Study On Three-dimensional Variational Assimilation Of Aerosol Optical Depth Based On Satellite Data

The structure and state of the atmospheric dynamics, thermodynamics provided by meteorological data is the indispensable condition for Numerical Weather Prediction. Compared with the conventional data, the satellite data is more consistent, covering wider area and having finer spatio-temporal resolution,Therefore,it is a meaningful topic that using satellite data effectively in mesoscale numerical model.Atmospheric aerosol is the important factors to air pollution and climate change,To promote the application of atmospheric aerosols satellite product in chemical composition simulation and atmospheric environment forecast. This paper focuses on realizing the directly 3D-Var assimilation of total aerosol optical depth (AOD) retrieval products using a 3D-Var method,together with the WRF-Chem model and a developed AOD observation operator. Three parallel experiments were designed to evaluate the impate of AOD DA on analyses and forecasts of aerosols.The main research results and conclusions of this paper are shown as follows:(1) It can simulate the regional background error convariance for China by using Ameriacn NMC method.And the B matrix can reasonably reflect the vertical characteristics of aerosol variable:Standard deviations varied substantially across the aerosol species and atmosphere altitude;Each variable is strongly local in the boundary layer,the influence of observations on analysis mainly concentrated in the local point around the observation location in the atmosphere of 1-2km.(2) A assimilation system for AOD has been constructed:an observation operator for AOD has been created based on CRTM.The statistical B martix is adoped in the assimilation system,which uses individual aerosol model variables as "control variables".Directly analyse on the 3-D mass concentration of 14 aerosol variables.(3) Assimilation experiments of both MODIS AOD and FY-3A/MERSI AOD data have shown:Assimilating AOD substantially adjusted aerosol background field when compared to satellite observations,and the results of assimilating FY-3A AOD has better agreement with the observations.Assimilation improved the extinction coefficientt below 500hpa,especially adjusted the most below 750hpa(2km),the effect largely depends on the location of observations that can produce a positive effect.(4) The assimilation system has made a dramatical change in background field from satellite observations, and improved the model initial field, produced a positive effect on the Numerical Weather Prediction.When compared with the control experiment,we found that after assimilation:the distribution of AOD from model is consisitent with MODIS AOD; The 24 hours prediction of the model can reflect the variation tendency of AOD with time,forcast is more exact when compared with ground-based AERONET observations.The case of FY-3A AOD data assimilation resulted a higher accuracy then MODIS in the west and north of China.