Flood forecasts are essential to issue reliable flood warnings and to initiate flood control measures on time. However, these forecasts comprise different uncertainties. Ensemble forecasts are means of describing different scenarios of the potential future development assuming different initial conditions, different model physics, or even different models. Hence they contribute to the estimation of the hydro-meteorological forecast uncertainty. This paper presents a methodology that is designed to support decision makers in issuing flood alerts, which are based on a probabilistic evaluation of flood forecast ensembles. First results from a case study at the mountainous Mulde catchment are shown using COSMO-LEPS meteorological ensemble forecasts for the extreme flood event in 2002. The rainfall-runoff model ArcEGMO was used to simulate discharge forecast ensembles.
