Antarctic station-based seasonal pressure reconstructions since 1905: 1. Reconstruction evaluation

摘要

Seasonal mean Antarctic pressures at 17 stations are reconstructed based on the method of principal component regression, employing midlatitude pressure data as predictors. Several reconstruction methods were performed in order to assess the stability and reliability of the reconstructions obtained, including performing the reconstructions over a shorter 30 year window and withholding the remaining data for an independent validation. Generally, there were small differences between the various approaches, but typically reconstructions conducted on data with the trends still present and over the full period of observations achieved the highest skill. Seasonally, reconstruction skill was high in austral summer across the entire Antarctic continent. Reconstructions that employed gridded pressure data over oceans as well as the observations (here termed “pseudoreconstructions”) also performed remarkably well in austral winter. Spatially, the reconstruction skill was highest near the Antarctic Peninsula in all seasons, and weakest in coastal East Antarctica and the Antarctic Interior during austral spring and autumn; the spatial variability of the skill in part reflects the distance to the nearest midlatitude predictor. Nonetheless, for nearly all seasons and locations the observed trends since 1957 were well captured by the reconstructions, as was the low-frequency decadal-scale variability. These results suggest Antarctic pressure observations can be extended throughout the twentieth century with high confidence, especially in summer, allowing for a more precise understanding of the role and magnitude of natural atmospheric circulation variability across Antarctica.