Premium
Error Assessment in the Universal Soil Loss Equation
Author(s) -
Risse L. M.,
Nearing M. A.,
Laflen J. M.,
Nicks A. D.
Publication year - 1993
Publication title -
soil science society of america journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.836
H-Index - 168
eISSN - 1435-0661
pISSN - 0361-5995
DOI - 10.2136/sssaj1993.03615995005700030032x
Subject(s) - soil loss , universal soil loss equation , environmental science , mathematics , statistics , soil science , ecology , biology , surface runoff
Abstract Although nearly three decades of widespread use have confirmed the reliability of the Universal Soil Loss Equation (USLE), very little work has been done to assess the error associated with it. This study was conducted to develop a set of statistics that would measure the performance of the USLE. Estimates of soil loss using the USLE were compared with measured values on 208 natural runoff plots, representing >1700 plot years of data, to assess the error associated with the USLE predictions. The overall Nash‐Sutcliffe model efficiency was determined to be 0.75 on an average annual basis and 0.58 when compared on a yearly basis. The USLE overpredicted soil loss on plots with low erosion rates while the plots with higher rates were underpredicted. Of the USLE parameters, the topographic factor ( LS ) and the cover and management factor ( C ) had the most influence on the model efficiency. Confidence intervals for USLE predictions were developed and showed that the accuracy of the USLE in terms of percentage difference between predicted and expected values increases with increasing values of total soil loss. It was also shown that there was no significant difference between the average magnitude of error for pre‐ and post‐1960 data sets and that the use of rainfall and runoff factor ( R ) values instead of calculated erosion index (EI) values resulted in a drop in model efficiency of 0.02. One must use caution in applying the results of this error analysis to conditions in which they may not be applicable, due to the limited nature of this data set.