Representation of different flood types in rainfall-runoff modelling

Flood characteristics vary due to different processes causing the floods. Classifying flood events according to their causative processes can help to improve the estimation accuracy of flood probabilities. Hydrological models can be used for derived flood frequency analyses when the length of observed climate and runoff data is insufficient. This also allows the estimation of uncertainties due to variable catchment characteristics, climate conditions, and runoff regimes. The aim of this work is to investigate whether rainfall-runoff models are capable of reproducing different flood types. Observed flood events of the period 1980 - 2020 are classified according to the approaches of Fischer et al. (2019) and Tarasova et al. (2020). The conceptual rainfall-runoff model HBV-IWW is operated for 120 meso- and macroscale (30km² - 1500km²) catchments in Germany. Observed and simulated flood events are compared to assess the model performance separately for different flood types. In addition, the model performance is evaluated for the pre-event phase to determine whether the preconditions of the flood event are met.  The results indicate that the model reproduces certain flood types better than others. The model performance is in particular poor for snow dominated flood events both for the pre-event phase and during the flood event. The outcome of these investigations should help to find and improve the deficits of the hydrological modelling.