Evaluating methods for predicting indoor residential volatile organic compound concentration distributions.

Accurate modeling of exposure to volatile organic compounds (VOCs) over a large study population depends on proper characterization of concentrations in the indoor residential environment. However, owing to the high expense of field sampling campaigns for determining indoor air concentrations, such studies have only been conducted for limited populations. Therefore, there is a need to determine the degree to which results can be extrapolated to unstudied settings through the use of models, the most appropriate information required to do so and the potential errors associated with the use of sub-optimal information. The goal of this analysis is to evaluate three different source indicators used to predict indoor VOC concentration distributions for a new study population. Data from two field studies are used. For each data set, source strength, indoor-outdoor (I-O) difference and indoor/outdoor (I/O) ratio, collectively referred to as source indicators, are calculated and fit with distributions. These distributions, as well as distributions for air exchange, volume and outdoor concentrations for the new study population, are used for predicting indoor concentrations using Monte Carlo simulations, which are then compared with actual distributions. As expected, the source strength often provides the most effective predictions (11 out of 20 instances), but is slightly outperformed by, although is still comparable with, the I-O difference on some occasions (4 out of 20). The I/O ratio generally has the greatest prediction errors, given its dependence on outdoor concentrations, but performs optimally in a limited number of cases (5 out of 20). When deciding between the source strength and I-O difference, one must consider the availability and fidelity of both current and future data. On the basis of our findings, exposure-monitoring studies should report the distribution statistics for I-O differences and, if the data are available, for source strengths.