Irrigation Scheduling Optimization for Cotton Based on the AquaCrop Model

• DOI: 10.1007/s11269-018-2087-1
• Publication Date: 2018-10-25T03:47:06Z
• Authors: Fawen Li; Dong Yu; Yong Zhao
• Keywords: Irrigation scheduling; Goodness of fit; Mean absolute error

To improve irrigation efficiency, it is important to optimize agriculture irrigation scheduling. The objectives of this study were to evaluate the AquaCrop model for its ability to simulate cotton in the North China Plain and optimize irrigation strategies. The AquaCrop model was calibrated using 2002–2009 data and validated using 2010–2014 data. Root mean square error (RMSE), mean absolute error (MAE) and residual coefficient method (CRM) were used to test the model performance. The model calibrated for simulating cotton yield had a prediction error statistic RMSE of 0.152 t hm−2, MAE of 0.123 t hm−2 and CRM of 0.120. On validation, the RMSE was 0.147 t hm−2, MAE was 0.094 t hm−2 and CRM was 0.092. The goodness-of-fit values for the calibration and validation data sets indicated that the model could be used to simulate cotton yield. The analysis of irrigation scenarios indicated that the highest irrigation water productivity could be obtained by applying one irrigation at the seedling stage in a wet year, two irrigations, at the seedling and squaring stages, in a normal year and three irrigations, at the seedling, squaring and flowering stages, in a dry year. These results could be useful to the government in determining reasonable, well-timed irrigation for agricultural regions.