Tropical Sensitivity of a Coupled Model to Specified ISCCP Low Clouds

Abstract

The seasonal cycle of SST observed in the eastern equatorial Pacific is poorly simulated by many ocean?atmosphere coupled GCMs. This deficiency may be partly due to an incorrect prediction of tropical marine stratocumulus (MSc). To explore this hypothesis, two basic multiyear simulations have been performed using a coupled GCM with seasonally varying solar radiation. The model?s cloud prediction scheme, which underpredicts tropical marine stratocumulus, is used for all clouds in the control run. In contrast, in the ?ISCCP? run, the climatological monthly mean low cloud fraction is specified over the open ocean, utilizing C2 data from the International Satellite Cloud Climatology Project (ISCCP). In this manner, the treatment of MSc clouds, including the annual cycle, is more realistic than in previous sensitivity studies. Robust surface and subsurface thermodynamical and dynamical responses to the specified MSc are found in the Tropics, especially near the equator. In the annual mean, the equatorial cold tongue extends farther west and intensifies, while the east?west SST gradient is enhanced. A double SST maximum flanking the cold tongue becomes asymmetric about the equator. The SST annual cycle in the eastern equatorial Pacific strengthens, and the equatorial SST seasonal anomalies migrate farther westward. MSc-induced local shortwave radiative cooling enhances dynamical cooling associated with the southeast trades. The surface meridional wind stress in the extreme eastern equatorial Pacific remains southerly all year, while the surface zonal wind stress and equatorial upwelling intensify, as does the seasonal cycle of evaporation, in better agreement with observation. Within the ocean, the thermocline steepens and the Equatorial Undercurrent intensifies. When the low clouds are entirely removed, the SST warms by about 5.5 K in the western and central tropical Pacific, relative to ?ISCCP,? and the model?s SST bias there reverses sign. ENSO-like interannual variability with a characteristic timescale of 3?5 yr is found in all simulations, though its amplitude varies. The ?ISCCP? equatorial cold tongue inhibits the eastward progression of ENSO-like warm events east of the date line. When the specified low cloud fraction in ?ISCCP? is reduced by 20%, the interannual variability amplifies somewhat and the coupled model responds more like a delayed oscillator. The apparent sensitivity in the equatorial Pacific to a 20% relative change in low cloud fraction may have some cautionary implications for seasonal prediction by coupled GCMs.