Researchers at the Ludwig-Franzius Institute (LuFI), in collaboration with international partner institutions, have developed an AI-based system that precisely maps flood-affected areas - even under dense vegetation, at night, and in cloudy conditions.
The system is based on radar images from the European Sentinel-1 satellite, whose typical system-related image artifacts are effectively reduced by a specially adapted neural network. By specifically linking this data with vegetation and land-use information, the new method significantly outperforms previous standard methods in the accuracy of flood mapping during prolonged crisis situations, as demonstrated by tests conducted during extreme events in Hanover (2023) and Valencia (2024).
The innovation developed by LuFI enables disaster management agencies to assess situations more quickly in emergencies and supports the long-term planning of effective climate adaptation measures.
The study was published in BIG EARTH DATA by Taylor & Francis:
Moghimi, A., Bahrami Mahtaj, A., Jamali, A., Welzel, M., Motagh, M., & Schlurmann, T. (2026). SAGE-UNet: A mixture-of-experts spatially-aware gated enhancement U-Net for flood water segmentation from Sentinel-1 SAR and post-event impact assessment—learning from flood events in Germany and Spain. Big Earth Data, 1–43. https://doi.org/10.1080/20964471.2026.2702216
