Assimilation Of EnSRF Aerosol Concentration And Lidar Extinction Coefficient Based On WRF-Chem Mode

An Ensemble Square Root Filter(EnSRF)data assimilation system for the aerosol concentration and extinction coefficient observation data in the WRF-Chem model with MOSAIC(Model for Simulating Aerosol Interactions and Chemistry)4bin aerosol parameterization scheme is established.The aerosol extinction coefficient is calculated by the IMPROVE(The Interagency Monitoring of Protected Visual Environments)equation,realizing the direct assimilation of the extinction coefficient.The ideal experiments of aerosol concentration and aerosol extinction coefficient were carried out to analyze the characteristics of assimilation increment.In the horizontal and vertical directions,the increment of assimilation spread and decreases gradually from the observation position,which shows the anisotropic characteristics of flow dependence.The experiment of covariance localization shows that horizontal and vertical length scales have a great influence on assimilation.In the middle and upper troposphere,a larger horizontal length scale is required for a better assimilation effect.Three groups of assimilation experiments with different observations(surface aerosol concentration(DA?aerosol),lidar extinction coefficient(DA?bet)and both of these observations(DA?all))were performed in the Beijing-Tianjin-Hebei region from 2019 February 28 to March 2.The results show that three groups of assimilation can effectively improve the surface PM2.5 distribution of the initial fields.The PM2.5 concentration of the initial fields are closer to the observation than that of the control experiment.After assimilation,the correlation coefficient(CORR)increases by at least 142.3%,the bias(BIAS)increases by at least 12.7%,and the Root Mean Square Error(RMSE)decreases by at least 22.1%.For the surface PM2.5 concentration forecast field,the improvement of the Da?all experiment is the most significant.Within 24 hours,the CORR of the DA?all experiment meanly increases by121.5%,the BIAS meanly decreases by 36.4%,and the RMSE meanly decreases by 27.4%.Compared with the vertical profile of the lidar extinction coefficient observation data,the aerosol extinction coefficients of the three assimilation experiments were improved in the Nanjiao and Haidian stations.The improvements of DA?bet and DA?all experiments are significant with a reduction of at least 63.8% in BIAS and 47.6% in RMSE.Besides,all three groups of assimilation experiments have improved the accuracy of the extinction coefficient prediction and the improvement of the DA?all experiment is the most significant.Within 24 hours,the average improvement of CORR,BIAS and RMSE of the DA?all experiment is 14.1%,5.5%,and 15.0%,respectively.