Partitioning the DU

Partitioning the DU into multiple SUs is one way to characterize variability on a smaller spatial scale. This can be useful for both exposure assessment (e.g., assessing risks to multiple receptors with different sized exposure units) and remedial design (e.g., delineation of remediation units smaller than a DU).When taken over the entire DU, replicates offer information on variability in the estimate of the mean provided by the ISM samples. They do not, however, provide any information on spatial variability of concentrations within the DU. Another approach is to divide the DU into multiple SUs and take one or more ISM samples from each. With this approach, ISM samples are not true replicates in that they are providing estimates of the mean for different subunits within the DU. Individually, they estimate the mean of a subarea, and collectively, they can be used to estimate the mean of the entire DU. Sampling designs with this method yield unbiased estimates of the mean.

  • The principal advantage of subdividing the DU is that some information on heterogeneity in contaminant concentrations across the DU is obtained. If the DU fails the decision criterion (e.g., has a mean or 95% UCL concentration above a soil action limit), information will be available to indicate whether the problem exists across the DU or is confined. This information can guide redesignation of the DU and resampling to further delineate areas of elevated concentrations.
  • If only one ISM sample is collected per SU, then it is important to understand that each result independently provides an estimate of the mean concentration within the respective SU. Just as a single ISM collected throughout the DU may over- or underestimate the mean by some magnitude (see Section 4.2.1), the information on heterogeneity at the scale of the SU is also subject to uncertainty. If greater certainty is needed at the scale of the SU, then additional increments and/or replicates should be collected at the scale of the SU.
  • Collectively, the results from each SU can be used to estimate the mean and 95% UCL at the scale of the DU.
  • Error estimates from partitioning a DU into SUs are larger than those from replicate data if the site is not homogeneous. Hence, 95% UCL estimates from a subdivided DU are as high as or higher than those obtained from replicate measurements collected across the DU (using the same number of total increments). The higher 95% UCLs improve coverage (generally attain 95% UCL) and increase the RPDA. These increases occur if unknown spatial contaminant patterns are correlated with the partitions.
  • It must be clearly understood by all that if the 95% UCL for the DU is below the action level, the entire DU passes, even if the ISM result for one or more of the partitioned areas is above the action level. Even with partitioning, the DU remains the unit over which a decision is made.

Note: “Row-column” is an additional sampling pattern proposed by Patil and Tallie (2001). This sampling pattern has not been widely discussed in the context of ISM and consequently was not explored in the simulation studies. However, this approach is discussed in the composite sampling literature and has the potential advantage of providing spatial information on localized areas of high concentration (see “oversized DUs” in Section 4.4.4).