ABSTRACT
Data concerning runoff and sediment yield in arid zones is of prime importance for hydrologists, geomorphologists, pedologists, ecologists and landscape engineers. For data comparison and extrapolations, runoff and sediment yield are often presented in mass per unit area. Runoff and sediment yield collected on dune slopes over a wide range of plot sizes during 1990–1994 in the Negev Desert, Israel, showed that the contributing area was mainly confined to a narrow belt at the bottom of the slopes. It was therefore hypothesized that the very short rain bursts, capable of runoff generation, may result in a scale effect (SE). Indeed, average duration of duration of consecutive medium and high rain intensities which are potentially above the surface infiltration rate ranged between 2.2 and 3.0 minutes, implying that flow connectivity is largely limited. Based on the intermittent character of the rain spells capable of runoff generation it is argued that SE is an inherent outcome of the rain properties. Yet, it is further argued that the magnitude of the SE is surface‐dependent. As a result, it is argued that the conventional way for runoff and sediment yield presentation as mass per unit area implies theoretical misconceptions and may cause gross overestimation in extrapolation and the presentation of runoff and sediment yield in mass per unit width of the slope is suggested. The accuracy of the two extrapolation methods are compared to the actual runoff and sediment yield collected in the field. The data show that extrapolation based on runoff (or sediment) yield per plot width deviates from the actual amounts collected by a factor of 1·1 to 1·3 only while deviating by a factor of 4·2 to 5·6 and 10·7 to 11·8 if the extrapolation is based on large and small plots, respectively. Theoretical and practical reasons for presentation of runoff and sediment yield as mass per unit width are discussed. Copyright © 2011 John Wiley & Sons, Ltd.