AURORa (AI-Driven Urban Flood Resilience: Integrating Earth Observation and Architectural Innovation)

Urban areas are becoming increasingly vulnerable to flooding as climate change intensifies extreme rainfall events and rapid urbanization reduces natural water retention while altering hydrological processes. Although measures such as green infrastructure and improved drainage systems are widely applied, a major challenge remains: how to systematically assess and enhance urban flood resilience under limited resources. Current approaches often address individual aspects of the problem in isolation, underscoring the need for an integrated and holistic solution.

This project aims to develop an urban flood analysis framework that supports flood-resilient urban planning and design. The framework integrates multi-sensor Earth observation (EO) data, hydraulic–hydrological models, urban digital twins and advanced AI methodologies to comprehensively analyze the factors influencing urban flood resilience.

The project focuses on four key objectives:

1. Enhanced flood detection by linking observed parameters to digital twins and explicitly addressing causal relationships.
2. Quantification of urban flood resilience through an AI-based surrogate hydrodynamic model.
3. AI-driven generation of what-if scenarios, enabling quality-assured digital twin simulations for future planning.
4. Automated scenario creation using representative learning from optical EO data.

By combining these components into a single, coherent system, the project aims to significantly improve and accelerate flood resilience planning while minimizing required investments. Calibration and validation will be performed using data from major flood events across multiple case study cities, ensuring robustness, reliability, and adaptability to diverse urban environments. To support transferability and interoperability, the framework will rely on open standards such as CityGML, 3DTiles, and the SensorThings API.

The project brings together expertise from four complementary research groups:

• Chair of Data Science in Earth Observation, specializing in AI and data science for Earth observation and remote sensing.
• Chair of Geoinformatics, focusing on spatio-temporal semantic modelling of urban environments and Urban Digital Twins.
• Chair of Algorithmic Machine Learning & Explainable AI, developing AI methods that learn causal relationships, provide explainable outputs, and adapt to new challenges.
• Chair of Hydrology and River Basin Management, contributing expertise in flood modelling, urban hydrology, and hydraulics.

By integrating these competencies within a single interdisciplinary team, the project is uniquely positioned to advance robust, interpretable, and transferable solutions for urban flood resilience.
 

AUROrA (AI-Driven Urban Flood Resilience: Integrating Earth Observation and Architectural Innovation)

Start date: 01.02.2026

End date: 31.01.2030

Project Partners: TUM Chair of Hydrology and River Basin Management (Prof. Dr.-Ing. Markus Disse); TUM Chair of Algorithmic Machine Learning & Explainable AI (Prof. Dr. Stefan Bauer); TUM Chair of Geoinformatics (Prof. Dr. rer. nat. Thomas H. Kolbe); TUM Chair of Data Science in Earth Observation (Prof. Dr.-Ing. habil. Xiaoxiang Zhu)

PhD Student: M.Sc. Teresa Jansen

Project Leader: Prof. Dr.-Ing. habil. Xiaoxiang Zhu

Funding: Georg Nemetschek Institute (GNI)