Identification of the high gold-potential zones: A case study of the Sabodala gold mine, Senegal, by the integration of the geostatistical and remote sensing tools

Gold exploration in West Africa increasingly demands innovative, cost-effective approaches that reduce ground-based interventions while maintaining spatial accuracy. This study introduces a novel integrative framework that combines geostatistical modeling and multispectral remote sensing to enhance gold targeting in the Sabodala region of southeastern Senegal—an area located within the prolific Birimian greenstone belt. Using a dataset of 3’103 drill-hole gold grade samples, three geostatistical interpolation methods—Ordinary Kriging (OK), Universal Kriging (UK), and Indicator Kriging (IK)—were applied to model subsurface gold distribution over a 30 km² area. Concurrently, Landsat 8 OLI/TIRS data were processed to derive surface indicators of mineralization, including the Normalized Difference Vegetation Index (NDVI), Normalized Difference Soil Index (NDSI), Silicate Index, and Land Surface Temperature. A key innovation of this study lies in the spatial overlap analysis between kriging-based predictions and remote sensing-derived indices, which revealed strong to moderate spatial concordance—particularly between IK and the Silicate Index (62%), and UK with NDSI (55%). These results suggest that lithological and alteration signatures captured by remote sensing can reinforce subsurface gold predictions. Moreover, NDVI and temperature anomalies exhibited additional but weaker associations (28–47%), supporting their value as indirect proxies in data-limited zones. This integrative, multi-scale methodology offers a replicable and scalable model for the beginning-stages of the mineral exploration in other terrains. By combining spatial statistics and earth observation tools, the study bridges the gap between surface and subsurface data, improving precision in gold prospecting while reducing financial and environmental costs.