SEED-FD advances global flood and drought forecasting by improving the representation of reservoirs, wetlands, and groundwater in the LISFLOOD hydrological model. These improvements bring forecasts more in line with real conditions, supporting better decisions for governments, businesses, and communities worldwide. Within the consortium, experts at IIASA are leading this part of the research.
Why Improving Hydrological Models Matters
Floods and droughts are among the most damaging natural disasters. Reliable forecasts depend on how well hydrological models represent water movement across landscapes. The LISFLOOD model, central to the Copernicus Emergency Management Service (CEMS) early warning system, does not yet capture processes such as reservoirs, wetlands, and groundwater flows well, leaving potential for improvement.
SEED-FD addresses this need by conducting research to better represent these processes within the model, resulting in more accurate flood and drought forecasts in global early warning systems. This is particularly relevant for data-scarce regions, where accurate modelling is most challenging.
Reservoir Improvements Lead to More Realistic Forecasts
Reservoirs play a central role in water management, serving purposes from flood protection to irrigation and hydropower. In the currently operational version of LISFLOOD, reservoir behaviour is captured only in broad terms. It does not represent how operators actually manage reservoir releases during wet and dry seasons.
Researchers in SEED-FD have developed a new modelling approach that simulates reservoir operations more realistically. By incorporating typical seasonal patterns of water storage and release, the model more accurately reflects real-world practices. Tested in India’s Bhima basin and parts of the Danube basin, it enables forecasts to capture storage and release dynamics with greater accuracy.
Wetland Improvements Strengthen Forecast Accuracy
Wetlands act as natural regulators, storing water and releasing it slowly, while also supporting ecosystems and local livelihoods such as fishing and grazing. Yet, in most large-scale hydrological models, wetlands have been underrepresented and their role in the water cycle has not been fully captured.
SEED-FD has introduced a dedicated wetland component into a project version of LISFLOOD, guided by Earth observation data from satellites. The improvements have been developed using the Inner Niger Delta in Mali as a case study. With the new approach, wetlands are represented more realistically by considering their seasonal extent and their high evaporation rates. This leads to forecasts that reduce the previous overestimation of river discharge and therefore match real-world observations more closely.
Image 1: Comparison of simulations with the new wetland approach (right) and without it (left). With the new wetland component, the simulations (red line) match the observations (blue line) more closely, and the previous overestimation of river discharge is noticeably reduced. © IIASA, all rights reserved.
Integrating Groundwater Dynamics into Forecasts
Groundwater plays a key role in long-term water availability, however, important processes such as lateral flow (the sideways movement of water underground) and capillary rise (water moving upward from groundwater into the soil) are not captured in the operational LISFLOOD model.
Researchers in SEED-FD are working to link a project version of LISFLOOD with the established MODFLOW groundwater model to include these dynamics. Since the two models rely on different map projections, data are converted on the fly from LISFLOOD to MODFLOW and vice versa. While this work is still in progress, it marks another step towards a more realistic representation of groundwater within large-scale hydrological modelling.
Global Impact of Hydrological Model Improvements
Better hydrological modelling translates directly into practical benefits. For public authorities, it means more reliable warnings to protect people and infrastructure. For farmers and industries, it supports smarter water allocation during dry periods. Humanitarian agencies gain clearer guidance for preparing relief operations.
These advances are especially valuable in regions where monitoring data is scarce, helping communities anticipate and respond to water extremes more confidently through more reliable information.
Conclusion and Outlook
SEED-FD has advanced hydrological modelling with more realistic reservoir operations, a dedicated wetland component, and progress toward integrating groundwater processes. The reservoir and wetland developments have recently been incorporated into a project version of LISFLOOD.
With the validation phase now underway, the results are being integrated alongside other research findings from SEED-FD into the new prototype of the CEMS Hydrological Forecast Modelling Chain. The aim is to test this prototype against historical extreme weather events in diverse regions, ensuring global applicability and assessing how far early warning performance can be improved.
If you would like to learn more about SEED-FD and stay up to date with the latest news, follow us on LinkedIn, Facebook, and X.