PM2.5 source apportionment using organic marker-based chemical mass balance modeling: influence of inorganic markers and sensitivity to source profiles

A Chemical Mass Balance (CMB) model has been applied to source apportionment of PM_2.5 in the Chinese megacity of Chengdu. The study explored the sensitivity of the CMB model to the adoption of different organic source profiles, and to the use of organic markers only (OM-CMB), compared with using a combination of organic and inorganic markers (IOM-CMB). A comprehensive comparison of OM-CMB and IOM-CMB shows that PM_2.5 mass concentrations from gasoline vehicles, diesel vehicles, industrial coal combustion, biomass burning, cooking, and SOA which shared same markers in the two methods are in fair to good agreement between the two methods, with the relative biases ranging from 2.2% to 17.3%. The average contributions of sulfate and nitrate sources are more sensitive to the choice of model because inorganic ions were not inputted directly into the OM-CMB. The temporal variations of PM_2.5 contributions from sulfate, nitrate, SOA, gasoline vehicles, and biomass burning, characterized by unique markers and low collinearity, were in good agreement between the OM-CMB and IOM-CMB results with the Pearson's r above 0.91 (p < 0.01). However, resuspended dust estimates from OM-CMB had a relatively weak correlation with that from IOM-CMB (Pearson's r = 0.73, p < 0.01), due to the different tracers used. When replacing the source profile for industrial coal combustion with that for residential sources, the contributions of resuspended dust and residential coal combustion were higher, and the contributions of other sources were lower compared with the result for the industrial coal combustion. Different source profiles for gasoline vehicles showed considerable sensitivity of the model to the choice of source profile, even when using data from within a single emissions study. Our results emphasize the value of combining inorganic and organic tracers in minimizing error, and in using up-to-date locally-relevant source profiles in source apportionment of PM.