Identifying air pollution characteristics, source apportionment methods, and air quality modeling approaches in transport hub settings: State-of-play and future directions

This review synthesises the current state-of-the-art in air quality (AQ) research relating to current monitoring and modeling methods focused on transport hub (TH) settings. Air pollution characteristics from monitoring studies revealed that higher concentrations of PM 2 . 5 (12.4-147.8 μg/m 3 ) and SO 2 (54.1-78.3 μg/m 3 ) dominated AQ issues in ports. Train terminals were dominated by NO 2 (52.2-472.5 μg/m 3 ), with VOCs (123-973 ppb) and UFPs (5×10 3 to 4.8×10 6 particles/cm 3 ) considerably higher at airports. Bivariate polar plots, data filtration techniques, and regression models were considered relatively simple, resource-efficient, and effective source apportionment methods to assess AQ (SO 2 , NO 2 and UFP) from sources in and around THs. Speciated receptor modelling is more expensive but is appropriate for complex environments to evaluate multi-pollutant (PM and VOCs) conditions. Gaussian models demonstrated better agreement than Eulerian and Lagrangian models at airports, with Eulerian models slightly outperforming Gaussian models in port settings. Additionally, Eulerian was the most effective methods to model secondary pollutants and over long distances. Limited AQ research focused on small-scale semi-enclosed THs, such as bus and train terminals, with an additional knowledge gap of indoor AQ in port and airport buildings. Improved characterisation of pollutants like VOCs, BC and PAHs in THs would benefit climate and health impact assessments, with the integration of AI offering a means to enhance monitoring and management AQ at THs in the future. • Local-scale THs (bus and train terminals) are underrepresented in current research • Selection of source apportionment method depends on cost, time and pollutant type • Emphasis for more extensive research of VOCs, BC, and PAHs in THs • Gaussian for local, Eulerian for long-range and secondary pollutant modeling • Data filtration, regression models and bivariate polar plots commonly used in THs