Conservation Drones Mapping and Monitoring Biodiversity Serge A Wich Full Chapter

Conservation Drones Mapping and
Monitoring Biodiversity Serge A. Wich

Visit to download the full and correct content document:
https://ebookmass.com/product/conservation-drones-mapping-and-monitoring-biodive
rsity-serge-a-wich/
OUP CORRECTED PROOF – FINAL, 24/4/2018, SPi
Conservation Drones
Conservation
Drones
Mapping and Monitoring
Biodiversity
Serge A. Wich, PhD

Liverpool John Moores University, UK, and University of Amsterdam, The Netherlands
Lian Pin Koh, PhD

University of Adelaide, Australia, and Conservation International, USA
1
3
Great Clarendon Street, Oxford, OX2 6DP,
United Kingdom
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK and in certain other countries
© Serge A. Wich and Lian Pin Koh 2018
The moral rights of the authors have been asserted
First Edition published in 2018
Impression: 1
All rights reserved. No part of this publication may be reproduced, stored in
a retrieval system, or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by licence or under terms agreed with the appropriate reprographics
rights organization. Enquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press, at the
address above
You must not circulate this work in any other form
and you must impose this same condition on any acquirer
Published in the United States of America by Oxford University press
198 Madison Avenue, New York, NY 10016, United States of America
British Library Cataloguing in Publication Data
Data available
Library of Congress Control Number: 2018933616
ISBN 978–0–19–878761–7
DOI:10.1093/oso/9780198787617.001.0001
Printed and bound by
CPI Group (UK) Ltd, Croydon, CR0 4YY
Links to third party websites are provided by Oxford in good faith and
for information only. Oxford disclaims any responsibility for the materials
contained in any third party website referenced in this work.
Preface
Serendipity plays a large part in science and was In a few days we flew 36 missions and collected
also crucial in how we started our drone work. a wealth of data that made us very enthusiastic
Both of us had been working in Southeast Asia’s about this technology. We shared our ideas and
forests to study its rich biodiversity for many years findings with colleagues and because of their pos-
when we first met in May 2011, over a coffee in the itive feedback started the conservationdrones.org
aptly named Greencafe of the Swiss Federal Insti- website and coined the term conservation drones.
tute for Science in Zurich. We planned to discuss Since then we have had the opportunity to work in
the ongoing impact of oil palm plantation devel- a large number of locations around the world and
opment on the species we studied and particularly posted these on our website in the hope that this
the orangutans. Soon enough we were discussing would inspire others to explore the use of drones
how costly and time intensive it is to obtain data for their work. We hope that we have stimulated a
at regular intervals for species such as orangutans few people to start using drones through the website
over large areas and how wonderful it would be if and our presentations and papers. The last seven
we could monitor the nests that orangutans build years have seen an enormous growth in the use of
from the air. Lian Pin had just started to fly remote drones for conservation work and a large amount of
control fixed wing aircraft so we decided to see exciting work has been done by researchers around
if we could fit a camera under a fixed wing, and the world. With this book we aim to share some of
fit an open source flight controller to it as well so our own work, but fortunately much more the work
that we could fly pre-programmed missions. A few of others, in the field of drones, conservation, and
months later, after some tests in Switzerland, we ecological research. We hope that this will further
found ourselves in Sumatra flying a Bixler fixed contribute to excite people about the opportunities
wing with an APM 1 and camera over the forests. that this field offers for conservation.
v
Acknowledgements
We owe a great amount of gratitude to many We specifically would like to express our grat-
colleagues and friends who contributed to this itude to Andy Goodwin, Dominique Chabot, Jeff
book in numerous ways. We apologize for those Kerby, and Owen McAree for kindly reviewing
who we have inadvertently forgotten to mention. most or all of the chapters. We appreciate Noémie
Some contributed by reviewing chapters, some by Bonnin’s tremendous help with going through some
providing discussion and ideas about certain topics, of the literature for Chapter 3. Perry van Duijnhoven
some by providing images and figures, some by has voluntarily designed many of the figures in this
collecting data, some by stimulating us to write this book and been very patient with the many ‘small’
book, some by helping us to build drones, some by amendments we asked for.
kindly sharing their drone expertise with us, some Finally SW is grateful for the love and support
by providing us with hardware, some by allowing from Tine, Amara, and Lenn during the writing of
us time at work to focus on drones when many this book and being patient with me being away
people thought this was ‘just a hobby’, some by during the many fieldtrips ‘droning’ around the
providing a supportive work environment, and world. Their patience has been enormous during
many by being on various data collection missions the weekends and evenings that were absorbed by
with us in the field and being a crucial part of writing. LPK dedicates this book to his lovely wife,
the experience we have collected over the years. Juanita, for holding the fort at home while he trav-
In alphabetical order we thank Christiaan Adams, elled around the world, ‘working’.
Marc Ancrenaz, Sander van Andel, Chris Ander- During writing LPK was supported by the
son, Brandon Basso, Frederic Bezombes, Noémie Australian Research Council, and Conservation
Bonnin, Hidde Boonstra, Claire Burke, David International. We thank the following organizations
Burton, Rhett Butler, Ahimsa Campos-Arceiz, Mike in no systematic order for supporting our work
Checkley, Adrienne Chitayata, Juanita Choo, financially or in other ways. National Geographic,
Kenneth Clarke, Chris Collins, David Dellatore, Waitt Foundation, Mongabay, US Fish and Wildlife
Brenden Duffy, Regina Frey, Andy Goodwin, Benoit Services, Denver Zoo, Philadelphia Zoo, Conserva-
Goossens, Molly Hennenkam, Christiaan van der tion International, Wildlife Conservation Society,
Hoeven, Jeff Kerby, Adam Kilpatrick, Francis Lilley, Chester Zoo, Ugalla Primate Project, RemoteIn-
Wee Siong Lim, Steve Longmore, Tony Lynam, sights, Google, 3DRobotics, WWF Netherlands and
David Mobach, Matt Nowak, Keeyen Pang, Ryan several other WWF offices, Orangutan Information
Pang, Alex Piel, Lilian Pintea, Maisie Rashman, Centre, ISTAT Foundation Humanitarian Grant,
Ramesh Raja Segaran, Carel van Schaik, Ahmed Orangutan Conservancy, Swiss Federal Institute of
Al-Shamma’a, Ian Singleton, Danica Stark, Andy Technology, PanEco, Jane Goodall Institute, Univer-
Tattersall, Gokarna Thapa, Graham Usher, Jaap van sity of Zurich, Liverpool John Moores University,
der Waarde, Peter Wheeler, Simon Wunderlin, and HUTAN, Indianapolis Zoo, Margot Marsh, Science
the large number of field staff that have supported and Technology Facilities Council (STFC), and the
our various missions. Sumatran Orangutan Conservation Programme.
vii
Contents
1 Deciding to use a drone 1
1.0 Introduction 1
1.1 Alternative technologies to acquire aerial data 1
1.1.1 Manned planes and helicopters 1
1.1.2 Satellites 3
1.1.3 Kites 5
1.1.4 Balloons and blimps 6
1.2 To drone or not to drone 6
1.2.1 Terrain 9
1.2.2 Weather 10
1.2.3 High altitude flying 11
1.2.4 Landing area 11
1.2.5 Regulations 11
2 Typology and anatomy of drones 13
2.1 Multirotor 13
2.2 Fixed wing aircraft 13
2.3 Hybrid VTOL 13
2.4 Features of multirotor vs fixed wing vs VTOL drones 14
2.5 Essential components 14
2.5.1 Power source 14
2.5.2 Flight controller 17
2.5.3 Ground control station 18
2.6 Summary 19
3 Sensors 20
3.0 Introduction 20
3.1 Sensor types 20
3.1.1 Visible spectrum sensors (RGB) 24
3.1.2 Multispectral cameras 36
3.1.3 Hyperspectral cameras 38
3.1.4 Thermal imaging cameras 38
3.1.5 Light Detection and Ranging (LiDAR) 38
3.1.6 Synthetic Aperture Radar (SAR) 39
3.1.7 Telemetry 39
3.1.8 Miscellaneous sensors 39
3.1.9 Live transmission 39
ix
x CONTENTS
3.2 Applications of sensor types 40

3.2.1 Animal studies 40
3.2.2 Locating radio-tagged animals 40
3.2.3 Land-cover classification 41
3.2.4 Land-cover change detection 42
3.2.5 Land-cover features 42
3.2.6 Water content/stress 42
3.2.7 Leaf area index 43
3.2.8 Fires 43
3.2.9 Miscellaneous applications 44
4 Surveillance 46
4.1 Terrestrial 46
4.2 Marine 46
4.3 Key issues 47
4.3.1 Technical challenges 47
4.3.2 Ethics and privacy 49
4.3.3 Regulations 50
5 Mapping 51
5.1 Land-cover mapping 51

5.2 Monitoring of vegetation condition 51
5.3 Biomass estimation 52
5.4 Key considerations 52
5.4.1 Mission planning 52
5.4.2 Data processing software 54
5.4.3 Environmental factors 54
6 Animal detection 55
6.1 Animal distribution and density data collection 55

6.1.1 Alternative methods 55
6.2 Drone usage for animal distribution and density 56
6.2.1 Comparing traditional surveys to drone surveys 57
6.2.2 Estimating density and population size 60
6.3 Anti-poaching efforts 61
6.4 Disturbance 65
7 Data post processing 73
7.0 Introduction 73
7.1 Photogrammetry basics 73
7.2 Basic process 74
7.3 Photogrammetry software packages 78
7.4 Analyses 79
7.4.1 General 79
7.4.2 Three-dimensional quantification of landscape and vegetation using
photogrammetry 80
CONTENTS xi
7.4.3 Analyses based on orthomosaics: land-cover and feature classification 81

7.4.4 Combining drone and satellite data 85
7.4.5 Automated analyses to detect and identify plant species 87
7.4.6 Automated animal detection, identification, and tracking 87
8 Future casting 94
8.1 Power systems 94

8.2 Autonomous systems 94
8.3 Platform integration 95
References 97
Index 115
CHAPTER 1
Deciding to use a drone
1.0 Introduction 1.1 Alternative technologies to acquire

aerial data
The recent proliferation of non-military applications
of drones in many aspects of our society, from Conservationists have a long history of using aerial
filming to asset management and farming, has and space platforms to obtain imagery for land-use
also spread into conservation (Koh and Wich 2012, mapping and monitoring, as well as for the study
Chabot and Bird 2015, Floreano and Wood 2015, of animal distribution and density. Here we discuss
Wich 2015, Christie, Gilbert et al. 2016, Crutsinger, the most common ones: manned aircraft, satellites,
Short et al. 2016). Within the conservation com- kites, and balloons. All of these come with a set
munity drones have mainly been used for three of advantages and disadvantages that are useful to
aspects of conservation efforts: 1) to determine the think through when considering collecting aerial
distribution and density of animal species; 2) for and/or space data for conservation projects (see
mapping of land-cover and land-cover change; 3) Table 1.1).
anti-poaching activities. These aspects will be dealt
with in more detail in later chapters of this volume 1.1.1 Manned planes and helicopters
along with related aspects of spatial ecology that are
relevant to conservation science and management. Aerial photography from manned planes started
In this chapter we will provide a brief introduction with one of the pioneers of flying, Wilbur Wright,
to these three main aspects and then focus on which in 1908 and seriously took off during World War I
considerations are important to take into account to provide imagery of the frontlines (Professional
when deciding to start applying drones for conser- Aerial Photographers Association 2016). After
vation or not. We will start, however, with providing World War I, mapping of areas for non-military
an overview of alternative aerial data-collection purposes started in earnest (Professional Aerial
technologies. Deciding which technology to use Photographers Association 2016). It has been and
for data gathering is not always straightforward still is an important aspect of natural resource
and will likely depend on multiple factors (Morgan, management (Paine and Kiser 2012). Manned
Gergel et al. 2010, Toth and Jóźków 2016). We will aircraft (planes or helicopters, Figure 1.1) have
give an overview of some aspects that are important been used widely by conservationists to map
to consider when starting the process of thinking land-cover and to examine land-cover change
about data collection. Then we will discuss which (Al-Bakri, Taylor et al. 2001, Plieninger 2006, Gerard,
equipment and infrastructural needs are important Petit et al. 2010). Similarly, aerial census work to
to consider when using drones and we will end by obtain data on animal distribution and density has
discussing some general considerations related to been extensively used (Caughley 1977, Buckland,
the usage of drones for conservation. Anderson et al. 2001, Buckland, Anderson et al.
Conservation Drones: Mapping and Monitoring Biodiversity. Serge A. Wich & Lian Pin Koh. Oxford University Press (2018).
© Serge A. Wich & Lian Pin Koh 2018. DOI:10.1093/oso/9780198787617.001.0001
1
2 C O N S E R VAT I O N D R O N E S
Table 1.1 Aerial and space data collection platforms and their constraints
Area covered Resolution Regulations∗ Wind Flight duration Costs Other

constraints
Satellites Large Relatively low NA∗∗ NA Multiple years High for high Cloud cover
resolution
Manned aircraft Large Medium Aviation Light-strong Hours High
authorities
Kites Small High Aviation Light-strong As long as there is Low
authorities wind
Balloons Small High Aviation Light Multiple days Low Hydrogen or
authorities helium might
not be locally
available or be
difficult to
transport to
the site
Drones Medium High Aviation Light-moderate Hours or parts Medium
authorities of hours
∗ Regulations for flying aerial platforms vary between countries, often depend on the weight of the platform, and the purpose of the flight (e.g. hobby, research, or
commercial). ∗∗ NA = not applicable.
Figure 1.1 Researchers climbing into a hovering helicopter above the Batang Toru river in Sumatra, Indonesia, and researchers preparing for a
flight with a Cessna fitted with gliders to find orangutan nests in the peat swamp forests of Indonesian Borneo. © Serge Wich.
2004). In some cases, such as that of the Cape fur photos after the flight. There have been some
seals (Arctocephalus pusillus pusillus) in southern concerns that aerial counts of large herbivore
Africa aerial census work has allowed for almost species are considerably lower than ground counts
four decades of population monitoring (Kirkman, (Jachmann 2002) and can be influenced by factors
Yemane et al. 2013). For instance aerial counts such as the low probability of spotting single
of animals in Africa have been conducted since animals, small groups of animals, and animals that
the mid-1950s (Jachmann 2002), but in contrast to were not visible because they were behind or under
surveys with drones the detection of animals is vegetation. In addition, animals that react to the
mainly done by human observers and not from aircraft and move are more likely to be recorded
D E C I D I N G TO U S E A D R O N E 3
(Jachmann 2002). Furthermore, there are potential This stands in contrast to data collection with a
biases introduced by factors relating to humans camera on a drone where after a drone flight there
such as visual acuity, concentration, and visual are potentially thousands of photos that need to be
attention (Fleming and Tracey 2008). Important, but examined to detect animals or their signs. Although
not often considered, is how the characteristics of inspecting images is conducted mostly manually at
different survey platforms (helicopters and fixed the moment, there are promising computer vision
wing planes) can affect the detection by observers methods being developed to (semi) automate this
on a manned aircraft (Noyes, Johnson et al. 2000, process (e.g. Chen, Shioi et al. 2014, van Gemert,
Fleming and Tracey 2008). Verschoor et al. 2014). An important advantage of
Although using manned aircraft has been widely collecting data with a camera on board a drone is
applied, there are also three challenges to using that there is a record of the data that can be re-
these. First, aerial surveys with manned aircraft analysed and shared if needed.
often face high costs. These costs are highly variable
for different parts of the world but can be sub-
1.1.2 Satellites
stantial and often prohibitive for conservationists,
particularly when regular flights are required. Classifying land-cover types and changes in land-
Second, availability of manned aircraft can be an cover types over time is usually conducted with
issue in certain parts of the world. For instance, the aid of satellite images (Horning, Robinson et al.
in parts of Indonesia where we have worked it 2010) and, particularly at the global scale, it will con-
is sometimes not possible to rent small planes tinue to be conducted in this way (Hansen, Potapov
to fly low over forest areas or remote areas can et al. 2013). Since 1972 satellites have been taking
be too far away for a small plane to reach. In images of the Earth at various resolutions (see Table
some areas with large river systems planes that 1.1), at various temporal intervals, and with various
can land on water can be an option. Third, there sensors (Table 1.1; for extensive lists of satellites
is substantial risk associated with aerial surveys and their specifications see Horning, Robinson et al.
using manned aircraft. Crashes are a leading cause 2010, Toth and Jóźków 2016). Analyses using these
of death among wildlife biologists (Sasse 2003). images have given us a wealth of information about
Fourth, flying manned aircraft for anti-poaching global land-cover and change as well as detailed
efforts is extremely risky and can lead to pilots being assessments of smaller areas.
killed.1 There are, however, four challenges with using
In many instances the census method applied satellite images. First, in the humid tropics, but
is substantially different from census work using also in the Arctic, areas can be covered by clouds
drones because much of the census work using for prolonged periods of time, which renders
manned aircraft is conducted directly by the satellite image acquisition of an area impossible or
observers on board the aircraft (Buckland, Ander- significantly delays acquisition (Hansen, Roy et al.
son et al. 2001, Buckland, Anderson et al. 2004), 2008, Mulaca, Storvoldb et al. 2011). In arctic regions
whereas with drones the most common method is passive sensors also do not provide information
to have a camera that faces down to take images on during the prolonged dark periods and therefore
which animals or their signs are counted later on do not provide scientists with the data they require
(Chabot and Bird 2012, Koh and Wich 2012, Chabot in a region that is experiencing rapid climate and
and Bird 2015, Wich 2015). Direct counts from a ecological change (Post, Forchhammer et al. 2009,
manned aircraft are time efficient because at the Serreze, Barrett et al. 2009, Post, Bhatt et al. 2013).
end of the flight the data is ready to be analysed Second, conservationists would often prefer to use
or close to being in the right format for analyses. high resolution imagery and, even though these
high resolution images are available, these are
1 https://www.theguardian.com/world/2016/jan/30/
costly and therefore often not within the budgets
british-pilot-killed-by-elephant-poachers-tanzania (accessed 7 of conservation workers (Table 1.2). Third, even
November 2017) though several satellites acquire imagery with short
Table 1.2 Selection of satellites
Satellite Pixel resolution Repeat frequency Estimated costs

(m)∗ (days) (US $) km2∗∗
Worldview-3 1.24 <1 14.50–17.50

GeoEye-I 1.65 2–8 14.50–17.50
Spot 4/HRVIR 20 2–3 0.8
Landsat/TM & 30 16 Free
ETM+
Sentinel 2 10, 20, 60 5 Free
MODIS 250, 500, 1000 1–2 Free
Table based on Horning, Robinson et al. 2010 with additional information. ∗ All pixel resolutions are for
the multispectral images. ∗∗ Price estimates for archived images and depends on exact bands required
(http://www.landinfo.com/satellite-imagery-pricing.html). Even though prices are per km2 the minimum area that
needs to be purchased is larger. For example for Spot 4 the minimum area is 1,000 km2 . See http://www.wmo-
sat.info/oscar/ for a complete list of Earth observation satellites that have been launched or are being planned
for launch in the near future.
Figure 1.2 Left panel: Drone georeferenced orthomosaic superimposed on a Landsat satellite image from an area in Sumatra. Landsat pixel
resolution is 30m per pixel, drone orthomosaic pixel resolution is 3cm per pixel. The orthomosaic was produced with Pix4Dmapper
(www.pix4d.com). Right panel: Single image from drone for the same area as the orthomosaic. © Serge Wich.
refresh rates (Table 1.2), the almost persistent cloud imagery and allow for a large flexibility in the
cover in some tropical areas and the high costs of timing of such data acquisition (Figure 1.2).
high resolution images make repeat data for land- Recently a number of companies (e.g. Planet
cover analyses prohibitively expensive for many Labs2 ) have been deploying small satellites (cube-
conservationists. Fourth, in cases where immediate sats, nanosats) into orbit with the aim of providing
data are necessary it is possible to task satellites to high resolution images (~1 m) at a much reduced
obtain data within a few days, but this comes at price and at daily refresh rates once the full
a premium price. Drones can therefore be a useful constellations of satellites are up.
addition to the use of satellite images because they
allow for the acquisition of very high resolution 2 www.planet.com (accessed 7 November 2017)
The World Meteorological Organization’s Obser- with some of the first stunning images from the air
ving Systems Capability Analysis and Review tool such as those of the devastation in the aftermath of
(OSCAR)3 is a complete list with current status the San Francisco earthquake (Professional Aerial
of the very many Earth observation satellites that Photographers Association 2016). Nowadays there
have been launched or are being planned in the are opportunities to 3D print aerial camera mount
near future. kits that pivot4 so that the camera remains in a
Although satellites have mainly been used by stable position under relatively small kites, or to
conservationists to obtain land-cover classifications purchase them online.5 Kite photography provides
and analyses of land-cover change, the high a good opportunity for a very low-cost aerial pho-
resolution satellite images have recently also been tography platform that allows for very high reso-
used to detect animals or their signs. An early lution mapping. Although kites are probably best
comparison of aerial counts from planes and counts suited for small-scale area mapping, area coverage
of several geese species using very high resolution can be substantial if large areas can be covered, for
satellite images showed promising results (Laliberte example by foot along a coastline or by boats in
and Ripple 2003). A pioneering study published an estuary or along rivers. This method has been
in 2012 managed to obtain a population estimate used in the aftermath of the BP Deepwater Horizon
of emperor penguins (Aptenodytes forsteri) from disaster off the coast of Louisiana to map coastal
space using a combination of medium and very areas.6 Kite photography is less suitable for areas
high resolution satellite images (Fretwell, LaRue that are forested because only the edge of the forest
et al. 2012). Subsequent studies have used satellite can be photographed. An additional benefit of kite
images (GeoEye-1) to obtain accurate counts of photography is that relatively high wind speeds
Burchell’s zebras (Equus quagga burchellii) and blue do still allow for operations, whereas for lighter
wildebeest (Connochaetes taurinus) in the Serengeti drones high winds can mean that the wind speed
National Park, in Tanzania (Yang, Wang et al. is higher than the speed at which the drone is safe
2014). Studies of marine mammals such as ele- to operate. Varying wind speeds have been argued
phant seals (Mirounga leonina) (McMahon, Howe to be less suitable for kite photography operations
et al. 2014) and Southern right whales (Eubalaena (Verhoeven 2009), but the development of kites with
australis) (Fretwell, Staniland et al. 2014) have been self-stabilizing cameras (Murray, Neal et al. 2013)
conducted as well. Although these approaches will probably allow for the usage of kites in more
are very promising in terms of their accuracy varying wind speeds as well because photos under
compared to other census methods, they have nadir can still be obtained.
several challenges such as cloud cover, difficulty Although kites have been used extensively in
or impossibility of detecting animals under veg- archaeology to map excavation sites and cultural
etation, and the necessity of complex analyses heritage sites (Verhoeven 2009) the use of kites
(Yang, Wang et al. 2014). for conservation research has been limited so far,
but recently researchers have been arguing that
1.1.3 Kites kites are suitable for environmental research as a
low-cost option that might have less constraints
In 1882 the English meteorologist E. D. Archibald on usage by regulations, are logistically easier
claimed to have taken the first aerial photos from in remote areas, and are useful in areas where
a kite, but this imagery no longer exists (Verhoeven wind speeds are often not conducive for flying of
2009). The earliest surviving images are from Arthur smaller drones (Anderson 2016, Anderson, Griffiths
Batut in 1888 (Batut 1890). Due to the weight of
cameras being used a series of kites had to be used 4 Fotr example, http://www.thingiverse.com/search/page:
to lift the cameras. This method provided the world 1?q=picavet (accessed 7 November 2017)
5 https://publiclab.myshopify.com/collections/mapping
(accessed 7 November 2017)

3 http://www.wmo-sat.info/oscar/ (accessed 7 November 6 https://publiclab.org/archive (accessed 7 November
2017) 2017)
et al. 2016, Duffy and Anderson 2016). An early development of advanced tethered balloons has
study assessed the possibility of using kites for recently also occurred for the police and military.9
shoreline mapping (Scoffin 1982), and subsequent Balloons can be very useful aerial platforms as
studies have continued to use kites for mapping they can stay aloft for prolonged periods of time
of vegetation, ice networks, and intertidal systems, (days rather than hours). They can either be fixed
and making orthomosaics to map vegetation types in one location by tethering them to the ground or
(Boike and Yoshikawa 2003, Smith, Chandler et al. they can be moved around. The fact that they can
2009, Bogacki, Giersz et al. 2010, Bryson, Johnson- be tethered has the advantage that the sensors can
Roberson et al. 2013). Organizations such as Public be powered from the ground which alleviates issues
Lab7 have been promoting kite and balloon aerial surrounding the batteries that are needed to power
photography and provide details on the kit that can sensors over prolonged periods of time. Tethering
be used on their website. also allows for a direct download of the data to a
Studies that have used kites for aerial census ground station.
of animals are sparse. Studies have applied kite
photography to count Adelie penguins (Pygoscelis
adeliae) (Fraser, Carlson et al. 1999) and New 1.2 To drone or not to drone
Zealand sea lions (Phocarctos hookeri) (Cawthorn
There are several aspects to consider before decid-
1993), and more recently beluga whales (Delphi-
ing whether a drone is the right platform. A key
napterus leucas).8
aspect in thinking about using a drone or not is to
determine exactly what the data are that one would
1.1.4 Balloons and blimps like to collect and whether these can be collected
The use of balloons to take aerial images has a long by an alternative aerial or space platform. A sim-
history that started in 1858 in Paris when the French ple decision tree can guide this process. The first
photographer Gaspard-Félix Tournachon flew with question in this tree can be whether the data that
a balloon over Paris (Newhall 1982). Unmanned one requires are collected in other ways or are not
balloons in archaeology have been in use since collected at all at present. If the data are not being
1932 using hot air, hydrogen or helium to provide collected at the moment and there are no alternative
lift (Guy 1932, Verhoeven 2009). Balloons lack a technologies (see section 1.1 on alternative technolo-
system to power them through the air and non- gies) to obtain the data then it is worth investi-
rigid airship systems (i.e. systems that completely gating whether using drones would be possible. If
rely on the internal pressure to maintain their the data are being collected by other methods or
shape) that do have a propulsion mechanism are can potentially be collected by other methods then
called blimps. Balloons and blimps of various kinds it is important to evaluate whether the costs, both
have been used to map vegetation and analyses money and time, of those methods are greater or
of such images have been applied to distinguish smaller than using drones. If the costs of current
various plant species assemblages (Miyamoto, methods or non-drone alternatives are higher than
Yoshino et al. 2004), mapping of agricultural fields using drones then the use of a drone is worth exam-
(Thornton, Fawcett et al. 1990, Inoue, Morinaga et al. ining further. At present there is unfortunately lit-
2000, Jensen, Apan et al. 2007), plant communities tle information available about the costs of differ-
(Pitt, Glover et al. 1996), vegetation mapping ent data collection methods (for a rare, but not all-
(Miyamoto, Yoshino et al. 2005), river corridors inclusive comparison of elephant surveys see Ver-
(Vericat, Brasington et al. 2009), and geomorphology meulen, Lejeune et al. 2013; see also Groves, Alcorn
and soils (see references in Verhoeven 2009). The et al. 2016 for an example on counting salmon nest-
ing sites). Both these studies indicated that drone
7
www.publiclab.org (accessed 7 November 2017) surveys were more costly than the manned aircraft
8
http://www.ifaw.org/united-states/news/how-
carefully-count-beluga-herd-including-every-baby-beluga 9 http://www.military.com/NewContent/0,13190, Defense
(accessed 7 November 2017) watch_072105_Helms,00.html (accessed 7 November 2017)

alternatives, but both do not include all costs, such of two observers (van Schaik, Priatna et al. 1995).
as those from data analysts. Although it is often Recently, researchers have been using drones to
known what the costs of an actual survey are in fly over the rainforests where orangutans occur
terms of renting a plane and car, obtaining a satellite and count their nests on photos taken from the
image, ground teams, food, travel, accommodation, drone. Both methods can yield density estimates
etc., there is little information available about costs (Hennekam 2015, Wich, Dellatore et al. 2016).
of the data analyses in terms of software, hard- For this simple comparison we will assume that
ware, and labour. If these were similar between the transect can be reached in a day walking
the various data collection options comparing these from the basecamp which is on the edge of the
would be relatively straightforward, but unfortu- forest. In addition, we assume that permissions
nately they are not. Although it is beyond the scope to collect both the ground and aerial data have
of this book to try to attempt to estimate the costs equal costs.
of the various methods in detail an example might The ground survey method requires two trained
help to understand the complexity of such compar- researchers to walk for a day to reach the transect,
isons (Table 1.3). spend one day on conducting the data collection,
To determine the distribution and density of and one day to walk back. Training researchers to
orangutans (Pongo spp.) researchers have tradition- collect such data usually takes 2–3 days. During a
ally relied on conducting ground surveys during transect data are recorded on a datasheet with a
which the sleeping platforms (here referred to as pen. Equipment needed during the survey includes
‘nests’) that orangutans build are counted along a a compass, global positioning system (GPS), a mea-
straight line transect that is walked by a minimum suring tape (25 or 50 m), a machete, and a pair of
Table 1.3 Comparison of ground and drone orangutan surveys
Item Ground survey Drone survey
Training data One week Two weeks

collection and
data analyses
Time to collect 3 days 0.5 days
data
Data processing Free US$549 and up
software costs
Computer Modest High processing
processing power
power
Survey ∼1500 US$∗ ∼6000 US$∗
equipment costs
Time to reach 4 days 2 days
density estimate
∗ For ground surveys two pairs of binoculars, one GPS, a laptop with modest processing power, and
the other equipment was estimated to be 1500 US$. For the drone survey we estimated the cost by
using a basic drone from www.hornbillsurveys.com (assessed on 11 February 2016: US$4200, costs of
shipping this to Indonesia, and a laptop with high processing power for data analyses (~1500 US$)).
The drone images can be processed into a georeferenced orthomosaic with several software packages
(see Chapter 7 for details). Although non-commercial options exist, these can be complicated to use
for non-experts and therefore a commercial user-friendly package is advisable. The most affordable
desktop commercial option for educational users (we assume here that the researchers are affiliated to
a university) at the moment is Agisoft Photoscan (www.agisoft.com, assessed on 11 February 2016).
Several cloud options are also available (details in Chapter 7).
binoculars. In addition, equipment to camp out in Once the number of nests is known the density
the forest is needed as well as food for three days. can be calculated by using a previously established
Once the forest team returns to camp, the data can relationship between the density established from
be entered on a computer in a free software package the ground surveys and aerial surveys (Hennekam
(Thomas, Buckland et al. 2009) and processed in a 2015). Thus after two days the drone team has a
few hours to yield a density estimate. Thus in four density estimate.
days a density estimate for that particular transect The above comparison is obviously somewhat
can be achieved. simplistic, but provides some of the aspects that are
For the drone surveys a team of two researchers relevant when comparing methods. The example
would need to be trained for two weeks or so is based on a single survey with one line transect.
(depending on the system) before they would be In this case the ground team would be less costly
sufficiently knowledgeable and confident to pilot to deploy than the drone team. This is due to
the drone. They would need a drone with a camera, the high upfront costs of the drone equipment, a
their survey equipment, and a laptop to act as high processing power computer, and costly data-
a ground control station. The data collection for processing software. In reality surveys are almost
the transect will involve some careful planning never restricted to a single line transect in a single
of the mission including preparation of the flight, area. Usually large-scale survey designs are used
the flight itself, and some post-flight checking over large areas. As scale increases the data collec-
and packing of the system. All combined, that tion with the drone system will become less costly
could take as little as three hours. The researchers per incremental area surveyed than the ground
have chosen to fly in a lawnmower pattern over surveys because the time needed for the ground
the area where the transect is located so that all surveys quickly becomes very large (Wich, Single-
the images can be processed into a georeferenced ton et al. 2016). There is a clear need for careful and
orthomosaic on which the nests can be counted, all-inclusive comparisons between survey methods
and the surface area is known, which is needed to so that conservationists can make decisions about
calculate density. After the flight the images will be which data collection method to use based on
downloaded to the computer and processed by the careful costs estimates for the various methods.
photogrammetry software to make the orthomosaic. Generally when considering whether to use
The images contain a geotag (containing latitude, drones there are several aspects to consider. There
longitude, and altitude) that comes from the GPS is the choice of drone system (e.g. multirotor or
in camera or from the GPS that is used by the fixed wing) and the ground control hardware and
autopilot (more details in Chapter 7). The time software associated with drones, but there are
needed to process the orthomosaic will depend several other aspects too (Table 1.4). There is a
on the specification of the computer, but will be a wide variety of drone options on the market (see
few hours at least. So realistically after day one the Chapter 2) for a variety of price brackets. Then
drone team has completed the data collection and there is a variety of sensors that can be fitted to
has a georeferenced orthomosaic, while the ground the drone depending on what the requirements are
team just arrived at the transect location. On day for the user’s data collection needs (see Chapter 3).
two the drone team will spend the first part of the Most of the sensors collect some sort of image and
day on carefully looking for nests on the images. very quickly these can run into many terabytes so
The camera on the drone is usually set at a short storage and back-up can become a non-negligible
interval, so typically one hour of flight will have cost. There is also the cost of batteries, although this
yielded 1800 photos. A trained observer can check is not high if the batteries are carefully maintained
an image for nests in 1 minute, which would result (see Chapter 2). Finally, there are costs of flight
in 30 hours of checking photos for nests if all the training and, depending on the regulations, costs
images need to be assessed. In this case only a subset of the qualification to fly as a pilot (see Chapter 4).
for the transect area would need to be assessed and In addition to the cost evaluation there are a num-
one morning for two observers would likely suffice. ber of other general considerations when planning
Table 1.4 Aspects to consider when considering using a drone
System Flight Sensors Telemetry/ Maintenance Logistics

parameters Transmission
Ease of software Flight duration Sensor Telemetry General user Ease of

user interface compatibility of distance support transportation
platform
System Flight velocity Legal aspects of Repair Can the batteries
complexity (cruise) telemetry (self/service) be transported
transmission internationally?
power
System material Flight velocity Photo/video Can spare parts
(maximum) transmission be easily
distance purchased?
System Maximum wind Remote control
propulsion speed range
Set up Required size of
complexity take-off area
Based on information from Patrick Ribeiro from https://openforests.com/
Table 1.5 Drone planning and operating aspects in different environments
Operating Safety & Societal Wind Fine Solar effects Spatial Telemetry Topography
environment/ regulations considerations particles (glint, constraints issues issues
specific shadows, (GNSS
challenge albedo) difficulties)
Coastal X X X X X X
Dry land X X X X X
Polar X X X X
Dense forest X X X X X X
High altitude X X X X
Based on (Duffy, Cunliffe et al. 2017).
to use drones and some of those vary on location A few key considerations are highlighted here
(Duffy, Cunliffe et al. 2017). The Duffy et al. study and in Chapter 4.
compiled the experiences of a group of researchers
that have been flying drones in a number of key 1.2.1 Terrain
environments to provide advice to the environ-
mental drone community. They describe a set of An important aspect of flying drones is the
aspects to consider when flying drones in different terrain. Drones are very capable of flying level
environments (Table 1.5). These relate to important above flat terrain and flight is efficient then in
considerations to make pre- and during flight terms of energy usage because the motor can turn
operations in terms of safety and regulations, factors at a relatively constant speed. Most drones can be
influencing safe operations, and successful data programmed to follow the terrain through the use
acquisition. of a digital elevation model or elevations that a
user programs in, but climbing does consume more system can be flown in and these should always
battery power so terrain-following flights will con- be adhered to. Measuring the wind on the ground
sume more energy than level flights. Although high is standard best practice and anemometers are
resolution digital elevation models exist, the digital widely available and affordable. It is important to
elevation models most often used in ground control be aware, though, that wind speed is usually much
software are often of relatively low resolution unless higher in the air than at ground level. So caution
a user provides higher resolution data. For instance is needed. But even below the maximum limit one
the standard global 30 m resolution SRTM (Shuttle should consider if there is a risk of quickly changing
Radar Topography Mission)10 has the advantage weather or whether wind and in particular gusts
of having global coverage but the disadvantage of could negatively influence data collection. For
relatively low resolution for areas that have very mapping missions some wind is not necessarily
undulating terrain. Such low resolution means that negative because a fixed wing drone can maintain
certain peaks or valleys might not be indicated a stable crab angle during stable wind speeds. This
by the digital elevation model and that the drone can actually improve the data collection as long as
therefore needs to fly at a higher elevation above the mission is carefully planned perpendicular to
ground to reduce the risk of crashes. the wind direction.
Drones have been widely used under very
different temperatures from tropical regions to
1.2.2 Weather polar conditions (Koh and Wich 2012, Nilssen,
Storvold et al. 2014, Ratcliffe, Guihen et al. 2015).
An important aspect of drone flying is to be aware Both can pose challenges. The tropics are usually
of the potential risks of adverse weather conditions. very humid and it is well known that prolonged
There are a growing number of specific apps that humidity can be damaging to electrical circuits and
help plan flights in relation to predicted and actual can also damage camera equipment by moisture
weather conditions (e.g. UAV Forecast). Adverse getting trapped between the lenses. Researchers in
weather can impact on drone flight itself and also humid areas often keep large air-tight boxes to store
have potential implications for data collection. Most electronic equipment in, with silica gel to absorb the
small drone systems on the market do not handle moisture. Therefore under such conditions drones
rain well. There are exceptions, and some systems and their sensors need careful storage to prevent the
can withstand some rainfall and can even land in damaging influence of moisture on the equipment.
water if needed. Irrespective of the capability of the Also in very hot areas where equipment is exposed
drone to fly in rain an important question is whether to the sun for prolonged periods of time it is
the sensor on the drone will still be able to collect advisable to use lightly coloured systems as much as
meaningful data during rain. Most data collection possible because they absorb less heat. For instance
by sensors would be negatively impacted by rain. black foam on fixed wing systems can become
Visual sensors either would get wet lenses or sim- very hot and lose its rigid structural properties
ply images would be hampered by rain between and adhesives can delaminate. Altitude estimates
the drone and the land surface. Acoustic sensors from the barometer in autopilot systems can also be
could also be negatively influenced by rain either in influenced by the aircraft being in the full sun due
terms of the sensor becoming humid and yielding to air temperatures rising in the aircraft. Although
different responses or by the rain influencing the flight time seems not to be affected much by the
transmission of sound that the sensor is trying to impact of tropical temperatures on batteries, this
record. Flying in the rain is therefore not advisable. is not the case in very cold conditions. In arctic
Wind is another factor that needs to be carefully conditions flight time is reduced due to shorter
assessed. Drone manufacturers usually provide an battery life. Some drone manufacturers supply
upper limit for wind speed that a particular drone battery warmers to get the batteries up to 15 ◦ C
and also provide upper limits for the usage of their
10 https://earthexplorer.usgs.gov/ drones. A number of multirotor aircraft have also
started to build heaters into the vehicle so that fixed wing are attempts to overcome this and reduce
batteries remain at a constant temperature during the take-off and landing area size constraints for
flight. Some autopilot systems (e.g. Pixhawk2) have fixed wing systems.
a heating element in the data processor to allow it to However, surveys over large forested areas where
work in colder regions. Under very cold conditions there are very few open areas will continue to be a
there is a chance that equipment becomes brittle challenge, particularly when these are not crossed
and breaks so care needs to be taken to avoid by roads that can be used for take-off and landing,
that. But generally flying under cold conditions is but only by river systems. Although it is possible
possible. to take off and land a multirotor on boats, this is
generally not possible with a fixed wing unless there
1.2.3 High altitude flying is a net in which the drone can be caught. Drone
surveys on seas have used nets or hooks from cranes
The last aspect to consider is that when flying at to catch drones during the landing phase (Koski,
high altitudes above sea level flight time is reduced. Abgrall et al. 2010). This requires considerable flying
At high altitudes the air is less dense so more skills though and is not without risk of damaging
power is used to keep the aircraft airborne and this the drone or having it fall into the water. Partic-
reduces flight time. Flying in such areas can often ularly when flying over the ocean the salt water
be improved by using a different set-up than used can quickly lead to corrosion and malfunction of
at standard altitudes, but even standard set-ups electrical systems. There are, however, fixed wing
for multirotors have been successfully flown above systems that have been designed to be able to land in
ground elevations as high as 5000 m. fresh and saltwater and recovered without damage
to the system.11 There are also multirotor systems
that claim to be waterproof and therefore be able to
1.2.4 Landing area
take off and land in water, but the usage of these
An important factor influencing the possibility of systems is not widespread and waves are likely a
using drones is the availability of landing areas. risk to such systems in addition to leaks. Salt air
Both fixed wing drones and multirotor drones need can also damage the ground control station and the
an open area to land in and these can be limited in transmitter. So particularly when flying from small
certain environments such as forests (Duffy, Cun- boats this can be a challenge because waves splash
liffe et al. 2017). The size of the area needed is larger over the boat. An additional challenge is getting the
for a fixed wing system than a multirotor system autopilot to initialize on a moving boat, especially a
because fixed wing systems generally cannot per- small boat that moves substantially. Gimbal systems
form a vertical take-off and landing (VTOL). As that stabilize the take-off area by maintaining it
a result, fixed wing systems need to glide down horizontally could facilitate this. It is also important
which often needs a considerable horizontal flight to realize that without a firm anchor to maintain the
path over which the flight altitude decreases at a position of the boat, the boat will move and there-
relatively shallow angle against the wind. Particu- fore any automated return to launch flight mode
larly when there is tall vegetation surrounding the will not land the drone on the boat, but in the water
landing area the size of the landing area itself needs due to the shifted position of the boat.
to be quite large to enable a safe landing. There are
also fixed wings (specifically delta-shaped wings) 1.2.5 Regulations
that can circle (loiter) down to the ground or close
to the ground and then do a horizontal glide down When considering using a drone for survey work it
over a small distance. But even those systems need is important to determine whether the flights would
considerably more space than multirotors that can be allowed within the regulatory framework of the
perform a true VTOL. Recent hybrid models that country the flight is planned to occur in (see Chapter
combine horizontal propellers for vertical take-off
and vertical propellers for horizontal flight on a 11 http://www.questuav.com (accessed 7 November 2017)
4, Duffy, Cunliffe et al. 2017). For instance, in many over large areas the VLOS requirements are often
countries the regulations stipulate that flights are limiting unless there are multiple take-off and land-
only allowed within visual line of sight (VLOS, 500 ing areas that one can use. In this situation a drone
m) or extended visual line of sight (750 m). Similarly survey might become cost prohibitive compared to
there often is a ceiling to how high it is permit- using alternative methods due to the need to move
ted to fly above the ground (400 ft). For surveys from one area to another.
CHAPTER 2
Typology and anatomy of drones
2.1 Multirotor movement of its control surfaces (ailerons, rudder

and elevator). For example, the ailerons that are
The most popular type of consumer-grade drones located near the tips of the main wings allow the
is a multirotor, which as the name implies, relies on aircraft to ‘roll’ left or right. Similarly, the rudder
one or more rotors that generate the lift required that is located on the ‘fin’ (or vertical stabilizer) of
for flight (Figure 2.1). The aircraft is able to move the tail assembly allows the aircraft to ‘yaw’ left or
forwards, backwards, left, and right, by adjusting right, whereas the elevator located on the tailplane
the relative speeds at which its rotors are spinning. (or horizontal stabilizer) allows the aircraft to ‘pitch’
The most common type of multirotor drone is the up or down (Marques and Ronch 2017).
quadcopter (four rotors). But there are hexacopters
(six rotors), octacopters (eight rotors), and other
configurations as well, each of which is used for 2.3 Hybrid VTOL
slightly different applications. For example, hexa- A hybrid VTOL is an aircraft design that combines
copters may be more appropriate than quadcopters the features of the multirotor and fixed wing aircraft
in situations where greater lifting power is needed. (Table 2.1). The two most common types of VTOL
Having a greater number of rotors also confers are the ‘tail-sitter’ and ‘tilt-rotor’. As the name sug-
redundancy to the aircraft such that there is a gests, a tail-sitter is usually a fixed wing aircraft that
higher chance of the aircraft making a successful rests on its tail when it is on the ground (Figure 2.3).
landing even if any one of its rotors should fail. At take-off, the vertically oriented rotors provide the
This book does not discuss the technical details of downward thrust to lift the aircraft off the ground.
aerodynamics as there are many existing resources The sophisticated flight controller carefully main-
that cover that topic (for example, Marques and tains the stability of the aircraft through constant
Ronch 2017). adjustments to the differential thrust of the two
rotors as well as the control surfaces on its wings.
2.2 Fixed wing aircraft When the aircraft has reached a safe altitude, the
operator can command it to transition to forward
This type of drone looks and functions just like a flight as the aircraft assumes a more conventional
manned aeroplane (Figure 2.2). It usually has one horizontal fixed wing profile.
or two horizontally oriented rotors that generate The tilt-rotor is the second type of VTOL air-
thrust to propel the drone forwards. The movement craft that is also based on a fixed wing airframe
of air over and under its wings generates the lift that (Figure 2.4). However, in this case the aircraft takes
allows a fixed wing drone to take off and maintain off from the ground in its typical conventional fixed
straight and level flight (Marques and Ronch 2017). wing profile with vertically oriented rotors that pro-
This aircraft changes the direction of its flight by the, vide downward thrust for lift. Once safely off the
13
Figure 2.1 A multirotor aircraft on a surveying mission. © Lian Pin Koh.
ground, the aircraft transitions into forward flight 2.5.1 Power source
by tilting its rotors 90 degrees to produce horizontal
Most drones are powered by rechargeable lithium-
thrust.
ion (Li-ion) or lithium-ion polymer (LiPo) batteries.
2.4 Features of multirotor vs fixed wing These are the same batteries commonly found
in computer laptops and other mobile devices.
vs VTOL drones
By packing large amounts of energy into a small
Multirotor and fixed wing drones have very differ- amount of material, Li-ion or LiPo batteries offer
ent capabilities that need to be carefully considered. excellent energy to weight ratios that make them a
The appropriate type of aircraft should be chosen for great energy source for drones (Figure 2.5).
each specific ecology or conservation application. However, the downside of Li-ion or LiPo batteries
The following table summarizes the most important is that they are rather unstable (Lisbona and
pros and cons of each aircraft type (Table 2.1). Snee 2011). Catastrophic failure can result from
overcharging, over-discharging, over-temperature,
short circuit, crushing, or penetration. The outcomes
2.5 Essential components
from mishandling or improper care can range from
In addition to the actual airframe and regardless of unreliable performance to vigorous battery fires.
whether it is a multirotor or fixed wing type, almost It is important to always closely adhere to the
all drones comprise a few key essential components. battery charging guidelines for each type of drone
These are invariably a power source, a flight con- model and battery charger. Most importantly, one
troller, and a ground control station. should never leave a charging battery unattended.
T Y P O L O G Y A N D A N ATO M Y O F D R O N E S 15
Figure 2.2 A fixed wing drone used for forest mapping in Malaysia. © Lian Pin Koh.
Table 2.1 Pros and cons of the three common types of conservation drones
Multirotor Fixed wing Hybrid VTOL
Launch locations Small take-off and landing area Requires large, open area for launch Small take-off and landing area
required; can launch in a wide range and landing. required; can launch in a wide range
of environmental conditions. of environmental conditions.
Payload Can support heavier payload (e.g. Can support lighter payload (e.g. less Can support lighter payload (e.g. less
several kilograms). than a kilogram), although gasoline than a kilogram).
powered fixed wings may carry heavy
payloads.
Flight time Shorter flight time (e.g. less Longer flight time (e.g. more than an Intermediate flight time (e.g. 30–60
than an hour). hour). minutes).
Pilot experience Can be flown with minimal or Substantial experience and training Intermediate pilot experience.
no training and experience. required to effectively pilot, especially
for take-off and landing.
Figure 2.3 A tail-sitter hybrid VTOL aircraft developed in Switzerland. © Wingtra.com.
Figure 2.4 A tilt-rotor hybrid VTOL aircraft developed in the United States. © BirdsEyeView Aerobatics.
2.5.2 Flight controller

There are a variety of different flight controllers
available on the market today that also range in
sophistication. The simplest flight controller is the
same system found in all remote-controlled model
aircraft. This would be a radio receiver that converts
command signals transmitted from a ground oper-
ator into proportional movements of the aircraft’s
control surfaces to direct its flight path.
The more typical flight controller used in drones
today is comprised of an array of electronic
components including the radio receiver, a space-
based radionavigation system such as the Global
Figure 2.5 A pair of LiPo batteries used in conservation drones. Positioning System (GPS) or Global Navigation
© Lian Pin Koh. Satellite System (GLONASS), and an Inertial
Measurement Unit (IMU). The IMU itself usually
is an integrated set of sensors that may include
Figure 2.6 An early version of the ArduPilotMega – an Open Source autopilot platform commonly used in conservation drones. © Lian Pin Koh.
an accelerometer, gyroscope, magnetometer, and 2.5.3 Ground control station

airspeed sensor. The modern flight controller also
A ground control station (GCS) is a ground-based
includes a microprocessor capable of processing
system that allows the operational control or pro-
data received from the IMU and other peripheral
gramming of a drone mission. In the context of
sensors to maintain straight and level flight of the
drones used in ecology and conservation applica-
aircraft without operator input (Figure 2.6). In more
tions, the GCS is most commonly a computer lap-
advanced systems, the flight controller is also
top running dedicated software. This GCS would
able to command the aircraft to fly a mission
also include a radio telemetry system, usually in
automatically to pre-programmed geographical
the form of a serial data modem, that transmits
coordinates. It is this feature that makes the modern
and receives flight data and commands between the
drone an extremely versatile tool for collecting
aircraft and the GCS.
remotely sensed data over a landscape or seascape.
Other functionalities of the GCS may include the
In fact, the cost of commercially available ready-
programming of a mission by specifying target
to-fly drones has dropped so dramatically in the
areas to survey, waypoints, and other essential
past five years that few drone operators actually
instructions such as flying height and speed
need to understand how a flight controller works
(Figure 2.7).
anymore.
Figure 2.7 The Mission Planner Ground Control Station used for programming drone missions. © Lian Pin Koh.
However, GCS are not restricted to the laptop 2.6 Summary

form, as other mobile devices such as tablets and
mobile phones can also run GCS applications to • The most popular aircraft types include mul-
perform similar functions. In fact, in some cases, tirotor, fixed wing, and ‘vertical-take-off-and-
such as when the operation is to be conducted in landing’ (VTOL) craft.
remote locations and without the ability to establish • The essential components of a drone are a power
a proper basecamp, having a more mobile GCS source, flight controller, and ground control sta-
would help to increase the chances of success. tion.
Recent advances in mobile GCS capabilities are • Each type of drone has its pros and cons. It is
increasing their versatility in conservation related important to use the appropriate type based on
applications, including the ability to ‘live’ stream the requirements and operating conditions of the
video data to the ground operator. mission.
CHAPTER 3
Sensors
3.0 Introduction or set of sensors will be suitable for their data

collection, we will approach the sensor issue from
Data collection for conservation purposes such the application end instead of vice versa.
as animal density and distribution or land-cover
change needs consideration of the various sensor
types that are available and their specifications 3.1 Sensor types
in relation to the data that are needed. For many
animal surveys a simple RGB (Red Green Blue) The number of sensors that can be fitted to drones
photo camera might suffice, but for crepuscular and/or have been specifically designed to be fitted
and nocturnal animals a thermal sensor might under drones is expanding rapidly (van Blyenburgh
be more applicable. Land-cover classification 2013). The compilation by van Blyenburgh (2013)
might be possible with RGB images, but could contains 406 sensors that range from the visual
be more accurate or allow for the classification of range to the microwave range. Several recent
more land-cover types when using more bands excellent overviews also provide lists of sensors
from the electromagnetic spectrum that capture used under drones, their specifications, and their
information outside of visible wavelengths (Figure specific applications (Everaerts 2008, Colomina and
3.1). Modified RGB cameras that allow data capture Molina 2014, Salamí, Barrado et al. 2014, Shahbazi,
in the near infrared or multispectral cameras are Théau et al. 2014, Linchant, Lisein et al. 2015, Pajares
often applied for the collection of such data. For an 2015, Christie, Gilbert et al. 2016, Toth and Jóźków
ecologist interested in vegetation structure it might 2016). Here we will restrict ourselves to those that
be sufficient to derive measurements from the DSM have specific conservation applications, although
(Digital Surface Model) and DTM (Digital Terrain several studies that focused on agriculture will also
Model) produced with photogrammetry software, be mentioned due to their potential application in
but in dense vegetation types such as tropical conservation. Due to the very rapid development in
rainforests using LiDAR (Light Detection and specific sensor models it is inevitable that many of
Ranging) would provide data that would allow for the specific sensor brands and models mentioned
more accurate results because there would be more in this chapter will be superseded shortly after
ground hits. Determining which sensor to use is not publication. But the information will still help the
always clear-cut and is often constrained by budget. reader to find further specifications on these sensors
In this chapter an overview of several sensors will that are useful for planning their own work. For
be given as well as specific applications that they such specific details readers are advised to consult
have been used for. Because most conservation the specific sources that have been used in the
researchers will start with a specific question and chapter, the above-mentioned overview papers, and
then subsequently will start exploring which sensor online search engines.
20
SENSORS 21
Visible short wave medium wave long wave

light infrared infrared infrared
gamma rays X rays ultraviolet rays infrared microwave radio waves
1pm 1nm 1um 1mm 1m 1km
high frequency low frequency
short wavelength long wavelength
Figure 3.1 An indication of the various parts of the electromagnetic spectrum and their wavelengths. © Perry van Duijnhoven.
Table 3.1 Ground sample distance and footprint for several sensors
Camera Canon S110 (RGB) Sony A6000 (RGB) Flir Vue (thermal)
Focal length (mm) 5.2 20 13

Number of pixels 4000 × 3000 6000 × 4000 640 × 512
Sensor size (mm) 7.6 × 5.7 23.5 × 15.6 9.0 × 6.8
Ground sample distance (GSD, cm)∗ 3.65 1.95 10.22
Footprint (m)∗ 146.2 × 109.6 117.5 × 78.0 69.2 × 52.3
∗ Above ground level (AGL) was 100 m for all calculations.
Table 3.2 Common sensors used in conservation research
Sensor type Brand examples Wavelength range∗
RGB (Red, Green, Blue) Canon, Nikon, Sony 390–700 nm

Multispectral Tetracam, Micasense, Parrot Sequoia, Slantrange, converted cameras 300–1100 nm
Hyperspectral Rikola Ltd, Headwall photonics, Resonon 380–2500 nm
Thermal FLIR, Xenics, Optris, TeAx-Technology 7–13.5 μm
LiDAR Riegl, Velodyne, YellowScan
∗ = The range is not for one camera specifically but indicates the total range that is generally available. Specific cameras might only use part of that range.
Because the focus of this chapter is on appli- discussed here. An important consideration when
cations and not on details of the various sensors using drones to collect data is to determine whether
we will only give a brief overview of common the system will need one sensor or multiple sensors
sensors that have been used in conservation work during the same flight, or whether data with
(Tables 3.1, 3.2, Figure 3.2). Details on specific multiple sensors can be collected over multiple
sensors will be given in the section on each sensor, flights. The latter is, for instance, likely to be an
but some common issues to think about will be option when flights are conducted for land-cover
Figure 3.2 Various sensors used under drones. Top left: GoPro © Andy Goodwin; top right: Canon RGB camera © Andy Goodwin; middle left:
Multispectral Parrot Sequoia © Serge Wich; middle right: Thermal sensor © Maisie Rashman; Lower left: A Vulcan Y6 multirotor with a
Velodyne HDL-32E LiDAR used in the DigiArt project. The DigiArt project has received funding from the European Union’s Horizon 2020 research
and innovation programme under grant agreement No 665066 © Frederic Bezombes; Lower right: Telemetry system tracking Noisy Miner birds
(inset) © Oliver Cliff.
SENSORS 23
classification because the objects of interest are AGL

motionless, but this approach is likely not applicable GSD = Pixel size ∗
Focal length
when studying animals that move. In that case
a system with multiple sensors to collect data at AGL is the above ground level in metres. For
the same time is likely necessary. There are not instance for a Canon S110 the focal length is 5.2 mm,
many systems on the market that are adapted the sensor resolution is 4000 × 3000 pixels, and the
for carrying multiple sensors at the same time. sensor size is 7.6 mm × 5.7 mm. Dividing either 7.6
Some others allow for multiple interchangeable by 4000 or 5.7 by 3000 gives the pixel size 0.0019 mm.
sensors, but with only one sensor on board during GSD when flying at 100 m (AGL) is then:
a flight.1 Although such off-the-shelf systems might 100
work well for certain applications, there might be GSD = 0.0019 ∗ = 0.0365
5.2
situation in which researchers will need to build
their own systems or order customized systems in The image footprint on the ground can then be
which researchers can choose sensors themselves calculated by multiplying the number of pixels by
and design the system around a sensor or set of the size of each pixel. In this case the width of the
sensors. Another important aspect is how images image is 146 m (4000 × 0.0365) and the height of
will be georeferenced and, in the case of multiple the image is 109.5 (3000 × 0.0365).
sensors, how images from the different sensors This assumes that the camera is placed top facing
will be co-registered. Although some sensors come forward. This is the usual camera position in drones
with a built-in GPS unit and IMU (see Chapter because it allows the distance between the flight
2 for details), many do not. There are several lines to be as wide as possible for a given side over-
workflows possible in which the images are either lap between the images. If the camera would be side
automatically geotagged or geotagged with the facing forward flight lines would have to be closer
aid of post-flight software on the ground control to each other for a similar side overlap between
station or other software packages.2 Some of these images. The two reasons to avoid this are that a) a
workflows also allow for the insertion of attitude fixed wing system might not be able to make tight
data from the autopilot (e.g. through Pix4D3 : details enough turns and building in overshoot to allow
in Chapter 7). the drone to line up with the flight lines is not
A key aspect of any mission to collect data should using flight time effectively, and b) it means flight
be whether the sensor and flying height combi- distance is larger to cover the same surface area
nation will yield images of sufficient resolution to on the ground which again is not effective. When
discern the object of interest, or in the case of land- considering ground sample distance of images it is
cover mapping, vegetation types of interest. Ground important to keep in mind that this differs widely
control software such as Mission Planner4 automat- among sensors (Table 3.1) and that, for instance,
ically calculates the ground resolution (often called thermal sensors generally have a much lower num-
GSD (Ground Sample Distance) which is the dis- ber of pixels than RGB cameras.
tance between the centre of two adjacent pixels in Examining the relationship between ground res-
an image) once the correct data for the camera have olution and flying altitude is important so that a
been entered or a sample image has been uploaded. researcher can assure that a specific sensor and fly-
However, it is straightforward to calculate this with- ing altitude combination will provide the ground
out software with the following formula: resolution needed. Obviously, smaller objects need a
higher ground resolution, which can be obtained by
flying lower, but whether that is an option depends
on a range of factors such as vegetation height,
1 eBee: https://www.sensefly.com/drones/ebee-ag.html the terrain, and how low or high one needs to fly
2 http://www.geosetter.de. to cover sufficient area for a survey. In addition,
3 https://pix4d.com/ (accessed 7 November 2017)
4 http://ardupilot.org/planner/docs/mission-planner- when flying low with a fast flying drone there is
overview.html (accessed 7 November 2017) a chance for blurred images if the shutter speed
is too slow compared to the speed at which the have fitted under drones, but common ones include
drone is flying at. A suggested rule of thumb to Canon, Sony, and Nikon. Although there are a large
avoid motion blur in images due to the speed of the range of cameras that have been used on drones
drone is that the distance covered during the time it is important to carefully determine whether the
the shutter is open to take the image is smaller than sensor size and number of pixels will be sufficient
half the size of the pixel (Bosak 2013). Using this rule for the particular research aim, because these factors
might still lead to images that are not sharp if there in combination with the flying height and focal
are sudden motions of the drone during the image length are important in determining the resolution
capture. Therefore a simpler option is to always of an image (see section 7.2). Another important
use a very fast shutter speed and use the ‘Shutter consideration when a drone is built and not
Priority’ mode with the ISO setting at auto so that purchased off-the-shelf is by which mechanism the
the exposure can be compensated when the aperture camera can capture images at a pre-defined interval.
reaches its maximum size. Generally a shutter speed This does not only apply to RGB cameras, but also
of 1/1000 sec works well during conditions when to multispectral and hyperspectral cameras. There
there is relatively low light (early and late in the are several common options for this depending
day and in cloudy and overcast conditions). A faster on the camera. First, cameras might have built-in
shutter speed can be used during brighter condi- time interval settings, but these often are restricted
tions. If flying lower is not an option it is sometimes to a few intervals. Cameras with a GNSS often
possible to use the optical zoom on a sensor, but also allow for the interval to be set at a flight
using this might come at the cost of more vibration distance interval (e.g. Parrot Sequoia). Second, some
of the camera and hence blurred images. Examin- cameras allow for the installation of certain apps
ing published papers and reports, and checking the on the camera that can be used to set the time
online DIY drone communities,5 to determine what interval that images will be taken. Third, specifically
previous researchers have been using, often helps for Canon cameras there is the opportunity to
with deciding which specifications are required to use CHDK (Canon Hack Development Kit).6 This
obtain sufficient ground detail. Related to this is allows for putting new firmware on the SD card
the question of whether the sensor will need to in the camera so that self-written scripts can be
be pointed straight down or at an angle. Whereas loaded that allow for images to be taken at a set
for mapping a sensor that faces straight down is interval. It is not known whether this will remain
required, this might not be the case for spotting available for future Canon cameras. Fourth, several
animals that are often found under trees in which autopilot systems (e.g. the 3DRobotics Pixhawk) can
case a camera at an oblique angle might be more trigger cameras at set intervals or set flight distances
beneficial. Experience has also taught us that it is with the ground control software.7 Depending on
important to set the camera focus to infinity and not the particular camera brand, different types of
on auto-focus because that will often lead to blurred specific cables and camera triggers will need to
images. To further improve exposure it is beneficial be used (e.g.8 ). Fifth, there are infrared triggers
to use the matrix (or pattern) metering mode when that can be connected to the autopilot and that
this is available so that the camera is using the full can trigger the camera through its infrared sensor
extent of its field to calculate the exposure setting. (e.g. stratosnapper9 ). Alternatively some drones
have triggers connected to a servo that physically
3.1.1 Visible spectrum sensors (RGB) presses the shutter button. Once the images have
been obtained these can be used for several pur-
Due to the, initially, high costs and relatively large
poses,
weight of multispectral and hyperspectral cameras
the most common sensors have been standard
consumer cameras that obtain imagery in the visible 6 http://chdk.wikia.com (accessed 7 November 2017)
part of the spectrum (Table 3.3; Lebourgeois, Bégué 7 Mission Planner: http://ardupilot.org/planner/docs/
et al. 2008, Rump, Zinke et al. 2011). There have mission-planner-overview.html
8 http://www.seagulluav.com/ (accessed 7 November
been a large number of brands that researchers
2017)
9 http://littlesmartthings.com/stratosnapper2/ (accessed
5 www.diydrones.com 7 November 2017)
Table 3.3 Applications, sensor types, and drone systems
Application Sensor(s) Species/topic Aim Drone (model if Detection Source

known)* method
Animals (birds)
Animal detection/birds RGB (Sony Nex5 camera) Black kites (Milvus migrans) Conspicuousness of bird MR (Hexacopter) Manual (Canal, Mulero-Pázmány
nests et al. 2016)
Bird detection RGB (GoPro Hero 4) Tristan Albatross (Diomedea dabbenena) Population count MR (DJI Phantom 2) Manual (McClelland, Bond et al.
2016)
Bird detection and RGB (Sony RX-1, Various waterfowl species Determining disturbance FW (UAVER Avian-P, Manual (McEvoy, Hall et al. 2016)
disturbance mvBlueCOUGAR-X, Sony on birds with various drone Skylark II, Drone Metrex
A7-R, Phase 1, GoPro Hero) models Topodrone-100), MR (DJI
Phantom, FoxTech
Kraken-130)
Animal detection/birds RGB (Canon EOS M) Lesser frigatebird (Fregata ariel), crested Counting animals MR (X8 by 3D Robotics) Manual (Hodgson, Baylis et al.
tern (Thalasseus bergii), royal penguin and FW (FX79 airframe) 2016)
(Eudyptes schlegeli)
Animal detection/birds RGB (GoPro Hero 1 and 3) Osprey (Pandion haliaetus), bald eagle Determining disturbance of MR (Draganflyer X-4) Manual (Junda, Greene et al. 2016)
(Haliaeetus leucocephalus), ferruginous ground and aerial surveys
hawk (Buteo regalis), red-tailed hawk
(B. jamaicensis)
Animal detection/birds RGB (GoPro Hero 1 and 3) Osprey (Pandion haliaetus), bald eagle Surveying raptor nests Multirotor (Draganflyer Manual (Junda, Greene et al. 2015)
(Haliaeetus leucocephalus), ferruginous X-4)
hawk (Buteo regalis), red-tailed hawk
(B. jamaicensis)
Bird count RGB (Canon Powershot Common tern (Sterna hirundo) Colony count FW (AI-Multi UAs) Automatic (Chabot, Craik et al. 2015)
S90)
Animal detection/birds RGB (DJI Phantom 2 Hooded crow (Corvus cornix) Check nest contents MR (DJI Phantom 2 Vision) Manual (Weissensteiner, Poelstra
built-in camera) et al. 2015)
Animal detection/birds RGB (Panasonic LUMIX G5) Adélie penguin (Pygoscelis adeliae) Measuring reaction to MR (MK ARF Okto XL) Manual (Rümmler, Mustafa et al.
drones 2015)
Animal detection/birds RGB (Canon EOS M) Gentoo penguin (Pygoscelis papua) Assessing detectability MR (Cinestar 6) Manual (Goebel, Perryman et al.
2015, Ratcliffe, Guihen
et al. 2015)
Animal detection/birds RGB (Sony NEX-5, Canon Gentoo penguin (Pygoscelis papua), Basic detection, wildlife 3 MRs (MK-OktoXL, Manual (Goebel, Perryman et al.
EOS-M and the Olympus chinstrap penguin (Pygoscelis antarctica) reaction, effectiveness APQ-18 and APH-22) 2015)
E-P1) comparison of 3 aircraft
and 3 cameras
(continued)
Table 3.3 Continued

known)* method
Animal reaction/birds RGB (GoPro Hero 3) Mallard (Anas platyrhynchos), greater Assessing behavioral MR (Phantom, Cyleone, Manual (Vas, Lescroël et al. 2015)
flamingo (Phoenicopterus roseus), common reactions to a multirotor France)
greenshank (Tringa nebularia) drone
Animal detection/birds Electro-optical/infrared Various seabird species, turtles, seals, Assessing detectability 2 FWs (MQ-9 Predator B Manual (Brooke, Graham et al.
(EO/IR) sensor and cetaceans and Puma AE) 2015)
Animal detection/birds RGB (Canon S100 and Various waterbird species Assessing detectability and Petrol-powered VTOL Manual (Dulava, Bean et al. 2015)
GoPro Hero3) behavioral reaction to UAS (Honeywell RQ-16 T-Hawk)
and FW (AeroVironment
RQ-11A)
Animal detection/birds RGB (GoPro Hero 2 and White stork (Ciconia ciconia) Assess power lines risk on FW (Easy fly St-330) Manual (Mulero-Pázmány, Negro
Panasonic LX3) birds et al. 2014)
Bird detection EO/thermal sensor and Greater sage-grouse (Centrocercus Counting and monitoring FW (Raven RQ-11A) Automatic (Hanson,
RGB (GoPro Hero 3 black urophasianus) Holmquist-Johnson et al.

edition) 2014)
Animal detection/birds RGB (GoPro 2) Steller’s sea eagle (Haliaeetus pelagicus) Nest check and birds MRs (APM and Naza-M Manual (Potapov, Utekhina et al.
reaction based) 2013)
Animal detection/birds RGB (Panasonic Lumix Black-headed gull (Chroicocephalus Detect changes in breeding FW (Multiplex Twin Star II, Manual (Sardà-Palomera, Bota
FT-1) ridibundus) populations no autopilot) et al. 2012)
Animal detection/birds RGB (Pentax Optio A20) Canada geese (Branta canadensis) and Comparing photographic FW (CropCam) Manual and (Chabot and Bird 2012)
snow geese (Chen caerulescens) counts from repeated automatic
flights to repeated visual
ground counts
Animal detection/birds Thermal Sandhill crane (Grus canadensis) Assessing detectability FW (Raven RQ11) Manual and (Hutt 2011)
automatic
Animal detection/birds RGB (Evoltmodel 420) Red knot (Calidris canatus), egret species Studying bird detection MR (Nova 2) Manual (Brush and Watts 2008)
(Ardea alba, Bubulcus ibis, Egretta sp.),
pelican (Pelecanus sp.), wood stork
(Mycteria americana)
Animal detection/birds RGB (Canon Elura 2) and White ibis (Eudocimus albus), Egret (Egretta Basic detection FW (FoldBat) Manual (Jones, Pearlstine et al.
near infrared sp.), wood stork (Mycteria americana) 2006)
Animals (aquatic)
Animal behaviour RGB (proprietary DJI Green turtle (Caretta caretta) Investigating green turtle MR (DJI Phantom 3) Manual (Schofield, Papafitsoros
camera) and fish cleaning et al. 2017)
Animal sex ratio RGB (proprietary DJI Green turtle (Caretta caretta) Determining the sex ratio MR (DJI Phantom 3) Manual (Schofield, Katselidis et al.
determination camera) of green turtles in press)
Animal density/marine RGB (GoPro Hero 3+) Blacktip reef sharks (Carcharhinus Density assessment MR (DJI Phantom 2) Manual (Kiszka, Mourier et al.
melanopterus) and pink whiprays 2016)
(Pateobatisfai)
Animal detection/fish RGB (GoPro 3+) Chinook salmon (Oncorhynchus Counting salmon spawning MRs (Aeryon Scout and Automatic (Groves, Alcorn et al. 2016)
tshawytscha) sites Mikrokopter)
Animal RGB (Canon S100) Estuarine crocodiles (Crocodylus porosus) Basic detection FW (Bormatec-MAJA) Manual (Evans, Jones et al. 2015)
detection/herptiles
Animal Electro-optical/infrared Sea turtles (Caretta caretta) Basic detection FWs (MQ-9 Predator B and Manual (Brooke, Graham et al.
detection/herptiles (EO/IR) camera Puma AE) 2015)
Animal detection RGB (GoPro Hero 3 Black Kemp’s ridley sea turtle (Lepidochelys Abundance and MRs (DJI Phantom 1 and 2) Manual (Bevan, Wibbels et al.
Edition) kempii) movements of turtles 2015)
Animal RGB (Sony NEX-5, Canon Antarctic fur seal (Arctocephalus gazella) Basic detection, wildlife MRs (MK-OktoXL, APQ-18 Manual (Goebel, Perryman et al.
detection/marine EOS-M and the Olympus and leopard seals (Hydrurga leptonyx) reaction, effectiveness and APH-22) 2015)
mammals E-P1) comparison of 3 UAVs and
3 cameras
Detection and RGB (Nikon D300) Spotted (Phoca largha) and ribbon seals Basic detection and FW (ScanEagle) Manual (Moreland, Cameron et al.
disturbance (Histriophoca fasciata) disturbance evaluation 2015)
Animal RGB (Sony HDR-CX760 Grey seals (Halichoerus grypus) and Basic detection, photo-ID MRs (DJI 450, Cinestar 6, Manual (Pomeroy, O’Connor et al.
detection/marine and PJ650) harbour seals (Phoca vitulina) and measurement, animal Vulcan 8) 2015)
mammals reaction, comparison of 3
UAVs
Animal Electro-optical/infrared Hawaiian monk seals (Neomonachus Basic detection FWs (MQ-9 Predator B and Manual (Brooke, Graham et al.
detection/marine (EO/IR) camera schauinslandi) Puma AE) 2015)
mammals
Animal RGB (Olympus E-PM2) Killer whales (Orcinus orca) Basic detection MR (APH-22) Manual (Durban, Fearnbach et al.
detection/marine 2015)
mammals
Animal RGB (GoPro cameras) Bowhead whales (Balaena mysticetus) Identification of bowhead FW (Brican TD100E) Manual (Koski, Gamage et al.
detection/marine whales and behavioral 2015)
mammals reaction to UAVs
(continued)
Table 3.3 Continued

known)* method
Animal RGB (Nikon® D90) Dugongs (Dugong dugon) Basic detection FW (ScanEagle) Manual (Hodgson, Kelly et al. 2013,
detection/marine Brooke, Graham et al.
mammals 2015)
Animal RGB (Nikon® D90) Sea turtles (Caretta caretta) Basic detection FW (ScanEagle) Manual (Hodgson, Kelly et al. 2013)
detection/herptiles
Animal detection/fish RGB (Canon EOS Kiss X) Wild pacific chum salmon Aerial census of adult Helicopter: gasoline-engine Manual (Kudo, Koshino et al. 2012)
(Oncorhynchus keta) chum salmon using (Voyager GSR260)
Animal RGB (IXUS 120) Green turtles (Chelonia mydas) Assess turtles’ spatial FW (Swinglet CAM) Manual (Winter 2012)
detection/herptiles distribution
Animal RGB (EVolt camera) Alligator (Alligator mississippiensis) Basic detection FW (University of Florida’s Manual (Watts, Perry et al. 2010)
detection/herptiles Nova 2)
Animal RGB (Video camera) Whale-like targets, harbour seals Basic detection and factors FW (Insight A-20) Manual (Koski, Allen et al. 2009,
detection/marine (Phoca vitulina), California sealions influencing detection rate Koski, Abgrall et al. 2010,
mammals (Zalophus californianus) Koski, Thomas et al. 2013,
Koski, Gamage et al. 2015)
Animal RGB (CMOS camera and Alligator (Alligator mississippiensis) Basic detection FW (MLB FoldBat) Manual (Jones, Pearlstine et al.
detection/herptiles Canon Elura 2) and 2006, Watts, Perry et al.
near-infrared sensor 2010)
Animal RGB (CMOS camera and Florida manatees (Trichechus manatus Basic detection FW (MLB FoldBat) Manual (Jones, Pearlstine et al.
detection/marine Canon Elura 2) and latirostris) 2006, Martin, Edwards
mammals near-infrared sensor et al. 2012)
Animals (terrestrial mammals)
Animal RGB (Nikon D3X) Caribou (Rangifer tarandus caribou) Assessing detectability FW (TD100E) Manual (Patterson, Koski et al.
detection/terrestrial using caribou targets 2016)
mammals
Animal RGB (Canon S100) Sumatran orangutans (Pongo abelii) Orangutans nest detection FW (Skywalker 2013) Manual (Wich, Dellatore et al.
detection/primate 2016)
Animal X American black bear (Ursus americanus) Behavioral and MR (Iris, 3DR) X (Ditmer, Vincent et al.
reaction/terrestrial physiological reaction to 2015)
mammals drone
Animal impact on RGB (Canon S100) Eurasian beaver (Castor fiber) Impact of beavers on MR (Y6, 3DR) Manual (Puttock, Cunliffe et al.
habitat ecosystem 2015)
Animal RGB (Canon Powershot Chimpanzees (Pan troglodytes) Chimpanzees nest FW (Maja, Bormatec) Manual (van Andel, Wich et al.
detection/primate SX230 HS) detection 2015)
Animal Thermal (Thermoteknix Giraffes (Giraffa camelopardalis) Rhinoceros anti-poaching FW (Easy Fly St-330) Manual (Mulero-Pázmány, Stolper
detection/terrestrial Micro CAM) et al. 2014)
mammals
Animal RGB (Panasonic Lumix Rhinoceros (Diceros bicornis and Rhinoceros anti-poaching FW (Easy Fly St-330) Manual (Mulero-Pázmány, Stolper
detection/terrestrial LX-3, GoPro Hero2), Ceratotherium simum) et al. 2014)
mammals Thermal (Thermoteknix
Micro CAM)
Animal RGB (Ricoh GR3) African elephant (Loxodonta africana) Basic detection and FW (Gatewing X100) Manual (Vermeulen, Lejeune et al.
detection/terrestrial animals reaction 2013)
mammals
Animal RGB (Canon IXUS 220 HS Sumatran orangutans (Pongo abelii) Assess the potential of FW (Bixler with APM) Manual (Koh and Wich 2012)
detection/primate and Pentax Optio WG-1 drone for conservation
GPS)
Animal RGB (Canon IXUS 220 HS; Sumatran elephants (Elephas maximus Demonstrate the potential FW (Bixler with APM) Manual (Koh and Wich 2012, Wich,
detection/terrestrial Pentax Optio WG-1 GPS) sumatranus) of UAVs for environmental Dellatore et al. 2016)
mammals and conservation
applications
Animal RGB (Pentax Opio A10 and Black bear (Ursus americanus), woodland Basic detection FW (CropCam) Manual (Chabot 2009)
detection/terrestrial A40) caribou (Rangifer tarandus caribou),
mammals white-tailed deer (Odocoileus virginianus),
grey wolf (Canis lupus), North American
beaver (Castor canadensis) lodges
Animal RGB Bison (Bison bison) Population survey of bison FW (Nova 2) Manual (Watts, Bowman et al.
detection/terrestrial 2008, Watts, Perry et al.
mammals 2010)
Animal RGB (Evoltmodel 420) Bison (Bison bison) Design of georeferencing FW (Nova 2) Manual (Wilkinson 2007,
detection/terrestrial techniques in the national Wilkinson, Dewitt et al.
mammals bison range 2009)
Automated detection
Automated detection Thermal (SenseFly LLC, Grey seal (Halichoerus grypus) Wildlife surveys FW (SenseFly eBee) Automatic (Seymour, Dale et al. 2017)
Thermomapper) and RGB
(Canon S110)
Automated detection Thermal (FLIR Tau 640) Cows (Bos taurus) and humans (Homo Wildlife survey and MR (Y6) Automatic (Longmore, Collins et al.
sapiens) anti-poaching 2017)
(continued)
Table 3.3 Continued

known)* method
Automatic animal Thermal (FLIR Tau 2-640) Koala (Phascolarctos cinereus) Wildlife automated MR (S800 EVO) Automatic (Gonzalez, Montes et al.
detection/terrestrial detection 2016)
mammals
Automatic animal X Dugong (Dugong dugon) Using deep convolutional X Automatic (Maire, Alvarez et al. 2015)
detection/dugong neural networks for
detection
Automatic animal Thermal (Flir Tau640) and American bison (Bison bison), red fallow Species detection Helicopter (Responder Automatic (Chrétien, Théau
detection/terrestrial RGB (Nikon D7000) deer (Dama dama),grey wolves (Canis INGRobotic Aviation, et al. 2015, Chrétien,
mammals lupus), white-tailed deer (Odocoileus Sherbrooke, QC, Canada) Théau et al. 2016)
virginianus) and elks (Cervuscanadensis)
Automatic animal Thermal (Tamarisk 640) Common hippopotamus (Hippopotamus Species detection FW (Falcon) Automatic (Lhoest, Linchant et al.
detection/terrestrial amphibius) 2015)
mammals
Automatic animal RGB (Sony Nex-5) Black-faced spoonbill (Platalea minor) Automating spoonbill FW (CropCam) Automatic (Liu, Chen et al. 2015)
detection/birds detection to facilitate
surveys
Automatic animal RGB (Olympus PEN Mini Pocket gopher (Thomomys talpoides) Pocket gopher mound FW (RQ-84Z Aerohawk) Automatic (Whitehead, Hugenholtz
detection/terrestrial E-PM1) feature detection et al. 2014)
mammals
Automatic animal RGB (Sony Nex-5 and Common gull (Larus canus) Automatic bird count, MRs (Falcon 8 and Automatic (Grenzdörffer 2013)
detection/birds Olympus Pen E2) determination of canopy MD4-1000)
height and DSM
Automatic animal RGB with polarizing filter Dugongs Automatically identify FW (ScanEagle) Automatic (Maire, Mejias et al. 2013)
detection/marine (Nikon 12 megapixel) marine mammals
mammals
Automatic animal Thermal (FLIR Tau 640) Roe deer (Capreolus capreolus) Rescue fawns from MR (Falcon 8) Automatic (Israel 2011)
detection/terrestrial mowing machines
mammals
Automatic animal RGB (Firefly MV USB) Whales Algorithms for object MR Automatic (Selby, Corke et al. 2011)
detection/marine identification and tracking
mammals
Automatic animal Multispectral (MANTIS-3) Belugas and other species of baleen Spectral detection of FW Automatic (Schoonmaker, Wells et al.
detection/marine whales marine mammals 2008)
mammals
Automatic animal RGB (Canon Elura 2 scan Various bird species and decoys Assess accuracy of FW (no autopilot) Manual (Abd-Elrahman 2005)
detection/birds video recorder and CMOS automatic identification
analog video chip)
Telemetry
Automatic animal BeagleBone VHF receiver Bicknell’s thrush (Catharus bicknelli) and Detecting VHF tagged birds MR (Sky Hero Spyder X8) Automatic (Tremblay, Desrochers et al.
detection/birds Swainson’s thrush (C. ustulatus) with a receiver on the 2017)
drone
Automatic animal Low-power RF tag Noisy miners (Manorina melanocephala) Real-time autonomous MR (Falcon 8) Automatic (Cliff, Fitch et al. 2015)
detection/birds localization localization
Automatic animal Tag localization Animals in general Radio tag simulation X Automatic (Posch and Sukkarieh
detection evaluation 2009)
Fire detection
Fire detection Multispectral X Fire detection and FW (Ikhana) Automatic (Ambrosia, Wegener et al.
(AMS-Wildfire sensor) monitoring 2011)
Fire detection Hyperspectral/thermal X Development of a system FW (based on Dornier Automatic (Rufino and Moccia 2005,
for fire detection and DO27) Martı́nez-de-Dios, Merino
monitoring et al. 2007, Merino,
Caballero et al. 2012)
Fire detection RGB/hyperspectral/fire X Fire detection and Helicopter and airship Automatic (Rufino and Moccia 2005,
sensor monitoring Karma Martı́nez-de-Dios, Merino
et al. 2007, Merino,
Caballero et al. 2012)
Land-cover studies
Land-cover RGB (Canon S100) X Vegetation structure MR (Y6, 3DR) Automatic (Cunliffe, Brazier et al.
classification mapping 2016)
Land-cover Hyperspectral (PIKA II) X Vegetation classification FW (Arcturus T-16) Automatic (Mitchell, Glenn et al.
classification 2016)
Land-cover RGB (GoPro Hero 3) X Grassland fraction MR (DJI Phantom 2) Automatic (Chen, Yi et al. 2016)
classification measurements
Land-cover RGB (Sony a6000) X Mapping and monitoring MR (Spyder PX8Plus 1000) Automatic (Fraser, Olthof et al. 2016)
classification of tundra vegetation
(continued)
Table 3.3 Continued

known)* method
Land-cover features RGB (Canon S110) X Top-of-canopy height and FW (Kestrel, Linn Automatic (Messinger, Asner et al.
above ground carbon Aerospace) 2016)
density
Land-cover Side-looking synthetic X Trees measurement MR (DJI Phantom 2) Automatic (Li and Ling 2016)
classification aperture radar (SAR)
Land-cover RGB (Sony NEX-5) X Forest monitoring (canopy MR (MD4-1000) Automatic (Zhang, Hu et al. 2016)
classification height and closure)
Land-cover Modified RGB (NIR) Canon Various peatland species Vegetation classification MR (MicrodroneMD4–200) Semi-automatic (Lehmann, Münchberger
classification PowerShot SD780 IS and CH4 fluxes et al. 2016)
Land-cover features RGB and modified RGB Various plant species Invasive plant species FW (Gatewing X100) Automatic (Michez, Piégay et al.
multispectral detection and health 2016)
Land-cover RGB (Mobius and GoPro x Map fish nursery areas MR (propriety prototype semi-automatic (Ventura, Bruno et al.
classification Hero 3) quadcopter) 2016)

Land-cover features Modified RGB camera Olive trees Canopy reconstruction FW (mX-SIGHT) Automatic (Dı́az-Varela, de la Rosa
(Panasonic Lumix et al. 2015)
DMC-GF1)
Land-cover RGB and modified RGB X Tree species identification FW (Gatewing X100) Automatic (Lisein, Michez et al. 2015)
classification
Land-cover features Modified RGB multispectral X Forest characteristics FW (SenseFly eBee) Automatic (Puliti, Ørka et al. 2015)
(Canon S110 near
infra-red)
Land-cover RGB (GoPro) and thermal X Polar region mapping MR (Tarot FY690) Automatic (Stuchlı́k, Stachoň et al.
classification (Optris PI) 2015)
Land-cover features RGB (Canon ELPH 520 HS) X Measure forest canopy MR Automatic (Zahawi, Dandois et al.
structure 2015)
Land-cover RGB (GOPRO 3+) X 3D models of trees MR (Iris, 3DR) Automatic (Gatziolis, Lienard et al.
classification 2015)
Land-cover features RGB X Map canopy gap FW (Carolo P200) Automatic (Getzin, Nuske et al. 2014)
Land-cover RGB and modified RGB X Vegetation classification MR (MD4-200) Automatic (Gini, Passoni et al. 2014)
classification multispectral
Land-cover features Modified RGB camera Orchards Tree height FW (mX-SIGHT) Automatic (Zarco-Tejada, Diaz-Varela
(Panasonic Lumix et al. 2014)
DMC-GF1)
Land-cover RGB (Canon S90) X Classify wetland vegetation FW (AI-multi) Automatic (Chabot and Bird 2013,
classification types to investigate Chabot, Carignan et al.
relationships between 2014)
habitat and breeding least
bitterns
Land-cover LiDAR X Tree DBH MR Automatic (Chisholm, Cui et al. 2013)
classification
Land-cover Hyperspectral X Land surface reflectance Not stated Automatic (Duan, Li et al. 2013)
classification
Land-cover Modified RGB camera Waterlogged bare peat, Sphagnum spp., Peat bog vegetation MR (Ifgicopter and MD Automatic (Knoth, Klein et al. 2013)
classification (Canon IXUS 400 and Eriophorum vaginatum and Betula pubescens 4-1000)
Panasonic LUMIX LX-3)
Land-cover features RGB X Assess forest understory FW (Carolo P200) Automatic (Getzin, Wiegand et al.
floristic biodiversity 2012)
Land-cover RGB (Pentax Optio A40) X Clustering tree species Multirotor (MD4-200) Automatic (Gini, Passoni et al. 2012)
classification and Modified RGB (Sigma
DP1)
Land-cover RGB (Canon PowerShot Tree species Automatically identify tree Remote controlled Manual (Peck, Mariscal et al. 2012)
classification A650) genera/species helicopter (no autopilot)
Land-cover mapping RGB (Panasonic Lumix Lesser kestrel (Falco naumanni) Mapping habitat under FW (ST-model Easy Fly) Automatic (Rodrı́guez, Negro et al.
LX-3) kestrel flight paths 2012)
Land-cover Multispectral X Land-cover classification Helicopter Automatic (De Biasio, Arnold et al.
classification (Condor-1000 MS5 5 CCD (CAMCOPTER,S-100 UAV) 2010)
multi-channel camera)
Land-cover RGB (Olympus stylus) X Measuring bare areas FW (RnR APV3) and Automatic (Breckenridge and Dakins
classification helicopter (Raptor Model 2011)
30)
Land-cover features Modified RGB multispectral Phragmites australis Invasive plant species FW (AggieAir) Automatic (Zaman, Jensen et al. 2011)
detection
Land-cover RGB Various plant species Feature extraction FW (based on J3 Cub) Automatic (Bryson, Reid et al. 2010)
classification
(continued)
Table 3.3 Continued

known)* method
Land-cover RGB (Canon SD 550 and Grass, shrub, bare Species level identification/ FW (MLB BAT 3) Automatic (Laliberte and Rango 2009,
classification Canon SD 900) vegetation classification Rango, Laliberte et al.
2009, Laliberte, Goforth
et al. 2011)
Land-cover Thermal olive orchards Canopy conductance and Helicopter (Benzin Automatic (Berni, Zarco-Tejada et al.
classification crop water stress index Acrobatic, Vario, Germany) 2009a)
Land-cover RGB (Canon PowershotG5 X Classify Riparian Radio controlled paraglider Automatic (Dunford, Michel et al.
classification and Canon EOS 5D) vegetation types (‘Pixy’ drone) 2009)
Land-cover Multispectral (MCA-6), X Vegetation monitoring Helicopter (frame from Automatic (Berni, Zarco-Tejada et al.
classification thermal camera (LAI, chlorophyll content, Benzin Acrobatic, Vario, 2009b)
(Thermovision A40M) and water stress detection) Germany)
Land-cover RGB (A101fc 1280x1024 Vineyards Creating vigour map of FW (RCATS/APV-3) Automatic (Johnson, Herwitz et al.
classification Bayer) and hyperspectral vineyards 2003)

camera
Miscellaneous
Songbird recordings Acoustic sensor (Zoom H1) Several song bird species in the US Determining the feasibility MR (DJI Phantom 2) Manual (Wilson, Barr et al. 2017)
of songbird recordings with
a drone
Vegetation monitoring Hyperspectral X Vegetation index MR (MK-OktoXL) Automatic (Aasen, Burkart et al. 2015)
Water sampling Water sampling equipment X Testing feasibility of aerial MR (AscTec Firefly) Manual (Ore, Elbaum et al. 2015)
water sampling
Miscellaneous/precision RGB (Panasonic Lumix X Crop biomass estimation MR (Flightcopter) Automatic (Jannoura, Brinkmann et al.
agriculture DMC-GF1) (NGRDI and LAI) 2015)
Vegetation monitoring RBG GoPro Hero3 and Vineyard Leaf area index MR (OnyxStar BAT-F8, Automatic (Kalisperakis, Stentoumis
VNIR hyperspectral sensor Altigator, Belgium) et al. 2015)
Vegetation monitoring Hyperspectral Maize, potato, and sunflower Leaf area index (Duan, Li et al. 2014)
Miscellaneous RGB (GoPro Hero 3) Green algae (Cladophora glomerata) Algal cover mapping MR Automatic (Flynn and Chapra 2014)
Vegetation monitoring Multispectral (ADC Air X Assessment of green leaf Helicopter (Predator Automatic (McGwire, Weltz et al.
digital camera) cover (NDVI) Gasser) 2013)
Vegetation monitoring Hyperspectral and Vineyard Estimating leaf carotenoid FW (mX-SIGHT and Viewer) Automatic (Zarco-Tejada,
multispectral content in vineyards Guillén-Climent et al.
2013)
Vegetation monitoring Multispectral (Tetracam Vineyard Vineyard management by MR (VIPtero, a modified Automatic (Primicerio, Di Gennaro
ADC-lite) NDVI Mikrokopter Hexa-II) et al. 2012)
Vegetation monitoring SAR X Development of SAR for FW (Aludra MK1) Automatic (Koo, Chan et al. 2012)
environmental monitoring
Miscellaneous/precision Multispectral X Crop biomass estimation Helicopter (based on Automatic (Honkavaara, Saari et al.
agriculture Mikado Logo 600) 2013)
Collection of blow Blow sample sensor Various marine mammal species Obtaining blow samples Helicopter Manual (Acevedo-Whitehouse,
samples Rocha-Gosselin et al.
2010)
Miscellaneous RGB, modified RGB and Antarctic moss mapping moss beds in MR Automatic (Lucieer, Robinson et al.
thermal camera Antarctica 2010, Lucieer, Robinson
et al. 2012, Turner, Lucieer
et al. 2012, Lucieer,
Malenovský et al. 2014,
Lucieer, Turner et al. 2014,
Turner, Lucieer et al. 2014)
Vegetation monitoring Multispectral Various types of vegetation Classify different land types Helicopter (CAMCOPTER Automatic (De Biasio, Arnold et al.
(Condor-1000 MS5 5 CCD) and calculate vegetation S100) 2010)
indices (NDVI)
Vegetation monitoring Modified RGB camera X Crop Monitoring (GNDVI FW (Vector-P) Automatic (Hunt, Hively et al. 2010)
and LAI)
Vegetation monitoring Modified RGB cameras Spring barley Environmental monitoring FW (CropCam) Automatic (Gay, Stewart et al. 2009)
(NDVI map)
Miscellaneous/precision Thermal Cotton (Gossypium hirsutum) and winter Plant water stress FW Automatic (Sullivan, Fulton et al.
agriculture wheat (Triticum aestivum) 2007)
Vegetation monitoring Multispectral Maize field Vegetation monitoring Helicopter Automatic (Sugiura, Noguchi et al.
(LAI, NDVI) 2005)
* FW = fixed wing drone, MR = multirotor drone.

Figure 3.3 Left panel: NIR image of boundary of oil-palm plantation and forest from modified Canon SX230, right panel: NDVI images as
generated with ImageJ script from (http://www.maxmax.com/). © conservationdrones.org.
such as animal monitoring or land-cover mapping measures can be calculated depending on which
(see Chapters 6 and 7). filter is being used (Hunt, Hively et al. 2010,
Berra, Gibson-Poole et al. 2015). The NIR band
in combination with the R and G bands (NIR-R-
3.1.2 Multispectral cameras
G) might also allow for more accurate land-cover
Modified RGB cameras classification than RGB and is therefore not only
Due to the cost of multispectral cameras many users useful for studies requiring vegetation indices, but
have resorted to converted standard RGB cameras also for researchers interested in land-cover.
as an alternative to true multispectral cameras There are two options to obtain a modified
(Lebourgeois, Bégué et al. 2008, Berra, Gibson- camera. The first is the do-it-yourself option of
Poole et al. 2015). In standard RGB cameras the removing the hot mirror filter and adding a
sensor is sensitive to part of the non-visual spectrum different filter. This is not necessarily a complicated
from 700–1250 nm. However, to restrict the camera procedure and can thus be done at home or in
to only the visible part of the spectrum the 700– the lab with proper instructions or online tutorial
1250 nm region is usually blocked by a hot mirror videos (e.g.10 ), but needs some care and work in a
filter. Theoretically, once the hot mirror filter is clean, dust-free environment. Second, there are also
removed the camera’s RGB channels provide counts off-the-shelf options to purchase modified cameras
of R+NIR, G+NIR, B+NIR. If then, for instance, a from vendors such as MaxMax that can convert a
longpass filter (a filter that only allows for longer multitude of camera brands and types from low-
wavelengths to be passed through) is placed in front cost digital cameras to high-end cameras.11 There
of the lens so that the wavelengths corresponding to are also companies such as Mapir12 that sell one
blue are blocked, the sensor only receives the NIR camera with various filter modifications that can
part of the spectrum on the blue channel. Once this either be used on their own or in a pack of four
NIR component is then subtracted from the digital cameras under a multirotor. These cameras are
counts of the other two channels (where R+NIR and also part of a user-friendly workflow from the
G+NIR have been recorded) one is left with the parts
of the spectrum that correspond to R, G, and NIR
10 https://publiclab.org/ (accessed 7 November 2017)
(Rabatel, Gorretta et al. 2014, Berra, Gibson-Poole
11 https://www.maxmax.com/ (accessed 7 November
et al. 2015). From these, measures related to plant
2017)
photosynthetic activity such as NDVI (Normalized 12 https://www.mapir.camera/ (accessed 7 November
Difference Vegetation Index, Figure 3.3) or related 2017)

Another random document with
no related content on Scribd:
pretty girls and music. Well a showman’s business is to give the
public what they want.”
“I think that this city is full of people wanting inconceivable
things.... Look at it.”
“It’s all right at night when you cant see it. There’s no artistic
sense, no beautiful buildins, no old-time air, that’s what’s the matter
with it.”
They stood a while without speaking. The orchestra began
playing the waltz from The Lilac Domino. Suddenly Ellen turned to
Goldweiser and said in a curt tone. “Can you understand a woman
who wants to be a harlot, a common tart, sometimes?”
“My dear young lady what a strange thing for a sweet lovely girl to
suddenly come out and say.”
“I suppose you’re shocked.” She didnt hear his answer. She felt
she was going to cry. She pressed her sharp nails into the palms of
her hands, she held her breath until she had counted twenty. Then
she said in a choking little girl’s voice, “Harry let’s go and dance a
little.”
The sky above the cardboard buildings is a vault of beaten lead. It
would be less raw if it would snow. Ellen finds a taxi on the corner of
Seventh Avenue and lets herself sink back in the seat rubbing the
numb gloved fingers of one hand against the palm of the other.
“West Fiftyseventh, please.” Out of a sick mask of fatigue she
watches fruitstores, signs, buildings being built, trucks, girls,
messengerboys policemen through the jolting window. If I have my
child, Stan’s child, it will grow up to jolt up Seventh Avenue under a
sky of beaten lead that never snows watching fruitstores, signs,
buildings being built, trucks, girls, messengerboys, policemen.... She
presses her knees together sits up straight on the edge of the seat
with her hands clasped over her slender belly. O God the rotten joke
they’ve played on me, taking Stan away, burning him up, leaving me
nothing but this growing in me that’s going to kill me. She’s
whimpering into her numb hands. O God why wont it snow?
As she stands on the gray pavement fumbling in her purse for a
bill, a dusteddy swirling scraps of paper along the gutter fills her
mouth with grit. The elevatorman’s face is round ebony with ivory
inlay. “Mrs. Staunton Wells?” “Yas ma’am eighth floor.”
The elevator hums as it soars. She stands looking at herself in the
narrow mirror. Suddenly something recklessly gay goes through her.
She rubs the dust off her face with a screwedup handkerchief, smiles
at the elevatorman’s smile that’s wide as the full keyboard of a piano,
and briskly rustles to the door of the apartment that a frilled maid
opens. Inside it smells of tea and furs and flowers, women’s voices
chirp to the clinking of cups like birds in an aviary. Glances flicker
about her head as she goes into the room.
There was wine spilled on the tablecloth and bits of tomatosauce

from the spaghetti. The restaurant was a steamy place with views of
the Bay of Naples painted in soupy blues and greens on the walls.
Ellen sat back in her chair from the round tableful of young men,
watching the smoke from her cigarette crinkle spirally round the fat
Chiantibottle in front of her. In her plate a slab of tricolor icecream
melted forlornly. “But good God hasnt a man some rights? No, this
industrial civilization forces us to seek a complete readjustment of
government and social life ...”
“Doesnt he use long words?” Ellen whispered to Herf who sat
beside her.
“He’s right all the same,” he growled back at her.... “The result has
been to put more power in the hands of a few men than there has
been in the history of the world since the horrible slave civilizations
of Egypt and Mesopotamia....”
“Hear hear.”
“No but I’m serious.... The only way of bucking the interests is for
working people, the proletariat, producers and consumers, anything
you want to call them, to form unions and finally get so well
organized that they can take over the whole government.”
“I think you’re entirely wrong, Martin, it’s the interests as you call
em, these horrible capitalists, that have built up this country as we
have it today.”
“Well look at it for God’s sake.... That’s what I’m saying. I wouldnt
kennel a dog in it.”
“I dont think so. I admire this country.... It’s the only fatherland I’ve
got.... And I think that all these downtrodden masses really want to
be downtrodden, they’re not fit for anything else.... If they werent
they’d be flourishing businessmen ... Those that are any good are
getting to be.”
“But I don’t think a flourishing businessman is the highest ideal of
human endeavor.”
“A whole lot higher than a rotten fiddleheaded anarchist agitator....
Those that arent crooks are crazy.”
“Look here Mead, you’ve just insulted something that you dont
understand, that you know nothing about.... I cant allow you to do
that.... You should try to understand things before you go round
insulting them.”
“An insult to the intelligence that’s what it is all this socialistic
drivel.”
Ellen tapped Herf on the sleeve. “Jimmy I’ve got to go home. Do
you want to walk a little way with me?”
“Martin, will you settle for us? We’ve got to go.... Ellie you look
terribly pale.”
“It’s just a little hot in here.... Whee, what a relief.... I hate
arguments anyway. I never can think of anything to say.”
“That bunch does nothing but chew the rag night after night.”
Eighth Avenue was full of fog that caught at their throats. Lights
bloomed dimly through it, faces loomed, glinted in silhouette and
faded like a fish in a muddy aquarium.
“Feel better Ellie?”
“Lots.”
“I’m awfully glad.”
“Do you know you’re the only person around here who calls me
Ellie. I like it.... Everybody tries to make me seem so grown up since
I’ve been on the stage.”
“Stan used to.”
“Maybe that’s why I like it,” she said in a little trailing voice like a
cry heard at night from far away along a beach.
Jimmy felt something clamping his throat. “Oh gosh things are
rotten,” he said. “God I wish I could blame it all on capitalism the way
Martin does.”
“It’s pleasant walking like this ... I love a fog.”
They walked on without speaking. Wheels rumbled through the
muffling fog underlaid with the groping distant lowing of sirens and
steamboat whistles on the river.
“But at least you have a career.... You like your work, you’re
enormously successful,” said Herf at the corner of Fourteenth Street,
and caught her arm as they crossed.
“Dont say that.... You really dont believe it. I dont kid myself as
much as you think I do.”
“No but it’s so.”
“It used to be before I met Stan, before I loved him.... You see I
was a crazy little stagestruck kid who got launched out in a lot of
things I didnt understand before I had time to learn anything about
life.... Married at eighteen and divorced at twentytwo’s a pretty good
record.... But Stan was so wonderful....”
“I know.”
“Without ever saying anything he made me feel there were other
things ... unbelievable things....”
“God I resent his craziness though.... It’s such a waste.”
“I cant talk about it.”
“Let’s not.”
“Jimmy you’re the only person left I can really talk to.”
“Dont want to trust me. I might go berserk on you too some day.”
They laughed.
“God I’m glad I’m not dead, arent you Ellie?”
“I dont know. Look here’s my place. I dont want you to come up....
I’m going right to bed. I feel miserably....” Jimmy stood with his hat off
looking at her. She was fumbling in her purse for her key. “Look
Jimmy I might as well tell you....” She went up to him and spoke fast
with her face turned away pointing at him with the latchkey that
caught the light of the streetlamp. The fog was like a tent round
about them. “I’m going to have a baby.... Stan’s baby. I’m going to
give up all this silly life and raise it. I dont care what happens.”
“O God that’s the bravest thing I ever heard of a woman doing....
Oh Ellie you’re so wonderful. God if I could only tell you what I....”
“Oh no.” Her voice broke and her eyes filled with tears. “I’m a silly
fool, that’s all.” She screwed up her face like a little child and ran up
the steps with the tears streaming down her face.
“Oh Ellie I want to say something to you ...”
The door closed behind her.
Jimmy Herf stood stockstill at the foot of the brownstone steps.
His temples throbbed. He wanted to break the door down after her.
He dropped on his knees and kissed the step where she had stood.
The fog swirled and flickered with colors in confetti about him. Then
the trumpet feeling ebbed and he was falling through a black
manhole. He stood stockstill. A policeman’s ballbearing eyes
searched his face as he passed, a stout blue column waving a
nightstick. Then suddenly he clenched his fists and walked off. “O
God everything is hellish,” he said aloud. He wiped the grit off his lips
with his coatsleeve.
She puts her hand in his to jump out of the roadster as the ferry
starts, “Thanks Larry,” and follows his tall ambling body out on the
bow. A faint riverwind blows the dust and gasoline out of their
nostrils. Through the pearly night the square frames of houses along
the Drive opposite flicker like burnedout fireworks. The waves slap
tinily against the shoving bow of the ferry. A hunchback with a violin
is scratching Marianela.
“Nothing succeeds like success,” Larry is saying in a deep
droning voice.
“Oh if you knew how little I cared about anything just now you
wouldnt go on teasing me with all these words.... You know,
marriage, success, love, they’re just words.”
“But they mean everything in the world to me.... I think you’d like it
in Lima Elaine.... I waited until you were free, didnt I? And now here I
am.”
“We’re none of us that ever.... But I’m just numb.” The riverwind is
brackish. Along the viaduct above 125th Street cars crawl like
beetles. As the ferry enters the slip they hear the squudge and
rumble of wheels on asphalt.
“Well we’d better get back into the car, you wonderful creature
Elaine.”
“After all day it’s exciting isnt it Larry, getting back into the center
of things.”
Beside the smudged white door are two pushbuttons marked

Night Bell and Day Bell. She rings with a shaky finger. A short
broad man with a face like a rat and sleek black hair brushed straight
back opens. Short dollhands the color of the flesh of a mushroom
hang at his sides. He hunches his shoulders in a bow.
“Are you the lady? Come in.”
“Is this Dr. Abrahms?”
“Yes.... You are the lady my friend phoned me about. Sit down my
dear lady.” The office smells of something like arnica. Her heart
joggles desperately between her ribs.
“You understand ...” She hates the quaver in her voice; she’s
going to faint. “You understand, Dr. Abrahms that it is absolutely
necessary. I am getting a divorce from my husband and have to
make my own living.”
“Very young, unhappily married ... I am sorry.” The doctor purrs
softly as if to himself. He heaves a hissing sigh and suddenly looks
in her eyes with black steel eyes like gimlets. “Do not be afraid, dear
lady, it is a very simple operation.... Are you ready now?”
“Yes. It wont take very long will it? If I can pull myself together I
have an engagement for tea at five.”
“You are a brave young lady. In an hour it will be forgotten.... I am
sorry.... It is very sad such a thing is necessary.... Dear lady you
should have a home and many children and a loving husband ... Will
you go in the operating room and prepare yourself.... I work without
an assistant.”
The bright searing bud of light swells in the center of the ceiling,
sprays razorsharp nickel, enamel, a dazzling sharp glass case of
sharp instruments. She takes off her hat and lets herself sink
shuddering sick on a little enamel chair. Then she gets stiffly to her
feet and undoes the band of her skirt.
The roar of the streets breaks like surf about a shell of throbbing
agony. She watches the tilt of her leather hat, the powder, the rosed
cheeks, the crimson lips that are a mask on her face. All the buttons
of her gloves are buttoned. She raises her hand. “Taxi!” A fire engine
roars past, a hosewagon with sweatyfaced men pulling on rubber
coats, a clanging hookandladder. All the feeling in her fades with the
dizzy fade of the siren. A wooden Indian, painted, with a hand raised
at the streetcorner.
“Taxi!”
“Yes ma’am.”
“Drive to the Ritz.”
Third Section
I. Rejoicing City That Dwelt Carelessly
T
here are flags on all the flagpoles up
Fifth Avenue. In the shrill wind of history
the great flags flap and tug at their
lashings on the creaking goldknobbed poles
up Fifth Avenue. The stars jiggle sedately
against the slate sky, the red and white
stripes writhe against the clouds.
In the gale of brassbands and trampling
horses and rumbling clatter of cannon,
shadows like the shadows of claws grasp at
the taut flags, the flags are hungry tongues
licking twisting curling.
Oh it’s a long way to Tipperary ... Over there! Over there!
The harbor is packed with zebrastriped
skunkstriped piebald steamboats, the
Narrows are choked with bullion, they’re
piling gold sovereigns up to the ceilings in
the Subtreasury. Dollars whine on the radio,
all the cables tap out dollars.
There’s a long long trail awinding ... Over there! Over there!
In the subway their eyes pop as they spell
out Apocalypse, typhus, cholera, shrapnel,
insurrection, death in fire, death in water,
death in hunger, death in mud.
Oh it’s a long way to Madymosell from
Armenteers, over there! The Yanks are
coming, the Yanks are coming. Down Fifth
Avenue the bands blare for the Liberty Loan
drive, for the Red Cross drive. Hospital ships
sneak up the harbor and unload furtively at
night in old docks in Jersey. Up Fifth Avenue
the flags of the seventeen nations are flaring
curling in the shrill hungry wind.
O the oak and the ash and the weeping willow tree
And green grows the grass in God’s country.
The great flags flap and tug at their
lashings on the creaking goldknobbed poles
up Fifth Avenue.
C
aptain James Merivale D.S.C. lay with his eyes closed while
the barber’s padded fingers gently stroked his chin. The lather
tickled his nostrils; he could smell bay rum, hear the drone of an
electric vibrator, the snipping of scissors.
“A little face massage sir, get rid of a few of those blackheads sir,”
burred the barber in his ear. The barber was bald and had a round
blue chin.
“All right,” drawled Merivale, “go as far as you like. This is the first
decent shave I’ve had since war was declared.”
“Just in from overseas, Captain?”
“Yare ... been making the world safe for democracy.”
The barber smothered his words under a hot towel. “A little lilac
water Captain?”
“No dont put any of your damn lotions on me, just a little
witchhazel or something antiseptic.”
The blond manicure girl had faintly beaded lashes; she looked up
at him bewitchingly, her rosebud lips parted. “I guess you’ve just
landed Captain.... My you’ve got a good tan.” He gave up his hand to
her on the little white table. “It’s a long time Captain since anybody
took care of these hands.”
“How can you tell?”
“Look how the cuticle’s grown.”
“We were too busy for anything like that. I’m a free man since
eight o’clock that’s all.”
“Oh it must have been terr ... ible.”
“Oh it was a great little war while it lasted.”
“I’ll say it was ... And now you’re all through Captain?”
“Of course I keep my commission in the reserve corps.”
She gave his hand a last playful tap and he got to his feet.
He put tips into the soft palm of the barber and the hard palm of
the colored boy who handed him his hat, and walked slowly up the
white marble steps. On the landing was a mirror. Captain James
Merivale stopped to look at Captain James Merivale. He was a tall
straightfeatured young man with a slight heaviness under the chin.
He wore a neat-fitting whipcord uniform picked out by the insignia of
the Rainbow Division, well furnished with ribbons and service-
stripes. The light of the mirror was reflected silvery on either calf of
his puttees. He cleared his throat as he looked himself up and down.
A young man in civilian clothes came up behind him.
“Hello James, all cleaned up?”
“You betcher.... Say isnt it a damn fool rule not letting us wear
Sam Browne belts? Spoils the whole uniform....”
“They can take all their Sam Browne’s belts and hang them on the
Commanding General’s fanny for all I care.... I’m a civilian.”
“You’re still an officer in the reserve corps, dont forget that.”
“They can take their reserve corps and shove it ten thousand
miles up the creek. Let’s go have a drink.”
“I’ve got to go up and see the folks.” They had come out on
Fortysecond Street. “Well so long James, I’m going to get so drunk
... Just imagine being free.” “So long Jerry, dont do anything I
wouldnt do.”
Merivale walked west along Fortysecond. There were still flags
out, drooping from windows, waggling lazily from poles in the
September breeze. He looked in the shops as he walked along;
flowers, women’s stockings, candy, shirts and neckties, dresses,
colored draperies through glinting plateglass, beyond a stream of
faces, men’s razorscraped faces, girls’ faces with rouged lips and
powdered noses. It made him feel flushed and excited. He fidgeted
when he got in the subway. “Look at the stripes that one has.... He’s
a D.S.C.,” he heard a girl say to another. He got out at
Seventysecond and walked with his chest stuck out down the too
familiar brownstone street towards the river.
“How do you do, Captain Merivale,” said the elevator man.
“Well, are you out James?” cried his mother running into his arms.
He nodded and kissed her. She looked pale and wilted in her
black dress. Maisie, also in black, came rustling tall and rosycheeked
behind her. “It’s wonderful to find you both looking so well.”
“Of course we are ... as well as could be expected. My dear we’ve
had a terrible time.... You’re the head of the family now, James.”
“Poor daddy ... to go off like that.”
“That was something you missed.... Thousands of people died of
it in New York alone.”
He hugged Maisie with one arm and his mother with the other.
Nobody spoke.
“Well,” said Merivale walking into the living room, “it was a great
war while it lasted.” His mother and sister followed on his heels. He
sat down in the leather chair and stretched out his polished legs.
“You dont know how wonderful it is to get home.”
Mrs. Merivale drew up her chair close to his. “Now dear you just
tell us all about it.”
In the dark of the stoop in front of the tenement door, he reaches
for her and drags her to him. “Dont Bouy, dont; dont be rough.” His
arms tighten like knotted cords round her back; her knees are
trembling. His mouth is groping for her mouth along one cheekbone,
down the side of her nose. She cant breathe with his lips probing her
lips. “Oh I cant stand it.” He holds her away from him. She is
staggering panting against the wall held up by his big hands.
“Nutten to worry about,” he whispers gently.
“I’ve got to go, it’s late.... I have to get up at six.”
“Well what time do you think I get up?”
“It’s mommer who might catch me....”
“Tell her to go to hell.”
“I will some day ... worse’n that ... if she dont quit pickin on me.”
She takes hold of his stubbly cheeks and kisses him quickly on the
mouth and has broken away from him and run up the four flights of
grimy stairs.
The door is still on the latch. She strips off her dancing pumps and
walks carefully through the kitchenette on aching feet. From the next
room comes the wheezy doublebarreled snoring of her uncle and
aunt. Somebody loves me, I wonder who.... The tune is all through
her body, in the throb of her feet, in the tingling place on her back
where he held her tight dancing with her. Anna you’ve got to forget it
or you wont sleep. Anna you got to forget. Dishes on the tables set
for breakfast jingle tingle hideously when she bumps against it.
“That you Anna?” comes a sleepy querulous voice from her
mother’s bed.
“Went to get a drink o water mommer.” The old woman lets the
breath out in a groan through her teeth, the bedsprings creak as she
turns over. Asleep all the time.
Somebody loves me, I wonder who. She slips off her party dress
and gets into her nightgown. Then she tiptoes to the closet to hang
up the dress and at last slides between the covers little by little so
the slats wont creak. I wonder who. Shuffle shuffle, bright lights, pink
blobbing faces, grabbing arms, tense thighs, bouncing feet. I wonder
who. Shuffle, droning saxophone tease, shuffle in time to the drum,
trombone, clarinet. Feet, thighs, cheek to cheek, Somebody loves
me.... Shuffle shuffle. I wonder who.
The baby with tiny shut purplishpink face and fists lay asleep on
the berth. Ellen was leaning over a black leather suitcase. Jimmy
Herf in his shirtsleeves was looking out the porthole.
“Well there’s the statue of Liberty.... Ellie we ought to be out on
deck.”
“It’ll be ages before we dock.... Go ahead up. I’ll come up with
Martin in a minute.”
“Oh come ahead; we’ll put the baby’s stuff in the bag while we’re
warping into the slip.”
They came out on deck into a dazzling September afternoon. The
water was greenindigo. A steady wind kept sweeping coils of brown
smoke and blobs of whitecotton steam off the high enormous
blueindigo arch of sky. Against a sootsmudged horizon, tangled with
barges, steamers, chimneys of powerplants, covered wharves,
bridges, lower New York was a pink and white tapering pyramid cut
slenderly out of cardboard.
“Ellie we ought to have Martin out so he can see.”
“And start yelling like a tugboat.... He’s better off where he is.”
They ducked under some ropes, slipped past the rattling
steamwinch and out to the bow.
“God Ellie it’s the greatest sight in the world.... I never thought I’d
ever come back, did you?”
“I had every intention of coming back.”
“Not like this.”
“No I dont suppose I did.”
“S’il vous plait madame ...”
A sailor was motioning them back. Ellen turned her face into the
wind to get the coppery whisps of hair out of her eyes. “C’est beau,
n’est-ce pas?” She smiled into the wind into the sailor’s red face.
“J’aime mieux Le Havre ... S’il vous plait madame.”
“Well I’ll go down and pack Martin up.”
The hard chug, chug of the tugboat coming alongside beat
Jimmy’s answer out of her ears. She slipped away from him and
went down to the cabin again.
They were wedged in the jam of people at the end of the
gangplank.
“Look we could wait for a porter,” said Ellen.
“No dear I’ve got them.” Jimmy was sweating and staggering with
a suitcase in each hand and packages under his arms. In Ellen’s
arms the baby was cooing stretching tiny spread hands towards the
faces all round.
“D’you know it?” said Jimmy as they crossed the gangplank, “I
kinder wish we were just going on board.... I hate getting home.”
“I dont hate it.... There’s H ... I’ll follow right along.... I wanted to
look for Frances and Bob. Hello....” “Well I’ll be ...” “Helena you’ve
gained, you’re looking wonderfully. Where’s Jimps?” Jimmy was
rubbing his hands together, stiff and chafed from handles of the
heavy suitcases.
“Hello Herf. Hello Frances. Isn’t this swell?”
“Gosh I’m glad to see you....”
“Jimps the thing for me to do is go right on to the Brevoort with the
baby ...”
“Isn’t he sweet.”
“... Have you got five dollars?”
“I’ve only got a dollar in change. That hundred is in express
checks.”
“I’ve got plenty of money. Helena and I’ll go to the hotel and you
boys can come along with the baggage.”
“Inspector is it all right if I go through with the baby? My husband
will look after the trunks.”
“Why surely madam, go right ahead.”
“Isnt he nice? Oh Frances this is lots of fun.”
“Go ahead Bob I can finish this up alone quicker.... You convoy
the ladies to the Brevoort.”
“Well we hate to leave you.”
“Oh go ahead.... I’ll be right along.”
“Mr. James Herf and wife and infant ... is that it?”
“Yes that’s right.”
“I’ll be right with you, Mr. Herf.... Is all the baggage there?”
“Yes everything’s there.”
“Isnt he good?” clucked Frances as she and Hildebrand followed
Ellen into the cab.
“Who?”
“The baby of course....”
“Oh you ought to see him sometimes.... He seems to like
traveling.”
A plainclothesman opened the door of the cab and looked in as
they went out the gate. “Want to smell our breaths?” asked
Hildebrand. The man had a face like a block of wood. He closed the
door. “Helena doesn’t know prohibition yet, does she?”
“He gave me a scare ... Look.”
“Good gracious!” From under the blanket that was wrapped round
the baby she produced a brownpaper package.... “Two quarts of our
special cognac ... gout famille ’Erf ... and I’ve got another quart in a
hotwaterbottle under my waistband.... That’s why I look as if I was
going to have another baby.”
The Hildebrands began hooting with laughter.
“Jimp’s got a hotwaterbottle round his middle too and chartreuse
in a flask on his hip.... We’ll probably have to go and bail him out of
jail.”
They were still laughing so that tears were streaming down their
faces when they drew up at the hotel. In the elevator the baby began
to wail.
As soon as she had closed the door of the big sunny room she
fished the hotwaterbottle from under her dress. “Look Bob phone
down for some cracked ice and seltzer.... We’ll all have a cognac a
l’eau de selz....”
“Hadn’t we better wait for Jimps?”
“Oh he’ll be right here.... We haven’t anything dutiable.... Much
too broke to have anything.... Frances what do you do about milk in
New York?”
“How should I know, Helena?” Frances Hildebrand flushed and
walked to the window.
“Oh well we’ll give him his food again.... He’s done fairly well on it
on the trip.” Ellen had laid the baby on the bed. He lay kicking,
looking about with dark round goldstone eyes.
“Isnt he fat?”
“He’s so healthy I’m sure he must be halfwitted.... Oh Heavens
and I’ve got to call up my father.... Isnt family life just too desperately
complicated?”
Ellen was setting up her little alcohol stove on the washstand. The
bellboy came with glasses and a bowl of clinking ice and White Rock
on a tray.
“You fix us a drink out of the hotwaterbottle. We’ve got to use that
up or it’ll eat the rubber.... And we’ll drink to the Café d’Harcourt.”
“Of course what you kids dont realize,” said Hildebrand, “is that
the difficulty under prohibition is keeping sober.”
Ellen laughed; she stood over the little lamp that gave out a quiet
domestic smell of hot nickel and burned alcohol.
George Baldwin was walking up Madison Avenue with his light

overcoat on his arm. His fagged spirits were reviving in the sparkling
autumn twilight of the streets. From block to block through the
taxiwhirring gasoline gloaming two lawyers in black frock coats and
stiff wing collars argued in his head. If you go home it will be cozy in
the library. The apartment will be gloomy and quiet and you can sit in
your slippers under the bust of Scipio Africanus in the leather chair
and read and have dinner sent in to you.... Nevada would be jolly
and coarse and tell you funny stories.... She would have all the City
Hall gossip ... good to know.... But you’re not going to see Nevada
any more ... too dangerous; she gets you all wrought up.... And
Cecily sitting faded and elegant and slender biting her lips and hating
me, hating life.... Good God how am I going to get my existence
straightened out? He stopped in front of a flowerstore. A moist warm
honied expensive smell came from the door, densely out into the
keen steelblue street. If I could at least make my financial position
impregnable.... In the window was a minature Japanese garden with
brokenback bridges and ponds where the goldfish looked big as
whales. Proportion, that’s it. To lay out your life like a prudent
gardener, plowing and sowing. No I wont go to see Nevada tonight. I
might send her some flowers though. Yellow roses, those coppery
roses ... it’s Elaine who ought to wear those. Imagine her married
again and with a baby. He went into the store. “What’s that rose?”
“It’s Gold of Ophir sir.”
“All right I want two dozen sent down to the Brevoort
immediately.... Miss Elaine ... No Mr. and Mrs. James Herf.... I’ll write
a card.”
He sat down at the desk with a pen in his hand. Incense of roses,
incense out of the dark fire of her hair.... No nonsense for Heaven’s
sake ...
Dear Elaine,
I hope you will allow an old friend to call on you and your
husband one of these days. And please remember that I am
always sincerely anxious—you know me too well to take this
for an empty offer of politeness—to serve you and him in any
way that could possibly contribute to your happiness. Forgive
me if I subscribe myself your lifelong slave and admirer
George Baldwin
The letter covered three of the florists’ white cards. He read it over
with pursed lips, carefully crossing the t’s and dotting the i’s. Then he
paid the florist from the roll of bills he took from his back pocket and
went out into the street again. It was already night, going on to seven
o’clock. Still hesitating he stood at the corner watching the taxis
pass, yellow, red, green, tangerinecolored.
The snubnosed transport sludges slowly through the Narrows in

the rain. Sergeant-Major O’Keefe and Private 1st Class Dutch
Robertson stand in the lee of the deckhouse looking at the liners at
anchor in quarantine and the low wharfcluttered shores.
“Look some of em still got their warpaint—Shippin Board boats....
Not worth the powder to blow em up.”
“The hell they aint,” said Joey O’Keefe vaguely.
“Gosh little old New York’s goin to look good to me....”
“Me too Sarge, rain or shine I dont care.”
They are passing close to a mass of steamers anchored in a
block, some of them listing to one side or the other, lanky ships with
short funnels, stumpy ships with tall funnels red with rust, some of

Conservation Drones Mapping and Monitoring Biodiversity Serge A Wich Full Chapter

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

Conservation Drones Mapping and Monitoring Biodiversity Serge A Wich Full Chapter

Uploaded by

Copyright:

Available Formats

Conservation Drones Mapping and

Monitoring Biodiversity Serge A. Wich

Serge A. Wich, PhD

Lian Pin Koh, PhD

1 Deciding to use a drone 1

2 Typology and anatomy of drones 13

3.2 Applications of sensor types 40

5.1 Land-cover mapping 51

6.1 Animal distribution and density data collection 55

7 Data post processing 73

7.4.3 Analyses based on orthomosaics: land-cover and feature classification 81

8.1 Power systems 94

Deciding to use a drone

1.0 Introduction 1.1 Alternative technologies to acquire

Area covered Resolution Regulations∗ Wind Flight duration Costs Other

Table 1.2 Selection of satellites

Satellite Pixel resolution Repeat frequency Estimated costs

Worldview-3 1.24 <1 14.50–17.50

(accessed 7 November 2017)

(accessed 7 November 2017) watch_072105_Helms,00.html (accessed 7 November 2017)

Table 1.3 Comparison of ground and drone orangutan surveys

Item Ground survey Drone survey

Training data One week Two weeks

Table 1.4 Aspects to consider when considering using a drone

System Flight Sensors Telemetry/ Maintenance Logistics

Ease of software Flight duration Sensor Telemetry General user Ease of

Based on information from Patrick Ribeiro from https://openforests.com/

Table 1.5 Drone planning and operating aspects in different environments

Based on (Duffy, Cunliffe et al. 2017).

Typology and anatomy of drones

2.1 Multirotor movement of its control surfaces (ailerons, rudder

Figure 2.1 A multirotor aircraft on a surveying mission. © Lian Pin Koh.

Multirotor Fixed wing Hybrid VTOL

Figure 2.3 A tail-sitter hybrid VTOL aircraft developed in Switzerland. © Wingtra.com.

2.5.2 Flight controller

an accelerometer, gyroscope, magnetometer, and 2.5.3 Ground control station

However, GCS are not restricted to the laptop 2.6 Summary

3.0 Introduction or set of sensors will be suitable for their data

Visible short wave medium wave long wave

gamma rays X rays ultraviolet rays infrared microwave radio waves

1pm 1nm 1um 1mm 1m 1km

high frequency low frequency

short wavelength long wavelength

Focal length (mm) 5.2 20 13

∗ Above ground level (AGL) was 100 m for all calculations.

Table 3.2 Common sensors used in conservation research

Sensor type Brand examples Wavelength range∗

RGB (Red, Green, Blue) Canon, Nikon, Sony 390–700 nm

classification because the objects of interest are AGL

Application Sensor(s) Species/topic Aim Drone (model if Detection Source

Application Sensor(s) Species/topic Aim Drone (model if Detection Source

RGB (GoPro Hero 3 black urophasianus) Holmquist-Johnson et al.

Application Sensor(s) Species/topic Aim Drone (model if Detection Source

Application Sensor(s) Species/topic Aim Drone (model if Detection Source

Application Sensor(s) Species/topic Aim Drone (model if Detection Source

classification Hero 3) quadcopter) 2016)

Application Sensor(s) Species/topic Aim Drone (model if Detection Source

classification Bayer) and hyperspectral vineyards 2003)

* FW = fixed wing drone, MR = multirotor drone.

Difference Vegetation Index, Figure 3.3) or related 2017)

There was wine spilled on the tablecloth and bits of tomatosauce

Beside the smudged white door are two pushbuttons marked

George Baldwin was walking up Madison Avenue with his light