A comparison of mean fields from the ECMWF operational analyses and the ECMWF Re-analysis (ERA) in high southern latitudes reveals significant differences in many variables, such as geopotential height, surface pressure, and winds. When the archive data are compared with Antarctic observations, results do not indicate that the ERA data are generally closer to observations. There are, in fact, many instances in which the ERA data are significantly farther from observations than the ECMWF operational analyses. It is possible that differences in the mean fields could be partially attributable to differences in the utilization of observations. In this regard, it is generally assumed that the ERA should be at an advantage, since communication problems periodically prevent operational analyses from including the optimal number of recorded observations. The ERA also has the advantage of consistency regarding the manner in which the data set was produced, whereas the operational data include periods with important model changes, such as in resolution. However, these advantages do not appear to have provided the ERA with an edge for use in climatological studies over the Antarctic.
Some scenarios that will be explored in order to help explain such unexpected results include the possibility that the ERA did not incorporate as many Antarctic data reporting sites as were used for the operational analyses. Another factor to be considered is that a different topographic representation is used in the ERA due to differences in model resolution. Thus, as these differences might help explain the differences in mean fields, they might also help explain the absence of a clear ENSO signal over the West Antarctic region in the ERA data. Such a signal is found in moisture flux convergence from the operational analyses for the domain 120° W to 180°W, 75°S to 90°S. These differences might also explain the absence of precipitation increases in the early 1990s in the ERA for the same region, increases seen in operational analyses and that are supported by precipitation estimates derived from snow pits and ice cores taken near Siple Dome in the center of the domain.