The study was conducted in extreme west part of Côte d’Ivoire, in the Tonkpi region, located between latitudes 7˚21' N and 7˚58' N and longitudes 7˚33' W and 8˚05' W ( Figure 1 ). This region covers an area of 12.3 km² and includes the departments of Man, Biankouma, Danané, Zouan-Hounien, and Sipilou. It is bordered to the west by Republics of Guinea and Liberia.

The study area is characterized by a humid tropical climate, known locally as the mountain climate. It is a cold climate with 2 seasons: a rainy season beginning in March and ending in October, and a dry season from November to February, with average annual rainfall of around 1800 mm. With regard to temperature, the annual average is 21˚C [11] . The relief of the Tonkpi region is subdivided into two morphological complexes of unequal extent: a mountainous complex (Toura and Dan ranges) and a plateau zone [12] . The mountainous terrain features several high peaks exceeding 1000 m, the highest of which are Mont Nimba (1753 m) and Mont Tonkpi (1189 m). The valleys lie between 300 and 700 m. This original ensemble owes much to the local variety of rocky material in the hypersthene granite family, but also to tectonic variations [13] . The geological environment consists exclusively of hypersthene granites and granito-gneisses with varied facies, derived both from the remains of an ancient basement and from major granitic intrusions [14] . The area is fairly well drained by various rivers, including the Cavally, Koué, Bafing and N’zo, tributaries of the Sassandra river.

The material consisted of Shuttle Radar Topographic Mission (SRTM) images data representing the digital elevation model (DEM), cartographic data (Man-Danané square degree geological map) and borehole data from data sheets for boreholes drilled in the study area. The SRTM images were used to extract the slope map and the drainage density map using Linwin.2.2 software.

The methodology adopted is based on multi-criteria analysis methods for mapping favorable areas for the siting of high-yield boreholes [15] - [18] and for estimating groundwater recharge [6] [19] [20] . The main stages of multicriteria analysis can be summarized in three: i) identification and elaboration of decision criteria; ii) classification and standardization of these criteria; iii) weighting of the criteria and their aggregation following the multicriteria approach.

In the fractured environment of the crystalline and crystallophyllous basement, the search for favorable zones for the installation of high-yield boreholes is based on the hydrogeological characteristics of the aquifers as well as the geological, physiographical and hydroclimatic aspects of the region [21] . The decision criteria for mapping water resources can be grouped into three quantitative indicators [15] [22] . These are: availability, accessibility and exploitability. These criteria are used to produce the thematic maps corresponding to the three indicators. The “availability” indicator refers to the existence of an aquifer and is the first condition to be known before any other activity . The availability map is produced from the combination of slope, effective infiltration, weathering thickness, drainage density and fracture density.

The slope map of the study area was generated in raster format from SRTM images with a resolution of 30 m.

Effective infiltration was obtained by calculating the water balance using the Thornthwaite method. The calculation method is illustrated in Equation (1):

$I_e=P-\left(AET\right)+R_s$ (1)

With,

I_e: Effective Infiltration (mm);

P: Precipitation (mm);

AET: Actual Evapotranspiration (mm);

R_s: Surface Runoff (mm).

Alteration thicknesses were extracted directly from drilling data sheets. Drainage density was determined from the hydrographic network extracted from SRTM images. Once the hydrographic network had been extracted, the area covered by this network was meshed by 5 km². The total number of meshes obtained was 1200. The drainage density was then calculated for each elementary surface. This was done using Linwin 2.2 software. Fracture density was obtained from fractures extracted from the work of [13] in the Man-Danané region. Like drainage density, fracture density was calculated in Linwin 2.2 after discretizing the study area into a 5 km² square mesh.

The “exploitability” indicator refers to the quantity of the groundwater reserve. It integrates the criteria of operating flow rate and static groundwater level. The exploitation rate depends on the quantity of water in the groundwater reserve, but also on the rate of renewal of this reserve in the event of high demand [16] . The two parameters of the exploitability indicator come from drilling data provided by the Hydraulics Regional Office of Man.

The “accessibility” indicator provides information on the conditions of access to the groundwater resource. These conditions are considered to be economic and social factors, as they promote or hinder access to the resource [18] . The main accessibility criteria are total depth and success index. Total depth is the most important criterion in the accessibility indicator, as it indicates the depth to be drilled to obtain the optimum flow rate. The total depth was provided by drilling data and the success index was determined according to Equation (2) considering that an exploitation yield of at least 10 m³/h is a large yield.

$S=\left(100*Q\right)/10$ (2)

With,

S: success index (%);

Q: operating yield (m³/h).

Classification involves grouping the values of each criterion into a number of classes. The boundaries chosen for these classes are not necessarily equidistant, but depend on the objectives set and the general context of the study . The choice of classes takes into account the variance of the available data and is inspired by the classification proposed by in the crystalline basement region. Thus, the classes Very Low, Low, Medium, High and Very High have been retained. In addition, as the criteria are measured on different scales, using different units, they need to be standardized for a good multi-criteria analysis. A common interval of 1 to 10 was selected for this operation, taking into account previous studies by in the Korhogo region and [24] in the Abidjan region, [17] in the Oumé region. The “very low” or “very heigh” classes are given a mark of 10, depending on whether they contribute to the excellent achievement of the indicator in question. If this is not the case, these classes are given a score of 1. Similarly, intermediate values are assigned to intermediate classes according to a linear distribution. The technique used to spatialize the criteria is interpolation, in particular the inverse distance weighting (IDW) interpolation method, and the areas obtained as a result of these operations have been categorized into five classes as shown in Table 1 .

The criteria were weighted using the pairwise comparison method developed by Saaty [25] [26] as part of the Analytical Hierarchy Process. It leads to standardized weighting coefficients whose sum is equal to 1. The weights used in this study are those obtained by [18] through the pairwise combination of criteria ( Table 2 ). Their study was conducted in the Denguélé District, north of the Tonkpi region. These two areas therefore have similar geological and hydrogeological contexts.

Once the weighting coefficients have been obtained, it becomes easier to combine the criteria. Thus, the thematic maps corresponding to slope, Effective Infiltration, drainage density, fracture density and weathering thickness were classified into four classes: Poor, Mediocre, Good and Excellent. These were then combined using layer addition (raster layer addition) to obtain the availability map. The process was repeated with total depth and success index to obtain the accessibility map, and with flow rate and static level to obtain the operability map. Finally, these three maps (availability, accessibility and exploitability) were combined to produce the map of potential areas for the siting of high-yield boreholes. All combinations were carried out in “Raster” mode, using the “raster calculator” tool.

The validation of the thematic maps consisted in calculating the uncertainties on the values of the parameters taken into account by each indicator. These uncertainties were determined on the averages of the various parameters as proposed by [18] and [27] translated by Equation (3).

$\Delta \bar{X}=\sigma /\sqrt{n}$ (3)

With,

$\Delta \bar{X}$: Uncertainty on the mean of data series,

$\sigma$: Standard deviation of the data series;

n: number of data sets.

In addition, an expansion factor K is calculated to determine the confidence interval. In fact, the K factor expressed by Equation (4), makes it possible to define an interval of sufficient scope having as its goal the certification of results [18] .

$K=\left|E-\Delta \bar{X}\right|/\sigma$ (3)

With,

E: extreme value of the data series, which can be the maximum or minimum of the series.

K = 1, indicates a confidence level of 68%.

K = 1, expresses a confidence level of 68%.

K = 2, indicates a confidence level of 95%.

The confidence level is 99% when K = 3.

Groundwater availability brings together the conditions required for the establishment of a probable aquifer. It is defined by the combination of slope, effective infiltration, drainage density, fracture density and alteration thickness. The purpose of the availability map is to guide the hydrogeologist in the design of water catchment structures, in order to obtain the best possible results. This map is made up of 4 classes distributed as follows ( Figure 2 ): The poor availability class occupies 43.77% of the region and the mediocre class covers 16%. These classes characterize areas with low fracturing density and high drainage density. They are not recommended for drilling, as a dense hydrographic network is unfavorable for groundwater recharge. Good and excellent availability classes represent 14.60% and 25.63% respectively. They characterize suitable areas for drilling. They reflect good groundwater recharge. Fracturing is intense and contributes to the effective infiltration of meteoric water.

The groundwater resource accessibility map was produced after combining the total depth and success index criteria. Figure 3 shows that poor and mediocre accessibility classes occupy more than half of the area (56.18%). They are located almost everywhere in the region ( Figure 3 ). These are areas characterized by very low flows and very high depths, ranging from 45.2 to 84 m. The good accessibility class represents 31.82% of the total area of the region and is characterized by structures of generally shallow depths and with a high success index (>50%). The excellent accessibility class represents 12% of the study area. The areas characterized by the good and excellent classes are more or less scattered throughout the region, particularly in the center.

The cross-referencing of the exploitation rate and the static level led to the elaboration of the exploitability map ( Figure 4 ). Analysis of the figure reveals four (4) exploitability classes, unevenly distributed across the study area. The poor class occupies 4.13% of the study area and is located in sectors where resources are generally not exploitable, although they are probably available. In fact, these areas are characterized by very low exploitation yields (<2 m³/h) and a slightly deep static level (>10 m). They represent the survival threshold for wells, and the largest pocket is located in the northern part of the region. The mediocre exploitability class is estimated at 11.2%. It expresses a relative difficulty in exploiting underground resources and is mainly found in the northern part of the study area. This class includes areas that are favorable to village hydraulics. The good exploitation class is characterized by medium and high yields accompanied by lower static levels (8 - 11 m) and occupies 35.15% of the study area. These localities are ideal for supplying drinking water to secondary centers and for urban water supply. Finally, areas of excellent exploitability account for 49.52% of the region. They are characterized by medium to high flow rates (>10 m³/h) and shallower static levels (<7 m). These zones are found to the west and east of the study area. They also constitute favorable areas for the supply of drinking water for urban hydraulics and secondary centers.

The map of potential zones was produced by combining the thematic maps of availability, accessibility and exploitability. Figure 5 shows four classes of zones suitable for borehole installation:

In short, in the Tonkpi region, more than half of the land (54.63%) is suitable high-yields boreholes.

The validation of the thematic maps obtained consisted in calculating the uncertainties of the parameters used. Analysis of Table 3 shows that the uncertainties of the parameters used to calculate groundwater availability are minimal (±0.06 to ±0.89), with the exception of groundwater recharge, which is ±54. This indicates a minimization of errors in the matrix calculations used to produce the availability map. However, the uncertainty on recharge is much higher, probably due to the low number of rainfall stations. On the other hand, the confidence levels are interesting, representing 95% for drainage density and 99% for groundwater recharge, weathering thickness and fracture density. In terms of accessibility parameters, the uncertainty is ±1.01 for the total depth and ±4.98 for the success index. Although these values are somewhat high, they are still acceptable, with respective confidence levels of 68% and 95%. With regard to the static level and yield used to assess groundwater operability, the errors are ±0.43 and ±0.95 for 99% confidence.

These uncertainty values and confidence levels show that errors were minimized in the creation of the various maps, thus demonstrating their validity.

Multicriteria analysis is a valuable technique for characterizing geospatial hydrogeological phenomena involving several criteria. Coupled with GIS, multicriteria analysis makes an undeniable contribution to the management of groundwater resources and guides decision-making in land-use planning. In fact, they provide a pre-prospecting tool that avoids the need for cumbersome, time-consuming and costly research phases [28] . In the Tonkpi region, these techniques have led to the production of thematic maps: availability, accessibility and exploitability on the one hand, and the map of potential areas for the installation of high-yield boreholes on the other. They have been successfully applied in several regions of Côte d’Ivoire to identify favorable areas, [15] in Korhogo in the north of the country, in Boundoukou in the east, [29] in the department of Oumé (Centre-west) and [18] in Odienné, in the northwest of Côte d’Ivoire. These techniques have also made it possible to select suitable sites for waste storage [26] and [24] . The potentiality map obtained shows that the Tonkpi region has good water potential, as do the regions of Korhogo [23] , Bondoukou , M’bahiakro [30] and Oumé [29] . Thus, the water shortages observed in the region could be attributed to poorly conducted prospecting studies (often based solely on geomorphology) and drilling installation work [9] .

Although multi-criteria analysis techniques are undeniable for their many advantages, they do have their limitations. In fact, parameter estimates often lack precision due to insufficient or totally missing data at certain points in the study area. In reality, the thematic maps produced only represent a relative assessment of the phenomena. However, these phenomena evolve over time and space. Multicriteria analysis is therefore error-prone, since it considers all the criteria involved in producing a thematic map to be perfectly comparable [21] . However, the technique of standardizing the criteria makes it easier to combine them. Several sources of error resulting from natural variations, numerical errors in the computer, the age of the data, etc. can somewhat alter the quality and reliability of the results. Despite all these shortcomings, it is interesting to note that GIS and multi-criteria analysis techniques offer results that are close enough to reality to facilitate decision-making in both groundwater research and exploitation.