Several indicators can be used to determine the coastline [12] . In this work, it corresponds to the boundary between dry and wet sand ( Figure 2 ).

In our case study, coastlines are extracted by digitally processing a series of Landsat 7 and 8 multi-date images for the periods 1985, 1997, 2009 and 2021. The downloaded images are imported into IDRISI Selva (a combination of three

spectral bands deemed sufficiently clear) for processing and classification. The classified images are subjected to a colour composition in order to obtain colour images based on the signature of the objects. This classified image will make it possible to highlight the clear boundary between the populations of pixels representing the open sea area (dark pixels) and the land area (light pixels) ( Figure 3(a) ).

The files recorded in shapefile format under IDRISSI as classified images are imported into ArcGIS for the extraction and digitisation of coastlines. The extraction of the coastline in Arcgis begins with an initial conversion of the raster image into a polygon. In an editing window, the population of pixels representing the terrestrial domain is removed. The polygon is then transformed into a polyline to obtain the marine domain. From the contours, we remove the contrasts on the polyline image and the contours to obtain the open sea/continent boundary. This corresponds to the digitized coastline. All of this is projected into a WGS_1985_UTM_Zone_28N coordinate system. These coastlines constitute the input data for the DSAS (Digital Shoreline Analysis System) extension, which will be used for statistical calculations of shoreline evolution rates ( Figure 3(b) ). The sum of the global coastline positioning errors for each time step is used to integrate the margin of error into the analysis results. These errors are calculated from Equations (1) and (2) [13] .

$Ept=\frac{\sqrt{{\left(Ep\right)}^2+{\left(Eg\right)}^2+{\left(Ed\right)}^2}}{1}$ Equation (1)

$E\alpha =\frac{\sqrt{{\left(Ept1\right)}^2+{\left(Ept2\right)}^2}}{\text{P}é\text{riode}}$ Equation (2)

Ept: Global coastline position error;

Ep: Pixel error (spatial resolution);

Eg: Georeferencing error (RMS);

Ed: Digitisation error;

Eα: Average overall error in m/year.

Once the data has been formatted in a personal geodatabase, a baseline is created and the shorelines imported into the personal catalogue ( Figure 4 ). Indices are calculated using two shorelines of different dates. An estimate of the uncertainty associated with the method and the choice of statistics for calculating rates of change are taken into account [14] . Transects perpendicular to the shoreline are generated at 50 meters’ intervals from the baseline. After calculation, the End Point Rate (EPR) and Linear Regression Rate-of-change (LRR) indices [2] , were selected from the transect attribute table to plot the variation diagrams for coastline change rates using Excel.

For monitoring purposes, a series of topographic profiles perpendicular to the shoreline are taken from the surf zone to observe the monthly morphological variations due to hydrodynamic agents [15] . Three beach profiles, P1, P2 and P3, were surveyed at each study site, covering the two main morphological units: the aerial beach (the foreshore, upper beach) and the underwater beach at each site ( Figure 5 ).

The data collected is integrated into Excel to determine the morphological units of the beaches and the volumes of sediment mobilized per meter of beach by the dynamic agents. Adobe Illustrator will be used to delineate and annotate these units. The average slope is defined by Equation (3). The volumes of sand eroded or accumulated in each profile are determined from Equation (4).

$P\left(\%\right)=\frac{\Delta H}{\Delta D}\ast 100$ Equation (3)

P = average slope in %;

ΔΗ = height difference (m);

ΔD = distance (m).

V = S × 1 Equation (4)

V = Volume eroded or accumulated in m³;

S = Eroded or accumulated surface area in m²;

1 = linear meter of beach.

The aim of the study of morphosedimentary evolution is to determine the monthly granulometric variations of the transverse profiles. Samples were taken along each topographic profile by simply scraping the surface layer of sediment to avoid having samples that were mixtures of different laminates [16] . Four sampling points were considered on each profile: the high beach, the mid-foreshore, the sea level, also known as the low foreshore, and the wave breaking zone located in the underwater beach ( Figure 6 ).

Considering all the profiles, it can be seen that the study sectors are characterized by sandy sediments with a medium to fine grain size [17] . Textural analysis

was carried out using the usual granulometric techniques, by dry sieving samples of the coarse fraction (∅ > 63 µm), washed and dried, on a series of sieves from the “AFNOR” range. This washing consists of removing salts, impurities and the fine fraction usually made up of silt and clay. After drying, a weight of 180 g of these dry sediments was taken and cold-attacked with 30% dilute hydrochloric acid to eliminate the carbonates. The energy of the medium was characterized using the mean particle size Mz (Equation (5)) supplemented by the Wentworth classification. The grading index σ (Equation (6)) defines the grain size distribution and the asymmetry index SK (Equation (7)) is used to assess grain spread [18] ( Figure 7 ).

$Mz=\frac{Q_{16}+Q_{50}+Q_{84}}{3}$ Equation (5)

Q represents the quartiles.

The formula used is as follows:

$\sigma =\frac{Q_{84}-Q_{16}}{4}+\frac{Q_{95}+Q_5}{6,6}$ Equation (6)

Skewness close to zero.

$SK=\frac{(Q_{16}+Q_{84})-2Q_{50}}{2\left(Q_{84}-Q_{16}\right)}+\frac{(Q_5+Q_{95})-2Q_{50}}{2\left(Q_{84}-Q_{16}\right)}$ Equation (7)

Between 1985 and 1997, the Djiffère and Joal sectors recorded respective coastline advances of 16.05 m ( Figure 8(a) ) and 18.6 m between Ngazobil and Joal ( Figure 8(b) ), corresponding to erosion. On the other hand, to the south of the Joal sector, a retreat of 6.12 m was recorded at Fadhiouth and Fadhiouth Island ( Figure 8(b) ). Between 1997 and 2009, the variation was irregular, with a maximum retreat of 11.2 m towards Palmarin Nguethie and an advance of 8.12 m at Djiffère ( Figure 8(a) ), a retreat of 12.6 m between Ngazobil and the village of Joal and an advance of 24.15 m for the island of Joal ( Figure 8(b) ). Between 2009 and 2021, the recession became more significant along the entire beach, with an average value of 14.21 m at Djiffère ( Figure 8(a) ) and an average advance of 19.25 m at Joal

( Figure 8(b) ). The EPR index between 1985 and 2021 shows a sediment deficit at Palmarin and on the Sangomar breach, with erosion rates of up to −2.13 m/year, although accretion is noted between 2009 and 2021 with an average rate of +1.43 m/year ( Figure 9 ). For the same period, the Joal EPR index shows that the north of the segment (from Joal fadhiouth to Ngazobil) is eroding at a rate of −2.17 m/year, while the south (the island of fadhiouth) is accreting at an average rate of +1.14 m/year. Erosion is more intense at Joal than at Djiffère ( Figure 10 ). This activity is the result of hydrodynamic factors. The accretion is partly linked to the ban on mining beach sand during this period.

Superimposition of the topographic profiles shows strong erosion activity with some accumulation points ( Table 1 ). The evolution of the profiles over time and space, as well as that of the morphological characteristics, confirms the erosion dynamics of the sites studied. Erosion dominates at P1 and P2 for Djiffère and P2-P3 for Joal, with respective accumulations of −13.74 m³/m of beach and −8.65 m³/m of beach at the aerial beach. Positive accumulations were noted at profile P3 for Djiffère and P1 for Joal, with +4.00 m³/m and +5.94 m³/m of beach respectively. These accumulations were noted on the aerial beach at Joal and the underwater beach at Djiffère ( Table 1 ). For both sectors (Djiffère and Joal), the volumes of sediment involved on the upper beach are greater than those on the foreshore ( Table 1 ). From a sedimentological point of view, the monthly change in the mean particle size at the stations is more significant at the foreshore level, regressing at the surf level between November 2022 and December 2022 and increasing very rapidly during the January-April 2023 period for P2 and P3. For P1, it is still regressive ( Figure 11 and Figures 12(a)-(c) ). The mean particle size Mz along the profiles shows that the upper beach has the coarsest sediments and their diameters gradually decrease towards the surf zone ( Figure 11 and Figure 12(d) ). Analysis of the morphosedimentary evolution shows a predominantly positive asymmetry indicating a spreading of the particles towards the fines. The Mz-sigma scatter plot shows the bimodal nature of the sediments, with energy variability ( Figure 13 ).