Spatial expansion of artisanal and small-scale gold mining nearby the Nile River, Sudan and its potential environmental impacts: Insights from Planetscope data and machine learning

Artisanal and small-scale gold mining (ASGM) has dramatically expanded along the Nile River, North Sudan; however, the rates and environmental impacts were not sufficiently assessed. We aimed to use PlanetScope data to detect and map ASGM and highlight its environmental impacts around the Nile River, North Sudan, using the random forest (RF) classifier in three steps. First, a visual inspection and analysis were performed to evaluate how distinguishable ASGM sites are from rock units/geological features in color composites; then, reference data were collected from processed images for training and testing, and supervised classification was conducted using binary and multiclass RF classifiers. RF and PlanetScope data were efficient in discriminating ASGM sites with high overall accuracy (0.84-0.92). The binary approach ensured higher accuracy over the multiclass approach, but the latter helped to understand the spatial distribution of illegal mining. Our findings showed that ASGM areas significantly expanded from 50 ha (2016) to 90 ha (2021) and 125 ha (2024). Additionally, we highlighted the environmental risks associated with the development of ASGM in the region. The results can help decision makers and stakeholders to obtain better information on the environment, and the methodology helps to monitor ASGM activities.