Identifying the factors influencing PM2.5 in southern Taiwan using dynamic factor analysis

Several heavily polluted industrial parks are located in the coastal area of Kaohsiung city, which the Taiwan EPA has declared to be the worst air quality region in Taiwan. This research used dynamic factor analysis (DFA) to investigate the source contributions of PM2.5 by monitoring data collected at the four aerosol supersites in Southern Taiwan throughout 2009. Dynamic factor analysis is a technique used to reduce or summarize the dimensions being studied, and is a proven useful technique for this type of study, which handles complex gaseous pollutant conditions. The results of the optimal DFA model showed that PM2.5 concentrations in the Kaohsiung metropolis were primarily influenced by explanatory variables that included sulfate (SO42−), nitrate (NO3−), carbonaceous aerosols, carbon monoxide (CO), sulfate oxides (SO2), nitrate oxides (NO2), and relative humidity (RH). The concentrations were also slightly affected by two common trends representing unexplained variables. Particulate sulfate was the primary variable among the identified explanatory variables. The optimal DFA model satisfactorily accounted for the fluctuations in PM2.5 for the four aerosol supersites (coefficient of efficiency = 0.93). That is, the extreme concentrations of PM2.5 could be successfully described by considering the selected explanatory variables. We used this DFA model successfully to research PM2.5, and future studies concerned with Kaohsiung air quality should consider gaseous pollutants and human activities that our model has identified.

► PM2.5 process investigated by DFA to reveal underlying space-time patterns.
► S/T PM2.5 associated with various aerosol and meteorological data are investigated.
► The most significant contributors to PM2.5 temporal process are identified.
► Two common trends characterize space-time PM2.5 variation are identified.
► DFA can successfully characterize the PM2.5 process in this case study.