本研究選用2008 年西南氣流實驗期間內IOP-8 個案的梅雨鋒面降水事件，分析密集觀測資料，並以WRF 模式進行模擬研究，模擬實驗中使用四個不同的微物理參數化方案(WDM6 方案、WSM6 方案、Morrison 方案和Thopmson 方案)，希望比較各參數化方案對台灣地區梅雨鋒面降雨表現的特性。 在個案期間內梅雨鋒面系統位於台灣上空，在6 月14 日至6 月15 日間在台灣西南部地區產生大量降水，並觀測到強回波於台灣西南沿海。在模擬結果上，綜觀環境場與觀測類似，但副熱帶高壓有偏弱的情形，但模式有表現出在台灣西南部的降水系統。雖然降水系統發展整體有延遲的現象，但四個微物理方法都模擬出與觀測類似的回波雨帶登陸位置及強度。比較S-POL 雷達的觀測分析，四組實驗在台灣西南部都可掌握到兩條回波雨帶的特徵。比較12 小時的台灣地區平均累計降水，以WDM6 及WSM6 方案的模擬降水量較接近於觀測，Morrison 和Thompson 方案則高估了降雨量。 我們在這段時間內選用雨帶強回波期及弱回波期兩個時間點，分別進行雷達回波以及垂直速度CFAD 的比較。雷達回波CFAD 在所有方案的模擬結果上，都有偏高的趨勢，其中WDM6 模擬亮帶的特徵較為明顯，與觀測最為相似。垂直速度CFAD 剖面顯示模式對於上升氣流的模擬表現與觀測相似，但是下降氣流的部分則都有低估的現象。 使用S-POL 雷達模糊邏輯以判斷水相粒子分類，並與模擬進行比較，發現在水相粒子分類(PID)的垂直分佈上，Thompson 方案對於雲冰和雪的與觀測相差較大，主要由於Thompson 方案本身的粒徑譜調整與其他方案不同，所以產生過多的雪。WSM6方案由於沒有粒子濃度的預報，因此對雨水混合比的高度有所偏差，整體而言，PID 的分析是以Morrison 及WDM6 方案是較接近觀測分析的結果。 模擬所使用的微物理參數方案中有三組是屬於Double moment 方案，共進行雨水粒子濃度的預報，因此與西南氣流實驗中的地面撞擊式雨滴儀(JWD)的雨滴觀測資料做比較分析。由時序變化圖的結果來看，模擬結果延遲了兩個小時，但WDM6 方案模擬的變化情形與觀測較為一致。最終本次個案的模擬結果上，以WDM6 方案表現得與觀測分析最為接近。 In this study, a typical case for heavy rainfall associated Mei-Yu front on 14-15 June 2008, also categorized as the IOP-8 event during the SoWMEX. We analyzed the observation data and compared with the WRF model simulations. One single moment microphysics scheme (WSM6) and three double moment microphysics schemes (WDM6, Morrison, Thompson) are chosen to perform sensitivity tests, in order to evaluate different schemes ability to simulate the Mei-Yu frontal rainfall near Taiwan area. This rainfall case produced heavy precipitation over southern and southwestern Taiwan, that associated with the interactions of the Mei-Yu front and the orographicover southern Taiwan. The WRF-simulated synoptic field, wind pattern and geopotential height were consistent but the subtropical high weaker than observation. Radar echoes show the precipitation system encountered southern Taiwan, and all the experiments simulated two rain bands were similar with S-pol (S band dual-polarization Doppler radar) radar observation but delayed 2 hours. Reflectivity CFAD (Contoured frequency with altitude diagrams) of four microphysics simulations compared with CFAD of S-pol radar echoes were all overestimated. WDM6 simulated bright bands were more obvious than otherschemes. Vertical velocity CFAD show that four schemes simulated updraft was similar to observation, but downdraft underestimated in all simulations. We compared with the PID (Particle Identification) vertical percentage distribution of observational analysis and simulations, the Thompson simulated produced unreasonable amount of snow because of power-law droplet size distribution; WSM6 scheme produced more amount of rainwater at high levels which without forecasting rainwater number concentration. Morrison and WDM6scheme is not biased toward a specific type of hydrometer. We analzed the JWD (Joss-Waldvogel disdrometer) data and compared with the model simulated rainwater number concentration time series. WDM6 schemes simulated time series show the most similar evolution to observation analysis. Finally, WDM6 simulation was the most reasonable, but still need to consider the model error, and the error of the observations.
