Ozone retrievals from Stratospheric Aerosol and Gas Experiment (SAGE) II Ve
rsion 5.96 (v5.96) below similar to 25 km altitude are discussed. This vers
ion of the algorithm includes improved constraints on the wavelength depend
ence of aerosol extinctions based on the ensemble of aerosol size distribut
ion measurements. This results in a reduction of SAGE ozone errors in the 2
years after the Mount Pinatubo eruption. However, SAGE ozone concentration
s are still similar to 10% larger than ozonesonde and Halogen Occultation E
xperiment (HALOE) measurements below 20 km altitude under nonvolcanic condi
tions (and by more than this in the tropics). The analysis by Steele and Tu
rco [1997] suggests that the SAGE ozone overpredictions are in the wrong di
rection to be explained by aerosol extinction extrapolation errors. Moreove
r, preliminary SAGE II v6.0a retrievals suggest that they are partially acc
ounted for by geometric difficulties at low altitudes in v5.96 and prior re
trievals. SAGE ozone trends for the 1979-1996 and 1984-1996 periods are cal
culated and compared, and the sources of trend errors are discussed. These
calculations are made after filtering out ozone data during periods of high
, local aerosol extinctions. In the lower stratosphere, below similar to 28
km altitude, there is shown to be excellent agreement in the altitudinal s
tructure of ozone decreases at 45 degrees N between SAGE and ozonesondes wi
th the largest decrease in both between 1979 and 1996 having occurred below
20 km altitude, amounting to 0.9 +/- 0.7% yr(-1) (2 sigma) at 16 km altitu
de. However, in contrast to the fairly steady decreases at 45 degrees N, bo
th SAGE measurements and Lauder ozonesondes show ozone increases at 45 degr
ees S over the period from the mid-1980s to 1996 of 0.2 +/- 0.5% yr(-1) (2
sigma) from 15 to 20 km altitude. The SAGE data suggest that this increase
is a wintertime phenomenon which occurs in the 15-20 km height range. Chang
es in dynamics are suggested as the most likely cause of this increase. The
se hemispheric differences in ozone trends are supported by ozone column me
asurements by the Total Ozone Mapping Spectrometer (TOMS).