水文網路的自動析取吸引了許多研究者的興趣。它不僅在理論上有許多值得研究的題目，也具有實際的應用價值。就地理資訊系統的技術發展而言，新方法的開發和現有方法的評估應該是不可偏廢的，但是前人研究大都偏重在自動析取方法的開發，較少就現有的方法進行評估。從現有文獻的分析中，本文作者認為水文法是現有方法中表現較好的一種，本研究乃針對GRASS 系統所提供的水文法進行準確性研究。研究中以四組臺灣地區的DTM資料進行測試，並且設計三個檢測方法，分別從宏觀和微觀的層面來檢測水文法自動析取結果的準確性。室內研究工作是以暈渲圖和人工判讀等方式來分析，室外檢測工作則利用全球定位系統來進行比對。綜合三個檢測的結果顯示，GRASS 系統所析取的水文網路，與人工判讀的結果非常接近，和野外實測的水系位置也非常接近。本研究肯定了水文法自動析取之準確性。後續研究將進一步探討水文法中所衍生的參數在地形學及水文學的應用。

　　　　　The automated extraction of draginage networks has attracted much research interest. It constitutes not only many profound research topics, but also great potential to pragmatic applications. For a healthful development of GIS, the evaluation of existing methods deserves the same amount of attention as the development of new methods does. However, most previous research focussed on the creation of new methods and overlooked the evaluation of existing methods. This paper aims to evaluate a popular method for the automated extraction of ridges and valley lines. The drainage networks of four DTM data sets were extracted by this author using the GRASS package. The resulting features were analyzed in micro and macro levels using three approaches: evaluation based on hill-shading maps, comparison to manual delineation, and field check using GPS. The result of these evaluations concludes that the automated extraction performed by GRASS correctly delineates drainage networks on a grid DTM. Future research will further explore the applicability of automated extraction to hydrologic and geomorphologic study.