Source identification of personal exposure to fine particulate matter (PM2.5) among adult residents of Hong Kong

Xiao Cui Chen, Tony J. Ward, Jun Ji Cao, Shun Cheng Lee, Ngar Cheung Lau, Steve HL Yim, Kin Fai Ho

Research output: Contribution to journalArticlepeer-review

14 Scopus citations


Epidemiological studies provide evidence of the harmful effects of source-specific fine particulate matter (PM2.5) on human health. Studies regarding relative contributions of multiple sources to personal exposure are limited and inconsistent. Personal exposure monitoring for PM2.5 was conducted in 48 adult subjects (ages 18‒63 years) in Hong Kong between June 2014 and March 2015. We identified seven sources of personal PM2.5 exposure using Positive Matrix Factorization (PMF). These sources included regional pollution (associated with coal combustion and biomass burning), secondary sulfate, tailpipe exhaust, secondary nitrate, crustal/road dust, and shipping emission sources. For personal PM2.5 exposure, one additional source related to individuals' activities was found: non-tailpipe pollution (characterized by Fe, Mn, Cr, Cu, Sr). We also applied principal component analysis (PCA) for PM2.5 source identification. The results revealed similar factor/component profiles using PMF and PCA, with some discrepancies in the number of factors. PCA/absolute principal component scores (PCA/APCs) coupled with a linear mixed-effects model (LMM) was applied to the same dataset for source apportionment, adjusting for temperature and relative humidity. Furthermore, stratified PCA/APCs-LMM models were applied to estimate season- and group-specific source contributions of personal PM2.5 exposure. A mixed source contributions of secondary sulfate, secondary nitrate, and regional pollution were shown (35.1–43.6%), with no seasonal or subject group differences (p > 0.05). Shipping emissions were ubiquitous, contributing 6.3–8.8% of personal PM2.5 exposure for all subjects. Tailpipe exhaust and traffic-related particles varied by season (p < 0.01) and subject group (p < 0.05). Caution should be taken when using source-specific PM2.5 as proxies for the corresponding personal exposures in epidemiological studies.

Original languageEnglish
Article number116999
JournalAtmospheric Environment
StatePublished - Dec 1 2019


  • Ambient air pollution
  • Mixed-effects model
  • Personal fine particles exposure
  • Positive matrix factorization
  • Principal component analysis/absolute principal component scores


Dive into the research topics of 'Source identification of personal exposure to fine particulate matter (PM2.5) among adult residents of Hong Kong'. Together they form a unique fingerprint.

Cite this