Sampling objectives and developing the decision unit

Challenge: Some project stakeholders are concerned about identifying potential hot spots.

Recommendation: Because the size of the DU sets the scale of the resolution of the investigation, it is critical to ask and resolve the following questions before dividing an area under investigation into DUs:

  • What is the overall objective(s)?
  • What is the amount of soil to be concerned about?
  • Do acute hazards need to be addressed? 

With discrete sample data, assertions about hot spots are typically made after looking at the results of the investigation. Claiming that a single discrete soil sample result represents a meaningful volume of highly contaminated soil is rarely defensible and more importantly often not practical. A hot spot identified by an individual discrete soil sample within a DU may simply denote the degree of heterogeneity of contaminant concentrations within the DU. It is sometimes assumed that a sample or samples from a data set which contain the highest contaminant concentrations represent the highest concentrations actually present at a site, when in fact, due to short-scale heterogeneity, an area only an inch away may contain higher concentrations. An extremely large number of samples is required to estimate and delineate with any degree of certainty relatively small areas with the highest concentrations that might be present at a site. Estimating the actual true maximum contaminant concentrations in soil is therefore often an impractical endeavor.

ISM cannot identify hot spots smaller than the DU.

ISM replicates provide a measure of variability in estimates of the mean concentration in the DU but do not provide information about the spatial distribution. To characterize spatial variability, either the scale of the DU or the scale of the areas sampled within the DU need to be adjusted. DUs can vary greatly in size from very small (e.g., much less than ¼ acre such as a sandbox) to very large (e.g., hundreds of acres). During systematic planning, options for either combining or splitting DUs may be considered to address multiple objectives. It is possible that prior site knowledge can be used to refine the sampling plan to account for potential source areas as separate DUs, but there must be high confidence that this information is accurate. If a DU is subdivided to evaluate smaller volumes of soil, it is likely that additional ISM samples will need to be collected before decisions can be made at the scale of the smaller DU. For example, a large DU may be subdivided into four areas with ISM samples collected from each subarea. The data would support calculations of an area weighted mean concentration and 95% UCL for the large DU, as well as preliminary estimates of means within each subarea. Additional ISM samples could then be collected to reduce uncertainty and delineate areas of elevated mean concentrations at the smaller scales. More detail on combining and splitting DUs and calculating area weighted means can be found in Sections 4.4.1 and 3.4.

As a general rule, misunderstandings about potential hot spots can be avoided with proper DU scaling during systematic planning and by keeping the investigation and evaluation focused on the DU identified, not to the scale of an individual sample.

There are two concerns regarding hot spots: sampling density and defining the DU. ISM effectively addresses compliance when action levels are based on the mean concentration within a DU. Concerns related to spatial resolution can be addressed only by changing the scale of the DU (i.e., DU equals hot spot size).

Challenge: Some project stakeholders are concerned that potential areas of higher concentration within a DU (i.e., hot spots) will be diluted out when combined through ISM with increments of soil from less-contaminated portions of the DU.

Recommendation: Even the best systematic planning and underlying CSM could result in a sampling design that fails to identify small areas of extreme high and low concentrations within the DU in the proper proportion compared to the total mass of the soil. This reality contributes to a concern about dilution of hot spots. There are at least two issues with the issue of hot-spot dilution:

  • Sampling density. If the small area(s) within a DU with extremely high concentrations are not represented by a sample in the proper proportions (compared to the total mass of soil in the DU), the estimate of the sample mean can be highly variable. Unless the ISM sampling process is repeated many times (replicates), there remains a concern about the performance of any one sampling event. Specifically, if the difference between the estimate of the mean and the true population is large, there is a greater chance of reaching the wrong conclusion regarding compliance with an action level (see performance metrics, Section 4). For this reason, it is very important to consider collection of replicates to ensure that the small areas of high concentration are collected in the proper proportion relative to the total mass.
  • Defining the DU. Different sampling designs provide different information concerning the location, spatial extent, and magnitude of subareas of high concentrations. Uncertainty regarding the toxicological significance of acute exposures leads to uncertainty in the definition of a hot spot. This uncertainty creates a challenge for any sampling design—without a clear definition of what constitutes a hot spot, it is difficult to delineate a DU and develop a sampling design that provides information to adequately address hot spots. For this reason, the definition of hot spot and size of the DU must be agreed on by the team during the systematic design phase.

The chance that any single sampling event will include subareas of high and low concentrations in the proper proportion is directly related to the number of samples collected within a DU. An advantage of ISM over other sampling designs is that it accommodates large sample sizes (i.e., large number of increments as well as multiple replicates). For this reason, while any individual sample collected in a hot spot is diluted within the larger group of samples, we are more likely to achieve an estimate of the mean that is representative of the true mean within the DU. This advantage of ISM addresses the first concern (compliance with action levels) but not the second concern (spatial resolution). If the DQO includes the identification and delineation of small areas of elevated concentrations, ISM sampling can address this objective only by changing the scale of the DU (i.e., DU must be the same size as the hot spot of concern).

Challenge: Developing the correct DU.

Recommendation: The DU is properly sized when knowledge about spatial variability/spatial patterns of contaminant concentrations within the DU are no longer of interest. Section 3.3 provides information to consider when developing the DU(s). Keep in mind that a site may be subdivided into multiple DUs to accomplish investigation objectives. Consider how DU sample results will be used. If the objective is to assess potential exposure concentrations over a ¼-acre residential lot, then the DU is ¼ acre, and subdivision into smaller DUs is not necessary. If the objective is to investigate and locate areas of potentially higher contaminant concentration within the site as a whole (e.g., source areas or separate exposure areas), using multiple smaller DUs is appropriate.

Although DUs are ideally sized no larger than the volume of soil in which the average concentration is sufficient to make a decision, there may be situations where DUs cannot be sized small enough for practical reasons. For example, the Florida Department of Environmental Protection has criteria for some contaminants in soil for protection from acute exposure. These criteria are based on a scenario in which a small child ingests, on a single occasion, a handful (10 g) of soil. For this scenario, the DU should be an area approximately the size of child’s hand. Obviously, it would be impractical to divide a site into DUs of this size, and use of a larger DU encompassing thousands of these exposure areas would raise legitimate questions of whether acute toxicity potential can be evaluated. This is a problem with the use of discrete data as well, but with discrete data, some information on variability of concentrations within an area is obtained with which to estimate what the concentration in the most contaminated exposure area might be. For sites in Florida in which acute exposure and toxicity are a concern, the regulatory acceptance issue is whether to allow use of a method such as ISM that appears incapable of providing reasonable assurance that acute-toxicity based criteria have been met. A similar issue may apply to ecological risk assessments, where the area of exposure for some species of interest is smaller than can be accommodated by an affordable number of DUs. Again, discrete sampling has its own set of problems dealing with this issue, but a sufficient number of samples can provide information on variability of concentrations over space from which predictions regarding worst-case exposure areas can be made.

A refinement of initial DUs flagged for remediation may be useful to better isolate areas of high contamination and optimize resources available for cleanup. See Section 3 for a thorough discussion on this topic.