The direct current (DC) electrical approach is typically used for aquifer exploration. On the resistivity maps, these aquifers show up as low resistivities and low chargeability anomalies [1]. The 3D models of the aquifers in the Messondo area are highlighted and characterized using these two techniques in conjunction with remote sensing.

Remote sensing involves the set of methods used to record electromagnetic radiation emitted or reflected by geolocatable bodies, in the form of images. In this work, we employ lineament mapping using Landsat 8-OLI scene (Path/Row 185/057) acquired on January 12, 2022. [18] defines lineaments as geomorphological structures inherited from previous topographies, or alignments of adjacent geological objects. The linear structures were improved on Landsat images, using SOBEL directional filters in Envi software, then manually extracted in ArcGIS software based on visual analysis as done by [19]. For reproducibility, the Landsat 8 OLI scene was converted to TOA reflectance. Bands 5 (NIR), 6 and 7 (SWIR) were selected for their sensitivity to moisture, iron oxides and clays [19] and combined into a false-color composite (R = 7, G = 6, B = 5). SOBEL directional filters (0˚,4 5˚, 90˚, 135˚) were applied in ENVI, followed by manual digitization in ArcGIS at 1:10,000 scale. Only lineaments greater than 50 m with alignment beyond 3 pixels were retained. The 50 m threshold balances regional fracture detection and rejection of anthropogenic features (roads, fields), consistent with regional studies [18][19]. Shorter features, roads and rivers (verified on Google Earth) were rejected.

The electrical characteristics of the soil are distributed both horizontally and vertically using electrical tomography. By sending an electric current through two electrodes in the soil and measuring the potentials that occur using a second pair of electrodes known as potential electrodes, it determines the resistivity of the ground [20]. The underlying material’s apparent resistivity (ρa) depends on the electrode geometry (K), recorded potential difference (V_MN), and current intensity (I_AB) [21]-[23]. It is provided by the relationship:

The grain matrix and water content regulate the aquifer’s resistivity [24]. Fracturing, saturation level, weathering, and porosity are among of the variables that affect the rock’s resistivity [25]. Resistivity often decreases as water quality and grain size increase.

By determining the earth material’s chargeability, the induced polarization approach outperforms electrical methods [26]. According to [1][27]-[29] and references therein, the chargeability (M) depends on a primary voltage (V_MN) and the secondary voltage (Vs) measured respectively immediately before and after the current stoppage at a date t.

After the current interruption, the secondary voltage can be measured once, but for a much more accurate measurement, it is preferable to integrate it over time [29]. Then, the chargeability is:

Comparing chargeability results between instruments is typically made possible by applying time-normalization [22][29].

The ERT/IP data were acquired with the Syscal Junior 48 resistivity meter along four SE-NW Schlumberger profiles of individual length 1.6 km, supposedly perpendicular to the structures of the targeted formations, in order to prevent differences caused by anisotropies in the geological formations [26]. Data were collected in a 100 m × 100 m grid (Figure 4) with 48 electrodes spaced at approximately 33 m.

Figure 4. Geoelectric profiles on a background showing the relief of the area.

The geometric factor (n) was increased from 1 to 8, providing 8 levels of investigation. Acquisition time was approximately 45 minutes per profile. The mean effective investigation depth was 140 m, limited by the resistive gneissic basement. The geoelectrical data are associated to depths relative to the distance between the current electrodes, can be interpreted qualitatively and quantitatively in terms of a hydrological model. A qualitative interpretation of the formations investigated was done from 2D pseudo-sections obtained by data inversion in Res2DInv software while adjusting the topography of the area. In practice, both Rho and IP inversions were performed with Res2Dinv v4.0 using a robust (L1-norm) scheme. After four iterations, the final absolute error was 1.5 for all profiles; noisy data were manually removed, and topography was included. 3D models were developed in Oasis Montaj software, to facilitate the visualization and quantification of the highlighted aquifers.

Structural analysis from remote sensing revealed more than 100 lineaments with lengths between 0.05 and 1 km (Figure 5(a)). The rosette in Figure 6(a) divides

Figure 5. (a) Lineament map; (b) lineament density map.

these lineaments into 17 groups of 10˚ based on their directions (NE-SW, ENE-WSW, SE-NW, and NNE-SSW). The longest lineaments (more than 0.5 km) travel diagonally in a NE-SW direction across the research region. The NyC Palaeoprotozoic shear zones align with this major direction.

Figure 5(b) shows the lineaments density map. The studied region is heavily worn and has a cluster of fractures in the middle. From a hydrogeological perspective, the shrub-covered basement areas (NDVI < 0.3; Figure 6(b)) are dominated by large fracture densities (more than 12 Km/Km²) (Figure 7). The shape of the aquifers in the research area is better understood thanks to them. The changed surface zones caused by the growth of massive tree roots (NDVI > 0.3) are reflected in the low densities. They favor surface water infiltration. Thus, the analysis of land use using the normalized vegetation index (NDVI) gives an understanding of the type of vegetation present. The bedrock undelaying that

Figure 6. (a) rosette; (b) land cover.

Figure 7. Overlay of linear density and land use map.

vegetation influences it. In response, it causes the fracturing that is necessary for aquifer development on this substratum. Thus, the NDVI examination offers important details about the existing fractures.

In this work, conductive anomalies associated with aquifers and those associated with iron ores were distinguished using the joint inversion of resistivity and induced polarization data. A bedrock composed of materials with resistivity values between 20 and 8000 Ohm.m and chargeability values between 0.5 and 40 mV/V, is depicted in the results presented as pseudosections (Figure 8 and Figure 9).

Figure 8. Pseudo-sections of resistivity (Top) and chargeability (bottom) of profiles 1 and 2.

Figure 9. Pseudo-sections of resistivity (Top) and chargeability (bottom) of profiles 3 and 4.

These sections display a zone with conductive sandy clays (700 - 1700 Ωm) and resistive duricrust (3000 Ωm) covering its surface. These materials exhibit high chargeability values (M > 16 mV/V), suggesting prospective mineralized targets that warrant further investigation [1]. This layer appears to be the outcome of a protracted process of mineral alteration and sedimentation of the intrusive material east of profiles 1 and 2. The chargeability values of this conductive incursion (Rho < 2500 Ωm) are extremely high (M > 22 mV/V). West of our profiles, different conductive alterites of varying thickness and limited chargeability (Rho < 2500 Ωm and M < 12 mV/V) are seen ten meters below the surface layer. Surface water may accumulate in these porous alterites.

The resistant gneissic basement (3600 - 7000 Ωm) is located in the centre, 39 m below the overburden. When one moves from profile 1 to profile 3, this bedrock appears to undergo a reworking towards the east, resulting in highly conductive granular materials whose low chargeability (Rho < 800 Ωm, M < 7.9 mV/V) indicates the presence of water. These materials continue on profile 4, which appears to be in a collapse zone where the base is obscured. Thus, a large aquifer that reaches a depth of more than 120 meters below the surface appears to be present in the eastern portion of the region covered by profiles 3 and 4.

A total of 11,407 sample blocks measuring 71 m × 47 m × 8 m were built, using the standard kriging technique in the Geosoft Oasis Montaj program. The overall volume of the area covered by the four electrical profiles is represented by their volume of 0.09 km³. Materials having resistivity values between 1216 and 1900 Ωm predominate on the conductive surface of the block model (Figure 10(a)). Underneath a resistive basement (Rho > 3000 Ωm) that is poorly permeable and susceptible to weathering, they make up the 39 m thick permeable weathered surface layer.

The chargeability contrasts remain unchanged when the conductive anomalies (Figure 10(b)) seen in the preceding figure are isolated. The conductive anomalies can be divided into two categories by observing these contrasts: (a) one potentially mineralized unit with high IP values (M > 15 mV/V) and (b) a prospective aquifer unit with low IP values (M < 12 mV/V) (Figure 10(c)). The prospective aquifer unit, isolated in Figure 10(d), corresponds to a volume of 0.068 km³ of geophysical cells interpreted as potential aquifer zones. This represents 75.5% of the total volume investigated. This volume refers to the interpreted subsurface volume where geoelectrical properties are consistent with water saturation, not to stored groundwater reserves whose estimate would additionally require porosity and specific yield, not available in this study. Given that the data were taken in a heavily fractured area that is known to have considerable water infiltration, this aligns with local environment.

In view of the thickness of the weathered layer, highlighted aquifers can be separated into surface aquifers (formed by the porous clay weathering) and deep fractured basement aquifers. The surface aquifers occupy a volume of 0.025 km³, i.e. 36.76% of the aquifers mapped in the area. They are span all-over the study area and give to it a marshy appearance. More than 64% of the water tables are therefore located at depth, precisely to the east of the zone in between profiles 3 and 4, where a collapse-structure is observable.