UC1- Flash Flood Forecasting

This dataset supports advancing flash flood prediction through AI-driven methods, addressing limitations of traditional hydrological models. Flash floods, rapid events from heavy rainfall, often cause significant damage or fatalities. Physical models simulate water flows but demand vast, diverse data and expertise, with long setup/inference times unsuitable for operations. We explore data-driven alternatives: hybrid AI-physical models to reduce input needs (volume, diversity, resolution) or AI surrogate models. Key preprocessing includes: AI-based building footprint extraction from point clouds. Upscaling coarse DEMs to high-resolution versions. The dataset covers data preparation, training, visualization, and explainability steps. Interactive, stakeholder-tailored visualizations enable personalized insights for alerts, planning, and analysis. UC1 – Flash Flood Forecasting (CS-Group): Users can get an overview of configured experiments and settings, see input data structure/examples, view processing/modelling organization, and access performance metrics. Part of the ExtremeXP project, co-funded by the European Union Horizon Program HORIZON CL4-2022-DATA-01-01 under Grant Agreement No. 101093164. Project website: https://extremexp.eu/.