Key points are not available for this paper at this time.
An important frontier in improving numerical weather prediction (NWP) is the introduction of radiatively interactive prognostic composition species. The CAMS (Copernicus Atmospheric Monitoring Service) configuration of the ECMWF model includes 123 chemical species and 16 aerosols as prognostic variables, but incurs a computational cost of around a factor of 10 compared to the standard NWP configuration with an aerosol climatology. In this talk we describe our work to test different approaches to capture the benefits of prognostic aerosols, but at much reduced computational cost. We first focus on our new hybrid aerosol configuration, in which selected species are made prognostic and the remainder are treated by a climatology. With seven prognostic aerosols (three sizes of mineral dust, and hydrophilic/hydrophobic black carbon and organic matter), the additional computational cost is reduced from around a factor of 10 to around a factor of 1.25. Evaluation of forecasts at tropical, sub-tropical and Mediterranean AERONET stations reveals that this combination of species is able to capture the daily variation in aerosol optical depth (AOD) almost as well as the full CAMS configuration at most sites. Global simulations have been performed at resolutions from 4 to 80 km, both with and without parameterized deep convection. While the atmospheric aerosol burden varies little with resolution, higher resolutions show much greater fidelity in capturing the spatial detail of aerosol plumes near isolated sources such as forest fires. A preliminary comparison will be presented between the hybrid aerosol scheme and an alternative dual configuration approach to reducing the cost of prognostic aerosols, which involves running the full CAMS chemistry scheme but at a much reduced resolution to the meteorological variables, with interpolation between resolutions whenever required. We have also used the full CAMS chemistry scheme to develop a new aerosol climatology that incorporates the long-term trends in anthropogenic aerosol load back to 1950; this is planned to be used in the next ECMWF reanalysis, ERA6. Finally, the impact of the various aerosol configurations on medium-range forecast scores will be presented.
Checa‐Garcia et al. (Mon,) studied this question.