以系集模擬方法探討不同寬度 moat 雙眼牆之成因-以海燕颱風(2013)為例

Abstract

在過去的研究中，已發現不同時長的眼牆置換會造成熱帶氣旋在眼牆置換過程中出現不同的強度和結構上的變化，而moat區域的寬度是影響眼牆置換時長的因子之一，但關於決定moat區寬度的因子仍缺乏相關研究，因此本研究的目標為試圖找出決定雙眼牆形成時moat寬度之因子。本研究了使用NCEP FNL模式的分析場資料作為初始場和邊界條件，搭配WRF 4.2.1模式模擬2013年海燕颱風眼牆置換過程。模擬結果顯示，海燕颱風的路徑基本與JTWC最佳路徑一致，但在強度演變方面仍有高估的情形。模擬重現的海燕雙眼牆形成時間與實際觀測接近，雙眼牆形成時的Moat寬度為25公里，眼牆置換時長約為12小時，均小於氣候平均值。 為探討海燕形成較窄Moat雙眼牆的原因，本研究進行系集模擬。透過WRF-VAR系統產生150個系集成員，其中40個成員出現完整的眼牆置換過程。初始Moat寬度與眼牆置換時長呈正相關，相關係數平方達0.5352。將系集成員中Moat寬度前25%之成員挑選出來作為寬組(Wide Group, WG)，後25%成員挑選出來作為窄組(Narrow Group, NG)進行分和比對。由內部動力場分析顯示，NG和WG組別在切向風場、慣性穩定度場和帶狀化的時間分布上無顯著差異，顯示內部初始動力結構的差異可能不是造成雙眼牆形成時不同寬度Moat的原因。在背景環境比對中， WG在雙眼牆形成20小時前有較強且較深厚的垂直風切，這個差異同樣通過統計顯著性檢定，並在雙眼牆形成前11小時消失。本研究將時間區分為兩組：Stage I（雙眼牆形成21小時至11小時以前）和Stage II（雙眼牆形成前10小時）。 在Stage I期間，WG在上風切象限有不活躍的上升運動和沉降運動分布，下風切則有較活躍的上升運動，符合受垂直風切影響的熱帶氣旋特徵。進入Stage II後，隨著垂直風切減弱，WG的垂直運動變得對稱且活躍。NG在Stage I期間的垂直運動一直維持對稱分布，且在雨帶對流分析中，WG在Stage I期間的對流不活躍且不對稱，而在Stage II期間逐漸活躍並形成較寬的Moat。NG的對流一直較為活躍且對稱，形成較窄的Moat。WG在由Stage I進入Stage II過程中，外圍雨帶轉變為活躍，低層徑向入流增強，可能是由於非絕熱加熱增加，增強了次環流。在軸對稱平均切向風場演變分析中，伴隨徑向入流增強，切向風場也出現明顯增強和擴張，邊界層頂也出現超梯度力加強的現象，推測徑向入流的增強確實引發切向風場擴張和邊界層非平衡動力機制，促成雙眼牆的形成。NG在Stage II時，外圍雨帶中的對流轉變為層狀降水過程中，非絕熱作用分布的改變引發Mesoscale Descending Inflow (MDI)-Like徑向入流，增強低層徑向入流，透過增強低層切向風或觸發邊界層非平衡動力機制來形成雙眼牆。由於MDI-Like徑向入流所造成的低層徑向入流較強，能更深入颱風內核區域，在更靠近中心的位置製造Forcing，形成較窄的Moat，縮短眼牆置換所需時間。 總結來說，背景垂直風切和外圍雨帶對流活動情形對雙眼牆的形成和Moat寬度有顯著影響。較弱的垂直風切和活躍的外圍雨帶造成較強的MDI-Like徑向入流，導致了較窄的Moat和較短的眼牆置換時間；反過來說，較強的垂直風切和不活躍外圍雨帶會產生較寬的Moat和較長的眼牆置換時間。

The duration of eyewall replacement cycles (ERC) is a critical factor influencing the intensity changes of TC during ERC. Previous studies have proposed relationships between the duration of ERC and the width of the moat (Fischer et al., 2020; Yang et al., 2021). There are also related studies investigating how the width of the moat affects the duration of ERC (Lai et al., 2019). However, as a key factor determining the duration of ERC, there is still a lack of literature discussing the factors that determine the width of the moat. The objective of this study is to investigate which internal dynamics or external environmental factors determine the size of the moat during the secondary eyewall formation (SEF) in tropical cyclones. The first part of this study utilizes the Weather Research and Forecasting (WRF) model to simulate the narrow moat concentric eyewall events observed in Typhoon Haiyan (2013) which involved a short-duration eyewall replacement during the intensifying phase, with a small secondary eyewall and no significant intensity weakening (Lin et al. 2021). In the second part, using the same model configuration as the first part, we employ WRF Variational data assimilation (WRF-VAR) to generate 40 ensemble members as control groups for comparison and analysis, aiming to identify factors that lead to a narrower moat width. The WRF simulations reveal that Haiyan exhibited a narrower initial moat (25 km) compared to the climatological mean (71 km). The WRF-VAR ensemble results indicate a positive correlation between the duration of eyewall replacement and the initial moat width, in accordance with the findings of Yang et al. (2021). Subsequently, the ensemble members were divided into two groups, narrow (NG) and wide (WG), based on the lowest 25% and the highest 25% moat width. Prior to the SEF, there were no significant differences between the two groups for the initial structural characteristics (e.g., tangential wind, inertial stability, and filamentation time). In terms of environmental factors, WG experienced greater environment vertical wind shear (VWS), which induces asymmetric structures and inactive rainbands. As VWS weakens, rainbands gradually become active and axisymmetric. Tangential wind tendency and agradient force analysis show that as rainband axisymmetrization, it generates forcing (Unbalanced dynamic, Wu et al., 2012; Huang et al., 2012) away from the primary eyewall, ultimately leading to a wider moat during SEF. In the NG group, weaker VWS results in more active and symmetric rainbands. Diabatic heating distribution and precipitation type analysis reveal that as convection within these rainbands gradually weakens into stratiform precipitation, an inflow band similar to mesoscale descending inflow (MDI, Didlake and Houze 2013b) develops from diabatic heating distribution appears. Agradient force analysis indicates that this MDI-like inflow creates forcing (Didlake et al., 2018) closer to the primary eyewall, ultimately leading to the formation of a narrower moat during SEF. The results indicate that varying VWS results in differences in rainband activity, leading to differences in the source and location of forcing, ultimately resulting in TCs exhibiting varying widths of moat during the SEF.