| 摘要: | 本研究的目的為了解不同微物理參數化方案(包括單矩量及雙矩量)在雲可解析模式對系集預報架構下系集離散度之影響及其特性。我們首先透過單一預報分析不同方案的基本特性，接著利用系集法分析不同方案在系集架構下的誤差結構並且探討這些方案對於同樣初始條件下的敏感度。我們選用 WRF 模式模擬西南氣流實驗期間 2008 年 6 月 16 日(SoWMEX-IOP8)在台灣 2 西南岸之豪大雨個案。在所有系集實驗中，系集初始場皆使用經由區域資料同化系統(WRF-LETKF)所得之分析系集。實驗共選用了四種不同的微物理方案來進行分析，包含 GCE、MOR、WSM6 和 WDM6 方案。
研究結果顯示雙矩量的微物理參數化方案(MOR, WDM6)並不一定能夠製造出比單矩量(GCE, WSM6)方案更大的系集離散度。而不同的微物理方案在水相粒子的分布上有著很大的差異，特別是在冰相粒子的部分又更為明顯。結果顯示在系集離散度的發展上 GCE 方案幾乎在每一個變數都有相對較大的離散度。主要是因為相比於其他方案，GCE 方案有著相對較強，且變動量大的垂直速度並又伴隨著非常有效率的相態轉換造成潛熱釋放的離散度增大。而這些微小尺度的放熱以及吸熱的過程又會進一步影響到更大尺度的溫度以及風場的離散度。雙矩量方案的效益在比較同一類型的微物理方(WSM6 和 WDM6)時才會顯現出來。WDM6 在所有變數上都有比 WSM6 大的離散度，尤其是在雨水以及雲水的部分，而即便兩方法使用相似的冰相設定，似乎雙矩量處理在暖雨過程所造成的離散度會進一步的影響到冰相變數的離散度。
因此，若以混成微物理法建立系集，本研究結果顯示混和 GCE 以及 WDM6方案將能比較有效的增加整體上的系集離散度以掌握不同方面的預報誤差。最後我們也發現不同發展者所設計的微物理方案最大的差異是在冰項微物理過程的處理上面，而這些差異會影響到不同的潛熱釋放特徵，進而再影響到更大尺度的溫度場以及風場。;In this study, we aim to understand the effects of different microphysics (MP)schemes [including single- (SM) and double-moment (DM)] on the ensemble spread under the framework of ensemble forecasts in the cloud-resolving model. We first analyze the basic features of these schemes through the deterministic forecasts then using the ensemble method to focus on the comparison of the ensemble-based error structures and investigate the sensitivity of the initial conditions to different MP schemes. The simulation for the heavy rainfall event during the Southwest Monsoon Experiment on June 16, 2008 (SoWMEX-IOP8) is examined with the WRF model. In all ensemble experiments, the initial conditions are obtained from the regional data assimilation system (WRF-LETKF). Four different MP schemes were selected for analysis, including GCE, MOR, WSM6, and WDM6 schemes.
Results show that the DM schemes (MOR, WDM6) do not necessarily produce a larger ensemble spread than the SM schemes (GCE, WSM6). Different MP schemes have great differences in the distribution of hydrometeors, especially with the icerelated variables. Results show that GCE has a relatively large spread in almost every variable. This is mainly because GCE has a relatively strong and variable vertical velocity associated with the efficient phase transition results in the spread of latent heat release increases and further affect the spread of larger-scale temperature and wind fields. The benefits of the DM scheme are only apparent when comparing similar MP schemes (WSM6, WDM6). Especially in the rainwater and cloud water, even if the two methods use similar ice processes, it seems that the spread caused by the DM treatment in the warm rain processes will further affect the spread of ice-related variables.
Therefore, if the ensemble is established by the multi-MP method, the results of this study show that the combination of GCE and WDM6 schemes would be more effective in increasing the overall ensemble spread to represent the forecast error in different aspects. Finally, we found that the ice-related processes are handled very differently with different MP scheme developers and these difference will affect the pattern of latent heat release which in turn affect larger-scale temperature and wind fields. |
