Antarctic gridded sea level pressure data: An analysis and reconstruction back to 1957

The reliability of the Australian (June 1972–April 1985) and NOTOS (1957–62) gridded monthly-mean, mean sea level pressure datasets over Antarctica is examined by comparison with station data from 29 sites over the continent. After rejecting about 30% of the months in both sets of gridded data, the remaining “good” months are used in a principal component motion technique to reconstruct gridded data from the station data for 1957 to 1985. The regression technique uses the “good” Australian data for calibration and verifies the statistical relationships developed between station and grid point pressure data with the “good” NOTOS data. The reconstructions are shown to be reliable over all of Antarctica between 60° and 75°S except in the area to the east of the Ross Sea and adjacent areas of the southern Pacific Ocean. The reconstructions are used to compare the NOTOS data with the more recent Australian gridded pressure data. Major differences between the two datasets are found over eastern Antarctica and the extreme southern Pacific and adjacent areas of western Antarctica. The first problem region was found to be related to extrapolation of the NOTOS data beyond their region of reliability as defined by the original published maps. The second problem region has a 10 mb difference between the two datasets, with the NOTOS data higher than the Australian. As this is the region of poorest data coverage anywhere in the world, the difference is difficult to resolve. In contrast, comparisons with the Taljaard et al. (1969) climatology show that this dataset contains fundamental spatial inconsistencies, and its further use cannot be recommended. A composite dataset linking the Australian, NOTOS and the reconstructed data can be produced for the whole region except for the southern Pacific and wet Antarctic region. This extended dataset is used to examine changes in pressure patterns between the January 1957–May 1972 and June 1972–April 1985 periods. Some of the changes in temperature that have occurred over this period can be explained by changes in surface circulation patterns