Impact of Vertical Variability captured by GPS Occultation on the Microwave Sounding Unit Stratospheric Record

The Microwave Sounding Units (MSU) aboard the NOAA TOVS series satellites have provided the best satellite-based record of tropospheric and stratospheric temperatures to date, because, in part, they used an absorption of a well-mixed gas and were mostly insensitive to clouds. The data set has an advantage over the radiosonde record because of its global coverage. This data set, though, is imperfect for a climate monitoring record because MSU was never intended for long-term calibration. Among recent controversies, two analyses of the same MSU data set give conflicting results regarding temperature trends in the mid-troposphere. This discrepancy is due to the complex correction necessary for multiple satellites in different orbits, non-linearity in the detector response, decaying orbits, etc. GPS occultation offers an ideal data set to quantify various uncertainties in the MSU climate records because GPS occultation is absolutely calibrated by highly precise atomic clocks rather than onboard blackbodies. We have used the occultation archive containing GPS/MET data from 1995-1997 and data from the CHAMP and SAC-C experiments from 2001 to the present to simulate the MSU stratospheric temperature channel and thus validate it. (GPS occultation is not well suited to validate the tropospheric MSU channel because GPS microwave refractivity cannot uniquely determine temperature in the lower troposphere.) We have performed a full radiative transfer calculation for MSU stratospheric brightness temperature for each occultation profile, mapped the brightness temperature globally using Bayesian interpolation, and compared directly to maps of MSU stratospheric brightness temperature. Initial results show that a full radiative transfer calculation differs from the use of a static weighting function by ∼ 1 K in brightness temperature, induced mostly by varying vertical structure as exhibited by the occultation data. This correction is of the same order as that associated with post-processing errors of MSU data. We will present an exploration of the difference between MSU and GPS occultation data as a function of season and latitude, and whether these structures contaminate estimation of the inter-annual trend. Reference is made to the ECMWF and the NCEP reanalyses.