TITLE:
Geospatial Modeling and Hierarchical Clustering of Hydrogeological Borehole Parameters: Application to the Centre-Ouest Area, Burkina Faso
AUTHORS:
Fabrice Ouoba, Vini Yves Bernadin Loyara, Hay Yoba Talkibing, Diakarya Barro
KEYWORDS:
Boreholes, Spatial Autocorrelation, Geostatistics, Hierarchical Clustering, Hydrogeology
JOURNAL NAME:
Open Journal of Applied Sciences,
Vol.16 No.9,
September
23,
2026
ABSTRACT: This study presents an integrated geostatistical and multivariate clustering framework to characterize hydrogeological variability across 261 boreholes in the crystalline basement aquifers of Burkina Faso. Spatial autocorrelation analysis based on Moran’s I indicates significant spatial dependence for borehole depth (I = 0.088, p = 0.0059) and static water level (I = 0.063, p = 0.0336), whereas discharge exhibits no significant spatial autocorrelation (I = −0.032, p = 0.7847). Hierarchical agglomerative clustering identifies two complementary organizational scales: a functional hydrogeological classification into three clusters (k = 3, s = 0.3045), corresponding to regolith reservoirs, hydrodynamically constrained media, and high-yielding structural fractures, and a broader macro-spatial regionalization into two clusters (k = 2, s = 0.4978). This scale-dependent organization highlights the dominant role of localized litho-structural heterogeneities in controlling borehole productivity. Variogram-based ordinary kriging captures broad spatial trends, with root mean square errors of 11.658 m for depth and 6.503 m for static water level. However, the low coefficients of determination (R2 = 0.068 for depth and R2 = 0.007 for static water level) indicate limited local predictive performance and substantial short-range variability. Accordingly, kriging is more appropriate for delineating regional spatial patterns than for precise point-scale prediction. Overall, the integrated framework provides a useful basis for regional hydrogeological characterization and supports groundwater management and borehole siting, while highlighting the need to incorporate auxiliary litho-structural, geophysical, and topographic information to improve local-scale prediction.