The principal mode of variability in the atmospheric circulation of the southern extra-tropics and high latitudes is the Southern Annular Mode or SAM, essentially a zonally symmetric structure with anomalies of opposite signs in Antarctica and the extra-tropics. Several studies have indicated a trend towards the positive phase of the SAM (which is when Antarctic pressures are lower), perhaps due to ozone loss above Antarctica. However the studies’ conclusions regarding long-term trends are potentially flawed because they utilise NCEP/NCAR reanalysis (NNR) data. This is known to have spurious decreases in pressure at high southern latitudes resulting from significant positive errors in pressure prior to the availability of satellite sounder data.

An empirical definition of the SAM exists based on zonal mean sea level pressure (MSLP) values at 40˚S and 65˚S. In this study we use MSLP observations from six stations located approximately at each of these two latitudes to provide a proxy zonal mean, which in turn allows us to compare the NNR (1958-2000) and ECMWF ERA-15 (1979-93) and ERA-40 (1958-66, 1973-79) reanalyses directly to observations. This study finds that:

1. At 65˚S there is significant bias (up to ~20 hPa) in monthly zonal MSLP in both the NNR and ERA-40 data in the 1950s and 60s. However, by the 1970s ERA-40 is very good—in fact better than NNR in the 1990s.

2. There is a marked annual cycle in the NNR bias at 65˚S with values accurate in summer but significantly too high in winter, resulting in weaker westerlies and circumpolar vortex at this time. The reduction of this positive winter bias through time has contributed to the apparent increase in the vortex strength (positive SAM) in the NNR.

3. The observations reveal a statistically significant increase (<5% level) in the difference between the zonal MSLP at 40˚S and 65˚S during 1958-2000. In the NNR data this increase is three times as great (and significant at <1% level) because of the aforementioned errors at the higher latitude. Thus, while studies using the NNR data are shown to have overestimated the change in SAM, observations establish the recent trend towards the positive phase to be a genuine indication of climate change.