218 Environmental Geoscience: Global Science, Local Focus

Wednesday, 21 October 2009
8:00 AM - 12:00 PM
Portland Ballroom 253

GEOSTATISTICAL ANALYSIS OF AMBIENT AIRBORNE CONTAMINANT VARIABILITY ACROSS AN INTERNATIONAL BORDER IN DETROIT AND WINDSOR

Paper number: 218-11
Program: Technical Sessions
Time: 10:30 AM – 10:45 AM

Authors

Abstract

Spatial variability of airborne contaminants was assessed in an international airshed spanning Detroit, Michigan, USA and Windsor, Ontario, Canada. In this study, refined geostatistical methods were employed to characterize air pollutant distributions measured by the Geospatial Determinants of Health Outcomes Consortium (GeoDHOC), a multidisciplinary, international effort aimed at understanding the health effects of air pollution in urban environments. The GeoDHOC team deployed a combination of active and passive air sampling devices for a two-week period during September 2008. Passive diffusion monitors were used to measure nitrogen dioxide (NO2), sulfur dioxide (SO2) and 26 volatile organic compounds (VOCs) at 100 sampling sites. Active samplers collocated at 50 of the passive sites sampled particulate matter (PM) and 23 polycyclic aromatic hydrocarbons (PAHs). Active and passive samplers were distributed at an approximate spacing of 1 per 10 km2 and 1 per 5 km2, respectively.

Statistical and geostatistical analysis was conducted for PM and NO2, as well as 14 VOC and 7 PAH species measured above method detection limits at ≥80% of sampling locations. Mean concentrations for all analyzed contaminants except the 2.5-10 μm PM size fraction were lower in Windsor by a factor of two or less. Moderate positive correlation coefficients (statistically significant at p<0.05) were observed between NO2 and analyzed PAHs (r = 0.603), analyzed VOCs (r = 0.529), and PM1-2.5 (r = 0.555) for the study area. Cross-variogram analysis was performed to describe cross-continuity between these pollutants. Co-kriging was used to generate PM and PAH concentration maps taking advantage of NO2 as secondary data to improve resolution between the more widely distributed active samplers. Concentration maps produced through both ordinary kriging and co-kriging were compared. Map patterns for NO2 and total VOCs, with highs associated with major roadways and border crossings, suggest mobile emissions as the primary source of these compounds. Conversely, the spatial distributions of PM and PAH concentrations primarily reflect the distribution of individual point sources. Observed patterns in spatial variability among pollutants and between cities underscore the importance of local anthropogenic sources over regional background.