本研究主要目的在於探討非線性時間序列模式之隨機係數自迴歸模式,簡稱RCA模式,並應用於台灣地區月流量資料,研析其統計特性之保有性及預測能力。研究結果顯示,當月流量資料經除勢後,RCA模式無論統計特性之保有性與預測能力均較AR模式為佳。 The major objectives of this study is to investigate the random coefficient autoregressive model of the nonlinear time series. The performance of the preservation of the characteristics of time series and the forecasting ability are studied by using the monthly riverflow data in Taiwan area. The results indicate that RCA model is superior to AR model, when the monthly data are detrended.