Improvements Of Soil NO_x Emissions And Aerosol Chemical Mechanisms And Their Impacts On Air Quality Based On WRF-Chem

Air quality models are often used to investigate the causes of air pollution,to simulate and predict air pollution.Due to the limitations in the understanding of air pollutant emissions,physical processes and chemical formation mechanisms,most air quality models,such as WRF-Chem,WRF-CAMQ,have large uncertainties in applications.The default version of WRF-Chem model generally underestimates soil NO_x emissions（SNO_x）,and poorly predicts the simulated chemical composition of PM_2.5（sulfate,nitrate,and ammonium,i.e.,SNA）,especially underestimates sulfate and overestimates nitrate.Here we improve the model performance by adding new SNO_xscheme and chemical mechanisms to the default version,as well as study the impact of SNO_x and different chemical mechanisms on simulating air quality.The main conclusions as follows:（1）Compared to the default SNO_x scheme in WRF-Chem,the new scheme Berkeley Dalhousie Soil NO Parameterization（BDSNP）can better represent the impact of different land cover types,soil temperature and moisture,and emission pulses on SNO_x,as well as includes the fertilizer N emissions and nitrogen deposition.The model with BDISNP shows a better agreement with TROPOMI NO₂ column densities（columns）,reproduces the observed pulsed emissions,and also improves the performance on simulating SO₂,NO₂ and PM_2.5 on the surface in both California and East China.（2）By using the WRF-Chem model with BDISNP scheme,we quantified the contribution of SNO_x to NO_x budget.For California,there are 40%of the state’s total NO_x emissions in summer are from soils,and SNO_x could exceed anthropogenic sources over croplands which accounts for 51%of NO_x emissions.Such considerable SNO_x enhance the monthly mean NO₂columns by 35%（53%）and surface NO₂ concentrations by 177%（114%）,leading to an additional 23.0%（23.2%）of surface O₃ concentrations in California（cropland）.For East China（cropland）,SNO_x contribute to 34%（39%）of the total NO_x budget in summer.Such amounts of SNO_x increase the monthly mean NO₂ columns by 33%（43%）and surface NO₂concentrations by 68%（69%）,resulting in surface O₃ concentrations in daytime（13:00-16:00）increased by 14.6%（14.8%）in East China（cropland）.Although the impact of SNO_x on O₃concentration in East China is less than that of California,high SNO_x could aggravate aerosol pollution in this region,leading to nitrate and PM_2.5 concentrations increased by 27%-33%and11%-16%,respectively.（3）WRF-Chem model can reasonably reproduce the observed total PM_2.5 and generally capture the haze episodes in Nanjing,but significantly underestimate sulfate concentrations and overestimate SO₂ concentrations,which is mainly caused by the much lower SO₂ oxidation rate（SOR）in the model.By conducting sensitivity studies,we explored the impacts of different chemical mechanisms on simulating SNA.The results show that tripling the gas-phase oxidation rate of SO₂ by OH does not significantly increase sulfate concentrations and show a flatten diurnal variation,indicating gas-phase oxidation may not be the main causes for the underestimations in the model.Compared to the reference run,inclusion of SO₂ heterogeneous reaction in aerosol water can obviously increase sulfate by nearly 1-2 times and better reproduce the observed diurnal variations of sulfate.It should note that the simulated sulfate is still 50%lower than the observations,though inclusion of heterogeneous reaction can substantially improve the simulation performance of SNA.（4）Aqueous-phase chemistry is one of important pathways of sulfate production,the underestimation of liquid water content leads to the insufficient contribution of in-cloud aqueous-phase chemistry to sulfate formation.By evaluating the model performance on simulating SNA of PM_2.5 in a heavy haze-fog event occurred in the Yangtze River Delta（YRD）,we found that the model underestimates sulfate concentrations and fails to reproduce the peak concentrations at noon,which corresponds to the timing of fog dissipation.Furthermore,liquid water content is significantly underestimated.Therefore,we constrained the simulated liquid water content in a sensitivity simulation based on MODIS Liquid Water Path（LWP）observations.Compared to the reference run,the simulation with MODIS-corrected liquid water content increases the sulfate by a factor of 3 and reproduces its peak concentrations occurring at noon.The improved sulfate simulation also reduces the modelled bias in nitrate and shows a consistent diurnal pattern with observations.Additionally,corrected liquid water content leads to a decrease of the modelled bias in SNA from 77%to 14%.