variability of measured runoff and soil loss from field plots

introduction: field plots are widely used in studies related to the measurements of soil loss and modeling of erosion processes. research efforts are needed to investigate factors affecting the data quality of plots. spatial scale or size of plots is one of these factors which directly affects measuring runoff and soil loss by means of field plots. the effect of plot size on measured runoff or soil loss from natural plots is known as plot scale effect. on the other hand, variability of runoff and sediment yield from replicated filed plots is a main source of uncertainty in measurement of erosion from plots which should be considered in plot data interpretation processes. therefore, there is a demand for knowledge of soil erosion processes occurring in plots of different sizes and of factors that determine natural variability, as a basis for obtaining soil loss data of good quality. this study was carried out to investigate the combined effects of these two factors by measurement of runoff and soil loss from replicated plots with different sizes. materials and methods: in order to evaluate the variability of runoff and soil loss data seven plots, differing in width and length, were constructed in a uniform slope of 9% at three replicates at koohin research station in qazvin province. the plots were ploughed up to down slope in september 2011. each plot was isolated using soil beds with a height of 30 cm, to direct generated surface runoff to the lower part of the plots. runoff collecting systems composed of gutters, pipes and tankswere installed at the end of each plot. during the two-year study period of 2011-2012, plots were maintained in bare conditions and runoff and soil loss were measured for each single event. precipitation amounts and characteristics were directly measured by an automatic recording tipping-bucket rain gauge located about 200 m from the experimental plots. the entire runoff volume including eroded sediment was measured on storm basis using the collection tanks. the collected runoff from each plot was then mixed thoroughly and a sample was taken for determining sediment concentration by weight. the per-storm soil loss was then obtained. results and discussion: a wide range of rainfall characteristics were observed during the study period.the results indicated that the maximum amount of coefficients of variation (cvs) for runoff and soil loss from replicated plots were 60 and 80 percent, respectively, which were considerably higher than the variability of soil characteristics from these plots. cv of runoff and soil loss data among the replicates decreased as a power function of mean runoff (r2= 0.661, p<0.01) and soil loss (r2= 0.428, p<0.01). empirical cumulative frequency distribution of the ratio between the cv of runoff and the corresponding soil loss cv in all of the plots data, showed that the cv of soil loss is higher than the cv of corresponding runoff for about 80 percent of cases. empirical cumulative frequency distribution was created for each size of plot separately, which showed that the difference between the cv ofrunoff and the corresponding soil loss decreases with decreasing the plot size. conclusion: this study provides estimates of variability for within runoff and soil loss data. a high amount of runoff and soil loss cvs emphasize the importance of the number of replications in erosion plot studies.there was a strong relationship between the cv of runoff and the average runoff amount from replicated plots. this relationship was also observed for soil loss and the cv from replications. these relationships create a useful tool to determine the number of replications on the basis of desired confidences of interval with known values of cvs. empirical cumulative frequency distribution of the ratio between the cv of runoff and the corresponding soil loss cv demonstrates that the mean runoff data from replicated plots were generally more representative than mean soil loss data, particularly in the large plots. runoff was generally less variable than soil loss. high variability of soil loss data in comparison with the corresponding runoff amounts may be related to the probability of measurement errors in soil loss data which increases with increasing plot size. therefore, an appropriate sampling scheme for plot soil loss should be used to yield representative values of both soil loss.