以空間資訊技術與布林粗糙集合分析雪霸崩塌地石礫土與壤土之差異性研究

山坡地發生崩塌後，因交通常被中斷或原本無道路通往以致救災或勘災人員無法到達該地進行勘測，而且其誘發崩塌的因子錯綜複雜，因此本研究利用空間資訊技術以獲取地表土地覆蓋之情形，並以布林粗糙集合論的集合案例分類判斷找出門檻值，建立其規則知識庫。研究中以遙感探測的方式來收集地表資料，進行研究區（雪霸國家公園）大範圍面積之監測，並用數值高程模型(DEM)表達地表型態之連續起伏，透過影像處理分析地表植被分佈之情形，且配合地理資訊系統強大之空間分析功能，依災害發生後之崩塌地進行災情評估，獲得有效且正確之災害量化資訊，並建立其規則資料庫，用以提高效率和緊急處理機制，並防範崩塌發生後，可能引發之土砂災害或不良影響，同時並考慮雪霸地區主要兩種土壤－石礫土與壤土之差異性研究：包括主要影響因子和知識界限圖。研究成果顯示，主要在研究區可得知最重要的影響因子依次為差異化常態植生指數(NDVI)、植生指數(VI)、高程、距道路之距離等重要屬性因子，並建立崩塌災害潛勢圖，對於判釋崩塌地有一定分類之準確率，以便能快速、即時在崩塌災害監測與管理上提出對策。

關鍵字

布林粗糙集合理訪；遙感探測；數值高程模型；差異化常態植生指數；植生指數

並列摘要

Landslides usually resulted in the broken down of the transportation where the personnel are unreachable. From the past investigations, governing factors of influencing landslide occurrences are complicate. To resolve the aforementioned problems, this research focused on the spatial information technology to attain the information of geology data. Firstly, we collect the surface material by remote sensing for monitoring the vast scope of study area on Shei-Pa National Park. The Boolean Rough Set information classification system is used as a tool to find the threshold with respect to occurrence of landslide in each attribute based upon the knowledge database. The GIS, remote sensing and digital elevation model (DEM) is used to attain the attribute values the surface of the earth. As the landslide occurred, this study offers a systematic solution on investigating the disaster. It can not only obtain effective factors but also compute the quantification values for analysis. On the other hand, the major soil component of loam and gravel on Shei-Pa area resulted in different landslide problem. The major factors and knowledge scope are successfully analyzed. Finally, the contribution of this research renders a rule-based knowledge database to provide better understanding on the occurrence of landslide. This rule-based knowledge database provides an effective and urgent system to manage landslide. The contributions of this study are shown as: (1) NDVI (normalization difference vegetation index), VI (vegetation index), elevation, distance from the road are the major influenced factors for landslide. (2) Meanwhile, the hazard landslide potential diagrams (landslide susceptibility map) are drawn and a rational accuracy rate of landslide is calculated. Thus, an instant and on-time monitoring to manage landslide strategy can be made.