Many works ( [3] [4] [15] [16] [17]) indicate that a number of factors would contribute to the definition of potential regional-scale aquifer recharge areas in the basement zone. These include: land use types, fracture density, aquifer permeability, slope, and drainage density.

1) Fracture density mapping

The elaboration of the fracture density map started with the production of the lineament map. This is done by satellite image processing. In this work, the Landsat 7 satellite scene 197-053 and 197-054 images acquired on February 02, 2018 were used. As the image was already geometrically rectified and georeferenced at acquisition, the treatments performed were mainly summarized in two steps (enhancement and validation). The enhancement of the discontinuities was done by applying the directional filters of Sobel. Their use allowed to manually extract the lineaments and to elaborate the detailed lineament map. This map was validated using the borehole flow rates of the two departments. The validation consisted in assessing the proximity of large flow boreholes to major lineaments. The validated fracture map was processed to determine quantifiable parameters such as fracture density. The production of the fracture density map is done with ArcGIS 10.4 software. Different zones are then identified according to the cumulative length of fractures on 5 × 5 km² grids.

2) Drainage density mapping

Drainage density is defined as the ratio of the total length of streams in a watershed to the area of the watershed [4]. The assessment of the hydrogeological properties of the drainage network was based on the characterization of density having a relationship with permeability. This map was developed by ArcGIS 10.4 software.

3) Land use map

The land cover map of the departments of Yamoussoukro and Toumodi was produced from a Landsat 7 satellite image acquired in 2018. The maximum likelihood algorithm was chosen for the supervised classification. This method is commonly used in the literature [18] [19]. This method was used to select five land use classes in the area. These are the class “forest”; “degraded forest”; “crop and fallow”, “habitat and bare soil” and “water”.

4) Slope mapping

The slopes are obtained from the DTM of Côte d’Ivoire from satellite images SRTM (Shuttle Radar Topography Mission) of 30 m resolution. The slopes calculated with ArcGIS 10.4 software are expressed in degrees. They were grouped into four classes.

5) Establishment of the permeability map or hydraulic conductivity of aquifers

The hydraulic conductivity was determined at the location of the boreholes that capture the fractures located in this area. This parameter expresses the capacity of a ground to let water pass through it. They were calculated considering the pumping tests. The deduction was made according to Equations (1) and (2).

T = Ke (1)

K = T / e (2)

with

K: hydraulic conductivity; T: transmissivity; e: thickness of the aquifer determined by the distance between the base and the top of the wells [20].

The reclassification step consisted in assigning scores to the different classes of the selected parameters according to their degree of infiltration by defining intervals. This process makes it possible to standardize the parameters by defining the degree of membership of each criterion in a common interval of 1 to 10. Indeed, the strongly unfavorable class always has the lowest score contrary to the strongly favorable class which has the highest score. However, the choice of scores depends on the influence of the class on infiltration. The result of the reclassification of the identified criteria, whose combination favors the good recharge or not of the groundwater is recorded in Table 1.

The decision criteria are weighted using the linear combination method based on the pairwise comparison technique according to the Analytical Hierarchy Process (AHP) of Saaty [13] improved and used by [4] [16]. It produces standardized weights whose procedure consists of comparing the relative importance of all elements belonging to the same level of the hierarchy taken two by two,

compared to the element of the level immediately above. This comparison was based on a numerical scale (Table 2).

Once all the combinations are made, it will be a question of deducing the eigenvector (Vp) (Equation (3)) and the weighting coefficient (Cp) (Equation (4)).

The principle of determination is as follows:

Vp = W 1 ⋯ Wk K (3)

With k = number of parameters compared; Wk = main values assigned to the variables;

Cp = Vp Vp 1 + ⋯ + Vpk (4)

The sum of the Cp of all the parameters of a square matrix must be equal to 1 (one). A high Cp, will tend to accentuate the recharge while a low Cp will reduce it. Table 3 summarizes the different binary combinations performed as well as the values of the corresponding eigenvectors and weighting coefficients.

In a study that relies on the subjectivity of values, it is imperative to determine the consistency of judgments [22]. Indeed, since the values assigned often present a certain degree of inconsistency, a synthesis of all the judgments proves wise. To verify the consistency of the judgments, [23] determined an assessment criterion which is the consistency ratio (CR). If this ratio is less than 10%, then the assessments are consistent; if it is more than 10%, they need to be revised. The verification principle is divided into several successive steps (Equations (5) to (11)):

· Normalization of the matrix [A]. It consists in dividing each element of a column by the sum of this column:

[ A ] = ∑ ​ Wn ∑ Wn (5)

K: alteration permeability, OS: land use, P: slope, DD: drainage density, Df: fracture density.

· Computation of the priority vector [C]. It consists in calculating the average of each line:

[ C ] = [ A ] × 1 n (6)

· Multiplying each note of the matrix by the corresponding priority vector allows to deduce the global priority [D].

[ D ] = ∑ Wn × [ C ] (7)

· Division of each global priority by its corresponding priority vector to find the rational priority [E].

[ E ] = [ D ] [ C ] (8)

· Calculation of the average of the rational priorities Ymax.

Ymax = [ E ] n (9)

· Calculation of the coherence index (IC).

IC = Ymax − n n − 1 (10)

· Deduction of the consistency coefficient (RC).

Ia = Ic Ia (11)

where (Ia) is the random index. The values of (Ia) are recorded in Table 4.

The computational method used in the GIS for overlaying the data is the operational approach of the single synthesis criterion of Roy in [22]. The principle is to sum the different layers according to their importance in groundwater recharge. The probability that the recharge in a zone is high is related to the values of the classes of parameters that intersect. In fact, it will depend on the index of the zones that favor recharge. Therefore, the calculation of the recharge index (Ir) is performed in GIS according to Equation (12).

Ir = 0.459 [ Df ] + 0. 264 [ K ] + 0.153 [ P ] + 0.080 [ OS ] + 0.0 43 [ DD ] (12)

where Ir: is the sum of the weighted scores calculated for each pixel of the raster from the score of the five parameters.

The probability of recharge is related to the values of the classes of the crossed parameters. This depends on the index of recharge-promoting areas. Thus, [24] defined a method (Equation (13)) to help interpret the classes to which the

recharge class values relate.

Irp = ( P ( Irmax − Irmin ) ) / 100 + Irmin (13)

With Irp = recharge index in percentage; P = corresponds to a percentage class; Imax = maximum recharge index; Irmin = minimum recharge index.

Following this same logic, the Ir recharge index values were interpreted (Equation (14)).

Since the weighting coefficient is fixed, the minimum rating that a parameter can have is 1 and the maximum rating is 10. Substituting the different values of maximum and minimum rating, the equation becomes (Equation (14)):

Ir = ( P × 6 . 2 ) / 100 + 1 (14)

All the operations performed to arrive at the mapping of potential recharge areas are grouped in the conceptual data model (Figure 2).

The detailed lineament field includes all lineaments surveyed (in total, 1718)

during the manual extraction phase. The cumulative length of the lineaments is equal to 14.694 Km with an average of 2.6 Km over an area of 4.789 Km² (Figure 3).

L’exploitation de la carte détaillée des linéaments permet d’élaborer de nombreux fichiers thématiques à partir desquels sont définies et caractérisée la géométrie des aquifers souterrains (densité de fracturation, distribution des longueurs et des orientations des linéaments …).

In order to better observe the spatial distribution of fracture density according to the number of fractures per unit area, the iso value map of distribution of this parameter was constructed (Figure 4).

The density is variable over the whole area. The fractures are therefore distributed differently over the entire study area. A global analysis of this map shows that the departments of Yamoussoukro and Toumodi are strongly fractured and therefore favourable to infiltration. Indeed, the medium, strong and very strong classes represent about 70.5% of the study area. On the other hand, areas of low fracture density occupy about 29.5% of the study area. Nevertheless, it should be noted that the zones of very high fracturing density occupy about 8.5% of this area and seem to be distributed over the whole area.

Data interpolation showed that almost the entire area is covered by hydraulic conductivities with values between 1.06 × 10⁻⁷ and 9.87 × 10⁻⁴ m/s (Figure 5).

More than 80% of the area is covered by low conductivity. However, there are high values in the south in the Kpouebo sub-prefecture and a few sites scattered throughout the rest of the area.

The relief of the Yamoussoukro and Toumodi departments is relatively hilly. Structuring into four classes indicates that the spatial coverage is characterized by 33.5% low slope; 48.7% medium slope; 12.5% steep slope and 6.3% very steep slope (Figure 6). The steepest slopes are found in the sub-prefectures of Attiégouakro, Angonda and Kpouèbo due to the existence of mountain ranges.

The spatial distribution map of land use shows six (6) classes (Figure 7). The class Habitats and Bare Soil represents 5.7% of the study area, the forest islands constitute 26.1%, the degraded forest represents 23.7%, in addition the crops and fallow land occupy 43.6% while water represents 0.9%.

The drainage density map shows (4) four surface classes (Figure 8). The low density class is 22.1%, the medium density class is 35.3% of the study area, the high density class is 25.4% and the very high density class is 17.2% of the area.

Table 5 presents the results of the judgment audit.

With an RC equal to 5.8%, the prioritization matrix presented (Table 5) is

considered consistent. Therefore, the assigned judgments can be used for the determination of recharge areas in Yamoussoukro and Toumodi departments.

Overlaying the different thematic maps obtained in a GIS made it possible to establish a map of groundwater recharge indices for the departments of Yamoussoukro and Toumodi (Figure 9), and from this map, the map of potential recharge areas.

The recharge index values range from 1 to 7.2. They highlight the areas where the degree of recharge is significant. Equation 17 yielded five recharge classes that range from “Poor” to “Excellent”. The values of these indices are recorded in Table 6.

The map of potential recharge areas is presented in Figure 10. Analysis of this map reveals five (5) classes (Bad, Poor, Good, Very good, Excellent) of potential groundwater recharge areas in the departments of Yamoussoukro and Toumodi. The “Poor” zones (26.2%) are concentrated in the sub-prefectures of Djékanou, Lolobo and Angonda. As for the “poor” class zones (30.69%), they are found throughout the zone. On the other hand, the “good”, “very good” and “excellent” classes represent 43.11% of the entire study area. However, the “Excellent” class covers 5.46% and seems to be scattered throughout the area.