Numerical Simulation And Optimization Improvement Of Key Physical Processes In Fog-haze Forecasting Model GRAPES?CUACE

Atmospheric chemistry models are the most important tools for fog-haze forecasting and warning.Accurate and timely numerical fog-haze forecasting is highly dependent on the reasonable description of key atmospheric physical-chemical processes by the models.Based on the GRAPES?CUACE of atmospheric-chemical online,two-way coupled model and combining multi-source observational data and weather science analysis,this paper investigated the influence of key physical processes,such as the evolution characteristics of meteorological elements and aerosols,near-surface scheme,planetary boundary layer(PBL)scheme,and PBL-aerosol-radiation interaction,on the numerical prediction of fog-haze in severe fog-haze,and deeply analyzed the relative contributions of relative humidity and PM_2.5 concentration to the low and very low visibility during the heavy fog-haze process.Based on this study,the coupling of the new near-earth layer scheme in the model and the improvement method based on the IMPROVE visibility parameterization scheme are completed.The main research findings are as follows.During the fog-haze,the significant increase in PM_2.5 concentration is often accompanied by high relative humidity(RH)and low visibility(VIS).Under low RH conditions,aerosol extinction is the main cause of VIS reduction,which shows a significant exponential decrease with increasing PM_2.5 concentration;under high RH conditions,the continued reduction of visibility depends on the atmospheric extinction caused by aerosols and their hygroscopic growth;when water vapor reaches near saturation,the direct extinction of fog droplets makes an important contribution to the further reduction of VIS.Compared with the pre-fog-haze,the scattered radiation exposure in the near-earth layer increases significantly and the total radiation exposure reaching the ground decreases significantly during the fog-haze,indicating the important influence of fog-haze on the atmospheric extinction and the radiation process in the near-earth layer.Based on the original IMPROVE equation for aerosol and its hygroscopic growth extinction calculation,the calculation of the direct extinction coefficient of fog droplets is added,so that the atmospheric extinction coefficient for VIS calculation consists of both aerosol extinction(including the effect of its hygroscopic growth)and fog droplet extinction.The simulation results show that,compared with the original VIS scheme,the new scheme reduces the overall monthly average VIS simulation error of the Beijing-Tianjin-Hebei(JJJ)region by about 30%,especially for the systematic overestimation of low VIS forecasts in heavy fog-haze pollution events,in which the simulation errors for low VIS below 5 km and 3 km are reduced by 65%and 88%,respectively.The new near-surface layer scheme Li is close to the original MM5 scheme for PM_2.5 simulations at low concentrations,and the difference between the two schemes increases with increasing PM_2.5 concentrations during the fog-haze pollution process.Under heavy pollution,the Li scheme simulates PM_2.5 concentrations closer to the observed values,indicating that the Li scheme is more advantageous in describing the ground-air exchange process during the transition from unstable to stable atmosphere.The simulation results using MRF,YSU and MYJ PBL schemes without aerosol radiative feedback(ARF)one-way model show that,overall,YSU scheme is the best,MYJ is the worst and MRF is in the middle,but the effects of the three PBL schemes on the fog-haze simulation are not very significant.The simulation results of all three PBL schemes of the two-way GRAPES?CUACE model including ARF show that,during the fog-haze,the aerosol significantly reduces the short-wave radiation reaching the surface,causing the cooling near the ground,the warming in the middle and upper levels,and the significant enhancement of atmospheric inversion,and the simulation of PM_2.5 concentration increases significantly and is closer to the observation,indicating that the online calculation of ARF plays an important role in the simulation of PM_2.5 concentration peaks in severe fog-haze.Further study shows that ARF has a significant effect on the simulation when the PM_2.5concentration is higher than a certain threshold value(about150?g m^-3 in this paper,which may vary according to different models,clouds and weather conditions).This result indicates that there are some limitations for ARF,i.e.,ARF is not effective in improving model errors when PM_2.5 concentrations are too low,which also ensures that ARF does not lead to high PM_2.5 simulations under clean or light pollution conditions while improving the low PM_2.5 simulations for heavy pollution.Although the effects of the three PBL schemes on PM_2.5 are not significant in the one-way model,the simulation results of the three PBL schemes in the two-way model show that ARF is highly dependent on the PBL schemes:MRF and YSU schemes have stronger responses to ARF than MYJ scheme in the simulations of meteorological elements,and the difference increases significantly with the increase of PM_2.5concentration.When severe pollution occurs,the PM_2.5 increment caused by ARF of MRF and YSU schemes is twice that of MYJ scheme,and the simulations are closer to the observations,indicating the complexity of aerosol-radiation-PBL interaction mechanism.