ROLE OF A CUMULUS PARAMETERIZATION SCHEME IN SIMULATING ATMOSPHERIC CIRCULATION AND RAINFALL IN THE 9-LAYER GODDARD LABORATORY FOR ATMOSPHERES GENERAL-CIRCULATION MODEL

A coarse (4-degrees x 5-degrees x 9-sigma level) version of the Goddar d Laboratory for Atmospheres (GLA) General Circulation Model (GCM) was used to investigate the influence of a cumulus convection scheme on t he simulated atmospheric circulation and hydrologic cycle. Two sets of integrations, each containing an ensemble of three summer (June, July , and August) simulations, were produced. The first set, containing co ntrol cases, included a state-of-the-art cumulus parameterization sche me in the GCM; whereas the second set, containing experiment cases, us ed the same GCM but without the cumulus parameterization. All simulati ons started from initial conditions that were taken from analysis of o bservations for three consecutive initial times that were only 12 h ap art, beginning with 0000 UTC 19 May 1988. The climatological boundary conditions-sea surface temperature, snow, ice, and vegetation cover-we re kept exactly the same for all the integrations. The ensemble sets o f control and experiment simulations are compared and differentially a nalyzed to determine the influence of a cumulus convection scheme on t he simulated circulation and hydrologic cycle. The results show that c umulus parameterization has a very significant influence on the simula ted circulation and precipitation. The influence is conspicuous in tro pical regions, interior of continents in the Northern Hemisphere, and some oceanic regions. The upper-level condensation heating over the in tertropical convergence zone (ITCZ) is much smaller for the experiment simulations as compared to the control simulations; correspondingly, the Hadley and Walker cells for the control simulations are also weake r and are accompanied by a weaker Ferrel cell in the Southern Hemisphe re. The rainfall under the rising branch of the southern Ferrel cell ( at about 50-degrees-S) does not increase very much because boundary-la yer convergence poleward reduces the local evaporation. Overall, the d ifference fields show that experiment simulations (without cumulus con vection) produce a cooler and less energetic atmosphere. The vertical profile of the zonally averaged diabatic heating also shows large diff erences in the tropics that are physically consistent with accompanyin g differences in circulation. Despite producing a warmer and wetter pl anetary boundary layer (PBL) in the tropics (20-degrees-S-20-degrees-N ), the control simulations also produce a warmer but drier 400-mb leve l. The moisture transport convergence fields show that while only the stationary circulation is affected significantly in the PBL, both the stationary and eddy moisture transports are altered significantly in t he atmosphere above the PBL. These differences not only reaffirm the i mportant role of cumulus convection in maintaining the global circulat ion, but also show the way in which the presence or absence of a cumul us parameterization scheme can affect the circulation and rainfall cli matology of a GCM.