Reducing Error of Rainfall-Runoff Simulation Using Coupled Hydrological SWAT Model and Data Assimilation Technique

Modeling conceptual rainfall-runoff procedure involves large number of parameters and climate data. Uncertainty in these input parameters are very likely which lead to output errors as well as impractical prediction of long-term impact of management policies. In this study Soil and Water Assessment Tool (SWAT) is implemented to simulate rainfall-runoff process in Chelgerd sub-basin. To develop appropriate model with acceptable and reliable performance, Ensemble Kalman filter (EnKF) as data assimilation technique is used to assimilate the variables of model which are known as sources of error product; these sources include model parameters and input data. The paper in concluded that EnKF as a data assimilation technique is capable of reducing the computational error inherited in the simulation model. Results of proposed model is evaluated by Nash-Sutcliff (NS) factor with value of 0.86 which have better performance than modeling without EnKF technique. Also developed model performance is improved with NS value of 0.82 for validation period.