Soil erosion is seen as both an agronomic problem and a geomorphological process, perhaps the most important erosional process for semi-arid areas in all but steepland settings. Research into soil erosion has focussed predominantly on soil conservation, following on from H.H. Bennett’s pioneering evangelism in the 1930s. Perhaps the major tool for estimating erosional loss is the 1960s Universal Soil Loss Equation, a simple factorial estimate based on a massive collation of plot experiments, that is still widely used despite its acknowledged methodological weaknesses. The work of Horton and Schumm helped to bridge from short term field-scale erosion estimates to the implications for landscape evolution, but there remain many issues about the most appropriate ways to scale up in space and time. For example, trade-offs are required to make best use of the available time/space resolution of past and forecast future climate, in combination with adequate spatial resolution of topography and soil characteristics, to adequately comprehend regional patterns and how they are changing over time as they interact with natural and anthropogenic drivers.

The interaction of runoff and erosion with vegetation cover, which is critical at all scales, has further implications at regional scales and over geological time periods. Vegetation not only helps to control erosion rates, but, through soil development, the balance between erosion and chemical denudation provide the substrate for vegetation, providing long-term feedbacks. On decadal to century times scales, many areas have experienced changes from grassland to shrubs, changing the spatial patterns and intensity of erosional activity. On Pleistocene time scales, and under present conditions of rapid climate change, colonisation rates may not be able to keep pace with changing climate drivers. Vegetation, and particularly grasses, has also evolved over the Tertiary, changing the relationships between rainfall, runoff and erosion.