Discussion of “NRCS Curve Number Method: Comparison of Methods for Estimating the Curve Number from Rainfall-Runoff Data”

The authors have reported on the important topic of estimating curve numbers (CN) from field measurements of rainfall and associated runoff.Their paper focuses on the performance of the curve-number method using curve numbers derived from various calibration methods.For each calibration method, two different initial abstraction ratios were considered (0.2 and 0.05), and also two different rainfall thresholds were considered (0 and 2.54 cm).The analyses were done using data collected at 31 USDA Agricultural Research Service (ARS) experimental watersheds.Based on their findings, the authors recommend using the least-squares (LS) method to estimate the curve number from field data, and also report that there was no significant improvement in the performance of the calibrated curve-number model in using an initial abstraction ratio of 0.05 versus the conventional value of 0.2.There are several concerns with the analyses and findings reported by the authors.First, the authors overlook the fact that curve numbers derived using different calibration methods can have fundamentally different definitions, and so calibrated curve numbers are not always to be interpreted and applied in the same way.Second, the authors report that the commonly used empirical exponential relation between the curve number and rainfall amount is fundamentally flawed and reflects spurious correlation; this assertion is not correct.Third, the authors have overlooked the fact that the initial abstraction ratio is a key determinant of the threshold rainfall amount for runoff to begin, and so could potentially be important when the curve-number method is applied to lowrainfall events.This discussion elaborates on the aforementioned concerns. Estimation Methods and Definitions