Sensitivity Analysis of AquaCrop Model for Winter Wheat in Different Water Supply Conditions

Abstract

AquaCrop, a water-driven model, has been developed to simulate the response of crops, including wheat, to the amount of irrigation water. To estimate crop yield using this model, the calibration stage is applied first, employing the available data. Calibration accuracy guarantees the validation accuracy of this model. For this reason, before the calibration stage, the response of the AquaCrop model to changes in input parameters is investigated using sensitivity analysis. Most researchers use additive-subtractive methods. However, these methods do not provide much information about model sensitivity. In this research, three methods were used to analyze the sensitivity of AquaCrop to simulate winter wheat grain yield under different irrigation requirements. The methods included (1) an increasing-decreasing method; (2) a limit method; and (3) a Gamma test that was based on the nonlinear relationship between inputs and outputs. The irrigation treatments were 100%, 75%, 50%, and 0% of the irrigation requirement and were designated as I1, I2, I3, and I4. Six input parameters consisting of normalized water productivity (WP*), maximum crop coefficient for transpiration (KCTR), initial canopy cover (CCo), crop canopy growth coefficient (CGC), crop canopy decline coefficient (CDC) and harvest index (HI) were evaluated for sensitivity analysis. The results showed that the sensitivity of the AquaCrop model was extremely high to WP* changes and moderate to CCo changes. An inverse relationship between wheat grain yield and CDC and a direct relationship between wheat grain yield and other input parameters were observed. The sensitivity of the AquaCrop model to the CCo parameter was the same in all irrigation treatments. The increase in water stress decreased the sensitivity of the AquaCrop model to the input parameters. Therefore, in the case of large differences between simulated and observed grain yield, it is suggested to change WP* and Kctr values. In the condition of moderate difference, it is better to change two parameters, HI and CDC. To reduce the slight difference between the simulated and observed grain yield, it is suggested to change the two parameters, CGC and CCo. It should be noted that the results of the sensitivity analysis are specific to the experimental conditions, such as plant density, soil texture, and water supply, and may vary when applied to different regions. Therefore, it is recommended to obtain region-specific results and determine the sensitivity of the AquaCrop model to input parameters.