Multipurpose Calibration of SWAT Model in Estimating Runoff, Evapotranspiration, and Crop Yield (A Case Study: Mahabad Plain)

In most studies, the SWAT model is calibrated based on surface runoff in hydrometric stations, and the evapotranspiration component, which has a great impact on the estimation of balance components, is not considered or less considered in the calibration process. Therefore, the aim of this study was to estimate the yield and evapotranspiration values of major crops in Mahabad plain in West Azerbaijan province, through multivariate calibration of SWAT model. Calibration and validation of the model based on the data of a hydrometric station (Gerdyaghoub station located at the exit of the plain), and crop yield and evapotranspiration of the major crops of the region including wheat, barley, corn, sugar beet, alfalfa, apple, and grape. A six year calibration period followed by a four-year validation period was used. In the multi-purpose calibration process, the coefficients of 1, 0.9, and 0.8, respectively, for hydrometric data, yield, and evapotranspiration were considered in the objective function of SWAT-CUP software and SUFI2 method. In general, the sensitivity analysis of the model led to the selection of 43 of the most sensitive parameters according to their sensitivity to river discharge, crop yield and evapotranspiration. Also, the heat unit potential required for full plant growth and Leaf area index (LAI) were the most important parameters to simulate the yield and evapotranspiration crops in the model calibration process. The results showed that the SWAT model was able to simulate the surface flow in the periods of calibration (NSE = 0.85; R2 = 0.87), and validation (NSE = 0.92; R2 = 0.89), as well. The model has high uncertainty in surface flow simulation, especially over-estimation in low flow during calibration period (PBIAS = 4.5%). In evaluating the performance of the SWAT model in estimating actual evapotranspiration, these values ​​were compared with the output of the WaPOR remote sensing model. The results showed that the RMSE criteria varies between 13.8 to 66.4 mm in the calibration period, and between 23.9 to 53.2 mm in the validation period. At the same time, the estimated actual evapotranspiration values ​​of the SWAT model were compared with the estimated crop evapotranspiration of the CROPWAT model. In this study, RMSE criteria in the calibration period between 24.4 (3.6%) to 49.4 (6.5%) mm, and in the validation period between 21.7 (6.2%) to 4.4 53 (7.2%) mm were obtained for different crops. PBIAS values ​​showed that the estimate of SWAT model evapotranspiration in both calibration and validation periods was lower than the estimate of CROPWAT crop evapotranspiration. In general, for most crops, the evapotranspiration values ​​obtained from the SWAT and WAPOR models are far less than the evapotranspiration values ​​estimated by the CROPWAT model that can indicate deficit-irrigation. Crops yield performance estimates were compared by SWAT model with crop yield performance values ​​reported in the region. The results showed a relative estimate of the model for crop performance, while the SWAT model somewhat over-estimated crop performance. The present study showed that multifunctional calibration of large-scale hydrological model SWAT in an irrigated area in addition to using the usual river flow index, taking into account additional parameters such as evapotranspiration and crop yield can increase the accuracy of model results.