Submitted to: Soil Science Society of America Journal
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/24/1999
Publication Date: N/A
Citation: Interpretive Summary: Nature is highly variable, and soil erosion is no exception! Scientists set up soil erosion plots in the field in order to measure rates of erosion and to understand what environmental factors and farming practices affect erosion. When they do so, they will often find that even when the treatments on the plots are the same, there are differences in the erosion rates from plot to plot. This can cause problems for interpreting the information from the field plots and for understanding and quantifying differences in the factors which they plan to study from the experiments. In the past we simply have not had a very good understanding of how much similar plots in the field do vary one from the other because it normally takes a great number of plot replications to quantify the variation. This study used new statistical methods to be able to quantify the variability, not from a large number of similar plots, but from a very large number of paired plots which were the same as one another. The results provide two important contributions to the science of soil erosion: 1) we now can make recommendations on the number of filed plots a scientist needs to set up in order to measure erosion accurately, and 2) we can now understand when soil erosion prediction models are predicting erosion at an acceptable level of accuracy.
Technical Abstract: The degree of unexplained variability in soil erosion measurements is generally considered to be large, but is not well documented. Understanding and quantifying variability in the data are critical for advancing the science of erosion, for evaluating soil erosion models, and for designing erosion experiments. The purpose of this study was to quantify variability between replicated soil erosion field plots under natural rainfall. Data from replicated plot pairs for 2061 storms, 797 annual erosion measurements, and 53 multi-year erosion totals were used. The relative differences between replicated plot pair measurements tended to decrease as the magnitude of the measured soil loss increased. Using an assumption that soil loss magnitude was the principal factor for explaining variance in the soil loss measurements, we were able to estimate the coefficient of variation of within-treatment, plot replicate values of measured soil loss. Variances between replicates decreased as a power function (r2=0.78) of measured soil loss, and were independent of whether the measurements were event-, annual-, or multi-year-values. These results have important implications for both experimental design and for using erosion data to evaluate prediction capability for erosion models.