圖書典藏及影音數位平台

本報告整理石門水庫集水區土壤沖蝕、崩塌體積、與泥沙遞移之估算工具和量測方法，含土壤沖蝕量量測方法(包括沖蝕釘法、放射性核種鈹7法與放射性核種銫137法)、土壤沖蝕量估算模式(包括USLE、指標法、MUSLE與WEPP模式)、崩塌地產砂量量測方法(包括航照圖與衛星影像判釋、全測站量測與光達量測)、崩塌地體積估算模式(包括冪次法則、打荻珠男經驗式、面積深度經驗式、坡度深度經驗式)、與崩塌地潛勢預測模式(邏輯迴歸模式)，總結出集水區在短期應變與長期管理適合的估算方式。崩塌地潛勢預測模式使用邏輯迴歸模型，在艾利颱風期間用了都卜勒天氣雷達回波率數據計算雨量，並使用最大三小時降雨強度與總降雨延時在邏輯迴歸模型中作為解釋變數，使用海棠颱風造成的崩塌地驗證結果準確率達到82%，由於在颱風期間可以計算出在一段時距中崩塌地發生的機率，故此網格為基礎的模型可以成為崩塌地預警與即時監測系統的一部分，文中並將此成果應用在決策支援系統中。由11個測站的現地採樣與實驗結果，在颱洪事件時於現地採取水樣做驗證，現地水樣與室內實驗之曲線大致符合，可提供各站泥沙濃度與濁度轉換公式供後續應變參考。此外，藉由歷史流量與泥沙濃度資料，可找出不同颱風事件下集水區泥砂來源。文中決策支援系統建置包含崩塌地預測模式、泥沙產量估算模式、崩塌災害緊急應變決策系統、集水區流量與泥沙濃度濁度推估模式、和管考系統。整治成效分為人命保全、崩塌復育、坡地土砂控制、河道泥砂濃度控制、與水庫平均淤積量控制成效五個面向來探討，在崩塌地復育成效方面，歷年衛星影像判釋結果顯示，95年整治計畫開始後，歷次颱風產生的新生成崩塌地有明顯減少的現象，總復育面積共計853.6公頃，至100年底復育率為80.4%。各年度檢討報告與第一和二階段成果報告以及經過各NGO團體、政府單位、與專家學者討論過的永續石門藍圖也完成在報告中。

In this project, we collected all related reports about estimation and measurements of soil erosion (erosion pins method, isotopes 7Be and 137Cs methods, USLE, soil erosion index method, MUSLE, and WEPP model), landslide volume (interpretation of air photo or satellite image, total station measurement, lidar measurement, power law method, Uchihugi empirical model, area-depth estimation method, slope-depth estimation, logistical regression model), and sediment delivery to find out the proper way for watershed management. The landslide prediction model derives a logit model to compute landslide occurrence probabilities. To demonstrate the effectiveness of this method, the model used radar rainfall estimates, 3-hours rainfall intensity, and rainfall duration as explanatory variables during Aere typhoon event and the same types of data associated with Haitang typhoon to validate the model. The model had a modified success rate of 82% for predicting landslides and stable areas. According to the estimation of landslide occurrence probabilities during typhoon events, it is better suited for decision-making in watershed management and also included in decision support system. Based on the experiment results, we provide regression formulae to conversion between sediment concentration and turbidity for 11 stations. The regression formulae fit well after the comparisons between experiment data and field measurements. Besides, based on the discharge and concentration of suspended sediment data, we identified the sediment sources in the watershed during typhoon events. To construct the decision supporting system, we built up landslide prediction model, sediment yield estimation model, response and recovery on landslide hazards system, discharge and concentration/turbidity of suspended sediment prediction model, and management system for managers in the watershed. The watershed management effectiveness, which included preservation of security of life, restoration of landslide, sediment control on the hillslope, sediment concentration control in the river, and sedimentation in the reservoir, was discussed. For the landslide restoration, the area of new landslides decayed from the analyses of interpretation of satellite images during 2006 to 2011. The area and percentage of landslide restoration amounted to 853.6 hectares and 80.4% until 2011, respectively. Annual performance reports during 2009-2011 were also represented. In addition, the blueprint of sustainable Shihmen watershed was accomplished from several discussions among non-government organizations, governments, experts, and professors.