Aerosol processes and activation in OpenIFS cycle 48r1 portable global aerosol-climate and weather prediction model

Aerosol-cloud interactions are complex processes that contribute significantly to uncertainties in predicting global climate. Cloud droplet formation is influenced by factors such as the hygroscopicity of soluble aerosol particles, aerosol processing within clouds, aerosol particle number concentration, and updraft velocity. Accurately understanding these factors is crucial for better representing cloud droplet numbers on a global scale. Updraft velocity is characterized by the probability density function (PDF) of the vertical velocity distribution. Earth system models (ESMs) typically assume a constant standard deviation (σ_w) for the vertical velocity PDF (van Noije et al. 2021; Mulcahy et al. 2020). However, this assumption may be inaccurate, as it has been shown that the σ_w exhibits significant diurnal variability (Bougiatioti et al. 2020). In this study, we implemented the turbulent kinetic energy (TKE) calculations to OpenIFS cycle 48r1 (OIFS48r1) following Bastak Duran et al. (2018). In addition, we implemented Morales and Nenes (2014) activation parameterization (M&N) to OIFS48r1 including the TKE to calculate the σ_w instead of using a constant σ_w for the vertical velocity PDF. OIFS48r1 is derived from the Integrated Forecasting System (IFS) developed by European Centre for Medium-Range Weather Forecasts (ECMWF) and it will be the main atmospheric model used in the newest version of European Community ESM (EC-Earth 4). First, we investigated the effects of using constant σ_w values of 0 and 0.8 m/s on the TKE-derived σ_w and their impact on column averaged cloud droplet number concentrations. The results showed that using σ_w of 0 m/s led to very low droplet numbers, as weaker updraft variability resulted in smaller supersaturations, causing fewer particles to activate. On the other hand, using σ_w of 0.8 m/s produced stronger activation, particularly in regions with high aerosol particle number concentrations. When using the TKE-derived σ_w, the differences in droplet numbers compared to the activation scheme with σ_w of 0.8 m/s were minimal. Next, we compared the monthly column-averaged and boundary-layer (BL) average number of activated particles obtained from the TKE-derived σ_w activation routine with the pre-existing Abdul-Razzak and Ghan (2000) activation parameterization (AR&G). The results showed that the AR&G scheme produced stronger activation than the M&N scheme, both in terms of the total column average and the BL-averaged regions. One of the reasons for this difference could be due to the activation calculations in the M&N scheme, which are only performed in areas where clouds are present, while in the AR&G scheme, activation is calculated for every gridbox.