Source apportioning of primary and secondary organic carbon in summer PM<sub>2.5</sub> in Hong Kong using positive matrix factorization of secondary and primary organic tracer data

Di Hu, Qijing Bian, Alexis K.H. Lau, Jian Zhen Yu

Journal of Geophysical Research Atmospheres · 2010 · 120 citations · 50 references

DOIFull text

Open access

Concepts

Abstract

The major inorganic constituents and organic tracer compounds in PM 2.5 were used in positive matrix factorization (PMF) and chemical mass balance (CMB) models to apportion the primary and secondary source contributions to organic carbon (OC) in Hong Kong during the summer of 2006. Secondary organic aerosol (SOA) tracers of several biogenic and anthropogenic hydrocarbons were included in the PMF analysis. Their inclusion allowed the identification of two components of SOA among seven factors resolved by PMF. One SOA component was mainly associated with secondary sulfate and nitrate. The other SOA component, characterized by biogenic SOA tracers and mixed with biomass burning and vegetative detritus particles, was biomass burning‐induced SOA. Secondary OC (SOC) apportioned by PMF (SOC PMF ) was on average 6.84 μ gC m −3 (65% of PM 2.5 OC) on high pollution days under influence of significant regional transport (i.e., regional days) and 0.70 μ gC m −3 (25% of PM 2.5 OC) on days under the influence of mainly local emissions (i.e., local days). The biomass burning‐induced SOA accounted for 20% of the total SOA on the regional days, underlining the importance of biomass burning aerosol source in this region. The average uncertainty for the SOC PMF estimates was ∼20% on the regional days and ∼120% on the local days. SOC PMF was compared with SOC determined by CMB (SOC CMB , i.e., unapportioned OC by CMB analysis) and a tracer‐based method (SOC TBM ) that apportioned SOC contributions by four hydrocarbon precursors including isoprene, monoterpenes, β ‐caryophyllene, and toluene. The three estimates of SOC closely tracked with each other among individual samples. The SOC CMB and SOC PMF estimates on the majority of the regional days differed from each other by less than 25%. Good correlations between contributions of SOC and individual primary OC sources apportioned by PMF and CMB further added to the credence to the PMF‐derived estimation of secondary and primary OC source contributions by using secondary and primary aerosol organic tracers as the fitting species.

References

50