Assessing traffic and industrial contributions to ambient nitrogen dioxide and volatile organic compounds in a low pollution urban environment

Abstract

Land use regression (LUR) modeling is an effective method for estimating fine-scale distributions of ambient air pollutants. The objectives of this study are to advance the methodology for use in urban environments with relatively low levels of industrial activity and provide exposure assessments for research on health effects of air pollution. Intraurban distributions of nitrogen dioxide (NO2) and the volatile organic compounds (VOCs) benzene, toluene and m- and p-xylene were characterized based on spatial monitoring and LUR modeling in Ottawa, Ontario, Canada. Passive samplers were deployed at 50 locations throughout Ottawa for two consecutive weeks in October 2008 and May 2009. Land use variables representing point, area and line sources were tested as predictors of pooled pollutant distributions. LUR models explained 96% of the spatial variability in NO2 and 75-79% of the variability in the VOC species. Proximity to highways, green space, industrial and residential land uses were significant in the final models. More notably, proximity to industrial point sources and road network intersections were significant predictors for all pollutants. The strong contribution of industrial point sources to VOC distributions in Ottawa suggests that facility emission data should be considered whenever possible. The study also suggests that proximity to road network intersections may be an effective proxy in areas where reliable traffic data are not available.

Keywords: Air pollution; Industrial emissions; Land use regression; Nitrogen dioxide; Traffic predictor; Urban; Volatile organic compounds.