Semi-distributed rainfall-runoff models and representation of losses

By using rainfall-runoff data observed on the Niccone watershed (136 Km2), evidence is given of the large errors involved in the application of the classical Soil Conservation Service-Curve Number (SCS-CN) method to basins of Central Italy. Errors in the total volume of direct runoff up to 100% in magnitude were obtained. Furthermore, the assumption of an initial abstraction as a different fraction, lambda, of the potential maximum retention is not sufficient to account for the inappropriate behaviour of the method. Lastly, by assuming the knowledge of the total volume of direct runoff, a modified estimate of the antecedent moisture condition class and a different lambda led to a simulation of the hydrograph comparable to that obtained, under the same conditions, by an infiltration loss represented by a Smith and Parlange based procedure. As the SCS-CN method was found to be very sensitive to the antecedent moisture condition, a detailed analysis on a large number of watersheds belonging to the same physiographic area should be carried out for adapting the classical relationships for assessing the initial soil moisture.