Recent years have seen a resurgence of interest in the structure of the upper tropical troposphere. One proposed climate feedback holds that the infrared emission temperature of tropical anvil clouds will remain constant with climate change (Hartmann and Larson, 2002). One basis for the feedback is the observation that convective clouds detrain preferentially at the altitude where clear-sky radiative cooling decreases most rapidly, approximately 200 hPa or 12 km (Hartmann et al., 2001a, 2001b). Other studies investigate the tropical tropopause layer (TTL) and its role as the source region for air entering the stratosphere. The TTL spans, approximately, altitudes of 12 to 17 km, although a consistent definition of its boundaries does not yet exist. Deep convection represents the potentially dominant source of lower tropospheric air into the TTL, but its exact role remains uncertain.
Most previous studies of the upper tropical troposphere have relied on satellite data (Hartmann et al., 2001; Gettelman et al. 2002; Soden 2002; Zhang 1993) or radiosonde or ozonesonde data (Folkins et al. 1999; Vomel et al. 2001). In this study, we examine a different dataset rarely examined within heavily-raining regions but of potential value: cloud radar data. Cloud radars can outline the cloud vertical structure of at least the lower half of the TTL more directly than either satellites or soundings, and they can comprehensively document diurnal variability in the cloud vertical structure. We examine the temporal and vertical variability of the upper troposphere during the three weeks of EPIC focused on deep convection studies, as observed by a 35 GHz cloud radar and soundings and clear-sky radiative heating rate profiles. The R/V Ron Brown was located at approximately 10 N and 95 W from September 10 through September 31.
A preliminary analysis examines peaks in the probability of cloud ("cloud" is defined as a radar reflectivity exceeding -50 dBZ) by height for cloud radar scans with and without precipitation (a radar scan is considered precipitating if almost all of the range gates below 4.5 km have Doppler velocities exceeding 3 m/s). For this heavily-raining region, many of the non-raining scans still occur near in time to precipitation. A broad maximum in cloud occurrence is seen between 8 and 12 km. A pronounced diurnal cycle is evident, with relative maxima in cloud occurrence at 8 and 12 km from 9 pm to 9 am LT, partially coinciding with a peak in convective activity observed at approximately 2 am LT within precipitation radar data. The soundings indicate enhanced detrainment into more stable layers is occurring at 8 and 12 km, as shown by maxima in the frequency of occurrence by height of lapse rates exceeding a set threshold. The peak at 12 km probably coincides with a near-zero clear-sky radiative cooling rate (this still needs to be determined) and is consistent with results from the GATE (Houze and Betts, 1981). The peak at 8 km in cloud and stability layer occurrence is more puzzling.