Spatial Distribution of Soil Attributes on Reconstructed Minesoils

Abstract Mining companies and regulatory agencies need clearly defined methods by which sample data of reconstructed minesoils can be interpolated to determine the spatial distribution and suitability of minesoils for reclamation. Objectives of this study were to model spatial distributions of minesoil attributes across reconstructed landscapes at the Rosebud Mine in southeast Montana, and to evaluate geostatistical procedures that account for spatial dependence in data. Minesoils at the Rosebud Mine have been routinely sampled in an irregular pattern of 100‐m sample intervals. The combined replacement depth of topsoil and subsoil material determined the depth of sampling. Topsoil and subsoil materials were analyzed separately for replacement depth, soil texture, soil pH, and electrical conductivity. Minesoil attributes in all cases were spatially independent at the 100‐m sample spacing. Four soil attributes, topsoil clay content, topsoil pH, topsoil electrical conductivity, and subsoil replacement depth, were used to described the spatial relationships found in the minesoil data. Application of kriging techniques to interpolate between data points was deemed unnecessary due to the uncorrelated nature of the data and lack of reasonable fit of any semivariograms. Trend surface analysis was used to develop a model that predicts minesoil properties across the site. For topsoil clay content, pH, and electrical conductivity, a constant surface through the overall means was the best predictive model for these properties at the site. A quadratic surface was fit to replacement depth data, which exhibited a trend across the reconstructed landscape. This data set provides an example of how geostatistical techniques can be used to evaluate spatial dependence in data. In the absence of spatial dependence, more traditional satistical techniques that rely on independent data assumptions can be used, such as regression techniques.