Application of Positive Matrix Factorization in the Identification of the Sources of PM 2.5 in Taipei City

Int J Environ Res Public Health. 2018 Jun 21;15(7):1305. doi: 10.3390/ijerph15071305.


Fine particulate matter (PM2.5) has a small particle size, which allows it to directly enter the respiratory mucosa and reach the alveoli and even the blood. Many countries are already aware of the adverse effects of PM2.5, and determination of the sources of PM2.5 is a critical step in reducing its concentration to protect public health. This study monitored PM2.5 in the summer (during the southwest monsoon season) of 2017. Three online monitoring systems were used to continuously collect hourly concentrations of key chemical components of PM2.5, including anions, cations, carbon, heavy metals, and precursor gases, for 24 h per day. The sum of the concentrations of each compound obtained from the online monitoring systems is similar to the actual PM2.5 concentration (98.75%). This result suggests that the on-line monitoring system of this study covers relatively complete chemical compounds. Positive matrix factorization (PMF) was adopted to explore and examine the proportion of each source that contributed to the total PM2.5 concentration. According to the source contribution analysis, 55% of PM2.5 can be attributed to local pollutant sources, and the remaining 45% can be attributed to pollutants emitted outside Taipei City. During the high-PM2.5-concentration (episode) period, the pollutant conversion rates were higher than usual due to the occurrence of vigorous photochemical reactions. Moreover, once pollutants are emitted by external stationary pollutant sources, they move with pollution air masses and undergo photochemical reactions, resulting in increases in the secondary pollutant concentrations of PM2.5. The vertical monitoring data indicate that there is a significant increase in PM2.5 concentration at high altitudes. High-altitude PM2.5 will descend to the ground and thereby affect the ground-level PM2.5 concentration.

Keywords: PM2.5; PMF; online monitoring; photochemical reaction; vertical profile.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Air Pollutants / analysis*
  • Air Pollution / analysis*
  • Air Pollution / statistics & numerical data*
  • Cities
  • Environmental Monitoring / methods*
  • Particle Size
  • Particulate Matter / analysis*
  • Seasons
  • Taiwan


  • Air Pollutants
  • Particulate Matter