Prediction of Indoor Air Quality from Soil‐Gas Data at Industrial Buildings

Abstract

AbstractSubslab or shallow soil‐gas data are often compared with indoor air concentration data in vapor intrusion (VI) evaluations. If no indoor air data are available or confounding sources are present, or if future scenarios are considered, the soil‐gas data may be used to estimate the indoor air concentrations due to VI. The typical approach in risk assessments is to use the 95th percentile values from a set of concentration data. For VI studies, however, this rarely is an option because the data sets tend to be quite small. Therefore, various guidance documents urge the use of maximum soil‐gas values. This may be reasonable for small residential buildings, but can lead to very conservatively biased estimates if applied to large industrial buildings with localized areas of contamination, especially given that the sampling locations may not be randomly selected and instead are biased toward worst‐case locations. By this approach, VI guidance implicitly tolerates a large percentage of false positive decision errors to minimize the number of false negative decision errors. In this paper, implications of using maximum values are discussed and illustrated with data sets from a number of large industrial buildings at various sites. An alternative approach to using maximum soil‐gas values is proposed that serves to reduce the number of false positive results while controlling the number of false negatives to an acceptable level.