Abstract: The new generation of Fengyun polar orbiters FY-3(02) satellite will be launched in 2012. Using Non-hydrostatic Modeling System (UW-NMS) cloud resolving model by University of Wisconsin, in conjunction with the VDISORT microwave radiative transfer model to simulate the upwelling brightness temperatures, the basic atmospheric parameters of Katrina are estimated, and the precipitation characteristics of sounding channels from two oxygen absorption complexes at 50—60 GHz and 118.75 GHz, which will be installed on FY-3(02) satellite are preliminarily analyzed. The two channels are combined to make use of their differential response to absorption and scattering by hydrometeors. To optimize the channel combination, four pairs of channels which show similar weighting functions in clear-sky atmospheric profiles are chosen (i.e., 50.3 GHz and 118.75±5.0 GHz, 51.76 GHz and 118.75±3.0 GHz, 52.8 GHz and 118.75±2.5 GHz, 54.40 GHz and 118.75±1.1 GHz), with the largest differences occurring at the lowest channels as a result of their different sensitivity to moisture that significantly affects observations below 500 hPa. To study the potential use of these frequencies, the relationships between the simulated TBs and the microphysical properties of the UW-NMS simulated precipitating clouds are analyzed. The sensitivity analysis shows that, all channels are very sensitive to hydrometeor species, and they are most sensitive to the graupel, then the snow, followed by the rain. Due to the scattering of the frozen hydrometeors, the TBs of 118.75 GHz decreases more than 50—60 GHz. After that, a Bayesian retrieval framework is used to retrieve the rainfall intensity and vertical structure of hydrometers, and three different sets of frequencies have been chosen to perform the retrieval: Using only the 50—60 GHz channels, using only the 118.75 GHz channels and a combination of all the channels respectively. Then the root mean square (RMS) and the correlation coefficient have been considered to analyze the behavior of the retrieval relative to the columnar liquid/ice water contents of rain, graupel, snow and surface rain rate. The results show that the 118.75 GHz channels are better in retrieving the columnar ice water contents of graupel, the 50—60 GHz channels are better in retrieving the columnar liquid/ice water contents of rain and snow, and the combination of all the channels always presents the highest correlation coefficient and the smallest RMS, so the combination of all channels will improve the precipitation retrieval precision. Then a second set of statistical indexes (correct rate, critical success index, probability of detection, false alarm rate) have been employed to evaluate the rain detection capability, discriminating between the raining and non-raining pixels. The results show that the correct rate and critical success index of all the channels are closer to 1, and the false alarm rate are closer to 0, so the combination of all the channels can better discriminate between the raining and non-raining pixels. After the launch of FY-3(02) satellite, this method will be further verified. And for global applications, more flexible retrieval approaches are required, which should be capable of constraining the algorithm according to the local situation.