Improvement of inorganic aerosol component in PM2.5 by constraining aqueous-phase formation of sulfate in cloud with satellite retrievals: WRF-Chem simulations

Abstract High concentrations of PM2.5 in China have caused severe visibility degradation and health problems. However, it is still challenging to accurately predict PM2.5 and its chemical components in numerical models. In this study, we compared the inorganic aerosol components of PM2.5 (sulfate, nitrate, and ammonium (SNA)) simulated by the Weather Research and Forecasting model fully coupled with chemistry (WRF-Chem) model with in-situ data in a heavy haze-fog event during November 2018 in Nanjing, China. Comparisons show that the model underestimates sulfate concentrations by 81% and fails to reproduce the significant increase of sulfate from early morning to noon, which corresponds to the timing of fog dissipation that suggests the model underestimates the aqueous-phase formation of sulfate in clouds. In addition, the model overestimates both nitrate and ammonium concentrations by 184% and 57%, respectively. These overestimates contribute to the simulated SNA being 77.2% higher than observed. However, cloud water content is also underestimated which is a pathway for important aqueous-phase reactions. Therefore, we constrained the simulated cloud water content based on the Moderate Resolution Imaging Spectroradiometer (MODIS) Liquid Water Path observations. Results show that the simulation with MODIS-corrected cloud water content increases the sulfate by a factor of 3, decreases the Normalized Mean Bias (NMB) by 53.5%, and reproduces its diurnal cycle with the peak concentration occurring at noon. The improved sulfate simulation also improves the simulation of nitrate, which decreases the simulated nitrate bias by 134%. Although the simulated ammonium is still higher than the observations, corrected cloud water content leads to a decrease of the modelled bias in SNA from 77.2% to 14.1%. The strong sensitivity of simulated SNA concentration to the cloud water content provides an explanation for the simulated SNA bias. Hence, uncertainties in cloud water content can contribute to model biases in simulating SNA which are less frequently explored from a process-level perspective and can be reduced by constraining the model with satellite observations.