Using CLIGEN to generate RUSLE climate inputs

Abstract

CLIGEN is a stochastic weather generator that produces continuous daily variables to drive process-based runoff and erosion prediction models such as WEPP. To test CLIGEN's ability to generate precipitation-related variables, which are particularly important to runoff and erosion prediction, algorithms were developed to compute the R-factor, its monthly distribution, and 10-year storm erosion index (EI) needed to apply the Revised Universal Soil Loss Equation (RUSLE). Measured R-factor and 10-year storm EI for 76 sites in the U.S. were used for calibration, and 89 additional sites were used for validation. It was found that the generated R-factor using CLIGEN is highly correlated with the measured R-factor for the calibration sites (r2 = 0.96), although the generated R-factor is systematically larger than the measured R-factor. The predicted R-factor for validation sites has a model efficiency (Ec) of 0.92 and a root mean squared error of around 600 MJ mm ha-1 h-1 year-1, or 24% of the average R-factor for the 89 sites. In addition, CLIGEN-generated precipitation data can also be used to predict 10-year storm EI (Ec = 0.75) and monthly distribution of rainfall erosivity for a wide range of climate environments (average discrepancy = 2.6%). This represents considerable improvement over existing methods to estimate R-factor and 10-year storm EI for locations with only monthly precipitation data, although the systematic over-estimation of the R-factor using CLIGEN-generated climate data suggests possible inadequacies in the assumed storm patterns in CLIGEN and WEPP.