You are here:
ESTIMATING CONTRIBUTIONS OF OUTDOOR FINE PARTICLES TO INDOOR CONCENTRATIONS AND PERSONAL EXPOSURES: EFFECTS OF HOUSEHOLD CHARACTERISTICS AND PERSONAL ACTIVITIES
Citation:
WALLACE, L. A., R. W. WILLIAMS, J. SUGGS, AND P. A. JONES. ESTIMATING CONTRIBUTIONS OF OUTDOOR FINE PARTICLES TO INDOOR CONCENTRATIONS AND PERSONAL EXPOSURES: EFFECTS OF HOUSEHOLD CHARACTERISTICS AND PERSONAL ACTIVITIES. U.S. Environmental Protection Agency, Washington, DC, EPA/600/R-06/023 (NTIS PB2006-113533), 2006.
Impact/Purpose:
The primary objectives are:
1) to complete the validation and development of exposure databases resulting from the NERL PM panel studies (data produced under TDs 5676 and 3937),
2) to perform analyses on these databases and identify key factors that have the potential of influencing human exposures to PM constituents, and
3) to summarize and report the findings of these additional analyses.
Description:
A study of personal, indoor, and outdoor exposure to PM2.5 and associated elements has been carried out for 37 residents of the Research Triangle Park area in North Carolina. Participants were monitored for 7 consecutive days in each of four seasons. One goal of the study was to estimate the contribution of outdoor PM2.5 to indoor concentrations and personal exposures on an individual basis. This contribution depends on the infiltration factor Finf, the fraction of outdoor PM2.5 remaining airborne after penetrating indoors. Relying on previous studies suggesting that sulfur has few indoor sources, we can estimate Finf for each house based on indoor/outdoor sulfur ratios. Since Finf depends on air exchange rates, we can also calculate the value by regressing the observed indoor/outdoor sulfur ratios on the measured air exchange rates. The two approaches are not completely independent but give excellent agreement (R2 = 0.96-0.99) in estimating the average infiltration factor for each house, thereby validating the air exchange rate measurements and providing considerable confidence in the reliability of Finf estimates. These estimates range from 0.26 to 0.89. Using the daily estimated infiltration factor for each house, we calculate the contribution of outdoor PM2.5 to indoor air concentrations, both for the full year and for the four seasons. We show that in summer when air conditioners were in use, the contribution of outdoor PM2.5 to indoor concentrations was at its lowest (averaging 0.50) for most homes, whereas the average infiltration factor for the other three seasons was 0.62-0.63. The indoor-generated contributions to indoor PM2.5 had a wider range (0-33 µg/m3) than the outdoor contributions (5-22 µg/m3). We then regress the calculated outdoor PM2.5 contribution to indoor air concentrations on measured residential outdoor concentrations. The regressions have moderately high R2 values (range 0.39-0.93), although we present evidence that these values should be treated as upper bounds rather than best estimates.
A second goal was to calculate the "outdoor exposure factor" Fpex, the fraction of outdoor PM2.5 contributing to personal exposure. This can be done directly using the ratio of personal/outdoor sulfur measurements, or indirectly using a model depending on the fraction of time spent indoors. The direct measurements result in an estimate of 0.54 for the mean outdoor exposure factor, whereas the model predicts a mean value of 0.63. We find that the time-weighted model adds no value to our estimate of Fpex and discuss the possible reasons for this. Using the direct estimates of Fpex based on the sulfur ratios, we calculate the contribution of outdoor PM2.5 to personal exposure. The range of outdoor contributions to personal exposure (6-18 µg/m3) is much smaller than the range of indoor contributions (6-34 µg/m3). We regress this contribution on measured outdoor PM2.5. This regression is desired by epidemiologists attempting to relate health effects to outdoor particle concentrations, since it shows the strength of the relationship between exposure to particles of outdoor origin and the outdoor concentrations. The regressions again have moderately high R2 values (range 0.19-0.88), although again we show that these values should be treated as upper bounds.
A third but somewhat subsidiary goal was to estimate two remaining parameters that affect Finf: the penetration coefficient P and the deposition rate k, where these parameters are assumed to refer to sulfur and particles in the same size range as sulfur. We use two approaches, which, however, are not completely independent. Both involve using the indoor/outdoor sulfur ratio (or its inverse) and the air exchange rate (or its inverse). While the two approaches generally agree well on predicting P and k for individual homes, each includes some nonphysical estimates (e.g., P >1). Also, many of the estimates for k are not significantly different from zero. We conclude that the lack of an independent method for estimating P and k results in our inability to validate these estimates for individual homes. On the other hand, the overall estimate of the average P (0.81-0.85) and k (0.22-0.23 h-1) for all homes may be more stable.
As a result of our ability to split indoor and personal exposures into indoor-generated and outdoor-generated particles, we were able to run regressions on these variables as well. This allows a better estimate of the impact of air exchange rates, since air exchange operates in opposite directions for these two variables. The results show that increases in air exchange (either measured or approximated by questionnaire responses referring to open windows) significantly affect each of these variables in the expected direction (increasing the impact of outdoor-generated particles, decreasing the impact of indoor-generated particles).
A fourth and final goal of this analysis was to identify household characteristics and personal activities that may affect exposure. A number of multiple regressions were performed on the personal, indoor, and outdoor PM2.5 and sulfur concentrations. A simple regression of indoor PM2.5 on outdoor PM2.5 resulted in an R2 value of only 9%; the multiple regression improved the R2 to 42% (using the outdoor monitors near the home) and 43% (using the central-site monitor). Important variables affecting indoor air PM2.5 concentrations included smoking and cooking, variables often found in other studies. The number of persons in the household contributed to PM2.5 and this has also been noted before. However, burned food, use of a kitchen exhaust fan, number of pilot lights, and duration of candle use were also important contributors, and these have not generally been noted in other studies.