4.2 Model application
The original SCS-CN and the proposed method (Methods 1, 2 and 3) with the calibrated and validated parameters were applied to predict runoff from the remaining three plots in the CBG watershed.
The predicted runoff values using the original SCS-CN and the proposed method plotted against the corresponding measurement values in plots of the CBG watershed was shown in Fig. 4. The original SCS-CN method continue to under-predict most of the large as well as small storm-runoff events, whereas Method 1 performed better than the original SCS-CN method with the former yielded a larger value of NSE(75.33%) and a lower value of RMSE (2.38 mm) (Fig. 4). However, Method 1 with λ =0.2 still underestimated some small storm-runoff events as compared with Method 2 (λ =0.001) which achieved higher NSEvalues of 82.29% and most of the data points were reasonably closer to the 1:1 line. Moreover, Method 2 using the parameters obtained from Huang et al.(2006) with data from slopes ranging from 14 % to 140 % seems perform better than Method 3 (NSE =79.22%) which using the parameters optimized with narrow range of slope (26.7% and 70%). Therefore, based on the results obtained from the experimental sites, the use of Method 2 is the most appropriate for runoff prediction with the SCS-CN method in the study area, and possibly in the Loess Plateau.