Soil fertility variograms from "true point sampling' on 20.90, 0.9, and 0.1 meter grids in two fields

Numerous soil fertility studies to determine the spatial variability of crop production fields have not explored sampling intensities less than 0.04 hectares. A result of such methods has led to very sparsely populated variograms for sampling distances less than 20 meters (m). The purpose of this research is to accurately define the spatial variability within normal production fields by collecting single core soil samples at three different intensities in two fields. Main level sampling in each field was in a diamond pattern grid measuring 20 m x 20 m. Two of these grids were sub-sampled in a 0.9 m x 0.9 m diamond pattern. Finally, three of these sub-grids were sub-subsampled in a 0.1 m x 0.1 m diamond pattern. Descriptive statistics and semivariograms were calculated for each sampling level. Variograms exhibited many cyclic patterns over the entire sampling distance with a general upward trend. Calculations indicate that if sampling is performed at the maximum limit of spatial correlation [5-6 m for pH, 0.1-0.2 m for phosphorus (P), 20 m for potassium (K), 6-125 m for calcium (Ca) and magnesium (Mg), and 34-111 m for organic matter (OM)], then precision of the data must be reported as +/- 0.3-0.7 pH, +/- 25-36 mg P kg-1, +/- 31-38 mg K kg-1, +/- 276-342 mg Ca kg-1, +/- 61-69 mg Mg kg-1, and +/- 0.4-0.6% OM, respectively. Only for sampling distances less than these does spatial correlation exist. Therefore, complete characterization of fields at minimum variance can only be approached at sampling distances less than or equal to 1 m for pH, 0.1 m for P, 1 m for K, 1-6 m for Ca and Mg, and 10 m for OM. Sampling at 0.1 m for all nutrients can reduce estimates of variance to levels that approach the precision in laboratory determinations. A question remains, "How precise do we need to be?"