New tools and methods for analysis of the sources and spatial distribution of air pollution
Loading...
Files
Full Text E-thesis
Date
2024
Authors
Byrne, Rósín
Journal Title
Journal ISSN
Volume Title
Publisher
University College Cork
Published Version
Abstract
Recent improvements in air quality sensors (AQSs) have presented an affordable and scalable way to enhance air quality monitoring. However, fully exploiting these new data sources requires novel data-driven methodologies to address the inherent uncertainties in the measurements reported by AQSs. This work goes beyond calibration exercises to highlight the potential of AQS networks for air quality assessments, particularly after demonstrating good sensor-reference correlations and minimal inter-sensor variation. Using three distinct datasets from AQS networks in Cork City and Dungarvan, Ireland, a range of analytical approaches have been developed and applied to understand the local spatiotemporal variability of PM2.5 in these urban areas.
A novel method leveraging high temporal resolution data from PurpleAir sensors linked rapid, short-lived concentration fluctuations to local emissions. The analysis revealed stark spatial contrasts in the proportion of local emissions across Cork City. On average, half of the total PM2.5 exposure originated mainly from local emissions and occurred in just 30% of the time. Discrete Fourier transform spectral analysis was also used to estimate local PM2.5 contributions, yielding comparable results when applied to the same dataset. Analysis of the temporal changes to the contribution of local emissions in Cork City between 2021 and 2023 suggested that legislative changes may be positively influencing the proportion of locally emitted PM2.5 in winter, however, this effect is not uniform across the city.
A new metric, the concentration similarity index (CSI), was developed and optimised for AQS network data. The CSI was used to assess the spatial representativeness of Environmental Protection Agency (EPA) monitoring locations in both Dungarvan and Cork City, revealing moderate exposure representation. However, significant spatial variability within each sensor network indicated strong differences in the degree of representation by the central monitoring locations and emphasised that location type, rather than geographical proximity, plays a key role in air quality representation.
Additionally, a frequency-based method was integrated with the analysis of winter EPA PM2.5 data and aethalometer-derived equivalent black carbon (eBC) data in a small Irish town, to investigate the changing impact of local emissions over time. Analysis of data from 2018 to 2024 showed considerable reductions in winter pollution levels, suggesting an overall improvement in air quality in the area. Analysis of the Ångström absorption exponent indicated an evolving emissions profile, with an increase in wood or peat/turf burning in recent years, while spectral analysis showed general reductions in both regional/background and local particulate pollution.
This research offers insight into cost-effective, spatially dense air quality monitoring solutions that can inform vital pollution mitigation strategies. The new methodologies developed across all datasets, demonstrate the potential for integrating AQS networks and data-driven techniques into air quality monitoring and assessment.
Description
Keywords
Air quality sensors , Particulate matter , Spatial distribution , Air pollution
Citation
Byrne, R. 2024. New tools and methods for analysis of the sources and spatial distribution of air pollution. PhD Thesis, University College Cork.