Developing a deep understanding of air pollution in complex source locations

Despite improvements in technology and increasing attention to air quality, air pollutant source identification remains a challenging issue. Identifying and understanding emission sources is important for informing policy to combat the adverse effects of air pollution. This Thesis presents a comprehensive exploration of air quality monitoring data with the objective of advancing source identification methodologies. Three interrelated research components are investigated, each contributing to a deeper understanding of air pollution. The importance of regression analysis in source identification is explored, with focus on the robustness of the York regression technique, which is not commonly used in atmospheric sciences. The results underscore the importance of taking account measurement uncertainty when calculating pollutant ratios. Additionally, meteorological normalisation, a well-established method in air quality analysis, is enhanced for source identification purposes. The important impact of particle re-suspension is revealed, which is shown to influence the concentrations of a range of metals. Inverse modelling using the ADMS dispersion model is applied to provide estimates of sources strengths for different source types. Of particular note is the development of weather-dependent emission rates, which explain much of the variation in concentrations seen for key metals, which has implications for the control of metal concentrations. Overall, this research contributes to the ongoing efforts in understanding and addressing air pollution challenges, particularly in complex industrialised settings, and sets the stage for further advancements in source identification methodologies using data science and modelling techniques.