Abstract: Based on the data of rice yield, the 10 day data of average air temperature, precipitation, and average sunshine hours in rice’s growth period from 1986 to 2009 in Chongqing, a dynamic prediction model was established for rice yield prediction by using statistic method. By the 5 year moving average method, the rice yield trend is isolated. Based on correlations between the meteorological yield of rice and the ten day meteorological elements in growth period, the dynamic prediction model was established for rice yield by using the key meteorological factors of midMarch mean air temperature, lateMay sunshine hours, mid July mean air temperature, mid July precipitation, mid July sunshine hours, late July mean air temperature, late July precipitation, early August mean air temperature. Rice yield was predicted dynamically with the model. Simulation with the model from 1986 to 2009 showed that average accuracy was more than 96%, and the accuracy was more than 90% in 95% of annual predictions.Prediction with the model for 2010 showed that the accuracy was 91.5% to 92.8%. The dynamic prediction model could be applied to quantitative prediction of rice yield, which could basically meet the demand for business services.

Key words: Mid season rice, Key meteorological factors, Meteorological yield, Dynamic prediction, Accuracy