STRATOSPHERIC AEROSOL AND GAS EXPERIMENT-II UMKEHR OZONE PROFILE COMPARISONS

This study compares 1789 pairs of ozone profiles derived from 1384 Umkehr observations at 14 different stations and 1163 Stratospheric Aerosol and Gas Experiment (SAGE) II profiles coincident within 1000 km and 14 hours between October 1984 and April 1989. The comparison indicates the following significant percentage differences (SAGE II-Umkehr)/Umkehr with 2 x standard errors of the mean: Umkehr layer 4, (18.3 +/- 0.8)%; layer 5, (-1.6 +/- 0.4)%; layer 6, (-6.2 +/- 0.5)%; layer 7, (0.8 +/- 0.6)%; layer 8, (7.7 +/- 0.6)%; and layer 9, (12.0 +/- 1.1)%. Differences in layers 4 and 6 are due at least in part, to inaccurate Umkehr climatologies. Average SAGE II/Umkehr differences in layers 5 through 9 at individual stations are generally less than 10%. While the Umkehr retrievals are known to be sensitive to aerosol interference, the mean layer 8 correction during the period of this study is estimated to be only 2% with large station-to-station variability. The correction in lower layers is smaller. We have chosen to ignore the small Umkehr aerosol correction in this study. The mean difference would decrease if Umkehr profiles were: corrected for a priori profile effects calculated by DeLuisi et al. (1989a). However, using the newer Bass and Paur (1985) ozone absorption cross sections would tend to increase the differences at most levels. The profile of mean differences is similar to previously observed differences between Umkehr and solar backscattered ultraviolet (SBUV) observations. Comparing SAGE II/Umkehr differences to SAGE I/Umkehr differences at Seven common stations shows a bias of -4% at the ozone peak (layer 4). This bias increases with altitude to 8% in layer 8 and 15% in layer 9, with SAGE II ozone partial pressures higher than or equal to those of SAGE I (version 6.1) relative to Umkehr in all layers above 4. A systematic upward reference-altitude shift between 0.25 and 0.50 km for SAGE I, similar to the quoted uncertainty, would increase SAGE I ozone 4% to 8% in layer 8 and would result in similar SAGE and Umkehr ozone trends during the 1980. Cross correlations of numerous variables associated with the Umkehr and SAGE II data sets show a minimum correlation between SAGE II and Umkehr ozone partial pressures in layers 5, 8, and 9. This correlation is a result similar to the one previously noted in other comparisons against Umkehr data. We discuss these minimum correlations in relationship to the seasonal cycle in ozone and synoptic scale variations at midlatitudes based on model results. Substantial differences between SAGE II and Umkehr exist in both the mean and the variability of ozone in layers 8 and 9. Substantial differences also exist in layer 6 where the Umkehr algorithm does not retrieve the low atone values periodically observed by SAGE II during winter.