2.2.2 Short-Scale Spatial Heterogeneity

If unaccounted for, the presence of short-scale heterogeneity can lead one to draw very different conclusions about contaminant concentrations simply depending on where a discrete sample happens to be collected.

"Short-scale heterogeneity" refers to differences in contaminant concentrations between colocated samples separated by short distances. These distances can be on the order of inches to a few feet. Some causes of short-scale heterogeneity are discussed in Hyperlink 6. Short-scale heterogeneity determines whether the same result is obtained if one happens to take the sample at placement A or placement B, which is close to A. Placement points A and B are equivalent in the sense that the probabilities of choosing one over the other are equal for a given sampling location. Consider the case where the sampling location is designated as the center of a 100 ft2 grid cell. Suppose that samples are collected with a 2-inch-diameter coring device. Within a 1 ft2 area at the center of the grid cell, there are 36 nonoverlapping placement points for a 2-inch corer, any of which might be sampled, as shown in Figure 2-5.

Colocated samples are considered equivalent in that roughly the same concentration would be expected from both placements because they are so close spatially. However, colocated samples often do not meet precision expectations.

High variability in colocated samples is illustrated in Figure 2-6, which presents data from a field investigation for uranium. The original sampling design called for one discrete sample per 270 ft2. In other words, the result from a single sample would be extrapolated to represent the concentration for an area centered on the sample and encompassing 270 ft2. Prior to the main investigation, a small pilot study was done to see how much short-scale heterogeneity was present at the 1 foot scale. Figure 2-6 displays the results from the pilot study. It is apparent that the concentration assigned to this 270 ft2 area could vary by an order of magnitude depending on where the technician happened to place the corer. If the sampler collected from sample placement #1, the concentration would be 30 mg/kg; if from placement #2 about 8 inches away, the result would be 496 mg/kg. In other words, if a single discrete sample were used to represent the sampling area, a conclusion of "contamination is low" vs. "contamination is high" is purely a matter of chance.

Results from a similar study involving arsenic variability along a transect covering just a few feet are presented in Hyperlink 7. The takeaway point is that one should not place a great deal of confidence in a single discrete sample result when little is known about the magnitude of short-scale heterogeneity.

Figure 2-5. A single square foot area of surface soil contains 36 possible 2-inch-diameter core sample locations.

Figure 2-5. A single square foot area of surface soil contains 36 possible 2-inch-diameter core sample locations.

 

Figure 2-6. Observed short-scale heterogeneity with colocated uranium sample results. Source: Unpublished data supplied by Robert Johnson, Argonne National Laboratories.

Figure 2-6. Observed short-scale heterogeneity with colocated uranium sample results.

Source: Unpublished data supplied by Robert Johnson, Argonne National Laboratories.