本文以1931∼1995年臺南及高雄地區5、6月梅雨季降水資料應用ARIMA入模式建立可行之定量降水預測模式。研究結顯示，適當的選用ARIMA模式可以有效的掌握臺灣南部地區梅雨季降水總量的年際變化趨勢，但在定量預報則因受基本統計方法影響，其極端振幅的幅度均較實測值為低，應用時應予適當的調整。
based on the May-June (Mai-Yu) precipitation data of 1931-1995 in cities of Taiwan and Kaoshoung, this paper applies and ARIMA intervention model to set up a feasible quantitative precipitation forecasting model. The paper shows that a carefully chosen ARIMA model can effectively predict the trend of annual change in precipitation during May and June in the Southern Taiwan. Neverthsless, due to the fact that the ARIMA model implies lower forecasted ranges of amplitude compared to actual data, proper adjustments would be needed when used for application.
