| 摘要: | 隨著全球暖化，極端天氣事件如豪大雨等事件引發的災害頻率正在不斷增加。劇烈風暴通常伴隨著快速且複雜的物理過程演變，而這些物理過程具有高度動熱力非線性的特性。雖然近年來，傳統氣象雷達廣泛應用於風暴的偵測中，然而傳統氣象雷達的空間和時間解析度不足，無法獲得足夠準確地描述快速變化的風暴發展的觀測資料，另一個限制在於其機械式掃描角度的設計限制了掃描速度和靈活性，並可能需要更頻繁的維護。相對之下，相控陣氣象雷達（phased-array weather radars, PAWR）能提供高空間和時間解析度，可以偵測風暴更詳細的演變過程，包括其發展、增強及減弱過程。
本研究利用2019年7月22日台北盆地午後對流事件建立觀測模擬系統實驗（observing system simulation experiment , OSSE），並假設PAWR設置於五股自行車道, 國立台灣大學及烏樹林里，並此架構探討 PAWR 資料同化對於台北盆地午後雷陣雨的影響。OSSE實驗中，真實大氣場是使用網格間距為200米的天氣研究與預報模式（Weather Research and Forecasting Model, WRF）所產生，並且以其模擬類PAWR 的觀測資料。接著再使用由日本理化學研究所（Institute of Physical and Chemical Research, RIKEN）開發適用於先進運算環境的局地系集轉置卡爾曼濾波（SCALE-LETKF）系統來同化雷達回波和徑向速度，以捕捉快速對流。本研究針對無雨資料、雷達數量以及同化頻率（5 分鐘和 1 分鐘）三種同化效益議題設計多組實驗，以探討其對台北都市區對流結構描述的影響。研究結果顯示較高的同化頻率能夠在短同化時間內，迅速地捕捉到回波的特性，並迅速減少誤差，但對降雨預報未能有明顯優勢，其中原因為同化過程中未能正確修正水氣場所導致。 在控制水氣場的錯誤修正下同化PAWR便可有效改善的降雨快速發展。
;With global warming, catastrophic disasters caused by atmospheric events, such as heavy rainfall, are continuously on the rise. Severe storms often evolve rapidly with complex dynamic and physical processes with high nonlinearity. Conventional weather radar has been widely used in monitoring storms. However, the spatiotemporal resolution of the conventional weather radar is not sufficient to analyze the development of the rapidly changing storms. Another limitation is mechanical scanning on elevations, which limits its scanning speed and flexibility and may require more frequent maintenance. By contrast, phased-array weather radars (PAWR) can provide high spatial and temporal resolution and thus detect the more detailed evolution process of convective storms, including development, intensification, and weakening processes.
This research examines how PAWR assimilation influences afternoon thunderstorms over the Taipei Basin with the framework of an observing system simulation experiment (OSSE). Based on the event on 22 July 2019, the nature run of the OSSE is generated using the Weather Research and Forecasting (WRF) Model with a 200-m grid spacing. Assuming radial velocity and reflectivity of three PAWRs are available at Fig. 5. The simulated PAWR observations are generated with the nature run. Then, the Local Ensemble Transform Kalman Filter for Scalable Computing for Advanced Library and Environment model (SCALE-LETKF), developed by the RIKEN Center for Computational Science (R-CCS), assimilates reflectivity and radial velocity data while using a 1 km analysis grid spacing. Three types of experiments are designed to investigate the assimilation impact, including the usage of no-rain data, the number of radars, and assimilation frequencies (5 and 1 min), on representing the convection structure in the Taipei metropolitan area. This study shows that a higher frequency can better capture the reflectivity characteristics and reduce error rapidly with the short assimilation period. However, its benefit on the rainfall forecast is not the greatest since the moisture adjustment is over-reduced in the assimilation due to the usage of no-rain data, which is too dry for the convection cells to keep growing. If the degradation from moisture adjustment can be controlled, assimilating PAWR data can have a great impact on predicting the rapid development of the precipitation. |
