- 學位論文
地形引發中尺度環流對燦樹颱風(2021)快速增強之影響
A Study on the Mesoscale Flows Induced by Taiwan Terrain and Their Impacts on the Rapid Intensification of Typhoon Chanthu (2021)
摘要
本研究主題為探討地形引發中尺度環流對熱帶氣旋內核結構演變之影響。2021年燦樹颱風在通過臺灣東部海面時強度快速增加，並伴隨眼牆對流顯著的波數1不對稱性，這些現象均透過臺灣密集的雷達網進行完整觀測。因此，藉由分析雷達回波、速度場以及地面測站資料，提供了本研究難得的機會，探討燦樹颱風內核結構的演變過程以及其與環境流場之關聯；並可針對此類結構極端扎實的颱風，進行雷達資料同化以探討模式渦漩初始化的議題。 為修正颱風強風造成雷達都卜勒速度的折錯問題，本研究首先發展vortex-based Doppler velocity dealiasing (VDVD)演算法，針對雷達都卜勒速度場進行品質控管。此演算法使用理想渦漩做為參考風場進行反折錯，並搭配內迴圈以ground-based velocity track display (GBVTD)技術調整颱風風場結構、外迴圈以GBVTD-simplex技術調整颱風中心的內外雙迴圈架構進行。結果顯示VDVD在理想個案或實際個案中，即使渦漩因為徑向風、平均流或渦漩本身切向風所造成的波數1不對稱風場結構，均可有效修正受折錯的速度場。燦樹颱風的雷達速度場資料經過VDVD與本研究所發展另一套修正局部折錯資料的方法處理後，可有效進行反折錯，以確保資料品質並進行後續的分析。 使用GBVTD反演燦樹颱風在3公里高度之切向風場顯示，其最大風速於增強階段時，在11小時內增加約18 m s-1，並於減弱階段在8小時內減少19 m s-1；顯示燦樹颱風在24小時內經歷了快速增強(rapid intensification)與快速減弱(rapid weaking)的過程，此類強度劇烈變化的個案為預報作業上的一大挑戰。在增強過程期間，燦樹颱風眼牆波數1不對稱性的最大值區域，由原本位於颱風中心東側，以氣旋式方向迅速移動至北側，此轉換過程恰好發生於地面測站資料觀測到地形繞流的訊號之後。另外，此波數1不對稱對流的軸向變化與雷達資料所反演的meso-β尺度垂直風切方向隨時間具有一致性；但由重分析資料計算而得的meso-α尺度垂直風切方向，則與眼牆對流分布不具一致性。顯示當外部強迫機制主要由地形所引發時， meso-β尺度的垂直風切可較meso-α尺度垂直風切更具代表性。利用此meso-β尺度垂直風切搭配地面測站資料進行分析，本研究推測燦樹颱風的快速增強過程與1) 地形引發在颱風南側之邊界層入流、2) 風向指向上風切左側象限的低層平均流以及3) 較弱的高層風切有關。 本研究進一步使用數值模式搭配雷達資料同化探討如何對於燦樹颱風此類風場結構極端扎實的颱風進行渦旋初始化。將雷達都卜勒速度內插至模式網格後進行資料同化的VRC實驗顯示其模擬之渦漩的強度變化與雷達觀測較為接近，從9月11日1100 UTC持續增強至9月12日0000 UTC，且眼牆對流演變亦與雷達觀測之圓形眼牆較為一致。VRP實驗模擬之渦漩雖然於9月11日1800 UTC前展現較VRC實驗更快的增強速率，但後續強度則呈現緩慢減弱的趨勢，加上眼牆對流的波數3的不對稱性，均與觀測資料不一致。上述實驗顯示VRC的同化策略可較為合理模擬渦漩的內核演變過程；然而，VRC實驗所模擬的垂直風切仍明顯較觀測資料為弱，不適合探討與風切相關的增強機制。因此，未來將藉由增加水平解析度與低層垂直解析度，對地形引發的meso-β尺度現象進行合理的模擬，以利後續定量評估造成燦樹颱風增強的主要物理機制。
並列摘要
The central theme of this study is to investigate the impacts of terrain-induced flow on the inner-core evolution of a tropical cyclone (TC) tracking northward along Taiwan's eastern coast. Typhoon Chanthu (2021) underwent rapid intensification (RI) and rapid weakening (RW) within the 24-hour analyzed period near Taiwan, posing challenges for intensity forecasts. Its intensification, characterized by a significant increase of 18 m s-1 at 3 km altitude within 11 hours and pronounced wavenumber-1 asymmetry in eyewall convection, was thoroughly observed by Taiwan's dense radar network. These comprehensive observations provide a valuable opportunity to explore the central theme of this study and the vortex initialization issue for such an extremely compact TC through radar data assimilation in numerical model simulation. To address the challenge of recovering aliased Doppler velocities caused by strong TC winds, this study proposes a vortex-based Doppler velocity dealiasing (VDVD) algorithm specifically for TCs. The algorithm employs an inner-outer iterative procedure, adjusting the reference vortex structure using the ground-based velocity track display (GBVTD) technique, and utilizing the GBVTD-simplex algorithm for center correction. Both the VDVD and a local dealiasing method, also developed in this study, were applied to the aliased Doppler velocities. These methods effectively corrected the velocities and ensured the data were suitable for accurate analyses. Radar analyses for Typhoon Chanthu suggest that radar-derived meso-β scale vertical wind shear (VWS), which aligns better with the observed rotation of eyewall asymmetry, is more representative than meso-α scale VWS when terrain-induced forcing predominates. Further examination of the radar-derived VWS indicates that the VWS profile provided a more favorable environment for typhoon intensification. Observational analyses reveal that Chanthu's RI was influenced by three factors: 1) terrain-induced boundary inflow from the south of the typhoon; 2) upshear-left-pointing low-level mean flow; and 3) weak upper-level VWS. To explore effective methods for initializing a compact TC vortex like Chanthu in numerical models, two radar data assimilation strategies were compared. The VRC experiment, which assimilates Doppler velocity interpolated onto the model grid, showed better consistency with radar observations regarding intensity change and eyewall evolution. In contrast, the VRP experiment, which assimilates thinned radar data in original coordinates, exhibited eyewall asymmetry dominated by a wavenumber-3 structure, differing from radar observations. However, the weak VWS in the VRC experiment, inconsistent with observations, hinders investigating intensification mechanisms. Future enhancements like increasing horizontal and vertical resolution are recommended to better capture meso-β scale features and assess the major mechanisms driving Chanthu's intensification.