Overview of UARS ozone validation based primarily on intercomparisons among UARS and stratospheric aerosol and gas experiment II measurements

Comparisons among Stratospheric Aerosol and Gas Experiment (SAGE) II, microwave limb sounder (MLS) (version 3), cryogenic limb array etalon spectrometer (CLAES) (version 7), improved stratospheric and mesospheric sounder (ISAMS) (version 10), and Halogen Occultation Experiment (HALOE) (version 17) ozone profiles are reported with the emphasis being on the periods of January 9-11, April 15-20, and August 25-30, 1992, which were selected for analysis at four UARS validation workshops. The differences with respect to SAGE II are consistent with the conclusions of those workshops, MLS values are found to be systematically approximately 5% larger than SAGE II values except at 1 mbar (where they are approximately 5% smaller) and at 46 mbar in the tropics where they are approximately 20% too small (perhaps with some time dependence to this difference). HALOE values are systematically approximately 5% lower than SAGE II values, with a tendency for this difference to increase slightly in the upper part of the altitude range. CLAES values are approximately 15% high near 4.6 mbar and at 46 mbar in the tropics and are approximately 20% low near 0.32 mbar but show small differences near 1 and 10 mbar. ISAMS produced considerably more variable differences with respect to the other sensors both in space and in time. There remains clear evidence of Pinatubo aerosol contamination of ISAMS values at 10 mbar (the lowest level of the v10 retrievals) in January. The standard deviations of the differences between the coincidentally measured ozone profiles are generally consistent with the prescribed profile error bars of better than 5% from approximately 46 to 0.46 mbar for HALOE and from 1.5 to 15 mbar for MLS (and to approximately 0.46 mbar fur the 183-GHz measurements). For ISAMS also, the error bars of 10-15% are consistent with the standard deviations of the differences, but for CLAES, the precision seems to be approximately 10% and would be approximately 20% if the relatively systematic, vertical structure differences were to be included in the error bars.