In this study we explore a database of CloudSat+AMSU-B coincident overpasses to determine the minimum threshold of passive microwave detection of snowfall using the high-frequency channels available on GMI.  Using the Advanced Infrared Sounder (AIRS) temperature and water vapor profiles along with a surface emissivity database developed from AMSU-B and MHS observations, clear-sky radiances are simulated and compared to AMSU-B observed radiances for all profiles with a maximum temperature less than 271 K.

The ability of a scattering signal (observed brightness tempertaure colder than clear-sky brightness temperature by a threshold T) to detect snowfall (CloudSat reflectivity greater than threshold Z) is quantified using the Heidke Skill Score. The 183+/-1 and 183+/-3 GHz channels have the highest skill scores, while those channels that are sensitive to the surface (89, 150 and 183+/-7 GHz) have zero or even negative skill (depending on Z and T), implying that an emission signal (presumably from cloud water) is as good or better for detecting precipitation than a scattering signal. These results emphasize the need for proper characterization of surface emissivity and adequate representation of cloud water in cold season precipitation profiles that form the databases used for Bayesian retrievals from GMI and other GPM constellation radiometers.