Traffic tracking and air quality: A holistic approach to predicting traffic-related air pollution
| dc.contributor.author | Deveer, Laura Ahinee | |
| dc.contributor.supervisor | Minet, Laura | |
| dc.date.accessioned | 2024-09-26T21:27:11Z | |
| dc.date.available | 2024-09-26T21:27:11Z | |
| dc.date.issued | 2024 | |
| dc.degree.department | Department of Civil Engineering | |
| dc.degree.level | Master of Applied Science MASc | |
| dc.description.abstract | On-road transportation has long been a significant contributor to air pollution in cities. Over the past few decades, the transportation sector, which has been targeted by major regulations, has undergone substantial changes. These changes include a shift in vehicle fleet composition and both natural and artificial alterations to traffic patterns. Despite the importance of on-road transportation for urban mobility, it remains a major source of air pollution and a public health challenge. Therefore, it is essential to accurately measure and model the temporal and spatial distribution of traffic-related air pollution. In that way, targeted implementations can be made to combat the adverse health effects associated to the exposure to these pollutants. In this thesis, we investigate the potential of low-cost methods to accurately estimate air pollutant concentrations. To this end, we employed modern traffic tracking technologies with low-cost sensors and machine learning techniques. The research addresses the effectiveness of leveraging these technologies to understand the factors and interactions that influence air quality. Chapter 2 predicts real-time air pollutant concentrations with high accuracy across pollutants using traffic videos and machine learning algorithms. The results reveal the superior performance of non-linear models over linear models. In addition, the Shapley additive explanation plots employed in this study effectively captured the intricate relationships between pollutants and their predictors. Chapter 3 examines the influences of traffic, particularly from cruise ship activities, on local air quality in James Bay, Victoria, BC. Results indicate that both emissions from traffic and cruise ship activities affected air quality. The integration of low-cost sensors with these traffic tracking technologies proves crucial for accurate air quality analysis and allows for context-specific and real-time assessments, providing valuable insights for policy makers and urban planners. | |
| dc.description.embargo | 2025-09-11 | |
| dc.description.scholarlevel | Graduate | |
| dc.identifier.uri | https://hdl.handle.net/1828/20458 | |
| dc.language | English | eng |
| dc.language.iso | en | |
| dc.rights | Available to the World Wide Web | |
| dc.subject | traffic-related air pollution | |
| dc.subject | machine learning | |
| dc.subject | air pollutant prediction | |
| dc.subject | computer vision | |
| dc.title | Traffic tracking and air quality: A holistic approach to predicting traffic-related air pollution | |
| dc.type | Thesis |