7. MAKING DECISIONS USING ISM DATA

7.1 Introduction

This section provides guidance on using data generated from ISM samples to make decisions about a DU. Since the data may inform one or more decisions, it is helpful to establish a structured approach to making decisions, referred to here as “decision mechanisms.”

“Decision mechanism” refers to the different ways that environmental concentration data can be used to make decisions at a site.

In the context of this discussion, decision mechanisms include the procedures, inputs, and algorithms that are used to aid decision making based on environmental concentration data. The simplest decision mechanism is a comparison of a single measured concentration to an action level. The inputs in this case are the concentration measured in the sample and the action level. The procedure may be, for example, to compare the concentration to the action level to determine whether further sampling, evaluation, or other action is needed. A common example of this type of decision mechanism might be the comparison of individual discrete soil sample results obtained during a CERCLA Preliminary Assessment to Regional Screening Levels (RSLs) for chemical contaminants at Superfund sites. Discrete soil sample results are often compared directly to the RSL benchmarks. Exceedance of a benchmark by one or more discrete soil sample results can be used to identify a contaminant as a COPC.

More complex decision mechanisms may involve procedures that include the use of advanced statistical analysis or numerical models. For example, a surface soil investigation may involve the use of geostatistical modeling or krieging to estimate the distribution and extent of contaminants across a site using high-density discrete soil sampling data. Decision mechanisms may involve a series of procedures that are iterative or progressively more complex.

The specific decision mechanisms that may be needed to make a final decision for a DU should be determined as part of the planning at the start of the investigation as noted below and in Section 3. As discussed below, decision mechanisms that apply to ISM are analogous to decisions with discrete data, and include the following:

  • comparison of a summary statistic (e.g., single ISM estimate of the mean, the mean of multiple ISM results, the 95% UCL of multiple results) to an action level
  • comparison of results of a quantitative risk assessment which used a summary statistic (typically a 95% UCL) to an acceptable risk range for carcinogens (e.g., 1 × 10-6 excess cancer risk to 1 × 10-5 excess cancer risk) or to an acceptable hazard threshold for noncarcinogens
  • comparison of site and background data sets
  • combination of data across multiple DUs
  • extrapolation of statistics across DUs

One of the primary benefits of ISM sampling is that the volume of media to which a decision will be applied must be determined prior to sample collection. It is also essential to have an understanding of the manner in which ISM samples will be used to make decisions during project planning. Decision mechanisms must be consistent with the rationale behind the sampling plan design, as discussed in Section 3, and should be based on the following:

  • CSM
  • goals of the project and end use of the data
  • scale of the decision
  • requirements for precision, total error, and decision quality
  • assumptions of the statistical method(s)
  • anticipated and/or measured degree of variability within the DU

Although the primary component of the decision mechanism is the actual procedure, algorithm, or statistical test employed to evaluate the data and make the decision, such variables as the location of the sample, the number of samples or increments involved, and the rationale behind the action level must be considered. The following are important aspects of decision mechanisms that must be included:

  • number of, rationale for, and size of DUsand SUs
  • number of SUs composing each DU(from one to many)
  • number of increments collected to form each ISM sample
  • bulk mass of ISM sample
  • mass of analytical subsample
  • aspects of “correct sampling” (Pitard 1993)
  • number of replicate ISM samples in each SU
  • particle size reduction or selection (where appropriate)
  • statistic calculated
  • source, nature, and numerical value of the action level

An ISM-based sampling project should be tailored to the decisions for which the data will be used. Careful planning is the key to ISM data usability.

ISM samples can be used for a number of different applications. The type of decision mechanism employed must be consistent with the type of decision being made. The specific size and location of DUs are guided by site knowledge regarding the spatial distribution of contaminant(s) and the movement or behavior of receptors that may contact different areas of the DU. Estimates of mean concentration provided by ISM can be useful in evaluating risk from direct contact with soil, where a DU is designated to correspond with a presumed or actual exposure area for human health or ecological risk assessment. Likewise, because most soil-to-groundwater leachate models assume a volume of contaminated soil as the source of contamination to the groundwater, estimates of mean concentrations in targeted volumes of soil are directly applicable to assessment of soil concentrations using soil-to-groundwater leachate models. Another useful application of ISM is when multiple decisions must be made for very large volumes of soil, for instance, when large former agricultural fields or dredge piles are intended for future residential uses. ISM has been used for the exploration of concentration gradients because, in the presence of small-scale heterogeneity, ISM provides a better understanding of contaminant distribution than a few widely spaced discrete samples. ISM samples may also be used over concentric SUs surrounding a suspected source area. Finally, a variety of different strategies may be used in subsurface investigations with ISM samples.

Regardless of the decision mechanism, the standard steps of data quality assessment as discussed in Section 3 apply. After data are collected, it is important to revisit the CSM and determine whether it is supported by the data or should be modified. Methods for statistical analysis of data should be selected based upon the sampling design and project objectives. Key underlying assumptions associated with the statistical test must be identified and determined to be valid for the data to be analyzed.