Rethinking the causes of extreme heavy winter PM2.5 pollution events in northern China.

It has been reported that air quality models largely underestimate PM2.5 concentrations during severe pollution events in China. In this study, the Models-3 Community Multi-scale Air Quality model (CMAQ) was employed to simulate PM2.5 concentrations in May-June (non-heating period) and in November-December (heating period) of 2013 in northern China, with a particular focus on determining the causes of the underestimation. Modeling results reproduced the mass concentrations of PM2.5 in approximately 50% of the non-heating and heating periods in Qingdao (referred to as the good periods), while the model performance was unsatisfactory during the remaining periods (the poor periods). In this respect, the overprediction of inorganic salts and the underprediction of organic matter in PM2.5 canceled each other out and resulted in a good simulation of PM2.5 concentrations during the good periods, whereas during poor periods, the bias of the planetary boundary layer height, wind direction, precipitation, and other factors caused inconsistencies between the simulated and observed PM2.5 concentrations. Sensitivity studies showed that the underestimation of primarily emitted particles from local emissions was likely the main cause of PM2.5 underpredictions during heavy haze days. Furthermore, our results implied that the assumption of the conditions of the gas-aerosol thermodynamic equilibria in the air quality model likely results in an overprediction of secondary PM2.5 inorganic salts (SO42- + NO3- + NH4+) during clear days. In contrast, during heavy pollution or heavy haze days, high concentrations of air pollutants theoretically rapidly leads to gas/particle chemical equilibrium and no overprediction of SO42-, NO3-, and NH4+ concentrations. Nevertheless, the underestimation of primarily emitted particles from local sources during heavy haze days is yet to be explained and needs further investigation.