Project Details
Projekt
High-resolution gravity fields for better flood forecasting
Subject Area
Hydrogeology, Hydrology, Limnology, Urban Water Management, Water Chemistry, Integrated Water Resources Management
Term
since 2025
Project identifier
Deutsche Forschungsgemeinschaft (DFG) - Project number 551141034
Floods are one of the most destructive natural hazards. Real-time monitoring, forecasting, and early warning of floods are thus essential for disaster mitigation. Besides rainfall characteristics, the wetness conditions of the river catchment prior to the event are important factors for runoff generation and thus for the magnitude of the flood event. With the advent of the Gravity Recovery and Climate Experiment (GRACE) twin satellites in 2002 and their successor mission GRACE Follow-On (GRACE-FO) in 2018, it became possible to retrieve terrestrial water storage (TWS) anomalies by measuring temporal variations of the Earth’s gravity field. This opens the possibility of estimating the wetness conditions before and during flood events. However, the use of these data in flood warning systems has largely been limited so far by their low temporal (monthly) and spatial (250 – 300 km) resolution. The goal of this project is to improve the prediction and monitoring of flood events by using daily TWS anomalies derived from GRACE/GRACE-FO, downscaled to a high spatial resolution of 50 km globally. In a first step, we will derive daily gravity fields using a Kalman filter approach. However, this improvement compared to the standard gravity field data comes at the cost of a worse spatial resolution. To achieve a high spatial resolution of the TWS data sets, we will develop new methods based on machine learning to fuse TWS anomalies from GRACE/GRACE-FO with those from hydrological simulations. Towards flood warning and monitoring applications, we develop a methodology to derive the TWS data in near real-time and predict them into the future with machine learning. From the high-resolution TWS anomalies, we will derive a wetness index that will serve as an (early-warning) indicator of flood-prone water storage conditions in river basins. We assess its value relative to other flood-generating factors for different environmental conditions in river basins worldwide of various sizes in the range of few 10,000 km² to several millions of km². A machine learning framework that utilizes this GRACE/GRACE-FO-based wetness index, precipitation forecasts, and other auxiliary data to issue flood warnings will be developed. Furthermore, the gravity-based water storage anomalies will be incorporated into an existing flood forecasting model in selected river basins in Lower Saxony, Germany. The performance of the forecasts based on both the machine-learning approach and the flood modelling system on the regional scale will be evaluated with benchmark hindcast experiments, i.e. by studying in detail its performance for flood events in the past. Our new approach has significant potential to improve the accuracy and reliability of flood forecasts. In addition, we also see further applications of the high-resolution TWS anomalies such as for quantifying groundwater or soil moisture dynamics.
DFG Programme
Research Grants
International Connection
Switzerland
Cooperation Partners
Dr.-Ing. Ulrich Meyer; Professor Dr. Benedikt Soja
