→ ERS MISSIONS

20 Years of Observing Earth

 SP-1326
July 2013

→ ERS MISSIONS
20 Years of Observing Earth
 Cover: Synthetic Aperture Radar multitemporal colour composite image
showing Naples and its bay. Naples, the largest city in southern Italy and
capital of the Campania Region, can be seen as the large bright area on the
northern tip of the Bay of Naples. The bay lies within an important volcanic
province, with the volcano Mount Vesuvius and the Phlegrean Fields (Campi
Flegrei) area being its active and ancient primary testimonies. The different
colours shown by the sea in the bay indicate sea surface roughness caused by
the winds which occurred during the dates of acquisition. The colour patches
to the north of Naples represent crop fields. The image is made from three ERS-2
SAR Precision Radar Image images acquired on different dates. A different RGB
colour is assigned to each date of acquisition: Red: 11  February 2004, Green:
21 April 2004, Blue: 30 June 2004. (ESA)

An ESA Communications Production

Publication ERS Missions: 20 Years of Observing Earth (ESA SP-1326, July 2013)

Production Editor K. Fletcher

Editing/Layout Contactivity bv, Leiden, the Netherlands

Publisher ESA Communications

ESTEC, PO Box 299, 2200 AG Noordwijk, the Netherlands
Tel: +31 71 565 3408 Fax: +31 71 565 5433 www.esa.int

ISBN 978-92-9221-424-1

ISSN 0379-6566

Copyright © 2013 European Space Agency

 Foreword

Foreword

Scientific and commercial remote sensing was still in its early stage when the
first European Remote Sensing satellite, ERS‑1, was launched in 1991, leading
to the outstanding achievements described in this book.
ERS‑1 was the most advanced and complex satellite of its day, carrying
three instruments (ERS‑2 had four instruments) and delivering what was at
the time an enormous volume of data to Earth. Ensuring the timely delivery of
products was a great challenge because of the low computing capacities and
the weak performance of telephone lines featuring ‘modern ISDN’ technology.
It should be considered that at the time, information technology for digital data
processing was very expensive and the majority of products were disseminated
using postal services. Consequently, the ERS programme was built with
exploitation goals based on the available ‘ground’ technology of the late 1980s
targeting specific user communities.
Concomitant with the development and launch of ERS‑1 a revolution took
place in computer and information technology. Computers and workstations
doubled in speed and capacity every year, and the World Wide Web took off,
boosting internet utilisation, supporting easy access to ERS data and bringing
computing costs down to a mass market level. The mission operations benefited
strongly from this technological revolution, removing processing limitations
and bottlenecks in accessing data and making them freely available online.
Earth observation applications based on space data also profited from these
developments as they could suddenly be processed on the desks of users
instead of only in computer centres.
Throughout the 20 years of ERS, special emphasis has been given to
the calibration of instruments, validation of products, cross-calibration of
instrumentation, information retrieval evolution, and to the documentation
of these processes. This has been a joint effort with specialised labs, user
communities and partner space agencies, keeping the mission an innovative
asset for research and services.
By enabling the dissemination of high-quality data and building on
continuous technological progress driven by the computer and internet
revolution, the ‘free and open’ data policy has been the ultimate accelerator
in ensuring the full exploitation of ERS data. Allowing an ever-growing
community to have easy and free access to satellite data enormously facilitated
the transformation of these unique data into valuable information, far beyond
the original scope of the programme.
The outstanding robustness of both satellites and instrumentation has
facilitated long time series of data overlapping with successor instrumentation
flown on ERS‑2, Envisat and MetOp, enabling the creation of ‘one mission’
across several platforms.
The environment of high-quality data being available easily to all users has
been the essence of an outstanding return on investment for ESA stakeholders, and
has also led to new national satellites and the Global Monitoring for Environment
and Security (GMES) programme, recently renamed Copernicus. ERS has served
as a precursor mission in Europe for acquiring key industrial expertise in areas
like radar technology and scientific application fields like interferometry. There are
thousands of projects throughout the world relying on ERS data.
This book describes the exceptional success of a pioneering satellite,
starting in a different technological era and founding communities that have
significantly advanced our knowledge of planet Earth, while bringing research
to operational services and business. The Sentinels will carry on this heritage,
making remote sensing a commodity for all citizens.

Volker Liebig
Director of Earth Observation Programmes

iii
 Contents

Contents

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

History of the ERS Programme . . . . . . . . . . . . . . . . . . . . . . . . . 7

G. Duchossois, G. Kohlhammer & W. Lengert

The ATSR Instruments and What They Have Led To . . . . . . . . . . . . . . . . 45

D.T. Llewellyn-Jones

ERS and Atmospheric Composition Measurements: GOME and the SCIAMACHY Project . . 63
J.P. Burrows

The ERS Scatterometer: Achievements and the Future . . . . . . . . . . . . . . 143

A. Stoffelen & W. Wagner

The ERS SAR Wave Mode: A Breakthrough in Global Ocean Wave Observations . . . . . 165
K. Hasselmann, B. Chapron, L. Aouf, F. Ardhuin, F. Collard, G. Engen, S. Hasselmann,
P. Heimbach, P. Janssen, H. Johnsen, H. Krogstad, S. Lehner, J.-G. Li, X.-M. Li,
W. Rosenthal & J. Schulz-Stellenfleth

Satellite Oceanography from the ERS Synthetic Aperture Radar and Radar Altimeter: . . 199
A Brief Review
J.A. Johannessen, B. Chapron, W. Alpers, F. Collard, P. Cipollini, A. Liu, J. Horstmann,
J. da Silva, M. Portabella, I.S. Robinson, B. Holt, C. Wackerman & P. Vachon

New Views of Earth from ERS Satellite Altimetry . . . . . . . . . . . . . . . . . 225

O.B. Andersen, P. Knudsen, P. Berry, R. Smith, F. Rémy, T. Flament, S. Calmant, P. Cipollini,
S. Laxon, A. Shephard, D. McAdoo, R. Scharroo, W.  Smith, D. Sandwell, P. Schaeffer,
J. Bamber, C.K. Shum, Y. Yi & J. Benveniste

20 Years of ERS Interferometry . . . . . . . . . . . . . . . . . . . . . . . 255

F. Rocca

ERS SAR over Land . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273

T. Le Toan

Acronyms and Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . 291

v
→→ Introduction
 Introduction

Introduction

The first European Remote Sensing (ERS‑1) satellite was launched in July 1991
and was followed by ERS‑2 in April 1995. Carrying suites of sophisticated
instruments to study the complexities of the atmosphere, land, oceans and
polar caps, these two missions were the most advanced of their time. Over 20
years the two satellites delivered a continuous stream of data and information
that changed our view of the world and placed Europe firmly at the forefront of
Earth observation.
In September 2011 ESA organised a scientific workshop to celebrate the
outstanding achievements of the ERS mission after 20 years of successful
operations. The main themes of the discussions are captured in each of the
chapters of this publication. Comprising contributions from key scientists
and their colleagues involved in the mission exploitation, this publication
highlights the outstanding scientific achievements unique to the ERS
mission. The authors also provide their personal perspectives on the mission,
addressing the human angle of the mission preparation and of its scientific
exploitation, as well as recalling successes and wider impacts resulting from
the unique achievements of the ERS mission and its exploitation.
The first chapter deals with the history of ERS from its conception in the
1970s, through the hectic days of the launches, the experiments that paved the
way for new research, the adaptations of ESA data policy and the deorbiting
phase. In particular it recalls the proactive role of ESA mission management for
the provision of data to the science community, the deployment of a worldwide
network of ground stations and early steps in international cooperation, as
well as the support to campaigns like the first ever ERS‑1/2 tandem campaign
and pilot projects for precursor applications demonstration.
The second chapter focuses on the Along-Track Scanning Radiometer
instruments, ATSR and ATSR-2, which were capable of measuring land and
sea surface temperatures with unprecedented accuracy. The first Sea Surface
Temperature measurements from ATSR proved to be better than actual buoy
measurements. The scientific exploitation results extended to global land
surface temperatures and cloud or aerosol measurements. ESA initiated an
innovative exploitation of the ATSR infrared channel properties to produce a
regularly updated world fire atlas. Thanks to the ERS ATSR/ATSR-2 instrument
data, we have a unique and continuous set of Sea Surface Temperature
measurements going back to 1991, which is used today in climate research.
The third chapter discusses the history of the development of the Global
Ozone Monitoring Experiment (GOME), including exploitation results. GOME
was the only new instrument on ERS‑2 as compared to the ERS‑1 payload,
adding atmospheric chemistry monitoring capabilities. Having a high spectral
resolution and broad-wavelength coverage, GOME measured ozone, but also
known ozone-depleting substances such as chlorine and bromine, clouds and
aerosols, and chemically reactive factors of air quality such as nitrogen and
sulphur dioxide. Based on the success of GOME, follow-on instruments such
as GOME-2 on MetOp and the Ozone Monitoring Instrument (OMI) on the Earth
Observing System satellite Aura have been realised, which are extending the
provision of long-term ozone data for the World Meteorological Organization’s
Scientific Assessment of Ozone Depletion. In addition, GOME has enabled the
development of air-quality applications based on satellite data.
The fourth chapter illustrates the contribution to science and operations
over land, sea and ice by observations made with the ERS Active Microwave
Instrument (AMI) scatterometer mode. The ERS scatterometer measures the
speed and direction of winds over the surface of the ocean. This information
is used for operational weather and wave forecasting, as well as climate
research. Other applications of ERS scatterometer backscatter measurements
include sea ice age, extent and melt. Over land the scatterometer is being used

3
SP-1326

operationally to retrieve soil moisture and soil water index globally. Follow-on
instruments are now carried on the European operational meteorology MetOp
satellites operated by Eumetsat.
The impact of the information provided by the ERS AMI wave mode on
oceanography, meteorology and climate research is discussed in the fifth
chapter. This chapter describes the motivation for incorporating the Synthetic
Aperture Radar (SAR) wave mode in the ERS‑1 instrument, in particular
thanks to its capability to image globally ocean waves and to the development
of numerical two-dimensional wave spectral models. The SAR wave mode
has become an established component of operational wave prediction and
has been continued on follow-on satellites such as Envisat and Sentinel-1.
Finally, the recent cross-spectral analysis of SAR wave data has allowed major
improvements in wave product inversion and subsequently a new application
on swell tracking has been demonstrated for operational use.
The sixth chapter reports some of the advances in studies of the upper
ocean, its monitoring and understanding with the use of ERS AMI Image Mode
and Radar Altimeter. SAR imaging over the ocean has enabled significant
advances in physical oceanography such as the observation of mesoscale
fronts and eddies, monitoring and understanding of internal waves, the
detection of oil spills and the measuring of wind speed. This period saw the
development of the first operational services in near-realtime, such as oil spill
monitoring and ship detection in the Norwegian Sea.
The ERS altimetric mission objective was to map global ocean wave heights,
roughness, ocean circulation and associated transfers of energy including
ocean–atmosphere interactions. With its 35-day repeat orbit and therefore
denser ground tracks, the ERS mission was an indispensable complement to
the NASA–CNES Topex–Poseidon altimetric mission, bringing high-resolution
mesoscale variability into the picture.
The seventh chapter illustrates the contribution of the Radar Altimeter to
geodetic research, ice and sea-ice studies, improvements in Digital Elevation
Models (DEMs), coastal oceanography, and river and lake hydrology. While ERS‑2
was in the final stage of development, the scientific community requested that
ERS‑1 be flown in a very dense orbit. The so-called geodetic phase, which lasted
from April 1994 to March 1995, made it possible to map the mean sea surface and
derive the marine geoid and seafloor bathymetry at unprecedented resolution
and accuracy. The ERS‑1 geodetic phase data acquired over land were used to
correct existing DEMs and led to the release of a new global and accurate DEM
called Altimetry Corrected Elevation (ACE). A new application emerged, based on
the capability to extract river and lake levels globally and in particular in regions
that are difficult access and poorly monitored. Ice sheet topography was a key
target of the ERS mission and was mapped with unprecedented coverage and
accuracy. Finally, the level of the sea ice above the water level could be estimated
and sea surface topography measurements extended in the coastal zone.
The eighth chapter explains the steps that have led to the birth of modern
SAR interferometry. Although not known at the time, and therefore not
specified in the ERS‑1 imaging radar development, scientists discovered that
it was possible to exploit the phase difference in radar pixels from consecutive
overpasses, which could be directly related to terrain height. ESA teams
developed a new processor to generate a new phase-preserved product,
organised a fringe group of science users, distributed new products widely and
set up proof-of-concept validation campaigns.
Today, SAR interferometry is used in most imaging radar applications
such as land cover classification, DEMs, surface deformation, earthquake
and volcano monitoring and ice flow velocity measurements. In addition,
the property of the radar over specific point targets to preserve the phase
information allowed measurements of local subsidence of a few millimetres.
Value-adding companies processed entire datasets from the ERS archive and
generated subsidence maps at regional and national scales.

4
 Introduction

In the ninth chapter the authors highlight the value of the ERS SAR
compared with optical imaging. SAR can image in all weather conditions,
irrespective of day or night, and is particularly useful for agriculture/forest
monitoring and snow and ice studies in the polar regions. In addition, the
ERS SAR demonstrated particular sensitivity to surface roughness, allowing
geological or topography mapping. Also sensitive to dielectric properties,
it allowed for soil moisture or biomass mapping. New applications such as
rice crop mapping and tropical forest monitoring were also developed using
multitemporal ERS imagery. Finally, combining radar interferometry and the
measurement of radar reflectivity allowed for biomass retrieval at regional
scales, such as the boreal forest Siberia project or the Chinese forest under the
Dragon cooperation.
In conclusion, we invite readers to evaluate whether ERS has fulfilled the
objective set at the origin of the mission: ‘ERS‑1 will be the European Space
Agency's first satellite devoted entirely to remote sensing from a polar orbit.
It will be a forerunner of a new generation of space missions planned for the
1990s which will make a substantial contribution to the scientific study of our
environment.’
Our opinion is a clear YES!
The contributions of the ERS missions to science, applications and
operational services is well illustrated by the more than 2000 scientific papers
exploiting the ERS data that have been published so far. The ERS archive is
freely and openly available for further exploitation – from science to climate
research – thus ensuring that the success story will continue.
We wish you all pleasant reading and look forward to the opportunity to
meet you at the next ESA Living Planet Symposium.

The organisers of the scientific workshop '20 years of ERS missions'

5
→→ History of the ERS Programme
 History of ERS

History of the ERS Programme

G. Duchossois1, G. Kohlhammer2 & W. Lengert3

1 ESA, ERS Mission Manager (1984–1994)

2 ESA, ERS Mission Manager (1994–2001)
3 ESA, ERS Mission Manager (2001–present)

1. The Emergence of Earth Observation at ESRO/ESA

in the 1970s
In 1971 a group of European experts from eight ESRO Member States was invited
by NASA to participate in an international colloquium on remote sensing
(3−14 May 1971) and in the 7th ERIM symposium at the Environmental Research
Institute of Michigan in Ann Arbor (17−24 May 1971). In their mission report to
the ESRO Director General, the group defined the requirements for a European
remote sensing programme and formulated a number of recommendations on
the possible roles of ESRO, including:

—— coordinating national activities at various levels in the ESRO Member States;

—— developing and implementing airborne remote sensing facilities in Europe
based on existing national capacities;
—— developing European reception facilities for the acquisition of data from the
Earth Resources Technology Satellite (ERTS-A) being developed by NASA and
planned for launch in 1972, and possible follow-on missions;
—— preparing and delivering information to the still embryonic community
of civilian users in Europe at that time (most existing expertise was in the
domain of military applications); and
—— defining and preparing a European space remote sensing programme
focusing on resources management, environmental monitoring and the
delivery of assistance to developing/emerging countries.

During the period 1972−79 ESRO/ESA carried out intensive exploratory

activities, including consultations with experts/advisory groups in existing
and potential areas of application, while European industry conducted studies
focusing on:

—— candidate space-borne remote sensing instrumentation, focusing on passive

microwave (Passive Microwave Radiometer Satellite, PAMIRASAT, in 1974)
and active microwave techniques (Synthetic Aperture Radar Satellite,
SARsat, in 1973; the Synthetic Aperture Radar on Spacelab, SARLab, in 1974;
followed by two dedicated SAR Phase A studies performed by Thomson-CSF
and Marconi Research Labs in 1977);
—— an inventory of European airborne facilities with studies performed by
the Société d’Études Techniques et d’Entreprises Générales (SODETEG) in
association with the French Institut Géographique National and Bureau de
Recherches Géologiques et Minières (IGN/BRGM), and EASAMS in association
with Fairey Air Survey, UK; and
—— the possible use of Spacelab as an intermediate step prior to the development
of free-flying remote sensing satellites (references to the post-Apollo group of
experts, the 1973 symposium on the use of Spacelab, the ESRO 1973 Summer
School in Nottingham, UK, etc.).

On the basis of the results of these consultations and industry studies, bodies
such as the Remote Sensing Ad Hoc Group (RSAHG) and its successor, the

9
SP-1326

Remote Sensing Programme Advisory Group (RESPAG), identified three main

priorities and possible lines of action:

—— In the short term (mid-1970s), the creation of a European network for the
acquisition, preprocessing, archiving and distribution of data from US/NASA
satellites such as Landsat, the Heat Capacity Mapping Mission (HCMM),
Nimbus-G and Seasat, based on existing national capacities. To this end, the
EarthNet programme was approved in 1977 and implemented based on the
use and upgrading of several European receiving stations (Fucino, Kiruna,
Oakhanger, Maspalomas, Lannion, etc.) and of the ESRIN centre for the
management of the programme. The creation of the European Association of
Remote Sensing Laboratories (EARSeL) in 1976 (chaired by Professor Preben
Gudmandsen of the Technical University of Denmark at Lyngby), under the
auspices of the European Commission, ESA and the Council of Europe, was
also a major step in the development of the community of scientific users in
Europe. Today, EARSeL members include more than 200 research institutes,
laboratories and universities in Western and Eastern Europe, the Middle East
and North Africa.

—— In the medium term (early 1980s), the use of Spacelab for the testing and
demonstration, as a priority, of high-resolution optical imaging sensors and of
all-weather remote sensing techniques. Germany proposed two experiments
for the First Spacelab Flight Payload (FSLP): the Metric Camera, a modified
Zeiss RMK A30/23 airborne camera, for cartographic applications and the
production of maps at scales down to 1/50 000, and the Microwave Remote
Sensing Experiment (MRSE) combining several modes of operation (SAR,
Wind Scatterometer and passive radiometer modes). The FSLP was flown on
28 November 1983; the Metric Camera was a full success but unfortunately
the MRSE SAR and Scatterometer modes failed.

—— In the longer term (late 1980s/early 1990s), the preparation of a European

space satellite programme focusing on all-weather techniques to account
for cloud cover conditions over northern Europe as well as day and night
operation capability. Particular attention was paid to the development of a
complex Synthetic Aperture Radar (SAR) instrument for which, following
numerous consultations and studies with European and American expert
groups, the C‑band frequency was preferred to the X- or L-bands.

2. The Genesis of ERS‑1: Mission Objectives and Payload

Industry studies performed in the late 1970s concentrated on the definition
of candidate satellite missions that would focus on land (Land Application
Satellite System, LASS), the open ocean (Global Ocean Monitoring Satellite
System, GOMSS) and coastal zones (Coastal Ocean Monitoring Satellite System,
COMSS), all of which included SAR instrumentation. These studies highlighted
the need for a Remote Sensing Preparatory Programme (RSPP) to develop the
critical technologies and techniques required for such ambitious projects
(optical and microwave technologies, very high data rate transmission, ground
data processing algorithms, fast processors, etc.). ESA Member States (and
Canada through a Cooperation Agreement) approved the RSPP in March 1979
for an initial period of two years, which was later extended to the end of 1981.
From consultations with the advisory groups, it appeared that most of the
requirements for GOMSS could be covered by the COMSS concept, so that,
as a consequence, only the LASS and COMSS Phase A studies proceeded on
the basis of instrument payloads comprising SAR and an Optical Imaging
Instrument (OII) for LASS, and an Ocean Colour Monitor (OCM) and a passive
Imaging Microwave Radiometer (IMR) for COMSS.

10
 History of ERS

An important element in the decision process for the ESA Remote Sensing
Satellite Programme was the launch of Seasat in July 1978, which provided, in
spite of its short lifetime (106 days), large amounts of data of great interest to the
global oceanographic community. The Seasat results encouraged consideration
of alternative COMSS payloads, and follow-on trade-off studies were performed
in 1980 to consider the inclusion of Radar Altimeter, Wind Scatterometer
and Earth radiation budget instruments. A high-level advisory committee
of European experts was set up in early 1981, chaired by John Houghton
(Director of the Rutherford and Appleton Laboratories). It was charged (by Eric
Quistgaard, Director General of ESA) with reviewing the mission objectives
and proposing a sound payload configuration for the new mission concept,
called ERS‑1, resulting from the merger of LASS and COMSS as it appeared that
a single mission could cover most land, ice and ocean objectives (see Table 1).
It is important to recall that the initial mission objectives defined in
the 1970s had emphasised commercial and operational exploitation and
applications of remote sensing. However, a gradual change occurred with the
emergence of global climate and ocean monitoring programmes (such as the
World Climate Research Programme, the World Ocean Circulation Experiment
and the Tropical Ocean Global Atmosphere programme) to respond to the
growing concerns of the public, decision makers and politicians about climate
variability processes and possible human interactions and contributions,
which required the development of an improved scientific understanding of the
climate system. This rebalancing of requirements from commercially oriented
applications to scientific research became effective in the early 1980s when the
ERS‑1 mission configuration and objectives were finalised as follows:

—— To increase scientific understanding of coastal zones, global ocean and polar

region processes and to provide support for government policies related to
the environment and climate change.

—— To develop and promote economic and commercial applications considering

ERS‑1 as a demonstration mission and a ‘market opener’ for downstream/
value-added industries.

—— To explore the potential of radar data for studies of land processes and
applications.

Table 1. Comparison of ERS‑1 and Seasat payloads (ESA, 1997b).

Observation targets Seasat ERS‑1

All-weather high-resolution imagery over
land and ice and global ocean wave spectra
Global ocean surface wind speed and
direction
Global sea surface temperatures
Global observations of land and water
features and identification of clouds
Global ocean wave heights and sea surface
topography
Global tropospheric water vapour content
Precise orbit determination
SAR L-band, HH polarisation
Microwave scatterometer
(Ku-band)
Scanning Multichannel Microwave Radiometer
(SMMR)
Visible and infrared radiometer
Radar altimeter
(Ku-band)
Combined use of SMMR and Visible Infrared
Radiometer
Laser Retroreflector array
SAR C-band, VV polarisation
Microwave scatterometer
(C-Band)
Along-Track Scanning Radiometer (ATSR)
ATSR
Radar altimeter
(Ku-band)
Dual-frequency microwave sounder
Precise Range and Range Rate Equipment
(PRARE) and Laser Retroreflector array

11
SP-1326

Figure 1. Evolution of the ERS payload

during Phase A. (ESA, 1997b; R. Francis)

In view of the technical payload accommodation constraints imposed by the

use of the multi-mission Système pour l’Observation de la Terre (SPOT) platform
(proposed by France as a ‘contribution in kind’), the financial implications and
the scientific merits of the various payload options, the advisory committee
proposed a payload configuration that no longer included an OCM and an IMR,
but consisted of:

—— An Active Microwave Instrument (AMI) including SAR and Wind Scatterometer

modes, both operating in the C-band (5.3 GHz) and using common timesharing
electronic hardware. The SAR mode was also designed to operate as a Wave
Scatterometer over the open ocean (SAR Wave mode) and as a high-resolution
imaging radar (SAR imaging mode) over land, coastal zones and ice.

—— A Radar Altimeter (RA) operating in the Ku-band.

To take advantage of the onboard capacities made available with the

elimination of the OCM and IMR, ESA released an Announcement of
Opportunity (AO), from which three instruments were selected. These three
nationally funded instruments included the Along-Track Scanning Radiometer
(ATSR) provided by the UK and Australia, the Microwave Radiometer (MWR)
provided by France, and the Precise Range and Range Rate and Experiment
(PRARE) provided by Germany.
This configuration was adopted by ESA Member States in 1981 as the basis
for the forthcoming industrial development activities. Also in 1981, a major
airborne campaign, using a Convair 580 provided by the Canada Centre for
Remote Sensing (CCRS), equipped with multi-frequency and multi-polarisation
SARs, was conducted in cooperation with the European Commission’s Joint
Research Council (JRC) over some 52 test sites across Europe, from Sweden
to Spain and Italy. The data collected were made available to the European
scientific community to promote their understanding the potential of SAR data
for various land, ice and ocean applications.
Figure 1 shows the evolution of the ERS payload during Phase A, which
began in 1978, during the operation of the Seasat mission.

3. ERS‑1 Detailed Definition Phase B

In 1981 a major milestone in the development of ERS‑1 was the approval by
10  ESA Member States, plus Canada and Norway, of Phase B (total financial
envelope: 25 million AU1). The Phase B industrial activities began in April 1982,

1 The ESA Accounting Unit (AU) was roughly equal to the European Currency Unit
(ECU). In 1995, 1 AU = 6.58 FF, £0.78, 1.93 DM, 1885 Lira.

12
 History of ERS

carried out by a very large consortium led by Dornier Systems GmbH, Germany,
as the Prime Contractor. The scope of the contract included the establishment of
the space segment specifications (platform and payload), the detailed definition
of the ground segment (payload data management and mission operations), the
continuation of critical technology development and breadboarding activities
(see Table 2) and the definition and realisation of experimental campaigns
to train users and demonstrate the soundness of the instrument concept and
performances required.
The objectives for the ground segment were ambitious, and it was sized
for the delivery a number of selected products within 3 h of observation on
the satellite. An additional condition was that maximum use must be made
of existing national ground facilities for data acquisition, preprocessing and
archiving. Phase B also included the preparation of technical and financial
proposals for Phases C/D/E on a design-to-cost basis.
A major difficulty encountered by the ESA project team at the beginning of
Phase B was the reluctance of industry to make contractual commitments on
the basis of geophysical performance requirements for the ERS‑1 instruments.
This was considered too risky, as it would rely on semi-empirical conversion
algorithms. As a consequence, the project team had to make a major effort
to produce technical and engineering specifications based on the agreed

Table 2. Summary of critical issues at end of ERS‑1 Phase B (ESA, 1997b).

Components Critical issues

Satellite
Active Microwave Instrument
Radar Altimeter
Payload Telemetry Equipment
(Instrument Data Handling and ——
Transmission, IDHT)
Platform
Ground Segment
Assembly, integration and
verification
—— Update of the coupled analysis in the light of work at the subsystem level
—— Detailed definition of the AMI power amplifier radiator
—— Power distribution subsystem optimisation
—— Definition of the titanium elements
—— Development of the payload harness mockup
—— SAR antenna testing procedures and detailed definition facilities
—— Scatterometer antenna: detailed design of hinges and release mechanism and analysis of mechanical/
thermal distortion
—— Multipaction effect for antenna waveguide interfaces
—— Harmonisation and verification of AMI internal interfaces for equipment and unit and cabling design
(in particular waveguides)
—— Procedure for selecting the power amplifier tube (choice between travelling wave tubes and a
klystron) and participation in tests
—— Influence of multipaction effect at the switching matrix level
—— Detailed design of matched filters
—— Optimisation of power requirements
—— Continuation of the development of the pulse generator (chirp)
—— Detailed design of the unit for locking the receiver onto the radar signal (tracker) and of the onboard
signal processor and associated software
—— Adaptation of the ultrastable oscillator specifications to meet the requirements of the Radar Altimeter
—— Continuation of the development of the radar return signal simulator
—— Detailed definition of the format structure to limit transmission of useless words
Mockup of the high-transmission rate interface in order to select appropriate components
—— Evaluation of the results of predevelopment contracts for the X‑band antenna and the QPSK modulator
—— Detailed definition of modifications to be made to the SPOT platform in the onboard control software area
—— Updating of a dynamic mathematical model of the satellite
—— Continuation of work needed to keep the development of the SAR data processing equipment on schedule
—— Monitoring the definition of the onboard software to assess its influence on the definition of the
Ground Segment
—— Detailed definition of the satellite/Electrical Ground Support Equipment (EGSE) interface specifications
—— Definition of the operational requirements of the EGSE and standardisation of the equipment

13
SP-1326

geophysical performance requirements before the industrial contract was

signed. An extension of Phase B to end of March 1984 was agreed (total
financial envelope: 8 million AU) to continue the definition of critical elements
of the programme and for the procurement of the long-lead-time components
(see Table 2).
It should be noted that an early definition of the ERS data policy guidelines
(see Annex C of the ERS‑1 Declaration) had to be elaborated during Phase B at
the request of delegations as a sine qua non condition prior to the final approval
of Phases C/D/E. This has to be seen in the international ‘commercial’ context
of the 1980s, which saw the adoption of the Landsat Commercialisation Act
in the United States, the definition of the French SPOT Data Policy and the
creation of the company Spot Image in July 1982.

4. ERS‑1 Development Phases C/D

In July 1984, Phases C/D/E were approved by 10 ESA Member States plus Canada
and two associated states (Norway and Austria) with a total financial envelope
of 606.63 million AU (502.88 million AU for Phases C/D and 105.75 million AU
for Phase E). The industrial work started in December 1984 with Dornier
Systems GmbH as the Prime Contractor, and involving about 80 firms, mostly
European (and Canadian) aerospace companies (see Fig. 2).

Figure 2. Members of the ERS‑1 industrial consortium during Phases C/D/E (ESA, 1997b).

14
 History of ERS

5. ERS‑1 Challenges
During Phases B/C/D, from 1982 to the launch of ERS‑1 in 1991, in addition
to the detailed definition of the ERS‑1 system, the management of industrial
development, testing activities and technical reviews typical of such an
ambitious project, the ESA project team faced a number of challenges, which
are summarised in the following.

5.1 Critical Technologies

ERS‑1 required the integration of key technologies, some of which were
available at national level within ESA Member States, while others had to be
developed within the programme. One should recall the high complexity of
the instrumentation and the lack of experience/know how in Europe to cope
with this extreme complexity. Some examples of critical technologies/issues
are listed below for each key payload instrument (note that there were also
stringent requirements for the platform subsystems such as the Instrument
Data Handling and Transmission (IDHT) and Low-Bit-Rate (LBR) tape recorders
for the satellite operations such as orbit determination and control, attitude
control and pointing accuracy, etc.):

—— The SAR mode of AMI: SAR antenna manufacturing: a 10 × 1  m slotted

waveguide array made of Carbon-Fibre-Reinforced Plastic (CFRP) with a
metallised inner surface and subdivided into five mechanical panels; major
issues with thermal stability behaviour, the metallisation process, the
deployment sequence, and the requirement that the planarity be better than
1.5  mm; developing and testing the high-power amplifier/Travelling Wave
Tube; the issue of arcing; overcoming the effect of multipaction (vacuum
discharge) for the antenna waveguide interfaces and switching matrix; and
developing a Surface Acoustic Wave (SAW) device for the chirp signal.

—— The Scatterometer mode of the AMI: three-beam scatterometer antenna

design and realisation as aluminium planar slotted waveguide arrays,
each one electrically subdivided into two panels; and stringent elevation
requirements for the sideways-looking mid-beam.

—— Radar Altimeter: stringent requirements for the chirp pulse generator (SAW
device).

—— ATSR: very high blackbody accuracy requirements; active Stirling cycle

cooler to maintain detector stability to support very accurate emissivity;
temperature uniformity; and temperature measurements with an error of less
than 33mK.

The ERS‑1 technical specifications are summarised in Table 3.

The geophysical validation of ERS‑1 derived products (e.g. surface
wind and wave fields, sea surface temperature fields), involving retrieval
algorithms and models (e.g. CMOD for the Wind Scatterometer), has required
the implementation of major dedicated campaigns (in Europe, Canada and
the United States) as well as statistical analysis of data, comparison of data
with meteorological and oceanographic model outputs, and with in situ
observations from ships, buoys and platforms.

15
SP-1326

Table 3. ERS‑1 technical specifications

Parameter Value
(ESA, 1997b).
Orbit
Type Near-circular, polar, Sun-synchronous
Altitude 782−785 km
Period about 100 min
Orbits per day 14.3
Repeat cycle 3-day, 35-day and 176-day
Instruments Active Microwave Instrument comprising SAR (image and wave
modes) and a Wind Scatterometer; Radar Altimeter; Along-
Track Scanning Radiometer; Microwave Radiometer; Precise
Range and Range-rate Equipment; and Laser Retroreflectors
Mass
Total mass 2157.4 kg
Total payload 888.2 kg
Total platform 1257.2 kg
AMI 325.8 kg
Radar Altimeter 96.0 kg
IDHT 74.0 kg
ATSR/M 55.3 kg
PRARE 12.0 kg
Laser Retroreflectors 2.5 kg
Electrical power
Peak power ≤2600 W
Permanent power ≤550 W
Power supply voltage 23−37 V
Onboard energy 2650 Wh max
Altitude and orbit control
Type 3-axis stabilised Earth-pointed
Data Handling
Onboard computer Word length: 16 bits
Payload memory capacity 20 K words max
Communications
Transponder Coherent S-band (2 kbit/s)
Transmission power 50−200 mW max
Telemetry rate 2048 bit/s
Telecommand rate 200 bit/s
Data downlink —— X-band (105 Mbit/s high rate link for SAR image mode)
—— X-band (15 Mbit/s low rate link for realtime and playback
of LBR data
—— Onboard recorders provide 6.5 Gbit storage
—— S-band telemetry links for housekeeping data

5.2 Calibration and Validation Approach

Particular attention was paid to the Calibration and Validation activities,
since these would be critical for the development of science and applications.
A dedicated delegate body, the ERS‑1 Operation Plan Advisory Group (EOPAG)
was set up to deal with ERS‑1commissioning phase activities, the choice of
the orbit scenario and the definition of the High-Level Operation Plan (HLOP).
EOPAG operated from1986 to 1992.
For each payload instrument, a calibration plan, including internal and
external calibration activities and a geophysical (where applicable) validation

16
 History of ERS

plan, was defined and implemented. This required the development of

innovative calibration techniques, including the use of:

—— reference targets such as ground transponders or Active Radar Calibration

(ARC), as well as the use of distributed, stable and homogeneous targets (e.g.
the Amazon rainforest) for the SAR and Scatterometer,
—— satellite laser ranging stations in Austria, Italy, Switzerland and the UK,
—— an oceanographic research platform (near Venice, Italy) for the Radar
Altimeter, and
—— incorporated black bodies and ground targets with known albedo (such as
White Sands, New Mexico) for the ATSR.

The geophysical validation, involving retrieval algorithms and models (e.g.

CMOD for the Wind Scatterometer), the required dedicated campaigns (in
Europe, Canada and the United States for the Scatterometer), statistical
analysis of data, comparison of data with meteorological and oceanographic
model outputs, and in situ observations from ships, buoys and platforms.

5.3 Selection of the Orbit Configuration Scenario

In order to meet the requirements of the various user communities (land, ice,
ocean, coastal zones, geodesy) in terms of revisit frequency, as well as those
of the commissioning phase activities (requiring a fast revisit frequency over
calibration sites), several trade-off studies were performed in particular with
the French Groupe de Recherches en Géodésie Spatiale (GRGS) in Toulouse
and the German Max-Planck Institute for Meteorology (MPI) in Hamburg. The
Programme Board for Remote Sensing (PB-RS) then evaluated the merits of
different repeat cycles and provided recommendations. The orbit configuration
selected included several repeat cycles, as shown in Table 4.
The selected sequence of repeat cycles generated lively discussions,
particularly in relation to the Geodetic Phase E, which would require a long
repeat cycle to allow dense spatial sampling of the marine geoid with the Radar
Altimeter. The concerns about the implications of such a repeat cycle for other
key priority user communities (land and ice in particular.) were reinforced
when a second Geodetic Phase (interleaved with the first) was proposed and
adopted, doubling the spatial sampling density of the geoid (16  km at the
equator). All this meant that the Mission Manager would have a hard time
reconciling the demands of all the user communities, but his task was made
easier when the existence of ‘pseudo-repeat cycles’ within this 168‑day cycle
was demonstrated.

Table 4. ERS‑1 mission phases (ESA, 1997b).

Phase Description Repeat cycle Mean altitude Dates

A Commissioning phase 3 days 785 km July to December 1991
B 1st Ice Phase 3 days 785 km December 1991 to March 1992
C 1st Multidisciplinary Phase 35 days 782 km April 1992 to December 1993
D 2nd Ice Phase 3 days 785 km January to April 1994
E 1st Geodetic Phase 168 days 770 km April to September 1994
F 2nd Geodetic Phase 168 days 770 km April to September 1994
G 2nd Multidisciplinary Phase 35 days 782 km April 1995 to end of mission

17
SP-1326

5.4 Definition of the High-Level Operation Plan

The definition of the High-Level Operation Plan for ERS‑1 was also an important
and difficult task for the definition of the instrument operation priorities, in
particular for the AMI, for which the SAR and the Wind Scatterometer modes
were mutually exclusive. Additional constraints that also had to be taken into
account included the maximum and minimum operating times per orbit for
the SAR mode, the maximum number of operations of the AMI switch matrix,
high-priority geographical zones for the SAR mode as a function of the time of
the year, the implications for coastal zone observations, etc.

5.5 Definition of the Overall Ground Segment Concept

The ERS‑1 ground segment included Telemetry, Tracking and Command (TT&C)
operations performed by ESOC, and payload data management by ESRIN. The
ground segment for the payload data handling was composed of two parts: the
Near-Realtime (NRT) part with responsibility for generating and distributing an
agreed set of products (geophysical parameters and SAR imagery) within 3 h of
observation on the satellite, and an offline part for generating more elaborate
products and for archiving and distributing them.
Studies of the NRT component in relation to the downloading, processing
and distribution of LBR tape-recorded data recommended a receiving station
network including Kiruna (Sweden), Maspalomas (Spain), Gatineau (Canada)
and Prince Albert (Canada). For the SAR data, the receiving stations at Kiruna,
Fucino and Maspalomas were selected. For offline activities, four national
Processing and Archiving Facilities (PAFs) were selected: the DFVLR (now the
German Aerospace Center, DLR) at Oberpfaffenhofen, Germany, the Research
Institute for Exploitation of the Sea (IFREMER) in Brest, France, the ASI facility
at Matera, Italy and the Royal Aircraft Establishment (RAE), Farnborough, UK,
each with well-defined responsibilities and mandates.

5.6 Definition and Implementation of Simulation Campaigns

Prior to the launch of ERS‑1, many airborne simulation campaigns related to the
C‑band SAR, the C-band Wind Scatterometer and the Ku-band Radar Altimeter
were carried out to provide samples of data to scientists for detailed analysis
and interpretation, as well as to develop processing algorithms, test calibration
techniques and generate sample products. Several of these campaigns were
funded by and executed in cooperation with the EC’s Joint Research Centre,
the Canada Centre for Remote Sensing and NASA/Jet Propulsion Laboratory
(JPL), such as the Convair 580 campaign over Europe in 1981, the Agriscatt and
Toscane-2 campaigns in 1988, and the airborne C-band campaigns over the
Mediterranean Sea and French Guiana in 1989.

5.7 Development of User Communities

ESA devoted considerable effort and resources to the development of user
communities, which in the early 1980s were still embryonic and not well
structured. These efforts included:

—— The release in 1986 of an AO for scientific experiments and application pilot

projects generated high interest and 140 replies, with 276 individual proposals
covering all ERS research themes and applications. More than 100 Principal
Investigators (PIs) were selected, each representing a team of scientists,
some of them very large. The team of the Programme for International Polar

18
 History of ERS

Ocean Research (PIPOR), for example, included several dozen polar ocean
and ice scientists. The PIs held their first meeting at ESRIN in May 1988
and an organisation had to be in place to manage such a large number of
scientists from all continents (Coordinating Investigators were introduced
for each theme and/or geographical region). The PIs played a key role in
the development of the user communities and in raising awareness of ESA’s
Earth observation (EO) programmes and the ERS in particular. Thousands
of peer-reviewed publications resulted from the AO process, which was
repeated several times during the lifetimes of ERS‑1 and ERS‑2, with a variety
of objectives, ranging from scientific research, application demonstration
projects to the exploitation of ERS‑1/ERS‑2 tandem data collected following
the launch of ERS‑2 in 1995. Regular ERS symposia were organised and
attracted increasing numbers of participants, from 400 at the first meeting in
Cannes, France, in 1992, to 1200 at the final symposium in Bergen, Norway,
in June 2010.

—— ESA set up instruments and product advisory expert groups for each ESA-
funded instrument (SAR, Wind Scatterometer, Radar Altimeter and a data
teams to contribute to the definition and preparation of sensor calibration
and product validation activities. These groups brought together the best
European and Canadian experts, who contributed to the definition of
processing algorithms to generate the agreed geophysical parameters derived
from engineering data. They also assisted the project teams in the definition
of airborne campaigns to support the calibration and validation activities.

—— EARSeL members (>200 laboratories and institutes) played an active role in

expanding the user communities by organising specialised studies financed
by ESA, as well as various workshops and symposia.

—— The creation of the European Association of Remote Sensing Companies

(EARSC) in 1989, a network of value-added companies and small and medium
enterprises (SMEs) provided an impetus for the development of operational/
commercial applications of ERS data. The EARSC now has 65 members.

—— ESA launched a range of activities to promote and raise awareness of the

ERS programme, including presentations at important events, conferences,
symposia, specialist workshops and thematic application events. ESA’s
cooperation with foreign countries and/or international organisations
provided opportunities to promote the ERS programme.

5.8 Definition of the ERS‑1 Data Policy

In 1998 the Programme Board for Remote Sensing set up a working group,
chaired by Marianne von Glehn, the Swedish delegate, to define the ERS‑1 data
policy. This was a major undertaking in view of the divergent views among the
delegates on the approach (science versus commercial) to be adopted regarding
the conditions of access to and distribution of ERS‑1 SAR and Low-Bit-Rate
data to users, and direct access to SAR data by national (from participating
states) and foreign ground stations. The policy that was finally adopted
in 1990 (after two years of discussion!) was complex, based on an open and
non-discriminatory approach and identified ‘user categories’ with specific
conditions in terms of pricing policy. Fortunately, this approach was later
simplified for Envisat, and was based on ‘usage categories’.

19
SP-1326

Figure 3. The launch of ERS‑1 (V44) on

17 July 1991. (ESA/Sigma, Noges)

6. ERS Launch and Early Exploitation Phase E

The ERS‑1 Launch Readiness Review took place in early 1990 but following
the failure of flight V36, Arianespace announced a delay and fixed a new
launch date in May 1991. Further problems with the third stage of Ariane 4
forced a rescheduling of the launch for 16 July 1991. ERS‑1 was launched at
the opening of the 10 min launch window, at 01:46 GMT on 17 July 1991 (Fig. 3)
and was delivered with high accuracy to the planned orbit so that only 10 kg
of hydrazine was required for final orbit acquisition, leaving a healthy fuel
margin to support a long lifetime (recall that ERS‑1 was designed for a nominal
two-year lifetime, with consumables sized for three years).
This section highlights some early results obtained with ERS‑1 data a few
weeks/months after launch when still in the commissioning phase. More
detailed examples of results are presented elsewhere in this volume.
The first SAR images of Spitsbergen and the Frisian islands, the Netherlands
(Fig. 4) were acquired on 27 July 1991, 10 days after launch and the successful
switch-on of all instruments (except PRARE, which failed after five days over
the South Atlantic Anomaly).
The first ERS‑1 image, after processing, was blurred and was not what was
expected (Fig.  5). The engineers at ESRIN, with remote support of the Prime

20
 History of ERS

Figure 4. One of the first SAR images

acquired by ERS‑1, covering the calibration
site of Flevoland, the Netherlands,
on 27 July 1991. (ESA)

Figure 5. The first ERS‑1 image with

inverted I/Q assignments, with original
annotation. (ESA)

Contractor of the ERS‑1 Ground Segment and the Canadian company MDA
that developed the processor, quickly identified that the data from ERS‑1 were
perfect, but that the I/Q assignments of the processor needed to be inverted.
MDA (Paul Lim) performed this remotely and less than 48 h after acquisition of
the first data, an image with perfect quality was processed.
Although the original mission requirements for ERS‑1 did not include
the provision of an interferometry capability, the high performance and
engineering quality of the ERS‑1 system allowed the development of this

21
SP-1326

Figure 6. The first interferograms and unwrapped phase of the Gennargentu mountains, Sardinia, from ERS‑1 (left) and ERS‑2 (right). (ESA/POLIMI)

Figure 7. Topographical map (top) and

3D view (bottom) of Alaska obtained from
ERS‑1 SAR data with SAR interferometric
methods. These images show the
extraordinary power of ERS‑1 to extract
topographical information from SAR data.
(C. Werner and R. Goldstein, ETH Zurich,
and JPL, Pasadena, CA, USA)

22
 History of ERS

technique, as reported by Fabio Rocca elsewhere in this volume. Philipp Hartl

and his team at the Italian Institute for Navigation (INS) and the Polytechnic
University of Milan (POLIMI) performed a ‘blind’ experiment using a set of
corner reflectors, some having been moved up between successive ERS‑1
passes. The interferometric technique allowed the detection of small,
centimetre-scale changes in the position of a few selected corner reflectors.
The following pages show a number of interesting results obtained by ERS.
Figure 8 shows a 10-km long oil slick detected off the coast of the South of
France near St Tropez, on 19 September 1991.
Figure 9 shows high-wave storm events detected by the ERS‑1 Radar Altimeter
in August 1991. The storm off the coast of South Africa caused serious damage to
giant oil tankers, which were forced to stop in Durban harbour for repair.

Figure 8. Image of an oil slick off the coast

of St Tropez, France, 19 December 1991.
(ESA)

Figure 9. Smoothed Significant Wave

Heights (SWH) from ERS‑1 orbits 208−251,
with ice edges added from a discriminator
using RA data, 1−3 August 1991. (ESA)

23
SP-1326
Figure 10. Geoid derived from ERS‑1 data.
(ESA/GFZ)
Figure 11. ATSR image of oil well fires in
Kuwait, 1990. (ESA/RAL)
Figure 12. Improvements in ECMWF
weather forecasts since 1981. (ECMWF)
Figure 13. Pictures of ERS‑1 taken from
SPOT-4. (CNES/SpotImage)
24
An improved marine geoid derived from ERS‑1 data (2 × 168-day cycle) is
shown in Fig. 10. This is not an early result but the generation of an improved
gravity field resulting from the exploitation of the two geodetic phase data
which led to the declassification of the US Navy (classified) marine geoid.
Figure 11 presents one of the first results provided by the ATSR, showing oil
well fires in Kuwait during the first Gulf War in 1990.
Wind scatterometer data are now used for weather forecasts, in cooperation
with the European Centre for Medium-Range Weather Forecasts (ECMWF).
Figure 12 shows the improvements in these forecasts, particularly for the
southern hemisphere.
Figure 13 shows an image of ERS‑1 taken from the French SPOT-4 satellite
over Ténéré in the Sahara.
Figure 14 shows a 3D image of the 1997−98 El Niño, created from simultaneous
ATSR and Radar Altimeter data acquired by ERS‑2 in November 1997. This El Niño
had very strong impacts worldwide, with increased rainfall causing flooding in
the United States and Peru, and drought in the western Pacific. This 3D image of
the Pacific shows the deviations from the ‘normal’ ocean state, with measured
changes in Sea Surface Height of between −40 cm and +40 cm and temperature
changes of −6°C (blue) to +8°C (red).
In September 2002, measurements of total ozone by the Global Ozone
Monitoring Experiment (GOME) on ERS‑2 showed an atypical split in the ozone
hole over the Antarctic due to the unusual meteorological conditions in that
year (see Fig. 15). GOME monitored the ozone layer and other atmospheric trace
gases for 16 years, and led the way in the use of space instrumentation for air
quality monitoring.
 History of ERS

Figure 14. 3D image of sea level and temperature anomalies in the Pacific during the 1997−98 El Niño event, based on simultaneous ATSR
and Radar Altimeter data acquired by ERS‑2 in November 1997. (ESA/CEOS)

Figure 15. An atypical split in the ozone

hole over the Antarctic, measured by
the GOME instrument on ERS‑2 in
September 2002. (ESA/DLR)

7. ERS‑1 Routine Exploitation

The ERS‑1 exploitation phase continued until the failure of the satellite on
10 March 2000 after nine years in orbit (45 000 orbits and 1.5 million individual
SAR scenes collected). In 1992 ESA and the ERS Consortium (which included
Spot Image, Eurimage and Radarsat International) reached agreement on ways
to promote and distribute ERS‑1 products, in which ESA retained responsibility
for the delivery of products to selected PIs free of charge. During that period
ESA released a number of AOs; the second AO in 1993 for application pilot
projects, for example, led to the selection of 80 projects. ERS symposia were

25
SP-1326

Figure 16. The global network of ERS receiving stations. (ESA)

held in Hamburg in 1993 (600 participants), in Toledo in 1994 (for application

projects only – 250 participants), in London in 1995 (again for application projects
only – 250 participants), and Florence in 1997 (700 participants).
In parallel, ESA negotiated numerous agreements for the acquisition
of ERS‑1 SAR data with Asian countries (Japan, China, Taiwan, Singapore,
Indonesia, Thailand), Africa (South Africa, Kenya), South America (Argentina,
Ecuador), Antarctica (Syowa, McMurdo, O’Higgins), the Unites States and
Israel. It also entered into international agreements regarding the deployment
of mobile/transportable stations (in Gabon, Mongolia and Bangladesh), and
for upgrading some 20 receiving stations (see Fig. 16). ESA also organised
major events, such as the Euro‑Asian Space week in Singapore and the Euro-
Latin America Space week in Mexico, to extend the use of ERS data and raise
awareness of ESA’s EO programmes.
ESA, either alone or in association with UN agencies, has organised
numerous education and training courses in developing countries and at
ESRIN, and has collaborated with international organisations, in particular
the UN Educational, Scientific and Cultural Organization (UNESCO), the
UN Environment Programme (UNEP), the UN International Drug Control
Programme (UNDCP), the UN Office for Outer Space Affairs (UNOOSA), the
World Meteorological Organization (WMO), the UN Food and Agriculture
Organization (FAO), the Intergovernmental Oceanographic Commission (IOC),
as well as the World Bank and the Asian and African Development Banks (the
Asian Development Bank, for example, provided funding for the upgrading of
the Bangkok receiving station). Such cooperation has greatly helped to promote
and extend the use of ERS data and products.

8. ERS‑2 Approval and Launch

In 1988 ESA felt the need for a second flight unit, ERS‑2, that would ensure the
continuity of ERS‑1 data, both for scientific research (e.g. for detecting small
climate changes) and for the development of applications. Past experience had
shown that it takes 5−10 years to develop operational application services. At that

26
 History of ERS

time, there were also increasing concerns within the scientific community about
ozone depletion and global warming. In November 1988, ESA consulted European
scientists and invited them to propose an atmospheric chemistry instrument
which would be embarked on ERS‑2. The selected proposal, by John Burrows and
Paul Crutzen (who was awarded the Nobel Prize for Chemistry in 1995), was called
‘SCIA-mini’, derived from the Scanning Imaging Absorption Spectrometer for
Atmospheric Chartography (SCIAMACHY) instrument selected for Envisat.
The instrument was further simplified to meet the ERS‑2 platform resource
constraints and was renamed the Global Ozone Monitoring Experiment (GOME).
The GOME Phase B started in July 1989 and the development phase in April 1992,
under the stringent condition that the addition of GOME should not delay the
development of the main system. Also, GOME should be accommodated within
the limited ERS‑2 system margins, and the project management and system
engineering should be provided by ESTEC staff. All of these conditions were
met and GOME was ready for launch three years later. This extremely short
development time represents outstanding performance for such sophisticated
space instrumentation. The GOME ground segment, including a GOME data
processing unit, was developed, implemented and operated by the German
Processing and Archiving Facility (PAF) with national funding.
In spring 1990 the Programme Board for Earth Observation (PB-EO)
approved the ERS‑2 programme budget, which totalled 440.33 million AU,

Figure 17. The launch of ERS‑2 on

21 April 1995 (Ariane 4/flight V72).
(ESA/Service Optique CSG, M. Hucteau)

27
SP-1326

including 370.98 million AU for Phases C/D and 69.35 million AU for Phase E.
With the addition of GOME for stratospheric ozone measurements, the ERS‑2
payload was further improved with the ATSR-2, with three additional VIS
channels, and PRARE-2, which had been completely redesigned and rebuilt
using Hi-Rel or military standard rather than commercial components. ERS‑2
was launched on 21 April 1995 at 01:44 GMT (Fig. 17).

9. ERS‑1/ERS‑2 Tandem Operations

Towards the end of 1993, it became clear that ERS‑1 would still be technically
operational when ERS‑2 was launched. The concept of ERS‑1/ERS‑2 tandem
operations started to take shape, considering that two identical SAR
instruments would be flying at the same time. The participants in several
ERS‑1 user meetings and symposia expressed their strong interest in the
simultaneous operation of ERS‑1 and ERS‑2 over a period of several months,
possibly up to one year.
To explore the potential benefits of such ERS‑1/ERS‑2 tandem operations
in more detail, ESA organised several consultation meetings with both
scientific and application-oriented users at ESRIN. They highlighted the
many advantages and benefits for both research and applications of flying
two identical sets of instruments in an appropriate orbital configuration;
in particular, a ‘unique SAR dataset’ could be collected for interferometry
applications (Fig. 18).
Following difficult negotiations with delegations (the extra operations
would cost some 5 million AU), on 23 March 1995, the ESA Council adopted a
resolution on ERS‑1/ERS‑2 tandem operations that would ‘keep ERS‑1 alive’
until April 1997 and allow such operations for a nominal duration of nine
months, beginning at the end of the ERS‑2 Commissioning Phase in autumn
1995 to April 1996 (this initial period was further extended in view of the
outstanding results obtained).
A primary objective of the tandem operation phase was to acquire, as
quickly as possible, complete Earth coverage with ERS‑1 and ERS‑2 SAR data
within the visibility of the existing ERS‑1 ground receiving stations that had
responded positively to this request. The collection of a unique tandem
dataset was considered to be an asset for the development of interferometry
applications such as Digital Elevation Models (DEMs) and differential

Figure 18. ERS‑1/ERS‑2 tandem orbital

configuration. (ESA)

28
 History of ERS

interferometry applications, allowing the detection of subcentimetre

displacements/ground motion.
Just a few days after the launch of ERS‑2, a number of SAR images were
acquired using ERS‑1 and ERS‑2 in tandem mode, observing the same region
with a one-day interval. These data were quickly processed and analysed
by European experts to demonstrate some of the possibilities offered by SAR
interferometry for the generation of DEMs (see Rocca, section 2.4, this volume).
An AO for the exploitation of the tandem dataset led to the selection of
61  projects. The first Fringe workshop on ASAR interferometry was held at
ESRIN in October 1996 and a second in Liège, Belgium, in November 1999.

10. The Failure of ERS-1

The failure of ERS‑1 in January 2000 was caused by a series of failures during
a blind orbit. First, the Central Control Unit (CCU) encountered a memory
block failure, and the satellite automatically took protective measures and
placed itself in fine acquisition mode. Shortly thereafter, one of the piloting
triplet gyros failed. Since this was the second major failure, the CCU took the
strongest protective measure and triggered the transition of the satellite to Safe
Mode. In this configuration, however, the switch-off of the failing piloting gyro
also introduced a switch-off of the electronic unit controlling a pair of gyros,
the failed gyro and the gyro that was in charge of the Attitude and Orbit Control
Subsystem (AOCS) in Safe Mode.
In this situation the satellite had to cope with two recovery measures within
a very short time, where the first measure has not been concluded when the
second measure to go to Safe Mode had already started. This combination of
acting on an uncompleted AOCS change prevented the correct entry into Safe
Mode. In the absence of piloting gyros, the satellite was no longer pointing in
an auto-controlled fashion. Unfortunately, the gyro failures and the attempt
to enter Safe Mode took place during the first blind orbit of the night, and the
situation of the satellite was only discovered at first visibility three orbits later.
At this point the satellite was already rotating around the pitch and roll axis
and the batteries were getting very low. In this configuration the satellite was
no longer controllable. Ground commands to force the satellite to enter Safe
Mode by changing the redundant Safe Mode gyroscope failed several times
because the S-band antenna was moving and could not receive the commands.
When one of the commands to enter Safe Mode was finally recognised,
the solar panel was facing away from the Sun, pointing towards deep space,
preventing the batteries recharging and power was decreasing very quickly.
Despite continued efforts to command the satellite from the ground, ERS‑1 was
unrecoverable.
ERS‑1 generated a wealth of observational data, including about 800  000
radar images acquired in its first three years of operation. The last SAR image
was taken on 7 March 2000, when ERS‑1 had been in service for about 8.5 years,
more than three times its nominal mission lifetime.

11. ERS‑2 Operations without Gyros

In mid-November 1999 the piloting gyroscopes, which ensure stable control
of the ERS‑2 satellite, showed signs of severe degradation. By the end of 2000,
three of the six gyros had stopped functioning and another two were showing
the same telemetry pattern of degradation as those that had failed. A team led
by the Post-Launch Support Office (PLSO/ESTEC), composed of ESOC, ESRIN
and Astrium, the company that provided the platform, elaborated work-around
solutions (‘zero gyro mode’) for piloting the satellite. They allowed the mission
to continue without gyros, while meeting most of the pointing accuracy

29
SP-1326

specifications. In this mode, precise roll and tilt piloting were ensured by the
Digital Earth Sensor and Sun Sensor, while the yaw variation was measured
based on the Doppler extracted from the Active Microwave Instrument Wave
Mode science data. These processed data, giving the yaw deviation over the
orbit, were provided by the ground stations to the Flight Operations Team at
ESOC for corrective command uplinks. The corrections had to be made during
several consecutive passes over Kiruna before they had an effect on the
pointing accuracy.
Thanks to the ERS ground station network, which is still based on
agreements and contracts established at the start of the ERS‑1 mission, a
global Low-Bit-Rate station network 2 was providing Wave Mode data to ESOC
within less than 40 min to be considered for attitude correction. A further
improvement in the attitude accuracy was made in mid-2003, when a new
Scatterometer Near-Realtime processor became operational that was able to
handle a relaxed attitude precision of up to ±2°. The Scatterometer, with its
three beams providing a significantly better yaw angle determination, replaced
the Wave Mode data for attitude control in 2004.
The outstanding success in keeping the satellite operational without
gyros was due to the reliable and timely support of the LBR stations that were
purposely integrated into the overall flight operations setup. However, an
important factor in this success was the availability of the internet, as a robust
infrastructure for global data flow, to support the ‘open-loop’ attitude control.
With this approach, the initial yaw uncertainty caused by the gyro failures
was reduced from ±10° to ±2°, ensuring the full functioning of all instrument
missions. At the end of this recovery exercise, only the interferometry mission
(which required very high pointing accuracy) was affected by the slight attitude
degradation, although it was still possible to use some 50% of the data.

11.1 Loss of the Tape Recorders

The ERS satellites were equipped with two conventional technology recorders
using magnetic tape. The mechanical recording and playback operations of
these recorders led to deterioration of the tape material. In December 2002 the
tape of the first recorder snapped, and six months later, so also did the second.
As a consequence, no onboard recordings could be made and observations
could only be performed within direct station coverage. Therefore, from
July  2003, the ERS‑2 LBR mission continued as a regional mission with an
initial global coverage of just 8%. Within a few weeks the LBR mission scenario
was modified to take maximum advantage of the existing ERS LBR station
network, which was gradually extended over the years to cover 45% of Earth’s
surface (see Fig. 19). Despite this reduced observation coverage, the ERS‑2
LBR data remained useful, in particular because of the special efforts that
were made to ensure full coverage of the North Atlantic, since this has strong
influence on European weather. Because of the long-term stability of the data
and the measurement accuracy, they have been used by the ECMWF in weather
forecasting models.

2 The LBR network includes ESA stations at Kiruna (Sweden), Maspalomas (Spain),
Matera (Italy), Gatineau (Canada) and Prince Albert (Canada); the national stations
O’Higgins (Antarctica), Chetulmal (Mexico) and West Freugh (UK); and foreign stations
in Miami (USA), Beijing (China), Hobart (Australia), Johannesburg (South Africa) and
Cuiaba (Brazil).

30
 History of ERS

Figure 19. The ERS‑2 LBR station network

used to compensate for the loss of the two
tape recorders. (ESA)

11.2 ATSR‑2 Failure

On 5 February 2008, after some 4 billion revolutions, the ATSR‑2 scan mirror
stopped turning following the mechanical failure of the mirror bearings.
For 13  years the instrument had provided high-precision data that met the
specifications with an outstanding stability. For Sea Surface Temperatures,
a retrieval accuracy of 0.1K was achieved thanks to an onboard calibration
system and the instrument’s excellent engineering that allowed the mission to
exceed by four times its designed lifetime.
The Advanced ATSR flown on Envisat was the successor to the ATSR-2
instrument on ERS‑2. The two instruments, observing the same target 28 min
apart, were intensively cross-calibrated, met the Global Climate Observing
System (GCOS) monitoring principles and represented significant assets for
understanding Earth and its evolution in the context of climate change.

11.3 One Mission across Three Platforms

The ERS instruments were the predecessors of the Envisat ASAR, Radar
Altimeter-2, AATSR and SCIAMACHY. The ERS Scatterometer and the successor
to GOME are now flying on MetOp-A. Ensuring long-term data continuity over
more than two decades, the cross-calibration requirements of the GCOS Climate
Monitoring Principle no. 12 (GCOS, 2003) were fulfilled. Throughout these
missions special emphasis has been given to determining and documenting
inter-satellite/instrument biases. Furthermore, the ERS and equivalent Envisat
instrument data were harmonised with respect to format and the same retrieval
algorithm was used. This was particularly important for reprocessing the entire
cross-satellite datasets. Data quality and homogeneity were operationally
assessed with the support of the ECMWF for the Scatterometer and Wave Mode
missions. For the ATSR missions, the data quality was jointly monitored by the
Rutherford Appleton Laboratory (RAL), the instrument provider, the PI team,
and the UK Natural Environment Research Council (NERC), which funded
the ATSR mission. The high quality of the GOME mission data was ensured
by DLR, which was responsible for operational processing and for improving
data retrieval. This led to five reprocessing to derive new trace gases with
progressively smaller error bars.
Long-term quality assurance and documentation were essential elements
of the ERS mission over two decades. For routine operational monitoring,
a consortium of specialists monitored instrument performance on a daily

31
SP-1326

basis. Quality Working Groups (QWGs), composed of European and Canadian

specialists, evaluated the evolution of the instruments, analysed the retrieval
algorithms and defined actions for further improvements. This setup ensured
continuous product evolution beyond the lifetime of the ERS missions. The
products also improved thanks to ESA’s tight working relationships with the
global user community whose members provided inputs at workshops and
symposia. For instruments that were jointly exploited by Eumetsat and the UK’s
Department of Energy and Climate Change, the mechanism of Science Advisory
Groups (SAGs) was used to advance the science and improve and develop
the instrument products. In the case of the Advanced Scatterometer (ASCAT)
on MetOp, for example, the user community, the QWG, the ASCAT SAG and
Eumetsat collaborated in the creation of a Scatterometer Soil Moisture product,
which after 20 years allowed ‘ocean data’ also to be used for land applications.
ERS, one of the longest multi-instrument missions ever flown, has in its
operations scenario also considered the GCOS Climate Monitoring Principles.
With the continuous mission extension approvals by Member States, ERS‑2
had a five-year overlap with ERS‑1, a nine-year overlap with Envisat and a
five-year overlap with MetOp-A for the Scatterometer and GOME‑2 mission.
Before deorbitation, an LBR station was installed in Brazil to allow concurrent
observations by the ERS Scatterometer and the MetOp ASCAT over the large
homogeneous target of the Amazon rainforest covering all three beams, which
contributed to the best possible cross-calibration of these twin instruments.
A further example of the evolution of the mission is described in section 2
of the WMO’s Scientific Assessment of Ozone Depletion, on the analysis of
measurements of the total ozone column retrieved from nadir sounding
instruments (WMO, 2010). The report described the GOME‑1 dataset as the most
reliable and accurate for the analysis of ozone trends for the morning orbit. The
dataset has therefore become the standard in terms of reliability and accuracy.
The SCIAMACHY and GOME‑2 data records are currently being combined
with those of GOME‑1 to provide full global coverage for trend analysis adapted
to GOME‑1.

11.4 ERS‑2–Envisat Tandem Operations

The inspiration to exploit cross-satellite interferometry capabilities came from
the Swiss company Gamma Remote Sensing AG. Envisat was flying 28 min ahead
of ERS‑2, observing the same ground targets. However, the two satellites had a
slightly different centre frequency of 31 MHz. In order to perform interferograms
taking advantage of this short observation interval, a perpendicular baseline of
about 2  km (see Fig. 20) was needed. Since ERS‑2 had ample fuel, and Envisat

Figure 20. ERS‑2−Envisat tandem orbital

configuration. (Gamma Remote Sensing)

32
 History of ERS

had stringent Global Monitoring for Environment and Security (GMES) service
commitments to fulfil, it was obvious that ERS‑2 needed to be manoeuvred to
implement the required baseline at specific observation targets.
Because the interferometry baselines could only be reached via slight
inclination changes, and thus only a range of observation targets within a
range of latitudes could be addressed, the ERS‑2–Envisat tandem mission
was divided into four campaigns. The second and the fourth campaigns were
to address change detection over the intermediate northern latitudes. The
campaigns took place over the following periods:

—— first campaign, September 2007 to February 2008, northern latitudes;

—— second campaign, November 2008 to April 2009, intermediate northern latitudes;
—— third campaign, February 2010 to April 2010, Antarctica;
—— fourth campaign, July 2010 to October 2010, intermediate northern latitudes.

The exploitation of the unique dataset was advanced and led by two Swiss
remote sensing companies, Gamma Remote Sensing AG and Sarmap SA.
The ERS‑2–Envisat tandem mission addressed numerous applications:

—— For cryospheric applications, changes in the location of ice shelf grounding

lines in 1992, 1994 and 2011 were identified and the dynamics of fast-moving
glaciers was studied.

—— Outstanding results were achieved in the derivation of high-resolution DEMs

of low-relief areas (see Fig.  21). The quality of these DEMs is equivalent to
those derived from of the Shuttle Radar Topography Mission (SRTM) flown
in February 2000, although the SRTM data were restricted to latitudes ±60°.

—— ERS‑2, a polar-orbiting satellite, acquired unique datasets that complemented

the SRTM data with respect to coverage and to change detection over certain
areas.

—— Outstanding results were achieved for forest mapping in Siberia, on the

largest scale so far observed.

—— The ERS‑2–Envisat satellite constellation supported new applications for

estimating biomass over low-relief areas.

Figure 21. Low-relief Digital Elevation

Model of the Ob river estuary, Siberia
(70.2N, 75.5E), derived from ERS-2/Envisat
data, 3 October 2007. (U. Wegmueller and
M. Santoro, Gamma Remote Sensing AG; ESA)

33
SP-1326

Figure 22. Synergistic exploitation of ERS‑2

and Envisat. (ESA)

For day-to-day operations, the centralised user service and mission planning
at ESRIN ensured that the synergy between ERS‑2 and Envisat was exploited
to the full. ERS‑2, carrying the predecessor instrumentation, acquired data
where Envisat conflicts could not be resolved. For example, ASAR on Envisat
with its modes and different polarisations created conflicting demands, so that
whenever Envisat Image Mode Swath 2 VV polarisation was requested within
the ERS station coverage, then that task was assigned to ERS‑2 while Envisat
addressed different requirements (see Fig. 22).

11.5 ERS‑2 Ice Phase

The ERS‑2 Ice Phase was intended to repeat the 1992 and 1994 ERS‑1 Ice
Phases (see Fig.  23), even though the satellite had already been in space for
16 years with its known and unknown problems. Therefore ERS‑2 could ensure
nominally only reduced duty cycles of some 4−5 min per orbit in order to
ensure the safety of the satellite in view of its degraded power system. During
the Ice Phase, however, the ESOC satellite operations team, in cooperation
with the ESRIN mission planners, optimised and closely monitored the power
utilisation of the satellite, allowing a maximum duty cycle of 10 min during
eclipse periods to avoid depleting the batteries.
The Ice Phase or ‘3-day repeat’ mission started on 10 March 2011, one day
before the disastrous earthquake in Japan. Since the main mission objective
was to observe the cryosphere, no pre-seismic acquisition was performed
over this region. However, ERS‑2, with its fast mission planning capacity,
was rescheduled to make observations of the Sendai area three days after
the earthquake until the end of the mission. The ESOC Flight Dynamics team
initiated special manoeuvres to ensure a minimal baseline of maximum 70 m
with respect to the reference scene. Thanks to the low level of solar activity, the
yaw accuracy was very stable, resulting in good data quality for interferometry
for nearly all passes during the last mission phase.
The Ice Phase could have started earlier (November 2010), but the orbit
change of Envisat in October 2010 bound all the necessary resources and
did not allow an ERS‑2 mission phase change at the time to ensure a better
seasonal overlap with ERS‑1 data. There was, however, a seasonal overlap
between ERS‑2 and ERS‑1 1994 data, which showed excellent results in terms of

34
 History of ERS

Figure 23. ERS‑2–Envisat tandem mission observation targets. (ESA)

Figure 24. Retreat of the grounding line

of the Petermann glacier in Greenland
between 1992 and 2011. (ESA/University of
Leeds/University of Strasbourg)

changes in the grounding lines and in the ice mass balances of major glaciers
over a period of 16 years. Figure 24, derived by A. Shepherd (University of
Leeds) and N. Gourmelen (University of Strasbourg), shows that the grounding
line of the Petermann glacier in Greenland retreated by 0.7–3.4  km between
1992 and 2011.

11.6 ERS‑2 Operations Supporting Data Exploitation

Throughout the two decades of the ERS missions, the Earth observation
environment has changed dramatically. This happened thanks to the
availability of high-quality ERS data, the involvement and integration of the
science community in the mission, as described elsewhere in this volume,
and also to the dramatic improvements in Information Technology (IT).
Since the early 1990s the capacity, cost and size of computers have changed
significantly. At the launch of ERS‑1 the single-look complex image was cut

35
SP-1326

into quarter scenes, since the VAX computer with its external array processors
could not handle the processing of an entire 100 × 100  km frame, and data
were disseminated via computer-compatible tape and later via Exabyte
tape shipment. The small mass-market laptops in use today have far greater
processing power than the SAR processing systems used at the start of the ERS
programme.
Likewise, at the beginning of the ERS‑1 mission in the early 1990s
the requirement to disseminate data 3 h after sensing was an enormous
challenge, for which a dedicated satellite-based data dissemination system,
the Broadband Data Dissemination Network (BDDN), was deployed. Now, in
the time of broadband internet, where mobile phones (LTE3 standard) have the
same data source capacity as EO satellites, the demands and expectations of
users have changed dramatically since the mission was defined.
Since 1995 the ERS‑2 operations scenario gradually evolved and adapted to
the changing environment. The Ground Segment, beside the data acquisition
systems located at the stations, was equipped with off-the-shelf computing
hardware. The ERS‑2 mission operations concept relied strongly on services
available on the internet, as the following two examples illustrate.
First, in 2007, ERS‑2 became a major contributor to the GMES Service
Element Maritime Security Service (MARISS) project. To support this project,
the mission planning procedure was tuned to allow satellite tasking within
less than 8 h after the request was communicated. This service started on a
trial basis thanks to the collaboration with DLR, which had the capacity to
manage the entire NRT service. For this operational service, the Neustrelitz
station (north of Berlin) ensured the provision of highly reliable ERS SAR
data by offering observations up to the Portuguese Atlantic coast and
the Mediterranean Sea, the timely processing of these data (initially at
Oberpfaffenhofen until a local processing system was set up at the Neustrelitz
station), and service support for the national authorities in their role as MARISS
consortium members.
Based on the success of this trial the service was quickly adopted by the
entire ESA SAR station network covering Europe and selected world regions.
In many cases the data were made available to the national authorities via the
internet within less than 15 min after sensing. Despite the fact that systematic
observations of the European coast were foreseen, in some cases fast tasking
and operations were still required. The best performance from a satellite tasking
request to the distribution of data, via the internet, to users was 6  h (mission
planning, satellite tasking, acquisition, data processing and dissemination).
Second, following the earthquake in Sendai, Japan, March 2011, ESA’s
cloud computing infrastructure, which became operational in early 2008,
was extended to Taiwan within 24 h, so that every three days ERS‑2 data over
the Sendai area were provided to the geohazard science community in less
than 3 h. In this ad hoc service setup, the Taiwan station (CSRSR) performed
the acquisition and quickly put the data on ESA’s cloud computing assets, for
which a Content Delivery Network storage facility was set up within a day on
the US West coast. The CSTARS station in Miami performed data processing,
putting the results back on ESA’s ‘virtual archive’, from which the National
Research Council (CNR/IREA) in Italy picked up the lower-level product to
create interferograms. These high-level products provided information on
ground deformation (Fig. 25) triggered by the almost daily aftershocks.
A further example of how the use of the internet allowed global
collaboration in an operational manner is the Scatterometer Near-Realtime
mission that followed the failure of the tape recorders on ERS‑2 in summer 2003.
As described elsewhere in this volume, many X-band stations around the world

3 Long Term Evolution (LTE) standard for wireless communication of high-speed data
for mobile phones and data terminals.

36
 History of ERS

Figure 25. Monitoring of the Japanese earthquake with ERS‑2 operated in a 3-day repeat cycle using the Small Baseline Subset−
Interferometric SAR (SBAS−InSAR) approach. (ESA, SAR data; CNR/IREA, SBAS–InSAR processing; GSI, GPS RINEX data; ARIA team at JPL
and Caltech, GPS data processing).

have acquired, processed and disseminated Scatterometer NRT data for attitude
control, as well as for science exploitation and operational meteorological
support. The priorities of the mission were to provide full observation coverage
of the North Atlantic, and to ensure that meteorological offices received timely,
high-quality data. The available station network with Gatineau, Maspalomas,
Kiruna, West Freugh and Matera stations, acquired nearly all possible passes,
going beyond their contractual commitments and providing large volumes
of overlapping data. The Royal Netherlands Meteorological Institute (KNMI)
used the easy availability of the Scatterometer data via the internet to compile
the products from the station network to form a homogeneous Scatterometer
product. These products were then made available to the science community
and to meteorological offices via the internet in NRT, allowing ERS data to be
considered for assimilation and contributing to improved weather forecasts.
A similar NRT service was set up for the GOME mission by the University of
Bremen, the Belgian Institute for Space Aeronomy (BIRA) and KNMI.
ERS‑1 (ESA) and the world wide web (CERN) were launched in 1991. Since
then the exploitation of mission data has been closely linked to the internet
and the evolution of mass-market IT assets. Today, the science community
is interconnected via the internet, and online storage of large datasets
(petabyte scale) is affordable allowing large-scale data sharing and supporting
interdisciplinary research. One example of this new way of working is the
Group on Earth Observation (GEO) Supersite infrastructure concept illustrated
in Fig. 26. This online infrastructure will contain 20 years of ERS SAR data,
as well as large-scale datasets derived from other space and in situ sources,

37
SP-1326

Figure 26. ERS data within the GEOSS

Supersite infrastructure. (ESA/EPOS/CEOS)

leading to new and outstanding science. These data, which currently represent
the bulk of data assets in this infrastructure, provide the basis for the Global
Earth Observation System of Systems (GEOSS) for monitoring geohazards.
In 2011, scientists’ requests for hundreds of thousands of SAR images
brought about significant changes in the operations concept since the start of
the ERS mission.
With respect to the future, the availability of cloud computing services at
mass market prices (a few cents per hour for processing), in combination with
the ‘open and free’ data policy of the ERS mission will drastically increase
the demand for ERS data, ensuring the continuation of the mission beyond
its active lifetime. Several areas of research require large-scale SAR data
(hundreds of thousands of images) and processing capabilities, including:

—— the quantification of the impacts of climate change and the efficacy of solutions;
—— studies of global earthquake cycles using the ERS‑1/ERS‑2 and Envisat SAR
data archives;
—— global volcano studies using ERS‑1/ERS‑2 and Envisat SAR data archives; and
—— orbital and atmospheric noise in InSAR data inferred from the global ERS‑1/ERS‑2
and Envisat SAR data archives.

11.7 The EO Information Services Sector

At the time of the launch of ERS‑1 in 1991, the commercial use of remote
sensing satellites was focused on the sale of data and basic image products.
Since then, the emphasis has changed from basic data to higher-level EO-based
information services as the main types of remote sensing products provided.
This trend has been accompanied by the emergence of many small, leading-
edge companies across Europe that provide such services. These companies

38
 History of ERS

often arise as start-ups from the university research and development sector,
and are a constant source of innovation for new techniques and methods
that find their way into commercial applications. Today, this industry has an
estimated annual turnover in revenues of €650 million to €700 million, with
the public sector still forming the largest source of demand (approximately
70% of revenues). The growth in revenues has been constant over the last two
decades at around 8% per year.
There has been no single breakthrough (or ‘killer app’), but continual
improvements in the technical capabilities of these small companies, and
an increasing awareness and acceptance of user communities of the benefits
that EO services can deliver. The ERS has played a significant role in that
development. It has led the way worldwide in the development of services based
on radar, and it is estimated that 60% of all EO information services produced
today are exploiting radar either entirely or in part. Some examples include oil-
spill monitoring (used on an operational basis by the European Maritime Safety
Agency, EMSA), ice mapping (used by national ice mapping agencies in Europe
and Canada), and forest monitoring (where radar is an essential supplement to
optical data for the cloud-covered regions of the tropics).
But perhaps the most striking example is that of precision land motion
monitoring using the technique of SAR interferometry. ERS has pioneered the
development of this sector and led to a small but growing base of specialist
companies in Europe that are continually expanding commercial activity with
clients in both the public sector (e.g. local and national governments) and the
private sector (e.g. oil & gas and civil engineering companies). Free access to
ERS data has also been pivotal in developing the prospects for higher-level EO-
based information services. It is planned to continue this policy with the next-
generation GMES Sentinels, where the heritage and capabilities developed
through ERS will come to full fruition.

11.8 ERS Data Policy: Adapting to a Changing Environment

The ERS data policy, adopted in 1990, was based on an open and non-
discriminatory approach and identifying ‘user categories’ with specific
conditions in terms of pricing policy. This approach has gradually evolved
to a new paradigm of an ‘open and free’ data policy, which was adopted in
2010. This evolution of the data policy was possible because of the IT mass
market, which has allowed significant reductions in satellite data production
costs; the internet, which has enabled data dissemination at very low cost;
modern means of communication and collaboration (no fax any more); and the
changing political environment for data sharing and global collaboration.
An example of the evolution of the data policy leading to a paradigm
change can be seen in the context of the GEO collaboration. Already in 2008,
for the GEO Geohazard Supersites initiative, large-scale ERS and Envisat SAR
data covering specific areas prone to severe earthquakes or volcanic eruptions
were made ‘open and free’ to the Geohazard community applying the rules
of ‘old’ data policy. The GEO Secretariat director, José Achache, was the Cat-1
PI and members of the global science community were GEO’s Co-PIs. Online
access to ERS and Envisat line data was granted to each Co-PI, who had
simply to sign the terms and conditions. Since 2008, tens of thousands of SAR
products have been made available using ESA’s ‘virtual archive’, allowing the
global community to download high-volume datasets at mass-market prices.
These large datasets were compiled by UNAVCO4, which coordinated the
demand for data from hundreds of co-PIs. Organising the demands of such

4 UNAVCO is a non-profit university-governed consortium that facilitates geoscience

research and education using geodesy: www.unavco.org

39
SP-1326

a very large group via one single interface greatly increased the efficient use
of ESA resources. This setup meant that the production performed for one
single user was fully accessible to several hundred others, thus significantly
reducing costs while improving the service. The availability of time series data
over certain observation areas over 20 years has also encouraged other data
providers to contribute their data and to allow open & free access, but in line
with their individual data policies.
ESA, with its medium-resolution satellites and its open & free data policy,
ensures an optimal outreach of remote sensing fostering further new research,
applications and services that will benefit commercial and dual-use satellite
missions.

11.9 Deorbiting Phase

After 16 years of successful operations, and a highly satisfactory output of
scientific data and applications, the ERS‑2 satellite was deorbited in summer
2011 with the final passivation on 5 September 2011.
The power system was the weakest point of the satellite, and already back
in 2007 special measures were implemented to ensure that the satellite had
enough power to recover from possible error conditions during eclipse periods.
Even though ERS‑2 had the same platform as SPOT-1 and SPOT-2, which
operated for more than 16 years, the problems with the ERS gyroscopes and
the ageing power system supporting the SAR, an instrument with a very high
power demand, were the main arguments for a controlled deorbitation in 2011.
For this, ESA had to:

—— minimise the satellite reentry time (limit 25 years), in accordance with the
Inter-Agency Space Debris Mitigation Guidelines;
—— free up the ‘busy’ orbits in the low-Earth orbit region (900–700 km); and
—— deactivate the satellite once it reached its disposal orbit.

The ERS‑2 deorbitation strategy, considering hardware constraints and

performance uncertainties, was developed under the leadership of ESTEC’s
Post-Launch Support Office with the ‘historical’ development team and the ERS
operations team.
The zero-gyro operations strategy, which had been operated since 2001,
could not be used for the deorbitation since the open-loop attitude control,
relying on Near-Realtime data from the LRB station network, did not work with
operational processing systems requiring a defined speed above ground. The
strategy identified, minimising the risk of losing the satellite, was to maintain a
Sun-synchronous orbit and to divide the deorbitation activities into three phases:

—— Phase 1: reducing altitude in a circular orbit using only one gyro having an
optimal position for measuring the yaw angle. This gyro had already shown
signs of degradation in the past, so the other gyro, which was still in perfect
shape, was selected in case of problems for Safe Mode operations. This phase
lasted three weeks.

—— Phase 2: further reducing altitude in a circular orbit, ensuring reentry well

within 25 years. This phase lasted also three weeks, with the satellite reaching
an altitude of 570 km.

—— Phase 3: exhausting the tanks, minimising the reentry time using a circular orbit
and deactivating the satellite. For this phase a one-day repeat orbit at 570 km
altitude was selected to allow a precise scheduling of the station network using
ESA’s Kiruna and Maspalomas stations, and also the stations at Kourou, Troll
in Antarctica, Hobart and Kumamoto. This phase lasted one week.

40
 History of ERS

Figure 27. The last image acquired by

ERS‑2, of the Rome area, 4 July 2011. (ESA)

The strategy was implemented via a detailed plan of activities, including

manoeuvre performance calibration and new trajectory calculations, global
station network activation, platform monitoring, etc., spanning a period of seven
weeks. This plan was followed precisely, without any deviation, until the end.
ERS‑2 is now flying in a circular orbit at about 570 km altitude; its reentry
into the atmosphere, where it will burn up, will take 15–20 years depending
on solar activity, well within the 25-year limit stated in the Inter-Agency Space
Debris Mitigation Guidelines.
The ERS‑2 satellite acquired data until 4 July 2011 (Fig.  27), with its last
months of activity dedicated to a specific Ice Phase (3-day repeat orbit). 

11.10 Exchange with Users − ESA Earth Observation Symposia

Even before the launch of ERS‑1 many scientists and operational data users
supported the ERS missions by participating in the definition phase, preparing
the Calibration and Validation activities and developing retrieval algorithms.
Many PhDs were earned in preparation for the ERS missions and during
20 years of operation. Since the beginning of the mission this community has
gathered at symposia organised by ESA every three or four years to exchange
the latest research findings but also to communicate their requirements so
that ESA could tune its operations and services as the mission advanced, and
address new challenges.
The contributions to the symposia, in terms of the numbers of participants
and presentations, increased steadily (see Fig.  28). The last ‘Living Planet’
symposium, held in Bergen in 2010, was the biggest since the launch of
ERS‑1, with some 1200 participants from 44 countries presenting more than
1000 papers and posters.

41
SP-1326

Figure 28. Participants and presentations

at ESA user meetings, 2003–11. (ESA)

12. Conclusions: Lessons Learnt and Follow-on Steps

Over the 20 years of successful operations, the ERS programme made major
advances and produced outstanding results, including:

—— The creation of archives of well calibrated, regularly reprocessed datasets

that are invaluable for climate studies and other applications. They represent
a major asset that is still to be fully exploited.

—— Demonstration of the operational capabilities and potential of ERS instruments

such as the Wind Scatterometer and GOME, which has enabled their transfer
to the operational Eumetsat MetOp series. It is important to recall the major
role played by the ECMWF in this demonstration with the assimilation of ERS
data into their global forecasting models, and in particular the contribution
of Tony Hollingsworth, Director of Research at ECMWF.

—— The confirmation of Europe’s leadership in the domain of Earth observation

from space with the development of well-calibrated active microwave
sensors requiring the definition of innovative calibration techniques, the
development of advanced technologies, the definition and implementation of
scientific algorithms for data processing and interpretation, the development
of industrial competence and expertise and of operational applications.

—— The reinforcement of the cooperation with the EU/European Commission and

in particular the Joint Research Centre (JRC) at the Institute for Environmental
Protection and Research (Ispra). The cooperation began in the early 1980s
with joint airborne campaigns and the exploitation of ERS data for various
projects. Within the Tropical Ecosystem Environment Observation by
Satellites (TREES) project, for example, the first SAR map of equatorial forests
of Central Africa was produced by Jean-Paul Malingreau (head of the MTV
unit) and his team, which may be regarded as a first or early step towards
GMES (the Baveno Manifesto establishing GMES was signed in May 1998).

—— The ERS mission can be regarded as an international cooperation tool that

led to the establishment and negotiation of agreements for a global network
of ground receiving stations. It encouraged regular exchanges of information
on issues such as data formats, and on standards within the Committee on

42
 History of ERS

Earth Observation Satellites (CEOS), a forum set up in 1984, through the

exchange of data with third-party missions JERS‑1), etc.

—— Concrete collaboration in the development of applications and scientific

research with the United States, China, Japan, India, Russia, Australia,
South Africa, etc., with international and European organisations (ECMWF,
FAO, UNESCO, UNDCP, UNOOSA, WMO, etc.), with the worldwide scientific
community, and with innovative SMEs.

—— With ERS, ESA (together with CNES and SPOT) played a major role in the
creation of the International Charter for Major Disaster Management initiated
in the aftermath of Hurricane Mitch, the most powerful of the 1998 hurricane
season with ~300 km h−1 winds that caused ~20 000 fatalities.

—— Innovations such as the internet and the dramatic evolution of information

technology have enabled an outstanding usage of the mission data, and have
catalysed global collaboration in operations and advancement of science
and services. Without this parallel evolution, fostered by the invention of the
world wide web (another European success story), Earth observation would
not have reached this level of maturity and would not now be part of the daily
lives of citizens and businesses.

—— ERS, in particular as a non-active mission, will continue to be a pioneer for

future satellite mission operations, collaborations and data exploitation
schemes that will benefit from its data archive covering all areas of Earth
science. Over two decades, it proved able to adapt efficiently to the changing
environment, building on technical and structural heritage and partnerships.

—— ERS, supported by the open and free data policy, made significant
contributions to innovative internet-based exploitation platforms with the
potential to create new business opportunities for commercial companies
and EO data providers, as well as for telecoms and IT industries.

In conclusion, one may say that the ERS programme has been a complete
European success that paved the way for the Envisat, MetOp and Sentinel
missions.

Acknowledgements
The authors wish to thank the thousands of engineers, technicians, scientists
from ESA establishments and from European and Canadian industry who
made ERS a reality and an outstanding worldwide success. Particular
acknowledgements are addressed to Jan Burger, the first ERS‑1 Project
Manager from 1984 to 1991 (when he unexpectedly passed away), and his
successor, Reinhold Zobl, who successfully continued his work. The authors
are also grateful to the Announcement of Opportunity Instrument Principal
Investigators and their teams who enabled their instruments to fly on the
ERS platform, namely David Llewellyn Jones and Chris Mutlow for the ATSR,
John Burrows for GOME, Philipp Hartl for PRARE, and René Bernard and his
successor, Laurence Eymard, for the Microwave Radiometer.
Special acknowledgements go to the worldwide user community, whose
members exploited the mission data with outstanding competence and
dedication throughout the exploitation life cycle (research, applications,
operational services), as well as to the ESA Member States, who have
recognised the excellent teamwork and the good return on the investment in
ERS, allowing the long-term continuation of the mission with Envisat and the
Sentinels, and of the Scatterometer and GOME instruments on MetOp.

43
SP-1326

References
GCOS (2003). GCOS Climate Monitoring Principles. World Meteorological
Organization, Geneva, Switzerland. www.wmo.int/pages/prog/gcos/
WMO (2010). Scientific Assessment of Ozone Depletion: 2010. Global Ozone Research
and Monitoring Project, Report No. 52. World Meteorological Organization,
Geneva, Switzerland. http://ozone.unep.org/Assessment_Panels/SAP/

In view of the thousands of publications and hundreds of services derived from

the ERS mission, the following list includes only ESA publications that provide
references to specific achievements:

ESA (1992). Space at the Service of our Environment. 1st ERS‑1 Symposium, Cannes,
France. ESA SP-359, vols 1 & 2. European Space Agency, Noordwijk, the Netherlands.
ESA (1993). Space at the Service of our Environment. 2nd ERS‑1 Symposium,
Hamburg, Germany, vols 1 & 2. ESA SP-361.
ESA (1994). First Workshop on ERS‑1 Pilot Projects, Toledo, Spain. ESA SP-365.
ESA (1995). New Views of the Earth: Scientific Achievements of ERS‑1. ESA SP‑1176/I.
ESA (1996a). New Views of the Earth: Applications Achievements of ERS‑1. ESA
SP‑1176/II.
ESA (1996b). Second ERS Applications Workshop, London, UK. ESA SP-383.
ESA (1997a). Third ERS Scientific Symposium, Florence, Italy. ESA SP-414.
ESA (1997b). New Views of the Earth: Engineering Achievements of ERS‑1. ESA
SP 1176/III.
ESA (1998). Further Achievements of the ERS Missions. ESA SP‑1228.
ESA (2000). ERS–Envisat Symposium, Göteborg, Sweden. ESA SP-461.
ESA (2004). Envisat and ERS Symposium, Salzburg, Austria. ESA SP-572.
ESA (2007). Envisat Symposium, Montreux, Switzerland. ESA SP-636.
ESA (2010). ESA Living Planet Symposium, Bergen, Norway. ESA SP-686.

44
→→ The ATSR Instruments →
and What They Have Led To
 ATSR

The ATSR Instruments and What They Have Led To

D.T. Llewellyn-Jones

AATSR Principal Investigator, Space Research Centre, Department of Physics &

Astronomy, University of Leicester, Leicester LE1 7QR, UK

1. Introduction
The Along-Track Scanning Radiometer (ATSR) instrument was carried on
the ERS‑1 satellite with the principal objective of providing global Sea-
Surface Temperature (SST) measurements at the levels of accuracy, coverage,
continuity and quality required by the climate research community. ATSR
was an Announcement of Opportunity instrument, nationally funded by the
UK, with additional contributions from Australia and France. It was followed
by its two successor instruments, ATSR‑2 and AATSR, flown on the ERS-2 and
Envisat satellites, respectively. The three ATSR instruments provided a near-
continuous dataset of global SSTs at the level of accuracy needed by the climate
research community, as well as high-quality data on land surfaces and, in the
atmosphere, on aerosols and clouds.
As an example of the completeness and the quality of the SST dataset from
the ATSR sensors, Fig. 1 shows a Global SST Field, averaged for the month of
January, from a new SST climatology derived exclusively from this 20-year
dataset. It can be seen that there are still some areas of the global oceans
where the quantity or quality of data were insufficient to enable a physically
meaningful climatological average SST value to be derived. Nevertheless, all
the main oceanographic features and currents can be seen with great clarity.
Work is continuing to refine and complete this climatology (Corlett, 2011,
private communication).

Figure 1. SST climatology for the month of

January, derived from ATSR‑1, ATSR‑2 and
AATSR SST data

47
SP-1326

2. The Origin of the ATSR

In the late 1970s, it was recognised by the climate research community that there
was an urgent need for accurate observations of global sea surface temperatures.
The requirements were demanding, including the need for accuracy of better
than 0.3°C, together with global coverage and continuous observations.
Although it was recognised at the time that space observations could
probably provide coverage and the continuity it was not widely believed that
the required level of accuracy could be achieved. However, a consortium of
scientists from four institutes – the Department of Atmospheric, Oceanic and
Planetary Physics, University of Oxford; the recently formed Space Science
Department, Rutherford Appleton Laboratory; the Mullard Space Science
Laboratory (MSSL), University College London; and the UK Met Office − devised
a scheme for achieving this level of accuracy by combining rigorous calibration
procedures, advanced technology and a novel two-angle view of Earth’s
surface in a single radiometric instrument.
At about this time, ESA scientists were finalising their plans for the
ERS-1 mission, the first polar-orbiting Earth-observing satellite. The envisaged
payload was similar to that of the highly successful but unfortunately short-
lived Seasat satellite that NASA had flown for just 100 days in 1978. The
proposed ERS‑1 payload was an advanced version of its NASA predecessor and
it included an ambitious Ocean Colour Monitor (OCM), which unfortunately
had to be dropped from the planned payload on account of resource
issues. In its place, ESA issued an Announcement of Opportunity (AO) for
nationally provided instrumentation to fill this gap, but weighing less than
50 kg, consuming no more than 50 W of power and with a data rate limited
to 200  kbit/s. In response to this AO, the ATSR instrument was proposed to
ESA and, following a competitive evaluation, was accepted for flight on ERS‑1.
It was duly built (Edwards et al., 1990) and launched on 17 July 1991 from the
European launch facility at Kourou, French Guiana, on an Ariane 4 rocket.
Figure 2 shows the cover of the ERS‑1 programme proposal that was presented
to ESA’s Programme Board for Earth Observation for approval, together with

Figure 2. Top panels: Some stickers from

the ERS era. Left: The first ERS-1 sticker
designed for the ERS team at ESTEC.
Centre: ATSR-1 sticker designed for the
ATSR team at RAL, and the AATSR sticker
designed by Sarah Llewellyn-Jones. Right:
Another ESA sticker for the ERS-1 mission.
Lower panels: The cover of the ERS‑1
programme proposal, which included
sketches of the satellite fully deployed and
inside the Ariane fairing (right).

48
 ATSR

Figure 3. The launch of ERS‑1 on 17 July

1991 (left), and a close-up of the fairing
from the same image (right) in which the
umbilical cord to the satellite can be seen,
still connected but about to part, signifying
the birth of the ERS‑1 mission. The giant
ERS‑1 sticker that decorated the fairing
can also be seen.

sketches of the proposed satellite included the proposal. Figure 3 shows the
launch of ERS-1, with a close-up of the fairing and of the sticker that the ERS-1
project team had designed for the occasion.
Soon plans were afoot for the successor instruments, ATSR‑2 and AATSR.
ATSR‑2, which was similar to ATSR‑1 but with the addition of visible and near-
infrared channels (Read et al., 1991), was launched on the ERS-2 satellite in
1995, while AATSR (Llewellyn-Jones et al., 2001), with a higher data rate for
more detailed data over land, was launched on Envisat in 2002.

3. Some Initial Results

Following its launch and commissioning, ATSR was soon producing very high-
quality thermal and reflected infrared images of Earth’s surface. It was in some
ways unfortunate that the launch of ERS‑1 coincided almost exactly with the
massive eruption of the Mount Pinatubo volcano in the Philippines, during
which considerable amounts of substantial particulate matter were injected
into the upper atmosphere, where the prevailing stratospheric winds quickly
distributed the dust, causing it to encircle the equatorial regions.
Although this made validation of ATSR very difficult, it provided an
immediate opportunity to demonstrate the power of the dual view, which
was the most innovative hallmark of the ATSR observing system. Figure 4
illustrates the difference between AATSR’s two types of retrieved sea surface
temperature: the SSTs retrieved from the generally used single view (nadir
only) brightness temperatures, which was known to be seriously degraded
by excessive atmospheric aerosol loading; and the other from ATSR’s novel
dual view (forward and nadir view brightness temperatures combined,
which prelaunch modelling studies had shown to be much less sensitive to
contamination by excessive atmospheric aerosols than the nadir-only view.
The difference between these two retrieved SST values was immediately
observed on account of the Mount Pinatubo aerosols. The global plot in Fig. 4
shows the differences between these two SST values, averaged over a period
of one month early in the ERS‑1 mission. Although the ATSR SST data were
not fully developed or validated at this stage, it was clear that the innovative
dual-angle viewing geometry would enable ATSR to generate SST values in the
presence of very adverse atmospheric conditions.
It can be seen from Fig. 4 that, throughout most of the mid-latitude regions,
the difference between the two SST values is of the order of one- or two-tenths
of a degree, but the difference is closer to 1°C throughout the tropics, where

49
SP-1326

Figure 4. A plot from the early days of

the ERS‑1 mission in 1991, showing the
difference between SSTs calculated from
dual-view observations and those from the
nadir view only. The differences are due to
the significant effects of the Mt Pinatubo
aerosols on the nadir-only (AVHRR-style)
estimates and the much less affected dual-
view SST estimates, and provide a clear
indication of the spatial distribution of the
dust cloud which, at this early stage, was
confined to tropical regions. The effects of
an outburst of Saharan dust can also be
discerned. (From Dundas, 1997)

the volcanic ash was known to be distributed during the months immediately
following the eruption.
This result immediately suggested that the difference between the dual-
view and nadir-only view SST values might be used as a proxy for the level
of atmospheric contamination. It can be seen from Fig. 4 that contamination,
presumed to be from Mount Pinatubo’s aerosols, encircled the equator, as
was already well known from other observations at the time. Figure 4 also
shows, slightly to the north of the Atlantic tropics, what is possibly aerosol
contamination due to a Saharan dust outburst, which is a regular occurrence.
Although it took several years to achieve the levels of accuracy required by
the climate research community, it was immediately clear from the Mt Pinatubo
aerosol contamination that ATSR’s unique dual-angle viewing geometry had the
potential to make consistent corrections for the effects of atmospheric aerosols.

4. Scientific Achievements of the ATSR Series

4.1 The Accuracy of AATSR
A comprehensive summary of ATSR’s scientific achievements over the oceans,
land surfaces and in the atmosphere, and meteorological applications, can be
found in the special issue of Remote Sensing of Environment (RSE), published
in January 2012 (Llewellyn-Jones & Remedios, 2012). This special issue covered
a wide range of scientific applications and demonstrated the high quality and
versatility of ATSR data. This chapter highlights only a few of ATSR’s scientific
achievements and the reader is referred to the RSE special issue and the rest
of the scientific literature for further details and a wider review of ATSR’s
achievements.
Following many years of algorithm refinement, combined with rigorous
validation programmes, the ATSR sensors met their objectives for the
measurement of Sea-Surface Temperatures. A recent assessment of the levels
of accuracy achieved, carried out by Corlett (2011, private communication),
involved a careful comparison of AATSR SST values and in situ measurements
from the global drifting buoy network. The result of an initial comparison,
employing around 10 000 matchups, is shown in Fig. 5.
The first aspect of this plot to note is that the agreement between the SST
values and the coincident in situ observations appears to fit a Gaussian curve
very well indeed; in fact it can be regarded as an exceptionally good fit for
geophysical observations. Table 1 shows the level of agreement between AATSR
SSTs and those from drifting boys when different statistical approaches are
applied. The least mean deviation is obtained using so-called robust statistics,
whereby exceptional outliers are disregarded from the daily analysis. This
analysis leads to an estimated accuracy of less than one-tenth of a degree with

50
 ATSR

Figure 5. The difference between the

best AATSR SST retrievals, which were
those using the dual-view 3-channel
retrieval algorithm, plotted against quality-
screened drifting buoy measurements for
10 000 matchups. (Corlett, 2011, private
communication)

Table 1. Mean values and standard

Method Mean value Standard deviation
deviations characterising the distribution
Simple arithmetical –0.09K 0.49K curve shown in Fig. 5, calculated by several
3-sigma –0.09K 0.35K generally accepted methods.
Dual-nadir difference (GHRSST L2P1) –0.08K 0.37K
Robust –0.08K (median) 0.33K
Gaussian fit –0.07K 0.2K
1 Group for High-Resolution SST, Level 2P dataset.

a standard deviation of just over one-quarter of a degree. It is probable that a

significant fraction of the standard deviation is due to variations in drifting boy
observations because, for example, of varying water depths of the temperature
sensors on account of the sea surface motion.
With this level of accuracy clearly demonstrated, and with over 20 years of
data collected, it is clear that the ATSR dataset has the potential to contribute
to the climate record.

4.2 Time Series of Global SST

Having established a high level of accuracy and in order to use SST observations
for investigating the behaviour of the global climate, it is an important starting
point to monitor the average global temperature of the sea surface. Following a
complex and detailed analysis of the dataset from the three ATSR sensors, taking
into account all known effects and processes that might affect the accuracy of the

51
SP-1326

Figure 6. Monthly global averages of

the dual-view 3-channel night-time SST
anomaly plotted over the time period
between the launch of ERS-2 in 1995 and
mid-2011. Note that there is a six-month
data gap near the beginning of the record
on account of a scan mirror anomaly in
ATSR‑2. When this plot was created, the
process of rigorous uncertainty analysis
was still being carried out (Veal et al.,
2012) and is therefore incomplete. For
this reason, an estimated uncertainty is
indicated, but this must be regarded as
provisional. (University of Leicester)

retrieved SSTs, monthly global averages of observed SST have been computed for
the duration of the three missions. A sample of the results is shown in Fig. 6.
It should be emphasised that this is a preliminary calculation and, in Fig. 6,
only the 3-channel dual-view data, which provided the most accurate SST values,
are shown. Certain aspects of the data processing, such as the identification and
screening of clouds in the field of view, are known to be less than optimal. For
such reasons, data from ATSR‑1 have not been included in Fig. 6, mainly but not
exclusively because that instrument lost the use of its powerful 3.7 µm thermal
infrared channel at an early stage in the mission (in May 1992) with the result
that, for the bulk of the ATSR‑1 mission, the most accurate SST retrievals could
not be implemented, nor could the most efficient night-time cloud-identification
scheme be used. Thus, continuous 3-channel dual-view data were not available
until the launch of ATSR‑2 on ERS-2 in April 1995.
The time series in Fig. 6, although it extends only from 1996 to mid-2011
shows strong variability. A detailed interpretation of these variations is beyond
the scope of this chapter, but the effects of the substantial El Niño effect in
1997, as well as the strong reverse effect (La Niña) in 2008, are very pronounced
and are perhaps the dominant features of this 15-year time series. Phenomena
such as these underline the importance of obtaining long time series of global
data in order to observe global changes without ambiguity. However, when all
the known attributes of the data are taken into account, the SST values shown
in Fig. 6 can be considered to have an uncertainty of approximately ±0.1K.
It was mentioned above that certain aspects of the current SST data
processing system are now known to be less than optimal. Many of these
aspects have recently been further analysed and more fully taken into account
in the course of a particularly important project, the ATSR Reprocessing
for Climate (ARC). This project is led by C. Merchant of the University of
Edinburgh and implemented by a consortium that includes the University of
Edinburgh, the University of Leicester, the UK Met Office and the University of
Southampton. As a result of this project, globally averaged SST data have been
recalculated between 1993 and the present. There are important similarities
and differences between the ATSR records and the Met Office Hadley Centre’s
own in-situ-based analysis, HadSST3, a discussion of which is beyond the
scope of this chapter. However, it is clear that the ATSR data have earned their
place as a key component of this climate record. It is worth noting that in the
course of the ARC project it has been established that, in the tropics at least,
the ATSR SST record exhibits the stability of 0.03°C per decade (Merchant et al.,
2012), which is well within the requirements previously set out by the climate
research community. Such levels of accuracy and stability clearly demonstrate
that the ATSR sensors certainly met their first objective.

52
 ATSR

4.3 ATSR Data over Land

In addition to the accurate SST measurements, the ATSR was applied to
measure the properties of land surfaces. The ATSR sensors were particularly
well suited to measuring the radiative temperatures of land surfaces because
the instruments were well calibrated radiometers, with channels selected to
provide accurate atmospheric corrections in the thermal infrared and with a
spatial resolution of 1 km.
Although Land Surface Temperature (LST) is not classified by the climate
community as an Essential Climate Variable (ECV), it has an important, even
critical, role to play in the investigation of climate processes, particularly those
involving soil moisture and the transpiration of water. Also, recent work by
Ghent et al. (2010, 2011) has demonstrated that satellite measurements of LSTs
have a positive impact on land surface characterisation models and could
therefore play an important role in future weather and climate forecasting
systems. Another possible application of LST data, especially in urban areas,
is in the area of public health, where there is a growing awareness of the
relationship between surface temperatures and the production of harmful
polluting compounds in the lower atmosphere. Increased effort is now being
devoted to optimising the LST data products generated from ATSR data.
An example of an AATSR LST data product is an image of the UK during the
exceptional heatwave of 2006, shown in Fig. 7. In order to make radiometric
measurements of land surface temperature from space, it is a major challenge
to characterise the land surfaces because they exhibit high variability in
structure and, implicitly, in actual temperature values. The characterisation
of land surfaces is thus one of the most challenging aspects of the data
processing system. Also, on account of the inhomogeneities and varying
topography of land surfaces, atmospheric behaviour is more heterogeneous
and therefore the atmospheric correction problem is more complex than in
most marine environments. In addition to this, the problem of unambiguous
cloud identification over land surfaces is even more difficult, with no ‘magic
bullet’ in sight at this time, making it perhaps the most challenging aspect
of land-surface remote sensing. ATSR LST data products are the subject of
ongoing research. The plot shown in Fig. 7 was produced in the course of the
development programme and should be regarded as an experimental product.
In 1995, ATSR‑2 was launched with additional channels in the visible and
reflected infrared regions. These had the new objective of using the unique
dual view to develop an atmospheric correction for the visible-wavelength
reflectance measurements and the land cover data products that would follow.
It had been shown that it was possible to use the dual-angle view as well as
the multi-wavelength channels to infer Aerosol Optical Depth (AOD) and then
correct for the presence of atmospheric aerosols over land. This meant that it
would be possible to derive land surface properties from the visible channels
of the ATSR instruments while utilising an atmospheric correction based on
optical depth determinations that utilise the ATSR unique two-angle viewing
geometry. Figure 8 shows some results from recent work (Bevan et al., 2011),
where global maps of AOD and derived surface reflectance, inferred from the
two-angle view in combination with a bidirectional reflectance model, can be
seen. It should be noted that this technique can be used over both land and
sea to produce the global plot shown in Fig. 8. This type of work opens up the
possibility for a wide range of atmospherically corrected land surface products
from the ATSR instruments.

53
SP-1326

Figure 7. Land-Surface Temperatures over

England, Wales and Scotland on 15 July
2006, during a heatwave, retrieved from
AATSR data. Significant variations and
spatial structure can be clearly observed.
Some of the values are exceptionally
high for a mid-latitude climate. The
major conurbations, including London,
Birmingham and Leicester, can be clearly
identified on account of their higher
surface temperatures.

Figure 8. Left: Aerosol Optical Depth (AOD) inferred from ATSR data using the two-angle view together with a bidirectional reflectance
model. Right: Inferred land surface reflectance, obtained by using the AOD to correct the measured top-of-atmosphere brightness.

54
 ATSR

4.4 ATSR Data and the Atmosphere

The processes of scattering, absorption and emission, by atmospheric
aerosols, of infrared and visible radiation are not only relevant to atmospheric
corrections for land surface reflectance measurements mentioned in the
previous section, but they also provide the means to investigate the properties
of atmospheric aerosols themselves. The ATSR instruments, with their well-
defined infrared and visible spectral channels, together with ATSR’s two-angle
viewing geometry, had considerable sensitivity to atmospheric aerosol content.
It must nevertheless be recognised that the problem of retrieving
comprehensive and accurate information about parameters such as particle
size, particle content, particle density and spatial distribution present complex
problems, both in the retrieval process and in validation of the data. This
is largely because there are undoubtedly more variables than measurable
quantities. However, there is an active community of strong research groups
working to develop new aerosol products from ATSR data and this is seen
as a major development area for the future. The universities and institutes
contributing to this work include the University of Bremen, the University
of Oxford, the Rutherford Appleton Laboratory, the Finnish Meteorological
Institute, the DLR and the University of Swansea. Looking to the future, the
development of new, robust and accurate aerosol products from ATSR data is
certainly both a major challenge and an exciting prospect for the Sentinel‑3
mission.

4.5 The ATSR Fire Atlas

A space sensor that produces high-quality images of Earth’s surface on a
regular basis is bound to have many applications, although reviewing every
one of them is certainly beyond the scope of this chapter. One spectacularly
successful application of ATSR data, which can claim tens of thousands of
users, involved monitoring large fires and miscellaneous hotspots on a global
basis. In the early 1990s ESA realised that ATSR was the only visible and near-
infrared imaging sensor in space that was continuously producing global data
at 1 km resolution, including data at 1.6 µm wavelength, which is particularly
well suited to the high temperatures of fires. Thus, the ATSR Fire Atlas was
created and has been operating continuously since 1996, and is now accessed
by about 60 000 users each year. Figure 9 shows two examples of the types of
observation made by ATSR: the existence of gas flares and the emergence of
global trends.

Figure 9. Left: Gas flare monitoring from space using the ATSR instrument (Casadio et al., 2011; Arino et al., 2011). Right: Global time series
of the number of hotspots observed each year, 1996–2010.

55
SP-1326

4.6 ATSR Image Data

When ATSR started working, it immediately produced images of remarkably
high quality, reflecting the excellent optical design and sensitive cooled
detection system that the instrument uses. From the point of view of measuring
sea surface temperature, the main function of the images was to identify
clouds and land masses so that the processing system could screen them
out, and also to monitor various aspects of data quality. It should be noted
that the ATSR sensors produce two kinds of image at 1 km spatial resolution:
namely, the generally familiar high-quality visible and near (reflected) infrared
wavelength images, as well as thermal images, which reveal remarkable
patterns of temperature variations that are invisible to the naked eye, forming
the foundation of ATSR’s accurate SST retrievals. Examples of both types of
image are shown in Figs 10 and 11, respectively.
Apart from the attractiveness of the images and the general fascination they
invoke, the ATSR image archive has an important role to play in environmental
monitoring. This is because the orbits of ERS-1/ERS-2 and Envisat were
highly stable for more than 20 years, and the ATSR sensors provided a global
image dataset that was acquired with unique consistency of viewing angle
and illumination angle throughout the three missions. This archive can be
used to detect and monitor quite large-scale environmental changes in such
parameters as ice coverage and the areas of lakes.

4.7 The Transition to Operational Status

Numerical Weather Prediction (NWP) requires a complete global SST field on a
daily basis. To provide SST data in this way presents substantial practical and
logistical challenges and, for that reason, analysis fields, based on available
observations and on past observations in combination with sophisticated
interpolation techniques, have generally been used for this purpose.
Until recently, although it had always been envisaged that ATSR’s accurate
SST observations would contribute to meteorological forecasting, there has

Figure 10. An unusually cloud-free view

of the UK, acquired by AATSR during
the early days of the Envisat mission,
showing the potential of visible or near-
infrared wavelengths for viewing the basic
geographical features of the land. Note
that the sea surface is uniformly dark, on
account of the low reflectivity of water at
visible wavelengths.

Figure 11. False colour image acquired by

ATSR‑1, showing a view of the UK in the
thermal infrared. The most striking aspect
is the appearance of temperature gradients
in the North Sea, in strong contrast with
the image in Fig. 10. Over land, major cities
can be identified as ‘urban heat islands’
and some of the geographical features also
have visible thermal signatures.

56
 ATSR

not been an available analysis field that includes ATSR data, for a variety of
reasons principally to do with the method of delivery. The situation has changed
recently on account of an ESA initiative to support the WMO’s Global Ocean Data
Assimilation Experiment (GODAE), which set out a requirement that operational
users of ocean data should be able to receive global data in an agreed format and
on an appropriate timescale (typically within a day of acquisition) and, most
important, with appropriate error and quality information for each data point
provided. This requirement was clearly not within the scope of the standard
Envisat ground segment and ESA, as a result, set up the ‘Medspiration’ project as
part of their Data User Element (DUE) programme. At the same time, the GODAE
High Resolution SST (GHRSST) project, since renamed the Group for High-
Resolution SST was formed, with crucial ESA support. GHRSST coordinated the
work of other SST data providers, including NASA, to ensure that data from other
satellite sensors was available to operational users on the same basis as data
from AATSR was provided through Medspiration.
The availability of global SST data on a daily basis, tailored to the needs
of operational users, presented a great opportunity for meteorological agencies
to access a wealth of excellent SST data from space and other sources. As an
example, the UK Met Office established a daily analysis scheme called the
Operational Sea Surface Temperature and Sea Ice Analysis (OSTIA),1 from
which the maps in Figs 12 and 13 were obtained. The OSTIA scheme, which
uses optimal interpolation of data from seven satellite sources as well as in situ
data, provides daily SST analyses field together with information regarding the
accuracy and consistency of the various datasets involved.
As an example, Fig. 12 shows a global SST analysis field from the OSTIA
scheme. The spatial resolution of the final product is obviously less than that of
the 1 km resolution satellite data from which it was compiled, because the daily
and weekly data will always contain gaps because of coverage and cloudiness
limitations, over which the analysis scheme has used interpolation techniques.

Figure 12. A global SST analysis field plotted for the week ending 24 September 2011. In this plot, data were obtained by six satellites plus
in situ sources. The satellite sensors include three IR sensors, two microwave sensors and one geostationary IR sensor. (UK Met Office;
Crown Copyright 2011)

1 See http://ghrsst-pp.metoffice.com/pages/latest_analysis/ostia.html for further

details and up-to-date examples of data.

57
SP-1326

Figure 13. The SST data from Fig. 12 plotted

as anomaly fields by subtracting the NCEP
analysis fields from the OSTIA data.
(UK Met Office; Crown Copyright 2011)

The information content of these SST fields is greatly enhanced when they
are plotted as anomaly fields. In Fig. 13, the corresponding National Centers for
Environmental Prediction (NCEP) climatology field has been subtracted from
the basic OSTIA SST field to provide the anomaly plot shown. It is interesting
to note the very high anomalies that existed in the Arctic regions at that time.
The OSTIA fields are of great potential interest to users of ATSR data because
they are spatially and temporally complete, combining the advantages of
coverage that the multiple sensors provide with the exceptional accuracy of the
AATSR. In fact, AATSR SST values are used in the OSTIA analysis scheme along
with the in situ data to derive bias corrections for the other satellite sensors.
There can be no doubt that the incorporation of AATSR data into an important
meteorological SST analysis scheme, on an operational basis, is one of the most
significant developments in the application of ATSR‑type data, because it has
overcome the shortcomings of coverage, which are important for mesoscale
process studies. It also allows for an objective comparative assessment of
AATSR’s very high data quality, as well as crossing the most challenging bridge
in the transition from scientific to operational observing systems.

4.8 AATSR on Show

On account of its importance as an instrument for quantitative monitoring
of a key climate parameter, the AATSR Engineering Model was included in
an exhibition on ‘Atmosphere: Exploring Climate Science’ that opened in
December 2010 at the Science Museum in London. Here, the AATSR instrument
is on display for the first time, suspended in such a way that its Earth-facing
side can be viewed from below, i.e. in its operational orientation, as seen in
Fig. 14.

58
 ATSR

Figure 14. The Structural/Engineering

Model of the AATSR on display as part of
the ‘Atmosphere: Exploring Climate Science’
exhibition at the Science Museum in
London. The individual in the left panel has
been included for scaling purposes.

Figure 15. Left: The exhibit on

measurement techniques, with the AATSR
Engineering Model suspended above and
including a yellow Argo Float, as well a
model of the ERS‑1 satellite lent by Chris
Rapley, director of the Science Museum.
Right: Noah, aged 3, and Freya, aged 5,
viewing an interactive commentary on the
AATSR.

In the Science Museum, the AATSR has been seen by thousands of visitors
of all ages. AATSR is seen in the context of its continued contribution to the
observations and measurements of climate properties, as illustrated in Fig. 15.

5. Summary and Conclusions

5.1 What ATSR has Shown Us
The first ATSR was conceived as the result of a strongly articulated need
expressed by the scientific user community and translated into a practical
solution by scientists and engineers, in response to an Announcement of
Opportunity from ESA. ESA then showed its confidence in ATSR by accepting
two successor instruments that were flown on ERS-2 and on Envisat. This was
very important because of the amount of time needed to reach the necessary
levels of data quality and general acceptance that AATSR enjoyed, and which
qualified it to achieve pseudo-operational status.
This process demonstrated the importance of strong coupling between
the work of data users, those who are active and creative in the science of
measurement processes and those who hold the key to the enabling technology
that provides the necessary competitive edge (for example, the ATSR coolers
and the blackbody reference targets), as well as the engineers and scientists
in research institutes and in industry, who actually designed, built and tested

59
SP-1326

the instruments. Not only were all these elements needed for a successful
progression, but there must be natural and productive communication
channels between parties. In addition, it must also be admitted that this
success was also made possible by the massive advances in data handling and
processing technologies during the ATSR missions and which were foreseen by
the proposers but not yet available to them.

5.2 ATSR and the Future

There is little doubt that, over the past 20 years the ATSR sensors met their
main objectives and there is still the clear prospect of further developments
of substantial scientific progress, in the form of new data products, new
applications and new discoveries. In addition, the AATSR data have found
their way into the daily SST analysis fields used by some of the world’s major
weather forecasting agencies.
The next step must be to ensure that an ATSR‑type sensor is part of
any future operational observing system that is used for meteorological
applications. This will be achieved when the Sentinel-3 satellite is launched by
ESA as part of the European GMES Space Element, because the payload will
include a fourth along-track scanning radiometer. This is to be the Sea and
Land Surface Temperature Radiometer (SLSTR), a new instrument designed
by ESA, which features all the main characteristics of the current AATSR with
substantial enhancements, including a significantly wider swath width and
additional channels, some with higher spatial resolution, features that will be
very much welcomed by users. This instrument will not only ensure the ATSR
system of a place in the future operational observing infrastructure of Europe,
but it will also, on account of these enhancements, open up new fields for
science and applications. Thus, in addition to meeting its scientific objectives,
the ATSR sensors will have also completed the transition from scientific to
operational functionality, perhaps for the first time.

Acknowledgements
The substantial contributions of several organisations and, perhaps more
crucially, of many individuals, were essential to the success of the ATSR
series. Although it is not practical to mention them all here, a few that must be
mentioned include:

—— UK Research Councils, which funded the early development and missions of

ATSR‑1 and ATSR‑2;
—— the UK Department of Environment (now the Department for Energy and
Climate Change), which funded AATSR;
—— many colleagues at RAL who designed, developed and built the first ATSR
and who remain the prime centre of expertise regarding these instruments;
—— my colleagues at the University of Leicester, whose efforts during the AATSR
era spearheaded many of the achievements in the calibration and exploitation
programmes of the ATSR sensors;
—— many industrial partners who contributed to all parts of the programme;
—— ESA, which had the confidence to accept ATSR for the first time and provided
steadfast support over all the missions; and, in particular,
—— the ERS project teams who designed, built and deployed ERS‑1, ESA’s first
pre-operational polar-orbiting Earth observation satellite, at a time when it
was the only such satellite available to the world’s EO community. They also
sustained and consolidated the achievement by maintaining continuity of
observations from 1991 onwards. Arguably, the main achievement of ERS‑1
has been to establish ESA as a world-leading agency in Earth observation.

60
 ATSR

It is too difficult to list all the individuals who made crucial contributions,
but it would, I believe, be appropriate to mention the contributions of Sir John
Houghton. As a distinguished academic scientist, having done pioneering
work in the development and use of atmospheric sounders in space, he later
became Director-General of the UK Met Office. Most important for ESA, he
was a particularly proactive chairman of the Earth Observation Advisory
Committee (EOAC, now ESAC), and was the founding chairman of the IPCC
Working Group 1. Together with the EOAC and ESA staff, he worked tirelessly
and effectively to define and present the ERS‑1 mission in such a way that it
was not only the optimum flagship for ESA’s applications programme, but also
of immense scientific value. Without this work, it is unlikely that the ERS‑1
mission would ever have met the diverse aspirations of ESA’s Member States
and won their full acceptance and enthusiastic support.

References
Arino, O., Casadio, S. & Serpe, D. (2011) Global night-time fire season timing and
fire count trends using the ATSR instrument series. Rem. Sens. Environ. 113,
408–420. doi:10.1016/j.rse.2011.05.025
Bevan, S.L., North, P.R.J., Los, S.O. & Grey, W.M.F. et al. (2011). A global dataset of
atmospheric aerosol optical depth and surface reflectance from AATSR. Rem.
Sens. Environ. 116, 119–21078. doi:10.1016/j.rse.2011.05.024
Casadio, S. Arino, O. & Serpe, D. (2011) Gas flaring monitoring from space using
the ATSR instrument series. Rem. Sens. Environ. 116, 239–249. doi:10.1016/j.
rse.2010.11.022
Corlett, G.K. (2011). Private communication.
Dundas, R.M. (1997): The use of ATSR data to measure the radiative properties of
aerosol particles, PhD Thesis, University of Leicester, UK.
Edwards, T., Browning, R., Delderfield, J., Lee, D.J., Lidiard, K.A., Millborrow, R.S.,
McPherson, P.H., Peskett, S.C., Toplis, G.M., Taylor, H.S., Mason, I.M., Mason,
G., Smith, A. & Stringer, S. (1990). The Along Track Scanning Radiometer:
Measurement of sea surface temperature from ERS‑1. J. Br. Interplanet. Soc. 43,
160–180.
Ghent, D., Kaduk, J., Ardo, J., Remedios, J. & Balzter, H. (2010). Assimilation of land-
surface temperature into the land surface model JULES with an Ensemble Kalman
Filter. J. Geophys. Res. – Atmospheres 115, D19112. doi: 10.1029/2010JD014392
Ghent, D., Kaduk, J., Remedios, J. & Balzter, H. (2011). Data assimilation into land
surface models: the implications for climate feedbacks. Int. J. Rem. Sens. 32(3):
617–632.
Llewellyn-Jones, D.T., Edwards, M.C., Mutlow, C.T., Birks, A.R., Barton, I.J. & Tait,
H. (2001). AATSR: Global-change and surface-temperature measurements from
Envisat. ESA Bull. 105, 10–21.
Llewellyn-Jones, D.T. & Remedios, J.R. (2012). The Advanced Along Track Scanning
Radiometer (AATSR) and its predecessors ATSR‑1 and ATSR‑2. Introduction to the
special issue on AATSR. Rem. Sens. Environ. 116. doi: 10.1016/j.rse.2011.06.002
Merchant, C.J., Embury, O., Rayner, N.A., Berry, D.I., Corlett, G.K., Lean, K., Veal,
K.L., Ken, E.C., Llewellyn-Jones, D.T, Remedios, J.J. & Saunders, R. (2012). A 20
year independent record of sea surface temperature for climate from Along-Track
Scanning Radiometers. J. Geophys. Res. 117, C12013. doi: 10.1029/2012JC008400
Read, P.D., Hardie, A.L., Magraw, J.E., Taylor, H.S. & Jelley, J.V. (1991). A Calibration
system, using an opal diffuser, for the visual channels of the Along-Track
Scanning Radiometer, ATSR‑2. Proc. 10th IOP Symp., IOP Conf. Ser. 121, 207–213.
Veal, K. et al. (2013). A time series of mean global skin SST anomaly using data from
ATSR-2 and AATSR. Rem. Sens. Environ. RSE-D-12-00191R2

61
→→ ERS and Atmospheric
Composition Measurements: →
GOME and the SCIAMACHY
Project
 GOME and SCIAMACHY

ERS and Atmospheric Composition→

Measurements: GOME and the SCIAMACHY
Project

J.P. Burrows

Institute of Environmental Physics, University of Bremen, Postfach 330440, 28334

Bremen, Germany. burrows@iup.physik.uni-bremen.de

1. Introduction
This chapter describes the evolution and development of the SCIAMACHY
project, and its spin-off, the GOME instrument, which was part of the
core payload on ERS‑2. GOME measured successfully from 1995 to the
decommissioning of ERS‑2 in July 2011. Figure  1 shows the GOME instrument
during testing at the Officine Galileo in Florence, Italy, and a schematic
diagram of GOME’s optical components. The ERS‑2 satellite was tested at
ESA’s European Space Research and Technology Centre (ESTEC) in Noordwijk,
the Netherlands, prior to being shipped for launch from Europe’s Space Port in
Kourou, French Guiana, in 1994.
The material in this chapter was presented at the symposium to celebrate
20 years of ERS in September 2011, and is structured as follows. Sections 2 to
4 describe briefly the evolution of Earth’s atmosphere, the physical structure
and chemical processes that determine its composition and the impact of
man on the Earth system. Section 5 describes key aspects of the history of
remote sensing from space. In section 6 the development of European remote
sensing and the need for global measurements of atmospheric composition,
the evolution of remote sensing prior to the ERS programme and the European
component. With increasing globalisation and urbanisation, coupled with an
increasing population, the need for this information is greater now than it was
30 years ago.
Sections 7−10 discuss some of the highlights of the development of
SCIAMACHY and its spinoff, GOME, the Phases C and D, the GOME instrument
and the contributions of some of the scientists, engineers and administrators
involved in this remarkable achievement of creating a unique atmospheric
composition observing capability. The excellent performance of GOME on
ERS-2 in orbit is documented in section 11. The development of the retrieval

Figure 1. Left: The GOME instrument during testing at Officine Galileo, Florence, Italy, showing the scan mirror and four spectral channels.
Right: Schematic diagram of the optical layout of GOME. (Officine Galileo)

65
SP-1326

algorithms to produce the GOME data products is explained in section 12.

Section 13 focuses on some of the key scientific applications and discoveries
made using the trailblazing experiment of GOME on ERS-2 and the follow-up
using GOME, SCIAMACHY and GOME‑2 data. Some of the challenges facing
Europe in establishing its contribution to a space segment providing adequate
global measurements of key trace constituents and parameters are addressed
in section 14. Finally, my conclusions and a summary of some of the issues
involved in the development GOME, and highlights of some GOME successes
are provided in section 15.

2. The Evolution of Earth and its Atmosphere

When Earth formed ~4.5 billion years ago its early atmosphere was in
physicochemical equilibrium, with carbon dioxide, CO2, as the most abundant
gas. Life began ~1 billion years later, and since then has evolved and
adapted. During the Achaean eon, primitive anaerobic organisms thrived,
utilising sulphate for their energy needs, and their descendants, known as
extremophiles, still survive. However, the emergence of cyanobacteria, or blue-
green algae, which use sunshine, water and CO2 to produce carbohydrates and
molecular oxygen, O2, changed Earth’s atmosphere. Approximately 2.45 billion
years ago, evidence that oxygen was becoming an important component of
Earth’s atmosphere was identified through changes in the isotopic ratio of
sulphur in rocks (Farquhar et al., 2000). Around the same time, oxidised
iron began to appear in ancient soils and bands of iron were deposited on the
seafloor, the latter being attributed to reactions with oxygen in seawater.
The release of molecular oxygen by photosynthesis to the atmosphere
resulted in feedback between the biosphere and the atmosphere. The
atmospheric composition in any geological era is determined by biogeochemical
photochemical processes, coupled with atmospheric dynamics. Plants evolved
and incorporated symbiotic cyanobacteria, known as chloroplasts, to undertake
photosynthesis. The release of O2 results in the production of ozone, O3, which
absorbs biologically damaging short-wavelength ultraviolet radiation in the
region known as UV-B. In addition, the biospheric modulation of atmospheric
composition provided conditions suitable for the emergence of new life forms.
Land plants are considered to have evolved from the charophycean lineage of
green algae. The first plants, liverworts and mosses, had probably evolved by
about 700 million years ago and land fungi by about 1300 million years ago.
This is earlier than previous estimates of around 480 million years, which
were based on the earliest fossils of those organisms. The recent examination
of Namibian fossils also extends the first appearance of animals to around
800 million years ago. The genus Homo appeared only about 2.3−2.4 million
years ago and evidence of the species Homo sapiens has been found from
approximately 0.2 million years ago.
At the Neolithic revolution, roughly 7000−10 000 years ago, human society
made the transition from hunting and gathering to establishing settlements
and agriculture. The population at that time is estimated to have been
approximately 4  million. Although fluctuating as a result of disease, natural
disasters and catastrophes, this population grew to 1 billion by 1800 AD. The
sources of energy fuelling this growth and being used for domestic heating,
agriculture and early industrial processes, was primarily biomass, but also
some energy was obtained from coal as well as wind and water power. The
earliest references to the use of coal date back to the ancient Greeks, and coal
and oil were being used in China and Europe in the Middle Ages.
Following the industrial revolution around 1800 AD, powered by the
use of fossil fuel, the pace of changes in atmospheric composition and land
use accelerated and both the human population and its standards of living
increased almost exponentially. Anthropogenic activities have now impacted

66
 GOME and SCIAMACHY

all aspects of the Earth system, with pollution at all scales from local to global.
Within a geologically very short period of 200 years, the human population
grew from 1 to 7 billion in 2011. To put this growth into context, it took until
about 1925 for the global population to reach 2 billion, but only 20 years, since
the launch of ERS‑1 in 1991, to add another 2 billion.

3. Physical Structure and Chemistry of the Atmosphere

Earth’s atmosphere comprises the gases released from its surface and produced
by photochemically induced reactions. After Earth was formed the temperature
of the surface was sufficiently high that gases and particles outgassing from
the surface had sufficient kinetic energy to escape the gravitational field. After
Earth cooled sufficiently, an atmosphere in geochemical equilibrium was
formed, with a chemical composition that was very different from that of the
present atmosphere: the latter being maintained in an evolving biogeochemical
stationary state by the presence of life.
Life on Earth produced reactive O2, which was released by bacteria. O2
does not have sufficient kinetic energy within the atmosphere to escape from
Earth’s gravitational field and thus started to build up in the atmosphere.
The photolysis of O2 protects Earth’s surface from short-wavelength UV solar
radiation. This photodissociation results in the production of ozone, O3, in the
upper atmosphere, which absorbs strongly UV radiation between 200 nm and
320 nm and weakly at longer wavelengths.
Some molecular nitrogen, N2, is geological in origin, being emitted from
volcanoes or outgassed from the mantle and crust. However, life also results in
the bacterial decomposition of ammonium nitrate, NH4NO3, in soils to produce
predominantly N2, some nitrous oxide, N2O, and nitric oxide, NO. The latter
is also produced in biomass burning where ionic processes and temperatures
in flames enable the equilibrium between N2 and O2 to produce significant
amounts of NO.
N2 and N2O are chemically long lived in the lower atmosphere, where they
are transported by convection and advection and are mixed by both turbulence
and diffusion. In the upper atmosphere they are photodissociated by short-
wavelength solar radiation. In this sense they provide a first protective layer
against biologically damaging radiation.
The atmospheric temperature and pressure as a function of altitude
are shown in Fig. 2. Temperature is expected to fall with altitude in any
planetary atmosphere, as a result of adiabatic expansion. This is the case in
the troposphere, which, as a result of convection and advection, is well mixed.
The O3 absorption heats the atmosphere creating a temperature inversion
layer, which begins at the tropopause and is called the stratosphere. At
the stratopause the heating by O3 absorption becomes negligible and the
temperature in the mesosphere falls once more with altitude. At the mesopause,
shortest-wavelength electromagnetic radiation from the Sun is absorbed by
atoms and molecules. As a result the temperature rises in the thermosphere
and the composition is no longer well mixed. The mesosphere marks roughly
the separation between the heterosphere and the homosphere where the bulk
composition of air is constant, comprising ~78% N2 and ~21% O2.

67
SP-1326

Figure 2. Atmospheric conditions in the

mid-latitudes, using the US Standard
Atmosphere defined in 1976. The
tropopause is shown for mid-latitude
conditions and the cumulonimbus cloud for
topical conditions where the tropopause is
at around 16 km. (WMO, 2010)

3.1 Thermospheric and Mesospheric Chemistry

The thermosphere is heated by the absorption of short-wavelength solar
radiation and the reaction of coronal mass ejections impacting in this region.
These interact with the upper atmosphere. In addition, meteoritic dust enters
the upper atmosphere. In the thermosphere, the photodissociation of O2
(λ < 214 nm) and N2 (λ < 100 nm) yield atoms that are in turn photoionised to
ions and electrons. The result is a soup of ions, electrons, free radicals and
molecules. The emissions from the upper atmosphere have been observed by
GOME on ERS‑2, in particular, emissions of the metal atoms in the mesosphere.

3.2 The Stratosphere and Stratospheric Ozone

Ozone plays a number of important roles in the stratosphere. By absorbing
biologically damaging UV radiation, including UV-B (280−390 nm), O3 protects
the biosphere and so was important in the evolution of life. The UV-B radiation
at the surface is absorbed by DNA, leading to mutations and cancer. By
absorbing the UV radiation the stratosphere warms causing the temperature
inversion known as the tropopause. The height of the tropopause influences

68
 GOME and SCIAMACHY

the weather in the troposphere, and thus the aviation industry. In the past
40 years it has been established that humans have influenced the amount and
distribution of stratospheric O3 in a number of ways.
Following the discovery of what turned out to be O3 around electrical
discharges by van Muren in the 18th century and its identification and naming
by Schönbein in 1839, significant progress was made during the late 19th and
early 20th centuries with the identification O3 in the atmosphere. In 1879, Cornu
observed a sharp cutoff around 300 nm in the UV solar spectrum. Subsequently,
Hartley measured the O3 absorption cross section in the laboratory and recognised
that the atmospheric UV cutoff is produced by the presence of ozone. The existence
of a temperature inversion layer in the stratosphere was discovered independently
by French meteorologist Teisserenc de Bort and German meteorologist Assmann
in 1902. Huggins reported several absorption bands in the observed spectrum of
the star Sirius between 320 nm and 360 nm in 1890. In 1913 Strutt (Lord Rayleigh)
showed that the UV absorption does not happen in the lower atmosphere. In
1917, the bands identified by Huggins were attributed to the presence of ozone in
the atmosphere by A. Fowler and R.J. Strutt. In 1920 at Marseilles, France, Fabry
and Buisson used UV absorption properties to deduce that the thickness of
atmospheric O3 was of the order of 3 mm STP, later to become 300 Dobson units.
Our understanding of the global nature of atmosphere began with
measurements of O3 in the stratosphere by Dobson and his contemporaries
in the 1920s. This led to our understanding of the dynamics of the upper
atmosphere, dominated by Brewer–Dobson circulation. Similarly, the
production and loss of O3 was a matter of much debate. In 1929 Chapman
explained the maximum of O3 by the competition production and loss of O3.
The production of O3 in the upper atmosphere proceeds by the photolysis of O2
and the molecular reaction of oxygen atoms with O2,

O2 + hν (λ < 242 nm) " O + O (1)

O + O2 + M " O3 + M (2)

This is followed by the loss of O3 via photolysis and the proposed ‘odd-oxygen’
reaction between O and O3:

O3 + hν (λ < 411 nm) " O(1D) + O2 (3a)

+ hν (λ < 1180 nm) " O(3P) + O2 (3b)
O + O3 " O2 + O2 (4)

Over the next 50 years the evolution of reaction kinetics enabled exact
measurements of the elementary reaction rate coefficients. The rate of reaction
(4) was found to be too slow to explain the loss of O3 in the upper atmosphere.
Following the observation of emissions of excited OH by Meinel (1950), which is
explained by the reaction of hydrogen atoms, H, with O3, Bates & Nicolet (1950)
recognised that reactions of HOx might catalytically destroy O3. This introduces
the generic odd oxygen loss mechanism, which is shown below:

Ozone catalytic destruction cycle 1-L:generic ‘odd oxygen’ cycle:

X + O3 " XO + O2 (5)
XO + O " X + O2 (6)
_____________________
Net: O + O3 " O2 + O2

where X is either a hydrogen atom, H, or the hydroxyl radical, OH.

69
SP-1326

Ozone catalytic destruction cycle 1a-L: HOx ‘odd-oxygen’ cycle:

H + O3 " OH + O2 (7)
OH + O " H + O2 (8)
_______________________
Net: O + O3 " O2 + O2

Ozone catalytic destruction cycle 1b-L: HOx ‘odd-oxygen’ cycle:

OH + O3 " HO2 + O2 (9)

HO2 + O " OH + O2 (10)
_______________________
Net: O + O3 " O2 + O2

It was subsequently shown that NO, chlorine atoms, Cl, bromine atoms, Br,
or even iodine atoms, I, are effective catalysts for this ‘odd-oxygen’ catalytic
cycle 1L. The HOx (= ΣH +OH + HO2 ) cycle largely determines the loss of O3
in the mesosphere and to a lesser extent in the stratosphere. In the late 1960s,
both Harold Johnston and Paul Crutzen recognised the potential importance of
emissions of oxides of nitrogen from high-flying supersonic aircraft proposed
at the time (Johnston, 1971, 1974; Crutzen, 1970, 1973).
Crutzen further proposed a global odd-oxygen catalytic cycle involving
oxides of nitrogen (X = NO) destroying O3. There are two sources of NO in the
stratosphere: the first being transport down from the upper atmosphere, where
it is produced by the interaction of short-wavelength solar radiation with the
bulk atmospheric constituents. The second is the reaction of N2O: photolysis
and reaction of excited oxygen atoms, O(1D), produced in the photolysis of O3
reaction (3):

O(1D) + N2O " NO + NO (11a)

" N2 + O2 (11b)

NO in the atmosphere is also produced by ion–molecule reactions in the upper

atmosphere and by lightning. At the high temperatures that occur in plasmas,
and jet and internal combustion engines, the equilibrium reaction of N2 and O2,
as expected from Le Chatalier’s principle, is shifted to the right-hand side and
produces NO.

N2 + O2 D NO + NO (12)

Currently the NOx odd-oxygen cycle is considered to determine approximately

70% of O3 loss in the stratosphere. The N2O is released at the surface by the
bacteriological reduction of nitrate ions, NO3–, and the oxidation of ammonium
ions, NH3+. The importance of N2O, whose atmospheric loading, as result of the
dramatic increase in the use of fertilisers since the development of the Haber–
Bosch process, both as a greenhouse gas and as a source of stratospheric NOx,
is well recognised.

Ozone catalytic destruction cycle 1c-L: NOx ‘odd-oxygen’ cycle

NO + O3 " NO2 + O2 (13)

NO2 + O " NO + O2 (14)
_______________________
Net: O + O3 " O2 + O2

The potential for a chlorine catalytic chain reaction in removing O3 was recognised
by Stolarski & Cicerone (1974), who studied the impact of Space Shuttle exhaust
on the upper atmosphere. However, Molina & Rowland (1974) proposed that the

70
 GOME and SCIAMACHY

release of long-lived chlorofluorocarbon compounds (CFCs) in the troposphere,

their transport to the stratosphere and the subsequent release of chlorine atoms,
Cl, which then participate in the catalytic odd-oxygen ClOx cycle removing O3.
Since the industrial production of CFCs began in the 1930s their concentrations
in the stratosphere had increased exponentially. Further, it was recognised by
Wofsey et al., (1975) and others that bromine atoms, Br, released from the halons
used in fire extinguishers, and methyl bromide, CH3Br, would also destroy O3 in
an efficient catalytic BrOx cycle. The international concern about the release of
halogens into the stratosphere eventually led to the introduction of controls on the
production of CFCs and other O3-depleting species.
The appearance of the phenomenon now known as the ozone hole – i.e. very
low values of the lower stratospheric O3 column in the polar vortex over Antarctica
in spring – was entirely unexpected, and a game changer. This was first reported
from ground-based observations by British Antarctic Survey scientists at the Halley
Bay station in 1985 (Farman et al., 1985). Some indication of this behaviour from
one season’s measurements had been reported by Shigeru Chubachi, a Japanese
scientist at an ozone symposium around the same time. The global O3 observations
from NASA satellite instruments, in particular the Solar Backscattered Ultraviolet
(SBUV) spectrometer and Total Ozone Mapping Spectrometer (TOMS) flown on
Nimbus‑7, showed the magnitude, extent and duration of this phenomenon, which
begins in winter and extends long into spring. They also showed the influence of
the stratospheric quasi-biennial oscillation, a tropical dynamical feature, on the
development and size of these holes.
After much international scientific endeavour, with aircraft campaigns led by
NASA and NOAA, it was established that the destruction of stratospheric O3 at
high latitudes in the austral spring was the result of a complex physicochemical
mechanism involving CFCs. During autumn at high latitudes a dynamical feature
called the polar stratospheric vortex spins up. This air mass has high potential
vorticity and is therefore effectively separated to a first-order approximation
from lower-latitude air masses and cools, enabling Polar Stratospheric Clouds
(PSCs) to form. During these and subsequent international expeditions, it was
established that these PSCs are complex mixtures. Sulphuric acid tetrahydrate,
H2SO4.4H2O, forms at the low temperatures present in the polar vortex and these
particles act as condensation nuclei for nitric acid trihydrate, HNO3.3H2O and
ice, which are stable at temperatures below ~194K and ~185K, respectively, under
the conditions present in the lower stratosphere.
The mechanism of producing the ozone hole by processing on the PSC is
complex and explained briefly below. During daylight in the lower stratosphere
where oxygen atoms react preferentially with oxygen molecules to form O3,
nitric oxide and nitrogen oxide participate in a ‘do-nothing’ cycle comprising
reactions (13), (15) and (2):

Ozone and oxygen atom exchange cycle: ‘do-nothing’

NO + O3 " NO2 + O2 (13)

NO2 + hν (λ < 390 nm) " NO + O (15)
O + O2 + M " O3 + M (2)

NO2 reacts slowly with O3 to form NO3, which in daylight is photolysed in the
visible:

NO2 + O3 " NO3 + O2 (16)

NO3 + hν (λ < 580 nm) " NO2 + O (17a)
+ hν (λ < 7320 nm) " NO + O2 (17b)

Lovelock & Maggs (1972) had discovered that CFCs, in particular CFC-11,
CF3Cl, were present in the troposphere in amounts similar to those produced
industrially. Stolarksi & Cicerone (1974) recognised that chlorine could

71
SP-1326

also catalyse O3 in an odd-oxygen cycle. However, Molina &Rowland (1974)

proposed that the CFCs released in the troposphere were a new source of Cl in
the stratosphere and would deplete O3.
In the unperturbed stratosphere outside of the vortex chlorine and
bromine atoms are produced as a result the release of, for example, CFC-11,
CF3Cl, CFC-12, CF2Cl2, halon 1211, CF2BrCl, and methyl bromide, CH3Br, and
in the meantime HCFCS, their transport to the stratosphere and subsequent
photolysis:

CF3Cl + hν (λ < 300 nm) " CF3 + Cl (18)

CF2Cl2 + hν (λ < 300 nm) " CF2Cl + Cl (19)
CH3Br + hν (λ < 548 nm) " CH3 + Br (20)
O(1D) + CF3Cl " CF3 + ClO (21)
O(1D) + CF2BrCl " CF2Cl+ BrO (22)

The Cl and Br atoms then participate in the following catalytic destruction

cycles for stratospheric O3:

Ozone catalytic destruction cycle 1d-L: ClOx ‘odd-oxygen’ cycle:

Cl + O3 " ClO + O2 (23)

ClO + O " Cl + O2 (24)
_______________________
Net: O + O3 " O2 + O2

Ozone catalytic destruction cycle 1e-L: BrOx ‘odd-oxygen’ cycle:

Br + O3 " BrO + O2 (25)

BrO + O " Br + O2 (26)
_______________________
Net: O + O3 " O2 + O2

Outside the vortex these atoms account for approximately 30% of the O3 removal.
These mechanisms currently remove about 6% of the total O3 at mid-latitudes
in the upper stratosphere and lower mesosphere. It is interesting to note that the
BrOx cycle has a chain length about a factor of 10 or more longer than that for
the ClOx cycle. As the industrial use of the cheaper CFCs was much greater than
the halons and methyl bromides, this results several parts per billion of Cl but
only 10–20 parts per trillion by volume of Br compounds. This turns out to be
fortunate, because if as much bromine had been released to the stratosphere by
anthropogenic activities as chlorine, then the O3 stratospheric layer would then
have been reduced globally by a factor of 10 or more. This would have had far
more dramatic consequences for the atmosphere and Earth’s surface, as well as
much larger doses of UV-B to the biosphere and humans.
At night, the equilibrium reaction (27), which is strongly temperature
dependent in the polar vortex, results in the formation of acid anhydride of
nitric acid, dinitrogen pentoxide, N2O5:

NO3 + NO2 + M D N2O5 + M (27)

During the growth of PSCs, complex heterogeneous reactions take place on and
in liquid and ice surfaces. These involve the temporary reservoir compounds
for the nitrogen and halogens such as N2O5 hydrogen chloride, HCl, chlorine
nitrate, ClONO2 and hypochlorous acid, HOCl. The actual mechanisms involve
the accommodation of species at the particular surface, diffusion across the
interface, diffusion through the liquid or solid PSCs, and then reactions within
the PSCs and the subsequent diffusion of products to the surface, diffusion
across the surface and release back to the atmosphere. For simplicity, these

72
 GOME and SCIAMACHY

complex elementary processes can be summarised by the following acid-

catalysed exothermic equilibrium reactions:

N2O5g + H2Ol D 2HNO3l (28)

ClONO2g + H2Ol D HNO3l + HOClg (29)
HOClg + HCll D CL2g + H2Ol (30)
ClONO2g + HCll D Cl2g + HNO3l (31)
ClONO2g + H2Ol D ClONOg + HOCll (32)
HOBrg + HCll D BrClg + H2Ol (33)
BrONO2g + HCll D BrCl2g + HNO3l (34)
BONO2g + H2Ol D HNO3l + HOBrg (35)
HOBrg + HBrl D Br2g + H2Ol (36)

The aerosol and PSC surfaces provide opportunities for chemical reactions,
bringing together chemical species for sufficient time that chemical reactions
take place, which do not occur rapidly enough during a gas phase collision. The
result, at the end of winter, is that the polar vortex in the southern hemisphere
is populated by PSCs and photolabile halogen compounds – molecular
chlorine, Cl2, bromine chloride, BrCl, and to a lesser extent molecular bromine,
Br2, and chlorine nitrite, ClONO – which are then photolysed at sunrise in the
polar vortex:

Cl2 + hν (λ < 493 nm) " Cl + Cl (37)

BrCl + hν (λ < 548 nm) " Br + Cl (38)
Br2 + hν (λ < 620 nm) " Br + Br (39)

The Cl and Br atoms released participate in catalytic destruction cycles of the

type:

Ozone catalytic destruction cycle 2a-L: ClO dimer

2x(Cl + O3 " ClO + O2) (23)

ClO + ClO + M D Cl2O2 + M (40)
Cl2O2 + hν (λ < 620 nm) " Cl + ClOO (41)
ClOO + M D Cl + O2 + M (42)
______________________________________
Net: 2O3 " 3O2

Ozone catalytic destruction cycle 2b-L: BrO ClO

Cl + O3 " ClO + O2 (23)

Br + O3 " BrO + O2 (25)
BrO + ClO " Cl + Br + O2 (43a)
____________________________
Net: 2O3 " 3O2

Ozone catalytic destruction cycle 2c-L: BrO

2x(Br + O3 " BrO + O2) (25)

BrO + BrO " Br + Br + O2 (44)
____________________________
Net: 2O3 " 3O2

The reaction (43) between ClO and BrO has several channels:

BrO + ClO " OClO + Br (43b)

BrO + ClO " BrCl + O2 (43c)

73
SP-1326

The OClO, which is itself rapidly photolysed, can be used to detect the
activation of halogens in the vortex:

OClO + hν (λ < 433.9 nm) " O + ClO (45a)

OClO + hν (λ < 4483 nm) " Cl + CO2 (45b)

An additional relatively slow catalytic destruction cycle for O3 by HOx is

feasible in the lower stratosphere and troposphere.

Ozone catalytic destruction cycle 2d-L: HOx

OH+ O3 " HO2 + O2 (9)

HO2 + O3 " OH + 2O2 (46)
_____________________________
Net: 2O3 " 3O2

The above processes are shown schematically in Fig. 3.

Figure 3. Schematic diagram of the production and catalytic destruction of stratospheric and mesospheric O3, showing the gas phase odd-
oxygen catalytic destruction cycles and the perturbed chemistry involving heterogeneous and multiphase reactions on polar stratospheric
clouds, the release of photolabile halogen compounds, their photolysis and the catalytic destruction of O3 cycles in the gas phase in the
absence of NOx. (J.P. Burrows and S. Noel, University of Bremen)

74
 GOME and SCIAMACHY

In the stratosphere the reactions between the different free radicals produce
temporary reservoirs by gas phase in addition to the multiphase reactions (28)
to (36):

OH + NO + M " HONO + M (47)

OH + NO2 + M " HNO3 + M (48)
Cl + CH4 " HCl + CH3 (49)
ClO + NO2 + M " ClONO2 + M (50)
2NO2 + H2O " HONO + HNO3 (51)

The HONO and ClONO are both rapidly photolysed during the day in both the
stratosphere and troposphere:

HONO + hν (λ < 579 nm) " OH + NO (52)

ClONO + hν (λ < 1498 nm) " ClO + NO (53)

Based on the measurements of changing stratospheric O3, scientific studies of

its origins and rising public concern, the Vienna Convention for the Protection
of the Ozone Layer was agreed in early 1985, prior to the discovery of the ozone
hole above Antarctica. The parties to the convention, which is implemented
by UN Environment Programme (UNEP) and the World Meteorological
Organization (WMO), then signed the Montreal Protocol in 1987, which banned
the production of ozone-depleting substances (ODSs). The Montreal Protocol
has very effectively reduced the production and release of CFCs and other
ODSs, although the long lifetimes of these substances in the troposphere mean
that levels of stratospheric halogens will not fall to pre-1980 levels before 2040.
Since the decision to include GOME on ERS‑2, the ozone hole above
Antarctica has developed and is now very stable. In the northern hemisphere
the polar vortex is not as strong or as long lasting. However, GOME was able
to measure the first and second clear ozone holes over the Arctic in 1997 and
2011. They occur when the dynamical conditions are suitable and produce cold,
long-duration polar vortices in the northern hemisphere.
One of the current issues in stratospheric science is the changing dynamics
produced by a changing climate resulting from both natural phenomena and
anthropogenic activities. Changing the circulation and the composition of
the stratosphere has large potential consequences, which is why we need
measurements of the total column amount and vertical profile of O3. The
GOME observations of the total column O3 averaged for the month of March
are shown in Fig. 4 for the two very large chemical depletions of O3 observed
thus far in the polar vortex above the northern hemisphere in 1997 and 2011.
These springtime ozone holes occurred in a cold stratosphere with stable
polar vortices that lasted into March. They are explained by the chemistry and
dynamics outlined above.
From the above, it can be seen that climate change (Houghton et al.,
1992, and subsequent IPCC assessment reports) is likely to have an impact on
stratospheric O3. It had already been noted by Tuck and co-workers (Groves et
al., 1978; Groves & Tuck, 1979) that the cooling of the stratosphere predicted
for increasing atmospheric CO2 would be offset to some extent by an increase
in the rate of production of O3 as reaction (2) has a negative temperature
dependence, a phenomenon now termed the super-recovery of ozone. However,
changes in the dynamics, coupled with changes in CH4, N2O and H2O, will
also impact. GOME and other data have provided evidence of changes in the
Brewer–Dobson circulation. It is early days in the study of this phenomenon
and its consequences are not yet clear. The importance of the increasing N2O
loading of the troposphere, resulting from anthropogenic fertilisation, both as
a greenhouse gas and more recently as an ozone-depleting substance has been
recognised (Ravishankara et al., 2009; for details of the basic mechanisms, see
Wayne, 1992; Holloway & Wayne, 2010).

75
SP-1326

Figure 4. Ozone holes above the northern

and southern hemispheres (NH/SH) in
spring, depicted using the merged total
column O3 retrieved in Dobson Units, DU,
from measurements by GOME in 1997 and
by GOME‑2 in 2011. (M. Weber and J.P.
Burrows, University of Bremen)

3.3 Tropospheric Chemistry, Air Pollution, Biogeochemistry→

and Feedback
Atmospheric pollution, its consequences and the need to legislate to control its
impact are not new phenomena. Even in the 14th century, the burning of fires
in London was prohibited, at least temporarily. The process of urbanisation
and industrial processes, powered by fossil fuels, generate both water and
air pollution. This has led to urbanisation with now over 50% of the world’s
population living in cities. Since the Haber–Bosch process was invented about
100 years ago to fix nitrogen from the air to produce ammonia, NH3, the use of
ammonium nitrate, NH4NO3, for fertiliser, together with industrially produced
pesticides, has resulted in huge increases in food production, but also changes
in land use, water and air pollution, and climate change. The impacts of human
activities now extend from the local to the global scale.
The meteorological conditions of a location often play an important role in
the type of pollution observed. In 1905 H.A. des Voeux first used the term smog
to describe the mixture of smoke and fog that occurred in London in winter.
The smoke from domestic fires and industrial processes led to acidic smog
particles containing a cocktail of hazardous materials. Visibility was so poor
that this type of smog was described by Cockneys as ‘pea-soupers’. London’s
great smog of 1952, although it was not considered particularly unusual at
the time, caused an estimated 12 000 premature deaths and 100 000 cases of

76
 GOME and SCIAMACHY

respiratory disease, and led the UK parliament to introduce the world’s first
clean air act in 1956.
Since then, governments around the world have introduced legislation
to control local emissions, while international institutions have supported
environmental agreements to attempt to manage the problem at the global scale.
In 1977, UNEP concluded a World Plan of Action on the Ozone Layer, which called
for international research and monitoring, and in 1981 drafted a global convention
to protect stratospheric ozone. As mentioned above, the Vienna Convention for
the Protection of the Ozone Layer, concluded in 1985, is a framework agreement in
which states agreed to cooperate in relevant research and scientific assessments of
the ozone problem, to exchange information, and to adopt ‘appropriate measures’
to prevent activities that harm the ozone layer. The obligations were general and
contain no specific limits on chemicals that deplete the ozone layer. However, the
latter changed with the Montreal Protocol, as described above.
Since 1979 the Convention on Long-Range Transboundary Air Pollution
(LRTAP), administered by the UN Economic Commission for Europe (UNECE),
has addressed some of the major environmental problems of the UNECE region
through scientific collaboration and policy negotiations. The Convention has
been extended by eight protocols that identify specific measures that states
need to take to cut their emissions of air pollutants. The LRTAP Convention
and its protocols, and the Vienna Convention and its Montreal Protocol and
amendments (see below), have had some success, but the UN Framework
Convention on Climate Change (UNFCCC) and its Kyoto Protocol have been
far less effective and are not reducing emissions of greenhouse gases to the
atmosphere from anthropogenic activity.
The LRTAP Convention addresses acidic deposition that is the result of
emissions of SO2 and NO2. During the day, SO2 is oxidised by OH radicals in the
gas phase:

OH + SO2 + M " HSO3 + M (54)

HSO3 + O2 " HO2 + SO3 (55)
SO3 + H2O " H2S04 (56)

Sulphuric acid, H2SO4, which has a low vapour pressure and is hygroscopic,
immediately forms aerosol condensation nuclei and aerosol particles result.
In addition, the SO2 is oxidised in the aerosol and cloud particles by a
complex reaction involving hydrogen peroxide, H2O2, formed in the gas phase
in the disproportionation of the free radical hydroperoxyl, HO2:

HO2 + HO2 " H2O2 + O2 (57)

H2O2g D H2O2l (58)
SO2g D SO2l (59)
H2O2g + SO2l D H2SO4l (60)

Nitric acid, HNO3, is formed primarily during the day by reaction (48) and at
night by reaction (28). It absorbs more strongly in the UV-B and UB-C regions,
and is only slowly photolysed in the troposphere compared to HONO. In the
troposphere, HNO3, which is soluble, dissolves in aerosol and cloud particles
and often reacts with NH3, which is also very soluble. The NH3 is released from
soils by the oxidation of NH4+ and the reduction of NO3–.
In the 1940s, residents of Los Angeles experienced a different type of smog
during summer (Haagen Smit, 1952; Haagen Smit et al., 1952). Again this was
associated with reduced visibility, but in this case O3 and a variety of organic
compounds, in particular peroxyacetyl nitrate, CH3CO.O2.NO2, were found to
be present in large amounts, as well as aerosols. It turns out that the processes
that produce this type of local smog also occur at the global scale.
Pollutants released from Earth’s surface are characterised as short or long
lived, according to their lifetimes in the troposphere. Air in the troposphere and

77
SP-1326

the stratosphere is exchanged on a timescale of 5−7 years. The combustion of

fossil fuels and biomass burning release both short- and long-lived pollutants.
The short-lived oxides of nitrogen, NOx (Σ ([NO] + [NO2]) and VOCs, which
include a large number of rapidly oxidised biogenic and man-made aliphatic
and aromatic gases, have natural tropospheric sources. NOx is produced from
the breakdown of NH4NO3 by bacteria in soils, and by lightning. Biogenic
emissions release only VOCs. Fire and biomass burning release both NOx
and VOCs. Fossil fuel combustion releases NOx, CO, CO2, SO2 and some
aromatics. Fossil fuel and biomass burning and land use change have modified
dramatically the inputs of NOx and VOCs. Methane CH4 and other VOCs are
released by oil refining and from coal mining. The new industrial practice of
fracking, used to release CH4 from shales, and the exploitation of tar sands are
likely to release CH4 and other VOCs to the atmosphere.
These substances, coupled with solar UV and visible radiation, are the
ingredients or fuel for the low-temperature chain reactions that oxidise the
VOCs and result in smog formation relatively close to their source release
region. In addition, the release of these short-lived species to the troposphere
results in their transformation into toxic and noxious secondary pollutants,
such as ozone, peroxyacetyl nitrate, CH3CO.O2.NO2 and aerosols and their
subsequent transport throughout the troposphere. These constituents and
their precursors have negative impacts on both human health and agriculture,
and may also result in significant radiative forcing.
The oxidation of CH4, which is long lived and is present throughout the
troposphere, results, in the presence of sufficient NOx, in a chain reaction that
produces O3 via the following mechanistic pathways:

Cycle 1-P: ozone production catalytic cycle in the oxidation of hydrocarbons

OH + CH4 " CH3 + H2O (61)

CH3 + O2 + M " CH3O2 + M (62)
CH3O2 + NO " CH3O + NO2 (64)
CH3O + O2 " HCHO + HO2 (65)
HO2 + NO " OH + NO2 (66)
2(NO2 + hν (λ < 411 nm) " NO + O) (15)
2(O + O2 + M " O3 + M) (2)
______________________________________________________
Net: CH4 + 3O2 + hν " HCHO + H2O + 2O3

The formaldehyde, HCHO, is photolysed and oxidised by OH, producing HO2

and CO:

HCHO + OH " H2O + CO (67)

HCHO + hν (λ < 331 nm) " H + CHO (68a)
HCHO + hν (λ < 361 nm) " H2 + CO (68b)
H+ O2 + M " HO2 + M (69)
CHO + O2 " CO + HO2 (70)

The larger VOCs have more complicated oxidation pathways. Catalytic O3

production cycles involving VOCs such as isoprene and terpenes, or the
anthropogenic release of VOCs such as aromatic or aliphatic compounds,
take place above the land close to their source. Their oxidation produces large
amounts of the peroxy radicals, HO2 and RO2, both of which react with NO via
reactions (64) and (66) and lead to the catalytic production of NO2, which in
turn is photolysed to produce O3.
Another catalytic cycle that produces O3 is that involving CO, cycle 2P. As
a result of the supply of CO from the oxidation of CH4 and VOCs, this cycle
operates throughout the troposphere, where sufficient NOx is present. This
cycle was first pointed out by Fishman & Crutzen (1978, and references therein).

78
 GOME and SCIAMACHY

Cycle 2-P: ozone production catalytic cycle in the oxidation of CO

OH + CO + O2 " HO2 + CO2 (71)

HO2 + NO " NO2 + OH (65)
NO2 + hν (λ < 425 nm) " NO + O (15)
O + O2 + M " O3 + M (2)
____________________________________
Net: CO + 2O2 " CO2 + 2O3

The oxidation of VOCs also produces not only aldehydes but also ketones such
as acetone, CH3.CHO, which is oxidised to form acetyl peroxy radicals, which
in turn produce peroxyacetyl nitrate and the characteristic components of
summer smog, along with acidic aerosols:

OH + CH3CHO " CH3.CO (72)

CH3CO + O2 + M " CH3.CO.O2 (73)
CH3.CO.O2 + NO2 + M " CH3.CO.O2.NO2 + M (74)

In the absence of sufficient NOx, O3 is removed globally by the HOx catalytic

destruction cycle:

Ozone catalytic destruction cycle 3-L: HOx

OH + CO + O2 " HO2 + CO2 (71)

HO2+ O3 " OH +2 O2 (46)
_______________________________
Net: CO + O3 " CO2 + O2

In addition to the above reactions, another source of halogen for the

troposphere and in particular the boundary layer are biogeochemical
mechanisms that release the organohalogens methyl chloride, CH3Cl,
methyl bromide, CH3Br, methyl iodide, CH3I, dichloromethane, CH2CL2,
dibromomethane, CH2Br2, bromochloromethane, CH2BrCl, bromoform, CHBr3,
and others. These compounds also absorb in the ultraviolet and visible and are
photolabile, releasing halogen atoms.

I2 + hν (λ < 792 nm) " I + I (75)

CH3I + hν (λ < 498 nm) " CH3 + I (76)
CH3Br + hν (λ < 406 nm) " CH3 + Br (20)
CH3Cl + hν (λ < 342 nm) " CH3 + Cl (77)

Chlorine atoms react rapidly with aliphatic VOCs, including CH4, to produce
HCl. Br and I, in contrast, do not react with aliphatic hydrocarbons. The
bimolecular disproportionation reaction of ClO is lower than that of BrO dimer
ClOOCl. As a result the chain length of the Cl catalytic removal of O3 is short
compared with that of bromine via a loss cycle similar or equivalent to that of
bromine atoms in cycle 3c-L above.
The Br atoms catalytically destroy O3, via the cycle 2c-L and the mixed
chain reaction 2e-L:

Ozone catalytic destruction cycle 2e-L: BrOx–HOx

OH + O3 " HO2 + O2 (9)

Br + O3 w " BrO + O2 (25)
BrO + HO2 " HOBr + O2 (78)
HOBr +hν (λ < 425 nm) " OH + Br (79)
_______________________________________
Net: 2O3 " 3O2

79
SP-1326

However, on the surface of sea salt aerosols and very efficiently on cold brine
under what is called potential frost flower conditions at high latitudes, this
reaction becomes the multiphase chain reactions 3a-L and 3b-L (Kaleschke et
al., 2004). 3a-L is autocatalytic in Br, provided there is a sufficient source of OH.

Ozone catalytic destruction cycle 3a-L: BrOx–HOx–HBr

OH + O3 " HO2 + O2 (9)

Br + O3 " BrO + O2 (25)
BrO + HO2 " HOBr + O2 (78)
HOBrg D HOBrl (80)
HOBrg + HBrl D Br2g + H2Ol (36)
Br2 + hν (λ < 620 nm) " Br + Br (39)
_________________________________________________
Net: OH + HBrl + 2O3 " 3O2 + H2O + Br

Ozone catalytic destruction cycle 3b-L: BrOx–HOx–HCl

OH + O3 " HO2 + O2 (9)

Br + O3 " BrO + O2 (25)
BrO + HO2 " HOBr + O2 (77)
HOBrg D HOBrl (79)
HOBrg + HCll D BrClg + H2Ol (33)
BrCl + hν (λ < 548 nm) " Br + Cl (38)
______________________________________________
Net: OH + HCll + 2O3 " 3O2 + H2O + Cl

This chain reactions 3a-L and 3b-L are accelerated at low temperatures, as
the equilibriums (36) and (33) have relatively large enthalpies of reaction
and are acid catalysed. Cations are lost by precipitation from the brine at low
temperatures, forming ikaite, CaCO3·6H2O, calcite (CaCO3) and other minerals.
This results in quasi-liquid layers of brine-covered frost flowers, surface
snow or aerosols that are readily acidified. These multiphase chain reactions
are thus accelerated in at least two ways at low temperature. As a result, sea
salt aerosols at mid-latitudes release Br but at much slower rates than in the
potential frost flower conditions at cold high latitudes.
Chlorine atoms participate in a similar cycle, but the chain length is not as
effective as that of bromine atom, Br, because the rate of reaction Cl with CH4
is much less than that of Br + CH4. Nevertheless, the release of Cl2 produces Cl,
thereby removing some O3 and CH4. This results in the formation of the methyl
peroxy CH3O2 and the hydoxyperoxyl radical HO2 at least in the troposphere.

Ozone catalytic destruction cycle 2f-L: IO

2 × (I + O3 " IO + O2) (81)

IO + IO " I + I + O2 (82a)
_________________________________
Net: 2O3 " 3O2

The iodine oxide, IO, disproportionation reaction has several channels and
produces not only iodine atoms but also OIO and the dimer I2O2. The oxides
of iodine then produce higher oxides, which lead to the production of the acid
anhydrides I2O4 and I2O5, which are hygroscopic and result in new particle
formation:

IO + IO " OIO + I (82b)

IO + IO + M " I2O2 + M (83)
IO + OIO + M " I2O3 + M (84)

80
 GOME and SCIAMACHY

These pathways are currently the subject of much research investigating the
importance of this natural source of aerosol condensation nuclei. The roles
of glyoxal, CHO.CHO, and methyl glyoxal, CH3CO.CHO, which are formed in
the oxidation of biogenic and aromatic VOCs and related compounds, in the
formation of secondary organic aerosols is also of much interest. Current
studies focus on understanding better their emission to the troposphere,
their natural biogeochemical cycling, the transport and transformation of the
emitted species and the anthropogenic modification of these processes.
The global measurements by GOME and subsequent instruments
(SCIAMACHY and GOME-2) have greatly improved our understanding of global
air pollution and its transport and transformation. Important new areas of
application include numerical predictions of the environment, and the impact
of and feedback between climate change, which changes the temperature fields
and momentum fluxes in the atmosphere, and atmospheric chemistry and
dynamics and/or the hydrological cycle. The global measurements of water
vapour, aerosols and cloud parameters are significant here. Figure 5 shows a
schematic representation of these processes.
NOx species, the sum of NO and NO2 are very important in tropospheric
chemistry. Through the do-nothing cycle mentioned above, NO and NO2 are
often coupled together during the day through the Leighton photostationary
state defined by reactions (2), (13) and (15), where

[NO2]/[NO] = k13[O3]/J15

At lower NOx, the rate of reaction of NO with peroxy radicals and halogen
oxides becomes significant and the ratio of NO to NO2 becomes more complex.

Figure 5. Schematic representation of chemical and physical processes as well as some biogeochemical interactions in the troposphere that
control ozone and aerosols amounts. (J.P. Burrows and S. Noel, University of Bremen)

81
SP-1326

It turns out that even HO2 has a pressure-dependent pathway in its reaction
with NO. This forms HNO3. As the organic peroxy radicals become larger, the
equivalent pathway, which produces organic nitrates, competes ever more
effectively with the channel producing NO2. Thus the rate of production of O3
is dependent on the rate of the reaction of HO2 and the smaller organic peroxy
radicals with NO for the branch producing NO2. HNO3 and the organic nitrates
in the troposphere are often hygroscopic and enter the liquid phase. Nitrites,
which can be carcinogens, sometimes result (for details of tropospheric
chemistry, see Finlayson-Pitts & Pitts, 1986).
Also depicted in Fig. 5 are the surface sources and fluxes. The interaction
of pollutants with the surface and the impact on and modification of the
natural biogeochemical cycling has become a focus of scientific studies. In
addition to the chemically short-lived atmospheric constituents, a variety of
long-lived gases are released by natural phenomena anthropogenic activity.
Species such as the CFCs, their hydrogenated shorter-lived substitutes, the
hydrochlorofluorocarbons (HCFCs), polyfluorinated compounds (PFCs) and
halons, have been created by industry for applications such as refrigeration,
fire extinguishers, electrical circuit board cleaners, etc., and were not present
in the pre-industrial atmosphere. They are also greenhouses gases but they
are infamous as they are transported to the stratosphere, releasing halogens,
in particular Cl and Br, which then catalytically deplete stratospheric O3. In
addition, as a result of fossil fuel combustion and land use change, copious
amounts of CO2, CH4 and N2O are being released to the atmosphere.
The changes in temperature and precipitation patterns result in feedback
with chemical processing and dynamics. Understanding the Earth system and
its atmosphere has been a scientific endeavour since the birth of the modern
scientific method during the Renaissance. Many famous scientists, including
Boyle and Hooke, conducted key experiments of relevance to the atmosphere
and the Earth system. Apparently, Einstein considered that his determination
of Avogadro’s number was likely to be more important than his papers on the
photoelectron effect or special relativity, which were published in the same year.
Climate change driven by anthropogenic activity is also not as novel as
it sometimes appears. Fourier had already noted in 1824 that Earth’s surface
is warmed by the presence of an atmosphere, and in the late 19th century
John Tyndall, following studies of the interaction of radiation with material,
demonstrated that individual gases had different warming effects. In 1896,
Svante Arrhenius calculated that a doubling of atmospheric CO2 would lead
to a surface warming of approximately 4°C. But it is only in the last four
decades that the significance and importance of global pollution and climate
change, driven by anthropogenic activity, has been recognised. This is now
underpinned by objective scientific evidence provided by observations of
changing atmospheric composition and more recently of the diminishing
extent of sea ice and rising sea surface temperatures.
The rate of development of understanding of the atmosphere and its
interactions with Earth’s surface, described briefly in this chapter, has
followed the exponential growth in the demand for energy since the industrial
revolution. In the search for knowledge, scientific curiosity endeavours to
understand the workings of the Earth system and its past behaviour and
to predict its future behaviour. It is recognised that achieving sustainable
development will require global management strategies. There are currently
three relevant international agreements: the UNECE Convention on Long-
Range Transboundary Air Pollution, the UN Vienna Convention for the
Protection of the Ozone Layer and its Montreal Protocol on Substances that
Deplete the Ozone Layer, and the UNFCCC and its Kyoto Protocol.
Since the dawn of the space age, Earth observation has evolved as the science
that meets the need for the global observations of the key parameters describing
Earth’s atmosphere and surface. In order to understand the atmosphere and its
changing behaviour, global measurements of key constituents (gases, clouds

82
 GOME and SCIAMACHY

and aerosols) are required to test our understanding. The development of space-
based measurements of atmospheric composition and surface parameters has
been one of the greatest societal, scientific and engineering feats of the past half
century. The following sections document some of the roles played by the ERS
satellites in this process.

4. The Earth System and the Impact of

Anthropogenic Activities
The Earth system comprises the Sun, the atmosphere, land and oceans,
and the physical, chemical and biological processes and their feedbacks,
determine the conditions experienced by the biosphere. The system is in a
complex biogeochemical equilibrium that has been evolving since life began
some 3.5 billion years ago.
As noted above, between the Neolithic revolution and about 1800, the
human population rose from several million to about 1 billion, powered
primarily by biofuel and alternative energy sources. Since the industrial
revolution there has been a rapid increase in the Earth’s population and
its standard of living. The explosion of economic growth in all its various
incarnations has relied on the availability of relatively inexpensive energy
produced from fossil fuels. This has led to unprecedented wealth and
previously unimaginable supplies of food and luxuries, at least for those in the
developed world.
Mankind has implicitly assumed that Earth’s resources – the land, oceans
and the atmosphere – are effectively inexhaustible. It is now recognised
that resources are not unlimited, and they are rapidly being used up. The
rise of humans has been accompanied by the destruction of ecosystems and
biodiversity, which impacts on ecosystem services.
The influence of human activities on Earth’s environment since the industrial
revolution has created a new geological epoch, called the Anthropocene. This
is generally considered to have begun with the industrial revolution, although
some argue that it started at the Neolithic revolution. One important societal
issue has been the creation and establishment of political and economic systems
that ensure an acceptable, fair and commensurate distribution of the growing
affluence during the Anthropocene. A new concern is the ability of mankind
to manage the human impact on the Earth system and thereby preserve the
environment and the biosphere, achieving sustainable economic development.
Perhaps the most obvious consequences of economic development have
been the rising levels of air and water pollution. Less obvious but equally
important has been the anthropogenic modification of the climate. Any
strategy to cope with the consequences of the release of atmospheric pollutants
requires changes in the way we use land and oceans. Pollution now extends
across all scales from the local to the global. Global measurements of key
atmospheric constituents are needed to manage the Earth system, and to
improve

—— our understanding of the role of biogeochemical cycling and natural

phenomena within the Earth system;
—— assessments of the impacts of anthropogenic emissions and modification of
the land and oceans;
—— our knowledge of feedback mechanisms within the Earth–atmosphere
system;
—— our ability to identify and provide early warnings of change;
—— the accuracy of predictions of weather and climate change; and
—— the evidence base for international policies aimed at mitigating and adapting
to climate change.

83
SP-1326

Remote sensing instruments on orbiting satellite platforms provide potentially

such objective global measurements, and complement the measurements made
by the sparse network of measurements made at ground stations.
The SCIAMACHY project and its spinoff, GOME, were responses by the
scientific community to the need for global measurements of atmospheric
constituents and the provision of an early warning of change, developed in the
early 1980s. GOME flew on ESA’s second Earth Research Satellite, ERS‑2, which
was launched on 20 April 1995, and its decommissioning was completed on
5 September 2011. ERS‑2 is now in circular orbit at around 570 km and will re-
enter the atmosphere in less than 15 years, in line with current agreements for
space debris management. Throughout and up to the end of the mission, GOME
functioned without significant problems. SCIAMACHY, the more ambitious
instrument, was flown on Envisat and, like GOME, has also worked excellently
in space. The following chapters describe some of the unique aspects of the
development of GOME and its success.

5. Atmospheric Remote Sensing from Space before

ERS‑2 and Envisat
The development of spaceborne measurements of atmospheric composition
began at the dawn of the space age. After an initial phase of technology
demonstration by both the Soviet space programme and NASA, focused
geophysical missions began to be developed. The retrieval of information about
trace gases by remote sensing techniques utilises knowledge of the absorption,
emission and scattering of electromagnetic radiation in the atmosphere. Gases
exhibit characteristic fingerprint spectra in emission or absorption:

—— rotational transitions are observed primarily in the far-infrared and

microwave spectral regions;
—— vibrational rotational transitions may be observed in the infrared; and
—— electronic transitions are mainly in the UV, visible and near-infrared spectral
regions.

This section discusses the advantages and disadvantages of each of these

spectral regions, and describes the various instruments utilising remote
sensing techniques to determine atmospheric composition. It documents
developments prior to ESA’s decision to build atmospheric measurements into
ERS‑2. The characteristics of some of the instruments that were flown prior to
or during the development of the SCIAMACHY project, and their objectives, are
summarised in Table 1 i.e. before and up to the early 1990s.

5.1 Remote Sensing of Trace Gases in the Far-Infrared and

Microwave Spectral Regions
All gases having a dipole moment exhibit active rotational transitions in
absorption and emission in the microwave (mm) and far-infrared (sub-mm)
spectral regions. The Doppler line widths of rotational spectra are relatively
narrow and the observed atmospheric emission or absorption is pressure-
broadened by collisions with air molecules. While on the one hand, this
enables the amount of gas at different pressures to be retrieved from an
understanding of the line shape of the spectral features, on the other, it
results in the broadening of the absorption or emission as the pressure in the
atmosphere increases, as shown in Fig. 2, leading to a loss of specificity in
the lower atmosphere. This limitation is compounded by strong atmospheric
absorption by water vapour and CO2 and results in saturation in many parts of
the spectral region.

84
 GOME and SCIAMACHY

Ground-based measurements are restricted to atmospheric windows

where the water vapour absorption is relatively small. For limb space-based
atmospheric sensing, this is less of a problem but retrievals of trace gases in
the upper troposphere are restricted to a limited number of candidates. For
many gases such measurements are optimal for sounding the stratosphere and
mesosphere. For the lower stratosphere and upper troposphere, there are some
interesting possibilities, but the range of applications is restricted.
Passive remote sensing instruments for both the millimetre, submillimetre
or far-infrared spectral regions have been developed for atmospheric remote
sensing. Some examples are briefly described in the following.

5.1.1 Microwave

Millimetre and submillimetre wave radiometers have been successfully used

from a variety of platforms to measure stratospheric gases. By using microwave
local oscillators and filter banks, or appropriate spectrometers, the observed
brightness temperatures can be measured at various frequencies or across an
emission line from a particular molecule.
Significant American and European efforts have gone into the development
of microwave and sub-mm sounding of the stratosphere. Some highlights of
these programmes have been measurements of elevated amounts of ClO using
ground-based (DeZafra et al., 1987), balloon-borne (Waters et al., 1981, 1984;
Stachnik et al., 1992) and aircraft-borne instruments (Crewell et al., 1993). OH
has been observed from an aircraft (Titz et al., 1995).
The Jet Propulsion Laboratory (JPL) research team have successfully flown
the Microwave Limb Sounder (MLS) on the Upper Atmospheric Research
Satellite (UARS), which has been measuring stratospheric profiles of ClO, O3,
H2O and HNO3 since 1992 (Waters et al., 1993) (see Table 1). The Microwave
Atmospheric Sounder (MAS), which flew three times on the Space Shuttle,
also measured ClO, O3 and H2O and provided a unique set of observations
(Hartmann et al., 1996). The Swedish, French Canadian and Finnish
collaboration Odin was initiated after Envisat and GOME and has made
very successful measurements since 2001. Its Sub-millimetre Microwave
Radiometer (SMR) instrument has retrieved many interesting species in the
stratosphere and mesosphere. The Microwave Limb Sounder (MLS) on NASA’s
Aura significantly extended the measurements of free radicals, in particular
OH. Aura was launched in 2004 and the MLS is continuing to work well.

5.1.2 Far-infrared

The spectral region 7−200 cm−1 has been used to study molecular species
in the stratosphere and mesosphere. In the far-infrared region, vibrations
having weak force constants and rotations of light molecules are active. For
these stratospheric measurements, Fourier Transform Spectrometers (FTS)
with relatively high spectral resolution have been used. Two balloon-borne
instruments have been flown a number of times (e.g. Carli et al., 1984; Chance
et al., 1989). More recently, an FTS instrument has been developed for flight on
a high-flying Safire aircraft (Table 1).
With the exception of H2O and O3, trace gas measurements made prior to
ERS-2 launch have exploited these long wavelength ranges exclusively for
stratospheric measurements.

85
SP-1326

Table 1. Some of the atmospheric remote sensing instruments flown prior to or during the development of the SCIAMACHY project , i.e. in or
before the early 1990s, and, the height range of measurements, target species and the satellite platform.

Instrument Height of Target species Platform

measurements*
ATMOS − Atmospheric Trace Molecule ST and upper TR O3, NOx, N2O5, ClONO2, Space Shuttle Spacelab-3 (1985); ATLAS-1, -2
Spectroscopy HCl, HF, CH4, CFCs, etc. and -3 (1992–94)
AVHRR − Advanced Very High Resolution TR Smoke, fire, clouds TIROS-N, NOAA-6, -8, -10 (four channels);
Radiometer aerosols, vegetation NOAA-7, -9, -11, -12, -13 (five channels)
(1978−present)
BUV − Backscatter Ultraviolet Ozone Experiment ST, TR, profiles O3 Nimbus‑4 (1970–74)
LIMS − Limb Infrared Monitor of the ST CO2, HNO3, O3, H2O, Nimbus‑7 (1978−79)
Stratosphere NO2
LRIR − Limb Radiance Inversion Radiometer ST CO2, O3 Nimbus‑6 (1975)
MAPS − Measurement of Air Pollution from TR CO STS-2 (1981); Space Shuttle (1984, 1994)
Satellites
MAS − Microwave Atmospheric Sounder ST ClO, O3, H2O Space Shuttle ATLAS-1, -2 and -3 (1992–94)
SAM II − Stratospheric Aerosol Measurement II ST Aerosols Nimbus‑7 (1979−90)
SAMS − Stratospheric and Mesospheric ST, ME CO2, H2O, CO, N2O, Nimbus‑7 (1979−90)
Sounder CH4, NO
SBUV − Solar Backscatter Ultraviolet Ozone ST, TR, profiles O3 Nimbus‑7 (1979−90)
Experiment
SBUV-2 − Solar Backscatter Ultraviolet Ozone ST, TR, profiles O3 NOAA-9 (1985–present),
Experiment 2 NOAA-11 (1989−95), NOAA‑14 (1995−present)
SCR − Selective Chopper Radiometer ST CO2, T Nimbus‑4−5 (1970−75)
SME − Solar Mesosphere Experiment ST, ME, profiles O3, O2(1∆), NO2 NASA (1983)
TOMS − Total Ozone Monitoring Spectrometer ST, TR, profiles O3 Nimbus‑7 (1979−92)
Earth Probe (1996−present)
Meteor (1992−94)
* ST – stratosphere; TR – troposphere; ME – mesosphere.

5.2 Remote Sensing of Trace Gases in the Thermal →

Mid-infrared
In the mid-infrared, molecules with a dipole moment have active vibrational
rotational transitions, which can be observed in both emission and absorption.
For the observation of tropospheric species, pressure broadening of the
spectral features is considerably less problematical than at longer wavelengths.
This is because the Doppler line width is proportional to the frequency of the
light and is therefore larger in the mid-infrared than at longer wavelengths.
Nevertheless, broadening of lines results in some loss of specificity of the
measurements. In addition, in the mid-infrared strong absorptions by H2O and
CO2 restrict the available spectral windows.
Passive remote sensing by mid-infrared spectroscopy has been applied
to the measurement of many stratospheric trace constituents and some
upper tropospheric constituents. Initially, measurements were made from
mountain tops (e.g. Zander, 1981), and balloon and aircraft experiments were
subsequently developed (Fischer et al., 1980; Murcray et al., 1975; 1979; Coffey
et al., 1981; Brasunas et al., 1988; Kunde et al., 1988).
Both absorption and emission experiments have been made using a
number of instruments. The Limb Radiance Inversion Radiometer (LRIR) and
Limb Infrared Monitor of the Stratosphere (LIMS) are infrared radiometers
that were flown on Nimbus‑6 and -7, respectively, and recorded data in 1978
and 1979 (Gille et al., 1980; Gille & Russell, 1984). The six channels of LIMS
observed emissions by CO2, HNO3, O3, H2O and NO2 from 15−65 km.

86
 GOME and SCIAMACHY

Starting with the Selective Chopper Radiometer (SCR) on Nimbus‑4 and -5,
the Department of Physics at the University of Oxford developed a series of
instruments to observe infrared emission from the atmosphere. A pressure-
modulated instrument was flown on Nimbus‑6 and the Stratospheric and
Mesospheric Sounder (SAMS) on Nimbus‑7 (Table 1). SAMS measured in limb-
viewing geometry the gases CO2, H2O, CO, N2O, CH4 and NO in the stratosphere
and mesosphere. An improved version SAMS was flown on the UARS and
added channels for NO2, N2O5, HNO3, O3, and H2O (e.g. Barnett et al., 1992)
(Table 1).
Fourier transform spectrometers had already been used to sound the
stratosphere and upper troposphere in the 1980s. One of the most important
successes was the Atmospheric Trace Molecule Spectroscopy (ATMOS)
project (Farmer et al., 1987; Gunson et al., 1996). The ATMOS instrument was
flown on Spacelab 3 and the Atmospheric Laboratory for Applications and
Science (ATLAS) on Space Shuttle missions (Table 1). ATMOS performed solar
occultation measurements and retrieved a variety of trace gases in the upper
troposphere and stratosphere.
From its launch in 1991 until 2005, the UARS circled Earth in a low-
Earth non-Sun-synchronous orbit and carried three infrared experiments.
In addition to ISAMS (described above), the Cryogenic Limb Array Etalon
Spectrometer (CLAES) and the Halogen Occultation Experiment (HALOE) made
infrared measurements designed to yield information about stratospheric
and tropospheric trace gas constituents. Some of the measurements were
of shorter duration. The CLAES used a high-resolution etalon to measure the
limb emission in the infrared (Roche et al., 1982; Roche & Kumer, 1989). The
target gases and parameters were N2O, NO, NO2, HNO3, CF2Cl2, CFCl3, HCl, O3,
ClONO2, CO2, H2O, CH4 and temperature (UARS, 1987). HALOE used broadband
filter radiometry to measure CO2, H2O, O3 and NO2 and gas filter correlation
radiometry to measure HF, HCl, CH4 and NO (Baker et al., 1986) (Table 1).
All four stratospheric remote sensing missions (MLS, ISAMS, CLAES and
HALOE) on the UARS achieved their goals. The HALOE and MLS were still
measuring after eight years and produced a unique record of the stratosphere.
The UARS eventually ceased operation in 2005 after 14 successful years.
The Japanese National Space Development Agency (NASDA) launched its
Advanced Earth Observing Satellite (ADEOS) in 1996. The payload included an
Improved Limb Atmospheric Spectrometer (ILAS), an Interferometric Monitor
for Greenhouse Gases (IMG) and a Total Ozone Monitoring Spectrometer
(TOMS) for atmospheric sensing. IMG and ILAS are nadir and limb sounding
infrared instruments (Table 1). ADEOS I was launched in August 1996 but
failed in June 1997. ADEOS II with ILAS‑II was launched in December 2002 but
communication was lost in October 2003.
Prior to the development of the atmospheric sensors on Envisat and ERS‑2,
the only experiment that specifically used infrared information to probe the
troposphere was the Measurement of Air Pollution from Satellites (MAPS)
experiment. This was a nadir sounding gas correlation, which made global
measurements of CO in the middle and upper troposphere. It flew three times
between 1981 and 1994 on the NASA Space Shuttle (Reichle et al., 1986; 1990;
Connors et al., 1991).

5.3 Remote Sensing in the UV, Visible and Near-IR

The source of radiation for passive remote sounding of the atmosphere in
the ultraviolet, visible and near- and short-wave infrared regions is the Sun.
The Sun’s maximum emission is around 580  nm. Beyond 300  nm the Sun
corresponds fairly well to a black body with a temperature around 5800K.
Absorption by atoms in the Sun produce the Fraunhofer lines, having many
strong features in the ultraviolet and visible but are less significant at longer

87
SP-1326

wavelengths. Below 300 nm the solar emission deviates from simple blackbody
behaviour and the solar output is determined to a significant extent by
processes occurring at the edge of the Sun and correlate with solar activity.
In 1957 it was proposed that satellite measurements of backscattered
ultraviolet (BUV) radiation from the terrestrial atmosphere could be used to
deduce ozone profiles on a global basis (Singer Wentworth, 1957). The method
relies on two effects, i.e. the scattering of light at short wavelengths and the
absorption of ozone. Rayleigh scattering of light by air molecules has a strong
dependence on wavelength, whereby the intensity of scattered light is a
function of the inverse of the fourth power of the wavelength. Similarly, ozone
absorption is strongly wavelength-dependent. These effects combine and as a
result the penetration depth of light in the atmosphere varies strongly between
the ozone maximum absorption in the Hartley band around 250  nm and its
minimum beyond 380 nm in the Huggins bands.
Numerical techniques for the determination of vertical profile information
were also studied by NASA for the determination of total ozone in the
atmosphere (Dave & Mateer, 1967; Mateer et al., 1971). The development of these
retrieval techniques has continued up to the present (Bhartia et al., 1996). The
earliest measurement utilising the BUV technique was undertaken by Rawcliffe
& Eliot (1966) using a photometer observing at 284  nm. Ozone distributions
were determined utilising measurements from the USSR Cosmos satellites,
which flew a double monochromator in 1965 and 1966 (Krasnopol’skiy, 1966;
Iozenas et al., 1969a,b).
The Backscatter Ultraviolet atmospheric ozone experiment (BUV) was the
first in a series of instruments produced by NASA and later NOAA, which has
successfully made long-term measurements of the BUV for the vertical profile
and total amount of ozone (Heath et al., 1973) (Table 1). BUV was launched on
the Nimbus‑4 satellite into a circular polar orbit at an altitude of 1100 km. This
orbit is Sun-synchronous and the satellite crosses the equator in an ascending
mode every 107 min close to local noon.
This instrument concept was developed and resulted in the Solar
Backscatter Ultraviolet (SBUV) and Total Ozone Mapping Spectrometer (TOMS)
launched on Nimbus‑7 (Heath et al., 1975). The SBUV instrument was further
improved to the SBUV-2 and has been flown by NOAA on a series of satellites
(Table 1).
After Nimbus‑7 (1979−92), TOMS was also flown on the Russian Meteor
platform (1992−94), on the Japanese Advanced Earth Observing Satellite
(ADEOS, 1996−97) and on Earth Probe (1996−present) (Table 1). These
measurements have been used to derive a unique ozone dataset. Readers are
referred to the literature about the TOMS data record for more details.
Recognising the importance of long-term calibration and validation of
space-based instrumentation, NASA developed the Shuttle SBUV (SSBUV),
which flew eight times on the Shuttle to calibrate radiometrically the BUV
instruments (Hilsenrath et al., 1988; 1996). The attention paid to the detail of
the calibration of the NASA and NOAA BUV instruments has established the
quality of these datasets (Hilsenrath et al., 1995).
The TOMS retrievals of total column O3 have been combined with data
from the SBUV or the second Stratospheric Aerosol and Gas Experiment
(SAGE II) to provide information about tropospheric O3. TOMS had also been
used to demonstrate the emissions of SO2 during volcanic eruptions into the
troposphere and the stratosphere.
The first SAGE instrument (SAGE I), flown on NASA’s Atmosphere Explorer
mission from 1979 to 1981 (Table 1), was a spectrometer that measured the
absorption of sunlight by ozone with four channels centred at 0.385, 0.45,
0.6 and 1.0 μm (McCormick et al., 1979; Chu & McCormick, 1979). SAGE II is a
seven-channel instrument from the same team (Maudlin et al., 1985), which
was launched on NASA’s Earth Radiation Budget Satellite (ERBS) and is still
in operation today. A third-generation SAGE has been developed, which was

88
 GOME and SCIAMACHY

launched on the Russian Meteor-3M in 1999 and deployed on the International

Space Station in 2002. SAGE determines atmospheric absorption in occultation
and measures at sunrise and sunset. So far the SAGE series of instruments has
provided reliable data for a number of atmospheric constituents.
An interesting set of measurements was obtained by the Solar Mesosphere
Explorer (SME), which was flown by NASA in 1983. It contained limb-scanning
ultraviolet, visible and near-IR channels for measuring stratospheric and
mesospheric O3, O2(1Δ) and NO2 (Rusch et al., 1984; Thomas et al., 1984; Mount
et al., 1984).

5.4 Remote Sensing Techniques for Aerosols and Clouds

Molecular or Rayleigh scattering of electromagnetic radiation is strongly
dependent on wavelength (~λ−4), whereas scattering by aerosols and clouds
obeys approximately the Mie theory and is relatively weakly dependent on
wavelength (~λ−(1−2)). Aerosols have been identified by their scattering effects
both actively and passively. Prior to the launch of ERS‑2 in 1995, the first
passive remote sensing experiment to measure successfully the abundance of
atmospheric aerosols from space was the Stratospheric Aerosol Measurement
(SAM II) on Nimbus‑7 (McCormick et al., 1979). This experiment was a single-
channel radiometer observing in solar occultation and was the forerunner
of SAGE. Stratospheric aerosols have also been measured by their infrared
absorptions (e.g. HALOE).
Tropospheric aerosols and smoke have been retrieved from several types of
remote sensing data. For example, algorithms to retrieve tropospheric aerosols
have been developed for data from the Advanced Very-High-Resolution
Radiometer (AVHRR) data over the oceans (Ignatov et al., 1995). Smoke from
burning biomass was observed by AVHRR and also from TOMS data (Hsu et al.,
1996).
In the ultraviolet, visible and near-infrared spectral regions, radiation is
strongly scattered by clouds, enabling its presence to be detected, provided
that the spatial resolution is sufficiently high and the difference between
the effective cloud albedo and the spectral reflectance of Earth’s surface is
sufficiently large. Several algorithms have been developed that aim to utilise
the oxygen absorption for the determination of cloud top heights (Guzzi et
al., 1996; 1998 and references therein). Clouds emit long-wavelength infrared
radiation, the spectrum of which depends on their temperature. Observations
of the infrared emissions by clouds with instruments on the Meteosat platform
and related meteorological satellites are routinely used to estimate the cloud
top height and cover.
The visible channels flown on the Along-Track Scanning Radiometer
(ATSR-2) were planned and used to determine aerosol and clouds. The French
Polarisation and Directionality of the Earth’s Radiance (POLDER) instrument
on the Japanese platform ADEOS-1 measured polarisation parameters (Table 2),
which were used to study polarisation, aerosols and clouds in the atmosphere
(Deschamps et al., 1990).
After ERS‑2 and GOME, in Europe the national agencies continued the
development of SCIAMACHY for Envisat, and ESA developed GOME‑2 for
Eumetsat’s MetOp series of platforms. NASA offered to fly the Dutch Ozone
Monitoring Instrument (OMI) instrument on its Earth Observing System
Chemistry (EOS-CHEM) mission, which was later renamed Aura.

89
SP-1326

6. The Development of European Observation of

Atmospheric Composition
To understand the development of the European component of global
measurements of atmospheric composition and Earth observation for Earth
system science, it is necessary to follow the evolution of modern atmospheric
science and the birth of the space age. While observing Montgolfier balloons
over Paris in 1783, Benjamin Franklin had noted and foreseen the potential
value of remote sensing of the Earth from elevated platforms. This was prior
to the development of spectroscopy by Fraunhofer and colleagues. The
identification of an ultraviolet absorber in the upper atmosphere by Cornu
in 1879 and its attribution by Hartley in 1880 to ozone may have been the
first modern application of remote sensing of atmospheric composition. This
was followed by the in situ discovery aboard balloons of the temperature
inversion at the tropopause and the stratosphere. Other important milestones
in the development of atmospheric Earth observation were the development by
Dobson of a spectrophotometer for the remote sensing of atmospheric ozone in
the 1920s and the observation of aurora phenomena from the ground.
The pioneering development of rocketry in the first half of the 20th century
was accelerated by the recognition of their use for military applications after
the Second World War. However, the superpowers that emerged in the post-
war period also engaged in a technological race, in part to demonstrate
their ideological superiority in the civil exploitation of space. The birth of
the space age began with the launch of Sputnik in 1957 by the Soviet Union
and the creation of NASA in the USA. The first American satellite, Explorer 1,
was launched in 1958 and carried instrumentation for studying the upper
atmosphere and magnetosphere, and led to the discovery of the van Allen
belts. In 1960 NASA began the Television Infrared Observation Satellite (TIROS)
programme and a series of platforms to test experimental television techniques
designed to develop a worldwide meteorological satellite information system.
In the 196Os the Soviet Union began the first measurements of atmospheric
ozone. At the same time NASA began to develop the Nimbus and other
research platforms carrying instrumentation designed explicitly to measure
atmospheric composition with increasing accuracy. At this stage, Europe was
still recovering economically and politically from the Second World War, and
its space exploits were limited to national programmes and collaborations
with NASA. However, the need for an overarching European dimension to
space exploration was recognised and both the European Space Research
Organisation (ESRO) and the European Launcher Development Organisation
(ELDO) were created in the early 1960s. ESA was formed by a merger of these
organisations in 1975. ESA immediately began to develop Earth observation
programmes to meet the evolving need for atmospheric research and
operational meteorological purposes. This resulted in the first European
geostationary meteorological satellite, Meteosat, whose operation was
transferred to the newly created European Organisation for the Exploitation of
Meteorological Satellites (Eumetsat) in 1986.
In the 1960s and 1970s Europe’s leading scientists had no direct access to space
for Earth observation other than through participation in national programmes,
which were linked to NASA or sometimes the Soviet space programme. While
international collaboration in space remains a cornerstone of activity, the creation
of ESA and later Eumetsat, and the broadening of national space agency objectives,
changed dramatically the role of European Earth observation.
In its early period, ESA’s Earth observation programme focused on the
development of satellite instrumentation for numerical weather prediction
and surface mapping. It was not concerned with atmospheric trace constituent
composition. The evolution of ESA resulted in the development of the European
Remote Sensing programme in the 1980s, with two satellites, ERS‑1 and
ERS‑2. The ERS team at ESA correctly recognised the need for continuity and

90
 GOME and SCIAMACHY

improved measurements, and this has been the most consistent and important
recommendation of all ESA scientific user consultations. The ERS programme
addressed the need to achieve long-term global datasets and to benefit from
technological developments to improve the spatial resolution and temporal
sampling of measurements.
The payload of ERS‑1 comprised Earth observation instruments that
gathered information about Earth (land, water, ice and atmosphere) using a
variety of measurement principles:

—— RA (Radar Altimeter) – a single-frequency nadir-pointing radar altimeter

operating in the Ku-band;
—— ATSR-1 (Along-Track Scanning Radiometer) – a 4-channel infrared radiometer
and microwave sounder for measuring temperatures at the sea surface and
at the top of clouds;
—— SAR (Synthetic Aperture Radar) – a radar operating in the C-band to detect
changes in surface heights with sub-millimetre precision.
—— Wind Scatterometer – to calculate information on surface wind speeds and
direction; and
—— MWR (Microwave Radiometer) – to measure atmospheric water, as well as to
provide a correction for the atmospheric water for the altimeter.

The ERS-1 payload also included the Precision Range and Range-Rate
Equipment (PRARE) and a Laser Retroreflector. The PRARE was non-
operational since launch. The Retroreflector was used for calibrating the Radar
Altimeter to within 10 cm. In summary, ERS‑1 had only one instrument, the
MWR, for measuring atmospheric composition.
Since the 1970s, the impacts of human activities on the atmosphere were
becoming increasingly apparent. In 1984 a team led by myself from the Max
Planck Institute in Mainz developed a concept called the measurement of
air pollution from space, MAP, to fly on ESA’s European Retrievable Carrier
(EURECA). This was submitted to the ESA call for EURECA instruments in
early 1985. Although the concept was unfortunately not selected, 1985 was a
momentous year for atmospheric science. Farman, Gardner and Shanklin from
the British Antarctic Survey published their discovery of the loss of ozone in
the polar vortex in spring above the Halley Bay research station in Antarctica
(Farman et al., 1985). The term ‘ozone hole’ results from the first observations
from NASA’s TOMS and SBUV O3 data, which showed the large extent and
evolution of this phenomenon, which filled the whole of the polar vortex. The
discovery shocked the public and demonstrated unambiguously the impact
of pollution, particularly the release of CFCs and other ozone-depleting
substances in the northern hemisphere, on a region as far from the point of
release as is possible.
Looking back, it is difficult to imagine the impact of the discovery of the
ozone hole and its importance. The scientific community, funded by the
American and European governments, then embarked on an ambitious set
of experiments to understand the precise origin of this phenomenon. Indeed,
as suggested by Farman et al., (1985), the ozone hole was the result of the
release of CFCs, halons, methyl bromide and other substances, although the
mechanism (described above) is much more complex.
The discovery of the ozone hole focused attention on the need for global
measurements of trace gas species. In Germany, an investigative commission
was established to provide the parliament with advice. The commission was led
by Bernd Schmidbauer and included prominent scientists such as Paul Crutzen,
Reinhardt Zellner and others. The commission’s interim report, ‘Protecting
the Earth’s atmosphere: an international challenge’, had a significant impact
on German scientific policy aimed at controlling ozone-depleting substances
and the measures needed to monitor and assess the impact of international
agreements. It recommended scientific programmes including studies of O3.

91
SP-1326

The commission specifically discussed the need for global measurements

from space using SCIAMACHY or equivalent instrumentation to provide
the information needed to facilitate the development of national, European
and international policy with respect to ozone-depleting substances and
tropospheric pollution. This was of key importance for the development of
GOME and SCIAMACHY.

7. The SCIAMACHY Project and the Spin-off of GOME

During the 1980s, ESA was growing and its mission evolving. In 1985 ESA
appointed a new director of Earth observation, Philip Goldsmith. He came
from the UK Met Office and had studied in Oxford under the guidance of the
atmospheric scientists Brewer and Dobson and had worked closely with
Sir John T. Houghton. He was therefore aware of the need for atmospheric
composition measurements from space. He brought with him to ESA another
atmospheric scientist, Christopher Readings, also from the UK Met Office.
While administering many different aspects of ESA, these men brought
additional expertise in atmospheric research and its needs to ESA. They
complemented the efforts of the both the then ERS Mission Manager Guy
Duchossois, who worked tirelessly for the development of European Earth
Observation throughout his career, and the ERS team at ESTEC.
European satellite-based meteorology began on the 23 November 1977
when Meteosat, developed by ESA, was launched from Cape Canaveral into a
geostationary orbit. Meteosat was also ESA’s first Earth observation satellite,
launched just two years after ESA had been established and nine years
before Eumetsat was created. In the intervening years ESA had promoted the
establishment of Eumetsat to address the needs of the European operational
community for numerical weather prediction. The provision of an adequate
space infrastructure was a key issue and Eumetsat was created in 1986 under
the leadership of John Morgan, also from the UK Met Office, to manage the
operational satellites, developed by ESA, for the meteorological community.
In late 1981 I decided to move from the UK Atomic Energy Research
Establishment at Harwell, Oxfordshire, and my collaboration with Richard
Wayne at the then Physical Chemistry Laboratory of the University of Oxford,
to take an opportunity to join the Max Planck Institute (MPI) for Chemistry in
Mainz in the new department of air chemistry led by Paul Crutzen. Initially I
worked with Geert Moortgat and David Griffith on laboratory kinetics and
spectroscopy of free radicals and reactive species. In 1982 Dieter Perner
(1934–2012) also joined Paul’s department at the MPI to establish an optical
spectroscopy group for atmospheric measurements. After three years in the
laboratory I joined the optical spectroscopy group as one of its leaders.
As a laboratory spectroscopist and kineticist researching atmospheric
reactions at the time, it became clear to me that the technique of Differential
Optical Absorption Spectroscopy (DOAS), which had been developed for active
long path measurements by Dieter Perner and Ulrich Platt at the Jülich Research
Centre in the 1970s, should work from space. From 1984 onwards, supported by
Paul Crutzen, Dieter Perner and Wolfgang Schneider, who worked closely with
me for many years, I began to develop a concept for measuring trace gases from
space. Our first effort, called Mapping of Atmospheric Pollutants (MAP), was
proposed prior to the discovery of the ozone hole in 1985, to ESA for flight on
EURECA.
DOAS uses the electronic vibrational rotational structure of trace gases to
identify in a fingerprint manner their presence in the atmosphere. It separates
the spectral features of gases from broad-band absorption and scattering by
a mathematical high-pass filter technique using a low-order polynomial to
capture the broad-band absorption and scattering. The technique utilises the
Beer−Lambert law, but requires explicit consideration of multiple scattering

92
 GOME and SCIAMACHY

in the atmosphere for the determination of amounts and distribution of trace

gases. This and more sophisticated retrieval approaches were planned from
the outset for SCIAMACHY and GOME.
In the mid-1980s ESA’s Earth Observation Directorate organised meetings
with the European research and operational meteorological community to
discuss and assess the needs of the different Earth observation disciplines.
At one such meeting in 1986, Paul Simon from the Belgian Institute for Space
Aeronomy (BIRA) and I were tasked with describing the requirements for
atmospheric trace gas measurements from a co-orbiting platform. We proposed
observation using scanning imaging remote sensing in the solar spectral
region for the retrieval of trace gases using the DOAS approach. At this time,
ESA’s Columbus programme had developed the Free Flyer or Polar Platform,
also known as part of the Polar Orbiting Earth Observation Mission (POEM-1).
Building on our preparations for the Announcement of Opportunity (AO)
call for EURECA, I began organising a science team at the MPI in Mainz.
International experts were invited to join the team, supported by the European
global atmospheric chemistry modelling community. In due course ESA issued
an AO for ‘the provision of Earth observation instrumentation on the European
Polar Platform of the Space Station’ in January 1988.
In response to this new AO, I led the development of a science team
proposal, supported by my managers and colleagues at the MPI. I invited
scientists from Europe and America to join our initiative. I christened our
proposed instrument the Scanning Imaging Atmospheric spectroMeter for
Atmospheric CHartographY, or SCIAMACHY, which in Greek means chasing
or hunting light and shadows, and in part has the meaning of undertaking
an impossible task. As part of the preparation of the proposal we contacted
aerospace companies and selected Dornier in Friedrichshafen, Baden-
Württemberg, Germany, to be the Prime Contractor for SCIAMACHY proposal
preparation. The Dornier department of optical instrument development,
led by Elmer Weber and supported by the Dornier management, agreed to
develop an industrial assessment of the scientific instrument concept for the
SCIAMACHY proposal. Edgar Hölzle, Rolf Mager and colleagues at Dornier then
began their analysis, design and cost assessment for developing a SCIAMACHY
instrument based on our concept. Dornier later became part of DASA, which
subsequently became part of Astrium, which is now part of EADS. Although
the team at Friedrichshafen evolved over the years, their hard work has been
one of the reasons for the success of SCIAMACHY, ERS-2 and Envisat.
Particularly valuable contributions at the beginning of the development
of the objectives and scientific justification for the proposal were made by
Wolfgang Schneider, Dieter Perner, Paul Crutzen and Geert Moortgat from the
MPI. Both the experimentalists Ulrich Platt, then at the Institute for Chemistry
of the Nuclear Research Centre in Jülich, and Jean-Pierre Pommereau at
CNRS in Paris, who with his colleagues had developed the SAOZ balloon and
ground-based instruments, joined our SCIAMACHY science team. In addition,
Paul Simon and Christian Muller from BIRA brought their solar physics and
aerosol expertise to the team. As well as Paul Crutzen, other scientists from the
European Global Modelling of Atmospheric Chemistry (GLOMAC) project such
as Ivar Isaksen of the University of Oslo, Han van Dop of the Royal Netherlands
Meteorological Institute (KNMI), Erik Rockner of MPI Hamburg, Hilding
Sundquist of the University of Bergen and Henning Rhode of the Meteorological
Institute of Stockholm University, joined the SCIAMACHY project becoming
the data user group within the science team. John Geary, an expert in detector
technology for space applications, and Kelly V. Chance, who provided
sensitivity analyses from the Harvard Smithsonian Astrophysical Observatory,
completed the first SCIAMACHY team. In addition, Jack Fishman of NASA,
USA, John Harries of the Rutherford Appleton Laboratory, UK, and Richard
Wayne of the University of Oxford agreed to act as scientific consultants to the
SCIAMACHY principal investigator and science team.

93
SP-1326

The SCIAMACHY proposal, conforming to the ESA guidelines for the AO for
POEM, was organised and written in late spring and summer 1988, by myself
with contributions from the science and industrial teams. I would like to
thank Wolfgang Schneider, Dieter Perner and Paul Crutzen for their help in the
preparation of the final proposal. In addition to excellent scientific objectives,
affordable and feasible instrument design, calibration and validation strategies,
which were of great importance for the potential success of the project, the
support of the German government was essential. Fortunately, the then German
Federal Ministry of Research and Technology (BMFT) in the department run
by Mr Spät and Mr Käsmeier, agreed that their department would support the
proposal, provided that SCIAMACHY was selected for Phase A studies for flight
on the Polar Platform by ESA’s technical and scientific review.
The SCIAMACHY instrument concept at this time comprised two
instruments both able to undertake nadir limb and solar or lunar occultation
measurements: one focused on limb observations and the other on nadir
observations during the Sun-synchronous orbit. Measurements were to be
made during solar and lunar occultation in both hemispheres. This approach
was and is a very attractive concept but was later descoped to one instrument,
making alternate limb and nadir measurements, as well as solar and lunar
occultation when available, in order to reduce costs.
During the time between proposing the MAP instrument for EURECA and
the winter of 1988, there were many developments. The publication of the first
observation of what came to be known as the ozone hole by Farman, Gardner
and Franklin from the UK Natural Environment Research Council; the British
Antarctic Survey had led to many campaigns to assess the origin of this unique
phenomenon. NASA traced the development of the ozone hole using maps
retrieved from both TOMS and SBUV instruments, and detected a large loss in
the vertical profile of O3 using the SAGE-II instrument.
To address the needs of users, ESA hosted a meeting with the EO user
community in Paris in November 1988, organised by Chris Readings, Guy
Duchossois, and others. One objective was to develop a set of priorities.
John Pyle of the University of Cambridge and I on behalf of the atmospheric
scientists presented our findings. One important conclusion was that a
European instrument observing atmospheric trace gas composition was
needed much more quickly than the schedule for the launch of the Polar
Platform would allow.
At the end of the meeting Jan Berger, then director of ESTEC, asked me
what sort of instrument the atmospheric community needed to address
atmospheric chemistry and composition. I explained that sounding
spectrometers could make a valuable contribution. Subsequently the ESA
Earth Observation Directorate invited proposals for ideas for an instrument to
measure atmospheric composition for ERS‑2, which was already being built.
The ERS team, led by Reinhold Zöbl, Peter Edwards and Peter Dubock, had
identified that ERS‑1, which had been built, had some valuable contingency.
They proposed that in ERS‑2, this contingency could be used to accommodate
an additional small instrument. As a result Philip Goldsmith, Director of Earth
Observation and Microgravity at ESA, supported by Guy Duschossois and Chris
Readings, issued a call for instrumentation for measurements of atmospheric
constituents for ERS‑2 in late November 1988.
In response to this call, I developed the SCIA-mini proposal, supported
by Paul Crutzen and selected members of the SCIAMACHY international
science team, and submitted it to ESA in December 1988. Part of the feasibility
assessment of SCIA-mini was again supported by the Dornier industrial team.
The instrument concept comprised two spectrometers, both capable of limb
nadir and occultation measurements. Like SCIAMACHY, one was focused on
nadir and one for limb. However, as ‘off-the-shelf’ technology was required
it was later restricted to using available detectors and was limited to the UV,
visible and near-IR spectral regions.

94
 GOME and SCIAMACHY

In February 1989, ESA concluded its review and selection procedures for
the payload instrumentation for the Polar Platform to enter Phase A studies at
a meeting in Paris. The meeting was fraught, as all the proposed instruments
were of unique value. In addition, the task of selecting in a transparent manner
an adequate compliment of instruments for the atmospheric composition
remote sensing payload was challenging. Many exciting concepts were
presented, including the Advanced Microwave Atmospheric Sounder (AMAS),
based on the successful Microwave Atmospheric Sounder, which flew on the
Space Shuttle, proposed by a team led by Klaus Künzi of the University of
Bremen. Another was the Doppler Wind Sounder (DWS), proposed by David
Rees and his team from University College London. The Dynamic Limb Sounder
(DLS) proposed by a team led by John J. Barnett (1948–2010) of the University of
Oxford, and based on pressure modulated experience with the Stratospheric
and Mesospheric Sounder (SAMS) on Nimbus‑7 and the Improved Stratospheric
and Mesospheric Sounder (ISAMS) on UARS. This was subsequently merged
with an American instrument to become the High-Resolution Dynamic Limb
Sounder (HRDLS) that flew as part of NASA’s Aura mission. The Michelson
Interferometer for Passive Sounding (MIPAS) was proposed by Karlruhe
University and the Medium-Resolution Imaging Spectrometer (MERIS) were
also presented. Finally, the Global Ozone Monitoring by Occultation of Stars
(GOMOS) proposal was presented by Jena-Louis Berteaux.
I presented the SCIAMACHY concept, supported by Paul Crutzen and
Wolfgang Schneider at this review. Jean-Pierre Pommereau of CNRS, an expert
in atmospheric remote sensing and a member of the science teams of both
GOMOS and SCIAMACHY was tasked to summarise and explain to the ESA
panel, the relative advantages and disadvantages of GOMOS and SCIAMACHY.
After its technical and scientific review, ESA announced that it planned to
initiate Phase A studies of atmospheric instrumentation for the Polar Platform.
Of the instruments in the competition, MIPAS and MERIS were selected to
become ESA-developed instruments, and GOMOS and SCIAMACHY were
selected as nationally funded AO instruments. The ESA Programme Board
for Earth Observation (PB‑EO) and subsequently the ESA Council agreed to
this proposal. The German BMFT, in the department led by Mr Spät and Mr
Käsmeier, then funded an industrial study of SCIAMACHY led by the Dornier
team. The funding was managed by the DLR Aerospace team, then at Köln
Portz, which later became Deutsche Agentur für Raumfahrtangelegenheiten
(DARA), and eventually the German Aerospace Centre (DLR) in 1997. This
enabled work to commence on the design and Phase A for SCIAMACHY.
In April 1989 the German BMFT initiated the Atmospheric Trace Molecule
Spectroscopy (ATMOS) project, coordinated by Wolfgang Mett. ATMOS was
a concept for a national satellite for atmospheric and ocean science, with an
instrument payload comprising the Advanced Millimeter Wave Atmospheric
Sounder (AMAS), the MIPAS, the Reflective Optics Imaging Spectrometer
(ROSIS) and SCIAMACHY. This provided a much-needed boost to the
development of the instrument and mission concepts for both MIPAS and
SCIAMACHY. Ultimately, after an ineffective collaboration with CNES, who
developed the Globsat concept, ATMOS and Globsat missions sadly were
not pursued. However, the ATMOS project then provided funding for the
development of SCIAMACHY for the Polar Platform.
After SCIAMACHY was selected for Phase A studies for the Polar Platform,
and for ATMOS, Albert Goede of the Netherlands Institute for Space Research
(SRON) and Roel Hoekstra (1932–5) of Netherlands Organisation for Applied
Scientific Research, Institute of Applied Physics (TNO-TPD) visited me in my
role as PI to discuss the potential Dutch interest and proposed that the Dutch
participate in the industrial consortium to build SCIAMACHY. After discussion
with Dornier, work packages were agreed for SRON and TNO-TPD who then
joined the industrial consortium with Dornier as the Prime Contractor. TNO-
TPD was tasked with optical design issues and the SRON team with the

95
SP-1326

design of the focal plane assemblies. R. von Konijnenberg of the Netherlands

Institute for Aerospace Programmes (NIVR), the Dutch delegate to ESA’s
PB-EO, actively supported this activity. Thus the Netherlands joined the
consortium supporting the SCIAMACHY design and industrial procurement.
Huib Visser and Kees Smorenburg of TNO-TPD began working with me and the
SCIAMACHY industrial team led by Edgar Hölze on key aspects of the optical
design. During Phase A the German company OHB System in Bremen also
joined the consortium, with responsibility for developing the electronic brain
of SCIAMACHY. Later, in Phase B, the Belgian Ministry joined the funders of
SCIAMACHY to help support some of the Dutch contributions. As a result the
polarisation monitoring device was provided by a Belgian company, and the
Dutch brought in Fokker Space, now called Dutch Space.
ESA introduced a fast-track review process for its plan to add an
atmospheric composition remote sounding instrument to ERS‑2. Although ESA
selected SCIA-mini, in the summer of 1989 the EO Programme Board and the
ESA Council believed there was a high risk that the instrument would not be
developed on time for flight on ERS‑2 in 1994. They therefore recommended
that a short Phase A study of a descoped version of SCIA-mini be produced. The
study was led from the ESA side by Peter Dubock and Chris Readings.
Initially, I called this instrument SCIAS, a Scanning Imaging Absorption
Spectrometer for ERS‑2 and, together with Wolfgang Schneider and Paul
Crutzen, submitted a proposal to ESA for a study of SCIAS. ESA’s condition was
that we were to design an adequate instrument that would fit the resources
available on ERS‑2 identified by the ESTEC team: volume 1 m3, mass 40 kg, a
data rate of 40 kbit/s and 40 W power. In order to achieve these goals it was
necessary to descope SCIA-mini and, after consultations within MPI, I proposed
that the nadir mode had the advantage that it could make both tropospheric
and stratospheric measurements. In the autumn of 1989, I proposed that TNO-
TPD join the study, and Huib Visser and Kees Smorenburg became part of the
Phase A industrial study team for the optics for GOME. The intention of the
ESTEC team was to obtain a design that, for reasons of geographical return,
could be developed in Phase B in Italy.
The study team included myself as lead scientist, W. Schneider from MPI
Mainz, who defined the performance and instrument requirements, P. Dubock,
A. Lefebvre, R. Zöbl, P. Edwards, B. Pieper, E. Leimann and B. Duesmann from
the ESTEC ERS team, and H. Visser and C. Smorenburg from TNO-TPD, who
developed the optical design based on the smaller, reduced-complexity version
of SCIAMACHY. Valuable inputs were provided by J. Geary from the SCIAMACHY
team on detector selection and G. Mount, then at NOAA, on the readout
electronics used for example in the NOAA instruments (Mount et al., 1987).
Nevertheless, the correlated sampling approach used in SCIAMACHY to reduce
the readout noise was not implemented. This team quickly developed a viable
concept using much of the development work undertaken for SCIAMACHY in
previous years. The instrument was then renamed the Global Ozone Monitoring
Experiment, GOME, by Chris Readings. Peter Dubock became the GOME
Instrument Manager for the ERS team at ESTEC, and I became the lead scientist
of GOME, complementing my role as the PI of SCIAMACHY.
The first meeting of the GOME Science Steering team took place at ESTEC on
15 February 1990, chaired by Chris Readings and P. Dubock. The participants
included members of the SCIAMACHY Science Team and some ESA nominees.
The initial team comprised myself as GOME Lead Scientist and SCIAMACHY
PI (MPI Mainz), Ulrich Platt (University of Heidelberg), J.-P. Pommereau (CNRS,
France), C. Muller (BIRA, Belgium) and A. Goede (SRON, the Netherlands) from
the SCIAMACHY science team, B.J. Kerridge (Rutherford Appleton Laboratory,
UK) and R. Guzzi (IMGA/CNR, Italy).
The second meeting of the now renamed GOME Science Advisory Group
(GSAG), took place under my leadership at the MPI in Mainz, with W. Schneider,
D. Perner and P. Crutzen (MPI) G. Fiocco/A. di Sarra (University of Rome), H. van

96
 GOME and SCIAMACHY

der Woerd (Netherlands Institute for Public Health and the Environment,
RIVM) and K.V. Chance (Smithsonian Astrophysical Observatory). Rudolfo
Guzzi and Adelbert Goede also joined the group in this first phase. Chris
Readings chaired the subsequent GSAG meetings until his retirement in 2001. I
acted as scientific secretary or organised this position for the next eight years.
Some 23 years later, GSAG still exists, with several of the original members.
GSAG now supports GOME‑2 on the MetOp series of platforms and is managed
by Rosemary Munro of Eumetsat and Tobias Wehr of ESA.
The SCIAMACHY project comprised the development of SCIAMACHY for
flight on the Polar Platform and of the SCIA-mini, later SCIAS−GOME proposals.
The research and technological development in this project provided the basis
for the GOME Interim Science report. The heritage of SCIAMACHY and GOME
included:

—— the European development of Differential Optical Absorption Spectroscopy

(DOAS) in the 1970s by Ulrich Platt and Dieter Perner;
—— the remote sensing research of John Noxon at NOAA and the Solar Mesosphere
Explorer (SME) team at the Laboratory for Atmospheric and Space Physics
(LASP) in Boulder, USA;
—— the development of ground-based and balloon measurements by Jean-Pierre
Pommereau and Florence Goutail of the Système d’Analyse par Observation
Zenithale (SAOZ); and
—— NASA’s BUV, SBUV and TOMS instruments that began measurements in 1974.

Early in the SCIAMACHY project, I recommended that D. Heath, who with

A. Kruger led the development of BUV, SBUV and TOMS at NASA, become a
consultant on the design and development of radiometric calibration of GOME
and SCIAMACHY. This led to a fruitful collaboration with the NASA team at
the Goddard Space Flight Center in Maryland. In particular, the contributions
of Earnest Hilsenrath, James Gleason and Scott Janz to the GOME and
SCIAMACHY Science Advisory Groups (GSAG/SSAG) were invaluable for the
two projects.
The SCIAMACHY science team worked largely on its own resources to
support the scientific development of both SCIAMACHY and GOME. ESA did
manage to use their general studies funding mechanism to support some
focused studies on key aspects of the science and the algorithm developments.
SCIAMACHY is the more substantive instrument, having nadir, limb
occultation and calibration viewing modes, whereas GOME, being a smaller
version of SCIAMACHY, could be much more readily realised.
Utilising the inputs from the preparation for SCIAMACHY, and the SCIA-
mini industrial proposal, the industrial Phase A study for GOME took place
in early 1990 by the ESA ERS team, led by P. Dubock and R. Zöbl, who were
responsible for the whole of ERS at the at the time, and myself as lead scientist.
The GSAG, under the leadership of Chris Readings and myself, wrote the
GOME Interim Science Report (Burrows et al., 1993), which, together with
the successful design studies developed by the above team, was used as the
basis for ESA’s proposal to the PB-EO and the ESA Council to develop GOME
for flight on ERS‑2, which at that time was planned to be launched in 1994.
This proposal was successful and in June 1990 the GOME ESA project began in
earnest. However ESA's PB-EO had demanded, the development of GOME was
not to interfere with the production of ERS‑2, and no delay in the development
of ERS‑2 to resolve GOME development issues would be allowed.
The GOME Phase B industrial study took place between July 1990 and
March 1991. The contractors for GOME Phase B comprised Officine Galileo,
in Florence, who were responsible for the GOME optical subsystem; Labem
Italia in Milan, who were responsible for the electronics; and TNO-TPD in
the Netherlands, who supported the development of the optical design and
prepared the optical calibration unit.

97
SP-1326

This distribution of effort reflected the demands of geographical return

on resources by ESA member states. This is one of the realities of European
development of its space segment. It has its merits, but obvious limitations.
In spite of the excellence of the subsequent work undertaken in Italy, the loss
of some key industrial partners of the SCIAMACHY industrial consortium,
including the Prime Contractor Dornier was, at least in my view suboptimal
with respect to the overall development. From that point onwards the
development of SCIAMACHY for the Polar Platform, later renamed Envisat,
and that of GOME began to diverge. The SCIAMACHY science team, which
participated within the GSAG, was able to remain fairly united with respect to
comparability of performance. The GOME baseline design review took place in
Noordwijk, the Netherlands on 12 and 13 March 1991, after which ESA led the
organisation of Phases C and D.
During this period the GSAG created subgroups that wrote specifications
and requirements for the characterisation and calibration, data and algorithm
and validation of GOME. These were endorsed and reviewed together with the
GOME instrument specification document, and provided ESA with a clear set of
requirements for the GOME development.

8. GOME Industrial Phases C and D

At the end of the GOME Phase B, Peter Dubock took on new responsibilities
within the Envisat project, later managing the Atmospheric Dynamics Mission
(ADM), and eventually becoming Inspector General of ESA before retiring.
Jörg Callies, who had previously worked on DOAS, joined the ERS team to
support GOME. Achim Hahne, who had been ESA’s Phase A study manager
for SCIAMACHY, became ESA’s GOME project manager, with Alain Lefebvre
becoming responsible for the development of the electronics. This team, who
reported to P. Edwards and R. Zöbl at ESA, managed the industrial development
of GOME at Officine Galileo in Florence, led by A. Mariani, R. Veratti, M. Fibbi
and M.S. Mazzoni, and supported by many others.
Under the guidance of the Officine Galileo team as the Prime Contractor, the
development of GOME proceeded well, meeting all the ERS‑2 project deadlines.
Officine Galileo were flexible and facilitated changes in their schedule needed
to obtain the required instrument performance. As GOME was new, the
engineering approach was to create a breadboard/engineering model and then
the flight models.
The Critical Design Review took place in May 1993. At this meeting the issue
of straylight in channel 1 was addressed and this resulted in the channel  1
and 2 boundary being adjusted. This had several unforeseen consequences,
including the loss of spectral overlap between channels 2 and 3. Later it
was discovered the spectral resolution, as measured by the full width half
maximum (FWHM) of the instrument spectral slit function, changed rapidly
between 314 nm and 425 nm in channel 2. Both a breadboard model and two
flight models of GOME were made. Overall the approach worked well with
the characterisation of the breadboard, providing essential knowledge for the
Flight Model (FM) development. The breadboard model was also used in the
validation campaigns, making zenith sky measurements.
The Officine Galileo and the entire industrial team are to be commended
for their efforts to produce and deliver the GOME FM on such a tight schedule.
As the GOME Lead Scientist, I continued to support ESTEC and the Officine
Galileo industrial team developing the GOME instrument. The key issue in
such collaboration is to maintain the performance of the instrument and the
resultant quality of the Level 1 and 2 data products, but to do so by staying
within the financial envelope foreseen for the project. This period was an
exhilarating, if challenging, phase of the development of GOME. In particular,
the analyses of the characterisation and calibration measurements with GOME,

98
 GOME and SCIAMACHY

which demonstrated the potential capability of the instrument inflight, were

enthralling.
Many of the GOME preflight characterisation and calibration activities,
which were designed to meet the GSAG Characterisation and Calibration
requirements document, were contracted to the department led by R. Hoekstra
at TNO-TPD. Carina Olij (later van Eijk-Olij), Erik Zoutmann from TPD-TNO and
colleagues organised this project. This was also supported by Officine Galileo
as the GOME prime. The TPD-TNO team later took on similar responsibilities
for the SCIAMACHY spectral and radiometric calibrations. The accuracy
of pre-flight and in-flight calibrations is of key importance for the Level 0 to
1 processing as they provides many parameters required for processing to
generate Level 1 and Level 2 data products.
Two GOME FMs were developed and the Breadboard Model was later
upgraded to a Flight Model. The GOME FM, selected for flight, was calibrated
and delivered on time for integration to the ERS‑2 platform in 1994. The
ESA and industrial GOME team are to be congratulated on their efforts. It is
a pity that there were no opportunities to fly the other FMs. The GOME FM
worked excellently for 16 years and was switched off when ESA decided to
decommission ERS‑2 in July 2011.

9. The Development of Retrieval Algorithms and the

GOME Ground Segment
When the ESA PBEO and the ESA Council agreed on the project GOME for
ERS‑2, funding was made available to support the industrial development
of the hardware. The LS for GOME supported by GSAG, and the GOME Team
at ESA, agreed that the development of Level 0, 1 and 2 data products from
GOME, provided to the scientific community, was the idealised strategy to
maximise data usage. The German BMFT agreed to develop a processor for
GOME data product generation within the DLR. The development of the Level
0 to Level  1 processor, which determines the extraterrestrial solar irradiance
and the Earth shine radiance upwelling at the top of the atmosphere as a
function of wavelength, required careful collaboration between the calibration
scientists and experts within the GSAG and DLR in particular with respect to
polarisation. Wolfgang Balzer of the DLR led this project.
The development of suitable algorithms for Level 1 to 2 processing required
the definition of the data products and the development of scientific prototype
by the science team. These could then be operationalized within the ERS‑2
GOME ground segment. Initially only a limited number of data products were
agreed by ESA but these grew over the years as the science team developed and
demonstrated its competency.
After the successes with GOME and SCIAMACHY, a larger remote sensing
team within the optical spectroscopy department of the MPI was required.
This group had the task of supporting the development of the GOME and
SCIAMACHY instruments and data product retrieval algorithms. After an
excellent collaboration over a decade, Wolfgang Schneider left the MPI to take
up a position at the DLR in 1991. Wolfgang has remained a strong supporter
of SCIAMACHY, GOME and atmospheric remote sensing and now advises the
German minister of economics on space issues.
In 1989 Vladimir V. Rozanov from the atmospheric physics research group
led by Yuri Timofeyev at Leningrad/St Petersburg State University, joined the
SCIAMACHY/GOME team at the MPI in Mainz. Using his expertise on radiative
transfer modelling, Rozanov addressed a key issue for the development and
processing of the Level 1 to 2 retrieval algorithms, i.e. the development of a
fast radiative transfer forward model. This we called GOMETRAN (Rozanov et
al., 1997) and later was later further developed to become SCIATRAN, which is

99
SP-1326

available from the University of Bremen. These models are suitable for use with
either optimal estimation or DOAS techniques (Platt & Perner, 1980).
One unique feature of the radiative transfer model GOMETRAN/SCIATRAN
was the linearisation procedure for calculating weighting functions developed
by Rozanov to meet the need for a rapid convergence during fitting iterations.
The development of SCIATRAN continued over the next 23 years. It now
comprises a family of programmes, which simulate all relevant absorption
scattering and emission phenomena in the atmosphere and at Earth’s surface
including Raman scattering, fluorescence from land plants , oceanic biomass,
and atmospheric emissions, for all relevant viewing geometries, e.g. nadir,
limb, solar or lunar occultation and ground. SCIATRAN calculates both as a
scalar or a vector model, where the polarisation is explicitly considered. The
SCIATRAN forward radiative transfer model and inversion programs have
become workhorses for atmospheric science. There is now a significant and
large group of SCIATRAN users.
Dorothy Diebel joined the MPI team in 1991 and provided valuable support
to the development of algorithms and management of the growing GOME and
SCIAMACHY activities, prior to her leaving to join Eumetsat, where she worked
initially on GOME‑2 issues in 1995. Robert Spur also joined my research group
in 1991, leaving in 1994 to join the DLR. He provided scientific secretary support
for GSAG and programming support for the development of GOMETRAN.
In 1991 I was elected Professor of Physics at the University of Bremen. As a
result part of the optical spectroscopy group at the MPI moved with me in early
1992 (D. Diebel, A, Ladstatter Weißenmayer, M. Weißenmayer, V.V. Rozanov
and R. Spur). Thereafter I continued a close collaboration with Paul Crutzen,
Dieter Perner and Geert Moortgat at the MPI in Mainz. At the University of
Bremen my research group and that of Klaus Künzi were initially part of the
Institute of Remote Sensing. In 1993 Klaus Künzi, Wolfgang Roether and I
formed the Institute of Environmental Physics (IUP), which included the
Institute of Remote Sensing as the section dedicated to long-term projects
using remote sensing using instrumentation from the ground, ship, aircraft or
satellite platforms.
In the summer of 1992 an international atmospheric conference was held
in Berlin, organised by Harold Schiff (1923–2003) from York University in
Canada and Ulrich Platt, who had moved to the University of Heidelberg in
1989. This provided an opportunity to showcase GOME and SCIAMACHY
scientific objectives and developments. I organised a session on atmospheric
remote sensing, which focused on DOAS applications and the potential of
GOME and SCIAMACHY. Ulrich Platt has been an excellent colleague over
the past 25 years, participating with his research team initially at the Jülich
Research Centre and from 1989 onwards at the University of Heidelberg, in
the development of the GOME and SCIAMACHY retrieval algorithms, the
validation of data products and data exploitation. Our research groups have
worked closely together on the development of ground-based, aircraft-borne
instrumentation and some balloon experiments. Over the past two decades,
the University of Bremen team has benefited from the scientific collaborations
and support of the Institute of Environmental Physics at the University of
Heidelberg and that of the Max Planck Institute for Chemistry.
From 1992 onwards, the University of Bremen became the Principal
Investigator or Lead Scientist Institution for both GOME and SCIAMACHY. A large
effort was invested by the University in this evolving team to provide scientific
support for these exciting, at the time, novel and challenging projects. In my
years in Bremen a number of young scientists, who would make outstanding
contributions to the development of GOME, SCIAMACHY and the remote sensing
of atmospheric constituents from space, ground or aircraft, began their research
careers. These included Andreas Richter (DOAS ground-based and satellite
measurements of trace gases), Michael Buchwitz (detector development, radiative
transfer model development and trace gas retrieval), Thomas Kuroso (cloud and

100
 GOME and SCIAMACHY

aerosol radiative transfer modelling), Bernd Deters and Stefan Himmelmann

(kinetics and spectroscopy of trace atmospheric constituents), Ricarda Hoogen
(retrieval of vertical profiles of O3), and Michael Eisinger (retrieval of trace gases
from ground-based and satellite observations; he later moved to ESTEC to work
on GOME and GOME‑2, and more recently EarthCare). In 1995 Mark Weber,
Johannes Orphal, Heinrich Bovensmann and Stefan Noel joined the IUP/UB
team to undertake research dedicated to the retrieval of O3 total column and
vertical profiles, laboratory spectroscopy and databases, sensor development
and calibration and characterisation, funded through the German Aerospace
Agency (DARA, now DLR). These activities addressed in part the development of
reterival algorithms for data products from GOME and their validation.
Another important aspect of the GOME calibration was the preparation of
the Calibration Apparatus for Trace Gas Absorption Spectroscopy (CATGAS).
Andreas Türk and then Angelika Dehn, now at ESRIN, worked for their
diploma dissertations on the development of the CATGAS, supported by the
experimentalists mentioned above. Although ESA originally wanted to restrict
this period of calibration, the team from the University of Bremen worked double
shift system over the two weeks given to them to measure at five temperatures
the O3 and NO2 spectra using GOME and spectra of SO2, and, very usefully as
it turned out, SO2 at room temperature. These absorption spectra and derived
absorption cross sections were then repeated and calibrated at the IUP/
UB and a paper was published. Twenty years later, that paper was one of two
milestone publications in the category ‘remote sensing’, selected by the Journal
of Quantitative Spectroscopy and Radiative Transfer to celebrate its 50th
anniversary in 2010. As a result of these initiatives, the GOME and SCIAMACHY
programmes also helped to improve our spectral database. Inadequate
knowledge of spectral parameters about the trace constituents in the atmosphere
or Earth’s surface are fundamental limitations for remote sensing. Further
spectroscopic improvements are needed to exploit the full potential of GOME.
In 1992 the SCIAMACHY Science Advisory Group was established and held
its first meeting at the University of Bremen. The SSAG and GSAG had in large
part complementary memberships. For the first years the Dutch co-PI was
Han van Dop of the University of Utrecht. Christian Muller (BIRA) became the
Belgian co-PI and has continued in this role up to the present. Han van Dop
was succeeded by Adellbert Goede, who led the SRON contributions to GOME
and SCIAMACHY prior to his moving to the Royal Netherlands Meteorological
Institute in 2001. He continued until 2007 when he was succeeded by Ilsa
Aben. After 20 years the SSAG is also continuing to support SCIAMACHY
and Envisat. For many years the GOME and SCIAMACY scientists organised
workshops for the development of data products, a tradition that continues
with the SCIAMACHY data and algorithm subgroup.
Between 1990 and 1994 ESA funded two general studies to support the
scientific development of aspects for Level 1 to 2 processing for GOME. One
was won in a competitive tendering procedure by a team the University of
Bremen and comprising the University of Heidelberg and Rutherford Appleton
Laboratory with support from the Smithsonian Astrophysical Observatory.
A study of aerosols was also undertaken by a team at the University of Lille
in France. Later, a study of aerosol and cloud retrieval, led by R. Guzzi at the
Italian National Research Council in Modena, aimed at developing the total
ozone column and vertical profile retrieval algorithms. All of these studies
provided key inputs for the development of the operational processor.
The operational processor was completed on time by the DLR and was ready
for the launch of GOME. At the same time, the GOME/SCIAMACHY scientists
were primed to investigate the full potential of the GOME data information
content, using their algorithms running under scientific control, and deliver
trace gas, aerosol and cloud data products, which had not been foreseen in the
first operational processor. These efforts to develop the retrieval algorithms
were key to the success of a mission.

101
SP-1326

10. The GOME Instrument on ERS-2

Figure 1 showed the ERS‑2 satellite, the GOME‑2 instrument during testing
at Galileo Officine and a schematic of the optical layout. The success of the
engineers, scientists and administrators in creating both ERS‑1 and ERS‑2
is a great credit to this team. After the launch of ERS‑2, the ESTEC ERS team
managed the operation of the satellite during its commissioning phase, while
ESOC was responsible for the operation of the instrument in orbit. After the
commissioning phase the ESRIN and ESOC team managed the ERS‑2 and
GOME project during its Phase E.
The primary molecules and parameters targeted for retrieval from the
nadir viewing observations of GOME and SCIAMACHY are shown in Fig. 6. The
spectral resolutions and coverage of the first four spectral channels of GOME
and SCIAMACHY are similar but not identical. For further details about the
GOME instrument, see the GSAG’s GOME Interim Science Report (Burrows
et al., 1993, 1999, and references therein). For comparison, the SCIAMACHY
instrument and targets are described by Burrows et al., (1990, 1995) and
Bovensmann et al., (1999).
Figure 7 shows details of the GOME instrument as originally configured
in Phase B, including the optical path of electromagnetic radiation collected.
GOME has only one moving part, the scan mirror. The instrument comprises
a scan mirror telescope and spectrometer, and equally important thermal
and electronic subsystems (not shown). A key issue for instruments aiming
to retrieve trace gas atmospheric absorptions and emissions is the thermal
stability of the instrument and the reduction of the electrical and digital noise
such that the signal to noise is dominated by the photon shot noise. For this
reason the SCIAMACHY instrument is cooled and its temperature actively
stabilised. For GOME the detectors are maintained at −37°C to reduce the dark
read signal to a minimum, and around the orbit the temperature change is of
the order of a few degrees. The impact of this can be readily observed in the
small shift of 0.008 nm around the orbit, which fortunately can be adequately
taken into account in Level 0 to 1 and Level 1 to 2 processing.
The solar electromagnetic radiation upwelling from the atmosphere is
collected by the GOME scan mirror and is directed via the telescope onto the
entrance slit of the spectrometer. The beam is then collimated and directed
onto a predispersing prism forming an intermediate spectrum. The two UV
channels are selected by a partially reflecting prism. The visible and near-

Figure 6. Left: Target molecules and parameters and the spectral coverage of the instruments GOME on ERS‑2 and SCIAMACHY on Envisat.
Right: Differential trace gas absorption cross sections of a selection of target trace gases. (TNO and University of Bremen)

102
 GOME and SCIAMACHY

Figure 7. Schematic diagrams of the GOME instrument (left) showing the four-channel spectral channel measuring simultaneously in nadir,
and the ERS‑2 in orbit with the GOME scan pattern in forward scan (right). (ESA)

IR electromagnetic radiation is then directed onto a dichroic mirror that

separates the radiation into two channels. Each spectral channel comprises
a transmission grating optics producing the spectrum focused at the
linear one-dimensional detector array with 1024 detectors. In this manner
the region between 230  nm and 793  nm is measured simultaneously. The
spectral resolution, defined as the Full Width at Half Maximum (FWHM), is
approximately 0.2 nm below 400 nm, and 0.4 nm above. The channel 1 and 2
boundary was changed after the review of the GOME breadboard model
performance and the new boundary was selected to be 315  nm. The original
pictures from Phase A in Fig. 7 do not show this change.
About 4% of the incoming radiation, which is polarised parallel to the
entrance slit, is reflected from the predispersing prism onto a small array of
detectors, corresponding to channels 2, 3 and 4, which are the polarisation
monitoring detectors. Combination of these measurements with those of the
spectrally resolved channels enables the polarisation of the upwelling radiation
to be determined and then accounted for using the radiometric calibration.
The data rate available to GOME limited its spatial resolution. A scan mirror
was used to direct the image of the entrance slit of the spectrometer onto the
ground. This resulted in an instantaneous field of view of the order of 40 km
(along track) × 7 km (across track). The Reticon diode array detectors in GOME
can be read out at different frequencies. This enabled channel 1 to be separated
into channel 1a and channel 1b beginning around 280  nm. The mirror was
scanned and the detectors read synchronously. This resulted in the channels
1b, 2, 3 and 4, having a spatial resolution of 40 × 320 km. Channel 1a was read
out less often to increase its signal to noise by on chip detector integration. The
scan pattern for the majority of the measurements after the commissioning
of ERS‑2 led to 320 × 40 km for channels 1b to 4, with channel 1a having a
200 × 960 km footprint. For three days in 30 the higher spatial resolution mode
was used by reducing the scan from 960 to 240 nm. This resulted in 80 × 40 km
for channels 1b to 4 and 240 × 200 km for channel 1a.
The extraterrestrial solar irradiance was observed once per day by turning
the scan mirror when the Sun came into the view of the calibration unit. For
about six months of the year lunar measurements are possible close to full
moon. These measurements, combined with those of the onboard emission
line lamp, the LEDs in each channel and the measurements of the dark signal
provided the basis for the inflight calibration of GOME. In addition vicarious
calibration was achieved using selected targets.

103
SP-1326

11. ERS‑2 and GOME in Orbit

ERS‑2 was launched at 22:30 h on 20 April 1995 from the European Spaceport
in French Guiana. My thanks go to ESA for inviting me to this event. Sadly,
the launch coincided with the Oklahoma City bombing and the world’s media
perhaps did not recognise the long-term importance of Europe beginning global
measurements of air pollution from space. The ESTEC ERS‑2 team were busy up
to the launch. For GOME as an optical instrument it was critical to maintain
the cleanliness of the instrument. The ERS‑2 was launched by the Ariane 4
V72 into orbit. This had the advantage of maximising the fuel on-board for the
mission. At the decommissioning of ERS‑2, which began in July 2011, there still
remained more than adequate fuel for many years of operation.
GOME was initially kept warm and allowed to outgas. A critical issue for all
optical instruments is the condensation of water and ice on the cold parts of the
instrument. Water is outgassed from the surface of the multilayer insulation
and the electrical boards, resulting in thin layers of ice that produce interfering
etalons and affect radiometric calibration. In spite of the efforts to outgas the
instrument indeed GOME detectors when cooled do grow a small layer of ice
from the atmosphere within ERS‑2 and GOME. Although this needed to be taken
into account for the absolute radiometric calibration, it did not significantly
reduce the performance of the instrument. This ice layer takes about two days
to stabilise after the cooling of the detectors is switched and can be monitored
by the observations of the Sun and the internal line lamp. The thin ice layer
combines with the intrinsic etalon produced by the passivating outer layer on the
diode array detectors. With hindsight the selection of a 3  µm passivating layer
was suboptimal and 1 µm would have been more favourable. This is because the
combined complex etalon forming the ice layer and the passivating outer Si layer
of the detector can have features that correlate with trace gas absorption.
Over time two issues did arise one of which was expected and one not.
The slow absorption of UV absorbing material was expected because the
phenomenon had been observed with all similar instruments in space. My own
explanation is that organic compounds released from the electrical boards of
the instrument under the hard ultraviolet condition present in space result in
low-volatility organic compounds having π orbitals, i.e. delocalised orbitals
that absorb in the 200−400  nm region. It is well known, for example, that
glycine and ketene, which exhibit such bonding, are present in gas clouds
in space. These types of compound, which have low vapour pressure, then
deposit on the instrument. In the case of GOME this deposition was identified
to be primarily on the scan mirror. This created an absorbing etalon that grew
over the years reducing the throughput in an oscillating manner. Ralph Snel
and Matthijs Krijger from SRON developed a physical model to explain this
behaviour for both GOME and SCIAMACHY, and the thickness of the layer can
be determined. Through inflight characterisation procedures, the impact of
this layer could be minimised.
The second and unexpected issue of this type was the outgassing of the
dichroic mirror, which resulted in a slowly shifting etalon particularly in long-
wavelength part of the channel 3 between 440 nm and 600 nm. This behaviour
could in part be accounted for in retrieval studies and removed in the final
reprocessing of Level 0 to 1 processing of the GOME datasets. As part of the
lessons learned from the SCIAMACHY and GOME‑2 instruments, the gratings
selected to have lower polarisation sensitivity.
Overall, GOME worked almost perfectly throughout its lifetime in space.
The errors in the instrument, such as the etalons described above, were design
risks , which had an impact but could in large part be accommodated for in the
retrieval algorithms. In June 2003, after eight years of data delivery, the second
backup tape recorder system failed after only six months in operation. After
this time, data from ERS‑2 could only be downlinked by direct broadcast. At the
recommendation of the Lead Scientist supported by GSAG, ESA built up a system

104
 GOME and SCIAMACHY

allowing about 30−40% of the GOME data to be downlinked. Wolfgang Lengert,

who later became the ERS‑2 Mission Manager, developed this system. Not all at
ESA were initially in favour of this approach. For example the then ERS‑2 Mission
Manager and others were concerned about the cost implications. As a result
it took some time to develop this system but ESA’s partners around the world,
including NASA, helped by using their ground stations to downlink the data.
Later, as a result of the failure of the ERS‑2 gyros, the ERS‑2 team developed
a unique steering approach for the platform. This reduced the observations of
the extraterrestrial solar irradiance for GOME. However the data obtained from
GOME was of excellent quality and the retrieval algorithms were successfully
adjusted to take these effects into account. GOME continued to work excellently
up to the time that ERS‑2 was decommissioned by ESA in July 2011. It showed
no substantial or unexpected behaviour throughout the 16 years of its
measurement period in space.
One important point to make with respect to Earth observations is that there
are two intrinsically unique and important aspects of making measurements
of atmospheric composition and surface parameters from space. The first is
their global or, in the case of geostationary satellites, regional character, and
the second depends on their duration. Measurements from the ground or
from aircraft provide snapshots or local temporal records, whereas those from
space provide global or regional data over potentially long periods of time. The
value of a dataset for many scientific applications from an orbiting platform
is proportional to a higher power of the number of years of well calibrated
and comparable data. The longer the dataset, the more valuable it becomes.
GOME, with its eight years of global coverage and an additional eight years of
30−40% coverage of the globe, has been a trailblazing success. The scientists
and engineers who built GOME and ERS‑2 and controlled its performance in
space are to be congratulated on this unique achievement. It is one of a limited
number of instruments that has worked for such a long time. I have no doubt
that GOME would have continued to work well for many years if ERS‑2 had not
been decommissioned.
After the launch of ERS‑2, and the GOME commissioning phase, formal
responsibility for GOME was transferred from ESTEC to ESRIN. The ESTEC
team began then to concentrate on the development of GOME‑2 as part of the
MetOp series of platforms built for Eumetsat. As expected, the data processing
of any instrument evolves during its life in space, for a number of reasons. One
significant factor has been the continuous improvement in computer storage
systems and the speed of the world wide web, providing an almost exponential
growth in calculation power and the ability to manipulate mathematically
and transfer large datasets even remotely. In retrospect, the data rate of GOME
at launch seems trivially small. But in 1995 a significant logistical effort was
involved in ensuring the delivery of data to the then small team of scientific
and pre-operational data users.
GOME provided what at the time were uniquely large amounts of data and
much attention was paid to their calibration. Given the necessary funding and
staff, the origins of systematic errors in retrieval algorithms and degradation
of the instrument inflight can be identified by instrument monitoring and
removed by improving the algorithms and reprocessing of the data. This
reprocessing of GOME data and the addition of new operational data products
is an evolving process. Similar to NASA projects, in the GOME/SCIAMACHY
project, data reprocessing typically takes place on a two-year cycle. The results
from the outset were outstanding but the improvements in precision and
accuracy through this reprocessing cycle were of key importance in providing a
dataset of value for environmental, atmospheric and climate change research.
In addition to their use in scientific applications, these data are required to
monitor objectively atmospheric pollution and thereby provide datasets for
use in the development of international environmental policy, in particular for
their transparency and verification.

105
SP-1326

12. Retrieval of Data Products from GOME

As a passive remote sensing instrument, GOME measured the upwelling
radiation from the atmosphere and the extraterrestrial solar irradiance.
The detectors used in GOME are monolithic linear diode arrays. GOME has
1024  detectors per spectral channel and three Polarisation Measurement
Devices (PMDs). In channel 1 about 75% of the channels are read out. The linear
diode arrays comprise 1024 Si semiconductor capacitances, which are charged
on reading. The interaction of electromagnetic radiation with the detectors
discharges the capacitance.
The linear array detectors are read serially by a multiplexer, which
recharges them, as it reads. An analogue to digital converter then generates
the Level 0 product, which is stored either on the tape recorder for periodic
downlink or, after June 2003, recovered by direct broadcast when close to
a network ground station. The latter system was established by Wolfgang
Lengert, who worked on GOME and was the last ERS Mission Manager at ESA.
The geophysical Level 1 data product is generated by the Level 1 processor.
The Level 1 data products for GOME are the extraterrestrial solar irradiance
and the Earthshine radiance at the solid angle of viewing. The calibration of
GOME involved both extensive preflight characterisation and calibration of
radiometric and spectral measurements and inflight calibration. The latter
requires the determination for the ~3840 individual diode array detectors and
the three PMDs of GOME of the following:

—— the radiometric calibration, which uses the dark signal during eclipse;
observations of the Sun via the diffuser plate on a daily basis; observations of
the full Moon, when available, via the diffuser plate, for long-term radiometric
calibration; and the LEDs in each spectral channel;
—— the spectral calibration, which uses periodic observations of the Pt/Cr/Ne
hollow cathode discharge line emission lamp, and observations of Fraunhofer
solar lines.

The Level 1 to 2 processor is used to invert the retrieved radiance and irradiance
into Level 2. Utilising the Beer–Lambert law and considering radiative transfer
through the atmosphere, the intensity of solar radiation reaching GOME can be
approximated by the following equation:

I^mh = cI0 ^mh exp $-/ v'i ^mh SCDi - / v * i ^mh SCDi - v Ray ^mh SCDRay - ^vhMie ^mh
i i
(1)
I^mh = cI0 ^mh exp $-/ v'i ^mh SCDi - / v * i ^mh SCDi - v Ray ^mh SCDRay - ^vhMie ^mh SCDMie .
i i

where I0(λ) is the intensity of solar radiation reaching Earth, i represents the
gas, c is a factor that depends on the surface spectral reflectance and multiple
scattering in the atmosphere; the absorption cross-section, σ(λ), of the
absorbers separated into the slowly varying low-pass contribution σ*(λ) and
strongly varying high-pass or differential part σ′(λ) σ(λ) = σ*(λ) + σ′(λ); SCDi is
the slant column density of the gas; SCDray is the slant column of air; SCDMie is
the slant column of particles. This can then be approximated by

I^mh = cI0 ^mh exp $-/ v'i ^mh SCDi - / a p m p . (2)

i p

where apλp is a polynomial that fits the broadband or low-pass scattering and
absorption terms, and the high-pass or differential structure term Σσ′i(λ)SCDi
yields the set of SCDi. The equation is written as
I0 ^mh
ln = / v'i ^mh SCDi + / a p m p (3)
I^ mh i p

106
 GOME and SCIAMACHY

Standard or simple DOAS works best for weak absorptions when

ln (I0(λ)/I(λ)) approximates well to (ΔI/I0(λ)), where ΔI = I0(λ) – I(λ). The I0(λ)
is the solar extraterrestrial spectrum measured by GOME over its diffuser plate
with the Doppler shift taken into account. The I(λ) signal is the Earthshine
radiance upwelling from the atmsophere measured by GOME. In the fitting it is
necessary to account for gaseous absortions, the Raman infilling of Fraunhofer
lines known as the ring effect and some undersampling structures. Examples
of this standard fitting of the gases NO2, SO2, HCHO and BrO are shown in
Figs 8−11 for a single ground scene at different locations around the world.
Having derived the Differential Optical Density (DOD) of a gas, the
slant column of the gas is obtained by dividing the DOD by the differential
absorption cross section. This is then converted into a vertical column amount
by dividing by the Air Mass Factor (AMF). These calculations are performed
using numerical models that solve the radiative transfer equation. For this
purpose the GOMETRAN/SCIATRAN was developed (Rozanov et al., 1997;
Buchwitz et al., 2001, and references therein).

Figure 8. Simple DOAS fit of NO2 reference

spectrum in the spectral region
425–450 nm from GOME measurements
above Europe. (A. Richter and J.P. Burrows,
University of Bremen).

Figure 9. Simple DOAS nonlinear least-

squares fit of SO2 in the spectral region
312–328 nm from GOME measurements of
the plume of the Nyamuragira volcano in DR
Congo, close to the equator, 4 December
1995. (M. Eisinger, A Richter and J.P.
Burrows, University of Bremen)

107
SP-1326

Figure 10. Nonlinear least-squares fit of

the differential optical density of HCHO for
a ground scene over Africa. (F. Wittrock,
A. Richter and J.P. Burrows, University of
Bremen)

Figure 11. Nonlinear least-squares fit of

the BrO differential optical depth, retrieved
from a single GOME scene. (A. Richter and
J.P. Burrows, University of Bremen)

For stronger absorptions, modifications of the DOAS approach are used to

fit the vertical column taking the radiative transfer into account. Similarly, the
optimal estimation approach can be used (Rozonov et al., 1998). This type of
retrieval is also called the full retrieval method. Briefly, in this case, a forward
radiative transfer model is used to generate the Earthshine expected at the top
of the atmosphere, and the vertical profiles and concentrations of trace gases
are adjusted until a convergence criterion is met. Vertical profiles of O3 were
retrieved in this manner from GOME observations (see Hoogen et al., 1998). For
further details of the preparations for GOME, see Burrows et al., (1994).
The GOME Level 2 data products envisaged at the beginning of the project
comprised the following:

—— O3 total column amount,

—— O3 vertical profiles,
—— total column amounts of NO2,
—— tropospheric column amount of NO2,
—— total column amount of SO2,
—— total column amount of HCHO,

108
 GOME and SCIAMACHY

—— total column amount of BrO,

—— total column amount of OClO,
—— total column H2O,
—— cloud fraction within a scene and cloud top height.

In addition the total column amounts of the trace gases IO and CHO.CHO, which
were retrieved first from SCIAMACHY and then GOME‑2, will be retrieved in
the future from GOME. The cloud fractions and cloud top heights are obtained
using the O2 A‑band absorptions around 760 nm.
The GOME data products have been produced and some interesting results
are highlighted below. First, however, one trivial lesson learned from GOME
is that the value of the data products is dependent on the investment in the
development of the data products and their distribution system. A significant
fraction of the investment in the development of the retrieval algorithms was
funded through and thus relied on the idealism of the scientific community.
There is clearly an imbalance between ESA’s mandate to invest in the
industrial procurement of instruments and the organisation of science teams to
exploit the data. NASA appears to work on a more systematic approach, creating
instrument support for validation and scientific exploitation. In this context
GOME and SCIAMACHY have been a voyage of discovery both for atmospheric
remote sensing and the development of human capacity and understanding the
limitations in the current ESA and national space agency approaches to deliver
scientific return as opposed to support for the development of hi-tech industry.
Although in many respects GOME established the template approach for trace
gas retrieval in European remote sensing, the approach did not provide adequate
resources for the scientific exploitation. The generation of data products and their
use in research and applications was in some senses fortuitous, relying on the
research infrastructure and the idealism of many researchers. This is particularly
challenging as in the case of GOME the experiment had a life of over 20 years.
The differences in the ability to fund the industrial contractors, the scientific
performance in space and that specifically targeted for the scientific exploitation
of the data are dramatic. Until this issue is addressed and the delivery of scientific
data products and their exploitation given adequate resources, the full value of the
investment of both resources and intellectual effort will not be realised.

13. Scientific Exploitation of GOME Data

This section provides examples of the unique set of data products from
the SCIAMACHY project, and in particular GOME. Solar, stratospheric
and tropospheric examples are provided. For tropospheric composition, a
comprehensive explanation of retrieval algorithms and compilation of results
was published in 2011 (Burrows et al., 2011).

13.1 Extraterrestrial Solar Spectral Irradiance, Proxies and

Solar Activity
Quantification of the extraterrestrial solar irradiance as a function of
wavelength is of fundamental importance for our understanding of the Earth
system and of the interactions between the Sun and Earth’s atmosphere.
During its calibration GOME made some unique measurements of the solar
spectral irradiance (SSI) between 230 nm and 793 nm at spectral resolutions of
between 0.2 from 230 nm to 400 nm, and 0.33 nm between 400 nm and 793 nm.
Absolute radiometric calibration requires extra care and complete
knowledge of the inflight degradation of the instrument in space, this remains
a challenge. The absolute calibration of the SSI from GOME requires further
focused study using measurements of the Moon and target regions of Earth of

109
SP-1326

Figure 12. Solar activity measured by GOME

and SCIAMACHY. A composite of the Mg II
index including the contribution of GOME.
(M. Weber and J.P. Burrows, University of
Bremen)

known surface spectral reflectance. GOME has nevertheless produced a unique

record for solar and climate science.
Proxies for solar activity are very valuable for deriving solar behaviour.
The Mg II index is one such proxy, which measures the depth of the solar Mg II
Fraunhofer line. The record of the Mg II index derived from GOME by Mark
Weber is shown in Fig. 12. The solar cycle is readily observed.

13.2 Stratospheric and Upper Atmospheric Science

In the 1990s a key issue for stratospheric science was the chemical depletion
of O3 above the northern hemisphere in spring. The complex Brewer–Dobson
circulation, coupled with wave activity, is such that fortunately O3 is more
abundant above the northern hemisphere and a warmer stratospheric polar
vortex. This results in much better protection for the biosphere from biologically
damaging radiation in the northern hemisphere than in the southern hemisphere.
Over the period 1995–2000 the research programme on stratospheric ozone
of the German Ministry of Science and Technology, through both its support
for the German Aerospace Agency (DARA, later DLR) and the national ozone
research programme, provided funding for our research using GOME data in
Bremen. R. Zellner (University of Duisberg-Essen), coordinator of the national
programme, supported by U. Schmidt (University of Frankfurt). They and their
colleagues were strong supporters of the GOME activities.
An additional source of funding was from the EU Framework Programmes.
The then Stratospheric Unit within the European Commission and its Ozone
Coordinating Unit, organised by John Pyle and Neil Harris of the University of
Cambridge, also supported the development of the GOME science applications.
My colleagues Klaus Künzi from the University of Bremen and Richard Wayne
from the University of Oxford also provided great support in these years. This
enabled the team in Bremen and our colleagues across Europe to begin using
GOME data for research. Unfortunately ESA did not or was not allowed to provide
direct support for scientific exploitation of GOME data. The Global Monitoring for
Environment and Security programme, now renamed Copernicus, was a double-
edged sword with respect to funding for scientific applications. This was and is
because research is explicitly not funded by this programme.

13.2.1 Total Column of O3

The total column of O3 is retrieved using O3 absorption in the Huggins bands

typically between 325 nm and 335 nm. This spectral window was recommended

110
 GOME and SCIAMACHY

by the IUP/UB team and its partners in 1994 as an optimal spectral window for
the first retrievals of the total column amount of O3 from GOME. This spectral
window has become a standard and has been used for the retrieval of total O3
from GOME, SCIAMACHY and GOME‑2.
There is more information about O3 available in the GOME measurements,
which can be used in the future to improve the precision and accuracy of the
measurements and thereby provide more information about O3 changes in the
atmosphere.
Figures 13 and 14 were produced using a merged dataset of GOME,
SCIAMACHY and GOME‑2 data, generated at IUP-UB. Figure 14 shows the
development of the ozone hole over the northern hemisphere, retrieved using
the so-called weighting function modified DOAS approach (Rozanov et al.,
1998). There are now several merged datasets for O3 from the European early
morning platform produced using different but related techniques and giving
similar results. There are also data assimilation schemes that use the different
measurements, but these are considered to combine model and measurement
information and are therefore less valuable for trend or change analyses.

Figure 13. The total O3 column amount above the northern hemisphere in March, 1996–2012, shown in a polar projection using the merged
GOME, SCIAMACHY and GOME‑2 data record. The background colours indicate the contributions of the different instruments. (M. Weber
and J.P. Burrows, University of Bremen)

111
SP-1326

Figure 14. The total O3 column amount above the southern hemisphere in October, 1995–2011, shown in a polar projection using the
merged GOME, SCIAMACHY and GOME‑2 data record. (M. Weber and J.P. Burrows, University of Bremen)

The colour scheme in Figs. 13 and 14 indicates the different instruments

that provided data for the northern and southern hemispheres. In Fig. 14 the
behaviour of the O3 hole over Antarctica over the period 1995–2011 was clearly
stable, except in 2002 when a sudden warming in the southern hemisphere was
first observed. This was studied extensively using GOME and SCIAMACHY,
both of which were fortunately working during the southern hemispheric
winter and spring of 2002. Many groups worked on this phenomenon. It
remains unclear whether this was an unusual event or an early warning of
increasing wave-driven circulation in the upper atmosphere.

13.2.2 Vertical Profiles of O3

The derivation of the vertical profile of O3 from GOME measurements is one

of the more challenging retrievals. The idea was first suggested by Singer
& Wentworth (1957). The concept utilises the wavelength dependence of
penetration depth of solar ultraviolet radiation to invert mathematically a
vertical profile for O3 (Hoogen et al., 1998, 1999a,b).

112
 GOME and SCIAMACHY

The retrieval approach to determine vertical profiles is explained schematically

in Fig. 15 together with some validation results. Different variants of this approach
are used by different research teams. The approach is to use a forward model to
determine the upwelling radiance in the ultraviolet and potentially also the visible,
i.e. covering the Hartley and Huggins bands. The use of the visible spectral region
with the backscattered radiation was found to have correlations between the
surface absorptions, primarily those of chlorophyll and the Chappuis O3 features.
Work on this problem is continuing. However, the chlorophyll absorption has been
used for type 1 waters to produce phytoplankton maps.
The NASA programme first developed the vertical profile retrieval for BUV
and SBUV. They later applied this to SBUV-2 together with the NOAA/NASA
SSBUV instrument. A total of nine teams developed different vertical profiling
algorithms of relevance for GOME. Figure 15 shows a schematic diagram of the
general approach and some results. The issue now is to develop consistent and
consolidated datasets using the data from the early-morning platforms.
Using GOME observations, the temporal evolution of the O3 concentration
at a given potential temperature within the vortex could be observed. Figure 16
compares the O3 concentrations at the potential temperatures of 475K, ~19 km,
and 550K, ~22 km, the total O3 and NO2 column amounts on 2 April 1997 for one
of the largest ozone holes observed above the northern hemisphere thus far.
The depletion of NO2 and O3 is readily observed, the latter being dominant in
the lower stratosphere. The right panel of Fig. 16 shows a transect along the
orbit of the vertical profile of O3 retrieved from GOME above Europe and the
Atlantic. The depleted O3 in the polar vortex is readily identified.
The chemical depletion and the increase in O3 resulting from subsidence
as the air mass sinks over the lifetime of the polar vortex are clear. Figure 17
shows this behaviour during winter and spring over the period 1997–2000.
In this manner, the chemical loss and the dynamical subsidence of O3 at two
potential temperatures can be studied during the evolution of the vortex. The
average O3 sonde values within the vortex during this period are also shown.
Good agreement between sonde measurements and satellite retrieved average

Figure 15. Left panel: The approach used for the retrieval of vertical profiles of O3. Right panel: Some comparisons of the retrieved profiles
with the initial guess or a priori measurements from an O3 balloon sonde and from the HALOE instrument on NASA’s UARS (M. Weber,
K.‑U. Eichmann, K. Bramstedt and J.P. Burrows, University of Bremen).

113
SP-1326

Figure 16. Left panels: Retrieval of the vertical profile of O3, showing the O3 concentration (molecules/m3) at the potential temperatures of
475K, ~19 km, and 550K, ~22 km, in the lower stratosphere together with the total column of O3 (in Dobson units) and NO2 (molecules/cm2),
retrieved from GOME measurements on 2 April 2002. Right: Latitudinal transect along the orbit, showing the high O3 concentration outside
the polar vortex and the chemical depletion of O3 within the vortex. (ESA/DLR/M. Weber, K.‑U. Eichmann, K. Bramstedt and J.P. Burrows,
University of Bremen)

Figure 17. Average O3 volume mixing ratio

(VMR) at 475K within the polar vortex,
determined using the GOME profile
retrieval for the Julian day of the year in
1997 1998, 1999 and 2000 (blue points),
and the values from sonde data (yellow
points). (K.‑U. Eichmann, M. Weber and J.P.
Burrows, University of Bremen

vortex value for O3 was achieved in these years. This was made possible by
European and German funding of polar ozone research.

13.2.3 Stratospheric OClO and NO2 observed by GOME

During late autumn and early winter the polar vortex literally spins up. In
the polar night the polar vortex cools, and NOx is converted first to N2O5 via
the chemical reactions (13), (16) and (27) described in section 3. N2O5 is then
transformed on aerosols and Polar Stratospheric Clouds to nitric acid by reaction
(28). Eventually, the PSCs become nitric acid trihydrate, HNO3.3H2O, which is
the most stable azeotrope. The consequence of these reactions is shown in the

114
 GOME and SCIAMACHY

observations of the total NO2 column amount above the northern hemisphere in
January and March 1997 and 2001, see Fig. 18. The loss of NO2 is readily visible.
This process has presumably always taken place, but the difference in the past
50 years is that the loading of halogens, resulting from the tropospheric release
of ozone-depleting substances, has resulted in a change in the chemistry and the
catalytic destruction of O3 in the lower stratosphere at high latitudes in spring.
A sudden stratospheric warming is defined as an event where the polar
vortex of westerly (eastward) winds in winter slows down abruptly (i.e. over
the course of a few days) or even reverses direction. This is accompanied
by a rise in stratospheric temperature by several tens of K and is a dramatic
meteorological event. In Fig. 19 a sudden warming, which resulted in the polar

Figure 18. Denitrification and loss of NO2

in the polar vortex during the development
of the first ozone holes over the northern
hemisphere in 1997 and in 2001.
(A. Richter and J.P. Burrows, University
of Bremen)

Figure 19. Observations of stratospheric

warming over Antarctica in 1997 and the
total NO2 column density. The polar vortex
was pushed away from the pole by a pulse
of energy from the tropics, shifting it to
lower latitudes where it broke up and
dissipated. (A. Richter and J.P. Burrows,
University of Bremen)

115
SP-1326

vortex being pushed to lower latitudes, is observed in the NO2 vertical amount.
The vortex air then mixes with the southern hemispheric air, diluting the low
O3 over the hemisphere.
An interesting test of our understanding of the stratosphere occurred in 2002,
during a major warming event, defined as when the westerly winds at 60°N and
10  hPa (geopotential height) reverse, i.e. become easterly (westward), above
Antarctica. This was a unique event in the southern hemisphere, whereas they are
regular features in winter and spring above the northern hemisphere, and were
first observed above Berlin in late January 1952. The question remains as to whether
this was a sign of changing dynamics, resulting from the increased amount of
energy in the atmosphere, the result of the absorption by the greenhouse gases
released by anthropogenic activity, or a phenomenon that occurs periodically but
had not been previously observed in modern times. This situation is similar to the
first hurricane observed in the South Atlantic off the coast of Brazil in 2004.
Figure 20 shows observations of NO2, OClO and O3, as well as estimates
of the potential vorticity and temperature from ECMWF and UK Met Office
data assimilation models. OClO is formed in the pathways of the reaction (43)
between BrO and ClO. The figure shows clearly that OClO and thus ClO are
activated as the NO2 is lost.
The temporal behaviour of the OClO slant column density at a solar zenith
angle of 90° and the NO2 vertical column between latitudes 70° and 80° is shown
in Fig. 21. The latter focuses on the unusual behaviour in 2002 and the range of
behaviours from 1996 to 2001. These data have now become diagnostics that
are used in the WMO Global Atmosphere Watch reports on the monitoring of the
ozone layer to characterise the behaviour of the polar vortex,1 and represent a
unique data product provided from GOME measurements. The datasets have been

Figure 20. Observations of NO2, OClO and O3 and estimates of potential vorticity and temperature in the polar vortex before 1 September
2002, during 25 September 2002 and after the first major sudden warming of the polar vortex in the southern hemisphere. (A. Richter, M.
Weber and J.P. Burrows, University of Bremen)

1 WMO Global Atmosphere Watch: www.wmo.int/pages/prog/arep/gaw/ozone

116
 GOME and SCIAMACHY

Figure 21. Left panels: OClO slant column densities, observed at a solar zenith angle of 90°, plotted for each day of the year in the northern and
southern hemispheres between 70° and 80° latitude. Right panels: The corresponding vertical column densities of NO2. Data retrieved from
GOME observations. The ranges of values for the years 1996–2001 are indicated in grey. (A. Richter and J. P. Burrows, University of Bremen)

extended by the IUP/UB team to include data from SCIAMACHY. Following the loss
of Envisat, these data products can be produced from GOME‑2, which flies an hour
earlier than GOME on MetOp-A, with an equator crossing time of 09:30.
OClO is an important indicator of the extent of chemical processing within
the polar vortices on polar stratospheric clouds. However OClO is rapidly
photolysed and as a result its stratospheric loading maximises at large solar
zenith angles close to dusk. The appearance of OClO is anti-correlated with the
amount of NO2 as this competes with BrO to react with ClO to form ClONO2.
As can be seen in Fig. 20, in late September 2002 the polar vortex above
Antarctica split into two parts as a result of the flow of energy from lower
latitudes. One of the vortices moved to lower latitudes while the second moved
back over the pole. This is clearly observed in the total O3, NO2 and OClO. In
the lower panels of Fig. 21, the behaviour of NO2 and OClO was clearly affected
in 2002 by the sudden stratospheric warming, which was observed for the first
time in the southern hemisphere in spring in 2002.

13.3 Tropospheric Science using GOME Trace Gas Composition

Prior to the launch of GOME, measurements of the upwelling UV radiation
by the TOMS instrument on NASA’s Nimbus‑7 had been used to observe
tropospheric SO2 from volcanic eruptions. This instrument was focused on
measurements of the total column of O3 and did not have sufficient sensitivity to
provide measurements of SO2 in the boundary layer. The combination of TOMS
and SBUV and TOMS and SAGE had been used to estimate the tropospheric O3
loading, and led to the discovery of elevated O3 in the tropics.
GOME, proposed as an approach to deliver rapidly an important subset of
objectives of the SCIAMACHY project, was the first remote sounding instrument
to probe explicitly for tropospheric trace gas composition. The SCIAMACHY
concept is intrinsically superior to that of GOME because it measures almost
simultaneously the upper atmospheric limb profiles and total column amounts of

117
SP-1326

key constituents such as O3, NO2, BrO and aerosols, which are present in the upper
atmosphere in significant amounts. When SCIAMACHY and GOME were proposed
in the late 1980s, many scientists were sceptical that valuable information about
tropospheric composition could be retrieved. For this reason the name GOME
focused attention on the instrument’s ability to observe global O3 from space.
One of the most important contributions made by the ERS project has been
to demonstrate unambiguously through GOME that accurate measurements of
key tropospheric trace gases can be made from space. The data rate available to
ERS‑2 for GOME was limited, and this in turn limited the spatial resolution of the
results. However, GOME’s ability to retrieve tropospheric trace gases emitted by
natural phenomena (e.g. biogenic emissions, volcanic eruptions, fires, surface
fluxes from the land, the ocean or ice, etc.) and resulting from human activities
(e.g. fossil fuel combustion, biomass burning, agriculture, etc.) represented a
breakthrough. The amounts and distributions of these constituents, which are
controlled by biogeochemical cycling, atmospheric chemistry and dynamics,
can be objectively measured from space using nadir observations or, even better,
the combination of nadir coupled with limb or occultation observations of the
upwelling electromagnetic radiation from the top of the atmosphere.
This limitation on GOME arising from the available resource on ERS-2
with respect to the data rate resulted in the poor spatial resolution of GOME of
320 × 40 km but global coverage at the equator in three days and more rapidly
at higher latitudes. We knew at the time that we needed much higher resolution
to resolve important tropospheric processes but had to accept this compromise.
The three-day global coverage was sufficient to catch many phenomena. In
addition, the somewhat higher spatial resolution of 80 × 40 km for 10% of the
GOME observation time enabled assessments of the impact of the lower spatial
resolution on our interpretation of trace gas retrievals.
Only 50 years ago it was thought that the behaviour of O3 in the troposphere
could be explained by transport from the stratosphere and deposition at the
surface. We now know that there are large catalytic sources and sinks for O3
in the troposphere, which determine the chemical weather or air quality
of a given air mass. These processes have been briefly described above. The
modification of these cycles as a result of human interventions can have large
impacts on air quality, agriculture and human health.
The retrieval from space of the total column amounts of the key trace
gases involved in the production and destruction of O3 and aerosols in the
troposphere, has enabled the quantification of natural biogeochemical cycles
and of human impacts on the lower atmosphere from local to global scales.
Some examples of the results and highlights of retrievals of tropospheric
species from GOME measurements are described in the following sections.

13.3.1 Tropospheric NO2 as a Surrogate for Tropospheric NOx

The oxides of nitrogen, NO and NO2, are key free radicals that play important
roles in the mesosphere and stratosphere. NO and NO2 also play special roles in
tropospheric chemistry, because, as explained above, they are chain carriers in
catalytic cycles producing tropospheric O3, react with OH to produce the acids
HONO and HNO3, and in the dark their reactions produce NO2, an oxidant,
andN2O5, the acid anhydride of HNO3. The sum of NO and NO2, NOx, and O3 in
the troposphere are often in the Leighton photostationary state as a result of
the ‘do-nothing’ cycle described in section 3.3. However, the reactions of RO2
and HO2 with NO to form NO2 are key propagation reactions in the catalytic
production of O3 in the troposphere and lower stratosphere.
NO has ultraviolet emissions, produced by the absorption of NO at short
wavelengths in the upper atmosphere, which were observed by GOME‑
However, this absorption by NO occurs below 230 nm and was therefore not
detected by GOME. As a result, NO cannot retrieve NO in the troposphere or

118
 GOME and SCIAMACHY

stratosphere. In contrast, NO2 turns out to be an ideal gas for detection by

passive atmospheric remote sensing techniques, such as DOAS, optimal
estimation and related approaches. Its complex ultraviolet and visible spectra
have many features suitable for its detection. One example was shown in Fig. 8.
The retrieval of the tropospheric column using only nadir measurements
requires the separation of the stratospheric column. This can be achieved by
assuming that the NO2 loading is negligible in clean regions of the troposphere
and attributing the column to that in the stratosphere. On average, assuming
longitudinal homogeneity is reasonable over a period of a month, this enables
the tropospheric slant column to be estimated. The latter is then converted
into a tropospheric vertical column for cloud-free regions by dividing by an
Air Mass Factor (AMF), which is calculated taking into account the path of
electromagnetic radiation through the atmosphere, accounting for the vertical
profile of NO2, absorption by gases and scattering by molecules and aerosols. If
the surface spectral reflectance is large, as in the case of snow, then the AMF
is close to the geometric factor of (1 + cosec(θ)), where θ is the angle between
the line of observation of the satellite to the ground scene and from the
ground scene to the Sun. Where surface absorption is strong then atmospheric
scattering becomes significant, for example in the UV.

13.3.2 Global Tropospheric NO2

In 1997 the strong El Niño–Southern Oscillation (ENSO) resulted in drought in

the Horn of Africa and in Indonesia. As a result fires, often ignited by humans,
became an environmental issue. It is interesting to note that on 26 September
1997, an Airbus A300 crashed into woodland 18 miles from Medan, Indonesia,
in low visibility, killing all 235 passengers and crew. The crash site was at an
altitude of 3000 feet (915 m).2
Figure 22 shows the global tropospheric NO2 vertical column retrieved from
GOME. This was sometimes also called the excess as it represents the difference
between the NO2 at a given latitude and longitude and that at the same latitude
around the international dateline of 180°. More sophisticated approaches
have been developed but the underlying concept remains the same, i.e. the
subtraction of the stratospheric column from the total column using regions
known to have negligible NO2 in the troposphere. The pollution centres in the
North America, Europe, Southern Africa and Asia are clearly visible in Fig. 22.
In addition, biomass burning in South America, Africa and Indonesia resulted

Figure 22. The excess or tropospheric

vertical column amount of NO2 in
September 1997 retrieved from GOME
measurements, generated by subtracting
total column NO2 values along the dateline
from those at the same latitude. The high
tropospheric NO2 close to Antarctica results
from the polar vortex being asymmetric,
and an excess NO2 results from reduced
amounts of stratospheric NO2 in the
polar vortex. (A. Richter and J.P. Burrows,
University of Bremen)

2 World Air Crash Disasters, Medan Indonesia: www.medanku.com/world-air-rash-

disaster/#ixzz22IDRZfSD

119
SP-1326

Figure 23. NO2 plumes correlated with fire

counts (top right), lightning estimates
(bottom left) and centres of population;
average values, September–November
1997. (ESA/NASA/A. Richter and
J.P. Burrows, University of Bremen)

Figure 24. Top: Annual cycle of tropospheric

NO2 above equatorial Africa, as derived
from SCIAMACHY in 2003, together
with fire counts from NASA’s MODIS and
lightning estimates from NASA’s OTD.
(A. Richter and J.P. Burrows, University of
Bremen) Bottom: Monthly values of NO2
over Montana, USA, as retrieved from
SCIAMACHY in 2004. (NASA/T.H. Bertram
and R.C. Cohen, University of California
Berkeley, and A. Heckel, A. Richter and
J.P. Burrows, University of Bremen)

in large NO2 plumes flowing out over the ocean. The fires, a consequence of
ENSO and anthropogenic activities, are also visible. The outflows of NO2 from
the Congo basin at low altitudes, and from East Africa at higher altitudes, can
also be observed. It turns out that the NO2 in ship tracks can also be retrieved
successfully from such pictures.

13.3.3 NO2 Emissions in the Tropics and Subtropics

In the tropics lightning, biomass burning, domestic heating, fossil fuel

combustion, the bacterial oxidation of NH4+ and reduction of NO3– in soils are
all sources of NO, which then reacts with O3 to produce NO2. It is important to

120
 GOME and SCIAMACHY

understand the strength of the different emissions and the human impact on
the production of NOx.
The emissions of NO2 from fires were assessed by using fire counts for example
from simultaneous measurements made by Envisat’s AATSR at night (Fig. 23).
Lightning estimates used GOME data and observations of lightning by NASA’s
Optical Transient Detector (OTD), carried as a secondary payload on Pegasus, an
Orbital Sciences Corporation rocket launched on 3 April 1995 into Earth orbit at an
altitude of approximately 710 km (446 miles), with an inclination of 70°.
Part of the NO2 over the rainforest and savannah during the monsoon or
rainy seasons can be attributed to the bacteriological reduction of NH4+ and
the oxidation of NO3–, which releases NO, N2O and N2. The NO reacts with O3 to
produce NO2 (Bertram et al., 2005).
Figure 24 (top) shows the annual cycle of NO2 above Africa, as derived from
SCIAMACHY in 2003 together with fire counts, in this case from NASA’s Moderate
Resolution Imaging Spectroradiometer (MODIS), and lightning estimates from the
OTD. Additional sources of NO2 attributed to soils and domestic fires are required
to explain the observations. Figure 24 (bottom) shows observations of similar
behaviour over Montana, USA, following the application of manmade NH4NO3
as fertiliser, using in this case the high spatial resolution of SCIAMACHY. The
production of NOx from the bacterial degradation of fertiliser in soils has also
been found to be significant over Asia and parts of Africa.

13.3.4 The Changing Tropospheric NO2 Column

As discussed above, the increase in tropospheric NO2 is the result of air

pollution in urban areas, from fossil fuel combustion for heating, industry
and transport. At the University of Bremen, Andreas Richter and I began
investigations of the spatial and temporal changes in tropospheric loading of
NO2. This led to the discovery of large increases in the industrialising areas of
Asia, particularly China. GOME was launched some 15 years after China began
its economic reforms under Deng Xiaoping in 1979. Between 1995 and 2011, the
lifetime of GOME, China’s GDP rose from US$0.7 trillion to US$8 trillion, fuelled
by the growing use of fossil energy, and rapid improvements in standards of
living. Over the same period, the introduction of legislation to curb pollution
in Western Europe, and changes in industry and the transport fleet in eastern
Europe following perestroika, resulted in reduced emissions of NOx and lower
tropospheric column amounts of NO2 over Europe and parts of North America.
The statistically significant annual changes in the tropospheric NO2 column
are shown in Fig. 25 using GOME data from 1996 to 2002 (Richter et al., 2005).
The large increase over China and the reduction over Europe during this period
are clearly visible. When these remarkable findings were published in 2005,
they were in disagreement with the estimated changes in bottom-up emission

Figure 25. Statistically significant linear

annual changes in NO2 determined on a
1° × 1° spatial grid using the retrievals
of NO2 column from GOME, 1996–2002.
(A. Richter and J.P. Burrows, University of
Bremen)

121
SP-1326

inventory data for Asia available at that time. In the meantime agreement is
now much better.
In Fig. 26 the monthly and annual average composite values for NO2
tropospheric columns are plotted for data from GOME, SCIAMACHY and GOME‑2
for the region 30−40°N 110–124°E in eastern China from 1996 to 2010. The
selection of the area is somewhat arbitrary. The ERS Envisat and MetOp A have
or had equator crossing times of 10:30, 10:00 and 09:30, respectively, so that
small changes as a result of the diurnal cycle of NO2 are also identifiable. The
global data from GOME and SCIAMACHY overlap, as do those from SCIAMACHY
and GOME‑2 data. The retrievals of tropospheric NO2 are in good agreement.
As expected, a seasonal cycle in NO2 is observed, with higher values in
winter than in summer. This is qualitatively explained by the seasonal cycles
in the photolysis frequency of NO2 and the combustion of fossil fuels. The
increase first observed in the GOME data continues. The impact of attempts
to reduce pollution for the Olympic Games in Beijing in summer 2008 and the
global economic recession in 2008 are clearly observable. Since then, the NO2
column over China has continued to increase.
Figure 27 shows the first 15 years of the tropospheric columns of NO2
derived from GOME and SCIAMACHY over six regions where statistically
significant annual changes could be determined from GOME data. The plot
shows increases over Asia and reductions over Europe and North America.
The improvement in air quality in Europe may now have stopped, while over
China the NO2 column continues to decline. The area shaded blue in the figure
indicates GOME data alone published some years ago.

Figure 26. Monthly changes in tropospheric NO2 over central eastern China (30°−40°N, 110–124°E) derived from GOME, SCIAMACHY and
GOME‑2 data for <10% cloud cover. (A. Richter and J.P. Burrows, University of Bremen)

Figure 27. Merged annual tropospheric

NO2 columns from GOME and SCIAMACHY
relative to 1996, observed over six regions:
central east coast of the USA, Western
Europe, Poland, east-central Europe and
Hong Kong. The area shaded blue indicates
GOME data already published.

122
 GOME and SCIAMACHY

Figure 28. Left: NO2 emissions from shipping retrieved from GOME‑2 data, 2007–9. (A. Richter and J.P. Burrows, University of Bremen)
Right: Increases in NO2 emissions from ships crossing the Indian Ocean between Sri Lanka and Indonesia, retrieved from GOME,
SCIAMACHY and GOME‑2 data, 2008. TEC, Total Electron Content. (A. Richter. V. Eyring, K. Franke and J.P. Burrows, University of Bremen)

13.3.5 NO2 in Ship Tracks and Inferences for NOx Emissions

One of the discoveries of the GOME and SCIAMACHY missions was the
observation of NO2 emissions from ships (e.g. Richter et al., 2005). The research
groups at the universities of Heidelberg and Bremen found clear evidence
of increased emissions from ships using the route between Sri Lanka and
Indonesia (Fig. 28; Richter et al., 2004; Franke et al., 2008/2009).

13.3.6 OVOCs from GOME as Surrogates for VOCs

Unfortunately, in the UVA–visible spectral regions, there is no strong

absorption of CH4 or VOCs such as isoprene, terpenes or aromatics. However,
the aldehydes and ketones do have absorption in these spectral regions as a
result of their delocalised bonds. Thus one of the original target gases for
GOME and SCIAMACHY were formaldehyde and HCHO, and glyoxal, CHO.
CHO, was added later.
HCHO is produced during the oxidation of CH4 and VOCs emitted from the
biosphere or by anthropogenic activities. In addition, as a result of biomass
burning, HCHO is injected in both directions and as a result of the oxidation
within the plumes from biomass burning (Ladstättter-Weißenmayer & Burows,
1998; Thomas et al., 1998).
Figure 29 shows the global average composite HCHO column amounts
retrieved from GOME from 1996 to 2002. Initially, this type of picture was
confusing but it was realised that both biomass burning and biogenic sources
release large amounts of VOCs that are oxidised to HCHO. Anthropogenic
pollution also adds VOCs.
Later, using SCIAMACHY, CHO.CHO was also identified (Wittrock et al., 2006).
The ratio of the columns of CHO.CHO to HCHO provides some sources speciation
information as aromatic and biogenics have different oxidation product ratios.
When coupled with NO2 the emissions to the air mass under investigation can be
synergistically characterised. A source of CHO.CHO discovered over the ocean is
considered to be a result of biogeochemical coupling between the ocean and the
lower troposphere.
Tropospheric O3 can also be retrieved from GOME measurements using a
variety of algorithms. As approximately 90% of the atmospheric O3 is typically
above the tropopause, it is necessary to obtain very accurate estimates of both
the total column density of O3 and its column above the tropopause.
In tropics and subtropics the tropopause is relatively well behaved and
tropospheric O3 can be retrieved using a residual technique similar to that used

123
SP-1326

Figure 29. Average composite HCHO column

amounts retrieved from GOME data,
1996–2002. (F. Wittrock, A. Richter and
J.P. Burrows, University of Bremen)

for the TOMS data (Fishman et al., 1990, 1991, 1992, 1996). The TTO algorithm
is particularly elegant (Hudson et al., 1996; Hudson & Thompson, 1998; Kim
et al., 1996). The TTO approach and an assessment of the profile regimes led
to the discovery that the tropical tropopause is expanding (Hudson, 2012, and
references therein). Another approach to derive tropospheric O3 used the cloud
top heights (Ziemke et al., 1998).
Figure 30 shows the tropospheric columns of HCHO, NO2 and the tropospheric
excess column of O3, the latter being the difference between the total column
density of O3 at a given location and that at the same longitude around the
international dateline, which is assumed to have low and reasonably constant
O3. The latter assumption is simple but useful for explanatory purposes.
The O3 is produced by the chain reactions described above, resulting from
both natural and anthropogenic emissions of NOx and VOCs. In the tropics,
lightning, biomass burning and biogenic emissions release the VOCs and NOx
required to produce this O3 photochemically. However, domestic heating and
industry cannot be ignored. As can be seen in Fig. 30, this results in a plume of
air producing O3 flowing from the Congo basin at low altitudes into the South
Atlantic and, conversely, a plume of air at higher tropospheric altitudes, sending
O3-rich air and its precursors towards Australia in the southern hemisphere.
There are also retrieval algorithms that derive tropospheric O3 outside the
tropics and subtropics, although this is more challenging as frontal systems
vary the height of the tropopause. An adequate discussion of these retrieval
approaches and their results is beyond of the scope of this chapter.

13.3.7 SO2 from Volcanoes and Pollution

As TOMS measurements had been used to derive SO2 from volcanic eruptions
(Krueger, 1983; Krueger et al., 1996, 1998), GOME with its much superior
spectral coverage was expected to measure the release of SO2 from volcanoes.
Michael Eisinger, who later joined ESTEC to work on GOME‑2 from the DOAS
group at IUP/UB, supported by Andreas Richter and myself, was the first to
retrieve SO2 columns from GOME data. The spectral measurements made by
the CATGAS team turned out to be invaluable as the GOME spectral channel
2 has some unique characteristics, including, as mentioned above, a spectral
resolution that varied between 314 nm and 325 nm, and the FWHM changing
from 0.6 nm at 314 nm to 0.2 nm at 325 nm (Eisinger & Burrows, 1998a,b).
In addition, the first space-based observations of SO2 from tropospheric
pollution were reported. The sources of this pollution include the burning
of lignite and other brown coals, metal smelters and the combustion of oil
containing sulphur. The sensitivity was not sufficient, however, to determine the

124
 GOME and SCIAMACHY

Figure 30. Average composite tropospheric

columns of HCHO, excess O3 and NO2,
September 1997. The excess O3 is different
from the O3 at a given longitude and
latitude and that at the international
dateline at the same latitude. (A Ladstätter-
Weißenmayer and J.P. Burrows, University
of Bremen)

background amounts of SO2 resulting from the oxidation of dimethyl sulphide,

released from the oceanic biosphere, which is below the detection limit.
The composite average SO2 column amounts retrieved from GOME
measurements over the period 1996–2002 are plotted in Fig. 31. Several active
volcanoes can be seen, including Nyamuragira (1.41°S, 29.2°E) in DR Congo, a
large shield volcano similar to Mauna Loa on Hawai’i, which is one of Africa’s
most active volcanoes. Popocatepetl (19.023°N, 98.622°W) and Colima (19.514°N,
103.62°W) in Mexico, and a number of erupting sea mounts can also be observed.
The SO2 from fossil fuel combustion is visible over eastern Europe,
the Middle East, the east coast of the USA and China. The monthly and
annual changes in SO2 above eastern China is shown for merged GOME and
SCIAMACHY data in Fig. 32. There is a clear seasonal cycle, with higher

125
SP-1326

Figure 31. Average composite tropospheric

SO2 vertical column amounts retrieved
from GOME measurements, 1996–2002.
(A. Richter F. Wittrock and J.P. Burrows,
University of Bremen)

Figure 32. Monthly and annual changes in SO2 vertical column amounts above eastern China, retrieved from GOME and SCIAMACHY
measurements. (A. Richter and J.P. Burrows, University of Bremen)

values in winter when more energy is produced for heating. Although the
desulphurisation of coal-fired power plants began in China in 1998, the
programme was rapidly expanded in 2007 to improve air quality for the
Olympic Games in Beijing in 2008. The decrease in 2008, which may also result
in part from the global recession, is readily visible. It appears that after 2010
the SO2 vertical column may have started to increase again.

13.3.8 Tropospheric Halogen Oxides: BrO and IO

One the most surprising discoveries in the early days of the analysis of GOME
data was associated with the retrieval of BrO vertical columns. Large clouds of
BrO at high latitudes in the Antarctic and the Arctic springs were discovered
by the groups at the universities of Heidelberg and Bremen, respectively
(e.g. Richter et al., 1998). Figure 33 shows examples for both hemispheres.
This natural phenomenon represents a coupling and feedback between the
cryosphere and the oxidising capacity of the troposphere, and has been the
subject of much recent research. This remains an important research topic as
the sea ice is changing as a result of global climate change and the impact on
the processes producing BrO, explained in more detail in section 3.3, and the
oxidising capacity of the troposphere is not yet clear.
These clouds of BrO are associated with tropospheric O3 and mercury
depletion events (Barrie et al., 1988; Simpson et al., 2007). Modulation of
the tropopause can also cause some apparent increases in BrO similar to

126
 GOME and SCIAMACHY

Figure 33. Total column BrO showing clouds

of BrO in the troposphere at high latitudes
above the sea ice in spring. (A. Richter and
J.P. Burrows, University of Bremen)

the changes in column O3 around frontal systems. The tropospheric release

is attributed to gas phase and cold brine chemistry and what is known as
a bromine explosion, which involves the acid catalysed reaction between
HBr and HOBr. This occurs globally on sea salt aerosols, but is triggered and
accelerated in low-temperature conditions on cold brine. Conditions termed
'potential frost flower' appear to be involved in the release mechanism
(Kaleschke et al., 2004 , and references therein; Sander et al., 2006).
Other sources of halogen compounds in the troposphere include the
production of halogen-containing compounds by the oceanic biosphere,
followed by exchanges between the ocean and the lower atmosphere, and
volcanic eruptions.
In seawater the ratio of [Cl–] : [Br–] : [I–] is roughly 1 : ~660 : ~15 000. However,
the biosphere concentrates the bromine and iodine, releasing the organohalogen
compounds described above into the oceanic troposphere. The organobromine
and iodine compounds are photolabile. In the case of iodine, molecular iodine is
released by various types of seaweed and algae. As a result of these processes a
background Br and BrO in the troposphere contributes to its oxidative capacity.
In the tropics even very short-lived species, such as halogen-containing
gases in the biosphere, are transported by rapid convection to the stratosphere.
Changes in the Brewer–Dobson circulation, which are expected in a warming
world, could therefore also change stratospheric composition and its halogen
loading.
Astonishingly, the very reactive IO has been retrieved from SCIAMACHY
and GOME‑2 data (Schönhardt et al., 2007/2008). Figure 34 shows an example
of a global estimate of the average composite for IO retrieved from SCIAMACHY.
The oceanic biosphere in the southern hemisphere appears to release more
iodine than in the northern hemisphere. Iodine is released from areas of algae
and kelp, and there are also significant releases from under or within the sea
ice in Antarctica There are also inorganic mechanisms involving O3 photo-
oxidation in the surface water layers..
Direct observations of ClO in the boundary layer are not possible in nadir
viewing because its structure absorptions are below 320 nm. As noted above,
OClO has strong structure absorption spectra, as do OBrO and OIO; the
spectrum of OIO was discovered as part of studies at the IUP/UB for GOME
and SCIAMACHY. However, OClO, OBrO and OIO have such short photolytic
lifetimes that their daytime concentrations are low. OClO has been measured in
the stratosphere close to the terminator during the day and at night by remote
sensing. Similarly, OIO has been observed in the troposphere over kelp beds
at night. At present no OIO or OBrO has been identified during the day from

127
SP-1326

Figure 34. Average composite IO

vertical column amounts, retrieved from
SCIAMACHY measurements, 2003–2010.
(A. Schönhardt, F. Wittrock, A. Richter and
J.P. Burrows, University of Bremen)

GOME, SCIAMACHY or GOME‑2. The role of halogens in the troposphere and

the feedback with climate change is currently a very active area of research.

13.3.9 The Hydrological Cycle: H2O and Cloud Data Products

The spectral resolution of GOME enable combination and overtone bands of

water vapour that absorb in the GOME spectral region to be identified. GOME
has strong and weak water vapour absorptions. The strong structure around
720 nm has been used together with the O2 B-band close to 670 nm to retrieve
the total water vapour.
Measurements of the H2O total column amount have been made over both
land and oceans. Some results for annual variations in the total column water
vapour are shown in Fig. 35. The monthly and seasonal variations, and the
impact of ENSO events, are all clearly visible and have been studied.
The annual changes in H2O were investigated on a 0.5° × 0.5° grid similar
to those of NO2 using the GOME and SCIAMACHY data. The statistically
significant changes derived for the first 10 years of data on H2O are shown

Figure 35. Global distribution of the total

column of H2O, retrieved from GOME and
SCIAMACHY, 1996–2006. (S. Noel and
J.P. Burrows, University of Bremen)

128
 GOME and SCIAMACHY

Figure 36. Statistically significant linear

annual changes in total column H2O derived
from GOME and SCIAMACHY on a 0.5° × 0.5°
grid, 1996–2006. (S. Mieruch, S. Noel and
J.P. Burrows, University of Bremen)

in Fig. 36 and have been validated by comparison with H2O sonde data. The
reason for these changes is not yet clear and may be in part the result of natural
variations in the hydrological cycle or global climate change. Changes in the
height of the tropopause and the tropical extent of the tropical tropopause will
both have impacts. Local changes attributable to water usage in urban areas
have been identified.
Measurements of reflectance using the higher-spatial-resolution Polarisation
Measurement Device and the O2 absorption are used to determine cloud cover
and cloud top heights. The O4 and the ring feature are also alternatives for
determining cloud top heights. Although the cloud data products were initially
intended to be used in the interpretation of the columns of trace gases retrieved
from GOME, they have become important data products in their own right.
Similarly, the absorbing aerosol index data products, first produced for TOMS,
generated from GOME, SCIAMACHY and GOME‑2 measurements have also
become useful data products.
The complexity of the hydrological cycle is such that accurate separation
of anthropogenic influences from natural occurring oscillations or changes
requires long-duration, comparable and consistent datasets. The early morning
measurements and resultant data products initiated with the launch of GOME
are now beginning to be of significant value in this respect.

14. Challenges for the Evolution of European and ESA

Earth Observation – Atmospheric Composition
Observations from space – global observations from low-Earth orbit
(LEO) satellites and regional observations from geostationary Earth orbit
(GEO) satellites – provide objective unique measurements of atmospheric
composition. For many applications, the true value of such data is proportional
to the number of months or years over which such measurements are made,
or the size of the consolidated dataset compiled from a series of instruments
on different satellites. These long-term datasets need to improve their spatial
resolution and temporal sampling in time as this becomes technically feasible.
With such datasets it is possible to distinguish between anthropogenic and
natural causes of change, provide early warning of change and increase the
accuracy of predictions about the amount of air pollution, the stability of the
stratospheric ozone layer and climate change.
Europe initiated the industrial revolution, has produced about 30% of the
ozone-depleting substances responsible for the loss of stratospheric O3 and
the ozone hole, and is a major source of greenhouse gases. Europe invented

129
SP-1326

the Haber−Bosch process, which brought about a revolution in agriculture

resulting in changes in land management and land use across the continent
and the world. European nations and the European Union were also the first
to recognise the consequences of air pollution and to introduce legislation to
mitigate its impacts at national, regional and international scales.
Europe has a complex system for developing and running long-term
atmospheric composition measurements from space, which is still in its infancy.
It is not yet clear how the system will evolve to maintain the key datasets required
for science and policy making. Eumetsat has a mandate from its member states to
make measurements needed for numerical weather prediction and some aspects
of climate services. ESA is an intergovernmental organisation with a mission to
provide for and to promote space science, research and technology applications
for peaceful purposes. However, its member states have not given ESA a mandate
to undertake science, equivalent to that given to NASA in its charter. One focus
of activities at ESA is on the development of industrial capability. The EU is also
involved in Earth observation through its funding of Copernicus, previously
known as the Global Monitoring for Environment and Security (GMES) programme,
and Galileo. The EU, ESA and Eumetsat are involved in Copernicus. Finally, many
EU member states have their own important and significant space agencies. It is
unclear to what extent any of these bodies has a mandate explicitly to undertake
and support science and the generation of knowledge of atmospheric composition.
To assess the needs of the science community, over the last 25 years
ESA has organised a series of user consultation meetings of relevance to
atmospheric composition or chemistry from space, and has published their
recommendations. The 1991 user consultation meeting, for example (ESA,
1991), pointed to:

—— the urgency of having adequate atmospheric composition measurements;

—— the need to ensure the continuity and validity of measurements; and
—— the need for satellite measurements as part of an integrated atmospheric
monitoring system.

Two decades later, the most recent ESA user consultation meeting was held
in Bruges, Belgium, in June 2012, for the atmospheric community following
the loss of ESA’s flagship Envisat. Its recommendations were similar to those
made in 1991. This is because without Envisat there are no European nadir
measurements of key greenhouse gases, no limb/occultation profiling of
atmospheric constituents and parameters, and more.
One of ESA’s programmatic objectives during the 1990s was to establish the
Earth Explorer and EarthWatch missions. The Earth Explorers were envisaged
as promoting innovation and the evolution of European Earth observation,
while the EarthWatch missions would yield the long-term measurements of
surface parameters and atmospheric composition required for applications
such as numerical weather prediction, air quality monitoring and climate
science. While the Earth Explorer missions for Earth observation have
flourished to some extent, with ESA releasing eight calls for missions since
1998, the need for long-term measurements of key atmospheric pollutants and
greenhouse gases as expressed in the EarthWatch approach has floundered.
The continuity of atmospheric composition has not been achieved adequately.
The loss of Envisat has resulted in the complete loss of European vertical profile
measurements and nadir sounding of the total columns of the greenhouse
gases CO2 and methane.
In the 1990s, however, ESA and Eumetsat were able to establish and define
the need for an ozone monitoring instrument for the MetOp series of platforms.
The ESA-developed GOME‑2 was selected as part of the core payload for MetOp.
This improved somewhat on the GOME design but ESA was unable to fund the
addition of 2D detector arrays, which would have significantly improved the
spatial resolution.

130
 GOME and SCIAMACHY

The EU−ESA–Eumetsat Copernicus programme is providing valuable

opportunities for the development of instrumentation to deliver knowledge of
surface parameters. Sentinel‑1, ‑2 and ‑3, currently planned for launch in 2013−14,
will provide mainly remote sensing of the surface parameters and, in the case of
Sentinel‑3, some aerosol and cloud parameters. Sentinel‑4 is an instrument that
will be launched on the Meteosat Third Generation Sounder, MTG-S, in 2019.
This will be the first geostationary instrument for measuring air quality, taking
its heritage from the SCIAMACHY/GOME, the Geostationary Scanning Imaging
Absorption Spectrometer (GeoSCIA)/GeoSCIA++/Geostationary Tropospheric
Pollution Explorer (GeoTROPE)/GeoSCIA-lite proposals submitted to ESA in
response to its calls for Earth Explorer missions in 1998 and 2002, a German
national call in 2005, and subsequent studies by ESA and Eumetsat.
A follow-on from GOME‑2, called Sentinel‑5, will extend the nadir
observations somewhat, and is currently being developed by ESA for the
Eumetsat Polar System (EPS) programme, which will deliver the MetOp Second
Generation polar-orbiting meteorological satellites from 2020 onwards. Both
Sentinel‑4 and Sentinel‑5 are funded by the EU’s Copernicus programme,
which aims to develop the European capacity for Earth observation. ESA is
also developing a Sentinel‑5 Precursor, which will include a Tropospheric
Monitoring Instrument (TROPOMI), a contribution in kind by the Netherlands,
and a shortwave infrared instrument that will measure CO, partly duplicating
the GOME‑2 measurements at higher spatial resolution.
As noted above, the European limb and occultation measurements and
the nadir measurements of CO2 and CH4, which have been such important
and successful aspects of Envisat, are now lost. To fill the data gap, the rapid
development of missions for limb occultation measurements and for nadir
measurements of greenhouse gases is required urgently. These missions
and their data are needed for fundamental Earth system science and for the
development of international environmental policy. In this context, the
possibility of using of the International Space Station – a platform that already
exists and to which Europe has access – should be investigated.
It is interesting to note that it took 6.5 years between the proposal of
SCIAMACHY and the launch of GOME. It will have taken 21 years to go from the
GeoSCIA proposal to the launch of Sentinel‑4 on MTG-S in 2019. So the question
is whether European remote sensing is moving forwards or backwards. The
time for process rather than the time to build an instrument or mission is
apparently now the dominant rate-determining step in developing a mission.
Europe needs to rejuvenate its intentions and establish space missions, aimed
at measuring atmospheric composition and addressing the key scientific issues
in the next phase of the Anthropocene.
An urgent issue is how to return and improve on the range of key
atmospheric measurements that were being made until 7 April 2012 by
European instrumentation on Envisat. Hopefully, European national bodies
will address the challenge and develop a system that provides both the data
and the scientific exploitation that are required to improve our knowledge
and build up the evidence base needed to assess environmental and climate
change, leading to the accurate attribution of change and measures to mitigate
appropriately the anthropogenic impact.

15. Summary and Conclusions

European and Chinese societies were the first to recognise that local air pollution
and later smog were related to human development and urbanisation. After
intense episodes of air pollution resulted in large negative epidemiological
impacts in cities both in winter and summer during the early postwar period,
public concerns led the governments of the UK, other European countries and
the United States to introduce legislation aimed at constraining and reducing air

131
SP-1326

pollution. The UNECE Convention on Long-Range Transboundary Air Pollution

was designed to manage and control the transport and transformation of
pollutants. In spite of this early warning of the potential global importance of
the emissions to the atmosphere by industry and society, European industry has
produced about 30% of the stratospheric ozone-depleting substances released
to the atmosphere. Europe has also modified agricultural systems through
its production and use of fertilisers and pesticides. It has undertaken land use
change on an unprecedented scale and is also one of the most important sources
of the greenhouse gases CO2, CH4 and N2O.
In response to the realisation of the importance of human impacts on the
environment and the world’s climate, the European scientific community and
the general public have recognised the need to understand, assess and manage
the human activities that are influencing the Earth system. As a first step in
this process, accurate, long-term measurements and global knowledge of
meteorological parameters, surface emissions and atmospheric composition
are required.
The SCIAMACHY team led by myself, working together with ESA, developed
the concept of GOME by descoping the successful SCIA-mini proposal in the
late 1980s. GOME made the first global contiguous measurements of the
backscattered and reflected solar radiation between 230  nm and 793  nm at
the top of the atmosphere from Sun-synchronous low-Earth orbit, in order
to retrieve the global amounts and distributions of key stratospheric and
tropospheric trace constituents in the mid-1990s.
Florence, the birthplace of the Renaissance, is where the GOME instrument
was designed in detail, utilising the instrument concepts from the science
team and the industrial Phase A studies, and then constructed by the Officine
Galileo and their sub-contractors. The GOME data products, developed by the
GOME scientists, have achieved one of the most important objectives of the
SCIAMACHY project, i.e. the initial demonstration and proof of concept that
the retrieval of accurate global measurements of the amounts and distributions
of key atmospheric constituents and pollutants from passive remotely sensed
measurements of backscattered and reflected solar radiation from space-based
instrumentation in low-Earth orbit is feasible. This has led to a unique record
of solar ultraviolet and visible output, and upper atmospheric and tropospheric
trace gas composition.
In orbit on ERS‑2, GOME performed flawlessly. Although the ERS-2 tape
recorders both failed in 2003 partial global coverage was still possible until
ESA began the decommissioning of ERS-2 in July 2011. GOME’s voyage
represents a milestone in the history of global passive remote sensing of
atmospheric composition. The close relationship between the scientists,
ESA and industry, and the short time taken to build the instrument, meant
that GOME was a relatively high-risk but in practice a low-cost instrument.
Approximately €25 million was required to construct the three GOME flight-
ready instruments. The scientific success of GOME and the SCIAMACHY project
is evident from a search of the ISI web of knowledge, which finds over 1200
scientific papers, with approximately 18 000 citations, yielding an h-index
of 57. The achievements of ERS and GOME may yet herald the arrival of a new
age of global enlightenment, where the sustainable management of the global
environment utilises objective evidence, provided in part by remote sensing
from space-based platforms.
GOME was originally selected for its ability to deliver total columns and
vertical profiles of O3, but made a pioneering contribution to improving our
knowledge of atmospheric chemistry and dynamics and quantifying air
pollution in both the stratosphere and the troposphere. Its data products have
provided early warning of environmental change produced by man, as well
as an evidence base for assessments of the effectiveness of the international
agreements such as UNECE’s Convention on Long-Range Transboundary Air
Pollution and the UN’s Vienna Convention for the Protection of the Ozone

132
 GOME and SCIAMACHY

Layer and its Montreal Protocol on Substances that Deplete the Ozone Layer.
The pictures of the amounts and distributions of the trace gases, aerosol and
cloud parameters resulting from GOME, SCIAMACHY and now GOME‑2 are
compelling. They document the changing atmospheric composition, which is
intrinsically fascinating, and demonstrate uniquely the increasing impact of
man during this early phase of the Anthropocene.
The success of GOME and SCIAMACHY and the value of their data products
have been internationally recognised. The importance of this breakthrough
in providing global, objective measurements of pollution and natural
biogeochemical cycling cannot be underestimated for both atmospheric
science and the development of global environmental policy. A review of
the development of modern tropospheric remote sensing has been provided
elsewhere (Burrows et al., 2011), and a book on SCIAMACHY is now in its
second edition (Gottwald & Bovensmann, 2011).
Turning to the European plans for the future, the picture is disappointing
and high risk. This is primarily because progress is too slow compared with the
growing need for accurate global information on atmospheric composition. The
Copernicus Sentinel‑4 will provide the first measurements of diurnal variations
in tropospheric trace gases over Europe from a geostationary orbit but not
simultaneously over the tropics or sub-Saharan Africa. It is interesting to note
that it took about six years from the proposal of SCIMACHY and SCIA-mini to
the launch of GOME on ERS‑2, whereas it will have taken ~21 years from the
proposal of GeoSCIA to the launch of Sentinel‑4. From 2020 onwards Sentinel‑5
will continue the early-morning measurement series currently being made by
GOME‑2, and will improve the spatial resolution and add targets CO and CH4,
like SCIAMACHY, but there are no plans to focus on CO2, according to ESA. In
addition, the Sentinel‑5 Precursor will make measurements similar to those of
OMI with the addition of a NIR/SWIR channel, but in the early afternoon, from
2017 onwards.
Sadly, as a result of the sudden and unexpected loss of contact with Envisat
on 8  April 2011, there will be no new data and therefore knowledge of many
important processes occurring in the atmosphere and at Earth’s surface.
Envisat had been in operation for twice its guaranteed lifetime, as defined by
the industrial consortium that built it. However, its instruments were working
excellently prior to failure, and in the case of SCIAMACHY no end-of-life issues
for components were foreseen prior to 2020. The lack of an overlap with or
plans for new European instruments to undertake nadir soundings of CO2,
the most important greenhouse gas, or limb and occultation observations of
the profiles of key atmospheric constituents from the upper troposphere to the
thermosphere shows a systemic problem in the development of the European
component of the required global measurement system.
This is not a new issue. The gaps in datasets and slow progress in
improving measurement spatial resolution and sampling is partly related to the
responsibilities and mandates given by governments to the various research,
space, meteorological and environmental agencies. There appears to be a
lack of joined-up thinking, which leads to insufficient funding to maintain
and deliver an adequate and integrated global observing system. The cost of
providing the research and space segment for the required global atmospheric
measurements is not trivial, but is, in my opinion, lower than that of not
obtaining the knowledge from this system. This is because it is not possible to
manage what is not being measured. I estimate that a 1 cent tax on a barrel of
oil, coal and gas equivalent in Europe (i.e. 1/10000 of the current cost) alone
would pay for much more than the global measurement systems required and
defined by expert groups.
The European EarthWatch concept, discussed in the 1990s, has not been
realised for atmospheric composition. After three pioneering decades, a golden
age, it appears that we are returning to a darker age when much less global
atmospheric information on atmospheric composition will be available for

133
SP-1326

scientists and policy makers. It appears that a global observing system that
builds on the pioneering achievements of the SCIAMACHY project, other
atmospheric composition measurements and in particular GOME on ERS‑2,
will remain a distant objective.
The creation of an adequate global observing system, associated ground
segment infrastructure, and the human capacity to run the system are needed
urgently to provide the information to assess human impacts on the evolution
of the Earth system during the next decades of the Anthropocene. The latter is
required to improve our understanding and the accuracy of prediction of the
behaviour of the Earth system and its responses in a warming, increasingly
urbanised world. This information would in turn provide the objective evidence
needed to develop further and monitor the effectiveness of international
environmental policy. The latter has the aim of achieving sustainable economic
development and optimised global environmental management, at a time
when the Earth’s population is heading rapidly towards and beyond 10 times
what it was at the outset of the industrial revolution.

Acknowledgements
This chapter has provided a subjective, partially historical review of key
aspects of the initiation and the development of GOME and SCIAMACHY, as
well as some of the scientific results obtained from GOME. It is not intended to
be a comprehensive objective analysis, but my personal view of the evolution
of GOME and European remote sensing of atmospheric composition.
Many individuals have contributed to the success of GOME and the
SCIAMACHY project that I have not had space or time to list in detail. I therefore
wish to acknowledge explicitly all the scientists, engineers and administrators
who contributed to the success of these projects and to the ERS and Envisat
programmes. These have been remarkable European civil achievements.
There are many fathers and mothers of success, and without the funding
agencies, who invested in this project it would not have happened. ESA, the
German Ministry of Science and Education (BMBF), the German Aerospace
Center (DLR), and the Dutch, Belgian and Italian space agencies are to be
thanked, as well as NASA (in particular E. Hilsenrath, J. Gleason and S. Janz)
for their collaborative support. My friends and scientific mentors Paul Crutzen,
Dieter Perner (1934–2012), Wolfgang Schneider, Geert Moortgat at the Max
Planck Institute for Chemistry in Mainz and Richard Wayne of the University of
Oxford were a great strength and support, particularly during the development
of the scientific objectives of SCIAMACHY and GOME. I would also like to thank
Ulrich Platt and his team at the University of Heidelberg for their many and
continuing contributions. Phil Goldsmith, Chris Readings, Guy Duschossois,
Peter Dubock, Peter Edwards, Achim Hahne, Alain Lefebvre, Jörg Calleis and
Reinhold Zöbl of ESA also made important contributions, without which GOME
would not have been possible. The teams at the then Officine Galileo have
had the pleasure of seeing their instrument becoming a world-beater. EADS
Astrium was responsible for key aspects of the ERS and Envisat platforms and
also the prime contractor for SCIAMACHY. I thank them all and many others
not mentioned by name for their outstanding work.
I am indebted to the State and University of Bremen for in part funding
my research. I thank the faculty of Physics for hosting our Institutes of
Environmental Physics and Remote Sensing. In particular I wish to thank the
members of the research team at the Institute of Environmental Physics of
the University of Bremen, IUP-UB, who have supported my efforts and made
many important and original contributions of their own to various aspects of
the GOME and SCIAMACHY projects. In particular, I thank my close colleagues
of many years at IUP-UB – Heinrich Bovensmann, Klaus Bramstedt, Michael
Buchwitz, Kai-Uwe Eichmann, Konstantin Gerilowski, Klaus Künzi, Alexander

134
 GOME and SCIAMACHY

Kokhanovsky, Annette Ladstätter-Weißenmayer, Stefan Noel, Justus Notholt,

Andreas Richter, Vladimir Rozanov, Wolfgang von Hoyningen Huene, Christian
von Savigny, Heiko Schellhorn, Heiko Schröter, Anja Schönhardt, Marco
Vountas, Folkard Wittrock, and Mark Weber − as well as my administrative
staff and the many postdoctoral and graduate students who passed through my
department at IUP-UB.
Finally, I would like to thank my wife and family for their patience and
support, which enabled me to invest fruitfully so much time and effort in
the development and evolution of GOME, SCIAMACHY, GOME‑2, GeoSCIA/
Sentinel‑4, Sentinel‑5, CarbonSat and more recently SCIA-ISS , not to mention
my studies of kinetics and spectroscopy.

References
Arrhenius, S. (1896). On the influence of carbonic acid in the air upon the
temperature of the ground. Phil. Mag. and J. Sci. 41, 237.
Baker, R.L., Maudlin, L.E. III & Russell, J.M. III (1986). Design and performance of
the Halogen Occultation Experiment (HALOE) remote sensor. Proc. SPIE 685,
181−191.
Barnett, J.J., Morris, P.E., Nightingale, T.J., Palmer, C.W.P., Peskett, G.D., Rodgers,
C.D., Taylor, F.W., Venters, P., Wells, R.J., Whitney, J.G., Ballard, J. & Knight,
J. (1992). The improved stratospheric and mesospheric sounder on the Upper
Atmosphere Research Satellite. Proc. SPIE 1715, 527.
Barrie, L.A., Bottenheim, J.W., Snell, R.C., Crutzen, P.J. & Rasmussen, R.A. (1988).
Ozone destruction and photochemical reactions at polar sunrise in the lower
Arctic atmosphere. Nature 334, 138−140.
Bates, D.R. & Nicolet, M. (1950). The photochemistry of atmospheric water vapour,
J. Geophys. Res. 55(3), 301–327. doi: 10.1029/JZ055i003p00301
Bertram, T.H., Heckel, A., Richter, A., Burrows, J.P. & Cohen, R.C. (2005). Satellite
measurements of daily variations in soil NOx emissions, Geophys. Res. Lett. 32
(24), L24812
Bhartia, P.K., McPeters, R.D., Mateer, C.L., Flynn, L.E. & Wellemeyer, C. (1996).
Algorithm for the estimation of vertical ozone profiles from the backscattered
ultraviolet technique. J. Geophys. Res. 101, 18793−18806.
Brasunas, J.C., Kunde, V.G. & Herath, L.W. (1988). Cryogenic Fourier spectrometer
for measuring trace species in the lower stratosphere. Appl. Opt. 27, 4964−4976.
Bovensmann, H., Burrows, J.P., Buchwitz, M., Frerick, J., Noël, S., Rozanov, V.V.,
Chance, K.V. & Goede, A.P.H. (1999). SCIAMACHY: Mission objectives and
measurement modes. Conf. on Global Measurement Systems for Atmospheric
Composition, Toronto, Canada. J. Atmos. Sci. 56(2), 127−150.
Buchwitz M., Rozanov V.V., Eichmann K.-U., de Beek, R. & Burrows, J.P. (2001).
SCIATRAN: A new atmospheric radiative transfer model for the ultraviolet,
visible, and near-infrared spectral regions. In: W.L. Smith & Y.M. Timofeyev (Eds),
IRS 2000, Current Problems in Atmospheric Radiation, Studies in Geophysical
Optics and Remote Sensing. A. Deepak Publishing, Hampton, VA, pp.365−368.
Burrows, J.P., Schneider, W., Geary, J.C., Chance, K.V., Goede, A.P.H., Aarts, H.J.M.,
de Vries, J., Smorenburg, C. & Visser, H. (1990). Atmospheric remote sensing
with SCIAMACHY. Dig. Mtg on Optical Remote Sensing of the Atmosphere No. 4.
Optical Society of America, Washington, DC, pp.71−75.
Burrows, K., Chance, V., Goede, A.P.H., Guzzi, R., Kerridge, B.J., Muller, C., Perner,
D., Platt, U., Pommereau, J.-P., Schneider, W., Spurr, R.J. & van der Woerd,
H. (1993). Global Ozone Monitoring Experiment: Interim Science Report, T.D.
Guyenne & C.J. Readings (Eds). ESA SP-1151, European Space Agency, Paris.
Burrows, J.P., Diebel, D., Kerridge, B.J., Munro, R., Pemberton, D., Platt, U. & Frank,
H. (1994). A Study of the Methods for the Retrieval of Atmospheric Constituents,
Final Report, ESA Contract 9687/91/NL/BI.

135
SP-1326

Burrows J.P., Hölzle, E., Goede, A.P.H., Visser, H. & Fricke, W. (1995). SCIAMACHY:
Scanning Imaging Absorption Spectrometer for Atmospheric Chartography. Acta
Astronaut. 35(7), 445−451.
Burrows, J.P., Weber, M., Buchwitz, M., Rozanov, V.V., Ladstätter, A., Weißenmayer,
A., Richter, A., DeBeek, R., Hoogen, R., Bramstedt, K. & Eichmann, K.U. (1999).
The Global Ozone Monitoring Experiment (GOME): Mission concept and first
scientific results, Conf. on Global Measurement Systems for Atmospheric
Composition, Toronto, Canada. J. Atmos. Sci. 56 (2), 151−175.
Burrows J.P., Borrel, P. & Platt, U. (Eds) and authors (2011). The Remote Sensing of
Tropospheric Composition from Space, Springer, Heidelberg. doi: 10.1007/978-3-
642-14791-3
Carli, B., Mencaraglia, F. & Bonetti, A. (1984). Submillimeter high resolution FT
spectrometer for atmospheric studies. Appl. Opt. 23, 2594−2603.
Chance, K.V., Johnson, D.G. & Traub, W.A. (1989). Measurement of stratospheric
HOCl: Concentration profiles, including diurnal variation. J. Geophys. Res. 94,
11059–11069.
Chu, W. & McCormick, M.P. (1979). Inversion of stratospheric aerosol and gaseous
constituents from spacecraft solar extinction data in the 0.38−1.0 micron region.
Appl. Opt. 18, 1404−1414.
Coffey, M.T., Mankin, W.G. & Golden, A. (1981). Simultaneous spectroscopic
determination of the latitudinal, seasonal, and diurnal variability of N2O, NO,
NO2 and HNO3. J. Geophys. Res. 86, 7331−7341.
Connors, V.S., Cahoon, D.R., Reichle, H.G. Jr. & Scheel, H.E. (1991). Comparison
between carbon monoxide measurements from spaceborne and airborne
platforms. Can. J. Phys. 69, 1128−1137.
Crewell, S., Künzi, K., Nett, H., Wehr, T. & Hartogh, P. (1993). Aircraft measurements
of ClO and HCl during EASOE 1991/92. Geophys. Res. Lett. 21, 1267−1270.
Crutzen P.J. (1970). Influence of the oxides of nitrogen on atmospheric ozone
content. Q. J. R. Meteorol. Soc. 96(408), 320−325. doi: 10.1002/qj.49709640815
Crutzen, P.J. (1973). A discussion of the chemistry of some minor constituents in
the stratosphere and the troposphere. Pure Appl. Geophys. 106−108, 1385−1399.
Dave, J.V. & Mateer, C.L. (1967). A preliminary study of the possibility of estimating
total atmospheric ozone from satellite measurements. J. Atmos. Sci. 24, 414−427.
Deschamps, P.Y., Herman, M., Podaire, A., Leroy, M., Laporte, M. & Vermande, P.
(1990). A spatial instrument for the observation of polarization and directionality
of Earth reflectances. Proc. IGARSS 90, IEEE, Washington, DC.
DeZafra, R.L., Jaramillo, M., Parrish, A., Solomon, P., Connor, R. & Barrett, J. (1987).
High concentrations of chlorine monoxide at low altitudes in the Antarctic spring
stratosphere: diurnal variation. Nature 328, 408−411.
Eisinger, M. & Burrows, J.P. (1998a). Observations of SO2 from volcanic plumes.
Earth Obs. Q. 58, 16−18.
Eisinger, M. & Burrows, J.P. (1998b). Tropospheric sulphur dioxide observed by the
ERS‑2 GOME Instrument, Geophys. Res. Lett. 25, 4177−4180.
ESA (1991) Report of the Earth Observation User Consultation Meeting, Barron, C. &
Battrick, B. (eds), May 1991. ESA SP-1143, European Space Agency, Paris.
Farman, J.C., Gardner B.G. & Franklin, J.D. (1985). Large losses of total ozone in
Antarctica reveal seasonal ClOx and NOx interaction. Nature 315, 207−210.
Farmer, C.B., Raper O.F. & O’Callaghan, F.G. (1987). Final report on the first flight
of the ATMOS instrument during the Spacelab 3 mission, Publication 87−32, Jet
Propulsion Laboratory, Pasedena, California.
Farquhar, J., Bao, H. & Thiemens, M. (2000). Atmospheric influence of Earth’s earliest
sulfur cycle. Science 289(5480), 756−758. doi: 10.1126/science.289.5480.756.
Finlayson-Pitts, B.J. & Pitts, J.N. Jr. (1986). Atmospheric Chemistry: Fundamentals
and Experimental Techniques. Wiley, Chichester, UK.
Fishman, J. and Crutzen, P.J. (1978). Origin of ozone in the troposphere. Nature
274(5674), 855−858. doi: 10.1038/274855a0
Fishman, J., Watson, C.E., Larsen, J.C. & Logan, J.A. (1990). Distribution of
tropospheric ozone determined from satellite data. J. Geophys. Res. 95, 3599−3617.

136
 GOME and SCIAMACHY

Fishman, J., Fakhruzzaman, K., Cros, B. & Nganga, D. (1991). Identification

of widespread pollution in the southern hemisphere deduced from satellite
analyses. Science 252, 1693−1696.
Fishman, J., Brackett, V.G. & Fakhruzzaman, K. (1992). Distribution of tropospheric
ozone in the tropics from satellite and ozonsonde measurements. J. Atmos. Terr.
Phys. 54, 589−597.
Fishman, J., Brackett, V.G., Browell, E.V. & Grant, W.B. (1996). Tropospheric ozone
derived from TOMS/SBUV measurements during TRACA-A. J. Geophys. Res. 101,
24069−24082.
Fischer, H., Fergg, F. & Rabus, D. (1980). Radiometric measurements of stratospheric
H2O, HNO3 and NO2 profiles. Proc. Int. Radiation Symp. Fort Collins, CO, August
1980.
Franke, K., Richter, A., Bovensmann, H., Eyring, V, Jockel, P., Hoor, P. & Burrows,
J.P. (2008/ 2009). Ship emitted NO2 in the Indian Ocean: Comparison of model
results with satellite data. Atmos. Chem. Phys. Disc. 8, 15997−16025; and 9(19),
7289−7301.
Gille, J.C., Bailey, P.L., Craig, R.A., House, F.B. & Anderson, G.P. (1980). Sounding
the stratosphere and mesosphere by infrared limb scanning from space. Science
208, 397.
Gille, J.C. & Russell, J.M. III (1984). The limb Infrared Monitor of the stratosphere:
Experimental description, performance and results. J. Geophys. Res. 89, 5125–
5140.
Gottwald, M. & Bovensmann, B. (eds) (2011). SCIAMACHY: Exploring the Changing
Earth’s Atmosphere. Springer Science+Business Media, Dordrecht, the Netherlands.
doi: 10.1007/978-90-481-9896-2
Groves, K.S., Mattingly, S.R. & Tuck, A.F. (1978). Increased carbon dioxide and
stratospheric ozone. Nature 273 (5665), 711−715. doi: 10.1038/273711a0
Groves, K.S. & Tuck, A.F. (1979). Simultaneous effects of CO2 and chlorofluoromethanes
on stratospheric ozone. Nature 280(5718), 127−129. doi: 10.1038/280127a0
Gunson, M.R., Abbas, M.M., Abrams, M.C., Allen, M., Brown, L.R., Brown, T.L.,
Chang, A.Y., Goldman, A., Irion, F.W., Lowes, L.L., Mahieu, E., Manney, G.L.,
Michelsen, H.A., Newchurch, M.J., Rinsland, C.P., Salawitch, R.J., Stiller, G.P.,
Toon, G.C., Yung, Y.L. & Zander, R. (1996). The Atmospheric Trace Molecule
Spectroscopy (ATMOS) experiment: Deployment on the ATLAS space shuttle
missions. Geophys. Res. Lett. 23, 2333−2336.
Guzzi R., Cattani, E., Cervino, M., Levoni, C., Torricella, F., Burrows, J.P. & Kurosu,
T. (1998). GOME Cloud and Aerosol Data Products Algorithms Development. Final
Report, ESA Contract 11572/95/NL/CN.
Guzzi, R., Cervino, M., Torricella, F., Burrows, J.P., Kurosu, T., Rozanov, V., Chance,
K. & Watts, P. (1996). A Study of Cloud Detection. Final Report, ESA Contract
10997/94/NL/CN.
Haagen-Smit, A.J. (1952). Chemistry and physiology of Los Angeles smog. Indust.
Eng. Chem. 44, 1342.
Haagen-Smit, A.J., Darley, E.F., Zailin, M., Hull, H. & Noble, W. (1952). Investigation
on injury to plants from air pollution in the Los Angeles area. Plant Physiol. 27, 18.
Hartmann, G.K., Bevilacqua, R.M., Schwartz, P.R., Kämpfer, N., Künzi, K.F., Berg,
A., Boogaerts, W., Connor, B.J., Croskey, C.L., Daehler, M., Degenhardt, W.,
Dicken, H.D., Goldizen, D., Kriebel, D., Langen, J., Loidl, A., Olivero, J.J., Pauls,
T.A., Puliafito, S.E., Richards, M.L., Rudin, C., Tsou, J.J., Waltmann, W.B.,
Umlauft, G. & Zwick, R. (1996). Measurements of O3, H2O, and ClO in the middle
atmosphere using the millimeter wave atmospheric sounder (MAS). Geophys.
Res. Lett. 23, 2313−2317.
Heath, D.F., Mateer, C.L. & Kruger, A.J. (1973). The Nimbus‑4 Backscatter Ultraviolet
(BUV) atmospheric ozone experiment – two years’ operation. Pure Appl.
Geophys. 106−108, 1238−1253.
Heath, D.F., Krueger, A.J., Roeder, H.A. & Henderson, B.D. (1975). The solar
backscatter ultraviolet and total ozone mapping spectrometer (SBUV/TOMS) for
Nimbus‑G. Opt. Eng. 14, 323−332.

137
SP-1326

Hilsenrath, E., Cebula, R.P., Deland, M.T., Laamann, K., Taylor, S., Wellemeyer, C.
and Bhartia, P.K. (1995). Calibration of the NOAA-11 solar backscatter ultraviolet
(SBUV/2) ozone data set from 1989 to 1993 using in-flight calibration data and
SSBUV. J. Geophys. Res. 100, 1351−1366.
Hilsenrath, E., Newman, P.A., Cebula, R.B., DeCamp, P.W., Kelly, T.J. & Coy, L.
(1996). Ozone change from 1992 to 1993 as observed from SSBUV on the ATLAS-1
and ATLAS-2 missions. Geophys. Res. Lett. 23, 2305−2308.
Hilsenrath, E., Williams, D. & Frederick, J. (1988). Calibration of long term data sets
from operational satellites using the space shuttle. Proc. SPIE 924, 215−222.
Holloway, A.M & Wayne, R.P. (2010). Atmospheric Chemistry. Royal Society of
Chemistry, London.
Houghton, J.T., Meira-Filho, L.G., Bruce, J., Hoesung Lee, Callander, B.A., Haites,
E., Harris, N. & Maskell, K. (Eds.) (1995). Climate Change 1994: Radiative Forcing
of Climate Change and an Evaluation of the IPCC IS92 Emission Scenarios.
Cambridge University Press, Cambridge, UK.
Hoogen R., Rozanov, V., Bramstedt, K., Eichmann, K.-U., Weber, M., de Beek, R.,
Buchwitz, M. & Burrows, J.P. (1998). Height resolved ozone information from
GOME data. Earth Obs. Q. 58, 9−10.
Hoogen, R., Rozanov, V.V., Bramstedt, K., Eichmann, K.-U., Weber, M. & Burrows,
J.P. (1999a). O3 profiles from GOME satellite data, I: Comparison with ozonesonde
measurements. Phys. Chem. Earth C – Solar, Terrestial and Planetary Science
24(5), 447−452.
Hoogen, R., Rozanov, V.V. & Burrows, J.P. (1999b). Ozone profiles from GOME
satellite data: algorithm description and first validation. J. Geophys. Res. −
Atmospheres 104 (D7), 8263−8280.
Hudson, R.D. (2012). Measurements of the movement of the jet streams at mid-
latitudes, in the northern and southern hemispheres, 1979 to 2010. Atmos. Phys.
Chem. 12(16), 7797−7808. doi: 10.5194/acp-12-7797-2012.
Hudson, R.D., Kim, J. & Thompson, A.M. (1995). On the derivation of tropospheric
column ozone from radiance measured by the total ozone mapping spectrometer.
J. Geophys. Res. 100, 11138−11145.
Hudson, R.D. & Thompson, A.M. (1998). Tropical ozone from total ozone mapping
spectrometer by a modified residual method. J. Geophys. Res. 103, 22129−22146.
Hsu, N.C., Herman, J.R., Bhartia, P.K., Seftor, C.J., Torres, O., Thompson, A.M.,
Gleason, J.F., Eck, T.F. & Holben, B.N. (1996). Detection of biomass burning
smoke from TOMSmeasurements. Geophys. Res. Lett. 23, 745−748.
Ignatov, A.M., Stiwe, L.L., Sakerin, S.M. & Korotaev, G.K. (1995). Validation of
the NOAA/NESDIS satellite aerosol product over the North Atlantic in 1989.
J. Geophys. Res. 100, 5123−5132.
Iozenas, V.A. et al., (1969a). Studies of the Earth’s ozonosphere from satellites. Izv.
Akad. Nauk SSSR Fiz. Atmos. Okeana 5, 77−82.
Iozenas, V.A. et al., (1969b). An investigation of the planetary ozone distribution
from satellite measurements of ultraviolet spectra. Izv. Akad. Nauk SSSR Fiz.
Atmos. Okeana 5, 219−233.
Johnston, H.S. (1971). Reduction of stratospheric ozone by nitrogen oxide catalysts
from supersonic transport exhaust. Science 173(3996), 517–521. doi: 10.1126/
science.173.3996.517
Johnston, H.S. (1974). Photochemistry in the stratosphere with applications to
supersonic transport. Acta Astronaut. 1 (1−2), 135−156. doi: 10.1016/0094-
5765(74)90013-7.
Kaleschke, L., Richter, A., Burrows, J.P., Afe, O., Heygster, G., Notholt, J., Rankin,
A.M, Roscoe, H.K., Hollwedel, J., Wagner, T. & Jacobi, H.W. (2004). Frost flowers
on sea ice as a source of sea salt and their influence on tropospheric halogen
chemistry. Geophys. Res. Lett. 31(16), L16114.
Kim, J.-H., Hudson, R.D. & Thompson, A.M. (1996). A new method for deriving
time averaged tropospheric column ozone over the tropics using total ozone
monitoring spectrometer (TOMS). radiances: Intercomparison and analysis
using TRACE A data. J. Geophys. Res. 101, 24317−24330.

138
 GOME and SCIAMACHY

Krasnopol’skiy, V.A. (1966). The ultraviolet spectrum of solar radiation reflected

by the terrestrial atmosphere and its use in determining the total content and
vertical distribution of atmospheric ozone. Russ. Geomagn. Aeron. 6, 236−242.
Krueger, A.J. (1983). Sighting of El Chichon sulfur dioxide with the Nimbus 7 total
ozone mapping spectrometer. Science 220, 1377−1378.
Krueger, A.J., Walter, L.S., Bhartia, P.K., Schnetzler, C.C., Krotkov, N.A., Sprod, I. &
Bluth, G.J.S. (1995). Volcanic sulfur dioxide measurements from the total ozone
mapping spectrometer (TOMS) instruments. J. Geophys. Res. 100, 14057−14076.
Krueger, A.J., Schnetzler, C.C. & Walter, L.S. (1996). The December 1981 eruption of
Nyamuragira Volcano (Zaire) and the origin of the ‘mystery cloud’ of early 1982.
J. Geophys. Res. 101, 15 191−15 196.
Kunde, V.G., Brasunas, J.C., Maguire, W.C., Herman, J.R., Massie, S.T., Abbas, M.M.,
Herath, L. & Shaffer, W.A. (1988). Measurement of nighttime stratospheric N2O5
from infrared emission spectra. Geophys. Res. Lett.15, 1177−1180.
Ladstätter-Weißenmayer, A. & Burrows, J.P. (1998). Biomass burning over Indonesia,
Earth Obs. Quarterly 58, 28−30.
Lovelock, J.E. & Maggs, R.J. (1972). Halogenated hydrocarbons in and over the Atlantic.
Nature 241(5386), 194−196. doi: 10.1038/241194a0
Mateer, C.L., Heath, D.F. & Kruger, A.J. (1971). Estimation of total ozone from satellite
measurements of backscattered Earth radiance. J. Atmos. Sci. 24, 414−420.
Maudlin, L.E. III, Zaun, N.H., McCormick M.P. Jr., Guy, J.H. & Vaughn, W.R.
(1985). Stratospheric aerosol and gas experiment II instrument: a functional
description. Opt. Eng. 24, 307−312.
Meinel, A.B. (1950). OH emission bands in the spectrum of the night sky 1950,
Astrophys. J. 112(1), 120–130. doi: 10.1086/145321
McCormick, M.P., Hamill, P., Pepin, T.J., Chu, W.P., Swissler, T.J. & McMaster, L.R.
(1979). Satellite studies of the stratospheric aerosol. Bull. Am. Meteorol. Soc. 60,
1038−1048.
Molina, M.J. & Rowland, F.S. (1974). Stratospheric sink for chlorofluoromethanes:
chlorine atom catalyzed destruction of ozone. Nature 249, 810−814.
Mount, G.M., Rusch, D.W., Noxon, J.F., Zawodny, J.M. & Barth, C. (1984). Measurement
of stratospheric NO2 from the solar mesospheric explorer satellite, 1. An overview
of the results. J. Geophys. Res. 89, 1327−1339.
Mount, G.M., Sanders, R.W., Schmetekopf, A.L. & Solomon, S. (1987). Visible and
near ultraviolet spectroscopy at McMurdo station Antarctica 1: Overview and
daily variations of NO2 and O3 during austral spring 1986. J. Geophys. Res. 92,
8320−8340.
Murcray, D., Barker, D.B., Brooks, J.N., Goldman, A. & Williams, G.W.J. (1975).
Seasonal and latitudinal variation of the stratospheric concentration of HNO3.
Geophys. Res. Lett. 2, 223−225.
Murcray, D.G., Goldman, A., Murcray, F.H., Murcray, F.J. & Williams, W.J. (1979).
Stratospheric distribution of ClONO2. Geophys. Res. Lett. 6, 857−859.
Platt, U. & Perner, D. (1980). Direct measurements of atmospheric HCHO, HONO,
O3, NO2, and SO2, by differential optical absorption spectroscopy in the near UV.
J. Geophys. Res. 85, 7453−7465.
Ravishankara, A.R., Daniel, J.S. & Portmann, R.W. (2009). Nitrous oxide (N2O):
The dominant ozone-depleting substance emitted in the 21st century, Science
326(5949), 123−125.
Rawcliff, R.D. & Elliot, D.D. (1966). Latitude distribution of ozone at high altitudes
deduced from satellite measurements of Earth’s radiance at 2840 Angstrom.
J. Geophys. Res. 71(21), 5077−5090.
Reichle, H.G. Jr., Connors, V.S., Holland, J.A., Hypes, W.D., Wallio, H.A., Casas,
J.C., Gormsen, B.B., Saylor, M.S. & Hesketh, W.D. (1986). Middle and upper
tropospheric carbon monoxide mixing ratios as measured by a satellite-borne
remote sensor during November 1981. J. Geophys. Res. 91, 10 865−10 887.

139
SP-1326

Reichle, H.G. Jr., Connors, V.S., Holland, J.A., Sherrill, R.T., Wallio, H.A., Casas,
J.C., Condon, E.P., Gormsen, B.B. & Seiler, W. (1990). The distribution of middle
tropospheric carbon monoxide during early October 1984. J. Geophys. Res. 95,
9845−9856.
Richter, A., Wittrock, F., Eisinger, M. & Burrows, J.P. (1998). GOME observations of
tropospheric BrO in the northern hemispheric spring and summer 1997. Geophys.
Res. Lett. 25, 2683−2686.
Richter, A., Eyring, E., Burrows, J.P., Bovensmann, H., Lauer, A., Sierk, B. &
Crutzen, P.J. (2004). Satellite measurements of NO2 from international shipping
emissions. Geophys. Res. Lett. 31 (23), L23110.
Richter, A., Burrows, J.P., Nüß, H., Granier, C. & Niemeier, U. (2005). Increase in
tropospheric nitrogen dioxide over China observed from space. Nature 437,
129−132.
Roche, A.E. & Kumer, J.B. (1989). Cryogenic Limb Array Etalon Spectrometer
(CLAES). Proc. SPIE 973, 324−335.
Roche, A.E., Forney, P.B., Kumer, J.B., Naes, L.G. & Nast, T.C. (1982). Performance
analysis for the cryogenic etalon spectrometer on the upper atmospheric
research satellite. Proc. SPIE 354, 46−54.
Rozanov, V., Diebel, D., Spurr, R.J.D. & Burrows, J.P. (1997). GOMETRAN: A
Radiative Transfer Model for the Satellite Project GOME: I Plane-Parallel Version,
J. Geophys. Res. − Atmospheres 102 (D14), 16683−16695 .
Rozanov, V.V., Kurosu, T. & Burrows, J.P. (1998). Retrieval of atmospheric constituents
in the UV-visible: A new analytical approach for the calculation of weighting
functions. J. Quant. Spectrosc. Radiat. Trans. 60(2), 277−299.
Rusch, D.W., Mount, G.H., Barth, C.A., Thomas, R.J. & Callan, M.T. (1984). Solar
mesospheric explorer ultraviolet spectrometer: measurements of ozone in the
1.0−0.1 mbar region. J. Geophys. Res. 84, 11677−11687.
Sander. R., Burrows, J.P. & Kaleschke, L. (2006). Carbonate precipitation in brine – a
potential trigger for tropospheric ozone depletion events. Atmos. Phys. Chem. 6,
4653−4658.
Schönhardt, A., Richter, A., Wittrock, F., Kirk, H. & Burrows, J.P. (2007/2008).
Observations of iodine monoxide (IO) columns from satellite. Atmos. Chem. Phys.
Discussions 712959−12999 (2007); Atmos. Chem. Phys. 8, 637−653 (2008).
Simpson, W.R., von Glasow, R., Riedel, K., Anderson, P., Ariya, P., Bottenheim,
J., Burrows, J.P., Carpenter, L., Frieß, U., Goodsite, M.E., Heard, D., Hutterli,
M., Jacobi, H.W., Kaleschke, L., Neff, B., Plane, J., Platt, U., Richter, A.,
Roscoe, H., Sander, R., Singer, S.F. & Wentworth, R.C. (2007). A method for the
determination of the vertical ozone distribution of ozone distribution from a
satellite. J. Geophys. Res. 62, 299−308.
Singer, S.F. & Wentworth, R.C. (1957). A method for the determination of the vertical
ozone distribution from a satellite. J. Geophys. Res. 62(2) 299–308. doi: 10.1029/
JZ062i002p00299
Stachnik, R.A., Hardy, J.C., Torsala, J.A., Waters, J.W. & Erickson, N.R. (1992). Sub-
millimeter heterodyne measurements of stratospheric ClO, HCl, O3 and HCl: First
results. Geophys. Res. Lett. 19, 1931−1934.
Stolarski, R.S. and Cicerone, R.J. (1974). Stratospheric chlorine – a possible sink for
ozone? Can. J. Chem. 52(8), 1610–1615. doi: 10.1139/v74-233.
Thomas, R.J., Barth, C.A., Rusch, D.W. & Sanders, R.W. (1984). Solar mesospheric
near-infrared spectrometer: measurements of 1.27 mm radiances and inference
for mesospheric ozone. J. Geophys. Res. 89, 9569−9580.
Thomas, W., Hegels, E., Slijkhuis, S., Spurr, R. & Chance, K. (1998). Detection of
biomass burning products in South East Asia from backscatter data taken by the
GOME spectrometer. Geophys. Res. Lett. 25, 1317−1320.
Titz, R., Birk, M. Hausamann, D., Nitsche, R., Schrier, F., Urban, J., Küllmann, H. &
Röser, H.P. (1995). Observation of stratospheric OH at 2.5 THz with an airborne
heterodyne system. Infrared Phys.Technol. 36, 883−891.

140
 GOME and SCIAMACHY

UARS (1987). Upper Atmosphere Research Satellite. Project Data Book April 1987.
Prepared by General Electric Astro-Space Division, Valley Forge, for the NASA
Goddard Space Flight Center.
Waters, J.W., Hardy, J.C., Jarnot, R.F. & Pickett, H.M. (1981). Chlorine monoxide
radical, ozone and hydrogen peroxide by microwave limb sounding. Science
214, 61−64.
Waters, J.W., Hardy, J.C., Jarnot, R.F., Pickett, H.M. & Zimmermann, P. (1984).
A balloon borne microwave limb sounder for stratospheric measurements.
J. Quant. Spectrosc. Radiat. Trans. 32, 407−433.
Waters, J.W., Froideveaux, L., Read, W.G., Manney, G.L., Elson, L.S., Flower, D.A.,
Jarnot R.F., & Harwood, R.S. (1993). Stratospheric ClO and ozone from the
microwave limb sounder on the Upper Atmospheric Research Satellite. Nature
362, 597−602.
Wayne, R.P. (1992). Chemistry of Atmospheres. Clarendon Press, Oxford, UK.
Wittrock, F., Richter, A., Oetjen, H., Burrows, J.P., Kanakidou, M., Myriokefalitakis,
S., Volkamer, R., Beirle, S., Platt, U. & Wagner, T. (2006). Simultaneous global
observations of glyoxal and formaldehyde from space. Geophys. Res. Lett. 33,
L16804. doi: 10.1029/2006GL026310
WMO (2010). Scientific Assessment of Ozone Depletion: 2010. Report No. 52,
World Meteorological Organization Global Ozone Research and Monitoring
Project. WMO, NOAA, NASA, UNEP, WMO and the European Commission.
http://ozone.unep.org/Assessment_Panels/SAP/Scientific_Assessment_2010/
Wofsey, S.C., McElroy, M.B. & Yung, Y.L (1975). Chemistry of atmospheric bromine.
Geophys. Res. Lett. 2(6), 215–218. doi: 10.1029/GL002i006p00215 
Zander, R. (1981). Recent observations of HF and HCl in the upper stratosphere.
Geophys. Res. Lett. 8, 413−416.
Ziemke, J.R., Chandra, S. & Bhartia, P.K. (1998). Two new methods for deriving
tropospheric column ozone from TOMS measurements: Assimilated UARS MLS/
HALOE and convective-cloud differential techniques. J. Geophys. Res. 103(D17),
22 115–22 127.

141
→→ The ERS Scatterometer:
Achievements and the Future
 Scatterometer

The ERS Scatterometer: Achievements and the

Future
A. Stoffelen1, W. Wagner2

1 KNMI Royal Netherlands Meteorological Institute, 3732 GK De Bilt, the Netherlands

2 Vienna University of Technology, 1040 Vienna, Austria

1. Introduction
The first Active Microwave Instrument was launched on the European Remote
Sensing satellite, ERS‑1, in July 1991. When operated in Wind Scatterometer
mode, the instrument acquired measurements of sea surface wind speed and
direction. ERS Scatterometer (ESCAT) data are now widely used for hurricane
prediction, Numerical Weather Prediction, marine ‘nowcasting’, oceanography,
hydrology and studies of the cryosphere, with societal and economic benefits
in areas such as marine safety, offshore activities, ship routing, wind energy
and climate change monitoring.
The ESCAT research mission was well prepared with aircraft campaigns,
including the setup of the RENE campaign for Calibration and Validation
(Cal/Val), the development of a Geophysical Model Function (GMF) and initial
preparations for user applications, such as the near-realtime availability of
the scatterometer winds. Nevertheless, initial comparisons of the ESCAT
winds with other wind information showed large inconsistencies. This led
to the development of novel analytical techniques using the measurement
space of ESCAT, which confirmed the expected internal consistency of the
ESCAT measurements and the low noise, but also provided an improved
characterisation of the sensitivity to wind speed and direction, resulting
in improved GMFs. Moreover, the visualisation of the ESCAT measurement
space provided insights into the nonlinear wind retrieval problem, which has
subsequently been improved, aided the discrimination of sea ice and water,
and provided important insights into wind quality control.
The ESCAT wind Cal/Val also led to the development of the triple
collocation methodology, which is now widely used in satellite algorithm
development. The lessons learnt from ESCAT were readily applied to the
NASA scatterometer missions at the Ku-band and provided similar benefits,
leading to improved rain screening and wind datasets. ESCAT has also been
successfully applied in soil moisture, sea ice and snow characterisation,
leading to operational applications. ESCAT’s heritage has also been important
in the development of the Advanced Scatterometer (ASCAT) on MetOp for the
Eumetsat Polar System (EPS). The EPS Second Generation ASCAT (ASCAT-SG)
is now being designed, where the ESCAT C‑band static fan beam concept is
being maintained due to its great success. This may well lead to over 40 years
of C-band fan beam scatterometer data, which will be a tremendous resource
for climate applications for studying wind climatologies, air–sea interactions,
atmospheric processes, etc. The demand for the unique ESCAT data and
services will therefore remain high in the years to come.

2. The ERS Scatterometer Instrument

Space-borne scatterometers provide unique global ocean surface vector wind
products at high spatial resolution. Since they operate at microwave (radar)
frequencies, these instruments are not hindered by cloud cover and hence are
able to reveal phenomena such as polar front disturbances and tropical cyclone
winds. Figure 1 shows mesoscale ESCAT winds (red) detected by the ESCAT

145
SP-1326

Figure 1. Scatterometer winds from ERS‑2

at around 12:00 GMT on 24 November
1999 at 70°N and 0°W (www.knmi.nl/
scatterometer). The grey shading is a
Meteosat IR image coherent with the
scatterometer winds. Blue mask: areas
where the sea surface temperature is below
zero and sea ice is probable. Grey mask:
land at 80°N and 20°W (top right). The red
contours indicate surface pressure from
the HIRLAM model at KNMI (3-h forecast).
Blue and purple are the spatially smooth
wind vectors from HIRLAM (the amount of
purple increases with wind speed). The red
wind vectors depict the spatially detailed
ERS‑2 winds. The red dots indicate where
winds were rejected because of a confused
sea state (bottom left) or the presence of
sea ice (top left). Scatterometers reveal
more coherent spatial detail than NWP
winds. (Eumetsat OSI SAF, 1999)

instrument on ERS‑2 and Numerical Weather Prediction (NWP) model winds

(blue wind vectors). The ESCAT mesoscale wind details in the area where the
cold northerly flow interacts with the warmer southerly flow line up well with
the geostationary satellite cloud bands at the time of the ERS‑2 overpass. The
NWP winds describe only the larger scales well.
During the Second World War, when radar was used to detect and track
hostile vessels, it was noted that this detection was hampered by increasing
wind speed. Naturally, the idea of measuring wind near the sea surface by
using microwaves, i.e. scatterometry (e.g. Moore & Pierson, 1967), was soon
developed. Wind scatterometers measure the radar backscatter from wind-
generated centimetre-sized gravity capillary waves and provide high-resolution
vector wind fields over the oceans. All-weather scatterometer observations
have proven accurate and important for the forecasting of dynamic and
severe weather. Oceanographic applications have been developed since
scatterometers provide unique forcing information on the ocean eddy scale.
Scatterometers also provide useful information on soil moisture, snow cover
and sea ice (see Fig. 2).

146
 Scatterometer

Figure 2. An ‘impressionist’ view based

on the radar map of Earth’s surface as
observed by the ERS scatterometer. Sea
ice is shown in violet, variable winds in
grey and high winds in white, but with
wind sensitivity depending somewhat on
swath position. The yellow and blue in
the northern and southern hemispheres
indicate trade winds and correspond
to higher signals in the fore and aft
antennae, respectively, and thus indicate
the directional stability of the wind in the
intertropical regions. (Data processing by
IFARS, Germany, for land surfaces, and by
the ERS Product Control Service at ESRIN
The first scatterometer in space was NASA’s Seasat-A Scatterometer System for oceans. (ESA)
(SASS), which flew for three months in 1978. SASS had four antennae, two on
each side of the satellite (see e.g. Stoffelen, 1998). Each set of two antennae
covered a swath – one to the right of the subsatellite (ground) track and one to
the left. In the horizontal plane, the fore and aft beams were pointing at 45° and
135° azimuth, respectively, with respect to the ground track. A location in the
swath was first hit by the fore beam, and a few minutes later by the aft beam.
Thus, each Wind Vector Cell (WVC) in the swath revealed two backscatter
measurements obtained with a 90° difference in azimuth. Figure 3 illustrates
the analysis of two such measurements. For each measurement it shows the
wind speed solution as a function of all possible wind directions. Given the
dependency of the backscatter signal on the basic harmonic wind direction,
four solutions exist in this general case. This ambiguity poses a strong
limitation on the usefulness of the SASS wind data, and extended manual
efforts were needed to remove the ambiguity in order to obtain an acceptable
wind product (Peteherych et al., 1984). The usefulness of this product has been
demonstrated (see, e.g. Stoffelen & Cats, 1991).
The scatterometers on ERS‑1 and ERS‑2 (here denoted ESCAT) were identical
and each had three antennae that illuminated the ocean surface from three
different azimuth directions (e.g. Stoffelen, 1998). A point on the ocean surface
would be hit first by the fore beam, then by the mid-beam and soon after by
the aft beam. Since this provided three measurements to determine two

Figure 3. Wind speed as a function of

wind direction for fore and aft beam
measurements of backscatter from SASS
data. The arrows indicate the four possible
solutions for this typical case. (Utrecht
University, Stoffelen)

147
SP-1326

parameters, i.e. wind speed and direction, the ESCAT wind retrieval problem
is overdetermined in principle. Moreover, the dominant harmonic azimuth
dependence of the radar backscatter is a double harmonic (Long, 1986), which
is ideally sampled by three azimuth angles 45° apart, enabling rather constant
wind direction sensitivity of the ESCAT instrument. It was therefore anticipated
(Attema, 1991) that this measurement geometry would generally result in (only)
two opposite wind vector solutions and some residual information.
Based on the available technology, the C-band wavelength was chosen
for the ERS microwave mission. With hindsight, this was a very fortunate
choice as it provided sufficient wind sensitivity and little sensitivity to rain
effects. Several campaigns were conducted to prepare the ESCAT mission (e.g.
PROGRESS and TOSCANE).
Long (1986) developed empirical GMFs to match backscatter measurements
with the measurement geometry and local wind conditions. Inversion of
this GMF at each WVC revealed two wind vector solutions. As the satellite
propagates along its orbit it thus produced a swath of WVCs with ambiguous
winds, so that statistical techniques that exploited prior information on the
spatial structure of the surface wind field were developed to resolve these
ambiguities (e.g. Offiler, 1987).

2.1 Launch of the ERS Scatterometer

At the launch of ERS‑1, the European Centre for Medium-Range Weather
Forecasts (ECMWF) was prepared to validate the ESCAT backscatter data and
winds. However, comparisons of ESCAT and ECMWF backscatter and wind
distributions (see Fig. 4), showed large differences depending on the radar
beam, incidence angle, wind speed and wind direction. Since the ECMWF
routinely monitors the quality of their winds and this had been intensified
prior to the ERS-1 launch, it was clear that there were some problems with the

Figure 4. Difference between backscatter

data from ESCAT and the ECMWF wind
analysis using the CMOD2 GMF in
September 1991. A large bias existed
depending on the ERS‑1 radar beam and
incidence angle. The blue, green and red
curves represent the fore, mid- and aft
beams, respectively.

148
 Scatterometer

ESCAT winds. Since the bias patterns were very stable and repetitive, there was
hope that they could be removed.

2.2 ESCAT Measurement Space

Cavanié et al. (1986) made a 3D plot, where radar cross sections of the fore beam
were plotted on the horizontal (x) axis, those of aft beam on the other horizontal
(y) axis and mid-beam backscatter on the vertical (z) axis, as depicted in Fig. 5.
They realised that due to wind speed and direction sensitivity, a two-parameter
manifold or surface should emerge in such a 3D space. Stoffelen (1998) set out
to make cross sections through this surface in order to observe whether the
measured backscatter triplets would indeed follow a manifold and whether
the manifold was well described by CMOD2. This led to a true revolution in
scatterometry, as described below.
The expected coherence in the measurement data was found and indeed
the backscatter triplets are arranged in close proximity to a conical surface,
as depicted in Fig. 5. The major axis of the ‘cone’ corresponds to variations
in wind speed, whereas the minor axes represent changes in wind direction,
confirming a basic sensitivity to the near-surface wind vector over the world’s
oceans (Stoffelen, 1997a, 1998).
The actual shape of the manifold has been characterised to high precision
and the GMF has been improved accordingly (Stoffelen, 1998). The resulting
C-band GMF, called CMOD5.N (Hersbach et al., 2007), is also used for Synthetic
Aperture Radar (SAR; Portabella, 2002) and ASCAT wind retrieval (Stoffelen &
Anderson, 1997a).
The measured triplets are spread across the ideal conical manifold and
this spread has been characterised in detail by Stoffelen & Anderson (1997a).
Later, Portabella & Stoffelen (2006) modelled this spread in the ERS data and
found that at high winds it is mainly determined by radar (speckle) noise,
but at low winds by local wind variability. Since the fore, mid- and aft beams
sample a WVC slightly differently, due to their different sampling sequences
and footprint shapes, variable winds may cause the mean wind in the spatially
integrated radar footprint over a WVC to be slightly different for each of the

Figure 5. Depiction of the CMOD5.N

GMF manifold at the outer ESCAT swath
in measurement space spanned by the
backscatter measurements of the fore,
aft and mid antennas (red, green and blue
axes, respectively). The colours indicate the
wind direction (red 0°, green 120°, blue
240°, all with reference to the mid-beam).
The GMF is split at wind speeds of 10 m s–1,
20 m s–1 and 30 m s–1, respectively, to
reveal its internal structure. (Vogelzang &
Stoffelen, 2012)

149
SP-1326

three beams. This effect is particularly relevant for low winds, as these show
the highest relative variation in surface backscatter. The inconsistency in the
spatial collocation of fore, mid- and aft beams and the resulting backscatter
uncertainty due to local (sub-WVC) wind variability has been called
‘geophysical noise’ by Portabella & Stoffelen (2006).
In the cross sections (as in Fig. 6) occasional triplets are seen that lie almost
in the middle of the cone (i.e. along its major axis) and rather far away from
the wind GMF manifold as measured by the nominal noise parameters. These
triplets do not correspond to good-quality winds and are rejected after wind
retrieval by a Quality Control (QC) step. For ERS and ASCAT, about 0.5% of data
over the open ocean (Portabella et al., 2011) could be rejected as being unlikely
wind triplets. The same QC methodology has been applied to the NASA
Scatterometer (NSCAT; Figa & Stoffelen, 2000) and SeaWinds instruments at
the Ku-band with great success (Portabella & Stoffelen, 2001, 2002). At this
wavelength about 5% of open-ocean WVCs are rejected by the QC procedure,
most often due to rain clouds.
It was also noted that in cross sections across the cone (Fig. 6, left panels)
the measurements appear to be triangular in shape. Stoffelen & Anderson
(1997a) realised that prior knowledge of this particular shape could be used in
the wind retrieval in order to avoid irregularities (attractors) in the retrieved
wind directions. They found a rather simple transformation of the backscatter
measurement space that results in a circular manifold. A circular manifold has
constant prior probability of each wind direction and is straightforward in the
wind retrieval. Later, Stoffelen & Portabella (2006) noted that the wind retrieval
in the transformed measurement space with a circular GMF manifold is regular
since it results in a wind vector sensitivity that is rather smooth and constant.
However, while the transformation works well for ESCAT-type scatterometers,
constant wind vector sensitivity cannot be obtained for rotating pencil-beam
scatterometers. So, due to rain and measurement geometry rotating pencil-
beam Ku-band scatterometers are difficult to handle, but the experiences with
ESCAT resulted in unprecedented-quality retrievals for QuikScat and OceanSat-2

Figure 6. Cross sections through the

manifold in the ESCAT measurement space
(see Fig. 5) across the manifold (left panels)
and along its main axis (right panels). (AMS,
Stoffelen & Anderson, 1997a)

150
 Scatterometer

Table 1. Wind component error standard

Dataset εν (m s–1)
deviations with respect to the (different)
Buoy ECMWF Scatterometer scales resolved by the different
ASCAT-12.5 1.21 1.23 1.54 1.55 0.69 0.82 scatterometer wind products as obtained
by triple collocation. The lower the buoy
ASCAT-25 1.24 1.30 1.42 1.45 0.65 0.74
error, the better the scatterometer product
SeaWinds − KNMI 1.40 1.44 1.19 1.27 0.79 0.63 resolves the buoy scales, and thus the
SeaWinds − NOAA 1.39 1.41 1.20 1.30 1.20 1.04 higher the resolution. (AGU, Vogelzang
et al., 2011)

(Stoffelen et al., 2013; Vogelzang et al., 2011). Table 1 provides evidence of the
high accuracy of the scatterometer products based on ESCAT heritage.
The final advance in the geophysical interpretation of the ESCAT backscatter
data was its use to model sea ice and to discriminate between water and sea ice
surfaces. Just like wind over water, sea ice has a very particular signature in
the measurement space. Sea ice surfaces are generally isotropic with varying
roughness. Indeed, sea ice points lie on a line in measurement space, where the
coordinate along the line depicts sea ice roughness (De Haan & Stoffelen, 2001).
Subsequently, Bayesian methods were used to compute the probability of water
in the WVC (Belmonte Rivas & Stoffelen, 2011). This method, first developed for
ESCAT, is now also in use for the QuikScat and the OceanSat-2 scatterometers.
In summary, the advances inspired by the ESCAT measurement space
revolutionised not only the ESCAT wind retrieval methodology, but also those
of all other wind scatterometers.

2.3 ESCAT Wind Quality

Besides the advances inspired by the ESCAT measurement space, the ERS era
brought other innovations as well. The first was ocean calibration by NWP
model wind fields (Stoffelen, 1998), which is still applied today to obtain
interbeam backscatter calibration and the highest-quality ASCAT winds
(Verspeek et al., 2012). Another statistical assessment showing the high quality
and consistency of the ESCAT and ASCAT instruments and wind retrievals was
provided by Hersbach (2008); the plot in Fig. 7 was produced after calibrating
both ESCAT and ASCAT data to the ECMWF model.
Another innovation was in triple wind collocation (Stoffelen, 1998), which
is used today for ASCAT wind quality assessment (see Table 1), but also in
several other Earth observation disciplines, such as quality assessments of soil
moisture, sea surface temperature, sea ice drift and altimeter wave height data.
Together, NWP Ocean Calibration (NOC) and triple collocation essentially tie
the scatterometer winds to the global buoy wind network and at the same time
provide statistical evidence of biases in the NWP model (e.g. ECMWF). NOC and
triple collocation are performed routinely to assess the stability of scatterometer
winds over time (see www.knmi.nl/scatterometer). Since the global moored
buoy wind network is considered as an absolute reference, the combination of
triple collocation and NOC may also be used to intercalibrate scatterometers,
in particular ESCAT and ASCAT, with the objective of obtaining decadal
scatterometer wind time series. These procedures could thus be applied to obtain
a high-quality Fundamental Climate Data record for the ESCAT data.
The excellent statistical verification of both ESCAT and ASCAT winds was
corroborated by the inspection of wind maps such as the one shown in Fig. 1
(see www.knmi.nl/scatterometer). The spatially smooth wind vectors from the
High-Resolution Limited Area Model, shown in blue and purple, depict a cold
northerly flow on the left, adjacent to a warmer southerly flow on the right
(the amount of purple increases with wind speed). The scatterometer winds
from ERS‑2 at around 12:00 GMT on 24 November 1999 in the Norwegian Sea
(indicated in red) show much more structure and detail. Note that the grey-

151
SP-1326

Figure 7. Joint distribution of collocated

ESCAT and ASCAT winds with a Pearson
correlation of 98.9% and standard deviation
of differences of 0.77 m s–1. Blue dots:
average ERS‑2 wind speeds in a 1 m s−1
ASCAT wind speed bin; red dots: average
ASCAT wind speeds in an ERS‑2 bin.
(From Hersbach, 2008)

shaded Meteosat IR image is coherent with the scatterometer winds, with wind
convergence patterns lining up well with cloud patterns, revealing details of
the local meteorological conditions.
Since C-band scatterometers are much less affected by rain than Ku-band
scatterometers, the representation of deep tropical convection will be much
improved. However, there may still be issues of concern due to effects of splash, i.e.
rain droplets roughening the sea surface, or wind downbursts. The latter are due
to air cooled by melting and evaporating precipitation, thus increasing its mass
density and falling to the sea surface, where the air is deflected in the horizontal
direction, often in a circular outflow pattern. The horizontal extent of downburst
patterns is generally much larger than that of the associated rain patterns.
Figure 8 shows a representative example of moored buoy winds in the
tropics in the presence of rain. Wind shifts as large as 10  m  s–1 are often
associated with local rain. However, convective areas also show wind shifts
when no rain is measured at the buoy. This is most likely because the horizontal
extent of downburst areas is larger than those of rain areas. ASCAT winds are
variable in convective areas, but prove a reasonable representation of the local
buoy winds, given the rather high wind variance in the spatial representation
difference of buoy and ASCAT. Triple collocation shows that the ECMWF model
winds are clearly further away from the buoy winds, as expected, given the
rather smooth wind fields in such convective areas (Portabella et al., 2011). It
may be clear that convective processes in the tropics have a rather large effect
on the surface winds and air−sea interaction processes, but also on vertical
exchange in the atmosphere and tropical circulation. It remains a challenge
today to assimilate the full spatial detail as measured by a scatterometer into
NWP models. Therefore, process studies using scatterometer data will remain
necessary in order to exploit scatterometer data to the full.

152
 Scatterometer

Figure 8. Wind speed (left axis) and rain rate

(right axis) as functions of time at a rainy
buoy location as explained in the legend.
Time is centred around an ASCAT overpass
(black dot) where the reference ECMWF
forecast winds have base time at –25 h.
(Portabella et al., 2011)

3. Benefits of the ERS Scatterometers

3.1 Wind
Retrieved ERS scatterometer winds have proven to be very useful for the
forecasting of dynamic weather (Le Meur et al., 1998; Isaksen & Stoffelen, 2000)
using the 4D-Var data assimilation system. This is mainly due to the capability
of the 4D-Var system to assimilate winds with dual ambiguity (Stoffelen &
Anderson, 1997b) and to propagate the ESCAT surface information vertically
(Isaksen & Janssen, 2004). Increased coverage, such as from tandem ERS‑1/
ERS‑2 measurements, clearly improve the forecasts of extreme events (e.g.
Stoffelen & Beukering, 1997; Le Meur et al., 1998); see Fig. 9. Severe storms that
hit Europe often originate over the North Atlantic, where sparse meteorological
observations are available. Consequently, the initial stages of severe storms are
often poorly analysed and their development poorly predicted (e.g. ESA, 1999).
As a result, occasional devastating ocean or coastal wind and wave conditions
remain a major challenge for NWP. In addition, NWP data assimilation systems
are cycled over observation windows, e.g. of 6 h, with cutoff times in the range
of 2–6 h. Therefore, particularly in case of fast weather developments, which
often have large societal impacts, the timely use of satellite observations is
complicated.
Scatterometer data are therefore used for nowcasting applications such
as hurricane warnings over sea (Sienkiewicz et al., 2010) for marine traffic
or offshore activities. Moreover, the near-surface wind conditions drive the
ocean circulation, which in turn plays a major role in the climate system and
in marine ecosystems (e.g. fisheries). Scatterometer winds have proven to be
very relevant in driving ocean circulation and air–sea interactions (Chelton et
al., 2004; Tokmakian, 2005; Blanke et al., 2005; Liu & Xie, 2006), which in turn
play a major role in the climate system and marine biology (Moore & Renfrew,
2005). An overview of the Ocean Vector Wind observation requirements,
capabilities and applications is provided in Bourassa et al. (2009).

153
SP-1326

Figure 9. Mean sea level pressure fields

from the ECMWF model valid at 12:00 UTC
on 6 September 1995. Top left: Five-day
forecast using the 4D-Var data assimilation
system without ERS scatterometer data.
Bottom left: Forecast using the 4D-Var data
system with scatterometer data.
Top right: Analysed field for verification of
the forecasts showing an excellent tropical
cyclone strength forecast with ESCAT winds
at the 5-day range. (Le Meur et al., 1998)

3.2 Land Surface Monitoring

Because ESCAT was designed for wind monitoring, it was initially not clear if
and how it might be useful for monitoring of the land surface. One particular
concern was that the spatial resolution of ESCAT is simply too coarse to be
of value for land applications due to the high heterogeneity of land surfaces.
Nevertheless, the global systematic coverage achieved by ESCAT attracted
several research groups who began to investigate the capabilities of ESCAT for
mapping of vegetation (Frison & Mougin, 1996), soil moisture (Wagner et al.,
1999b), freeze/thawing (Wismann, 2000) and snow (Drinkwater et al., 2001).
One of the scientific challenges faced by all initial land surface studies was
the strong dependency of the backscattering coefficient on the incidence angle,
which varies from 18° for the mid-beam in the near-range, to 59° for the fore and
aft beams in the far range. Therefore, the backscattering coefficient changes
significantly over the image swath, and from acquisition to acquisition,
making the interpretation of the images, or time series, elusive without a prior
correction or normalisation of the backscatter data. Thus many of the initial
studies addressed this dependency by fitting a linear function to the ESCAT
backscatter measurements collected over a long period such as a month
(Mougin et al., 1995; Schmullius, 1997), following a procedure previously
developed for the SASS scatterometer by Kennett & Li (1989). This allowed
studies of the relationship between the ESCAT measurements and global land
cover, and seasonal land surface dynamics due to vegetation phenology or
freeze/thaw processes in high-latitude areas. Nevertheless, this monthly fitting
procedure suppresses short-term signal fluctuations caused, for example, by
changes in soil moisture, short-term freeze/thaw events, or changes in snow
morphology, rendering a more physically based interpretation of the ESCAT
data impossible.

154
 Scatterometer

To improve our capabilities to analyse ESCAT data over land, more

advanced modelling approaches have been developed in which the ESCAT
signal can be decomposed into its different backscatter components. One such
approach is the method developed by Wagner et al. (1999b), where in one of
the first processing steps the incidence angle dependency of σ0 is determined
for each land surface pixel and for each day of the year by comparing the
quasi-simultaneous mid- and aft/fore beam backscatter measurements. This
knowledge of the incidence angle behaviour then not only allows extrapolation
of the backscatter measurements to a given reference angle (40°), but also
separating the backscatter contributions due to seasonal vegetation growth
and decay from the shorter-term soil moisture fluctuations (Wagner et al.,
1999a). Furthermore, several semi-empirical backscatter models typically
composed of bare soil backscatter- and volume scattering formulations for
simulating backscatter of vegetation- and snow-covered surfaces have been
developed (Magagi & Kerr, 1997; Pulliainen et al., 1998; Wen & Su, 2003;
Woodhouse & Hoekman, 2000; Zribi et al., 2008). These semi-empirical models
can be used to improve our theoretical process understanding and, at a more
practical level, the simultaneous retrieval of soil moisture, vegetation and
other land surface parameters (e.g. roughness) using iterative least-squares
matching procedures. These models have been implemented over selected
world regions with varying degrees of success, but not yet on a global scale.
These research efforts led to the derivation of the first global long-term
(1992−2000) soil moisture database using ESCAT data from both ERS‑1 and ERS‑2
(Scipal et al., 2002; Wagner et al., 2003; see Fig. 10). This dataset was released
in 2002 by the Vienna University of Technology (TU Wien) and freely shared
with the scientific community. Fortunately, the dataset raised considerable
interest and several validation studies carried out by independent research

Figure 10. Mean monthly soil moisture derived from ESCAT backscatter data (1992−2000) for January, April, July and October. Brown tones
indicate dry conditions (wilting level), blue tones indicate wet conditions (field capacity). Tropical forest, desert regions with strong azimuth
effects, and areas affected by snow and frost have been masked out.

155
SP-1326

teams quickly demonstrated the high quality of the data (Dirmeyer et al.,
2004; Drusch et al., 2004; Pellarin et al., 2006). This was unexpected because
active C-band measurements had not been considered to be well suited for the
retrieval of soil moisture. This was mainly because, first, active measurements
are more sensitive to surface roughness compared with passive microwave
measurements and, second, the C-band (λ = 3.8−7.5 cm) is less able to penetrate
vegetation and the soil than the L-band (λ = 15−30 cm) (Wagner et al., 2007).
Exactly for these reasons, passive long-wavelength measurement concepts
have been selected for ESA’s Soil Moisture and Ocean Salinity (SMOS) mission
and NASA’s Soil Moisture Active Passive (SMAP) mission – the first two satellite
missions developed specifically for the retrieval of soil moisture – utilising
passive sensors operating at the L-band (Entekhabi et al., 2004; Kerr, 2007).
Nevertheless, the positive results from the validation studies, and
the increasing use of the ESCAT soil moisture database in such diverse
applications as hydrology (Parajka et al., 2006; Scipal et al., 2005), Numerical
Weather Prediction (Scipal et al., 2007; Zhao et al., 2006), agronomy (De Wit
& van Diepen, 2007) and climate monitoring (Künzer et al., 2009) led to the
recommendation to build up a global near-realtime soil moisture service for
ASCAT. This operational service was developed by Eumetsat in cooperation
with TU Wien, and went fully operational in December 2008 (Bartalis et al.,
2007; Wagner et al., 2010).
In retrospect, it can be said that this unexpected success was possible
because of ESCAT’s unique characteristics:

—— the multi-incidence angle viewing capability, which allows the separation of

vegetation and soil moisture effects;
—— the high temporal sampling rate, which allows researchers to exploit the
advantages of change detection; and
—— the excellent radiometric accuracy, which results in a suitable signal-to-noise
ratio for the task of soil moisture retrieval.

Now that the ERS‑2 mission has come to a successful end, there is an
opportunity to reprocess the complete ESCAT data archive to 25 km to create
a Fundamental Climate Data Record (FCDR) that will provide crucial inputs
to any effort dealing with the creation of an Essential Climate Variable (ECV)
record on soil moisture (Wagner et al., 2009). The first efforts in this direction
are already underway and it is planned to release an improved 25-km resolution
ESCAT soil moisture dataset in 2013. Furthermore, by merging ESCAT and
ASCAT soil moisture data with a suite of soil moisture products derived from
passive microwave sensors (SMMR, SSM/I, TMI, AMSR-E) it will be possible
to create a 30+ year soil moisture ECV data record (Dorigo et al., 2010; Liu et
al., 2011). Such a first soil moisture ECV dataset, where ESCAT is essential for
characterising the climate in the 1990s, is also planned for release in 2013.

3.3 Sea Ice and Drift

The extent of sea ice over the polar oceans is a critical parameter for
understanding and forecasting ocean circulation and climate change.
No satisfactory conventional observing system is available and sea ice is
difficult to observe using satellite optical sensors because of the low level of
illumination and the frequent cloud cover in polar regions.
Microwave instruments and in particular satellite-borne radars, such
as the ESCAT scatterometers, are of great interest because of their large
temporal and spatial coverage and their all-weather measurement capability.
The development of specific data analysis methods for these radars led to the
detection of sea ice and to estimates of its extent, drift and age (see Fig. 11).
A routine ESCAT processing has been developed at the ERS Processing and

156
 Scatterometer

Figure 11. Images of the backscatter

coefficient over the Arctic region, observed
at an incidence angle of 40° during October
1995 (ERS‑1, left) and October 1996
(ERS‑2, right). Dark blue: unprocessed
areas; grey: open water. The highest values
of the backscatter coefficient are observed
for multi-year ice (ice that has survived
at least one summer), shown in green
or yellow. This pair of images shows how
the sea ice situation can differ from year
to year. In 1996, new and multi-year ice
covered the coastal zones of Siberia where
large areas of open water were apparent in
the 1995 image. (CERSAT, 2011)
Archiving Facility of the French Research Institute for Exploitation of the Sea
(CERSAT/IFREMER) to produce fields of sea ice age, drift and extent, with polar
coverage, on a weekly and monthly basis.
Over polar oceans, values of backscatter measurements depend on the
dielectric properties of the observed materials − sea water, first-year ice and
multi-year ice − on their relative concentrations and on surface topography. As
sea ice ages, its dielectric properties, linked to volume and surface scattering,
as well as its surface roughness are modified. Sea ice characteristics are
derived from backscatter levels at prescribed incidence angles (40°), and from
backscatter variations with incidence angle. In summer, surface melting of sea
ice changes the backscatter signature and renders interpretation of the data
more difficult.

4. The Future
4.1 C-band Fan Beam Scatterometer Heritage
ESCAT was followed by ASCAT, a clear heritage instrument, building on
the success of the ESCAT instrument and ground processing. The Eumetsat
Polar System programme lists the Second Generation ASCAT as a high-
priority instrument. Specific requirements include slightly increased spatial
resolution, extended swath, radiometric stability with reference to ASCAT,
and the addition of horizontal polarisation. The C-band was chosen because it
provides good sensitivity, all-weather capability and continuity of the ESCAT
and ASCAT series. Rotating pencil beam concepts were discarded due to poorer
simulated performance, while rotating fan beam concepts were considered to
be technically too complex, since ASCAT-SG would share a satellite platform
with several other Earth-viewing microwave instruments (Lin et al., 2012).
The addition of horizontal polarisation could potentially result in an
improved capability to sense hurricane winds, since the vertical emit and receive
polarisation of ESCAT and ASCAT saturates at around 40 m s–1. Since Eumetsat,
together with many international sister agencies, is involved in building a global
constellation of ocean vector wind missions (CEOS VC) and obtaining extreme
winds is a prerequisite for these sister agencies, the requirement for measuring
extreme winds becomes more important. Horizontal polarisation in addition
to vertical polarisation, however, requires an extended-capability antenna,
which may be best accommodated on the mid-beams that are fixed on the body
of the satellite platform. When emitting and receiving horizontal polarisation
(HH) with the mid-beam, the additional sensitivity to hurricane winds would
be limited as the geophysical dependency of HH is only favourable for extreme
winds at incidence angles above 35° (see Fig. 12). Therefore, ASCAT-SG incidence
angles from 20° to 35° would not profit much from the extended HH capability.

157
SP-1326

Figure 12. C-band horizontal polarisation

backscatter as a function of wind speed.
The solid lines represent HH at varying
incidence angles, while the red dotted
line represents anticipated VH (see text).
For extreme winds the sensitivity to wind
speed appears to be most favourable for VH
backscatter measurements.

Another, more convenient, option would be to use the vertically emitted pulses
and use the dual-polarisation mid-beam antennas to receive both the vertical
(VV) and horizontal polarisations (VH).
Recently, Vachon & Wolfe (2011) reported on the Radarsat SAR, which has
cross-polarisation or VH capability, a linear dependency of VH backscatter
on wind speed. No significant dependency on azimuth or incidence angle
was found. Preliminary verification of the Radarsat SAR VH backscatter
dependency in tropical hurricanes confirmed a linear dependency at extreme
winds as well, but with somewhat reduced slope. NOAA’s aircraft campaigns
and Eumetsat’s Satellite Application Facility for Numerical Weather Prediction
(NWP SAF) performance assessments are planned to verify the VH geophysical
dependency at extreme winds.
Cross-polarisation capability would furthermore improve vegetation
determination and, as a consequence, soil moisture index retrieval. Advantages
for sea ice detection are also anticipated.
Following the successful ERS mission at the C-band and successful NASA
missions at the Ku-band, several agencies plan to develop scatterometer
systems combining both wavelengths. In this way, resolution enhancements
may be combined with the C‑band all-weather capability.

4.2 Other Wind Resources

Besides using scatterometers, components of the wind field can be measured
with SAR (e.g. Portabella et al., 2002) or passive radar measurements (WindSat,
2011; SSMIS, 2011). As a scatterometer, a SAR also measures the radar
backscatter but at higher resolution (down to ~100 m) and in only one azimuth
direction. The GMF derived for ESCAT is often used for C-band SAR wind
retrieval. The latter is a problem since it is not possible to resolve both wind
direction and speed at a matching spatial resolution without prior information,
while the sensitivities to wind speed and direction (in m  s–1 per dB) are
comparable (underdetermined problem). For passive radar measurements a
similar underdetermination applies, but here the wind direction sensitivity
of the measurements is so small that wind speed can be retrieved with
sufficient accuracy. However, a more significant problem is that the microwave
measurement sensitivity to rain, cloud liquid water and humidity has to be

158
 Scatterometer

appropriately accounted for by using multiple-frequency measurements. This

limits the performance of passive radar winds in dynamic weather conditions.

4.3 Doppler Scatterometer

Scatterometer winds are measured relative to the ocean, whereas buoy and
NWP model winds are provided with reference to a fixed Earth reference.
Comparisons of these datasets therefore require that ocean currents are
known. Direct current measurements would greatly improve ocean modelling,
particularly in the tropics. ESA recently launched a project that aims to assess
the potential of scatterometer instruments for sea surface current retrieval in
addition to surface wind vector estimation. Besides new conceptual ideas, the
project will provide recommendations on signal processing techniques and
GMF development for surface current retrieval from the Doppler estimates.
Since the wind forces a net motion on the ocean waves, called the Stokes drift,
the Doppler capability will also provide wind information, which may be
combined with backscatter data.

5. Conclusions
C-band scatterometer data are now widely used for hurricane prediction,
Numerical Weather Prediction, marine nowcasting, oceanography, hydrology
and studies of the cryosphere, providing societal and economic benefits in
areas such as marine safety, offshore activities, ship routing, wind energy
and climate change monitoring. The ESCAT research mission anticipated such
a wide range of applications, with for example the near-realtime availability
of the scatterometer winds. The launch of ERS‑1 inspired the development of
novel analytical techniques using the measurement space of ESCAT, which
confirmed the expected internal consistency of the ESCAT measurements and
the low noise, but also provided improved characterisation of the sensitivity
to wind speed and direction, resulting in improved GMFs. Moreover, the
visualisation of the ESCAT measurement space provided insights into the
nonlinear wind retrieval problem, which has subsequently been improved.
It has also aided the discrimination of sea ice and water and has provided
important insights into the wind retrieval quality control.
The ESCAT wind Cal/Val led to the further development of the triple
collocation methodology that is now widely used in Earth observation. The
lessons learnt from ESCAT were readily applied to the NASA scatterometer
missions at the Ku-band and provided similar benefits, leading to improved
rain screening and wind datasets. ESCAT has also been successfully applied
in soil moisture, sea ice and snow characterisation, leading to operational
applications. The ESCAT heritage has been important for the development of
ASCAT on MetOp for the Eumetsat Polar System. The EPS Second Generation
ASCAT is now being designed, where the ESCAT C-band static fan beam concept
is being maintained due to its great success. This may well lead to over 40 years
of C-band scatterometer data, which will be a tremendous resource for climate
applications to study wind climatologies, air–sea interactions, atmospheric
processes, etc. The demand for the unique ESCAT wind, soil moisture, snow
and sea ice data will therefore remain high in the years to come, and continued
ESA data and processing services would provide great benefit to society.

159
SP-1326

Acknowledgements
The NWP SAF scatterometer wind processors for SeaWinds (SDP) and ASCAT
(AWDP) are freeware, and can be obtained upon registration at www.nwpsaf.org.
Access to near-realtime scatterometer wind products can be obtained upon
registration at www.osi-saf.org. The NWP SAF and OSI SAF are sponsored by
Eumetsat.
TU Wien would like to acknowledge the support of the Austrian Science
Fund (FWF), the Austrian Space Application Programme (ASAP) and ESA for
developing the first ESCAT soil moisture database. The support through the
enhanced-resolution ERS‑2 Scatterometer Soil Moisture Product project funded
by ESA is also gratefully acknowledged.

References
Attema, E.P.W. (1991). The Active Microwave Instrument on-board the ERS‑1
satellite. Proc. IEEE 79(6), 791–799. doi: 10.1109/5.90158
Bartalis, Z., Wagner, W., Naeimi, V., Hasenauer, S., Scipal, K., Bonekamp, H., Figa,
J. and Anderson, C. (2007). Initial soil moisture retrievals from the MetOp-A
Advanced Scatterometer (ASCAT). Geophys. Res. Lett 34, L20401.
Belmonte Rivas, M. & A. Stoffelen, 2011, New Bayesian algorithm for sea ice
detection with QuikSCAT. IEEE Trans. Geosci. Rem. Sens. I 49(6), 1894–1901.
doi:10.1109/TGRS.2010.2101608
Blanke, B., Speich, S., Bentamy, A., Roy, C. & Sow, B. (2005). Modelling the
structure and variability of the southern Benguela upwelling using QuikSCAT
wind forcing. J. Geophys. Res. 110, C07018. doi: 1029/2004JC002529
Bourassa, M.A., Stoffelen, A., Bonekamp, H., Chang, P., Chelton, D.B., Courtney,
J., Edson, R. Figa, J. He, Y., Hersbach, H., Hilburn, K., Jelenak, Z., Kelly, K.A.,
Knabb, R., Lee, T., Lindstrom, E.J., Liu, W.T., Long, D.G., Perrie, W., Portabella,
M., Powell, M., Rodriguez, E., Smith, D.K., Swail, V. & Wentz, F.J. (2009).
Remotely sensed winds and wind stresses for marine forecasting and ocean
modelling. Proc. OceanObs ’09: Sustained Ocean Observations and Information
for Society, Community White Paper. www.oceanobs09.net/proceedings/cwp/
Bourassa-OceanObs09.cwp.08.pdf
Cavanié, A., Demurger, J. & Lecomte, P. (1986). Evaluation of the different
parameters in Long’s C-band model. Proc. Workshop on ERS‑1 Wind and
Wave Calibration, Schliersee, Germany. ESA SP-262, European Space Agency,
Noordwijk, the Netherlands, pp.47−51.
CERSAT (2011). Monitoring Sea Ice Using Scatterometers of the ERS Satellites. Centre
ERS d’Archivage et de Traitement, IFREMER, France. http://cersat.ifremer.fr/
Science-Applications/Sea-ice/Sea-Ice-by-ERS
Chelton, D.B., Schlax, M.G. Freilich, M.H. & Milliff, R.E. (2004). Satellite
measurements reveal persistent small-scale features in ocean winds. Science
303, 978–983.
De Haan, S. & Stoffelen, A. (2001). Ice Discrimination using ERS Scatterometer.
Document external project:  2001, SAF/OSI/KNMI/TEC/TN/120, Eumetsat.
www.knmi.nl/publications/fulltexts/safosi_w_icescrknmi.pdf
De Wit, A. & Van Diepen, C. (2007). Crop model data assimilation with the ensemble
Kalman filter for improving regional crop yield forecasts. Agric. Forest Meteorol.
146(1–2), 38–56.
Dirmeyer, P.A., Guo, Z.C. & Gao, X. (2004). Comparison, validation, and
transferability of eight multiyear global soil wetness products. J. Hydrometeorol.
5(6), 1011–1033.
Dorigo, W.A., Scipal, K., Parinussa, R.M., Liu, Y.Y., Wagner, W., de Jeu, R.A.M. &
Naeimi, V. (2010). Error characterisation of global active and passive microwave
soil moisture datasets. Hydrol. Earth Syst. Sci. 14(12), 2605–2616.

160
 Scatterometer

Drinkwater, M.R., Long, D.G. & Bingham, A.W. (2001). Greenland snow accumulation
estimates from satellite radar scatterometer data. J. Geophys. Res. D: Atmospheres
106(D24), 33 935–33 950.
Drusch, M., Wood, E., Gao, H. & Thiele, A. (2004). Soil moisture retrieval during
the Southern Great Plains Hydrologic Experiment 1999: A comparison between
experimental remote sensing data and operational products. Water Resources
Res. 40(W02504). doi:10.1029/2003WR002441.
Entekhabi, D., Njoku, E.G., Houser, P., Spencer, M., Doiron, T., Yunjin, K., Smith, J.,
Girard, R., Belair, S., Crow, W., Jackson, T.J., Kerr, Y.H., Kimball, J.S., Koster, R.,
McDonald, K.C., O’Neill, P.E., Pultz, T., Running, S.W., Jiancheng, S., Wood, E.
& van Zyl, J. (2004). The Hydrosphere State (Hydros) satellite mission: An Earth
system pathfinder for global mapping of soil moisture and land freeze/thaw.
IEEE Trans. Geosci. Rem. Sens. 42(10), 2184–2195.
ESA (1998). Proc. Emerging Scatterometer Applications Workshop, ESTEC. ESA
SP‑424, European Space Agency, Noordwijk, the Netherlands.
ESA (1999). Atmospheric Dynamics Mission: Core Earth Explorer Mission Selection
Report. ESA SP-1233(4), European Space Agency, Noordwijk, the Netherlands.
Eumetsat OSI SAF (1999). www.knmi.nl/scatterometer/ers_prod/
Figa, J. & Stoffelen, A. (2000). On the assimilation of Ku-band scatterometer winds
for weather analysis and forecasting, IEEE Trans. Geosci. Rem. Sens. 38(4),
1893–1902.
Frison, P.L. & Mougin, E. (1996). Monitoring global vegetation dynamics with ERS‑1
wind scatterometer data. Int. J. Rem. Sens. 17(16), 3201–3218.
Hersbach, H. (2008). The usage of scatterometer data at ECMWF. OVWST meeting,
Seattle. http://coaps.fsu.edu/scatterometry/meeting/docs/2008/othersci/hersbach.pdf
Hersbach, H., Stoffelen, A. & de Haan, S. (2007). An improved C-band
scatterometer ocean geophysical model function: CMOD5. J. Geophys. Res. 112.
doi:10.1029/2006JC003743.
Hollmann, R., Merchant, C.J., Saunders, R., Downy, C., Buchwitz, M., Cazenave, A.,
Chuvieco, E., Defourny, P., de Leeuw, G., Forsberg, R., Holzer-Popp, T., Paul, F.,
Sandven, S., Sathyendranath, S., van Roozendael, M. & Wagner, W. (2013), The
ESA Climate Change Initiative: satellite data records for essential climate variables.
Bull. Am. Meteorol. Soc. doi: http://dx.doi.org/10.1175/BAMS-D-11-00254.1
Isaksen, L. & Stoffelen, A. (2000). ERS scatterometer wind data impact on ECMWF’s
tropical cyclone forecasts. IEEE Trans. Geosci. Rem Sens. 38, 1885–1892.
Isaksen, L. & Janssen, P.A.E.M. (2004). Impact of ERS scatterometer winds in
ECMWF’s assimilation system. Q. J. R. Meteorol. Soc. 130(600), 1793–1814, Part
A. DOI: 10.1256/qj.03.110.
Kennett, R.G. & Li, F.K. (1989). Seasat over-land scatterometer data, I: Global
overview of the Ku-band backscatter coefficients. IEEE Trans. Geosci. Rem. Sens.
27(5), 592–605.
Kerr, Y.H. (2007). Soil moisture from space: Where are we? Hydrogeol. J. 15(1),
117–120.
Künzer, C., Zhao, D., Scipal, K., Sabel, D., Naeimi, V., Bartalis, Z., Hasenauer, S.,
Mehl, H., Dech, S. & Wagner, W. (2009). El Niño Southern Oscillation influences
represented in ERS scatterometer-derived soil moisture data. Appl. Geog. 29,
463–477.
Le Meur, D., Isaksen, L. Hansen, B., Saunders, R. & Janssen, P. (1998). Global
Validation of ERS Wind and Wave Products. Final report to the European Space
Agency, ESA contract No. 8488/95/NL/CN, ECMWF, Shinfield Park, Reading,
Berkshire, UK.
Lin, C.C., Betto, M. Belmonte Rivas, M. Stoffelen A. & de Kloe, J. (2012). Post-
EPS wind scatterometer concept trade-offs and wind retrieval performance
assessment. IEEE Trans. Geosci. Rem. Sens. 50(7), 2458−2472. doi: 10.1109/
TGRS.2011.2180393

161
SP-1326

Eumetsat Meteorological Satellite Conf., September 2010, Cordoba, Spain.

www.eumetsat.int/Home/Main/AboutEUMETSAT/Publications/
ConferenceandWorkshopProceedings/2010/groups/cps/documents/document/
pdf_conf_p57_s2_06_lin_v.pdf
Liu, W.T. & Xie, X. (2006). Measuring ocean surface wind from space. In J. Gower
(Ed), Remote Sensing of the Marine Environment, Manual of remote Sensing.
3rd edn, vol. 6, American Society for Photogrammetry and Remote Sensing,
pp.149–178.
Liu, Y.Y., Parinussa, R.M., Dorigo, W.A., de Jeu, R.A.M., Wagner, W., van Dijk, A.,
McCabe, F.M. & Evans, J.P. (2011). Developing an improved soil moisture dataset
by blending passive and active microwave satellite-based retrievals. Hydrol.
Earth Syst. Sci. 15 (2), 425-436.
Long, A.E. (1986). Towards a C-band radar echo model for the ERS‑1 scatterometer.
3rd Int. Coll. Spectral Signatures in Remote Sensing, Les Arc, France. ESA SP-247,
European Space Agency, Noordwijk, the Netherlands, pp.29–34.
Magagi, R.D. & Kerr, Y.H. (1997). Retrieval of soil moisture and vegetation
characteristics by use of ERS‑1 wind scatterometer over arid and semi-arid areas.
J. Hydrol. 188–189, 361–384.
Moore, G.W. & Renfrew, I.A. (2005). Tip jets and barrier winds: a QuikSCAT
climatology of high wind speed events around Greenland, J. Climate 18, 3713–
3725.
Moore, R.J. & Pierson, W.J. (1967). Microwave determination of winds at sea. Proc.
IEEE 67, 1504–1521.
Mougin, E., Lopes, A., Frison, P.L. & Proisy, C. (1995). Preliminary analysis of ERS‑1
wind scatterometer data over land surfaces. Int. J. Rem. Sens. 16(2), 391–398.
Offiler, D. (1987). ERS‑1 Wind Retrieval Algorithms. UK Met Office, MET O 19, Branch
Memorandum, No. 86.
Parajka, J., Naeimi, V., Blöschl, G., Wagner, W., Merz, R. & Scipal, K. (2006).
Assimilating scatterometer soil moisture data into conceptual hydrologic models
at the regional scale. Hydrol. Earth Syst. Sci. 10, 353–368.
Pellarin, T., Calvet, J.-C. & Wagner, W. (2006). Evaluation of ERS scatterometer soil
moisture products over a half-degree region in southwestern France. Geophys.
Res. Lett. 33(17), L17401.
Peteherych, S., Wurtele, M.G., Woiceshyn, P.M., Boggs, D.H. & Atlas, R. (1984). First
global analysis of SEASAT scatterometer winds and potential for meteorological
research. NASA Conf. Publ. CP2303, pp.575–585.
Portabella, M. (2002). Wind Field Retrieval from Satellite Radar Systems. Thesis,
University of Barcelona, Spain. www.knmi.nl/publications/fulltexts/phd_thesis.pdf
Portabella, M. & Stoffelen, A. (2001). Rain detection and quality control of sea
winds. J. Atm. Ocean Techn. 18(7), 1171–1183.
Portabella, M. & Stoffelen, A. (2002). A comparison of KNMI quality control and JPL
rain flag for seawinds. Can. J. Rem. Sens. 28(3), 424–430.
Portabella, M. & Stoffelen, A. (2006). Scatterometer backscatter uncertainty due to
wind variability. IEEE Trans. Geosci. Rem. Sens. 44(11), 3356–3362. doi:10.1109/
TGRS.2006.877952
Portabella, M., Stoffelen, A. & Johannessen, J.A (2002). Toward an optimal inversion
method for SAR wind retrieval. J. Geophys. Res. 107(C8), 1–13.
Portabella, M., Stoffelen, A., Verhoef, A. & Lin, W. (2011). Rain effects on ASCAT
retrieved winds: towards an improved quality control, IEEE Trans. Geosci. Rem.
Sens. 50(7), 2495–2506.
Pulliainen, J.T., Manninen, T. & Hallikainen, M.T. (1998). Application of ERS‑1 wind
scatterometer data to soil frost and soil moisture monitoring in boreal forest
zone. IEEE Trans. Geosci. Rem. Sens. 36(3), 849–863.
Schmullius, C.C. (1997). Monitoring Siberian forests and agriculture with the ERS‑1
wind scatterometer. IEEE Trans. Geosci. Rem. Sens. 35(5), 1363–1366.

162
 Scatterometer

Scipal, K., Drusch, M., Hasenauer, S., Wagner, W. & Jann, A. (2007). Towards the
assimilation of scatterometer derived soil moisture in the ECMWF numerical
weather prediction system. Joint 2007 Eumetsat Meteorological Satellite Conf. and
15th Satellite Meteorology & Oceanography Conf. of the American Meteorological
Society, Amsterdam, the Netherlands.
Scipal, K., Scheffler, C. & Wagner, W. (2005). Soil moisture–runoff relation at the
catchment scale as observed with coarse resolution microwave remote sensing.
Hydrol. Earth Syst. Sci. 9(3), 173–183.
Scipal, K., Wagner, W., Trommler, M. & Naumann, K. (2002). The Global Soil
Moisture Archive 1992–2000 from ERS scatterometer data: First results. Proc.
IGARSS 2002.
Sienkiewicz, J., Brennan, M.J., Knabb, R., Chang, P.S., Cobb, H., Jelenak, Z.J.,
Ahmad, K.A., Soisuvarn, S., Kosier, D. & Bancroft, G. (2010). Impact of the loss of
QuikSCAT on NOAA MWS marine warning and forecast operations. 1st Int. Ocean
Vector Winds Science Team Meeting, Barcelona, Spain.
SSMIS (2011). www.ncdc.noaa.gov/oa/rsad/ssmi/swath/index.html
Stoffelen, A. (1998). Scatterometry, PhD Thesis, Utrecht University, the Netherlands.
http://igitur-archive.library.uu.nl/dissertations/01840669/inhoud.htm
Stoffelen, A. & Anderson, D. (1993). Wind retrieval and ERS‑1 scatterometer radar
backscatter measurements. Adv. Space Res. 13(5), (5)53–(5)60.
Stoffelen, A. & Anderson, D. (1997a). Scatterometer data interpretation: measurement
space and inversion. J. Atm. Ocean Techn. 14(6), 1298–1313.
Stoffelen, A. & Anderson, D. (1997b). Ambiguity removal and assimilation of
scatterometer data. Q. J. R. Meteorol. Soc. 123, 491–518.
Stoffelen, A. & van Beukering, P. (1997). Implementation of Improved ERS
Scatterometer Data Processing and its Impact on HIRLAM Short Range Weather
Forecasts. HIRLAM Technical Report No. 31, KNMI, the Netherlands.
Stoffelen, A. & Cats, G. (1991). The impact of SeaSat-A scatterometer data on high-
resolution analyses and forecasts: The development of the QEII storm. Mon. Wea.
Rev. 119, 2794–2802.
Stoffelen, A., & Portabella, M. (2006). On Bayesian scatterometer wind inversion,
IEEE Trans. Geosci. Rem. Sens. 44(6), 1523–1533. doi:10.1109/TGRS.2005.862502
Stoffelen, A. & Verhoef, A. (2011). Validation of Oceansat-2 scatterometer data,
Proc. Eumetsat Satellite Conf., Oslo, Norway, www.eumetsat.int/Home/Main/
AboutEUMETSAT/Publications/ConferenceandWorkshopProceedings/2011/
groups/cps/documents/document/pdf_conf_p59_s5_06_stoffele_v.pdf
Stoffelen, A., Verhoef, A., Verspeek, J., Vogelzang, J., Driesenaar, T., Risheng, Y.,
Payan, C., De Chiara, G., Cotton, J., Bentamy, A., Portabella, M. & Marseille,
G.J. (2013). Research and Development in Europe on global application of
the OceanSat-2 Scatterometer winds, Document extern project: 2013, NWP
SAF report number: NWPSAF-KN-TR-022 OSI SAF report no. SAF/OSI/CDOP2/
KNMI/TEC/RP/196, KNMI, De Bilt, the Netherlands. www.knmi.nl/publications/
fulltexts/oceansat_cal_val_report_final_copy1.pdf
Tokmakian, R. (2005). An ocean model’s response to scatterometer winds. Ocean
Modeling 9, 89–103.
Vachon, P. & Wolfe, J. (2011). C-band cross-polarization wind speed retrieval. IEEE
Trans. Geosci. Rem. Sens. Lett. 8(3), 456–459. doi: 10.1109/LGRS.2010.2085417.
Verspeek, J., Stoffelen, A. Verhoef, A. & Portabella, M. (20112. Improved ASCAT
wind retrieval using NWP ocean calibration. IEEE Trans. Geosci. Rem. Sens.
50(7), 2488–2494. doi: 10.1109/TGRS.2011.2180730
Vogelzang, J., Stoffelen, A., Verhoef, A. & Figa-Saldaña, J. (2011). On the
quality of high-resolution scatterometer winds. J. Geophys. Res. 116, C10033.
doi:10.1029/2010JC006640.
Vogelzang, J. & Stoffelen, A. (2012). Scatterometer wind vector products for
application in meteorology and oceanography. J. Sea Res. special issue on
Physics of Sea and Ocean, 74, 16–25.

163
SP-1326

Wagner, W., Bartalis, Z., Naeimi, V., Park, S.-E., Figa-Saldana, J. & Bonekamp, H.
(2010). Status of the METOP ASCAT soil moisture product. Proc. IEEE Geoscience
and Remote Sensing Symp., IGARSS 2010, Honolulu, USA, pp.276–279.
Wagner, W., Blöschl, G., Pampaloni, P., Calvet, J.-C., Bizzarri, B., Wigneron, J.-P.
& Kerr, Y. (2007). Operational readiness of microwave remote sensing of soil
moisture for hydrologic applications. Nordic Hydrol. 38(1), 1–20.
Wagner, W., de Jeu, R. & van Oevelen, P. (2009). Towards multi-source global soil
moisture datasets for unravelling climate change impacts on water resources.
Proc. 33rd Int. Symp. on Remote Sensing of Environment (ISRSE). Joint Research
Centre of the European Commission, Stresa, Italy, p.3.
Wagner, W., Lemoine, G., Borgeaud, M. & Rott, H. (1999a). A study of vegetation
cover effects on ERS scatterometer data. IEEE Trans. Geosci. Rem. Sens. 37(2),
938–948.
Wagner, W., Noll, J., Borgeaud, M. & Rott, H. (1999b). Monitoring soil moisture over
the Canadian Prairies with the ERS scatterometer. IEEE Trans. Geosci. Rem. Sens.
37(1), 206–216.
Wagner, W., Scipal, K., Pathe, C., Gerten, D., Lucht, W. & Rudolf, B. (2003).
Evaluation of the agreement between the first global remotely sensed soil
moisture data with model and precipitation data. J. Geophys. Res. D: Atmospheres
108(D19), Art. No. 4611.
Wen, J. & Su, Z.B. (2003). A time series based method for estimating relative soil
moisture with ERS wind scatterometer data. Geophys. Res. Lett. 30(7), 1397.
WindSat (2011). www.nrl.navy.mil/WindSat/
Wismann, V. (2000). Monitoring of seasonal thawing in Siberia with ERS
scatterometer data. IEEE Trans. Geosci. Rem. Sens. 38(4 II), 1804–1809.
Woodhouse, I.H. & Hoekman, D.H. (2000). Determining land-surface parameters
from the ERS wind scatterometer. IEEE Trans. Geosci. Rem. Sens. 38(1), 126–140.
Zhao, D., Su, B. & Zhao, M. (2006). Soil moisture retrieval from satellite images and
its application to heavy rainfall simulation in eastern China. Adv. Atmos. Sci.
23(2), 299–316.
Zribi, M., André, C. & Decharme, B. (2008). A method for soil moisture estimation in
Western Africa based on the ERS scatterometer. IEEE Trans. Geosci. Rem. Sens.
46(2), 438–448.

164
→→ The ERS SAR Wave Mode:→
A Breakthrough in Global
Ocean Wave Observations
 SAR Wave Mode

The ERS SAR Wave Mode: A Breakthrough in

Global Ocean Wave Observations
K. Hasselmann1, B. Chapron2, L. Aouf 3, F. Ardhuin4, F. Collard5, G. Engen6,
S. Hasselmann7, P. Heimbach8, P. Janssen9, H. Johnsen10, H. E. Krogstad11,
S. Lehner12, J.-G. Li13, X.-M. Li14, W. Rosenthal15, J. Schulz-Stellenfleth16

1 Max Planck Institute of Meteorology, 20146 Hamburg, Germany.

Klaus.Hasselmann@zmaw.de
2 IFREMER, 29280 Plouzané, France. bertrand.chapron@ifremer.fr
3 Météo-France, Division Marine et Océanographie, 31100 Toulouse, France.
lotfi.aouf@meteo.fr
4 IFREMER, 29280 Plouzané, France. fabrice.ardhuin@ifremer.fr
5 OceanDataLab, 29280 Plouzané, France. dr.fab@oceandatalab.com
6 Norut, P.O. Box 6434, 9294 Tromsø, Norway. geir.engen@norut.no
7 Max Planck Institute of Meteorology, 20146 Hamburg, Germany.
susanne.hasselmann@zmaw.de
8 Massachusetts Institute of Technology, Cambridge, MA 02139, USA.
heimbach@mit.edu
9 ECMWF, Marine Aspects Section, Shinfield Park, Reading, RG2 9AX, UK.
peter.janssen@ecmwf.int
10 Norut, PO Box 6434, 9294 Tromsø, Norway. harald.johnsen@norut.no
11 Norwegian University of Science and Technology, 7491 Trondheim, Norway.
Harald.Krogstad@math.ntnu.no
12 German Aerospace Center (DLR), Remote Sensing Technology Institute,
Oberpfaffenhofen, 82234 Wessling, Germany. Susanne.Lehner@dlr.de
13 Met Office, Exeter, Devon, EX1 3PB, UK. jian-guo.li@metoffice.gov.uk
14 German Aerospace Center (DLR), Remote Sensing Technology Institute,
Oberpfaffenhofen, 82234 Wessling, Germany. Xiao.Li@dlr.de
15 Helmholtz-Zentrum Geesthacht, 21502 Geesthacht, Germany.
wolfgang.rosenthal@hzg.de
16 Helmholtz-Zentrum Geesthacht, 21502 Geesthacht, Germany.
Johannes.Schulz‑Stellenfleth@hzg.de

1. Introduction
The first European Remote Sensing satellite, ERS‑1, launched on 17 August
1991, was conceived in the 1980s, at a time of growing scientific and public
awareness of the need to better understand, monitor and manage the Earth
system. The dangers of environmental degradation, the alarming increase in
the rate of loss of species, and the threat of irreversible climate change had
emphasised the need to treat Earth as an integrated system. Satellites provided
a unique opportunity to obtain the necessary global, continuous datasets. And
an ocean satellite would clearly be needed as a key component of an integrated
satellite Earth observation system in order to monitor the oceans, which cover
more than two-thirds of Earth’s surface, and the ocean−atmosphere interface.
To achieve a 24-hour, all-weather global capability, it was decided that
the main instrument on ERS‑1 would be a multi-component Active Microwave
Instrument (AMI), consisting of a scatterometer, a radar altimeter and a
Synthetic Aperture Radar (SAR). Since they can see through clouds and are
independent of sunlight, microwave systems have the great advantage of being
able to operate under all conditions. The proof of concept of satellite-based
microwave measurements of the ocean had been provided by the NASA−NOAA
satellite Seasat, launched in August 1978. Although Seasat unfortunately failed
through a power short-circuit after only 100 days of operation, initial analyses
had demonstrated that microwaves, reflected or backscattered from the sea

167
SP-1326

surface, could provide valuable data on sea surface heights, sea surface winds,
wave heights and many other important features.
In particular, it had been shown that the SAR could reveal detailed
structural information on a wide variety of ocean phenomena. The SAR is an
advanced radar imaging instrument that achieves high resolution not only in
the range direction, through the standard radar technique of measuring the
travel time of short microwave pulses, but also in the along-track (azimuthal)
direction, by exploiting the Doppler shift induced by the satellite motion. The
achievable azimuthal resolution is then of the order of the antenna length, as
opposed to the inverse dependency on the antenna length for a conventional
radar. The ocean surface features imaged by the Seasat radar included the
propagation directions and wavelengths of surface waves and internal waves,
current-induced surface signatures of bottom topography, and the distribution
of oil slicks, eddies and fronts.
Seasat not only provided a major technical impetus for an improved follow-
on European satellite ERS‑1, but also created a strong motivation for European
scientists to engage in satellite remote sensing of the oceans. The collaboration
with American colleagues had demonstrated to European scientists the
exciting opportunities that all-weather global satellite measurements offered
for studies of the ocean, the coupled ocean−atmosphere system and the
cryosphere. Particularly influential in this respect was John Apel (1930−2001),
a long-standing advocate of ocean remote sensing and a key player in the
Seasat project.
One of the many important applications of ERS‑1 (1991–2000) and its
successors ERS‑2 (1995−2011) and Envisat (2002−12) was in the field of ocean
waves. The launch of ERS‑1 came at an opportune time for ocean wave
research. In the three decades prior to the launch of ERS‑1, detailed theoretical
studies and extensive field programmes had led to major advances in our
understanding of the dynamics and propagation of wind-driven ocean waves.
Numerical wave models, such as the WAM model (WAMDI, 1988), had been
developed to compute the space–time evolution of the two-dimensional
ocean wave spectrum, for given wind field forcing, as a function of period and
propagation direction (for a comprehensive summary of ocean wave research
and modelling, see Komen et al., 1994). What were needed now to drive the
models were good surface wind data, and, to initialise and validate the models,
good wave data. The combined modes of the Active Microwave Instrument of
ERS‑1 were able, in principle, to provide all the data needed: surface winds,
from the scatterometer and Radar Altimeter; mean wave heights, also from the
Radar Altimeter; and the full two-dimensional surface wave spectrum, from
the SAR.

2. The SAR Wave Mode

The SAR, as the most sophisticated of the AMI systems, provided potentially
the most detailed ocean wave information. To reduce the so-called clutter
noise, four single-look SAR images, with a theoretical resolution of 4 m, were
averaged to produce multi-look images of about 20 × 16 m resolution. The SAR
instrument produced millions of images of the ocean surface at this extremely
high spatial resolution. But it also presented two major challenges: achieving
global coverage and retrieving calibrated two-dimensional ocean wave spectra
from the SAR images.
The difficulty in achieving global coverage was that, in continuous
operation, the available data generation rate of the SAR was far too large
for onboard storage of the data and later transmission to a ground receiving
station, after the satellite came in sight of one of the relatively few ground
stations. The problem was resolved by introducing two operating modes for
the SAR: a standard broad-swath mode, which could be switched on when

168
 SAR Wave Mode

Figure 1. ERS‑1 orbits and footprints of the nadir-looking Radar Altimeter and the full-swath SAR of the AMI. The wave-mode SAR was
switched on briefly every 200 km along the satellite track, producing a 5 × 10 km SAR imagette (top right). (From Lehner et al., 2000; DLR)
Bottom right: Example of the standard SAR wave mode product delivered by ESA. (Brooker, 1995)

the satellite was in sight of a ground station, to which the SAR data could
then be transmitted directly, and a global SAR wave mode, designed with
a periodic on/off operation cycle. The SAR wave mode was activated only for
a short period every 200 km along the orbit, producing 5 × 10 km rectangular
‘imagettes’ (see Fig. 1, top right). (In Envisat, which succeeded ERS‑1/ERS‑2, the
separation between imagettes was reduced to 100 km).
The wave mode data were sufficiently reduced relative to the full-swath
data that they could be stored aboard the satellite until the data could be
transmitted to a ground station in sight of the satellite. Although small, the
imagettes, were nevertheless sufficiently large compared with the wavelengths
of ocean waves to retrieve a meaningful statistical ocean wave spectrum.
They were also sufficiently closely spaced to be compatible with the spatial
resolution of current numerical global ocean wave prediction models, as
applied, for example, in the daily global wave forecasts of the European Centre
for Medium Range Weather Forecasts (ECMWF).
The SAR wave mode imagettes were obtained at a mean incidence angle of
23°, within the low-incidence-angle region of the broader scatterometer swath
(see Kerbaol et al., 1998, for a detailed description and analysis). Together with
the Radar Altimeter wave heights measured directly below the satellite orbit
and the winds retrieved from the scatterometer, these data provided a valuable
and unique synergistic dataset that could be used both as inputs to numerical
wave prediction models and as data for model validation.

169
SP-1326

The second basic problem in the SAR imaging of ocean waves was the
interpretation of the images. Seasat had clearly demonstrated that the SAR
could see ocean waves (see Johannessen et al., this volume, for similar full-
swath examples for ERS). But what was the relation between the waves seen
in the image and the real waves in the ocean? Since the standard SAR image
processing assumed that the Doppler was produced only by the satellite
motion, the imaging theory could not be directly applied to random time-
varying surfaces such as the sea surface. In particular, doubts had been
expressed that SAR images could be used for quantitative measurements
of ocean waves due to the orbital motions of the long waves that transport
the short Bragg waves (in the 20 cm wavelength range) responsible for the
microwave backscatter. The additional Doppler shift induced by the long-
wave orbital velocities leads to a misplacement of the inferred position of the
backscattered surface element, resulting in a distortion of the inferred wave
spectrum. In one of the first analyses of these effects, Alpers & Rufenach (1979)
pointed out that, depending on the wave steepness and propagation directions,
some wave components would thereby be enhanced, while others would be
smeared out.
A first convincing demonstration that, despite these concerns, SAR wave
image data could nevertheless provide useful information on wave spectra
was provided by the analysis of the full wave dataset from a Seasat orbit (Fig.
2, left). The inferred centre of high winds that produced the swell agreed well
with the wind field reconstructed from weather data (Fig. 2, right).
Yet this successful reconstruction of the swell sources from SAR wave data
glossed over some of difficulties emphasised by Alpers & Rufenach (1979).
Since the Seasat SAR wave data were uncalibrated, only the wavelengths
and propagation directions of the wave retrievals were used, not the wave
amplitudes. Moreover, the swell fields were well dispersed, corresponding to

Figure 2. Left: Principal propagation directions of swell derived from the SAR wave images of the SEASAT orbit 757, acquired at 22:47 UTC
on 18 August 1978, with the position of the inferred extratropical cyclone. (From Lehner, 1984) Right: Contour plot of surface wave heights
(in metres) derived from a later reanalysis of the ECMWF data (ERA-40) at 18:00 UTC on 17 August 1978.

170
 SAR Wave Mode

narrow directional spectra. In this case, the different weightings assigned by

the SAR to different propagation directions produced no significant distortions.
Finally, the long swell waves were propagating in the near-range direction
(i.e. orthogonal to the satellite flight direction) and had low steepness. In this
case, the distortions induced by the wave orbital motions were small and, if
anything, enhanced the imaging.
ERS SAR data are now widely used to track waves emanating from storm
regions, using both the SAR wave mode and full-swath data (see below). The
SAR enables the routine reconstruction of the high-wind source region of the
swell and decay rates of the swell as it propagates over many thousands of
kilometres, using the basic spectral wave propagation relations first applied by
Barber & Ursell (1948) to infer the locations and times of the distant storms that
produced the swell. The SAR wave retrieval methods have meanwhile become
more sophisticated, yielding well calibrated wave spectra. However, care must
still be taken to allow for the distortions induced by the nonlinearity of the
imaging mechanism, as discussed in the next section.

3. Retrieval of Ocean Wave Spectra from SAR Images

In anticipation of the difficulties of retrieving quantitative wave spectral
data from SAR images not only in individual case studies, but also for the
more general application of operational wave prediction, an extensive field
programme, the Marine Remote Sensing Experiment (MARSEN), was carried
out in 1979. The experiment, involving numerous research aircraft, ships,
wave measurement stations, etc., from Europe and the US, was originally
planned as a Seasat underflight project. However, when Seasat failed in 1978,
it was rescheduled as a field programme in support of the planned European
successor to Seasat.
MARSEN yielded invaluable insights. Among other results, it led to an
extensive multi-author collective analysis of the theory of the SAR imaging of
ocean waves (Hasselmann et al., 1985). This provided the basis for the design
of the ERS‑1 SAR wave mode instrument and the later retrieval of ocean wave
spectra from the wave mode imagettes.
Figure 3 illustrates the impact of velocity bunching on the SAR imaging
of ocean waves. For a small azimuthal component of the wave steepness, the
velocity bunching can be represented by a linear transfer function. This can be
simply added to the two other linear transfer functions characterising the radar
imaging of ocean waves: the hydrodynamic and tilt modulation of the short
backscattering Bragg waves (in the centimetre range) by the imaged long ocean
waves (in the range 10−1000 m). The hydrodynamic and tilt transfer functions
apply equally to real and synthetic aperture radars and are reasonably well
understood. However, the velocity bunching mechanism rapidly becomes
nonlinear when the wave steepness increases and the displacement direction,
which is in the satellite flight direction, is parallel to the propagation direction
of the waves. When these two effects are superimposed, in particular for short
steep waves travelling parallel to the flight (azimuthal) direction, the wave
image can become completely smeared out (see lower part of Fig. 3).
Thus, in order to retrieve quantitative ocean wave spectra from SAR image
spectra, two developments were necessary: a theory of the nonlinear mapping
of an ocean wave spectrum into a SAR image spectrum, and a method for
inverting the nonlinear mapping relation to infer the wave spectrum from the
measured SAR wave spectrum. Both problems were successfully solved, the
first rigorously, the second more approximately, leading to different inversion
schemes (Hasselmann & Hasselmann, 1991; Krogstad et al., 1994; Engen et al.,
1994; Hasselmann et al., 1996; Mastenbroek & de Valk, 2000; Chapron et al.,
2001; Schulz-Stellenfleth et al., 2005; Collard et al., 2009).

171
SP-1326

Figure 3. Velocity bunching: Impact of

wave orbital motions on the SAR-imaged
position of a backscattering surface
element of a wave field. The vertical axis
refers to the average wave steepness. For a
given steepness, the imaged position of an
element on the sea surface is given by the
intersection of the associated ray with the
horizontal line corresponding to the given
wave steepness. p and n indicate the orbital
motions of water particles toward and away
from the SAR platform. (Courtesy of X.-M.
Li, after Hasselmann et al., 1985)

Figure 4. Contour plots of observed and

computed 2D wavenumber spectra.

The original inversion method of Hasselmann & Hasselmann (1991) is

illustrated in Fig. 4. It is based on a maximum likelihood matching of the first-
guess (prior) information available from a wave model and the data provided
by the SAR wave image spectrum. From the first-guess wave spectrum, the
forward transform is applied first to compute the associated SAR wave image
spectrum. This will generally differ from the observed SAR wave image
spectrum. One then constructs a maximum likelihood resultant wave spectrum
and an associated SAR wave image spectrum by linearly combining the two
(generally inconsistent) sets of information. The final maximum likelihood
(posterior) wave spectrum depends on the weights assigned to the two separate
sets of information. It is then possible to effectively disregard the regions of the
SAR image spectrum in which the wave components are strongly distorted by
the velocity bunching mechanism, the unreliable information being replaced
by the predicted first-guess spectrum. The method necessarily depends on the
subjective assessment of the reliability of the first-guess wave spectrum, which
is derived from integrated information on winds and waves over a larger area
around the location of the retrieval, compared with the reliability of the SAR
wave spectral image (which may well have problems apart from the distortions
induced by velocity bunching).
The first generation of retrieval algorithms yielded a wave spectrum
corresponding to a frozen image of the sea surface. They were thus unable to
resolve the 180° directional ambiguity of the wave propagation directions of

172
 SAR Wave Mode

a frozen image without additional information, such as from the first-guess

wave model prediction. However, the required information was, in principle,
available in the multi-look SAR data itself. Instead of simply averaging
four successive looks (in order to reduce the clutter noise), it was possible to
determine small changes between successive looks due to the propagation of
the waves. Technically, the directional ambiguity was removed by computing
not only the spectrum for the averaged looks, but also the complex cross-
spectrum between successive looks (Vachon & Raney, 1991; Vachon & West;
1992). The analytic nonlinear ocean SAR transform based on cross-spectra
and the corresponding inversion was published by Engen & Johnsen (1995).
Following a proof-of-concept using reanalysed ERS‑2 data, cross-spectra were
incorporated in the standard ESA SAR wave mode product for the following
satellite Envisat (Johnsen et al., 1998; Lehner et al., 2000).
The original maximum-likelihood method of Hasselmann & Hasselmann
(1991) (Fig. 4) was generalised accordingly, such as by ECMWF (Abdalla et al.,
2004) and in the PARSA retrieval algorithm of Schulz-Stellenfleth et al. (2005),
to derive two-dimensional wave spectra free from directional ambiguities.
The need to augment SAR wave image data using independent wave data
from other sources – preferably from a state-of-the-art wave model used
for operational global wave prediction – nevertheless remains an inherent
limitation of SAR wave spectral data in all cases in which the azimuthal
spectral cutoff due to velocity bunching cannot be neglected. Attempts to
circumvent this limitation have used two approaches.
The first approach is simply to limit the retrieval at the outset to the
spectral range not seriously affected by the azimuthal cutoff. There exists a
broad range of interesting problems, in particular regarding the propagation of
swell (as illustrated in Fig. 2 and further examples discussed below), for which
this approximation is quite acceptable. The second approach is to substitute
the detailed spectral information provided by a state-of-the-art wave spectral
model by auxiliary information obtained, for example, from scatterometer-
derived surface winds or Radar Altimeter wave heights. This leads to the
difficulty, however, that the missing azimuthal cutoff data are of a specific
spectral structure, whereas the substitute data necessarily represent spectral
integrals that are unable to resolve the missing spectral information.
Thus, in summary, the various proposed SAR wave spectrum retrieval
algorithms can be classified with respect to two characteristics. First, whether
they are first-generation retrievals (‘1’), based only on the standard image
variance spectrum, or second-generation retrievals (‘2’), using inter-look
cross-spectral data. Second, whether they use state-of-the-art wave model
predictions as first-guess input data (M), are limited to swell-type spectral
data (S), in which the spectral components affected by the azimuthal cutoff
can be neglected, or, finally, use auxiliary data instead of a first-guess model
to augment the spectral region affected by the cutoff (A). An example of
retrieval type 1S is the swell study of Lehner presented in Fig. 2; examples of
the application of the other retrieval types will be discussed in subsequent
sections.
Parallel to these different development strands, all retrieval algorithms
have benefited from continuous technical improvements. These include
improved formulations of the imaging transfer functions, special retrieval
algorithms for selected integrated spectral properties, analysis of image
pixel statistics, and investigations of the signal-to-noise and image contrast
problems associated with the inter-look cross-spectral analysis method used to
remove the directional ambiguity (Schulz-Stellenfleth et al., 2005).

173
SP-1326

4. Comparison of Retrieved SAR Wave Spectra with

Operational Wave Model Predictions
A comprehensive comparison of three years of ocean wave image spectra
retrieved from ERS‑1 SAR with the wave spectra predicted by the ECMWF
operational global wave model was undertaken by Heimbach et al. (1998);
see Fig. 5. This provided the first quantitative assessment of the quality of the
ECMWF wave forecasts over a continuous period, covering all seasons and
on a global scale. A similar study on a regional scale, using the Norwegian
operational wave prediction model, was undertaken by Breivik et al. (1998). In
both studies a type 1M retrieval algorithm was applied.
The availability of a first-guess wave field was important for the retrieval of the
wind sea spectrum (Fig. 5, left panels), but not critical for the swell components
(right panels), which consisted largely of well dispersed wave systems, in which
the velocity bunching could be linearised. The retrieval algorithm was upgraded
in this analysis by subdividing the wave spectrum into a number of separate
wave systems, which were then inverted individually (Hasselmann et al., 1996).
In a complementary study, Bauer & Heimbach (1999) compared SAR-retrieved
significant wave heights (using the same retrieval algorithm) against those
obtained independently from the altimeters on ERS‑1 and Topex/Poseidon. As in

Figure 5. Seasonally and ocean-basin averaged polar diagrams of wave heights and directions (defined analogously to the standard wind
vector polar diagrams of operational meteorology). Left: Wind sea spectra. Right: swell. Solid lines: ERS‑1 SAR wave mode data; dashed
lines: ECMWF-WAM model predictions. (From Heimbach et al., 1998)

174
 SAR Wave Mode

other applications of type 1M retrieval algorithms using first-guess wave spectra

from a wave model, the directional ambiguity presented no serious limitation, as
this could be resolved by the first-guess data from the wave model.
The overall agreement between the SAR-retrieved and predicted wave data as
well as with independent wave measurements was clearly very encouraging. At
this stage, no attempt had been made to assimilate the retrieved SAR wave data
in the wave model. Thus, the investigation was a test of the mutual consistency
of the wave model prediction, driven by winds produced by the standard ECMWF
weather analysis, and the wave spectra retrieved from the SAR. Despite the good
overall agreement, systematic seasonally and regionally dependent deviations
between the model and retrieved polar wave height diagrams are clearly visible.
At that time, a tendency was revealed for the ECMWF model to predict too high
wind seas and too low swell, indicating that both the wind input and the swell
dissipation source terms in the wave model needed to be modified. This has led
to successive improvements in the wave model.
With the later model updates and the assimilation of SAR wave spectral data
and Radar Altimeter wave heights, the ECMWF operational wave prediction
model has now reached a level of maturity (see Figs 6–9) that the current

Figure 6. Global comparison of inverted

Envisat/ASAR Level 1b data and WAM model
results for significant wave heights, 2009
(Abdalla et al., 2010). (Peter Janssen,
ECMWF)

Figure 7. Global comparison of inverted

Envisat/ASAR Level 1b data and WAM
model results for mean wave period, 2009
(Abdalla et al., 2010). (Peter Janssen,
ECMWF)

175
SP-1326

Figure 8. Global comparison of inverted

Envisat/ASAR Level 1b data and WAM model
results for mean directional spread, 2009
(Abdalla et al., 2010). (Peter Janssen,
ECMWF)

Figure 9. Impact of various assimilation

setups (ERS‑2 RA and SAR) on the bias
and rms error of a one-dimensional wave
spectrum, verified against in-situ buoy
observations, 1−29 May 2001.
(Abdalla et al., 2004)

operational SAR wave data assimilation scheme has only a minor impact on
the skill of the prediction (Fig. 9). However, this level of maturity could not
have been reached without the two-dimensional wave spectra provided as
verification data by the global SAR wave mode system.
The above comparisons of retrieved and predicted wave spectra were still
based on a first-generation type 1M retrieval algorithm, in which the directional
ambiguity of the SAR image spectra had not been removed. Figures 10 and
11 show similar comparisons using the type 2M PARSA retrieval algorithm.
Figure 10 illustrates that in this case it is meaningful to compare not only the
directional spread (as in Fig. 8), but also the mean direction itself. The impact
of combining model and retrieved fields in the simultaneous determination of
both the retrieved and predicted wave spectra is illustrated in Fig. 11, which
shows comparisons of retrieved and predicted wave heights for two models,

176
 SAR Wave Mode

Figure 10. Comparisons of the PARSA mean frequency (left) and mean wave direction (right) with those of the ECMWF reanalysis wave
model. (Li et al., 2010)

Figure 11. Comparisons of the PARSA significant wave heights with the results of the ECMWF reanalysis wave model (left) and the DWD
forecast wave model (right). (Li et al., 2010)

ECMWF and the German Weather Service (DWD), both of which are based on
the same basic physics.
The examples differ in two respects, however. First, as input first-guess
spectra, the same predicted ECMWF spectra were used in both cases. Thus in
the DWD case, the input data came from another model. Second, in the ECMWF
prediction, in contrast with the DWD prediction, earlier SAR spectra had been

177
SP-1326

assimilated in initialising the wave spectrum for the subsequent wave model
integration. Thus, it is not surprising that the deviations between the predicted
and retrieved wave spectra are almost 50% smaller for the ECMWF model than
for the DWD model. Attention must clearly be given to the relative weights
assigned to the predicted first-guess spectrum and the SAR image spectrum
both in the maximum likelihood retrieval algorithm and in the assimilation
scheme when comparing predictions and observations. We return later to
the important question of the advantages and disadvantages of combining
prediction, retrieval and assimilation in a single operation, but first we present
results in which no first-guess input wave spectra are used.

5. SAR Retrievals of Swell

In addition to the retrieval of the full wave spectrum, valuable information
on the dynamics of ocean waves can be obtained from investigations of swell
waves emanating from high-wind source regions. These are particularly
suitable for global monitoring with SAR, as they propagate over large oceanic
distances. Furthermore, if not too near the source of origin, they normally
have sufficiently small steepness that the nonlinearities induced by velocity
bunching can be neglected. Thus retrieval algorithms of type 1S or 2S can be
used, independent of first-guess model predictions.
Following the early Seasat example of Lehner (1984; Fig. 2), Holt et al.
(1998) reported similar results for ERS‑1, in which swell waves emanating from
an intense storm were tracked over a period of several days. Comparisons of
SAR and buoy data indicated that SAR-derived peak wavelength and direction
measurements could be used reliably to predict arrival times and propagation
direction over large distances. A classic field experiment conducted by
Snodgrass et al. (1966), who studied swell propagation along a great circle, and
its attenuation during its crossing of the Pacific, was repeated by Heimbach
& Hasselmann (2000). This time the authors used ERS‑1 SAR retrievals over a
10-day period (Fig. 12, top right panel), together with spectra from ECMWF’s
WAM model collocated in time and space (top left panel) to assess the simulated
swell properties. The retrieved and simulated spectra were traced back to
their region of generation by a major storm off New Zealand (Fig. 12, bottom
panels). A detailed comparison of the travel times of simulated and retrieved
swells (in terms of their implied group velocities) revealed an underestimation
of ECMWF’s wind speed used to drive the wave model, a result that was
confirmed independently by scatterometer data.
After 2002, the availability of operational complex SAR wave mode spectra
from the Advanced SAR (ASAR) on Envisat stimulated the wider use of type 2S
wave spectral retrieval algorithms, in which the directional ambiguity had been
removed and the nonlinearities of velocity bunching (beyond the linear response
approximation) were simply ignored. Since a first-guess wave spectrum, as
in a type 2M algorithm, was now no longer needed to remove the directional
ambiguity, but only to supply the information lost through the azimuthal cutoff
by velocity bunching, the latter was simply ignored by restricting the application
to spectra, such as swell, in which these effects were small.
Extensive comparisons of buoy data and the Envisat/ASAR wave mode
Level 2 products (based on a mixed type 2S–2A retrieval algorithm; see
Johnsen et al., 2006) were also carried out by Li & Holt (2009) using the UK
operational wave model to interpolate between the locations and times of SAR
and buoy measurements (Fig.  13). Figure 14 shows the comparison of model
and Envisat/ASAR wave mode Level 2 products wave height at the Christmas
Island (CI) buoy site in different spectral bins, i.e. wave periods. The agreement
was satisfactory except for a median range of typical wind sea wave periods
(10−16 s), in which the SAR retrieved wave height in this case was 25% higher
than the buoy data.

178
 SAR Wave Mode

Figure 12. Simulated swell (top left) and

SAR wave mode (SWM) retrieved swell
(top right) propagating across the Pacific
over a 10-day period. In the bottom panels,
both simulated and observed swells are
traced back to their regions of generation.
(From Heimbach & Hasselmann, 2000)

Similar comparisons and cross-validations of the operational wave model

against the ASAR wave mode Level 2 products have also been carried out by
Météo-France. The Météo-France spectral retrieval is based on a partitioning
and optimal interpolation algorithm applied to individual wave systems. The
assimilation of one year (September 2005 to August 2006) of ASAR Level  2
wave products in the wave model WAM cycle 4 (Aouf et al., 2008) reduced
significantly the rms error of the significant wave height. The impact of
the assimilation is strongest in the tropics, where the sea is dominated by
swell, and the weakest component of the wave model. Compared with the
assimilation of Radar Altimeter wave heights, the impact of the assimilation of
ASAR Level 2 wave spectra is most pronounced for longer-range forecasts from
12 to 60 h (Fig. 15).
Figure 16 shows a further example of the application of a type 2S retrieval
algorithm, in this case to investigate the decay properties of long swells with
periods of 13–18 s (Arduin et al., 2009). These wave components were and still
are the largest source of error in global numerical wave models. The spectra
from individual imagettes are sufficiently accurate to infer the dissipation
rates of swell systems. Moreover, the large-scale consistency of the swell fields
can be exploited to reach even more accurate estimates of characteristic wave
parameters, by averaging wave properties in space and time. Ardhuin et al.
(2009) found that the dissipation rates of swells were surprisingly nonlinear,
with a stronger dissipation of steeper swells, and no clear dependence on the
wave period or wind speeds (Fig. 16). This nonlinearity is consistent with the
estimated strong dissipation rates for very steep swells derived by Högström et
al. (2009).

179
SP-1326

Figure 13. ASAR-derived versus buoy swell

partition heights after bias correction, for
the years 2004−8. The solid line joins the
median values from SAR observations in
each 0.1 m class of buoy-measured heights.
(From Collard et al., 2009)

Figure 14. Comparison of model and Envisat

ASAR wave heights at the CI buoy site in
four spectral bins. (From Li & Holt, 2009)

Building on this observation, Ardhuin et al. (2010) used an analogy with

the bottom boundary layer to parameterise nonlinear swell dissipation. The
parameterisation is now used operationally by Météo-France and the Argentine
Navy, and is being implemented by NOAA’s National Weather Service and the
National Centers for Environmental Prediction (NCEP) in the United States
(Tolman et al., 2011).

180
 SAR Wave Mode

Figure 15. Variation in the assimilation

index over the forecast period compared
with Jason-1 wave heights. The red
line indicates the assimilation of Radar
Altimeter RA-2 wave heights only, while
the green line refers to the assimilation of
both RA-2 and ASAR Level 2 wave spectra.
The assimilation index is the reduction
of the rms error relative to the reference
run without assimilation. (From Aouf et al.,
2008)

Figure 16. Swell dissipation for 22 swell

events in the Pacific: estimated linear
attenuation coefficient as a function of
the swell significant slope (ratio of the
swell significant wave height and peak
wavelength, s = 4 Hs/L), taken 4000 km
from the storm centre, for a variety of peak
swell periods (colours). (Ardhuin et al.,
2009)

6. Globally Integrated versus Local Retrieval

Algorithms
It is undisputed that second-generation retrievals, which use the complex
information of the image cross-spectra to remove the directional propagation
ambiguity, are inherently superior to first-generation retrievals using only SAR
image variance spectra. However, this advantage is not pronounced in the
combinations 1M and 2M, in which the first-guess input wave spectra from a
state-of-the-art wave model can provide the necessary directional information
(see, for example, Fig. 9, based on a type 1M retrieval algorithm).
The main advantage of second-generation retrievals comes to bear either
in retrievals of type 2S, applied to swell-type spectra in which the azimuthal
cutoff due to velocity bunching is insignificant (Figs 13−16), or in retrievals of
type 1A and 2A, where the information lost through the azimuthal cutoff is

181
SP-1326

replaced by auxiliary data from sources other than a wave model. But even
in retrievals of type 1S or 2S, the directional ambiguity is often not a serious
problem, if the propagation directions (as in the examples presented above)
can be clearly inferred from the overall geometry.
The main issue in the choice of retrieval algorithm therefore arises when
the azimuthal cutoff becomes important, namely for steep, high-wind seas. The
question is whether it is best in this case to use a first-guess wave spectrum
provided by a state-of-the-art wave model (type 1M and 2M retrievals), or to
try to substitute for the missing azimuthal cutoff information using other data
(type 1A and 2A retrievals), such as from a radar altimeter, a scatterometer
(although this instrument was no longer available on Envisat) or from wave
buoys. The answer depends on whether the goal is to provide the best possible
estimate of the wave spectrum, or to improve the wave model.
If the goal is to provide the best possible estimates of the wave spectrum
for operational predictions or hindcasting studies, the answer is that one
should clearly attempt to make optimal use of all available data in a global
data assimilation scheme, including not only state-of-the-art wave models,
but also state-of-the-art operational weather forecasting models, and all forms
of observational data, including from satellites and conventional observing
systems.
As an example, Figs 17 and 18 (Li et al., 2010) show the impact of
assimilating first-guess model data on the prediction of a complex sea state
in an intense storm region. It is clear that without a reasonably realistic first-
guess input, it would have been impossible to derive the complex spectra
shown on the right in Fig. 18 from the ESA WVW wave mode products shown
in the middle column, which exhibit a pronounced azimuthal cutoff. But it
is equally evident that the first guess for this complex storm system exhibits
systematic errors that could not have been corrected without the information
from the SAR. Thus a retrieval of the wave spectrum from the complex SAR
image spectrum requires simultaneous information from both the SAR and the
wave model.
On the other hand, if the goal is to test and improve the model, one should
clearly strive to use independent data that have not been used in the model
prediction. Unfortunately, this is feasible for SAR wave data only if one restricts
the investigation, as in the examples of Figs 13 and 16 above, to type 1S and 2S
retrievals, in which the nonlinear azimuthal cutoff region is excluded. In the
case of strong wind sea spectra, as in the example just shown, the important
spectral information lost beyond the azimuthal cutoff cannot be provided by
model-independent, spectrally integrated observations. For this reason, most
of the various proposed type 1A and 2A retrieval algorithms − e.g. Mastenbroek
& de Valk (2000), CWAVE (Schulz-Stellenfleth et al., 2007), CWAVE_ENV (Li
et al., 2011) − were not developed to test and improve state-of-the-art wave
models, but rather simply to retrieve characteristic wave parameters.
Although these retrieval algorithms cannot therefore compete with type
1M and 2M state-of-the-art retrieval algorithms for detailed spectral wave
measurements, or with type 1S and 2S algorithms for testing and improving
state-of-the-art wave models, they nevertheless represent a valuable
complement to these more sophisticated retrieval algorithms in providing
generally accessible, simple tools for useful exploratory studies. As an
example, Fig. 19 shows the validation of the operational wave model of the
German Weather Service (DWD) against a combination of Envisat RA-2 data
and ASAR-retrieved wave heights using the CWAVE_ENV algorithm.
In the long term, the SAR wave mode data could be seen as but one
component of a highly complex satellite. The satellite, in turn, is only conceived
as one component of an integrated Earth observation system, consisting of
several satellites and an extensive ground-based conventional observing
system. To make full use of this extensive global dataset, we need to develop
sophisticated integrated models that also treat Earth as a complex coupled

182
 SAR Wave Mode

Figure 17. Locations of predicted and

observed wave spectra at stations A, B
and C within and south of an intense
cyclone in the Pacific on 12 April 2009.
(Li et al., 2010)

Figure 18. ECMWF predicted spectra (left),

ESA SAR wave mode spectra (WVW) (centre)
and PARSA retrieved spectra (right) for the
three stations A (top), B (middle) and
C (bottom) shown in Fig. 17. (Li et al., 2010)

system. In particular, the ocean wave field should be treated as an integrated

component of the global weather system. The long-term goal should therefore
be to assimilate the SAR wave data together with all other relevant sources of
data (radar altimeter wave heights and winds, standard meteorological data,
ocean buoy data, etc.) within a state-of-the-art integrated global atmospheric
and ocean wave forecasting system. Within this general framework, a future
ERS−Envisat−Sentinel integrated assimilation system should be extended to
include also ocean currents (see section 8).

183
SP-1326

Figure 19. Validation of the DWD forecast wave model using simultaneous wave measurements of the Envisat/ASAR and RA-2. Centre:
The forecast wave field at 0:00 UTC on 20 January 2007 (background) and significant wave heights (SWH) retrieved from the ASAR wave
mode data using the CWAVE_ENV algorithm (right track), and the collocated RA-2 SWH data (left track), both acquired between 23:39 and
23:51 UTC on 19 January 2007. The right and left panels show the model predicted SWH compared with the ASAR and RA-2 retrievals,
respectively. (Courtesy of X.-M. Li)

In summary, the 20 years of SAR wave mode data provided by ERS‑1

(1991−2000), ERS‑2 (1995–2011) and Envisat (2002−2012) have had a major
impact on ocean wave research and operational numerical wave prediction.
The satellites produced invaluable data on the two-dimensional ocean wave
spectrum that no other instrument was able to provide at a comparable level
of spectral resolution and global, continuous coverage. The dataset has been
indispensable for improving and validating the global wave model WAM
(WAMDI, 1988), which is now applied in over 200 research and operational
forecasting centres worldwide, as well as for improving similar wave prediction
models operated by the French, UK and US weather services. Implemented
together with the radar altimeter wave heights in the data assimilation mode,
the wave mode SAR data have significantly improved, and can be expected to
further improve, the quality of wave prediction.

7. Ongoing and Future Research

Apart from their invaluable contribution to the determination of global ocean
wave spectra, SAR wave mode data also have other applications that are the
subject of current and planned future research. These include investigations
of extreme sea states in intense storms, studies of individual exceptional
wave states, such as freak (monster) waves, and the extension of numerical
wave models in the general context of Earth system modelling, including
the improved representation of wave–current interactions, air–sea fluxes,
and second-order spectral quantities related to infra-gravity and microseism
generation.

184
 SAR Wave Mode

7.1 Extreme Sea States

Both tropical cyclones, which are associated with the most intense winds
and air–sea fluxes, and extratropical cyclones, which produce most of the
highest sea states, are difficult to observe directly due to sampling and signal
saturation issues. Yet, these storms define the design criteria for most ocean
engineered structures and coastal defences. Satellites provide a unique
opportunity for overcoming these sampling problems. For instance, the largest
significant wave height recorded by a satellite altimeter reached 20.1 m when
storm Quirin crossed the North Atlantic in February 2011. A joint analysis of the
remotely sensed winds and waves, including Envisat/ASAR data, confirmed
the existence of very long waves, with peak periods reaching 25 s, supported
by in situ buoy measurements and the analysis of seismic noise at land-based
stations (Hanafin et al., 2012).
To understand the dynamics of such extreme wave systems, it is important
to observe both the conditions within the high-wind region itself and the
long swell components emanating from the storm, which allow conclusions
to be drawn on the processes within the storm itself (see examples presented
in Figs 14–16). Data for both regimes can be provided by the global SAR wave
mode observations. A careful combination with seismic noise data should
yield further insights into the evolution and statistics of these extreme events
(Grevemeyer, 2000; Aster et al., 2008; Ebeling & Stein 2011).
The evolution of the wave spectrum within an intense storm region is
governed by four processes: wave propagation, the energy transfer due to
resonant nonlinear wave–wave interactions, wind generation and dissipation
by white capping and turbulence (Komen et al., 1996; Janssen, 2008). Of these,
the first two are completely understood theoretically from first principles
and are free from empirical constants, while the last two are represented in
wave models by empirical expressions tuned against extensive observed
wave growth data. The most important process for the structure of the wave
spectrum is the nonlinear energy transfer, which redistributes the energy
within the spectrum through resonant interactions between wave quadruplets
and is responsible for the formation of a sharp spectral peak that gradually
shifts to lower frequencies (longer wavelengths) in a growing wind sea.
Unfortunately, although the nonlinear energy transfer is well understood
theoretically, and has been computed numerically in a number of case
studies, the exact transfer expression is represented by a three-dimensional
integral that is impossible to compute exactly for each frequency, direction
and spacetime point of a numerical wave model. Most models therefore use
a strongly simplified expression, the Discrete Interaction Approximation
(DIA) based on a single dominant wave quadruplet (Hasselmann et al., 1985)
that reproduces the principle properties of a growing wind sea in a uniform
wind field, but is known to exhibit serious deficiencies under more complex
wind conditions. With the advances in computing power since the DIA was
first introduced, a number of improved versions of DIA including more basic
interaction quadruplets have now been proposed (e.g. van Vledder, 2001;
Tolman, 2004, 2009, 2011; Ueno & Kohno, 2004; Tamura et al., 2008, 2010). The
data derivable from the ESA SAR wave mode using type 2M retrieval algorithms
provide excellent new opportunities for testing the impact of these suggested
improvements, in combination with the assumed wind input dissipation
source functions, under realistic extreme wind conditions (as illustrated by the
example in Figs 17 and 18).
Nonlinear interactions, wind action and dissipation processes can still
be important a few storm diameters outside an active generating area. They
lead to a redistribution of the wave energy leaving the storm (Snodgrass et
al., 1966), which can then be detected in the swell propagating away from the
storm over very large distances. The swell systems can be readily observed
with the SAR using simpler type 2S retrieval algorithms. Apart from avoiding

185
SP-1326

the complexities of 2M retrievals associated with the admixture of predicted

and observed data, these measurements have the advantage that the swell
systems can be tracked over much larger areas of the ocean. As well as
reconstructing the location and time of the storm, the measurements can then
provide valuable indirect information on the processes active within and in
the vicinity of the storm area. Preliminary investigations (Delpey et al., 2010)
have found, for example, that swell can be detected with the SAR wave mode
propagating away from the storm in directions perpendicular to the principal
swell propagation direction. It is not yet clear whether this is controlled by the
variability of the wind fields in a storm, by the variable directional spectral
shapes in the storm, or by directional spreading through nonlinear wave–wave
interactions. A combination of storm-related swell data using type 2S SAR
retrievals and type 2M retrievals for the storm area itself could yield important
new insights into the nature of extreme sea states in severe storms.

7.2 Freak Waves

The occurrence of extreme individual waves, called freak or monster waves, has
become a major concern because of the rapid increase in global trade and the
concurrent growth in global shipping using ever larger container vessels. It has
been estimated that in recent years an average of about one super-container ship
carrying many thousands of containers has been lost per month. Although the
cause of the loss is often unknown, it is suspected that in many cases it was due
to an unexpected encounter with an individual monster wave (Fig. 20).
Although long attributed to the narrative lore of seamen, the reality
of monster waves has now been clearly established through many well
documented reports. While they are obviously difficult to interpret directly due
to the intrinsic nonlinear imaging property of the SAR instrument, unusual
wave formations have also been identified in SAR wave mode images, opening
up the possibility of obtaining information on the probability of the occurrence
of freak waves using the global observing capability of the SAR wave mode.
However, to relate SAR data to freak waves, three major problems must first
be resolved: What is the physical structure of a freak wave? Given the physical
structure of a freak wave, how is this very nonlinear, rapidly moving individual
surface feature mapped into a SAR image? How can one invert the mapping to
recover the original freak wave from the SAR image?
Regarding the first question, the simplest hypothesis is that a freak wave
can be described to first order as the straightforward linear superposition of

Figure 20. Monster wave encountered by

the NOAA ship Discoverer in the Bering Sea,
1979. (Commander Richard Behn, NOAA
Corps)

186
 SAR Wave Mode

the individual wave components of a random sea state. Freak waves would
then obey Gaussian statistics. However, the available evidence suggests that
observed monster waves can be considerably higher than can be reasonably
expected from Gaussian statistics, and must thus therefore represent a
basically nonlinear phenomenon. Various nonlinear models based on the
Benjamin−Feir wave−wave interaction instability mechanism (Janssen, 2003)
or on wave–current interactions have been proposed, but it is likely that a
number of different nonlinear interaction configurations can produce freak
waves (Dysthe et al., 2008).
Given the structure of a freak wave, including the orbital velocity field and
its interaction with the short backscattering Bragg waves superimposed on the
large-scale orbital velocity field, the mapping into a SAR image is in principle
well defined (although for ERS SAR wave mode incident angles of only 23°, the
standard Bragg backscattering theory must presumably be augmented for steep
range-travelling waves by a specular reflection model). However, the computation
of the resultant image is computationally demanding. Unfortunately, the statistical
closure methods applied in the fully nonlinear forward-mapping algorithm of
Hasselmann & Hasselmann (1991) and Krogstad (1992) for wave spectra cannot be
simply carried over to the case of individual wave representations.
However, an approximation using a linearised Fourier transform relation
for individual images, in which a filter is applied to the azimuthal cutoff
region (corresponding to a type 2S algorithm, but applied now to the image
itself, not to the image spectrum), has been proposed by Schulz-Stellenfleth
& Lehner (2004). Figure 21 shows an example of the sea surface elevation
field (B) derived from the calibrated ERS‑2 wave mode data (A) using this
algorithm. The algorithm has been applied to two years (September 1998 to
November 2000) of reprocessed ERS‑2 SAR wave mode data to derive a global
map of maximum (SAR estimated) single wave heights (Fig. 22). Although not
applicable for quantitative assessments of individual wave heights, due to the
neglect of the nonlinear distortions induced by the SAR imaging system, this
provides nevertheless a useful tool for arriving at first estimates of the statistics
of exceptional individual wave occurrences as seen by the SAR.
Another application of SAR-derived wave images is the investigation of
wave groupiness (Borge et al., 2004; Niedermeier et al., 2005). Figure 23 shows
an example of pronounced wave groupiness occurring in the vicinity of the
Ekofisk oil platform in the North Sea.
The third and final challenge is to invert the freak wave → SAR image
mapping relation in order to recover the properties of the original monster wave
from the SAR freak wave image. This will depend, of course, on the freak wave
model and will presumably involve some form of parametric inversion method.
Without progress in addressing the first two problems, however, predictions on
advances on this question must remain speculative.
Even after a complete three-stage freak wave model → SAR freak wave
image → SAR image inversion scheme has been developed, one then faces
the further difficulty that the probability of simultaneously observing a
freak wave both in reality and in a SAR wave image in order to validate the
model is for all practical purposes negligible. However, there is an interesting
exception: the monster waves that occur at a few locations famous for their
exceptional surfing conditions (Fig. 24). These huge surf waves are produced
by refraction at subsurface reefs and are generated under well defined,
reproducible and measurable wave conditions. Although representing a special
type of monster wave, they may provide an ideal dataset for testing theories
of the mapping of monster waves into SAR images, as well as for developing
the associated inversion algorithms. Under less extreme conditions, coastal
wave transformation due to complex bathymetry changes can also be used to
interpret and better understand SAR observations involving wave focusing and
refraction effects (e.g. Collard et al., 2005).

187
SP-1326

Figure 21. SAR ocean wave retrieval in the

spatial domain. A: A 5 × 10 km normalised
ERS‑2 SAR wave mode imagette acquired at
48.45°S, 10.33°E on 27 August 1996.
B: The retrieved sea surface elevation field.
(From Schulz-Stellenfleth & Lehner, 2004)

Figure 22. Estimated maximum individual

wave heights inferred from ERS‑2 SAR wave
mode data acquired between September
1998 and November 2000. (Koenig et al.,
2007; DLR)

In summary, most SAR wave observations have been motivated by the

desire to improve our understanding and prediction of wave spectra, with
less attention paid to the imaging of the individual wave fields themselves.
However, the representation of individual waves can yield important
information not only on freak waves, but also on other phenomena, such as
wave grouping, the transformation of waves in the surf zone, and non-Bragg
microwave returns from fast scatterers or reflectors, for example, from white
caps or steep waves.

188
 SAR Wave Mode

Figure 23. Analysis of wave groupiness in

the area of the Ekofisk oil platform in the
North Sea. (From Borge et al. 2004)

Figure 24. A monster wave off Hawaii. Such

surf waves pose no immediate danger to
shipping, but are ideal for testing theories
of the mapping of monster waves into SAR
wave images, together with their associated
inversion algorithms. (EPA)

7.3 Wave–Current Interactions

Ocean waves propagating through strong, variable currents can undergo
dramatic changes in wave height and steepness. In a steady shear current,
wave rays are bent, the bending being proportional, in the approximation of
geometric optics, to the ratio of the vertical component of the current vorticity
to the wave group velocity. Thus refraction effects become important whenever
surface gravity waves propagate through major ocean currents exhibiting
high gradients. Particularly critical are frontal zones, which can lead to wave

189
SP-1326

reflections, trapping and waveguide propagation of the trapped wave within

the current. The resulting wave field can become very complex, with wave
systems following different paths, producing crossing seas and caustics. The
resultant strong spatial variations in wave energy can seriously endanger
navigation. The effects have been observed in numerous SAR images; Liu et
al. (1994), for example, carried out an empirical analysis of wave ray refraction
patterns inferred from an ERS‑1 SAR image over an ocean eddy.
Most of these data have not been supported by synchronous measurements
of surface currents.
Today, however, routine ocean circulation models are available that assimilate
multi-satellite altimeter sea surface height data, thereby better characterising
the mesoscale upper-ocean dynamics and enabling the incorporation of wave
refraction effects in wave models. Wave–current investigations using SAR
observations can furthermore be enriched through an emerging analysis
capability to infer ocean surface velocities directly from the Doppler information of
the SAR data themselves (Chapron et al., 2005; Johannessen et al., 2008).
As an example, Figs 25 and 26 illustrate the deflection of waves by large
surface current gradients in the Agulhas current off the coast of South Africa,
detected using the Envisat wide-swath SAR data. The currents (Fig. 25) were
determined using the SAR Doppler information, while the wave spectra
(Fig. 26) were obtained using a 1S retrieval algorithm (2S retrieval algorithms
free from directional ambiguities were not directly available for the full-swath
SAR data, as ESA provides the required complex image spectra only for the SAR
wave mode imagette data). The incoming swell propagation rays are seen to
be strongly deflected by the large current gradients, producing crossing-swell
systems from an initially uniform incoming swell system within only 200 km.
The problem of wave–current interactions is closely related to that of
wave–wave interactions, and both can be invoked to explain the generation of
spacetime caustics and anomalous wave systems, including freak waves as an
extreme case. In general, waves are scattered and amplified when propagating
in an inhomogeneous random medium, which can represent a nonuniform,
slowly varying current, as in the above example, or the low-frequency, low-
wavenumber random currents that are continuously generated in a random sea
state by quadratic difference interactions between different wave components.
From this perspective, the Benjamin–Feir instability invoked to explain the
generation of extreme waves (Janssen, 2003) can be placed within the general
theory of unstable wave–wave interactions (Hasselmann, 1967) in a form
analogous to the wave–current interaction mechanism considered above.
Finally, quadratic wave–wave interactions between wave components of nearly
the same frequency propagating in opposite directions generate microseisms
(Longuet-Higgins, 1950; Hasselmann, 1963). Precisely measured microseismic
records can be used to determine the arrival times of different swell components.
Using geometric optics, the source can be determined with high accuracy by
triangulation from several seismic stations (Fig. 27; Husson et al., 2012).
Waves propagating in opposite directions with the same frequency occur
only infrequently in the open ocean and are usually an indication of an
unusual storm event, such as an intense tropical or mid-latitude cyclone, or
possibly effects of multiple wave–current refractions. As recently demonstrated
(Ardhuin et al., 2010), numerical model of microseism generation by random
ocean waves, including ocean wave reflections, are becoming increasingly
realistic. Yet, the development of microseism observations into a reliable
early warning system and/or the full exploitation of available 50 years of
seismic data archives still require more detailed knowledge of the directional
properties of ocean wave spectra than is presently achievable using the energy
transfer parameterisations (as discussed above) applied in current state-of-the-
art wave prediction models. The extensive global directional wave information
made available by the two-dimensional ERS−Envisat wave mode data should
again prove invaluable in investigating this attractive option.

190
 SAR Wave Mode

Figure 25. Envisat/ASAR-derived currents and wave observations in the Agulhas current (left), confirmed by the altimetry-derived surface
flow field (right). A southwesterly swell system interacts first with the Agulhas current, flowing toward the west, and subsequently with
the Agulhas return current, flowing and meandering towards the east. The principal swell propagation directions (left) are displayed as
background below the sketched ASAR Doppler-derived current pattern (full curves). (CLS)

Figure 26. Wave spectra for three imagettes corresponding to the three locations shown in Fig. 25: A (left), B (middle) and C (right).
The 300 m swell system is strongly influenced by the large surface current gradients, and the two swell patterns of comparable energy
eventually cross. (CLS)

191
SP-1326

Figure 27. Location of an expected storm

source from independent SAR wave mode
products and seismic noise measurements.
For each seismic station, a 1000-km wide
blue disc has been plotted with a radius
equal to the distance to the storm source,
as estimated from the differential arrival
time of the swell at the seismic station.
The different discs intersect southwest of
New Zealand, in agreement with the storm
location given by the SAR (the red dot is the
location at which backward-propagated SAR
measurements converge). (From Husson
et al., 2012)

8. Conclusions and Outlook

The sequence of ESA satellites ERS‑1, ERS‑2 and Envisat deployed over the
last 20 years has had an immeasurable impact on oceanography, meteorology
and climate research. We have focused in this chapter on the dynamics and
prediction of ocean waves, in particular on the information provided by the
ERS SAR wave mode component of the Active Microwave Instrument (AMI).
However, since ocean waves are generated by the wind, interact with ocean
currents, generate microseisms, affect coastal and offshore structures, and
influence shipping and navigation, an improved understanding and prediction
of ocean waves is important for many other aspects of the Earth system. Thus,
the ERS SAR wave mode, designed to provide global measurements of the two-
dimensional ocean wave spectrum, must be seen as an integral component of
a global Earth observation scheme, consisting not only of the full complement
of ERS instruments, but also of many other satellites providing data on
ocean currents, surface winds and other oceanic and atmospheric variables,
including also the extensive system of conventional in situ measurements.
The impressive series of publications based on the ERS SAR wave mode
instrument (of which only a selected subset has been cited here) illustrates
the complexity of the problems that need to be addressed in order to exploit
the full potential of the instrument. The ERS satellite was conceived in the late
1970s, when the oceanic community had just developed realistic numerical
models of the two-dimensional ocean wave spectrum for operational global

192
 SAR Wave Mode

wave prediction and eagerly welcomed the opportunity to feed these models
with appropriate wave spectral data from a globally observing satellite.
However, although Seasat had already demonstrated that a SAR could image
ocean waves, it soon became apparent that the imaging process was highly
complex. For a quantitative interpretation of the ERS SAR wave images,
detailed models of the electromagnetic return from a random, moving surface,
including, in particular, the interactions between short backscattering
Bragg waves in the decimetre range and the imaged long waves in the 100–
1000 m range were needed. These were soon developed. Methods were then
needed to invert the mapping relation, and retrieve the two-dimensional
wave spectra from the satellite image spectra. This task was also achieved.
However, because the general forward-mapping relation was nonlinear, the
inverse mapping relation involved the unavoidable loss of some information:
wavenumbers beyond an azimuthal high wavenumber cutoff could not be
resolved. Thus, for a satisfactory general retrieval algorithm, the missing
spectral information needed to be supplied from another source. In practice,
the only available source was a two-dimensional spectral wave model, as all
conventional measurement systems (other than wave buoys at a few isolated
locations) are unable to resolve the specific spectral bands of interest. Thus
retrieval algorithms in the general nonlinear case necessarily require an input
first-guess wave spectrum from a wave model.
This results, however, in an unavoidable intertwining of three separate
aspects of the ERS SAR wave mode data: the retrieval of ocean wave spectra
from the SAR data, the prediction of ocean wave spectra, and the assimilation
of observational data in wave prediction models. The last two aspects imply
that the application of the SAR wave mode data becomes coupled also to the
problem of weather prediction, involving in addition the measurement and
assimilation of atmospheric data. The separability of the three problems is
feasible only for swell systems, outside an active wave generation region, in
which the nonlinearities of the SAR imaging mechanism become negligible.
The application of SAR wave mode data in this intertwined mode, together
with improvements in weather prediction and associated atmospheric data
assimilation methods, has led to the continuous improvement of global wave
prediction models to a level at which, for most state-of-the-art wave models
today, no systematic deviations between observed and modelled wave spectra,
beyond the expected variability of observations, is found in the overall
statistics of the comparisons of model predictions and SAR observations.
However, significant deviations can still be found in individual cases,
such as in intense cyclonic storms. Apart from the difficulties of an accurate
prediction of wind fields in such extreme situations, it is well known from
theoretical studies that the numerical approximations of the important source
function describing the nonlinear energy transfer exhibit serious shortcomings
in most wave models in complex wind seas. The SAR wave mode data
provide an excellent opportunity for revisiting this problem and significantly
improving wave models in these important cases.
A related problem in which the application of SAR wave mode data could
lead to significant progress is in the occurrence of individual extreme wave
formations known as freak or monster waves. Although monster waves occur
infrequently, they represent a serious danger for the rapidly expanding global
shipping industry, causing significant losses or damage to container ships
and supercarrier vessels. Global SAR wave mode observations of individual
extreme waves could provide a valuable statistical database for determining
the frequency and conditions of occurrence of these poorly understood wave
formations. However, further work is needed on the nonlinear SAR imaging of
individual waves before a quantitative determination of the statistics of real
freak waves rather than freak SAR wave images can be provided.
Another process that leads to the formation of extreme wave conditions is
the interaction between waves and currents, an important problem that is also

193
SP-1326

ripe for a more detailed exploration using SAR wave mode (and in this case
also full-swath SAR data). Strong current gradients can result in a focusing of
waves, producing abnormal wave conditions. Previously regarded as largely
unpredictable, the simultaneous measurements of both wave spectra and
currents using SAR wave imaging, the SAR Doppler for current measurements,
and currents computed from radar altimeter sea level measurements, now
enable predictions of the anomalous wave spectra produced by wave–current
interactions.
Finally, interesting data on the location of unusual systems of ocean
waves travelling in opposite directions can be derived from triangulations
of microseism measurements at different stations. Microseisms of twice the
wave frequency are generated by quadratic interactions between ocean waves
travelling in opposite directions in the open ocean, caused by moving cyclonic
storm systems, or in the vicinity of coasts through wave reflections at steep
shorelines. Important indirect information on processes within a storm can
also be derived from observations of swell propagating away from a distant
storm.
These and other examples presented in this chapter demonstrate that
despite the considerable advances in our understanding and prediction of
ocean waves achieved through the ERS SAR wave mode data, many interesting
problems that could be successfully addressed using this instrument still await
a solution.
It is therefore very encouraging that there are plans to continue the ERS-1/
ERS-2–Envisat satellite system with the launch in 2013 of Sentinel-1, the first in
a series of next-generation satellites, as an important component of the Global
Monitoring for Environment and Security (GMES) system. Sentinel-1 will again
carry a SAR wave mode instrument with enhanced capabilities (20 × 20 km
imagettes every 100 km). The continuation of this satellite series is important
not only for the upgrading of the present ocean observing system and the
resolution of the open problems mentioned, but also for the creation of a long-
term time series in support of climate change monitoring.

References
Abdalla, S., Bidlot, J. & Janssen, P. (2004). Assimilation of ERS and ENVISAT wave
data at ECMWF, Proc. 2004 Envisat & ERS Symposium, Salzburg, Austria. ESA
SP-572, European Space Agency, Noordwijk, the Netherlands.
Abdalla, S., Janssen, P. & Bidlot, J.-R. (2010). Envisat ASAR wave mode spectra
global validation and assimilation. Proc. SeaSAR 2010, Frascati, Italy. ESA
SP-679, European Space Agency, Noordwijk, the Netherlands.
Alpers, W.R. & Rufenach, C.L. (1979). The effect of orbital motions on synthetic
aperture radar imagery of ocean waves. IEEE Trans. Antennas Propagat. 27,
685–690.
Aouf, L., Lefevre, J.-M., Chapron, B. & Hauser, D. (2010). Some recent improvements
of the assimilation of upgraded ASAR L2 wave spectra. Proc. SeaSAR 2010,
Frascati, Italy. ESA SP-679, European Space Agency, Noordwijk, the Netherlands.
Ardhuin, F., Chapron, B. and Collard, F. (2009). Observation of swell dissipation
across oceans. Geophys. Res. Lett. 36(L06607), 10.1029/2008GL037030.
Ardhuin, F., Rogers, W.E., Babanin, A.V., Filipot, J., Magne, R., Roland, A., van
der Westhuysen, A., Queffeulou, P., Lefevre, J., Aouf, L. & Collard, F. (2010).
Semiempirical dissipation source functions for ocean waves. Part I: Definition,
calibration, and validation. J. Phys. Oceanogr. 40, 1917–1941.
Aster, R.C., McNamara, D.E. & Bromirski, P.D. (2008). Multidecadal climate
induced variability in microseisms. Seismol. Res. Lett. 79, 194–202. doi:10.1785/
gssrl.79.2.194

194
 SAR Wave Mode

Barber, B.F. & Ursell, F. (1948). The generation and propagation of ocean waves
and swell. I. Wave periods and velocities. Phil. Trans. R. Soc. Lond. A, 240(824),
527–560.
Bauer, E. & Heimbach, P. (1999). Annual validation of significant wave heights
from ERS‑1 synthetic aperture radar wave mode spectra using TOPEX/
POSEIDON and ERS‑1 altimeter data. J. Geophys. Res. 104(C6), 13  345–13  357.
doi:10.1029/1999JC900014
Borge, J.C.N., Lehner, S. Niedermeier, A. & Schulz-Stellenfleth, J. (2004). Detection
of ocean wave groupiness from spaceborne synthetic aperture radar. J. Geophys.
Res. 109, C07005. doi:10.1029/2004JC002298
Breivik, L.A., Reistad, M. Schyberg, H. Sunde, J. Krogstad, H.E. & Johnsen, H. (1998).
Assimilation of ERS SAR wave spectra in an operational wave model. J. Geophys.
Res., 103(C4), 7887−7900.
Brooker, G. (1995). UWA Processing Algorithm Specification, version 2.0, Tech. Rep.,
ESTEC/NWP, Noordwijk, The Netherlands.
Chapron, B., Collard, F. & Ardhuin, F. (2005). Direct measurements of ocean surface
velocity from space: interpretation and validation, J. Geophys. Res. 110, C07008.
doi:10.1029/2004JC002809
Chapron, B., Johnsen, H. & Garello, R. (2001). Wave and wind retrieval from SAR
images of the ocean. Ann. Telecommun. 56, 682−699.
Collard, F., Ardhuin, F. & Chapron, B. (2005). Extraction of coastal ocean wave
fields from SAR images. IEEE J. Oceanic. Eng. 30(3), 526—533.
Collard, F., Ardhuin, F. & Chapron, B. (2009). Monitoring and analysis of ocean
swell fields from space: New methods for routine observations. J. Geophys. Res.
114, C07023. doi: 10.1029/2008JC005215
Delpey, M.T., Ardhuin, F. Collard, F. & Chapron, B. (2010). Space–time structure of
long ocean swell fields. J. Geophys. Res. 115, C12037. doi:10.1029/2009JC005885
Dysthe, K., Müller P. & Krogstad, H.E. (2008). Oceanic rogue waves. Ann. Rev. Fluid
Mech. 40, 287−310.
Ebeling, C. W., & Stein, S. (2011). Seismological Identication and characterization of
a large hurricane. Bull. Seismol. Soc. Am. 101(1), doi: 10.1785/0120100175
Engen G. & Johnsen, H. (1995). SAR-ocean wave inversion using image cross
spectra. IEEE Trans. Geosci. Rem. Sens. 33(4), 1047−1056.
Engen G., Johnsen H., Krogstad, H.E. & Barstow, S.F. (1994). Directional wave
spectra by inversion of ERS‑1 synthetic aperture radar ocean imagery. IEEE
Trans. Geosci. Rem. Sens. 32(2), 1994.
Grevemeyer, I., Herbert, R. & Essen, H.H. (2000). Microseismological eveidence for
a changing wave climate in the northeast Atlantic Ocean. Nature 408, 349–352.
Grob, M., Maggi, A. & Stutzmann, E. (2011). Observations of the seasonality of the
Antarctic microseismic signal, and its association to sea ice variability. Geophys.
Res. Lett., 38, L11302. doi: 10.1029/2011GL047525
Hanafin, J.A., Quilfen, Y., Ardhuin, F., Sienkiewicz, J., Queffeulou, P., Obrebski, M.,
Chapron, B., Reul, N., Collard, F., Corman, D., de Azevedo, E. B., Vandemark, D.
& Stutzmann, E. (2012). Phenomenal sea states and swell from a North Atlantic
storm in February 2011: A comprehensive analysis. Bull. Am. Meteorol. Soc.
93(12), 1825–1832. doi: http://dx.doi.org/10.1175/BAMS-D-11-00128.1
Hasselmann, K. (1963). A statistical analysis of the generation of microseisms. Rev.
Geophys. 1, 177–210.
Hasselmann, K (1967). A criterion for nonlinear wave stability. J. Fluid Mech. 30,
737–739.
Hasselmann, K. & Hasselmann, S. (1991). On the nonlinear mapping of an ocean
wave spectrum into a Synthetic Aperture Radar image spectrum and its
inversion. J. Geophys. Res. 96, 10713−10729.
Hasselmann, K., Raney, R.K. Plant, W.J. et al. (1985). Theory of synthetic aperture
radar ocean imaging: A MARSEN view. J. Geophys. Res. 90, 4659−4686.
Hasselmann, S., Brüning, C. Hasselmann K. & Heimbach, P. (1996). An improved
algorithm for the retrieval of ocean wave spectra from SAR image spectra.
J. Geophys. Res. C101, 16615−16629.

195
SP-1326

Heimbach, P. & Hasselmann, K. (2000). Development and application of satellite

retrievals of ocean wave spectra, In: D. Halpern (Ed.) Satellites, Oceanograghy
and Society. Elsevier Oceanography Series vol. 63, Elsevier Science, Amsterdam,
pp.5−33. doi:10.1016/S0422-9894(00)80003-3
Heimbach, P., Hasselmann, S. & Hasselmann, K. (1998). Statistical analysis and
intercomparison with WAM model data of three years of global ERS‑1 SAR wave
mode spectral retrievals, J. Geophys. Res. 103, 7931–7977.
Högström, U., Smedman, A., Sahleé, E., Drennan, W.M., Kahma, K.K., Pettersson,
H. & Zhang, F. (2009). The atmospheric boundary layer during swell: a field
study and interpretation of the turbulent kinetic energy budget for high wave
ages. J. Atmos. Sci. 66, 2764–2779. doi: http://dx.doi.org/10.1175/2009JAS2973.1
Holt, B., Liu, A.K. Wang, D.W., Gnanadesikan, A. & Chen, H.S. (1998). Tracking
storm-generated waves in the northeast Pacific Ocean with ERS‑1 synthetic
aperture radar imagery and buoys. J. Geophys. Res. 103(C4), 7917−7929.
Husson, R., Ardhuin, F., Collard, F., Chapron,B. & Balanche,A. (2012). Revealing
forerunners on Envisat's wave mode ASAR using the Global Seismic Network.
Geophys. Res. Lett.. 39, L15609, doi: 10.1029/2012GL052334.
Janssen, P.A.E.M. (2003). Nonlinear four wave interaction and freak waves. J. Phys.
Oceanogr. 33, 863–884.
Janssen, P.A.E.M. (2008). Progress in ocean wave forecasting, Comput. Phys. 227,
3572−3594.
Johannessen, J.A., Chapron, B., Collard, F., Kudryavtsev, V., Mouche, A., Akimov,
D. & Dagestad, K.-F. (2008). Direct ocean surface velocity measurements from
space: Improved quantitative interpretation of Envisat ASAR observations.
Geophys. Res. Lett. 35, L22608. doi:10.1029/2008GL035709
Johannessen, J.A., Chapron, B., Alpers, W., Collard, F., Cipollini, P., Liu, A.,
Horstmann, J., da Silva, J., Portabella, M., Robinson, I.S., Holt, B., Wackerman,
C. & Vachon, P. (2013) Satellite oceanography from the ERS Synthetic Aperture
Radar and Radar Altimeter: A brief review. This volume.
Johnsen, H., Engen, G. & Lauknes, I. (1998). The Envisat ASAR wave mode cross-
spectra algorithm, Proc. CEOS SAR Workshop, February 1998, ESTEC/ESA.
Johnsen, H., Engen, G. Collard, F., Kerbaol, V. & Chapron, B. (2006). Envisat ASAR
wave mode products: Quality assessment and algorithm upgrade. Proc. SEASAR
2006, Frascati, Italy. ESA SP-614, European Space Agency, Noordwijk, the
Netherlands.
Kerbaol, V., Chapron, B. & Vachon, P.W. (1998) Analysis of ERS‑1/2 synthetic
aperture radar wave mode imagettes. J. Geophys. Res. 103, 7833–7846.
Koenig, T., Lehner, S. & Schulz-Stellenfleth, J. (2007). Global analysis of a 2 year
ERS‑2 wave mode dataset over the Ocean, Proc. IGARSS 2007, Barcelona, Spain.
Komen, G.J., Cavaleri, L., Donelan, M., Hasselmann, K., Hasselmann, S. & Janssen,
P.A.E.M. (1996). Dynamics and Modelling of Ocean Waves, Cambridge University
Press, Cambridge, UK.
Krogstad, H. E. (1992). A simple derivation of Hasselmann’s nonlinear ocean-SAR
transformation. J. Geophys. Res. 97(C2), 2421−2425.
Krogstad, H. E., Samset, O. & Vachon, P.W. (1994). Generalisations of the non-linear
ocean−SAR transform O. and a simplified SAR inversion algorithm. Atmos.
Ocean 32(1), 61-82.
Lehner, S., (1984). The Use of SAR for Large Scale Wind Measurement over the
Ocean. Hamburger geophysikalische Einzelschriften. M.L. Wittenborn Söhne,
Hamburg.
Lehner, S., Schulz-Stellenfleth, J., Schättler, B., Breit, H. & Horstmann, J. (2000).
Wind and wave measurements using complex ERS‑2 wave mode data. IEEE
Trans. Geosci. Remote Sens. 38(5), 2246−2257.
Li, J.G & Holt, M. (2009). Comparison of ENVISAT ASAR ocean wave spectra with
buoys and altimeter data via a wave model. J. Atmos. Oceanic Technol. 26, 593−614.
Li, X.-M., Lehner, S. & Bruns, Th. (2011). Ocean wave integral parameter
measurements using ENVISAT ASAR wave mode data, IEEE Trans. Geosci. Rem.
Sens. 49(1), 155−174. doi: 10.1109/TGRS.2010.2052364

196
 SAR Wave Mode

Li, X.-M., König, Th., Schulz-Stellenfleth J. & Lehner, S. (2010). Validation and
intercomparison of ocean wave spectra retrieval scheme using ASAR wave mode
data. Int. J. Remote Sens. 31(17), 4969–4933. DOI: 10.1080/01431161.2010.485222
Liu, A.K., Peng, C.Y. & Schumacher, J.D. (1994). Wave−current interaction study in
the Gulf of Alaska for detection of eddies by synthetic aperture radar. J. Geophys.
Res. 99(C5), 10 075−10 085. doi: 10.1029/94JC00422
Longuet-Higgins, M. (1950) A theory of the origin of microseisms. Phil. Trans. R. Soc.
London, A, 243, 1–35.
Mastenbroek, C. & de Valk, C.F. (2000). A semi-parametric algorithm to retrieve
ocean wave spectra from synthetic aperture radar. J. Geophys. Res. 105,
3497−3516. doi:10.1029/1999JC900282
Niedermeier, A., Borge, J.C.N., Lehner, S. & Schultz-Stellenfleth, J. (2005). A wavelet-
based algorithm to estimate ocean wave group parameters from radar images.
IEEE Trans. Geosci. Rem. Sens. 43(2), pp. 327–336. doi: 10.1109/TGRS.2004.836873
Schulz-Stellenfleth, J. & Lehner, S. (2004). Measurement of two-dimensional sea
surface elevation fields using complex synthetic aperture radar data, IEEE Trans.
Geosci. Rem. Sens. 42, 1149−1160.
Schulz-Stellenfleth, J., Lehner S. & Hoja, D. (2005). A parametric scheme for the
retrieval of two-dimensional ocean wave spectra from synthetic aperture radar
look cross spectra. J. Geophys. Res, 110, C05004. doi: 10.1029/2004JC002822
Schulz-Stellenfleth, J., König, Th. & Lehner, S. (2007). An empirical approach for the
retrieval of integral ocean wave parameters from synthetic aperture radar data.
J. Geophys. Res. 112, doi: 10.1029/2006JC003970
Snodgrass, F.E., Groves, G.W., Hasselmann, K., Miller, G.R., Munk, W.H. & Powers,
W.H. (1966). Propagation of ocean swell across the Pacific. Phil. Trans. R. Soc.
London 249A, 431–497.
Tamura, H., Waseda, T. & Miyazawa, Y. (2010). Impact of nonlinear energy transfer
on the wave field in Pacific hindcast experiments. J. Geophys. Res. 115, C12036.
doi: 10.1029/2009JC006014
Tamura, H., Waseda, T., Miyazawa, Y. & Komatsu, K. (2008). Current-induced
modulation of the ocean wave spectrum and the role of nonlinear energy
transfer. J. Phys. Oceanogr. 38, 2662–2684.
Tolman, H.L. (2004). Inverse modeling of discrete interaction approximations for
nonlinear interactions in waves. Ocean Modelling 6, 405–�422.
Tolman, H.L. (2009). Practical nonlinear interaction algorithms. Proc. 11th Int.
Workshop on Wave Hindcasting and Forecasting, and 2nd Coastal Hazards Symp.,
Halifax, Nova Scotia, Canada.
Tolman, H.L. (2011). On the impact of nonlinear interaction parameterizations
in practical wave models. Proc. 12th Int. Workshop on Wave Hindcasting and
Forecasting, and 3rd Coastal Hazards Symp., Hawaii, USA.
Tolman, H.L., Banner, M. & Kaihatu, J. (2011). The NOPP Operational Wave
Model Improvement Project, Proc. 12th Int. Workshop on Wave Hindcasting and
Forecasting, and 3rd Coastal Hazards Symp., Hawaii, USA.
Ueno, K. & Kohno, N. (2004). The development of the third generation model MRI-III
for operational use, Proc. 8th Int. Workshop on Wave Hindcasting and Forecasting
G2, pp.1–7.
Vachon, P.W. & West, J.C. (1992). Spectral estimation techniques for multilook SAR
images of ocean waves. IEEE Trans. Geosci. Rem. Sens. 30(3), 568–577.
Vachon, P.W. & Raney, R.K. (1991). Resolution of the ocean wave propagation
direction in SAR imagery. IEEE Trans. Geosci. Rem. Sens. 29(91), 105–112.
Van Vledder, G.P. (2001). Extension of the discrete interaction approximation for
computing nonlinear quadruplet wave–wave interactions in operational wave
models, Proc. 4th ASCE Int. Symp. on Ocean Waves, Measurements and Analysis,
San Francisco, CA, USA, pp.540–549.
WAMDI Group (1998). The WAM model – a third generation ocean wave prediction
model. J. Phys. Oceanogr. 18, 1775−1810.

197
→→ Satellite Oceanography from
the ERS Synthetic Aperture
Radar and Radar Altimeter: →
A Brief Review
 SAR & Altimeter

Satellite Oceanography from the ERS Synthetic

Aperture Radar and Radar Altimeter: A Brief
Review

J.A. Johannessen1, B. Chapron2, W. Alpers3, F. Collard2, P. Cipollini4, A. Liu5,

J.  Horstmann6, J.C.B. da Silva7, M. Portabella8, I.S. Robinson9, B. Holt10,
C. Wackerman11, P. Vachon12

1 Nansen Environmental and Remote Sensing Center, Bergen, Norway

2 L’Institut Français de Recherche pour l'Exploitation de la Mer (IFREMER),
Plouzane, France
3 Center for Marine and Atmospheric Sciences, Institute of Oceanography, University
of Hamburg, Hamburg, Germany
4 National Oceanography Centre, Southampton, UK
5 Department of Ocean Science and Engineering Zhejiang University, Hangzhou,
China
6 NATO Centre for Maritime Research and Experimentation (CMRE), Viale San
Bartolomeo, La Spezia, Italy
7 Department of Geosciences, Environment and Spatial Planning, University of
Porto, Porto, Portugal.
8 Institut de Ciències del Mar (ICM−CSIC), Pg. Marítim de la Barceloneta, Barcelona,
Spain
9 Ocean & Earth Science, University of Southampton, National Oceanography
Centre, Southampton, UK.
10 Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, USA
11 General Dynamics Advanced Information Systems, Ypsilanti, MI, USA
12 Radar Applications and Space Technologies, Defence R&D Canada, Ottawa,
Canada

1. Introduction
The cloud-independent, all-weather, day-and-night active microwave Synthetic
Aperture Radar (SAR) and Radar Altimeter (RA) operated continuously on
ERS‑1 from July 1991 to March 2000 and on ERS‑2 from October 1995 to July
2011. In addition, the two ERS satellites also operated the active microwave
C-band scatterometer (see Stoffelen & Wagner, this volume). The side-looking
C-band (5.6 GHz) SAR was operated to obtain two-dimensional spectra of ocean
surface waves to reveal ultra-high-resolution surface wind field patterns, to
identify numerous oceanic features, as well as to detect oil spills, ship and
detailed sea ice conditions and deformation. The nadir-pointing Ku-band
(13.8  GHz) RA aimed to measure sea surface height, significant wave heights
and wind speed, and thanks to the choice of orbit, it also covered high latitudes
for sea ice signature observations including freeboard heights.
Table 1 summarises the applications of the ERS‑1/ERS‑2 SAR and RA data
in the field of oceanography. The SAR surface wave retrievals, in particular the
SAR wave mode capabilities and the altimeter mean sea level retrievals, are
addressed by Hasselmann et al. and Andersen et al., respectively, elsewhere
in this volume. A comprehensive overview of the key early scientific findings
based on ERS data is presented in the special issue the Journal of Geophysical
Research: Oceans on advances in oceanography and sea ice research using
ERS observations (Johannessen et al., 1998).

201
SP-1326

2. Surface Currents from SAR

The ERS SAR operated in both Wave Mode (WM) and Image Mode (IM) and
achieved a high spatial resolution of 26 m in range (across-track) and between
6  m and 30 m in azimuth (along-track). As a first approximation, the side-
looking ERS SAR formed an image from the Bragg resonance between the
transmitted vertically (V) polarised radar pulses and the short gravity ocean
waves (Wright, 1978; Valenzuela, 1978; Ulaby et al. 1982; Elachi, 1987). These
short gravity waves, including small-scale breakers, form in response to the
local wind stress and are further modulated by the orbital motion of the longer
waves, by intense wave breaking at longer waves, and by the surface current
and/or Sea Surface Temperature (SST) gradients. The ERS SAR provided unique
observations of near-surface winds, waves and surface current features.

2.1 Current Shear, Convergence/Divergence

Surface density fronts are ubiquitous features of the upper ocean dynamics.
The surface signatures of meandering fronts and eddies can be quite
spectacular and have been regularly observed and documented in SAR images.
Wave–current interactions, the suppression of short wind waves by natural
films, and the varying wind fields resulting from atmospheric boundary layer
changes across an oceanic temperature front all contribute to the radar image
manifestations of such mesoscale features. A clear example of the frontal
features associated with the mesoscale circulation patterns expressed in SAR
images is shown in Fig. 1.
In Fig.  1 the changes in surface roughness detected by the SAR apparently
correspond to changes in the near-coincident SST field imaged by NOAA’s
Advanced Very-High-Resolution Radiometer (AVHRR; Johannessen et al., 1996a).
In the latter, the structure of the SST field with the curvilinear temperature fronts
represents a mesoscale variability of 10−50  km, characteristic of the unstable
Norwegian coastal current (Johannessen et al., 1989).
The configuration and orientation of the frontal features contained in the
radar image are in good qualitative agreement with those in the AVHRR image,
where the sharp radar cross-section gradients are strongly collocated with the
SST gradients. Other comparable image expressions using near-coincident
ERS‑1/ERS‑2 SAR and AVHRR acquisitions of surface frontal features were
documented by Nilsson & Tildesley (1995) for the East Australian current, and
by Mitnik et al. (2005) for the southwestern Okhotsk Sea. In order to interpret
the observed surface roughness changes, numerous further investigations
have built on the SAR internal wave imaging capability (see section 2.3).

Table 1. Overview of the oceanographic applications of ERS‑1/ERS‑2 SAR and RA data.

SAR (image mode) RA

Surface currents Mesoscale fronts and eddies via Surface geostrophic currents along fronts
divergence−convergence and eddies
Internal waves Planetary waves
Shallow-water bathymetry
Range Doppler surface velocity
Oil spill, natural film and ship detection Distinct contrast in radar cross-section
Wind field High-resolution wind speed and in some Along-track wind speed
cases also wind direction
Atmospheric boundary layer processes
Sea level Mean Sea Surface Heights relative to a
reference surface (ellipsoid or geoid)

202
 SAR & Altimeter

Figure 1. Left: The Norwegian coastal

current front in a NOAA/AVHRR 1-km
resolution image of the SST field (white,
14°C; dark blue, 12°C; land is masked with
green) acquired at 14:20 UTC on 3 October
1992. Right: ERS SAR 100-m resolution
image of the surface roughness field
acquired at 21:35 UTC (after Johannessen
et al., 1996a). Both images cover
approximately the same 100 × 300 km
region. (© ESA/NERSC)

Kudryavtsev et al. (2005) and Johannessen et al. (2005) used a forward-imaging

simulation model to demonstrate that geostrophic surface current convergence
and divergence along the frontal boundaries are the principal source of strong
modulation of the intermediate short-scale surface roughness that, in turn,
contributes to the image contrast.
Mesoscale to submesoscale eddies were often examined with ERS‑1/ERS‑2
SAR in conjunction with SST data. These eddies are most often detected in
coastal zones and are largely cyclonic in a rotational sense (DiGiacomo &
Holt, 2001; Johannessen et al., 1996; Nilsson & Tildesley, 1995; Liu et al.,
1994). They are identified in the SAR images by a series of dark (reduced
backscatter), narrow, curvilinear bands arranged within the eddy into a spiral
appearance, as shown in Fig.  2. The dark bands are composed of surfactants
brought together by the complicated flow patterns within the eddies. They
appear within a limited range of wind speeds (3−7 m s−1) and are likely short
lived. There are probably several forms of generation mechanisms related to
current shear and frontal convergence, and are often associated with coastal
topography. The importance of meso- to submesoscale eddies is related

203
SP-1326

Figure 2. ERS‑1 SAR image of the Santa

Monica Bay/Santa Monica Basin, California,
at 1834 UTC on 19 December 1994.
The labels A−F indicate the locations of
submesoscale cyclonic eddies. (From
DiGiacomo & Holt, 2001)

to biological pumping of nutrients into the upper ocean and their role as
significant components of dissipation within the ocean circulation energy
regime.

2.2 Internal Waves

Internal Solitary Waves (ISWs) are thought to play an important role in
global mixing in the upper layers of the ocean (Shroyer et al., 2010), and they
contribute considerably to shelf dynamics in terms of vertical heat fluxes and
ocean shelf mass transport. Internal tides and ISWs have also been shown to
interact with upwelling jets, driving intensified turbulence and dissipation
(Avicola et al., 2007; Kurapov et al., 2010). From SAR imaging principles,
imaging of internal waves in a stratified ocean is then connected with the
interaction between the internal wave-induced surface current fields and
wind-driven ocean surface waves (Alpers, 1985; Apel et al., 1985). In a surface
current convergence zone, small gravity waves are squeezed together and
the surface is roughened compared with the surroundings. In contrast, in
the surface current divergence zone, small gravity waves are dispersed. The
signature of a rough surface in a SAR image is correspondingly bright, while a
smooth surface appears dark.
The interaction of internal waves, the induced surface current with wind
waves, and the resultant SAR image intensity variations are illustrated
schematically in Fig. 3. When the thickness of the mixed layer is thinner than
that of the bottom layer (e.g. H1 < H2), only depression waves can originate. In
contrast, for cases when the bottom layer H1 > H2, only elevation waves evolve,
as illustrated in Fig.  3 (Liu, 1998). In deep water with a shallow mixed layer,
the nonlinear internal solitons are therefore typically depression waves. The
depression internal wave packet can be identified as a bright band immediately
followed by a dark band. Conversely, an elevation internal wave packet can

204
 SAR & Altimeter

Figure 3. Schematic diagram of interaction

of internal waves, induced surface current
with wind waves, and the resultant SAR
image intensity variations. (After Liu, 1998)

be identified as a dark band immediately followed by a bright band. Thus

the observation frequency of internal waves depends on the induced surface
current strain rate, sea state, wind speed and environmental conditions. In the
open ocean with deep water and a shallow mixed layer, the nonlinear internal
solitons are generally mode-one depression waves, while elevation waves
can exist in shallower water near to the shore in cases when the bottom layer
becomes thinner than the mixed layer.
The Strait of Gibraltar and the Strait of Messina were among the first
sites of internal wave generation that was investigated using ERS‑1/ERS‑2
SAR images (Alpers et al., 2008), although the first SAR image showing the
generation of internal waves in the Strait of Messina was captured in 1978 by
the SAR onboard the American Seasat satellite (Alpers & Salusti, 1983). In the
Strait of Gibraltar the SAR images reveal that internal waves only propagate
eastwards into the Mediterranean Sea (Fig. 4, left). This is due to the complex
topography of the sill region with which the predominantly semi-diurnal tide
interacts as shown in model calculations (Brandt et al., 1996). In contrast, the
ERS‑1/ERS‑2 SAR images acquired over the Strait of Messina show internal
waves propagating northwards as well as southwards from the sill, although
internal waves propagating northwards are much less frequent (Brandt et al.,
1997, 1999). In most ocean regions, the internal wave generation processes are
repeated with tidal periodicity. With a large enough dataset covering different
phases of the complete (semi-diurnal) tidal cycle, it is therefore possible to
investigate and reconstruct the propagation ‘history’ of waves generated by a
given source (e.g. Moum et al., 2003; New & da Silva, 2002; Klimak et al., 2006;
da Silva & Helfrich, 2008; da Silva et al., 2007, 2009, 2011).
The largest internal waves (peak to trough values of more than 60 m) are
encountered in the Andaman Sea (Fig. 4, right), between the Malay Peninsula
and the Andaman and Nicobar Islands. These internal waves are associated
with strong roughness bands (ripplings) that can even be observed from ships.
This was described in a book published in 1861 by Maury (quoted in Osborne
& Burch, 1980): ‘The ripplings are seen in calm weather approaching from a
distance and in the night their noise is heard over considerable time before
they arrive. They beat against the sides of the ship with great violence, and
pass on, the spray sometimes coming on deck.’

205
SP-1326

Figure 4. Left: ERS‑1 SAR image acquired on 20 January 1994 revealing the surface manifestation of an internal soliton close to its
generation area in the centre of the Strait of Gibraltar and an internal wave packet east of the strait generated during the previous tidal
cycle. Right: ERS‑2 SAR strip acquired over the Andaman Sea on 11 February 1997 showing a surface manifestation of internal solitary
waves. The land area visible at the bottom is the island of Sumatra. The image area is 100 × 300 km. The smaller circular pattern in the
upper third of the SAR image is the surface signature of a secondary internal wave generated by the interaction of the first internal solitary
wave with the Dreadnought underwater bank, 241 m deep. (Vlasenko & Alpers, 2005)

2.3 Shallow Water Bathymetry

Knowledge of sea bottom bathymetry and corresponding depths is of vital
importance to ship routing in shallow water, for fisheries and for offshore
operations. While traditional bathymetric surveys are time and cost-consuming,
the imaging capabilities with SAR have demonstrated a promising capability to
retrieve bathymetric maps over large areas at relatively low cost, provided the
average water depth does not exceed 20–30 m. Under such conditions, bottom
topographic features may then become visible on radar images of the sea
surface if there is a current (usually tidal current) flowing over these features.
They cause local perturbations to the surface current, which in turn modulates
the sea surface roughness, hence making it visible in the SAR images.
Figure 5 shows a typical ERS SAR image acquired over a shallow tidal area
around Shanghai off the east coast of China. A theory relating the modulation
patterns visible in such SAR images to the underwater bottom topography was
developed by Alpers & Hennings (1984) using weak hydrodynamic interaction
theory. Although this theory describes the modulation qualitatively quite
well, it cannot be used to invert SAR image intensity modulation patterns into
bathymetric maps.
When used in combination with acoustic sounding data the SAR images are
of great value in generating bathymetric maps. Off the coast of the Netherlands
large-scale topographic structures are regularly imaged under favourable
moderate (3–7 m s−1) wind conditions. These, in turn, are used to validate and
initiate morphodynamic models that are developed to forecast changes in
bottom shapes linked with sediment transport, river deposition and coastal
erosion. The results show that by using the SAR information in combination

206
 SAR & Altimeter

Figure 5. ERS‑1 SAR image of the Xinchuan

Gang shoals off the east coast of China
north of Shanghai. Parts of the area
become dry during ebb tide (dark areas off
the coast).

with ship-based depth soundings the number of ship tracks can be reduced
by a factor of 3–5, while still maintaining an rms error of the depth map of
less than 30 cm (Hasselmann et al., 1997), which is the accuracy required by
the Dutch Ministry of Transport and Public Works (Rijkswaterstaat). This
conclusion has been corroborated by the results of depth assessments in other
areas. It demonstrates that the use of SAR imaging can lead to substantial
reductions in the cost of traditional sounding campaigns. Capitalising on these
findings, the Dutch company ARGOSS has developed a system, called the
Bathymetry Assessment System (BAS), which is commercially available (see
Alpers et al., 2004).

2.4 Wave–Current Interactions

Ocean waves propagating through strong, variable currents can undergo
dramatic changes in wave height and steepness. In a steady shear current,
wave rays are bent, the bending being proportional, in the approximation of
geometric optics, to the ratio of the vertical component of the current vorticity
to the wave group velocity (see Fig. 6).
For more information on wave-current interactions, see section 7.3 in the
chapter by Hasselmann, et al., page 190.

207
SP-1326

Figure 6. ERS‑2 SAR image of the Agulhas

current from 12 February 2010. Significant
wave heights are given by the colour code.
Local swell systems can be detected over
this image and multiple wave rays derived.
The white arrows correspond to the length
(100–500 m) and propagation direction of
the main detected swell system over a given
subimage tile. Using the local direction
angles, the reconstructed propagation rays
(red and green) have been reconstructed
and clearly exhibit refraction effects, with
increasing bending corresponding to
stronger shear current areas. (CLS)

2.5 Range Doppler Surface Velocity

The SAR-received signal in amplitude and phase can be used to infer the first-
order moment of the Doppler distribution, called the Doppler centroid shift. The
reflected signals over the illuminated ocean scene in the cross-range direction
systematically exhibit different phase shifts compared to the nominal expected
phase history associated with a motion-free scene such as obtained over land.
To first order, this mean shift is directly connected to an overall radar line-of-
sight mean velocity relating to the dominant roughness scales contributing to
the backscatter power. These scales are predominantly affected by the wind
speed and direction, orbital wave motion and underlying surface current. After
correctly taking into account the local platform orbital geometry and kinematics
as well as Earth’s rotation, the measured mean Doppler correlates well with the
local ocean surface motion as well as roughness (see Fig. 7). To achieve this with
adequate accuracy, a very precise orbit is needed as well as instrument pointing.
The wind direction in this case is offshore (e.g. the well-known Tramontane
wind in the Gulf of Lion, France), and the Doppler centroid shift over the
ocean is consequently positive, implying that the scatterers move towards
the antenna as expected for this descending track. Based on this innovative
method, presented in the Marine SAR Analysis and Interpretation System
(MARSAIS) toolkit using ERS SAR data (Chapron et al., 2005), a new Envisat
wide-swath single-look complex product provided the opportunity to derive
potentially highly accurate surface current estimates over the entire ASAR
400 km swath in the coastal zones as well as open oceans.

208
 SAR & Altimeter

Figure 7. Left: Complex ERS SAR image of

the Gulf of Lion, France. Centre: Doppler
centroid anomaly from the complex image.
Right: Doppler anomaly along the blue
profile in the centre image. A transition
between land and ocean, with positive
Doppler over the ocean is clearly depicted.

Figure 8. Mean radial surface speed derived from ascending Envisat/ASAR wide-swath mode acquisitions for the Gulf Stream (left) and the
Agulhas current (right). The colour bars indicate the velocity (in m s−1) revealing opposite flow directions for the two current systems. (CLS)

An example of this is shown in Fig.  8, using an average of more than

600 ascending acquisitions of the Gulf Stream and the Agulhas current
(Johannessen et al., 2008). As noted, the maximum mean radial Doppler velocity
reaches about 1.5 m s−1, with a mean current width of the order of about 100 km.
In the Agulhas current a distinct sign shift in the range Doppler velocities,
followed by the gradual manifestation of moderately weaker (~0.75 m s−1) and
broader eastward velocities, are manifested in the semi-permanent meandering
of the Agulhas return current (Boebel et al., 2003). These robust Doppler
velocity signatures are striking. Although the Doppler velocity is not a direct
surface current measurement, it inevitably demonstrates that the use of Doppler
observations, first discovered in the ERS SAR data, can help to derive new and
innovative estimates of mesoscale dynamics associated with intense surface
current regimes. For more details, visit http://soprano.cls.fr

209
SP-1326

Figure 9. Spatial distribution of oil pollution

in the waters around Singapore derived
from ERS SAR images. (CRISP, Singapore)

3. Oil Spills, Natural Films and Ship Detection

Imaging radars are very suitable for monitoring oil spills notably in regions of
intense shipping traffic and in offshore areas of extensive oil and gas production
(Pedersen et al., 1996). Oil and natural slicks (film) dampen the capillary short
gravity waves. In turn, the radar cross-section is lowered and the imaged area
appears dark. For most users, the basic problem is how to distinguish between
man-made oil spills from natural films and other ‘lookalike’ features. For the
practical determination of slick type (whether natural or man-made) and origin
using a single-frequency and polarisation SAR, empirical and conceptual and
interactive algorithms, employing wind history and drift models, appear to be
most useful (Espedal et al., 1998).
On 19 August 1996 the beaches of Singapore were found polluted with oil.
Using an ERS SAR image, investigators were able to find the ship responsible,
and the ship owner ended up in court, where he was convicted and fined
US$0.6 million. Building on repeat acquisitions, ERS SAR images have also
been used to monitor areas with intense shipping traffic and frequent oil spills,
either accidental or deliberate. The map in Fig.  9 shows the number of oil
slicks per 10 000 m2 derived from quick-look ERS SAR images downloaded and
processed by the Singapore receiving station.

4. High-Resolution Wind Field and Boundary Layer→

Retrievals
The ERS‑1/ERS‑2 SARs used the same radar frequency (C-band) and polarisation
(VV) as the scatterometers on these platforms. Hence, it was rapidly postulated
that the CMOD Geophysical Model Function (GMF) relationship between wind
and radar backscatter could also be used for the SAR. Given the very high
spatial resolution of the SAR instruments, this opened the way for the retrieval
of very high-resolution wind (~1 km) fields. However, with just one SAR antenna
the near-surface wind field cannot be simply partitioned into the contributions
from the wind speed and the direction. Several algorithms developed during
the late 1990s tackled this deficiency with satisfactory results, at least under
good imaging conditions (Etling & Brown, 1993; Johannessen et al., 1996b;
Chapron et al., 1995; Wackerman et al., 1996; Vachon & Dobson, 1996; Fetterer

210
 SAR & Altimeter

et al., 1998; Kerbaol et al., 1998; Lehner et al. 1998; Wackerman et al., 2003;
Horstmann et al., 2000, 2004; Horstmann & Koch, 2005).

4.1 Wind Retrieval

4.1.1 Wind Direction Retrieval

Besides signals of coastal shadowing effects, the SAR wind direction

retrievals are based on the possibility of detecting linear features, such as
those associated with wind streaks and atmospheric boundary layer rolls,
at scales larger than typically 800  m. The determination of the orientation
of these linear features can be carried out either in the spectral domain or in
the spatial domain and resulted in high-resolution wind direction estimations
ranging from 1 km to 10 km. The wind directions, being perpendicular to the
orientation of the linear features, are then determined with an accuracy of
about 15−20° but with a directional ambiguity of 180° (Monaldo et al., 2003;
Drobinski & Foster, 2003; Koch, 2004; Sikora et al., 2006). This ambiguity can,
however, be removed if wind shadowing is manifested in the image, which
is often the case in the vicinity of coastal regions. In other cases where there
are no manifestations of linear features in the image, the wind directional
information can be derived from weather maps, atmospheric models or in situ
measurements. A very nice documentation of the wind retrieval capabilities of
SAR is presented in Beal & Pichel (2000).
Attempts have also been made to explore in more detail the along-azimuth
complex SAR signals (Engen & Johnsen, 1995). The proposed methodology
is based on the processing of single-look-complex SAR data to estimate the
decorrelation time and phase from the cross-spectra between different SAR
looks. Such a methodology is closely related to the Doppler centroid method
described above, and can be used to help to remove the wind ambiguity more
directly as demonstrated by Mouche et al. (2012).

4.1.2 Wind Speed Retrieval

Given the wind direction, the wind speed retrieval then relies on the global
empirically based model function developed for the ERS scatterometer for
fully developed seas (Stoffelen et al., 1997a,b; Quilfen et al., 1998). This model
function relates the Normalised Radar Cross Section (NRCS) to the local near-
surface wind speed u, wind direction versus antenna look direction Φ, and the
incidence angle θ. The general form of the function is:

v0 - A^ih u v(i) ^1 + B^u, ih cos U + C^u, ih cos 2Uh (1)

where A, B, C and ν are coefficients that generally depend on radar frequency

and polarisation. In the case of the models available for the C-band VV
polarisation, these coefficients were determined empirically by evaluating
ERS‑1 scatterometer data and independent wind fields from models (e.g.
Numerical Weather Prediction (NWP) models, ECMWF) or wind-measuring
buoys (Monaldo et al., 2001) (for more details, see Stoffelen & Wagner, this
volume). In coastal regions this model function should be applied with care
as the wind can be fetch-limited, implying that the seas may not be fully
developed.
In an attempt to avoid the inherent inversion problem for SAR-based
wind vector retrievals, Portabella et al. (2002) proposed combining the SAR-
observed σ0 and/or wind streaks, with other background information from
NWP models and/or buoy data. Assuming Gaussian errors, and that both the
observation and the background errors are uncorrelated, the general Bayesian

211
SP-1326

approach (Lorenc, 1986) was then applied to SAR observations. This method
allows retrievals of an optimum wind vector from the best combination of
SAR and other wind information, and has shown promising results, although
deficiencies remain in cases of abrupt shifts in wind direction associated with
wind fronts (Portabella et al., 2002). The methodology is now being extended
to include the SAR Doppler information (Mouche et al., 2012) in order to help
constrain front locations.

4.1.3 Extreme Wind Events

Under extreme environmental conditions, the air–sea exchanges are thought

to represent the main kinetic energy sources necessary to maintain the deep
ocean stratification and to strengthen ocean stirring processes. Thanks to
satellite-based observations, including SAR, extreme weather events such
as tropical cyclones, hurricanes and explosive mid-latitude storms and polar
lows are now more often reported and analysed (Katsaros et al., 2000; Quilfen
et al., 1998). These measurements are critical for short-term forecasting, but
also offer a better means to question the role of extreme conditions for the state
of the ocean at local and global scales and how they affect ocean circulation
and ocean heat transport. In particular, the imaging SAR-based observations
provide high-resolution to delineate the fine-scale structures associated with
extreme wind conditions. The early observations of these extreme wind fields
associated with hurricanes and polar lows spurred the systematic collection of
all possible acquisitions during the hurricane seasons.

4.2 Expressions of Atmospheric Boundary Layer Phenomena

Mesoscale and microscale atmospheric phenomena in the Marine Atmospheric
Boundary Layer (MABL) become visible on SAR images because they are
associated with variations in the wind stress at the sea surface (Beal et al.,
1997; Elfouhaily, 1997; Thompson & Beal, 2000). The wind stress depends on
wind speed and direction at the sea surface, possibly on the relative surface
motion, but also strongly on the stability of the air–sea interface, which is a
function of the temperature difference between the water and air (Keller et al.,
1989), and on the coverage of the sea surface with surface-active material that
attenuates the short surface capillary gravity waves. Variations in the wind
stress disturb the short surface waves and thus give rise to ‘imprints’ on the
sea surface that become visible in a radar image as variations in the SAR image
intensity.
A collection of ERS SAR images acquired over the tropical and subtropical
oceans showing imprints of atmospheric phenomena on the sea surface can
be viewed at the ESA website earth.esa.int/applications/ERS‑SARtropical or
www.ifm.zmaw.de/~ERS‑sar. Some of these MABL phenomena are discussed
in the following sections.

4.2.1 Katabatic Winds

Katabatic winds are cold winds that blow down a sloping terrain (‘gravity flow’)
and, at a coast, out over the adjacent sea surface. These winds are generated
because during the evening and at night, air near the surface cools faster over
land than over the sea, resulting in a downhill flow of cold air. A large number
of ERS SAR images show sea surface signatures of katabatic wind fields in
various coastal regions adjacent to mountainous areas. This is shown in Fig. 10
for the katabatic winds blowing down from 1800 m high mountains through a
broad valley into the Gulf of Gioia Tauro in the Tyrrhenian Sea. The roughness

212
 SAR & Altimeter

Figure 10. ERS‑1 SAR image acquired at

21:14 UTC on 8 September 1992, over
the Strait of Messina and adjacent sea
areas. Over the Tyrrhenian Sea (northern
section) the sea surface signatures of
katabatic wind fields are visible. Note the
mushroom-like pattern caused by katabatic
winds blowing through a broad valley on the
Calabrian coast onto the sea.

pattern has the form of a mushroom. The wind field has been simulated
using a non-hydrostatic mesoscale atmospheric model and then compared
with the wind field derived from the ERS‑1 SAR image using the C-band wind
scatterometer model CMOD4 (Alpers et al., 1998). The comparison shows that
the atmospheric model reproduces quite well the mushroom-like form of the
wind field pattern, while the wind speed is somewhat lower than the one
inferred from the SAR image.

4.2.2 Atmospheric Gravity Waves and Boundary Layer Rolls

Atmospheric Gravity Waves (AGWs) exist in the presence of layered

atmospheres, and occur as quasi-periodic waves, solitary waves or undulating
bores. They are often generated behind mountain ranges, in which case they
are called lee waves. In the steady state lee waves are stationary with respect to
the terrain feature, but they are propagating relative to the mean air flow above
the surface. Lee waves are very common in visible remote sensing imagery
where they manifest themselves as wave-like cloud patterns. However, they
are also manifested in SAR images since they are associated with varying near-
surface wind speeds and thus varying short-scale surface roughness. Examples
of SAR observations of atmospheric waves have been reported by Vachon et al.
(1995), Alpers & Stilke (1996) and Li et al. (2004). Expressions of atmospheric
gravity waves can usually be separated from expressions of atmospheric
boundary layer rolls or wind streaks (Vachon & Dobson, 1996). The latter
provide a reliable means for wind direction retrieval via determination of the
orientation of the linear features (e.g. imprint of the rolls).

213
SP-1326

Figure 11. ERS‑1 SAR image of atmospheric

waves acquired on 30 May 1995 near the
Kamchatka Peninsula.

Figure 11 shows an ERS‑1 SAR image of atmospheric waves acquired on

30 May 1995 near the south of Kamchatka Peninsula in far eastern Russia.
The mountain range along the peninsula is clearly visible. It looks like
two atmospheric wave packets coming out of two sections (northwest and
north−south) of the mountain range. However, the two wave packets are
well connected with an indentation showing at the intersection area. Based
on the study by Li et al. (2004), this wave pattern is most likely the upstream
atmospheric waves generated by flow impeded by the mountain range on the
Kamchatka Peninsula.

4.2.3 Coastal Wind Fronts

In coastal seas bordered by mountain ranges the sea surface signatures of

wind fronts are often expressed in ERS SAR images. They can be caused by
barrier jets, katabatic winds or by synoptic-scale winds impinging on a coastal
mountain range causing recirculation of airflow, such as seen in the ERS SAR
image off the east coast of Taiwan (Fig. 12, left) (Alpers et al., 2007).
This distinct SAR image expression has been assessed in comparison with
model calculations obtained by a high-resolution atmospheric (MM5) model,
as seen in Fig.  12 (right). The observed fontal line is indeed suggested to be of
atmospheric origin and seems to be generated by the collisions of the airflows
moving in opposing directions: one is associated with a weak easterly synoptic-
scale wind blowing against the high coastal mountain range on the east coast of
Taiwan, and the other with a local offshore wind. At the convergence zone, air is
forced to move upwards, which often gives rise to the formation of precipitating
cloud bands parallel to the coast. This has been confirmed by the complementary

214
 SAR & Altimeter

Figure 12. Left: ERS‑2 SAR image acquired

on 18 January 1999 showing a distinct wind
front over the east coast of Taiwan. Right:
Schematic plot of the airflow showing the
recirculation of air causing the formation
of the atmospheric frontal line parallel to
the coast. The two vertical lines with the
horizontal red arrows denote the offshore
components of the wind vector at the
shoreline and east of the convergence line.

Japanese Geostationary Meteorological Satellite, the American Terra satellite and

rain-rate maps from ground-based weather radars (Alpers et al., 2010).

5. Currents from RA
The Radar Altimeters on ERS‑1/ERS‑2 measured the transit time and
backscatter power of the transmitted pulses. The transit time is proportional
to the satellite’s altitude above the ocean and sea ice surfaces, provided that
adequate corrections for time delays are applied. To ensure reliable atmospheric
and ionospheric corrections of the RA signals and accurate knowledge of
their position in space, Microwave Radiometer (MWR) and Precise Range and
Range-Rate Equipment (PRARE) were operated simultaneously to measure the
integrated atmospheric water vapour column and cloud liquid water content as
well as particle density in the ionosphere. Over ocean surfaces the measured
range is accurate to about 5 cm at an along-track resolution of about 5  km,
with the largest contribution to the uncertainty arising from the sea state bias
(Gaspar & Florens, 1998). Higher-range accuracy is then achieved by detailed
analysis of the received signal resulting from averaging a large number of
echoes at the expense of reduced spatial resolution. The magnitude and shape
of the returned echoes also contain information about the characteristics of the
reflecting surface, from which it is possible to retrieve geophysical parameters
such as significant wave heights, wind speeds and the freeboard height and
location of the edges of sea ice.
Measurements of the sea surface height have been routinely obtained from
satellite altimeter missions in the last 20 years. Today the annual mean sea
surface (MSS) height derived from altimetry is known with millimeter accuracy
(e.g. Cazenave et al., 2009) in the open ocean. As most changes in ocean surface
currents (on timescales of a few days or more) result in a geostrophic balance,
the sea surface height gradients derived from the ERS Radar Altimeters were
employed to investigate temporal and spatial changes in the large basin-scale
circulation patterns and mesoscale ocean dynamics. At seasonal and inter-
annual time scales the RA also detects phenomena such as the westward-
propagating Rossby waves and El Niño−Southern Oscillation that are manifest
in the sea surface topography anomalies (see also Andersen et al., this volume).

215
SP-1326

5.1 Mesoscale Fronts and Eddies

By measuring the departure of the sea surface height from its long-term mean
level at that location, the Sea Surface Height Anomaly (SSHA) obtained from
the ERS altimeters was used to detect the presence of frontal boundaries
mesoscale eddies. From the along-track gradients of the SSHA the cross-track
geostrophic velocity was then calculated at a spatial resolution of about 10 km.
This estimates the time-varying part of the surface velocity field, relative to
a mean circulation, except in a zonal band of 1–2° of the equator where the
Coriolis parameter is zero and the geostrophic assumption does not apply.
It is not possible to optimise the sampling of any single satellite mission to
observe all oceanic processes and regions, in particular at the mesoscale of
10–100 km. Since 1992 there have been at least two and sometimes as many as
four altimeters in orbit at the same time. When the orbits of these instruments
are harmonised the along-track data from each of them can be interpolated to
a common grid and thus achieve a much higher-quality product (Le Traon et
al., 1998). For the combination of Topex/Poseidon and ERS, for example, the
sea level mean mesoscale mapping error was reduced by a factor of 4 and the
standard deviation reduced by a factor of 5 (Le Traon & Dibarboure, 1999)
compared with those for Topex/Poseidon only.
Gridded maps of SSHA data provide details of the streamlines of the
geostrophic flow anomalies depicted along contours of constant height and
directed with higher values of the SSHA to the right/left in the northern/
southern hemisphere, respectively. The time-variable part of the surface
current field can thus be derived and used to study and monitor changes in
the location and strength of current fronts and eddies. Such a time sequence
of altimeter-derived SSHAs over several years yields valuable information
about the relationships between eddies on different spatial scales and their
dependence on the strength of surface currents as they vary over the seasons
and years. By merging such altimeter mission data, the mean and eddy
kinetic flow behaviours have also been characterised and investigated for the
Gulf Stream and its extension into the North Atlantic, and for the Kuroshio,
Malvinas and Agulhas currents (e.g. Le Traon et al., 1998; Ducet et al., 2000;
Okkonen et al., 2001; Fu, 2007).

5.2 Westward Propagating Planetary Waves

In most regions the ocean surface height variability at the meso- to large scale
(i.e. at wavelengths greater than a few tens of  km) is dominated by features
that propagate towards the west (a meridional, i.e. north−south propagation
component, may be present, but it is normally smaller than the westward
velocity). These features are planetary waves, also known as Rossby waves,
and nonlinear, rotating eddies. Following the seminal paper by Chelton &
Schlax (1996), many studies of large-scale westward propagation based on
satellite altimetry in the 1990s have interpreted the observed features in terms
of planetary waves (for a review, see Fu & Chelton, 2001). Although most of
those papers were based on data from the Topex/Poseidon mission, it was
nevertheless soon demonstrated that the ERS altimeters could resolve the
waves similarly well (Cipollini et al., 1997a). Moreover, accurate infrared data
from the ERS‑1/ERS‑2 Along-Track Scanning Radiometer (ATSR) instruments
provided the first evidence of the presence of these features in the sea surface
temperatures (Cipollini et al., 1997b), which were finally shown by Hill et al.
(2000) to be nearly ubiquitous.
The satellite observations allow estimations of the characteristics of
propagating features, such as westward speed, amplitude and wavelength.
It was immediately found that westward-propagating features travel faster
than predicted by the classical (‘linear’) theory for Rossby waves. This finding

216
 SAR & Altimeter

in turn stimulated a number of efforts to extend the theoretical framework to

more representative cases, for instance by including a mean flow (Killworth et
al., 1997) and realistic bottom topography (Killworth & Blundell, 2004, 2005).
Such theoretical advances have greatly reduced the discrepancies between
predicted and observed speeds.
Figure 13 (left) demonstrates how clear the signature of westward-
propagating features is in the ERS RA dataset showing a longitude/time section
(at 28°S, 50°− 80°E in the Indian Ocean) of Sea Surface Height Anomaly data
from the Radar Altimeter on ERS‑2, covering a time span of more than eight
years. The diagonal stripes in the plot (diagonal alignments of positive and
negative anomalies) are the westward propagating features, i.e. eddies and/or
Rossby waves. The wavenumber/frequency spectrum of the data (Fig. 13, right)
shows a ridge of energy broadly aligned with the line at the propagation speed
predicted by the extended theory for planetary waves (black dashed line),
while it is clear that the dispersion curve from the old ‘linear’ theory (magenta
curve) could not explain the observed spectrum.

Figure 13. Left: Longitude/time plot of

SSHAs (in m) at 28°S in the Indian Ocean,
from the ERS‑2 RA. Right: Wavenumber/
frequency spectrum of the plot on the left.
The diagram has been flipped horizontally
so that westward-propagating signals are
in the bottom left quadrant. The magenta
curve is the approximate (flat bottom and
no background flow) dispersion relationship
for planetary waves with a Rossby radius
of deformation of 42.9 km, the typical
value in the area (Chelton et al., 1998). The
black dashed line corresponds to a speed
of 4.7 cm s−1, which is the median speed
from the extended theory of Killworth et al.
(1997) over the area. (Data extracted from
the RA Database System (RADS) archive:
http://rads.tudelft.nl)

217
SP-1326

The advent of optimally interpolated maps of SSHA from the merging of a

10-day orbital repeat mission (first Topex/Poseidon, and then Jason-1 and -2)
with data from the ERS satellites (and then Envisat) finally made it possible to
detect and analyse westward-propagating features with significantly increased
spatial resolution (Chelton et al., 2007), and thus allowed the characterisation
of these features at scales of the order of 100−500 km. In these higher-resolution
data it is apparent that a large part of the variability at extratropical latitudes
takes the form of westward-propagating, eddy-like features, nonlinear and
with spatial scales close to the lower limit for planetary waves (Chelton et al.,
2011). As a consequence, many of the smaller-scale features seen in Fig. 13 (left)
can be interpreted as nonlinear eddies.

6. Summary and Outlook

Since 1991 the ERS active microwave sensors on ERS‑1/ERS‑2, notably the
SAR and Radar Altimeters, and their successors, have clearly contributed to
advances in the study, monitoring and understanding of the upper ocean. By
2011 these two sensors together had contributed to more than 1000 publications
in peer-reviewed journals. Key products today include merged global ocean
surface topography using the data available from various altimeter missions
(www.aviso.oceanobs.com) and merging of data related to the sea state from
various instruments and platforms (www.globwave.org) with particular efforts
to provide consistent tracking of travelling swell systems and crossing sea
statistical occurrences, as well as the emerging range Doppler velocity products
at both Level 2 and Level 3 (http://soprano.cls.fr/L3/fireworks.html). Accordingly,
analyses are now under way to characterise more precisely the variability
in the global ocean at scales of tens to hundreds of kilometres. Also routinely
assimilated by numerous forecasting systems, these efforts serve both research
and applications such as those associated with marine services established
under the global monitoring for environment and security (www.gmes.info) and
presently routinely operated in MyOcean (www.myocean.eu.org).
The ERS missions, extended with the Envisat mission, have therefore
been invaluable in the preparations for the launch and routine availability
of Sentinel‑1 SAR and Sentinel‑3 altimeter data with tentative launch in
October 2013 (Sentinel‑1) and November 2014 (Sentinel‑3). Building on existing
processing methods and retrieval algorithms developed for and applied to the
ERS and Envisat SAR and altimeter data, combined with available in situ data
and numerical simulations, it is anticipated that the Sentinel‑1 and Sentinel‑3
missions will further advance our capability to monitor, quantify and describe
the high-resolution upper ocean dynamical state.
The Sentinel missions will be able to capitalise on the valuable archive of
satellite active microwave SAR and Radar Altimeter data acquired over 25 years,
with global coverage, excellent quality and easy and user-friendly data access.
The next generation of Sentinel active microwave data will stimulate new and
more research, increase the range of applications and broaden and strengthen
marine services with subsequent societal benefits. In particular, the next
decade of regular SAR and altimeter observations should lead to:

—— advances in quantitative understanding of upper ocean dynamics at the

meso- to sub-mesoscale, including frontal zone instabilities, eddy generation
and decay, and the importance of convergence and divergence;
—— regular high-resolution information on surface waves and surface currents, as
well as their interactions, which is invaluable for oil spill monitoring as well
as managing responses to emergencies and search and rescue operations;
—— intensified studies and greater understanding of planetary waves and their
role in the zonal transport of heat across ocean basins;

218
 SAR & Altimeter

—— strengthening the consistent use of SAR and altimeters in synergy with high-
resolution optical sensors; and
—— better scientific and operational use of merged high-resolution SAR imaging
sensor data and nadir-profiling altimeters taking into account new approaches
for measuring surface velocities and topography directly from satellites such
as NASA’s Surface Water and Ocean Topography (SWOT) mission and ESA’s
Wavemill instrument mission.

Acknowledgements
The advances in satellite oceanography based on the ERS Synthetic Aperture
Radar and Radar Altimeter observations over the last 20 years have been
possible thanks to the involvement and efforts of a large range of scientists,
engineers and technicians with support from national and international
funding agencies, as well as the steady supply of data and scientific support
studies initiated by ESA as well as other space agencies.

References
Alpers, W. (1985). Theory of radar imaging of internal waves. Nature 314, 245−247.
Alpers, W. & Salusti (1983). Scylla and Charybdis observed from space. J. Geophys.
Res. 88, 1800−1808.
Alpers, W. & Stilke, G. (1996). Observation of a nonlinear wave disturbance in the
marine atmosphere by the synthetic aperture radar aboard the ERS‑1 satellite,
J. Geophys. Res. 101, 6513−6525.
Alpers, W., Brandt, P. & Rubino, A. (2008). Internal waves generated in the Straits of
Gibraltar and Messina: Observations from space, in V. Barale & M. Gade (Eds.),
Remote Sensing of the European Seas. Springer, Heidelberg, pp.319−330.
Alpers, W., Campbell, G. Wensink, H. & Zheng, Q. (2004). Bottom Topography:
Synthetic Aperture Radar Marine User's Manual. Chapter 10, Underwater
topography. Sponsored by the NOAA/NESDIS Office of Research and Applications.
www.sarusersmanual.com
Alpers, W., Chen, J.-P. Lin, I.-I. & Lien, C.-C. (2007). Atmospheric fronts along the
east coast of Taiwan studied by ERS synthetic aperture radar images. J. Atmos.
Sci. 64, 922−937.
Alpers, W., Chen, J.P., Pi, C.-J. & Lin, I.-I. (2010). On the origin of atmospheric frontal
lines off the east coast of Taiwan observed on space-borne synthetic aperture
radar images, Mon. Wea. Rev. 138, 475−496. doi:10.1175/2009MWR2987.1
Alpers, W. & Hennings, I. (1984). A theory of the imaging mechanism of underwater
bottom topography by real and synthetic aperture radar. J. Geophys. Res. 89,
10529−10546.
Alpers, W., Pahl, U. & Gross, G. (1998). Katabatic wind fields in coastal areas studied
by ERS‑1 synthetic aperture radar imagery and numerical modeling. J. Geophys.
Res. 103, 7875−7886.
Apel, J.R., Holbrook, J.R., Liu, A.K. & Tsai, J. (1985). The Sulu Sea Internal Soliton
Experiment. J. Phys. Oceanogr. 15, 1625−1651.
Avicola, G.S., Moum, J.N., Perlin, A. & Levine, M.D. (2007). Enhanced turbulence
due to the superposition of internal gravity waves and a coastal upwelling jet.
J. Geophys. Res. 112, C06024. doi: 10.1029/2006JC003831
Beal, R., Kudryavtsev, V., Thompson, D., Grodsky, S., Tilley, D.D., Dulov, D. &
Graber, H. (1997). The influence of the marine atmospheric boundary layer on
ERS 1 synthetic aperture radar imagery of the Gulf Stream. J. Geophys. Res.
102(C3), 5799−5814.
Beal R.C. & Pichel, W.G. (2000). Coastal and marine applications of wide swath
SAR. Johns Hopkins APL Tech. Dig. 21, 176.

219
SP-1326

Boebel, O., Rossby, T., Lutjeharms, J.R.E., Zenk, W. & Barron, C. (2003). Path and
variability of the Agulhas return current. Deep Sea Res. 50, 35−56.
Brandt, P., Alpers, W. &. Backhaus, J.O. (1996). Study of the generation and
propagation of internal waves in the Strait of Gibraltar using a numerical model
and synthetic aperture radar images of the European ERS 1 satellite. J. Geophys.
Res. 101, 14 237-14 252.
Brandt, P., Rubino, A., Alpers, W. & Backhaus, J.O. (1997). Internal waves in the
Strait of Messina studied by a numerical model and synthetic aperture radar
images from the ERS‑1/2 satellites. J. Phys. Oceanogr. 27, 648−663.
Cazenave, A., 2009. Sea level budget over 2003–2008: A reevaluation from GRACE
space gravimetry, satellite altimetry and Argo. Global and Planetary Change
65(1–2): 83–88.
Cazenave, A., Dominh, K., Guinehut, S., Berthier, E., Llovel, W., Ramillien,
G., Ablain, M. & Larnicol. G. (2009). Sea level budget over 2003−2008: A
reevaluation from GRACE space gravimetry, satellite altimetry and Argo. Global
Planet. Change 65(1−2), 83−88.
Chapron, B., Collard, F. & Ardhum, F. (2005). Direct measurements of ocean surface
velocity from space: Interpretation and validation. J. Geophys. Res. 110, C07008.
Chapron, B., Fouhaily, T.E. & Kerbaol, V. (1995). Calibration and Validation of ERS
Wave Mode Products, Document DRO/OS/95-02, IFREMER, Plouzane, France.
Chelton, D.B. & Schlax, M.G. (1996). Global observations of oceanic Rossby waves.
Science 272, 234–238.
Chelton, D.B., Schlax, M.G. & Samelson, R.M. (2011). Global observations of
nonlinear mesoscale eddies. Prog. Oceanogr. 91(2), 167−216. doi:10.1016/j.
pocean.2011.01.002
Chelton, D.B., Schlax, M.G., Samelson, R.M. & de Szoeke, R.A. (2007). Global
observations of large oceanic eddies. Geophys. Res. Lett. 34, L15606.
doi:10.1029/2007GL030812
Chelton, D.B., de Szoeke, R.A., Schlax, M.G., Naggar, K.E. & Siwertz, N. (1998).
Geographical variability of the first baroclinic Rossby radius of deformation.
J. Phys. Oceanogr. 28, 433–460.
Cipollini, P., Cromwell, D., Jones, M.S., Quartly, G.D. & Challenor, P.G. (1997a). The
potential of ERS for the detection of Rossby waves in the Northeast Atlantic.
Proc. 3rd ERS Symp., Space at the Service of our Environment, Florence, Italy,
17−21 March. ESA  SP-414, Vol. 3, European Space Agency. Noordwijk, the
Netherlands, pp.1473−1478.
Cipollini, P., Cromwell, D., Jones, M.S., Quartly, G.D. & Challenor, P.G. (1997b).
Concurrent altimeter and infrared observations of Rossby wave propagation
near 34°N in the Northeast Atlantic. Geophys. Res. Lett. 24, 889–892.
da Silva, J.C.B. & Helfrich, K.R. (2008). Synthetic Aperture Radar observations of
resonantly generated internal solitary waves at Race Point Channel (Cape Cod).
J. Geophys. Res. 113, C11016. doi:10.1029/2008JC005004
da Silva, J.C.B, New, A.L. & Azevedo, A. (2007). On the role of SAR for observing
local generation of internal solitary waves off the Iberian Peninsula. Can. J.
Remote Sens. 33, 388−403.
da Silva, J.C.B., New, A.L. & Magalhaes, J.M. (2009). Internal solitary waves in the
Mozambique Channel: observations and interpretation. J. Geophys. Res. 114,
C05001. doi:10.1029/2008JC005125
da Silva, J.C.B., New, A.L. & Magalhaes, J.M. (2011). On the structure and
propagation of internal solitary waves generated at the Mascarene plateau in the
Indian Ocean. Deep-Sea Res. 58, 229−240.
DiGiacomo, P.M. & Holt, B. (2001). Satellite observations of small coastal ocean
eddies in the Southern California Bight. J. Geophys. Res., 106 (C10), 22 521–22 544.
Drobinski, P. & Foster, R. (2003). On the origin of near-surface streaks in the
neutrally-stratified planetary boundary layer. Bound.-Layer Meteorol. 108(2),
247–256.

220
 SAR & Altimeter

Ducet, N., Le Traon, P.-Y. & Reverdin, G. (2000). Global high-resolution mapping of
ocean circulation from TOPEX/Poseidon and ERS‑1 and -2. J. Geophys. Res. 105,
19 477−19 498.
Elachi, C. (1987). Spaceborne Remote Sensing: Applications and Techniques. IEEE
Press, New York.
Elfouhaily, T. (1997). Physical Modelling of Electromagnetic Backscatter from the
Ocean Surface: Application to retrieval of wind fields and wind stress by remote
sensing of the marine atmospheric boundary layer. PhD dissertation, Département
d’ Océanographie Spatiale, IFREMER, France.
Engen, G. & Johnsen, H (1995), SAR-Ocean wave inversion using image cross
spectra. IEEE Trans. Geo. Rem. Sens. 33, 1047−1056.
Espedal, H.A., Johannessen, O.M., Johannessen, J.A., Dano, E., Lyzenga, D. &
Knulst, J.C. (1998). Coastwatch ’95: A tandem ERS‑1/2 SAR detection experiment
of natural film on the ocean surface. J. Geophys. Res.: Oceans 103, 24 969–24 982.
Etling, D. & Brown, R. (1993). Roll vortices in the planetary boundary layer: A
review. Bound. Layer Meteorol. 18(3), 215–248.
Fetterer, F., Gineris, D. & Wackerman, C. (1998). Validating a scatterometer wind
algorithm for ERS‑1 SAR. IEEE Trans. Geosci. Rem. Sens. 36(2), 476–492.
Fu, L.-L. (2007). Interaction of mesoscale variability with large-scale waves in the
Argentine Basin, J. Phys. Oceanogr. 37, 787−793.
Fu, L.-L. & Chelton, D.B. (2001), Large scale ocean circulation, in L.-L. Fu & A.
Cazenave (Eds), Satellite Altimetry and Earth Sciences. Academic Press, San
Diego, CA.
Gaspar, P. & Florens, J.P. (1998). Estimation of the sea state bias in radar altimeter
measurements of sea level: Results from a new non parametric method.
J. Geophys. Res. 103, 15 803−15 814
Gerling, T. (1986). Structure of the surface wind field from Seasat SAR. J. Geophys.
Res. 91(C2), 2308–2320.
Hasselmann, K., Calkoen, C. & Wensink, H. (1997). Mapping on seabed topography
to and from SAR. Proc. 3rd ERS Symp., Space at the Service of Our Environment,
ESA SP-414, European Space Agency, Noordwijk, the Netherlands, pp.1055−1058.
Hersbach, H., Stoffelen, A. & de Haan, S. (2007). An improved C-band scatterometer
ocean geophysical model function: CMOD5. J. Geophys. Res. 112, C03006−18.
doi:10.1029/2006JC003743
Hill, K.L., Robinson, I.S. & Cipollini, P. (2000). Propagation characteristics
of extratropical planetary waves observed in the ATSR global sea surface
temperature record. J. Geophys. Res. 105(C9), 21 927–21 945.
Horstmann, J. & Koch, W. (2003) High-resolution ocean surface wind fields
retrieved from spaceborne synthetic aperture radars operating at C-band. Proc.
2nd Workshop on Coastal and Marine Applications of SAR, Svalbard, Norway.
ESA SP‑565, European Space Agency, Noordwijk, the Netherlands, pp.61–67.
Horstmann, J., Koch, W., Lehner, S. & Tonboe, R. (2000). Wind retrieval over the
ocean using synthetic aperture radar with C-band HH polarization, IEEE Trans.
Geosci. Remote Sens. 38(5), 2122–2131.
Horstmann, J., Schiller, H., Schulz-Stellenfleth, J. & Lehner, S. (2004). Global wind
speed retrieval from SAR, IEEE Trans. Geosci. Remote Sens. 41(10), 2277–2286.
Johannessen, J.A., Chapron, B., Collard, F., Kudryavtsev, V., Mouche, A., Akimov,
D. & Dagestad, K.-F. (2008). Direct ocean surface velocity measurements from
space: Improved quantitative interpretation of Envisat ASAR observations,
Geophys. Res. Lett. 35(22), 1−6.
Johannessen, J.A., Janssen P., Minster J.F., Robinson, I., Rothrock, D. & Vachon,
P.W. (1998). Preface to special section on advances in oceanography and sea ice
research using ERS observations. J. Geophys. Res.: Oceans 103(C4), 7753. doi:
10.1029/97JC03412
Johannessen, J.A., Kudryavtsev, V., Akimov, D., Eldevik, T., Winther, N. & Chapron,
B. (2005). On radar imaging of current features, Part 2: Mesoscale eddy and
current front detection. J.Geophys. Res. 110, C07017.

221
SP-1326

Johannessen, J.A., Shuchman, R.A., Digranes, G., Lyzenga, D., Wackerman, C.,
Johannessen, O.M. & Vachon, P.W. (1996a). Coastal ocean fronts and eddies
imaged with ERS‑1 SAR. J. Geophys. Res. 101(C3).
Johannessen, J.A., Svendsen, E., Sandven, S., Johannessen, O.M. & Lygre, K. (1989).
Three-dimensional structure of mesoscale eddies in the Norwegian coastal
current. J. Phys. Oceanogr. 19(1), 3−19.
Johannessen, J.A., Vachon, P.W. & Johannessen, O.M. (1996b). ERS‑1 SAR imaging of
marine boundary layer processes, Earth Observation Quarterly no. 46. European
Space Agency Publications Division, ESTEC, Noordwijk, the Netherlands.
Katsaros, K.B., Vachon, P.W., Black, P.G., Dodge, P.P. & Uhlhorn, E.W. (2000). Wind
fields from SAR: Could they improve our understanding of storm dynamics?
Johns Hopkins APL Tech. Dig. 21(1), 86−93.
Keller, W.C., Wismann, V. & Alpers, W. (1989). Tower-based measurements of the
ocean C-band radar backscattering cross section. J. Geophys. Res. 94, 924−930.
Kerbaol, V., Chapron, B. & Vachon, P.W. (1998). Analysis of ERS‑1/2 synthetic
aperture radar wave mode imagettes. J. Geophys. Res. 103(C4), 7833-7846.
Killworth, P.D. & Blundell, J.R. (2004) The dispersion relation for planetary waves
in the presence of mean flow and topography. Part I: Analytical theory and one-
dimensional examples. J. Phys. Oceanogr. 34, 2692–2711.
Killworth, P.D. & Blundell, J.R. (2005). The dispersion relation for planetary
waves in the presence of mean flow and topography. Part II: Two-dimensional
examples and global results. J. Phys. Oceanogr. 35, 2110–2133.
Killworth, P.D., Chelton, D.B. & de Szoeke, R. (1997). The speed of observed and
theoretical long extra-tropical planetary waves. J. Phys. Oceanogr. 27, 1946–1966.
Koch, W. (2004). Directional analysis of SAR images aiming at wind direction.
IEEE Trans. Geosci. Remote Sens. 42(4), 702–710.
Kudryavtsev, V., Akimov, D., Johannessen, J.A. & Chapron, B. (2005). On radar
imaging of current features, Part 1: Model and comparison with observations.
J. Geophys. Res. 110, C07016.
Kudryavtsev, V., Grodsky, S., Dulov, V. & Bol'shakov, A. (1995). Observations of wind
waves in the Gulf Stream frontal zone. J. Geophys. Res. 100(C10), 20715−20727.
Kurapov, A.L., Allen, J.S. & Egbert, G.D. (2010). Combined effects of wind-driven
upwelling and internal tide on the continental shelf. J. Phys. Oceangr. 40,
737−756. doi:10.1175/2009JPO4183.1
Lehner, S., Horstmann, J., Koch, W. & Rosenthal, W. (1998). Mesoscale wind
measurements using recalibrated ERS SAR images, J. Geophys. Res. 103(C4),
7847–7856.
Le Traon, P.-Y. & Dibarboure, G. (1999). Mesoscale Mapping Capabilities of Multiple-
Satellite Altimeter Missions. J. Atmos. Oceanic Technol. 16, 1208–1223.
Le Traon, P.-Y., Nadal, F., & Ducet, N. (1998). An improved mapping method of
multisatellite altimeter data. J. Atmos. Oceanic Technol. 15, 522−534.
Li, X., Dong, C., Clemente-Colon, P., Pichel, W.G. & Friedman, K.S. (2004).
SAR observations of sea surface imprints of upstream atmospheric solitons
generated by flow impeded by an island. J. Geophys. Res. 109, C02016.
doi:10.1029/2003JC002168
Liu, A.K. (1998). Analysis of nonlinear internal waves in the New York Bight.
J. Geophys. Res. 93, 12 317−12 329.
Liu, A., Peng, C. & Schumacher, J.D. (1994). Wave−current interaction study in the
Gulf of Alaska for detection of eddies by synthetic aperture radar. J. Geophys.
Res. 99(C5), 10 075−10 085. doi: 10.1029/94JC00422
Lorenc, A.C. (1986). Analysis methods for numerical weather prediction, Q. J. R. Meterol.
Soc. 112, pp. 1177-1194,
Mitnik, L.M., Dubina, V.A. & Shevchenko, G.V. (2005). ERS SAR and Envisat ASAR
observations of oceanic dynamic phenomena in the southwestern Okhotsk Sea.
Proc. Envisat and ERS Symp., Salzburg, Austria, September 2004. ESA SP-572,
European Space Agency, Noordwijk, the Netherlands.

222
 SAR & Altimeter

Monaldo, F., Thompson, R., Beal, D., Pichel, W. & Clemente-Colon, P. (2001).
Comparison of SAR-derived wind speed with model predictions and ocean buoy
measurements. IEEE Trans. Geosci. Remote Sens. 39(12), 2587–2600.
Monaldo, F.M., Kerbaol, V. & the SAR Wind Team (2003). The SAR measurement of
ocean surface winds: An overview. Proc. 2nd Workshop on Coastal and Marine
Applications of SAR. 8-12 September, Svalbard, Norway.
Mouche, A., Hauser, D., Daloze, J.-F. & Guerin, C. (2005). Dual polarization
measurements at C-band over the ocean: Results from airborne radar observations
and comparison with Envisat ASAR data, IEEE Trans. Geosci. Remote Sens. 43(4),
753–769.
Mouche, A.A., Collard, F., Chapron, B., Dagestad, K.-F., Guitton, G., Johannessen,
J.A., Kerbaol, V. & Hansen, M.W. (2012). On the use of Doppler shift for sea
surface wind retrieval from SAR, IEEE Trans. Geosci. Rem. Sens. 50, 2901−2909.
doi: 10.1109/TGRS.2011.2174998.
Moum, J.N., Farmer, D.M., Smyth, W.D., Armi, L. & Vagle, S. (2003).Structure
and generation of turbulence at interfaces strained by internal solitary waves
propagating shoreward over the continental shelf. J. Phys. Ocean. 33, 2093−2112.
doi: 10.1175/1520-485(2003)033<2093:SAGOTA>2.0.CO;2
New, A.L. & da Silva, J.C.B. (2002). Remote-sensing evidence for the local generation
of internal soliton packets in the central Bay of Biscay. Deep Sea Res., Part I, 49,
915−934. doi: 10.1016/S0967-0637(01)00082-6
Nilsson, C.S. &. Tildesley, P.C (1995). Imaging of oceanic features by ERS1 synthetic
aperture radar. J. Geophys Res. 100, 953−967.
Okkonen, S.R., Jacobs, G.A., Metzger, E.J., Hurlburt, H.E. & Shriver, J.F. (2001).
Mesoscale variability in the boundary currents of the Alaska Gyre. Cont. Shelf
Res. 21, 1219–1236.
Osborne, A.R. & Burch, T.I. (1980). Internal solitons in the Andaman Sea. Science
208, 451−460.
Portabella, M., Stoffelen, A., & Johannessen, J.A. (2002). Toward an optimal
inversion method for SAR wind retrieval. J. Geophys. Res. 107 (C8). doi:
10.1029/2001JC000925.
Quilfen, Y., Chapron, B., Elfouhaily, T., Katsaros, K. & Tournadre, J. (1998).
Observation of tropical cyclones by high-resolution scatterometry, J. Geophys.
Res. 103(C4), 7767–7786.
Shroyer, E.L., Moum, J.N. & Nash, J.D. (2010). Vertical heat flux and lateral mass
transport in nonlinear internal waves. Geophys. Res. Lett. 37, L08601. doi:
10.1029/2010GL042715
Sikora, T.D., Young, G.S. & Winstead, T.S. (2006). A novel approach to marine wind
speed assessment using synthetic aperture radar. Wea. Forcasting. 21, 109−115.
Stoffelen, A. & Anderson, D. (1997a). Scatterometer data interpretation: Estimation
and validation of the transfer function CMOD-4. J. Geophys. Res. 102(C3),
5767−5780.
Stoffelen, A. & Anderson, D. (1997b). Scatterometer data interpretation: Measurement
space and inversion. J. Atm. Ocean. Technol. 14(6), 1298−1313.
Thompson, D. & Beal, R. (2000). Mapping of mesoscale and submesoscale wind
fields using synthetic aperture radar. John Hopkins APL Tech. Dig. 21, 58–67.
Ulaby, F.T., Moore, R.K. & Fung, A.K. (1982). Synthetic aperture side-looking
airborne radar systems. Microwave Remote Sensing, Active and Passive, Vol. II:
Radar Remote Sensing and Surface Scattering and Emission Theory. Addison-
Wesley, Reading, MA, pp.630–745.
Vachon, P.W. & Dobson, F. (1996). Validation of wind vector retrieval from ERS‑1
SAR images over the ocean. Global Atmos. Ocean Syst. 5, 177–187.
Vachon, P.W., Johannessen, J.J. & Browne, D.P. (1995). ERS‑1 SAR images of
atmospheric gravity waves. IEEE Trans. Geosci. Rem. Sens. 33(4), 1014−1025.
Vachon, P.W., Johannessen, O.M. & Johannessen, J.J. (1994). An ERS 1 synthetic
aperture radar image of atmospheric lee waves. J. Geophys. Res. 99(C11),
22483−22490.

223
SP-1326

Valenzuela, G.R. (1978). Theories for the interaction of electromagnetic and oceanic
waves: A review. Bound. Layer Meteorol. 13(1–4), 61–85.
Vlasenko, V. & Alpers, W. (2005). Generation of secondary internal waves by the
interaction of an internal solitary wave with an underwater bank. J. Geophys.
Res. 110, C02019. doi: 10.1029/2004JC002467.
Wackerman, C., Pichel, W.G. & Clemente-Colon, P. (2003). A projection method
for automatic estimation of wind vectors with Radarsat SAR imagery, Proc. 2nd
Workshop on Coastal and Marine Applications of SAR, Svalbard, Norway. ESA
SP‑565, European Space Agency, Noordwijk, the Netherlands, pp.55−60.
Wackerman, C., Rufenach, C., Schuchman, R., Johannessen, J. & Davidson, K.
(1996). Wind vector retrieval using ERS‑1 synthetic aperture radar imagery.
J. Geophys. Res. 34, 1343–1352.
Wright, J.W. (1978). Detection of ocean waves by microwave radar: The modulation
of short gravity-capillary waves. Boundary-Layer Meteorol. 13(1−4), 87−105.

224
→→ New Views of Earth using →
ERS Satellite Altimetry
 Altimeter

New Views of Earth using ERS Satellite Altimetry

O.B. Andersen1, P. Knudsen1, P. Berry2, R. Smith2, F. Rémy3, T. Flament3, S.

Calmant3, P. Cipollini4, S. Laxon5, A. Shephard6, D. McAdoo7, R. Scharroo7,
W. Smith7, D. Sandwell8, P. Schaeffer9, J. Bamber10, C.K. Shum11, Y. Yi11,
J. Benveniste12

1 Danish National Space Institute (DTU Space), Copenhagen, Denmark

2 De Montfort University, Leicester, UK
3 Laboratoire d’Etudes en Géophysique et Océanographie Spatiales, Toulouse,
France
4 National Oceanography Centre, Southampton, UK
5 University College London, UK
6 University of Leeds, UK
7 National Oceanic and Atmospheric Administration (NOAA), Washington, DC, USA
8 Scripps Institute, La Jolla, CA, USA
9 Collecte de Localisation Satellite (CLS), Toulouse, France
10 University of Bristol, UK
11 Ohio State University, Columbus, OH, USA
12 ESA Centre for Earth Observation (ESRIN), Frascati, Rome, Italy

1. Introduction
The first European Remote Sensing satellite, ERS‑1, launched in 1991, carried a
comprehensive payload including an imaging Synthetic Aperture Radar (SAR)
designed to map the cryosphere, a radar altimeter to measure the height of the
ocean and other powerful instruments to measure ocean surface temperatures
and winds at sea. 
Satellite altimetry works conceptually as follows. The satellite transmits a short
pulse of microwave radiation with known power towards the sea surface, where it
interacts with the surface and part of the signal returns to the altimeter, where the
travel time is accurately measured. The determination of the height of the surface
from the altimeter range measurement involves a number of corrections to account
for the behaviour of the radar pulse through the atmosphere (Fu & Cazenave,
2001). In the case of altimetry over the ocean, corrections for the sea state and
other geophysical signals are also applied. (Andersen & Scharroo, 2011)
Some of the most spectacular results of the ERS‑1 mission, such as the
ice sheet topography maps, the global marine gravity field or the global
bathymetry map, were obtained when the satellite was operating in a number
of ambitious mission phases:

—— First ice phase, from December 1991 to March 1992, optimised for Arctic
ice experiments characterised by a 3-day repeat cycle with a ground track
spacing of roughly 930 km at the equator.
—— First multidisciplinary phase, from April 1992 to December 1993, with a 35‑day
repeat cycle with a ground track spacing of around 80 km.
—— Second ice phase, from January 1994 to March 1994, with features identical to
those of the first ice phase.
—— The geodetic phase, from April 1994 to March 1995, in which the satellite
performed interleaved 168-day repeat track cycles resulting in a final ground
track spacing of 8 km.
—— Second multidisciplinary phase, from April 1995 until the end of the mission
on 10 March 2000.

The ERS‑1 168-day interleaved geodetic phase mission was proposed and
designed by the ERS‑1 international science team members, led by Guy

227
SP-1326

Duchossois and Richard Francis. However, the geodetic phase was planned to
last only until the launch of the ERS‑2 satellite in April 1995, giving only 1.5
cycles of 168-day repeat observations. The decision to continue the geodetic
phase and to complete the second 168-day repeat, resulting in a ground-
track spacing of 8 km, was not made until the end of 1994. Shortly before the
December 1994 American Geophysical Union (AGU) meeting there was a launch
failure at Arianespace. During the meeting Anny Cazenave, Dave Sandwell
and Richard Francis drafted a letter to ESA, which was supported by many
scientists at the meeting. In the letter they asked that, if the launch of ERS‑2
was delayed, then perhaps the second 168-day phase could be extended to
completion before resuming the 35-day operations. This was finally approved
by ESA and the scientific community had two complete 168-day cycles before
ERS‑1 was returned to its multidisciplinary phase.
With two cycles/phases of ERS‑1 Geodetic Mission (GM) data, the spatial
density of these data now equalled that of the Geodetic Mission data from the
older Geosat mission, which had been classified by the US Navy. This may
have convinced the US Navy to release all the Geosat GM data in July 1995,
providing a unique opportunity to cross-validate the two datasets and to create
in particular global marine gravity fields of unprecedented accuracy, much
higher than could have been achieved using just one satellite.
ERS‑2 was launched in 1995 to replace the older ERS‑1, and was operated
in a 35‑day repeat multidisciplinary phase throughout the mission. By the
time of the launch of ERS‑2, the original ERS‑1 was still in very good health
and went on to remain in operation until 1999, providing the first tandem
mission altimetry phase valuable to construct climate profiles. ERS‑2 remained
in operation until 2011 and was one of the longest Earth observation satellites
missions ever flown.

2. Altimeter Data Improvement

Initially, the ERS‑1 altimeter provided sea surface height observations
that were noisier than the data from the NASA/CNES Ocean Topography
Experiment (Topex) satellite launched in 1992. There were several reasons for
this, including the satellite’s Sun-synchronous orbit and higher inclination.
Consequently, many post-launch improvements were made throughout the
mission, which improved the data quality significantly, and the applicability of
the ERS‑1 mission for climate studies.

2.1 ERS Retracking

Due to the unique nature and spatial coverage of the Geodetic Mission data,
much effort was devoted to improving this important dataset. Maus (1998) and
subsequent authors had shown that the noise in retracked range data can be
reduced significantly by imposing the constraint that the Significant Wave
Heights (SWHs) must be correlated from waveform to waveform along the track
(Fig. 1). This greatly improved the quality of the ERS‑1 GM dataset.
Another way to improve the value of the ERS‑1 GM dataset was achieved by
increasing the amount of data that could be made available from the mission.
This was done by using multiple tolerant retrackers adapted to local conditions
(Fig. 2). This additional step allowed the recovery of an additional 6–10% more
data for shallow water and polar regions. In particular, in polar regions, only a
few echoes are ocean-like in shape and so most fail to be retracked by a standard
Brown waveform model. Previously, these echoes had been rejected by the space
agencies (i.e. ESA), but they could be recovered using more tolerant retrackers.

228
 Altimeter

Figure 1. Profiles of along-track sea surface

slope for six repeat cycles crossing the
South Pacific Ocean in a region of generally
high SWHs. Top: Profiles derived from the
onboard tracker available in the waveform
data record (rms = 8.23 µrad). Middle:
Profiles derived using a weighted least-
squares 3-parameter retracking algorithm
(rms = 6.45 µrad). Bottom: Profiles derived
from a 1‑parameter retracking algorithm
constrained by smoothing the rise-time and
amplitude parameters (rms = 4.01 µrad).
(Smith & Sandwell, 2005)

2.2 ERS Orbit and Range Correction Improvements

The radial orbit had long been a major source of error in ERS‑1 altimetry, which Figure 2. Altimetric height observations
was made worse by having only satellite laser ranging for precise tracking and in ice-covered regions east of Greenland.
having to rely on insufficiently accurate general-purpose gravity field models. Top: The number of data points that can
However, it was realised that altimeter crossovers can be used very effectively be retrieved using standard ESA retracked
as additional tracking data to laser ranging. The ERS‑1/ERS‑2 tandem mission 1 Hz data. Bottom: The amount of 1 Hz data
even provided the unique possibility to determine simultaneously the orbits of that can be retrieved using more tolerant
two similar satellites flying in the same orbit. Altimeter crossovers between the retrackers. (Andersen et al., 2009)
two satellites then linked the two orbits in a common reference frame. Tailoring
the Joint Gravity Model 3 (JGM-3) was another way to reduce orbit errors.
The resulting Delft Gravity Model (DGM-E04) reduced this part of the
orbit error by a factor of 2, and performed even better with respect to the
ESA-provided orbits, and also the recent Earth Geopotential Model (EGM96).
The ERS‑1 and ERS‑2 orbits for the entire tandem mission were computed
and studied in detail, and the orbit errors due to the gravity field and non-
conservative forces are identified. The analyses systematically show that the
orbits computed with JGM-3 had a radial rms orbit accuracy of 7 cm, while
those of DGM-E04 were 5 cm (Fig. 3).
In 2003, ESA funded the Reprocessing of Altimeter Products for ERS
(REAPER) project to perform homogeneous and full reprocessing of the ERS‑1
and ERS‑2 orbit files from the period July 1991 to June 2003 using up-to-date
standards that were subsequently made available to users.
During the ERS era, the space agencies established the Radar Altimetry
Data System (RADS) archive for storing, cross-validating and distributing high-
level, consistent altimetry data from all available satellites as an extension to
the data distribution system.

229
SP-1326

Figure 3. Locally averaged ERS−Topex dual

satellite crossover height differences for
the ERS‑1/ ERS-2 tandem mission. The four
panels show the most common ERS orbit
solutions: (a) GFZ PGM055, (b) DTU JGM-3,
(c) EGM96 and (d) DGM-E04.
(Scharroo et al,, 1998)

The ERS‑1 Altimeter (Phases A–G) Value-Added GDR Data Product (Berwin,
2001), was jointly developed by Ohio State University, the JPL Physical
Oceanography Distributed Active Archive Center (JPL PO.DAAC) and the
University of Texas Center for Space Research. Improved orbits, media and
ocean tide corrections have been derived and applied to the data products
(Shum et al., 1993; Anzenhofer et al., 1999; Urban et al., 2001; Ogle, 2003).

2.3 Coastal Altimetry

Among the many contributions that the ERS satellites have made to altimetric
science, one of the highlights has been the establishment and development of
coastal altimetry.
Although open-ocean altimetry is a mature technique that is now routinely
used in operational oceanography, until recently altimetric measurements of
sea level and significant wave heights in the coastal zone were flagged as bad
and discarded for technical reasons. These included the degradation of the
waveforms owing to the presence of land in the altimetric footprint, and the
inadequacy of some corrections for geophysical and atmospheric effects such
as the path delay induced by atmospheric water vapour. In recent years several
research groups have attempted to overcome such problems, and have given
rise to the new field of coastal altimetry (Vignudelli et al., 2011).
The idea that meaningful measurements in coastal zones can be retrieved is
not new. Deng et al. (2002), for instance, demonstrated, using ERS‑2 data, that
the statistics of the altimetric signal are encouraging up to just a few kilometres
from the coast. This is illustrated in Fig. 4, where the shape conforms to the
open-ocean model, and the standard deviation drops quickly with increasing
distance from the coast.
The seeds of this exciting new development for altimetry were sown in the
1990s with a number of studies based on ERS data. The first attempt to retrieve
nearshore altimetric data by custom processing was made by Manzella et al.
(1997), who recomputed the wet tropospheric correction for ERS‑1 altimeter data
over the Corsica channel by recalibrating the model correction with the closest
available radiometric estimate from the onboard ERS Microwave Radiometer,
EMR-1. This was a precursor to the Dynamically Linked Model, a technique
that is still used today (Obligis et al., 2011). Manzella et al. (1997) also recovered
measurements that had been flagged as bad because the nadir backscatter was
high due to a smooth sea surface, using another technique that has now been
adopted to improve the recovery rate of altimetric estimates in coastal regions.

230
 Altimeter

Figure 4. Mean power of the waveform

(left) and the standard deviation of the
mean waveform (right) for ERS‑2 in five
2-km-wide bands from the Australian coast.
(Deng et al., 2002)

An extensive study of coastal altimetry, taking into account all the

issues in the reprocessing, was carried out by Anzenhofer et al. (1999). They
described the generation of coastal altimetry data and were the first to analyse
in detail the various retracking algorithms and their implementation. They
showed some examples based on the intermediate ERS waveform data, and
concluded with some recommendations for better (local) tidal modelling,
careful screening of the data, improvement of the wet tropospheric correction
and retracking, which still represent the main areas in which coastal altimetry
research is concentrating its efforts (Vignudelli et al., 2011). This pioneering
work with ERS altimetry data prompted ESA and CNES to launch two projects,
COASTALT and PISTACH, to study and further exploit coastal altimetry data.

3. Mean Sea Surface Mapping

The Mean Sea Surface (MSS) is an important parameter in geodesy and physical
oceanography. It is the time-averaged physical height of the ocean surface. In
principle, a complete separation of the ocean mean and variable part requires
uninterrupted infinite sampling in both time and space. The challenge in MSS
mapping is to achieve the most accurate filtering of the sea surface variability
over a limited time span and simultaneously obtain the highest spatial
resolution. This is normally achieved by combining data from the highly
accurate Exact Repeat Mission (ERM) with that from the ERS‑1 non-repeating
Geodetic Mission and the older (1985) Geosat Geodetic Mission (Fig. 5).
Since 1996, several determinations of the Mean Sea Surface have been
made at the Collecte de Localisation Satellite (CLS) in Toulouse. For these
surfaces data from the two 168-day interleaved ERS‑1 Geodetic Mission cycles
were extremely important because they provided the spatial resolution of
the finest geophysical structures at the shortest wavelengths with a track-to-
track distance of 8 km at the equator and a sample of 7 km along tracks. As a
complement, data from the Exact Repeat Mission of ERS‑1, ERS‑2 and Envisat
were used to calculate mean profiles over a long period, from 1992 to the

231
SP-1326

Figure 5. An early Mean Sea Surface

derived from three years of satellite
altimetry data from Topex and ERS‑1.
(Yi, 1995)

Figure 6. Mean Sea Surface models CLS-SHOM 1998 (left) (Schaeffer et al., 1998) and CNES-CLS11 (right) (Schaeffer et al., 2012).

present. The first European MSS was calculated at CLS in 1998, with funding
from the Hydrographic and Oceanographic Service of the French Navy (SHOM).
In 2001 the CLS01 MSS was computed and in 2011, 20 years after the launch
of ERS‑1, CLS determined the CNES-CLS11 MSS using 16 years of altimetry data
from Topex/Poseidon/Jason-1 and, respectively, 14 years of data from ERS/Envisat,
7  years of data from the Geosat Follow-On (GFO), 3 years of data from Topex/
Poseidon interlaced and the two 168-day non-repeat ERS‑1 subcycles (Fig. 6).
MSS models were also developed at the Danish National Space Center
(DNSC/DTU). The DTU10 Mean Sea Surface is one of the most recent geometric
descriptions of the averaged height of the ocean surface derived from a
combination of 17 years of altimetry data from eight satellites covering the
period 1993–2004 (Fig. 7). This is a truly global MSS derived at 1 min resolution,
and includes all of the Arctic Ocean by including laser altimetry from Icesat
and a sophisticated extension towards the North Pole using the geoid.

232
 Altimeter

Figure 7. The differences between the

Mean Sea Surface models CNES-CLS11 and
DTU10, of the order of 1–3 cm rms, clearly
demonstrate the high level of consistency.
Locally (at high latitudes, in areas of strong
variability) the differences can exceed
10 cm, particularly in the Arctic Ocean
where the CNES-CLS11 surface coverage
is limited. These differences provide good
indicators of the consistency and relative
accuracy of the MSS models derived at
various institutions. (Andersen & Knudsen,
2009)

4. Gravity and Tectonics

The sea surface mimics the geoid, which is an equipotential surface of Earth’s
gravity field. Matter is not evenly distributed on or below the seafloor (or
further within the crust and mantle), causing variations in the pull of gravity
that produce tiny variations in ocean surface height. By accurately mapping
these variations, the process can be inverted and the pull of gravity can be
deduced from observations of sea surface heights (Fig. 8).
Today, several global marine gravity fields at 1 min resolution are available
for free download from the internet. These include the NTU fields (Hwang et
al., 2000), the Sandwell & Smith fields (Sandwell & Smith, 1997), the KMS02
fields (Andersen & Knudsen, 1998), the GSFC fields (Wang, 2001), and the CLS-
SHOM fields (Hernandez & Schaeffer, 2000); see Fig. 9.
The derivation of a marine gravity field from satellite altimetry over
permanently ice-covered regions of the Arctic Ocean has provided much new
geophysical information about the structure and development of the Arctic
seafloor. Because of its remote location and permanent ice cover, the Arctic
Ocean, from a tectonic point of view, remains the most poorly understood ocean
basin. A gravity field has been derived with data from the ERS‑1 radar altimeter,
including permanently ice-covered regions. The gravity field shown in Fig. 10
clearly delineates sections of the Arctic Basin margin along with the tips of the

Figure 8. One of the first examples of

regional gravity field mapping in the North
Atlantic between Norway, Iceland and
Greenland, using ERS‑1 radar altimeter
data from the 35‑day multidisciplinary
phase. Left: Map showing known geological
features; right: gravity anomalies given in
mGal. (Knudsen et al., 1992)

233
SP-1326

Figure 9. Three ERS‑1 GM derived global

marine gravity fields. Top: DNSC08 global
free air gravity anomalies (Andersen et al.,
2009); middle: the Sandwell & Smith field
v18.2; bottom: the CLS-SHOM1999 gravity
anomaly field.

Lomonosov and Arctic mid-ocean ridges. Several important tectonic features of

the Amerasia Basin are clearly expressed in this gravity field. These include the
Mendeleev Ridge; the Northwind Ridge; details of the Chukchi Borderland; and
a north–south trending, linear feature in the middle of the Canada Basin that
apparently represents an extinct spreading centre that ‘died’ in the Mesozoic era.
A corresponding marine gravity field for the Antarctic region from ERS‑1
altimetry (Fig. 11) reveals the southernmost traces of fracture zones (FZs), e.g. the

234
 Altimeter

Pahemo and Endeavour FZs in the southern Amundsen Sea, all the way to their
intersection with the west Antarctic continent (see McAdoo & Laxon, 1997). These
traces of fracture zones place constraints on the early history (65–80 million years
ago) of seafloor spreading between the Antarctic and the New Zealand micro-
continent and support the existence of a palaeo-tectonic plate, the Bellingshausen.
For tectonics, the gravity field indicates a strong anomaly within the
Ross Sea that may constrain models of recent ice sheet history in the Ross
embayment. It has been suggested that one-half of this gravity low could
be attributed to the effect of the loading of the crust by the weight of a much
larger sheet during the early Holocene, 10 000–20 000 years ago. It has been
proposed that such an ice sheet may have covered the Ross Sea during that
period. The rheology of the oceanic lithosphere in this area is dominated by
the inner distribution of temperature, the lithosphere cooling slowly from the
ridge where it is formed to the old abyssal basins. Owing to the global coverage
offered by satellite altimetry, the equivalent elastic thickness of the oceanic
lithosphere has been investigated on a global scale (Calmant & Baudry, 1996)
and regional anomalies of the thermal structure of the oceanic lithosphere Figure 10. The regional altimetric gravity
in the Central Pacific have been reported (Calmant et al., 2002). Recently, a field for the Arctic Ocean derived from
derivation of retracked gravity anomalies over smaller inland water bodies like ERS‑1 radar altimetry. This is one of the
the Great Lakes was presented by Kingdon et al. (2008). first examples of gravity determined under
the ice-covered Arctic Ocean. (Laxon &
McAdoo, 1994)
5. Bathymetry
A detailed knowledge of the topography of Earth’s surface is fundamental for
understanding most physical processes. On land, weather and climate are
controlled by topography on scales ranging from large continental landmasses
to small mountain valleys. Since the land is shaped by tectonics, erosion and
sedimentation, an understanding of topography is essential for any geological
investigation.
In the oceans, a detailed knowledge of bathymetry is also essential for
understanding physical oceanography, biology and marine geology. Currents
and tides are steered by the overall shapes of the ocean basins as well as by
smaller sharp ocean ridges and seamounts. Recent reports suggest that the
interaction of tides and currents with the rugged seafloor mix the ocean to
provide global overturning. Sea life is abundant where rapid changes in ocean
depth deflect nutrient-rich water upwards towards the surface.
Because erosion and sedimentation rates are low in the deep oceans,
detailed bathymetry also reveals mantle convection patterns, plate boundaries,
the cooling/subsidence of the oceanic lithosphere, oceanic plateaus and the
distribution of volcanoes (Fig. 12). Figure 11. The first complete altimetric
Because it is not possible to map the topography of the ocean basins directly gravity field for the entire Southern
from space, most seafloor mapping is a tedious process that has been carried Ocean derived from ERS‑1 radar altimetry,
out over a 40-year period by research vessels equipped with echo sounders. So including ice-covered seas and extending
far, only a few percent of the oceans have been surveyed at 200 m resolution. to the southernmost ocean limit of 79°S,
It has been estimated that 125–200 ship-years of survey time will be needed i.e. excluding the Ross and Ronne−Filchner
to map the deep oceans, costing several billion dollars. Mapping shallow seas ice shelves (McAdoo & Laxon, 1997; Laxon
would take even more time and funding. Therefore, the geodetic missions & McAdoo, 1998) The southern limits of
such as those of Geosat and ERS‑1 included high-resolution mapping of the ERS, Icesat and Cryosat-2 altimetry are
oceanic domain that has been of primary importance in seafloor mapping and shown by the green, blue and red circles,
seamount detection (Baudry & Calmant, 1996). respectively, overlying the grey Antarctic
In the wavelength band 10–160 km, variations in gravity anomalies are continent.
highly correlated with seafloor topography and thus, in principle, can be used to
recover topography in an inversion process (Baudry & Calmant, 1991; Calmant &
Baudry, 1996; Calmant et al., 2002; Andersen & Knudsen, 2009). For longer and
shorter wavelengths, isostatic compensation and noise are limiting factors.

235
SP-1326

Figure 12. Two examples of global

bathymetric models derived from satellite
altimetry. Top: Sandwell & Smith v7.1;
bottom: DNSC08.

6. Sea Ice
6.1 Arctic Sea Ice Thickness
ERS satellite altimetry has been instrumental in determinining the mean
thickness of Arctic ice and its variability between 65°N to 82°N over the period
1993–2001. The ERS data reveal a high-frequency inter-annual variability
in mean Arctic ice thickness that is dominated by changes in the amount
of summer melt, rather than by changes in circulation. To deduce the ice
thickness from ice elevation, the source of the echoes scattered from snow-
covered sea ice must be determined. Laboratory experiments show that, under
dry cold snow conditions, a normal-incidence 13.4‑GHz radar reflection from
snow-covered sea ice originates at the snow–ice interface. The ERS radar
altimeter measurements of ice elevation therefore provide the level of the
snow–ice interface above the water level, that is, the ice freeboard. The ice
freeboard measurements are subsequently converted to thickness and mass
(Fig. 13).

236
 Altimeter

Figure 13. Left: Average winter Arctic sea ice thickness from October 1993 to March 2001 from ERS altimeter measurements of the
ice freeboard. Right: Comparison of satellite altimeter- and submarine-derived ice thicknesses in the Beaufort Sea during the 1990s.
Submarine thicknesses are shown for each of the 50-km segments gathered during four missions during the 1990s. Altimeter thickness
estimates are generated from observations within 15 days and 100 km of the submarine draught sections. The submarine thicknesses
exclude thin ice (0.5 m) and open water, because (owing to difficulties in discriminating thin ice from open water) progressively more
altimeter data are excluded once the ice thickness falls below 1 m. The error bars show uncertainties in altimeter thickness due to
measurement errors and snow depth variability.

6.2 Antarctic Sea Ice Loss

By combining ERS measurements and the results of a coupled ice–ocean
model, Shepherd et al. (2010) were able to provide the first estimates of changes
in the quantity of ice floating in the global oceans and their consequent
contribution to changes in sea level.
Their results show that rapid losses of Arctic sea ice and small Antarctic
ice shelves are partially offset by thickening of Antarctic sea ice and large
Antarctic ice shelves (Fig. 14). Altogether, 746 ± 127 km3 yr–1 of floating ice was
lost between 1994 and 2004, a value that exceeds considerably the reduction
in grounded ice over the same period. Although the losses are equivalent to
a small (49  ±  8  mm  yr−1) rise in mean sea level, there may be large regional
variations in the degree of ocean freshening and mixing.

237
SP-1326

Figure 14. Average rate of rate of change

in the thickness of Antarctic ice shelves,
1994–2008, determined from ERS and
Envisat radar altimetry and a model of
accumulation fluctuations. (Shepherd et al.,
2010)

7. Ice Sheet Elevation Models

Ice sheet research was one of the key targets for the ERS‑1 programme.
Compared with other satellite and field data, ERS‑1 allowed significant
advances, particularly in polar ice sheet monitoring. The key factors were the
improved spatial coverage and improved latitudinal coverage compared with
previous satellites. Improved microwave sensor technology also made a major
contribution. Finally, the ERS‑1 altimeter was designed with a specific mode for
mapping ice sheet topography and investigating volume changes.
The main physical processes (climatic and dynamic) that act on an ice sheet
induce particular signatures on its free surface. From the small scales to the
larger ones, topography contains important information about local anomalies
or general trends in behaviour (Rémy et al., 1999). Accurate knowledge of
the topography of an ice sheet also provides an initial condition on which to
base analyses of its future evolution. At the large scale, the surface is quasi-
parabolic and is governed by ice viscosity and by flow boundary conditions
(Fig. 15). The topography is mostly controlled by the distance from the coast
(Vaughan & Bamber, 1998), except where basal conditions, such as sliding,
affect the topography in the upslope direction. The large-scale topography
controls the flow direction and its mapping allows the calculation of balance
velocities.
At largest scale, networks of anomalies of the surface topography
perpendicular to the steepest slope direction can be pointed out. They are due
to boundary flow conditions of outlet glaciers that are propagated from the
coast up to the dome (Rémy & Minster, 1997). They indicate drainage patterns,
upstream glacier positions or the flowline directions, but also the role of the
outlet flow conditions on the whole shape.
At the 10 km scale, topographic signatures of the flat ice shelf make it
possible to delimit adjacent glaciers and to visualise lateral stress effects
(Fig.  16). ‘En echelon‘ structures with a tilted orientation of 45° with respect

238
 Altimeter

Figure 15. Complete elevation map of the Greenland ice sheet derived mainly from ERS‑1 altimetry data. Left: The topographic relief has
been spatially filtered to highlight the shorter-wavelength features in the ice sheet and to illustrate ice sheet dynamics. (Bamber et al.,
1998). Right: Topographic relief map of the Antarctic ice sheet (Bamber et al., 2009).

Figure 16. High-resolution map of Antarctic

ice sheet topography from the ERS‑1
Geodetic Mission. The elevation reaches
4000 m. Note the ice shelves surrounding
most of the continent, surface undulations,
flat areas reflecting subglacial lakes
and elongated scars due to hydrological
networks. (Rémy et al., 1999)

239
SP-1326

to the flow direction have been found but have not yet been explained. In
the same way, undulations at the 10 km scale wavelength with a metric scale
amplitude seem to be the main ice sheet surface features. They have been
attributed to ice flow above an irregular bedrock. Up to now, they have been
modelled as symmetrical dome-shaped features. Their spatial characteristics
may allow a better understanding and modelling of these structures; they have
been found to be elongated, frequently with an orientation of 45° with respect
to the flowline direction. The 10 km scale seems to be a characteristic scale of
ice flow processes, as discussed in section 8.

8. Ice Sheet Mass Balance

Understanding ice sheet dynamics is a key objective of glaciological work in
both Antarctica and Greenland, along with the motion and deformation of
the ice and relating these to the mass balance and ice volume. Very early into
the mission changes in the elevation of the Antarctic Ice Sheet were reported
(Wingham et al., 1998). The mass balance for the Antarctic ice sheet is mapped
in Fig. 17 for the ERS and Envisat periods 1992–2003 and 2002–6, respectively.
Rémy & Parouty (2009) and Flament & Rémy (2012) extracted the temporal
trend by correcting for the radar penetration with the help of the backscattering
coefficient and the whole waveform shape. The amplitude of local fluctuations
can be as large as 15 cm yr−1 and are attributed to small temporal changes in
precipitation. The difference between the two maps in Fig. 17 has the same
ranges of variation as the average map, meaning that more than 50% of the
signal is not constant over time. Indeed, one can see differences in the patterns
of change between the ERS and Envisat periods. As soon as the distance
to the coast is greater than 200 km, the amplitude of the variations falls to
a few centimetres per year. The centre of the east Antarctic ice sheet is then
stationary over a few thousand kilometres.
Most of the studies of the Antarctic ice sheet mass balance using altimetry
(see Zwally et al., 2005; Wingham et al., 2006; Rémy & Parouty, 2009) have
found few areas of thickening in the eastern part of Wilkins Land or in the
Antarctic Peninsula, and few areas of thinning near the Pine Island Bay
sector and near Law Dome. Most changes in ice surfaces near the coast have
been attributed to changes in ice dynamics due to perturbation of the outlet
glaciers, as previously explained. Depending on the kind of corrections made,
especially to account for penetration issues, various authors find this part to be
slightly thinning or slightly thickening.
The behaviour of the western part of Antarctica is quite different. The
West Antarctic Ice Sheet (WAIS) is a marine ice sheet, since most of its
bedrock is below sea level, and it is therefore very sensitive to increases in

Figure 17. Mass balance of the Antarctic ice

sheet (in m yr–1) for the ERS period (1992–
2003, left) and the Envisat period (2002–6,
right). Note the significant thinning over a
large sector of the western ice sheet. The
eastern part of the continent exhibits less
impressive signals, but also shows local
fluctuations that depend on the period.
(Rémy & Parouty, 2009)

240
 Altimeter

ocean water temperature. It receives a relatively higher geothermal flux than

the eastern part, so that the basal temperature is higher (Llubes, 2006). As
a consequence, the temperature of most of the ice lying on the bedrock rises
close to melting point and the ice slides down at higher velocity. This sector
exhibits volume fluctuations over a wide range of timescales. Mass balance
observations support the different behaviours of these two parts. Indeed, all
attempts to derive the mass balance from altimetry measurements agree that
there is a clear thinning of the drainage basin of the Pine Island and Thwaites
glaciers in the western part of the WAIS. One team found that the whole of
western Antarctica is thinning of at a rate of 47 Gt  yr–1 (Zwally et al., 2005),
corresponding to a rise in sea level of 0.12 mm yr–1. Another team (Shepherd et
al., 2003; 2006; Wingham et al., 2006) detected dynamic responses of the ice
sheet leading to either thickening or thinning, as shown in Figs 18 and 19.
In summary, Antarctica happens to be close to equilibrium but strong
regional signals suggest that the local dynamical responses of ice may
accelerate in the future. The glaciers of both east and west Antarctica may
provide a substantial contribution to global sea level change (Shepherd et al.,
2001, 2008). The altimeter is not only a useful tool for mapping global trends,
but it can also identify some local imbalances. However, the discrepancies in
studies of some areas may be explained by the limitation of retrieving height
variations due to the penetration of the altimetric radar wave in the snowpack.
Such an effect has been pointed out in the Amery ice shelf. Densification
processes also yield critical errors because they act directly on the ice surface
and indirectly via changes in snowpack characteristics. For some authors,
the changes in the magnitudes of firn depth are comparable with observed
changes in ice sheet elevation (Sundal et al., 2012).

Figure 18. A: Changes in the elevation

of the Antarctic (left) and Greenland ice
sheets (right), 1992−2003, from ERS radar
altimetry (from Shepherd & Wingham,
2007). The insets show the bedrock
geometry, including floating (light grey),
marine-based (mid-grey) and continental-
based (black) sectors of the ice sheets.
B: Changes in elevation (flow in excess of
50 m yr–1) of the trunks of glaciers such as
Pine Island (basin GH in panel A), Thwaites
(basin GH), Totten (basin C′D), and Cook
(basin DD′). (Shepherd & Wingham, 2008)

241
SP-1326

Figure 19. Rate of change in surface

elevation of the Larsen ice shelf determined
from ERS radar altimeter measurements,
1992−2001 (colour scale and 0.1 m yr−1
white contours). The elevation data are
superimposed on a mosaic of Advanced
Very-High-Resolution radiometer (AVHRR)
satellite imagery (grey scale). Data points
used in the interpolation are shown as
black dots. Also shown are the locations of
meteorological stations (blue dots), stakes
used to measure changes in snow height
(red dots) and surface mass balance (5-km
grid, centred on the green dot).
The 1990 boundaries of the Larsen-A, -B
and -C ice-shelf sections are highlighted in
blue, green and red, respectively. Larsen-A
collapsed before the ERS measurements,
Larsen-B has since disintegrated, and only
Larsen-C remains intact. (Shepherd et al.,
2003)

9. Global Digital Elevation Models

The Earth and Planetary Remote Sensing Laboratory at De Montfort University,
UK, generated a global Digital Elevation Model (DEM) at 30 arcsec resolution by
combining the best available ground truth data with a unique global database
of ERS‑1 altimeter-derived heights over land, reprocessed with a novel expert
system. This model, called Altimetry Corrected Elevations (ACE), presents
radically improved spatial accuracy and resolution over large regions of the
globe, in particular over Africa and South America, but has suffered so far from
a serious lack of in situ data.
The ACE-2 dataset was created by merging the Shuttle Radar Topography
Mission (SRTM) dataset with satellite radar altimetry within the region
bounded by the 60° parallels (Fig. 20). For areas lying outside the SRTM’s
latitude limits other data sources were used, such as the Global Land One-km
Base Elevation (GLOBE), the original ACE DEM, together with new matrices
derived from reprocessing the ERS‑1 GM dataset with an enhanced retracking
system, and data from other satellites. The main altimetry dataset used in
the generation of the ACE-2 dataset was the ERS‑1 Geodetic Mission, which,
because of its small across-track spacing, presents a uniquely dense spatial
distribution of tracks over land surfaces. Data from the ERS‑2 and Envisat
Ku-band were also included where appropriate. All of the altimetry data were

242
 Altimeter

Figure 20. The ACE-2 model. Over the oceans the elevation model has been supplemented with altimeter-derived bathymetry. (Smith et al., 2007)

reprocessed using the Berry expert system to retrack the waveforms (Smith et
al., 2007).
Over 11 billion pixels from the SRTM dataset have been adjusted using this
unique network of control arcs of altimeter-derived height data. In addition, an
extensive investigation over the rainforests has shown that the altimetry data
are returning ground values whereas the SRTM signal bounces off somewhere
within the canopy. Therefore, areas of rainforest in both the Amazon and the
Congo have been completely replaced by altimetry-defined surfaces.

10. River and Lake Levels for Hydrology

The launch of the ERS satellites began a new era of applications of altimetry for
monitoring river and lake levels. An obvious advantage of using data derived
from satellites is the global coverage and regular temporal sampling of the
processed data, although there are difficulties in interpreting radar altimeter
measurements made over inland water bodies (Frappart et al., 2006).
The primary reason for studying altimetry over lakes was to validate
altimeter measurements (i.e. Cretaux et al., 2009) as lakes have minimal tides
and little dynamic variability compared with the oceans, so that the spatial
changes in lake levels closely follow the geoid.
The great potential of altimetry for monitoring inland water levels rapidly
became apparent (Birkett, 1994; Calmant & Seyler, 2006; Calmant & Birkett,
2006). This has been applied several times over the North American Great
Lakes (Frappart et al., 2006, Morris et al., 1994), the Caspian Sea (Cazenave et
al., 1997); Lake Baikal Kouraev et al., 2007), the East African lakes (Birkett et
al., 1999) and the largest rivers, including the Amazon (Alsdorf, 2001; Medina
et al., 2008; Santos da Silva et al., 2011). The potential for generating river and
lake heights on a global scale by reprocessing individual altimeter echoes was
demonstrated by (Berry et al., 2005, 2007). This subsequently became ESA’s
River and Lake project (http://earth.esa.int/riverandlake/).
The accuracy of altimeter measurements has greatly improved over the past
decade as a result of advances in the instrumentation and in the precision of

243
SP-1326

Figure 21. Detected river and lake targets

using ERS and Envisat altimetry globally
(left) and over the Amazon delta (right).
The colour coding indicates the quality
of the time series that can be extracted
from ERS altimetry; green indicates nearly
uninterrupted time series.
(http://earth.esa.int/riverandlakes)

satellite orbit calculations. Satellite altimetry coverage over land surfaces has
also been greatly improved by ESA’s inclusion of additional tracking modes on
the ERS and Envisat altimeters, which enable the instruments to track rapidly
changing surfaces. This has led to substantial advances in altimeter research
over ice, land and inland water bodies (Frappart et al., 2006).
Several teams around the world are now involved in satellite altimetry
over inland water bodies, including those led by C. Birkett (NASA, USA),
A. Cazenave (Centre National d’Etudes Spatiales, CNES, France), P. Berry
(De Montfort University, DMU, UK) and J. Santos da Silva (Universidade do
Estado do Amazonas, Manaus, Brazil). At DMU, the team has studied the land
surface response to the altimeter pulse, and this has allowed characterisation
of the relationship between the underlying terrain and the individual echo
shapes (Figs 21 and 22).
Because of the ERS and Envisat missions it is now possible to obtain decadal
time series over a number of targets globally. These targets are often large
enough to present Brown model waveforms and therefore ocean retracking can
be performed. However, the vast majority of inland water bodies do not return
these types of signal.
The use of retracking means that it is possible to monitor a significantly
larger number of targets. Even so there are a number of factors that limit the
capability of an altimeter to obtain the correct heights. The current generation
of databases, such as ESA’s River and Lake database and the LEGOS Hydroweb
(Cretaux et al., 2011c), have hundreds of targets where the results are deemed to
be of high enough quality. The current system can obtain results globally from
both rivers and lakes throughout the year.
Satellite altimeter data provide water resource planners with an additional
source of information about remote areas such as Africa, for which the
availability of in situ data is limited. One example of a lake time series is that
for Lake Kainji in western Nigeria (Fig. 23). This important lake is fed by the
Niger River and its water levels have knock-on effects along the length of the
river, which is the main source of water in the area.
An altimeter can perform just as well over rivers as over lakes, although
more complicated processing is required. A major benefit is that it is possible
to obtain time series from along the entire length of a river system to provide
as complete a picture of water flow as possible. As an example, Fig. 25 shows
the ERS–Envisat time series of the Brahmaputra River that flows through India
and Bangladesh.

244
 Altimeter

Figure 22. Bottom: Time series of water

levels from ERS‑2 and Envisat in the
Amazon Basin, 1995−2009.
(Santos da Silva et al., 2010)
Top: Differences between the altimetry
and the gauge series. Histograms of the
residuals are shown in the inset
(bottom right) with rms differences (in mm).

Figure 23. Lake Kainji in western Nigeria (left) and joint ERS-2−Envisat time series of lake levels, 1995−2008 (right). (Berry et al., 2005)

245
SP-1326

Figure 24. Lake Balbina, Brazil, monitored

by satellite altimetry. This is an important,
reservoir with a dam that provides energy
for the 2 million inhabitants of Manaus,
Brazil (Santos da Silva et al., 2010). Note
that gauge data from the reservoir itself
or from the Uatuma River that feeds the
lake are not publicly available; the gauge
series is displayed by courtesy of Amazonas
Energia.

Figure 25. The Brahmaputra River

in Bangladesh. Location of the ERS−Envisat
time series (left) and observed river
level time series (right)
(http://earth.esa.int/riverandlake).

11. Global and Large-Scale Ocean Signals

One of the objectives of the ERS altimetric mission was to map the global ocean
circulation and its variations and associated transfers of energy, together
with ocean−atmosphere interactions and ocean tides. ERS also proved to
be a great asset for studying climate change and for mapping the impacts of
unpredictable events such as the 2004 tsunami in the Indian Ocean once the
quality of the altimetric observations was enhanced, as described in section 2.

11.1 Ocean Tides

The ERS and Envisat satellites operated in Sun-synchronous orbits, and so
were far from ideal for accessing ocean tides because they always observed the
same phase of the solar tidal wave. However, ocean tides can be derived from
the ERS−Envisat radar altimeter data using the correlation between the tidal
constituents and/or by combining the data with hydrodynamic modelling.
In polar seas, very few data are available for tidal analysis as tide gauges
are sparse and the coverage of the Topex/Jason-1 satellites is limited to ±66°
latitude. Satellite altimetry from ERS and Envisat was one way to observe
the ocean tides. From early in the ERS mission satellite altimetry was used to
derive regional and global ocean tide models (Figs 26; Andersen, 1994, 1995),
as well as to determine ocean tides around and beneath large ice shelves and
comparing them with observations of their vertical movements (Figs 27 and 28).

246
 Altimeter

Figure 26. Ocean tides in the North Atlantic

and adjacent seas derived from ERS‑1
radar altimeter data using the correlation
between the tidal constituents for the Sun-
synchronous S2 constituent. The amplitudes
are contoured and the phases are colour
coded. (Andersen, 1994, 1995)

Figure 27. The major tidal signal in the Southern Ocean. The largest semidiurnal constituent M2 (left) and the largest diurnal constituent K1
tides (right) determined using multiple satellite altimetry data. (Yi et al., 2004, 2006)

247
SP-1326

Figure 28. Amplitudes and phases of the

four most energetic tidal constituents
(K1, O1, S2, and M2) for the ice shelves on
the west coast of the Antarctic Peninsula,
determined from ERS radar altimetry and
tide gauge recordings (squares). Ice shelf
boundaries were delimited from digital
coastline data (British Antarctic Survey,
Scott Polar Research Institute and the
World Conservation Monitoring Centre).
Tidal amplitude and phase (colour) are
superimposed on an AVHRR mosaic (grey)
of grounded ice. Open water is white.
Circles: locations (and footprint sizes) of
ERS satellite ground track crossing points
from which the tidal model was derived;
squares: tide gauges. (Shepherd & Peacock,
2003)

11.2 Tsunamis
Tsunamis are waves triggered by earthquakes and landslides or, rarely,
unusually large seafloor volcanic eruptions. A large tsunami can generate huge
waves that endanger people and damage property in low-lying coastal areas.
The Indian Ocean tsunami, on 26 December 2004, killed more than 200  000
people and left millions homeless.
The radar altimeters on the Jason-1, Topex, Envisat and Geosat Follow-On
satellites obtained profiles of sea surface heights along transects across the
Indian Ocean between two and nine hours after the earthquake (e.g. Song et
al., 2005; Scharroo, 2007; see Fig. 29). Such data are usually received hours to
days after the event, and so are too late to be used in the detection of tsunamis
or to issue warnings.

248
 Altimeter

Figure 29. A–C: Tsunami wave heights

generated by the Indian Ocean earthquake
on 26 December 2004, computed by the
MOST (Method of Splitting Tsunami) model
(A) 2:00 h, (B) 3:15 h, and (C) 8:50 h after
the earthquake. These times coincide with
overflights of the Topex/Jason-1, Envisat
and GFO satellites, respectively. D–G:
Comparison of sea height deviations as
measured by the satellite altimeters and
the modelled wave heights. (Scharroo et al.,
2005)

11.3 Sea Level Change

During the last two decades satellite altimetry has demonstrated its ability
to measure temporal changes in global mean sea level with a precision of
the order of 0.4 mm  yr–1 or better. Such changes in mean sea level are most
frequently reported using altimetry from the Topex and Jason-1 satellites, and
those obtained using ERS and Envisat satellites are nearly identical (Fig. 30).
Changes in sea level outside the 66° parallels can be estimated using the joint
ERS and Envisat time series (Fig. 31) providing estimates even for polar regions
(Fig. 32).

Figure 30. Global sea level trend from

all available satellites for the period
1991−2012. Note that GIA or IB
corrections have not been applied.
(Scharroo et al., 2012)

249
SP-1326

Figure 31. Changes in sea level from ERS

and Envisat altimetry over the period
1992–2008. Yellow and red indicate a rise
in sea level, while blue and green indicate a
fall. (Andersen et al., 2002, 2009)

Figure 32. Trends in sea level in the Arctic

and Antarctic oceans on a 1°×1° grid
based on the Aviso MSLA delayed-time
data product (0.25°×0.25°). The mean of
the linear trend within the red boundary
(55°N–82°N/45°W–60°E) is 2.9 mm yr–1.
(Updated from Kuo, 2006; Shum & Kuo,
2011)

Figure 33. Sea level anomalies in the

equatorial Pacific Ocean on 30 November
1997 (top), and 30 August 1998 (bottom),
based on ERS−Envisat radar altimeter data
over a 16-day period centred on those days.
(DEOS; http://rads.tudelft.nl/results)

250
 Altimeter

11.4 El Niño
The radar altimeters on the ERS satellites measured sea surface heights between
July 1991 and 2011. One area of interest is the equatorial Pacific Ocean where the
famous El Niño roars every few years (Fig. 33). El Niño events are characterised by
a relatively high sea level along the west coast of Central America accompanied
by radical changes in the regional climate with heavy rainfall. At the same time,
the sea level drops in the western equatorial Pacific, where extreme droughts
devastate crops. The opposite extreme is called La Niña.

References
Alsdorf, D., Birkett, C.M., Dunne, T., Melack, J. & Hess, L. (2001). Water level
changes in a large Amazon lake measured with spaceborne radar interferometry
and altimetry. Geophys. Res. Lett. 28(14), 2671−2674.
Andersen, O.B. (1994). Ocean tides in the northern North Atlantic Ocean and
adjacent seas from ERS-1 altimetry, J. Geophys. Res. 99(C11), 22557−22573.
Andersen, O.B. (1995). Global ocean tides from ERS-1 and TO­PEX/POSEIDON
altimetry, J. Geophys Res. 100(C12), 25 249−25 259.
Andersen, O.B. & Knudsen, P. (1998). Global marine gravity field from the ERS‑1 and
Geosat Geodetic Mission altimetry. J. Geophys. Res. 103, 8129–8137.
Andersen, O.B. & Knudsen, P. (2009). The DNSC08 mean sea surface and mean
dynamic topography. J. Geophys. Res. 114, C11. doi: 10.1029/2008JC005179
Andersen, O.B., Knudsen, P. & Beckley, B. (2002). monitoring long-term changes in
sea level using ERS satellites. Phys. Chem. Earth 27, 1413–1418.
Andersen, O.B., Knudsen, P. & Berry, P. (2010). The DNSC08GRA global marine
gravity field from double retracked satellite altimetry. J. Geodesy 84(3). doi:
10.1007/s00190-009-0355-9
Andersen, O.B. & Scharroo, R. (2011). Range and geophysical corrections in coastal
regions, in S. Vignudelli, A. Kostianoy, P. Cipollini & J. Benveniste, J. (Eds),
Coastal Altimetry. Springer, Berlin.
Anzenhofer, M., Shum, C.K. & Rentsh, M. (1999). Costal altimetry and applications,
Tech. Rep. 464, Geodetic Science and Surveying. Ohio State University, Columbus,
USA.
Bamber, J.L., Ekholm, S. & Krabill, W.S. (2001). A new, high-resolution digital
elevation model of Greenland fully validated with airborne laser altimeter data.
J. Geophys. Res. 106(B4), 6733–6745.
Bamber, J.L., Gomez Dans, J.L. & Griggs, J.A. (2009). A new 1 km digital elevation
model of the Antarctic derived from combined satellite radar and laser data.
Part I: Data and methods. Cryosphere 3(2), 101–111.
Baudry, N. & Calmant, S. (1991). 3-D modelling of seamount topography from
satellite altimetry, Geophys. Res. Lett. 18, 1143–1146.
Baudry, N. & Calmant, S. (1996). Seafloor mapping from high-density satellite
altimetry. Marine Geophys. Res. 18, 135–146.
Becker, M., Llowel, W., Cazenave, A., Güntner, A., Crétaux, J.-F. (2010). Recent
hydrological behaviour of the East African Great Lakes region inferred from
GRACE, satellite altimetry and rainfall observations. C. R. Geosci. doi: 10.1016/j.
crte.2009.12.010
Berry, P.A.M., Garlick, J.D., Freeman, J.A. & Mathers, E.L. (2005). Global inland
water monitoring from multi-mission altimetry. Geophys. Res. Lett. 32 (16),
L16401.
Berry, P.A.M., Freeman, J.A., Rogers, C. et al. (2007). Global analysis of Envisat RA-2
burst mode echo sequences. IEEE Trans. Geosci. Rem. Sens. 45(9), 2869−2874. 
Berwin, R. (2001). ERS‑1 (Phases A−G) Altimeter Value-added GDR Data Product,
Version 1. JPL DVD-ROM, generated by Ohio State University (C.K. Shum and
Yuchan Yi), JPL (R. Berwin, R. Benada and V. Zlotnicki), and the University of
Texas Center for Space Research.

251
SP-1326

Birkett, C.M. (1994). Radar altimetry − A new concept in monitoring global lake level
changes. Eos Trans. AGU 75(24), 273–275.
Birkett, C., Murtugudde, R. & Allan, T. (1999). Indian Ocean climate event brings
floods to East Africa's lakes and the Sudd Marsh. Geophys. Res. Lett. 26(8),
1031–1034. 
Calmant, S. & Baudry, N. (1996). Modelling bathymetry by inverting satellite
altimetry data: A review. Marine Geophys. Res. 18, 123–134.
Calmant, S., Berge-Nguyen, M. & Cazenave, A. (2002). Global seafloor topography
by least-squares inversion of satellite altimetry and shipboard bathymetry.
Geophys. J. Int. 151, 795–808.
Calmant, S. & Seyler, F. (2006). Continental surface waters from satellite altimetry.
Géosciences 338, 1113–1122.
Cazenave, A., Bonnefond, P., Dominh, K. & Schaeffer, P. (1997). Caspian sea level
from Topex-Poseidon altimetry: Level now falling. Geophys. Res. Lett. 24(8),
881–884.
Cretaux, J.-F., Jelinski, V., Calmant, S., Kouraei, A., Vuglinski, V., Bergé-Nguyen,
M., Gennero, M.-C., Nino, F., Abarco Del Rio, R., Cazenave, A. & Maisongrande,
P. (2011). SOLS, a lake database to monitor in near real time water level
storage variations from remote sensing data. Adv. Space Res. doi: 10.1016/j.
asr.2011.01.004
Cretaux, J.-F. & Birkett, C. (2006). Lake studies from satellite altimetry. C. R. Geosci.
338, 1098–1112. doi:10.1016/J.crte.2006.08.002
Deng, X., Featherstone, W.E., Hwang, C. & Berry, P.A.M. (2002). Estimation of
contamination of ERS‑2 and POSEIDON satellite radar altimetry close to the
coasts of Australia. Marine Geodesy 25, 249–271.
Flament, T. & Rémy, F. (2012). Dynamic thinning of Antarctic glaciers from along-
track repeats radar altimetry. J. Glaciol. 58(211), 830–840.
Frappart, F., Calmant, S., Cauhopé, M., Seyler F. & Cazenave, A. (2006). Results
of ENVISAT RA-2 derived levels validation over the Amazon basin. Rem. Sens.
Environ. 100, 252–264.
Fu, L.L. & Cazenave, A. (2001). Satellite Altimetry and Earth Sciences: A Handbook of
Techniques and Applications. International Geophysics Ser. 69, Academic Press,
San Diego, CA.
Hernandez, F. & Schaeffer, P. (2000). Altimetric Mean Sea Surfaces and Gravity
Anomaly Maps Intercomparisons. AVI-NT-011-5242-CLS, CLS Ramonville St Agne,
France.
Hwang, C., Hsu, H. & Jang R. (2000). Global mean sea surface and marine gravity
anomaly from multi-satellite altimetry: Applications of deflection-geoid and
inverse Vening−Meinesz formulae. J. Geodesy 76(8), 407−418.
Kingdon, R., Hwang, C., Hsiao, Y.S. et al., (2008). Gravity anomalies from retracked
ERS and Geosat altimetry over the Great Lakes: Accuracy assessment and
problems. Terr. Atmos. Ocean. Sci. 19(1−2), 93−101. 
Knudsen, P., Andersen, O.B. & Tscherning, C.C. (1992). Altimetric gravity anomalies
in the Norwegian-Greenland Sea: Preliminary results from the ERS‑1 35-day
repeat mission, Geophys. Res. Lett. 19(17), 1795–1798.
Kouraev, A.V., Semovski, S.V., Shimaraev, M.N. et al. (2007). Observations of Lake
Baikal ice from satellite altimetry and radiometry. Rem. Sens. Environ. 108(3),
240−253. 
Kuo, C. (2006). Determination and Characterization of 20th Century Global Sea Level
Rise. Geodetic Science Report No. 478, Ohio State University.
Laxon, S. & McAdoo, D.C. (1994). Arctic ocean gravity field derived from ERS‑1
satellite altimetry. Science 265, 621–624.
Laxon, S. & McAdoo, D.C. (1998). Satellites provide new insights into polar
geophysics. EOS Trans. AGU 79(6), 69–73.
Llubes, M., Lanseau, C. & Rémy, F. (2006). Relations between basal condition,
subglacial hydrological networks and geothermal flux in Antarctica. Earth
Planet. Sci. Lett. 241, 655–662.

252
 Altimeter

Manzella, G.M.R., Borzelli, G.L., Cipollini, P., Guymer, T.H., Snaith, H.M. &
Vignudelli, S. (1997). Potential use of satellite data to infer the circulation
dynamics in a marginal area of the Mediterranean Sea. Space at the Service of our
Environment, Proc. 3rd ERS Symp., Vol. 3. ESA SP-414, European Space Agency,
Noordwijk, the Netherlands pp.1461–1466.
Maus, S., Green, C.M. & Fairhead, D. (1998). Improved ocean-geoid resolution from
retracked ERS‑1 satellite altimeter waveforms. Geophys J. Int. 134(1), 243–253.
McAdoo, D.C. & Laxon, S. (1997). Antarctic tectonics: Constraints from a new ERS‑1
satellite marine gravity field. Science 276(5312), 556–561.
Medina, C.E., Gomez-Enri, J., Alonso, J.J. et al. (2008). Water level fluctuations
derived from ENVISAT Radar Altimeter (RA-2) and in-situ measurements in a
subtropical water body: Lake Izabal (Guatemala). Rem. Sens. Environ. 112(9),
3604−3617
Morris, C.S. and Gill, S.K. (1994a). Variation of Great Lakes water levels derived from
Geosat altimetry. Water Resources Res. 30(4), 1009–1017.
Obligis, E., Desportes, C., Eymard, L., Fernandes, J., Lázaro, C. & Nunes, A. (2011).
Tropospheric corrections for coastal altimetry, in S. Vignudelli, A. Kostianoy. P.
Cipollini & J. Benveniste, J. (Eds), Coastal Altimetry. Springer, Berlin.
Ogle, J. (2003). Orbital Error Analysis for ERS‑1 and ERS‑2. Geodetic Science Report,
Ohio State University, USA.
Rémy, F. & Minster, J.F. (1997). Antarctica ice sheet curvature and its relation with
ice flow and boundary conditions. Geophys. Res. Lett. 24, 1039–1042.
Rémy, F. & Parouty, S. (2009). Antarctic ice sheet and radar altimetry: A review.
Rem. Sens. 1, 1212–1239. doi: 10.3390/rs1041212
Rémy, F.D., Shaeffer, P. & Legresy, B. (1999). Ice flow physical processes derived
from the ERS‑1 high-resolution map of the Antarctica and Greenland ice sheets.
Geophys. J. Int. 139, 645–656.
Sandwell, D.T. & Smith, W.H.F. (1997). Marine Gravity anomaly from Geosat and
ERS‑1 satellite altimetry. J. Geophys. Res. 102, 10039–10054.
Sandwell, D.T. & Smith, W.H.F. (2005). Retracking ERS‑1 altimeter waveforms for
optimal gravity field recovery, Geophys. J. Int. 163, 79–89. doi: 10.1111/j.1365-
246X.2005.02724
Sandwell, D.T. & Smith, W.H.F. (2009). Global marine gravity from retracked Geosat
and ERS‑1 altimetry: Ridge segmentation versus spreading rate. J. Geophys. Res.
114, B01411. doi: 10.1029/2008JB006008
Santos da Silva, J., Calmant, S., Seyler, F., Rotunno Filho, O. C., Cochonneau, G. &
Mansur, W.J. (2010) Water levels in the Amazon basin derived from the ERS -2
and ENVISAT radar altimetry missions, Rem. Sens. Environ.114, 2160–2181.
Schaeffer, P., Faugère, Y., Legeais, J.F., Ollivier, A., Guinle, T. & Picot, N. (2012).
CNES_CLS11 global mean sea surface computed from 16 years of satellite
altimeter data. Marine Geodesy. Special Issue on Jason-2, 35, 3–19.
Schaeffer, P., Hernandez, F., Le Traon, P.-Y., Mertz, F. & Bahurel, P. (1998). A mean
sea surface dedicated to ocean studies: Global estimation, Poster Session,
European Geophysical Society, Nice, France.
Scharroo, R. (2007). Tsunamis, in M.D. King et al. (Eds), Our Changing Planet: The
View From Space. Cambridge University Press, Cambridge, UK. pp.196–201,
Scharroo, R., Leuliette, E., Lillibridge, J., Neije, M., Schrama, E. & Andersen, O.
(2012). Experiences with the use of RADS for Cal/Val and data distribution.
Presented at the S3 CalVal meeting, Frascati, Italy.
Scharroo, R., Smith, W.H.F. & Lillibridge, J.L. (2005). Satellite altimetry and the
intensification of Hurricane Katrina, Eos Trans. AGU, 86(40), 366.
Scharroo, R. Visser, P.N.A.M. & Mets, G.J. (1998). Precise orbit determination
and gravity field improvement for the ERS satellites. J. Geophys. Res., 103(C4),
8113−8127.
Shepherd, A. & Peacock, N.R. (2003). Ice shelf tidal motion derived from ERS
altimetry, J. Geophys. Res. 108, 3198. doi: 10.1029/2001JC001152
Shepherd, A. & Wingham, D. (2007). Recent sea-level contributions of the Antarctic
and Greenland ice sheets. Science 315, 1529–1532.

253
SP-1326

Shepherd, A. & Wingham, D. (2008). Antarctic glacier thinning, 1992–2003. Scott.

Geogr. J. 124, 154–164.
Shepherd, A., Wingham, D., Payne, T. & Skvarca, P. (2003). Larsen ice shelf has
progressively thinned. Science 302, 856–859.
Shepherd, A., Wingham, D.J., Mansley, J.A.D. & Corr, H.F.J. (2001). Inland thinning
of Pine Island Glacier, West Antarctica. Science 291, 862–864.
Shepherd, A., Wingham, D., Wallis, D., Giles, K., Laxon, S. & Sundal, A.V. (2010).
Recent loss of floating ice and the consequent sea level contribution. Geophys.
Res. Lett. 37, L13503.
Shum, C. & Kuo, C. (2011). Observation and geophysical causes of present-day sea
level rise, in R. Lal, M. Sivakumar, S. Faiz, A. Rahman & K. Islam (Eds.), Climate
Change and Food Security in South Asia, Part 2, ch. 7, pp.85–104. Springer, Berlin.
doi: 10.1007/978-90-481-9516-9_7
Shum, C., Tapley, B., Kozel, B., Visser, P., Ries, J. & Seago, J. (1993). Precise orbit
analysis and global verification results from ERS‑1 altimetry. Space at the Service
of Our Environment, Proc. 2nd ERS‑1 Symp., vol. 2, pp.747–752. ESA SP-361, ESA,
Paris.
Smith, R.G, Berry, P.A.M. & Benveniste, J. (2007). Representation of rivers and lakes
within the forthcoming ACE-2 Global Digital Elevation Model. Proc. Hydrospace
2007, Geneva, Switzerland. www.congrex.nl/07M19/
Song, T., Chen, J., Fu, L.-L., Zlotnicki, V., Shum, C., Yi, Y. & Hjorleifsdottir, V. (2005).
The 26 December 2004 tsunami source estimated from satellite radar altimetry
and seismic waves. Geophys. Res. Lett. 32, L20601. doi: 10.1029/2005GL023683
Sundal, A.V., Shepherd, A., Nienow, P., Hanna, E., Palmer, S. & Huybrechts, P.
(2011) Melt-induced speed-up of Greenland ice sheet offset by efficient subglacial
drainage. Nature 469, 522−U83.
Urban, T., Pekker, T., Tapley, B., Kruizinga, G. & Shum, C. (2001). A multiyear
comparison of wet troposphere corrections from TOPEX/Poseidon, ERS‑1 and
ERS‑2 microwave radiometers and the European Centre for Medium-range
Weather Forecasts model. J. Geophys. Res. 106(C9), 19 657–19 669.
Vaughan, D.G. & Bamber, J.L. (1998). Identifying areas of low-profile ice sheet and
outcrop damming in the Antarctic ice sheet by ERS‑1 satellite altimetry. Ann.
Glaciol. 27, 1–6.
Vignudelli, S., Kostianoy, A.G., Cipollini, P. & Benveniste, J. (Eds) (2011). Coastal
Altimetry, Springer, Berlin. doi: 10.1007/978-3-642-12796-0
Wang, Y.M. (2001). GSFC00 mean sea surface, gravity anomaly, and vertical gravity
gradient from satellite altimeter data. J. Geophys. Res. 106(C12), 31 167−31 174.
Wingham, D., Ridout, A.J., Scharroo, R., Arthern, R.J. & Shum, C.K. (1998). Antarctic
elevation change from 1992 to 1996. Science 282(5388), 456–458. doi: 10.1126/
science.282.5388.456
Wingham, D.J., Siegert, M.J., Shepherd, A. & Muir, A.S. (2006). Rapid discharge
connects Antarctic subglacial lakes. Nature 440, 1033–1036. doi: 10.1038/
nature04660
Yi, Y. (1995). Determination of Gridded Mean Sea Surface from TOPEX, ERS‑1 and
Geosat Altimeter Data. Ohio State University Geodetic Science Report 434, Ohio
State University.
Yi, Y., Matsumoto, K., Shum, C., Wang, Y. & Mautz, R. (2006). Advances in southern
ocean tide modelling. J. Geodynamics 41(1–3), 128–132.
Yi, Y., Shum, C., Andersen, O. & Knudsen, P. (2004). Extreme southern ocean tide
modelling, in C. Hwang, C. Shum & J. Li (Eds), Satellite Altimetry for Geodesy,
Geophysics and Oceanography. Int. Assoc. Geodesy Symp. Series 126. Springer,
Berlin.
Zwally, H.J., Giovinetto, M.B., Li, J., Cornejo, H.G., Beckley, M.A., Brenner, A.C.,
Saba, J.L. & Yi, D.H. (2005). Mass changes of the Greenland and Antarctic ice
sheets and shelves and contributions to sea-level rise: 1992–2002. J. Glaciol. 51,
509–527.

254
→→ 20 Years of ERS →
Interferometry
 SAR Interferometer

20 Years of ERS Interferometry

F. Rocca

Dipartimento di Elettronica e Informazione, Politecnico di Milano, 20133 Milano, Italy.

1. Introduction: Interferometry for Beginners

For the inexperienced reader, an interferogram is the pixel-by-pixel product of
two complex conjugate SAR images of the same scene (Curlander & McDonagh,
1991; Franceschetti & Lanari, 1999; Ferretti et al., 2007). The modulus is
dependent on the power reflected from that pixel and the phase is the difference
of the phase delays due to the two way travel path and to the scattering process
in the two instances. Hence, if the scatterers stay the same, only the travel path
changes are recorded, be they due to the motion of the target or to the parallax
of the takes (called baseline), or both.
In fact, if the scatterers did not change at all their reflectivity from one pass
to the next, the interferogram phase would show the changes of the travel path
of the scatterer to the satellite and back, in terms of number of wavelengths.
The interferometric fringes then would tell to a small fraction of the wavelength
(5.77  cm, in the C-band) the travel path changes, and they would depend
mostly on the terrain topography, for any given baseline (parallax). Thus, we
can obtain Digital Elevation Models (DEMs) from interferograms. However,
the two images should be taken at the same time to avoid the effects of the
ever-changing atmospheric delays due to water vapour that could spoil the
otherwise satisfactory DEMs.
If we wish to get rid of the atmospheric disturbances, interferometry has to
be single pass, but then two well-separated receivers are needed to create the
parallax (baseline). Owing to the simultaneity of the passes, there is no change
in reflectivity (the coherence of the two images is high), and the atmospheric
delay is the same for the two images, and thus it is totally subtracted. The two
receivers should be distant enough to achieve a good parallax (several tens of
metres, in the C-band) and this is costly, in space. On the other hand, if the
images are taken in successive passes (multipass), only one simple satellite is
needed, but then the number of passes should be sufficient to be able to reduce
the atmospheric disturbances by averaging, and this only for those scatterers
that did not change their reflectivity in the meantime.
Another problem that needs to be solved is Phase Unwrapping (PU), i.e. to
enumerate properly the interferometric fringes. With PU we find the total travel
path change by counting the number of interferometric fringes from any pixel
to any other or to some reference point. This problem is solvable in theory
and practice, but only if we are given many more than one interferogram.
Otherwise, between any two pixels, we need to find a path encountering only
well separated and noiseless fringes, so that we do not miss any. This entails
smooth topography and good coherence, i.e. small changes in the reflectivity
from one pass to the next.
In conclusion, we need either complex single-pass experiments such as
the NASA/DLR/ASI Shuttle Radar Topography Mission (SRTM), or the DLR
TerraSAR‑X add-on for Digital Elevation Measurement (TanDEM-X; see below),
using two SAR receivers tens or hundreds of metres apart, or we may use
unmanned satellites that periodically revisit the same area many times at short
revisit intervals.
Again, interferometry has to be either single pass, to remove the
atmospheric bias as it is just the same for the two images and to achieve utmost
stability of the reflectivity of the targets, or multipass. In this case, the number
of passes should be sufficient to reduce the ever-changing atmospheric bias

257
SP-1326

by exploiting statistics. This could also be helpful to solve Phase Unwrapping.

Therefore, unmanned satellites are the key for interferometric advances as they
can produce a very high number of interferograms at low cost.
The first studies of SAR interferometry were carried out at NASA’s Jet
Propulsion Laboratory (Graham, 1974; Zebker & Goldstein, 1986, Goldstein
et al., 1988) and led to seminal papers by Jack Curlander, Richard Goldstein,
Scott Hensley, Paul Rosen, Jacob Van Zyl, Howard Zebker, and many others.
Attention was centred on short-term interferometry as the observations had
been carried out using aircraft or manned space platforms. It is a useful
simplification to state that the studies carried out in the US dealt mostly with
single-pass interferometry. Europeans used mostly unmanned satellites, in the
tradition of ERS‑1, and therefore used multipass interferometry and excelled in
long-term differential interferometry studies. However, European airplane data
were plentiful and well studied, and US activities using European, Canadian
and Japanese unmanned SAR satellites were also numerous and important.
The first civilian SAR satellite was the L-band Seasat, launched in 1978.
In its short life, just 78 days, it produced numerous splendid images and
was shown by JPL to have interferometric capabilities. Seasat’s important
interferometric results were well exploited by the JPL team, even if they were
obtained with just three serendipitous images. It was followed by the important,
interferometric, shuttle-based Spaceborne Imaging Radar (SIR) missions: SIR‑A
(1981, analogue), SIR‑B (1984, digital), and SIR‑C (1994, multiband L, C, X, in
cooperation with DLR and ASI). In these cases, the orbiting radar deliberately
imaged the same areas on two consecutive days, showing the predicted
interferograms and allowing the development of satellite SAR interferometry,
i.e. InSAR.
The many important results obtained with Seasat (not interferometry,
though) had meanwhile persuaded ESA to consider a satellite for Earth
resources, to be named ERS‑1, hosting many instruments including a C-band
radar. The choice of the C‑band was mostly due to the availability of the
technology and its suitability for observations of sea waves and ice. In fact,
interferometry was not considered at all in the planning.

2. SAR in Milan
2.1 Phase Preserving Focusing
Meanwhile, in Milan, we had developed interferometry algorithms to a
satisfying extent. Our studies of SAR had started in 1983 when Fulvio Marcoz
and Alfonso Farina of Alenia proposed that our department participate in
a European research project on SAR, named Adaptive Real-time Strategies
for Image Processing (ARTS‑IP). Alenia had a good experience with radar; in
Milan, together with Claudio Prati who had just graduated, we had experience
in signal processing and our colleague Maria Giovanna Sami, who managed
the team, was an expert in digital design. A team at Imperial College London
also participated in the research. It is interesting to note that the SAR system
we wished to design at the time does not exist even now, even if it starts being
considered. The idea was to design a naval search machine: the onboard
antenna would squint forward to image ships at low resolution. Then, an
onboard focusing system would focus data in realtime; adding fantasy to
science, we proposed to have a realtime pattern recognition system (the task
of the Imperial College team) to detect ships. After, the very same antenna
was supposed to be squinted aft to zoom in on the detected ships and finally
downlink the result.
We learned that SAR was a complex system, and that paper satellites
were not adequate. However, we also learned that SAR technicians, owing to
their realtime habits, had their own approximate way to focus SAR data. We

258
 SAR Interferometer

proposed to focus the data as geophysicists did, i.e. using the exact transfer
function that could even be written in a closed formula (Cafforio et al., 1991),
the ω −k approach. Among other things, I had developed this algorithm for the
Italian oil company AGIP back in 1977. The technique is efficient, but as the
appearance of a SAR image would not change much, at the time the innovation
was not considered essential. However, as it is precise and phase preserving,
the technique proved to be an important tool for advancing interferometry.

2.2 Interferometry in Milan

In autumn 1987 I was on sabbatical at Stanford, and together with Claudio Prati
visited JPL, at the invitation of Jack Curlander, to inform them about the ω −k
focusing technique, which was new to them. In exchange, Richard Goldstein
treated us with a talk on interferometry, and he showed us the interferograms
that he had obtained with Seasat. His attention was mostly devoted to Phase
Unwrapping. He had noticed that any 2π phase terrace added to the topography
would be invisible, as it would correspond to just one cycle of the interferogram
and thus disappear. However, at the beginning (source) and the end (sink)
of that terrace, traces are left in the phases that look like positive or negative
charges, just like the smile of the Cheshire cat remains even after the cat has
gone.
Once we had traded presents, we gave the JPL team our ω −k code, and
they gave us two tapes, one containing the raw data of the three available
interferometric Seasat passes, and the other with airplane interferometric
data. Claudio got good interferograms in a few days. Using their approximate
focusing methodology, phase preservation was hard to get and their
interferograms were noisier than they should have been. Using instead
the phase-preserving ω −k technique, we got our first, very clean, Seasat
interferograms of Death Valley in California, which we used on our Christmas
cards in 1987.
Not in JPL indeed, but interferometry was rather new to others. At a
seminar at the Canada Centre for Remote Sensing (CCRS) in June 1988, at the
invitation of Keith Raney, I discussed ω −k focusing, phase preservation and
interferometry. I had hoped to trade presents as we had done in JPL, but the
audience, composed of people who were in charge of designing the forthcoming
Radarsat‑1, was not as favourable to interferometry. One participant voiced
the idea of storing only the moduli of the images, as it appeared that the
Radarsat data would be too bulky to store them all. I did my best to show that
phases were even more important, as our interferograms showed. Eventually,
Radarsat‑1 archived complex values. After the launch of ERS‑1, and the take-off
of interferometry, a paper describing an alternative phase-preserving focusing
system, named extended chirp scaling, was published by K. Raney, R. Bamler
and others who had developed it in 1992−94.
Returning home to Milan in 1988, I started to look for funds to develop
the technology. Until then, we had been funded only by the initial European
contract. Then, thanks to Livio Marelli, ESA gave us our first contract to write
a code for focusing ERS‑1 SAR data using the ω −k technique. At the time, we
were trying to convince ESA to store what would later be called the Single-Look
Complex (SLC) images, and Livio also tried to help us to convince the Italian
Space Agency (ASI) of the validity of the ω −k technology. I remember a meeting
on that with Claudio Prati, Livio Marelli, Mark Doherty, L. Guerriero, the pro
tempore president of ASI, C. Albanesi and G. Milillo.
Even Mark’s vivacious support could not gain us any funds. With hindsight,
we were lucky: we will always thank ESA people for their quest for innovation
and precision, and for fostering competition with other capable European
teams. Further, the ω −k technology was very slow to be accepted by others; our
first paper on ω −k focusing was submitted to IEEE Trans. Aerospace Electronic

259
SP-1326

Systems, and duly rejected, and was then presented at the International
Geoscience and Remote Sensing Symposium (IGARSS) meeting in Vancouver in
1989. But later, at the Washington IGARSS meeting in 1990, I remember a very
enjoyable talk on SAR focusing with Richard Bamler and Helmut Runge; they
understood the ins and outs of the question, and together we started discussing
and proposing the phase preservation tests that would thereafter characterise
all ESA processors, making focusing transparent to interferometry users. The
paper by Bamler comparing ω −k and range Doppler focusing systems appeared
in 1992 (Bamler, 1992). Chirp scaling also followed, as noted above. But we
were still short of funds, and desperately needed another contract, and it could
only be from ESA and on something new that could not be anything else than
interferometry. Stefano Bruzzi was approached and was finally convinced to
give us all the information we needed to be able to simulate the raw data from
ERS‑1, yet to be launched, and a contract. So, by the end of the commissioning
phase, in the summer of 1991, we were ready.

2.3 ERS-1 interferometry

We received the first images from ERS‑1 on 25 August 1991, a few weeks after
launch., and less than a week later (I remember recalling Claudio and Andrea
Monti Guarnieri, who had recently joined the group, from their summer
vacations) we produced the very first interferogram of the Gennargentu area
of Sardinia, Italy, which is usable even today (Fig. 1). ERS‑1 interferometry was
born, and it was full of surprises.
The first surprise was the coherence between the two images forming the
interferogram, as we called it, again following the geophysicists. It tells about
the mechanical stability of the imaged targets in the time interval between
the takes. We gave it its geophysical name, as we had no other precedent for
SAR. In the US, the same thing is still at times bravely called the modulus of
the complex correlation coefficient, notwithstanding the word count cost. The
coherence proved to be complementary to image amplitudes: restless water
is always black, especially on windy days, when its radar reflectivity is high
and is therefore bright in the amplitude images. Changing vegetation and

Figure 1. The Gennargentu data (August 1991). Left: amplitude; right: interferogram. The Su Gurruppu gorge is visible at top right.

260
 SAR Interferometer

forests were dark, towns bright, and so on, adding an invaluable new tool for
researchers who wanted to classify radar images. Luckily enough, the revisit
interval was just three days, and the coherence very high.
The second surprise was the precision with which we could estimate the
baselines, and the first, indeed approximate, DEMs. We discovered in the data
(no Google then) and later on maps, the Su Gurruppu gorge and the Grotta del
Bue Marino in Sardinia, and then the Patagonia and the Norway data, the fjords
and the shores, and so many views of the Etna and of the Vesuvius, at all times.
We derived the baselines from the data themselves, as the orbital data were not
available and not precise enough. We were able to tell the timing of the orbit
corrections as we could measure from the interferogram DEMs the changes in
the baselines in successive passes and see their parabolic behaviour, just like
a bouncing ball. Not gravity pull but solar wind push, not ground bounces but
orbit corrections: anyway, intersections of parabolas.

2.4 Fringe Meetings and the Tandem Experiment

Steve Coulson at ESRIN got really interested, and so he immediately convened
a meeting in Frascati in autumn 1991, to present the results that were so easy to
get using the splendid ERS‑1 data. The need for subpixel coregistration and the
cove of information to be found in single-pixel phases were hard to accept to
people used to combining many pixels to evaluate their amplitudes for speckle
reduction (Prati et al., 1990, 1994; Prati & Rocca, 1993, 1994).
With hindsight, we can see that the perfection of the instrument, designed
by Dornier, then Astrium, made our task much easier. The long-term stability
of the local oscillator, so necessary for us beginners, was much better than
needed, as it was necessary only during the formation of the synthetic antenna
and during the revisit time. Now we know that the millimetre stability of parts
of the solid world allows for the correct parameterisation and estimation of the
rather few unknowns, be they orbital corrections, exact oscillator frequencies,
atmospheric biases or, in the modern multi-baseline tomography, vertical
forest structures or slow motions of the targets. We just need an energy source
and a crude receiver, as parasitic interferometry would show, and astronomers
using very-long-baseline interferometry know so well.
Immediately we saw that in quite a few places the coherence became too
low even after a few days, so that, when we entered the 35-day phase after the
long geodetic interval and the initial three-day revisit interval, we knew that
something new and different was needed. On the other side of the ocean, SRTM
was being conceived to be carried out much later, in February 2000. Bistatic,
single-pass interferometry, as noted above, would be a cure for atmospheric
biases, but that cure would be expensive and hard to get. Looking to the
forthcoming launch of ERS‑2 in 1995, during that first meeting on interferometry
convened by Steve Coulson in Frascati in October 1991, we proposed what was
later called the tandem experiment, namely, the imaging of the same scene
with the two satellites in rapid sequence in order to create highly coherent
interferometric pairs. The multiplicity of passes would compensate for
atmospheric delays, small to start with as the delay was supposed to be short.
Being unfamiliar with the real problems of orbit engineering, we would have
liked some sort of temporal tomography, to be able to change progressively
the revisit interval from a few minutes to a few days, in order to elucidate the
mechanics of the scatterers. But this was still science fiction, as was our initial
radar programme, although it still would be an interesting experiment, maybe
doable with a geosynchronous machine.
Anyway, the joint use of the two satellites would have many applications,
and a meeting in Frascati in early 1995 recommended to ESA the experiment
that was indeed carried out, but only for a nine-month period in 1996, due to
budget limitations. This splendid data set has been exploited again and again.

261
SP-1326

Figure 2. The first ERS‑1/ERS‑2 tandem

interferogram, central Italy, May 1995.
(ESA/POLIMI)

The DEMs that could be derived using multiple baselines to solve unwrapping
problems as well as atmospheric biases would be comparable with and, some
believed, exceed in quality those yielded by the much more expensive SRTM,
to be carried out five years later. There were doubts, voiced by an eminent
researcher, that the two local oscillators and the different orbits could create
unwanted unacceptable fringes. They did, but it was an easy correction to
carry out. Further, the lifetime of ERS‑1 could be impaired if it had to follow
ERS‑2 (maybe the same, one said). But the experiment was carried out
successfully, and the very first joint interferogram was published on the ESA
website in May 1995, a few days after the end of the Commissioning Phase of
ERS‑2. Our only problem was to adapt for different pulse repetition frequencies
(Fig. 2), but the fact that tandem operation was not easy was confirmed by the
following comments that appeared in an ESA Bulletin (Duchossois & Martin,
1995; emphasis added):
‘The passes of the two satellites over the ground receiving stations last
about 10  min. The reconfiguration of the stations between the end of one
satellite’s pass and the beginning of the pass of the next takes about 15 min;
since the orbital period is approximately 100 min, the time interval between
the two satellites has to be between 25 and 75 min. With the current orbital
configuration, ERS‑2 follows ERS‑1 with an approximate delay (called the
orbit phasing of the two satellites) of 35 min. Because of this delay and Earth’s
rotation, the ground-track patterns of ERS‑2 are shifted westwards with respect
to those of ERS‑1. The orbit phasing has been adjusted to ensure that ERS‑2’s
track over the Earth’s surface coincides exactly with that of ERS‑1 24 h earlier.
Within the repeat cycle of 35 days, the opportunity to observe any point on the
ground under identical conditions (altitude, incidence angle, etc.) is therefore
doubled during the tandem operations. Different ground-site revisiting intervals
can be achieved by changing the orbital phasing of the two satellites.’

2.5 Interferometric Applications

One of the most important applications of the tandem dataset was forestry.
The data were exhaustively studied by several experts: Jan Askne of Chalmers
University of Technology in Sweden, and his students Patrick Dammert and
Maurizio Santoro, and then Lars Ulander, from the CARABAS team who

262
 SAR Interferometer

independently rediscovered the ω−k focusing technique in 1987; Urs Wegmüller

and Charles Werner (Wegmüller, 1997) who used Gamma, the symbol of
coherence, as the name of their startup company in 1995; Christiane Schmullius
and her team; and many others within the splendid SIBERIA activity who were
able to exploit the tandem dataset to the full (Schmullius et al., 1999).
In 2000, SRTM came and the results, published after quite a while, were
very satisfactory. The SRTM DEM is used by all of us. Meanwhile, important
results on Phase Unwrapping were published by Dennis Ghiglia (Ghiglia &
Romero, 1996) and then Mario Costantini (1998). Costantini used minimum
flow methodologies for PU, and minimised the global average height (the cost
of the flow of the charges from sources to sinks) of the walls of the terraces
to be added to the DEM to make it as continuous as possible. As the height of
these walls corresponds to an integer number of altitudes of ambiguities (the
topography change corresponding to one fringe), it is invisible from the data as
to add it to the DEM does not change the interferogram phases. The minimum
flow methodology is now well known and is used in many codes. Further, the
great effort of Michael Eineder and his team to unwrap and create the DEMs
from the SRTM X‑band data using multi-baseline and ascending/descending
views should also be remembered.
We obtained further results in interferometry in 1994 by introducing the
wavenumber shift paradigm, i.e. showing that centre frequency and incident
angle played complementary roles. By using multipass observations with
different baselines one could achieve wider range resolution, if the targets were
stable. The quest for stable targets would then lead to the selection of permanent
scatterers, to be seen later. Again using the wavenumber shift paradigm, we
introduced ScanSAR interferometry, if the azimuth bursts could be spatially
synchronised in order to have Doppler spectrum overlap (Gatelli et al., 1994).
Beautiful examples were provided, thanks to the efforts of Betlem Rosich
(Rosich & Laur, 1996). Results on real data were given by DLR and CCRS in 1999,
and then, with ESA−ESOC working hard to achieve synchronisation, beautiful
examples could systematically be provided with Envisat, yielding interferograms
400 km wide that proved very useful for seismological applications (Fig. 3).
The operational mode of Sentinel‑1 is Terrain Observation by Progressive Figure 3. The 2011 Japan earthquake.
Scans (TOPS), a slightly different and more efficient sort of ScanSAR that An example of ScanSAR interferometry
we proposed later. Using once again the same paradigm, we proposed obtained with ALOS PALSAR data. (© JAXA/
interferometric quick looks and ERS‑2/Envisat interferograms, compensating METI)
for the 30 MHz frequency change with a −2 km baseline. We introduced what is
now called Cross-Interferometric SAR (CInSAR). The large baseline allows very
small altitude ambiguities and therefore very good DEMs of flat areas such as
coastal plains and marshes (Fig. 4).
Meanwhile ESA provided ever more products for interferometry: SLC
images and a catalogue of normal baselines were made available to improve
services. Two important papers on SAR interferometry were published (Bamler
& Hartl, 1998; Rosen et al., 2000) just in time for SRTM. Ramon Hanssen’s book
Radar Interferometry (and atmospheric effects) was published in 2001, and our
ESA Manual on InSAR Principles, written together with Didier Massonnet, was
published in 2007 (Ferretti et al., 2007). With regard to the scientific audience
(following Harzing’s Publish or Perish rule), by September 2011 the papers by
Goldstein et al. (1988) on 2D phase unwrapping, the one by Massonnet et al.
(1993) on the Landers earthquake, and ours on permanent scatterers (Ferretti et
al., 2001; to be discussed later), had received a total of 850 citations.
The data available were growing and growing. New satellites were available
for interferometry, like Radarsat-1 and the Japanese Earth Resources Satellite 1
(JERS‑1), and all showed the familiar interferograms. Beautiful images, but
then what? The impairment due to the atmospheric vapour was significant,
and only the now concluded tandem mission, but carried out with a very short
revisit interval, could provide a cure.

263
SP-1326

Figure 4. Digital Elevation Model of

the Po delta, Italy, an example of cross
interferometry between ERS‑2 and Envisat;
area covered 50 × 30 km. (From Santoro et
al., 2008; ESA/Gamma Remote Sensing AG)

2.6 Differential Interferometry: single pass and multipass

But the best was still to come. The idea of Differential Interferometric SAR
(DInSAR) had been proved by Richard Goldstein, who showed that Seasat
measured vertical millimetre motion of agricultural fields in California,
and surmised that this could be due to the changing humidity of the terrain.
Further studies were made by the team of Laurence Gray at the Canadian CCRS.
Using an airborne SAR in C- and X‑bands, with eight repeated passes, they
measured the millimetre motion of corner reflectors in 1990. The results, first
presented at the Progress in Electromagnetics Research Symposium in 1991,
were published in 1993 (Gray & Farris-Manning, 1993).
The first studies of ground motion using satellite data were carried out on the
landslide in St Étienne de Tinée, near Nice, France, in 1992 (Fig. 5). Francesco

Figure 5. Detection of the landslide at

St. Étienne de Tinée, France, 1992, using
SAR interferometry and modelling. (From
Fruneau et al., 1996)

264
 SAR Interferometer

Parizzi and Bénédicte Fruneau, then a PhD student with Josè Achache, but for
a time in Milan, observed the first landslide fringes in February 1992. They had
just three images. Then the 3‑day repeat phase in winter 1992 provided the data
needed to separate the topographic from the motion fringes. The atmospheric
effects were reduced by the fact that the motion was local. The splendid
interferogram of the co-seismic motion of the Landers earthquake in California,
published by Massonnet et al. (1993, 1994; see Fig. 6) generated considerable
interest among seismologists.
That paper was exemplary, in that not only was the interferometry part
significant but also the seismological analysis was very detailed. Co-seismic
motion indeed proved to be easy to see, and many other examples followed.
But glacier motion was also made visible as the many beautiful examples by
Gilles Peltzer and Lawrence Gray showed. But then came the real questions:
ok for co-seismic motion, glaciers, etc. but how about pre-seismic motion? Is
it there, can we see it? Where are the signatures of the faults in motion? Could
this information be useful for understanding earthquakes?

3. The Bonn Experiment Figure 6. The interferogram yielding the

co-seismic signal of the Landers
The first experiment to measure millimetre ground motion from a satellite was earthquake, 1992. (Massonnet et al., 1993)
the Bonn experiment, a collaboration between ESA, Milan and the University
of Stuttgart team led by Philip Hartl and Karl-Heinz Thiel. We proposed to
install 18 Corner Reflectors (CRs) along a line in February–March 1992, and
then to move one or two by a couple of centimetres. In a blind experiment,
we would tell, from the satellite data, which ones had been moved. The final
meeting was in Milan in May 1992, and Steve Coulson was the umpire. The
results were clear; the two CRs that had been moved were identified and their
motion established. As Steve said, it was not just a success for these two; the
fact that the other 16 were seen not to have moved had been a result too. The
experiment was not that easy. To position so many CRs in a short time had been
difficult enough. Even worse, a thunderstorm eradicated most of the corners, so
that the duration of the experiment was noticeably reduced. Nonetheless, we
had the proof that DInSAR did work.

3.1 Corner Reflectors Everywhere?

But without CRs what to do? First, install as many as you can, we thought. At
the time, people were happy if they could see their CRs, whereas now they
become anxious if they move by more than a millimetre! So, we initiated a
study for the EU with Jose Achache and Hartl, and we placed a few CRs in the
Pozzuoli Solfatara and on Mount Vesuvius near Naples. We were helped by the
late V. Murino, from the University of Naples. But too many other Neapolitans
were too curious and liked to touch the objects; at times, they liked them so
much they carried them away. In fact, we got nothing out of these efforts. The
CRs were visible for a few months and then either they moved or disappeared
altogether. Germany is indeed far from Naples, and the Bonn experiment had
a short life anyway. Then, expecting strange and unguarded objects to remain
untouched for a long time appeared to be, and still is a very difficult problem
to solve, as Ramon Hanssen at TU Delft found out later in the much quieter
Netherlands. So, we began to feel the need for natural persistent or Permanent
Scatterers, as we later called them, a long time before we found them.

265
SP-1326

3.2 Long-term Interferometry

Meanwhile our team had been joined by Alessandro Ferretti who had written
his PhD thesis on multi-baseline DEMs (Ferretti et al., 1999). It was nice, it
worked, but the coherent combination of multiple datasets with multiple
baselines was a hard problem to solve. For one, we had to compensate
for atmospheric biases. We were not alone in that research: we would be
competing with a research group from the National Research Council (CNR/
IREA) in Naples, coordinated by Riccardo Lanari. Many more groups followed
and, as all research can and should be replicated, that was what they did.
Our two groups followed two different tracks that are now merging. The
IREA team was more interested in spatial continuity and, due to the vagaries
of the baselines of ERS satellites, could not find coherence in temporally
contiguous datasets if the baselines were too far apart: the coherence on the
distributed scatterers was lost due to the wavenumber shift. Thus, the total
baseline span had to be divided into groups of Small Baselines (SBAS) so to
create space continuous interferograms, but then the different baseline groups
would be temporally discontinuous. A smooth motion could be identified by
the joint analysis of the complete set, optimising smoothness using singular
value decomposition. They started publishing on this topic in 2000.
In Milan, we followed the other track of time continuity, and thus space
discontinuity. We looked for points that were there all the time, natural CRs,
which we named Permanent Scatterers (PSs). These points did not change, or
changed in a predictable way the phase and amplitude of their reflection with
time or baseline. Thus, we could use them in all the interferograms as pivots on
which to determine the local atmospheric bias (the Atmospheric Phase Screen,
APS). If their spatial density was high enough, we were able to find the APS,
peel it and then find many more PSs. The problem was to find the first ones,
and this we did by looking into their amplitudes, independent of the APS: if the
phases are stable, the amplitudes should be as well, shouldn’t they?

3.3 The First Results

By the end of 1998 we had obtained many quantitative results of the PS
technology, over the Phlegraean Fields (subsiding), Naples (subsiding), Vesuvius
(not moving), Etna (cooling), and then Ancona, where there had been a gigantic
landslide a few years ago and the terrain was still sliding, down mostly, but up
in places. This we found out, and we were comforted that it was not a mistake
as Alfredo Mazzotti and the municipality provided us with historical data and
levelling analyses. Before any publication, ESRIN set up a video conference to
inform the key Earth observation value-added companies in February 1999.
Anyway, we were able to patent the concept for our school in May 1999.
In the following months, we recognised the incredible stability of the
terrain, and that millimetre motions could be measured so easily and so
accurately across seismic faults, landslides and buildings in motion, even
though none of the imaged objects was marked with a precise reference
point. Maybe this was not a killer application, but it was a way to use remote
sensing to find things that were not visible to bystanders. We were competing
with the ancient art of optical levelling, but our advantage was that, thanks to
the ERS archives, we could see things that had happened in the past, and no
one else could. At IGARSS ’99 we published the results of a study of Pomona,
California, where we could see the subsidence due to water extraction and, for
the first time, a fault in motion (Fig. 7). Then we processed results for the Los
Angeles basin (Fig. 8), and we started seeing faults moving everywhere. The
competition with GPS was rapidly followed by a fruitful cooperation as the
two systems are complementary: GPS is absolute, space discontinuous, time

266
 SAR Interferometer

Figure 7. Subsidence in Pomona, California.

The blue uplift (top left) is due to a moving
fault. (Ferretti et al., 2000)

Figure 8. Subsidence in the Los Angeles

basin. (TRE)

continuous and regional, while DInSAR is relative, space continuous, time

discontinuous and local (Ferretti et al., 2000, 2001).
The extraction of information from a set of coherent points was also
proposed by Stefania Usai, from TU Delft, at IGARSS ’99. Following on from
her work, a significant research group was formed in Delft, headed by Ramon
Hanssen who had spent part of his PhD studies with Howard Zebker at Stanford
from JPL since 1995.
Another interesting result was obtained for the disaster of Via di Vigna
Jacobini in Rome in December 1998, when a building had collapsed because
too heavy equipment had been installed without sufficient prior engineering
analysis, as the judge later said. We could see the building going down, and
down, until it collapsed. The experiment, made in February 1999 to prove
that we were actually looking at that building, was carried out in cooperation
with Fabrizio Ferrucci, then of the Italian Civil Protection Department, who
mobilised the assistance of Roman firemen, Francesco Posa of the University of
Bari who gave us three gigantic CRs to position on the ground, and also Marco
Anzidei of the Istituto Nazionale di Geofisica e Vulcanologia (INGV), who used
GPS technology to locate the reflectors. Indeed, we had been watching the
building before it collapsed.

267
SP-1326

Figure 9. Italy’s Permanent Scatterers from ERS ‑1, ERS‑2 and Envisat data. (TRE)

After that, the technologies of PS, SBAS, and that of interferogram stacking
proposed by Sandwell & Price (1998) provided many results all over the world
(Fig. 9). ERS, and then Envisat and Radarsat‑1, provided so many datasets that
all researchers could fulfil their subsidence dreams.

4. Fringe Meetings, PSIC4 and the Market

Steve Coulson and then Yves Louis Desnos contributed to interferometry
by organising Fringe Meetings in 1991 and 1992 (both in Frascati), and then
in Zurich in 1996, Liège in 1999, and in Frascati every two years after 2003.
Each meeting was attended by a couple of hundred people, 150 papers or
posters, and lots of interaction. The last one, in September 2011, attracted 320
participants. The main topics discussed at the Fringe Meetings included Phase
Unwrapping, conventional DInSAR, DEMs, new missions, etc., but after 2003
there were more papers related to PS-like algorithms. It was then decided, in
agreement with ESA, that the growing family of second-generation DInSAR
algorithms aiming at exploiting natural radar targets would be referred to as
Persistent Scatterer Interferometry (PSI).
Also during Fringe 2003, with Richard Bamler, we proposed to establish a
comparison of different PS techniques: the PSI Codes Cross-Comparison and
Certification (PSIC4); see http://earth.esa.int/psic4/). Although it would be
unfair to say that the commercial exploitation of SAR interferometry started
with the PS technique, PSI was a boost for space geodesy and created a new
market sector for of SAR data distribution companies eager to find new civilian

268
 SAR Interferometer

applications of radar data, which is still used primarily for intelligence and
security applications. ESA had supported the creation of a market for PSI
applications from the beginning and financed several projects based on PSI
products and services. Among them, it is worth mentioning Terrafirma, an ESA
GMES project, managed by Fugro NPA, which aims to create ‘a pan-European
ground motion hazard information service’ (www.terrafirma.eu.com).
In order to speed up the expansion of PS applications, ESA focused on two
key concepts, validation and standardisation, neither of which turned out to be
at all easy to reach. The former is needed to fulfil the need to involve different
scientific communities and create a common language for comparing results.
The latter is needed for the simple reason that this time there was indeed a
market (!) for this technology, and differentiation rather than standardisation
is a must in any marketing plan.
Looking back over the last 10 years, both lines of action by ESA were
successful, since they forced all actors involved in PSI to improve the quality
and reliability of PS results, figure out any possible applications of PS data, and
learn how to speak in layman’s terms about this remote sensing technology.
We conclude this section by pointing out, once again, the importance
of SAR data availability and data acquisition policies for the development of
SAR interferometric technologies/applications. All PSI algorithms owe much
to ESA’s ERS missions. The ERS‑1 and ERS‑2 satellites were uncomplicated
and manageable using simple acquisition policies. Since interferometric data
stacks were available over many areas of the world, with dozens of image
acquisitions, the development of the second-generation DInSAR algorithms
became possible. The existence of an up-to-date archive of speculative
scheduled raw data acquisitions has helped to build the confidence of the
user community in using InSAR. Also, the market potential of speculative
scheduled SAR has encouraged satellite owners and distributors to become
interested in such programmes.

5. Today and Tomorrow

The current constellations of X-band satellites, the future Sentinel‑1 A and B,
the Argentinean L-band system, the Chinese S-band ones, etc., are all systems
that will offer interferometry in the short and the long term. The key parameter,
the revisit time, will shorten progressively and geosynchronous systems have
been proposed to reduce that interval to 12 h. Maybe their time will come too.
Indeed, observations of dykes, oil and gas fields (Fig. 10) and cities will
continue and the creation of worldwide DEMs will improve. The mechanisms
of interferometry are now well known, and indeed the next steps, polarimetry
and tomography, also very well understood. Lower-frequency radiation like
P-band will penetrate the targets and multi-baseline systems will be able
to create cross-track resolution, thus allowing the retrieval of the vertical
structure of forests, for example.

Figure 10. Surface deformation of a gas

field.

269
SP-1326

The future will tell if systematic global monitoring will become operational,
but there are many indications that it will. Microwaves will do their job
illuminating targets that beams of light cannot penetrate, and scientists will
find other interesting topics to study. The adventure of interferometry as a
science is basically concluded, and it is now an industrial and commercial
engineering activity.

References
Bamler, R. (1992). A comparison of range-Doppler and wave-number domain SAR
focusing algorithms, IEEE Trans. Geosci. Rem. Sens. 30(4), 706−713.
Bamler, R. & Hartl, P. (1998). Synthetic Aperture Radar interferometry, Inverse
Problems 14, R1–R54.
Cafforio, C., Prati, C. & Rocca, F. (1991). SAR data focusing using seismic migration
techniques. IEEE Trans. Aerospace Electron. Syst. 194−207.
Costantini, M. (1998). A novel phase unwrapping method based on network
programming. IEEE Trans. Geosci. Rem. Sens. 36(3), 813–821.
Curlander, J.C & McDonough, R.N. (1991). Synthetic Aperture Radar: Systems and
Signal Processing. Wiley, New York.
Duchossois, G. & Martin, P. (1995). ERS‑1 and ERS‑2 tandem operations. ESA Bulletin
83, 54–60.
Engdahl, M., Borgeaud, M. & Rast, M. (2001). Use of ERS-1/2 tandem interferometric
coherence in the estimation of agricultural crop heights. IEEE Trans. Geosci.
Rem. Sens. 39(8), 1799–1806.
Ferretti, A., Monti Guarnieri, A., Prati, C., Rocca, F. & Massonnet, D. (2007). INSAR
Principles: Guidelines for SAR Interferometry Processing and Interpretation. ESA
TM-19, European Space Agency, Noordwijk, the Netherlands.
www.esa.int/esaMI/ESA_Publications/SEM867MJC0F_0.html
Ferretti, A., Prati, C. & Rocca, F. (1999). Multibaseline INSAR DEM reconstruction,
the wavelet approach. IEEE Trans. Geosci. Rem. Sens. 37, 705-715.
Ferretti, A., Prati, C. and Rocca, F. (2000). Nonlinear subsidence rate estimation
using permanent scatterers in differential SAR. IEEE Trans. Geosci. Rem. Sens. 38,
2202–2212.
Ferretti, A., Prati, C. and Rocca, F. (2001). Permanent scatterers in SAR interferometry.
IEEE Trans. Geosci. Rem. Sens. 39(1), 8–20.
Franceschetti, G. & Lanari, R. (1999). Synthetic Aperture Radar Processing, CRC
Press, Boca Raton, FL.
Fruneau, B, Achache, J. & Delacourt, C. (1996). Observation and modelling of the
Sant- Étienne-de-Tinée landslide using SAR interferometry. Tectonophysics 265,
181–190.
Gatelli, F., Monti Guarnieri, A., Parizzi, F., Pasquali, P., Prati, C. & Rocca, F. (1994).
IEEE Trans. Geosci. Rem. Sens. 32, 855−865.
Ghiglia, D.C. & Romero, L.A. (1996). Minimum Lp-norm two-dimensional phase
unwrapping. J. Opt. Soc. Am. A 13(10), 1−15.
Goldstein, R., Zebker, H. & Werner, C., (1988). Satellite radar interferometry: two
dimensional phase unwrapping, Radio Sci. 23(4), 713–720.
Graham, L.C. (1974). Synthetic interferometer radar for topographic mapping. Proc.
IEEE 62(6), 763−768.
Gray, A.L. & Farris-Manning, P.J. (1993). Repeat-pass interferometry with airborne
synthetic aperture radar. IEEE Trans. Geosci. Rem. Sens. 31(1), 180−191.
Hanssen, R.F. (2001). Radar Interferometry: Data Interpretation and Error Analysis,
Kluwer Academic, Dordrecht, the Netherlands.
Massonnet, D., Feigl, K., Rossi, M. & Adragna, F. (1994). Radar interferometric
mapping of deformation in the year after the Landers earthquake. Nature 369,
227−230.

270
 SAR Interferometer

Massonnet, D., Rossi, M., Carmona, C., Adragna, F. , Peltzer, G., Feigl, K. & Rabaute,
T. (1993). The displacement field of the Landers earthquake mapped by radar
interferometry. Nature 364, 138–142.
Pasquali, P., Stebler, O., Small, D., Holecz, F. & Nuesch, D. (1998). SAR polarimetric
interferometry experiments. Workshop on Advances in Radar Methods, Progress
in Electromagnetics Research Symp., Baveno, Italy, 20–22 July.
Prati, C. & Rocca, F. (1993). Improving slant-range resolution with multiple SAR
surveys. IEEE Trans. Aerospace Electron. Syst. 29(1), 135−144.
Prati, C. & Rocca, F. (1994). Process for Generating Synthetic Aperture Radar
Interferograms. US Patent No. 5 332 999.
Prati, C., Rocca, F., Guarnieri, A.M. & Damonti, E. (1990). Seismic migration for SAR
focusing: Interferometrical applications. IEEE Trans. Geosci. Rem. Sens. 28(4),
627−640.
Prati, C., Rocca, F., Monti Guarnieri, A. & Pasquali, P. (1994). Interferometric
Techniques and Applications. ESA Study Contract Report. Contract No. 3- 7439/92/
HGE-I.C.
Rosen, P., Hensley, S., Joughin, I., Li, F., Madsen, S., Rodriguez, E. & Goldstein, R.
(2000). Synthetic aperture radar interferometry. Proc. IEEE 88(3), 333–379.
Rosich, T.B. & Laur, H. (1996). Phase preservation in SAR processing: The
interferometric offset test. IGARSS 1996, Lincoln, NE, USA, May 1996.
Rott, H. & Siegel, A. (1996). Glaciological studies in the Alps and in Antarctica using
ERS interferometric SAR. Proc. Fringe 1996, ESA SP-406, vol. 2, European Space
Agency, Noordwijk, the Netherlands, pp.149–159.
Sandwell, D.T. & Price, E.J. (1998). Phase gradient approach to stacking
interferograms. J. Geophys. Res. 103, 30 183−30 204.
Santoro, M., Wegmüller, U., Strozzi, T., Werner, C., Wiesmann, A. & Lengert, W.
(2008). Thematic applications of ERS–Envisat cross interferometry. IGARSS 2008.
Schmullius, C., Holz, A., Marschalk, U., Roth, A., Vietmeier, J. & Wagner, W. (1999).
Operational readiness of ERS SAR interferometry for forest mapping in Siberia.
Fringe 99, ESA SP-478, European Space Agency, Noordwijk, the Netherlands.
SRTM Shuttle Radar Topography Mission: Mapping the World in Three Dimensions.
http://srtm.usgs.gov/
Wegmüller, U. (1997). Soil moisture monitoring with ERS SAR interferometry. Proc.
3rd ERS Symp. ESA SP-414, European Space Agency, Noordwijk, the Netherlands,
pp.47-52.
Zebker, H.A. & Goldstein, R.M. (1986) Topographic mapping from interferometer
synthetic aperture radar observations. J. Geophys. Res. 91(B5), 4993–4999.

271
→→ ERS SAR over Land
 SAR Image Mode

ERS SAR over Land

T. Le Toan

Centre d’Etudes Spatiales de la Biosphère, CNRS-CNES Université Paul Sabatier,

Toulouse, France

1. Introduction
With the launch of the first Landsat in 1973, mapping of land surfaces from
space became a reality. In the late 1970s, following increasing concerns that
human activities are starting to affect the ecological balance of the planet,
environmental monitoring and conservation became major objectives, to
be addressed by global and repeated observations. After the short Seasat
mission in 1978, it was clear that radar missions were needed to undertake
measurements of areas not covered by existing missions. The first European
Remote Sensing satellite, ERS‑1, launched in 1991, was designed primarily to
monitor the sea state, surface winds, ocean circulation and sea/ice levels. All-
weather imaging of ocean, ice and land was regarded as an additional benefit.
However, the land community was very interested in using Synthetic
Aperture Radar (SAR) for land observations, primarily because imaging radar
was the only means available for mapping areas under persistent cloud cover
and for monitoring land surface changes. To assess the use of SAR data for land
applications, research was undertaken well before the launch of ERS‑1 to develop
methods of land surface mapping and monitoring land surface parameters.
Following the launch of ERS‑1 in 1991 and of ERS‑2 in 1995, ERS data were the
basis of many scientific projects and operational services, which continued with
Envisat’s Advanced SAR (ASAR) and will continue with Sentinel-1.
This chapter recounts the journey from the era before ERS until today,
outlining the scientific and operational achievements and their impacts
on the public, and on new developments. This chapter is not intended to be
comprehensive, but focuses on the use of ERS and Envisat SAR for agriculture
and forest research. The observations and opinions presented here are also
subjective since they represent a personal view of the history of ERS.

2. The Era before ERS

Until the 1970s, radar remote sensing research was focused on understanding
geometric factors associated with SAR imaging, and on the generation of
geological and geomorphological maps of cloud-covered areas. The use of SAR
data for a more diverse range of land applications required an understanding
of radar backscatter from natural surfaces such as soil, water and vegetation.
For this purpose, ground-based scatterometer experiments were initiated by
Fawwaz T. Ulaby of the University of Kansas. Between 1970 and 1984, extensive
measurements were made and many innovative experiments were conducted
in the USA and in Europe, most notably in the Netherlands (P. Hoogeboom,
E. Attema) and in France (T. Le Toan). Other teams working on ground-
based systems (active and passive) included those at the University of Bern,
Switzerland (E. Shanda), the National Research Council (CNR/IREA) in Florence,
Italy (P. Pampaloni) and the European Joint Research Centre, JRC (A. Sieber).
In these ground-based experiments, the primary targets were agricultural
fields of bare soil and various crops (e.g. wheat, corn, soybean, alfalfa, grass) that
were monitored throughout the growing season. The results, published in several
papers and summarised in books (Ulaby et al., 1986), indicated that at frequencies
of 1–18 GHz the backscattering coefficient σ0 is sensitive to variations in soil and

275
SP-1326

vegetation parameters. An example of the early results is shown in Fig. 1, where

the backscattering coefficient σ0 (at the C-band, 20° incidence angle and VV
polarisation) of bare soil surfaces for three roughness states of agricultural fields
was found to be sensitive to the soil moisture content. These examples provided
the first stimulus for the use of radar remote sensing for agricultural applications.
In the late 1980s, European airborne campaigns were organised to collect
data for diverse test sites and conditions. The first campaign, AgriSAR (1986),
used the Varan-S from CNES over agricultural sites three or four times during
the growing cycle. The MAESTRO-1 campaign (1989) was organised jointly
by the JRC and ESA (P.  Churchill and E.  Attema). The NASA/JPL AirSAR was
deployed to collect and evaluate P, L and C data over five European test sites,
mostly forests (Churchill & Attema, 1994). The MAC Europe campaign in
1994 was also conducted over forest sites. Airborne scatterometers have been
developed in Finland (M. Hallikainen) and in France (R. Bernard).
The diverse scenes observed by airborne systems revealed much more complex
relations between C-band data and vegetation and soil parameters than had the
results obtained during ground-based studies. For bare soil under a wide range of
tillage conditions, for example, it was found that the effects of surface roughness
needed to be accounted for in the soil moisture retrievals (Beaudoin et al., 1990).
Because of the diverse crop types encountered, it was found to be difficult to
derive robust methods of crop classification or retrieval of crop parameters using
single SAR measurements (in terms of frequency, polarisation and incidence
angle). There was one exception, however, and that was the rice observed during
the AgriSAR campaign in the Camargue, France. Because of the dominant
vegetation−water interaction mechanism, rice can be monitored based on its
unique temporal and polarisation signature (Laur, 1989; Le Toan et al., 1989). For
forests, the campaign data using multifrequency SAR showed that C-band data
had only weak sensitivity to forest parameters, and the lowest frequency (P-band)
was found to be the most suitable for biomass measurements (Le Toan et al., 1992).
With the growth of experimental databases, important progress in modelling
was achieved, leading to a significant expansion of our capability to interpret
radar signals. A simple cloud model gave a first key to understanding the σ0
dependence on the main soil and vegetation parameters (Attema & Ulaby, 1978),
while more complex models simulated σ0 using radiative transfer models (A.K.

Figure 1. Measured backscattering

coefficient σ0 (left scale) as a function of
soil moisture content for three surface
roughness states. The solid curve is the
reflectivity (right scale) calculated on the
basis of dielectric measurements. (Le Toan,
1982; Ulaby et al., 1986)

276
 SAR Image Mode

Fung, University of Kansas) and random medium models (J.A.  Kong, MIT).
Modelling studies were later undertaken by teams at the University of Rome Tor
Vergata, Chalmers University of Technology in Sweden, and the CESR (now the
Centre d’Etudes Spatiales de la Biosphère (CESBIO) in France.
In parallel, several European teams intensified their activities devoted to
SAR image processing, including those at the Marconi Research Centre in the
UK (S. Quegan), the University of Stuttgart in Germany (P. Hartl), and at CESR
(A. Lopes, R. Touzi) and IRESTE in France (J. Saillard, E. Pottier), to name but a
few. Their achievements included the development of methods for calibration,
speckle filtering, and geometric and radiometric correction. SAR polarimetry,
first demonstrated in mid-1980s, was capturing the interest of an increasing
number of researchers. However, beyond a few applications, polarimetry
for land did not lead to the full development of applications as many had
predicted. In contrast, interferometry, which in the mid-1980s had been
regarded as a tool only for topographic mapping, became an indispensable
observation technique for many land applications (see Rocca, this volume).
It is interesting to note that in 1981, before the era of SAR satellites, the first
International Geoscience and Remote Sensing Symposium (IGARSS) was devoted
to radar remote sensing, and attracted over 400 participants from 20 countries.

3. The ERS years: research and operational results

Although ERS‑1 was designed for observation of ocean processes and polar
ice regions, the use of SAR data for land observations soon attracted a large
number of scientific projects, most of which involved studies of renewable
resources such as agriculture and forests. In 1998, about 39% of ESA’s
Announcements of Opportunity (AO) projects were devoted to oceanography,
sea ice and ice sheet analysis, 9% to studies of atmospheric processes, 40% to
renewable resources and environmental monitoring in land applications, and
11% to geology and hazards (Rémondière & Laur, 1998).
The research on the use of SAR for land applications has been reported in
various symposia and workshops, and in most remote sensing journals. The
results of the ERS AO investigations were presented at ESA’s international
symposia and workshops, from the first in Cannes in 1992, to the most recent
Living Planet symposium in Gothenburg, Sweden, in 2010. The first symposium
dedicated to land applications, on the retrieval of bio- and geophysical
parameters from SAR data, was held in Toulouse in 1995 (T. Le Toan), and
the series of meetings continued in Noordwijk, the Netherlands, in 1998
(M. Borgeaud), Sheffield, UK, in 2001 (S. Quegan), Innsbruck, Austria, in 2004
(H. Rott) and Bari, Italy, in 2007 (F. Mattia).
It would be an enormous task to summarise all that we learned during
those years about how to make use of SAR data for our environment. Instead,
this chapter outlines the approaches used, the results obtained and the lessons
learned in just a few research and application domains: soil moisture retrieval,
forest mapping, crop monitoring and hazard mapping.

3.1 Soil Moisture Retrieval

The first ERS experiment was conducted in the Ghard area of western Morocco in
1992. It clearly showed that while irrigated and non-irrigated agricultural fields
could be detected in the ERS‑1 SAR images, the relationships between the SAR
signal and soil moisture are influenced by surface roughness (Fig. 2) and vegetation
cover. Accounting for these effects proved to be a major scientific challenge when
using single VV, 23° C-band SAR data to measure soil moisture (Le Toan et al.,
1993). Meanwhile, theoretical model simulations of σ0 failed in comparison
with ERS SAR data (Borgeaud et al., 1995), and the reason, we found, was the

277
SP-1326

scale effect in in situ measurements for providing model inputs. The measured
roughness correlation length with 3 m profiles differed from the measurements at
the SAR spatial resolution, as found by measurements with a longer profiler (>10
m), for which a specific setup was developed (Mattia & Le Toan, 1999; Davidson et
al., 2000). To assimilate the soil moisture estimates in a flood forecasting scheme,
a model inversion of ERS data based on a priori knowledge of the roughness state
was proposed (Bach & Mauser, 2003; Davidson et al., 2003). However, we know
that for agricultural and hydrological applications the most important requirement
is the data short repeat cycle. With such repeated data over a few days it would be
possible to track changes in soil moisture, since the changes in other parameters
(surface roughness, vegetation structure and biomass) take place at longer
temporal scales. This was later demonstrated with data from the Envisat Advanced
SAR (ASAR) wide-swath mode in a 3-day repeat cycle, which showed changes
in soil moisture content at a coarse spatial resolution (1 km) (e.g. Vista GmBH).
Overall, the lack of systematic data acquisitions was a serious limitation on the use
of ERS (and later Envisat) SAR data for estimating soil moisture for applications
such as agricultural monitoring and hydrological modelling.
For global observations, methods based on change detection using dense
time series and coarse-resolution data such as ERS scatterometer data had
already been developed and used in process models (W. Wagner). More
recently, the Soil Moisture and Ocean Salinity mission (SMOS) mission was
launched in 2009 for global soil moisture monitoring (Y. Kerr). But for many
applications in agriculture and hydrology, which require higher spatial

Figure 2. Left: ERS‑1 image of the agricultural area of Gharb, Morocco. Irrigated and non-irrigated fields, and fields where irrigation was under
way can be identified. Right: ERS backscattering coefficient as a function of soil moisture for fields at two types of roughness status (old and new
fallow bare fields of the same rms height s with different surface correlation lengths, l). (Le Toan et al., 1993, in Massonnet & Souyris, 2008)

278
 SAR Image Mode

resolution, expectations are now focused on the Sentinel-1 mission, which will
provide an opportunity to put to effective use the knowledge accumulated and
methods developed so far (Satalino et al., 2003; Balanzano et al., 2011).

3.2 Forest and Biomass Mapping

Among the ERS projects on renewable resources, the most popular theme
was forests. By that time, it had been recognised that deforestation in tropical
regions was contributing to increased global carbon emissions. The SAR has
proved itself to be the only realistic source of geographical information on
tropical forests, where the mean annual cloud cover can reach 90%. Mosaics of
ERS images have provided unprecedented maps of large tropical regions, such
as that of Central Africa generated by the JRC as part of the ESA–EU Tropical
Ecosystem Environment Observation by Satellites (TREES) project (1994).
But the most remarkable example was the map of French Guiana (Rudant
et al., 1997), an area that is covered by rainforest and mostly inaccessible,
and for which, except for the coastline, existing maps were incomplete and
inaccurate. Figure 3 shows the ERS mosaic image using 1992–94 ERS data.
The mosaic image was used by IGN, the French mapping agency, to derive the
first cartographic products such as land cover and topographic maps showing
forested areas, clearings, agricultural fields and settlements.

Figure 3. Mosaic image of French Guiana,

South America, based on ERS‑1 data,
1992–94. (J.P. Rudant)

279
SP-1326

Quite soon after the beginning of data acquisition, it was recognised that the
multitemporal ERS data provided more valuable information than single images.
Methods based on time series data for speckle filtering and for change detection
were developed and used for deforestation mapping (e.g. in Sumatra by CESBIO
and ESA in 1998). To assess the use of SAR data for monitoring forests, the EU’s
European Forest Observations by Radar (EUFORA) project (fourth Framework
Programme, 1996−98) brought together several teams to work on a collaborative
basis (Le Toan et al., 1998). Many of the project’s findings, including the
application of multitemporal ERS data to forest mapping (Quegan et al., 2000)
and methods based on SAR interferometry, are still being used.
While the most striking land applications of SAR interferometry have been
measurements of terrain displacements following earthquakes and subsidence
monitoring, the findings for other applications have received much less attention.
In the EUFORA project, for example, J. Askne and his team from Chalmers
University assessed the potential of interferometry for forest applications (Askne
et al., 1997). The Swiss company Gamma Remote Sensing (Wegmüller & Werner,
1997; Strozzi et al., 2000) initiated the operational use of interferometric coherence
in land use and forest mapping. The coherence is a measure of the unwanted
phase noise of the interferogram when dealing with terrain displacements. But
as an interesting point for vegetation observations, the coherence decreases with
volume scattering and temporal changes in the target area. Water and forests
show low coherence, while bare soil and urban areas show high coherence,
and agricultural fields intermediate coherence. Figure 4 presents one of the first
examples of the simplest and most popular method of land surface mapping
using backscatter intensity and interferometric coherence.
Beyond forest mapping using interferometry, the next step involved the
retrieval of forest parameters. The idea was simple: the volume scattering and
temporal changes due to the motion of scatterers increase with the number
of scatterers in a forest canopy. Hence the interferometric coherence was
expected to decrease with forest biomass. Experimental results relating ERS

Figure 4. RGB colour composite of

interferometric coherence (red),
backscatter intensity (green), and
backscatter change (blue), in an area near
Bern, Switzerland. The data were obtained
by ERS‑1, 24−27 November 1991. (Gamma
Remote Sensing)

280
 SAR Image Mode

repeat pass coherence at different time intervals have been reported. The
sensitivity to biomass was found to decrease after a certain value of biomass
was reached, e.g. 40−60 t ha−1 for a temperate forest (Le Toan & Floury, 1998),
or for the Siberian forest (Wagner et al., 2003). Some exceptions were observed
in boreal forests in winter conditions, where the sensitivity can remain high
with biomass values up to 300 t ha−1 (Askne et al., 2003).
The interferometric phase information from ERS SAR was also examined
as a possible source of information to derive forest canopy heights. When the
height of the phase centre in a forest canopy was compared with the canopy
height (Floury et al., 1997), however, it was found that more information was
needed to retrieve both the canopy height and the ground topography. This
was later realised with single-pass interferometry for ground topography
with the Shuttle Radar Topography Mission (SRTM) and the emergence of
the polarimetric interferometry (Pol-InSAR) technique, based on Spaceborne
Imaging Radar SIR-C/X-SAR data (S. Cloude and K. Papathanassiou).
Building on the use of interferometric coherence and intensity, the SAR
Imaging for Boreal Ecology and Radar Interferometry Applications (SIBERIA)
project, led by C. Schmullius, was one of the largest remote sensing campaigns
conducted in Europe. It used three satellites to acquire data for the central
Siberian taiga, an area of global ecological importance. In an unprecedented
joint effort involving the German DLR, ESA and the Japanese National Space
Development Agency (NASDA), ERS‑1/ERS‑2 and the Japanese Earth Resources
Satellite (JERS‑1) were used to acquire data over an area of more than
1.2 million km2 during autumn 1997 and summer 1998. The project generated
96 maps (100 × 100 km) of forest cover and biomass (Fig. 5), enabling Russian
foresters to update their information on the status of central Siberian forests.
More recently, in the ESA and the Chinese Ministry of Science and Technology
(MOST) Dragon programme, coordinated by Y.-L. Desnos and Li Zeng Yuan,
the archived ERS‑1/ERS‑2 tandem data were used to generate forest cover and
biomass maps of China, in a forestry project also led by C. Schmullius.
The SIBERIA-1 data were later used to test a process-based dynamic global
vegetation model. Such models attempt to capture the best representations of
ecosystem processes, estimate the global spatial distribution of carbon fluxes and,
when coupled to a climate model, predict current and future carbon and water
fluxes. The model results and measurements in Fig. 6 show differences in both
the range and spatial distribution of biomass values (Le Toan et al., 2004). The
differences were suggested to be partly the result of forest disturbance contained

Figure 5. The SIBERIA project. One of

the 96 forest cover and biomass maps
generated using ERS‑1/ERS‑2 and JERS‑1
data acquired over central Siberia in 1997
and 1998. (SIBERIA‑1 project, led by
C. Schmullius)

281
SP-1326

Figure 6. Measured and model-based

values of above-ground biomass in central
Siberia (t C ha−1). Top left: The SIBERIA-1
map degraded to pixels of 0.5°; bottom
left: the results of the Sheffield Dynamic
Global Vegetation Model (SDGVM). Right:
Scatterplot comparing the model results
and SIBERIA-1 measurements of above-
ground biomass. (Adapted from Le Toan et
al., 2004)

in the data but not considered by the models, and the saturation in model values
as being due to incorrect assumptions about tree mortality). It was clear that the
model needed to be improved by taking into account the reality revealed by such
biomass maps, and that global maps that cover a much wider range of biomass
than the SIBERIA-1 data are required for testing the model across all biomes. Such
work paved the way for ESA’s future Biomass mission (Le Toan et al., 2011).

3.3 Rice Monitoring

As stated in section 2, agricultural crops were the first study targets for which the
all-weather capability of the ERS SAR was expected to provide a unique means
of crop monitoring. Many papers have been published on crop classification
and on the retrieval of crop biomass using ERS data. The analysis confirmed
earlier airborne campaign results showing that the C-band backscatter signal is
sensitive to both vegetation and soil parameters (e.g. Picard et al., 2003; Mattia
et al., 2003).
It was also found that using a single SAR measurement (VV, 23°) it is difficult
to decouple from the signal soil moisture, surface roughness, vegetation
structure and biomass. This is the reason why few effective applications have
been derived from ERS data. The only exception was inundated rice fields,
where the signal is governed mainly by vegetation parameters. This is because
of the dominant vegetation–water interaction scattering mechanism, which
also explains why the ERS backscatter showed a significant increase (10 dB)
during the plant growing phase (Le Toan et al., 1997; Wang et al., 2004). This
temporal behaviour has been reported in a number of studies using ERS (and
also ASAR and Radarsat) data in Japan (Kurosu et al., 1995), Indonesia (Le Toan
et al., 1997; Ribbes & Le Toan, 1999), China, India, the Philippines, Thailand
(Aschbacher et al., 1995) and Vietnam (Liew et al., 1998). In most rice-growing
regions, where fields are small and do not have the same crop calendar, the
rice mapping methods have been developed based on the unique temporal
behaviour of the backscatter. Such methods have proven to be effective at a
number of sites (Le Toan et al., 1997). Figure 7 shows a map of rice fields near
Samarang, Java, Indonesia, distinguished into early and late rice, according to
the irrigation scheme in the region.
Regional projects for rice monitoring have also been pursued beyond the
research stage such as for the Mekong delta in Vietnam. However, an important
factor that has limited the development of rice applications has been the difficulty
in acquiring systematic multitemporal SAR data during the growing season.
Meanwhile, rice cultivation has raised concerns about its impacts on various
environmental aspects, from water management to climate change. Rice is the
primary food source for more than half of the world’s population; according to
the UN Food and Agriculture Organization, the demand is likely to rise by nearly

282
 SAR Image Mode

Figure 7. Map of rice fields (22 × 22 km)

derived from multitemporal ERS data. Cyan,
late rice; magenta, early rice; yellow, other
crops. Region of Semarang, Java, Indonesia
(6°55 S, 110°41 E). (Le Toan et al., 1997)

70% by 2025. At the same time, growing populations and the rapid pace of
economic development are leading to a decline in the rice-growing area. Changes
in the distribution of rice paddies and in farm management towards more
intensive production systems (multi-cropping, water management, fertilisers
and high-yield cultivars) are likely to intensify over the coming decades. These
changes in the rice-growing area and farming practices are likely to have
significant impacts on the global climate, since irrigated rice fields are among
the major sources of methane emissions. Estimates of seasonal changes in
atmospheric methane concentrations based on observations by the SCIAMACHY
instrument on Envisat (Fig. 8) clearly show the importance of methane emissions
from the rice-growing regions of Asia (Schneising et al., 2009).
One of the first ESA–China Dragon projects was ‘Rice monitoring in China’
in 2003. China is the world’s largest rice producer, accounting for 32–35%
of global production in 2000. This makes China the preferred country for

Figure 8. Seasonal column-averaged methane mole fraction from Envisat/SCIAMACHY in 2003. The highest atmospheric methane
concentrations were observed over Asian countries during the rice-growing season. (Schneising et al., 2009)

283
SP-1326

studies of rice production systems and their impact on global climate change.
The area harvested has declined since the mid-1970s, and at the same time,
the increased demand for rice has been met by raising yields and improved
farming practices. China has introduced novel rice irrigation techniques that
allow the paddy fields to dry out intermittently, reducing water consumption
without reducing yields. The shift from continuous flooding to midseason
drainage water management has resulted in a significant reduction in methane
emissions from China’s paddy fields (Huang et al., 2004). Figure 9 shows
the model calculation of methane emissions from rice fields based on recent
statistics and input data on irrigation systems and farming practices). A major
objective of the Dragon project has been to develop and improve remote sensing
methods to derive model input data to refine the model calculation results.
Methods using the Envisat ASAR Alternating Polarization Precision (APP)
and wide-swath mode to map rice fields, rice varieties and irrigation status
over a study region have been developed (Le Toan et al., 2005) to provide
inputs for methane and CO2 emission models. Figure 10 shows a map of rice
fields in China, where flooded rice and drained fields can be distinguished.
However, the increasing interest in the Envisat data (and increased conflict
data requests) in China has limited the availability of data to test the methods
for the entire rice-growing season in the regions under study.
To demonstrate large-scale rice mapping, studies were conducted in
Vietnam in 2007, where the demand for ASAR data was much less important.
Mapping rice over large areas (regional to continental scales) would require
the acquisition and processing of a daunting amount of high-resolution data.
Instead, the wide-swath sensors on Envisat (and the future Sentinel-1 system)
have opened the way to the adaptation of these methods to medium resolution
(50–100 m) mapping for regional statistical and environmental studies.
Figure 11 (left) shows a map of two rice seasons in the Mekong delta in Vietnam
in 2007 (Bouvet & Le Toan, 2011). The map can be compared with the salinity
map of the region (right), indicating a general correlation between salt-free soil
and areas under rice cultivation. However, local exceptions could be observed,
either by use of irrigation water, or by change of rice into aquaculture.
Studies are under way to gain an understanding of the overall environment
of the Mekong delta and the impacts of human activities and climate change.
Sentinel-1 will be a major source of data for such studies.

Figure 9. Distribution of methane

emissions from rice fields in China, based
on the CH4MOD model at the Institute of
Atmospheric Physics, Chinese Academy of
Sciences, Beijing; averages for the period
2001−5. (Huang Yao)

284
 SAR Image Mode

Figure 10. Map of rice fields in Jiangsu

province, China, using Envisat ASAR.
Magenta: flooded rice fields; green: rice
fields with mid-season drainage. (A. Bouvet)

Figure 11. The Mekong delta, Vietnam. Left: Rice map (one and two crops a year in 2007) derived from Envisat ASAR data.
(Bouvet & Le Toan, 2011) Right: Soil salinity map of the same period. White, no salinity; magenta, through blue, green to red: decreasing
salinity. (Southern Institute of Water Resources Research, Vietnam, 2009)

285
SP-1326

3.4 Hazard Mapping

The demand for hazard mapping has increased due to the intensification of
natural disasters that affect human lives and infrastructure, and the increasing
use of risk-prone areas by growing populations. The interest is mainly in
assessments of both the extent and the level of damage, which is important for
the planning of protection measures. Data acquired during or after such events
can be combined with archive data to produce maps showing the changes. SAR
is particularly useful because of its capability to detect changes regardless of
the weather during the day and at night.
ERS‑1 and ERS‑2 have frequently been used for mapping storm damage to
forests, for which tandem InSAR data have proved effective in detecting abrupt
changes characterised by decreased coherence. For example, in the three
months before the end of ERS‑1, the two satellites gathered a large volume
of tandem InSAR data that was used (e.g. by Gamma Remote Sensing and
Sarmap) for mapping the storm damage to European forests in December 1999.
With ERS‑1 and ERS‑2, major advances were made towards the operational
application of SAR data for flood mapping and monitoring. One of the first results
was the mapping of 400 km of flooded banks of the River Thames in the UK in
December 1992, which highlighted the capability of ERS data to provide valuable
information (Blyth et al., 1993). Since then, flood products have been generated
for various flood events around the world (e.g. Badji & Dautrebande, 1997; Yésou
et al., 2001). During the dramatic flooding of the River Oder in central Europe in
summer 1997, the capability of acquiring data over an area of more than 110 000
km2 was demonstrated (De Roo et al., 1999). In addition to the exploitation of
multitemporal ERS or Envisat data, interferometric coherence has also been used
to derive flood maps, such as those produced by Gamma Remote Sensing during
the River Adour floods in southwest France in 2009, based on ERS‑2/Envisat
tandem observations. One of the last uses of ERS for flood mapping in near-
realtime was during the 2010 spring floods in Poland (Tholey et al., 2011).
In 2000, ESA and CNES, the French space agency, initiated the International
Charter ‘Space and Major Disasters’ to make it easier for emergency services
to access satellite data in the event of natural or man-made disasters. By 2012,
the number of Charter members had risen to 21. Between 2000 and 2010, the
charter was invoked about 300 times, nearly half (46.5%) of them for flood
events, 17% for windstorms in forests and 12% for earthquakes. The first
mapping products were delivered under the Charter in 2001, in response to the
flooding in the Saône River valley in France. Figure 12 shows images produced

Figure 12. The first image products

delivered to emergency services under the
International Charter on Space and Major
Disasters, following the floods in the Saône
River valley in France, 27 March 2001.
Left: ERS radar map of the flooded area;
centre: map of flooded areas obtained by
co-registering the ERS image with a SPOT
image; right: land affected by floodwaters.
(SERTIT)

286
 SAR Image Mode

by SERTIT, at the University of Strasbourg (H. Yésou), that were quickly

delivered to the emergency services to enable them to assess of the extent of
the flooding and plan their response activities. The images were also used in
the development of plans to prevent such floods in the future.
The use of SAR data for hazard mapping, starting with ERS, has
undoubtedly had a significant impact on the public. With Sentinel-1, mapping
of windstorm damage in forests and monitoring of the extent of floods are
expected to be fully operational.

4. Conclusions
ERS‑1, launched in 1991, carried an imaging SAR and other instruments to
measure ocean surface temperatures and winds at sea. ERS‑2 was launched in
1995 with an additional sensor for atmospheric ozone research. In March 2000,
ERS‑1 finally ended its operations, after far exceeding its planned lifetime, and
in July 2011 ERS‑2 was retired.
At their time of launch the two ERS satellites were the most sophisticated
Earth observation spacecraft ever developed by Europe. They collected a
wealth of valuable data on Earth’s oceans, polar ice caps and land surfaces.
The initial objective of land surface observations was to map cloud-covered
regions. But most of the achievements have been based on the capability of the
data to monitor changes in land surfaces used either for agriculture or forestry,
and to monitor natural disasters such as severe flooding or earthquakes in
remote parts of the world.
Twenty years after the launch of ERS‑1, followed by ERS‑2 and Envisat, we
have learned a great deal about how SARs see land surfaces, and how to make
best use of the data for applications and science. Some unexpected techniques
also emerged, such as the use of interferometry for land observations, as well
as outstanding applications such as rice monitoring, and use of the data for
flood monitoring operations. We have gained a better understanding of how
to use Earth observation for environmental monitoring and conservation. The
achievements of ERS have clearly paved the way for further investigations of
how future missions can benefit both science and applications. In the future,
moving from case studies to the global environment will require missions with
global coverage, systematic data acquisition and short repeat cycles, coupled
with the provision of long-term data. This will be achieved with the missions
such as Sentinel-1.

Acknowledgements
This chapter reflects many years of research during which I have had the chance
to work with and learn from many friends and colleagues. My very special thanks
go to the CESR-CESBIO members who worked with me on ERS SAR research
programmes: Henri Laur, Eric Mougin, RidhaTouzi, André Beaudoin, Florence
Ribbes, Nicolas Floury, Ghislain Picard, Jean- Michel Martinez, Francesco
Mattia, Jean Claude Souyris, Malcolm Davidson and Alexandre Bouvet.

References
Aschbacher, A., Pongsrihadulchai, S., Karnchanasutham, C., Rodprom, D., Paudyal,
R. & Le Toan, T. (1995). Assessment of ERS‑1 SAR data for rice crop mapping and
monitoring. IGARSS, Florence, Italy, pp.2183−2185.
Askne, J., Dammert, P.B.G., Ulander L.M.H. & Smith, G. (1997). C-band repeat-pass
interferometric SAR observations of forest. IEEE Trans. Geosci. Rem. Sens. 35(1):
25−35.

287
SP-1326

Askne, J., Santoro, M., Smith G. & Fransson, J. (2003). Multitemporal Repeat Pass
SAR interferometry of forests. IEEE Trans. Geosci. Rem. Sens. 41(7), 1540−1550.
Attema, E.P.W. & Ulaby, F.T. (1978). Vegetation modeled as a water cloud. Radio Sci.
13(2), 357−364.
Bach, H. & Mauser, W. (2003). Methods and examples for remote sensing data
assimilation in land surface process modeling. IEEE Trans. Geosci. Rem. Sens.
41(7), 1629−1637.
Badji, M. & Dautrebande, S. (1997). Characterisation of flood inundated areas and
delineation of poor drainage soil using ERS-1 SAR imagery. Hydrol. Processes 11,
1441−1450.
Balenzano, A., Mattia, F., Satalino, G. & Davidson, M. (2011). Dense temporal series
of C-and L-band SAR datafor soil moisture retrieval over agricultural crops. IEEE
Trans. Geosci. Rem. Sens. 4(2).
Beaudoin, A., Le Toan, T. & Gwyn, Q.H.J, (1990). Observations and modelling of the
C-band backscatter variability due to multiscale geometry and soil moisture.
IEEE Trans. Geosci. Rem. Sens. 28(5).
Borgeaud, M., Attema, E., Salgado-Gispert, G., Bellini, A. & Noll, J. (1995). Analysis
of bare soil surface roughness with ERS‑1 data. Proc. 1st Int. Symp. on Retrieval of
Bio- and Geophysical Parameters from SAR Data for Land Applications, Toulouse,
France.
Bouvet, A. & Le Toan, T. (2011). Use of ENVISAT/ASAR wide-swath data for timely
rice fields mapping in the Mekong river delta. Rem. Sens. Environ. 115(2011),
1090−1101.
Blyth, K., Biggin, D.S. & Raga, R. (1993). ERS‑1 SAR for monitoring soil moisture and
river flooding. Space at the Service of our Environment, Proc. 2nd ERS‑1 Symp.,
Hamburg. ESA SP-361, European Space Agency, Noordwijk, the Netherlands,
pp.839-844.
Davidson, M.W.J., Le Toan, T., Mattia, F., Satalino, G., Manninen, T. & Borgeaud, M.
(2000). On the characterization of agricultural soil roughness for radar remote
sensing studies. IEEE Trans. Geosci. Rem. Sens. 38(2), 630−640.
Davidson, M.W.J., Satalino, G., Verhoef, N., Le Toan, T., Borgeaud, M. & Louis, J.
(2003). Joint statistical properties of rms height and correlation length. IEEE
Trans. Geosci. Rem. Sens. 41(7), 1651−1658.
De Roo, A., van der Knijff, J., Horritt, M., Schmuck, G. & De Jong, S. (1999). Assessing
damage of the 1997 Oder flood and the 1995 Meuse flood. Proc. 2nd Int. Symp. on
Operationalization of Remote Sensing, Enschede, the Netherlands.
ESA (1998) ERS and its Applications. BR-128/11, Remote Sensing Exploitation
Division, ESRIN.
ESA (2008). Candidate Earth Explorer Core Mission − Biomass: Report for Assessment.
ESA SP-1313/2. European Space Agency, Noordwijk, the Netherlands.
http://esamultimedia.esa.int/docs/SP1313-2_BIOMASS.pdf
Floury, N., Souyris, J.C., Bruniquel, J. & Le Toan, T. (1997). A study of SAR
interferometry over forested areas: Theory and experiment. Proc. IGARSS 1997,
Singapore, pp.1868−1870.
Henry, J.B., Chastanet, P., Fellah, K. & Desnos, Y.L. (2006). Envisat multi-polarized
ASAR data for flood mapping. Int. J. Rem. Sens. 27(10), 1921−1929.
Huang, Y., Zhang, W., Zheng, X., Li, J. & Yu, Y. (2004). Modelling methane emission
from rice paddies with various agricultural practices. J. Geophys. Res. 109,
D08113.
Kurosu, T., Masaharu, F. & Chiba, K. (1995). Monitoring of rice crop growth from
space using the ERS‑1 C-band SAR. IEEE Trans. Geosci. Rem. Sens. 33, 1092−1096.
Laur, H. (1989). Analyse d’images radar en télédétection: Discriminateurs
radiométriques et texturaux. PhD thesis, Université P. Sabatier, Toulouse, France.
Le Toan, T. (1982). Active microwave signatures of soil and crops: Significant results
of three years of experiments. IGARSS 1982 Digest, vol. 1.
Le Toan, T., Askne, J., Beaudoin, A., Hallikainen, M., Quegan, S., Ulander, L. &
Wegmüller, U. (1998). EUFORA – European Forest Observations by Radars. FP4
Project, final report.

288
 SAR Image Mode

Le Toan, T., Beaudoin, A., Riom, J. & Guyon, D. (1992). Relating forest biomass to
SAR data. IEEE Trans. Geosci. Rem. Sens. 30, 403-411.
Le Toan, T., Bouvet, A., Tan Bingxiang, Li Zengyuan, He Wei, Li Bingbai, Zhang
Pingping & Bondeau, A. (2005). Rice Monitoring in China. ESA Dragon Project
Mid-term Report.
Le Toan, T. & Floury, N. (1998). On the retrieval of forest biomass from SAR data.
Proc. 2nd Int. Workshop on the Retrieval of Bio-Geophysical Parameters from SAR
Data for Land Applications. ESA SP-441, European Space Agency, Noordwijk, the
Netherlands, pp.595−600.
Le Toan, T., Laur, H., Mougin, E. & Lopes, A. (1989). Multitemporal and dual
polarisation observations of agricultural vegetation covers by X-band SAR
images. IEEE Trans. Geosci. Rem. Sens. 27(6), 709−717.
Le Toan, T., Quegan, S., Davidson, M., Balzter, H., Paillou, P., Papathanassiou, K.,
Plummer, S., Saatchi, S., Shugart, H. & Ulander, L. (2011). The BIOMASS mission:
Mapping global forest biomass to better understand the terrestrial carbon cycle.
Rem. Sens. Environ. 115, 2850−2860.
Le Toan, T., Quegan, S., Woodward, I., Lomas, M., Delbart, N. & Picard, G. (2004).
Relating radar remote sensing of biomass to modelling of forest carbon budgets.
J. Climatic Change 67, 379−402.
Le Toan, T., Ribbes, F., Wang, L.-F., Floury, N., Ding, K.-H., Kong, J.A., Fujita, M.
& Kurosu, T. (1997). Rice crop mapping and monitoring using ERS‑1 data based
on experiment and modelling results. IEEE Trans. Geosci. Rem. Sens. 35, 41−56.
Le Toan, T., Smacchia, P. & Souyris, J.C. (1993). On the retrieval of soil moisture from
ERS‑1 SAR data. Proc. 2nd ERS‑1 Symp., Hamburg. ESA SP-361, European Space
Agency, Noordwijk, the Netherlands, pp.883−886.
Liew, S.C., Kam, S.-P., Tuong, T.-P., Chen, P., Minh, V.Q. & Lim, H. (1998). Application
of multitemporal ERS‑2 synthetic aperture radar in delineating rice cropping
systems in the Mekong River Delta, Vietnam. IEEE Trans. Geosci. Rem. Sens. 36,
1412-1420.
Massonnet, D. & Souyris, J.C. (2008). Imaging with Synthetic Aperture Radar. EPFL
Press, Lausanne, Switzerland.
Mattia, F. & Le Toan, T. (1999). Backscattering properties of multiscale rough
surfaces. J. Electromagn. Waves Applic. 13, 491−526.
Mattia, F., Le Toan, T., Picard, G., Posa, F., D’Alessio, A., Notarnicola, C., Gatti, A.M.,
Rinaldi, M., Satalino, G. & Pasquariello, G. (2003). Multitemporal C-band radar
measurements on wheat fields. IEEE Trans. Geosci. Remote Sens. 41, 1551–1560.
Mattia, F., Le Toan, T., Souyris, J.C., De Carolis, G., Floury, N., Posa, F. &
Pasquariello, G. (1997). The effect of surface roughness on multifrequency
polarimetric SAR data. IEEE Trans. Geosci. Rem. Sens. 35(4), 954−966.
Moreau, S. & Le Toan, T. (2003). Biomass quantification of Andean Wetland forages
using ERS SAR data. Rem. Sens. Environ. 84(4), 477−492.
Picard, G., Le Toan, T. & Mattia, F. (2003). Understanding C-band radar backscatter
from wheat canopy using a multiple scattering coherent model. IEEE Trans.
Geosci. Remote Sens. 41, 1583–1591.
Quegan, S., Le Toan, T., Yu, J.J., Ribbes, F. & Floury, N. (2000). Estimating temporate
forest area with multitemporal SAR data. IEEE Trans. Geosci. Rem. Sens. 38(2),
741−753.
Rémondière, S. & Laur, H. (1998). The use of ERS SAR data for land applications: an
overview of ESA ERS Announcement of Opportunity achievements. Proc. 2nd Int.
Workshop on the Retrieval of Bio-Geophysical Parameters from SAR Data for Land
applications. ESA SP-441, European Space Agency, Noordwijk, the Netherlands,
pp.27−33.
Ribbes, F. & Le Toan, T. (1999). Rice field mapping and monitoring with Radarsat
data. Int. J. Rem. Sens. 20, 745−765.
Rudant, J.P., de Hauteclocque, H. & Penicand, C. (1997). Potential of radar image
maps: The example of French Guiana. Bull. Société Française de Photogrammétrie
et de télédétection 148, 44−56.

289
SP-1326

Satalino, G., Mattia, F., Davidson, M., Le Toan, T., Pasquarello, G. & Borgeaud,
M. (2003). On current limits of soil moisture retrieval from ERS SAR data. IEEE
Trans. Geosci. Rem. Sens. 40(11), 2438−2447.
Schneising,  O., Buchwitz,  M., Burrows,  J.P., Bovensmann,  H., Bergamaschi,  P. &
Peters, W. (2009). Three years of greenhouse gas column-averaged dry air mole
fractions retrieved from satellite, Part 2: Methane. Atmos. Chem. Phys. 9(2),
443−465.
Strozzi, T., Dammert, P., Wegmüller, U., Martinez, J.M., Askne, J., Beaudoin, A. &
Hallikainen M. (2000). Land use mapping with ERS SAR interferometry. IEEE
Trans. Geosci. Rem. Sens. 48(2), 766−775.
Tholey, N., Yésou, H., Battiston, S., Clandillon, S., Allenbach, B., Belabbes, S.,
Uribe, C., Caspard, M., Montabord, M. & De Fraipont, P. (2011). Exploitation
of middle to very high resolution SAR data for emergency flood mapping and
monitoring. Geophys. Res. Abstr. 13, EGU2011-8625.
Ulaby, F.T., Moore, R.K. & Fung, A.K. (1986). Microwave Remote Sensing, Active and
Passive. Artech House, Dedham, MA.
Wagner, W., Luckman, A., Vietmeier, J., Tansey, K., Balzter, H., Schmullius, C.,
Davidson, M., Gaveau, D., Gluck, M., Le Toan, T., Quegan, S., Shvidenko, A.,
Wiesmann, A. & Yu., J. (2003). Large-scale mapping of boreal forest in Siberia
using ERS tandem coherence and JERS backscatter data. Rem. Sens. Environ.
85(2), 125−144.
Wang, L.F., Kong, J.A., Ding, K.-H., Le Toan, T., Ribbes-Baillarin, F. & Floury, N.
(2005). Electromagnetic scattering model for rice canopy based on Monte Carlo
simulation. Progr. Electromagn. Res. 153−171.
Yésou, H., Chastanet, P., de Fraipont, P., Dossmann, P., Stock, N. & Béquignon J.
(2001). Mapping floods in France. Backscatter 12(3), 23−26.
Wegmüller, U. & Werner, C.L. (1997). Retrieval of vegetation parameters with SAR
interferometry. IEEE Trans. Geosci. Rem. Sens. 35(1), 18−24.

290
→→ Annex

Acronyms and Abbreviations
ACE Altimetry Corrected Elevation
ADEOS Advanced Earth Observing Satellite (Japan)
AGU American Geophysical Union
AGW Atmospheric Gravity Wave
AirSAR Airborne Synthetic Aperture Radar
ALOS Advanced Land Observing Satellite (JAXA)
AMI Active Microwave Instrument
AMSR-E Advanced Microwave Scanning
Radiometer for EOS
AO Announcement of Opportunity
AOCS Attitude and Orbit Control Subsystem
AOD Aerosol Optical Depth
AMAS Advanced Millimetre Wave Atmospheric
Sounder
AMF Air Mass Factor
AMS American Meteorological Society
AMSR-E Advanced Microwave Scanning
Radiometer for EOS
APS Atmospheric Phase Screen
ARC Active Radar Calibration
ARC ATSR Reprocessing for Climate project
ARTS-IP Adaptive Real-time Strategies for Image
Processing (EU)
ASAP Austrian Space Application Programme
ASAR Advanced Synthetic Aperture Radar
ASCAT Advanced Scatterometer (MetOp)
ASCAT-SG ASCAT Second Generation
ASI Italian Space Agency
ATLAS Atmospheric Laboratory for Applications
and Science
ATMOS Atmospheric Trace Molecule
Spectroscopy project
ATSR Along-Track Scanning Radiometer
AU Accounting Unit (ESA)
AVHRR Advanced Very-High-Resolution
Radiometer
AWDP ASCAT Winds Data Processor
BAS Bathymetry Assessment System
BDDN Broadband Data Dissemination Network
BIRA Belgisch Instituut voor Ruimte-
Aëronomie / Belgian Institute for Space
Aeronomy
BRGM Bureau de Recherches Géologiques et
Minières / Bureau of Geological and
Mining Research (France)
BMFT Bundesministerium für Forschung und
Technologie / German Ministry of Science
and Technology
BUV Backscattered Ultraviolet
Cal/Val Calibration and Validation
CARABAS Coherent All RAdio BAnd Sensing
CATGAS Calibration Apparatus for Trace Gas
Absorption Spectroscopy
Acronyms and Abbreviations
CCRS Canada Centre for Remote Sensing
CCT Computer-Compatible Tape
CCU Central Control Unit
CDN Content Delivery Network
CEOS Committee on Earth Observation
Satellites
CERN Organisation Européenne pour la
Recherche Nucléaire / European
Organization for Nuclear Research
CERSAT Centre ERS d’Archivage et de Traitement
/ ERS Processing and Archiving Facility
(IFREMER, France)
CESBIO Centre d’Études Spatiales de la Biosphère
/ Centre for the Study of the Biosphere
from Space (France)
CESR Centre d'Etude Spatiale des
Rayonnements / Centre for the Study of
Radiation in Space (France)
CFCs Chlorofluorocarbons
CFRP Carbon-Fibre-Reinforced Polymer
CInSAR Cross-interferometry SAR
CLAES Cryogenic Limb Array Etalon
Spectrometer
CLS Collecte de Localisation Satellite (France)
CNES Centre National d’Etudes Spatiales
(France)
CNR/IREA National Research Council, Institute
for Electromagnetic Sensing of the
Environment (Italy)
COASTALT Coastal Altimetry Project (ESA)
COMSS Coastal Ocean Monitoring Satellite
System
CR Corner Reflector
CSTARS Center for Southeastern Tropical
Advanced Remote Sensing (Miami, USA)
DARA Deutschen Agentur für
Raumfahrtangelegenheiten /German
Aerospace Agency (now DLR)
DEM Digital Elevation Model
DEOS Delft Institute for Earth-oriented Space
Research
DGM Delft Gravity Model
DIA Discrete Interaction Approximation
DInSAR Differential Interferometric SAR
DLR Deutsches Zentrum für Luft- und
Raumfahrt / German Aerospace Center
DLS Dynamic Limb Sounder
DNSC Danish National Space Center
DOAS Differential Optical Absorption
Spectroscopy
DOD Differential optical density
DTU Technical University of Denmark
DUE Data User Element
293
SP-1326
DWD Deutsche Wetter Dienst / German
Weather Service
DWS Doppler Wind Sounder
EARSC European Association of Remote Sensing
Companies
EARSeL European Association of Remote Sensing
Laboratories
ECMWF European Centre for Medium-Range
Weather Forecasts
ECV Essential Climate Variable
EGM Earth Geopotential Model
EGSE Electrical Ground Support Equipment
ELDO European Launcher Development
Organisation
EMSA European Maritime Safety Agency
ENSO El Niño–Southern Oscillation
EO Earth Observation
EOAC Earth Observation Advisory Committee
EOPAG Operation Plan Advisory Group
EOS Earth Observing System (NASA)
EM Engineering Model
EMR‑1/-2 ERS Microwave Radiometer 1/2
EPS Eumetsat Polar System
ERBS Earth Radiation Budget Satellite (NASA)
ERM Exact Repeat Mission
ERS‑1/-2 European Remote Sensing satellites 1/2
ERSC ERS Consortium
ERTS Earth Resources Technology Satellite
ESA European Space Agency
ESAC European Space Astronomy Centre
(Madrid, Spain)
ESCAT ERS Wind Scatterometer
ESRIN European Space Research Institute
(Italy)
ESRO European Space Research Organisation
EU European Union
EUFORA European Forest Observations by Radar
project (EU)
Eumetsat European Organisation for the
Exploitation of Meteorological Satellites
EURECA European Retrievable Carrier (ESA)
FCDR Fundamental Climate Data Record
FM Flight Model
FSLP First Spacelab Flight Payload
FTS Fourier Transform Spectrometer
FWHM Full Width at Half Maximum
FZ Fracture Zone
GCOS Global Climate Observing System
GDR Geophysical Data Record
GEO Group on Earth Observation
GEOSS Global Earth Observation System of
Systems
GeoSCIA Geostationary Scanning Imaging
Absorption Spectrometer for Atmospheric
Cartography
294
GeoTROPE Geostationary Tropospheric Pollution
Explorer
GFO Geosat Follow-On
GFSC Goddard Space Flight Center (NASA)
GFZ Deutsches GeoForschungsZentrum −
Helmholtz Centre Potsdam / German
Research Centre for Geosciences
GHRSST Global Ocean Data Assimilation
Experiment (GODAE) High-Resolution
Sea Surface Temperature pilot project
GIA Glacial Isostatic Adjustment
GLOBE Global Land One-km Base Elevation
Project
GLOMAC Global Modelling of Atmospheric
Chemistry
GM Geodetic Mission
GMES Global Monitoring for Environment and
Security
GMF Geophysical Model Function
GMT Greenwich Mean Time
GODAE Global Ocean Data Assimilation
Experiment (WMO)
GOME Global Ozone Monitoring Experiment
GOMOS Global Ozone Monitoring by Occultation
of Stars
GOMSS Global Ocean Monitoring Satellite System
GPS Global Positioning System
GRGS Groupe de Recherches en Géodésie
Spatiale (France)
GSAG GOME Scientific Advisory Group
GSFC Goddard Space Flight Center (NASA)
HALOE Halogen Occultation Experiment
HCFCs Hydrochlorofluorocarbons
HCMM Heat Capacity Mapping Mission
HIRLAM High-Resolution Limited Area Model
HLOP High-Level Operation Plan
HRDLS High-Resolution Dynamic Limb Sounder
(NASA)
IB Inverted Barometer
IDHT Instrument Data Handling and
Transmission
IGN Institut Géographique National /
National Institute of Geographic and
Forest Information (France)
IFARS Institute for Applied Remote Sensing
(Germany)
IFREMER Institut Français de Recherche pour
l’Exploitation de la Mer / French Research
Institute for Exploitation of the Sea
ILAS Improved Limb Atmospheric
Spectrometer
IM Image mode
IMG Interferometric Monitor for Greenhouse
Gases

IMGA Istituto per lo Studio dello Metodologie
Geofisiche Ambientali / Institute for the
Study of Geophysical and Environmental
Methodologies (Italy)
IMR Imaging Microwave Radiometer
INGV Istituto Nazionale di Geofisica e
Vulcanologia (Italy)
InSAR Interferometric SAR
IOC Intergovernmental Oceanographic
Commission
IPCC Intergovernmental Panel on Climate
Change
IR Infrared
IREA Istituto per il Rilevamento
Elettromagnetico dell'Ambiente /
Institute for Electromagnetic Sensing of
the Environment (Italy)
IRESTE Institut de Recherché et d'Enseignement
Supérieur aux Techniques de
l’Electronique (France)
ISAMS Improved Stratospheric and mesospheric
Sounder
Ispra Institute for Environmental Protection
and Research (Italy)
ISW Internal Solitary Wave
IT Information Technology
IUP/UB Institute of Environmental Physics,
University of Bremen
JERS‑1 Japanese Earth Resources Satellite 1
JGM-3 Joint Gravity Model 3
JPL PO.DAAC Jet Propulsion Laboratory − Physical
Oceanography Distributed Active Archive
Center (USA)
JRC Joint Research Council (EC)
KNMI Koninklijk Nederlands Meteorologisch
Instituut / Royal Netherlands
Meteorological Institute
LASP Laboratory for Atmospheric and Space
Physics (Boulder, USA)
LASS Land Application Satellite System
LBR Low Bit Rate
LED Light-emitting diode
LEGOS Laboratoire d'Etudes en Géophysique et
Océanographie Spatiales (France)
LIMS Limb Infrared Monitor of the
Stratosphere
LRIR Limb Radiance Inversion Radiometer
LRR Laser Retroreflector
LRTAP Convention on Long-Range
Transboundary Air Pollution
LST Land Surface Temperature
MABL Marine Atmospheric Boundary Layer
MAC-Europe Multisensor Airborne Campaign Europe
(NASA/ESA)
Acronyms and Abbreviations
MAPS Measurement of Air Pollution from
Satellites
MARISS Maritime Security Service project (GMES)
MARSAIS Marine SAR Analysis and Interpretation
System
MARSEN Marine Remote Sensing Experiment
MASL Mean altitude above sea level
MERIS Medium-Resolution Imaging
Spectrometer (Envisat)
MIPAS Michelson Interferometer for Passive
Sounding
MLS Microwave Limb Sounder (Aura)
MODIS Moderate-Resolution Imaging
Spectroradiometer (NASA)
MOST Method of Splitting Tsunami model
(NOAA)
MoU Memorandum of Understanding
MPI Max Planck Institute (Germany)
MRSE Microwave Remote Sensing Experiment
MSLA Maps of Sea Level Anomalies (Aviso,
France)
MSS Mean Sea Surface
MSSL Mullard Space Science Laboratory (UK)
MWR Microwave Radiometer
NASA National Aeronautics and Space
Administration (USA)
NASDA National Space Development Agency
(now JAXA, Japan)
NERC Natural Environment Research Council
(UK)
NERSC Nansen Environmental and Remote
Sensing Center (Norway)
NIVR Nederlands Instituut voor
Vliegtuigontwikkeling en Ruimtevaart
/ Netherlands Institute for Aerospace
Programmes
NOAA National Oceanic and Atmospheric
Administration (USA)
NOC NWP Ocean Calibration
NRCS Normalised Radar Cross-Section
NRT Near-Realtime
NSCAT NASA Scatterometer
NWP Numerical Weather Prediction
NWP SAF Satellite Application Facility for
Numerical Weather Prediction (Eumetsat)
OCM Ocean Colour Monitor
ODS Ozone-depleting substance
OII Optical Imaging Instrument
OMI Ozone Monitoring Instrument
OSI SAF Ocean and Sea Ice Satellite Application
Facility
OSTIA Operational Sea Surface Temperature
and Sea Ice Analysis (UK Met Office)
OTD Optical Transient Detector (NASA)
OVOC Oxygenated volatile organic compound
295
SP-1326
PALSAR Phased Array type L-band Synthetic
Aperture Radar
PAN Peroxyacetyl nitrate
PAMIRASAT Passive Microwave Radiometer Satellite
PARSA Partition Rescale and Shift Algorithm
PB-EO Programme Board for Earth Observation
(ESA)
PB-RS Programme Board for Remote Sensing
(ESA)
PFC Polyfluorinated compound
PI Principal Investigator
PIPOR Programme for International Polar Ocean
Research
PISTACH Prototype Innovant de Système de
Traitement pour l’Altimétrie Côtière et
l’Hydrologie / Coastal Altimetry project
(CNES)
PLSO Post-Launch Support Office (ESTEC)
PMD Polarisation Measurement Device
POEM Polar Orbiting Earth Observation Mission
POLDER Polarisation and Directionality of the
Earth’s Radiance (France)
POLIMI Polytechnic University of Milan
Pol-InSAR Polarimetric Interferometric SAR
PRARE Precise Range and Range Rate Equipment
PS Permanent Scatterer
PSC Polar Stratospheric Cloud
PSI Persistent Scatterer Interferometry
PSIC4 PSI Codes Cross-Comparison and
Certification
PU Phase Unwrapping
QC Quality Control
QPSK Quadrature Phase Shift Keying
QWG Quality Working Group
RA Radar Altimeter
RADS Radar Altimetry Database System
RAL Rutherford Appleton Laboratory (UK)
REAPER Reprocessing of Altimeter Products for
ERS
RESPAG Remote Earth Sensing Programme
Advisory Group
RIVM Rijksinstituut voor Volksgezondheid
en Milieu / National Institute for
Public Health and the Environment
(the Netherlands)
rms Root mean square
ROSIS Reflective Optics System Imaging
Spectrometer
RSAHG Remote Sensing Ad Hoc Group
RSPP Remote Sensing Preparatory Programme
SAG Science Advisory Group
SAGE I Stratospheric Aerosol and Gas
Experiment I
SAM Stratospheric Aerosol Measurement
SAMS Stratospheric and Mesospheric Sounder
296
SAO Smithsonian Astrophysical Observatory
(USA)
SAR Synthetic Aperture Radar
SARLab Synthetic Aperture Radar on Spacelab
SARSat Synthetic Aperture Radar Satellite
SASS SeaSat-A Scatterometer System (NASA)
SBUV Solar Backscatter Ultraviolet experiment
(NOAA)
ScanSAR Scanning Synthetic Aperture Radar
SCD Slant Column Density
SCIAMACHY Scanning Imaging Absorption
Spectrometer for Atmospheric
Chartography (Envisat)
SCIAS Scanning Imaging Absorption
Spectrometer
SCR Selective Chopper Radiometer
SDP SeaWinds Data Processor
SERTIT Service Régional de Traitement d'Image
et de Télédétection, University of
Strasbourg (France)
SHOM Service Hydrographique et
Océanographique de la Marine /
Hydrographic and Oceanographic Service
(France)
SIBERIA SAR Imaging for Boreal Ecology and
Radar Interferometry Applications
project (EU/ESA/NASDA)
SIR Spaceborne Imaging Radar
SLC Single-Look Complex
SLSTR Sea and Land Surface Temperature
Radiometer
SMAP Soil Moisture Active Passive mission
(NASA)
SME Solar Mesosphere Explorer (NASA)
SMEs Small and medium enterprises
SMOS Soil Moisture and Ocean Salinity mission
(ESA)
SMMR Scanning Multichannel Microwave
Radiometer
SMR Sub-millimetre Microwave Radiometer
(Odin)
SOAZ Système d’Analyse par Observation
Zenithale
SODETEG Société d’Études Techniques et
d’Entreprises Générales
SPOT Système Pour l’Observation de la Terre
SRON Netherlands Institute for Space Research
SRTM Shuttle Radar Topography Mission
(NASA)
SSAG SCIAMACHY Science Advisory Group
SSHA Sea Surface Height Anomaly
SSI Solar Spectral Irradiance
SSM/I Special Sensor Microwave/Imager
SST Sea Surface Temperature
SSBUV Shuttle SBUV
STS Space Transportation System (NASA)
SWH Significant Wave Height
SWM SAR wave mode

SWOT Surface Water and Ocean Topography
satellite (NASA)
SZA Solar Zenith Angle
TanDEM-X TerraSAR-X add-on for Digital Elevation
Measurement (DLR, Germany)
TEC Total electron content
TIROS Television Infrared Observation Satellite
Program (NASA)
TNO-TPD Netherlands Organisation for Applied
Scientific Research, Institute of Applied
Physics
TMI TRMM Microwave Imager
Topex Ocean Topography Experiment (NASA/
CNES)
TOMS Total Ozone Mapping Spectrometer
(NASA)
TOPS Terrain Observation by Progressive Scans
TRE Tele-Rilevamento Europa (Italy)
TREES Tropical Ecosystem Environment
Observation by Satellites project (EU)
TRMM Tropical Rainfall Measuring Mission
(NASA/JAXA)
TROPOMI Tropospheric Monitoring Instrument
TT&C Telemetry, Tracking and Command
TWT Travelling Wave Tube
UARS Upper Atmosphere Research Satellite
(NASA)
Acronyms and Abbreviations
UB University of Bremen
UNDCP United Nations International Drug
Control Programme
UNECE United Nations Economic Commission
for Europe
UNEP United Nations Environment Programme
UNESCO United Nations Educational, Scientific
and Cultural Organization
UNFCC United Nations Framework Convention
on Climate Change
UNOOSA United Nations Office for Outer Space
Affairs
UTC Coordinated Universal Time
UV Ultraviolet
VLBI Very-Long-Baseline Interferometry
VC Virtual Constellation (CEOS)
VOC Volatile organic compound
VMR Volume mixing ratio
WAIS West Antarctic Ice Sheet
WAM Wave Atmospheric Model (ECMWF)
WSM Wide-Swath Mode
WM Wave Mode
WMO World Meteorological Organization
WVC Wind Vector Cell
297
ESA Member States

Austria
Belgium
Czech Republic
Denmark
Finland
France
Germany
Greece
Ireland
Italy
Luxembourg
Netherlands
Norway
Poland
Portugal
Romania
Spain
Sweden
Switzerland
United Kingdom

An ESA Communications Production

Copyright © 2013 European Space Agency