Abstract
Monthly rainfall estimates by two spaceborne microwave sensors TRMM/TMI and TRMM/PR are compared. The originality of this study is that the difference between them is decomposed into retrieval error and rain/no-rain classification (RNC) error. Overestimation by the standard algorithm for TRMM/TMI against the standard algorithm for TRMM/PR has been known, but the difference decreased after the version-up of both algorithms. Still, it should be noted that retrieval error shows larger overestimation than total difference. Our developed RNC methods can replace the original RNC method of the standard algorithm for TMI. It is discussed how to set the global parameter k0 in our methods.
