Funded by GNI

Project Leader at TUM
Prof. Dr.-Ing. Xiaoxiang Zhu

Cooperation Partners
TUM Chair of Algorithmic Machine Learning & Explainable AI (Prof. Dr. Stefan Bauer)
TUM Chair of Hydrology and River Basin Management (Prof. Dr. -Ing. Markus Disse)
TUM Chair of Geoinformatics (Prof. Dr. rer. nat. Thomas H. Kolbe)

Runtime
2026 – 2030

Climate change and rapid urbanization are increasingly exposing cities to flood risks by reducing natural water retention and altering hydrological cycles. Existing measures such as green infrastructure and improved drainage are insufficient if applied in isolation, highlighting the need for an integrated framework to systematically assess and enhance urban flood resilience. This project aims to develop a multi-disciplinary urban flood analysis framework that combines Earth observation (EO), hydrological–hydraulic models, urban digital twins, and advanced AI methods. The framework will analyze the causal drivers of flooding, quantify resilience with AI-based surrogate hydrodynamic models, and generate what-if digital twin scenarios through quality-assured, data-driven methods. Furthermore, representative learning from optical and multi-sensor EO data will automate scenario creation, accelerating resilience planning with limited resources. The framework will be designed on open standards such as CityGML, 3DTiles, and SensorThings API, enabling broad applicability and scalability. The project unites expertise from four research domains: AI-driven Earth observation (Chair of Data Science in Earth Observation), flood modelling and urban hydrology (Chair of Hydrology and River Basin Management), semantic urban modelling and digital twins (Chair of Geoinformatics), and algorithmic machine learning with causal and explainable AI (Chair of Algorithmic Machine Learning & Explainable AI). This integration of complementary strengths will enable innovative, transferable solutions to support flood-resilient urban design.