| 摘要: | 台灣北部於2018年9月8日受到鋒面系統、熱帶性低氣壓與午後熱對流的多重尺度天氣系統，造成當日有不同特性的強降雨事件發生。本研究利用WRF-LETKF雷達資料同化系統(WRF-LETKF Radar Assimilation System, WLRAS)建立每15分鐘一筆之快速更新分析場(Rapid Update Analysis)，並以此探討當日一系列降雨事件的生成因素。除了使用標準的對流尺度局地化法於WLRAS，在本研究中更使用逐次協方差局地化法(Successive Covariance Localization; SCL)進行多重尺度修正，以強調各種尺度間交互作用的重要性。此外，本研究也分別額外使用地面測站資料並搭配「無降雨」資料(“No-rain” data)建立分析場以測試近地面特性對於此個案的影響。
由於資料品質會影響多重尺度分析的表現，本研究利用RAKIT對雷達資料進行更完整的資料檢定，並以此探討SCL的最佳同化策略。利用最佳的實驗組我們進一步討論大尺度調整對鋒面及地面風場結構的重要性，以及這些特性如何影響台北盆地內的局地深對流發展。對於鋒面系統所在區域，最重要的是鋒前西南風與鋒面上環流，大尺度修正可以使其建立結構更完整且位置正確的鋒面系統。然而，相較於單一對流尺度修正，SCL對於二次環流掌握不足，使鋒面對流胞強度無法完整呈現。對於局地對流胞，經由大尺度動力場的修正，強對流可以更容易維持在雪山山脈的山頂，而後造成冷池在台北盆地快速激發局地雷雨胞。結果顯示，即便不額外使用地面測站風場資料，使用SCL所獲得的四維分析場便可以建立適合對流發展的低層環境。相較而言，SCL方法可以提供環境的動力結構，使多重尺度的交互作用得以展現。而地面風場資料則可修正部分環境近地面風場結構。本研究顯示對流尺度的高時空解析度分析場建立除須考慮局地對流胞本身動力及熱力結構，亦須描述環境的動力場以完整顯示局地對流發展的成因。
;The occurrence of the heavy rainfall event over northern Taiwan on 8 September 2018 involves different weather systems, including a front, a tropical depression, and afternoon thunderstorms. This study investigates the establishment of a series of precipitation events by performing rapid update analyses with the WRF-LETKF Radar Assimilation System (WLRAS). In addition to using a standard convective-scale localization for radar observation, we apply a multi-scale correction scheme, the Successive Covariance Localization approach (SCL), to emphasize the importance of interaction between different scales. Moreover, we also assimilate the surface station data and the “no-rain” type of radar observation to improve the near-surface analysis.
First, the influence of radar data quality on the analysis using SCL is discussed. Using the quality-controlled radar data, we examine the optimal settings for SCL. With the optimized settings, the main focuses include the large-scale adjustment and its impact on the frontal wind structure, the importance of the surface wind structure, and how these factors further lead to the development of the local deep convections over the Taipei Basin. For the area with the frontal system, the magnitude of the prefrontal southwesterly flow and frontal circulation is essential. When the SCL method is applied in the radar assimilation, the large-scale adjustment can be achieved, which can help establish a well-structured frontal system with a more accurate position, but the intensity of convections is slightly underestimated due to insufficient secondary circulation. Through the adjusted large-scale dynamics, strong convection can be better maintained over the top of the Xueshan Range, and the resulting outflow of the cold pool further triggers the rapid development of the local thunderstorm over the Taipei Basin. Consequently, both the experiments using SCL and surface wind data can provide a suitable 4-D rapid update analysis field: The SCL can provide an environmental structure that facilitates the multiscale interaction, and the surface wind data can give a better representation of the near-surface environmental structure. This study suggests that to well represent the local convection development in a high-resolution convective scale data assimilation analysis, it is necessary to analyze not only the dynamic and thermodynamic structure of local convections but also the environment’s dynamic field. |
