PM_2.5 Assimilation Forecast And Its System Development Based On Satellite-ground Aerosol Data

At present,the problem of haze pollution in our country is still severe,especially for compound pollution in the central and eastern regions,and there is still a lack of mechanism research on it.Atmospheric chemical model is an essential means for atmospheric aerosol business forecasting and scientific research.Assimilation technology combines observation and model.It improves the simulation and forecast of the model by adjusting the initial field of atmospheric chemical model,which has broad application prospects.This paper firstly introduces the principle and structure of the WRF-Chem chemical transport model and discusses in detail the aerosol parameterization scheme involved in the construction of assimilation systems.Because of the public version of the GSI 3D variational assimilation system,it can only achieve assimilation for the GOCART aerosol solution,and the problem of poor applicability in the prediction of composite smog pollution.Here,the observation operator for the MOSAIC aerosol solution is targeted.Establish a method,combining the calculation principle of 3D variational assimilation,the construction method of background error covariance,and the pre-processing process of observation data.In this paper,the three-dimensional variational assimilation of near-surface PM_2.5 concentration observation data and satellite AOD data for the MOSAIC aerosol scheme is implemented.Then,assimilation experiments were conducted for the pollution process in central and eastern China in November 2018.The results show that when the PM_2.5concentration data near the ground is assimilated separately,the simulation of the PM_2.5analysis field is significantly improved,and the root mean square error of the analysis field is reduce from 57.11?g/m³ to 27.34?g/m³.Correspondingly,when the MODIS AOD data is individually assimilated,the simulation effect of the AOD analysis field is improved,and the root mean square error is reduced from 0.57 to 0.39.When assimilating PM_2.5 and AOD data at the same time,the root mean square error of the PM_2.5 analysis field is 24.58?g/m³,the root mean square error of the AOD analysis field is 0.42,and the total relative error of the mode analysis field is the smallest,which is 71%.At this time,the near-surface concentration information and the entire layer of optical information work together during the assimilation process to obtain the most reasonable analysis field.AOD assimilation will cause the accuracy of the near-ground PM_2.5 analysis field to decrease(root mean square error is 80.5?g/m³),but the areas with a significant decrease in accuracy are mainly areas not covered by AOD data.At the same time,the assimilation of AOD data in the coastal waters can improve the analysis field accuracy of PM_2.5 in the coastal areas(the root mean square error is reduced from 62.66?g/m³ to 45.20?g/m³).The assimilation of AOD data can bring aerosol optical information and whole-layer observation information to the analysis field,make up for the lack of observations in scarce areas on the ground station,and bring observation data over the ocean,which makes up for the data lack of PM_2.5 ground concentration in assimilation.The forecast test shows that the timeliness of the assimilation improvement forecast is about 48h,and the first 24h of the forecast test has the most significant assimilation improvement effect.In the forecast of severe pollution(q>150?g/m³),the simultaneous assimilation can obtain a better overall forecast effect.Based on the above analysis,this article builds an aerosol parameter assimilation commercialization system and conducts a commercial operation test.The system adds ground PM_2.5 concentration observation data and DPC AOD data every six hours and uses the analysis field after assimilation as the initial field to restart model forecast.The results show that the system can realize the automatic and stable operation of the assimilation forecasting process,and the spatial and temporal distribution of the output PM_2.5 and AOD analysis field is intuitive and reasonable.The original forecast without data assimilation has a systematic overestimation of PM_2.5 concentration.In contrast,the assimilation forecast can suppress the systemic overestimation and reduce the average error of the forecast by more than 60%.The assimilation forecast of this system has the most apparent improvement in heavy pollution.The operational effect of the system reflects the positive effects of assimilation on haze forecasting of atmospheric chemical models and the application prospect of new observation techniques in numerical forecasting.