在本研究中我們將探討若干與同化都卜勒雷達資料以改進短期數值天氣預報相關的科學議題，我們準備使用的資料同化方法為系集卡曼濾波器或三維變分法。主要的研究目標在建立一套適合台灣特殊地理條件的0-6 小時定量降水預報的方法。在本兩年期計畫中吾人預計要完成的工作項目為評估在不同的條件下，在使用系集卡曼濾波器或三維變分法同化雷達資料後，對模式預報的影響。這些條件包括：雷達與分析區域的相對位置、雷達資料的觀測誤差與相關、不完整的資料覆蓋、雷達掃描策略、最佳的同化窗口、以及資料同化的策略等。第二年則進行後續之規畫，計畫內容為使用真實個案來測試系集卡曼濾波器，測試的重點在於短期定量降水預報的改善，而真實個案的資料可以是在2008 年西南氣流實驗當中，由NCAR S-POL、TEAM-R、氣象局七股雷達或墾丁雷達所觀測的劇烈降水系統。 ; In this research we will explore several scientific issues related to short-term numerical weather prediction using Ensemble Kalman Filter (EnKF) or 3DVAR to assimilate Doppler radar data into a numerical model. The primary purpose is to establish a 0-6 hr QPF algorithm suitable for Taiwan’s geographic characteristics. The research topic we would like to accomplish in this 2-year project is to study the performance of EnKF or 3DVAR under different scenarios, such as the number of ensembles, radar location relative to the analysis domain, observations and background field errors covariances, incomplete data coverage, optimal assimilation window, and so on. In the second year the plan will be testing EnKF or 3DVAR with real data sets such as the cases collected during TIMREX/SOWMEX by NCAR S-POL, TEAM-R, RCCG or GCKT. The focus of these real case studies will be the improvement of the short-term quantitative precipitation forecast (QPF). ; 研究期間 9708 ~ 9807
