Integrating groundwater models in catchment runoff models in a multi-model ense mble approach for improved low-flow prediction

Several studies indicate that future climate change and increased urbanization will result in an increase of low-flow conditions along rivers in Belgium. A reliable quantification of the occurrence and magnitude of low-flows is required to support impact analysis on water supply, navigation, water availability for natural ecosystems and river water quality, and the design or evaluation of adaptation measures. Current river modeling approaches, however, face difficulties in low-flow simulation. Different types of hydrological models predict very different changes in low-flow occurrence and magnitude under changing climate conditions. These differences may be largely attributed to differences in conceptualization of the groundwater component. It may seem logical that hydrological models with a more detailed physically-based groundwater component result in better predictions but these models often suffer from overparameterization. Additionally, computational times may be so large that these models become unfit for several practical purposes. In this project, a solution for combining the advantages of complex fine-scale spatially-distributed models with simple fast rainfall-runoff models is sought in a multi-model approach where different models of increasing complexity and spatial detail can be combined.