Abstract
In this paper, the second part of a two‐part study, we use the UK Met Office's large‐eddy‐simulation model with bin‐resolved cloud microphysics (the ‘BR‐LEM’) to investigate the semi‐direct aerosol effect. In particular, we focus on the interaction between the semi‐direct effect that results from absorbing aerosol in the boundary layer and the indirect aerosol effects in non‐precipitating marine stratocumulus. We show that, irrespective of the initial concentration of cloud condensation nuclei (CCN), the presence of absorbing aerosol, i.e. black carbon (BC), in the boundary layer and the associated warming cause a decrease in liquid‐water path (LWP) and a positive radiative forcing, i.e. a positive semi‐direct forcing. Analysis of the change in the semi‐direct forcing with increasing CCN concentration shows that as CCN increases from 100 cm^−3^ to 500 cm^−3^ the semi‐direct forcing is reduced by 20%. In contrast, further increasing CCN concentration to 1000 cm^−3^ causes the semi‐direct forcing to increase by 75% relative to the value at 100 cm^−3^. We show that the semi‐direct forcing increases with increasing CCN concentration because of a coupling between the BC warming of the boundary layer and an evaporation–entrainment feedback that results from increasing CCN concentration, which further warms the boundary layer. This coupling increases the impact of BC, causing enhanced BC‐induced reduction of LWP with increasing CCN. It causes marine stratocumulus formed in high‐CCN regimes to be more sensitive to warming by absorbing aerosol, and hence produce a stronger semi‐direct forcing.
We compare the BR‐LEM results with results from simulations performed with the standard large‐eddy‐simulation model (the ‘Bulk‐LEM’), employing a simple single‐moment bulk scheme, which does not simulate the dynamical feedback. This comparison demonstrates that for a case of very high pollution, the failure of the bulk microphysics to capture an evaporation–entrainment feedback results in a semi‐direct forcing that is 40% lower than that of the clean control run. As a result, the Bulk‐LEM significantly underestimates the semi‐direct forcing estimate relative to the BR‐LEM. We conclude that, irrespective of the microphysics scheme employed, the semi‐direct aerosol effect is sensitive to the indirect aerosol effect. However, failure to simulate the evaporation–entrainment feedback associated with increasing CCN concentration results in a significant underestimation of the magnitude of the semi‐direct effect and an overestimation of the total aerosol effect. Copyright © 2008 Royal Meteorological Society