Advanced Flood Prediction with Physics-Guided Deep Learning: Combining UNet, FNO, and SAR/Optical Imagery

This paper presents a physics-guided deep learning framework for flood prediction that integrates multi-modal remote sensing data (Sentinel-1 SAR, Sentinel-2 optical imagery, and DEM-derived terrain features) with constraints from the depth-averaged shallow water equations. The proposed hybrid architecture combines UNet for feature extraction from imagery with FNO (Fourier Neural Operators) for learning PDE-based physical dynamics, enabling accurate and scalable flood mapping across heterogeneous terrain conditions.