Numerical Simulation of Stratiform Clouds Aerosol and Radiative Properties

Results from a numerical simulation of a stratocumulus cloud-topped boundary layer (CTBL) based on a 3-D LES model that includes explicit formulation of cloud microphysics are presented. The results reveal such interesting features of CTBL evolution as decoupling of the stratiform cloud layer and formation of bimodal spectra due to the processes of droplet evaporation and reactivation. The effect of CCN regeneration on cloud properties has been studied in a set of 4 experiments. It is shown that the process of CCN regeneration is quite important in formation and evolution of cloud microphysical and radiative properties, as well as in the long time evolution of cloud; dynamical structure. On the other hand, the parameters of stratocumulus cloud are rather insensitive to the exact form of CCN spectrum regenerated after droplet evaporation.

We have also evaluated the effect of spatial inhomogeneities in cloud macro and microstructure on the performance of commonly used in large-scale models parameterizations of optical depth and transmittance of direct solar radiation. It is shown that .an accurate parameterization of the grid average optical depth alone is not sufficient for correct detemination of cloud transmittance to solar radiation due to the nonlinear dependence between these two variables.

The problem can be solved by introducing the "equivalent" value of optical depth that differs from the ordinarily defined mean optical depth by a factor a_t, that depends on the degree of cloud inhomogeneity and ranges from about 2 in the cumulus case to about 1.3 in the stratiform case.

Finally, the accuracy of cloud radiative parameterizations commonly employed in large- scale models has been evaluated using the data from the explicit microphysical model as a benchmark for comparison. It has been shown that in the cumulus cloud case the parameterized expressions can error by as much as 100%. The error is smaller for more uniform stratiform clouds, where for some parameterizations varied in the 10-40% range. The best results are given by parameterizations that account for vertical stratification of parameters on which they are based. However. the error given by a particular parameterization varies and is different at cloud and surface Ievels.