An estimate of the strength of the Austral Summer Antarctic Oscillation using station sea level pressure records for the period 1877-2000 is presented, the first to our knowledge. The reconstruction was obtained by relating the Antarctic Oscillation intensity derived from NCEP/NCAR reanalysis data to 10 station records using canonical correlation analysis. Particular effort has been made to fit the model in a way that is robust to the questionable trends in the Southern Hemisphere in the NCEP/NCAR data. The trends in the reconstruction are therefore not trends from the NCEP data, rather trends from the stations. Cross-validation with the NCEP data and comparison with other analyses of the AAO over the late instrumental period gives us confidence that this station-based reconstruction can be regarded as trustworthy, enabling the circulation of the last few decades to be considered in a longer-term context. Station-based reconstructions for the other seasons are also in progress.

To extend this record further back, a reconstruction using tree-ring chronologies back to 1743 has also been undertaken. Comparison of this reconstruction with the station-based reconstruction shows moderate agreement on interannual and decadal timescales, but the comparison also points towards the inherent uncertainties of proxy-based climate reconstructions. In particular it was found that this tree-based reconstruction may have been influenced by a warming that is not attributable to changes in the Antarctic Oscillation Index during the twentieth century. Comparison of the tree-based reconstruction with a published reconstruction of zonal flow over New Zealand before the twentieth century shows common features.

The surface temperature and precipitation signals of the Antarctic Oscillation have been calculated and show that the response of the chronologies to Antarctic Oscillation variability is physically-consistent. In addition it was shown that the observed warming over much of Antarctica between the late 1950s and the 1980s is attributable to a substantial fraction to changes in the Antarctic Oscillation, whereas the observed warming over New Zealand is due to other influences.