近年來微波頻道解析度提昇及其可透雲性，對於反演颱風降雨率之準確性有不錯的結果。Kidder et al.(2005)利用微波頻道優點發展一個快速實用方法估算降雨。Kidder et al.(2005)利用SSM/I、TMI及AMSU等被動式微波資料估算熱帶氣旋降雨，參考官方預報熱帶氣旋路徑資料，利用平移方法預測24小時後熱帶降水潛勢(Tropical Rainfall Potential，TRaP)，但此方法沒有考慮熱帶氣旋旋轉所造成的降雨空間分布改變，且忽略了熱帶氣旋登陸後強度衰減產生降雨變化的事實。 Kaplan and DeMaria (1995)發展出一套簡易模式以exponential decay equation關係來推估熱帶氣旋登陸平原後強度變化，本研究進ㄧ步應用上述關係式並加以改進成可預測颱風登陸後有地形影響之強度變化。研究結果顯示:登陸平原颱風個案，估算強度之平均絕對誤差為4.48kt ;登陸丘陵地形颱風個案，估算強度之平均絕對誤差為6.73kt。本研究使用SSM/I衛星觀測資料反演降雨率，推估1990~2004年期間西北太平洋颱風登陸中國大陸華南地區陸地後的強度變化，再加入Kidder et al.(2005)考慮颱風平移過程與Liu et al.(2008)考慮颱風旋轉對降雨造成的影響，以預測未來6小時的累積降雨，並以中國大陸地面測站降雨資訊做驗證。研究結果發現:只考慮颱風平移時，其6小時累積降雨和測站觀測資料的相關係數為0.68，若同時考慮颱風平移、旋轉及強度變化相關係數增加為0.75，故考慮颱風的旋轉及強度變化有助於提昇降雨估算之準確度。 In the past decades, microwave sensors have become an important tool in typhoon rainfall monitoring due to the cloud transparency capacity and resolution improvement. Kidder et al. (2005) used passive microwave data, such as SSM/I, TMI and AMSU, to predict the 24-hour later tropical cyclone rainfall potential (TRaP) via a simple cyclone shift-motion assumption, along with the official cyclone track prediction results. However, their method didn’t take into account the cyclone intensity and cyclone rotation changes, which could alter the rainfall spatial patterns . Based on the TRaP method, this research factors in the cyclone intensity changes after making landfall via an exponential decay equation(Kaplan and DeMaria, 1995). The intensity estimations reveal that the mean absolute errors (MAE) are 4.48kt and 6.73kt for plain and hill regions, respectively. In addition, the cyclone rotation effect on the rainfall patterns is considered (Liu et al., 2008), as well. The landfalling tropical cyclones in the southeastern China between 1990 to 2004 were analyzed, where their respective 6-hour later rainfall rates were predicted. The projected rainfall was verified by ground weather station data. Results reveal that the correlation is 0.68 when the shift-motions are only considered, and is 0.75 when the tropical cyclones’ intensity and rotation changes are also taken into account, it can further improve the TRaP method accuracy.
