10th International Aerosol Conference September 2 - September 7, 2018 America's Center Convention Complex St. Louis, Missouri, USA
Abstract View
Neighborhood-scale Spatial Variability of PM Mass and Number and Exposure Misclassification in an Eastern US City
Hugh Li, Peishi Gu, Qing Ye, Naomi Zimmerman, Ellis Shipley Robinson, R. Subramanian, Joshua Apte, Allen Robinson, ALBERT A. PRESTO, Carnegie Mellon University
Abstract Number: 1272 Working Group: Aerosol Exposure
Abstract Long and short-term exposure to airborne pollutants are associated with adverse health effects. Regulatory monitors can be used to determine if their surrounding regions meet regulatory standards of air pollution. Measurements at monitors describe surrounding residents’ air pollution exposure, help derive exposure dose response functions in epidemiology studies, and are essential input for chemical transport model and remote sensing. But evaluation is needed to assess spatial representativeness of monitors in different environments. We measured CO (traffic source marker), NO2, ultrafine particle concentration (UFP), and PM1 with a mobile laboratory in a post-industrial U.S. Eastern city (Pittsburgh, PA) in 2016. We sampled in ~1 km2 areas in 15 neighborhoods representing different land use and exposure regimes (e.g., urban and rural, high and low traffic). Mobile sampling was conducted on up to 15 distinct days in each neighborhood to study fine-scale spatial variation in pollutant concentrations. Each 1 km2 neighborhood was subdivided into 50 m cells for spatial analysis. CO and NO2 exhibited within-neighborhood (~1 km2) spatial variation, with hotspots elevated by up to a factor of 5 above the regional background. UFP was the most variable, with spatial variations up to an order of magnitude higher than background. PM1 showed the least spatial variability. Spatial variability is driven by local sources, such as traffic at urban sites.
Mobile measurements convolve spatial and temporal variations in concentration, and we use multiple methods to separate these sources of variability. First, we compare observed spatial variations within each 1 km2 neighborhood to typical hour-to-hour variations measured with a stationary monitor placed in the neighborhood. Spatial differences outside the interquartile range of hourly temporal differences are deemed significant. This analysis indicates that using a single monitor to represent the surrounding ~1 km2 areas could introduce an average daily exposure misclassification of 46 ppb (S.D. = 26) for CO (30% of regional background), 3 ppb (S.D. = 2) for NO2 (43% of background), 4000 #/cm3 (S.D. = 1900) for UFP (64% of background), and 1.2 µg/m3 (S.D. = 1) for PM1 (13% of background). Exposure differences showed fair correlation (R2 = 0.5) with traditional land use covariates such as traffic volume and restaurant density. Secondly, we used Wilcoxon signed-rank test to determine which 50 m cells represent the same underlying distribution of concentrations, and which are different. This test allows us to examine the robustness of our exposure misclassification calculations, and to determine the number of monitors required in each neighborhood to reduce pollutant exposure misclassification within 10%.