SEED-FD use case datasets are now publicly available, providing a tangible resource to explore how the project’s scientific innovations can improve flood and drought forecasting. Featuring simulations for a range of river basins, the datasets offer researchers, operational forecasters and emergency providers a direct opportunity to assess the benefits of SEED-FD’s developments.
Why improved forecasting matters
Floods and droughts threaten lives, livelihoods, infrastructure and ecosystems worldwide. Reliable forecasts support preparedness for these events by helping decision-makers understand developing risks and act before severe impacts occur.
The SEED-FD research project develops scientific innovations to improve flood and drought forecasting, building on the Copernicus Emergency Management Service (CEMS). The project explores how additional data, advanced hydrological modelling and new forecasting methods can strengthen the anticipation of extreme hydrological events.
To test and bring together all developments, SEED-FD has built a prototype version of the CEMS Hydrological Forecast Modelling Chain. This prototype allows the new components to be evaluated, helping to assess their added value and support further fine-tuning.
What the SEED-FD use case datasets provide
Using the SEED-FD prototype, scientists carried out a series of simulation experiments in specific use case regions to assess how the project’s new developments affect flood and drought forecasting. A selection of these simulations is now publicly available as SEED-FD use case datasets. The simulation results cover five river basins with varied hydrological and geographical conditions: the Danube in Europe, the Bhima in India, the Juba-Shebelle in East Africa, the Niger in West Africa and the Paraná in South America.
For each development, the prototype was run both with and without the new component. This allows users to compare results directly and assess the effect of each innovation on the simulated variables, such as river discharge.
The datasets cover the following SEED-FD developments:
- Assimilation of additional data from Earth observation and local measurements
- Improved hydrological modelling for reservoirs and wetlands
- Forecast error correction through AI-based post-processing
- New forecasting products for flash floods, flash droughts, seasonal hydrological drought, and seasonal drought based on soil moisture tracking
Benefits of the SEED-FD use case datasets
A key feature of SEED-FD’s use case datasets is that the simulations were generated in the same environment as the operational CEMS Hydrological Forecast Modelling Chain, which underpins widely used services such as GloFAS and GDO. This makes it easier to compare outputs with operational forecasts and assess the added value of SEED-FD developments.
The data assimilation experiment makes it possible to examine whether the combination of the model’s background with observational data can bring simulated river discharge closer to truth conditions. The resulting output provides insight into how satellite data and local measurements can be used to contribute to improved hydrological forecasting.
The reservoir module allows users to assess how the enhanced representation of reservoir operations and outflows in the model affects the resulting forecasts. The wetland module offers the chance to test whether accounting for wetland processes can reduce the overestimation of river discharge, a known limitation of existing models.
The use case datasets also include results from forecast error correction through AI-based post-processing. These methods are designed to correct forecast errors by learning from past forecasts, observations and additional data sources, aiming to improve the reliability of hydrological predictions.
Finally, the datasets support the evaluation of new forecast products for extreme events that were not previously covered by the CEMS Hydrological Forecast Modelling Chain. By addressing risks such as flash floods, flash droughts, and seasonal drought conditions, a broader range of flood and drought events can be forecast, providing a basis for better preparedness.
Who can use the SEED-FD use case datasets
The SEED-FD use case datasets are freely available, lowering barriers to using these reference simulations for hydrological forecasting research and the evaluation of SEED-FD’s practical value.
For researchers, the datasets provide a basis for benchmarking the new developments against their own methods and approaches. For operational forecasters, they offer a way to explore how the innovations may affect hydrological forecast skill. For emergency providers, they can help assess how improved forecasts for a broader range of extreme events could contribute to better preparedness and response.
Outlook: from datasets to future forecasting improvements
The published use case datasets serve as case studies demonstrating the potential added value of SEED-FD developments. They support the ongoing validation and refinement of the prototype components during the project’s final year.
The developments have been designed with future integration into the Copernicus Emergency Management Service in mind. At the same time, they may also be adopted independently in other forecasting and early warning systems, depending on user needs and operational contexts. Institutions interested in the SEED-FD developments and their potential use in different forecasting contexts are welcome to contact the project team.
To receive the latest updates on SEED-FD results, follow the project on LinkedIn, X and Facebook, or subscribe to the newsletter via the website footer.
Access the SEED-FD use case datasets
To explore SEED-FD’s use case datasets, download the accompanying document with further background information. Public access to the data is explained on page 32. Please note that the datasets contain prototype and experimental data that may change during the course of the project and should therefore be used with caution. For any questions about the datasets or their use, please contact the SEED-FD project team.