Investigation of the Impact of Precipitation Uncertainty on Spatially Distributed Hydrological Modelling

Hydrological models account for the storage and flow of water and are tools for runoff calculation and can be used for forecasting. Physically-base spatially distributed models are capable to use high-resolution input data. However, the high uncertainties of input and catchment properties introduce a substantial amount of error into the model predictions. With a distributed hydrological model SHETRAN applied to the upper Neckar catchment in South-West Germany, different spatial and temporal resolutions of input data were used. Higher resolution input data do not necessarily produce better predictions. The relationship between cumulative portion of overall modeling error and the number of modeling days shows that errors occurring on very few days contribute to more than 80% of the total error. Simulations considering precipitation uncertainty can be used to quantify the uncertainty of meteorological and hydrological data and can lead to more efficient models. Precipitation data which are simulated using Random Mixing approach fulfill the statistical properties of observed data were used as inputs and aided for the improvement of model.