Scale dependent variability of soil electrical conductivity by indirect measures of soil properties

Knowledge of scale dependent variation of soil properties is important where upscaling and generalization from plot scale studies to field and larger scale is desired. We used conventional statistics, geostatistics, and fractal analysis to characterize and compare the apparent soil electrical conductivity (ECa) of six contiguous agricultural fields each ranging between 9.5 and 14.0 ha in size. Factor analysis revealed that ECa was strongly related to ammonium extractable K, organic matter (OM), pH and Bray-2 Phosphorus, but not to ammonium extractable Ca and sum of bases. All six fields were spatially structured and well described by exponential semivariograms. Fractal dimensions estimated from the linear portion of the semivariogram using a linear plateau model were statistically different (p = 0.05) among some of the fields, and the differences may have been caused by management differences. Fractal analysis identified at least two scales of variation for the fields. The first scale of variation, common to all six fields, was for distances less than 9 m. The second scale of variation was for distances ranging between 9 and 46 m (field NC), 9 and 79 m (fields SC and SW); and 9 and 126 m (field SE). Two of the fields (fields NW and NE) did not have a plateau on the log-log plot of the semivariograms, indicating a scaling behavior at larger distances. The study showed that although the semivariogram forms are similar among the six fields, the rate of change of the semivariograms (as indicated by the fractal dimension) differs for some of the fields at distances greater than 9 m.