熱帶氣旋的強降雨常對台灣造成極大的災害，因此熱帶氣旋降雨潛勢預報對台灣地區相當重要。而利用衛星即時的觀測資料進行降雨潛勢預報則是目前國際上最常用的方式，如 TRaP (Tropical Rainfall Potential Technique)、eTRaP (Ensemble Tropical Rainfall Potential)及 I-TRaP (Improved Tropical Rainfall Potential Technique) 可用來預估颱風的累積降雨分佈，以掌握颱風可能造成的影響，達到防災及減災的目的。尤其 I-TRaP為 Liu et al. (2008)針對台灣地區歸納颱風降雨特性，進行降雨潛勢估算之修正，包含降雨分佈、總降雨量及環境風場之輻合效應，用來減少降雨潛勢估算的誤差，大幅提升台灣地區颱風降雨潛勢之準確度。 由於 I-TRaP 在颱風與環境風場輻合效應所歸納之個案相對較少，在部分極端降雨的颱風個案中(如2004Mindulle)，常有低估之情形，主要受環境風場輻合效應(西南氣流或東北季風)在總降雨量估算上的影響。為了修正I-TRaP在颱風降雨總量上的偏差，本研究選取受西南氣流與東北季風的案件，針對 I-TRaP 模式在潛勢的估算結果，配合再分析場資料 (European Centre for Medium-Range Weather Forecasts；ECMWF) 之風向與水氣含量，探討並修正颱風與環境風場的輻合/輻散效應對颱風降雨之影響。 結果顯示，輻合/輻散效應的強弱與水氣含量的多寡為造成預報降雨差異的主要原因，其中季風環流的區域測試也是本研究探討的重點之一，根據颱風歷史資料選取預報降雨差異大的區域作為修正總雨量的依據，針對這兩個參數進行修正後的颱風降雨潛勢預報，西南氣流個案之均方根差約有24% 之改善，而東北季風的個案則有22%，顯示配合水氣含量針對輻合效應在台灣地區颱風降雨修正之方法可行性很高，後續仍須更多個案分析，以建立較適用之修正模式。 ;Based on the retrieval of satellite microwave observations, a useful approach to estimate the typhoon rainfall potential (TRaP) has been proposed by Kidder et al. (2005). However, the uncertainty of TRaP may be induced by the topographic effect of the Central Mountain Range (CMR) in Taiwan, because only the simple shift-motion is considered along with the typhoon track. By considering the topographic effect, variation of typhoon intensity and convergence effect, the uncertainty has been much improved for the area around Taiwan Island (I-TRaP technique). However, the potential of typhoon rainfall is usually underestimated for the extreme precipitation cases which coupled with the strong monsoon (southwesterly or northeast monsoon) around Taiwan, that is to say the effect of convergence. Therefore, this study aims at the assessment of convergence effect on typhoon precipitation by means of I-TRaP technique associated with the European Centre for Medium-range Forecasts datasets. The results of case studies show that the uncertainty of I-TRaP model is caused by convergence effect. The sensitivity test of rainfall area is also important point of this study. This study used ECMWF datasets to construct the I-TRaP and estimate new rainfall patterns, and then were validated with by CWB automatic rainfall station observations. There is about 24% improvement in the RMSE of southwesterly flow influenced cases, while with 22% in the northeast monsoon influenced cases. It shows that I-TRaP technique can be significantly improved by considering the convergence effect and water vapor content in Taiwan area.
