Sensitivity of a Global Climate Model to the Specification of Convective Updraft and Downdraft Mass Fluxes

Abstract

We examine the response of the GISS global climate model to different parameterizations of moist convective man flux. A control run with arbitrarily specified updraft mass flux is compared to experiments that predict cumulus mass flux on the basis of low-level convergence, convergence plus surface evaporation, or convergence and evaporation modified by varying boundary layer height. An experiment that includes a simple parameterization of saturated convective-scale downdrafts is also described. Convergence effects on cumulus mass flux significantly improve the model's January climatology by increasing the frequency of occurrence of deep convection in the tropics and decreasing it at high latitudes, shifting the ITCZ from 12°N to 4°5, strengthening convective heating in the western Pacific, and increasing tropical long-wave eddy kinetic energy. Surface evaporation effects generally oppose the effects of convergence but are necessary to produce realistic continental convective heating and well-defined marine shallow cumulus regions. Varying boundary layer height (as prescribed by variations in lifting condensation level) has little effect on the model climatology. Downdrafts, however, reinforce many of the positive effects of convergence while also improving the model's vertical humidity profile and radiation balance. The diurnal cycle of precipitation over the West Pacific is best simulated when convergence determines cumulus mass flux, while surface flux effects are needed to reproduce diurnal variations in the continental ITCZ. In each experiment the model correctly simulates the observed correlation between deep convection strength and tropical sea surface temperature; the parameterization of cumulus mass flux has little effect on this relationship. The experiments have several implications for cloud effects on climate sensitivity. The dependence of cumulus mass flux on vertical motions, and the insensitivity of mean vertical motions to changes in forcing, suggests that the convective response to climate forcing may be weaker than that estimated in previous global climate model simulations that link convection only to moist static instability. This implies that changes in cloud cover and hence positive cloud feedback have been overestimated in these climate change experiments. Downdrafts may affect the feedback in the same sense by replenishing boundary layer moisture relative to cumulus parameterization schemes with only dry compensating subsidence.