Physical and hydraulic characterization of a clay soil at the plot scale

Abstract

The soil physical and hydraulic properties have to be determined for interpreting and simulating many hydrological processes. An investigation was carried out to determine the physical and hydraulic characteristics of a clay soil at the plot scale. An intensive sampling of the surface soil layer of two plots of 4x11 m2 was carried out by measuring, for each plot, dry soil bulk density, b, and antecedent soil water content, i, at 88 sampling points and field-saturated soil hydraulic conductivity, Kfs, at 176 sampling points. A wide range of Kfs values (0.7-5107 mm h-1) were measured by the Simplified Falling Head (SFH) technique. For each variable, the two plots yielded very similar results in terms of summary statistics (mean values of b, i and Kfs equal to 1.02-1.03 Mg m-3, 0.27-0.28 m3m-3 and 1181-1244 mm h-1, respectively; associated coefficients of variation equal to 9.8-10%, 19.2-21.1%, and 75.9-84.4%) and empirical frequency distributions. For both plots, the hypothesis that data were normally distributed was not rejected for the untransformed b and i, and only for Kfs2/3 among many mathematical transformations of this variable. However, differences in the spatial distribution of a given variable were detected between the two plots. It was concluded that the SFH technique is suitable for measuring a wide range of Kfs values with a relatively low experimental effort, and that the Kfs2/3 transformation was the best transformation for characterizing the spatial variability of Kfs. In addition, the sampling strategy should vary with the objective of the investigation. For the considered clay soil, a small sample size (i.e., < 19) is enough to determine representative means for both b and i. Obtaining representative mean results for Kfs needs a much larger sample size (i.e., > 84). A large sample size (i.e., 56-93, mainly depending on the variable) is also necessary to deduce representative standard deviations and to study the spatial structure of the variables included in this investigation. Finally, mean and standard deviation of b, i and Kfs obtained by an intensive sampling of a single plot may be used to characterize another plot close to the sampled one.