Integrated geophysical, hydrochemical, and machine learning framework for sustainable groundwater management in West African coastal aquifers: Insights from Kribi, Cameroon

Study region: Douala-Kribi-Campo Basin, Cameroon, West Africa. Study focus: This study developed an integrated physics-informed machine learning framework for characterizing data-scarce, fractured coastal aquifers. The approach combines vertical electrical sounding, seismo-electromagnetic tomography, and hydrochemical sampling with a neural network trained solely on synthetic geophysical responses eliminating the need for pumping tests. The resulting hydraulic property estimates were fused with remote sensing and 3D geological modeling to generate policy-ready groundwater vulnerability and potential indices. New hydrological insights for the region: The aquifer system is multi-layered and fracture-controlled, with principal conductive zones at 6–15 m, 20–45 m, and 85–95 m depths. Hydrochemical tracers (Cl⁻, NO₃⁻) distinguish anthropogenic contamination from saltwater intrusion in low-resistivity coastal zones. The 3D geological model confirms a hydraulically connected weathered–fractured basement, where groundwater productivity is governed by discrete fracture networks. The integrated conceptual model and spatial indices provide the first actionable tools for sustainable well siting, abstraction limits, and monitoring in Kribi, offering a transferable framework for coastal crystalline aquifers in data-scarce regions globally.