Data Assimilation for Calibration-Prediction using SWAT Model

This paper deals with parameter estimation of SWAT model by means of streamflow data assimilation and application of calibrated model for hydrological simulation of Mahabad River which leads to Urmia Lake. Data assimilation algorithem is compared with SUFI2 algorithem. SUFI2 is an uncertainty-based optimization method first developed for auto-calibration of environmental and water resource models and due to availablity in SWAT-CUP package is usually used for calibration of SWAT. To illustrate capabilities of data assimilation for calibration of the model and prediction of the river discharge, Ensemble Kalman Filter (EnKF) is utilized in a joint state-parameter estimation framework. Both coding EnKF and calling SWAT is done in MATLAB environment. Results show joint state-parameter estimation using EnKF for SWAT, lead to improvement of accuracy of simulation and prediction of Mahabad River’s monthly discharge at Bitass hydrometery gauge compared to parameter estimation of the model using SUFI2.