The impact of AOD assimilation on surface PM analysis and forecast with the global CAMS system.

Surface air pollution is a major hazard for society because of its negative impact on human health, crops yield and on other aspects of the economy. Monitoring the air quality with in-situ instruments is routinely carried out in many countries. These air quality networks have limited coverage. They  are often very sparse or are not even present in many parts of the world, that suffer from the worst air quality.  Satellite observations of atmospheric composition provide the unique prospect to contribute to a more spatially complete monitoring of surface air pollution. However, among several other limitations, only vertically integrated measures of the pollution concentration can usually be retrieved from satellites, and not the surface concentration.  The relation between total column information and surface concentrations depends on the shape of the vertical profile. The correlation may be especially poor in the case of lofted air pollution plumes originating from long range transport.   

Data driven methods (machine learning, statistic etc.) are used to infer surface concentrations from satellite observations based on in-situ surface observations. An alternative approach is data assimilation of satellite retrievals in an atmospheric model, a method which has been developed for numerical weather forecasting.  The global forecasting system of atmospheric composition of the Copernicus Atmosphere Monitoring Service (CAMS) applies 4D-VAR data assimilation to satellite retrievals of Aerosol optical depth (AOD), Ozone, Carbon Monoxide and Nitrogen Dioxide (NO2) to correct the model initial conditions and consequently also the PM2.5 and PM10 surface concentrations.  In our presentation, we will give an overview of the usefulness of atmospheric composition (AC) satellite data assimilation for monitoring of surface air quality with the global CAMS system. We will specifically show to what extent AOD assimilation improves the analysis and forecast of surface PM2.5 for different regions and at different temporal and spatial scales. In particular we will discuss the correlation between AOD and PM2.5 in the observations and in the model. We will also discuss the influence of the model performance and aspects of the data assimilation procedure on the impact of AC satellite data assimilation on surface concentrations.