侵台颱風之時空特性與災害個案分析

Abstract

颱風是台灣最主要的氣象災害，台灣國土面積雖然不大，但由於各縣市自然與社經條件的不同，使得颱風災害有很明顯的空間差異。「空間」向來是地理學重要的傳統，「災害風險研究」更是近年來國內外方興未艾的重要研究主題。本研究以侵台颱風為對象，分析過去五十年侵台颱風數量、強度與路徑特性及其與死亡人數的定量關係。再由災害風險的構成要素，災害、暴露與脆弱性針對近八年的颱風死亡災害進行綜合風險分析與評估。 研究結果發現，過去五十年來各年代颱風造成的死亡人數呈現減少、持平、然後再增加的趨勢，顯示颱風死亡人數並未隨著科技與經濟的進步持續減少。單從颱風的強度、數量與路徑無法完全解釋死亡人數的變化。利用集群分析選出死亡人數較高的重大颱風事件，並濾除平均值以下的颱風事件後，死亡人數與颱風個數的相關係數可達0.9 以上。顯示重大颱風事件在統計上的關鍵影響。此外相較於過去，近十年侵台颱風的數量大幅增加，路徑也有明顯的變化。 本研究進一步從整體統計數據中解析出兩大集群，即一般颱風事件與重大颱風事件。一般颱風事件中，死亡個案多發生於戶外，且死亡人口屬於較易外出暴露的中壯年男性人口。這部份可以從男女的災害識覺差異、文化背景與職業特性上來解釋。對於重大颱風事件而言，經常是以家戶或整個村落為受災單位，因此在年齡與性別上的分布上就較為平均。 一般颱風事件的死亡災害多分散於各縣市，即無明顯的區域特徵，且死亡風險最高的階段幾乎都不是在風雨最強的時候，而是在風雨較弱的階段。重大颱風事件則可分為都會型與山地鄉鎮型兩類。山地鄉鎮型的死亡風險與高災害潛勢地區的重疊性較高，與降雨的相關性也較好。相較於鄉鎮地區，都會區則由於較佳的防災工程與應變能力，災害潛勢與死亡風險的相關性則較低。Typhoon is the major meteorological disaster in Taiwan. Although Taiwan’s territory is not large, yet because the natural and social condition difference in each county and city, Typhoon disaster has very obvious spatial difference in Taiwan. “Space” is always very important tradition in geography, and “disaster risk research” is even an ever-increasing and important domestic and oversea research topic in recent years. In this study, typhoon in Taiwan is used as the research target to analyze the typhoons in Taiwan for the past five decades in terms of quantity, strength and path characteristic and its quantitative relationship with the number of death it caused. Moreover, from the compositional factors of disaster risk, for example, hazard, exposure and vulnerability, general risk analysis and assessment for the typhoon death disaster in the recent eight years is performed. It can be seen from the research result that in the past five decades, the number of death caused by typhoon in each year shows a trend of decrease, flatness and then increase, which shows that the death caused by typhoon does not show a continuous decreasing trend along with the advancement in technology and economy. If we just simply consider from the point of view of typhoon’s strength, quantity and path, it would be impossible to explain the change in the death number completely. If we use Cluster Analysis to sort out major typhoon accidents with higher death number and filter out typhoon accidents that are below the average value, then the correlation coefficient between death number and the number of typhoon can be above 0.9. This shows the key influence of major typhoon accidents statistically. In addition, as compared to the past, the quantity of the typhoon that invaded Taiwan in the recent decade shows great increase, and the path also shows obvious change. In this study, two major clusters are come out further from the entire statistical data, namely, general typhoon accident and major typhoon accident. In general typhoon accident, the death case usually occurs outside, and the death population belongs to medium age male persons who go out frequently and have the chance to expose themselves. This part can be explained in terms of disaster consciousness difference, cultural background and vocational characteristic difference between male and female. For major typhoon accident, the disaster-receiving unit is usually a family or an entire village; hence, the distribution is more even in terms of age and gender. The death disaster of general typhoon accident usually distributes among many cities and counties, that is, there is no obvious area characteristic; moreover, the stage with highest death risk is almost not at the time when the rain and wind are strongest, instead, it is at stage with weaker rain and wind. Furthermore, major typhoon accident can be divided into metropolitan and mountain and rural type. The death risk for mountain and rural type has higher overlap with the potentially higher disaster region, and it also has better correlation to the rainfall. As compared to the rural area, the metropolitan area, due to better anti-disaster engineering and accident-handling capability, usually has lower correlation between disaster potential and death risk.