A data-driven remote sensing approach for VMS mineralization mapping: Integrating Sentinel-2 imagery with geology data in the Asmara Belt, Eritrea

Mineral resources play a critical role in the sustainable economic development of countries. In Eritrea, conventional mineral exploration and geological mapping methods are expensive, time-consuming, and in some inaccessible areas, difficult to implement. Advances in remote sensing and data-driven analytical techniques now provide efficient alternatives for mineral exploration. This research applies a remote sensing and data-driven approach, combining multispectral Sentinel-2 imagery with field geological data, to identify volcanogenic massive sulfide (VMS) deposits and map lithology in the arid Asmara mineralized belt of Eritrea. Image processing techniques, including band ratios and Feature Oriented Principal Components Selection (FPCS), were combined with supervised classification algorithms such as Maximum Likelihood, Minimum Distance, and Spectral Angle Mapper to derive geological classes. Field data, including rock samples and GPS locations of known VMS gossans, were integrated for model training, thin-section validation, and performance assessment. The results demonstrate that hydrothermal alteration zones associated with VMS deposits, expressed as oxidized gossans, were effectively distinguished from widespread but unmineralized hematitic laterites using the Chica-Olma ratio method and supervised classification algorithms. The derived lithological and alteration maps show strong agreement with existing geological maps, and the locations of known VMS deposits, underscoring the potential of combining Sentinel-2 imagery and geological field data for mineral exploration in the Arabian–Nubian Shield.