本計畫希望能針對不同特性之強對流發展型態改善對流尺度資料同化系統在豪大雨預報上之應用，並建立適當的多尺度降雨預報評估工具。為了後續能納入氣象局雷達資料網，亦將強化同化系統使用高時空解析度資料之能力。針對綜觀-中尺度交互作用下的強對流事件，將發展多網格雙局地化法以拓展雷達及地面觀測資料在多重尺度下之修正量，以改善相對應之長延時強降水特性。將透過梅雨期間個案測試了解適合台灣多重尺度分析之特徵尺度，並建立網格雙局地化自動化流程。此外，針對快速發展的小尺度對流胞及所對應的短延時強降水則需仰賴高解析度觀測資料及模式網格的使用。為此，我們將透過在同化系統中加入雷達資料的誤差相關性以期能改善高解析度觀測資料的使用。相關工作將包含誤差相關性估計及同化系統計算流程之改進。 ;The project aims to advance the WRF-LETKF radar assimilation system (WLRAS) to provide multi-scale corrections and further improve heavy rainfall at different scales. WLRAS will also be optimized to assimilate large volume of data from the CWB radar network.For the heavy rainfall associated with synoptic-meso scale, we plan to implement the dual-localization for multi-domains to derive the multi-scale analysis corrections from the radar and surface data and expect improvement for predicting the long-life heavy rainfall. On the other hand, for improving the small-scale characteristics and short-life heavy rainfall events, we plan to better use the high-resolution radar observations by increasing the resolution of the superobs. This will be achieved by using a correlated observation error covariance in WLRAS.
