台灣地處環太平洋地震帶上及亞熱帶氣候區,由於地理環境特殊,因此颱風帶來的豪雨及地震的頻繁是台灣無法避免也必須面對的天然災害。加上台灣面積三萬六千平方公里,平地僅占全台面積的26%,都市的人口過度密集,土地利用有限,便逐漸往山區遷移,大量的開發山坡地,導致山坡地的超限利用及不當開發,颱風、地震隨著山坡地的大規模發展,造成許多的人員及財產的傷亡與損失。 截至目前為止台北市政府產業發展局山坡地歷史災害記錄共收集達994筆(1959~2008),由資料可得知,由於現今山坡地人口快速增加,災害影響市民生活甚鉅,所以山坡地災害的預測及防災是極為重要的課題。本研究將台北市分割成1km×1km的網格,並利用坡地災害相關因子及坡地歷史災害資料,處理製作成GIS圖層,利用MapInfo進行各災害相關因子圖層與網格圖層的疊加,並計算各個圖層疊加後的指標分數。 圖層資料經過前置處理後,利用兩種分析方法:多變數迴歸分析及類神經網路分析評估綜合指標模型的有效性,再將坡地災害相關因子及坡地歷史災害資料進行迴歸分析,探討迴歸分析出的數據,並將坡地歷史災害資料點位做空間分布統計分析,討論這些歷史災害點位之分布是否為均勻,本研究以Google Earth作為整合平台,將最後分析出的結果以視覺化展示,以提供給區域計畫或土地利用者參考。
Taiwan is located in the Pacific Rim earthquake belt and sub-tropical climate zone. This special geographical environment renders Taiwan vulnerable to natural disasters, notably typhoons and earthquakes. Moreover, plains account for only 26% of Taiwan’s total area of 36,000 square kilometers, resulting in the dense population of the island’s major cities. An increased number of hillside communities have been developed, and excessive and improper hillside development aggravates into a serious problem causing great casualties and property loss in times of typhoons and earthquakes. According to the official statistics published by the Department of Economic Development, Taipei City Government, 994 hillside disasters have been recorded during the period from 1959 to 2008 As suggested by the historical data, forecast and prevention of hillside disasters have become a crucial issue for the safety and wellbeing of residents in Taipei City that has witnessed a rapid growth of hillside population in recent years. This study accordingly divides Taipei City into 1km × 1km grids and consults the triggering factors and historical data of hillside disasters so as to develop GIS layers by utilizing the MapInfo analysis application. Layers related to the triggering factors of hillside disasters can then be obtained and placed on the grid map to help calculate the index scores of the triggering factors. After the preliminary processing of layers data, two methods - multi-variable regression analysis and neural network analysis model – were adopted to measure the effectiveness of the integrated indicators. Regression analysis was then performed to examine the triggering factors and historical data of hillside disasters, followed by a statistical analysis on the spatial distribution of hillside disaster locations based on historical data to check if the distribution was even. The final analysis results were visualized via the integration platform of Google Earth to provide regional planning and land use references.