Improving River Discharge Forecasting With the Hymod Conceptual Rainfall-Runoff Model Using Data Assimilation

Predicting discharge prediction through modeling is inherently associated with important uncertainties.Then uncertainty in hydrological modeling is mostly reduced by increasing the quality of inputs, improving structure of models, and data assimilation. Even if we assume that the physical structure of the model is perfect, uncertainties in parameters, forcing variables and initial conditions will be reflected in the simulation results through complex error propagations. One of the actions that can be taken toward reducing uncertainty in hydrologic predictions is data assimilation. It provide a superior hydrologic state estimate by considering input and observation uncertainties. In the current study, the efficiency of assimilating stream-flow into a hydrologic model using the Ensemble Kalman Filter (EnKF) in the Roudak catchment is investigated. Four evaluation criteria including NSE, KG,LNSE, DCpeak are applied to estimate the predictive performance of results. Results show that EnKF improved estimated stream-flow compared to an offline calibration with SCE-UA as NSE, KG,LNSE, DCpeak are increased by 13%, 5%, 17% and %94 respectively. Also one-day ahead prediction of stream-flow could be estimated by acceptable accuracy.