A comparison of mineral-magnetic and distributed RUSLE modeling in the assessment of soil loss on a southeastern U.S. cropland

As geospatial technologies increasingly figure into resource management activities, there is a corresponding need to provide commensurately detailed high resolution spatial data. This study addresses the capacity of rapidly and cheaply acquired mineral-magnetic data to provide detailed spatially distributed assessments of long-term cumulative soil loss from agricultural fields. Model output from a simple distributed implementation of the Revised Universal Soil Loss Equation (RUSLE) and from a field-data-oriented soil magnetism-based erosion model are compared for a small Alabama (USA) farm lot. An undisturbed reference soil toposequence is used as input to the magnetism model, in contrast with earlier studies that relied on a single reference soil profile. Results from this procedure, while likely to be more dependable, proved primarily to further emphasize spatial patterns noted in prior studies. In addition, the use of a toposequence decreased the area over which RUSLE and magnetism models were in agreement. RUSLE underpredicted denudation relative to magnetism-model results over upper slopes, and overpredicted it on lower slopes. The locations of former access roads may explain underprediction on the upper slopes. The differences between the two method outcomes are discussed with regard to the potential for spatial variability in parent materials, the efficacies of non-fluvial soil redistribution processes, and the availability of detailed land use records for the 100+ years of agricultural activities at the site.