© Author(s) 2015. CC Attribution 4.0 License.

ISSN 2047-0371

2.1.7. Unmanned Aerial Vehicles (UAVs) and their application

in geomorphic mapping
Christopher Hackney1 and Alexander I Clayton1
1
Geography and Environment, University of Southampton, UK ([email protected])

ABSTRACT: Detailed topographic surveys are a pre-requisite for many studies into Earth surface
processes and dynamics. Often such surveys are required for large (>10 km2) areas and at a
relatively high temporal resolution (sub-daily to daily) for use in hazard monitoring, monitoring
ecological change, and detailed process studies. Techniques such as Terrestrial Laser Scanning,
Total Stations and low-level aerial photography via chartered light aircraft flights may provide the
spatial resolution required, but are often costly and time-intensive, making them less viable in
obtaining the temporal resolution necessary. Further still, satellite imaging platforms often produce
products whose image resolution is too coarse to resolve fine scale topography. Recent
technological advances have seen the development of Unmanned Aerial Vehicles (UAVs) as a
platform from which to acquire aerial photos over large spatial scales at high temporal resolution.
These photos may then be combined as orthophotos for spectral analysis, or used to generate
useful digital terrain models through Structure from Motion (SfM) photogrammetric techniques. The
tandem development of low-cost, rapid deployment UAV platforms and SfM algorithms has seen
the rapid growth in in the application of UAVs for generating high-resolution topographic data. Here
we detail some of the considerations needed before deployment of UAV systems, before showing
how UAVs may be used to collect high resolution aerial photos to enable generation of pro-glacial
topography in Iceland.
KEYWORDS: UAVs; fixed-wing platform, rotor-wing platform, aerial surveys; topographic models

Introduction Previously the domain of the military, UAVs

have seen an increase in civilian and
The acquisition of high resolution topographic academic use, partly driven by the
data is key to many studies in Earth science. improvements in affordable miniature GPS
For mapping studies requiring data at high and Inertial Measurement Units (IMUs) which
temporal (hourly, daily), and large spatial (>5 enable accurate operation of UAV systems.
km2), scales traditional surveying methods UAVs come in a range of designs. Large
are often costly and time intensive. Recent fixed-wing platforms have been adapted from
advances in technology have seen the military-grade platforms and are typically 5 m
advent of digital photogrammetry as a viable of more in wingspan and may carry payloads
means of obtaining such high resolution greater than 200 kg. These systems may
topographic data (Smith et al., 2009; Rosnell have an extended range of ~500 km but
and Honavaara, 2012; Fonstad et al., 2013). require full aviation clearance and need a
In tandem, the development and increased large ground operations team (Anderson and
affordability of Unmanned Aerial Vehicles Gaston, 2013). Smaller UAV systems may
(UAVs) as a novel platform with which to come as either fixed-wing or multi-rotor
collect the low-level aerial photography systems. At this scale, many off-the-shelf
needed for such photogrammetry has seen a designs and user-built kit systems are
rapid increase in their usage in available. Small fixed-wing UAVs may be
geomorphological studies (Lejot et al., 2007; only a couple of meters wide. Small rotor-
Hugenholtz et al., 2013). wing platforms may have up to eight rotors
and may only weight one or two kilograms. At

British Society for Geomorphology Geomorphological Techniques, Chap. 2, Sec. 1.7 (2015)
 Unmanned Aerial Vehicles (UAVs) 2
the even smaller scale, Micro-dones may
weigh less than kilogram, however they have Applications of UAVs in geomorphology are
limited flight durations (~10 mins) and wide ranging and include, for example,
payloads (<1 kg). surveying fluvial bathymetry to map and
monitor gravel bar location and change using
Of particular interest to the geomorphological a small remote controlled motorized vehicle
community are the small, mini- and micro- (Lejot et al., 2007). d’Oleire-Oltmanns et al.
UAV systems (Anderson and Gaston, 2013).
(2012) deployed a small fixed-wing platform
The flexibility in operation and shorter
to monitor rates of soil erosion over a 6km2
response times afforded by small UAV
area in Morocco, showing how small UAVs
systems means they enable rapid
may be used to bridge the gap between field
deployment and the collection of high
scale and satellite imagery. Repeat
spatial- and temporal-resolution datasets
topographic surveys over a 2.5 km2 gully
where traditional aerial techniques, including
allowed Grellier et al. (2012) to constrain
larger UAV systems, may not.
rates of gully erosion and vegetation change
allowing them to elucidate subsurface
The use of UAVs in geomorphological
processes controlling gully evolution.
mapping is often facilitated by the application
Niethammer et al. (2012) deployed a quad-
of photogrammetric techniques such as
copter, rotary-wing platform to monitor
Structure from Motion (SfM) (e.g. Harwin and
landslides in the Southern French Alps,
Lucieer, 2012; Westoby et al., 2012;
mapping failures, source and sink zones from
Micheletti et al. 2015). SfM utilises
orthophotos and digital terrain models
overlapping imagery acquired from multiple
(DTMs) generated from the UAV imagery.
viewpoints to reconstruct the camera position
UAVs have also been deployed in glacial
and camera geometry. From these
environments, with Whitehead et al. (2013),
reconstructed camera locations it is then
using a fixed-wing UAV system carrying a
possible to generate spatial relationships
Lumix LX3 camera to monitor glacial ablation
between common feature points and thereby
in consecutive ablation seasons on the
generate a feature’s structure (Westoby et Fountain Glacier, Canada.
al., 2012; Fonstad et al. 2013; Micheletti et al.
2015). Given correct deployment and Additionally, and as technology and
attainment of accurate ground control data, capabilities evolve, the ability of UAVs to
the horizontal accuracy and precision of carry variable payloads will open up the
resultant aerial imagery and Digital Elevation possibility of using multispectral sensors to
Models (DEMs) generated through SfM can add value to mapping projects, for example it
be better than satellite imagery and aerial is already possible to detect the
LIDAR (±0.2 m; Fonstad et al., 2013; geomorphological controls on crop production
from combined multi-spectral and traditional
Hugenholtz et al., 2013), whilst vertical
photogrammetric techniques using UAV
accuracy is typically better than ±0.1m systems (Dunford et al., 2009; Martinez-
(Fonstad et al., 2013). Casasnovas et al., 2013). More recent
advances have seen the application of
After some years of development survey-grade Lidar equipment on small fixed-
commercially available small fixed-wing and wing UAV platforms (Lin et al., 2011). This
rotor-wing UAV systems now enable the low- advance will allow the acquisition of
cost acquisition of aerial photos over large topographic data beneath vegetation which
areas at high temporal resolution. Coupled current SfM algorithms do not.
with the concurrent development of SfM
techniques (Micheletti et al., 2015) and a The rest of this article will focus specifically
greater appreciation of the potential errors on the use of small UAVs (c. Anderson and
introduced by these methodologies (James Gaston, 2013) in obtaining aerial imagery for
and Robson, 2014; Nouwakpo et al., 2014), the purpose of geomorphological mapping. It
low-cost, rapid, high resolution topographic will outline considerations when selecting a
data for use in geomorphological mapping suitable UAV platform. It will then provide
collection is now becoming common place in some background as to the legal framework
the geosciences.. within which such studies must be conducted
in the EU. Finally, it will provide a case study

British Society for Geomorphology Geomorphological Techniques, Chap. 2, Sec. 1.7 (2015)
3 Christopher Hackney and Alexander Clayton
example of the deployment of a small fixed- those which may be operated under CAA
wing UAV in a pro-glacial environment in requirements (see below for more details).
Iceland, which details good practice
workflows and site-specific considerations for
operations in remote and topographically
Fixed-Wing Platforms
restrictive environments. Fixed-wing platforms (Figure 1A) are perhaps
the most common form of UAV. Wings and
bodies are normally constructed from
Considerations lightweight polystyrene with a wingspan
Hardware typically <2m. The lift characteristics of a
flying wing mean that, relative to rotor-wing
UAVs commonly used for geomorphological platforms, their fuel efficiency is high. The
surveying are predominately built around two limited number of moving parts and
types of airframe; fixed-wing and rotor-wing. lightweight design also means that damage
Both platforms are used in geomorphological inflicted by hard landings is more limited than
applications working off of the same with rotor-wing designs (although it is less
theoretical standpoint; the acquisition of likely that hard landings will occur with rotor-
overlapping, photographs which can later be wing designs). They are launched from the
used with SfM (Fonstad et al., 2013; ground either by hand or, more commonly
Micheletti et al., 2015) algorithms to now, with use of a catapult. Whilst the
reconstruct topography. Recent work has apparatus is small, the necessity for a non-
shown how topographic datasets derived vertical climb to altitude means they require
from UAV derived aerial photos are improved more space to take off than their rotor-wing
by having images captured from non-parallel counterparts. They are larger in size than
viewing locations (James and Robson, 2014). their rotor-wing counterparts (Figure 1) and
That is to say, it may be beneficial to collect as such require more space to operate. They
imagery from more unstable platforms which often also have more associated peripheries
enable more photo acquisition form non- (additional laptop and launching gear, Figure
uniform camera locations. A summary of 1A) required for their operation than rotor-
typical small fixed-wing and rotary-wing wing UAVs. However, recent advances in
systems is provided in Table 1. These fixed-wing platforms have seen an increase
systems reflect those commonly used in in their flight endurance such that they can
geomorphological studies and do not now fly longer than rotor-wing platforms
represent the actual maximum values enabling greater areas to be covered with
obtained by larger UAV systems (Anderson more ease. They are often controlled in-flight
and Gaston, 2013). Note the specifications
and details provided below are limited to

Figure 1: A) A fixed-wing UAV system with associated peripheries required for its operation. B) A
quad-copter rotor-wing UAV system and associated operational peripheries. The same 1.5 m2
ground control target has been used a background to provide consistent scale between both A)
and B).

British Society for Geomorphology Geomorphological Techniques, Chap. 2, Sec. 1.7 (2015)
 Unmanned Aerial Vehicles (UAVs) 4
Table 1: Technical specification comparison between typical small fixed-wing and multi-rotor UAV
platforms which may be used in geomorphological studies. Values in italicised parentheses relate
to CAA operating regulations as per CAP393 and CAP722. Values as reported in Hugenholtz et al.
(2013), Mancini et al. (2013) and Anderson and Gaston (2013).

Fixed-wing UAVs Multi-rotor UAVs

Wingspan (m) 1–3 <1

Flight time (mins) 20 - 60 20

Max payload Weight (kg) 30 (7) 15 (7)

Max. Speed (km/h) 50 – 80 (130) 30 – 50 (130)

Operating Range (km) 1 - 5 (0.5) 1 – 2 (0.5)

Altitude Range (m) > 2000 (121) 400 (121)

by built-in autopilots, with flight plans pre- blades, Figure 1B), hexa-copters (with six
programmed before deployment. This means rotary blades), and octa-copters (with eight
they require less user interaction in flight, and rotary blades) which are becoming the
are more stable, than rotor-wing platforms. standard for heavy lift photography work.
However, this has implications for Their footprint is typically smaller than fixed-
topographic data sets derived from photos wing platforms (~0.8 m). Vertical take-off and
obtained from fixed-wing platforms, as the landing means that they do not require
increased stability of the platform and pre- extensive unconstrained landing sites and
programmed, often parallel flight lines may can be deployed from relatively inaccessible
introduce errors into the topographic dataset areas. This favours rapid deployment and
(James and Robson, 2014). enables access to areas previously
unfeasible with traditional survey techniques
Fixed-wing airframes inevitably require more and fixed-wing UAV platforms.
space than rotor-wing options. Often a ~100
m strip is sensible to allow for overrun and Due to their smaller footprint, rotor-wing
variations in headwind strength. There is also UAVs have limited flight endurance (typically
a need for the survey area to be less less than 20 mins) when compared to fixed-
constricted (e.g. from valley side walls) than wing platforms, requiring many battery packs
is required with rotor-wing UAVs as they have and/or recharging units to accomplish the
a larger footprint and may be operated at a same spatial coverage as fixed-wing UAVs.
greater range from the operator. This increases the likelihood of inclement
weather impacting the survey. However, as
Fixed-wing UAVs may be more suited to rotor-wing platforms typically fly at slower
topographic surveys over larger spatial speeds than fixed-wing UAVs, and often
scales due to their longevity in flight. contain better gimbals, they are more stable
However, their application to locations at higher wind speeds than fixed-wing UAVs.
bounded by terrain, trees or other This permits a trained pilot greater control of
obstructions may limit their successful the UAV, and facilitates the collection of non-
operation. Likewise, their relative lack of parallel survey lines under a wider range of
maneuverability may provide challenges in wind conditions. It is therefore likely that
certain situations. topographic datasets derived from rotor-wing
systems are likely to be of higher quality than
equivalent data-sets produced from fixed-
Rotor-wing Platforms wing systems as their ability to be more
Rotor-wing platforms comprise a suite of flexible in their survey lines will permit the
designs ranging from common helicopter acquisition of a more non-uniform set of
designs to quad-copters (with four rotary photos (James and Robson, 2014).

British Society for Geomorphology Geomorphological Techniques, Chap. 2, Sec. 1.7 (2015)
5 Christopher Hackney and Alexander Clayton
enable greater detail of the surface to be
Additionally, unlike fixed-wing platforms, captured.
rotor-wing platforms are able to hover over Although these techniques would be less cost
objects and locations. This facilitates higher effective than photogrammetric techniques
precision photogrammetry over features of for obtaining topographic data, they will still
interest and allows for complete 3D require careful planning of flight lines,
inspection of stationary objects, whilst also acquisition of a detailed ground control
opening the possibility of at-a-point temporal network and will still have to operate within
sequences in measurements. Similarly, it the legal restrictions.
enables users to survey the same feature at
different altitudes, thereby assessing issues Ground Control
of pixel resolution and photo quality as a
function of varying altitude. For the successful registration and alignment
of photos collected from any UAV platform
and sensor, an accurate and precisely
Sensing Applications
located network of Ground Control Points
Both fixed-wing and rotor-wing UAVs have (GCPs) must be acquired. To ensure that no
payloads capable of carrying small to lens warping, or ‘doming’ (c. James and
medium size digital cameras and video Robson, 2014) is present in the final product,
recorders (and are restricted to payloads of 7 it is vital the GCPs are distributed throughout
kg under CAA regulations; see below for the study area. Of particular importance is the
further details). These cameras can be placement of GCPs close to the edges of the
simple RGB digital cameras suitable for the survey area, where potential doming may be
acquisition of high resolution aerial photos exacerbated. The number of GCPs deployed
which can subsequently be used in the depends on the overall aim of the survey. For
generations of georectified ortho-photos or example, for geomorphic mapping it is more
digital elevation models (e.g. Fonstad et al., important to distribute GCPs across the
2013; Hugenholtz et al., 2013). Alternatively, survey area to ensure doming and
these sensing unit can be hyper-spectral registration errors are constrained. However,
cameras which can be used to enable an if the survey is designed for obtain detailed
assessment of the local water stress (Zarco- topography, then it is important that subtle
Tejada et al., 2012) or agricultural and variations in topography are accurately
forestry health (e.g. Saari et al., 2011). resolved. As such a denser network of GCPs
in locations of interest may be necessary to
More recently, UAVs have been deployed ensure such fine-scale topography is
with small Lidar sensors (Lin et al., 2011; captured. It is worth noting, that it is still
Wallace et al., 2012). This growing area important to maintain a good distribution of
allows for high resolution topographic surveys GCPs in this case, to avoid doming of the
from UAVs to be conducted without recourse final DEMs.
to SfM software. It should be noted however
that the current suite of Lidar sensors which GCPs normally consist of brightly coloured
may be deployed on UAVs are less high targets which will be visible within the aerial
powered than traditional Lidars and have a photographs, laid out by the operator evenly
higher signal-to-noise ratio. In addition, the across the survey area. The GCPs should be
development of smaller Lidar sensors larger than the pixel resolution of the sensor
capable of being lifted by UAVs could onboard the UAV such that they are clearly
possibly enable more data to be obtained visible within the photographs. As pixel
than tradition photogrammetric techniques, resolution varies in size with survey elevation,
for example Lidar intensity return data can be surveys conducted at higher altitudes require
used to identify moisture variability and larger GCPs. The pixel size (PR) for a sensor
surface geology variability more easily than with a given focal length (LF) and pixel
would be visible from RGB camera images. dimension (PD), at a given altitude (A) can be
Furthermore, one of the current limitations calculated using the following calculation
with SfM is that the cameras used are
PR = PD(A/LF) (1)
passive sensors, i.e. they cannot penetrate
through vegetation. Having active Lidar, and Equation 1 can therefore be used to
phased-based sensors onboard UAVs would determine the required size of GCPs for a
survey of any given altitude. For example,

British Society for Geomorphology Geomorphological Techniques, Chap. 2, Sec. 1.7 (2015)
 Unmanned Aerial Vehicles (UAVs) 6
Smith et al. (2009) deployed 30 GCPs across
a 40 x 50 m survey area. Their GCPs EU law states that UAVs operated in any EU
consisted of 20 cm laminated sheets country airspace must be kept within the
containing a black and white target design. visual line of sight. This equates to
Smith et al. (2009) flew their survey at 50 m approximately 500 m in the horizontal and
altitude. For surveys flown at higher altitudes 400 feet in the vertical, although this is
(> 50m) the size of GCPs deployed are heavily dependent upon local weather
recommended to be larger than 1 m2 to be conditions and terrain. This has practical
captured in the aerial photos. Alternatively, it implications for survey and flight line design,
is possible to use unique ground features as limiting the operational space achievable with
GCPs, for example, Fonstad et al. (2013) each flight. As such, it is likely that multiple
identified features on the ground to use as launches will need to be made if your area of
GCPs which provide fixed features in the interest covers a substantial area. Further,
survey photos (e.g. field corners, lake edges, this increases the importance of a dense
road junctions). ground control network (see above) to aid
stitching of photos and georectification during
Once a network of GCPs has been post-processing.
constructed, the position of each GCP must
be obtained. This can either be a real world Additionally, UAVs may not be flown within
coordinate of the GCP obtained through the 50 m of a member of the public (with the
use of a differential Global Positioning exception of the operating crew), thus limiting
System (dGPS) or a relative position to an their use within public spaces and over tourist
arbitrary coordinate grid obtained through a locations. Similarly, they may not be flown
total station or similar surveying equipment. over or within 150 m of any organised open-
Either technique should result in positional air assembly of more than 1,000 people.
accuracies of approximately ± 0.05 m. The
technique used depends upon the ultimate
goal of the survey being conducted. If the Case Study
data is to be used in conjunction with other Science questions and aims
data sets and registered within a real-world
context then global coordinates will be Proglacial zones are highly dynamic regions
necessary. However, if this is not the case which are subject to seasonal variations in
then it is possible to use an arbitrary, relative, energy regime and thus geomorphological
coordinate system. activity. These regions are often inaccessible
and remote. The foreland of Skalafellsjokull,
Legal Limits Iceland (Figure 2) includes a series of well-
preserved push moraines, the spacing of
The legal requirements for UAV flight vary which relate to local climatic conditions
between countries. It is necessary to (Boulton, 1986; Bennett, 2001). Yet, our
research country specific regulations before understanding of how these features respond
any survey work is undertaken to ensure you to climatic variations at Skalafellsjokull is
meet the requirements. poorly understood. In order to map these
features and determine accurately the inter-
Many legal considerations exist when seasonal spacing, a high resolution DEM was
planning a survey with a UAV. Within the EU required (see Chandler et al., In Prep, for
flights undertaken with a UAV must adhere to details). These features are located in a
central legislation (and the reader is guided topographically constrained region which
towards the Civil Aviation Authority (the does not easily facilitate high resolution
governing body within the UK) protocols mapping with terrestrial laser scanning, Lidar
CAP393 and CAP722 for full details of the or satellite radar mapping. As such, the use
legislation which is applicable across the of a UAV presented a lower cost, more time
entire EU). Here (for brevity) we will cover a effective option than standard aerial surveys
few key points arising from this legislation and due to the size of the site (2 km2) a
which require consideration when planning terrestrial approach would not have been
geomorphological surveys. It is stressed that suitable. In addition, satellite imagery is of too
you check the legislation before deployment coarse a resolution to resolve the spatial
at each new site to ensure you adhere to the scales of the push moraines.
rules.
British Society for Geomorphology Geomorphological Techniques, Chap. 2, Sec. 1.7 (2015)
7 Christopher Hackney and Alexander Clayton

Figure 2: Google Earth image of Heinbergjokull and Skallafellsjokull in Icealand. The UAV flight
lines are depicted in yellow and blue dots represent the GCPs used to register the aerial photos.

Methodology and Results ±0.05 m. in the horizontal and vertical,

To that end, during 2013 a Quest 200 fixed- respectively.
wing UAV carrying a Panasonic Lumix LX5 Fifteen ground control point targets were
off-the-shelf, point-and-shoot camera (see created in the field from orange plastic
Table 2 for details) was used to survey material and measured 2 m 2. Tape was used
proglacial and ice marginal geomorphology at to indicate the centre of the target. The use of
Skalafellsjokull and Heinbergjokull in Iceland tape larger than the pixel resolution enabled
(Figures 2 and 3). The aim of the surveys a precise determination of target centres. The
was to obtain low level aerial photography targets were deployed in a grid network
which could subsequently be processed with spaced approximately 0.5 km apart and a
SfM software to produce high resolution density of 1.2 points/km2; this results in ~0.03
topographic surveys of the study area (Figure
GCPs per image (Figure 2). GCPs were
3). The UAV was used in conjunction within
deployed before the UAV was launched to
Leica dGPS deployed in Real-Time
ensure they would be visible in the survey.
Kinematic (RTK) mode to limit distortion in
the final photogrammetric product. RTK
The resulting orthophoto (Figure 3A) and
corrections were accurate to ±0.01 m and DEM (Figure 3B) have spatial resolutions of
0.05 and 0.1 m respectively, and are of
suitable resolution to be able to determine
British Society for Geomorphology Geomorphological Techniques, Chap. 2, Sec. 1.7 (2015)
 Unmanned Aerial Vehicles (UAVs) 8
sub-seasonal variations in push moraine beyond known limits, experience and ability
location (Chandler et al., In Prep). It is noted in the field.
that the resolution of the orthophoto and DEM
Table 2: Attributes of the Skalafellsjokul and
vary with topography, particularly as this site
Heinbergjokull surveys conducted in 2013.
displays large relief (~100 m).
Attribute Value
Pre-departure checks UAV Platform Quest UAV 200
Prior to departing to the field, the flight Fixed-wing
regulations for Iceland were checked and
accounted for in all pre-flight planning. The Sensor Panasonic
legal framework in Iceland is different to Lumix LX5
those in place within the EU (see above for
EU restrictions). The Icelandic CAA stated Survey altitude (m) 100
that as long as the UAV was <5 kg there
were no specific requirements on altitude and Photo endlap (%) 80
range. This relatively relaxed regulatory
stance is a function of the sparsely populated Photo sidelap (%) 60
area and allowed for more freedom in survey
planning than would otherwise have been No. of images captured 1980
available within the EU. However, during the
field season surveys were flown within the Image resolution (m) 0.05
EU limitations (see above) to avoid any
issues which may arise with operating DEM resolution (m) 0.1

Figure 3: A) Final registered orthophoto of Skallafellsjokull generated from aerial photos captured
using the UAV. B) DEM of the Skallafellsjokull proglacial foreland derived from Agisoft Photoscan.
The push moraines are identified by the dashed circles in both A and B.

On-site considerations changeable, with the possibility of

unpredictable wind and precipitation.
Access, logistics and changeable weather Accordingly, the survey was carried out as
limited the survey time available, as such the series of small surveys to avoid the prospect
deployment methodology was designed to be of having to cut a longer survey short, thus
flexible and quick. As in many locations potentially compromising the acquisition of
worldwide, surveying conditions are highly data. This is facilitated by the rapid

British Society for Geomorphology Geomorphological Techniques, Chap. 2, Sec. 1.7 (2015)
9 Christopher Hackney and Alexander Clayton
deployment of the fixed-wing UAV, and is one turbulence meant that it was impossible to
of the advantages of using such technology remove this issue from the survey operation.
in these challenging areas. Each survey Therefore, to limit any potential damage to
lasted approximate 20 minutes and covered the UAV and operation team, a sensible
an area of 2 km2. In total an area of 15 km2 emergency rally point for the aircraft was set
was surveyed over a total of 7 days. After on the on board auto-pilot and constant visual
each survey was conducted the ground on the UAV was kept during the survey. If it
control point targets were registered with the was felt the atmospheric conditions were
dGPS and collected. The above process was exceeding the pilot’s operational abilities, the
repeated over a total of 9 surveys, to ensure
UAV was recalled to its emergency recall
the entire survey area was captured.
point and the survey was halted until
conditions became more favourable. At all
The study site (Figure 2) is characterised by
times vigilance on behalf of the operating
rugged landscapes and highly changeable
crew was essential.
weather conditions. Accordingly, the selection
of a suitable operating base which permits
Terrain, in any mountainous environment,
accessibility, safe take-off and landing sites
presents difficulty to a UAV operator. In
and enough space to permit safe operating
Iceland the expansive unpopulated glacial
practices is necessary. Similarly it places
valleys made judging distances difficult, safe
more emphasis on all pre-flight checks
landing spots are rare and maintaining line of
including obtaining accurate weather
sight to the aircraft can require limiting
forecasts the day before deployment, as well
surveys. Limiting the impacts these issues
as detail on foot (or by vehicle)
requires careful survey planning. Initially,
reconnaissance of the field site(s) to identify
reconnaissance trips to the field site were
access points and areas of shelter and safety
conducted prior to any deployment to
should the weather turn all of a sudden.
familiarise the operational team with the
survey location and surround topography.
As with any UAV survey conducted in a
Prior to departing on the field trip, an
maritime climate at high latitudes, weather
accurate and recently geo-referenced image
was the primary concern. Despite being in
was obtained such that in situations when
Eastern Iceland frequent rain storms were
communication with the UAV is lost, and the
possible and high winds probable. There is
operator is reliant on visual cues, they have a
little that can be done to mitigate these
detailed map of the surrounding area. It was
factors, but test flights in the area were an
also decided to identify numerous landing
essential part of our preparation for the actual
sites which would enable to the UAV to be
surveys. These helped establish a rough
landed when conditions became operationally
guide to the vertical profile of wind speeds
difficult. As it is beneficial for the pilot to circle
and give an indication as to safe flying
into a landing, topographically unconstrained
conditions.
sites of different orientation were selected to
cover a range of possible wind directions.
However, whilst wind speeds provided a
good indication of flying conditions and
The most difficult landings occurred when the
incoming precipitation was simple to spot, the
survey area moved onto the glacier foreslope
turbulence caused by mixing air masses
and the UAV was landing on the bare ice of
proved difficult to predict and was a major
Skalafellsjokull. Despite flying from the
issue to the operational safety of the UAV.
flattest area there remained approximately
Whilst this is probably common in many
0.5 m of relief in the ice topography. Whilst
valley confined environments, the mixing of
landing on snow proved simple, it was difficult
the katabatic winds off Vatnajokull with air
to mitigate the damaging landings on the
masses blowing off the Atlantic frequently
hard ice. In low wind conditions, even by
resulted in unpredictable UAV behavior in
flaring the aircraft almost into a stall before
apparently stable conditions. Telemetry
landing, the impact was hard and caused
recorded on the UAV logger revealed that the
some damage to the airframe. This issue
time of day, local valley topography and
could be avoided if a parachute option (now
valley floor wind direction all appeared to
built in to some UAV systems) was available.
correlate with turbulence at height. However,
the lack of capability to monitor the upper air

British Society for Geomorphology Geomorphological Techniques, Chap. 2, Sec. 1.7 (2015)
 Unmanned Aerial Vehicles (UAVs) 10
As well as an increased chance of inclement Although set in a remote, constricted location,
weather, the higher latitudes also presented the methodology and work-flow adopted in
issues with the quality of the photos obtained Iceland can be applied to the majority of
from the surveys; namely light angle and geomorphological settings UAVs are likely to
quality. However, the low angle of the sun at be deployed within.
high latitudes resulted in a very flat lighting
angle combined with high levels of shading Acknowledgements
due to the confined valley environments. Christopher Hackney was supported by
Clearly this presents an issue when image award NE/JO21970/1 to the University of
quality, and specifically detail, are of Southampton from NERC. Alexander Clayton
paramount importance in a photogrammetric gratefully acknowledges PhD studentship
survey. funding from NERC. We wish to extend our
thanks to Tom Bishop and James O’Dwyer
The available options were limited as weather for their assistance in the field. The
conditions were the primary restraint on comments of two reviewers and the editor,
surveying. To that end it is highly worth Lucy Clarke, have greatly improved the
investing in a camera with good low light quality of this paper.
sensitivity for flat light situations. Ideally one
should be used that has the option to
programme a range of apertures, exposures References
and ISO settings in order to best capture the
Anderson K, Gaston KJ. 2013 Lightweight
image with the minimal amount of distortion
unmanned aerial vehicles will revolutionize
but best detail. This functionality should
spatial ecology, Frontiers in Ecology and the
enable the user to optimise the camera
Environment 11: 138-146. DOI:
settings for a range of lighting conditions and
10.1890/120150.
facilitate a much broader range of
deployment conditions with the same result in Bennett MR. 2001 The morphology, structural
image quality. evolution and significance of push moraines.
Earth-Science Reviews 53(3-4): 197-236.
DOI: 10.1016/S0012-8252(00)00039-8.
Conclusion Boulton GS. 1986 Push-moraines and
Unmanned aerial vehicles (UAVs) provide a glacier-contact fans in marine and terrestrial
low-cost, rapid deployment method of environments, Sedimentology 33(5): 677-
obtaining high-resolution aerial photography 698. DOI: 10.1111/j.1365-
over areas of varying size. Whether fixed- 3091.1986.tb01969.x
wing or rotor-wing, UAVs provide a viable Chandler BMP, Ewertowski MW, Evans DJ,
alternative to traditional surveying techniques Roberts DH, Clayton A. (In Prep) Glacial
which can deployed in a range of situations
and locations. Although strict restrictions geomorphological mapping of “annual”
apply with regards to their use and moraines on the foreland of Skalafellsjokul,
deployment in many locations, their S.E. Iceland, Journal of Maps.
application in a wide range of
geomorphological environments (glacial, d’Oleire-Oltmanns S, Marzolff I, Peter KD,
fluvial, hillslope, coastal) means UAVs are Ries JB. 2012. Unmanned aerial vehicles
becoming more and more popular in (UAVs) for monitoring soil erosion in
geomorphological research. Morocco. Remote Sensing 4: 3390-3416.
DOI: 10.3390/rs4113390.
Here, we outline some of the considerations Dunford R, Michel K, Gagnage M, Piegay H,
and regulations which must be adhered to Tremelo ML. 2009. Potential and constraints
when operating UAVs in many situations. It is of unmanned aerial vehicle technology for
vital that weather conditions are researched characterization of Mediterranean riparian
and that the operating team have scouted the forest. International Journal of Remote
study site prior to deployment. We use the Sensing 30(19): 4915-4935.
example of an aerial survey of pro-glacial
push moraines in Iceland to detail a Fonstad MA, Dietrich JT, Courville BC,
suggested best practice when operating Jensen JL, Carbonneau PE. 2013.
UAVs in challenging and remote locations. Topographfonic structure from motion: a new

British Society for Geomorphology Geomorphological Techniques, Chap. 2, Sec. 1.7 (2015)
11 Christopher Hackney and Alexander Clayton
development in photogrammetric Agriculture ’13. Wageningen Academic
measurement. Earth Surface Processes and
Publishers, Wageningen, The Netherlands.
Landforms 38: 421-430. DOI:
10.1002/esp.3366 Micheletti N, Chandler JH, Lane SN, 2015.
Section 2.2: Structure from Motion (SfM)
Grellier S, Kemp J, Janeau JJ, Florsch N,
Photogrammetry. In: Clarke LE, Nield JM.
Ward D, Barot S, Podwojewski P, Lorentz S,
(Eds.) Geomorphological Techniques (Online
Valentin C. 2012. The indirect impact of
Edition). British Society for Geomorphology;
encroaching trees on gully extension: A 64
London, UK. ISSN: 2047-0371.
year study in a sub-humid grassland of South
Africa. Catena 98: 110-119. DOI: Niethammer U, James MR, Rothmund S,
10.1016/j.catena.2012.07.002. Travelletti J, Joswig M. 2012. UAV-based
remote sensing of the Super-Sauze landslide:
Harwin S, Lucieer A. 2012. Assessing the
Evaluation and results. Engineering Geology
accuracy of georeferenced point clouds
128: 2-11. DOI:
produced via multi-view stereopsis from
10.1016/j.enggeo.2011.03.012.
unmanned aerial vehicle (UAV) imagery.
Remote Sensing 4(6): 1573-1599. DOI: Nouwakpo SK, James M, Weltz MA, Huang
10.3390/rs4061573. C, Chagas I, Lima L. 2014 Evaluation of
structure from motion for soil
Hugenholtz CH, Whitehead K, Brown OW,
microtopography measurement.
Barchyn TE, Moorman BJ, LeClair A, Riddell
Photogrammetric record 147: 297-316.
K, Hamilton T. 2013. Geomorphological
DOI: 10.1111/phor.12072.
mapping with a small unmanned aircraft
system (sUAS): Feature detection and Rosnell T, Honkavaara E., 2012. Point cloud
accuracy assessment of a generation from aerial image data acquisition
photgrammetrically-derived digital terrain by a quadcopter type micro unmanned aerial
model. Geomorphology 194: 16-24. DOI: vehicle and a digital still camera. Sensors 12:
10.1016/j.geomorph.2013.03.023. 453-480. DOI: 10.3390/s120100453.
James M, Robson S. 2014 Mitigating Saari H, Pellikka I, Pesonen L, Tuominen S,
systematic error in topographic models Heikkila J, Holmlund C, Makyen J, Ojala K,
derived from UAV and ground-based image Antila T. 2011. Unmanned aerial vehicle
networks. Earth Surface Processes and (UAV) operated spectral camera system for
Landforms 39(10): 1413-1420. forest and agriculture applications.
DOI: 10.1002/esp.3609. Proceedings SPIE 8174, Remote sensing for
Lejot J, Delacourt C, Piegay H, Fourier T, agriculture, ecosystems and hydrology XIII.
Tremelo ML, Allemand P. 2007. Very high DOI: 10.1117/12.897585.
spatial resolution imagery for channel Smith M, Chandler J, Rose J. 2009. High
bathymetry and topography from an spatial resolution data acquisition for the
unmanned mapping controlled platform. geosciences: kite aerial photography. Earth
Earth Surface Processes and Landforms 32: Surface Processes and Landforms 34: 161-
1705-1725. DOI: 10.1002/esp.1595. 255. DOI:10.1002/esp.1702.
Lin Y, Hyyppa J, Jaakkola A. 2011. Mini- Wallace L, Lucieer A, Watson C, Turner D.
UAV-Borne LIDAR for fine-scale mapping. 2012. Development of a UAV-LIDAR system
IEEE Geoscience and Remote Sensing with application to forest inventory. Remote
Letters 8(3): 426-430. Sensing 4: 1519-1543.
Mancini F, Dubbini M, Gattelli M, Stecchi F, Westoby MJ, Brasington J, Glasser NF,
Fabbri S, Gabbianelli G. 2013. Using
Hambrey MJ, Reynolds JM. 2012. ‘Structure-
unmanned aerial vehicles (UAV) for high-
resolution reconstructions of topography: the from-Motion’ photogrammetry: A low-cost,
structure from motion approach on coastal effective tool for geoscience applications.
environments. Remote Sensing 5(12): 6880- Geomorphology 179: 300-314, DOI:
6898. DOI: 10.3390/rs5126880. 10.1016/j.geomorph.2012.08.021.
Martinez-Casasnovas J, Ramos M, Balasch Whitehead K, Moorman BJ, Hugenholtz CH.
C. 2013. Precision analysis of the effect of 2013. Brief communication: Low-cost, on-
ephemeral gully erosion on vine vigour using demand aerial photogrammetry for
NDVI images. In Stafford, J. (Eds.) Precision

British Society for Geomorphology Geomorphological Techniques, Chap. 2, Sec. 1.7 (2015)
 Unmanned Aerial Vehicles (UAVs) 12
glaciological measurement. The Cryosphere
7: 1879-1884. DOI: 10.5194/tc-7-1879-2013.
Zarco-Tejada PJ, Gonzalez-Dugo V, Berni
JAJ. 2012. Flourescence, temperature and
narrow-band indices acquired from a UAV
platform for water stress detection using a
micro-hyperspectral imager and thermal
camera. Remote sensing and the
Environment 117: 322-337.

British Society for Geomorphology Geomorphological Techniques, Chap. 2, Sec. 1.7 (2015)