Kidder et al.(2005) 提出 TRaP 方法，將衛星反演之降雨分佈平移，迅速估算熱帶氣旋未來可能帶來的強降雨，但TRaP僅適用於地形較平坦之區域。陳(2010)修正TRaP方法，考量地形效應的影響，並根據測站降雨資料重新分配估算颱風降雨，以改善TRaP 方法之結果，稱為I-TRaP。 分析I-TRaP對2012年颱風個案的預報結果，發現颱風強度的改變、預報路徑的誤差、衛星反演雨量的準確性會造成I-TRaP預報結果的誤差，其中以衛星反演颱風完整雨帶影響最大。本研究使用多衛星反演降雨產品GSMaP進行I-TRaP颱風降雨潛勢預報，以便改善微波觀測颱風不完整時所造成的誤差。此外利用GSMaP逐時降雨資料統計2000到2011年所有颱風的降雨強度趨勢時，發現颱風登陸台灣後，因受地形影響，強度會逐漸減弱，颱風中心降雨強度也會逐漸降低，因此根據中央氣象局路徑分類統計各路徑在南、北台灣時的中心降雨強度趨勢，進行總降雨量之修正，期改善I-TRaP的預報能力。 將GSMaP資料放入I-TRaP估算2000~2012年間81個颱風個案之24小時累積降雨，並和測站觀測資料比較，對於SSM/I、SSMIS觀測缺乏完整雨帶的颱風個案，使用GSMaP資料相關係數由0.67提升至0.71，在Mean Error的部分由-17.23mm改善至-15.53mm，整體均方根誤差由81.89mm降低至61.66mm，減少24%的均方根誤差，有明顯的改善。而在降雨的趨勢方面，將GSMaP逐時降雨資料所獲之颱風降雨強度趨勢，加入I-TRaP颱風降雨潛勢中，整體結果並無明顯之改善，主要是由於強烈颱風之風速較強，因地形輻合及抬升造成大量的降雨，使I-TRaP原本就低估降雨，加入降雨強度趨後有更大的誤差。整體而言，結果顯示應用GSMaP全球降雨資料進行I-TRaP降雨預報，能提升SSM/I及SSMIS未觀測到颱風完整雨帶個案之預報精準度。 Analysis shows that the accuracy of the rainfall potentials from the Improved Tropical Rainfall Potential (I-TRaP) technique are influenced by the changes of typhoon intensity, the accuracy of the predicted typhoon paths, the satellite-retrieved rainfall algorithms, and so on. One of the non-negligible impacts is from the integrity of typhoon rain bands. Whereas many microwave radiometer images from the polar orbiting platforms (such as DMSP satellites) are often lack of data in some areas and periods because of the satellites’ orbits and revisiting rates. That poses a difficulty when using those microwave observation data to estimate a typhoon’s rainfall rate. In the other hand, the analysis result from the hourly GSMaP data shows that the mountainous terrains of Taiwan Island often cause typhoons’ weakening and further rainfall forecasting errors. In this study, the Global Satellite Mapping of Precipitation (GSMaP) , a temporal-and-spatial-continuous precipitation data set from integrated microwave and infrared images , are used to improve I-TRaP for better typhoon rainfall predications over Taiwan. The results suggest that there is a better agreement to station-measured rainfalls than TRaP and I-TRaP results did. It gives us an opportunity to use the combined microwave and infrared data for getting more accurate typhoon rainfall predications in the future. Moreover, the accuracy of the forecasting rainfall rates can be further handled and improved with considering the change of typhoon intensity. Then, use the long-term trends to correct the I-TRaP-derived rainfall rates. For instance, due to the typhoon circulations for stronger typhoons generally interact with the mountains significantly and trend to bring more topographic rainfall, and then weaken themselves, making I-TRaP under estimate the rainfall by stronger typhoons. With the historical data, such rainfall estimation errors due to the topographic effect can be reduced again.
