Beneficial uses of compositing

Several USEPA guidance documents, as referenced above, describe composite sampling designs that can be used for various purposes, such as finding areas of high concentration and estimating a population proportion. Only two simple and beneficial uses of compositing will be discussed in this section and contrasted with ISM. For additional information on composite sampling design options, consult the USEPA documents.

With discrete sampling designs there is often the implicit assumption that each discrete sample represents some volume of soil surrounding the area where it was collected. As shown in Figure 2-6, this may be a faulty assumption when short-scale heterogeneity is significant. Recall that short-scale heterogeneity is what occurs at the scale of colocated samples, where kneeling down in one location can give a radically different concentration than kneeling down in a location one foot away. Composite sampling can be used to reduce errors due to short-scale heterogeneity. For example, in Figure 2-6, instead of collecting a single discrete sample from only one of the five placements, an increment of soil could be taken from each of the five placements and composited into a single sample. This method results in a sample more representative of the 1 ft2 area shown in the figure as compared with any single discrete sample. This process could be repeated in other 1 ft2 areas, resulting in a number of composite samples. Note that, as with any sampling design, subsequent steps still need to be taken with each sample to address microscale (within sample) heterogeneity to reduce this source of sampling error as discussed in Section 2.4.1. This process is different from an ISM sampling design since it may not include the prior establishment of a specific DU and the goals may not necessarily be limited to estimating the mean concentration. Also note that the analytical cost of such a composite sample design will typically be much larger than with ISM since many samples will be submitted for separate lab analysis.

A second similar application of compositing is for grid sampling. Instead of taking a single discrete sample from the center of a number of grid cells laid out across a site, a series of composite samples could be taken. The result of a composite sample consisting of several increments of soil collected from across the grid cell is likely to produce a better estimate of the true concentration for that grid cell than will a single discrete result. Note that analytical costs are approximately the same for either the discrete sample/grid center or the composite sample/grid cell design. Again, this approach differs from ISM in that one may have goals in addition to estimating a mean concentration within a predefined volume of soil.

The actual dimensions and number of increments to composite depends, of course, on the spatial scale(s) of the decisions and the degree of short-scale heterogeneity. These can be derived judgmentally or statistically. In either case, it is a good idea to verify that the design is accomplishing its intended goals.