This presentation provides the first comprehensive report on how well the five facility TRMM precipitation algorithms designed for use with measurements from either the 13.8 GHz TRMM Precipitation Radar (PR), the 5-frequency/9 channel TRMM Microwave Imager (TMI), or their combination, have converged at a hierarchy of time-space scales. All five algorithms are physical in nature and use various forms of the radiative transfer equation (RTE) to relate rainrate to either passive microwave brightness temperatures (TBs) and/or radar reflectivities (Zs). The five algorithms consist of:
(1) a TMI-only rainmap solution that presumes an underlying log-normal rain distribution over a monthly-5 degree grid box and takes advantage of the varying dynamic ranges of the rain signals associated with different TMI frequencies;
(2) a TMI-only profile solution that uses cloud model output as a microphysical solution-basis and is cast in Bayesian form to expedite the RTE inversion calculations;
(3) a PR-only profile solution akin to Hitschfeld-Borden inversion that uses a PR-based climatology of cloud-free surface radar reflectance to assign total path attenuation and an observational database for defining the underlying allowable rain microphysics (i.e., drop size distribution information);
(4) a combined profile method based on a Bayesian modification to the PR-only scheme and augmented with an additional estimate of total path attenuation derived from the TMI 10.7 GHz channels; and
(5) a combined profile "tall vector" solution, also in Bayesian form, in which all radar-measured Zs and radiometer-measured TBs are used in an "instrument-balanced" inversion scheme, and in which the microphysical database is based on both observations and cloud model output.
The convergence statistics will be studied at three different time scales (monthly, daily, and instantaneously), at two different space scales (5-degree and pixel-level), and will be separated into ocean and land categories. Intercomparisons to ground-based volume scan radar data and specially selected raingage data will also be discussed. Preliminary analysis indicates that algorithm agreement has surpassed that found in prior algorithm intercomparison studies involving SSM/I-based algorithms, although more work is needed to carefully quantify the convergence statistics. Quantitative summaries will be provided in conference