最优气候值法与ARIMA模型在气温预报中比较研究

he paper applies optimal climate normal and ARIMA model to separately forecast the average temperature of July from 2001---2004 of YanAN according to the date of YanAn forty years (1961---2000). The result shows that both optimal climate normal method and ARIMA model perform well in temperature prediction, especially compared to the ARIMA model, the optimal climate normal method had higher prediction accuracy. The max absolute error of the optimal climate normal method is just only 1.800℃, but the max absolute error of the ARIMA model is 2.317 ℃, however, the ARIMA model was better for extreme temperature prediction. Except for the objective limitation of forecasting method, choosing different areas and different periods of time might result in prediction consequences. That is to say, for a different time and space point, different forecasting methods would produce different forecasting results, which required to adopt a pertinent forecasting method in order to obtain the best prediction.