A sensitivity study of the subtropical ocean surface energy balance to the parameterization of precipitation from stratocumulus clouds

In the 'First Lagrangian' of the Atlantic Stratocumulus Experiment (ASTEX), a cloudy air mass was tracked as it was advected by the trade winds toward higher sea surface temperatures. In this study, a full diurnal cycle observed during this experiment is simulated and the impact of the precipitation parameterization is examined. The model we use is the one dimensional version of the hydrostatic primitive equation model MAR (Modele Atmospherique Regional) developed at the Universite catholique de Louvain (UCL). It includes an E-epsilon turbulence closure, a wide-band formulation of the radiative transfer, and a parameterized microphysical scheme allowing partial condensation. The model realistically reproduces the diurnal clearing of the cloud layer as well as the formation of cumulus clouds under the stratocumulus deck. Nevertheless, as the surface warms and the boundary layer becomes more convective, the simulation progressively differs from the observed evolution. Further experiments are carried out with different precipitation parameterizations commonly used in mesoscale models and general circulation models (GCMs). A strong sensitivity of the simulated liquid water path evolution is found. The impact on the surface energy flux and the solar flux reflected by the cloud is also examined. For both fluxes averaged over 24 hours, differences as large as 20 W m(-2) are obtained between the various simulations. Low cloudiness covers large areas over the ocean and such errors on the reflected solar flux may strongly affect the Earth's radiative budget in GCM simulations. We estimate that the impact on the globally averaged outgoing solar flux could be as large as 5 W m(-2). Furthermore, when atmospheric models are coupled to ocean models, errors in the surface energy exchanges may induce significant drift in the simulated climate.