1

Remote Sensing of the Environment with Small Unmanned Aircraft Systems

(UASs), Part 2: Scientific and Commercial Applications

Ken Whitehead1*; Chris H. Hugenholtz1; Stephen Myshak2; Owen Brown2; Adam LeClair1; Aaron

Tamminga3; Thomas Barchyn1; Brian Moorman1, Brett Eaton3

1 Department of Geography, University of Calgary, 2500 University Drive NW, Calgary AB T2N 1N4

2 Isis Geomatics, 3582 30 St. N., Lethbridge AB T1H 6Z4

3 Department of Geography, University of British Columbia, 1984 West Mall, Vancouver BC V6T

1Z2

* Corresponding author: kwhitehe@ucalgary.ca

Abstract

Small Unmanned Aircraft Systems (UASs) are often suited to applications where the cost, resolution,

and/or operational inflexibility of conventional remote sensing platforms is limiting. Remote sensing with

small UASs is still relatively new, and there is limited understanding as to how the data are acquired and

used for scientific purposes and decision making. This paper provides practical guidance about the

opportunities and limitations of small UAS-based remote sensing by highlighting a small sample of

scientific and commercial case studies. Case studies span four themes: (i) mapping, which includes case

studies to measure aggregate stockpile volumes and map river habitat; (ii) feature detection, which

includes case studies on sandhill image classification and detection of agricultural crop infection; (iii)

wildlife and animal enumeration, with case studies describing the detection of fish concentrations during

a major salmon spawning event, and cattle enumeration at a concentrated animal feeding operation; (iv)

landscape dynamics, with a case study of arctic glacier change. Collectively, these case studies only

2

represent a fraction of possible remote sensing applications using small UASs, but they provide insight

into potential challenges and outcomes, and help clarify the opportunities and limitations which UAS

technology offers for remote sensing of the environment.

Key words: UAS, remote sensing of the environment, case studies, unmanned aerial vehicles, remotely

piloted aircraft, remote sensing

1. Introduction

Remote sensing from satellites and from manned aircraft is a valuable tool that can provide summary

measurements of land cover, surface conditions, and related changes at a variety of spatial and temporal

scales. Remote sensing is continuously and rapidly evolving with the advent of new technology. In the

past, most remote sensing relied upon medium-resolution imagery from sensors such as Landsat ETM and

SPOT (ground resolutions ≈ 5-30 m). The latest generation of optical satellites (e.g., GeoEye,

WorldView) have ground resolutions in the range of 0.5-1 m, which is starting to rival traditional aerial

photography (resolutions typically 0.1-0.5 m). This “resolution creep” is also blurring the distinction

between photogrammetry and remote sensing. Satellite imagery is now often used in traditional

photogrammetric roles, such as the generation of digital elevation models (DEMs) and orthoimages.

Further evidence of the integration of photogrammetry and remote sensing can also be seen in

technologies such as airborne LiDAR (e.g. Glennie et al. 2013) and Synthetic Aperture Radar (SAR) (e.g.

Crosetto and Aragues 2000). These are active systems, which allow the production of DEMs through

completely different workflows than those used in traditional photogrammetry. Frequently spectral

information for analysis is being extracted from aerial photography (e.g. Hunt et al. 2010; Knoth et al.

2013). Similarly, the development and refinement of Structure from Motion (SfM) packages such as

Bundler (e.g. Fonstad et al. 2013) and Photoscan (e.g. Turner et al. 2012) now make it possible to produce

accurate Digital Surface Models (DSMs), Digital Terrain Models (DTMs) and orthoimage mosaics

quickly and affordably.

3

The resolutions and capabilities of different sensors have seen major improvements over the last

few years. However until recently, remote sensing surveys were largely carried out from satellites or

manned aircraft. This is changing with the proliferation of small (< 25 kg) Unmanned Aerial Systems

(UASs). Such systems typically consist of a lightweight Unmanned Aerial Vehicle (UAV), imaging

payload, controller, navigational computer, and ground-based pilot (and spotters, if required). Distinction

of the “small” class of UASs came about through the U.S. Federal Aviation Administration’s

Modernisation and Reform Act of 2012 (cf. Hugenholtz et al. 2012). The use of UASs smaller than 5 kg

is expected to expand rapidly for civil, commercial and scientific uses, because they are less expensive

and more versatile than larger UASs. They are also capable of acquiring high-resolution imagery safely

from much lower altitudes than those associated with manned aircraft and larger UASs.

There has been a great deal of interest and speculation about the potential applications of small

UASs in the scientific community (cf. Hugenholtz et al. 2012). However much of the literature published

to date only addresses potential applications, rather than analysing completed case studies and presently-

viable uses. Other published studies relate to experimental applications of new sensors such as UAS-

mounted LiDAR, (e.g. Lin et al. 2011), which are of interest for the future but are insufficiently mature at

present to allow routine operation. To address this, we present seven case studies derived from our

collective experience. These studies represent applications for which the current generation of small

UASs are well suited, and which can be carried out on a routine basis without the need for customisation.

These applications can therefore be considered as being representative of the current range of applications

for which small UASs are being used. Although the examples presented here are all from Canada, similar

applications are possible anywhere in the world (subject to local regulations). The case studies focus on

four broad themes:

1) Mapping: Case studies include the determination of aggregate stockpile volumes and mapping of

river habitat.

4

2) Feature detection: This section includes case studies on image classification of pocket gopher

mounds and the detection of agricultural crop infection.

3) Wildlife: Examples include a survey of a major salmon spawning event and the use of thermal

infrared imagery for cattle enumeration.

4) Landscape dynamics: This section features a case study in which imagery from a small UAS was

used to study glacier dynamics.

While these case studies illustrate only a small number of the remote sensing applications that are

possible using small UASs, they serve to illustrate both the advantages and the challenges of working

with this type of data. Successful integration of data gathered by small UASs will ultimately depend on

whether the advantages of such data (e.g. high spatial resolution, affordability, flexibility of acquisition)

outweigh the disadvantages (e.g. variable illumination, inconsistent geometry, low quality imagery).

This review is the second part of a two part paper. Part 1 (Whitehead and Hugenholtz, in review)

addresses the technology and present issues associated with spectral and topographic data collected from

UASs. For more basic primers on the use of UASs and complementary perspectives, we refer readers to

reviews by Hardin and Jensen (2011) or Watts et al. (2012).

2. Regulations governing the use of small UASs in Canada

As described in the companion paper (Part 1), regulations are in place in many countries which govern

non-military uses of small UASs. These regulations influence the remote sensing data collected by UASs.

Since this review features case studies performed in Canada, in this section we outline the regulatory

framework and some of the rules that distinguish remote sensing data acquired by small UASs to provide

necessary context.

In Canada, the operation of UASs falls under the Canadian Aviation Regulations, which are

administered by Transport Canada – the federal agency overseeing Canada’s transportation systems. In

order to carry out non-recreational remote sensing surveys with small UASs, an individual must hold

5

insurance and a Special Fight Operations Certificate (SFOC) (Transport Canada 2008). This permits the

operation of a UAS, subject to several conditions, some of which are described below. Although

Transport Canada may issue blanket SFOCs for organisations to operate in certain parts of the country,

new applications are always dealt with on an individual basis.

The paperwork and time involved in completing SFOC applications can slow entry for many

researchers and professionals wishing to become involved in UAS operations. To meet the demand from

such users, Transport Canada makes provision for a simplified application process (Transport Canada

2008). This process streamlines the application and is designed to meet the needs of the majority of such

users. However users who require greater operational flexibility, such as police and search and rescue,

may need to apply through the standard SFOC process, as the conditions imposed by the streamlined

application process may be too restrictive to meet their needs. The following discussion refers to SFOCs

issued under the simplified application process, unless otherwise stated.

One important restriction that impacts remote sensing with small UASs is the regulatory

requirement that they are operated within visual range at all times, even though the telemetry between the

UAS and ground control system may reach well beyond this limit. In practice, this means that UASs must

remain relatively close to the takeoff point, limiting the area that can be surveyed in a single flight. The

maximum distance will vary according to the size, shape and colour of the UAS, as well as the weather

conditions and lighting. Although Transport Canada may specify otherwise, for many UASs the

maximum visual range is < 1 km, which yields a maximum surveyable area of ≈ 3 km2 from one

location.

A secondary restriction affecting remote sensing surveys is that the flying height of UASs is

typically limited by Transport Canada to a ceiling of 400 feet (122 m) above ground level (Transport

Canada 2008). This allows for very high ground resolutions in the imagery, but makes it necessary to

collect many more images than would otherwise be necessary, which can in turn introduce problems at

the processing stage. At low flying heights, the effects of relief displacement and tree lean are often

greatly exaggerated, making it extremely difficult to produce good orthoimages for some areas. In areas

6

of high relief, the 400 feet limitation may not be practical, and may thus rule out the use of a UAS. Under

such circumstances, Transport Canada may allow a higher flying height; however the additional time

involved in obtaining the certification may impact the viability of a project.

The simplified SFOC application process currently limits UAS operation in Canada to daylight

use only. For photogrammetric and most remote sensing applications this is not a problem. However the

daytime-only requirement can be a limitation for thermal surveying. This particularly affects applications

such as search and rescue operations, and heat loss and energy efficiency audits (e.g. Martinez-De Dios

and Ollero 2006) as night enhances thermal contrasts. With the development of reliable sense and avoid

systems, night flights may become more commonplace in the future. However at present night time

operation is only permitted when a full SFOC application has been lodged.

Transport Canada also imposes restrictions on the size of UAS platforms that can be flown. In

general, only aircraft with a take-off weight of less than 35 kg are permitted (Transport Canada 2008).

While aircraft below this weight have the advantages of portability and safety, this restriction effectively

limits the duration of flights, as well as impacting aircraft stability. However, the most serious impact is

on the size of remote sensing payloads that can be carried. Payload limitations restrict most UASs to

carrying only lightweight compact cameras. Larger sensors used on piloted aircraft, such as metric

cameras, hyperspectral scanners, Synthetic Aperture Radar (SAR) and light detection and ranging

(LiDAR) systems, are generally too heavy at present to be carried on most UAS platforms meeting this

weight restriction, and although some proof-of-concept research has mounted some of these sensors to

lightweight UASs (e.g. Turner et al. 2012; Wallace et al. 2012), the flying time is generally too short for

practical applications. The development of miniaturised “Micro-Electro-Mechanical (MEMS) systems”

may usher in a new generation of smaller, lighter sensors in the next few years, but at the present the

takeoff weight restriction imposes practical limitations on the type of remote sensing data that can be

acquired from small UASs.

Due to safety considerations, there are also restrictions on where small UAS surveys can be

carried out. Currently in Canada, no UAS overflights are allowed in urban areas, or within 100 feet of

7

people or inhabited structures unless consented (Transport Canada 2008). The high spatial resolution that

can be obtained from small UAS surveys would be very useful in many urban settings; however, for

public safety and perhaps privacy concerns, this restriction is likely to remain in place for research and

commercial applications. Further restrictions on location apply in restricted airspace, and close to airports.

A recent report by the Canadian UAV Systems Program Design Working Group proposes major

changes for regulation of lightweight UASs under 25 kg in weight (Transport Canada 2012). In particular,

this report proposes that UASs classified as “low energy” should be exempted from SFOC regulations.

The proposed definition of low energy is that the kinetic energy imparted by the platform to a stationary

person or object in the event of a crash is less than 12 J/cm2, which is not considered to be a dangerous

impact. The implications of this change are that many small UASs could potentially be operated without

restriction, particularly those optimised for surveillance. This recommendation has been criticised for

ignoring privacy concerns, and has prompted calls for a wider public debate on the implications of

regulatory change (Gersher 2014).

3. Case studies

In this section we outline seven case studies. While not exhaustive, these applications represent a cross

section of the types of applications for which small UASs are well suited. Many proposed applications of

small UASs are experimental, or require expensive customisation of the UAS platform. However the case

studies described here were all carried out using commercially-available systems, equipped with standard

“off the shelf” components and cameras. As such, they can be considered to be representative of the

current state of the industry. In all cases, photography or video is the primary data source. While there are

a number of more advanced sensors that have been developed for use with small UASs, these are still

largely experimental. The primary use of small UASs to date has been for the production of digital

elevation models (DEMs) and orthoimage mosaics. This represents an extension of the traditional

photogrammetric mapping process. It is expected that this emphasis will change as the industry matures

8

and new sensors and analysis techniques become available. However the case studies presented reflect

our intention to portray the current state of the industry.

For each survey we used one of four battery-operated UASs (Figure 1), which included one

rotary-wing platform (Aeryon Scout Pro quadcopter) and three fixed-wing UASs (RQ-84Z Aereohawk,

Outlander UAS, Ebee). The rotary-wing platform has a maximum flying time of ≈ 20 minutes or less,

depending on wind, while the fixed wing UASs range from 40 (Ebee) to 90 minutes (Areohawk). The

cameras used to acquire the imagery included consumer-grade RGB, near-infrared filters, and a FLIR

thermal. The purchase price for these UASs ranges from $7,000 CAD for the Outlander UAS to $107,500

CAD for the Aeryon Scout Pro.

3.1. Mapping

3.1.1 Case study 1: Aggregate (sand and gravel) volumetrics

One of the leading commercial applications of UAS-based remote sensing is measuring aggregate

stockpile volumes (e.g. Eisenbeiss 2009; Haala et al. 2011; Sauerbier et al. 2011). In Canada, aggregate

companies have been early adopters of UAS surveys for several reasons. Aggregate quarries are usually

self-contained, and encompass areas that are within or close to the LOS limit. The area involved can

easily be covered by a small UAS, although larger quarries may require two or three flights for complete

coverage. Quarries are also usually free of vegetation, making the production of accurate DEMs

comparatively straightforward.

For this case study, two UAS surveys were carried out of a small stockpile in June and again in

November 2012, using the Aeryon Scout Pro quadcopter. This UAS uses a proprietary RGB camera

(Photo3S), which has a focal length of 7.5 mm and a field of view of 37° by 29°. For both surveys, 62

images were collected, spread over 6 flight lines, at a flying height of 100 m. In both cases, the aircraft

flew at a speed of 5 m s-1 and the survey was completed in approximately 20 minutes. The same flight

plan was used for both surveys in order to ensure that the imaging geometry remained as consistent as

possible. For each survey, 10 ground control points (GCPs) were surveyed around the stockpile using a

9

real time kinematic Global Navigation Satellite System (GNSS). On each occasion the stockpile was also

surveyed using the GNSS, in order to assess the vertical accuracy of the UAS-derived DEMs relative to

the conventional field-based survey approach.

Processing for both jobs was carried out using EnsoMOSAIC software. Initial estimates of image

centres and camera orientations were obtained from the aircraft log file. After aerial triangulation (AT)

was completed, a 3.5 cm resolution DEM was constructed for each survey. Both DEMs were edited to

remove the effects of vegetation and were then used to produce final orthoimage mosaics. The DEMs and

orthoimage mosaics generated are shown in Figure 2. Volumetric estimates were produced in each case

using ArcGIS.

For the June survey, the RMSE of vertical difference between the UAS-derived DEM and 126

GNSS points surveyed on the stockpile was 0.106 m, with the UAS-derived volume estimate being 2.6%

lower than that estimated from the GNSS survey of the stockpile. For the November survey the RMSE of

the difference between the UAS-derived DEM and 107 surveyed GNSS points was 0.097 m, with the

UAS and GNSS-derived volumes differing by 3.9%. The volume removed from the stockpile between the

two surveys was estimated at 1521 m3, which was 2.5% greater than that estimated from the haul weight.

Conversion of haul weight to volume necessitates the application of a conversion factor, so haul weight

derived estimates are unlikely to agree exactly with the UAS data. Overall, this case study demonstrates

that photogrammetric surveying using a small UAS can yield measurements that are comparable to

conventional methods of measuring stockpile volume and volume changes.

3.1.2. Case study 2: River habitat mapping

One of the key advantages of remote sensing with small UASs is that the data can bridge an important

gap between ground-based surveys and remote sensing data from satellites or piloted aircraft. For

example, ground-based techniques are very effective for developing high-resolution maps over small

areas, but impractical for large areas due to the time and physical effort required. Conversely, remote

sensing data from satellites and piloted aircraft can be used to map large areas, but they either don’t have

10

adequate spatial resolution or are too expensive to map fine-scale features within small to intermediate

areas on a routine basis. Small UASs are particularly well-suited to mapping at an intermediate spatial

scale (i.e., 1-10 km2).

The case study for this application was carried out along a 1 km segment of the Elbow River,

approximately 21 km west of Calgary, Alberta. The UAS survey was conducted during low-flow

conditions (5.9 m3s-1) in late September 2012. The purpose of the UAS survey was to evaluate the ability

of the imagery to measure key metrics of reach-scale river morphology. At the field site (50.988°N,

114.509°W), the river flows through a ~200 m wide active channel belt with extensive gravel bars. In a

number of places the river splits into separate channels. The channel belt consists of mixed sediment (silt-

boulders) and was covered in places by Large Wood (LW), which included individual logs, trees,

rootwads and large jams. LW has an important influence on the morphological and biological quality of

aquatic ecosystems and as such, is a key component of fish habitat.

The Aeryon Scout Pro quadcopter was used for the survey with its proprietary Photo3S RGB

camera. A total of 45 highly visible GCPs were distributed throughout the study area in order to assist

with the photogrammetric processing. In total, the survey yielded 192 images, which were used to

develop a DEM and orthoimage. One of the unique aspects of this survey was that the water in the active

channels was relatively clear, which allowed extraction of the bathymetry. We then applied a two-

dimensional hydrodynamic model (River2D) to estimate spatial variability in the flow (see details in

Tamminga et al. 2014). We mapped the median grain size of surface sediment (D50) using an empirical

relation between image texture and field measurements of D50 (R2 = 0.82).

Results from the UAS survey are shown in Figure 3. The 5 cm spatial resolution orthoimage

mosaic (Figure 3a) and DEM (Figure 3b) allowed for the detailed characterisation of channel morphology

and aquatic habitat features (Figure 3c). Based on 297 check points, the vertical RMSE of the DEM was

calculated to be 0.088 m in dry, exposed areas, and 0.119 m in submerged areas, following a correction

procedure for the refractive effects of the water-air interface (Westaway et al. 2000). This topographic

information was sufficient to apply the River2D model, which allowed for determination of depths and

11

velocities throughout the study reach (Figure 3d, 3e). When superimposed on the base orthoimage, these

depth and velocity distributions provide an intuitive method with which to assess reach-scale flow

patterns and their relationship to geomorphic context.

To evaluate the utility of the UAS survey data for aquatic habitat assessment, we then combined the

modeled hydraulic data with features that provide visual overhead cover for fish species, including LW,

overhanging vegetation, undercut banks, water surface turbulence, and pools. Such features are strong

determinants of fish habitat availability and were digitised directly from the orthoimage. Using preference

curves developed for the nearby, ecologically-similar Kananaskis River that describe the theoretical

suitability of depths, velocities, and cover accessibility for adult brown trout, we were able to map out

spatial patterns of habitat suitability (Figure 3f) in terms of a composite suitability index that ranges from

0 (least suitable) to 1 (most suitable). Although not validated directly, this method highlights

distributions of potentially desirable habitat hotspots and can be used to estimate the total habitat

availability for the reach. It also demonstrates how the perspective offered by UAS data can show the

spatial complementarity of habitat features: deep, low velocity pools with abundant overhead cover

sources were identified as highly suitable areas and easily mapped throughout the reach. Overall, the

combination of the high-resolution orthoimage and detailed topographic data provided by small UAS-

based remote sensing provides an ideal method for intermediate scale analyses of river morphology and

aquatic habitat.

3.2. Feature Detection

3.2.1. Case study 3: Precision agriculture

Precision agriculture is a sophisticated method of actively managing within-field variability in crops,

which often makes use of remotely-sensed imagery in order to provide data on spatio-temporal variability

in crops (Mulla 2013). Remote sensing is especially useful for identifying areas needing additional

treatment with water, chemicals, pesticides and herbicides. However, widespread adoption of remote

12

sensing in precision agriculture is contingent on affordability, accessibility, and the resolution (spatial,

temporal, spectral and radiometric) of the imagery. For small-scale operations, the cost of data from

conventional remote sensing platforms may be too high. Small UASs potentially offer a low-cost

alternative to conventional remote sensing platforms, and research to date shows promise (e.g. Berni et al.

2009; Baluja et al. 2012; Zhang and Kovacs 2012; Bellvert et al. 2013; Calderon et al. 2013; Corcoles et

al. 2013; Garcia-Ruiz et al. 2013; Gonzalez-Dugo et al. 2013; Huang et al. 2013; Urbahs and Jonaite

2013; Zarco-Tejada et al. 2013; Duan et al. 2014).

The case study for this application involved using imagery from a UAS survey in order to identify

and track infections due to late blight (a pathogenic organism which occurs in high value potato crops).

We used visible (VIS) and Near-InfraRed (NIR) imagery to identify areas of late blight. The primary

characteristic of late blight is chlorosis, which occurs when leaves produce insufficient chlorophyll. To

detect late blight we used the Normalised Difference Vegetation Index (NDVI), defined by:

( )( )VISNIR

VISNIRNDVI

+−= (1)

We used the Ebee fixed-wing UAS equipped with a Canon IXUS/ELPH 125HS compact camera

(16.1 MP resolution). In order to calculate and map NDVI, the UAS was flown twice over the ≈ 1 km2

field located in southern Alberta: once with the camera setting for visible (RGB) light, and once with a

NIR filter. A total of 12 flight lines were used to cover the field, resulting in 162 images. The altitude of

the UAS was 120 m, yielding a ground resolution of 4.5 cm, which was resampled to 25 cm during the

processing. To expedite the image processing and provide the farmer with quicker results, direct

georeferencing, using GNSS positions from the UAS log file, was used instead of the more rigorous

approach with GCPs. Typical levels of accuracy obtainable from direct georeferencing without

differential correction are in the range of 2-5 m (Turner et al. 2013).

13

The first survey was flown in mid July, with a follow up survey two weeks later. The image

mosaics of the visible and NIR imagery from both surveys are shown along with the computed NDVI

maps in Figure 4. Imagery from the first flight in mid July showed that late blight was present in a small

patch located in the southeast corner of the field. Identification of late blight is fairly straightforward

because it typically begins in small patches, whereas water stress is not as isolated. The presence of late

blight was later confirmed by the farmer, who then applied fungicide in an attempt to manage the

infection and limit its spread. A second UAS survey was then performed in late July in order to determine

if the fungicide had been successful. This survey was identical to the first and used the flight parameters

and configuration from the first survey, two weeks earlier. In this case the imagery revealed that late

blight had spread to other patches, thus requiring more intensive fungicide applications to limit further

spread.

3.2.2 Case study 4: Pocket gopher mound feature detection and image classification

One of the unique characteristics of remote sensing imagery from small UAS is that it is able to resolve

small-scale features that are otherwise difficult or impossible to resolve with imagery from satellites and

manned aircraft. For example, after processing it is not uncommon to obtain ground resolutions of 5 cm

or less with RGB imagery acquired from a UAS equipped with an off-the-shelf digital camera. This is

about an order of magnitude smaller than the ground resolution of the highest-resolution satellite imagery

that is available commercially (e.g., GeoEye-1: 0.5 m). However, the high ground resolution also brings a

whole host of challenges in terms of adapting image classification techniques developed for low-

resolution imagery (Laliberte et al. 2010; Laliberte and Rango 2011).

This case study illustrates feature detection of pocket gopher mounds in a grassland region of

southwestern Saskatchewan. In June 2012 we acquired RGB imagery from the RQ-84Z Aerohawk UAS

over an area of roughly 2 km2. The camera used was an Olympus PEN Mini E-PM1with a 14 mm lens.

This camera had been pre-calibrated prior to the survey. A total of 280 images were gathered along 14

14

flight lines, with the overlap and sidelap between strips both set to 65% to ensure full stereo coverage.

The flight was completed in 50 minutes under ideal conditions: clear skies and light winds.

Ground control for the survey consisted of 28 yellow plastic targets distributed throughout the

site. Processing was carried out using the Match-AT module of Trimble’s Inpho software. Following

Aerial Triangulation (AT), 1 m resolution DEMs and DSMs were created for the survey area. The process

used by Inpho to create DEMs involves interpolation between points within a sparse point cloud

(Hugenholtz et al. 2013). Consequently it is less sensitive to the effects of vegetation than a surface model

produced using a dense point cloud, and will often provide a better representation of the terrain in

vegetated areas. In contrast a DSM better represents bare areas and areas which have low contrast

variations. The DSM covering the bare areas, and the DEM covering the vegetated areas, were combined

to form a composite terrain model, which was used to generate a set of 0.1 m resolution orthoimages. The

final step was to combine these to form a seamless orthoimage mosaic (Figure 5).

While the primary goal was to test the vertical accuracy of a DEM derived from the UAS imagery

relative to airborne LiDAR data (Hugenholtz et al. 2013), the final mosaiced orthoimage revealed the

pervasiveness of a distinct feature that was otherwise unremarkable from the ground. Littered extensively

across the survey area were small mounds of sand, usually 1 m in diameter or less, caused by the digging

activities of fossorial mammals like the northern pocket gopher (Thomomys talpoides) and thirteen-lined

ground squirrel (Ictidomys tridecemlineatus). These are highly active burrowing mammals that leave

numerous, but relatively small, mounds of excavated sand above their burrow networks.

While there has been a significant amount of research on the impact fossorial mammal

disturbance has on ecosystems at a localised scale, there is little in the literature to suggest how

widespread such ‘bioturbation features’ actually are. For the most part, their intensity of disturbance in

any one landscape has largely relied on proxies such as population censuses or rough estimates of mound

coverage, with 5-25% being commonly reported numbers (see for example Richens 1966; Turner et al.

1973; Foster and Stubbendieck 1980; Grant et al. 1980; Hobbs and Mooney 1985; Spencer et al. 1985;

Huntly and Inouye 1988; Reichman et al. 1993). The majority of mounds are too small to distinguish

15

using traditional remote sensing imagery, limiting all research to what is logistically feasible to achieve at

the scale of a research plot (i.e., < 1 acre). With 10 cm or better resolutions over relatively large areas,

imagery from small UASs presents an opportunity to directly measure the footprint of fossorial mammal

disturbance at the scale of a landscape for the first time.

An automated workflow was developed using object-based image classification to avoid

manually digitizing the mounds (estimated to take > 3 months). Using the feature extraction module in

ENVI 5, groups of similar pixels in the image were segmented in order to create image objects at a scale

fine enough to recognise the mounds. Spatial, spectral, and textural attributes for each image object were

then calculated based on both the RGB bands in the original mosaic, and several custom bands generated

in a pre-processing step that enhanced different aspects of mound characteristics (edges, brightness, etc.).

Classification rule sets to isolate the mound objects from the background grassland matrix were then

constructed based on the object attributes. Due to variability in focus and issues with bidirectional

reflectance, no single rule-set could accurately separate mounds across the entire mosaic. As a result, the

image was broken up in to five locally similar sub-zones, with separate classification rule-sets built for

each.

The final classification produced a binary vector map of mounds at study site (Figure 5). This

allowed total area of disturbance to be assessed at a landscape scale, showing that while mounds can be

locally extremely abundant, in aggregate they only cover ~ 1% of the landscape, one fifth of the area than

has previously been estimated using ground based methods.

3.3. Animal and wildlife monitoring

3.3.1. Case study 5: Monitoring a salmon spawning event

Due to logistical challenges, cost, and the requirement for high resolution imagery, small UASs can fill an

important niche for wildlife researchers. The imagery or video collected by the UAS can be used for

wildlife enumeration or detecting the presence or absence of wildlife. This case study describes efforts to

map the annual salmon run in October 2010 along the Adam’s River in southern British Columbia,

16

Canada. The goal was to provide an overview of salmon locations; salmon have important ecological and

socio-economic implications in the region.

The October 2010 sockeye salmon run along the Adam’s River was believed to have been the

largest in the last 100 years, with an estimated four million salmon returning to the river (Adams River

Salmon Society 2013). During the peak of this run, a UAS survey was carried out along a 1 km segment

of the river near Shuswap Lake. The survey involved the Outlander UAV, which is a lightweight fixed-

wing UAV with a 2.5 m wingspan. The aircraft carried a Panasonic Lumix-LX3 camera, which is

lightweight, compact, and has a retractable lens assembly. The camera was used at its widest angle

setting, giving it an effective focal length of 5.1 mm.

The main purpose of the survey was to produce an image mosaic that would allow the major

concentrations of salmon, and individual salmon on the spawning grounds to be clearly identified. The

river survey was flown in four strips at a height of 150 m above ground level, which gave a spatial

resolution of better than 5 cm. The aircraft was launched from a sandy beach adjacent to the river, and it

was recovered after the survey by flying it into a net.

Five GCPs were placed at accessible locations along the river. For this application, high spatial

accuracy was not required, and horizontal errors of up to 2 m were considered acceptable. Processing was

carried out using Trimble’s Inpho software, with all GCPs being given the same nominal elevation.

Rather than using a DEM to orthorectify the images, a simple flat surface was used to rectify the imagery.

This process is sufficient to remove scale and perspective distortions, but residual relief distortions within

the imagery are not corrected for. The final mosaic was assembled from the rectified images and is shown

in Figure 6.

The imagery acquired provides a unique perspective of the distribution of salmon within the river,

and clearly shows individual salmon occupying the spawning beds. Conventional aerial photography

would not be able to provide the level of resolution, nor the flexibility in the timing of image acquisition

that was achieved using the small UAS. When combined with external information from sources such as

fish tagging, UAS imagery could potentially be used to arrive at an estimate of salmon numbers at the

17

time of the survey. In this case there was no such external information available, but it is estimated that

the salmon shown in Figure 6 number in the hundreds of thousands.

3.3.2 Case study 6: Cattle enumeration

For larger mammals, imagery acquired from small UASs has been quite successful at detecting

individuals, for example: elephants (Vermeulen et al. 2013), rhinoceros (Mulero-Pazmany et al. 2014),

manatees (Jones et al. 2006), birds (Jones et al. 2006) (Chabot and Bird 2012) (Sarda-Palomera et al.

2012), and dugong (Hodgson et al. 2013). In most published examples the enumeration has been

conducted with RGB imagery, but in order to improve detection against a complex background, Mulero-

Pazmany et al. (2014) suggest it may be preferable to use thermal imagery.

Our animal enumeration case study was carried out with the Aeryon Scout Pro quadcopter and an

integrated thermal FLIR camera. The application involved beef cattle enumeration at a concentrated

animal feeding operation (CAFO) or feedlot. The CAFO consisted of separate holding pens where

different groups of cattle were held. The flight was performed in late fall 2012, which provided

reasonable temperature contrast between the ground and the cattle. The UAS was flown approximately

100 m above the ground along five flight lines. Individual frames from the thermal FLIR video were

extracted and stitched together with Microsoft’s Image Composite Editor (ICE) in order to produce the

final mosaic. Given the application, georeferencing was not required. Individual cows were then

identified from manual interpretation of the imagery.

A portion of the CAFO is shown in Figure 7a, with individual animals denoted by points in

Figure 7b. In most cases, the animals are clearly resolved from the background soil, which makes

identification straightforward. There is some distortion in parts of the image caused by animals moving in

between flight lines, which introduces some uncertainty to the counts. Capturing a single frame over each

pen would likely reduce uncertainty, but this would also require an intervention by the operator in order to

position the UAS manually, since the footprints of images acquired during autonomous operation are

unlikely to be optimal for this purpose.

18

3.4. Landform dynamics

3.4.1. Case study 7: Repeat survey of an arctic glacier

Small UASs are ideal for multi-temporal remote sensing aimed at identifying and quantifying

environmental changes. Under appropriate weather conditions, the baseline interval between UAS

surveys can be specifically tailored to match the pace of the environmental changes under investigation,

or can match the interval required for decision making. Furthermore, the ability to develop multi-temporal

DSMs and DTMs enables researchers and professionals to track and quantify morphodynamic changes in

three dimensions. Already, researchers have used multi-temporal imagery from small UASs to track

landslides (Niethammer et al. 2012; Lucieer et al. 2013; Rothmund et al. 2013), monitor crop growth

(Bendig et al. 2013), measure changes in stockpile volume (Hugenholtz et al.. 2014), count birds (Sarda-

Palomera et al. 2012), monitor crop water stress (Stagakis et al. 2012), and monitor changes in various

vegetation indices (Zarco-Tejada et al. 2013).

For this case study, a UAS survey was carried out in the summer of 2010 over the terminus of an

arctic glacier on Canada’s Bylot Island. A follow up survey was carried out a year later, using imagery

captured from a piloted helicopter. Further details on this study are described by Whitehead et al. (2013).

The resulting DEMs and orthoimage mosaics allowed detailed estimates to be made of surface thinning,

marginal retreat, and flow rates at the glacier terminus, over the one year interval between the surveys.

The 2010 UAS survey was carried out using the Outlander fixed-wing UAS, which carried a

Lumix LX3 compact camera with an effective focal length of 5.1 mm. The aircraft flew 16 north-south

flight lines, and collected 148 photos from a nominal altitude of 300 m above the glacier surface. The

survey covered the lower 1.5 km of the glacier adjacent to the terminus and took approximately 30

minutes to complete. Prior to the aerial survey, 16 targets were placed across the glacier surface, and in

the adjacent moraine regions.

This job was processed using Trimble’s Inpho software, with the photos from every second strip

being used, except for the steeply-sloping valley sides, where extra photos were added from the unused

19

strips in order to ensure full stereo coverage. Eight of the targets were used as horizontal and vertical

GCPs, with the remaining targets being used as check points. The accuracy assessment showed RMS

errors for the check points of 0.18 m, 0.21 m, and 0.42 m in X, Y, and Z, respectively. After the

triangulation was completed, a 1 m resolution DEM was created. This was used to orthorectify each

image, prior to the creation of an orthoimage mosaic, with a resolution of 10 cm.

For the 2011 survey, a Lumix GF-1 camera, equipped with a 14 mm lens was attached to the

landing gear of a manned helicopter. The helicopter flew over the glacier in a series of north/south flight

lines, with the camera being triggered manually every four seconds. In this case the helicopter had a

nominal altitude of 400 m above the glacier surface and preliminary photo centre coordinates were

estimated from the 2010 mosaic. The processing chain was similar to that for the previous year, although

in this case natural features were used for GCPs and check points, because of time constraints. Horizontal

RMS accuracies at the check points were lower than for the previous year at 0.63 m, 0.52 m, in X, and Y

respectively, likely because target positions could not be as accurately identified. However the vertical

RMS accuracy of 0.19 m was better than for the previous year. As with the previous year, a 1 m

resolution DEM and a 10 cm resolution orthoimage mosaic were produced. The orthoimage mosaics

produced from the 2010 and 2011 surveys are shown in Figure 8.

Using the DEMs generated from the two surveys, it was possible to quantify the thinning of the

glacier from 2010 to 2011. The orthoimage mosaics also allowed changes to the terminus extents to be

directly measured, making it possible to quantify glacial retreat over a one year period. The magnitude

and direction of surface flow was also measured, by comparing the positions of approximately 400 points

on the glacier surface. This study shows how comparative measurements can be made of rapidly-

changing landscape features. It is likely that glaciology is one field which will benefit significantly from

the availability of low-cost, on-demand imagery acquired from small UASs.

4. Discussion and conclusions

20

While the case studies described above are selective in their application, they are sufficiently diverse to

illustrate many of the major benefits and challenges currently associated with the use of small UASs.

They also provide a good snapshot of the present state of the industry. It can be seen that

photogrammetric and mapping-type applications are well suited to the capabilities of the current

generation of small UASs. The concomitant development of structure from motion software packages,

such as Photoscan and Pix4D is also helping improve the application of small UASs for mapping.

The use of small UASs for photogrammetric surveying can meet a need for detailed, high-

accuracy surveys of areas in the range of 1-10 km2. This has traditionally marked the boundary between

ground-based surveys and photogrammetric surveys from manned aircraft. For areas of this size, ground

surveying is typically too time-consuming to be cost effective. The point density for ground surveys is

generally much lower than that collected from a UAS, and while the accuracy of individual GNSS points

can be very high, a survey with a small UAS will typically produce a DEM that contains much more

detail and higher accuracy overall. Traditional photogrammetric surveying is typically inflexible in terms

of scheduling. It is also very expensive, and only becomes economical for high value projects, or where

large areas need to be covered. Because of the higher flying height, the ground resolution of photography

acquired from manned aircraft is typically much lower than that gathered by small UASs. Surveying

using small UASs is cost effective, flexible in terms of scheduling, and provides the highest resolution

imagery available.

What these case studies also show is that currently the main use of the orthoimagery from small

UASs is for qualitative analysis. There are many challenges, some of which are outlined in Whitehead

and Hugenholtz (in review), that limit that ability to straightforwardly adapt automated mapping

techniques developed for satellite and manned aircraft. Thus far, the imaging technology adapted for

small UASs has not kept pace with the advances in UAS platforms and capabilities. Problems relating to

image quality and illumination differences between images generally make imagery acquired from sUASs

unsuitable for quantitative analysis. While the land classification study described above did involve some

automated image classification, illumination differences meant that the job had to be separated into five

21

separate zones prior to classification, and unique classification rules developed for each zone. This

example also serves to illustrate how object-based classification algorithms, which take account of shape,

textural, and contextual attributes, are likely preferable to pixel-based classification for the analysis of

imagery from small UASs.

With the continued expansion of small UAS for remote sensing and mapping, it is likely that new

sensors will be developed that better match the needs of researchers and the market. For imaging sensors,

key areas that need to be addressed include size and weight, image quality, sensor stability, signal to noise

ratios, and the ability to record imagery across multiple wavebands simultaneously. The next generation

of sensors, designed specifically for incorporation within small UASs, are likely to address many of these

shortcomings. The capability for automated image analysis will greatly increase the utility of small UASs

for many applications, and is likely to lead to greater uptake in many sectors.

In the case studies described here, we have attempted to show the types of applications for which

small UASs are currently suitable. At present these are heavily biased towards photogrammetric

applications. While small UASs do not provide a universal solution to all mapping requirements, they

offer a cost-effective alternative to ground surveying and traditional photogrammetry for many jobs. With

the continued evolution of platforms and sensors, it is likely that the range of applications for which small

UASs are suitable will continue to expand.

Acknowledgements

The authors are grateful for the financial support received for the case studies. Funding was provided by

the Natural Sciences and Engineering Research Council of Canada, Cenovus Energy, Alberta Innovates,

the Canadian Foundation for Innovation, and the University of Calgary. Jordan Walker (Isis Geomatics) is

acknowledged for his role in provided assistance with some of the data acquisition and image processing.

22

References

Adams River Salmon Society. 2013. 2010 Late Run Sockeye Salmon Preliminary Escapement Estimates.

Available from http://www.salmonsociety.com/DFO%20salmon%20count.pdf [Accessed 26

June 2014]

Baluja, J., Diago, M.P., Balda, P., Zorer, R., Meggio, F., Morales, F. and Tardaguila, J. 2012. Assessment

of vineyard water status variability by thermal and multispectral imagery using an unmanned

aerial vehicle (UAV). Irrig. Sci. 30(6): 511-522. doi: 10.1007/s00271-012-0382-9.

Bellvert, J., Zarco-Tejada, P.J., Girona, J. and Fereres, E. 2013. Mapping crop water stress index in a

Pinot-noir vineyard: comparing ground measurements with thermal remote sensing imagery from

an unmanned aerial vehicle. Precision Agriculture. 15(4): 361-376. doi: 10.1007/s11119-013-

9334-5.

Bendig, J., Bolten, A. and Bareth, G. 2013. UAV-based Imaging for Multi-Temporal, very high

Resolution Crop Surface Models to monitor Crop Growth Variability. Photogrammetrie-

Fernerkundung-Geoinformation. 2013(6): 551-562.

Berni, J., Zarco-Tejada, P.J., Suarez, L. and Fereres, E. 2009. Thermal and narrowband multispectral

remote sensing for vegetation monitoring from an unmanned aerial vehicle. IEEE Transactions

on Geoscience and Remote Sensing. 47(3): 722-738.

Calderón, R., Navas-Cortés, J.A., Lucena, C. and Zarco-Tejada, P.J. 2013. High-resolution airborne

hyperspectral and thermal imagery for early detection of Verticillium wilt of olive using

fluorescence, temperature and narrow-band spectral indices. Remote Sensing of the Environment.

139: 231-245. doi: 10.1016/j.rse.2013.07.031.

Chabot, D. and Bird, D.M. 2012. Evaluation of an off-the-shelf Unmanned Aircraft System for Surveying

Flocks of Geese. Waterbirds. 35(1): 170-174.

Corcóles, J.I., Ortega, J.F., Hernández, D. and Moreno, M.A. 2013. Estimation of leaf area index in onion

(Allium cepa) using an unmanned aerial vehicle. Biosystems Engineering. 115(1): 31-42.

23

Crosetto, M. and Aragues, F.P.: Radargrammetry and SAR Interferometry for DEM Generation

Validation and Data Fusion. 2000. Proceedings of the CEOS SAR Workshop, Toulouse, 26–29

October 1999, ESA SP-450. Available from

http://icc.cat/index.php/content/download/3860/12916/file/radargrammetry_and_sar_interferomm

etry.pdf [accessed 26 June 2014].

Duan, S.B., Li, Z.L., Wu, H., Tang, B.H., Ma, L., Zhao, E. and Li, C. 2014. Inversion of the PROSAIL

model to estimate leaf area index of maize, potato, and sunflower fields from unmanned aerial

vehicle hyperspectral data. International Journal of Applied Earth Observation and

Geoinformation. 26: 12-20. doi: 10.1016/j.jag.2013.05.007.

Eisenbeiss, H. 2009. UAV Photogrammetry. Eidgenössische Technische Hochschule, Zürich.

Fonstad, M.A., Dietrich, J.T., Courville, B.C., Jensen, J.L. and Carbonneau, P.E. 2013. Topographic

structure from motion: a new development in photogrammetric measurement. Earth Surf.

Processes Landforms. 38(4): 421–430.

Foster, M.A. and Stubbendieck, J. 1980. Effects of the plains pocket gopher (Geomys bursarius) on

rangeland. J. Range Manage.. 33(1): 74-78.

Garcia-Ruiz, F., Sankaran, S., Maja, J.M., Lee, W.S., Rasmussen, J. and Ehsani, R. 2013. Comparison of

two aerial imaging platforms for identification of Huanglongbing-infected citrus trees. Computers

and Electronics in Agriculture. 91(0): 106-115.

Gersher, S.: Canada's domestic regulatory framework for RPAS. 2014. A call for public deliberation,

Journal of Unmanned Vehicle Systems, 2(1): 1-4.

Glennie, C.L., Carter, W.E., Shrestha, R.L. and Dietrich, W.E. 2013. Geodetic imaging with airborne

LiDAR: the Earth's surface revealed, Rep. Prog. Phys. 76(8): 086801.

Gonzalez-Dugo, V., Zarco-Tejada, P., Nicolás, E., Nortes, P.A., Alarcón, J.J., Intrigliolo, D.S. and

Fereres, E. 2013. Using high resolution UAV thermal imagery to assess the variability in the

water status of five fruit tree species within a commercial orchard. Precision Agriculture. 14(6):

660-678.

24

Grant, W.E., French, N.R. and Folse Jr, L.J.1980. Effects of pocket gopher mounds on plant production in

shortgrass prairie ecosystems. Southwestern Naturalist. 25(2): 215-224.

Haala, N., Cramer, M., Weimer, F. and Trittler, M. 2011. Performance test on UAV-based

photogrammetric data collection. Proceedings of the International Archives of the

Photogrammetry, Remote Sensing and Spatial Information Sciences. 38(1/C22): 7-12.

Hardin, P.J. and Jensen, R.R. 2011. Small-Scale Unmanned Aerial Vehicles in Environmental Remote

Sensing: Challenges and Opportunities. GIScience & Remote Sensing. 48(1): 99-111.

Hobbs, R.J. and Mooney, H.A. 1985. Community and population dynamics of serpentine grassland

annuals in relation to gopher disturbance. Oecologia. 67(3): 342-351.

Hodgson, A., Kelly, N. and Peel, D. 2013. Unmanned Aerial Vehicles (UAVs) for Surveying Marine

Fauna: A Dugong Case Study. PLoS One, 8(11): e79556.

Huang, Y., Thomson, S.J., Hoffmann, W.C., Lan, Y. and Fritz, B.K.. 2013. Development and prospect of

unmanned aerial vehicle technologies for agricultural production management. Int. J. Agric. Biol.

Eng. 6(3). 1-10.

Hugenholtz, C.H., Moorman, B.J., Riddell, K. and Whitehead, K. 2012. Small unmanned aircraft systems

for remote sensing and earth science research. Eos, Transactions American Geophysical Union.

93(25): 236-236.

Hugenholtz, C.H., Whitehead, K., Barchyn, T E., Brown, O.W., Moorman, B.J., LeClair, A., Hamilton, T.

and Riddell, K. 2013. Geomorphological mapping with a small unmanned aircraft system

(sUAS): feature detection and accuracy assessment of a photogrammetrically-derived digital

terrain model. Geomorphology. 194: 16-24.

Hunt, E.R., Hively, W.D., Fujikawa, S.J., Linden, D.S., Daughtry, C.S.T. and McCarty, G.W. 2012.

Acquisition of NIR-green-blue digital photographs from unmanned aircraft for crop monitoring

Remote Sens. 2(1): 290-305.

Huntly, N. and Inouye, R. 1988. Pocket gophers in ecosystems: patterns and mechanisms. BioScience.

38(11): 786-793. doi: 10.2307/1310788.

25

Jones, G.P., Pearlstine, L.G. and Percival, H.F. 2006. An Assessment of Small Unmanned Aerial Vehicles

for Wildlife Research. Wildlife Society Bulletin. 34(3): 750-758.

Knoth, C., Klein, B., Prinz, T. and Kleinebecker, T. 2013. Unmanned aerial vehicles as innovative remote

sensing platforms for high-resolution infrared imagery to support restoration monitoring in cut-

over bogs. Applied Vegetation Science. 16(3): 509-517.

Laliberte, A.S., Herrick, J.E., Rango, A. and Winters, C. 2010. Acquisition, orthorectification, and object-

based classification of unmanned aerial vehicle (UAV) imagery for rangeland monitoring.

Photogramm. Eng. Remote Sens. 76(6): 661-672.

Laliberte, A.S. and Rango, A. 2011. Image processing and classification procedures for analysis of sub-

decimeter imagery acquired with an unmanned aircraft over arid rangelands. GIScience &

Remote Sensing. 48(1): 4-23.

Lin, Y., Hyyppa, J. and Jaakkola, A. 2011. Mini-UAV-borne LIDAR for fine-scale mapping. IEEE

Geoscience and Remote Sensing Letters. 8(3): 426-430.

Lucieer, A., de Jong, S. and Turner, D. 2014. Mapping landslide displacements using Structure from

Motion (SfM) and image correlation of multi-temporal UAV photography. Progress in Physical

Geography. 38(1): 97-116. doi: 10.1177/0309133313515293.

Martinez-De Dios, J.R. and Ollero, A. 2006. Automatic Detection of Windows Thermal Heat Losses in

Buildings Using UAVs. Automation Congress, 2006. WAC '06. doi:

10.1109/WAC.2006.375998.

Mulero-Pazmany, M., Stolper, R., van Essen, L.D., Negro, J.J. and Sassen, T. 2014. Remotely Piloted

Aircraft Systems as a Rhinoceros Anti-Poaching Tool in Africa. PLoS One. 9(1): e83873.

Mulla, D.J. 2013. Twenty five years of remote sensing in precision agriculture: Key advances and

remaining knowledge gaps. Biosystems Engineering. 114(4): 358-371.

Niethammer, U., James, M.R., Rothmund, S., Travelletti, J. and Joswig, M. 2012. UAV-based remote

sensing of the Super-Sauze landslide: Evaluation and results. Eng. Geol. 128(0): 2-11.

26

Reichman, O.J., Bendix, J.H. and Seastedt, T. R. 1993. Distinct animal-generated edge effects in a

tallgrass prairie community. Ecology. 74(4): 1281-1285. doi: 10.2307/1940496.

Richens, V.B.: Notes on the digging activity of a northern pocket gopher. 1966. J. Mammal. 47(3): 531-

533. doi: 10.2307/1377704.

Rothmund, S., Niethammer, U., Malet, J.P. and Joswig, M. 2013. Landslide surface monitoring based on

UAV-and ground-based images and terrestrial laser scanning: accuracy analysis and

morphological interpretation. First Break. 31(8): 81-87.

Sarda-Palomera, F., Bota, G., Vinolo, C., Pallarés, O., Sazatornil, V., Brotons, L., Gomariz, S. and Sarda,

F. 2012. Fine-scale bird monitoring from light unmanned aircraft systems. Ibis. 154(1), pp:177-

183. doi: 10.1111/j.1474-919X.2011.01177.x.

Sauerbier, M., Siegrist, E., Eisenbeiss, H. and Demir, N. 2011. The practical application of UAV-based

photogrammetry under economic aspects, International Archives of the Photogrammetry. Remote

Sensing and Spatial Information Sciences. ISPRS Zurich 2011 Workshop, 14-16 September 2011.

38(1/C22): 45-50.

Spencer, S.R., Cameron, G.N., Eshelman, B.D., Cooper, L.C. and Williams, L.R. 1985. Influence of

pocket gopher mounds on a Texas coastal prairie. Oecologia. 66(1): 111-115.

Stagakis, S., González-Dugo, V., Cid, P., Guillén-Climent, M.L. and Zarco-Tejada, P. J. 2012.

Monitoring water stress and fruit quality in an orange orchard under regulated deficit irrigation

using narrow-band structural and physiological remote sensing indices. ISPRS Journal of

Photogrammetry and Remote Sensing. 71: 47-61.

Tamminga, A., Hugenholtz, C., Eaton, B. and Lapointe, M. 2014. Hyperspatial remote sensing of

channel reach morphology and hydraulic fish habitat using an unmanned aerial vehicle (UAV): a

first assessment in the context of river research and management. River Research and

Applications. DOI: 10.1002/rra.2743.

Transport Canada. 2008. The review and processing of an application for a Special Flight Operations

Certificate for the Operation of an Unmanned Air Vehicle (UAV) System. Available from

27

http://www.tc.gc.ca/eng/civilaviation/opssvs/managementservices-referencecentre-documents-

600-623-001-972.htm [Accessed 26 June 2014].

Transport Canada. 2012. UAV systems program design working group phase 1 final report. Available

from http://www.h-a-c.ca/UAV_REPORT.pdf [Accessed 26 June 2014].

Turner, D., Lucieer, A. and Wallace, L. 2014. Direct Georeferencing of Ultrahigh-Resolution UAV

Imagery. IEEE Transactions on Geoscience and Remote Sensing. 52(5): 2738 - 2745. doi:

10.1109/TGRS.2013.2265295.

Turner, D., Lucieer, A. and Watson, C. 2012. An automated technique for generating georectified

mosaics from ultra-high resolution Unmanned Aerial Vehicle (UAV) imagery, based on Structure

from Motion (SfM) point clouds. Remote Sens. 4(5): 1392-1410.

Turner, G.T., Hansen, R.M., Reid, V.H., Tietjen, H.P. and Ward, A.L. 1973. Pocket gophers and

Colorado mountain rangeland. Colorado State University Agriculture Experimental Station

Bulletin. 554S: 90p.

Urbahs, A. and Jonaite, I. 2013. Features of the use of unmanned aerial vehicles for agriculture

applications. Aviation. 17(4):170-175.

Vermeulen, C.D., Lejeune, P., Lisein, J., Sawadogo, P. and Bouché, P. 2013. Unmanned Aerial Survey of

Elephants, PLoS One 8(2): e54700.

Wallace, L., Lucieer, A., Watson, C. and Turner, D. 2012. Development of a UAV-LiDAR system with

application to forest inventory. Remote Sens. 4(6):1519-1543.

Watts, A.C., Ambrosia, V.G. and Hinkley, E.A. 2012. Unmanned aircraft systems in remote sensing and

scientific research: Classification and considerations of use. Remote Sens. 4(6): 1671-1692.

Westaway, R.M., Lane, S.N. and Hicks, D.M. 2000. The development of an automated correction

procedure for digital photogrammetry for the study of wide, shallow, gravel-bed rivers. Earth

Surf. Processes Landforms. 25(2): 209-226.

Whitehead, K., Moorman, B.J. and Hugenholtz, C.H. 2013. Low-cost, on-demand aerial photogrammetry

for glaciological measurement, The Cryosphere. 7(6): 1879-1884.

28

Zarco-Tejada, P.J., Guillén-Climent, M.L., Hernández-Clemente, R., Catalina, A., González, M.R. and

Martín, P. 2013. Estimating leaf carotenoid content in vineyards using high resolution

hyperspectral imagery acquired from an unmanned aerial vehicle (UAV). Agricultural and Forest

Meteorology. 171: 281-294.

Zhang, C. and Kovacs, J.M. 2012. The application of small unmanned aerial systems for precision

agriculture: a review. Precision Agriculture. 13(6): 693-712.

29

Figure 1: Four small UASs were used in the seven case studies presented herein: (a) RQ-84Z Aerohawk (Hawkeye

UAV Ltd., New Zealand), (b) Outlander UAS (CropCam Inc., Canada), (c) Aeryon Scout Pro quadcopter

(Aeryon Labs Inc., Canada), and (d) Ebee (senseFly Inc., Switzerland).

30

Figure 2: UAS orthomosaics and DEMs used for determining changes in stockpile volume following excavation

(Case study 1). The images in (a) and (b) show the orthomosaic for the June survey and the corresponding DEM,

respectively, while the images in (c) and (d) show the orthomosaic for the November survey and the

corresponding DEM, respectively. The outline of the stockpile prior to excavation is denoted by the dashed line

in each image.

Figure 3: River morphology and aquatic habitat analyses of the Elbow River, 2012 (Case study 2), showing (a) the

orthoimage, (b) the DEM, (c) D50 values and digitised cover features, (d) modeled depth, (e) modeled velocity,

and (f) composite suitability index values for adult brown trout habitat.

River morphology and aquatic habitat analyses of the Elbow River, 2012 (Case study 2), showing (a) the

alues and digitised cover features, (d) modeled depth, (e) modeled velocity,

and (f) composite suitability index values for adult brown trout habitat.

31

River morphology and aquatic habitat analyses of the Elbow River, 2012 (Case study 2), showing (a) the

alues and digitised cover features, (d) modeled depth, (e) modeled velocity,

32

Figure 4: The application of UAS-based remote sensing to identify late blight in a potato crop (Case study 3). The

image sequences (a-c) and (d-f) are from two separate flights in mid and late July 2013, respectively. For the mid

July flight the visible imagery is shown in (a), the NIR imagery is in (b), and the calculated NDVI is shown in

(c). For the late July flight the visible imagery is shown in (d), the NIR imagery is in (e), and the calculated

NDVI is shown in (f). Areas with confirmed or suspected late blight are circled. In the NDVI images the area

outside the field was masked.

33

Figure 5: Object-based classification of gopher mounds in the Great Sand Hills, Saskatchewan (Case study 4).

Image (a) is the colour composite of the 1.92 km2 study site. Image (b) is a detail close-up of a roughly 60x60m

area with a linear stretch to highlight the bright yellow mounds while (c) shows the result of the binary object-

based feature extraction (mounds are in black). (d) shows a cluster of 1-2 year old mounds as they appear on the

ground.

34

Figure 6: Georeferenced image mosaic of the lower Adams River, British Columbia (Case study 5). Insets show

large concentrations of salmon. Spawning salmon can be seen as individual dots adjacent to the main

concentrations.

35

Figure 7: Cattle enumeration based on thermal imaging from the Aeryon Scout Pro UAS (Case study 6).

The image in (a) shows a portion of the CAFO, while (b) shows an oblique photo with individual cows

identified. Two half-ton pickup trucks are circled in (a) for scale.

36

Figure 8: Fountain Glacier orthoimage mosaics from 2010 (a), and from 2011 (b) (Case study 7). Both images have

10 m contours superimposed.