Estimation and Uncertainty Analysis of Secondary Organic Carbon Using 1 Year of Hourly Organic and Elemental Carbon Data

Hourly particulate matter 2.5 organic carbon (OC) and elemental carbon (EC) ambient concentrations were measured from February 2012 to January 2013 at a suburban site in the city of Guangzhou, China. The major primary OC and EC sources at this site are vehicular exhaust and regional combustion sources. The annual average OC and EC concentrations (±1SD) were 7.30 ± 5.68 and 2.67 ± 2.28 μgC/m³, respectively. Taking advantage of this large data set of more than 7,000 measurements, we examined the diurnal variation dynamics of the primary OC-to-EC ratio ([OC/EC]_pri), derived using the minimum R squared (MRS) method, which deploys an algorithm that calculates (OC/EC)_pri by exploring the assumed independency between EC and secondary OC (SOC). The MRS-derived hour-by-hour (OC/EC)_pri was the lowest (1.8–2.0) for the morning hours while higher (2.4–3.0) for the afternoon and evening hours. Cumulative evidence from other pollutant measurements suggests that the increase in MRS-derived (OC/EC)_pri in the afternoon and evening was a result of SOC formation from photooxidation of precursors that covary with EC (e.g., vehicular exhaust) and thereby inflate (OC/EC)_pri. We subsequently separate SOC into two categories based on their correlations with EC, that is, SOC_a that covary with EC and SOC_b that is unassociated with EC variations. SOC_a and SOC_b make comparable contributions to SOC, demonstrating that their separation is essential for a more accurate SOC estimation. Otherwise, SOC_a would be mistakenly accounted as primary OC. This approach in estimating SOC may underestimate SOC_a and subject to bias stemming from OC emissions from cooking and biomass burning.