Vehicular emissions have a significant impact on roadside pollutants concentration in urban areas, especially in densely populated cities like Hong Kong characterize by high population and vehicle densities. Managing roadside air quality in such settings is challenging due to the complex interplay of emission and dispersion factors. However, comprehensive roadside air quality monitoring and actual emission-related data are relatively rare. Moreover, real-world vehicle emission measurements, which are essential for emission management, are costly and labour-intensive, limits their implementation, especially in developing countries. To address these challenges, this thesis presents an innovative, cost-effective roadside sensor network (RSN) for monitoring air quality and estimating emissions. A deconvolution method was developed and applied to improve sensor response times so that they could capture rapid variations in roadside air pollution concentrations. Monitoring campaigns were carried out at five typical roadside sites in Hong Kong, representing urban and suburban areas with different land use and traffic patterns. The concentrations of kerbside NOx and CO were found to be highly skewed. Pollutant levels were related to fleet composition, and the main contributor varied between sites. Higher concentrations reflected plume segments from larger vehicles. In street canyons, airflows tended to move with vehicles along the canyon. Sensors placed at 2-5m high above the kerbside showed more auto-correlation, indicating less variability in low-turbulence areas away from traffic. Concentrations also exhibited cyclic changes corresponding to nearby traffic signal timings. This study successfully demonstrated the use of RSN to establish emission factors (EFs) for individual vehicles and the overall vehicle flow. Individual plume segments captured by sensor nodes were isolated, and the ratio of specific pollutant concentrations to CO
2 was used to estimate EFs. A customized deep learning model classified vehicle types from photos, revealing significant differences in emission between vehicle groups. Data were collected from observations of 856 plume groups emitted by more than 590 individual vehicles. Open-source bus data, including fuel type, emission standards and manufacturing year, allowed further detailed analysis. The EFs exhibited skewness with EF
NOx for vehicle types ranked as follows: trucks > vans (goods vehicles) > buses > private cars > taxis, and diesel > petrol > LPG, with larger engine sizes dominating. The EF
CO demonstrated the following classification: minibus > taxis > double-decker buses > private cars > vans and trucks. Moreove, the primary emission of NO
2 were also estimated from the plumes of passing vehicles, indicating decreasiing trend as vehicles get older. Further, two key climate forcing pollutants were included in this study. EF
N2O and EF
CH4 were first estimated from roadside monitoring, showing high variation and comparable to previous dynamometer tests. Diesel trucks and vans emit much higher EF
N2O, while LPG-fueled minibuses and taxis were the main emitters of CH
4. This research contributes to a better understanding of roadside air quality and provides a cost-effective method for estimating EF. It bridges the gap in roadside air quality management and emission estimation, enabling more efficient and targeted measures to mitigate the adverse effects of vehicular emissions on urban air quality. Further, it introduces two important climate forcing pollutants for emissions estimation.
| Date of Award | 2024 |
|---|
| Original language | English |
|---|
| Awarding Institution | - The Hong Kong University of Science and Technology
|
|---|
| Supervisor | Zhi NING (Supervisor) & Zhe WANG (Supervisor) |
|---|