Drill cuttings were collected at regular 1-m intervals and macroscopically described according to color, grain size, texture, the lithological nature of the rocks intersected by the boreholes, and groundwater inflows. These observations were used to construct lithological logs and determine the thickness of the weathered layer, the depth to the top of the fresh basement, the position of productive fractured horizons, and the total depth of each borehole. The integrated interpretation of weathering and fracturing patterns was used to define the vertical organization of the aquifer system and to identify the horizons likely to provide groundwater storage and flow. Similar combinations of borehole logs and geophysical information have been used to characterize weathered-rock aquifers [15][17][18].

Figure 4.Boxing of crushed samples collected at one-meter intervals in the Dikodougou area, Northern Côte d’Ivoire.

Borehole productivity was assessed using air-lift yields and operational pumping rates. The air-lift yield, measured after borehole development, was used as a preliminary indicator of productivity and was not considered equivalent to the sustainable operational yield. Borehole productivity in the Poro Region was classified according to the adapted [19] classification presented in Table 1.

Table 1.Operational classification of air-lift yields adapted from [19].

Hydrodynamic parameters were determined from pumping tests conducted at a discharge rate maintained as constant as practicable. Before pumping began, the static water level was measured from a fixed reference point. During both the drawdown and recovery phases, time, discharge rate, and water levels were recorded. The resulting datasets were subsequently checked to identify outliers, discharge-rate variations, and inconsistencies in time records or measurement units. Such quality control is important when interpreting pumping and recovery tests in heterogeneous formations [7][20][21].

Transmissivity was estimated using the Cooper-Jacob semi-logarithmic method, which represents a late-time approximation of the Theis solution. In basement aquifers, where the assumptions of homogeneity, isotropy and lateral continuity are rarely fully satisfied, the method was applied only to the approximately linear late-time segment of the drawdown-time curve. The effects of heterogeneity on this approximation have been examined in pumping-test studies [20][21]. Transmissivity was calculated as follows:

where T is transmissivity (m²/s), Q is the pumping rate (m³/s), and Δs is the change in drawdown over one logarithmic cycle of time.

Recovery data were used as an independent control, while recognizing that non-ideal test conditions can affect their interpretation [21]. Equivalent hydraulic conductivity was calculated using the following relationship:

where e is the saturated thickness of the aquifer. In a fractured medium, this value represents the overall hydraulic behaviour of the tested interval rather than the intrinsic hydraulic conductivity of each individual fracture. Specific capacity was calculated as follows:

where Q is expressed in m³/h and sₘ is the maximum drawdown in meters. To ensure comparability among boreholes, discharge rates and drawdowns were considered at the same stage of the test or at the end of the final stabilized step.

A standardized Principal Component Analysis (PCA) was subsequently performed to identify associations among the variables and the main factors controlling borehole productivity. Depending on data availability, the input matrix included total borehole depth, air-lift yield, operational pumping rate, maximum drawdown, specific capacity, transmissivity, and hydraulic conductivity. Measurement units were harmonized, after which the variables were centred and standardized. Strongly skewed variables were logarithmically transformed where necessary. The components were extracted from the correlation matrix. The number of retained components was determined by combining eigenvalues greater than 1, the scree plot, the cumulative proportion of explained variance, and hydrogeological interpretability. Data completeness and selection bias should be considered when interpreting the borehole database [4][22].

The PCA was interpreted using factor loadings, variable contributions, squared cosines (cos²), correlation circles, and borehole projections on the factorial planes. Components characterized by high contributions from transmissivity, hydraulic conductivity, specific capacity and discharge rates were interpreted as hydraulic-productivity gradients. Components dominated by borehole depth, weathered-layer thickness or drawdown were associated with reservoir geometry, degree of weathering or operational constraints. The analysis of relationships among the parameters was based on descriptive statistics, bivariate correlations and standardized PCA; related approaches have been applied to fractured aquifers in Côte d’Ivoire [6][7][23].

Finally, the results obtained from the lithological logs, pumping tests, and PCA were integrated into a combined interpretation. This approach made it possible to distinguish boreholes whose productivity was primarily controlled by hydraulically effective fracturing from those whose performance was associated with a thick weathered layer or specific local conditions. It therefore provides an objective basis for classifying boreholes and guiding future siting campaigns in crystalline basement aquifers.

The lithological logs reveal a typical succession comprising, from the top to the base, a lateritic duricrust, lateritic or variegated clays, a granitic saprolite and a fissured granite. The superficial horizons mainly provide a storage function, while the fractures in the granite control the preferential circulation and exploitable groundwater inflows (Figure 5). Comparable vertical weathering profiles and fractured horizons have been described in crystalline aquifers [1]-[4][17].

This interpretation should be considered alongside evidence that the geometry of the weathered profile can influence borehole productivity [4][9].

Analysis of the first water strikes indicates that several of them occur in the fissured granite, mainly between 27 and 60 m depth (Figure 5). This result confirms that productivity depends more on the interception of open and connected fractures than on the presence of a significant thickness of weathered materials alone.

Figure 5.Lithological logs and position of water inlets in the boreholes studied.

Table 2 presents all the hydrogeological and hydrodynamic parameters. The total depths vary from 59.77 to 104.90 m, with a mean of 79 m. Static levels range from 2.75 to 18.88 m, with an average of 10.51 m. The first water inflow is observed on average at 23.22 m. The average pumping-test discharge rate is 8.09 m³/h, which indicates substantial potential for rural water supply, although the dispersion of values remains high. Specific capacities range from 0.04 to 0.61 m³/h/m, illustrating the heterogeneity of fractured-basement aquifers. The mean transmissivity is 2.38 × 10⁻⁴ m²/s, with minimum and maximum values of 3.84 × 10⁻⁵ m²/s and 1.05 × 10⁻³ m²/s, respectively (Table 3). The variability of transmissivity is comparable in principle to the heterogeneous hydraulic behaviour reported in other Ivorian fractured aquifers, although direct numerical comparison requires consistent test conditions and units [7][23].

Table 2.Parameters analyzed and hydrogeological purpose.

Table 3.Statistical synthesis of hydrogeological parameters.

The total depths vary from 59.77 to 104.90 m, with a mean of 79 m. The average static level is 10.51 m, while the first water inflow is observed on average at 23.22 m (Figure 6). The average pumping-test discharge rate is 8.09 m³/h, which indicates substantial potential for rural water supply. The mean transmissivity is 2.38 × 10⁻⁴ m²/s, but the values span more than one order of magnitude, illustrating the typical heterogeneity of fissured aquifers.

Figure 6.Statistics of drilling parameters (TD: Total depth; WSL: Static level (m); WI: First water inflow; Q: Pumping-test discharge rate in m³/h; Q_sp Specific capacity in m³/h/m; WT: Weathering thickness (m)).

The statistical analysis of the regional pumping-test discharge presented in Table 4 and Figure 6 shows the positions of the minimum, mean and maximum values on the CIEH class scale. The regional average (7.6 m³/h) places the region in the “high” category. The observed minimum (0 m³/h) falls into the “very low” class, while the observed maximum (20 m³/h) reaches the “very high” class (Figure 7).

Table 4.Regional hydrodynamic characteristics.

Figure 7.Histogram of the pumping-test discharge classes of the Poro region according to the regional average and the CIEH classification.

In accordance with the CIEH classification, regional pumping-test discharges are shown in Figure 8. The study carried out in the Poro region indicates that 85.29% of the boreholes exceed 1 m³/h, although a significant proportion still exhibits low discharge rates. This reflects an average productivity that is generally acceptable, but spatially heterogeneous.

Figure 8.Position of the region’s pumping-test discharge on the CIEH scale (1978).

The average productivity increases as successive aquifer levels are intercepted: the average discharge increases from 2.34 m³/h at WI1, to 4.28 m³/h at WI2, then to 5.87 m³/h at WI3 (Table 5, Figure 9). This progression suggests that the deeper horizons intercepted by drilling often correspond to better-developed or better-connected fractures, without, however, implying that systematic deepening is always effective. Comparable studies caution against treating greater drilling depth as a general guarantee of higher yields [4][8][9][24].

Table 5.Depths and average flows of successive water inflows.

Figure 9.Evolution of mean yield with intercepted aquifer levels and depth.

The major correlations between the hydrodynamic parameters are presented in Table 6. The strongest correlations are observed between transmissivity and hydraulic conductivity (r = 0.95), between specific capacity and transmissivity (r = 0.80), and between pumping-test discharge and air-lift flow (r = 0.79). These relationships confirm that the most productive structures are associated with more transmissive and hydraulically better connected environments. Conversely, the negative correlations between discharge and depth of water inflow (r = −0.74) and between specific capacity and total depth (r = −0.70) show that depth alone is not a reliable indicator of productivity.

Table 6.Major correlations between hydrodynamic parameters.

The linear relationship between transmissivity and specific capacity is expressed as T = 1.04e−03 × Q_sp − 1.61e−05, with R² = 0.678 (Figure 10). This relationship confirms the usefulness of specific capacity as an indicator of the transmissive capacity of fractured horizons, even if the dispersion of the points reflects the effect of lithological and structural heterogeneity. Empirical relationships between specific capacity and transmissivity have also been examined in the Lacs and San-Pedro regions, but their coefficients are site-dependent [7][23].

Figure 10.Relationship between transmissivity and specific capacity of boreholes.

Table 7 summarizes the variation in productivity and lithological architecture. The first three axes of the PCA explain 79.87% of the total variance. Axis F1, which explains 47.92% of the total variance, is mainly associated with the hydrodynamic productivity variables (Figure 11): transmissivity, hydraulic conductivity, specific capacity and flow rate. Axis F2 is more closely linked to the lithological architecture (Figure 12) and to the depths, in particular to the weathering-depth relationship (Table 7). This analysis confirms the separation between direct hydraulic control of productivity and geological control that modulates the probability of intercepting productive fractures. The PCA should be read as an exploratory description of covariation rather than proof of causal controls [4][6][22].

Figure 11.Circle of PCA correlations on the F1-F3 plane.

Table 7.Eigenvalues and variance explained by the first three axes of the PCA.

Figure 12.Circle of PCA correlations on the F1-F2 plane.