Capturing and interpreting the Southern Annular Mode’s positive shift in large ensembles using the snapshot approach

In this study we apply a recently developed, ensemble-based method to track the changes of the Southern Annular Mode (SAM) and its climate impacts under a changing climate. The advantage of this method is that it does not rely on oscillation patterns tacitly assumed to be constant over a given time period, as it is the case with traditional methods based on single time series.

SAM is the leading mode of atmospheric variability of the Southern Hemisphere’s extratropics. One of the traditional definitions of the SAM is that it is the first mode of the empirical orthogonal function (EOF) analysis of mean sea-level pressure (SLP) for 20°S–90°S for a given time period. SAM index time series is then computed by projecting the SLP anomalies on this loading pattern and standardized for the time period. The strength of the linkages associated with SAM can be calculated as correlation coefficients between time series of the SAM index and other meteorological variables.

Studies over the last decades showed a trend of the SAM towards the positive phase. Since the traditional, time series-based definition of SAM calculates the oscillation pattern for a chosen time period, this pattern and the SAM-related correlations are treated as constant for that time period. However, in a changing climate stationarity cannot be assumed. Consequently, positive shift inferred using traditional methods are questionable. Therefore, in this study, using different climate models’ large ensembles, we apply a recently developed method, the snapshot EOF (SEOF) analysis to calculate SAM. This method performs the EOF analysis across the ensemble dimension at each time instant, utilizing the fact that members of a sufficiently large ensemble correctly cover the distribution of the possible climate states at each time instant after a certain convergence time. Instantaneous ensemble-based SEOF loading patterns represent the spatial structure of the SAM that characterizes the potential variability in the climate states of the given time instant, and the corresponding SEOF-based SAM indices reveal the phases in which the ensemble members are in that very moment. In this way, beyond a correct characterization of the SAM at each time instant, the time-dependence of its pattern can also be monitored. Furthermore, instantaneous correlation coefficients between the instantaneous indices and other variables can be computed across the ensemble to reveal the correct instantaneous connection strengths and their time-dependence.

By means of the SEOF analysis, we show that the recent and future positive trend in the SAM for 1950–2100 seen with the traditional methods is the consequence of the change in the ensemble mean SLP field (mean state of the oscillation), with decreasing SLP in Antarctica and increasing SLP at the mid-latitudes. Besides this, SEOF-based SLP regression maps reveal that the absolute value of the typical amplitudes of SAM-related anomalies will decrease at most of the geographical locations, and the explained variance also shows a significant decrease of 5-10%. Correlation coefficients with surface temperature changes even 0.3-0.4 over the 150 years in certain regions.