Coupling duo-assimilation to hydrological model to enhance flood forecasting

Model parameters play important roles in the accuracy of model results, especially parameters characterizing initial states of the basin need to be updated regularly with the most recent ground station data. This study applied ensemble the Kalman filter (EnKF) coupled with the HYdrological MODel (HyMOD) hydrological model to identify the optimal parameterization for flood forecasting and provided assessments of its impacts on the Chu River Basin of Vietnam. The results showed significant improvements with the application of EnKF techniques, however the parameterization depends on types and characteristics of flood events. The parameters were converged by the end of the simulated time periods to the most accurate model parameterization. As a result, the Nash-Sutciffe coefficient was 0.9 for EnKF, whereas it was only 0.3 for non-EnKF; mean absolute bias of 70 m(3)/s in applied-EnKF compared with 400 m(3)/s in non-EnKF, the Brier Score and spread values in applied-EnKF are much closer to zero than that in non-EnKF.