Recent updates in the aerosol model of the ECMWF IFS and their impact on skill scores

Abstract:

The Integrated Forecasting System (IFS) of ECMWF, as well as being used for numerical weather prediction, is also used within the Copernicus Atmosphere Monitoring Service (CAMS) to provide global analyses and forecasts of atmospheric composition, including aerosols as well as reactive trace gases and greenhouse gases. The aerosol model of the IFS is a sectional-bulk scheme that now forecasts seven species: dust (three size bins), sea-salt (three size bins), black carbon, organic matter, sulfate, nitrate (two bins) and ammonium. Nitrate and ammonium are recent additions, which have only been included operationally since July 2019. The developments that have recently been carried out and tested and are now contemplated for implementation in the next operational version (known as cycle 48r1) are presented here.

The dry deposition velocities are computed as a function of roughness length, particle size and surface friction velocity, while wet deposition depends mainly on the difference between precipitation fluxes (liquid or solid). The parameterizations of both dry and wet deposition have been upgraded with more recent schemes, which have been shown to improve the simulated deposition fluxes for several aerosol species. The impact of this upgrade on the skill scores of simulated aerosol optical depth (AOD) and surface particulate matter (PM_2.5 and PM₁₀) concentrations against a range of observations is very positive.

The simulated surface concentration of nitrate and ammonium are frequently strongly overestimated over Europe and the United States in the current version. Nitrate, ammonium, and their precursors nitric acid and ammonia, were evaluated against field campaign data and it was found that the recently-implemented gas-particle partitioning scheme is too efficient in producing nitrate and ammonium particles. A series of small-scale changes, such as adjusting nitrate dry deposition velocity, direct particulate sulphate emission, and limiting nitrate/ammonium production by the concentration of mineral cations, have been implemented and shown to be effective in improving the simulated surface concentration of nitrate and ammonium.

The representation of secondary organic aerosol (SOA) has been overhauled with the introduction of a new SOA species, separate from primary organic matter, with anthropogenic and biogenic bins each with associated precursor gases. The partitioning between gaseous and particulate species is modelled using a simple Volatility Basis Set (VBS) approach. The implementation of this new species leads to a significant improvement of the simulated surface concentration of organic carbon. An evaluation of simulated SOA concentrations at the surface against climatological values derived from observations using Positive Matrix Factorisation (PMF) techniques also shows a reasonable agreement.