Variability of Runoff and Soil Loss from Fallow Experimental Plots

Abstract The amount of unexplained variation in runoff and soil loss studies often limits the interpretation of data. In this study, variability in runoff and soil loss from 40 essentially uniform experimental plots was examined for 25 natural rainfall events occurring during a 155‐d period. Plots had been maintained uniformly for the prior 3 yr and were kept fallow with periodic cultivation during the study period. Except for events producing low runoff and soil loss, event coefficients of variation were relatively constant for both runoff and soil loss at about 20%. Differences in runoff and soil loss among plots varied with event. Only minor amounts of observed variability could be attributed to any of several measured plot properties, and plot differences expressed by the 25 events did not persist in prior or subsequent runoff and soil loss observations at the site. The relatively large amount of unexplained variability shows that several replications of treatments are needed to confidently estimate mean runoff or soil loss for comparison purposes and that effects of factors having relatively minor effects on runoff or soil loss may be difficult to detect statistically. The fact that most variability is unexplained indicates important effects of factors or processes that are not currently understood.