廣大的熱帶洋面上有些雲簇會發展成熱帶氣旋，有些卻不會形成，為了 分析熱帶氣旋的發展與否，需藉由客觀的方法來分析發展與非發展系統間 的特性差異。本研究主要的目的為探討發展與非發展雲簇的海氣環境參數 進行系統性的分類，進而全面且完整的歸納兩者間環境條件差異，提供熱 帶氣旋生成預報之參考。 本研究利用西北太平洋2000~2009 年間的JTWC 颱風最佳路徑和 ECMWF 再分析資料，計算七項海氣環境參數的發展與非發展差異指數 (BDI)，使用參數包括：1）850hPa 相對渦度、2）850hPa－200hPa 垂直風切、 3）900hPa 輻合度、4）總降水量、5）850hPa 相對溼度、6）可感熱通量與 潛熱通量相加值、7）海表面溫度，計算並結合各熱帶氣旋生成閾值，計算 權重總分後轉換成熱帶氣旋發展機率形式呈現。 研究結果以2010~2012 年的獨立個案進行驗證，結果顯示發展個案有命 中率（Hit rate）達0.95 的高分表現，能夠準確判斷生成位置且平均提早JTWC 發布警報前49.9 小時達系統判斷發展機率超過60%，系統達到90%發展機 率時也有提早37.1 小時的表現；而若以60%發展機率為誤報門檻值，系統 的假警報率（False Alarm Ratio；FAR）為0.24。 In order to analyze the development of tropical cyclone, it is necessary to investigate the difference characteristic between the development and non-development systems by using objective methods. The main objective of this study is to discuss the environment parameters between the development and non-development cloud clusters which can be further processed systematic classification. Furthermore, it can fully and completely generalizes the differences between environmental conditions and provides as a reference to the tropical cyclone forecast. In this study, we analyzed the typhoon cases over the Northwest Pacific Ocean from 2000 to 2009 by the JTWC best track and ECMWF Interim Reanalysis Data. And we determine the environmental air-sea parameters’ thresholds by Box Difference Index (BDI). The environmental air-sea parameters include: 1) 850hPa relative vorticity, 2) Sea surface temperature, 3) 850hPa relative humidity, 4) 900hPa divergence, 5) 850-200hPa Vertical wind shear, 6) Thermal energy, 7) Total precipitation. The results adopt the weights method and show as probability of tropical cyclone development. This study results are verified by using the independent cases from 2010 to 2012 and show the good performance of the Hit rate up to 0.95. Moreover, these results can determine precisely the prediction accuracy reached above 60% and averagely 49.9 hours ahead of warning form JTWC. If the system set the development possibility about 60% as false alarm threshold, the False Alarm Ratio (FAR) of the systems are 0.24.
