Evaluation of the ability of the EC tracer method to estimate secondary organic carbon

The elemental carbon (EC) tracer method has often been used to estimate the primary and secondary organic aerosol (OA) fractions using field measurements of organic carbon (OC) and EC. In this observation-based approach, EC is used as a tracer for primary OC (POC), which allows for the estimation of secondary OC (SOC). The accuracy of this approach is evaluated using concentrations generated by PMCAMx, a three-dimensional chemical transport model that simulates the complex processes leading to SOC formation (including evaporation and chemical processing of POC and chemical aging of semivolatile and intermediate volatility organics). The ratio of primary organic to elemental carbon [OC/EC]p is estimated in various locations in the Eastern United States, and is then used to calculate the primary and secondary OC concentrations. To estimate the [OC/EC]p from simulated concentrations, we use both a traditional approach and the high EC edge method, in which only values with the highest EC/OC ratio are used. Both methods perform best on a daily-averaged basis, because of the variability of the [OC/EC]p ratio during the day. The SOC estimated by the EC tracer methods corresponds to the biogenic and anthropogenic SOC formed during the oxidation of volatile organic compounds. On the other hand, the estimated POC corresponds to the sum of the fresh POC, the SOC from oxidation of the evaporated POC and the intermediate volatility organic compounds, and the OC from long-distance transport. With this correspondence, the traditional EC tracer method tends to overpredict primary OC and underpredict secondary OC for the selected urban areas in the eastern United States. The high EC edge method performs better, especially in areas where the primary contribution to OC is smaller. Despite the weaknesses of models like the one used here, the conclusions about the accuracy of observation-based methods like the EC-tracer approach should be relatively robust due to the internal consistency of the data and the approach.