A satellite infrared technique for estimating ''deep/shallow'' precipitation

This paper describes the statistical characteristics of the Japanese geosynchronous satellite, Geostationary Meteorological Satellite-5 (GMS-5), infrared (IR) data for estimating both ''deep/shallow'' precipitation. For this study, geographically matched data sets of GMS-5 IR data and composite digital radar data were prepared. By using those data sets, three-dimensional (3-D) matrices of Probability of Rain (PoR) and Mean Rain Rate (mRR) were calculated, where the input variables consisted of three parameters: IR 11 mu m brightness temperature (TB11); IR TB difference between 11 mu m and 12 mu m (TB11-12); and IR TB difference between 11 mu m and 6.7 mu m (TB11-6.7). The resulting statistical characteristics from those matrices are as follows: TB11-12 is an useful parameter for the removal of thin cirrus with no precipitation TB11-6.7 is an useful parameter for the extraction of deep convective cloud with heavy precipitation By using those matrices for looking up PoR and mRR, an empirical algorithm for estimating precipitation was developed. The potential of this technique (denoted as 3-D looking-up table (LUT)) as a now casting tool for severe weather was tested in the case of typhoon ''RYAN (T9514)''. The error and scatter of the 3-D LUT estimations were relatively large, but they captured peak rainfalls and accumulative rainfall in good agreement. (C) 1997 COSPAR.