土壤沖蝕造成的水庫泥沙淤積是集水區管理的重要課題。欲解決問題須先了解集水區泥砂來源和哪些因子影響集水區泥沙移動情形,由於有限的資料,往往很難訂定有效的決策。另一個替代方法是用土壤中放射性核種銫137來估算集水區土壤沖蝕量與分布情況。 藉著石門水庫白石溪集水區現場288個土壤樣本採集,估算出集水區總淨沖蝕量108,346噸/年、平均沖蝕量與淨沖蝕量為10.1和9噸/年-公頃;由3個坡頂到坡底的採樣設計,158個樣本點回歸分析結果顯示土壤沖蝕和堆積空間分布情形與樣本點相對位置、高程及土壤密度有關;由樣本點相對位置與土壤密度建置的多變數回歸模式可以解釋其中30%的變異性。 在三光溪集水區,484個土壤樣本資料建置出集水區土壤重新分布情形、標示出沖蝕嚴重區和釐清集水區泥砂來源。崩塌地、表土沖蝕和蝕溝沖蝕各佔三光溪集水區泥砂來源71%、23%和6%。此外,研究中使用土壤重新分布率與地層上升速率模擬出集水區未來100年、1,000年與10,000年的地形地貌變化,結論出影響地貌變化的因子是崩塌地而不是表土沖蝕。 在玉峰集水區405個土壤樣本分析,得到最大土壤沖蝕率167噸/年-公頃、最大堆積率155噸/年-公頃與平均沖蝕率37噸/年-公頃,且土壤重新分布情形與農地位置有關。土壤沖蝕分布情形可以當作決策支援系統輸入的重要資訊,來擬定集水區優先整治的順序。此決策支援系統也可以模擬情境分析,展示整治經費和整治效益,更能幫助決策者節省時間擬定整治保育之對策。
Sedimentation from soil erosion is a critical reservoir watershed management issue. The formulation of watershed management strategies to protect water resources threatened by soil erosion and sedimentation requires a thorough understanding of sediment sources and factors that drive soil movement in the watershed. Due to limited data sources, informed decisions are not always easy to make. An effective alternative to estimating soil erosion is to analyze the spatial variation of 137Cs inventory in the soil. In the Baishi watershed, the net soil loss estimate from 288 soil samples is 108,346 t yr-1 and the gross erosion and net erosion rates are 10.1 and 9 t ha-1 yr-1 respectively. With 158 samples collected in three transects on hillslopes, correlation analysis results show that the spatial variability of soil erosion and deposition rates is associated with the relative position of a sample site to the nearest ridge, elevation, and soil bulk densities (with or without particles smaller than 2 mm). A multiple regression model with the relative position and soil bulk densities as the explanatory variables explains 30% of the total variation in soil redistribution rates. In the Sanguang watershed, the 137Cs measurements of 484 soil samples were taken to construct soil redistribution pattern, locate seriously eroded sites, and identify sediment sources. Landslide, surface soil loss, and gullies account for 71%, 23%, and 6% of the sediment sources respectively in the Sanguang watershed. In addition, this study simulates landscape evolution for 100, 1000, and 10000 years by using soil redistribution and tectonic uplifting rates and suggests that landscape evolution is dominated by landsliding, rather than surface soil loss. By taking 405 soil samples in the Yufeng watershed, the average soil loss is -37 t ha-1 yr-1, with the maximum erosion rate of -167 t ha-1 yr-1 and the maximum deposition rate of 155 t ha-1 yr-1. The soil redistribution pattern appears to be associated with farm locations. This information is then used as input to a decision support system (DSS), a geographic information system- (GIS-) and knowledge-based model for prioritizing areas for protection in watershed management. The DSS can evaluate different scenarios, present saved sediments, and perform cost and effect analysis to save the managers time and help them target specific areas for effective watershed protection and conservation.