5 Discussion
From the above results, the original SCS-CN method did not exhibit good calibration and validation performances for the three plots in the XDG watersheds, it still underestimated multiple soil loss events, and it also performed poorly when applied to the plots of CBG watershed. Although Huang et al.(2006) and Huang et al.(2007) method with the single factor of slope and soil moisture were considered respectively, could enhance the performance to some extent, the improvement is limited. Overall, the above three methods did not perform as well as the proposed method. The superior performance of the proposed method suggests it can accurately predict runoff for different vegetation using the optimized (calibration) parameter for the six plots in the XDG and CBG watersheds. The results also provide some insight into the physical processes represented by incorporating slope, soil moisture, and storm duration factors in the proposed method, which eliminated the major shortcomings of the traditional SCS-CN model.