依據106 年度科技部自然司防災科技學門專題研究計畫課題重點說明(氣象領域，學門代碼：M1710)之研究課題1-2(即時與極短期暴雨強風預報技術之建立與應用)，本計畫打算建置與發展一個電腦運算時間極短，且又較為精密之即時預報系統( 0-3小時)。其目的主要在發展即時預報系統(符合研究內容1)以及進行極短期降雨預報與災害潛勢分析，並發展可支援防災應變決策作業系統之產品(符合研究內容4 & 5)。利用雷達網聯之回波觀測，使用變分回波追蹤技術(Variational Echo Tracking technique)，估計天氣系統之氣流運動場(motion field)，進而利用semi-Largrangian backward scheme 的方式平流回波，達到外延推估天氣系統下一刻可能所在位置，進行預報。此即時預報特點在於能夠獲得具有旋轉、形變(deformation) 特性之運動向量，掌握天氣系統移動、變化的能力。以劇烈天氣現象所產生之對流性降雨系統為主要研究對象(梅雨、鋒面系統、西南氣流，午後對流)，由probability of detection (POD)、false alarm rate (FAR)、critical success index (CSI)以及equitable threat score (ETS)等技術得分(skill score)的指數評估與驗證其預報能力表現。期望改進與提升此類天氣系統所帶來短延時、強降雨之即時預報能力，進而達到防災/減災之目的。 ;The purpose of the research is to build up and develop a sophisticated nowcasting system, but costs less computational resource. By using radar reflectivity from the network over Taiwan, with variational echo tracking technique and semi-Largrangian backward scheme﹐ the motion field of the weather system is able to obtain and extrapolate to the next time step for nowcasting. The advection based on the motion field is allowed for large-scale rotation, deformation, and differential motion between different systems. These features can better capture the displacement of the weather system. The main targets of the nowcasting system are strong convections with heavy precipitation such as Mei-Yu front, frontal system, south-west flow and thunderstorm. The performance of the nowcasting will be examined by using different skill scores (POD, FAR, CSI, and ETS). We expect to improve 0-3 h weather prediction, so that it is able to prevent disasters and reduce damages.
