Source apportionment of fine organic carbon (OC) using receptor modelling at a rural site of Beijing: Insight into seasonal and diurnal variation of source contributions

Abstract

This study was designed to investigate the seasonal characteristics and apportion the sources of organic carbon during non-haze days (<75 μg m−3) and haze (≥75 μg m−3) events at Pinggu, a rural Beijing site. Time-resolved concentrations of carbonaceous aerosols and organic molecular tracers were measured during the winter of 2016 and summer 2017, and a Chemical Mass Balance (CMB) model was applied to estimate the average source contributions. The concentration of OC in winter is comparable with previous studies, but relatively low during the summer. The CMB model apportioned seven separate primary sources, which explained on average 73.8% on haze days and 81.2% on non-haze days of the organic carbon in winter, including vegetative detritus, biomass burning, gasoline vehicles, diesel vehicles, industrial coal combustion, residential coal combustion and cooking. A slightly lower percentage of OC was apportioned in the summer campaign with 64.5% and 78.7% accounted for. The other unapportioned OC is considered to consist of secondary organic carbon (SOC). During haze episodes in winter, coal combustion and SOC were the dominant sources of organic carbon with 23.3% and 26.2%, respectively, followed by biomass burning emissions (20%), whereas in summer, industrial coal combustion and SOC were important contributors. Diurnal contribution cycles for coal combustion and biomass burning OC showed a peak at 6–9 pm, suggesting domestic heating and cooking were the main sources of organic aerosols in this rural area. Backward trajectory analysis showed that high OC concentrations were measured when the air mass was from the south, suggesting that the organic aerosols in Pinggu were affected by both local emissions and regional transport from central Beijing and Hebei province during haze episodes. The source apportionment by CMB is compared with the results of a Positive Matrix Factorization (PMF) analysis of ACSM data for non-refractory PM1, showing generally good agreement.