Textbook Hyperspectral Imaging Remote Sensing Physics Sensors and Algorithms 1St Edition Dimitris G Manolakis Ebook All Chapter PDF

Hyperspectral Imaging Remote Sensing
Physics Sensors and Algorithms 1st

Edition Dimitris G. Manolakis
Visit to download the full and correct content document:
https://textbookfull.com/product/hyperspectral-imaging-remote-sensing-physics-senso
rs-and-algorithms-1st-edition-dimitris-g-manolakis/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
Hyperspectral Remote Sensing: Fundamentals and

Practices 1st Edition Ruiliang Pu
https://textbookfull.com/product/hyperspectral-remote-sensing-
fundamentals-and-practices-1st-edition-ruiliang-pu/
Hyperspectral remote sensing of vegetation. Volume IV,

Advanced applications in remote sensing of agricultural
crops and natural vegetation Second Edition Huete
https://textbookfull.com/product/hyperspectral-remote-sensing-of-
vegetation-volume-iv-advanced-applications-in-remote-sensing-of-
agricultural-crops-and-natural-vegetation-second-edition-huete/
Polarization Remote Sensing Physics Lei Yan
https://textbookfull.com/product/polarization-remote-sensing-
physics-lei-yan/
Remote sensing of vegetation. Hyperspectral indices and

image classifications for agriculture and vegetation
Second Edition Huete
https://textbookfull.com/product/remote-sensing-of-vegetation-
hyperspectral-indices-and-image-classifications-for-agriculture-
and-vegetation-second-edition-huete/
Fuzzy Machine Learning Algorithms for Remote Sensing
Image Classification 1st Edition Anil Kumar
https://textbookfull.com/product/fuzzy-machine-learning-
algorithms-for-remote-sensing-image-classification-1st-edition-
anil-kumar/
Novel Sensors and Sensing First Edition Jackson
https://textbookfull.com/product/novel-sensors-and-sensing-first-
edition-jackson/
Image analysis, classification and change detection in

remote sensing: with algorithms for Python Fourth
Edition Canty
https://textbookfull.com/product/image-analysis-classification-
and-change-detection-in-remote-sensing-with-algorithms-for-
python-fourth-edition-canty/
LiDAR Remote Sensing and Applications 1st Edition

Pinliang Dong
https://textbookfull.com/product/lidar-remote-sensing-and-
applications-1st-edition-pinliang-dong/
LiDAR Remote Sensing and Applications 1st Edition

Pinliang Dong
https://textbookfull.com/product/lidar-remote-sensing-and-
applications-1st-edition-pinliang-dong-2/
Hyperspectral Imaging Remote Sensing
A practical and self-contained guide to the principles, techniques, models, and tools
of imaging spectroscopy. Bringing together material from essential physics and digital
signal processing, the book covers key topics such as sensor design and calibration,
atmospheric inversion and model techniques, and processing and exploitation algo-
rithms. Readers will learn how to apply the main algorithms to practical problems, how
to choose the best algorithm for a particular application, and how to process and inter-
pret hyperspectral imaging data. A wealth of additional materials accompany the book
online, including example projects and data for students, and problem solutions and
viewgraphs for instructors.
This is an essential text for senior undergraduate and graduate students looking
to learn the fundamentals of imaging spectroscopy, and an invaluable reference for
scientists and engineers working in the field.
Dimitris G. Manolakis is a Senior Member of Technical Staff at MIT Lincoln Laboratory

in Lexington, Massachusetts. He is the co-author of Applied Digital Signal Processing
(Cambridge University Press, 2011), and has taught at various institutions including
Northeastern University, Boston College, and Worcester Polytechnic Institute. He is an
IEEE Fellow, and in 2013 he received the IEEE Signal Processing Society Education
Award.
Ronald B. Lockwood is a Member of Technical Staff at MIT Lincoln Laboratory. He
previously worked at the Air Force Research Laboratory, where he developed imaging
spectrometers for both space-based and airborne applications. He has also developed
vicarious calibration techniques in collaboration with colleagues at the University of
Arizona and the NASA Goddard Space Flight Center.
Thomas W. Cooley is the Air Force Research Laboratory Senior Scientist for Space Sit-
uational Awareness, and has made significant contributions to the fields of atmospheric
compensation and spectral data analysis. He developed the ARTEMIS sensor program,
which was successfully launched in 2009, and he has published over 70 research papers.
“The authors have done a masterful job of integrating and presenting the diverse subjects that
form the foundation of the field of hyperspectral imaging and applications. This comprehensive
textbook will clearly become one of the standard references for all who wish to learn about both
fundamentals and advanced applications in this important field.”
Charles Bachmann, Rochester Institute of Technology
“An extraordinarily comprehensive treatment of hyperspectral remote sensing by three of the

field’s noted authorities. An indispensable reference for those new to the field and for the seasoned
professional.”
Ronald G. Resmini, George Mason University
Hyperspectral Imaging Remote
Sensing
Physics, Sensors, and Algorithms
DIMITRIS G. MANOLAKIS
Massachusetts Institute of Technology
Lincoln Laboratory
RONALD B. LOCKWOOD
Massachusetts Institute of Technology
Lincoln Laboratory
T H O M A S W. C O O L E Y
University Printing House, Cambridge CB2 8BS, United Kingdom
Cambridge University Press is part of the University of Cambridge.

It furthers the University’s mission by disseminating knowledge in the pursuit of
education, learning and research at the highest international levels of excellence.
www.cambridge.org
Information on this title: www.cambridge.org/9781107083660

c Cambridge University Press 2016
This publication is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2016
Printed in the United Kingdom by TJ International Ltd. Padstow Cornwall
A catalogue record for this publication is available from the British Library
Library of Congress Cataloging-in-Publication Data
Names: Manolakis, Dimitris G., author. | Lockwood, Ronald B., author. |
Cooley, Thomas W., author.
Title: Hyperspectral imaging remote sensing : physics, sensors, and
algorithms / Dimitris G. Manolakis (Massachusetts Institute of Technology,
Lincoln Laboratory), Ronald B. Lockwood (Massachusetts Institute of
Technology, Lincoln Laboratory), Thomas W. Cooley
Description: Cambridge, United Kingdom : Cambridge University Press, 2016.
Identifiers: LCCN 2016025909 | ISBN 9781107083660 (hardback) |
ISBN 1107083664 (hardback)
Subjects: LCSH: Multispectral imaging. | Remote-sensing images. | Imaging
systems. | Image processing–Digital techniques.
Classification: LCC TR267.733.M85 M36 2016 | DDC 771–dc23
LC record available at https://lccn.loc.gov/2016025909
ISBN 978-1-107-08366-0 Hardback
Additional resources for this publication at www.cambridge.org/9781107083660
Cambridge University Press has no responsibility for the persistence or accuracy
of URLs for external or third-party internet websites referred to in this publication,
and does not guarantee that any content on such websites is, or will remain,
accurate or appropriate.
To Anna
DGM
To Susan
RBL
To Claire, Kate, and Mary

TWC
Contents
Preface page xi
1 Introduction 1
1.1 Introduction 1
1.2 Infrared Sensing Phenomenology 3
1.3 Hyperspectral Imaging Sensors 8
1.4 Data Preprocessing 15
1.5 Data Exploitation Algorithms 19
1.6 Applications of Imaging Spectroscopy 25
1.7 History of Spectral Remote Sensing 28
1.8 Summary and Further Reading 33
1.9 Book Organization 33
2 The Remote Sensing Environment 36

2.1 Electromagnetic Radiation 36
2.2 Diffraction and Interference 41
2.3 Basic Radiometry 50
2.4 Radiation Sources 55
2.5 Quantum Mechanical Results 61
2.6 Spectral Line Shapes 76
2.7 Atmospheric Scattering Essentials 84
2.8 Optical Thickness 95
2.9 Properties of the Atmosphere 102
3 Spectral Properties of Materials 117

3.1 Introduction 117
3.2 Geometrical Description 118
3.3 Directional Emissivity 126
3.4 Volume Scattering of Materials 128
3.5 Elements of Mineral Spectroscopy 131
3.6 Organic Spectroscopy 144
3.7 Man-Made Materials 148
viii Contents
3.8 Long Wave Infrared Spectra 150

4 Imaging Spectrometers 154

4.1 Telescopes 155
4.2 Imaging Spectrometer Common Concepts 161
4.3 Dispersive Imaging Spectrometer Fundamentals 171
4.4 Dispersive Imaging Spectrometer Designs 195
4.5 Interference Imaging Spectrometer Fundamentals 209
4.6 Data Acquisition with Imaging Spectrometers 222
5 Imaging Spectrometer Characterization and Data Calibration 228

5.2 Application of the Measurement Equation 229
5.3 Spectral Characterization 231
5.4 Radiometric Characterization 238
5.5 Spatial Characterization 261
5.6 Advanced Calibration Techniques 262
5.7 Error Analysis 263
5.8 Radiometric Performance Modeling 269
5.9 Vicarious Calibration 280
6 Radiative Transfer and Atmospheric Compensation 295

6.1 Radiative Transfer 295
6.2 General Solution to the Radiative Transfer Equation 303
6.3 Modeling Tools of Radiative Transfer 312
6.4 Reflective Atmospheric Compensation 322
6.5 Estimating Model Parameters from Scene Data 330
6.6 Reflective Compensation Implementation 344
6.7 Atmospheric Compensation in the Thermal Infrared 351
7 Statistical Models for Spectral Data 360

7.1 Univariate Distributions – Variance 360
7.2 Bivariate Distributions – Covariance 363
7.3 Random Vectors – Covariance Matrix 367
7.4 Multivariate Distributions 371
7.5 Maximum Likelihood Parameter Estimation 383
7.6 Statistical Analysis of Hyperspectral Imaging Data 387
7.7 Gaussian Mixture Models 394
Contents ix
8 Linear Spectral Transformations 406

8.2 Implications of High-Dimensionality 408
8.3 Principal Components Analysis: Theory 411
8.4 Principal Components Analysis: Application 421
8.5 Diagonalizing Two Different Covariance Matrices 424
8.6 Maximum Noise Fraction (MNF) Transform 428
8.7 Canonical Correlation Analysis (CCA) 429
8.8 Linear Discriminant Analysis 432
8.9 Linear Spectral-Band Estimation 437
9 Spectral Mixture Analysis 443

9.1 Spectral Mixing 443
9.2 The Linear Mixing Model 446
9.3 Endmember Determination Techniques 451
9.4 Fill-Fraction Estimation Techniques 453
9.5 The Method of Least Squares Estimation 454
9.6 Least Squares Computations 463
9.7 Statistical Properties of Least Squares Estimators 466
9.8 Generalized Least Squares Estimation 468
9.9 Maximum Likelihood Estimation 470
9.10 Regularized Least Squares Problems 471
9.11 Consequences of Model Misspecification 475
9.12 Hypotheses Tests for Model Parameters 477
9.13 Model Selection Criteria 480
9.14 Variable Selection in Linear Signal Models 482
9.15 Linear Spectral Mixture Analysis in Practice 487
10 Signal Detection Theory 494

10.1 A Simple Decision-Making Problem 494
10.2 Elements of Statistical Hypotheses Testing 496
10.3 The General Gaussian Detection Problem 503
10.4 Gaussian Detectors in the Presence of Unknowns 511
10.5 Matched Filter and Maximization of Deflection 517
10.6 Performance Analysis of Matched Filter Detectors 522
10.7 Detectors for Signals in Subspace Clutter and Isotropic
Noise 533
10.8 Eigenvector Matched Filters 539
10.9 Robust Matched Filters 541
10.10 Adaptive Matched Filter Detectors 547
x Contents
11 Hyperspectral Data Exploitation 551

11.1 Target Detection in the Reflective Infrared 551
11.2 Target Detection Performance Assessment 570
11.3 False Alarm Mitigation and Target Identification 577
11.4 Spectral Landscape Classification 581
11.5 Change Detection 586
11.6 Unique Aspects of Spectral Exploitation in the Thermal Infrared 591
11.7 Remote Sensing of Chemical Clouds: Physics 595
11.8 Remote Sensing of Chemical Clouds: Algorithms 605
Appendix Introduction to Gaussian Optics 621

A.1 The Optical Path 621
A.2 Ideal Image Formation 624
A.3 The Paraxial Approximation 628
A.4 The Limiting Aperture 633
A.5 Example: The Cooke Triplet 640
A.6 Afocal Systems 642
A.7 Aberration Theory 644
A.8 Summary and Further Reading 652
Bibliography 654
Index 678
Color plate section between pages 338 and 339.

Preface
T Q̃ − ’š•©Kš’− £ J̃ J© ¡Kš—’ ©˜’š’š š›¡’,

K £Jo •J© ›’¡ oJ ”œ¤›¤.
K
The roots of education are bitter but the fruit is sweet.
Isocrates, 436-338 BC
Greek orator and rhetorician
Sir Isaac Newton invented the first monochromator with two slits and a prism in 1666,
and is largely credited with founding the field of spectroscopy based on his publica-
tion of Opticks in 1704. Over a century after Newton’s optics experiments, Joseph von
Fraunhofer made the significant advance in the field with the use of a diffraction grating
to replace the prism as the dispersion element. His ground-breaking studies of emissions
from the sun, in which he cataloged the series of absorption features in the solar spec-
trum that bear his name, established the discipline of spectral remote sensing. An early
demonstration of the power of spectroscopy for identification was when Joseph Fou-
cault showed that two of the Fraunhofer lines corresponded to sodium emission. The
precision afforded by the diffraction grating facilitated the field of spectroscopy becom-
ing a mainstay for the fields of chemistry, astronomy and, in recent decades, remote
sensing. Imaging spectroscopy is a relatively recent discipline which has been made
possible by progress in several technology areas including detectors, optical design,
optical components, atmospheric radiative transfer, and computer processing, to name
just a few. While spectroscopy and imaging techniques have advanced independently,
imaging spectroscopy is the merger of both, using spectroscopic techniques to form a
multi-band-imaged scene or “hyper-spectral” image.
The hallmark of imaging spectroscopy is the utilization of spectroscopic techniques in
which measurements can be used for unambiguous direct and indirect identification of
surface materials and atmospheric trace gases. The measurement of the relative concen-
trations of materials and the ability to distinguish between subtle absorption features of
differing materials for a two-dimensional scene are examples of imaging spectroscopy.
The main objective of this book is to present the fundamentals of remote sensing
imaging spectroscopy from the perspective of the user. There is a strong interconnection
between foundational phenomenology, sensing technology, atmospheric propagation
effects, and exploitation algorithms. Successful application of remote sensing imag-
ing spectroscopy requires a working knowledge and fundamental understanding of all
xii Preface
these aspects. Providing the reader this foundation is the objective of this book. Imag-
ing spectroscopy is also used by meteorologists and atmospheric scientists to directly
characterize the properties of the atmosphere. Our focus is on the use of imaging
spectrometer data to identify attributes of the surface.
The book covers material from different disciplines ranging from physics, optics,
spectroscopy, linear algebra, statistics, and signal processing. Integrating material from
such disparate areas is not easy, but we have tried to provide smooth and coherent cover-
age. The reader is expected to have a basic background in physics, chemistry, electrical
engineering, or a related field. We attempt to proceed by introducing concepts from first
principles to provide both foundational ideas and consistent nomenclature throughout
the text.
We have tried to provide balanced coverage of topics of general interest. However, the
selection and emphasis has been, we hope only slightly, biased by our own experience
and backgrounds. We have selected material and references to reflect fundamental work
and provide useful and practical techniques, insights, and applications. We apologize
for any omissions or oversights of areas which may be sought by the reader, but the
foundations found throughout the book can be applied to a variety of applications. We
provide seminal bibliographic citations to provide historical context for the evolution
of the technology both from our direct experience in the field and through judicious
examination of the literature.
Special thanks go to Eric Truslow and Mike Piper for their cautious and thorough
review, helping with figures, and preparing the solutions manual. Michael Chrisp, Steve
Miller, and Kurt Thome lent their considerable expertise in reviewing several chapters.
Their contributions are gratefully acknowledged.
Also, RLB was privileged to study atomic physics under the guidance of Chun C.
Lin and in close collaboration with L. Wilmer Anderson and James Lawler at the Uni-
versity of Wisconsin at Madison. The presentation in Chapter 4 is extensively based
on the course in applied optics that was part of the curriculum there in the late 1980s,
co-developed by Edward Miller and Frederick Roesler. In particular, the instrumental
spectroscopy formalism here parallels Roesler’s notes and has been adapted for the mod-
est performance of an imaging spectrometer in comparison to the instruments used in,
for example, atomic and molecular physics or astrophysics. Fred Roesler collaborated
with Pierre Jacquinot early in his career, and his thinking, and therefore this presentation,
was heavily influenced by the French spectroscopy school.
RBL would also like to thank his parents, Robert and Sharon Lockwood. Both are
educators and farmers who instilled a keen interest in the natural world in each of their
children, leading to careers in science and mathematics. Having raised us on a family
farm in West Texas, they also ensured that we all have healthy work ethics! This book
is a result of their support and encouragement.
Most importantly, we must thank our families for their patience and enduring support
for this project. No undertaking of this magnitude and duration is accomplished without
a foundation of support to see it through. Anna, Susan, and Mary know the effort and
time invested in this undertaking, and each is a pillar of forbearance for the time we have
taken to complete this text. We are grateful for their wisdom, support, and endurance.
Preface xiii
This undertaking required significant encouragement from our families, friends, and
colleagues who recognized the opportunity to provide this compilation for the broader
community of scientists, researchers, and students entering the field. We are grateful for
the many voices of encouragement throughout this project. We wish also to thank the
team at CUP, including Phil Meyler and Heather Brolly, who have patiently awaited the
completed manuscript and provided valuable comments and guidance throughout. We
hope you find benefit from our enterprise.
Acknowledgments
We have been fortunate to build upon the work of many other researchers, and to have
access and help from many colleagues leading up to and through this project. We would
like to thank the many scientists and researchers who have contributed directly and
indirectly to this undertaking, and specifically want to thank: Gail Anderson, Peter Arm-
strong, Gregory Berthiaume, Stuart F. Biggar, William Blumberg, Rebecca N. Breiding,
Michael Brueggeman, Thomas G. Chrien, James Dodd, David Driscoll, Richard Dunn,
Janet Fender, James Gardner, Mary Ann Glennon, Steven Golowich, Peggy Grigsby,
Lynn Hatfield, David Hardy, John Jacobson, Angela Kim, Steven Lipson, Joaquin
Mason, Karis W. Miles, Dan Mooney, Richard Nadile, Sidi Niu, Martin Pilati, Ryan
M. Pospisal, C. Richard Quade, Frederick L. Roesler, Stephanie Schieffer, John Schum-
mers, Vittala Shettigara, Krista Steenbergen, Stanley Straight, Johnny L. Touchstone,
Klaes Wandland, Christopher J. Ward, Andrew Weisner, Benjamin Worker, and Randall
Zywicki.
1 Introduction
This chapter provides an introduction to the basic principles of hyperspectral remote

sensing. The main objective is to explain how information about the earth’s sur-
face is conveyed to a remote hyperspectral imaging sensor, which are the key factors
determining the nature and quality of the acquired data, and how the data should be
processed to extract meaningful information for practical applications. By definition,
hyperspectral imaging systems collect co-aligned images in many relatively narrow
bands throughout the ultraviolet, visible, and infrared regions of the electromagnetic
spectrum.
1.1 Introduction
The term “remote sensing” has several valid definitions. In the broadest sense, accord-
ing to Webster’s dictionary, remote sensing is “the acquisition of information about a
distant object without coming into physical contact with it.” For our purposes, remote
sensing deals with the acquisition, processing, and interpretation of images, and related
data, obtained from aircraft and satellites that record the interaction between matter and
electromagnetic radiation.
The detection of electromagnetic radiation via remote sensing has four broad compo-
nents: a source of radiation, interaction with the atmosphere, interaction with the earth’s
surface, and a sensor (see Figure 1.1). The link between the components of the system
is electromagnetic energy transferred by means of radiation.
Source The source of electromagnetic radiation may be natural, like the sun’s
reflected light or the earth’s emitted heat, or man-made, like microwave radar. This leads
to a classification of remote sensing systems into active and passive types. Active sys-
tems emit radiation and analyze the returned signal. Passive systems detect naturally
occurring radiation either emitted by the sun or thermal radiation emitted by all objects
with temperatures above absolute zero. With active systems, like microwave radar, it is
possible to determine the distance of a target from the sensor (range); passive systems
cannot provide range information.
Atmospheric interaction The characteristics of the electromagnetic radiation propa-
gating through the atmosphere are modified by various processes, including absorption
and scattering. This distortion is undesirable and requires correction if we wish to study
the earth’s surface, or desirable if we wish to study the atmosphere itself.
2 Introduction
Figure 1.1 Pictorial illustration of how the atmosphere, the solar illumination, and the spectral response of
the sensor affect the relationship between the observed radiance spectrum and the wanted
reflectance spectrum of the ground resolution cell. (A black and white version of this figure will
appear in some formats. For the color version, please refer to the plate section.)
Earth’s surface interaction The amount and spectral distribution of radiation

emitted or reflected by the earth’s surface depend upon the characteristics of the
involved “target” materials. The interaction between matter and electromagnetic energy
is determined by (a) the physical properties of the matter, and (b) the wavelength of
electromagnetic energy that is remotely sensed.
Sensor The electromagnetic radiation, which has interacted with the surface of the
earth and the atmosphere and has undergone some form of spectral discrimination as it is
transmitted through a spectral detector, is recorded by an electro-optical detector which
converts electromagnetic radiation into an electrical signal directly related to the radia-
tion in a particular spectral band from the scanned scene. The electrical signal is ampli-
fied, converted to digital data, and organized into a data structure for further processing.
Once the data have been collected by the sensor, they must be analyzed in real time
or offline. However, before we start talking about analysis techniques, we should under-
stand what information about the target material is conveyed by the reflected or emitted
electromagnetic radiation and how this information has been modified by the sensor.
The radiation received by the sensor can be measured at different locations of a
scene, over a range of wavelengths, and sometimes at different times (see Figure 1.2).
This leads to rich data sets which can be organized and explored in many different
ways. We consider electro-optical systems that include the visible, near infrared, and
thermal infrared regions of the electromagnetic spectrum. We are interested in remote
sensing imaging systems designed to form a two-dimensional representation of the
two-dimensional distribution of radiant energy across the target at one or multiple
spectral bands. Traditionally, data from a single wavelength band are used to form
a digital image, where each pixel represents the radiant energy of a corresponding
Figure 1.2 Principle of hyperspectral imaging sensing. The resulting “data cube” can be viewed as a set of
co-registered images at multiple wavelength regions or as a set of spectra corresponding to each
pixel of the image. (A black and white version of this figure will appear in some formats. For the
color version, please refer to the plate section.)
ground element area. Each individual image can then be processed and analyzed using
established image processing techniques that exploit geometrical or shape informa-
tion. However, if we collect radiation measurements at different wavelength bands for
the same spatial sample, and arrange them in a vector, we can analyze the resulting
“spectral” data using multivariate statistical techniques.
In the remainder of this chapter, we discuss the basic principles underlying each
part of the remote sensing system, we outline the key ideas of data preprocessing and
exploitation techniques, and we summarize some common applications.
1.2 Infrared Sensing Phenomenology
The main sources of radiation in passive remote sensing is the electromagnetic radiation
emitted by the sun and the self-emission of objects in the scene due to their temperature.
Electromagnetic radiation is the means by which electromagnetic energy is propagated
in the form of waves.
Electromagnetic waves are characterized by their location within the electromagnetic
spectrum. The division of the electromagnetic spectrum into different regions, such as
visible, infrared, or microwave, has more to do with the different methods used for
sensing it rather than the nature of the radiation itself. Figure 1.3 shows the regions used
in electro-optical remote sensing and the typical applications for each region.
1.2.1 Sources of Infrared Radiation

All objects with temperatures greater than absolute zero emit radiation whose amount
changes as a function of wavelength. An object that absorbs all incident energy, converts
4 Introduction
Figure 1.3 Typical examples of spectral information present at different spectral regions of the
electromagnetic spectrum for different applications.
it to heat energy, and transforms the heat back into radiant energy at the maximum
possible rate allowed by thermodynamic laws, is a perfect thermal emitter known as a
blackbody. The spectral radiant emittance of a blackbody is given by Planck’s law.
For our present purpose, the spectral radiant exitance, the power per unit area emitted
by the sun, as seen from above the earth’s atmosphere, can be approximated by a black-
body curve at a temperature of 5800 K. Thus the sun, with a temperature of 5800 K,
will have a maximum emittance at a wavelength of 0.50 m. In contrast, the earth’s
ambient temperature on a warm day, due largely to heating by the sun, is about 300 K
or 27 ı C and the maximum spectral radiant emittance from earth’s features occurs at
a wavelength of about 9.7 m. This radiation is known as “thermal infrared” energy
because it is related to terrestrial heat.
In the visible and near infrared regions of the electromagnetic spectrum we study
the radiation from the sun as it is modified by the earth’s atmosphere and surface. In
the thermal infrared region we study the radiation emitted by the earth’s atmosphere and
surface. The dividing line between reflective and emissive wavelengths is approximately
4 m. Reflective energy predominates below this wavelength and emissive energy
predominates above it. This difference in the nature of radiation also has profound
effects on the sensors used in the reflective and emissive regions of the electromagnetic
spectrum.
Figure 1.4 shows the contrast between the solar spectral irradiance and the spectral
irradiance from a blackbody at 300 K. Irradiance will be discussed in more detail in
Chapter 2, but it describes the spectral energy per second that flows through a real or
104 1
Irradiance (W/m2/μm)
0.8
Transmission
102
0.6
0.4
100
0.2
10−2 0
0.2 1 10 25
Wavelength (μm)
Figure 1.4 Irradiance and atmospheric absorption for the wavelength range 0 to 25 m. The black curve is
the exoatmospheric spectral irradiance and the dashed curve is the spectral irradiance from a
blackbody at a temperature of 300 K with both referenced to the left vertical scale. The gray
curve is the atmospheric transmission referenced to the right vertical scale. Note the presence of
atmospheric windows in the thermal wavelength regions 3 to 5 m and 8 to 14 m.
imaginary aperture of unit area. The two curves cross at slightly longer than the 4 m
dividing line defined above.
1.2.2 Atmospheric Propagation

All radiation observed by remote sensing systems passes through some distance or path
length of atmosphere, which has a profound effect on its radiant energy and spectral
composition. The main mechanisms causing these effects are atmospheric scattering and
absorption. Atmospheric scattering is the unpredictable diffusion of radiation by parti-
cles in the atmosphere. Unlike scattering, atmospheric absorption results in the effective
loss of energy from the radiation field as it is converted to other forms.
The transmission curve, illustrated in Figure 1.4, also defines the spectral regions
where remote sensing can occur. Aerosol scattering and molecular absorption limit the
transmission range at the shortest wavelengths. The spectral region from about 0.35 to
2.5 m defines the reflective range that is dominated by solar illumination and where
the atmosphere is broadly transmissive. The large absorption features throughout the
0.2 to 25 m range, where light transmission is severely attenuated or completely
absorbed, are largely due to absorption by water although minor constituents such as
carbon dioxide play an import role as well.
The regions of the electromagnetic spectrum in which atmospheric absorption is low
are called atmospheric windows and it is through these “windows” that remote sensing
of the earth’s surface takes place. We note that the peak of the solar spectrum coincides
with the visible atmospheric window at 0.4–0.9 m, whereas the peak of emitted “heat”
from the earth extends through the atmospheric windows at 3–5 m and 8–14 m.
The important point to note from Figure 1.4 is that remote sensing of radiation is not
possible if there is not sufficient energy from the radiation source within the atmospheric
windows.
6 Introduction
1.2.3 Reflectance and Emissivity Spectra

When electromagnetic radiation interacts with earth’s surface, various fractions of the
energy are reflected, absorbed, and/or transmitted. In general, these fractions vary as
a function of wavelength depending on the type and condition of the surface material.
Therefore, it is important to describe those variations and investigate whether they can
be used to characterize different materials.
In the reflective part of the spectrum, where solar illumination predominates, the
reflectance properties of the earth’s features are used to describe the surface character-
istics. Depending on the surface roughness, there are two general classes of reflectors.
Specular reflectors are flat surfaces that act like mirrors; that is, the angle of reflection
equals the angle of incidence. In contrast, diffuse or Lambertian reflectors are rough
surfaces that reflect uniformly in all directions. The behavior of real surfaces, which
is between these two cases, is dictated by the surface’s roughness in relation to the
wavelength of the incident electromagnetic radiation.
In remote sensing, we are primarily interested in diffuse reflectance because this is
the dominant type of reflectance for most materials, although there are exceptions such
as water. The reflectance characteristics of an earth surface material may be quantified
by its reflectance spectrum, which measures the percentage of incident radiation, typ-
ically sunlight, that is reflected by a material as a function of the wavelength of the
radiation. Reflectance spectra are measured by special instruments called reflectometers
in the laboratory or the field. Figure 1.5 shows typical reflectance spectra of green veg-
etation, dry vegetation, and soil. Dips of the spectral curve represent absorption of the
incident radiation and are known as absorption features. In contrast, upward excursions
Visible Near Infrared Middle Infrared

Chlorophyll Cell structure Water Water
Grass Walnut tree canopy

Reflectance (%)
60
Fir tree
40
Dry, yellowed
20 grass
0
0.4 0.8 1.2 1.6 2.0 2.4
Wavelength (micrometers)
Figure 1.5 Typical spectral reflectance curves for different types of green and yellowed (dry) vegetation.
The spectral curves for green vegetation at different regions are shaped by the components
shown at the top.
1
Grass
Sandy Loam Soil
0.9
Emissivity
0.8
Granite
0.7
7 8 9 10 11 12 13
Wavelength ( m)
Figure 1.6 Typical emissivity spectra of grass, granite, and sandy loam soil materials in the thermal infrared
window from 7 to 13 m.
are called reflectance peaks. These spectral features provide valuable information for
the identification of different materials.
Real materials are not perfect blackbodies, but instead emit radiation in accor-
dance with their own characteristics. The ability of a material to emit radiation can
be expressed as the ratio of its radiant emittance at a given temperature to the radiant
emittance of a blackbody at the same temperature. This ratio is known as the spectral
emissivity of the material. The plot of emissivity as a function of wavelength is called
the emissivity spectrum of a material. Figure 1.6 shows examples of emissivity spectra
with unique spectral features; the data were obtained from the ASTER spectral library
described by Baldridge et al. (2009a).
Radiance in the thermal infrared, the mid-wave infrared and the long-wave infrared
(MWIR and LWIR) is determined by both the emittance and temperature of the
materials. Variations in temperature can cause both large radiance variations and
large spectral variations at the input of a remote sensor. Furthermore, it is not
possible to measure separately the spectral emissivity and temperature using hyper-
spectral sensing alone. Emissivities of most materials vary between 0.8 and 0.95
in the 8 to 14 m thermal IR window. Thermal emission is generally the dom-
inant term; the variations caused by variations in spectral emittance are small by
comparison. This causes additional variability in the statistics of target and back-
ground clutter, which makes detection and discrimination in the thermal infrared more
challenging.
In conclusion, the signal of interest in spectral remote sensing applications, that is,
the information-bearing signal, is the reflectance or emissivity spectrum of the imaged
material. The details and relationships between spectral reflectance, emittance, and
temperature will be fully developed in subsequent chapters.
8 Introduction
1.3 Hyperspectral Imaging Sensors
Hyperspectral sensors, more properly known as imaging spectrometers, col-

lect simultaneously digital images in many relatively narrow, contiguous and/or
noncontiguous spectral bands of the ultraviolet, visible, and infrared regions of the elec-
tromagnetic spectrum. The collection of spectral reflectance and emittance information
in hundreds of spectral bands for each spatial location requires the design of special
sensors. Hyperspectral imaging sensors consist of a scanning mechanism, an imaging
system, and a spectrometer. Figure 1.7 illustrates the basic elements of a hyperspec-
tral data collection sensor. All imaging spectrometers require some sort of scanning
and pointing system in order to accumulate data and assign position coordinates. The
imaging system, or fore optic, collects and images radiant energy from a location on
the earth’s surface (spatial sampling) with a certain spatial resolution. The image plane
of the fore optic becomes the optical input, or the object, for the spectrometer. Finally,
the spectrometer measures the radiant energy at a number of spectral bands (spectral
resolution) for each spatial sample with a certain accuracy (radiometric resolution).
From a signal processing perspective, to acquire hyperspectral imaging data requires
three sampling operations: spatial, spectral, and radiometric. Temporal sampling, which
refers to how often data are obtained for the same area, is related to operational rather
than signal processing considerations. Therefore, we concentrate on the physical imple-
mentation and the key parameters required for the characterization of spatial, spectral,
and radiometric sampling operations.
1.3.1 Spectral–Spatial Data Collection and Organization

It is helpful at this point to discuss how the data from an imaging spectrometer are
organized. Consider the data in the raw format of digital numbers as it is read from the
aa
a
l l a
l
ll
Figure 1.7 Basic principle of a hyperspectral remote sensor

individual detector elements. The data are stored in a data structure, known as a data
cube, consisting of two spatial dimensions and one spectral dimension. Each datum in
the cube is associated with a particular spatial location of a certain size and its value
is proportional to the spectral radiant energy contained within that particular spectral
sample and area.
The order and procedure by which each datum in the cube is populated are deter-
mined by the details of the imaging spectrometer. The challenge faced by the optical
designer is to record a three-dimensional data set using the signal generated by either a
two-dimensional array of detectors, a line array of detectors, or a single detector. Addi-
tionally, there is a limited number of ways that light can be separated into its spectral
components.
As described above, the fore optic presents an image of a scene location at the input of
a spectrometer. The spectrometer input is often a spatial mask of some sort. For example,
a slit spectrometer employs a two-dimensional opening that is long and narrow. One can
think of this imaging process in the reverse sense with the spectrometer input mask
being imaged onto the scene. The spatial dimensions of the cube are then assembled as
this projected mask is scanned across the scene of interest.
Figure 1.8 illustrates some of the different methods by which the data cube is popu-
lated depending upon the type of imaging spectrometer. The data subset marked with an
“A” is from a slit spectrometer, which relies on a dispersing element to spatially separate
the light into the different wavelength samples which are recorded by a two-dimensional
detector array. A scanning mechanism is utilized to move the projected slit image across
Figure 1.8 Three-dimensional illustration of a data cube showing the spatial dimension .x, y/ and the
spectral dimension . The numbers of spatial and spectral samples will depend upon the details
of the sensor design. The regions, labeled A, B, and C, illustrate the portion of the data cube that
is acquired during one acquisition cycle as described in the text.
10 Introduction
2-D FPA
Spectral
Focusing
optics
Collimating Sp
optics ati
al
al
ati
Optics Sp
Dispersing
element
Slit
Scan
(a)
line
Spatial
2-D FPA
Spatial
Focusing Sp
ati
optics al
FFT al
e ctr
Optics
Sp Sp
ati
al
FTIS OPD
(b)
Figure 1.9 Illustration of two types of imaging spectrometers. Illustration (a) is a slit spectrometer with each
data frame composed of spatial and spectral dimensions and the second spatial dimension
generated through scanning the projection of the slit onto the scene. Illustration (b) is of a FTIS,
with each data frame corresponding to both spatial dimensions at a particular OPD. The spectral
information is recovered by taking the fast Fourier transform (FFT) of the interference data for
each spatial location. (A black and white version of this figure will appear in some formats. For
the color version, please refer to the plate section.)
the scene. The spatial image recorded by a single data frame has m 1 samples with m
being the number of spatial samples along the long axis of the slit. The spectral infor-
mation is recorded in the perpendicular direction by n detector elements. The entire
data cube is assembled by scanning the slit along the scene and reading out the array
of detector elements at an appropriate rate. This is often referred to as pushbroom scan-
ning, with the forward motion of the airborne or space-based platform providing the
slit scanning motion. The diagram in Figure 1.9(a) sketches this common and highly
successful design.
A second example illustrated in Figure 1.8 and labeled as “B” is a staring system that
utilizes a series of spectral filters mounted in a filter wheel. A system of this type “stares”
at an area of the scene and cycles through a series of filters. A two-dimensional array
records an area of the scene at the various filter wavelengths and the cube is populated
as the filter wheel cycles. Once the spectral data corresponding to the spectral bandpass
from each filter in the wheel has been recorded, the scanning mechanism transitions to
the next portion of the scene and the process is repeated until the full scene has been
imaged. This is not a commonly employed imaging spectrometer, but it does introduce
the concept of a staring sensor.
A staring imaging spectrometer that is employed, particularly for remote sensing of
the atmosphere, is a Michelson interferometer, or Fourier Transform Imaging Spectrom-
eter (FTIS), which is illustrated in the sketch of Figure 1.9(b). This spectrometer utilizes
interference of light to record the spectral information. The interference pattern is intro-
duced by splitting the light into two unequal paths and recombining it. The light will
constructively and destructively interfere after recombining depending upon the optical
path difference (OPD) and the wavelength. A moving mirror produces the necessary
OPD and the interference pattern is recorded by repeatedly reading out the data from
each detector element in a two-dimensional array at a particular mirror position. Each
element of the two-dimensional array records the spectral information for a particular
spatial sample. In this case, the spectrum is recovered by taking the Fourier transform
of the interference signal for each spatial location. Once the full mirror travel has been
completed the spatial scanning mechanism transitions to the next portion of the scene
and another area corresponding to the projection of the two-dimensional detector array
is recorded, as was done for the filter wheel system. The details of this type of imaging
spectrometer will be addressed in Chapter 4; it is mentioned here as a practical staring
system for some applications.
The final example, labeled as “C”, is an imaging spectrometer that measures a com-
plete spectrum for a single spatial sample. A system of this type employs either a line
array, as in the case of a slit spectrometer, or a single detector element, if the imaging
spectrometer is a Michelson interferometer. The full data cube is assembled by raster-
scanning the single spatial element until the full scene area has been collected. This
is often referred to as whiskbroom scanning, since the forward motion of the platform
generates one axis of the scanning motion and a rotating mirror “whisks” the projected
spatial sample in the perpendicular direction. An example slit imaging spectrometer
that utilizes this scanning method is the widely used Airborne Visible/Infrared Imaging
Spectrometer (AVIRIS) system built at the Jet Propulsion Laboratory (Vane et al., 1993).
Armed with this background in the collection of a full hyperspectral data cube, we
will now introduce the concepts critical to our discussion of the details of the sensor
systems. The spatial, spectral, and radiometric figures of merit are defined. These ideas
are critical in order to fully understand the details of each spatial-spectral datum in the
image cube.
1.3.2 Spatial Sampling

Hyperspectral imaging data are acquired by sensors that can systematically scan parts of
the earth’s surface, recording radiation reflected or emitted from a scene location, known
as a ground resolution element, at multiple regions of the electromagnetic spectrum. A
key component of the sensor is an optical system which collects the radiation originating
12 Introduction
Figure 1.10 Illustration of the ground resolution element and the geometrical figures of merit for an imaging
spectrometer. The IFOV is alternatively defined as the ratio of the pixel width p to the focal
length f .
from the ground surface area and focuses the intercepted rays onto the surface of a
detector element, as in the filter wheel case, or onto a spatial mask such as a slit. For
those systems with a spatial mask, the area is further segmented depending upon the
number of spatial samples as was described above for a slit spectrometer. Figure 1.10
illustrates this concept for a single spatial element from either case. The area of the
detector or segment of the spatial mask projected through the system’s optics onto the
earth’s surface coincides with the ground resolution element. Each ground resolution
element is mapped onto one picture element or pixel of the image collected by the sensor
as the surface is properly scanned.
The minimum solid angle through which radiant energy originating from the scene is
accepted by the optical system and reaches a single detector element or mask segment is
called the sensor’s instantaneous field of view (IFOV). The IFOV is a strictly geometric
factor used as a rule of thumb to describe the smallest spatial resolution element that
is sampled by the imaging spectrometer. The true spatial resolution is determined by
an analysis of the optical aberrations and the effects of diffraction. The angular field of
view is the full angular extent of all the IFOVs. This is often restricted to the cross track
dimension, as in the case of a slit spectrometer that is pushbroom scanned. The cross
track direction corresponds to the projection of the long axis of the slit onto the scene.
In this case, the along track dimension is accumulated by scanning as described above
and is arbitrary in its extent.
The ground sample distance or GSD is the simple geometrical representation of a
detector element or mask segment projected onto the scene. It is calculated by mul-
tiplying the IFOV by the range, the distance from the sensor to the scene location.
However, the true spatial sampling is somewhat different from this idealization. As
mentioned above, the light from a given scene point is modified due to optical aber-
rations and diffraction as it is sampled by the imaging optics. This smearing effect is
described by a weighting function, known as the point spread function of the sensor.
The spatial resolution is further smeared by transmission through the atmosphere and
the scanning process. The resulting spatial resolution is known as the ground resolution
distance or GRD. The GSD defines the smallest area viewed by the sensor and estab-
lishes a lower limit for the level of spatial detail that can be represented in a digital
image.
Although scanning mechanisms vary according to sensor type, after geolocation pro-
cessing, each measurement corresponds to a location in the scene and is presented as
a pixel in an image. A sensible geometrical correspondence exists between scene loca-
tions and image pixels, such that the image can be considered as a projection of the
earth’s surface onto an image plane.
1.3.3 Spectral Sampling

An imaging spectrometer is designed to recover the spectral information inherent in
the electromagnetic radiation from a given ground sample. This recovery is by its
nature imperfect due to finite spectral sampling with the data binned into particular
spectral bands. The spectral response function (SRF) is the weighting function that
describes the range of wavelengths that are transmitted to a particular spectral sam-
ple. The width of the SRF, measured at half of the peak transmittance, is a measure
of the spectral resolution of the spectrometer. The narrower the width, the more spec-
tral details that can be resolved, such as the narrow absorption features inherent in
the transmission of light through the atmosphere. Naturally, the signal-to-noise ratio
is also proportional to the SRF, with less light transmitted for a narrow SRF. Figure 1.11
shows a comparison of the spectrum recovered by spectrometers with high and low
resolutions.
There are various ways that this spectral sampling can be achieved, but they fall into
two broad categories. The first relies upon the interaction of an electromagnetic wave
with matter and the second utilizes the wave nature of the light itself. The familiar prism,
where light of different colors is refracted at different angles due to the wavelength
dependent index of refraction of the prism material, is an example of the first cate-
gory. The second category uses the physics of diffraction and interference to recover
the spectral information and is represented by instruments that employ diffraction grat-
ings or interferometry as examples. The number of bands recorded can range from a
few to a few hundred depending upon the application with the spectral distance between
samples on the order of 5 to 25 nm. For a well designed and manufactured imaging
spectrometer, the shape of the function that describes the spectral sampling will be uni-
form for all spectral channels for every ground sample. This will minimize the spectral
artifacts introduced by the spectral imager that could compromise the exploitation of
the data.
1.3.4 Radiometric Sampling

The function of the detector is to transform radiant power into an electrical signal.
The electrical signal is converted into a number by an analog-to-digital converter.
14 Introduction
15
10
5
Radiance( W/cm2sr/nm)
0
500 1000 1500 2000 2500
15
10
0
500 1000 1500 2000 2500
Wavelength(nm)
Figure 1.11 The two graphs depict MODTRAN R

5 spectral radiance simulations from 400 to 2500 nm for
the same scene radiance. The broad absorption features at 940 and 1115 nm are due to water
absorption. The opaque absorption features at about 1400 and 1900 nm are also due to water
with the scene radiance completely absorbed by the atmosphere. The deep and narrow absorption
feature at 760 nm is due to oxygen absorption. The upper graph is a high-resolution model that
shows the details of the molecular absorption features at a resolution of 1 nm. The lower graph
shows a similar result but at a resolution of 10 nm.
Detectors are classified on the basis of the physical mechanisms that cause the
conversion from radiant to electrical energy. The two main classes are thermal detec-
tors and quantum (or photon) detectors with imaging spectrometers employing the
later type.
The line or area arrays employed in an imaging spectrometer are composed of individ-
ual photoconductors. These semiconductor devices work quantum-mechanically with
photons being absorbed to produce free charge carriers. Photoconductors function over a
broad spectral range and are employed throughout our spectral region of 0.350 to 14 m.
The wavelength range is determined by the details of the semiconductor quantum struc-
ture with the quantum efficiency, the ratio of a quantum of light input to the device to a
charge carrier generated within the device, varying as a function of wavelength.
The resulting current generated by applying a voltage across the detector element
is collected and stored in a capacitor for measurement. This voltage is converted to a
digital number (DN) using an analog-to-digital converter with a set number of bits B,
typically 12 or 14. A B-bit converter records 2B levels. The number of bits determines
the radiometric resolution. For example, if the maximum range of the electrical signal
is R, then the radiometric resolution is R 2B .
Another random document with
no related content on Scribd:
the next evening, if the weather permitted. This programme had
been drawn up by my husband and myself two or three weeks
before Mr. Quinton’s arrival, but it has since come to light that the
Jubraj suspected us of treachery in asking him to arrange and be
present at these nautches.
NATIVES OF THE MANIPUR HILLS.

We had never seen so many people in the Residency at once as
there were that Sunday night at dinner—fifteen in all. I felt rather
forlorn, being the only lady present, and wished that I had even one
familiar spirit in the shape of another woman to keep me company.
The band was very much appreciated, and everything seemed
very bright and cheery. No thought of evil troubled any of us, for little
we knew that it was the last evening we were to spend in peace
there all together.
The next day Mr. Melville was to leave us. I had tried to persuade
him to stay longer, as he had only been two days with us before, and
had seen nothing of the place; but his time was precious, and he had
his work to do in a great many other places, so we could not get him
to alter his arrangements. He agreed to compromise matters by
remaining until the afternoon of the following day, the 23rd, and
about eleven the party broke up and retired to rest.
CHAPTER XIV.
Up early on the eventful morning—The Jubraj does not attend the Durbar—Visit of
Mr. Grimwood to the Jubraj—Finds him in high fever—Matters assume a
serious aspect—Thoroughfares deserted—Terrific thunderstorm—Our
servants take French leave—My ayah deserts—Melancholy thoughts—Lovely
moonlight night—A Manipuri arrives to spy out our doings—The night before
the outbreak—Attack on the Residency—Capture of the Jubraj’s house—
Anxiety about Lieutenant Brackenbury—Stray bullets find their billet in the
Residency—Attack gets hot, and big guns play on the Residency—We have
to take to the cellars—The Regent invites Mr. Quinton to an interview.
It is now some time since the events took place which I am

recording here, and not one vestige of the past remains to help me in
my work, not a single scrap of writing or note of any kind; yet the
smallest detail of those few terrible days is engraved so indelibly
upon my memory that it seems but the occurrence of yesterday, and
I need no reference to help me in my description of a catastrophe
which almost outrivals some of the horrors of the Mutiny.
We were all up early on the morning of the 23rd. The durbar was
fixed to take place at eight, and the rooms had to be prepared for the
ceremony. But when eight o’clock came, it brought only a message
from the palace, saying that the Jubraj was too ill to leave his house,
and therefore the regent had not come; so the red cloth arranged for
his reception was put away, and a consultation took place between
the Chief and my husband as to what the next move should be.
It was decided to make one last attempt to get the princes to
attend, and then if that failed, other measures were to be resorted to.
But twelve o’clock brought no better results, and about four my
husband was sent to the palace to see the Jubraj, and convey to him
personally the orders of the Government, and use all his influence to
persuade the prince to give himself up quietly, telling him at the
same time that the proposed banishment was not to last for ever, but
that it would depend chiefly on his good behaviour, and eventually, at
the death of his brother, the regent, he (the Jubraj) would be allowed
to return to Manipur, and ascend the throne as Maharajah. It was a
veritable hornets’ nest into which my husband ventured that
afternoon, accompanied only by his friend Mr. Simpson. He would
not take even a single orderly with him, knowing in what an excited
state the whole palace was at that time. It was crowded with Sepoys,
collected, the regent told him, for the review which we had desired to
witness. I got very anxious about them both when more than an hour
had passed and they had not returned, but when my husband did
come back I knew at a glance that his mission had failed. He said
the Jubraj was certainly very unwell. He had had some difficulty at
first in persuading the prince to come and see him at all, but after
finding out from his people that the two Sahibs had come quite
alone, the Jubraj had himself carried down to see them in a litter.
The exertion caused him to faint, and my husband said that there
was no doubt as to his illness, and that he found him in high fever.
Shortly before this visit to the palace took place, Mr. Melville
started off on the first stage of his journey to Kohima. He intended
travelling as far as Sengmai that day, a distance of ten miles. My
husband tried to make him reconsider his decision to go, and I
added my persuasions to his. We did not like the look of things at all,
and how matters would end was, to say the least, very uncertain.
I remember so well our all standing on the steps of the Residency
that afternoon, watching the coolies collecting Mr. Melville’s luggage,
and begging him to remain even one day longer with us, for fear of
anything going wrong; and I remember equally well his answer:
‘Thank you very much, Mrs. Grimwood,’ he said, ‘but do not fear for
me. I am not important enough to be captured by these Manipuris. I
shall get on all right, never fear’; and in a few more minutes he had
left us. But the thoughts of his going away like that, alone, without
any guard to protect him, troubled me more than once that
afternoon, and I could not get it out of my head. Matters assumed a
serious aspect indeed when my husband returned about six o’clock
from the palace with the news that he had been unsuccessful with
regard to the Jubraj. There was only one way, then, out of the
difficulty, and that was to place the affair in the hands of the military,
and apply force where persuasion had failed. It was a council of war,
indeed, and everything seemed to combine to fill me with sensations
of dread for what was going to happen. I could not feel the
excitement that took possession of the men when the chances of a
probable fight became known. Such an idea filled me with alarm and
horror. The place had a deserted look about it, and on the principal
road, as a rule crowded with people at that hour, not a soul was
visible.
The clouds had been gathering up all the afternoon, and about
seven o’clock a terrific thunderstorm occurred, and darkness set in,
which was only lit up now and then by brilliant flashes of lightning. I
busied myself about the house, where I found a state of confusion
reigning. A number of the servants had taken French leave and
departed, scenting danger.
My old ayah was among the first to go. She had been with me four
years, and had followed me about faithfully till now; but at the first
sign of danger she packed up her belongings and went off. I
wondered where she had taken refuge, for she had a good many
enemies, and was not a native of Manipur, so had no home or
relations in the place to whom she could fly for protection. I felt her
desertion very much. She was only a native, but she was at any rate
a woman, and better than no one in a case like that. However, there
was no good to be got out of thinking over her departure, and I had
as much as I could do as it was to keep the other servants up to the
mark, and get them to understand that dinner that evening would
have to be gone through the same as usual. Mr. Quinton and three
of the others amused themselves by playing whist until dinner-time,
as, of course, going out was an impossibility; and I went to the
kitchen to superintend the arrangements there, and to make
preparations for the next day, as I knew that if there were fighting
going on, I should be left without a single servant, and so resolved to
get as much work out of them while they were there as was possible.
We made a quantity of soup that night, as I thought it would be
useful, and cooked some fowls to provide us with something to eat
the next day in case of accidents. And then we had dinner.
No one seemed inclined to speak much that evening. With me
conversation was almost an impossibility, and the rest were too
excited about the morrow to be able to talk and laugh as they had
done the day before. It was a relief to me when dinner was over. I felt
nervous and low-spirited, and very lonely, quite out of place amongst
those men whose profession it was to fight, and who were longing
for the next morning.
Thoughts of England and of all whom I loved there, flocked
through my mind, and I wondered what they would say if they could
see us then, and know the possible danger that threatened us and
our home. My husband was troubled at the thoughts of my being in
the place at such a time, and he blamed himself for having agreed to
my staying, though I had done so of my own free will. Even then we
did not dream of any really serious ending. We expected that the
Jubraj would fight well—in fact, the officers and Sepoys were hoping
that the resistance would be strong, and my husband was afraid that
the house might get knocked about, and some of our property
destroyed; but serious alarm for our own safety never entered our
heads. This was the night of the 23rd, the date that we had originally
fixed for the Manipuri nautch to take place; but under the
circumstances we did not think it likely that the girls would come. Mr.
Brackenbury amused us by singing comic songs, accompanying
himself on his banjo after dinner, and all went to bed early, as
everyone had to be up at three the next morning.
It was a lovely moonlight night, and my husband and I walked up
and down in the garden for some time after our guests had gone. I
felt restless and unhappy, but he did his best to reassure me and
make me believe that we should all be perfectly safe. Just before we
were preparing to go in, the sentry challenged at the gate, and
appeared a few minutes afterwards with a Manipuri, who had been
sent from the palace to inquire whether we wished to have the
nautch or not, saying at the same time that the girls who were to
dance were waiting outside in the road if we wanted them. Of course
we told him it was much too late to think of such a thing at that time,
and the man left the place. We believed that he had really been sent
to spy out the land, and find out what preparations, if any, we had
been making. If that were his mission, he must have been seriously
disappointed, for the whole place, and everyone in it, was wrapt in
slumber, with the exception of my husband and myself, and we very
often walked about the grounds late on moonlight nights, so there
was nothing unusual in our doing so on this occasion. There were a
few extra sentries on guard, but chiefly at the back of the house, and
the presence of the Chief Commissioner was quite sufficient cause
for a larger guard than usual.
At last we turned in too, one to sleep as unconcernedly as ever,
knowing not that it was his last night upon earth; the other to lie
awake, listening to the hours as they were struck out on the gong
down at the quarter-guard, and wonder what the ending of the next
day would bring. I never closed my eyes all through the watches of
that night—the last I was destined to spend in Manipur—and when
three o’clock came I woke my husband, and told him that the hour
had come when we were all to get up, and the work of day
commenced.
It was a bitterly cold morning, and quite dark. I dressed quickly,
putting on a warm, tight-fitting winter dress. We had a sort of scratch
breakfast of eggs and bread-and-butter about 3.30 a.m.; but most of
the officers had theirs at the camp, and started from there, so that I
did not know when they actually left to commence the attack. My
husband accompanied Colonel Skene, much to my distress, as I
thought he would have stayed at the Residency, being a civilian; but
he seemed just as keen on going as the others, and I had to make
the best of it. Mr. Quinton, Mr. Cossins, and I all went off to the
telegraph-office, which was situated at the end of the drive, about
three hundred yards from the Residency. It was well built and fairly
strong, the basement being made of stone, and there was a similar
building on the opposite side, which my husband used as an office
for himself, the lower half of which contained the treasury. Here we
took up our position, going up into the telegraph-office first to send a
message to the Government of India, giving details of all that had
occurred up to date. Mr. Cossins, who was acting as secretary to the
Chief Commissioner, brought the telegram down, and while we were
waiting and watching the Baboo[15] despatch it, we heard the first
shot fired in the palace, which was followed up quickly by others, and
we knew then that the fight had begun.
By this time the dawn was breaking, and streaks of daylight were
dispelling the darkness around. It seemed difficult to me to realize
what was really taking place. I had heard firing in the palace so often
that it seemed almost impossible to understand that a sterner game
was being played now, and one which was to cost both sides so
dear. Only half the telegram had been sent, when we were startled
by the sudden advent of a bullet through the office window at our
elbows. It crashed through the glass, breaking it to pieces, and went
into the wall opposite. My heart went to my mouth with fright, and I
left the place with considerable rapidity, taking up a more secure
position below, where I was fairly protected by the stone basement of
the building. Mr. Cossins occupied his time in taking several journeys
up to the house, where he mounted to the roof to discover whether
he could see what was going on inside the palace wall; but it was
impossible, as the Jubraj’s house, which we knew was being
attacked, was some distance off, and hidden from us by intervening
buildings. It was situated near the outer wall of the palace, and our
men seemed to have taken a very short time in getting up to it.
The whole palace was fortified. Five walls surrounded the
Maharajah’s enclosure. The outer of these was much broken, and of
no great height; but the inner ones were very strong, built of brick
and supplied with bastions, and they surrounded the inner palace on
all four sides. On three sides of the outer wall was a canal, very deep
and wide. It was here that the great boat-races took place every
year, and the water was always kept weeded and clean for those
events. The whole citadel was built with a view of resisting attack in
the time before Burmah was annexed, when armies of raiders used
to come down upon Manipur with hostile intent; and it was a place
which could easily be held against an attacking force, provided big
guns were not brought to bear upon it. The Manipuris were well
armed, and supplied with ammunition. The Maharajah had four
mountain guns which had been presented to his father by our
Government in return for services which he had rendered in times
gone by. The Jubraj understood perfectly how to work these guns.
We had seen him fire them himself for our amusement on an
occasion already described, and we knew he would be perfectly
cognizant of their powers of destruction when the opportunity
occurred to bring them into play against us. Of course we, who were
left at the Residency, did not know what was going on round the
Jubraj’s house, where all the firing seemed to come from. From time
to time stray bullets came over our heads where we sat down at the
telegraph-office. I thought it was very exciting then, and the little
Ghoorkas, who had remained to keep guard over the place, were
constantly running out on to the road in front of our entrance-gate, to
see whether they could discover what was happening. They did not
like being inactive at all.
About half-past ten my husband returned, and came to the
treasury to get out some of the reserve ammunition which had been
stored there. He only stayed a few minutes, talking to me before
rejoining Colonel Skene. He told me that the Jubraj’s house had
been captured after a good fight, and that our men were in
possession of it, and the principal gateway besides, and had taken a
good many prisoners. I asked if anyone had been hurt, and he said
there were grave rumours about Lieutenant Brackenbury. No one
seemed to be certain of his whereabouts, while some affirmed that
he had been wounded, and others that he had been killed. We were
very anxious about him, but my husband said that it was all
uncertain, and he might be perfectly safe all the time; and of course
we hoped he was all right.
About twelve Mr. Quinton and I went up to the house, but long
before going he had made another attempt to get the telegram to the
Government of India despatched, and had found that the wires had
been cut on all sides, so all hope of communication from that source
was abandoned. We were rather hungry by twelve, as no one had
eaten much at the hasty repast at three in the morning, and we were
very glad of some hot tea and sandwiches now. I went on a voyage
of discovery round the house. One or two servants still remained, but
they seemed very frightened, and were saying many prayers to their
gods for their safety. A stray bullet or so had hit the walls of the
house, knocking off some of the plaster, but otherwise everything
looked the same as usual.
SKETCH MAP OF MANIPUR.
We returned to the office in about an hour, after I had seen that all
the preparations for lunch were made. The cook had departed, but
the bearer and I between us managed to get things ready in a
fashion. I took a book to read with me, and busied myself in that
manner until, about one o’clock, Colonel Skene and some of his
officers, with my husband, returned from the scene of action. Our
first inquiries were for Mr. Brackenbury, and then it became evident
that something serious had happened to him, and all our fears were
aroused. After that, things seemed to assume quite a different
aspect, for the officers were all talking so gravely together, and did
not seem quite satisfied with the way things were going.
However, we went back to the Residency to get something to eat.
All had returned with the exception of Mr. Simpson and Captain
Butcher, who were still at the Jubraj’s house, and Mr. Brackenbury,
whose exact whereabouts were unknown. We had commenced
lunch, when my husband asked me if I would give orders that some
food should be sent to the two officers who were not able to leave
their posts, and I went away to a little room adjoining the dining-room
and commenced cutting sandwiches for them, as the servants had
disappeared, and one had to get everything for one’s self or go
without.
I had been busily engaged for about ten minutes, when I heard a
sound which filled me with alarm, and a bullet crashed through the
window above my head. It frightened me more than the one at the
telegraph-office had done, and I dropped my knife, left the
sandwiches as they were, and rushed into the dining-room. All the
officers meanwhile had gone out, and had found that the Manipuris
had crept round to the back of the Residency and commenced an
attack upon us, using as cover the Naga village which lay between
our grounds and the river. This was a clever move on their part, and
it was some time before the troops could drive them back, as most of
our men were engaged in holding the posts inside the palace
captured early in the morning, and this left only a small guard for the
Residency, treasury offices, and Sepoys’ camp. Eventually our party
set fire to the Naga village, and drove the Manipuris out. Bullets had
made their way through the window-panes and doors of the dining-
room, and had smashed some of the breakfast-things and the glass
on the sideboard. It was difficult to find out the most secure place in
the house, as the firing was hot in the front of the Residency by this
time, and the walls, being only lath and plaster, were little or no
protection.
My husband suggested my descending to the cellars, which were
under the house and built of stone; but I did not like the idea, and
remembered how scornful I had been when we had talked over
matters weeks before, and he had joked about the snug corner he
would make ready for me in the basement of the house. So I made
up my mind to remain above-board, so to speak, until the worst
came to the worst. It was heart-rending to see the work of
destruction which was proceeding in the different rooms meanwhile.
The windows were broken, and every now and then bullets crashed
into the rooms, smashing different things—first a picture, then a
vase, then a photograph. All my beloved household gods seemed
coming to grief under my very eyes, and I was powerless to save
them. We did try to collect some of the most valuable of our
belongings together and put them away in a heap in the durbar
room, which at that time had escaped with only one broken pane; but
it was dangerous work going into the front rooms to remove them, for
as the afternoon went on the firing became hotter, and bullets rained
into the house at every second.
It must have been about half-past four that the big guns began to
be played against us. It had been found necessary to concentrate
the whole of our force on the Residency and out-buildings, such as
the treasury and offices, and this entailed abandoning all the
positions captured in the early part of the day inside the outer wall of
the palace, and bringing all the men together. The wounded had to
be recovered from all directions and conveyed to the hospital, which
was some distance from the Residency.
Lieutenant Brackenbury had been discovered lying on the bank of
the river which flowed north of the palace, where he had fallen
shortly after the attack was made early in the morning. He had
mistaken the direction, having got the wrong side of the wall near the
Jubraj’s house, from which point he had been exposed to a heavy
fire from the enemy. It was only a marvel that he was still alive when
eventually discovered, for he had remained where he fell the whole
of that day, and the Manipuris had never ceased firing at him as he
lay. When his exact whereabouts did become known, it was a difficult
and dangerous task to remove him. Efforts had been made by some
of the Sepoys to drag him away, and a native officer had been
mortally wounded in the attempt. At last, about four o’clock in the
afternoon, he was rescued and brought into the hospital, and it was
found that he had received terrible injuries, being wounded in several
places.
The sound of the first shell which whizzed over the Residency
made me speechless with terror. I had heard the boom of the guns in
the morning, and knew that they had been used to try and drive
Captain Butcher’s party out of the Jubraj’s house, which had been
captured; but they had sounded some distance off, and I had not
realized how terrible they could be until they were turned against our
own house.
The cellars were by this time unavoidable. My husband told me
that we should have to make some sort of rough hospital in one of
them, as the Residency hospital, where the wounded had been
taken, was built of plaster and would not be bullet-proof; so we set to
work to get blankets and sheets down from the house, and
everything we thought might be useful.
Meanwhile shells were doing dreadful damage over our heads,
and we were afraid they might set fire to the thatch and force us out
of our temporary shelter. Luckily most of them went over the house
into the garden at the back, where they could not do such serious
damage; but the noise the guns made, added to the other firing,
which had never ceased, was deafening.
There was not the slightest doubt by this time that our position was
about as bad as it could very well be. I seemed paralyzed with fear,
and it was only by forcing myself to do something, and never thinking
or imagining for one moment what the end of it all might be, that I
kept my senses sufficiently to be able to make an effort to help the
rest. I heard that the wounded were to be brought up to the house
immediately, as the hospital was getting too hot for them to remain in
it. Poor fellows! they had endured so much as it was in getting there,
that it seemed very hard to be obliged to move them again so soon,
and take them up to the Residency.
There were a good many of us in the cellar by this time—Mr.
Quinton, Colonel Skene, my husband and myself, Mr. Cossins, and
Mr. Gurdon. It was about seven o’clock, and a lovely evening. The
sun was just setting, and the red glow of the sky seemed to
illuminate the landscape around and the faces of the colonel and my
husband as they stood in the doorway talking together in low tones.
It was no difficult matter to read what was written on both their faces,
and I did not dare ask what was going to happen.
At last my husband came and told me that we were to leave the
Residency, and try and find our way to Cachar. It seemed worse to
me to think of going out of the house than to remain there; but
whatever was to take place had to be at once, and there was no time
to spend in giving way to the terrible fear which possessed me.
However, a further consultation was held, and it was decided to
make a truce with the regent, and put an end to hostilities by coming
to some terms with him. A letter was written, which the Chief
Commissioner signed. It ran as follows:
‘On what condition will you cease firing on us, and give us
time to communicate with the Viceroy, and repair the
telegraph?’
While this letter was being written, the colonel had ordered our
buglers to sound the ‘cease fire,’ which they did at once; but it was
some time before the Manipuris followed suit. At last their guns
ceased, and all was quiet. Then my husband went out with the letter,
and called a Manipuri off the wall to take it to the Jubraj. The man
went away with it, and my husband returned to the Residency.
FRANK ST. CLAIR GRIMWOOD.
FROM A PHOTOGRAPH BY VANDYK.
LONDON: RICHARD BENTLEY & SON, 1891

Some minutes later a message came to say that the regent
wished to see Mr. Quinton and talk over matters with him; and this
message was followed by a letter written in Bengali, which contained
an acknowledgment of the Chief’s letter, and a proposal to the effect
that we should surrender our arms if the Manipuris agreed to cease
firing. There was some discussion about the translation of part of this
letter, and Mr. Quinton proposed that the Jubraj should be called
upon to explain the meaning of the passage in question, and asked
whether it would be possible to see him.
Meanwhile the Chief Commissioner’s party, consisting of himself,
Colonel Skene, Mr. Cossins, Lieutenant Simpson, and my husband,
had gone down to the office at our entrance gate, and waited there
while the regent’s letter was being translated. Mr. Simpson had gone
of his own accord, as he wished to accompany my husband, and I
had begged to be allowed to go with him too; but he said I was safer
where I was, and bade me good-bye, telling me to keep a brave
heart, that the firing was at an end, and peace about to be restored;
and he told me to try and get some rest, as I looked so tired. I never
saw him again.
CHAPTER XV.
Mr. Brackenbury—Scenes in the little cellar—Destruction of our home—Another
moonlight night with a difference—Reopening of the attack on the Residency
—Death of Mr. Brackenbury—Preparations to escape.
I remained where he had left me, alone for some minutes, though
some of the officers were standing just outside the door of the cellar
where I was sitting. It seemed so hard that I could not go with my
husband. I feared being left alone without him, and felt very lonely
and broken-hearted among so many men, mostly strangers to me. I
knew, too, that they would look upon me as an extra burden, and
wish me very far away.
I was roused to action by the doctor, who had taken advantage of
the truce to get his wounded brought up from the hospital to the
house, and had come up first to see what kind of a place could be
got ready. I showed him the cellars, for there were several, which all
communicated with each other, and formed the entire basement of
the house.
Shortly afterwards the Kahars[16] arrived, carrying poor Mr.
Brackenbury on a mattress, and the others followed fast, so that the
small cellar was very soon quite full of men lying side by side on the
stone floor. The blankets and sheets that we had already collected
were very useful, and I made several journeys up to the house, and
gathered up every kind of covering from every direction, and all the
pillows I could find. A little cooking-stove proved of great service. I
fixed it securely upon a table in one corner which I reserved for
cooking operations.
The soup we had made on the previous day was in great request.
Fortunately there was a large quantity of it, to which I added the
contents of five or six tins which I found in the store-room. Milk was
the difficulty. All the cows were out in the grounds, and many of them
had strayed away altogether and we could not get any milk from
them, so were obliged to fall back on condensed milk, of which we
also had several tins.
Some of the men were terribly wounded, but poor Mr. Brackenbury
was by far the worst. His legs and arms were all broken, and he had
several other injuries besides. It seemed a marvel that he was still
alive and fully conscious to all that was going on around him. The
doctor attended to him first of all, and had bound up his broken
limbs, and done as much as possible to alleviate his sufferings; but it
was a terrible sight to see the poor lad in such agony, and be so
powerless to lessen it in any way. He was very thirsty, and drank a
good deal of soup and milk, but we could not get him into a
comfortable position. One minute he would lie down, and the next
beg to be lifted up; and every now and then his ankle would
commence bleeding, and cause him agony to have it bound up
afresh. His face was gray and drawn, and damp dews collected on
his forehead from the great pain he was suffering.
That scene will never be forgotten—the little cellar with a low roof,
and the faces of the wounded lying together on the floor. We did not
dare have a bright light for fear of attracting attention to that
particular spot, and the doctor did his work with one dim lantern.
Such work as it was, too! Every now and then he asked me to go
outside for a few moments while the dead were removed to give a
little more space for the living.
There were some terrible sights in the cellar that night—I pray I
may never see any more like them; but being able to help the doctor
was a great blessing to me, as it occupied my attention, and gave
me no time to think of all the terrible events of the day, and the wreck
of our pretty home. I was very weary, too—in fact, we all were—and
when at about half-past ten I asked everybody to come and get
some sort of a dinner, they seemed much more inclined to go to
sleep, and no one ate much.
The dinner was not inviting, but it was the best that could be got
under the circumstances, for I had had to do it all myself. One or two
of the servants still remained, but they cowered down in corners of
the house, and refused to move out or help me at all. Perhaps had
we known that it was our last meal for nearly forty-eight hours, we
should have taken care to make the most of it; but no thought of
what was coming entered our minds, and long before the melancholy
meal was ended most of the officers were dozing, and I felt as
though I could sleep for a week without waking.
We all separated after dinner about the house. I went back to the
hospital for a little, and found the doctor wanted more milk, so I
returned to the dining-room, where I was joined by Captain Boileau,
and we sat there for some time mixing the condensed milk with
water, and filling bottles with it, which I took downstairs. It was
quieter there than it had been. The wounded had all been attended
to, and most of them had fallen asleep. Even Mr. Brackenbury was
dozing, and seemed a little easier, and only one man was crying out
and moaning, and he was mortally wounded in the head. So finding I
could do no more there, I went upstairs again, resolving, if possible,
to go to my room and lie down for a little while and sleep, for I was
very tired.
I went sorrowfully through our once pretty drawing-room, where
everything was now in the wildest confusion, and saw all the
destruction which had overtaken my most cherished possessions.
There are those who imagine that in a case like this a woman’s
resource would be tears; but I felt I could not weep then. I was
overwhelmed at the terrible fate which had come upon us, and too
stunned to realize and bewail our misfortunes.
Perhaps the great weariness which overcame me may have
helped me to look passively on my surroundings, and I walked
through the house as one in a dream, longing only to get to some
haven of rest, where I could forget the misery of it all in sleep.
I wended my way to the bedroom through a small office of my
husband’s, but when I reached the door I found it would not open,
and discovered that part of the roof had fallen in, caused by the
bursting of a shell. So I gave up the idea of seeking rest there, and
retired to the veranda.
I went down the steps and stood outside in the moonlight for a few
minutes. It was a lovely night, clear and bright as day! One could
scarcely imagine a more peaceful scene. The house had been
greatly damaged, but that was not apparent in the moonlight, and the
front had escaped the shells which had gone through the roof and
burst all round at the back. The roses and heliotrope smelt heavy in
the night air, and a cricket or two chirped merrily as usual in the
creepers on the walls.
I thought of the night before, and of how my husband and I had
walked together up and down in the moonlight, talking of what the
day was to bring, and how little he had thought of such a terrible
ending; and I remembered that poor lad lying wounded in the cellar
below now, who only twenty-four hours ago had been the life and
soul of the party, singing comic songs with his banjo, and looking
forward eagerly to the chances of fighting that might be his when the
morning came.
I wondered where my husband was, and why they had been away
so long. They would be hungry and tired, I thought, and might have
waited to arrange matters till the next day, as they had apparently
been successful in restoring peace. I had an idea of wandering as far
as the gate to see whether the party was visible, but on second
thoughts I went back into the veranda, and resolved to wait there
until my husband should return.
There was one of the officers asleep in a chair close to me, and I
was about to follow his example, when Captain Boileau came out,
and I went to him and asked him if he would mind going down to the
gate and finding out whether he could hear or see anything of the
Chief Commissioner’s party, and if he came across any of them to
say I wanted my husband. He went off at once, and I fell into a doze
in the chair.
It was about twelve o’clock at this time. I do not know how long I
had been asleep, when I was awaked suddenly by hearing the
deafening boom of the big guns again, and knew then that it was not
to be peace.
For a few seconds I could not stir. Terror seemed to have seized
hold of me, and my limbs refused to move; but in a minute I
recovered, and ran through the house down to the cellar again,

Textbook Hyperspectral Imaging Remote Sensing Physics Sensors and Algorithms 1St Edition Dimitris G Manolakis Ebook All Chapter PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Textbook Hyperspectral Imaging Remote Sensing Physics Sensors and Algorithms 1St Edition Dimitris G Manolakis Ebook All Chapter PDF

Uploaded by

Copyright:

Available Formats

Hyperspectral Imaging Remote Sensing

Physics Sensors and Algorithms 1st

Hyperspectral Remote Sensing: Fundamentals and

Hyperspectral remote sensing of vegetation. Volume IV,

Polarization Remote Sensing Physics Lei Yan

Remote sensing of vegetation. Hyperspectral indices and

Novel Sensors and Sensing First Edition Jackson

Image analysis, classification and change detection in

LiDAR Remote Sensing and Applications 1st Edition

LiDAR Remote Sensing and Applications 1st Edition

Dimitris G. Manolakis is a Senior Member of Technical Staff at MIT Lincoln Laboratory

“An extraordinarily comprehensive treatment of hyperspectral remote sensing by three of the

Cambridge University Press is part of the University of Cambridge.

To Claire, Kate, and Mary

2 The Remote Sensing Environment 36

3 Spectral Properties of Materials 117

3.8 Long Wave Infrared Spectra 150

4 Imaging Spectrometers 154

5 Imaging Spectrometer Characterization and Data Calibration 228

6 Radiative Transfer and Atmospheric Compensation 295

7 Statistical Models for Spectral Data 360

8 Linear Spectral Transformations 406

9 Spectral Mixture Analysis 443

10 Signal Detection Theory 494

11 Hyperspectral Data Exploitation 551

Appendix Introduction to Gaussian Optics 621

Color plate section between pages 338 and 339.

T Q̃ − ’š•©Kš’− £ J̃ J© ¡Kš—’ ©˜’š’š š›¡’,

The roots of education are bitter but the fruit is sweet.

This chapter provides an introduction to the basic principles of hyperspectral remote

Earth’s surface interaction The amount and spectral distribution of radiation

1.2 Infrared Sensing Phenomenology

1.2.1 Sources of Infrared Radiation

1.2.2 Atmospheric Propagation

1.2.3 Reflectance and Emissivity Spectra

Visible Near Infrared Middle Infrared

Grass Walnut tree canopy

Sandy Loam Soil

1.3 Hyperspectral Imaging Sensors

Hyperspectral sensors, more properly known as imaging spectrometers, col-

1.3.1 Spectral–Spatial Data Collection and Organization

Figure 1.7 Basic principle of a hyperspectral remote sensor

1.3.2 Spatial Sampling

1.3.3 Spectral Sampling

1.3.4 Radiometric Sampling

Figure 1.11 The two graphs depict MODTRAN R

NATIVES OF THE MANIPUR HILLS.

It is now some time since the events took place which I am

FRANK ST. CLAIR GRIMWOOD.

FROM A PHOTOGRAPH BY VANDYK.

LONDON: RICHARD BENTLEY & SON, 1891

You might also like