Extrapolation between and within DUs

In some situations, the area to be evaluated is larger than can be effectively sampled. This is typically the case for large tracts of land where available resources may preclude sampling each properly sized DU. One approach in this situation is to sample a portion of the area to be evaluated using ISM and extrapolate data to other, unsampled areas. This can take a number of forms, including (a) dividing the area into DUs, sampling some fraction of the DUs, and extrapolating the mean and/or variance of ISM data from sampled to unsampled DUs and (b) creating SUs within a DU that cover some but not all of the area. Results from the SUs are used to make a decision on the DU.

Justification for extrapolation from sampled to unsampled areas is usually based on a CSM that predicts a similar distribution of concentrations in both areas. Generally, this assumption is based solely on judgment and can be associated with considerable uncertainty. This issue is not unique to ISM and applies equally if extrapolation is considered using data from other sampling strategies such as composite and discrete. It is also important to note that ISM offers no special advantage in reducing this uncertainty. Regulatory acceptance of uncertainty associated with extrapolation can vary considerably, depending on the agency and sometimes site-specific circumstances such as the intended use of the property (e.g., agricultural vs. residential). Discussions within the ITRC ISM Team and feedback from states indicate that extrapolation will be accepted by some states, under some circumstances, but not by others.