| 摘要: | 台灣由於地理環境特殊，每當夏、秋交際時分常常會有颱風侵襲導致淹水災害發生。近年來的颱風帶來的損失與日俱增，尤以2009年的莫拉克颱風，造成全台灣的總損失至少34億美元當具代表性。由此可見，建立一套完善的颱洪災害風險評估模型是非常重要的。 迄今，國內外已有許多專家學者建構了颱洪災害的風險評估模型，整體來說可歸納出數個不同的主題，如颱風洪水事件資料庫、洪水危害度分析、淹水損失分析及財務損失分析等模組，在相關的文獻中皆有詳盡的介紹及範例分析。儘管如此，其選用之台灣地區颱洪災害颱事件資料庫係以歷史事件為主，根據中央氣象局有發布警報之颱風資料庫之內容，自1961年到2010年共紀錄有239筆資料，略嫌不足；再者，因應氣候變遷，近年來颱風所帶來的最大雨量有時即超越以往之紀錄，若僅依據歷史事件較無法考量此現象。 為了改善上述的問題，本研究將參考既有的歷史事件，根據各流域之間的降雨量相關性、對數常態分佈的統計模型，進行雨量事件之推測。並以一範例呈現在不同事件資料庫背景下之風險評估結果。 Due to the unique geographic conditions, Taiwan is an area where frequent natural catastrophes occur. Whenever summer turns to autumn, the attacks of typhoon often lead to floods. The losses caused by typhoon in recent years grow with each event. Among all typhoon events, the Morakot typhoon in 2009 is the typical case, which brought about at least USD 3,400 million of the total loss. This proves that to build a sound Typhoon/Flood Risk Assessment Model is very important. Until now, throughout the world, many experts and scholars have established the Typhoon/Flood Risk Assessment Models. As a whole, it generalize many subjects, such as Typhoon/flood events database, Flood Hazard Module, Vulnerability Analysis and Financial Analysis Module etc, Nevertheless, the adopted Taiwan typhoon/flood event database are based on historical events. According to the typhoon database that Center Weather Bureau issued typhoon warnings, it is slightly insufficient to have 239 typhoon and flood events in the record from 1961 to 2010. Furthermore, because of the climate change, the maximum rainfall that caused by recent typhoons sometimes exceed the historical records. This situation can’t be taken in consideration if merely standing on historical events. In order to improve the above problems, this research refers to the essential historical events, to proceed stochastic rainfall events based on the statistic model of log-normal distribution and the rainfall correlation of each basin. Moreover, this research provides an example to present the results of risk assessment under different event database. |
