From a cartographic synthesis, geological description, to the use of modelling through the analysis of satellite images, of DTMs, various types of data sources were used to process the variables retained for the IVC model ( Table 1 ).

A spatio-temporal analysis of the evolution of the coastline was developed in order to calculate the erosion rate. In this case, the analysis first consists of a classification supervised by digital processing of a series of multi-date Landsat 7 and 8 images over the period 1985 and 2021. The uploaded images are imported into IDRISI Selva for processing and classification. Classified images are subjected to a color composition in order to obtain color images based on the signature of the objects. This classified image will highlight the clear boundary between the pixel populations representing the open ocean domain (dark pixels) and the terrestrial area (light pixels). Files saved in shapefile format under IDRISSI as classified imagery are imported into ArcGIS for extraction and digitization of coastlines. The work of extracting the coastline on Arcgis consists first of all in making an initial conversion of the raster image into a polygon. On an edit window, we delete the population of pixels representing the terrestrial domain. Then we transform the polygon into a polyline to obtain the marine domain. From the contours, we remove the contrasts on the polyline image and the contours in order to obtain the boundary between the open sea and the continent. It thus corresponds to the coastline after digitization. These coastlines constitute the input data for the DSAS (Digital Shoreline Analysis System) extension that will allow statistical calculations of the coastline evolution rates [8] . After formatting the data in a personal geodatabase, a baseline is created and the shorelines are imported into the personal catalog edition. The calculation of indices is carried out using the 1985 and 2021 coastline. Transects perpendicular to the coastline are generated with a measurement step of 50 meters from the baseline. After calculation, the End Point Rate (EPR) and the Linear Regression Rate-of-change (LRR) indices are retained from the transect attribute table to plot the variation of the rates of change of the coastline with Excel.

The data used to analyse the degree of vulnerability linked to the geomorphology of the coasts studied are taken from Digital Terrain Models (DTMs). These DEMs are obtained through the United States Geological Survey (USGS) online portal and on the open topography DEM extension on QGIS. The DEMs were merged with the data from the topographic maps and processed in QGIS to extract the study areas. On the layer style menu, a color palette has been created of type cpt-city catalog and the topography model is selected. A number of five classes are defined with an equal interval display mode. On raster, contour lines were generated with a spacing of 50m between them. The processing of the data made it possible to recognize the geomorphological facies of the coastline confirmed by field observations.

The slopes of the coastal zone of our study areas were calculated from a grid of topographic elevations. These are obtained by processing the DTM by projecting it into a coordinate system WGS_1984_UTM_Zone_28N covering the study area. On the raster menu, a slope analysis was done to calculate the edges. On analysis, a Pseudo-color banded rendering is chosen with a shading factor of z = 5 for a multiple blending mode. This data was merged and processed with QGIS to establish a slope map.

A numerical model of the MIKE21-DHI is used in this study to calculate significant wave heights at the coast (Hs). We will first delineate the study area on ArcGIS using the Google Earth map. All the data necessary for the simulation (bathymetric, meteorological and sea level data) are collected. Return periods of 8 s and 14 s, respectively, for the Dakar and Djiffère to Rufisque sectors respectively allow us to assess the sensitivity related to this variable.

The land cover of the study area was determined through QGIS processing of NTMs, as well as Pleiades images uploaded through the livingatlas.arcgis.com online portal to the Land Cover Explorer menu for a directed classification approach. On raster, the data collected were superimposed with the administrative map of Senegal to extract our NCD study areas. Then a shading factor z = 25 is applied to the layer style window. A validation stage was undertaken to establish the occupancy map.

The data used to analyse the degree of vulnerability related to coastal geology are taken from the World Geologic Maps and the administrative map of Senegal. These maps are obtained through the United States Geological Survey (USGS) online portal. They were merged and processed under QGIS to extract the study areas. The processing of digital and field data made it possible to establish the geological map of the study area.

The coastal vulnerability index was obtained by processing each variable in a GIS model by applying a weighting system to 6 degrees of scoring (a code from 1 to 6) according to a 30 m*30 m cell calculation grid [9] . Each variable is assigned to a degree depending on the potential magnitude of its exposure and its contribution to physical changes in the coast. The 6 degrees obtained are calculated using Equation (1) [6] .

$CVI=\sqrt[\cdot ]{\frac{a.b.c.d.e.f}{6}}$ Equation (1)

a = Erosion rate; b = geomorphology; c = slope; d = wave heights, e = land use; f = geology.

The rates of change during the period studied reveal significant erosion in several parts of the sandy coast from Djiffère to Joal [10] and Rufisque (more than 2 m/year) ( Figure 2(c) ), therefore, a weighted value of 5 has been assigned to them. This could be linked to the massive construction on the dune ridges and the unregulated extraction of sand (looting) [6] . For the same period, the EPR index shows moderate to low erosion on the coast of Dakar (maximum 1.02 m/year) ( Figure 2(a) ), a weighted value of 4 has been assigned to them. This low erosion noted in Dakar is linked to the presence of hard rocks that are resistant to sea forcing.

Figure 3 shows that the gradient of the slope of the coastal reaches studied varies between 0.1% and 90%. At the level of Djiffère, Joal and Rufisque, on this part consisting mainly of beaches with slope gradients of less than 5% ( Figures 3(a)-(c) ) are more likely to undergo sediment leaching against the current. Therefore, at low slopes, the coast is expected to recede faster than steep slopes. These zones are assigned weighted values of 4. In contrast, the rocky coasts of Dakar ( Figure 3(d) ), which have gentle slopes with a gradient of up to 89%, are more resistant to erosion, hence a weighting of 2.

The study areas have alternating beaches, dunes, d’une ridges, cliffs, capes and rocky bays ( Figure 4 ). Depending on the relief encountered, the risks of coastal erosion are more and more important at the level of the Djiffère section in Rufisque and more or less important at the level of the Dakar coast. Coastal landforms that are at low altitudes (<10 m) ( Figures 4(a)-(c) ), such as sandy beaches, are assigned a value of 5 because they have a high potential for flooding or marine submersion. The coasts in the Dakar sector with cliffs and rocky shores have an altitude (>40 m) ( Figure 4(d) ) and are considered less vulnerable because they are more resistant to sea forcing and sea level rise. As a result, they were assigned a weighted value of 2.

The output models show that almost the entire coastline from Djiffère to Rufisque and especially to Dakar is exposed to wave and swell action ( Figure 5 ). This high exposure is linked to the fact that coastal and underwater morphology dissipates little marine energy. The coastal areas of the Dakar study area show an increase in wave height of more than 2.8 m ( Figure 5 ), releasing more energy on the coast, resulting in significant erosion or subsidence of beaches. As a result, a weighted value of 5 was assigned to these ratings. The low-energy waves from Djiffère to Rufisque ( Figure 5 ) produce a lower wave height (<1.6 m) which leads to accretion processes due to the accumulation of sediments. Therefore, a weighted value of 3 was assigned to these low wave height areas.

Figure 6 illustrates the spatial distribution of the land cover type of the sectors studied. This analysis made it possible to detect several land cover classes varying from one sector to another. The overall analysis revealed that the coasts studied are highly urbanized and industrialized at the level of Dakar ( Figure 6(d) ), mainly in its parts close to the coast. Agricultural activity, flooded vegetation and trees, are more frequent from Djiifère to Joal ( Figures 6(a)-(c) ), particularly in areas close to the sea. These natural and anthropogeni+639-c actions impact the plant tissue and the encroachment of the dune barriers. The latter, which are considered as a natural protective barrier, are experiencing a significant rate of decrease. They were assigned a weighted value of 5.

From a geological point of view, the study areas present several geological facies, as a result, their exposure and resistance to marine agents. Indeed, local geology plays a decisive role in the rate of coastal erosion. Coasts composed of soft rocks or loose sediments are more prone to erosion than those made of hard rocks. The low resistance of the loose formations that make up the coast from Djiffère to Rufisque ( Figure 7(a) and Figure 7(b) ) causes a significant retreat of the beaches. Therefore, a weighted value of 5 was assigned to these ratings. The coasts in the Dakar sector with hard and soft cliffs ( Figure 7(a) ) are considered less vulnerable because they are more resistant to sea forcing and sea level rise. In addition, some places have moderate intensities of vulnerability due to the geological structure of the coherent and resistant rocks. This tectonic structure is identified by a few faults in the region, and is characterized by a sub-horizontal tabular structure. Nevertheless, it partly influences the morphology of the coastal zone, the division of the coastline and the dynamics. In addition, limestone cliffs are particularly susceptible to erosion, as limestone is more soluble in water, resulting in the formation of caves, arches, and other spectacular coastal formations. As a result, they were assigned a weighted value of 3.

The results show that the IVC values vary from 6.13 to 21.4 in the Djiffère sector in Rufisque and from 1.72 to 14.8 in the Dakar sectors. Based on these ranges of values, coastal areas are categorized into four classes ranging from low to very high vulnerabilities to moderate and high. The VIC maps ( Figure 8 ) show the degree of coastal vulnerabilities along the sectors studied. The coastal linear areas (Km) are shown in Figure 9 . The linear area studied on the Joal Spit ( Figure 9(a) ) extends over 18.5 km, of which 33.08% to 40.54% has a high to very high degree of vulnerability, 16.64% of this linear is characterized by medium intensity. On the Djiffère spit ( Figure 9(b) ), the average intensity dominates with 36.37% compared to 20.45% to 34.29% from strong to very strong over a 24 km linear study. Over the 16 km studied on the Rufisque line ( Figure 9(c) ), from 16.25% to 26.87% is characterized by respective degrees of vulnerability from very high to high compared to 43.13% average and 13.75% low intensity. These summary maps of the IVC reveal that the high intensity characterizes above all the highly urbanized sectors, as well as the low-lying reliefs and sandy parts on the coasts from Djiffère to Rufisque. The coast studied in Dakar ( Figure 9(d) ) extends over a 42 km line, of which 4.28% to 13.57% is a strong to very high degree of vulnerability, 41.42% of this line is characterized by medium intensity. In addition, 35.95% of this linear area has a low degree of vulnerability. The same goes for rocky and cliff coasts (elevation greater than 10.0 m) which are better protected.

The variables in the CVI model can be selected based on the location and characteristics of the coasts. In the model developed here, six variables were processed by methods of geomatics, remote sensing and geology. Thus, their effectiveness in making coastal vulnerability measurable has been proven thanks to the rating system of 6 levels according to the intensity intervals of each variable that have heterogeneous units, by homogenizing their dimensions. The results of the variables through modeling and image processing were demonstrated and validated by in-situ observations. In addition, to analyse the level of exposure of socio-economic activities to the risk of erosion, the DVI model developed here has integrated the type of land use and the geology.