Automatic geological mapping using remote sensing data: case of the Zgounder deposit (Anti-Atlas, Morocco)

Remote sensing is an effective tool used in a variety of earth science applications including extracting geological details at unmapped terrain. In this research, the support vector machine (SVM) algorithm is applied to geological classification of a Zgounder silver deposit (Anti-Atlas, Morocco) using a combination of data acquired by Sentinel-2 and a digital elevation model (DEM). The aim of this study is to identify potential zones of mineralization on the basis of automatic geology mapping using the SVM classification. Image enhancement and processing techniques have been used to improve geological and lithological discrimination. Also, principal component analysis (PCA) was applied to reduce data redundancy by choosing bands that retain more than 90% of the original information. The results showed an overall classification accuracy of > 97%, with a kappa index > 0.96. The results obtained allow us to conclude that the classification approach produced an image containing lithological units, allowing easy identification of potential mineralization zones. The study shows that the approach presented can help construct a first-pass geological map for areas where information on the lithology is unavailable. It saves time and resources compared to geological mapping in the field.