Mapping Artisanal and Small-scale Mining using Machine Learning: impacts of training sample collection strategies in light of point-based and spatially explicit testing

This repository contains the data used in the article “Mapping Artisanal and Small-scale Mining using Machine Learning: impacts of training sample collection strategies in light of point-based and spatially explicit testing.” The collection includes: - Landsat-9 multispectral imagery (acquired November 2024) - PRISMA hyperspectral imagery (acquired October 2024) Training shapefiles – five versions prepared with different data volumes and collection strategies: - orig (95 points) - ext1 (126 points) - ext2 (232 points) - ext_12 (263 points) - aug (794 points, generated through a region-growing algorithm in R 4.4.3) Testing shapefiles used for both point-based and spatially explicit accuracy assessments These datasets support the comparative analysis of multispectral and hyperspectral data for detecting ASM-related alteration zones in arid environments. Related analytical tools are available at: - https://github.com/ud-geoai/Spatial-Accuracy-Assessment-Tool - https://github.com/ud-geoai/Statistical-region-growing-by-vectorseeds.