Abstract

The difficult problem of parameterizing tropical convection in large‐scale models of the atmosphere led to the Global Atmospheric Research Program’s Atlantic Tropical Experiment (GATE), whose goal was to improve basic understanding of tropical convection and its role in the global atmospheric circulation. A dense network of instrumented ships equipped with upper air sounding equipment and quantitative weather radars were located over the Atlantic Ocean, in the intertropical convergence zone (ITCZ), just west of equatorial Africa. The ship network was supplemented by a fleet of research aircraft and a geosynchronous meteorological satellite. The data obtained show that the deep convection in the ITCZ was concentrated in two types of ‘cloud clusters,’ rapidly moving squall clusters, and slowly moving nonsquall clusters. The clusters were characterized by large mid‐to‐upper level cloud shields, or ‘anvil clouds,’ that emanated from penetrative cumulonimbus convection. Accompanying the deep cumulonimbus in each cluster was a log normal spectrum of smaller convective features ranging from moderate cumulonimbus down to tiny nonprecipitating cumulus. The large cumulonimbus were typically grouped within a cluster into one or more mesoscale precipitation features (or MPF’s), which were apparently triggered in mesoscale regions of intensified low‐level convergence. As an MPF matured it developed a region of stratiform precipitation adjacent to its active deep convective cells. The stratiform precipitation fell from the anvil cloud. Associated with the stratiform precipitation were a mesoscale downdraft below the anvil cloud and an apparent mesoscale updraft within the anvil cloud itself, above the mesoscale downdraft. These mesoscale drafts were distinct from the convective‐scale updrafts and downdrafts of the cumulus and cumulonimbus cells of the cluster. Downdrafts, both convective scale and mesoscale, filled the planetary boundary layer in the vicinity of cumulonimbus with stable air of low moist static energy. These wakes of downdraft air exerted a strong control on where future convection broke out. The results of GATE show that to simulate the effects of tropical convection in large‐scale numerical models of the atmosphere a variety of phenomena must be accounted for, including not only convective‐scale updrafts and downdrafts but anvil clouds with mesoscale updrafts and downdrafts, downdraft‐induced boundary layer transformations, and mesoscale convergence patterns. Experimentation with ways of including some of these features of tropical convection in large‐scale diagnostic and prognostic studies is under way, but much work remains to be done.