Interaction between Input and Model Uncertainty for Distributed Rainfall-Runoff-Models

Abstract

The work at hand deals with the uncertainty in the hydrological modeling. The aim of this work is to implement more than one uncertainty analysis method (MCMC, GLUE and a new developed concept INPUT) and to compare them. The interaction between input (different spatial resolutions and/or rainfall uncertainty) and model (WaSiM-ETH) uncertainties for distributed rainfall-runoff models as well as the effect of the catchment characteristics and size on the interaction have been studied. The new concept (INPUT) was developed to incorporate input uncertainty in the total uncertainty analysis procedure. Three different catchments were chosen, the Weise Elster catchment till gauge Gera-Langenberg, the Weida catchment and the Dollnitz catchment. Regardless of the uncertainty analysis method used, similar results (Nash Sutcliffe and water balance) were obtained for all catchments. Using the MCMC with autocorrelation improved the overall results, and precisely the water balance. The reason for this is the fact that the modeling error in previous time step is removed from the next time step. The disadvantage of this method is that it cannot be used to predict future events. The new concept INPUT allows studying the interaction between rainfall point measurement uncertainty and the parameter uncertainty. The INPUT method promises new insight in the interaction between input and model uncertainties. Assumptions made about the form of the input uncertainty have a large influence on the uncertainty analysis results. Furthermore no considerable difference between the spatial resolutions for each catchment was observed. One downside of such methods is the difficulty of distinguishing and separating the diverse sources of uncertainty. For the rainfall measurement uncertainty a different behavior is observed. For the smallest catchment (Weida), the rainfall uncertainty plays an important role, on the contrary to the Gera-Langenberg (largest catchment).