A Procedure for State Updating of SWAT-Based Distributed Hydrological Model for Operational Runoff Forecasting

The foundation of operational management of river flows (regarding water resources exploitation, hydropower potential use, flood management, etc.) is accurate and reliable runoff forecast. The systems for operational management support are based on hydrological/hydrodynamic simulation model that uses data on up-to-date states of catchment and prognostic values of model forcing (e.g. precipitation) for prediction of water levels and flows in streams and reservoirs. The best estimate of model forcing (temperature, quantity and distribution of precipitation, etc.) derived from ground measurements and remote sensing is crucial for success of simulation model use. This paper discusses the application of "data assimilation" procedure for updating simulation model state variables using mainly model forcing data. The term itself relates to a set of mathematical methods designed for use of measured data and differences between simulated and measured values in up-to-date model state estimation and forecast of future states in physical systems. These methods allow for better parameterization, structural and sensitivity analysis of the model, as well as design of more efficient observation networks and improving forecasts and quantitative measures of their reliability. The paper presents an approach to operational use of SWAT-based rainfall/runoff hydrological model, using an algorithm for rainfall distribution estimation based on data from few automated weather stations. The implementation of presented algorithm in "Drina" hydroinformation system confirms the sensitivity of the simulation model to forcing data, and provides improvements of the runoff forecasting, bringing the mentioned model into operational use.