Computed tomography data on soil structural and hydraulic parameters assessed for spatial continuity by semivariance geostatistics

Abstract

Visual observations on the spatial distribution, at 1-cm intervals, of bulk density (ρ), porosity (ε), fractal dimension (D), water content (θ), and unsaturated hydraulic conductivity (Kus) in uniformly packed soil columns showed randomness. We explore the use of semivariance geostatistics to clarify the issue of randomness and continuity on the spatial distribution of ρ, ε, D, θ, and Kus data obtained using a custom-built gamma scanner and computed tomography technique. Semivariance increased with increasing lag distance and plots of semivariance v. lag distance produced spherical semivariograms for most of the soil parameters investigated. This indicated that even though randomness existed in the spatial distribution of the soil parameters, there existed specific trends in their spatial continuity. Higher spatial continuity, in water stable aggregates, was characterised by smaller values of semi-, sill-, and nugget-variances and larger values of span. Opposite trends were observed for unstable aggregates. Wetting in unstable aggregates produced further reductions in span increases for other geostatistical parameters, indicating that wetting decreased spatial continuity. The results indicate that geostatistical analysis is useful to clarify the issue of randomness at very small scales and to quantify and discriminate the influence of differences in structural stability and wetting-induced changes in the spatial continuity of soil parameters, particularly ε .