Assessing groundwater favourability using multi-scale spatial regression techniques within a GIS framework

Abstract

Study region Raya Azebo Woreda in northern Ethiopia faces increasing groundwater stress due to variable rainfall, heterogeneous lithology, and growing agricultural and domestic water demand, making reliable groundwater-potential assessment essential for sustainable water management. Study focus This study integrates Multiscale Geographically Weighted Regression (MGWR) with GIS to map groundwater potential using lithology, rainfall, land cover, distance to lineaments, and distance to rivers. The model achieved an adjusted R² of 0.8796, with RMSE = 0.3412, MAE = 0.2736, and AUC = 0.964. MGWR identified broad-scale effects for lithology, rainfall, lineaments, and rivers (bandwidth = 101,990.79 m), while land cover exhibited a localized influence (bandwidth = 48,037.78 m). New hydrological insights The MGWR–GIS integration advances conventional groundwater mapping by capturing the scale-dependent spatial influence of individual hydrological controls rather than assuming uniform relationships across the landscape. The resulting classification into poor, moderate, high, and very high groundwater-potential zones provides spatially targeted information for groundwater exploration, recharge planning, and sustainable resource management.