Deep U-Net Framework for Flood Hazard Mapping in Wupper Catchment

Researchers developed a deep-learning surrogate model for flood hazard mapping, using a U-Net architecture to predict maximum water levels efficiently. The model was tested on hydraulic simulations of the Wupper catchment in North-Rhein Westphalia, Germany, achieving results comparable to traditional computationally expensive simulations. The study optimized U-Net architecture, patch generation, and data handling to approximate hydraulic models, aiming to provide rapid and reliable flood prediction tools amid increasing flood frequency and severity.