Abstract
Artisanal and small-scale gold mining (ASGM) represents a significant economic activity for communities in developing countries. In southeastern Senegal, this activity has increased in recent years and has become the main source of income for the local population. However, it is also associated with negative environmental, social, and health impacts. Considering the recent development of ASGM in Senegal and the difficulties of the government in monitoring and regulating this activity, this article proposes a method for detecting and mapping ASGM sites in Senegal using Sentinel 2 data and the Google Earth Engine. Two artisanal mining sites in Senegal are selected to test this approach. Detection and mapping are achieved following a processing pipeline. Principal component analysis (PCA) is applied to determine the optimal period of the year for mapping. Separability and threshold (SEaTH) is used to determine the optimal bands or spectral indices to discriminate ASGM from other land use. Finally, automatic classification and mapping of the scenes are achieved with support vector machine (SVM) classifier. The results are then validated based on field observations. The PCA and examination of spectral signatures as a function of time indicate that the best period for discriminate ASGM sites against other types of land use is the end of dry season, when vegetation is minimal. The classification results are presented as a map with different categories of land use. This method could be applied to future Sentinel scenes to monitor the evolution of mining sites and may also be extrapolated to other relevant areas in the Sahel.