Variation of Selected Properties across a Soil Boundary

Abstract

AbstractThis study is a statistical analysis of the changes in the distributions of soil properties at the contact between two soil series. Samples were collected at 10‐m intervals on an 18 by 40 grid positioned over the mapped boundary between two contrasting soil series in eastern Kansas. Samples collected from the B2 horizons of both soils were analyzed for gravel, sand, and silt content, for pH, and for Munsell color. These measurements were used to calculate a discriminant function which was used to define the boundary between the two soils. The variations of sand, silt, and pH measurements were compared with each of three models describing changes within the study area. The first model is a random distribution of the measurements. Simple trend surfaces were calculated for each of the three sets of measurements as a second model. A third model consisted of two uniform categories separated by an abrupt boundary. The fit of each model was tested against the distributions of the three sets of measurements using autocorrelation functions to examine the distributions of the residuals. The simplest model having random variation in the residuals was selected as the most efficient descriptor for each variable. Trend models were selected for the pH and silt measurements; and the two‐category model was selected for the sand data. This indicates that measurements of pH and silt change slowly within the study area, whereas the sand measurements exhibit a very sharp change at the contact between the two soils.