Rainfall estimation using a combination of geosynchronous infrared and TRMM observations

An improved version of the Convective/Stratiform Technique (CST) and its application over Brazil are presented. Keeping the major components of traditional CST, we have modified the scheme for classing convective and stratiform rainfall and tested various schemes to eliminate non-raining cirrus clouds. The parameters of the technique are calibrated using TRMM multi-sensor observations, including TMI, PR and VIRS. This technique takes advantage of the high temporal sampling rate of geosynchronous infrared channel and the better instantaneous rain observation of TMI and PR. Moreover, sparse TMI rain estimates are used to dynamically calibrate the parameters of the technique to improve its performance. The technique is applied to make rainfall estimates on various time scales and to study rainfall statistics such as the distributions of rain intensity and storm duration over a four-month period beginning January 1999. These rainfall statistics are needed to improve land-surface parameterization for current GCMs and are important parameters themselves for hydrological models. The study period coincides with the TRMM/LBA ground validation experiment, and observations from the LBA radar are used to validate the technique. Using an earlier version of this technique, promising results were found in the intercomparison between instantaneous estimates and independent passive microwave-based rainfall. The diurnal variation of rainfall in early 1996 in Brazil was well described by the technique, as was the variation of the convective and stratiform components of the rainfall.