See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/360644916

Satellite, drone and video camera multi-platform monitoring of coastal

erosion at an engineered pocket beach: A showcase for coastal management at
Elmina Bay, Ghana (West Africa)

Article  in  Regional Studies in Marine Science · May 2022

DOI: 10.1016/j.rsma.2022.102437

CITATIONS READS

6 250

8 authors, including:

Donatus Angnuureng Emmanuel Kwadzo Brempong

University of Cape Coast University of Cape Coast
40 PUBLICATIONS   365 CITATIONS    9 PUBLICATIONS   32 CITATIONS   

SEE PROFILE SEE PROFILE

Philip-Neri Jayson-Quashigah Olusegun A. Dada

University of Ghana Federal University of Technology, Akure
21 PUBLICATIONS   354 CITATIONS    48 PUBLICATIONS   558 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Measuring and Simulating Shoreline Morphodynamics in the Volta Delta View project

Beach response to multi-scale sea level variations View project

All content following this page was uploaded by Philip-Neri Jayson-Quashigah on 20 May 2022.

The user has requested enhancement of the downloaded file.

Journal Pre-proof

Satellite, drone and video camera multi-platform monitoring of coastal

erosion at an engineered pocket beach: A showcase for coastal
management at Elmina Bay, Ghana (West Africa)

D.B. Angnuureng, K.E. Brempong, P.N. Jayson-Quashigah,

O.A. Dada, S.G.I. Akuoko, J. Frimpomaa, P.A. Mattah, R. Almar

PII: S2352-4855(22)00141-4
DOI: https://doi.org/10.1016/j.rsma.2022.102437
Reference: RSMA 102437

To appear in: Regional Studies in Marine Science

Received date : 22 January 2022

Revised date : 12 April 2022
Accepted date : 7 May 2022

Please cite this article as: D.B. Angnuureng, K.E. Brempong, P.N. Jayson-Quashigah et al.,
Satellite, drone and video camera multi-platform monitoring of coastal erosion at an engineered
pocket beach: A showcase for coastal management at Elmina Bay, Ghana (West Africa). Regional
Studies in Marine Science (2022), doi: https://doi.org/10.1016/j.rsma.2022.102437.

This is a PDF file of an article that has undergone enhancements after acceptance, such as the
addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive
version of record. This version will undergo additional copyediting, typesetting and review before it
is published in its final form, but we are providing this version to give early visibility of the article.
Please note that, during the production process, errors may be discovered which could affect the
content, and all legal disclaimers that apply to the journal pertain.

© 2022 The Author(s). Published by Elsevier B.V. This is an open access article under the CC
BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
 Journal Pre-proof

Highlights

1
2
3
4 HIGHLIGHTS
5
6
 We assessed short-time beach changes using a multi-platform data collection

of
7
8 strategy at Elmina Bay, Ghana.
9
10  Sentinel satellite data are a potential alternative tool but inadequate for
11
12
monitoring daily and intraseasonal event-based beach changes.


pro
13 Frequent and local video cameras and drones are effective for monitoring
14
15 shoreline changes at all time scales.
16  Combining sentinel satellite, video camera and drone systems platforms would
17
18 allow large scale, high frequency, and regular observation of the beach for
19 management purposes.
20
21  Accuracy and inadequacy of the various methods should be considered when
22 being deployed for coastal management.
23
24
25
26
27
re-
28
29
30
31
lP
32
33
34
35
36
37
38
39
rna

40
41
42
43
44
45
46
47
48
49
Jou

50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
 Journal Pre-proof

of
Graphical Abstract Click here to access/download;Graphical Abstract;Graphical Abstract.png

pro
re-
lP
rna
Jou
 Journal Pre-proof

Manuscript File Click here to view linked References

Satellite, drone and video camera multi-platform monitoring of coastal erosion at

1
2 an engineered pocket beach: a showcase for coastal management at Elmina Bay,

of
3 Ghana (West Africa)
4
5
6
7 Angnuureng D.B1*., Brempong K.E. 1., Jayson-Quashigah P.N2., Dada O.A3., Akuoko
8 S. G.I. 1., Frimpomaa J1., Mattah P.A1., Almar R4.
9

pro
10 1
Africa Centre of Excellence in Coastal Resilience, Centre for Coastal Management,
11
12 SBS, University of Cape Coast.
13 2 IESS, University of Ghana, Legon.
14
15 3
Department of Marine Science & Tech., Federal University of Technology, Akure,
16
17 Nigeria
18 4
LEGOS (University of Toulouse/CNRS/IRD/CNES), Toulouse, France.
19
20 * Corresponding author, e-mail: donatus.angnuureng@ucc.edu.gh, (+233246630489)
21
22
23
24
ABSTRACT
re-
Regular monitoring of coastal areas is a prerequisite to evade any imminent erosive
25
26 disaster. However, data-driven decisions become more uncertain when a monitoring
27
28 platform is unable to capture events of a certain frequency. Erosion along Ghana's
29
lP
30 coastline is endemic as in most of the Gulf of Guinea countries in West Africa. The
31 current challenge is how to document and understand the dimensions of erosion
32
33 despite limited human and logistical capacity. In the present study, shoreline change
34
35 was assessed by an intensive multi-platform data collection strategy deployed for a
36 year at Elmina Bay, Ghana through the use of drones, a shore-based camera, Sentinel
37
rna

38 satellite images and a dumpy level. The potential causes and areas of erosion at Elmina
39
40 were clearly identified at a very fine scale by our sediment budget calculations. While
41 a section of the beach in front of the Elmina Castle was adequately protected by the
42
43 presence of jetties , downdrift of the larger unprotected portion of the beach was out
44
45
of balance with high erosion rates. Furthermore, the results revealed that frequent,
46 local video cameras and drones are more effective for operational monitoring of
47
48 shoreline changes at all time scales. Satellite imagery is also a potential alternative tool,
Jou

49
but its low temporal resolution, compared with video cameras, makes it unsuitable
50
51 for detecting daily or event-based beach changes, and is of low accuracy for
52
53 practitioners and management decisions.
54
55
56 Keywords: Remote sensing, Coastal management, Shoreline erosion, Shoreline
57
58 monitoring tools, Pocket beaches
59
60
61
62
63
64
65
1
 Journal Pre-proof

1
2 1 Introduction

of
3
4
5 As in many other regions of the world (Luijendijk et al., 2018), coastal recession in
6 West and Central Africa can occur as a result of natural or anthropogenic factors, or
7
8 both (Angnuureng et al., 2013; Mimura, 2013), and when it occurs, it may continue if
9

pro
10 not firmly addressed. Sediment loss due to drift alongshore and across the shore
11 facilitates rapid shoreline recession (Bonou et al., 2018). It is more challenging to
12
13 evaluate shoreline change behaviour when the beach is anthropised. This can be
14
15
exacerbated by activities such as sand mining and the presence of sea defence
16 structures (Ndour et al. 2018; Jayson-Quashigah et al., 2019; Alves et al., 2020;) that
17
18 could result in high erosion, linked to the ‘knock-on effect’ of the groynes (Abessolo
19
et al., 2021). The understanding of the effect of sea defence requires pre-and post sea
20
21
22
23
24
re-
defence assessment (e.g. Angnuureng et al., 2013; Taveneau et al., 2021) from intensive
hydrodynamic, submerged beach bathymetry, and aerial topography measurements
(Anthony et al., 2016; Angnuureng et al., 2020).
25
26
27 The coast of West and Central Africa has been considered largely vulnerable to natural
28
29 hazards and the actions of human interventions (Ndour et al. 2018; Alves et al., 2020).
lP
30
31 Despite the implementation of various measures at different but local scales, erosion
32 has not ceased, suggesting the need to improve management by working
33
34 collaboratively at the regional level to develop regional-specific solutions. The
35
36
development of a holistic plan (Alves et al., 2020) is seen as a sustainable approach to
37 management that seeks to link users/processes together rather than focus on a single
rna

38
39 particular issue and solution. To attain this, regular monitoring, promoting
40 observation networks, national data generalization, centralization and exchange for a
41
42 better understanding of coastal dynamics and pressures are encouraged. The
43
44 availability of consistent data can yield a new methodology for effective coastal
45 management.
46
47
48 Among the diversity of beaches, pocket beaches are prevalent among many coastlines
Jou

49
50 of the world (Daly et al., 2011; Turki et al., 2013), yet the understanding of their
51
52
morphodynamics remains lagging. Pocket beaches are generally relatively short (< 2
53 km) beaches of sand, gravel or both, locked between bedrock headlands, although
54
55 artificial pocket beaches between engineering work also exist. These lateral bounding
56
conditions commonly constrain alongshore sediment supply to pocket beaches which
57
58 are, thus, generally relatively narrow beaches that may show little progradation and
59
60 may flank backshore bedrock, although some pocket beaches also bound lagoons, as
61
62
63
64
65
2
 Journal Pre-proof

in the case of Elmina. This constrained sediment supply and the short and narrow
1
2 nature of pocket beaches may, thus, constitute an important management problem by

of
3 rendering these beaches vulnerable to sediment loss and erosion caused by human
4
5 interventions such as sand mining, maladapted engineering solutions, and in the
6
7 future, sea-level rise (e.g., Brunel and Sabatier, 2007).
8
9

pro
10 Presently, monitoring strategies use remote sensing and in situ platforms in isolation
11 even though they can be optimally used complementarily. One challenge is the
12
13 frequency at which they can be deployed (Angnuureng et al., 2020). Table 1 illustrates
14
15
some pros and cons of some of these platforms. In the last three decades, high-
16 resolution video camera systems (VCS) have advanced the monitoring of shoreline
17
18 changes (e.g., Almar et al., 2012), intertidal bathymetry (e.g., Aarninkhof et al., 2003;
19
Angnuureng et al., 2020) and measurement of wave properties (e.g., Yaniv et al., 2011;
20
21
22
23
24
re-
Angnuureng et al., 2016). Video-based remote sensing is particularly well-suited for
cross-shore shoreline monitoring because it is fixed in one place and covers timescales
from seconds to years, albeit at short spatial scales (Almar et al., 2012). Its application
25
26 can be found in the aspects of assessing natural beach behaviour (e.g. Davidson et al.,
27
28 2007; Smit et al., 2007), artificial nourishment efficiency (e.g. Kroon et al., 2007), and
29
lP
beaches dedicated to recreational use and safety (e.g. Jiménez et al., 2007). In addition
30
31 to video-based remote sensing, Unmanned Aerial Vehicles (UAVs or drones) can be
32
33 used to acquire good data to determine high-resolution sediment loss (Appeaning
34 Addo et al., 2018a; Jayson-Quashigah et al., 2019), particularly in areas where a video
35
36 camera cannot be set up. The spatial coverage of UAV and video systems suffices for
37
the monitoring of bay or micro beaches. These remote techniques have aided the
rna

38
39 acquisition of morphological information to expand research works at sites where
40
41 access to data has been challenging. A summary of some of these coastal monitoring
42
approaches is indicated in Table 1. VCS and UAV are well suited for micro spatial spot
43
44 checks over desired timescales. Satellite images on the other hand could be used for
45
46 regional measurement of coastal changes at monthly to yearly timescales. In addition,
47 satellite image data has been in existence for over half a century, hence providing data
48
Jou

49 on most beaches that could be used as preliminary information for new

50
51 measurements, particularly in less documented areas. Previously, satellite images
52 data was challenged by poor resolution (>20-30 m) but that has changed in recent
53
54 times (with metric to sub-metric resolution, such as the Pleiades (Taveneau et al.,
55
56 2021), making it competitive and complementary to other methods.
57
58
59
60
61
62
63
64
65
3
 Journal Pre-proof

Therefore, in this paper, the main goal is to test the feasibility of using multiple sources
1
2 of remote sensing platforms to assess the dynamics of Elmina Bay beach, Ghana (Fig.

of
3 1) as a showcase for local management.
4
5
6 2 Materials and Methods
7
8
9 2.1 Study site

pro
10
11
12 In contrast to the eastern section of the Ghanaian coast, which is sandier and dynamic
13
14 with more open beaches (Anthony et al., 2016; Appeaning Addo et al., 2018b; Jayson-
15 Quashigah et al., 2019;), the central section is characterised by many rocks (Boye and
16
17 Fiadonu, 2020) and pocket beaches. There is little information on this section of the
18
19 Ghanaian coastline even though it currently experiences various degrees of erosion.
20 Elmina is a town situated along the central coast of Ghana. The beach is located
21
22
23
24
re-
between latitude 5o4’41.67” N and longitude 1o22′ 41.67″ W to the west and latitude
5o5’52.59″ N and longitude 1o18′50.63″ W to the east along the Gulf of Guinea. It has
25 discontinuous beaches, fragmented into rocky and sandy segments. One of these
26
27 beaches is featured by the Elmina Castle adjacent to it and a fishing port built on the
28
29
inlet of the Benya lagoon as well as sea defences constructed (Table 1) on each side of
lP
30 the inlet (Jonah et al., 2016) of the lagoon. Waves approach the coast from a south-
31
32 south-west direction (Fig.1) and the waves reaching the shoreline are weak due to its
33 concave shape and the influence of jetties (Jonah et al., 2016). The lagoon inlet serves
34
35 as a medium for fishermen to move to land and sea. Nearly 40 castles (Osei-Tutu, 2004)
36
37 and forts built between the 15th and 18th centuries by the Europeans are situated along
rna

38 the coastline of Ghana. The Elmina beach accommodates three of these monuments
39
40 one of which is the UNESCO’s world heritage sites (i.e., the Cape Coast Castle, the
41
42 Elmina Castle and Fort St. Jago) but this beach is currently facing various degrees of
43 erosion (Boateng, 2012; Dadson et al., 2016). The presence of erosion along this stretch
44
45 of the coastline requires regular monitoring, especially in that the beach has received
46
47 management intervention through the construction of various coastal defence
48 systems.
Jou

49
50 The beach under the present investigation is about a 1.5 km stretch, starting
51
52
from the Elmina Castle in the west to the Elmina Beach Resort in the east (Fig. 1). The
53 grain size for the Elmina beach is in the range of 1 mm (Anim, 2012). This is principally
54
55 due to the presence of rocks amidst the sand. Despite the coarse sediment, the beach
56
is reported to be experiencing chronic erosion (Boateng, 2012). Historically, the Elmina
57
58 beach has been identified as a hotspot of erosion with reported rates ranging between
59
60 -1.14 and -3.4 m/year (Boateng, 2012; Jonah, 2016). This erosion has led to the
61
62
63
64
65
4
 Journal Pre-proof

destruction of houses and properties, which subsequently led to the construction of

1
2 jetties and revetment in 2017 to protect the Castle and nearby community.

of
3 At the moment, the source of sediment supply to many parts of the beach seems
4
5 to have significantly reduced, maybe due to the jetties/revetments and other human
6
7 activities. Generally, jetties trap sediments up drift, in this case on the western side of
8 the Elmina beach while depriving sediment on the downdrift side, in this case,
9

pro
10 towards the beach resort on the east. There are two main different sea defence types
11
12 on the beach:
13  Two jetties (Fig. 1): At the east of the Benya Lagoon is a 130-m jetty, made of
14
15 boulders aligned across-shore. It is in a good state and located at the mouth of
16
17 the lagoon. There is also a 170-m jetty to the west of the Benya lagoon mouth
18 as well, made of boulders aligned across-shore. These defences are in good state
19
20 and located at the mouth of the lagoon. These were constructed in the years
21
22
23
24

re-
2007 and 2011, respectively.
A revetment (Fig. 1): Behind the Castle, a more recent revetment defence of
about 1 km was constructed in 2017 using aligned boulders alongshore. It is in
25
26 good state and protects the Castle on the West side.
27
28 As jetties and revetments exist on this coast, considerable amounts of sediment can be
29
lP
30 trapped on the updrift side, which can lead to major coastline setbacks on the
31 downdrift side.
32
33 Generally, depths in the nearshore zone decrease towards the east, with significant
34
35 shallow depths in the cove, which could have been influenced by eastward-directed
36 sediment transport, and a substantial sediment exchange likely with the Benya lagoon
37
rna

38 inlet. Previously, it was reported that the Benya lagoon supplies about 226 m3/year of
39
40 sediment to the beach (Boateng et al., 2012). Maybe this sediment has further reduced
41 along most parts of the beach. There are areas of high depths, specifically on the
42
43 western side of the castle where sediment sink. Because of rough nearshore wave
44
45
conditions, a temporal bathymetry survey is a challenging task close to the shore.
46 Nonetheless, nearshore bathymetry has been measured to an offshore distance of
47
48 about 1 km away from the castle on a total area of about 2.5 km2 during this study
Jou

49
which revealed depths ranging between 0 and 10 m.
50
51 Human activities (Boateng, 2012) such as sand mining, inappropriate management
52
53 interventions (Ndour et al. 2018; Alves et al., 2020; Angnuureng et al., 2020) and global
54 climate changes (Mimura, 2013) have been identified as major factors responsible for
55
56 coastal environment challenges in Elmina. For instance, sand mining is widespread
57
58 across the Elmina-Cape Coast-Moree coastline (Jonah et al., 2016) and takes place in
59 several forms, with the magnitude of sand taken from the beach being dependent on
60
61
62
63
64
65
5
 Journal Pre-proof

the transportation medium and the purpose for which sand is to be deposited. This
1
2 seems to take significant tow on the beach stability with a great challenge to address

of
3 along various sections of the coastline.
4
5 Despite the expected threat to life and properties, available data on nearshore
6
7 physical processes, bathymetry, and human interventions that could support critical
8 analysis and management of coastal erosion are fragmented, and incomplete.
9

pro
10
11
12
13
14
15 3 Materials and methods
16
17
18 3.1 Beach profile measurements
19
20 On August 14, 2019, we carried out beach profile measurements using a dumpy level
21
22
23
24
re-
and handheld GPS (Fig. 2). Five beach transects/profiles at about 40 m apart were
measured on a 200 m wide coastline in front of the castle (Fig. 2). The profiles start
from the backshore to the foreshore and were demarcated perpendicular to the beach
25
26 and the baseline based on the slope. They were also used to measure the horizontal
27
28 distance between slope breaks (in meters) and the values were used to determine the
29
lP
30
slope of the beach. A base station of 1.5 m high was established to serve as a point of
31 reference elevation datum with respect to the mean sea level for the study. Relative
32
33 heights were read every 5 m seaward along the transects so that many data points
34 could be collected in less intensive time.
35
36
37
rna

38 3.2 Multi-platform remote sensing beach monitoring techniques

39 3.2.1 UAV (drone)
40
41 Monthly beach surveys were carried out on this beach using DJI Phantom 4 Pro drone
42
43 flights from January to December 2019. However, there was no flight in September
44 2019 due to some technical constraints. Ten permanent ground control points were
45
46 established along the beach and coordinated to the national grid with a D-GPS. To
47
48 avoid the influence of tides, drone flights were, as much as possible, conducted at low
Jou

49 tide levels. The ground control points (GCPs) were used in geo-referencing the drone
50
51 imagery. Averagely, 300 images were captured each month during each flight with a
52
53
total of 11 flights in 2019. These images were manually filtered to remove blurred or
54 overexposed shots.
55
56 A 3D scene reconstruction was performed using structure from motion (SfM)
57 and Multi-View Stereo (MVS) algorithms (Snavely, 2011) implemented in Agisoft
58
59 Metashape software. Analysis of SfM, involves (1) image alignment (Mancini et al.,
60
61
62
63
64
65
6
 Journal Pre-proof

2013), (2) pixel-based dense stereo reconstruction using the aligned data and the GCPs
1
2 and (3) building of orthophoto and digital elevation map (DEM). Where artificial

of
3 objects such as fishing boats, coastal defences and buildings exist along the beach, the
4
5 DEM was classified, as much as possible, to extract only the bare ground; thereby
6
7 reducing exaggeration of topographic heights.
8 The 1.5 km beach was divided into four sections at random distances in relation
9

pro
10 to the presence of defences, sand and rock outcrops. Beach profiles were extracted for
11
12 each section from the drone DEM for each of the months in an ArcGIS platform. The
13 beach profiles were extracted from the lower foreshore to the berm crest. Total errors
14
15 estimated on the vertical elevations (Z) and horizontal locations (XY) were 0.6 and 7.0
16
17 cm, respectively.
18 Orthophotos generated from the UAV images were uploaded into ArcMap.
19
20 Shorelines were manually digitized using the high water mark (HWM) as a proxy
21
22
23
24
re-
(Boak and Turner, 2005) from each of the orthophotos, in order to be consistent with
the proxies used for shoreline studies in Ghana (Jayson-Quashigah et al., 2013; Wiafe
et al., 2013; Angnuureng et al., 2020). Orthogonal transects were cast at 5 m intervals
25
26 to cross the shoreline positions, and these were used to compute the rates of change.
27
28 Shoreline rates of change and horizontal shoreline mobility were computed
29
lP
30 statistically following the Linear Regression Rates (LRR) method of the Digital
31 Shoreline Analysis System (DSAS; Thieler et al. 2009). LRR allows the computation of
32
33 the rates of change using all shoreline data and it was adopted for the rates of change
34
35 estimation along the entire coastline under study.
36
37
rna

38 3.2.2 Video camera system (VCS)

39
40 A video camera system (VCS) with a 180° field view was mounted on the about
41 40-m-high edge of the Elmina Castle wall in November 2018. The VCS was installed
42
43 about 80 m from the shoreline. The final processed beach area on each video image
44
45
was approximately 250 m wide in front of the castle. Timex images, defined as the
46 average of a 15-min set of snapshots (Holland et al., 1997; Angnuureng et al., 2020),
47
48 were selected and the shoreline positions were automatically extracted following the
Jou

49
minimum shoreline variability method (Almar et al. 2012). Almar et al. (2012) have
50
51 shown that beach sand displays high red colouration (R) and low green (G) pigments,
52
53 thus yielding a high ratio of R to G onshore and low R to G in water. From their work,
54 based on some bimodal distribution, the local minimum of this distribution curve
55
56 represents the transition between water and beach, that is, the shoreline position. The
57
58 shoreline is represented by the time-averaged waterline where the ratio was unity. To
59 improve accuracy, where the ratio is below 0.98 or greater than 1.1, no shoreline was
60
61
62
63
64
65
7
 Journal Pre-proof

extracted. These pixel data were converted to metric values by geo-referencing with
1
2 ground control points collected with differential GPS. The shoreline change rates were

of
3 then estimated with the median point rate (MPR). For the MPR method, the shoreline
4
5 change rate (∆𝑦) for all months was estimated by simply finding the difference
6
7 between alongshore-averaged shoreline positions and the overall median position.
8
9

pro
10
11
12 3.2.3 Satellite images
13 Coastsat toolkit (Vos et al., 2019), an open-source software toolkit written for Python
14
15 3.6 or higher, can be used to obtain time-series of shoreline positions at any sandy
16
17 coastline worldwide from publicly available satellite images imagery. It employs the
18 functionality of several freely available Python software packages. In the current
19
20 study, Sentinel-2 satellite images were accessed from January 2019 to December 2019
21
22
23
24
re-
from the GEE archive. Using the functionalities provided within the CoastSat toolkit;
(i) pre-processing of the multispectral images (cloud masking, pan sharpening and
down-sampling); (ii) sub-pixel resolution shoreline extraction; and (iii) shoreline
25
26 positional change measurements along shore-normal transects were conducted.
27
28 The Google Earth Engine (GEE) Python API package was used to access the
29
lP
30 satellite images imagery of Elmina, while machine learning and image processing
31 packages, namely scikit-learn (Pedregosa et al., 2011; Vos et al., 2019) and scikit-image
32
33 (van der Walt et al., 2014; Vos et al., 2019), were employed to automatically extract the
34
35 position of the shoreline from the multispectral imagery. Four transects, as shown in
36 Fig. 1, were drawn across-shore and the shoreline positions were extracted at each
37
rna

38 transect for all the months of all available sentinel images. The MPR approach is
39
40 inbuilt within the Coastsat toolkit. To estimate the shoreline changes, the MPR
41 approach was applied to remove median positions from the alongshore-averaged
42
43 shoreline positions of all transects for all months. The shoreline changes were then
44
45
analysed for the four sections to represent the temporal changes.
46
47
48 3.3 Waves and tide
Jou

49
ERA-5 Reanalysis (C3S, 2017) wave data was used to assess the wave climate of the
50
51 area and its potential influence on coastal dynamics. ERA5 is the 5th generation
52
53 ECMWF reanalysis for the global climate and weather for the past 4 to 7 decades (C3S,
54 2017). These wave data are produced and archived on a different grid to that of the
55
56 atmospheric model, namely a reduced latitude and longitude grid. The horizontal
57
58 resolution of the wave data of ERA5 is about 52 km (C3S, 2017; Bruno et al., 2020).
59
60
61
62
63
64
65
8
 Journal Pre-proof

This data is at nine grid locations of the entire coastline of Ghana. The grid size
1
2 is coarse in spatial coverage and does not fall within the length of Elmina beach, which

of
3 means the ERA5 data is only used here as a representative of regional wave forcing.
4
5 The closest grid swath is around Komenda, between 1.15-1.50° W and 5.03-5.13° N,
6
7 which is about 20 km from Elmina. Significant wave heights of combined wind waves
8 and swells (Hs), peak wave periods (Tp) and wave direction (wdir) were downloaded
9

pro
10 at six-hourly intervals. Wave condition variations were assessed in line with the time
11
12 of image acquisition and shoreline changes.
13 There is no measured water level data for Elmina. Water levels were extracted
14
15 from the WXTide model (Flater, 2010) at 15-minute intervals. This is a free Windows
16
17 tide prediction program. The tide data was used to correct the shoreline variations
18 which were extracted from Sentinel-2 satellite images, UAV flights and VCS.
19
20 As shown in Fig. 3 a-c, the computed monthly mean significant wave height,
21
22
23
24
re-
Hs (Fig. 3b) increased gradually from about 0.8 m in January (dry season) to a peak of
about 1.2 m in July (wet season), and eventually decreased to 0.84 m in December (dry
season). Generally, the mean wave periods, Tp (Fig. 3a) varied between > 10 s and <
25
26 12 s with the peak in May and July (wet season). The wdir (Fig. 3c) are generally
27
28 southwesterly and ranged between 180o and 186o, with the highest wdir value (more
29
lP
30 SW) observed in September 2019 and the lowest (more southerly) in April 2019. Fig.
31 3d presents the astronomical tides that were extracted. The beach is a microtidal zone
32
33 with average heights of 1 m at daily intervals.
34
35
36
37
rna

38 3. Results
39
40 3.1 Description of the seasonal beach profile evolution
41 The cross-shore beach profiles that were measured with the UAV are presented in Fig.
42
43 4. The slopes of the beach were estimated as the ratio of the average profile elevation
44
45
to the seaward distance. The sectional variations of the beach presented a dominant
46 reflective beach with quite steep slopes that averagely ranged from 0.04 to 0.13. The
47
48 beach width decreases from sectors 1, 3, 4, to 2 (Fig. 4). The most accreted profiles
Jou

49
varied from sector to sector probably because this is a bay beach, with a diverse
50
51 direction of wave conditions incidence. For all the monthly profiles (Fig. 4) that were
52
53 extracted from the UAV flights, sectors 1 and 3 profiles present highly dynamic
54 accretion-erosion patterns. These profiles seem to suggest averaged accreting and
55
56 eroding beaches, respectively.
57
58
59
60
61
62
63
64
65
9
 Journal Pre-proof

In Section 1, the survey of the beach profile using dumpy levels revealed gentler or
1
2 dissipative beaches with an average annual slope of 0.04. The vertical range of the

of
3 profile envelope data shows that the topography of Section 1, on about 200 m
4
5 longshore beach length, has changed by approximately 1 m in elevation on the
6
7 foreshore while the backshore remains unchanged within the year. This elevation is
8 however subject to uncertainties due to the presence of several parked fishing boats
9

pro
10 as well as sea defence structures on the beach, which results in topography
11
12 undulations. In December 2019, accretion occurred on the foreshore and the beach
13 widened to about 80 m (Fig. 4a). Overall, the January profile was the most eroded.
14
15 During the field surveys, most of the beach users were found on this side of the beach,
16
17 mostly for recreation.
18 Section 2 (Fig. 4b) is shorter in width, steeper and longer in longshore length,
19
20 and about 50% larger and dominantly sandier than the rest of the Sections. The profile
21
22
23
24
re-
variation in this section (Fig. 4b) was small and uniform compared to adjacent profiles
and the beach was steeper and more reflective. The mean slope of profiles in Section
2 was estimated to be 0.13.
25
26 Profiles in Section 3 were wider and more variable than those in Sections 2 and
27
28 4. Section 3 varied across the entire profile. Sections 2 and 3 were both sandy and open
29
lP
30 to the waves. Fig. 4 suggests that in sections 2 and 3, the December profiles rather were
31 more eroded in contrast to section 1. The average slope of Section 3 is 0.08. The beach
32
33 fronting Section 3 has a large variation between months, especially at the foreshore.
34
35 Sections 2 to 4 are more open to the sea unlike Section 1 which was in between
36 jetties. However, section 4 is backed by rocks, and as shown in Fig. 4d, the profile
37
rna

38 presents several bar-like features. These are rocky outcrops at the Section. At the time
39
40 the data was collected for this study, the beaches in Sections 2 to 4 were unprotected
41 and open to larger wave attacks compared to Section 1 but are now protected with
42
43 rocky sea walls.
44
45
46 3.3 Capturing seasonal shoreline position from multiple remote sensing sources
47
48 As shown in Fig. 5a, the alongshore-averaged (<X>) shoreline positional changes for
Jou

49
the platforms used were computed with data for all months as earlier explained. VCS
50
51 shoreline positions ranged between 25 m and 43 m, excluding a distance of 80 m from
52
53 the camera location. UAV shorelines were digitized in ArcMap with a baseline created
54 at about 20 m from the first shoreline. Thus, the UAV <X> varied between 20 m and
55
56 40 m. Sentinel data obtained with the Coastsat toolkit varied between 20 and 50 m.
57
58 Shoreline change rates were computed by subtracting the shoreline position of the
59
60
61
62
63
64
65
10
 Journal Pre-proof

oldest shoreline from the median of the youngest shoreline. Monthly shoreline
1
2 changes that were estimated showed erosion of -8 m and accretion of about 17 m.

of
3 As shown in Fig. 5a, the shoreline highly advanced during October, November
4
5 and December 2019 while retreat began in February and peaked in April. The average
6
7 rates of shoreline changes were found to be 1.7, 2.8, and 1.0 m for Sentinel, UAV and
8 VCS platforms, respectively, over the entire period. Considering individual platforms,
9

pro
10 changes from the Sentinel satellite images data had a wider range of variations (Fig.
11
12 5a). The monthly cross-shoreline changes observed from the UAV were mostly
13 consistent with those changes observed from the Sentinel, particularly from October
14
15 to December. Despite the sectoral variations in erosion and accretion by transects after
16
17 a year of observations, the entire beach followed an accreting trend though most
18 months recorded sediment loss. The monthly cumulative rate of change was
19
20 approximately 2 m of accretion in 2019. This result is a consequence of activities such
21
22
23
24
re-
as the presence of the defences at some portions (Angnuureng et al., 2013) and the
fluctuations of sediment between the lagoon estuary and the sea.
The influence of tide on the cross-shore shoreline positions was computed for
25
26 all three methods (Fig. 5b). The various approaches reacted differently to tides
27
28 particularly when data collection coincided with spring tides. From our data, the
29
lP
30 shoreline positions from sentinel varied between 20 and 50 m but the accuracy of
31 sentinel shoreline positions was very poor. The accuracy ranged between 0 and 15 m
32
33 due to spatial resolution, tides and image processing. Shoreline positions from UAV
34
35 and VCS on the other hand, with the same range of tide values, were less affected,
36 with much higher accuracy or lesser error. The accuracies were estimated by a ratio of
37
rna

38 the tide to beach slope at the time of image collection. The averaged uncertainties due
39
40 to tides were respectively, 6.7, 3.0 and 3.5 m for sentinel, VCS and UAV on horizontal
41 shoreline position. The difference in the values of accuracies highlights the importance
42
43 of the various approaches.
44
45
The seasonal shoreline positions extracted from the sentinel-2 satellite images,
46 Timex images of the fixed VCS and monthly UAV flights at Section 1 are presented in
47
48 Fig. 5 c-e. The shoreline evolution showed seasonal change patterns. The seasonality
Jou

49
as observed from the Sentinel satellite images, VCS and UAV indicate that the largest
50
51 variations occurred during the Harmattan season (November to March). These
52
53 periods were characterized by both increasing and decreasing hydrodynamic
54 conditions experienced in the study area (Fig. 3), which could result in 3D features
55
56 and non-uniform beaches. The Elmina shoreline position showed larger variations
57
58 from the satellite images and low variation in the VCS and UAV data within the year
59 (Fig. 5 c-e). We observed that the shoreline positions moved more seaward from
60
61
62
63
64
65
11
 Journal Pre-proof

October to December (Dry season) and more landward between March to August
1
2 (Wet/ Rainy season). On the other hand, the data shows that the Rainy season which

of
3 is characterized by moderate to high energy conditions present less variable shoreline
4
5 patterns, probably due to dissipative beach conditions.
6
7
8
9

pro
10 4. Discussion
11
12 4.1 Beach change and the potential drivers
13 The overall beach changes presented were computed using all shorelines in the
14
15 alongshore and cross-shore directions using the linear regression rate (LRR) method
16
17 (Thieler et al., 2009), as shown in Fig. 6a. Along the entire study area, the maximum
18 beach accretion was about 15 m, particularly close to the jetties while the highest
19
20 erosion during the period of study was 3 m. Specifically, LRR results revealed that the
21
22
23
24
re-
observed erosion was highest between transects 500-m and 700-m as one moves
downdrift away from the castle (Fig. 6a). This result is consistent with erosion rates of
3.4 m/year observed by Boateng (2012) in the same area.
25
26 In all, 20 % of the entire study area experienced erosion at rates of about 3 m. As
27
28 shown in Fig. 6a, large portions of sections 3 and 4 are in disequilibrium and will
29
lP
30 probably be eroded. In Fig. 6b, we present sediment volume changes for all months at
31 the various sections. The volume changes were usually estimated by finding the
32
33 difference between sediment volumes of consecutive months.
34
35  Section 1 is observed to be generally accreting. For Section 1, which is protected
36 by a revetment and jetties, the beach is more stable with no record of erosion,
37
rna

38 but very high rates of accretion (~3 m/year). This is revealed from the UAV
39
40 profiles and VCS analyses with the beach building up from January to
41 December, a seaward advancement of the beach for this section. Though
42
43 erosion may have occurred in Section 1, the accretion is larger. The results in
44
45
Fig. 6a suggest that at the frontage of the castle, the shoreline is almost at the
46 same position and thus it may be in a stable state for a long time while the rest
47
48 of the beach would likely be unstable. Based on our observation, if there will
Jou

49
be shoreline instability in Section 1 in the future, it would be due to the
50
51 topography which is low and of gentle slopes. Our findings revealed that
52
53 sediment volumes accumulated throughout most parts of the year (over 54%).
54  In Section 2, the results (Fig. 6a) indicate no erosion near the jetty. Shoreline
55
56 changes of most individual transects show that the shoreline is advancing
57
58 cumulatively seaward, which could be explained by the sediment released by
59 the Benya lagoon that supplies about 226 m3/year of sediment to the beach
60
61
62
63
64
65
12
 Journal Pre-proof

(Boateng et al., 2012). However, on moving eastward away from the jetty, the
1
2 beach begins to experience sediment losses. Averages over the months (Fig. 6b)

of
3 show some losses of sediment in this Section. The whole period of March to
4
5 July showed no recovery or accumulation of sand on the beach. The cause of
6
7 these high changes observed in these months is not clear but it may be due to a
8 combination of any of the following: cross-shore sediment transport, sand
9

pro
10 mining at the beach, the image acquisition time, intermittent storm events,
11
12 wave seasonality owing to seasonal variations (Fig. 3; Dada et al., 2016) and
13 tide-induced errors (Fig. 3d).
14
15 A moderate correlation (r~0.45) was observed between ocean forcing (waves
16
17 and tides) and beach change. This confirms that the beach change may not be
18 solely due to waves. For instance, the area in Section 1 in front of the castle, 0-
19
20 200 m (Fig. 7), between the two jetties was poorly correlated (r~0.0; p>0.6) with
21
22
23
24
re-
waves. The results suggest that wave energy is low, obviously because of the
geometry of the beach as a pocket beach or the effects of the jetties. Section 2 is
the sandiest part of the beach where illicit sand mining occurs by the
25
26 inhabitants (Jonah, 2015). Thus, sand mining in combination with the presence
27
28 of the sea defences and nonlinear wave propagation owing to irregular
29
lP
30 coastline shape could play a significant role in the beach changes.
31
32
33  Downdrift, Sections 3 and 4, experience higher erosion up to a -3 m/year (Fig.
34
35 6a). This high rate of erosion compares well with rates observed by previous
36 studies in the area and along the Eastern Coast of Ghana (e.g., Boateng, 2012;
37
rna

38 Jonah, 2016; Appeaning Addo et al., 2018b; Jayson-Quashiah et al., 2019). The
39
40 main cause of the changes is unclear but based on our findings, it may be due
41 to the wave conditions, the effects of the sea defence located about 400 m to the
42
43 west and sand mining. These factors are heavily resident on this side of the
44
45
beach. Away from the jetty sea defence, waves seem to have a high influence;
46 the correlation between shoreline variation and waves was observed to be r>0.5
47
48 with p<0.05. These parts of the beach are open to wave breaking, yet there is no
Jou

49
defence system. Thus, it was exposed to direct wave action with no
50
51 management intervention at the time of the study. The erosion here has led to
52
53 the destruction of previous seawalls, houses and infrastructure along this
54 stretch of the coast. As in Section 2, in Section 3, most months recorded
55
56 sediment losses. Toward Section 4, where the beach is a bit rocky, sediment loss
57
58 reduces but the shoreline retreat is high. Even though the government has a
59 plan to protect this beach with gabion revetments, such an effort could further
60
61
62
63
64
65
13
 Journal Pre-proof

lead to future instability of the beach and cause erosion of large portions of the
1
2 beach in the downdrift sections, from Sections 2 to 4, if not properly

of
3 implemented. Overall, Fig. 6b shows that March, June and December are the
4
5 months when sediment losses occurred in all Sections. We observed more
6
7 sediment loss during harmattan (November to March), and also from June to
8 September.
9

pro
10
11
12 Elmina beach is characterized by natural drivers and human interventions in various
13 ways, including buildings, sand mining, fishing activities, canoes, and sea defences
14
15 like the jetties. Even though we did not evaluate the precise impact of human
16
17 interventions, they have implications for the study area. The VCS shows that the
18 updrift part of the beach is mostly used to dock canoes and for recreational activities.
19
20 By occupying the beach with boats, these activities have the potential to reduce the
21
22
23
24
re-
width of the beach, influence the sediment distribution, and possibly account partially
for beach erosion. As shown in Fig. 7, the beach was largely used to dock canoes in
May, and this could have affected the estimated width and slope. Quarterly, the beach
25
26 slope changed between 0.07 and 0.13 during the year; JFM - 0.12; AMJ - 0.07; JAS –
27
28 0.13 and OND – 0.08. In terms of width, the beach was widest between October and
29
lP
30 December, and the narrowest between March and May. Presently, erosion occurs at
31 some parts of the beach while others are accreting. Overall, the three platforms used
32
33 here have shown the beach changes in all parts.
34
35
36 4.2. Performance of Sentinel satellite images, UAV, and VCS platforms for
37
rna

38 monitoring beach evolution

39
40 Sentinel satellite and UAV data were provided on a weekly to monthly scale while the
41 video camera system (VCS) was obtained daily. The performance of these platforms
42
43 can be based on their ability to capture all major events that occur on the beach. As
44
45
shown in Fig. 8, sentinel data, acquired on a weekly to monthly basis, is not frequent
46 enough and cannot adequately detect intermittent –storm events- evolution, while
47
48 high-frequency video-derived data can. On seasonal to interannual scales, satellite
Jou

49
data is only able to capture about 80% of the signal at an increasing uncertainty (Fig.
50
51 8c) compared to video-derived data that shows greater significance levels. In addition,
52
53 observed seasonal shoreline variations at Elmina are within the pixel resolution of
54 Sentinel-2 (10 m) compared with UAV and VCS (0.1 m). This currently restricts the
55
56 use of satellite images data for local beach management (Taveneau et al., 2021).
57
58 However, a combination of large scale satellite image-derived information and the
59 high-frequency temporal performances of the VCS has a brighter future.
60
61
62
63
64
65
14
 Journal Pre-proof

1
2 4.3 Sentinel satellite images, UAV, and VCS platforms as management tools

of
3 The potential areas and causes of erosion are clearly identified at a very fine scale by
4
5 our sediment budget calculations. There is a need for more careful consideration, by
6
7 any hard engineering management option, particularly based on what we discussed
8 in section 4.2 with regards to the type of data and the accuracy that would be obtained.
9

pro
10 These platforms have outlined the eventual sources of vulnerability of this beach as
11
12 the beach profile, its slope, topography, erosion, sand mining, and the geometry of the
13 beach. At Elmina, our results inform management about gauging the potential pitfalls
14
15 associated with the emplacement of coastal housing and infrastructure. The present
16
17 solutions at Elmina seem to be fighting coastal erosion, rather than managing it. The
18 results inform management on the need for hybrid solutions such as beach
19
20 nourishment to counter erosion in some areas, stopping beach sand mining and
21
22
23
24
re-
implementing hard engineering to reinforce the coastline. These methods and the
potential results could be useful to the management of pocket beaches all along the
central and eastern coasts of Ghana, and the approach will also be important in
25
26 gauging the response of these vulnerable beaches to sea-level rise (Brunel & Sabatier,
27
28 2007).
29
lP
30
31 5 Conclusions
32
33 This paper analysed the existing regular monitoring platforms as the key to the
34
35 management and mitigation of potential risks with a showcase at Elmina beach,
36 Ghana (West Africa). The potential areas and causes of erosion have been clearly
37
rna

38 identified at a very fine scale by our sediment budget calculations. When compared
39
40 with a daily VCS, monthly sentinel satellite images data and UAV are effective, but
41 insufficient to accurately detect storm events and intra-seasonal beach evolution. We
42
43 have emphasized the accuracy and disadvantages of the various methods while
44
45
presenting the different options to choose from for any desired study. There is a need
46 for more careful consideration, by any hard engineering management option, of the
47
48 type of data to use and the uncertainty to expect.
Jou

49
Our results show that due to the presence of the sea defence structures on the beach
50
51 in front of the castle, no landward recession was observed at this section of the coast.
52
53 The usual ‘knock-on effects’ from hard structures were adequately detected in the
54 non-protected sections which showed various degrees of erosion. Beaches situated
55
56 down the drift are eroding up to -3 m over the study period. The management of the
57
58 beach with only hard engineering may only be fighting the erosion, but the addition
59 of softer solutions such as beach nourishment to counter erosion would eventually
60
61
62
63
64
65
15
 Journal Pre-proof

build the beach. The results of these methods could be useful to the management of
1
2 other pocket beaches all along the coast of Ghana, and the approach will also be

of
3 important in gauging the response of these vulnerable beaches to sea-level rise.
4
5 Overall, this show-case study illustrates the potential of using multi-platform remote
6
7 sensing observations to monitor coastal evolution for local coastal management.
8
9 Author Contributions

pro
10
11 For this research article, the contribution of the authors is as follows:
12
13 Donatus B. Angnuureng: conceptualization, methodology, Writing - Original Draft,
14 Software, formal analysis, funding acquisition, Project administration
15
16 Emmanuel K. Brempong: Investigation, Validation, formal analysis,
17
18 Philip-Neri Jayson-Quashigah: Validation, formal analysis
19
20 Dada A. Olusegu: Validation, Writing - Review & Editing
re-
21
22 Ivy S.G. Akuoko: Investigation, Writing - Review & Editing
23
24 Janet Frimpomaa: Investigation, Writing - Review & Editing
25
26 Precious A. Matta: Writing - Review & Editing, supervision, Project administration
27
28 Rafael Almar: Writing - Review & Editing, Supervision, funding acquisition
29
lP
30
31
32 Conflicts of Interest
33
34 The authors declare no conflict of interest. The funders had no role in the design of
35 the study; in the collection, analyses, or interpretation of data; in the writing of the
36
37 manuscript, or in the decision to publish the results.
rna

38
39
40 Funding
41 This research received funds from the National Geographic Society (CP-107T-17),
42
43 The World Academy of Science (17-429 RG/PHYS/AF/AC_IFR3240300132) and the
44
45 Institute of Research for Development (IRD-JEAI).
46
47
48
Jou

49
50 References
51 1. Abessolo G. O., Hoan L. X., Larson M., Almar R. 2021. Modeling the Bight of
52
53 Benin (Gulf of Guinea, West Africa) coastline response to natural and
54
55 anthropogenic forcing. Regional Studies in Marine Science, 48, p. 101995 [12 p.].
56 ISSN 2352-4855
57
58
59
60
61
62
63
64
65
16
 Journal Pre-proof

2. Aarninkhof, S.G.J., Turner, I.L., Dronkers, T.D., Caljouw, M., Nipius, L., 2003. A
1
2 video-based technique for mapping intertidal beach bathymetry. Coast. Eng.,

of
3 49(4), 275–289. doi:10.1016/s0378-3839(03)00064-4.
4
5 3. Almar, R, Ranasinghe, R, Catalan, PA, Senechal, N, Bonneton, P, Roelvink, D,
6
7 Bryan, KR, Marieu, V, Parisot, JP. 2012. Video-based detection of shorelines at
8 complex meso-macro tidal beaches. Journal of Coastal Research, 28(5), 1040-1048.
9

pro
10 4. Alves, B., Angnuureng, D.B., Morand, P., Almar, R., 2020. A review on coastal
11
12 erosion and flooding risks and best management practices in West Africa: what
13 has been done and should be done. J Coast Conserv 24, 38 (2020).
14
15 https://doi.org/10.1007/s11852-020-00755-7
16
17 5. Angnuureng, D.B., Jayson-Quashigah, P-N., Almar, R., Stieglitz, T.C., Anthony,
18 E.J., Aheto, D.W., Appeaning Addo, K., 2020. Application of shore-based video
19
20 and unmanned aerial vehicles (drones): complementary tools for beach studies.
21
22
23
24
6.
Remote Sens. 12:394–413
re-
Angnuureng, D.B., Almar, R., Appeaning Addo, K., Senechal N., Castelle, B.,
Laryea S.W., Wiafe, G., 2016. Video observation of waves and shoreline change on
25
26 the microtidal James Town beach in Ghana. J. Coast. Res. 75:1022–1026
27
28 7. Angnuureng, B.D., Appeaning Addo, K., Wiafe, G., 2013. Impact of sea defence
29
lP
30 structures on downdrift coasts: The case of Keta in Ghana. Acad. J. Environ. Sci. 1,
31 104–121, doi:10.15413/ajes.2013.0102.
32
33 8. Anim, M., 2012. Lithological responses to sea erosion along the coastline from
34
35 Gold Hill (Komenda) to Amissano (Saltpond). University of Cape Coast, MPhil
36 Dissertation. Pg 106
37
rna

38 9. Anthony, E., Almar, R., Aagaard, T.,2016. Recent shoreline changes in the Volta
39
40 River delta, West Africa: The roles of natural processes and human impacts. Afr.
41 J. Aquat. Sci. 41, 81–87.
42
43 10. Appeaning Addo, K., Nicholls, R.J., Codjoe, S.N.A., Abu, M. 2018b. A biophysical
44
45
and socioeconomic review of the Volta Delta, Ghana. J. Coast. Res 34(5):1216–1226
46 11. Appeaning Addo, K., Jayson-Quashigah, P.-N., Codjoe, S.N.A., Martey, F., 2018a.
47
48 Drone as a tool for coastal flood monitoring in the Volta Delta, Ghana.
Jou

49
Geoenviron. Disasters 5, 17. https://doi.org/10.1186/s40677-018-0108-2.
50
51 12. Boak, E.H., Turner, I.L., 2005. Shoreline definition and detection: a review. J. Coast.
52
53 Res. 214, 688–703. https://doi.org/10.2112/03-0071.1.
54 13. Boateng, I., 2012. An application of GIS and coastal geomorphology for large scale
55
56 assessment of coastal erosion and management: a case study of Ghana. J. Coast.
57
58 Cons. 16:383–397
59
60
61
62
63
64
65
17
 Journal Pre-proof

14. Bonou, F., Angnuureng, D.B., Sohou, Z., Almar, R., Alory, G., Du Penhoat, Y.,
1
2 2018. Shoreline and Beach Cusps Dynamics at the Low Tide Terraced Grand Popo

of
3 Beach, Bénin (West Africa): A Statistical Approach. J. Coast. Res. 81, 138–144.
4
5 15. Boye, C. B., Fiadonu, E.B., 2020. Lithological effects on rocky coastline stability.
6
7 Heliyon, 6(2020) e03539, 1–10
8 16. Brunel, C., Sabatier, F, 2007. Pocket beach vulnerability to sea-level rise. Journal
9

pro
10 of Coastal Research, Special Issue 50, 604-609.
11
12 17. Bruno, M.F., Molfetta, MG., Totaro, V., Mossa, M.(2020). Performance
13 Assessment of ERA5 Wave Data in a Swell Dominated Region.
14
15 18. Copernicus Climate Change Service (C3S), 2017. ERA5: Fifth generation of
16
17 ECMWF atmospheric reanalyses of the global climate . Copernicus Climate
18 Change Service Climate Data Store (CDS),
19
20 https://cds.climate.copernicus.eu/cdsapp#!/home. Accessed on 14 th August, 2020.
21
22
23
24
re-
19. Dada, O.A., Li, G.X., Qiao, L.L., Ding, D., Ma, Y.Y., Xu J.S., 2016. Seasonal shoreline
behaviours along the arcuate Niger Delta coast: Complex interaction between
fluvial and marine processes. Continental Shelf Research, 122, 51-67.
25
26 20. Dadson, I.Y., Owusu, A.B., Adams, O., 2016. Analysis of shoreline change along
27
28 Cape Coast-Sekondi Coast, Ghana. Geo. J., 2016, 1–9
29
lP
30 http://dx.doi.org/10.1155/2016/1868936
31 21. Daly, C.J., Bryan, K.R., Roelvink, J.A., Klein, A.H.F., Hebbeln, D., Winter, C., 2011.
32
33 Morphodynamics of Embayed Beaches: The Effect of Wave Conditions. J. Coast.
34
35 Res. SI 64 (Proceedings of the 11th International Coastal Symposium), 1003 – 1007.
36 22. Davidson, M., Van Koningsveld, M., de Kruif, A., Rawson, J., Holman, R.,
37
rna

38 Lamberti, A., Medina, R., Kroon, A., Aarninkhof, S., 2007. The Coast-View project:
39
40 Developing video-derived coastal state indicators in support of coastal zone
41 management, Coast. Eng., 54(6 – 7), 463 – 475, doi:10.1016/j.coastaleng.2007.01.007.
42
43 23. Flater, D., 2010. WXTide32 - a free Windows tide and current prediction program.
44
45
www.wXtide32.com. Last accessed on 22/11/2020.
46 24. Holland, K., Holman, R., Lippmann, T., Stanley, J., Plant, N., 1997. Practical use of
47
48 video imagery in nearshore oceanographic field studies. IEEE J. Ocean. Eng., 22,
Jou

49
81–92.
50
51 25. Jayson-Quashigah, P.N., Appeaning Addo, K., Amisigo, B., Wiafe, G., 2019.
52
53 Assessment of short-term beach sediment change in the Volta Delta coast in
54 Ghana using data from unmanned aerial vehicles (drone). Ocean Coast Manage
55
56 161:104952. https://doi.org/10.1016/j. ocecoaman.2019.104952
57
58 26. Jayson-Quashigah, P.N., Appeaning Addo, K., Kufogbe, S.K., 2013. Medium
59 resolution satellite images imagery as a tool for monitoring shoreline change. Case
60
61
62
63
64
65
18
 Journal Pre-proof

study of the Eastern coast of Ghana. J. Coast. Res. 65, 511–516, doi:10.2112/SI65-
1
2 087.1.

of
3 27. Jiménez, J.A., Osorio, A., Marino-Tapia, I., Davidson, M., Medina, R., Kroon, A.,
4
5 Archetti, R., Ciavola, P., Aarnikhof, S.G.J., 2007. Beach recreation planning using
6
7 video-derived coastal state indicators. Coast. Eng., 54(6–7), 507–521.
8 https://doi.org/10.1016/j.coastaleng.2007.01.012
9

pro
10 28. Jonah, F.E., Boateng, I., Osman, A., Shimba, M.J., Mensah, E.A., Adu-Boahen K.,
11
12 Chuku, E.O., Effah, E., 2016. Shoreline change analysis using end point rate and
13 net shoreline movement statistics: An application to Elmina, Cape Coast and
14
15 Moree section of Ghana’s coast. Reg. Studies in Mar. Sci. 7, 19–31.
16
17 http://dx.doi.org/10.1016/j.rsma.2016.05.003
18 29. Jonah, F.E., 2015. Managing coastal erosion hotspots along the Elmina, Cape Coast
19
20 and Moree area of Ghana. J. Ocean Coast. Manag.
21
22
23
24
re-
doi:10.1016/j.ocecoaman.2015.02.007
30. Kroon, A., Davidson, M.A., Aarninkhof, S.G.J., Archetti, R., Armaroli, C.,
Gonzalez, M., Medri, S., Osorio, A., Aagaard, T., Holman, R. A., Spanhoff, R., 2007.
25
26 Application of remote sensing video systems to coastline management problems.
27
28 Coastal Engineering, 54(6–7), 493–505.
29
lP
30 https://doi.org/10.1016/j.coastaleng.2007.01.004
31 31. Luijendijk, A., Hagenaars, G., Ranasinghe, R. et al. The State of the World’s
32
33 Beaches. Sci Rep 8, 6641 (2018). https://doi.org/10.1038/s41598-018-24630-6
34
35 32. Mancini, F., Dubbini, M., Gattelli, M., Stecchi, F., Fabbri, S., Gabbianelli, G., 2013.
36 Using Unmanned Aerial Vehicles (UAV) for High-Resolution Reconstruction of
37
rna

38 Topography: The Structure from Motion Approach on Coastal Environments.

39
40 Remote Sens. 5, 6880–6898.
41 33. Ndour, A., Laïbi, R., Sadio, M., Degbé, C.G.E., Diaw, A.T., Oyédé, L.M., Anthony,
42
43 E.J., Dussouillez, P., Sambou, H., Dièye, E.H.B., 2018. Management strategies for
44 coastal erosion problems in West Africa: Analysis, issues, and constraints drawn
45
46
from examples from Senegal and Benin. Ocean & Coastal Management, 156, 92-106.
47 https://doi: 10.1016/j.ocecoaman.2017.09.001.
48
Jou

49
50 34. Mimura, N., 2013. Sea-level rise caused by climate change and its implications for
51 society. Sea-level rise caused by climate change and its implications for society.
52
53 Proc. Jpn. Acad. Ser. B Phys. Biol. Sci., 89 (7) (2013), pp. 281-301
54
55
35. Osei-Tutu, B., 2004. "African American reactions to the restoration of Ghana's
56 'slave castles' ", in: Public Archaeology; 3/4, 2004, pp. 195–204. ISSN 1465-5187
57
58 36. Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O.,
59
Blondel, M., Prettenhofer, P., Weiss, R., Dubourg, V., Vanderplas, J., Passos, A.,
60
61
62
63
64
65
19
 Journal Pre-proof

Cournapeau, D., Brucher, M., Perrot, M., Duchesnay, E., 2011. Scikit-learn
1
2 machine learning in Python. J. Mach. Learn. Res. 12, 2825–2830.

of
3 https://doi.org/10.1007/ s13398-014-0173-7.2.
4
5 37. Smit, M.W.J., Aarninkhof, S.G.J., Wijnberg, K.M., González, M., Kingston, K.S.,
6
7 Southgate, H.N., Ruessink, B.G., Holman, R.A., Siegle, E., Davidson, M., Medina,
8 R., 2007. The role of video imagery in predicting daily to monthly coastal
9

pro
10 evolution. Coast. Eng., 54(6–7), 539–553.
11
12 https://doi.org/10.1016/j.coastaleng.2007.01.009.
13 38. Snavely, N., 2011. Scene Reconstruction and Visualization from Internet Photo
14
15 Collections: A Survey. IPSJ Trans. Comput. Vis. Appl. 3, 44–66.
16
17 39. Taveneau A., Almar Rafaël, Bergsma E. W. J., Sy B. A., Ndour A., Sadio M., Garlan
18 T. (2021). Observing and predicting coastal erosion at the Langue de Barbarie sand
19
20 spit around Saint Louis (Senegal, West Africa) through satellite-derived digital
21
22
23
24
re-
elevation model and shoreline. Remote Sensing, 13 (13), p. 2454
40. Thieler, E.R., Himmelstoss, E.A., Zichichi, J.L., Ergul, A., 2009. The Digital
Shoreline Analysis System (DSAS) Version 4.0 - An ArcGIS Extension for
25
26 Calculating Shoreline Change. Open-File Report. US Geological Survey Report
27
28 No. 2008- 1278: http://woodshole.er.usgs.gov/projectpages/dsas/version4/
29
lP
30 41. Thomas, T., Rangel-Buitrago, N., Phillips, M.R., Anfuso, G., Williams, A.T., 2015.
31 Mesoscale Morphological Change, Beach Rotation and Storm Climate Influences
32
33 along a Macrotidal Embayed Beach. J. Mar. Sci. Eng. 3, 1006-1026;
34
35 doi:10.3390/jmse3031006
36 42. Turki I , Medina R., Gonzalez M., Coco G. (2013). Natural variability of shoreline
37
rna

38 position: Observations at three pocket beaches. Marine Geology, 338, 76-89,

39
40 http://dx.doi.org/10.1016/j.margeo.2012.10.007
41 43. van der Walt, S., Schonberger, J.L., Nunez-Iglesias, J., Boulogne, F., Warner, J.D.,
42
43 Yager, N., Gouillart, E., Yu, T., 2014. scikit-image: image processing in Python.
44
45
PeerJ 2, e453. https://doi.org/10.7717/peerj.453.
46 44. Vos, K., Splinter, K.D., Harley, M.D., Simmons, J.A., Turner, I.L., 2019. CoastSat:
47
48 A Google Earth Engine-enabled Python toolkit to extract shorelines from publicly
Jou

49
available satellite imagery. Environmental Modelling and Software 122 (2019)
50
51 104528. https://doi.org/10.1016/j.envsoft.2019.104528
52
53 45. Wiafe, G., Boateng, I., Appeaning Addo, K., Quashigah, P.N., Ababio, S.D.,
54 Laryea, S.W., 2013. Handbook of Coastal Processes and Management in Ghana; The
55
56 Choir Press: Gloucester, UK, 2013; p. 274.
57
58
59
60
61
62
63
64
65
20
 Journal Pre-proof

46. Yaniv G., Browne M. and Lane C.(2011). Automatic Estimation of Nearshore Wave
1
2 Height from Video Timestacks. International Conference on Digital Image Computing:

of
3 Techniques and Applications. DOI: 10.1109/DICTA.2011.68
4
5
6
7
8
9

pro
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
re-
25
26
27
28
29
lP
30
31
32
33
34
35
36
37
rna

38
39
40
41
42
43
44
45
46
47
48
Jou

49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
21
 Journal Pre-proof

Figure Click here to access/download;Figure;Figures.docx

1
2

of
3
4
5
6
7
8
9

pro
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
re-
25
26
27
28
29
lP
30
31
32
33
34
35
36
37
rna

38
39
40
41
42 Fig. 1. a) Map of the study site showing Elmina, inset the map of Ghana showing Elmina (a
43 red box). b) The orthophoto shows the four sections, sections 1 to 4 of the Elmina beach where
44 analysis was done. Sea defenses are indicated in blue rectangles.
45
46
47
48
Jou

49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
 Journal Pre-proof

1
2

of
3
4
5
6
7
8
9

pro
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
re-
25
26
27
28
29
lP
30
31
32
33
34
35
36
37
rna

38
39
40 Fig. 2. Beach profile measurements in front of the Elmina Castle using a dumpy level.
41
42
43
44
45
46
47
48
Jou

49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
 Journal Pre-proof

1
2

of
3
4
5
6
7
8
9

pro
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
re-
25
26
27
28
29
lP
30
31
32
33
34
35
36
37 Fig. 3. Hydrodynamic forcing at Elmina Bay. a) Peak periods of the wave, Tp b)
rna

38 significant wave heights, Hs c) wave direction, wdir of ERA 5 reanalysis at Elmina, d)

39
40 tide range, TR; Red circles indicate time of drone flights and blue squares indicate
41
42 the time of Sentinel imagery.
43
44
45
46
47
48
Jou

49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
 Journal Pre-proof

1
2

of
3
4
5
6
7
8
9

pro
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
re-
25
26
27
28
29
lP
30
31
32
33 Fig. 4a-d. Monthly UAV topographic profiles variations between January and
34
35 December 2019 for Sections 1, 2, 3 and 4, respectively (see Fig. 1) on the beach;
36
37
rna

38
39
40
41
42
43
44
45
46
47
48
Jou

49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
 Journal Pre-proof

1
2

of
3
4
5
6
7
8
9

pro
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
re-
25
26
27
28
29
lP
30
31
32
33
34 Fig. 5. Seasonal shoreline position from remote sensing platforms. a) Monthly
35
36 shoreline positions from Sentinel satellite, VCS and UAV in 2019; b) Monthly error
37
rna

38 induced by the tide on shoreline positions; c) Mean seasonal shoreline positions

39 obtained from Sentinel satellite d) Mean seasonal shoreline positions obtained from
40
41 VCS; e) Mean seasonal shoreline positions obtained from UAV; From c to e, the boxes
42
43 and error bars represent the standard deviation from mean positions.
44
45
46
47
48
Jou

49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
 Journal Pre-proof

1
2

of
3
4
5
6
7
8
9

pro
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
re-
25
26
Fig. 6. a) Shoreline change rate obtained from all monthly shorelines using LRR;
27 Negative values means the sea is advancing landward; b) Monthly sediment volume
28
29 changes at the sectors along the beach
lP
30
31 a)
32
33
34
35
36
37 b)
rna

38
39
40
41
42
43
44 c)
45
46
47
48
Jou

49
50
51 d)
52
53
54
55
56
57 Fig. 7. Beach occupation during the year 2019. From top to down, a) March, b) May, c)
58
59 September and d) December.
60
61
62
63
64
65
 Journal Pre-proof

1
2

of
3
4
5
6
7
8
9

pro
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
re-
25
26
27
28
29
lP
30
31
32
33
34
35
36
37 Fig. 8. The estimated performance of daily (VCS data) and monthly data (UAV and
rna

38
39 Sentinel) and the signals of VCS, UAV and Sentinel platforms.
40
41
42
43
44
45
46
47
48
Jou

49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
 Journal Pre-proof

Table Click here to access/download;Table;Table.docx

Table 1. Review of data collection methods.

1
2 Item Dumpy level UAV VCS Satellite

of
3
4 Longshore > 1 km > 1 km < 1 km >100 km
5 coverage
6
7
8
9 Best temporal Daily Daily Minutes Weekly

pro
10 resolution
11
12 Age > 3 decades < 2 decades <3 decades > 3 decades
13
14 Orientation to Perpendicular Dynamic Oblique Dynamic
15
16 shoreline
17
18 Data coverage Spot checking Micro (Spot Micro (Fixed Macro (Global
19 checking) spot) or large data)
20
21
22
23
24
Best accuracy

Data collection
cm

>2 persons
re-<10 cm

2 persons
≈1m

1 person
>3 m

Remotely
25
26 Risk to user Risky at high Spinning blades Safe Safe
27 tides
28
29 Remote use not possible Secured Not secured Secured
lP
30
31
32
33 Processing time Short Long (dense Short (light Long
34
35 map) image)
36
37 Georeferencing no applicable Manual Automatic Required GPS
rna

38 of images Coordinates
39
40
41
42
43
44
45
46
47
48
Jou

49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
 Journal Pre-proof

Author Contributions
For this research article, the contribution of the authors is as follows:

of
Donatus B. Angnuureng: conceptualization, methodology, Writing - Original Draft,
Software, formal analysis, funding acquisition, Project administration
Emmanuel K. Brempong: Investigation, Validation, formal analysis,

pro
Philip-Neri Jayson-Quashigah: Validation, formal analysis
Dada A. Olusegu: Validation, Writing - Review & Editing
Ivy S.G. Akuoko: Investigation, Writing - Review & Editing
Janet Frimpomaa: Investigation, Writing - Review & Editing
Precious A. Matta: Writing - Review & Editing, supervision, Project administration
Rafael Almar: Writing - Review & Editing, Supervision, funding acquisition
re-
lP
rna
Jou
 Journal Pre-proof

Declaration of interests

of
☒ The authors declare that they have no known competng fnancial interests or personal relatonships
that could have appeared to influence the work reported in this paper.

☐ The authors declare the following fnancial interests/personal relatonships which may be considered
as potental competng interests:

pro
re-
lP
rna
Jou

View publication stats