3种时间序列分析模型在水稻稻瘟病预测中的应用

Application of time series analysis in the prediction of rice blast

【目的】研究3种时间序列分析模型对水稻稻瘟病的预测效果,为该病害的预测提供参考。【方法】以1980-2007年四川省剑阁县水稻种植面积和稻瘟发病面积为原始数据,分别采用时间序列分析中的滑动平均、指数平均和方差分析周期外推法建立水稻稻瘟病发病情况的预测模型,分析其预测效果,并在2008年的稻瘟病预测中进行了验证应用。【结果】滑动平均、指数平均和方差分析周期外推法3种时间序列线性模型均能较好地预测水稻稻瘟病的发病趋势,1980-2007年其预测值与实测值的拟合度分别为97.7%,96.1%和99.8%;2008年预测值与实测值的相对误差分别为14.5%,18.1%和8.4%。【结论】3种时间序列分析模型均可用于水稻稻瘟病的长期预测。

【Objective】 The prediction effect of rice blast was studied to provide reference to control rice blast by time series analysis model.【Method】 Time series analysis model was established through moving average,exponential average and period extrapolated method by variance analysis,according to the incidence of plant areas of rice and apperance of rice blast in Jiange district Sichuan province from 1980 to 2007.The forecast result was analysed and used in the appearance predication of rice blast in 2008.【Result】 It was shown that the time series analysis linear model including moving average,exponential average and period extrapolated method by variance analysis can preferably fit the trend of rice blast.Comparing the measured value with the estimated value,their accuracy of testing-forecast from 1980 to 2007 was 97.7%,96.1% and 99.8%,respectively;relative errors in 2008 were 14.5%,18.1% and 8.4%,respectively.【Conclusion】 Time series analysis model can be applied in a long-term forecasting for rice blast.