Scale effects on runoff and soil erosion in rangelands: Observations and estimations with predictors of different availability

- Large dataset on rainfall simulations on coupled large and small plots was analyzed.
- Large scale effects on runoff coefficients and sediment yield were quantified.
- Best runoff and erosion predictors were not the same for large and small plots.
- Site-specific data did not improve predictions obtained with basic soil and rainfall properties.

Runoff and erosion estimates are needed for rangeland management decisions and evaluation of ecosystem services derived from rangeland conservation practices. The information on the effect of scale on runoff and erosion and on the choice of runoff and erosion predictors remains scarce. The objective of this work was to evaluate the effect of scale on the selection of runoff and erosion predictors with the data from rich National Range Study database containing data from 444 coupled large (3.05 m by 9.1 m) and small (0.61 m by 1.22 m) plot field experiments. With data from both plot sizes, we assessed the usefulness of adding site-specific soil surface information to basic soil and rainfall data in order to estimate runoff and erosion in rangelands. We observed the scale-dependence of the runoff coefficient and the sediment yield. Smaller values of both variables were found at large plots as compared to small plots. Regression trees were used to build predictive relationships and evaluate the relative importance of predictors. Rainfall and basic soil properties were identified as the major predictors of runoff coefficients and sediment yields at both scales. Differences in the importance of predictors were observed between the two plot sizes and between predictions of runoff and sediment yield at the same plot sizes. The antecedent soil water content was not as important as rainfall parameters. Overall, including site-specific soil surface properties did not improve the predictability of the runoff coefficient and the sediment yield. The difference in runoff and sediment yield between small and large plots was found most likely because the small plots only contained a single soil/vegetation expression, whereas there was a matrix of vegetation clumps and bare interspaces arranged in a non-uniform pattern at the large plots. The variability of runoff and sediment yield may depend on how the latter pattern expresses itself in each of the large plots. More research or a different approach is required to account for vegetation-driven spatial hydrologic processes and their influence on rangeland runoff and soil erosion processes.