Use of geostatistics to describe the spatial variability ofpH and epipedon thickness in two soils

Soil scientists traditionally use classification and mapping to partition systematic variability. The range in observed properties is described quantitatively. Descriptions of spatial variability are generally qualitative. This study tests geostatistical parameters, along with the fractal dimension, to quantitatively describe spatial variability. The spatial distribution of two properties; epipedon thickness, and pH at a depth of one meter; was described at three locations each for Herrick silt loam (fine, montmorillonitic, mesic Aquic Argiudolls), and for Yutan silt loam (fine-silty, mixed, mesic Mollic Hapludalfs). Semi-variograms were constructed for each site. Pooled semi-variograms from the combined data for each soil were also constructed. Cross-validation confirms that a pooled semi-variogram model can satisfactorily represent a soil series from one location to another for predictive purposes. The Herrick series has greater overall variability in pH, and the maximum variation tends to occur over about twice the distance as in the Yutan soil. For both soils, a large proportion of total pH variability (30-80%) occurs within the shortest lag spacings used in the study (10 and 12.5 meters). The Yutan soil has greater overall variability than Herrick for epipedon thickness. The distance over which the maximum variability occurs (37 to 87 meters) for the Yutan soil is related to the cycle of the landform. In the Herrick soil, the distance over which maximum variation of epipedon thickness occurs ranges from 35 to 140 meters. Overall, a greater proportion of total epipedon thickness variability occurs within the first lag distance for Herrick than for Yutan. The fractal dimension is related more closely to the property being studied than to the specific soil type. For pH, the fractal dimension ranged from 1.68 to 1.94. For epipedon thickness it ranged from 1.53 to 1.72. This implies processes responsible for pH variation are acting over shorter distances than are those responsible for epipedon thickness variation. The parameters describing the semi-variogram vary from site to site and should be reported with a low, high and representative value just as is done for other soil properties