莫拉克颱風以超過2500mm 雨量侵襲台灣地區，對中南部和東部山區造成嚴重崩塌及堰塞
湖，對災害治理機關而言，如何有效地評估崩塌地重點區位為重要課題。本研究結合遙測技術
及非監督式類神經網路，發展快速萃取風災崩塌區位之模式，並以保全對象為評估因子，計算
崩塌規模、其與建物、橋樑、道路及水系距離等因子，輔以多目標決策模式－PROMETHEE，
探討莫拉克風災嚴重之小林村崩塌地重點治理區位，以集水區為地形單元，探討十大重點治理
崩塌區位之影響因子，以作為崩塌地治理及防災之參考。

Morakot typhoon with over 2500 mm rainfall caused massive landslides and several landslide dams in southern and eastern Taiwan. Therefore, it became an important issue for monitoring and assessing the treatment area of the triggered landslide. In this study, multi-temporal SPOT satellite
images coupled with unsupervised neural network model were developed to extract the landslide sites.Five factors of the protected targets, such as landslide area, the relative distances to buildings, bridges,
roads and streams, in the landslide watershed were calculated and used to assess the priority of the treatment sites via PROMETHEE technique. Xialin villiage with serious landslide occurring, located in southern Taiwan, was chosen. Ten priorities of treatment landslide sites were assessed for discussing the affecting factors. The analyzed results can provide useful information for decision making in the landslide area.