Eroded aggregate size distributions from disturbed lands

A characterization of eroded soil properties, especially size, is required for proper usage of modern process-based soil erosion models. In a study addressing reclaimed mine land erodibility, over 350 measured eroded aggregate size distributions were collected from erosion plots created on reclaimed and nearby unmined soils at three separate strip mine sites. The experimental design allowed for a comparison of eroded aggregate size distributions as a function of land slope, run sequence, and reclamation treatment. No influence of slope was apparent, as distributions collected from plots constructed at 6% slope were coarser than those collected from 18% slope, but finer than those collected from 12%. Samples collected during extra inflow runs, in which runoff rates, flow velocity, and rill erosion were greatest, had coarser distributions than samples collected during rainfall only runs, which tended to have similar distributions. Distributions from reclaimed topsoil treatments were similar to nearby unmined soils and were distinct from subsoil reclamation treatments. At two sites, topsoil treatments had coarser distributions, while the converse was true at the third. Distributions taken from topsoil and unmined soil treatments were used to develop five regression equations to predict the percentage of total eroded soil mass failing into each of five size classes-clay, silt, and three sand sizes. A stepwise, multivariate regression technique was used to select the best independent variables from 30 soil parameters. Independent variables that proved to be significant in at least one equation include primary particle and water stable aggregate size distributions, bulk density, organic carbon, and soil moisture. The number of significant independent variables ranges from 11 to 15 and correlation coefficients (R2) range from 0.77 to 0.53, which is a general improvement over previously published prediction methods.