Controls on a scale explicit analysis of sheet erosion

Although the impact of sheet erosion on the evolution of soils, soil properties and associated ecosystem services across landscapes is undisputed, there are still large uncertainties in the estimation of sheet erosion, as the results obtained are highly scale dependent. Consequently, there is a need to develop a scale‐explicit understanding of sediment erosion yields, from microplot to hillslope through to plot, to surmount actual erosion modelling flaws and to improve guidance for erosion mitigation. The main objective of this study was to compare sediment yields from small and large plots installed under different environmental conditions and to interpret these results in terms of the main mechanisms and controlling factors of sheet erosion. Fifteen 1 × 1 m² and ten 2 × 5 m² plots were installed on a hillslope in the foothills of the Drakensberg, South Africa. Data of runoff, sediment concentration (SC), soil loss (SL) and rainfall characteristics obtained during the 2009–2010 rainy season at the two spatial scales and from different soils, vegetation cover, geology and topographic conditions were used to identify the main controlling factors of sheet erosion. Scale ratios for SC and SL were subsequently calculated to assess the level of contribution of rain‐impacted flow (RIF) to overall sheet erosion. The average runoff rate (n = 17 events) ranged between 4.9 ± 0.4 L m^‐2 on 1 m² and 5.4 ± 0.6 L m² on 10 m², which did not correspond to significant differences at P < 0.05 level. Sediment losses were significantly higher on the 10 m² plots, compared with the 1 m² plots (2.2 ± 0.4 vs 1.5 ± 0.2 g L^‐1 for SC; 9.8 ± 1.8 vs 3.2 ± 0.3 g m^‐2 for SL), which illustrated a greater efficiency of sheet erosion on longer slopes. Results from a principal component analysis, whose two first axes explained 60% of the data variance, suggested that sheet erosion is mainly controlled by rainfall characteristics (rainfall intensity and amount) and soil surface features (crusting and vegetation coverage). The contribution of RIF to sheet erosion was the lowest at high soil clay content (r = 0.26) and the highest at high crusting and bulk density (r = 0.22), cumulative rainfall amount in the season and associated rise in soil water table (r = 0.29). Such an explicit consideration of the role of scale on sediment yields and process domination by either in situ (soil and soil surface conditions) or ex situ (rainfall characteristics and antecedent rainfall) factors, is expected to contribute to process‐based modelling and erosion mitigation. Copyright © 2012 John Wiley & Sons, Ltd.