The data used for this study were drawn from several complementary sources. Satellite data include a Sentinel-2 image from 2023 for land-use mapping and SRTM data at 30-metre resolution for generating the Digital Elevation Model (DEM). Climate data were obtained from the Société d’Exploitation et de Développement Aéroportuaire, Aéronautique et Météorologique (SODEXAM) for the period 1985-2025. Demographic data were taken from the General Population and Housing Census, at the neighbourhood level in San Pedro.

The identification of criteria is based on previous studies focusing on San Pedro. These include physical environmental constraints as well as human activities that expose the town to coastal hazards. The criteria selected are: slopes, land use, drainage density, population density, soil type, precipitation and elevation.

A GIS was used to produce thematic vulnerability maps based on the criteria mentioned above. These maps represent the spatial indicators of coastal vulnerability within the study area.

By applying the AHP, the research aimed to identify vulnerable areas along the San Pedro coastline. The weighting process involves comparing each criterion in pairs and assigning a weight to each one.

Using thematic maps and the AHP methodology, a coastal vulnerability assessment was carried out, covering both physical and human factors in San Pedro. This section aims to select and analyse the criteria in order to identify vulnerable areas along the San Pedro coastline. The methodology utilises various datasets, remote sensing tools, GIS techniques and the AHP method to provide meaningful information on the coastal vulnerability of the study area, as shown in Figure 2.

In this study, the hierarchical multi-criteria analysis method devised by the mathematician [17] Saaty (1980) is used. AHP is a systematic technique commonly used in complex decision-making contexts, which enables difficult decision-making processes to be broken down into comprehensible elements. It follows the steps set out below.

Step 1: Define the decision-making problem

The aim is to analyse and map the coastal vulnerability of San Pedro by combining seven physical and human criteria.

Step 2: Create a hierarchy

A hierarchical structure is created, representing the main objective, the coastal vulnerability map, the seven selected criteria and the five vulnerability classes associated with each.

Step 3: Pairwise comparisons

The criteria are compared in pairs using value scale [17]. Decision-makers assess the elements on a scale of 1 to 9, where 1 represents equal importance between two elements and 9 represents the absolute importance of one element over another.

Step 4: Calculating weights

The weights of the criteria and sub-criteria are calculated on the basis of the pairwise comparisons. These weights represent the relative importance of each element within the hierarchy. The weighting coefficient (Cp) is calculated as follows:

∑ V p is the sum of all eigenvectors. The sum of the Cp values must equal 1. Table 1 presents the value scale used for the pairwise comparisons.

Figure 2. Flowchart of methodology.

Table 1.Saaty’s value scale.

Step 5: Checking consistency

The consistency index is used to check whether the judgements are logical, i.e., whether the pairwise comparisons are reliable and do not contradict one another. A widely used measure is the consistency ratio (CR). If the CR exceeds a certain threshold (usually 0.1 or 10%), the judgements may need to be reviewed. The formulas are as follows:

where λ max is the largest eigenvalue of the pairwise comparison matrix, k is the number of criteria and Ia is the random consistency index. Table 2 shows the values of Ia as a function of the number of criteria.

Table 2.Values of the random consistency index (Ia) [17].

Step 6: Aggregation and vulnerability assessment

Aggregation involves combining the criterion weights with the vulnerability scores for each thematic layer to produce the composite coastal vulnerability map.

Step 7: Developing the coastal vulnerability map

The criterion weights are used to create the final coastal vulnerability map, which assists managers and decision-makers in their coastal risk planning and management processes.

Slope is essential for mapping flood vulnerability [21][22] as it determines the trajectory and velocity of water. It is slope that transforms rainfall into rapid runoff on slopes or into dangerous stagnation in low-lying areas [23]. Areas with a gentle slope, typically found in low-lying floodplains, are more susceptible to flooding and are therefore assigned higher scores, whilst steeper slopes are less prone to flooding and receive lower scores.

Slope is classified into five categories using a reclassification tool (Figure 3).

The “very gentle” class corresponds to a gradient of between 0˚ and 5˚, whilst the “very steep” class extends to over 30˚. Gradient is a key factor in understanding the direction of water flow from upstream to downstream areas [24].

The identification of vulnerable areas takes into account factors such as land cover and land use. Land cover and land use are essential for understanding the interactions between human activities and the natural environment. Rainwater runoff is increased when it falls on impervious surfaces such as residential areas and roads, whilst bare fields can lead to soil erosion and significant downstream runoff. Conversely, areas with dense vegetation pose fewer risks, as dense vegetation promotes the infiltration of rainwater into the soil, thereby reducing the impact of flooding compared with open land. The study area comprises various types of land cover, such as forest, cropland, settlements, bare ground and water bodies. Figure 3 shows the land-use map of the area, which is divided into five classes. Land-use and land-cover patterns have a significant influence on the frequency of flooding; urbanised and developed areas generate persistent runoff that takes time to dissipate, making them more vulnerable to flooding than areas with bare soil or vegetation cover. The “settlement” class is the most dominant, covering 27.61 km², or 51.05 per cent of the total area. This is followed by “crops”, which cover 15.36 km² (28.40 per cent). Forest accounts for 8.31 per cent of the study area (4.49 km²). The “water” and “bare soil” classes account for 6.73 per cent (3.64 km²) and 5.51 per cent (2.98 km²) respectively.

Precipitation is a key factor that contributes significantly to flood vulnerability. Floods occur when increased surface runoff, caused by heavy rainfall, exceeds the capacity of watercourses to discharge the excess water [25].

Precipitation in the study area is unevenly distributed (Figure 3). Average annual precipitation in the town of San Pedro ranges from 1774.66 to 1812.57 mm and is categorised into five classes. Precipitation is classified as very low (1774.66 - 1787.84 mm), low (1787.84 - 1792.95 mm), average (1792.95 - 1797.79 mm), high (1797.79 - 1802.62 mm) and very high (1802.62 - 1812.57 mm). The areas with the highest rainfall levels exhibit the greatest vulnerabilities. Areas with very low vulnerability are those where rainfall is very low and include the districts of Corniche, Nitoro, Mohikakro and Rade. Areas of low vulnerability include Balmer, Lac Sonouko and Soweke. Areas of medium vulnerability include Zimbabwe, Bardot and part of Sotref. Areas of high vulnerability include the Colas neighbourhood, part of Sotref, Victor Ballet and Soleil. Areas of very high vulnerability are located in the north-west of the study area.

The figure shows the map of reclassified elevations for the study area (Figure 3). Higher elevations are less vulnerable, whilst lower elevations are more vulnerable to flooding. Elevations in the study area are divided into five classes: very low (5 - 17 m), low (17 - 38 m), medium (38 - 60 m), high (60 - 81 m) and very high (81 - 103 m). The town of San Pedro is thus characterised by a relatively flat terrain with low elevations. Areas of very low vulnerability cover 0.18 per cent, or 0.10 km², and are located in Mohikakro and Rade. Areas of low vulnerability cover 2.83 per cent, or 0.47 km², and are located in Mohikakro. Areas of medium vulnerability cover 0.87 per cent, or 1.53 km², and include Cathedral 1 and Victor Ballet. Areas of high vulnerability cover 21 per cent of the study area, or 11.37 km², and are located near the cathedral. Areas of very high vulnerability account for 75.12 per cent, or 40.66 km², and include the neighbourhoods of Colas, Soweke, Bardot, Zimbabwe Lac Sonouko and Nabo.

(a) (b)

(c) (d)

(e) (f)

(g)

Figure 3. Spatial distribution of the flood-vulnerability factors used in the analysis: (a) Slope, (b) LULC, (c) Rainfall, (d) Elevation, (e) Drainage density, (f) Population density and (g) Soil types.

The density of the drainage network is a key Hydromorphological parameter for modelling and mapping coastal vulnerability to flooding. The establishment and evolution of a drainage network depend on multiple physical variables, notably topography, the lithology of the substrate (permeability and porosity), and the configuration of local and regional fracture systems. Areas with low soil permeability tend to have high drainage density, leading to increased runoff of precipitation, and vice versa [26]. The drainage density map of the town of San Pedro, as shown in the Figure 3, is classified into five categories: very low (0 - 1.34 km/km²), low density (1.34 - 2.68 km/km²), medium density (2.68 - 4.01 km/km²), high (4.01 - 5.35 km/km²) and very high (5.35 - 6.69 km/km²). The town of San Pedro is predominantly characterised by areas of very low to low drainage density.

The higher the density, the greater the number of people vulnerable to flooding, leading to more significant socio-economic losses, including damage to infrastructure, businesses and homes. Urban sprawl destroys wetlands, forests and green spaces that naturally absorb excess water. Figure 3 shows the population density map of the town of San Pedro. Based on these values, the classification into five distinct categories is as follows: very low, low, medium, high and very high. The highest weighting, assigned a value of 5, is given to the category ranging from 132.67 to 165.83 inhabitants per km², which corresponds to the highest flood risk. Conversely, the category ranging from 33.18 to 66.35 inhabitants per km², which corresponds to the lowest flood risk, is assigned the lowest weighting.

Water absorption capacity varies according to soil type [27], making infiltration a key factor in storm water management. Infiltration directly regulates surface runoff during rainfall. The soil’s capacity to absorb water depends primarily on its porosity, grain size, structure, aggregation and organic matter content. The soil map of the study area shown in Figure 3 reveals a predominance of ferralitic soils. These soils are characteristic of humid tropical regions and result from intense weathering of rocks under the influence of heavy rainfall and high temperatures. They are generally deep, well-drained and exhibit low cohesion when subjected to severe water erosion.

The coastal vulnerability assessment map for San Pedro was created by incorporating factors contributing to flooding, such as slope, rainfall, density of the river network, population density, elevation, land use/land cover and soil type. A decision-making approach based on the Analytic Hierarchy Process (AHP) and a Geographic Information System (GIS) was employed, with weights assigned to these parameters based on expert opinion and previous studies. To assess the vulnerability of coastlines to flooding, a pairwise comparison matrix was constructed using the AHP method to evaluate the relative importance of the selected socio-economic and infrastructural indicators (Table 3). The matrix was then normalised by dividing each element by the sum of its column, thereby providing a consistent scale of comparison (Table 4). Subsequently, Table 5 shows the weights for each indicator, which were calculated by averaging the values in each row of the normalised matrix. A consistency check was carried out by calculating the consistency ratio (CR), which was found to be 0.047, thereby confirming that the estimates fell within the acceptable consistency range (CR ≤ 0.10).

RC = 0.047 = 4.70%, so RC < 10%; therefore, the matrix is consistent.

The final weights were integrated into a GIS environment, where each parameter layer was spatially processed and reclassified. A weighted overlay analysis was then carried out to produce the flood vulnerability assessment map (Figure 4), classifying the study area into zones of very low vulnerability, low vulnerability, medium vulnerability, high vulnerability and very high vulnerability. The weighted overlay method standardised and combined these factors into a unified model using the equation below.

Vulnerability map = 0.145 × slope + 0.270 × land use + 0.048 × drainage density + 0.329 × population density + 0.070 × soil type + 0.028 × precipitation + 0.112 × altitude.

Analysis of Table 6 shows that the study area includes zones classified as having very low flood vulnerability, which cover 11.91 per cent, or 6.43 km², and are scattered across the area, notably in Rade and to the south-west of the port area. Areas with low vulnerability account for 20.94 per cent, or 11.31 km², and are situated to the north of Sotref and Château, as well as in Mohikakro. Areas with medium vulnerability cover 22.97 per cent, or 12.40 km², and occupy part of Soleil and Château. High-vulnerability areas account for 21.91 per cent, or 11.83 km², and include Corniche and Cathedral 1. Very high-vulnerability areas cover 22.27 per cent, or 12.03 km², and include the neighbourhoods of Bardot, Colas, Victor Ballet, Citi, Lac Sonouko, Nabo and part of Sotref. The most vulnerable areas are those with the highest population density. These classifications were determined on the basis of expert assessments of the contributing factors.

Table 3. Pairwise comparison matrix of factors contributing to coastal vulnerability to flooding.

Table 4. Normalised pairwise comparison matrix of factors contributing to flood vulnerability.

Table 5. Calculation of eigenvectors and weights for each vulnerability parameter relating to the consistency of factors contributing to flood vulnerability.

Table 6. Distribution of coastal vulnerability zones in San Pedro.

In total, 44.18 per cent of San Pedro’s area is classified as having high to very high vulnerability. These areas correspond to the most densely populated areas, the lowest elevations and land uses dominated by residential areas. Conversely, only 11.91 per cent of the territory is classified as having very low vulnerability, corresponding to preserved natural areas and sparsely urbanised zones. Local authorities should monitor these areas continuously, particularly during the rainy seasons, in order to prevent disasters linked to flooding and coastal inundation. The geospatial model based on the AHP method developed in this study can help policy-makers identify and prioritise high-risk areas, thereby enabling targeted interventions, optimised resource allocation and effective spatial planning. It can also facilitate the integration of flood risks into local development plans, infrastructure projects and climate change adaptation strategies.

Figure 4.Flood vulnerability map of San Pedro.