FLOOD WARNING SYSTEM IN THE KARST CATCHMENT OF THE RIVER LAUCHERT (SWABIAN ALB, GERMANY, EUROPE)

River floods can be considered a natural hazard imposing a serious risk for inhabitants of river planes. Flood forecasting is a measure to reduce damages associated to floods. Nevertheless, flood forecasts in karst catchments are challenging. Karst aquifers, providing for spring discharge usually show a heterogeneous structure simultaneously allowing for long term storage, quick transport, and discharge of groundwater. Due to these particularities, stream flow simulations with common precipitation-runoff models are challenging and often doomed to fail. Consequently, flood warnings in karst catchments are equally challenging.

In order to overcome these challenges, a groundwater based approach was applied to the well studied karst catchment of the river Lauchert in south west Germany (Europe), taking transport and storage in the soil, unsaturated and saturated zone into account. Insights from field data were used for the development of distributed and lumped numerical models allowing for the simulation of spring and stream flow of sub catchments on a sub daily time scale. The developed models are capable of using radar based precipitation data as input. Since topographic data is implemented at the same resolution, height induced temperature variations, snow accumulation and melt can be resolved. Land use information was introduced. A distributed approach was chosen for the soil zone, since here, detailed and spatially distributed data are available and of importance. A lumped approach on a sub catchment scale was in contrast chosen due to limited information on the spatial distribution of hydraulic properties within the karst aquifer. By coupling both model frameworks for stream flow simulations, a purely groundwater based approach is followed for the simulation of stream flow, adapted to the special hydrology of karst catchments. With the developed models it was possible to simulate stream flow with a solely groundwater focused approach. The models can be used for flood warnings since now forecast data from weather services can directly serve as input for the simulations.