Land Use Regression of Particulate Matter in Calgary, Canada

Abstract

GIScience 2016 Short Paper Proceedings Land Use Regression of Particulate Matter in Calgary, Canada S. Bertazzon 1 , F. Underwood 1 , M. Johnson 2 , J. Zhang 2 University of Calgary, Department of Geography, 2500 University Dr. NW, Calgary, AB, Canada, T2N 1N4 Email: {bertazzs, feunderw}@ucalgary.ca Health Canada, Air Health Science Division, 269 Laurier Ave West, Ottawa, ON, Canada, K1A 0K9 Email: markey.johhnson@hc-sc-gc.ca; joyce.zhang@alumni.utoronto.ca Abstract Two-week integrated samples of particulate matter (PM 1.0 , PM 2.5 , PM 10 ) were collected in summer and winter in Calgary, Canada. PM concentrations were higher in summer for all size fractions. In both seasons, spatial variation and clustering were moderate. Land use regression (LUR) models were estimated for each PM size fraction and season, yielding R 2 > 0.75 for PM 2.5 and PM 10 in summer, and R 2 > 0.45 for PM 1.0 in summer and for all winter models. Summer models yielded consistent predictors across size fractions, representing industrial emissions, local traffic, and major arterial traffic. Winter predictors included industrial emissions, major arterial traffic, and distance from open, snow-covered parks. The models suggest industrial pollution covered large areas in both seasons, and was affected by prevailing winds in summer, whereas traffic-related pollution decayed rapidly as distance from roads increased. 1. Introduction Particulate matter (PM) is a mixture of small particles: acids, organic chemicals, metals, and dust particles (EPA 2016). Coarse particles (PM 10 ) are 2.5−10 micrometers in diameter; fine particles (PM 2.5 ) are less than 2.5 micrometers. Particulate pollution is associated with reduced visibility, environmental degradation, and adverse health effects, e.g., respiratory and cardiovascular morbidity and mortality (Ruckerl et al. 2011), with evidence that health impacts and chemical composition vary by size fraction (Kelly and Fussell 2012). Land use regression (LUR) yields air pollution estimates at fine spatial resolution based on the relationship between air pollution values and land use variables observed at sampled points (Henderson et al., 2007). Most LUR literature focuses on NO 2 , with a few studies modelling PM 2.5 , ultrafine particles, and PM components (e.g., Henderson et al., 2007, Zhang et al., 2015). This paper is the first study comparing models for three PM size fractions. Further novel elements in the well-established LUR literature are the inclusion of prevailing winds and the use of GIScience to advance spatial understanding of air pollution: an example of best practice for a spatial turn in health and environmental research (Richardson et al., 2013). 2. Methods Air monitoring campaigns were conducted in Calgary in August 2010 and January-February 2011. A network of 50 monitors was deployed in each campaign (Bertazzon et al. 2015). Due to power outages and equipment failures, the campaigns yielded 27 valid summer PM samples and 29 winter samples. Predictor variables were defined on circular buffers from each sampling point. In addition, windrose variables were defined on buffers modified according to the prevailing winds in each season (Zhang et al. 2015). Getis G and Moran’s I spatial statistical tests were conducted to assess spatial clustering and autocorrelation in the variables, based on a row-standardized 3-nearest-neighbours spatial