| 摘要: | 隨著衛星觀測技術與解析度的進步，併隨新降雨反演方法的開發，衛星降雨資料越來越具多元及應用潛力，有系統地校驗個種多元衛星降雨資料能更瞭解資料的特性與精度，也能拓廣其應用，例如用於熱帶降雨潛勢的雨量疊加及系集，以估算颱風降雨潛勢。由於傳統熱帶降雨潛勢模式基於過度簡化假設，易導致可觀誤差及限制其實用性，特別是如台灣此類地形高聳地區，颱風降雨強烈受到地形影響，更為細緻的基礎假設對改進潛勢模式是必需的。Liu et al. (2012)主要將地形效應及颱風雨帶旋轉等效應納入假設，大幅提昇模式合理性，所得颱風降雨分佈也更貼真實降雨。隨著衛星觀測資料的多元化，又進一步提供改良式熱帶降雨潛勢模式更多驗證機會及再改進可能，此亦即本兩年期計畫目的，主要工作重點在於校驗及改進衛星微波降雨反演方法，以資料融合出發，建立時間解析度更高且更適用於台灣及周遭地區的多衛星颱風降雨系集預報，以增進改良式熱帶降雨潛勢模式的精度及實用性。 第一年計畫的工作項目主要於多元衛星微波影像的降雨校驗，探討比較數種常用微波降雨反演機制理論及演算法，其中主要涉入降雨放射/散射機制、空間解析力及客觀分析內差方法影響、不同颱風類型、及颱風雨帶內/外圈等降雨型態差異，與選用不同反演法的優劣分析，以找出不同觀測資料對應的較佳/最佳反演法，據而進一步建立多元衛星熱帶降雨潛勢系集模式；第二年工作著重量化改進多元衛星資料空間及時間融合的最佳權重組合選擇。另外，颱風隨時間前置的降雨變化率及路徑偏移變化，特別是地形效應重分配因子的調整及改進，將是系集模式導入改良式熱帶降雨潛勢模式過程中另一探討重點。 ;The traditional Tropical Rainfall Potential (TRaP) is rather intuitive and easy for operating. With satellite-retrieved rainfall and a time-lagging superimposing process, the typhoon rainfall potentials can be obtained straightforwardly. However, unsuited errors could be induced also due to its oversimplified assumptions and that thus weakens its robustness, especially for mountainous region where the terrain effect always is strong. Definitely finer assumptions are necessary for the improvements of a further TRaP model. Liu et al. (2012，2015) consequently put the terrain effect, systematic errors in rainfall retrievals, and typhoon rainfall band rotations into account to improve the TRaP base. Their work revealed that not only the improved model have a better reality approximation to in theoretical base, but also in real application results. Following the more satellites and sensors were launched in the few recent years, meaning that there are more opportunities to verify and improve the TRaP-like models. That is here the goal for this two-year project. The project is designed to verify and improved the rainfall estimation algorithms mainly with satellite microwave imagery, and the data fusion perspective will be emphasize, to develop a higher time-resolution satellite rainfall ensemble forecast system with multi-satellite observation for the improved TRaP, and for Taiwan area.The first year of the project will gather multiple and newer satellite data, and pay attention to investigate and compare several main rainfall retrieved algorithms, involving the different rainfall conditions versus their appropriated or optimal algorithms. For instance, such as emitting/scattering rainfall processes, statistical/physical rainfall retrieval algorithms, typhoon category types, typhoon inner/outer core rainfall types. Then the result will be integrated into the improved TRaP; the second year will keep gathering data. The ensemble weightings for multiple satellite data and in different time resolutions will be optimized if possible. Meanwhile, the possible change in rainfall intensity and typhoon paths over time will be examined to understand their sensitivities and estimate their potential errors. Meanwhile, the redistribution factor for the terrain-induced rainfall amount will be another very important research topic in this task. |
