本研究將1950至2005年間之183個颱風依侵台近中心最低氣壓、侵台近中心最大風速、全程近中心最大風速、逼近台灣時移動速度、路徑分類及雨量等屬性,建立起資料庫,依序以案例式推理法推估各颱風雨量,將推估雨量與觀測雨量比較,求其相對誤差。 文中以歐幾里得距離法及權重法,作為案例式推理求相似度的準則。以歐幾里得距離法求相似度時,所得推估雨量與觀測雨量之相對誤差總平均為38.52%;以權重法求相似度時,所得推估雨量與觀測雨量之相對誤差總平均為34.37%。若取兩方法所得之推估雨量平均值與觀測雨量作比較,其誤差總平均為28.29%。 本研究結果可當作未來颱風降雨預測之參考。
In this study the 183 typhoons’ category data from 1950 to 2005 were set up in an information file with the minimum atmospheric pressure and the maximum wind velocity in the nearby of the typhoon center, the moving velocity of the typhoon during landing Taiwan, rainfall quantity, typhoon paths classification etc. The Case-Based Reasoning method was used to estimate the rainfalls of typhoon and then compared the estimated rainfalls with the observed rainfalls to find the relative er-rors. The Euclidean distance and weighted methods were considered as the criteria for the Case-Based Reasoning to get the ways of similarity. When the Euclidean distance method was used to find the similarity, the total average error between the estimated rainfalls and the observed rainfalls is 38.52%. When the weighted method was used to find the similarity, the total average error between the estimated rainfalls and the ob-served rainfalls is 34.37%. When the average estimated rainfall of the two methods was compared with the observed rainfall, the total average error is 28.29%. The results of this research can be used as a reference for typhoon rainfalls pre-diction.