Integrating Remote Sensing and Artificial Intelligence for Structural Analysis of Fractured Aquifers in the Beni-Mellal Atlas, Morocco

Laboratory of Applied Geophysics, Geotechnics, Engineering Geology and Environment (L3GIE), Mohammadia School of Engineers (EMI), Mohammed V University in Rabat, Morocco

^* Corresponding author: z.elkamel@research.emi.ac.ma

Abstract

The Beni Mellal Atlas is a complex karstic mountain range where faults and tectonic fractures highly control groundwater flow. The objective of this research was to map geological lineaments using radar remote sensing data — Sentinel-1 (C-band) and ALOS PALSAR 1.1 (L- band) — and artificial intelligence (Deep learning), highlighting their impact on hydrogeological behavior. The study was carried out using a semi- automatic extraction approach using PCI Geomatica and an automatic segmentation using a U-Net neural network. The NE-SW and E-W fracture systems dominate the direction of underground water flow and the outlets of the main springs. The proposed U-Net approach produces smoother, more geologically reasonable fracture networks than the customary model. It demonstrates that the use of artificial intelligence in radar remote sensing is feasible for structural and hydrogeological research. This approach is of great interest for the reproducible and sustainable exploration of groundwater resources in semi-arid mountain areas.