Geological mapping using Random Forests applied to Remote Sensing data: a demonstration study from Msaidira-Souk Al Had, Sidi Ifni inlier (Western Anti-Atlas, Morocco)

Geological mapping plays a very important role in the exploration of oil, mineral and water resources, as well as in identifying and monitoring natural hazards. It is an indispensable means for the economic development of the country. Remote sensing data provides critical support by reducing the costs and increasing the precision. This research work evaluates the use of Random Forests, a supervised machine learning algorithm, for geological mapping of the Msaidira-Souk Al Had region, a part of the sidi Ifni inlier situated in southern Morocco. By integrating the spectral and textural features of Sentinel-2A with the morphometric attributes of Digital Elevation Model (DEM) of ALOS/PALSAR. The experiment revealed that the overall accuracy reaches ≈ 91% while the kappa coefficient is 88%. As the final result of this research, the Random Forest method is an effective tool that geoscientists can use to produce a new map or to update existing geological maps.