A Next Generation Microwave Rainfall Retrieval Algorithm for Use by TRMM and GPM

ABSTRACT
The Global Precipitation Mission (GPM), aside from a powerful core satellite, contains a conceptual element that combines many independent radiometers into a coherent framework dedicated to providing a global, 3-hourly rainfall product. This requires a modification of the current algorithm paradigm to one which (a) is independent of detailed sensor characteristics; (b) has an open architecture that allows the community to test and incorporate new procedures; (c) is transparent so that validation activities can help to refine the actual assumption in the algorithm; and (d) has a complete error model for both random and systematic errors. This proposal is aimed at building the necessary infrastructure to accomplish the above goals in a systematic fashion before the launch of GPM. The paradigm uses the GPM core satellite and cloud resolving model simulations to retrieve rainfall structures around the globe for an a priori database of clear and raining scenes. Radiative transfer calculations are then used to relate this a priori database to the expected radiances for each of the constellation satellites, thus forming a priori databases for each radiometer that are consistent in the sense that they are all based upon the same retrieval from the core satellite. A Bayesian inversion scheme is used to derive the rainfall from each constellation member. To implement this scheme, we will use the precipitation radar (PR) and the Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI) to simulate the GPM core satellite, while the Special Sensor Microwave Imager (SSM/I) and subsets of its observations will serve as constellation satellites. In the process of building the infrastructure needed for GPM, we will also build the next version of the TRMM TMI algorithm (2A12-v7), which should serve as a transition between the two missions. Unlike previous algorithm development efforts that concentrated solely on results, this effort will focus more on the algorithm characteristics needed for GPM. Nonetheless, we expect this effort to make significant improvements to the a priori database used in the current algorithm. As such, we expect version 7 of the current TRMM 2A12 rainfall algorithm to represent a substantial improvement over previous versions, while providing an architecture that is open, transparent, and amenable to multiple sensors.