Based on the data since 1961,the ARMI-TR(auto-regressive integrated moving average and trend regression)model was used to predict the average yield and top yield per unit area of world rice before 2030.The actual yield per unit area in 2020 and 2021 was used to test the prediction reliability,and the GS(grey system)model was used to verify the prediction validity.The effects of global mean temperature and land precipitation on average yield and top yield of rice in the world were analyzed by unary regression models.The results showed that the RMSE(root mean square error)value of ARIMA(1,1,1)model was greater than that of TR quadratic model,and the RMSE value of ARIMA(1,1,2)model was greater than that of TR quadratic model.That is,the predicted value of the TR model could be used as the prediction result of the ARMI-TR model.In 2030,the average yield per unit area of world rice would reach 5 195 kg/hm2,and the highest yield per unit area would reach 10 269 kg/hm2,the former reaching 50.6％of the latter.The effect of global warming on rice yield per unit area is negative,and the negative effect on the average yield per unit area is less than the top yield per unit area.There was no obvious upward or downward trend of global land precipitation change,and the positive effect on the average yield per unit area was slightly higher than the top yield per unit area.The findings indicate that the gap between the average yield and the top yield of world rice promoted by climate change will gradually narrow.
NETL
NSTL
万方数据
