Skip to main content
ARS Home » Pacific West Area » Tucson, Arizona » SWRC » Research » Publications at this Location » Publication #164084


item Nearing, Mark

Submitted to: Book Chapter
Publication Type: Book / Chapter
Publication Acceptance Date: 10/20/2004
Publication Date: 6/1/2006
Citation: Nearing, M.A. 2006. Can soil erosion be predicted? In: Soil Erosion and Sediment Redistribution in River Catchments. P. Owens editor. CABI Publishing. p. 145-152.

Interpretive Summary: Soil erosion is hard to predict accurately because of the fact that erosion itself is quite variable in nature. Trying to predict soil erosion is like shooting at a moving target, and attempts to test the capabilities of models can only give us a general idea of their accuracy. This is particularly true when the erosion rates themselves are quite small. The less the erosion, the harder it is to predict. This causes problems for areas such as rangelands, where erosion rates are relatively small compared to cropped fields. Another concern with predicting erosion is that it takes a long time in nature to measure average erosion rates, because erosion varies so much in time. Again, this makes prediction particularly difficult in rangelands where rainfall is highly variable, anyway. This means that a person has to run a model to simulate as much as 200 years of erosion in order to quantify what the long term erosion rates might be, even if the model is correct. In spite of these problems, if one understands these limitations of erosion models and uses the models correctly, they can be used to design engineering structures, manage conservation programs, help people make land use decisions, and quantify the amount of erosion that occurs over broad geographical areas.

Technical Abstract: Variability in soil erosion data from replicated plots is large. One might think of the replicated plot as the best "real-world, physical model" of soil erosion, and that the physical model represented by the replicate plot represents a best-case scenario in terms of erosion prediction. In this study, replicated plot pairs for 2061 storms, 797 annual erosion measurements, and 53 multi-year erosion totals were used to estimate the natural variance of erosion data. Coefficients of variation ranged on the order of 14% for a measured soil loss of 20 kg/m2 to greater than 150% for a measured soil loss of less than 0.01 kg/m2. The r2 for the fit for the replicate plot model was 0.76. This fit sets a benchmark for what one can expect for soil erosion models in general. This paper also discusses the critical nature of continuous simulation modeling in predicting erosion reliably. Results of simulation testing with the WEPP model indicate that 60 to 200 years of continuous simulation are required in order to quantify erosional response to plus or minus 10%. Single storm models do not have the capacity to accurately characterize erosional response of the complex and dynamic erosional system.