Project objective

The project aims to increase the accuracy and to estimate the uncertainty of forecasts by utilizing all available information through the joint use of hydrological models with nonlinear time series models and data driven models from hydro-informatics.

Research efforts will be directed toward proposing a hybrid modelling framework for flow forecasting with different time steps for typical application scenarios in water resources management. This framework will be based on the combination of deterministic and stochastic models or deterministic and data driven models with the aim to make the best use of available domain and process physics knowledge in hydrological rainfall-runoff models and information gained from the study of the uncertainties associated with the propagation of model input errors and forecasting errors by data dictated tools and nonlinear and nonstationary time series models.

For the deterministic part of the hybrid modelling system the parameterizations of the rainfall-runoff and flow routing models developed in previous projects of the research team will be studied with respect to their influence on the modelling errors and error propagation. Parameterizations for ungauged sites will be also sought. The joint use of deterministic rainfall-runoff and flow routing models with nonlinear time series models and/or data driven models from the field of hydroinfomatics will utilize all available information for making forecasts more accurate and will also enable to supply information on the uncertainty of the forecast. It is expected, that guidelines will be developed for a hydrologically based selection and parameterization of data driven methods (e.g. with respect to regime switching) and that the methods will be developed, which will be transferable to poorly gauged basins.