Author: delobbe L. Gallée H.
Publisher: Springer Publishing Company
ISSN: 0006-8314
Source: Boundary-Layer Meteorology, Vol.89, Iss.1, 1998-10, pp. : 75-107
Disclaimer: Any content in publications that violate the sovereignty, the constitution or regulations of the PRC is not accepted or approved by CNPIEC.
Abstract
Marine stratocumulus observations show a large variability in cloud droplet number concentration (CDNC) related to variability in aerosol concentration. Changes in CDNC modify the cloud reflectivity, but also affect cloud water content, cloud lifetime, and cloudiness, through changes in precipitation. In mesoscale models and general circulation models (GCMs), precipitation mechanisms are parameterized. Here we examine how the precipitation parameterization can affect the simulated cloud. Simulations are carried out with the one-dimensional version of the hydrostatic primitive equation model MAR (Modéle Atmosphérique Régional) developed at the Université catholique de Louvain. It includes a E- turbulence closure, a wide-band formulation of the radiative transfer, and a parameterized microphysics including prognostic equations for water vapour, cloud droplets and rain drops concentrations. In a first step, the model is used to simulate a horizontally homogeneous stratocumulus deck observed during the Atlantic Stratocumulus Transition Experiment (ASTEX) on the night of 12–13 June 1992. The observations show that the model is able to realistically reproduce the vertical structure of the cloud-topped boundary layer. In a second step, several precipitation parameterizations commonly used in mesoscale models and GCMs are tested. It is found that most parameterizations tend to overestimate the precipitation, which results in an underestimation of the vertically integrated liquid water content. Afterwards, using those parameterizations that are sensitive to CDNC, several simulations are performed to estimate the effect of CDNC variations on the simulated cloud. Based upon the simulation results, we argue that currently used parameterizations do not enable assessment of such a sensitivity.