Artificial intelligence driven aeromagnetic and satellite data fusion for mineral and groundwater resource mapping in sub Saharan Africa

Abstract The integration of spatial Artificial Intelligence (AI) with aero-magnetic and satellite remote sensing data is redefining subsurface resource exploration across mineral-rich, data-constrained regions. This review synthesizes recent advances in geospatial artificial intelligence, including machine learning, deep learning, and hybrid approaches, with particular emphasis on their applications in mineral prospectivity and groundwater potential mapping. Using a comparative framework, the study categorizes AI models based on learning architecture, data source fusion, and deployment context, highlighting emerging techniques such as Federated Learning, Graph Convolutional Networks, Autoencoders, and Vision Transformers. A novel taxonomy is proposed to evaluate model maturity and field readiness across four African case contexts: Nigeria, Ghana, South Africa, and the Democratic Republic of Congo. The review highlights how these techniques improve spatial resolution, predictive accuracy, and automation in subsurface characterization, while addressing regional challenges such as geological heterogeneity, limited labeled data, and infrastructure constraints. In addition to summarizing the state of AI adoption, this review identifies systemic gaps in policy alignment, data governance, and cross-border collaboration. By drawing cross-regional comparisons and proposing a roadmap for scalable AI-driven geoscientific exploration, this work supports the development of resilient and inclusive technologies for sustainable resource management across sub-Saharan Africa and other developing regions.