In Antarctica, most weather stations did not start collecting meteorological observations until the late 1950s. To extend these records and provide a more complete picture of the pressure variability at each station, this thesis reconstructs the pressure records at 18 different Antarctic stations back to 1905, based on the Principal Component Regression (PCR). The PCR model uses only Southern Hemisphere mid-latitude pressure observations used as predictors. Several independent validation techniques are used to examine the level of accuracy of the PCR models, such as calibration correlation, validation correlation, reduction of error (RE), coefficient of efficiency (CE), and a comparison of the observed and reconstructed trends (1957-2011). The results have shown that austral summer and austral winter have higher reconstruction skill, whereas austral fall and austral spring are slightly weaker. Nevertheless, the reconstructions of all seasons are still better than using the climatological mean alone, and as such, all reconstructions are deemed to have some scientific use.
Since the reconstructions extend nearly double the observed record at each station and provide a more thorough depiction of the scope of pressure variability at each site, the uniqueness of changes throughout the historical record (1905-2011) can be examined. To accomplish this, several methods were used, including a 7-/ 11-year smoothing, 4 comparison of decadal means, and comparison of trends, using both fixed (1905-1956 vs.1957-2011) and temporally varying time periods (30-/40-/50-year running trends). In austral summer and autumn, a significant (p<0.05) decreasing trend can be seen in East Antarctica from 1990-2010, with these decades recording the lowest station pressures over the last century. These dramatic changes are believed to be related to stratospheric ozone depletion and/or greenhouse gases, especially in austral summer. In austral winter and spring, a shift to a most positive trend is observed from 1970-2011 in West Antarctica (particularly the Antarctic Peninsula). The recent (1957-2011) positive trends along the Antarctic Peninsula in austral spring are statistically significant (p<0.05). Interestingly, the lowest decadal mean pressures occurred in East Antarctica during the 1930s in austral spring, autumn, and winter. Although these reconstructions highlight many unique events since 1905, future research is still required to understand the mechanisms that influence many of these trends and fluctuations.