On the spatial variability of soil hydraulic properties in a Holocene
coastal farmland
J. Bevington
College of Engineering, University of Georgia, Athens, United States
D. Piragnolo, F. Morari
Dept. Environmental Agronomy, University of Padova, Legnaro, Italy
P. Teatini
Dept. Civil, Environmental and Architectural Engineering,
University of Padova, Padova, Italy
G. Vellidis
Dept. of Crop and Soil Sciences, University of Georgia, Tifton, United States
ABSTRACT
Understanding water movement in the vadose zone is critical for accurate climate and crop modeling, precision
agriculture, soil-atmosphere gas exchanges, and contamination mitigation. A major reason for the difficulty of
performing effective hydraulic measurements is because they are scale dependent due to the inherent heterogeneity
of the soil. A better understanding of the spatial variability and underlying processes responsible for this
variability could lead to amore accurate modeling. The goal of this study is to investigate the scale dependencies
of soil hydraulic properties. Factorial kriging analysis (FKA) is a geostatistical technique which identifies scale
dependent spatial relationships and common sources of variability. FKA was applied on a number of soil properties
acquired from samples collected at 4 depths ranging from 8 to 68 cm in a 20.8 ha field in the Po River delta.
The farmland is characterized by the presence of paleo-channel structures and highly heterogeneous soil.
Texture, bulk density, Ks (saturated hydraulic conductivity),
and the van Genuchten-Mualem parameters α
(inverse of air entry), n (shape parameter), ϑr (residual water content),
and ϑs (saturated water content) were
included in the analysis. Two nested spherical models with ranges around 105 m and 235 m plus nugget fit
the experimental variograms and cross-variograms best. Regionalized correlation coefficients and regionalized
PCA revealed many strong, scale dependent relationships which were not obvious from descriptive statistics,
such as the effect of interaction between texture and bulk density on n and Ks,
and the stronger influence of
bulk density than texture on Ks. The first principal components (PCs) of the regionalized PCA explained the
majority of the variability and the second PCs were rarely informative. The spatial distributions of the first PCs
resembled bulk density at short scale and the paleo-channels and texture at long scale. The decoupling of bulk
density and texture is likely caused by differences in soil structure. The influence of the short scale PCs is greater
than the long scale PCs near the surface but becomes less important as depth increases. This suggests that depth
plays an important role and should be considered more often in spatial analysis.
