Using Machine Learning to Map Artisanal Mining across Five Countries

Abstract

Artisanal and small-scale mining (ASM) is a significant but often unregulated economic sector in sub-Saharan Africa. The lack of comprehensive monitoring of ASM activity hinders efforts to address its environmental and social impacts or promote its sustainable development. We address this gap by creating a large, high-resolution dataset of manually labeled ASM sites across five countries in the region. Using these data, we train machine learning models that integrate geographic features and publicly available satellite imagery to map ASM at scale. Our results demonstrate that these models can outperform on-the-ground surveyors and successfully detect ASM in previously unmonitored areas. We show that standard machine learning approaches applied to this problem fail to spatially generalize, but that simple adjustments to model selection can improve model generalization in policy-relevant ways. Finally, we predict ASM prevalence across our sample countries, highlighting areas where ASM activity may overlap with biodiversity hotspots, protected areas, and critical watersheds.