You are here:
MODELING RELATIONSHIPS BETWEEN MOBILE SOURCE PARTICLE EMISSIONS AND POPULATION EXPOSURES
Impact/Purpose:
For year 6 of the project, we had proposed extending our intake fraction (iF) methodology to address motor vehicle emissions, as a way of informing PM control decisions and future analyses. Our specific objectives were to:
- Evaluate geographic patterns in primary and secondary particulate matter iFs for mobile sources, using a national-scale source-receptor (S-R) matrix
- Determine the relative contributions of near-source and long-range populations to particulate matter iFs for mobile sources in different geographic locations
- Develop predictive regression equations for iFs to explain geographic patterns as a function of population density and meteorological covariates.
Description:
Results from this analysis were recently published (Greco, et al., 2007b). For primary fine particulate matter emitted from mobile sources, the intake fractions varied across source counties from 0.14 to 23 per million (median of 1.2 per million). These values were highly correlated with near-source population density; the population in the source county explained 43% of the variability in the above estimates, and a multivariate regression model with population at various radii from the source explained 86% of the variability. Spatial analyses of residuals indicated generally strong model performance, with greater errors along the coasts, where wind fields are more difficult to characterize and downwind populations may be less significant.
For secondary ammonium sulfate formed from SO2 emissions, the median intake fraction (0.43 per million) was somewhat lower than for primary PM. The variability was similar to that for primary PM, but with more regional variability rather than small-scale spatial variability. In spite of the regional influence on atmospheric chemistry, multivariate regressions with only population terms had an R2 of 0.78, indicating the significance of population patterns even in this context. However, there was relatively greater statistical significance for population beyond 200 km from the source, relative to primary PM, and relatively lower statistical significance for population within 200 km, reflecting expected concentration patterns.
Secondary ammonium nitrate formed from NOx emissions had an even lower median intake fraction (0.072 per million), with spatial variability driven somewhat by population patterns (R2 of 0.63 in multivariate regression model) but also by relative ambient concentrations of sulfate, nitrate and ammonium. Higher values tended to be found in the Midwest, where there is adequate ammonia to neutralize nitrate (and lower ambient sulfate), versus higher levels in the Ohio River Valley and Northeast for secondary sulfate and primary PM.
We also quantified the extent to which SO2 controls might free up ammonia to react with nitrate, thereby increasing ammonium nitrate concentrations. We determined that the public health benefits of SO2 emission controls (due to sulfate reductions) would be offset by ammonium nitrate increases by an average of 9%, ranging from 1% to 29% across U.S. counties.
As mentioned above, one of our primary objectives was to determine the relative importance of near-source and long-range populations. The median distances within which half of the total intake fraction was realized was about 150 km for primary PM, 450 km for secondary sulfate, and 390 km for secondary nitrate. However, these values varied substantially by setting (i.e., range for primary PM from 0 km, indicating that more than 50% of the iF was realized in the source county, to 1800 km). In dense urban areas, often a majority of the intake fraction was realized within the source county, indicating that more geographically resolved dispersion modeling may be warranted.
Comparing our results with the published literature, the magnitude of our estimates appear reasonable, and this analysis remains the first attempt to characterize spatial variability in mobile source intake fractions and to derive conclusions about the model scope and resolution needed to accurately estimate public health benefits of pollution control from mobile sources. Specifically, we concluded that a national-scale county-resolution dispersion model is likely sufficient for secondary particulate matter or primary particulate matter in rural areas with substantial downwind populations, but that more resolved models should be explored in dense urban areas or less-populated areas without significant downwind populations.
Based on the findings from Greco, et al. (2007b) we proceeded with follow-up work addressing potential within-county heterogeneity in primary PM mobile source intake fractions, as well as the questions of the spatial extent of the iF for sources within urban areas and the potential biases in estimates based on county-level resolution. We used the CAL3QHCR dispersion model (in the CALINE family of models) to simulate the influence of line-source emissions on concentrations on 23,000 road segments in the Boston area. A year’s worth of hourly intake fractions were determined for each road segment using actual meteorological conditions and residential population patterns. The annual average values for the road segments range from 0.8 to 53 per million, with a mean of 12 per million. On average, 46% of the total exposure is realized within 200 m of the road segment, though this varies from 0–93% across road segments, largely due to variable population patterns. Our findings indicate the likelihood of substantial intra-urban variability in mobile source primary PM2.5 iF, especially as taking into account population dynamics, localized meteorological conditions, and street-canyon configurations might all increase the variability in iF. These results were published as part of a doctoral thesis, and a manuscript has been submitted to Environmental Science & Technology (Greco, et al., 2007a).