應用群集分析探討侵臺颱風之分類特性

Cluster analysis on the paths and scales of typhoons driving to Taiwan

賴瑩嬛

指導教授 ： 張憲國

國立交通大學/工學院/土木工程系所/碩士(2012年)

https://doi.org/10.6842/NCTU.2012.00783

摘要

本文利用二階段群集分析法以侵臺颱風形成及結束的位置、路徑長度、最大風速及延時為變數探討颱風的特性分類。若依颱風位置為變數時分成四類最為合適，其中兩類為東北行，一類為西行，一類為臺灣附近海域生成之颱風。若依颱風位置及路徑長度為變數時則分成六類為佳，可將路徑長短區分出來，其中三類為東北行，兩類為西行，一類為臺灣附近海域生成之颱風。若再加入風速及延時為變數時皆以分成四類為最佳，其中三類有轉彎之特性，一類為臺灣附近海域生成之颱風。研究結果顯示隨著經度越偏東，颱風強度也會隨之增加。
本研究應用此分類方法提供一個較直接且明顯的方式進行颱風分類的作業研究，並從分類結果探討每一類型颱風路徑、規模及延時之特性。未來可用此法先將颱風特性加以分類，其結果可提供颱風預報及颱風防災模式之前期輸入使用，以獲得更加計算模擬，以期對未來颱風研究及防災有所貢獻。

關鍵字

群集分析 ； K平均法 ； 華德法

並列摘要

This paper investigates the classification of typhoons driving to Taiwan depending on their paths and scales using the two-stage cluster analysis method. According to the positions of typhoons’ starting and terminal points four categories are separated to indicate that two groups of typhoons move northeastwards, one group goes westwards and one group moves near Taiwan. When the whole length of typhoon’s path is considered as an additional variable to the positions of typhoons’ starting and terminal points, six groups are classified. Those show that three groups of typhoons move northeastwards, two groups go westwards and one group moves near Taiwan. If maximum wind speed and duration of the whole typhoon are included in the other two variables three separated groups are classified to have different turning paths and other one group moves near Taiwan.
The cluster analysis is proposed to more directly and clearly classify the paths, scales and duration of the whole typhoon than the previous methods. The proposed method can afford future researches, such as typhoon wave forecasting or prevention model of typhoon disaster, an early classification of typhoons for highly accurate predictions.

並列關鍵字

cluster analysis ； K-means ； Ward’s method

參考文獻

11. 吳佳蓉(2008)，「颱風路徑之客觀分析方法」，國立中央大學大氣物理研究所碩士論文。

15. Bezdek, J. C. (1981) Pattern recognition with fuzzy objective function algorithms, New York, Plenum.

16. Camargo, S. J. , Robertson, A. W., Gaffney, S. J., Smyth, P. and Ghil, M. (2007a) Cluster analysis of typhoon tracks. Part I: General properties, Journal of Climate, 20, 3635-3653.

17. Camargo, S. J. , Robertson, A. W., Gaffney, S. J., Smyth, P. and Ghil, M. (2007b) Cluster analysis of typhoon tracks. Part Ⅱ: Large-scale circulation and ENSO, Journal of Climate, 20, 3654-3676.

18. MacQueen, J. B. (1967) Some methods for classification and analysis of multivariate observations, Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability, 1, 281-297.