本研究以南投縣水里鄉新山村為崩塌潛勢分析之研究對象,利用桃芝颱風、七二水災、莫拉克颱風不同時期之SPOT-5衛星影像結合數值高程模型(DEM)資料,應用地理資訊系統(GIS)、常態化差異植生指標(NDVI)及影像分類法(非監督式分類)進行崩塌地判釋。再將崩塌判釋結果與所選定之因子進行套疊,統計出各因子中的崩塌百分比及權重值,最後將本區域分為五個等級潛感程度,繪製崩塌潛感圖。 研究發現其高潛勢密集度以沿著河道周圍之坡地最為顯著,集中在陳有蘭溪主流河道及土石流潛勢溪流:平坑(DF170),南平坑(DF169)周圍。其崩塌因子對研究區域內邊坡穩定性影響較大者依序為植生覆蓋、坡向、地質因子。
The research was conducted to study the natural hazards of regional landslide potential in Hsin-shan village in Nantou, Taiwan, using the SPOT-5 satellite images produced at selected time periods before and after several floods resulted from storm/ typhone in recent years. These tropical cyclones include Toraji, Mindulle, and Morakot which caused catastrophic damages in many areas near the central mountain ridge in Taiwan including the study area. In the study, the satellite images were combined with the data from Digital Elevation Model (DEM), Geographic Information System (GIS), Normalized Difference Vegetation Index (NDVI), and the unsupervised image classification to evaluate the landslide potential of the study area which was divided into a 20m x 20m grid system. Results of the landslide potential evaluation were then super-positioned and analyzed with the maps representing 5 selected factors related to geographic and site conditions to reveal the weight and % area of landslide due to these studied factors. Based on the results of the study, the target area was then categorized into a 5 landslide potential level system from which a landslide potential map was developed as final product of the research. The developed hazard potential map can be used for future reference and use in preparedness/rescue plans in the event of severe floods from storms or typhones in region. The results of the study revealed that the areas on or near mountain slope along the major rivers have the highest landslide potential in the study area, which include Chen Yu-Lan river, Pen-Kang (DF170), and south Peng-Kang (DF169), where landslide has been a long term threat for several decades. In addition, site factors such as surface coverage, slope orientation and geological condition also proven to be significantly affecting landslide potential in the study area.