Investigation of the uncertainty of hydrological modelling for climate change impact assessment

Besides the uncertainties of input data and model parameters, the uncertainties of process representations within a hydrological model are an additional source of error when climate change impact is assessed. Different evapotranspiration (ET) submodels included one-by-one in a rainfall-runoff model were investigated. A split sampling type approach was used to test the reaction of the models to climate signals in the past. By this, the transferability of the models with respect to their performance in different climatic conditions was investigated. The application of the methodology shows that a great number of catchments with different land uses has to be considered to find the appropriate candidates for modelling ET in hydrological models for climate change scenarios.