s)]

where V horizontal wind speed,

6 k1 (Aa )5  k2 (Az)

2.4.6 Isentropic charts

is a shape parameter estimated from:

f (x) = 冤(a) (b)冥

0°–90° sin cos

so that is deﬁned by:

If variations in the Coriolis parameter (f ) are ignored, in an isobaric surface is propor-

where 2 the two-dimensional Laplacian operator (see Appendix 2.1).

where dp/dt. Accordingly, the continuity equation here becomes:

It is generally assumed that 0 at the surface pressure value. This is satisfactory as

w= 冢(∂T/ ∂t) V· T 冣 = 冢T/t冣

[(∂/ ∂t) V (∂/ ∂s)]

6 k1 (Aa )5 k2 (Az)

*(h) = 0.5[z(x1 h) z(x1)]2

冤兺 a cos kx 兺 b sin kx冥

where Ak (ak2 bk2)1/2, the amplitude of the harmonic, and:

Note that the vertical relative vorticity, , is:

S (1 p) r2 = Te4 4r 2

differences (T) expressed as a function of (TG), (i.e. T/TG) display a consistent

M" = " r2 cos2

[uv] = [u ][v ] [u]′[v]′ [uv] [u′v′] (1)

[(cpvT )] cp[v][T] cp [vT ] cp [v′T ′]

where # Q is the mean horizontal divergence of moisture in the air column,

Model Ts (K) Ta (K) P (cm/yr) E (cm/yr) Reference