NASA - National Aeronautics and Space Administration

This study uses measurements of radiation and cloud properties taken between January and August 1998 by three Tropical Rainfall Measuring Mission (TRMM) instruments, the Clouds and the Earth’s Radiant Energy System (CERES) scanner, the TRMM Microwave Imager (TMI), and the Visible and Infrared Scanner (VIRS), to evaluate the variations of tropical deep convective systems (DCSs) with sea surface temperature and precipitation. The authors find that DCS precipitation efficiency increases with SST at a rate of ϳ2% KϪ1. Despite increasing rainfall efficiency, the cloud areal coverage rises with SST at a rate of about 7% KϪ1 in the warm tropical seas. There, the boundary layer moisture supply for deep convection and the moisture transported to the upper troposphere for cirrus anvil cloud formation increase by ϳ6.3% and ϳ4.0% KϪ1, respectively. The changes in cloud formation efficiency, along with the increased transport of moisture available for cloud formation, likely contribute to the large rate of increasing DCS areal coverage. Although no direct observations are available, the increase of cloud formation efficiency with rising SST is deduced indirectly from measurements of changes in the ratio of DCS ice water path and boundary layer water vapor amount with SST. Besides the cloud areal coverage, DCS cluster effective sizes also increase with precipitation. Furthermore, other cloud properties, such as cloud total water and ice water paths, increase with SST. These changes in DCS properties will produce a negative radiative feedback for the earth’s climate system due to strong reflection of shortwave radiation by the DCS. These results significantly differ from some previously hypothesized dehydration scenarios for warmer climates, partially support the thermostat hypothesis but indicate a smaller magnitude of the negative feedback, and have great potential in testing current cloud-system-resolving models and convective parameterizations of general circulation models.