The southeast Atlantic Ocean (SEA) provides an excellent natural laboratory to study smoke-cloud interactions, a large driver of uncertainty in climate projections. This region features near-permanent stratocumulus clouds and a massive persistent smoke source during the agricultural burning in the Austral winter. The physicochemical evolution of the smoke, its transport, as well as its tendency to function as CCN in warm clouds all present important targets to constrain uncertainty in predictions of aerosol’s climate impacts.

In this work, we compare observations from the ORACLES, CLARIFY, and LASIC field campaigns to study representation of these processes in two earth system models (CESM and E3SM) and a regional climate model (WRF-Chem-CAM5). Using airborne measurements of 3D-winds, clouds, and aerosols, we find that there is a strong sensitivity of modeled cloud droplet number concentration (CDNC) to marine boundary layer (MBL) turbulence, both in the mean turbulent updraft velocity and its distribution, which together drive model CDNC biases. We also find model uncertainties in the representation of smoke aging in the free-troposphere, with models not displaying the observed loss of organic aerosol (OA) and decrease in mean particle diameter, biases which manifest in MBL smoke once it entrains. Finally, models overpredict smoke concentrations in the MBL, likely associated with too-weak wet removal. Sensitivity simulations will be shown aimed at improving model representation of these processes in the multiple modeling systems, such as modifying turbulent updrafts, OA loss parameterizations, and updated cloud microphysics schemes.

These results have cascading effects on representation of cloud brightness and lifetime in different cloud regimes and the stratocumulus-to-cumulus transition in this key region. Additionally, they may help inform modeling of aerosol-cloud-radiation interactions on a climatic scale by targeting these major sources of uncertainty.