Representing aerosol-cloud interactions constrains aerosol representation and processes in the ECMWF model

The ECWMF Integrated Forecasting System (IFS) is a numerical weather prediction (NWP) model used to perform operational meteorological forecasts. It also supports a configuration with online chemistry and prognostic aerosols, which is used to produce air quality analysis and prediction within the Copernicus Atmosphere Monitoring Service (CAMS). The IFS’s single-moment cloud scheme diagnoses liquid droplet number concentrations (N_d) and effective radius, which are used by the radiation scheme to calculate cloud radiative properties and thereby how reflective the clouds are to sunlight. In the atmosphere, N_d values are strictly related to the amount of available cloud condensation nuclei (CCN), and this is known as the first indirect (or Twomey) effect of aerosols. However, the IFS currently estimates N_d values using a wind-dependent parametrization that differs over land and sea, but which lacks realistic variability in space and time. Simplified approaches such as this are common in NWP models, because reliable aerosol quantities (numbers, sizes, hygroscopicity) are usually not available in an NWP context. However, such limited realism has been shown to contribute to biases in top-of-atmosphere (TOA) shortwave (SW) fluxes, which motivates the interest towards more realistic aerosol-cloud interactions (ACI). We will present how we introduced a method to use the CAMS aerosols to diagnose optimal N_d values from bulk mass concentration fields, with the final aim of improving all-sky TOA SW fluxes for NWP applications. We will show how this procedure promotes consistency across model components, by exploiting the fact that aerosol number concentrations are heavily under-constrained by visible aerosol optical depth. Using ACI to pass aerosol information to cloud fields also opens up exciting opportunities to validate aerosol properties and processes otherwise poorly constrained at global scales by direct observations like AOD from satellites. As a case in point, we will illustrate how the beneficial impact of more realistic aerosol scavenging in mixed-phase clouds can be assessed through simulated N_d and all-sky SW fluxes. At the same time, we will show how several observation sources, including size spectra from sun photometers, can be used together for more thorough evaluation of aerosols than traditionally done, providing valuable evidence supporting future development opportunities for the aerosol forecasting system.