Ocean and Coastal Management 160 (2018) 103–116

Contents lists available at ScienceDirect

Ocean and Coastal Management

journal homepage: www.elsevier.com/locate/ocecoaman

High-resolution monitoring of beach topography and its change using T

unmanned aerial vehicle imagery
Benqing Chen∗, Yanming Yang, Hongtao Wen, Hailin Ruan, Zaiming Zhou, Kai Luo,
Fuhuang Zhong
Third Institute of Oceanography, State Oceanic Administration, Xiamen, 361005, China

A R T I C LE I N FO A B S T R A C T

Keywords: Beaches suffer from considerable degradation due to the dual influences from natural and human-induced
UAV factors. Higher resolution topographic data are frequently needed for assessment of the rapid changes and for
Remote sensing building the models that can predict the evolution of these natural environments. In this paper, a low-altitude
Topographic monitoring quad-rotor unmanned aerial vehicle (UAV), equipped with a non-metric camera, was used to acquire the two
Digital surface model
periods of UAV images in a complex beach of Wujiao Bay where the significant geomorphic changes occurred
Beach
mainly due to the strong scouring of the drainage gullies. A detailed practical technique workflow was presented
for deploying the low-altitude UAV for high-resolution monitoring the topography and the changes of the beach.
The high resolution orthomosaic with geo-reference accuracy less than 1 pixel and the Digital Surface Model
data with high vertical accuracy of approximately 10 cm were achieved by applying the UAV photogrammetry
process based on Structure from Motion method to numerous overlapped UAV images. These data were further
applied to the topographic change monitoring and analysis of the beach. This study greatly enhanced the ap-
plicability of the UAV remote sensing and enriched the existing tools for the high resolution topographic survey
and geomorphic change detection in the complex beach.

1. Introduction technology for monitoring topography and erosion in beaches.

Beaches are located in areas with strong land-sea interaction, which
are also relatively concentrated areas of human development and ac-
tivities. The natural erosion process of beaches is frequently accelerated
or augmented by anthropogenic actions, which cause significant topo-
graphic changes, even over short periods of time. Currently, approxi-
mately 50% sandy coastlines in China are suffering from the severe
coastal erosion(Luo et al., 2013). To efficiently manage and maintain
the beaches, accurate and high-resolution topographic surveys are re-
quired to monitor and quantify associated morphological changes.
These surveys are typically conducted by using a Real-Time Kinematic
or differential Global Position System (DGPS) (Brasington et al., 2000b;
Squarzoni et al., 2005), total stations (Fuller et al., 2003; Moss et al.,
1999), Terrestrial Laser Scanning (TLS) (Brasington et al., 2012; Fairley
et al., 2016; Nagihara et al., 2004) and Aerial Laser Scanning (Dewitte
et al., 2008; McKean and Roering, 2004; Notebaert et al., 2009). While
these acquisition techniques are increasingly used in geomorphology
and result in rich observational datasets, they each have their own
limitations.
GPS measurement was the traditional and most commonly used
However, this method is limited to only acquiring points and profile
data, which cannot present the complete topographic spatial pattern
and changes in features, especially in the cases of complicated topo-
graphy and severely dangerous areas such as steep and unconsolidated
slopes. The recently developed TLS can be very accurate, but often
requires the arrangement of several scanning stations across several
kilometers of narrow beaches due to its limited scanning range of
several hundred meters. Although the airborne LIDAR technology is
capable of obtaining the spatial elevation data for large-scale coastal
areas (Sallenger et al., 2003; White and Wang, 2003), it is high costly
and cannot provide data with a comparable spatial resolution and
vertical accuracy with respect to TLS and GPS.
With the availability of reliable, low-cost and lightweight un-
manned aerial vehicle (UAV) and the rapid development of computer
vision algorithms such as the structure-from-motion (SfM), the UAV is
becoming a promising technical tool. Furthermore, UAV-based photo-
grammetry is increasingly being adopted to produce high resolution
topography for the study of surface processes (Colomina and Molina,
2014), as a significant supplement to satellite and aerial remote sensing
technology. SfM allows for the generation of topography from

∗
Corresponding author.
E-mail address: chenbenqing@tio.org.cn (B. Chen).

https://doi.org/10.1016/j.ocecoaman.2018.04.007
Received 6 December 2017; Received in revised form 27 February 2018; Accepted 8 April 2018
Available online 19 April 2018
0964-5691/ © 2018 Elsevier Ltd. All rights reserved.
B. Chen et al.
randomly oriented and distributed photographs (Fonstad et al., 2013;
Smith et al., 2016; Westoby et al., 2012). Compared with other ap-
proaches, the UAV has the advantages of maneuverability, flexibility,
and capability in acquiring images with consumer-grade cameras at
very high spatial resolution (centimeter) and high temporal frequency.
Recent applications of UAV in the different dynamic environments in-
cluded vegetation monitoring (Dandois and Ellis, 2010; Jaakkola et al.,
2010; Messinger et al., 2016), Antarctic moss beds (Lucieer et al.,
2014b; Turner et al., 2014), gully erosion (Gómez-Gutiérrez et al.,
2014), glacial dynamics (Immerzeel et al., 2014), landslide (Lucieer
et al., 2014a; Niethammer et al., 2012; Turner et al., 2015), fluvial
dynamics (Javernick et al., 2014; Miřijovský and Langhammer, 2015),
and river bank and gorge erosion(Prosdocimi et al., 2015) and coastal
topographic survey and geomorphic change (Clapuyt et al., 2016;
Delacourt et al., 2009; Gonçalves and Henriques, 2015; Long et al.,
2016; Mancini et al., 2013; Turner et al., 2016).
Particularly in the beach, exploratory researches have shown that
the UAVs are very suitable for the high resolution topographic survey
and geomorphic change detection by achieving the orthomosaic and
digital surface model (DSM) data. Delacourt et al. (2009) developed a
set of UAV photogrammetry helicopters equipped with automatic na-
vigation systems to obtain the high-resolution orthomosaic of Pors-
millin Beach (French Brittany). Mancini et al. (2013) demonstrated that
the accuracy of the elevation data from UAVs was comparable with that
obtained by TLS technology in the Ravenna sand dune (Italy). Based on
the UAV photogrammetry and accuracy analysis in two beaches,
Gonçalves and Henriques (2015) believed that UAVs can replace the
conventional flights within the regular coastal monitoring program in
the regions, with considerable gains in the cost of the data acquisition
and without any loss in the quality of topographic and aerial imagery
data. Clapuyt et al. (2016) also assessed the reproducibility of the UAV
topographic measurement based on the SfM method, and found the
UAV can be used for the monitoring of the multi-temporal topo-
graphical changes of beaches if the very high-resolution (centimeter)
DGPS are used for field measurements.Long et al. (2016) used UAV
images to monitor the topography of a dynamic tidal inlet. Turner et al.
(2016)deployed UAV to monitor and assess the post-storm coastal
erosion spanning a full coastal embayment. However, in order to ex-
tensively assess the accuracy, repeatability and usability of UAVs for
monitoring the topography of the beaches, the various cases under
different geographical environments are needed.
This study aims at evaluating the applicability of the low-altitude
UAV remote sensing for the high resolution topographic survey and
geomorphic change detection in the complex beach. The two dates of
Oct. 2013 and July 2014 were chosen due to the significant geomorphic
changes that occurred as a result of scouring by the gullies in the coast
beach during the period. A practical workflow of the UAV photo-
grammetry method based on SfM was presented to generate a high
resolution orthomosaic and extract highly accurate DSM data of the two
periods. These data then were used to analyze the geomorphic changes
over the nearly one-year period.
2. Materials and methods
2.1. Study area
The beach of the Wujiao Bay (Fig. 1), located in the Dongshan Is-
land, the sixth largest island in China, is concerned in this study. The
south-central part of the beach coast (black rectangle region in Fig. 1),
an area with 1.5 km length and approximately 200 m average width at
low tide, was selected for the UAV monitoring due to its typical geo-
morphic features. The area was typically characterized by the drainage
gullies which were caused by the waste discharge from the aquacultures
near the coast, and numerous drainpipes across the beach for allowing
the passage of sea water into the aquacultures near the coast. Some of
the gullies extend for several hundred meters across the beach, with
104
Ocean and Coastal Management 160 (2018) 103–116
depths of up to 1 m and widths of 3 m (Fig. 2a), which result in more
serious erosions on the beaches. The main hydrodynamic forces for
Wujiao Bay are generally tide and seawave. The tidal type is irregular
semidiurnal tide, with the mean tidal range of about 2.3 m and the
average flow rate of approximae 1.0 m/s. The seawave nearshore is
dominated by the wind-driven wave. In the past years, the beach geo-
morphic features changed quickly as a consequence of several factors
including the drainage gullies, strong waves and tide. The beach
coastline was also severly damaged (Fig. 2b), which had posed a great
threat to the power windmills close to the coast.
2.2. UAV remote sensing system
The MD4-1000 quad-rotor UAV system made by the Microdrones
Company was used to obtain the UAV images of the beach. The quad-
rotor UAV (Fig. 3) has the advantages of not requiring a runway, low
demand of taking off and the landing site, and is suitable for acquiring
the UAV images in harsh environments, such as the coasts and the sea-
islands. The MD4-1000 system is mainly composed of an unmanned
aircraft, remote controller, ground control station and non-metric
camera by integrating with many advanced sensors, such as DGPS, in-
ertial navigator, barometer and magnetometer. The UAV can imple-
ment the functions of spot hover and GPS position hold in addition to
allowing flight by an autopilot along the flight lines determined in the
flight plan. The max cruising speed is about 12 m/s. Power is supplied
by a Lithium-Polymer battery that is specified for lasting for approxi-
mately 88 min. In actual use, the flight duration was controlled for less
than 30 min for security with recommended load of 800 g, especially
with moderate winds, low temperatures or low air pressure con-
ditions.To monitor larger area , more than one flight would be required
with demanding a battery replacement. The MD4-1000 can operate in a
wind speed less than 12 m/s, but less than 6 m/s in order to stabilize the
flight enough to obtain the high-qualified images with no coverage gap.
The ground control station is used to monitor most of the flight para-
meters and real-time photos by the radio linkage. The main specifica-
tions of the quad-rotor UAV system are shown in Table 1.
2.3. Acquisition of UAV images
Prior to the UAV flight operation, a preliminary field investigation is
needed to obtain more knowledge about the beach topographic fea-
tures, distribution of important facilities as well as the existence of any
interference to the GPS and magnetometer of UAV. A handheld GPS was
also used to determine the take-off and landing position, approximate
location of the coastline and boundary of the monitoring area, in order
to facilitate the mission planning. A specific waypoint planning package
was adopted to determine the flight route and several main flight
parameters such as flight height, flight speed, and image overlaps
(Table 2). The flight is performed in the direction of the smaller sensor
side of the UAV camera.
To study the spatial variability of the topography in the beach in
Wujiao Bay, two campaigns were executed to acquire the UAV images
in Oct., 2013 and July 2014. As the beach was relatively narrow with a
width of about 200 m, a single flight is satisfactory to cover the whole
monitoring area for both periods. To ensure the overlap of UAV images
and enough coverage for the first time, one more flying route within a
flight was conservatively designed in Oct., 2013, compared to the two
flying routes in July 2014.
Fig. 4 shows the representation of the overlaps in the two UAV
campaigns, together with the camera locations. Most of the areas are
imaged in four or more photos. This multi-ray intersection may result in
an improvement in accuracy for the extraction of point clouds. The
central part of the monitoring area has more overlaps because it was
covered in both flying routes.
The day with low wind and tide was chosen for UAV flight in order
to acquire the higher quality images in the largest possible area. Before
B. Chen et al.
Fig. 1. Unmanned aerial vehicle (UAV) monitoring area (black rectangle region) in Wujiao Bay.
Fig. 2. Field pictures demonstrate the degradations in the beach of Wujiao Bay. (a) beach erosion by the drainage gullies; and (b) seriously destroyed coastline.
the data acquisition procedures, the advanced placement of artificial
signboards and the assemblage of the UAV system should be conducted.
Once all of the signboards were placed, the flight was conducted and
the DGPS were surveyed. In Oct. 2013 and July 2014, a total of 435
high-overlapping UAV images were acquired in the beach of Wujiao
Bay. The typical coastal features, such as the coastline and drainage
gullies, were clearly represented in the acquired UAV single image with
a high resolution of approximately 5 cm in the two periods (Fig. 5).
2.4. Ground control point and check point measurement
The ground control points (GCPs) are required to conduct the image
orientation. In the beach of Wujiao Bay, the surface were mostly cov-
ered by sand and sparse vegetation, that do not provided well
105
Ocean and Coastal Management 160 (2018) 103–116
recognizable points to be used as GCPs. Therefore it was decided to put
artificial points in the monitoring area. For each period, a total of about
10 GCPs were arranged spatially along the beach (Fig. 6) with a uniform
distribution by using 50 × 50 cm artificial signboards. The Trimble
ProXRT GPS receiver was used to perform the high-accuracy real-time
DGPS measurement based on the local Continuously Operating Re-
ference Station (CORS) with the predicted accuracy of 3 cm in plani-
metry and 5 cm in height. Meanwhile, the field pictures of the sign-
boards and their surrounding environment were taken to aid the
placement of the corresponding GCPs in the UAV images during the
image orientation process.
To assess the planimetric accuracy of the orthomosaic and the
vertical accuracy of the final DSM data of the two periods, the common
six artificial structures such as the embankment were selected for the
B. Chen et al. Ocean and Coastal Management 160 (2018) 103–116

Fig. 3. Picture of MD4-1000 quad-rotor UAV used in this study.

Table 1
Main specifications of the quad-rotor UAV system.
Weight of 2650 g Recommended load 800 g
unmanned
aircraft
Dimensions 1030 mm (from rotor Max. Flying time 88 min
hub to rotor hub)
Max. cruising speed 12.0 m/s Max. climb rate 7.5 m/s
Max. wind tolerance < 12 m/s Take-off altitude 4000 m
Temperature −10–50 °C Humidity Max. 90%
r.H.

Table 2
Main parameters for the UAV flight operation in the beach of Wujiao Bay.
Flying height 150 m Cruise speed 4 m/s
Flight control Waypoints Shooting 2 - 4s
navigation interval
Take-off mode Vertical take-off Landing mode Vertical landing
Camera Pentax Option Camera Pixels 12 M pixels
A40 (4000 × 3000)
Camera pixel size: 1.88 um Camera focal 7.9 mm
lenght
Minimum heading 60% Minimum side 50%
overlap overlap

orthomosaic as independent check points (ICPs) and four totally

random elevation profiles of ICPs for the DSM were chosen (Fig. 6).
These ICPs were also surveyed by the DGPS receiver at an interval of
2–3 m. The profiles were usually chosen in the areas where the drainage
gullies lied or the beach topography obviously changed. In October
2013, two random profiles, a total of 73 ICPs, lied at the two ends of the
monitoring area were surveyed, while a total of 110 ICPs of the two Fig. 4. Camera central locations (black points) and image overlaps in (a) Oct.,
profiles lied at the south and middle of the monitoring area were sur- 2013 and (b) July 2014.
veyed in July 2014. During the surveys of GCPs and ICPs, a thin iron
flat was always carefully placed on the beach surface to prevent the GPS 2.5. UAV photogrammetric process based on SfM
pole inserting into the sand, which would lead to a measurement error
in height of several centimeters. To acquire coastal topography from 2-D UAV images, the UAV
photogrammetric process based on SfM algorithm is required for 3-D
coastal topography reconstruction. The SfM algorithm allows re-
constructing a 3D scene geometry by solving the geometry of the scene,

106
B. Chen et al.
Fig. 5. The acquired single UAV image for the typical topographic features in the beach of Wujiao Bay in the two times of October 2013 and July 2014.
camera positions and orientation simultaneously using a highly re-
dundant, iterative bundle adjustment procedure, based on a database of
features that are automatically extracted from a set of multiple over-
lapping images (Snavely, 2008). Recently, several photogrammetric
software packages based on the SfM method that bring knowledge from
computer vision and traditional photogrammetry together, have been
developed and are more dedicated to the processing of UAV images.
One commercial package, Agisoft Photoscan, was chosen for this study
due to its essentially automatic process based on multi-view 3D re-
construction technology and being suitable for UAV image processing.
The general workflow for the UAV images processing is divided into
different steps and described in Fig. 7.
In the first stage, a dataset of distinctive points were extracted in
each UAV image which are stable under viewpoint and lighting varia-
tions and generates descriptors for each points based on its local
neighborhood. These descriptors are used later to detect corre-
spondences across the images to form a large number of conjugate
points. This is implemented by an algorithm that is similar to the well-
known SIFT approach, but with a slight improvement in the alignment
quality (Semyonov, 2011). In order to speed up this process, the image
pair pre-selection option is needed. The camera locations will be used if
they are presented, otherwise the overlapping pairs of photos are se-
lected by matching photos using lower accuracy setting firstly. The
focal distance and the sensor size of each image are read automatically
from the EXIF header of the JPEG images, which are needed to facilitate
the search for conjugate points.
Based on the conjugate points, a rough 3D model of the scene with
no information about the spatial scale is produced when there was no
independent position information for UAV images. After this initial
scene construction, the GCPs can be identified and marked manually on
at least two overlapped images. With the setting of the coordinate and
projection parameters as well as use of the Brown camera distortion
107
Ocean and Coastal Management 160 (2018) 103–116
model, a greedy algorithm and block bundle adjustment method are
used to achieve the accurate interior and exterior orientation para-
meters by minimal optimization of projection error between the mea-
sured ground control point and image point. The higher accuracy set-
ting helps obtain more accurate estimates of the orientation parameters.
After this process, a sparse point cloud appropriate with the GCP ac-
curacy will be generated within the pre-defined geometry projection.
Fig. 8a shows the sparse point cloud data of part of the beach of Wujiao
Bay Beach in July 2014.
By rotating the view of point cloud data, some obvious outliers and
points with large projection error are checked and deleted manually,
and then the optimization is repeated. Generally the accurate image
orientation will be achieved by 2–3 iteration of optimizations. Once the
accuracy of the UAV image orientation is satisfactory, a dense point
cloud reconstruction processing algorithm is applied to generate the
dense point cloud with RGB colors, which is used to generate a full DSM
in the form of a triangular mesh after deleting the points with obvious
error. Fig. 8b–d respectively present the extracted DSM of the dense
point cloud data, triangular mesh, and mesh filling of part of Wujiao
Bay Beach in July 2014.
The DSM is later used to build the orthomosaic of the entire mon-
itoring area, and both datasets can be exported in the Geotiff format in
32 bits and 8 bits. The coordinate system engine is based on the PROJ4
open source library, which allows for the choice of map projections
based on EPSG codes. Data export takes into account the changes in-
troduced in the exterior orientation by the parameters of the map
projection.
2.6. Topographic change mapping method
Basically, the topographic change was analyzed simply based on the
subtraction of the multi-temporal topography data. However, the
B. Chen et al. Ocean and Coastal Management 160 (2018) 103–116

uncertainty generally exist accounting for the instrument imprecision

and the quality of the individual topographic data used to determine
the topographic changes. Several authors introduced a level of detec-
tion (LoD) (Brasington et al., 2000a; Lane et al., 2003; Neugirg et al.,
2016; Wheaton et al., 2010) to separate measurement errors from the
real geomorphic changes. For the multi-temporal analysis, the com-
bined errors of both Digital Elevation Models (DEMs) for the DEM of
difference (DoD) can be calculated using the simple Gaussian error
propagation (Lane et al., 2003). Firstly, the resulting standard deviation
of the differences (δdiff ) of the two resulting DEM data can be attributed
to the individual standard deviation of the DEM elevation error.

δdiff = δ12 + δ22

Where δ1represents the DEM error of time period 1 and δ2 is the DEM
error of time period 2. Following the assumption that the errors are
statistically independent, the change in z-value for each grid cell can be
converted into a t-statistics:
z1 − z2
t=
δdiff

The t-values can be compared to a critical t-value (tcrit ) from the

student’s t-distribution. For this study we used a 95% significance level,
which corresponds to a tcrit value of 1.96. We assumed that all |t| > 1.96
represented statistically significant surface changes. All other grid cells
were treated as cells with no surface change. The above mentioned LoD
value can be calculated according to Lane et al. (Lane et al., 2003).
LoD = tcrit ∗δdiff

In this study, the ICPs were surveyed to qualify the UAV DSM data
of two periods respectively, therefore the standard deviation calculated
from the elevation errors between the DSM and the ICPs were used as
the proxy for the errors of the DSMs. The standard deviations of the two
periods were 0.11 m and 0.07 m respectively, which led to the LoD
value of 0.26 m.

3. Data processing and results

Fig. 6. Distribution of the ground control points (GCPs) and independent check
points (ICPs) for the orthomosaic and DSM in the monitoring area. 3.1. DSM extraction and orthomosaic generation

For beach topographic monitoring, the DSM data is very important

for the topographic assessment (profiles, volumes, etc.) and the change

Fig. 7. General workflow for the UAV images processing, starting from the preparation of UAV campaign to the determination of DSM and orthomosaic accuracy by
GPS survey.

108
B. Chen et al.
Fig. 8. Example of an extracted digital surface model (DSM) of part of Wujiao Bay Beach: (a) sparse point cloud with average point density of 0.9 pts/m2 ; (b) dense
point cloud with RGB color with average point density of 155.1 pts/m2; (c) triangular mesh and (d) the mesh filling created from dense point cloud. (For inter-
pretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
analysis. The orthomosaic is also important for interpreting the image
content and complementing the three-dimensional perception given by
the DSM data. Some 2D indicators of migration, such as coastline and
vegetation lines, can also be delineated from the image.
In this study, the two periods of DSM data and orthomosaic images
were achieved by using the UAV photogrammetric method based on
SfM. Before the DSM extraction, certain point clouds need to be care-
fully deleted if they are less accurate due to failure to match feature
points in the areas of the drainage gullies covered by water. Following
this, the TIN-interpolation method is adopted to interpolate these pat-
ches using the data from neighboring point clouds. The final DSM data
were projected into the UTM zone 50 map projection in the WGS-84
geographic coordinate system and exported by being re-sampled with
grid size of 50 cm. Fig. 9 showed the DSM data and an enlarged sample
of the beach of Wujiao Bay in October 2013 and July 2014. Most of the
topographic characteristics of the beach such as the shoreline slope,
bank collapse and erosions by the drainage gullies, were clearly and
accurately represented in the two periods of the DSM data.
Based on the DSM data, the orthomosaic were generated and ex-
ported by being re-sampled with a pixel size of 10 cm using the same
projection and coordinate system with the DSM data. Fig. 10 showed
the orthomosaic and an enlarged sample of the beach of Wujiao Bay in
the two periods. A qualitative assessment of the orthomosaic revealed
that cut lines were not detectable and image color balancing was also
good, which implied that the final orthomosaic was satisfactory. Based
on the orthomosaic, some important features such as the shoreline
destruction, drainpipes on the beach, drainage gullies and small sand
wave were clearly distinguished.
The planimetric geo-referenced accuracy of the two periods of or-
thomosaic was evaluated based on the six orthomosaic ICPs, which
showed that the planimetric RMS errors were respectively 7.8 cm and
7.7 cm. Besides, for monitoring the beach topographic changes, the co-
registration of the two periods of images is more important, because the
109
Ocean and Coastal Management 160 (2018) 103–116
mis-registration would result in significant errors in the comparisons of
the two periods of topographic elevation. The co-registration error
between the two orthomosaic was assessed by selecting the 59 common
object features from the two orhtomosaic with a uniform distribution.
The RMS error of co-registration was 6.3 cm, less than the pixel size of
10 cm, which showed the co-registration was good. As for the DSM
vertical accuracy assessment, it is detailed in the following section.
3.2. DSM vertical accuracy assessment
The DSM vertical accuracy assessment is the most important pro-
cedure for the monitoring of the beach topography and its changes. The
ICPs surveyed at the time of UAV images acquisition were used for that
purpose. For these points, the heights calculated from the DSM by linear
interpolation were compared to the corresponding measured heights in
the two periods. The statistics of these height differences were calcu-
lated (Table. 3). The RMS errors were both 11.0 cm for the two periods
of DSM data with the average errors of 2.0 cm and 7.0 cm respectively.
Fig. 11 showed the scatter diagram and error histogram between the
DSM elevation and the measured elevation for the two periods. It was
shown that the two elevation data had a good agreement, and most of
the errors were less than 10 cm. These results indicated that the accu-
racy of the two period of DSM from UAV is good in the beach of Wujiao
Bay, and met the requirements for the high-resolution topographic
monitoring, as well as the analysis of its changes.
Although the UAV DSM vertical accuracy was generally satisfactory,
there were some obvious discrepancies with some errors of 10–20 cm
and few errors that were larger than 30 cm. Some comments may be
made for these discrepancies. First, the spatial scale for measurements
is different between the two compared elevation data. The UAV DSM
elevation represents the spatial averaged value of 50 cm grid size, and it
was linearly interpolated when compared to the ICPs, which would
introduce large errors in the area where the beach topography changes
B. Chen et al. Ocean and Coastal Management 160 (2018) 103–116

Fig. 9. Two periods of the DSM data in the beach of Wujiao Bay and the corresponding enlarged data of the black rectangle.

sharply (Fig. 12). Some larger errors had been found near the coastline
and close to the drainage gullies (see gray ellipse area in Fig. 16).
Secondly, all the GCPs and ICPs were surveyed using the DGPS with a
vertical accuracy of about 5 cm. This error may be transmitted into the
UAV DSM and contribute to the error estimation. The last comment
would refer to the characteristics of the terrain, which is essentially
sand. The measurement of the ICPs was performed with the all-possible
care using a flat base under the GPS pole, but inevitably errors can be
introduced due to movement of the sand.
3.3. Topographic changes of the beach
The topographic changes in the beach of Wujiao Bay were mon-
itored based on the two periods of DSM data and the corresponding
orthomosaic. The two periods of orthomosaic were used for the changes
by quantitatively interpreting and analyzing the coastline and the beach
morphology. The coastline was continuing to retreat due to being da-
maged by the sea waves and anthropological activities. There was a
serious collapse and obvious retreat in part of the artificial coast banks,
which directly threatened the windmills close to the coastline.
Moreover, as seen from the two periods of the orthomosaic, the beach
morphology changed rapidly mainly due to the direct discharge of
waste water from the aquaculture farms in the coast. In the monitoring
area, the number of the drainage gullies changed from 4 to 6, with an
increase of nearly double, from an area of 13,113.0 m2 to 22,849.0 m2
between Oct., 2013 and July 2014.
By applying the DoD method under the ArcGIS software, a DoD map
was produced for the beach of Wujiao Bay. The areas affected by the
ocean waves were eliminated near the waterline. As shown in Fig. 13,
the positive value of the DoD map indicated an increase in beach
elevation, while the negative value represented the decrease from Oct.,
2013 to July 2014.
The topographic change of the beach in the Wujiao Bay showed
spatial complexity, which made it difficult to comprehensively detail
the changing pattern just only by traditional GPS measurements. The
beach generally maintained an accretive status except in some local
areas with vertical erosion. The magnitude of the accretionary was
extensively less than 0.5 m. In the northern part of the beach, the ac-
cretion could reach 0.5–1.0 m and even 1.0–1.5 m in some local pat-
ches. The vertical erosion of the beach mainly occurred in the drainage
gullies and areas at both sides with a magnitude of most erosion areas
being less than 0.5 m. In some areas of the drainage gullies, erosion had
a magnitude of 0.5–1.0 m, with a maximum erosion of 2.0 m in the
middle part of the beach where a new drainage gully was formed in
July 2014. Additionally, the topographic changes could also be verified
by the comparison of the two periods of the orthomosaic. In some ac-
cretion areas, the drainpipes on the beach in Oct., 2013 were partly
covered by the sand and was invisible in the orthomosaic of July 2014
(Fig. 14). In contrast, due to wave erosion, the originally covered
drainpipes in Oct., 2013 were partly exposed in July 2014 (Fig. 15).
Based on the result of the DoD method, a volumetric analysis for the
whole monitoring area, was also performed by computing erosion and
deposition as a simple product of DoD grid-cell values and areas. The
volumetric change due to erosion was 22,719.1 m3, which was mostly
from the newly occurred gullies and their surrounding areas. However,
the total accretion was 37,499.3 m3, which means the net sediment
exchange for this area was about 14,780.2 m3 of accretion over a period
of nearly nine-month in this monitoring area. Furthermore, the in-
tensity of erosion and accretion was defined as the elevation change per
square meter, which was calculated as the ratio of the total volumetric

110
B. Chen et al. Ocean and Coastal Management 160 (2018) 103–116

Fig. 10. Two periods of UAV orthomosaic in the beach of Wujiao Bay and the corresponding enlarged image of the black rectangle.

Table 3 features and changes in detail, especially for a complicated beach such
Descriptive statistics of the elevation errors of the two periods of DSM assessed as the Wujiao Bay. In contrast, UAV acquires the whole image for the
by the ICPs (all values in m) beaches and topographic elevation data in terms of overlooking, which
Oct., 2013 July 2014 overcomes the limitation of conventional GPS measurements.
Furthermore, it also overcomes the deficiency of the TLS as it is unable
Number of PCPs 73 110 to acquire effective data in the topographical sink and shielded areas
Average 0.02 0.09
due to the adoption of horizontal optical scanning. The low-altitude
Std. Dev. 0.11 0.07
RMS 0.11 0.11 UAV significantly improves the working efficiency of monitoring the
Min. −0.44 −0.19 beach with a lower cost and becoms an alternative tool for the topo-
Max. 0.24 0.34 graphic survey of the beach. The feature points in the beach usually are
smaller and undetectable easily compared with the other surface fea-
tures. However, benefiting from the robustness of the SfM algorithm
change to the corresponding areas. The intensity of the erosion was and sufficient high resolution (centimeter level) of acquired UAV
0.63 m, which was larger than the intensity of the accretion (0.49 m). images, a large number of subtle objects and tonal change character-
istics of the beach are usually reflected on the UAV images. Thus these
4. Discussion objects and characteristics can be more easily detected and then mat-
ched as feature points, which guarantees the accuracy of image or-
4.1. Relevance of UAV application and accuracy assessment ientation. This is an important basis of applying the UAV to monitor the
beach topography and its changes.
In the beach of Wujiao Bay, the geomorphic features are spatially In addition to the accuracy of image orientation in the SfM algo-
variable and have drastic temporal changes in the form of erosion and rithm, the spatial distribution, number and measurement accuracy of
accretion. The erosion is mainly caused by the combined action of the the GCPs are important factors in determining the vertical accuracy of
drainage scouring and the strong ocean waves, while the accretion is the UAV DSM in the beach. Due to the difficulty of moving on the sand
the result of the transport of the terrigenous sediment from the coast. and obstacles of the drainage gullies, it will take more than half of the
Generally, the highly accurate DGPS measurement is the first choice for entire UAV monitoring time to arrange the artificial signboards and
surveying the coastal topography and temporal changes. However, the perform the GPS survey. In particular, the measurement of the GCPs
DGPS survey only acquires the topographic points or profile data of the was impossible in the inaccessible or hazardous areas. In this study,
beach, which is really not sufficient to characterize the topographic approximately 10 GCPs were evenly distributed and measured in the

111
B. Chen et al.
Fig. 11. (a, b) Scatter diagram and (c, d)errors histogram for the DSM elevation and DGPS measured elevation for the two periods of time.
Fig. 12. Large errors induced by the sharp changes in topography.
beach for each UAV campaign, although we knew that a higher accu-
racy for the UAV DSM would be achieved with a higher number of GCPs
(Agüera-Vega et al., 2017; Harwin and Lucieer, 2012). Furthermore,
due to concern regarding the tide wave washing the GCPs signboards
away and undetectable by UAV images during the flight, the GCPs were
generally located further than 20 m from the waterline. This would
result in a loss in vertical accuracy in the areas outside the control of
GCPs due to reduction in support of control points during the process of
UAV DSM extraction. To make it clear, the northern part of the ICPs in
2013 and the middle U-shape of the ICPs in 2014 (Fig. 8) were chosen
respectively for the transect of the elevation data from the ICPs as well
as the elevation difference between ICPs and UAV DSM data. As shown
in Fig. 16, the elevation differences of the points beyond the GCP
control (gray rectangle area) of the two periods, with the RMS errors of
0.12 m and 013 m, were obviously larger than that under the GCP
control having errors of 0.08 m and 0.11 m. There are exceptions in
several points where there are sharp change in topography (gray ellipse
areas). Therefore, to solve this limitation, the direct geo-referenced
UAV mapping is a promising technique for the topographic monitoring
112
Ocean and Coastal Management 160 (2018) 103–116
of the beach. By integrating the high-precision DGPS into the UAV
system, it is possible to directly acquire the images with accurate 3D
coordinate information and to further achieve the geo-referenced or-
thomosaic, point cloud and DSM elevation data without using the GCPs.
4.2. Preliminary interpretations of the geomorphic changes
The coast of Wujiao Bay belongs to the Cape-bay coast type, where
most of the sediment comes from the sand removed from the coastal
erosion. Furthermore, the coast of Wujiao Bay may be directly damaged
by the strong ocean waves and tides from the Taiwan Strait as there is
no buffer from the barriers outside.
According to Zhou et al. (2016), the hydrodynamic process of Wu-
jiao Bay was dominated by the ocean waves, followed by the tide
during the period of UAV survey. Induced by the leading seaward wind
with an average speed of 8.0 m/s from the north-east and east-north-
eastern, the wind-driven waves were generally dominant. In this case,
the coast of Wujiao Bay is seriously eroded by the strong waves and the
coast damaged severly (Fig. 2b), which result in a larger quantity of
sediment deposition near the coastline. This could be confirmed by
comparing three datasets of DGPS elevation data (Fig. 17) that were
surveyed repeatedly in one of the ICPs profiles (location in Fig. 13) in
the periods of the two UAV campaigns and an additional date of Feb.,
2014. The elevation dramatically increased during the period from Oct.,
2013 (autumn) to Feb., 2014 (winter), which could be attributed to the
deposition of the sediments from the damaged coast by the strong
waves during autumn and winter. However, the elevation decreased in
July 2014 (summer) because a new balance of the sedimentation pro-
cess was obtained during the period of the low-wave interactions (such
as in the spring). These large quantities of sediments by the coastal
erosion, combined with the debris coming from the aquacultures near
B. Chen et al. Ocean and Coastal Management 160 (2018) 103–116

Fig. 13. DoD map in the beach of Wujiao Bay from October 2013 to July 2014 (overlaid on the orthomosaic of 2014) and the enlarged image of the black rectangle
for the accretion and erosion.

the coast, would be transported into the ocean or re-deposited by the
wave and tide during the ebb tide. However, the kinetic energy of
waves decreased more in the areas where numerous drainage pipelines
distributed due to the increase of frictional forces between waves and
beach surfaces,which resulted in the more deposition and accretion in
the in these areas. Addtionally, the slope of the near-shore bathymetry
in the northern part of the monitoring area was larger than that in the
southern part (see Fig. 10 in the lecture of Liu et al. (2016)). The kinetic
energy of the waves dissipated more quickly in the northern part during
the moving shoreward due to the steeper bottom and a greater

Fig. 14. Comparison of the orthomosaic in the accretion area (rectangle area in Fig. 13) from Oct., 2013 to July 2014.

113
B. Chen et al.
Fig. 15. Comparison of the orthomosaic in the erosion area (rectangle area in Fig. 13) from Oct., 2013 to July 2014.
Fig. 16. Transect of the elevation data from the ICPs and the elevation difference between ICPs and UAV DSM data in (a) 2013 and (b) 2014.
Fig. 17. Comparison of the DGPS elevation measurements of a ICPs profile
repeatedly three times.
frictional effect. Therefore, the kinetic energy of the wave to the
northern coast of the monitoring area was less than that to the southern
coast, which resulted in a larger extent of sediment deposition in the
northern coast. With the interactions of the factors but not limited to
those described above, the beach had specific spatial features with areas
114
Ocean and Coastal Management 160 (2018) 103–116
of the accretion being more frequently located in the northern part of
the monitoring area and the area around the drainage pipelines.
In contrast to the accretion, the vertical erosion of beach was mainly
caused by the gullies due to the continuous scouring of the drainage
from the aquacultures near the coast. The large quantity of sediments
on the beach was removed by the discharged water, which would result
in a steep slope at the edge of the gullies. Under gravity, these sedi-
ments in the edge would collapse and removed again. Repeating these
processes, the gullies would become wider and deeper until a new
balance was achieved. With the extensive development of aquacultures,
the area of the aquacultures in the coast of Wujiao Bay significantly
increased from 7.8 × 105 m2 to 2.1 × 106 m2 during the period from
2003 to 2013. There was an average increase of 1.32 × 105 m2/a ac-
cording to the interpretation from the satellites, which would increase
the amount of wastewater discharge. In this situation, the range and the
intensity of erosion by the gullies would inevitably increase. As shown
in Fig. 18, the maximum vertical erosion along the gully (see the lo-
cation of the gully in Fig. 13) had reached about 2.1 m in the UAV
monitoring area from Oct., 2013 to July 2014.
In this study, only two periods of UAV imagery were used to analyze
the topographic changes and the preliminary interpretation. However,
the hydrodynamics resulting in the geomorphic changes for the beach
near the shore is very complicated. More temporal UAV campaigns are
essential in the future for improving our understanding of the geo-
morphic process in the beach of Wujiao Bay.
B. Chen et al.
Fig. 18. Transect of the vertical erosion along the gully.
5. Conclusions
High resolution topographic data and quantitative erosion assess-
ment, further the updating knowledge of beach evolution and of the
factors influencing coastal environment, are most important and ne-
cessary in the development of coastal management policies. As a newly
developed technique, UAVs have extended and enriched the measure-
ment tools for monitoring the coastal environment. Compared to the
points or profile topographic data by the traditional procedure using
DGPS measurement, the UAVs can achieve not only the high-resolution
(centimeter levevl) orthoimages but also the high-resolution topo-
graphy data, which is especially needed in some complicated and hard-
to-reach coastal area such as Wujiao Bay. The multi-temporal UAV
orthoimages and topographic data also present more detailed in-
formation about the dynamic changes of the coastal environment, such
as the coastline condition, vegetation distribution and gullies partern,
as well as the accretion and erosion. All these are critical information
needed for the decision-making of the coastal management, such as for
coastal change analysis, wetland mapping, ecosystem monitoring, nat-
ural disaster prevention, and damage assessment.
This study presented the practical technique workflow and de-
monstrated the applicability of the UAVs for high-resolution monitoring
the topography and the changes in the complicated beach. The high
resolution orthomosaic with the geo-reference accuracy of less than 1
pixel and the DSM data with a high vertical accuracy of 10 cm were
achieved in the beach of Wujiao Bay during two periods by applying the
SfM method to numerous overlapped UAV images. These data could be
further applied for high-frequency monitoring of the topographic
changes. Based on the high-accurate products, a DoD map was com-
puted, which revealed that the topographic changes of the beach in the
Wujiao Bay were spatially complicated and generally maintained an
accretive status except in some local areas with vertical erosion due to
the drainage gullies. The net sediment exchange in the monitoring area
was about 14,780.2 m3 of accretion during the period of the two UAV
campaigns, while the intensity of the erosion was larger than that of the
accretion. The accretion might be attributed to the interactions among
the sediments from the destructed coast, the waves and tide, the beach
surface features, as well as near-shore bathymetry. In contrast to the
accretion, the vertical erosion of beach was mainly caused by the gullies
due to the continuous scouring of the drainage from the aquacultures
near the coast.
Although UAV is increasingly being applied to the coastal en-
vironment, there remain some limitations to be solved and challenges
facing in the future. Firstly, to solve the loss in vertical accuracy in-
duced by the outside control and the limited numbers of the GCPs, the
direct geo-referenced UAV mapping technique is needed for the
115
Ocean and Coastal Management 160 (2018) 103–116
topographic monitoring. Secondly, steming from the strict safety re-
quirements for manned aircraft, the operation of UAVs is generally
strictly to use by regulating the maximum weight, flight height and
speed, which hinder the UAV wider application. In addition, there
needs a breaking development of electric battery for longer enduration
of fight for motor-driven UAVs to cover lager monitoring area. With the
solution of the above challenges, the UAVa will be the most extensively
used technique for the coastal management.
Conflicts of interest
The authors declare no conflicts of interest.
Acknowledgments
This work was supported by the Science Foundation of Two Sides of
the Strait (Grant No. U1405234), the Public Science and Technology
Research Funds Projects of Ocean (Grant No.20100504), the Scientific
Research Foundation of Third Institute of Oceanography, SOA
(2009053) and the Public service platform project of Southern Marine
Research Center, Xiamen (14PST63NF27).
References
Agüera-Vega, F., Carvajal-Ramírez, F., Martínez-Carricondo, P., 2017. Assessment of
photogrammetric mapping accuracy based on variation ground control points
number using unmanned aerial vehicle. Measurement 98, 221–227.
Brasington, J., Rumsby, B.T., McVey, R.A., 2000a. Monitoring and modelling morpho-
logical change in a braided gravel-bed river using high resolution GPS-based survey.
Earth Surface Processes and Landforms 25, 973–990.
Brasington, J., Rumsby, B.T., McVey, R.A., 2000b. Monitoring and modelling morpho-
logical change in a braided gravel-bed river using high resolution GPS-based survey.
Earth Surface Processes and Landforms 25, 973–990.
Brasington, J., Vericat, D., Rychkov, I., 2012. Modeling river bed morphology, roughness,
and surface sedimentology using high resolution terrestrial laser scanning. Water
Resources Research 48.
Clapuyt, F., Vanacker, V., Van Oost, K., 2016. Reproducibility of UAV-based earth to-
pography reconstructions based on Structure-from-Motion algorithms.
Geomorphology 260, 4–15.
Colomina, I., Molina, P., 2014. Unmanned aerial systems for photogrammetry and remote
sensing: a review. ISPRS Journal of Photogrammetry and Remote Sensing 92, 79–97.
Dandois, J.P., Ellis, E.C., 2010. Remote sensing of vegetation structure using computer
vision. Remote Sensing 2, 1157.
Delacourt, C., Allemand, P., Jaud, M., Grandjean, P., Deschamps, A., Ammann, J., Cuq, V.,
Suanez, S., 2009. DRELIO -an Unmanned Helicopter for Imaging Coastal Areas,
Proceedings of the 10th International Coastal Symposium. Portugal, Lisbon.
Dewitte, O., Jasselette, J.C., Cornet, Y., Van Den Eeckhaut, M., Collignon, A., Poesen, J.,
Demoulin, A., 2008. Tracking landslide displacements by multi-temporal DTMs: a
combined aerial stereophotogrammetric and LIDAR approach in western Belgium.
Engineering Geology 99, 11–22.
Fairley, I., Thomas, T., Phillips, M., Reeve, D., 2016. Terrestrial laser scanner techniques
for enhancement in understanding of coastal environments. In: Finkl, C.W.,
Makowski, C. (Eds.), Seafloor Mapping along Continental Shelves: Research and
Techniques for Visualizing Benthic Environments. Springer International Publishing,
Cham, pp. 273–289.
Fonstad, M.A., Dietrich, J.T., Courville, B.C., Jensen, J.L., Carbonneau, P.E., 2013.
Topographic structure from motion: a new development in photogrammetric mea-
surement. Earth Surface Processes and Landforms 38, 421–430.
Fuller, I.C., Large, A.R.G., Milan, D.J., 2003. Quantifying channel development and se-
diment transfer following chute cutoff in a wandering gravel-bed river.
Geomorphology 54, 307–323.
Gómez-Gutiérrez, Á., Schnabel, S., Berenguer-Sempere, F., Lavado-Contador, F., Rubio-
Delgado, J., 2014. Using 3D photo-reconstruction methods to estimate gully headcut
erosion. CATENA 120, 91–101.
Gonçalves, J.A., Henriques, R., 2015. UAV photogrammetry for topographic monitoring
of coastal areas. ISPRS Journal of Photogrammetry and Remote Sensing 104,
101–111.
Harwin, S., Lucieer, A., 2012. Assessing the accuracy of georeferenced point clouds
produced via multi-view stereopsis from unmanned aerial vehicle (UAV) imagery.
Remote Sensing 4, 1573.
Immerzeel, W.W., Kraaijenbrink, P.D.A., Shea, J.M., Shrestha, A.B., Pellicciotti, F.,
Bierkens, M.F.P., de Jong, S.M., 2014. High-resolution monitoring of Himalayan
glacier dynamics using unmanned aerial vehicles. Remote Sensing of Environment
150, 93–103.
Jaakkola, A., Hyyppä, J., Kukko, A., Yu, X., Kaartinen, H., Lehtomäki, M., Lin, Y., 2010. A
low-cost multi-sensoral mobile mapping system and its feasibility for tree measure-
ments. ISPRS Journal of Photogrammetry and Remote Sensing 65, 514–522.
Javernick, L., Brasington, J., Caruso, B., 2014. Modeling the topography of shallow
B. Chen et al.
braided rivers using Structure-from-Motion photogrammetry. Geomorphology 213,
166–182.
Lane, S.N., Westaway, R.M., Murray Hicks, D., 2003. Estimation of erosion and deposition
volumes in a large, gravel-bed, braided river using synoptic remote sensing. Earth
Surface Processes and Landforms 28, 249–271.
Liu, Y., Chen, B., Liu, L., Li, P., Du, J., Feng, A., 2016. Coastal erosion and its cause
analysis in different spatial temporal scales based on multi sources data in Dongshan
Island of Fujian Province. Haiyang Xuebao 38, 98–110.
Long, N., Millescamps, B., Guillot, B., Pouget, F., Bertin, X., 2016. Monitoring the topo-
graphy of a dynamic tidal inlet using UAV imagery. Remote Sensing 8, 387.
Lucieer, A., de Jong, S.M., Turner, D., 2014a. Mapping landslide displacements using
Structure from Motion (SfM) and image correlation of multi-temporal UAV photo-
graphy. Progress in Physical Geography 38, 97 null.
Lucieer, A., Turner, D., King, D.H., Robinson, S.A., 2014b. Using an unmanned aerial
vehicle (UAV) to capture micro-topography of Antarctic moss beds. International
Journal of Applied Earth Observation and Geoinformation 27 (Part A), 53–62.
Luo, S., Feng, C., Wang, H., 2013. Development of coastal erosion and management.
Advances in Earth Science 28, 1239–1247.
Mancini, F., Dubbini, M., Gattelli, M., Stecchi, F., Fabbri, S., Gabbianelli, G., 2013. Using
unmanned aerial vehicles (UAV) for high-resolution reconstruction of topography:
the structure from motion approach on coastal environments. Remote Sensing 5,
6880–6898.
McKean, J., Roering, J., 2004. Objective landslide detection and surface morphology
mapping using high-resolution airborne laser altimetry. Geomorphology 57,
331–351.
Messinger, M., Asner, G., Silman, M., 2016. Rapid assessments of Amazon forest structure
and biomass using small unmanned aerial systems. Remote Sensing 8, 615.
Miřijovský, J., Langhammer, J., 2015. Multitemporal monitoring of the morphodynamics
of a mid-mountain stream using UAS photogrammetry. Remote Sensing 7, 8586.
Moss, J.L., McGuire, W.J., Page, D., 1999. Ground deformation monitoring of a potential
landslide at La Palma, Canary Islands. Journal of Volcanology and Geothermal
Research 94, 251–265.
Nagihara, S., Mulligan, K.R., Xiong, W., 2004. Use of a three-dimensional laser scanner to
digitally capture the topography of sand dunes in high spatial resolution. Earth
Surface Processes and Landforms 29, 391–398.
Neugirg, F., Stark, M., Kaiser, A., Vlacilova, M., Della Seta, M., Vergari, F., Schmidt, J.,
Becht, M., Haas, F., 2016. Erosion processes in calanchi in the Upper Orcia Valley,
Southern Tuscany, Italy based on multitemporal high-resolution terrestrial LiDAR
and UAV surveys. Geomorphology 269, 8–22.
Niethammer, U., James, M.R., Rothmund, S., Travelletti, J., Joswig, M., 2012. UAV-based
116
Ocean and Coastal Management 160 (2018) 103–116
remote sensing of the Super-Sauze landslide: evaluation and results. Engineering
Geology 128, 2–11.
Notebaert, B., Verstraeten, G., Govers, G., Poesen, J., 2009. Qualitative and quantitative
applications of LiDAR imagery in fluvial geomorphology. Earth Surface Processes and
Landforms 34, 217–231.
Prosdocimi, M., Calligaro, S., Sofia, G., Dalla Fontana, G., Tarolli, P., 2015. Bank erosion
in agricultural drainage networks: new challenges from structure-from-motion pho-
togrammetry for post-event analysis. Earth Surface Processes and Landforms 40,
1891–1906.
Sallenger Jr., A.H., Krabill, W.B., Swift, R.N., Brock, J., List, J., Hansen, M., Holman, R.A.,
Manizade, S., Sontag, J., Meredith, A., Morgan, K., Yunkel, J.K., Frederick, E.B.,
Stockdon, H., 2003. Evaluation of airborne topographic lidar for quantifying beach
changes. Journal of Coastal Research 19, 125–133.
Semyonov, D., 2011. Algorithms Used in Photoscan [Msg 2]. Retrieved May 3. Message
posted to. www.agisoft.ru/forum/index.php?topic=89.0.
Smith, M.W., Carrivick, J.L., Quincey, D.J., 2016. Structure from motion photogrammetry
in physical geography. Progress in Physical Geography 40, 247–275.
Snavely, N., 2008. Scene Reconstruction and Visualization from Internet Photo
Collections. University of Washington, USA.
Squarzoni, C., Delacourt, C., Allemand, P., 2005. Differential single-frequency GPS
monitoring of the La Valette landslide (French Alps). Engineering Geology 79,
215–229.
Turner, D., Lucieer, A., de Jong, S., 2015. Time series analysis of landslide dynamics using
an unmanned aerial vehicle (UAV). Remote Sensing 7, 1736.
Turner, D., Lucieer, A., Malenovský, Z., King, D., Robinson, S., 2014. Spatial Co-
Registration of ultra-high resolution visible, multispectral and thermal images ac-
quired with a micro-UAV over Antarctic moss beds. Remote Sensing 6, 4003.
Turner, I.L., Harley, M.D., Drummond, C.D., 2016. UAVs for coastal surveying. Coastal
Engineering 114, 19–24.
Westoby, M.J., Brasington, J., Glasser, N.F., Hambrey, M.J., Reynolds, J.M., 2012.
‘Structure-from-Motion’ photogrammetry: a low-cost, effective tool for geoscience
applications. Geomorphology 179, 300–314.
Wheaton, J.M., Brasington, J., Darby, S.E., Sear, D.A., 2010. Accounting for uncertainty in
DEMs from repeat topographic surveys: improved sediment budgets. Earth Surface
Processes and Landforms 35, 136–156.
White, S.A., Wang, Y., 2003. Utilizing DEMs derived from LIDAR data to analyze mor-
phologic change in the North Carolina coastline. Remote Sensing of Environment 85,
39–47.
Zhou, Z., Yang, Y., Chen, B., 2016. Spatial and temporal characteristics of surface sedi-
ment grain size in Wujiao Bay, Dongshan Island. Marine Sciences 40, 82–90.