Improved SCS-CN Methodology Incorporating Storm Duration and Temporally Decaying Retention for Enhanced Runoff Prediction

Abstract

This study presents novel mathematical formulations of the Soil Conservation Service curve number (SCS-CN) method that incorporate both temporally decaying retention parameters and storm intensity/duration. To evaluate its performance, we compared it with the existing versions of the SCS-CN model using a large data set of 35,546 storm events of 113 different US watersheds. Obtained results indicate that the proposed model outperforms other models in almost all 113 US watersheds with the highest Nash-Sutcliffe efficiency (NSE). Furthermore, the results are supported by the percent bias (PBIAS) being close to 0 and the lowest root mean square error (RMSE), RMSE-observations standard deviation ratio (RSR), normalized root mean square error (NRMSE), and mean absolute error (MAE) statistics. The general form of the proposed model performed particularly well in clayey and sandy soils with different land uses and catchment areas larger than 1 ha. Rainfall (P) and the coefficient (β) parameters are identified as the most and least sensitive parameters of the proposed model, respectively.