media. Off-site contamination and human exposure have been modeled rather than measured directly in off-site sampling of groundwater, drinking water, soil, and air. The estimates of off-site contamination of media derived from these models are then used as surrogates for actual exposure measurements in the calculation of risk assessments for each contaminant. The lack of validation studies for many of these modeling approaches may introduce errors in exposure estimates, which in turn introduce further uncertainty in the risk-assessment process. In ATSDR's review of 951 NPL sites, 75-80 percent were found to have reached the NPL without adequate off-site sampling, such as sampling of nearby homeowners' wells for drinking-water contamination suggested by migration models or sampling of soil in near off-site locations where lead contamination was likely (H. Emmett, ATSDR, personal communication, 1990). Groundwater was most likely to be monitored off-site; surface water, air, and soil releases were even more rarely sampled, although citizen reports of odors and concerns about airborne chemicals frequently are the complaints that lead to site evaluations (Layefsky et al., 1988). Recently, EPA has begun to direct more resources to off-site sampling, and the agency has incorporated more direct measures of population exposure in the revised HRS (EPA, 1990b).
Minimum data requirements for sampling pathways of exposure