In order to join virtual sessions, you must be registered and logged-in(Were you registered for the in-person meeting in Denver? if yes, just log in.) 
Note: All session times are in Mountain Daylight Time.

The Effectiveness of Using Different Mobile Monitoring Methods to Predict the Spatial Variation of Ambient Air Pollution in Canadian Cities

Authors: Felix Massey*, University of Toronto, Jérémy Gelb, Institut national de la recherche scientifique, Philippe Apparicio, Institut national de la recherche scientifique, Vincent Jarry, Institut national de la recherche scientifique, Matthew Adams, University of Toronto
Topics: Spatial Analysis & Modeling, Geography and Urban Health, Land Use
Keywords: Land-use regression, mobile monitoring, air pollution, neural networks
Session Type: Paper
Presentation File: No File Uploaded

This study investigates the effectiveness of using various mobile monitoring methods to generate land-use regression (LUR) models to predict the spatial variation of ambient air pollution concentrations in Canadian cities: Mississauga, Hamilton, Montréal and Québec. This research is significant because there are only a few passive air monitoring stations located throughout the region which may not capture the temporal and spatial complexity of air pollution. LUR provides a convenient approach for pollutant exposure modelling while also offering the benefit of accounting for small-scale variability in urban environments. Mobile monitoring techniques were used to obtain spatially-varying air pollution estimates across a mix of land-use conditions and socioeconomic zones. The pollutants that were monitored in this study include nitrogen dioxide (NO2) and ozone (O3) which were sampled using bicycles equipped with low-cost air pollution sensors and a mobile air pollution laboratory with research-grade instruments. Separate models were created for both monitoring methods to compare the accuracy and efficiency of using low-cost air pollution sensors to predict the spatial variability of nitrogen dioxide and ozone. Cross-validation was used to evaluate model performance and artificial neural networks were integrated to increase the explained variance. These LUR models can be used to generate interpolated continuous pollution surfaces across the region to identify exposure levels at unobserved locations while also assessing the effectiveness of pollution modelling using low-cost sensors. This research provides an evidence-based assessment of nitrogen dioxide and ozone pollution exposure which could be used for administrative purposes and designing local air pollution monitoring systems.

To access contact information login