Author(s): Peter Godiksen, Marc-Etienne Ridler, Henrik Madsen
Linked Author(s): Henrik Madsen
Keywords: Data assimilation, Ensemble Transform Kalman Filter, hydrological modeling, water management
Abstract: Hydrodynamic-hydrological modeling is affected by various sources of uncertainty, which degrades its performance and predictive capabilities for use in different applications such as water resources management, flood forecasting, real time control and etc. Combining observations with model predictions using data assimilation can improve model prediction skills. State-of-the-art ensemble based Kalman filter algorithms are implemented for hydrodynamic-hydrological data assimilation in the MIKE HYDRO River modeling system. The focus of the paper is on data assimilation in the hydrological model, which is evaluated using a case study from Murrumbidgee River in New South Wales, Australia. It is demonstrated that the hydrological model states are updated consistently and the state updates benefit from the hydrological memory of the catchment for increasing forecast lead time. The rainfall-runoff is routed through a hydrodynamic river model and the combined improvement of rainfall-runoff estimates from upstream catchments leads to an improved estimate of the river discharge further downstream, and thus the predictive capabilities of the whole modelling system are increased
Year: 2017