Testing assumptions and procedures to empirically predict bare plot soil loss in a Mediterranean environment

Empirical prediction of soil erosion has both scientific and practical importance. This investigation tested USLE and USLE-based procedures to predict bare plot soil loss at the Sparacia area, in Sicily. Event soil loss per unit area, Ae, did not vary appreciably with plot length, l, because the decrease in runoff with l was offset by an increase in sediment concentration. Slope steepness, s, had a positive effective on Ae and this result was associated with a runoff coefficient that did not vary appreciably with s and a sediment concentration generally increasing with s. Plot steepness did not have a statistically detectable effect on the calculations of the soil erodibility factor of both the USLE, K, and the USLE-M, KUM, models but a soil-independent relationship between KUM and K was not found. The erosivity index of the USLE-MM model performed better than the erosivity index of the CSI (Central and Southern Italy) model. In conclusion, the importance of an approach allowing soil loss predictions that do not necessarily increase with l was confirmed together with the usability of already established and largely applied relationships to predict steepness effects. Soil erodibility has to be determined with reference to the specific mathematical scheme and conversion between different schemes seems to need taking into account the soil characteristics. The USLE-MM shows promise for further developments. The evolutionary concept applied in the development of the USLE should probably be rediscovered to improve development of soil erosion prediction tools.