Establishing the spatial variability of ambient nitrogen dioxide in Windsor, Ontario

This study is aimed at identifying the determinants of intra-urban variation in ambient concentrations of nitrogen dioxide (NO2) in Windsor, Ontario, Canada, and to develop a model to predict NO2 in the city. NO2 samples were collected over a two-week period during winter 2004 by using passive samplers in 50 locations across Windsor. A land use regression (LUR) model was constructed to generate independent variables that could best predict NO2 concentrations across the city. We tested 71 variables of land use, traffic, road characteristics, population and dwelling counts, and physical geographic variables such as elevation. The final model contained five variables: distance to the Ambassador Bridge, daily traffic on Class 1 and Class 2 arterial roads within 400 m, industry within 600 m, and total population within 700 m. All the variables were statistically significant and had the expected direction of correlation. The final regression model yielded a coefficient of determination, R2, of 0.88. The spatial variations captured in this analysis are being used to guide the selection of participants for a number of Health Canada studies.