Stability assessment of the curve number methodology used to estimate excess rainfall in forest-dominated watersheds

Abstract

The credibility and wide acceptance of the curve number (CN) method are rooted in its simplicity, convenience, authoritative agency support, absence of a strong competitor, and versatility to estimate runoff especially from ungauged watershed. Numerous studies have reported that the application of the tabulated CN (CNTab) and the fixed initial abstraction ratio (lambda = 0.20) to estimate the runoff volume is inferior. In the current study, we investigated the applicability of lambda = 0.20 and its successful modification to lambda = 0.05 using the CNTab and other calibrated CNs based on the measured rainfall-runoff data. In addition, with a fixed CNTab, the value of. was reduced to 0.01 and 0.0 in order to evaluate the runoff variability response for 587 storm events from 10 South Korean forest-dominated mountainous watersheds. Both for lambda = 0.20 and lambda = 0.05, the application of CNTab exhibited poor results in all of the watersheds based on quantitative statistics using four performance evaluation indicators. The application of the calibrated CN increased the runoff prediction capability of the CN method both for lambda = 0.20 and lambda = 0.05. Using lambda = 0.05 and its adjusted calibrated CN depicted a modest improvement in the results. Among the different calibrated CN variants, the asymptotic method was comparatively appropriate. However, the method still needs improvement for a consistent CN. Interestingly, for the unaltered CNTab, the lambda = 0.01 and lambda = 0.0 exhibited very consistent results for the estimation of runoff from watersheds. Therefore, it is more tempting for application in ungauged watersheds without deterioration of the CN model's versatility and simplicity.