Table 1. Weight assignment.

Table 2. Saaty weighted task importance scale.

Table 3. Binary comparison matrix.

The binary comparison matrix is a preferential intensity assignment (Table 2) of two parameters. This pair-by-pair comparison of layers is assigned an associated value showing the importance of the layer in relation to the other on the outcome of the phenomenon.

The elements of the judgment matrix are obtained by dividing the value of each cell (Table 3) by the sum of the values corresponding to the column of that cell.

To obtain the weights of the factors, the sum per line (Table 4) corresponding to each factor was divided by the total sum of the lines.

The consistency of judgments was assessed by multiplying the associated value of the binary comparison matrix (Table 3) column by the weight of the corresponding factor.

Consistency was calculated by dividing each sum of the rows in the judgement consistency score (Table 6) by the weight corresponding to the criterion.

Table 4. Result of judgment matrix.

Table 5. Results of factor weights.

Table 6. Results of the assessment consistency of judgments.

Table 7. Consistence.

Table 8. Random index.

Consistency was calculated by dividing each sum of the rows in the judgement consistency score (Table 6) by the weight corresponding to the criterion.

We can then determine:

With: γ m a x : max value for 4 criteria; n: number of factors; IC: consistency index; IA: random index linked to the number of factors; RC: consistency ratio.

The matrix is sufficiently coherent as RC < 0.1

A soil pit was dug manually, a profile described and soil samples collected. The horizon’s thickness was measured with a tape, and the colour was determined using the Munsell soil colour chart. Samples are collected and stored in plastic bags previously labelled. Accessible past landslides were described, and geographical coordinates were recorded using Garmin 73 GPS navigation devices. Morphometric measurements were also taken using a double decameter and a clinometer. In the laboratory, granulometry was carried out using the Robinson pipette method after dispersion with sodium hexametaphosphate.

The analysis results are presented in Table 9.

Table 9. Physical characteristics of Foréké soils.

The profile made in the Foréké village shows from top to bottom:

0 - 20 cm: A dark-brown humus horizon (10YR4/3), clay-loam, friable, plastic, lumpy structure and dense root.

200 - 135 cm: A reddish-brown horizon (7.5YR5/6), clayey, porous, plastic, coarse polyhedral structure.

135 - 530 cm: Red alloterite (2.5R3/8), silty-clayey, porous, small weakly indurated rock fragments representing about 30% of the horizon.

530 - 750 cm: Silty isalterite with juxtaposition of yellowish (2.5Y5/3), reddish (5R3/6) and whitish-gray (10YR8/3) domains, presence of rock fragments.

750 to more than 1200 cm: Whitish-grey isalterite (10YR8/3), sandy-loamy, quartz grains up 40% of the horizon, many rock fragments.

4.2.1. Characterization of the Double Landslide on the Foréké Escarpment

The double landslide of November 05, 2024, is a major landslide that occurred at the “Nteh-Gwang” steep hill (56˚). The crater is subcircular, with a diameter of over 70 m and a depth of 40 m. Around 130,000 m³ of material was mobilized, and the toe is over 350 m (Figure 2Figure 2 and Figure 3Figure 3).

Figure 2. Characteristics of the first landslide.

Figure 3. Starting of the second landslide.

4.2.2. Landslide Hazard along the Dschang_Santchou Road

The slope inclination on the site of November 05, 2024, double landslide reaches 296% (Figure 4(a)), the profile of curvature reaches +0.03 (Figure 4(b)),

(a)

(b)

(c)

(d)

Figure 4. (a) Slope; (b) Profile of curvature; (c) Terrain ruggedness; (d) Terrain ruggedness.

the terrain ruggedness index reaches 25 (Figure 4(c)) and the altitude above the channel reaches +153 m (Figure 4(d)). The Ménoua River, the main collector in the study area, is located at 1031 m from the “Nteh-Gwang” hill. About 35% of the Dschang_Santchou road presents a high hazard (Figure 5Figure 5).

Figure 5. Risk map along the Foréké escarpement. (Source: ArcGIS 10.8 (ESRI) software, from the topographic base sheets of Cameroon produced by the National Institute of Cartography)

In the West Cameroon region, the slope inclination varies from 0 to 692% with a maximum along the escarpments bordering the highlands (Figure 6(a)). The profile of curvature varies from −0.09 to +0.08, with the minimum value on the Bambouto volcanic massif (Figure 6(b)). The terrain ruggedness index varies from 0 to 64, with a maximum along the escarpments bordering the highlands (Figure 6(c)). The altitude above the channel varies from −275 to +891 m with a maximum along the escarpments bordering the highlands (Figure 6(d)). High-hazard zones represent 6.40% of the Western Region, medium-hazard zones 8.47% and low-hazard zones 25.14% (Figure 7Figure 7). High-hazard zones are located in the highlands (between 1200 and 1800 m).

All control cases (Ngouache, Magha’a, Fongo-Tongo, Dschang, Fossong-Wentche, Santchou, Kekem) are located perfectly in high-hazard zones (Figure 8Figure 8).

(a)

(b)

(c)

(d)

Figure 6. (a) Slope; (b) Profile of curvature; (c) Terrain ruggedness; (d) Terrain ruggedness. (Source: ArcGIS 10.8 (ESRI) software, from the topographic base sheets of Cameroon produced by the National Institute of Cartography).

Figure 7. Risk map along the Foréké escarpement. (Source: ArcGIS 10.8 (ESRI) software, from the topographic base sheets of Cameroon produced by the National Institute of Cartography).

Figure 8. Past landslides map in the West Cameroon region. (Source: ArcGIS 10.8 (ESRI) software, from the topographic base sheets of Cameroon produced by the National Institute of Cartography).

Hazard defines natural conditions that combine to create a danger [35]-[37]. The hazard map is produced using Saaty’s statistical spatial analysis model. In the literature, the selection of distribution center locations is based on multi-objective decision support metaheuristics (MMODM), multi-objective combinatorial optimization methods (MMOCO) and Multi-Criteria Decision support Methods (MCDM) [38]. MMODM methods optimize the situation after converting it into a single-objective problem, but they are very costly [39]. MMOCO methods solve discrete multicriteria problems for which the alternatives are not explicitly known [40]. The MCDM method evaluates alternatives and scenarios in order to select the most appropriate choice according to the objectives and results sought [41]. They fall into two categories, namely Multi-Objective Decision Support Methods (MODM) and Multi-Attribute Decision Support Methods (MADM). Unlike MADM, MODMs do not include quantitative criteria in the decision-making process [42]. However, for MADM, several approaches are not suitable for mapping natural hazards. For example, the Weight Sum Method (WSM) is recommended for one-dimensional problems, the Weighted Product Method (WPM) is dimensionless, and the TOPSIS technique (Technique for Order of Preference by Ideal Solution Similarity) is only suitable for decision-making for certain criteria [43]. ELECTRE multi-criteria methods take into account quantitative and qualitative criteria, the importance of criteria, and validate the solutions retained by simultaneous tests of agreement and disagreement, but they do not take into account interactions between criteria [44].

To overcome these limitations and meet the requirements of landslide hazard mapping in humid tropical mountain areas, we opted for Thomas Saaty’s hierarchical multicriteria analysis method. Based on the comparison of pairs of options and criteria, it offers the advantages of hierarchical structuring (classes—criteria—weights), priority structuring (sub-criteria—ranks), and logical consistency [45]. It therefore leads to a justified choice, and the decision is then said to be rational, systematic and correctly made [46]. In terms of combining variables and formalizing expert rules, this method is the most comprehensive [47]-[50]. It retains the flexibility of the geomorphological approach while being more objective [51][52]. It is widely used because of its generalizability and reproducibility [25][53]. Previous landslides served as the basis for validation of the hazard map for the West Cameroon region. This field validation means that the model can be considered reliable [54][55]. An inventory of past events is a prerequisite for any hazard assessment, whatever the scale of analysis and approach adopted [56][57]. This is the principle of cross-validation, which uses non-overlapping spatial subsets of the mapped landslide to assess the accuracy of the prediction.

Availability of high-resolution DTMs offers landslide cartographers the possibility of accurately detecting landslide-prone areas [55][58]. DEM enables precise visual analyses or automatic processing to facilitate certain geomorphological interpretations. In landslide studies, weightings can be transposed from one site to another if the geomorphological context is similar and if the input data are exactly the same in terms of resolution and variable classes [59]. Based on the natural factors predisposing to landslides on the Foréké escarpment, slope appears as the most important because it influences from 44% to 77% of landslides in humid tropical mountainous regions [60]. Soil loss increases exponentially with slope steepness [61]. Profile curvature is the second important factor, expressing the rate of change of the slope gradient and variations in flow velocity along the slope. A negative value signifies the surface is upwardly convex, while a positive value indicates it is upwardly concave, and a value of zero means the surface is linear [62]. On the Foréké escarpment, it varies between −0.03 and +0.03, characteristic of rugged relief [63]. According to [61], a convex slope tends to concentrate rainwater and runoff at its top, increasing the speed and volume of water and promoting the transport of materials, while a concave slope reduces material transport [63]. Terrain ruggedness index is the third factor. It measures the topographical complexity and increases as the terrain becomes more rugged, as observed on the study site. The altitude above the canal is the last factor. The higher altitude increases the force of gravity acting on the ground, increasing the risk of landslides.

The 15 km Santchou_Dschang road has a vertical drop of 737 m. The double landslide is the result of an imbalance created by the construction of a road on this very hilly site [64]-[68]. It has been demonstrated that, in the case of low mobilizable shear strength [69]-[72], several landslides can occur on the same site in clay soil [73][74]. This finding is supported by the clay content, which varies between 69.35 and 82.69% in Foréké soils. On similar soils in the nearby locality of Kékem, [12] describe a very clayey material with high porosity (>29%), low cohesion (<0.5 bar) and a high angle of friction (15˚ - 22˚). The high clay content therefore significantly influenced the residual friction angle by decreasing the residual strength of the soils [75]-[77]. Creep movements have thus evolved into progressive landslides [74][78]. This process is exacerbated by water infiltration, which accelerates the reduction in shear strength [79] due to the seasonal transition of the soil from a compact to a plastic state [80]. As a result, microcracks opened up in the clay material, altering cohesion [12][64] and accelerating a landslide by creep. The impact of runoff is amplified by deforestation, which reduces rainwater infiltration [75]-[77][81]. It should be noted that the conditions that prevailed at the end of the first landslide still prevail at the end of the second landslide, and further landslides are likely to occur on this hill (Figure 9Figure 9).

5.3.1. Anthropogenic Constraints

The traffic on Dschang_Santchou is heavy, and the low speed of machinery impacts the amplitude of vibrations. These vibrations, caused in particular by trucks,

Figure 9. Schematic diagram of the process that led to the double landslide on the Foréké escarpment.

give rise to stress waves that propagate through the ground [82]. In addition, during construction of the P17 road, unsterilized cuttings embankments were built at mid-slope on very steep hillsides. [83] note that in such cases, due to slow and complex variations in interstitial overpressure, the most critical stability conditions may only appear in the long term. Indeed, the stiffening of the slope by the excavated material on its flank contributes significantly to increasing driving forces [84]. The second landslide, which occurred a few hours after the first, is a perfect illustration of anthropogenic action in triggering disasters [85][86].

5.3.2. Natural Constraints

Landslides are very common in mountainous areas [37][87][88], on hills [12][21] and along the high coasts [89]. The West Cameroon region is naturally predisposed to landslide risk due to its very rugged relief [27], the nature of its soils and its high rainfall [26]. The plains are linked to the plateau by escarpments [31][33] such as Kékem and Foréké [12][32], on which very clayey soils have developed [90][91]. The CU triaxial shear performed by [12] on similar soils in the near locality of Kékem shows variations between 4 and 12 for apparent internal friction angle (φcu), 15 and 22 for effective internal friction angle (φ’cu), 0.15 and 0.50 bar for apparent cohesion (Ccu) and 0.10 and 0.38 bar for effective cohesion (C’cu). With low cohesion and a high angle of internal friction, the geotechnical characteristics are unfavorable. Similarly, in the Kekem locality, [12] shows that the soil is saturated with water (80%), taking the material beyond the Atterberg limits (Wp = 28% - 39% and Wi = 52% - 81%).

5.3.3. Mitigation of Landslide Risk in the West Cameroon Region

In the West Cameroon region, landslides generally occur in July, August and September [28]. For the landslide of August 26, 1978, at Santchou, out of a total of 389 mm of water recorded during the month of August 1978, 88 mm of water were recorded on August 25, 1978, and 71 mm on August 26, 1978 [17]. Similar observations were made during the July 20, 2003 disaster in the Bambouto massif, where for a monthly average of 280 mm of water in July 2003, 84 mm of water were recorded during the two days preceding the landslide [19]. More recently, in the town of Bafoussam, 250 mm of rain fell continuously during the two days preceding the landslide on October 28, 2019 [92]. Monitoring rainfall data will alert people to potential landslide risks, and closing the Dschang-Santchou road to traffic during critical periods would be a good mitigation measure. In addition, to prevent landslides along the Dschang-Santchou line, all embankments over 2 meters high with a slope greater than 40% must be stabilized, as recommended by [93]. Combating creep landslides requires a comprehensive approach combining prevention and ground stabilization measures. Regular monitoring helps to anticipate landslides and assess the danger. Retaining walls or the insertion of steel bars in the ground are recommended along road embankments. The installation of drain sub-drains is necessary to evacuate water from the ground. Revegetation of embankments with deep-rooted plants is recommended to stabilize the soil and reduce water erosion. In the West Cameroon highlands, it is essential to carry out geotechnical studies before building any structures.