In recent years, portable air pollution monitors have been increasingly used to measure air pollutant concentrations in various microenvironments. Typically, traditional methods take the measured PM2.5 concentrations as the microenvironmental pollution level without taking the spatiotemporal variability of ambient PM2.5 concentration into consideration. This thesis research develops one method for apportioning the measured PM2.5 concentration into ambient and microenvironment-related contributions.
Field sampling campaigns were conducted to quantify PM2.5 exposure concentration in several typical Hong Kong transport microenvironments (TMEs), including Mass Transit Railway (MTR), minibus, double-decker bus, tramcar (on-road). The analysis results complemented the understanding of pollution level and variability in the TME exposure concentration. In particular, the in-cabin spatial variability of PM2.5 concentration was observed for MTR and double-decker buses mainly due to the change of ventilation condition (e.g., MTR train door open versus close). In addition, a large fraction of variability in PM2.5 exposure concentration (e.g., more than 90% of tramcar (on-road) PM2.5 concentration) can be explained by ambient PM2.5.
Furthermore, a bootstrapping-based minimum sample size method was developed to obtain a precise estimation of statics (e.g., mean, median, percentiles) of the data set. Simulation case studies were conducted using synthetic data set and measured microenvironmental PM2.5 concentration. The results highlight a considerably less intensive sampling data set could provide similar PM2.5 concentration estimates with high precision and low bias.
To conclude, this thesis research has focused on characterizing TME-related PM2.5 concentration. The results provide insights on the contributions of ambient PM2.5 and microenvironment-related part to TME PM2.5 exposures, and can be used to improve the design of microenvironmental air pollution exposure study regarding sampling interval and duration. Future research is needed to validate the proposed apportioning method using a larger sampling data set.