作者: L. M. Risse , M. A. Nearing , J. M. Laflen , A. D. Nicks
DOI: 10.2136/SSSAJ1993.03615995005700030032X
关键词: Statistics 、 Erosion 、 Data set 、 Expected value 、 Hydrology 、 Plot (graphics) 、 Surface runoff 、 Confidence interval 、 Mathematics 、 Universal Soil Loss Equation 、 Magnitude (mathematics) 、 Soil science
摘要: Although nearly three decades of widespread use have confirmed the reliability Universal Soil Loss Equation (USLE), very little work has been done to assess error associated with it. This study was conducted develop a set statistics that would measure performance USLE. Estimates soil loss using USLE were compared measured values on 208 natural runoff plots, representing >1700 plot years data, predictions. The overall Nash-Sutcliffe model efficiency determined be 0.75 an average annual basis and 0.58 when yearly basis. overpredicted plots low erosion rates while higher underpredicted. Of parameters, topographic factor () cover management had most influence efficiency. Confidence intervals for predictions developed showed accuracy in terms percentage difference between predicted expected increases increasing total loss. It also shown there no significant magnitude pre- post-1960 data sets rainfall instead calculated index (EI) resulted drop 0.02. One must caution applying results this analysis conditions which they may not applicable, due limited nature set.