A STATISTICAL CLOUD SCHEME FOR USE IN AN AGCM

We have designed a diagnostic subgrid scale cloud scheme that brings s ome coherence to the various representations of clouds within the GCM. It takes into account interactions between cloud fraction coverage, l iquid water content and turbulence (for stratiform clouds). The scheme has been developed on the basis of a statistical cloud scheme and a l evel 2 subgrid scale turbulence scheme (Yamada and Mellor) in which we introduce the effects of water vapor and liquid water. The scheme has been implemented in the T42-20 level version of the French spectral G CM used for climate studies. Simultaneously, large-scale precipitation has been modified according to Smith's scheme (with different paramet erizations for liquid and ice water). The GCM has been run with and wi thout the new parameterizations. We have studied the impact of the imp lementation of two new parameterizations on the results of the French Emeraude GCM in January. The model simulations are validated with sate llite data from the ERB experiment (total cloudiness field) and from t he ERBE experiment (visible and IR radiation fields at the top of the atmosphere),