| 摘要: | 本研究利用WRF-LETKF Radar Assimilation System (WLRAS, Tsai et al., 2014)分析場與預報場結果，評估宜蘭冬季降水個案同化差異反射率觀測之效益，並比較莊(2021)提出的新更新法：Mean Diameter Update (MDU) Approach，與傳統變數更新法的異同，以尋找最佳的微物理分析場。同化實驗共有四組設置，第一組(VrZ)僅同化傳統雷達觀測資料(ZH, Vr)，第二組(VrZZ)則多同化差異反射率(ZDR)並用傳統方法進行更新，其餘兩組(NwDm與qrDm)則多同化ZDR並使用MDU Approach中兩種不同設定進行變數更新。研究結果顯示，同化ZDR觀測資料能調整模式分析場雨水混合比與水氣混合比，並改變整體雨滴平均粒徑大小，改善分析場微物理結構與近地表水氣表現，但仍存在參數化方案與觀測算符造成的偏差。使用新方法進行變數更新時，微物理結構調整更多且更快速，使得回波與差異反射率更加接近真實觀測表現。短期定量降水預報分析上，未同化的系集(noDA)有一定的能力描述降雨分佈與極值位置，但無法描述降雨生成、消散與移動的過程。同化傳統雷達觀測時，能改善第一個小時的降水表現，但嚴重低估第二個小時降水，使得2~6小時累積降水表現較noDA差。相較之下，同化ZDR在分析場增加的近地表水氣、平均雨滴粒徑與雨水混合比，其效益能維持至預報第2~3小時，並大幅改善定量降水預報表現。此外使用qrDm法更新模式變數，在預報表現上能調整局部降水強度，增加降水表現改善幅度。綜合所有研究結果，在2021YESR IOP2冬季個案中同化ZDR資料，能改善分析場微物理與熱力結構，提升短期定量降水預報能力，並且利用MDU Approach更新模式變數，能有更好的分析場微物理結構，減少與觀測之差異。;In this study, the WRF-LETKF Radar Assimilation System (WLRAS, Tsai et al., 2014) analyses and forecasts of the wintertime rainfall case have been evaluated and confirmed the impact of assimilating differential reflectivity (ZDR) with different approaches. Two sets of experiments that VrZ assimilates reflectivity and radial velocity, and VrZZ assimilates reflectivity, radial velocity and ZDR are examined. In addition, the impact of assimilating ZDR with the two updating strategies in the Mean Diameter Update (MDU) approach, NwDm and qrDm, has been investigated to validate the performance between different update methods.
Results of assimilating ZDR show that the water vapor and rain water mixing ratios are enhanced in analyses, which adjust the mean drop size and modify the ZDR structure toward observations. Besides, the MDU approach has much more improvement than the traditional update method. However, due to configuration of the observation operator and double-moment microphysics scheme, the bias of the ZH-ZDR structure still remains. The first hour Quantitative Precipitation Forecast (QPF) has been improved after assimilating traditional radar data; however, the underestimation in the second hour causes a the worse accumulated rainfall performance than ensemble forecasts without data assimilation (noDA) after the first hour. By contrast, after ZDR assimilation, the enhancement of water vapor and rain water mixing ratios continues to the second hour in the forecasts, which leads to a better Probabilistic QPF (PQPF) and a lower underestimation. Also, using the qrDm strategy may enhance partial rainfall. To sum up, properly assimilating additional ZDR observations can not only have a better description of the uniform ZDR structure but lead to a better precipitation forecast in the wintertime rainfall case. |
