Thermal imaging techniques to survey and monitor animals in the wild : a methodology

361 Pages·2016·14.25 MB·English

by  Havens, Kirk J., Sharp, Edward J

Thermal Imaging
Techniques to Survey
and Monitor Animals
in the Wild
A Methodology
Kirk J Havens
Edward J Sharp
AMSTERDAM • BOSTON • HEIDELBERG • LONDON
NEW YORK • OXFORD • PARIS • SAN DIEGO
SAN FRANCISCO • SINGAPORE • SYDNEY • TOKYO
Academic Press is an imprint of Elsevier
Academic Press is an imprint of Elsevier
125, London Wall, EC2Y 5AS, UK
525 B Street, Suite 1800, San Diego, CA 92101-4495, USA
225 Wyman Street, Waltham, MA 02451, USA
The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, UK
Copyright © 2016 Elsevier Inc. All rights reserved.
No part of this publication may be reproduced or transmitted in any form or by any means, elec-
tronic or mechanical, including photocopying, recording, or any information storage and retrieval
system, without permission in writing from the publisher. Details on how to seek permission, further
information about the Publisher’s permissions policies and our arrangements with organizations
such as the Copyright Clearance Center and the Copyright Licensing Agency, can be found at our
website: www.elsevier.com/permissions.
This book and the individual contributions contained in it are protected under copyright by the
Publisher (other than as may be noted herein).
Notices
Knowledge and best practice in this field are constantly changing. As new research and experience
broaden our understanding, changes in research methods, professional practices, or medical
treatment may become necessary.
Practitioners and researchers must always rely on their own experience and knowledge in evaluating
and using any information, methods, compounds, or experiments described herein. In using such
information or methods they should be mindful of their own safety and the safety of others, including
parties for whom they have a professional responsibility.
To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume
any liability for any injury and/or damage to persons or property as a matter of products liability,
negligence or otherwise, or from any use or operation of any methods, products, instructions, or
ideas contained in the material herein.
ISBN: 978-0-12-803384-5
British Library Cataloguing-in-Publication Data
A catalogue record for this book is available from the British Library
Library of Congress Cataloging-in-Publication Data
A catalog record for this book is available from the Library of Congress
For information on all Academic Press publications
visit our website at http://store.elsevier.com/
Dedication
To my wife, Karla, who only occasionally raised an
eyebrow and rarely questioned the late night trips to “study
wildlife.” To my son, Kade, who understands the wisdom in
questioning everything and to my parents, Bill and Ginny,
who gave me the childhood freedom to explore.
Kirk J Havens
Preface
Over the past few decades there has been a marked increase in areas of remote
sensing, including thermal imaging, to study and count wildlife in their natural
surroundings. While much of the work with thermal imagers to date has been
devoted to testing equipment during surveys, little advancement has actually
been achieved. This is primarily due to three basic problems:
1. Early field studies were conducted with cryogenically cooled thermal imag-
ers (photon detectors) with sensitivities an order of magnitude lower than
those available today. With few exceptions, the new and improved models
of thermal imagers with superior sensitivities and resolution have not been
used in the field because of the perceived difficulty in data acquisition and
to some extent limited availability and cost. The more recent fieldwork has
been for the most part confined to the use of uncooled bolometric cameras
that use thermal detectors as opposed to photon detectors.
2. A pervasive misunderstanding of what thermal imagers detect and record
and what ultimately constitutes ideal conditions for conducting thermal im-
aging observations.
3. The promulgation of results that have erroneously compared survey data
collected with thermal imaging equipment to that obtained with standard
techniques such as spotlighting or visual surveys.
In this volume, we spend considerable effort reviewing the literature and
pointing out fallacies that have been built upon as a result of these problems.
This book presents a methodology for maximizing the detectability of both ver-
tebrates (homotherms and poikilotherms) and invertebrates during a census or
survey when using proper thermal imaging techniques. It also provides details
for optimizing the performance of thermal cameras under a wide variety of field
conditions. It is intended to guide field biologists in the creation of a window of
opportunity (a set of ideal conditions) for data gathering efforts. In fact, when
thermal imaging cameras are used properly, under ideal conditions, detectivity
approaching 100% can be achieved.
Recent attempts of researchers and field biologists to use thermal imagers
to survey, census, and monitor wildlife have in most cases met with limited
success and while there are a number of good books that treat the theory and
applications of remote sensing and thermal imaging in significant detail for
applications in land mapping, construction, manufacturing, building and ve-
hicle inspections, surveillance, and medical procedures and analyses (Barrett
xi
xii Preface
and Curtis, 1992; Budzier and Gerlach, 2011; Burney et al., 1988; Holst, 2000;
Kaplan, 1999; Kozlowski and Kosonocky, 1995; Kruse et al., 1962; Vollmer
and Mollmann, 2010; Williams, 2009; Wolfe and Kruse, 1995), they contain
very little on how wildlife biologists should go about using this equipment in
the field to survey and monitor wildlife. This book provides detailed informa-
tion on the theory and performance characteristics of thermal imaging cam-
eras utilizing cooled quantum detectors as the sensitive element and also the
popular uncooled microbolometric imagers introduced into the camera market
in the past decades, which rely on thermal effects to generate an image. In
addition, there are numerous excellent texts devoted to survey design and sta-
tistical modeling to aid in the monitoring and determination of wildlife popula-
tions (Bookhout, 1996; Borchers et al., 2004; Buckland et al., 1993; Buckland
et al., 2001; Caughley, 1977; Conroy and Carroll, 2009; Garton et al., 2012;
Krebs, 1989; Pollock et al., 2004; Seber, 1982, 1986; Silvy, 2012; Thompson
et al., 1998; Thompson, 2004; Williams et al., 2001), but they do not include the
treatment of thermal imaging capabilities to help achieve these tasks. This book
is being offered as a bridge between the two technologies and the teachings
presented in these excellent volumes so that their combined strengths might be
united to improve upon past efforts to assess animal populations and to monitor
their behavior.
Even though there has been a technological disconnect since the earliest field
experiments, there has still been a considerable amount of work carried out by
biologists using thermal imagers to study and monitor wildlife. These studies be-
gan in the late 1960s and early 1970s when cryogenically cooled thermal imagers
using photon detectors were first used for surveys and field work (Croon et al.,
1968; Parker and Driscoll, 1972) and this phenomena continued to grow as ther-
mal imagers became more readily available to field biologists. At the time, these
early cameras were acknowledged as being only marginally sensitive for the task
of aerial surveying. The more recent introduction of the low-cost uncooled bolo-
metric cameras generated a new wave of experimentation with thermal imagers
in the field. The sensitivity and range of bolometric cameras are limited due to
the fact that they rely on a thermal process to generate an image. So we see at the
start that all thermal imagers are not the same and if they are used in the field they
must be used to exploit the strengths of the particular imaging camera so that
reliable data can be obtained. There are appropriate uses for imagers utilizing
photon detectors where high sensitivity and long ranges are characteristics mak-
ing them suitable for surveying applications. There are also applications suitable
for imagers fitted with thermal detectors that have lower sensitivities and ranges.
Their advantages are their availability, cost, and that they are uncooled. Field
applications favoring bolometric cameras that do not require long ranges or high
sensitivity will also be addressed in this book.
The process of using thermal imagers as a tool to collect field data has been
compared with other data collection techniques; however, in nearly all cases the
thermal imager was not used correctly and perhaps was even inadequate for
Preface xiii
the task. This practice has led to a number of misconceptions about the basic use
of a thermal imager and the correct interpretation of the results. There is a big
distinction between thermal imagers that utilize quantum detectors as the sensi-
tive element and detectors that rely on thermal effects to generate an image. The
differences are enormous as far as fieldwork goes for censusing and surveying,
particularly on a landscape scale. Unfortunately, a text describing the use of
3–5 and 8–12 mm photon detectors for animal surveys and field studies has
not emerged. This is probably due to the fact that 3–5 and 8–14 mm imagers
were not widely used since the first field experiments. These experiments used
cryogenically cooled units typically borrowed from military installations. These
robust units are now becoming available at a reasonable cost and should see in-
creased use by field biologists. An excellent text describing the practical use of
pyroelectric and bolometric imagers for a wide range of applications has been
written (Vollmer and Mollmann, 2010) and a number of distinctions are pointed
out between these imagers and those using photon detectors as the focal plane.
Past work using thermal imagers in the field has mainly been carried out so
that comparisons could be made with other data gathering methods. From the
outset we see that comparing the results obtained with thermal imagers with that
of data collected with other methods such as spotlighting and visual surveys must
necessarily be skewed and these efforts, while commendable, do not allow for
a fair comparison of the data collection capability of the compared techniques.
Thermal cameras are suitable for surveys and counts throughout the 24-h diurnal
cycle while other methods are not. These studies by their nature and design mean
that the results of data collected with a thermal imager will be compared with
data collected using a method that was optimized for the conditions of the survey
at hand. For example, consider the comparison of data collected during a visual
survey and the data collected via thermal imagery using the same temporal and
spatial conditions. Note that the survey must be conducted during daylight hours
because the visual spotters need daylight to see the animals of interest. Thermal
cameras can also detect the animals of interest during daylight hours but there are
concomitant conditions required for the optimization of the thermal survey if it is
conducted during daylight hours. These conditions can be met in a relatively easy
manner but were not generally addressed during these past comparisons so the
results reported were skewed and in some cases grossly inaccurate. We review
many of these comparisons and offer alternatives. A variety of statistical meth-
ods, such as distance sampling and mark recapture, among others, were used for
estimating the abundance of animal populations in these comparisons and the
results of these studies were built upon by others. We do not treat these statisti-
cal methods here but point out that each of them has strengths and weaknesses
(Borchers et al., 2004), depending on the species of the animal being surveyed.
All will benefit from data collection methods that produce a detectability (see
Chapter 1) that approaches ∼100%.
The widespread dissemination of these results is the existing foundation that
later work has been built upon and it has led to a confusing and widespread
xiv Preface
misunderstanding of the capabilities of thermal imaging as a powerful survey
tool in these applications. This distribution of erroneous or badly skewed in-
formation regarding the performance of thermal imaging for these tasks needs
to be rectified and it is one of the major goals of this book to start that process.
The work of Romesburg (1981, p. 293) pointed out the fallacies of building
on unreliable knowledge: “Unreliable knowledge is the set of false ideas that are
mistaken for knowledge. If we let unreliable knowledge in, then others, accept-
ing these false laws, will build new knowledge on a false foundation.” We still
overlook important aspects of the scientific inquiry to gain reliable scientific
knowledge. All the statistical methods applied to data gathered in the field are
better predictors when the count is completely random and the sample is large.
It is also known that the general methods used to count animals in the field dur-
ing a survey are usually biased and yield animal counts less than what is actually
there; however, in some cases there will be more counted than are actually there.
These statistical losses or gains are presumably accounted for in the statisti-
cal formulation being used. The problems arise when the estimated parameters
to account for losses or gains in populations, along with other parameters to
account for such things as species mingling, group sizes, mortality rates, and
sometimes double counting, are folded into the calculations. Even though these
parameters are often very good guesses, they all come with systematic and
random errors attached and cannot predict valid outcomes except by chance
(Romesburg, 1981, p. 309). This is because the more parameters a model con-
tains that are guesses the more they are amplified by their interaction with one
another through the calculations, such that the resulting errors can be quite large
at the output of the calculations.
It is essential for wildlife management and the preservation of healthy popu-
lations that we seek and promulgate reliable knowledge regarding the current
status of animals in the wild. Ratti and Garton (1996) advance the important re-
alization put forth by Romesburg by showing that in order for wildlife research
to be useful to wildlife managers and their varied programs, it must be founded
on high-quality scientific investigations that are in turn based upon carefully
designed experiments and methodologies. Limitations to achieving the desired
high quality and reliable knowledge must be identified and rectified. We postu-
late that the single most important thing to do at the present time to mitigate the
unreliable knowledge stemming from skewed and distorted animal surveys and
counts is to look very carefully at the detectability possible by different count-
ing methodologies.
The components of science required for meaningful and reliable outcomes
are mingled together in a relatively complex way. Wildlife managers and field
biologists must incorporate biology, chemistry, atmospheric science, physics,
and climatology, as well as the behavioral ecology and physiology of the animals
surveyed or studied. All must be considered when forming a research plan for a
species. The best window of opportunity for collecting data must be determined
based on the best science available. To this end, a detailed methodology for using
Preface xv
infrared thermal imaging to conduct animal surveys in the field and other stud-
ies requiring nondisruptive observation of wildlife in their natural surroundings
is developed in this book. We show that ∼100% detection can be achieved for
surveys if the methodology is formulated to take full advantage of the infrared
cameras used for observation and if it is coupled with the details of the behav-
ioral ecology and physiology of the animals being surveyed or studied.
In this book we address the primary difficulty with surveying or censusing
animals and demonstrate that it is not the sampling methodology (i.e., distance
sampling, aerial transect sampling, quadrat sampling, etc.) or the statistical
model being used on the collected data, but rather lies with the detectability
that can be achieved with any particular sampling or data collecting technique.
This suggests that more work needs to be done on comparing factors that influ-
ence the detectability of a species of interest rather than the statistical methods
to compensate for the inadequacies of over or undercounting. There are many
other details of a research plan that could grossly skew or render the resulting
survey invalid (Thompson et al., 1998; Lancia et al., 1996; Krebs, 1989) but the
visual observation (or other counting methods) are well-known to be skewed
by a number of factors and limit data collection to daylight hours or when the
landscapes or transects are artificially illuminated. It is also known that artificial
illumination introduces behavioral modifications that can adversely influence
the detectability and introduce bias (Focardi et al., 2001). There are various
treatments proposed to deal with known biases. They are adjustments to the
calculations to deal with under- or overcounting animals during surveys re-
sulting from biased detectability. In this work, we will concentrate on the task
of increasing detectability by eliminating bias in the data collection aspect of
wildlife monitoring.
Because thermal imaging can be conducted at any time during the diurnal
cycle and can be conducted from various aerial or ground-based viewing plat-
forms, it offers a host of configurations to observe animals of interest while
using their preferred habitat. If performed correctly, the observations can be
conducted from a distance that precludes disturbances to the animals under
study, thus reducing the possibilities of skewing the counts or surveys caused
by anthropogenic-produced behavioral changes or double counting. Each vari-
able introduced by some recognized uncertainty in the counting or observation
techniques used must be accounted for and if it is done statistically the results
become more and more questionable. If an uncertainty in the counting tech-
nique can be fixed at the field level, the resulting counts are closer in line with
the true situation because there is one less layer of data manipulation to perform
due to under- or overcounting.
As noted earlier, there is already a significant amount of up-to-date infor-
mation available on methods for treating collections of field data with various
statistical formulations and appropriate assumptions. These mathematical tools
allow the evaluation of field data (if correctly collected) so that meaningful es-
timations of the abundance and/or the density of wildlife populations can be
xvi Preface
determined. As a result, we do not delve into these methods but rather focus on
the details of establishing a technique for correctly collecting data and achiev-
ing the highest detectability possible when conducting field work. Applications
other than those dealing with wildlife will not be treated here unless we need
to make a specific point about some aspect of the workings of a thermal imager
or if the application would clarify some aspect of the proposed methodology.
Applications such as military, surveillance, police work, fire detection, manu-
facturing, and building inspection have been well-treated by others and can be
found in the references mentioned earlier. The results of many studies of animal
behavior, thermoregulation, pathology, and physiology are also reviewed.
In order to appreciate the advantages that thermal imaging has to offer we
must recognize that our eyes are sensors that are limited in a number of ways
that limit their utility as effective detectors of wildlife in their preferred habitat.
Our eyes are confined to the visible region of the spectrum and at low-light lev-
els they do not collect enough data so that our brain is able to form images that
are recognizable; however, there are a number of ways that we can easily extend
their functional range for our applications. For example, binoculars greatly en-
hance the probability of observing an object when faced with low-light levels
and long viewing ranges. If we can use various technologies and instrumenta-
tion to aid our vision by seeing in the dark and seeing at longer ranges, then we
need to add these things to our set of observational tools. In short we need to
detect objects in order to count them and we need to see them in some fashion
to detect them. The acquisition of images in the infrared region of the spectrum
can be provided by thermal imagers and as such serve as an aid to our overall vi-
sual capability. By utilizing thermal imagers we can create images of very high
contrast so that objects of interest are clear and distinct from their backgrounds,
allowing us to extend our visual capability into the dark portion of the diurnal
cycle. Once this is accomplished, the brain can process the images that the eyes
see. In fact, in recent work at Cal Tech and UCLA, researchers found that indi-
vidual nerve cells fired when subjects were shown photos of well-known per-
sonalities. The same individual nerve cell would fire for many different photos
of the same personality and a different single nerve cell would fire for many dif-
ferent photos of another personality. Follow-up research suggests that relatively
few neurons are involved in representing any given person, place, or concept,
which makes the brain extremely efficient at storing and recalling information
after receiving visual stimulation.
Without going into a detailed mathematical description of thermal imaging
and the complex principles behind the operation of thermal imagers (thermal
cameras) we instead introduce basic laws and principles that allow us to set the
stage for data collection with thermal imagers. However, field biologists need
to have a basic understanding of the physics governing heat transfer processes
in the environment (Monteith and Unsworth, 2008) and the effects of local me-
teorological changes on the performance of a thermal imager. The proper use of
a thermal imager requires a basic knowledge of how an imager works, why we
Preface xvii
see what we see with a thermal imager, and how we can optimize those images
for the tasks at hand. Simple “point-and-shoot” infrared imagery for data col-
lection will not work nor will using someone else’s “point-and-shoot” imagery
in sophisticated statistical calculations. What the imagery actually represents
and how it was acquired must be known for it to be useful. While the perfor-
mance capability of uncooled thermal imagers has improved remarkably over
the last decade and the cost of these cameras has become reasonable for most
researchers, field biologists must understand how they work, how to use them,
and what they are actually recording as imagery. Unfortunately, for the most
part, the rapid technological advancement and availability of thermal imagers
has outpaced the knowledge and understanding required of the specialists using
them in the field (Vollmer and Mollmann, 2010, p. xv). This sad commentary
regarding the use of thermal imagers stems, for the most part, from applications
associated with monitoring inanimate objects in fixed backgrounds. Our appli-
cations, as we have already pointed out, are much more difficult and complex
so we need to be particularly careful and thorough in our understanding of a few
basic principles regarding thermal imaging and wildlife ecology.
This book is about formulating a methodology to optimize a window of
opportunity so that wildlife can be observed and studied in its natural habitat.
This requires that biologists and program managers get together and formulate
a sound survey design, which assumes that they know the ecology of the spe-
cies of interest plus all mitigating factors that could possibly distort the outcome
of a thermal imaging survey. The methodology presented here is logical and
simple yet it demands a detailed understanding and incorporation of critically
interlinked disciplines arising from biology, physics, micrometeorology, ani-
mal physiology, and common sense. Thermal imaging is a technique that forms
images from heat radiating from objects and their backgrounds, so much of
the information contained in this book is devoted to managing the interplay
of the heat transfer processes of conduction, convection, and radiation between
the objects of interest (animals) and their backgrounds to obtain the best thermal
images. We will see that creating this window of opportunity is not as restrictive
as one might think. Data can be collected from ground- or aerial-based plat-
forms at any time during the diurnal cycle without compromising detectivity,
disturbing the animals, or altering their behavior. Even though the methodology
used to obtain meaningful data brings together a wide range of criterion and re-
quirements that must be met concomitantly, it boils down to creating a window
of opportunity that will allow researchers to conduct surveys with near 100%
detectability by properly using thermal imagers as a detection tool.