氣候變遷下牡丹水庫集水區坡地災害評估

Abstract

臺灣地區雨量豐沛，常造成山區崩塌，以致水庫淤積嚴重，無法有效儲存水資源，且當坡地災害發生，對於生命財產損失或環境破壞都相當嚴重。近年氣候變遷影響加劇，逐漸改變臺灣的降雨時空分布，水資源議題和用水安全也備受關注，因此坡地災害評估成為預防崩塌災害的重要工作。本研究應用地理資訊系統(GIS)技術，萃取內政部20公尺網格數值地形模型(DTM)資料獲取坡度、坡向、高程，開放街圖(OpenStreetMap, OSM)之距水系距離、距道路距離，經濟部中央地質調查所「五萬分之一台灣區域地質圖」地質和地質敏感區，常態化差值植生指標(Normalized Difference Vegetation Index , NDVI)，搭配GrADS分析時雨量、降雨動能和等10個因子並搭配極端降雨指標進行相關研究。再以不安定指數計算其變異係數、標準差、權重，推論其中的關鍵因子。最後以現有崩塌資料，透過地理資訊系統進行風險區的繪製分析，再以接受者作業特徵曲線(Receiver Operating Characteristic curve , ROC曲線)來計算模型準確率達 80%~90%，並以預報驗證方式進行多個指標的計算。研究結果發現，因子中影響程度最大者為NDVI，之後依序為、坡度、地質敏感區、地質、降雨動能、坡向、距道路距離、距河流距離、降雨強度、高程。利用不安定指數和風險評估概念建立基期坡地災害風險評估模型後，本研究利用動力降尺度WRF-HiRAM、WRF-MRI計算氣候變遷下基期和世紀末極端降雨發生的頻率和強度變化，進一步探討世紀末RCP8.5情境下(表示各國對於溫室氣體並無任何減量措施)極端降雨對牡丹集水區坡地災害風險的影響，此結果可供未來坡地管理及水土保持之參考。

Abundant rainfall in Taiwan often causes mountainous areas to landslide, resulting in serious reservoir siltation, unable to effectively store water resources, when slope disasters occur, the loss of life and property or environmental damage are quite serious. In recent years, the impact of climate change has intensified, gradually changing the temporal and spatial distribution of rainfall in Taiwan, water resource issues and water security have also attracted much attention. Therefore, landslide hazard assessment has become an important task in preventing landslide disasters. In this study, the geographic information system (GIS) technology was used to extract the 20-meter grid numerical terrain model (DTM) data of the Ministry of the Interior to obtain the slope, aspect, and elevation. Open Street Map (OSM) distance to water system, distance to road. Geology and Geologically Sensitive Area from Central Geological Survey, Ministry of Economic Affairs"1/50,000 Taiwan Regional Geological Map" geology. Normalized difference vegetation index (Normalized Difference Vegetation Index, NDVI), with GrADS to analyze hourly rainfall, rainfall kinetic energy, and daily rainfall, 10 factors in total and to conduct related research with extreme rainfall indicators. Then calculate its coefficient of variation, standard deviation, and weight with the instability index, and deduce the key factors. Finally, based on the existing landslide data, ArcGIS is used to draw and analyze the risk area, and then the receiver operating characteristic curve (ROC curve) is used to calculate the model accuracy rate of 80% to 90%, and the calculation of multiple indicators is carried out by means of forecast verification. The research results found that NDVI is the most influential factor, followed by, slope, geological sensitive area, geology, rainfall kinetic energy, slope aspect, distance from road, distance from river, rainfall intensity, and elevation.After using the instability index and the concept of risk assessment to establish a risk assessment model for landslide disasters in the base period, this study uses Dynamical downscaling data such as WRF-HiRAM and WRF-MRI to calculate the frequency and intensity of extreme rainfall in the base period and the end of the century under climate change, and further explore the impact of extreme rainfall on slope disaster risk in the Mudan reservoir watershed area under the RCP8.5 scenario (indicates that countries do not have any reduction measures for greenhouse gases) at the end of the century, the results can be used as a reference for future slope management and soil and water conservation.