Implementing the first indirect radiative effect of aerosols in the ECMWF model

For many years, aerosol-cloud-radiation interactions (ACI) have been incorporated in climate models, due to the high sensitivity of the Earth’s climate to changes in cloud microphysical properties. Despite this, such representations are still fraught with uncertainty, strongly limiting the capacity of such models to precisely predict climate change. For numerical weather prediction (NWP), the impact of ACI on forecast skill is controversial, though regional radiative fluxes can be affected significantly. Their representation in NWP models has therefore usually been neglected. We will present early results from an ongoing investigation to calculate the number of droplets (Nd) in liquid-phase clouds online from global aerosol fields in the ECMWF global Integrated Forecasting System (IFS), i.e. the first indirect radiative effect of aerosols. Our ACI scheme uses a lookup table, produced from offline cloud parcel model simulations, to estimate the number of aerosol particles activated into cloud droplets. This approach allows the effect of large sea salt to suppress the activation of sulfate aerosols to be included. We will then constrain the representation of ACI by verifying weather forecast against available satellite and station-based observations. Ultimately, the impact of such simulated processes on the model can be assessed in different configurations of the IFS, spanning from the NWP medium range (10-15 days) to seasonal forecasts to climate projections. We will show how such a computationally inexpensive approach can effectively increase realism of the simulated liquid-phase cloud microphysics in the model, resulting in improved representation of global and regional top-of-the-atmosphere radiative fluxes. We can apply the same approach both to the climatological and prognostic aerosol representations supported by the IFS (the latter being used operationally for air quality forecasts by the Copernicus Atmospheric Monitoring Service, CAMS), allowing sensitive comparison between different configurations of the model. We will conclude by showing advantages of the new ACI scheme when aerosols are prognostic variables, and how such experiments could inform our work towards introducing ACI into the operational IFS forecasts, where currently a climatological representation of aerosols is used.