Simulations of aerosol-cloud-dynamical feedbacks resulting from entrainment of aerosol into the marine boundary layer during the Atlantic Stratocumulus Transition Experiment

[1] Numerical simulations of a cloudy marine boundary layer (MBL) observed during the Atlantic Stratocumulus Transition Experiment were performed to study the influence of entrainment of free tropospheric cloud condensation nuclei (CCN) on cloud microphysics, dynamics, and radiative properties. The initial CCN concentration is 100 cm 3 in one simulation, while in the second simulation it varies from 100 cm 3 below the cloud top to a peak of 1200 cm 3 at the inversion. In the clean case, cooling from evaporating drizzle destabilizes the layer just below cloud base (not the entire subcloud layer) with respect to the surface, and promotes stronger penetrating cumulus. In the case with the elevated pollution layer, reduced drizzle at the cloud base results in weaker penetrating cumulus and a less effective supply of surface moisture to the cloud. This results in a much lower liquid water path (LWP) relative to the clean case that offsets the cloud albedo enhancement due to higher drop concentrations. Thus, although entrained CCN enhance the droplet concentration, the net effect on the cloud albedo is small. Additional simulations were performed to study the sensitivity of the MBL to varying levels of largescale subsidence. The change in large-scale subsidence has a large effect on boundary layer dynamics, cloud microphysics, and the radiative budget. The simulations are used to separate the effects of enhanced albedo due to enhanced drop concentrations at constant LWP and those where LWP is modified due to dynamical feedbacks. For this case study, weaker subsidence results in a cloud with higher LWP and a cloud albedo that is enhanced over and above that due to enhanced droplet concentration. The simulations point to the complex dynamical-microphysical-radiative feedbacks in the MBL and how elevated polluted layers can change cloud radiative forcing in ways that would not be easily predicted by large-scale models.