A systematic analysis of PM2.5 in Beijing and its sources from 2000 to 2012

Particulate matter with an aerodynamic diameter of 2.5 μm or less (PM_2.5) is the main air pollutant in Beijing. To have a comprehensive understanding of concentrations, compositions and sources of PM_2.5 in Beijing, recent studies reporting ground-based observations and source apportionment results dated from 2000 to 2012 in this typical large city of China are reviewed. Statistical methods were also used to better enable data comparison. During the last decade, annual average concentrations of PM_2.5 have decreased and seasonal mean concentrations declined through autumn and winter. Generally, winter is the most polluted season and summer is the least polluted one. Seasonal variance of PM_2.5 levels decreased. For diurnal variance, PM_2.5 generally increases at night and decreases during the day. On average, organic matters, sulfate, nitrate and ammonium are the major compositions of PM_2.5 in Beijing. Fractions of organic matters increased from 2000 to 2004, and decreased afterwards. Fractions of sulfate, nitrate and ammonium decreased in winter and remained largely unchanged in summer. Concentrations of organic carbon and elemental carbon were always higher in winter than in summer and they barely changed during the last decade. Concentrations of sulfate, nitrate and ammonium exhibited significant increasing trend in summer but in reverse in winter. On average they were higher in winter than in summer before 2005, and took a reverse after 2005. Receptor model results show that vehicle, dust, industry, biomass burning, coal combustion and secondary products were major sources and they all increased except coal combustions and secondary products. The growth was decided both changing social and economic activities in Beijing, and most likely growing emissions in neighboring Hebei province. Explicit descriptions of the spatial variations of PM_2.5 concentration, better methods to estimate secondary products and ensemble source apportionments models to reduce uncertainties would remain being open questions for future studies.