本研究以臺灣新店溪流域寶橋集水區為例，搜集前人應用SPOT衛星影像所判釋出1989、1990、1993、1994、1995、1997、1998、1999、2001年研究區各種土地利用型態後，採用美國農業部水土保持局所研發的逕流曲線指數法，分析出上述九個時期中，不同土地利用型態下的標準逕流曲線值，再依據歷年雨量資料整理出上述各年的前期五日降雨型態，將逕流曲線值加以修正後，求出年逕流係數值。本研究更進一步運用多元迴歸分析方法，探討研究區土地利用型態與逕流特性二者間的關係。結果顯示，利用林地、農地、草地、都市用地、水體這五種地表土地利用型態，所能說明的年逕流係數值的總變異程度，達80%以上。改變集水區土地利用的各種狀態，對逕流量造成不同程度的影響。大面積砍伐森林，會引起地表逕流量大幅增加，若變更現有林地為草地或都市用地，逕流係數也有增加的趨勢，但增加幅度比變更為農地較少。隨著農地面積的減少，逕流係數也會變少。尤其是農地變更為林地，其次為都市用地，再其次為草地。林地土地利用型態的增加，有助於地表逕流量、沖蝕量、泥砂產量及養分流失的減少，確保水土資源持續開發利用。
his study uses Paochiao Watershed which is a sub-watershed of Hsintien Stream Watershed as the experimental site. Nine sets of SPOT satellite images for the year 1989, 1990, 1993, 1994, 1995, 1997, 1998, 1999, and 2001 are gathered and analyzed to determine the land use pattern at the study site. Curve Number (CN) values, determined according to the US Soil Conservation Service, are computed for each time period under different land use pattern. These values are further adjusted based on 5 days antecedent rainfall pattern derived from long-term historical annual and seasonal rainfall records. Annual runoff coefficients are also computed for each time period. Multivariate regression analysis is performed between these runoff coefficients and the land use pattern. Results indicate that significant regression relationships exist between runoff coefficient and the area size of forest-, agricultural-, grass-, urban-land, and water body. Over 80% of the variation can be explained by the regression relationship.
