10th International Aerosol Conference September 2 - September 7, 2018 America's Center Convention Complex St. Louis, Missouri, USA
Abstract View
Estimation of Human Exposure to Near Road Emission Sources Using a Hybrid Modeling Framework
Fatema Parvez, KRISTINA WAGSTROM, University of Connecticut
Abstract Number: 1502 Working Group: Aerosol Modeling
Abstract Traffic related air pollution is considered one of the major challenges for a large number of urban population. The rapid growth of the world’s motor-vehicle fleet due to population growth and economic improvement causes a significant negative impact on public health. As pollutants from roadway emission sources reach background concentration levels within a few hundred meters from the source, it is very challenging to implement a model that captures this behavior. Currently available air quality modeling approaches can compute the source specific pollutant fate on either a regional or a local scale but still lack effective ways to estimate the combined regional and local source contributions to exposure. Temporal variabilities in human activities and differences in pollutant dispersion pattern in stable and unstable atmospheric conditions greatly influence the exposure. Estimating air pollution exposure from local sources such as motor vehicles while considering all the variables impacting the dispersion make the process computationally intensive.
In this study, we develop a hybrid modeling framework combining a regional model, CAMx - Comprehensive Air Quality Model with Extensions, and a local scale dispersion model, RLINE, to estimate concentrations of both primary and secondary species from roadway emission sources. We utilize all chemical and physical processes available in CAMx and use the Particulate Matter Source Apportionment Technology (PSAT) to quantify the concentrations from onroad and non-road emission sources. We employ RLINE to estimate pollutant distribution from onroad emission sources at a finer resolution. Combining these two models, we estimate combined concentrations at a finer spatial resolution and at hourly temporal resolution.
We further conduct an operational model evaluation of our hybrid modeling framework for the year 2011 for NO2 using both satellite data and regression model data at census block resolution. We find that our hybrid modeling framework performs well with a mean fractional bias 0.15 and a mean fractional error 0.4 when compared to the land-use regression model. We have applied this modeling framework to three major cities in Connecticut (Hartford, New Haven, and Windham) and quantified human exposure to NOx, PM2.5, and elemental carbon (EC). We quantify exposure considering census tract population density and temporal and spatial variability in concentrations. Our approach using a dispersion model is unique as it uses the mass fraction of the total dispersed pollutant at different receptor points and hence is not dependent on extensive roadway emissions data or extensive model runs. Overall, this modeling approach overcomes two major challenges facing hybrid modeling for near roadway exposures - double counting emissions and a lack of temporal variability in estimating concentrations.