請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/9275
標題: | 颱風路徑與結構同化研究─系集卡爾曼濾波器 Assimilation of Tropical Cyclone Track and Structure Based on the Ensemble Kalman Filter |
作者: | Guo-Yuan Lien 連國淵 |
指導教授: | 吳俊傑(Chun-Chieh Wu) |
關鍵字: | 颱風初始化,資料同化,系集卡爾曼濾波器,T-PARC, tropical cyclone initialization,data assimilation,ensemble Kalman filter,T-PARC, |
出版年 : | 2009 |
學位: | 碩士 |
摘要: | 近30年來,颱風的路徑預報有穩定的進展,但要在模式中建構出具有正確中心位置、移速、以及合理渦旋結構的颱風初始場,一直是颱風數值模擬的一大挑戰,渦旋植入、虛擬渦旋資料同化、或是渦旋重新移位等方法皆設計用以改善颱風的初始化。隨著EnKF的發展,以EnKF直接同化渦旋位置的方法亦被提出,透過一計算渦旋中心位置的觀測算符,可使模式中的渦旋保持在觀測路徑上。
在本研究中,我們進一步定義更多與颱風路徑和結構相關的觀測算符,包括中心位置、渦旋移速與海表面軸對稱風速等。前二者的觀測量可使用作業單位依據衛星資料所做的颱風定位,軸對稱風速剖面則可由經驗公式擬合衛星與如DOTSTAR或T-PARC的飛機觀測資料而來。 我們在配置EnKF資料同化系統的WRF模式中進行了颱風渦旋初始化和快速更新週期同化分析等兩類實驗。在初始化實驗中僅同化颱風特殊觀測量,未採用任何現有的渦旋植入方案。颱風的路徑與軸對稱平均風速剖面可在初始化時段末期與給定的觀測資料相符,並且在僅同化最低層風速剖面的狀況下,整個垂直結構可被完善建立。本方法最大的優點在於EnKF得到之分析場平衡性佳且相容於使用的模式,因此後續預報的颱風強度得以穩定維持,不會在預報初期有劇烈調整的狀況。而在快速更新週期同化分析實驗中,除了同化颱風特殊觀測量外,常態性探空儀與投落送資料也被一併同化,以進行數天的長時段模擬分析。海表面軸對稱風速結構主要由2008年T-PARC實驗中C-130飛機連續4次穿越颱風中心偵察任務的資料而來,此資料對颱風軸對稱風速同化有相當大的助益。實驗結果顯示,包括眼牆置換的過程在內,颱風的路徑與結構的演變可被完善掌握。 由此實驗結果,我們認為這個新概念與技術提供了一個颱風初始化的有效方法,並有潛力用以設計探討颱風結構的高解析度實驗,以及改善作業颱風模式的路徑與強度預報。本研究已成功運用此方法,有效同化2008年T-PARC四架飛機所聯合觀測之寶貴颱風資料,可為颱風動力探討帶來新的突破契機。 Over the past 30 years, track forecasts of tropical cyclones (TC) have been in steady progress, but initializing a realistic vortex in the correct location and with the correct storm motion and structure remains a challenging task. Some techniques, including vortex bogusing, bogus data assimilation, and relocation, have been designed to improve the TC initialization. With the progress of ensemble Kalman filter (EnKF), assimilating vortex position based on the EnKF given an “observation operator” that computes the vortex position has also demonstrated capability in keeping a vortex along the observed track. In this study, some new and effective observation operators related to the TC track and structure are proposed, including center position, velocity of storm motion, and sea surface axisymmetric wind structure. The observational quantities of first two parameters can be available from operational centers mainly based on the satellite analysis, and the radial wind profile can be evaluated through curve fitting using empirical formula, along with the aircraft surveillance data such as from DOTSTAR and T-PARC. By assimilating these parameters based on the EnKF, we carry out two types of experiments in high-resolution WRF model: one for TC initialization, and the other for update cycle analysis. In initialization experiments, only special parameters for TCs are assimilated in a 24-hour period, without any extant bogus scheme. The TC track and axisymmetric wind profile well follow the specified observation data at final time of the initialization period. The overall vertical structure can be suitably constructed by assimilating only one-level wind profile. Moreover, one important benefit of this method is that almost no subsequent adjustment follows, indicating that the initial condition for forecast simulation after initialization period is dynamically balanced, as well as model-compatible. In update cycle experiments, both special parameters for TCs and conventional radiosonde and dropwindsonde data available from GTS are continuously assimilated, in order to perform an update cycle analysis of several days. The sea surface axisymmetric wind structure is determined from 4 continuous reconnaissance flights by C-130 aircraft during T-PARC in 2008. The result shows that the track and structure evolution of TCs, including the eyewall replacement cycle, can be captured in this simulation, indicating the usefulness of observations from reconnaissance missions in this method. Our results suggest that this new technique provides an effective means of improving TC initialization and has good potential to help conducting some high-resolution numerical experiments to better understand the dynamics of TC structure, and to improve the operational TC model forecast. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/9275 |
全文授權: | 同意授權(全球公開) |
顯示於系所單位: | 大氣科學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-98-1.pdf | 14.45 MB | Adobe PDF | 檢視/開啟 |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。