Groundwater potential zone assessment and well network optimization under hydrogeological complexity: application of fuzzy-analytical hierarchical process and evidential belief function in the Kabul Aquifer, Afghanistan

Abstract As groundwater demand rises in arid and semi-arid regions, identifying high potential areas is essential for sustainable management. The Kabul aquifer, Afghanistan, is severely stressed by urbanization and over extraction, with groundwater declining at − 0.8 m/year and 33% of wells non-operational. This study integrates GIS with two models, Fuzzy Analytical Hierarchical Process (FAHP) and Evidential Belief Function (EBF), to map groundwater potential zones (GWPZs) and optimize the well network in the Kabul shallow aquifer. Nine hydrogeological layers were used as predictors: aquifer media, aquifer thickness, unsaturated zone thickness, groundwater drawdown, hydraulic gradient, vadose zone type, aquifer recharge areas, geology, and distance from drainages. Hydraulic conductivity was incorporated as a key layer for the well network optimization phase. The target variable comprised 149 wells with hydraulic conductivity above 59 m/day (median), split into training (70%, 104 wells) and validation (30%, 45 wells) subsets. The EBF model outperformed FAHP, achieving a validation AUC of 62.5% (training: 69.1%) versus FAHP’s 51.1% (training: 50.2%). Sensitivity analysis identified geology as the most influential factor; its removal reduced EBF performance from 62.5% to 59.4%. For the primary optimization (1000 m grid), 122 wells are proposed for removal, 263 retained, and 45 added in high potential zones, resulting in a network of 308 wells. Secondary scenarios (400 m and 300 m grids) propose 264 additional wells for extreme urbanization contingencies. This tiered framework offers a practical decision support tool for sustainable groundwater extraction, while emphasizing that recommendations are preliminary and require field validation.