A USLE and SCS-CN coupled approach for design sediment yield prediction

Knowledge of sediment yield is essential for predicting and mitigating the impact of natural disasters such as floods and landslides as well as for managing water resources and ecosystems of a region. It has been found that a considerable portion of sediment yield is sometimes generated from extreme rainfall events of high magnitude and intensity compared to that from myriad small rain events. Therefore, it is vital to accurately predict sediment yield resulting from extreme storms of varying durations, especially from data-scarce regions. This study proposes an empirical approach based on the Universal Soil Loss Equation (USLE) and Soil Conservation Service-Curve Number (SCS-CN) methods integrated with a sediment yield model to predict sediment yield resulting from a storm of desired duration (d) and recurrence interval (T). To this end, the potential erosion (A) is empirically related to ‘d’ and ‘T’ and the empirical relation is calibrated and validated on the data of ten sub-watersheds of Ashti catchment, India, involving annual maximum rainfall (observed), runoff (daily observed) and sediment (daily SWAT simulated). The model performance is evaluated using Nash-Sutcliffe Efficiency (NSE), Coefficient of Determination (R²), Percent Bias (PBIAS), Normalized Root Mean Square Error (nRMSE), and visually by scatter plots. The model was calibrated with high NSE, low nRMSE and PBIAS values in all the sub-watersheds (NSE>0.85, PBIAS< ±10% and 0.156< nRMSE <0.216). In validation, the performance was also excellent (0.77≤ NSE ≤0.98 mean value = 0.86, PBIAS ≤ ±20%, and 0.86≤ R²≤ 0.99 mean value = 0.95) in 9 out of 10 sub-watersheds. Additionally, a correlation matrix between catchment physiographic characteristics (terrain slope, stream length and size) and calibrated empirical parameters (‘α’, ‘β’, ‘m’ and ‘n’) was developed, indicating stream length influences these parameters more than size and slope.