本研究分別採用 (1) 降雨完全入滲與 (2) 降雨部份入滲兩種不同入滲模擬情境之土壤水分演算模式 (Soil Moisture Routing Model, SMR)，以南投蓮華池4號與5 號集水區為例，進行1993-1994 年逐日流量估算與集水區內土壤體積含水量空間分布模擬，並對照量水堰實測流量與現地土壤含水量調查資料，以檢核兩種降雨入滲情境之模擬效度。數據指出，兩種降雨入滲情境之Nash-Sutcliffe 流量模擬效度指數 (R2) 差異不大 (降雨完全入滲之模擬效度為0.87 與0.80；採降雨部份入滲情境之模擬效度為0.87 與0.78)，但採降雨完全入滲情境之土壤體積
含水量空間分布推估結果較符合現地實測值。而土壤水分演算模式無論採用何種入滲情境進行模擬，於濕季與土壤水分消退時段均具較高之流量預測能力；但乾季時其演算過程則可能受集水區地表複雜度與土壤層含水量急劇變化影響，而產生較差之模擬效果。綜合上述研究成果，可發現降雨完全入滲之設定較適用於蓮華池4 號與5 號集水區。

Two different circumstances, total rainfall and partial rainfall infiltrating into soil layers, are applied in this research. We apply the Soil Moisture Routing Model (SMR)for hydrologic simulation in Lien-Hwa-Chi Watershed No. 4 and No. 5 during the period between 1993 and 1994, and compare the quantity of streamflow and spatial distribution of soil water content counted from both different circumstances. Results
show that different circumstances produce almost equal accuracy of streamflow simulation. The values of Nash-Sutcliffe efficiency are 0.87 and 0.80 for the total rainfall infiltrating into soil circumstance for Watershed No. 4 and No. 5 respectively;while the values are 0.87 and 0.78 based on the partial rainfall infiltrating into soil circumstance. Comparing the different seasons, these two circumstances produce
better efficiencies of streamflow simulation during wet season. Moreover, comparing the distribution of soil moisture content on watersheds, the total rainfall infiltrating into soil circumstance is more reasonable. Overall, the total rainfall infiltrating into soil mechanism is more suitable for the hydrologic simulation in the Lien-Hwa-Chi watersheds, including quantity of runoff generation and soil moisture content.