在過去針對定量降雨預報的數值模擬實驗中，我們知道對於降雨事件之總降雨量具有一定程度 的預報能力，但是對於降雨事件在時間與空間上分布的預報能力卻十分欠缺。在許多的可能造成預 報誤差的機制中，關於不同天氣系統所具有的降雨特性差異卻從未被深入研究。也造成過去的降雨 數值預報校驗或模式物理過程之微調時，多是利用單一型態的降雨個案作為校驗與調整的基準，而 沒有考慮不同天氣型態的降雨特性差異。而這樣的差異，其實是由於綜觀尺度與大尺度的大氣系統 與導致極端降雨現象的中小尺度對流系統的交互作用所導致。所以如何系統性的分析與研究不同季 節、不同天氣型態的降雨特性，並利用這些研究資訊進一步針對搭配可以解析多重時空尺度交互作 用之降雨過程的數值模式進行校驗與分析，便是目前改進降雨預報技術的重要課題之一。 本研究依照研究工具與課題的不同，區分為兩個主要的方向。其中一個研究的主軸是利用台灣 地區長時間的觀測資料與分析資訊，搭配長時間大尺度的數值模式再分析資料，針對台灣地區不同 天氣型態下之季節性極端降雨事件的時空分布特性，進行系統性的統計分析。研究過程中，我們將 會建立主觀天氣分析檢索系統與客觀天氣系統分類技術。另一方面，我們將借助於最新的非靜力高 解析度全球模式，利用數值模擬進一步探討多重時空尺度下，極端降雨事件的發生機制。利用理想 化數值實驗，探討不同天氣系統環境下降雨之機率密度函數。而這兩方面的研究成果，亦可以相互 驗證與支援，作為改善降雨之數值模擬能力或降雨預報技術的重要參考指標。 According to the experience for a long term systematic evaluations of numerical quantitative precipitation forecasts (QPF), it has some skill of the rainfall amount for an individual event. It also shows the spatial and temporal distribution of precipitation could not be proper performed by numerical models. This kind of problem may be caused by the differences of rainfall characteristics in various weather systems, which relate to the atmospheric multi-scale interactions and was never be deeply studied before. Therefore, the systematic analysis to study the rainfall characteristics in various seasonal weather systems is one of the important issues to improve the QPF. This project will focus on two major research topics. The first part is combining the advantages of the large scale numerical re-analysis outputs and a long term observational data in Taiwan. We will study the extreme rainfall characteristics in various seasonal weather systems via a systematic statistical analysis. We will establish a subjective weather pattern retrieval system, and will also developing an objective weather system classification technique. The other hand, this project will also use the newest high resolution non-hydrostatic global model to study the interactions of multi-scale weather systems which are associated with extreme rainfall events. The idealized numerical experiments will be used to discuss the probability density function (PDF) of various weather systems. The research results of these two parts can be verified and support by each other. It will help us to improve the QPF skill of numerical models and also can be a guidance of extreme rainfall forecasts.
