Remote Sensing Prospecting Method Based on Random Forest in Eastern Botswana Mining Area

In this study, multi-source remote sensing data and machine learning algorithms were used to delineate the prospect area of remote sensing geological prospecting in eastern Botswana. Landsat 8 remote sensing images were used to produce iron stain and hydroxyl anomaly maps, ASTER remote sensing images were used to extract chalcopyrite mineral distribution maps, and Microsoft high-resolution remote sensing data were used to extract lithology and structure maps to comprehensively analyze regional metallogenic information. Then, the random forest, classification regression tree (CART) and gradient Lift Tree (GBT) classification algorithms were used to compare the models. The results showed that the random forest algorithm had the best performance in identifying mineralization potential areas, and its accuracy reached 0.95. Finally, the remote sensing geological prospect area of eastern Botswana was delineated based on random forest algorithm, which provided important technical support for mineral resource exploration in this area. This study shows that the combination of multi-source remote sensing data and efficient classification algorithm has great potential in geological prospecting, and provides scientific methods and technical means for the follow-up remote sensing prospecting research.