Dynamic Prediction of Cotton Regional Yield Based on the COSIM Model - A Case Study of Akesu City,Xinjiang

Objective] Dynamic prediction of crop yield using a crop growth simulation model is the focus of increasing research attention. Method] Based on meteorological, cotton yield, and cotton phenology data recorded at Akesu in Xinjiang from 1991 to 2014, this study aimed to improve the accuracy of crop yield prediction by the COSIM model. The average sowing date for each study year, as well as multiple sowing dates during the suitable sowing period, was imported into the COSIM model, and the two yield prediction methods were compared and analyzed. Result] The accuracy of both yield prediction methods was higher than 90.0%, indicating that the two methods showed good applicability at Akesu. However, the method using multiple sowing dates during the suitable sowing period showed higher prediction accuracy when cotton yield was dynamically predicted in each month and the actual sowing date was uncertain. Conclusion] The two prediction methods based on the crop growth simulation model are suitable for prediction of cotton yield at Akesu. In addition, according to the characteristics of different forecast years, the appropriate forecasting method can be used to improve the accuracy of prediction. The results also provide a reference for dynamic prediction of cotton yield in other cotton-producing areas.