Evaluating the Performance of Machine Learning and Deep Learning Techniques to HyMap Imagery for Lithological Mapping in a Semi-Arid Region: Case Study from Western Anti-Atlas, Morocco

Accurate lithological mapping is a crucial juncture for geological studies and mineral exploration. Hyperspectral data provide the opportunity to extract detailed information about geology mineralogy of Earth’s surface. Machine learning (ML) deep (DL) techniques an accurate effective various types lithologies in arid semi-arid regions. This article discusses use machine algorithms, specifically Support Vector Machines (SVM), one-dimensional Convolutional Neural Network (1D-CNN), random forest (RF), k-nearest neighbor (KNN), complex area with strong hydrothermal alteration. The study evaluates performance four algorithms three different zones Ameln valley shear zone (AVSZ) at eastern Kerdous inlier, Moroccan western Anti-Atlas. results demonstrated that 1D-CNN achieved best classification most units. Additionally, LK-SVM good compared other SVM models, as well RF KNN. Our concludes combination CNN HyMap can lithologic selected region, overall accuracy ~95%. However, this highlights challenges identifying units using remotely sensed due spectrum similarities induced by similar chemical mineralogical compositions. emphasizes importance carefully considering evaluating ML DL methods studies, then recommends high-resolution hyperspectral models results. implications would be fascinating exploration geologists Mineral Prospectivity Mapping (MPM), especially selecting appropriate highly metallogenic provinces.