Calibration and Validation of AquaCrop Model for Simulation of Spring Wheat Yield under Simultaneous Salinity and Water Stress

Crop growth simulation models have been developed for predicting the effects of water and salinity on grain and biomass yields and water productivity of different crops. These models are calibrated and validated for different regions using the data generated from field. This study was carried in Mashhad for two years (2010 and 2011) in order to evaluate the AquaCrop model under simultaneous salinity and water stress. Calibration was done using the data of 2009-2010 and validation with the data of 2010-2011.Results indicated that AquaCrop successfully simulated yield, biomass, water productivity, harvest index, soil moisture and salinity profiles for spring wheat under salinity and water-limiting treatments with high accuracy, although simulation of harvest index  and soil salinity profiles were less accurate. Average value of normalized root mean square error (NRMSE), maximum error (ME), index of agreement (d), coefficient of the residual mass (CRM) and coefficient of determination (R²) in both the calibration and verification were 13.3 %, 36.1 %, 0.95, -0.072, and 0.87, respectively, for grain yield, while these measures were 12.59%, 34.46%, 0.92, 0.057, and 0.77, respectively, for biomass. Also, value of NRMSE, ME, d and CRM were 11.84 %, 25.72 %, 0.93, and 0.032, respectively, for soil moisture, while these measures were 26.25%, 58.5%, 0.91 and -0.12, respectively, for soil salinity. Sensitivity analysis revealed that crop transpiration coefficient (K_C-Tr), normalized crop water productivity (WP^*), reference harvest index (HI_O), volumetric water content at field capacity, soil water content at saturation[S1] , and air temperature were the most sensitive parameters. Although the accuracy of the model simulation decreased with increasing salinity and water stress, AquaCrop can be a valuable model for simulating spring wheat yield and soil water content and salinity in Mashhad region, because the model requires few input data which can be readily available or easily collected.



 [S1]This is probably”initial conditions” and not saturation.