Aerosol–cloud interactions studied with the chemistry-climate model EMAC

Abstract. This study uses the EMAC atmospheric chemistry-climate model to simulate cloud properties and estimate cloud radiative effects induced by aerosols. We have tested two prognostic cloud droplet nucleation parameterizations, i.e., the standard STN (osmotic coefficient model) and hybrid (HYB, replacing the osmotic coefficient by the κ hygroscopicity parameter) schemes to calculate aerosol hygroscopicity and critical supersaturation, and consider aerosol–cloud feedbacks with a focus on warm clouds. Both prognostic schemes (STN and HYB) account for aerosol number, size and composition effects on droplet nucleation, and are tested in combination with two different cloud cover parameterizations, i.e., a relative humidity threshold and a statistical cloud cover scheme (RH-CLC and ST-CLC).

The use of either STN and HYB leads to very different cloud radiative effects, particularly over the continents. The STN scheme predicts highly effective CCN activation in warm clouds and hazes/fogs near the surface. The enhanced CCN activity increases the cloud albedo effect of aerosols and cools the Earth's surface. The cooler surface enhances the hydrostatic stability of the lower continental troposphere and thereby reduces convection and convective precipitation. In contrast, the HYB simulations calculate lower, more realistic CCN activation and consequent cloud albedo effect, leading to relatively stronger convection and high cloud formation. The enhanced high clouds increase greenhouse warming and moderate the cooling effect of the low clouds. With respect to the cloud radiative effects, the statistical ST-CLC scheme shows much higher sensitivity to aerosol–cloud coupling for all continental regions than the RH-CLC threshold scheme, most pronounced for low clouds but also for high clouds. Simulations of the short wave cloud radiative effect at the top of the atmosphere in ST-CLC are a factor of 2–8 more sensitive to aerosol coupling than the RH-CLC configurations. The long wave cloud radiative effect responds about a factor of 2 more sensitively.

Our results show that the coupling with the HYB scheme (κ approach) outperforms the coupling with STN (osmotic coefficient), and also provides a more straightforward approach to account for physicochemical effects on aerosol activation into cloud droplets. Accordingly, the sensitivity of CCN activation to chemical composition is highest in HYB. Overall, the prognostic schemes of cloud cover and cloud droplet formation help improve the agreement between model results and observations, and for the ST-CLC scheme it seems to be a necessity.