A High-Resolution Artificial Neural Network-Based Model for Predicting Urban Flooding in Hanover, Germany

Urban flooding poses a significant challenge to cities, requiring the development of advanced predictive models to mitigate potential risks and enhance urban resilience. In the present study, we test an artificial neural network (ANN)-based model for predicting urban flooding in the city of Hanover, Germany. The model provides high-resolution spatial analysis on a 5 x 5 meter grid, providing detailed insights into potential flood-prone areas. With a temporal resolution of 5 minutes, the ANN model uses radar-based precipitation data to predict water levels during extreme weather events. The study is part of the FURBAS project (Forecasting urban floods and strong rainfall events, 2022-2025). This research project is a cooperation of the Institute for Technical and Scientific Hydrology (itwh) GmbH, the Institute of Fluid Mechanics and Environmental Physics in Civil Engineering, Leibniz University of Hanover and the municipal operation for Hanover city drainage. The project is funded by the Federal Ministry for the Environment, Nature Conservation, Nuclear Safety and Consumer Protection under grand number 67DAS224. The data driven flood prediction model uses pre-simulated flood scenarios from a physically based model as training data. The model approach of Berkhahn and Neuweiler (2023) was adapted for the present study to cope with the large catchment area of about 260 km. The proposed model could improve timely decision making for urban planning and emergency response in the future. Despite the focus on the specific challenges of the city of Hanover, the chosen modeling approach could also be applied to flood forecasting and management in other cities. With this conference contribution we want to highlight the challenges of real-time forecasting of pluvial urban floods in large catchments and present first preliminary results. Berkhahn, S., Neuweiler, I. (2023). Data driven real-time prediction of urban floods with spatial and temporal distribution. Journal of Hydrology X, 100167.