Uncertainty from synergistic effects of multiple parameters in the Johnson and Ettinger (1991) vapor intrusion model

Migration of volatile chemicals from the subsurface into overlying buildings is known as vapor intrusion (VI). Under certain circumstances, people living in homes above contaminated soil or ground water may be exposed to harmful levels of these vapors. VI is a particularly difficult pathway to assess, as challenges exist in delineating subsurface contributions to measured indoor-air concentrations as well as in adequate characterization of subsurface parameters necessary to calibrate a predictive flow and transport model. Often, a screening-level model is employed to determine if a potential indoor inhalation exposure pathway exists and, if such a pathway is complete, whether long-term exposure increases the occupants’ risk for cancer or other toxic effects to an unacceptable level. A popular screening-level algorithm currently in wide use in the United States, Canada and the UK for making such determinations is the “Johnson and Ettinger” (J&E) model. Concern exists over using the J&E model for deciding whether or not further action is necessary at sites as many parameters are not routinely measured (or are un-measurable). Many screening decisions are then made based on simulations using “best estimate” look-up parameter values. While research exists on the sensitivity of the J&E model to individual parameter uncertainty, little published information is available on the combined effects of multiple uncertain parameters and their effect on screening decisions. This paper presents results of multiple-parameter uncertainty analyses using the J&E model to evaluate risk to humans from VI. Software was developed to produce automated uncertainty analyses of the model. Results indicate an increase in predicted cancer risk from multiple-parameter uncertainty by nearly a factor of 10 compared with single-parameter uncertainty. Additionally, a positive skew in model response to variation of some parameters was noted for both single and multiple parameter uncertainty analyses. From these results, an example order of data collection is presented to reduce output uncertainty. An on-line version of the model with automated uncertainty analyses is available to the public on EPA's Office of Research and Development website (http://www.epa.gov/athens/onsite).