Southern Ocean is a highly dynamic component of the global ocean circulation that plays a key role in the transport of heat, the uptake of carbon, and the global climate system [1
]. The Southern Ocean circulation is largely wind driven. Changes in the Southern Hemisphere wind field drive sea surface temperature (SST) gradients that can support a feedback mechanism and influence both the latitude of the Antarctic Circumpolar Current (ACC) and the distributions of heat and nutrients [9
]. Southern Hemisphere atmospheric variability exhibits a large number of modes, mostly influenced by large-scale low-frequency patterns [14
] and has been known to play a major role in Southern Hemisphere weather and climate [15
]. Due to the wave patterns associated with large-scale teleconnections, the largest temperature anomalies occur in the Amundsen-Bellingshausen Sea [17
]. For a more in-depth analysis into the relationship between large-scale atmospheric teleconnections and Southern Ocean SST, refer to [17
]. To analyze trends of Southern Hemisphere air-sea interactions, two patterns of atmospheric variability are compared with SST: the Antarctic Oscillation (AAO) and the Southern Oscillation (SO).
The AAO (also referred as the Southern Annular Mode) is a large-scale low-frequency pattern and is the dominant mode of atmospheric variability in the Southern Hemisphere Westerlies [16
]. Previous studies found that the westerly winds have shifted south due to the increasing frequency of the AAO and the growing Antarctic ozone hole [18
] and significant changes have been observed in both the temperature and salinity of the Southern Ocean [4
]. However, despite the increasing frequency of the AAO and the shifting Westerlies, statistical evidence indicates an insensitivity of the ACC and sloping isopycnals to decadal changes in wind stress [22
]. Within the Southern Ocean and Antarctic waters, previous studies have compared SST with the AAO [23
] on weekly to monthly scales, noting the significant anomalies induced by the atmospheric pattern.
The SO is a measure of the Walker circulation in the tropical South Pacific. This circulation relates to variations in ocean temperatures and atmospheric pressure across the broad expanse of the tropical Pacific. The SO is also a metric for the El Niño-Southern Oscillation (ENSO) index. Negative (positive) SO index signifies warm (cold) SST across the eastern tropical Pacific and can therefore describe patterns similar to El Niño (La Niña) events. Additionally, the SO and El Niño (La Niña) were found to be negatively (positively) related to the Pacific-South American (PSA) wave pattern [26
], which can have strong influences on SST, sea-ice, and atmospheric temperatures along Antarctica [27
]. Reference [29
] describes the strengthened influence of the AAO and SO during in phase periods, indicating the strong relationship between the atmospheric circulation and the effect on ocean dynamics. In this analysis, the statistical significance of the influence of teleconnections on Southern Ocean SST is investigated.
Hypothesizing that the SO has a greater influence than the AAO in the Southern Ocean and that both teleconnections drive warming (cooling) trends during positive (negative) phases of the oscillations, this analysis compares atmospheric teleconnections to Southern Ocean SST in both space and time. Investigation of the relationships between SST patterns and atmospheric variability with which they are associated provides a basis for improved understanding of Antarctic sea-ice and air-sea dynamics within the Southern Ocean.
We begin by addressing the relationship between SST and the large-scale AAO and SO teleconnections during the period 1982–2016. This period is used as it includes all full years in the OISST v2 data product. As described above, to reduce seasonal bias, standardized anomalies are compared against the large-scale patterns of AAO and SO indices. Correlations between SST and the teleconnections (Figure 1
a,b) relate positive anomalies in the mid-latitudes (30°S–50°S) and negative anomalies in the high latitude regions during positive phases. A positive AAO is associated with a poleward shift in the Westerlies, while a positive SO describes anomalously cold temperatures in the eastern Pacific (similar to La Niña events).
During positive AAO years, the shift in the Westerlies correlates with anomalously warm surface waters in the mid-latitude Atlantic and Indian basins and a broad negative anomaly in the South Pacific (Figure 1
a). A similar spatial pattern exists during positive SO years (Figure 1
b). Positive SO years correlate with large-scale negative standardized SST anomalies in the high latitude Pacific and positive anomalies in the mid-latitude Pacific, along the west coast of Australia, and south of 45°S in the Atlantic. The spatial extent of significant SST anomaly correlations with the SO is larger than the correlations between SST and the AAO. Between the mid-2000’s and 2014, both indices have been mostly in a positive phase (Figure 1
c). When both indices are in the positive phase, spatial correlations relate increasing negative anomalies along the Antarctic coast with increasing positive anomalies in portions of the Ross and Weddell Seas and the sub-tropical Southern Ocean, similar to [24
The influences of teleconnections are further explored through a more detailed analysis of the standardized SST anomalies (Figure 2
). Spatial linear regressions from 1982 to 2016 in the Southern Ocean (Figure 2
a) show significant large-scale surface warming (positive) in the mid-latitude Southern Ocean, the southern Indian Basin (60°E–120°E), and the Amundsen Sea sector. In addition, significant negative trends in the South Pacific Basin and Drake Passage region are seen. An estimate of the Southern Ocean trends suggests a mean value of 0.0142 per year, a median of 0.0154 per year, and a distribution negatively skewed towards negative trends.
Comparing one-year averaged anomalies, we look to analyze two recent examples: a year with opposing index values and another with similar in-phase index values. In doing so, we explore the spatial regions and magnitudes of positive and opposing index years. Using Figure 1
c, the years 2016 and 2010 were selected for each comparison respectively. The 2016 indices are described by a strong negative SO in the beginning of the year, changing to a strong positive SO by the end of the year with a consistently positive AAO (see Figure 1
c). That is, the phases of the indices are mostly out of sync. The pattern of standardized SST anomalies (Figure 2
b) from 2016 does show the strong similarities to the positive phases of both AAO and SO (i.e., negative sub-polar and positive sub-tropical anomalies), but also from the negative SO phase. The potency of the strong negative SO value (Figure 1
b) is seen in the anomalously warm SSTs in the eastern and central sub-tropical Pacific Ocean (Figure 2
b), both of which overshadow the values forced by the positive SO months.
The 2010 mean standardized SST anomalies (Figure 2
c) were the start of successive La Niña years (denoted in Figure 1
c) and a positive AAO. Most notable are the anomalously warm temperatures in the sub-tropical Pacific, South Atlantic, and western Indian Basins. Moreover, there are strong negative anomalies in the South Pacific and in the South Indian Ocean (near 60°E). The pattern of 2010 positive and negative anomalies (Figure 2
c) shares strong similarities with the positive and negative SST and SO correlations (Figure 1
b), showing the strong influence of the in-phase oscillations on SST.
Temporal analysis of the mean Southern Ocean SST (Figure 2
d) suggests warming at a rate of 0.0142 per year between 1982 and 2016 (0.0092 °C per year if not standardized), with a standard error of 0.0008. The average was taken between 30°S and 70°S for each month. Throughout the 35-year period, there was a net warming of approximately 0.50 based on standardized anomalies (approximately 0.32 °C if not standardized). A spatial comparison indicates that positive index months correlate to broad-scale warming in the mid-latitudes that supports the warming trend. To show the more recent relationship of in phases indices, Figure 3
depicts the standardized SST anomalies for AAO positive and SO neutral, AAO neutral and SO positive, and both AAO and SO positive indices. In this comparison, the opposing relationships of neutral-phase indices mitigate anomalies. However, during in-phase months of AAO and SO indices, the anomalies are greater in magnitude and broader spatially compared to either out-of-phase relationship, similar to the results of [29
A comparison of monthly and yearly averaged standardized SST anomalies (Figure 4
) mark contrasting differences between the temporal scales of AAO and the SO. The monthly (Figure 4
a) and yearly (Figure 4
b) averaged standardized SST anomalies during positive AAO events both have negative anomalies in the Pacific Basin and positive anomalies in the Atlantic and Indian Basins. On longer time-scales, this yearly pattern of anomalous temperatures is stronger than monthly averaged. Differences in the yearly and monthly AAO anomalies (Figure 4
c) show the stronger (red) yearly signal, particularly east of the Greenwich Meridian and south of Australia to the dateline. Yearly anomalies are weaker (blue) in the high latitude Indian, the mid to high-latitude Atlantic and mid-latitude Central Pacific.
Although global air-sea interactions are heavily influenced by large-scale teleconnections, seasonal regressions derive a response in SST to changes in index values (Figure 5
). In both instances, the largest coefficients of regression occur in austral summer and autumn, while the weakest coefficients arise in austral winter. Similar results for SST and ENSO were previously described in [27
] and [29
]. Based on these findings, spatial changes in austral summer indices can be used to depict linear changes in temperature. However, minimally significant regressions between austral winter anomalies and the indices suggest a potential non-linear or lag relationship.
Analyzing correlations and mean standardized anomalies allow for the interpretation of how large-scale teleconnections influence the Southern Ocean. Strong positive indices of both the AAO and SO (La Niña) induce broad-scale cooling in the high latitudes of the Southern Ocean and warming in the mid-latitudes and Weddell Sea. Each index is seen to drive spatial anomalies on a magnitude of 0.5 and anomalies approaching 1 when both indices are in-phase. Within Figure 1
a,b, the spatial correlations of the SO are more highly correlated spatially and based on magnitude, displaying the overlaying influence of the SO on the Southern Ocean SST. This result is important as the mid-2000’s through 2012 were largely in-phase positive oscillations, associated with large-scale negative anomalies in the high latitudes and positive anomalies in the mid-latitudes and Weddell sea. More importantly, the spatial pattern in the temporal linear regression of SST is spatially similar to SST anomalies in 2010, a year marked by a strong La Niña period and a positive AAO index value, showing the importance of in-phase oscillations on SST. The continual positive indices within the 21st century would therefore be strongly contributing to the positive trend in Southern Ocean mid-latitude SST and negative trend within the Southern Hemisphere high latitudes.
The patterns of SST anomalies seen during positive SO monthly (Figure 4
d) and yearly (Figure 4
e) periods are similar, with intensification throughout the year in both the Indian and Pacific Basins (Figure 4
f). This temporal disparity is most likely due to the time required for an Equatorial-Tropical Pacific phenomenon to influence large regions. A distinguishing feature of these patterns is that the magnitude of positive SO events in the Pacific Basin is greater than those associated with the AAO. The potential implication is that the SO plays a dominant role in the Southern Ocean. Furthermore, differences in yearly and monthly averaged anomalies are comparatively large. Thus, the longer duration of AAO and SO events could be an effective means of supporting or driving long-term trends.
SST anomalies can still be influenced by localized processes and feedback mechanisms. In years where the AAO and SO are increasingly positive, the Westerlies shift poleward and anomalously cooler temperatures are found in the high latitudes, with warmer temperatures in the mid-latitudes. There are breaks in the trend when strong El Niño events occur (negative SO), driving temperature changes that oppose the existing trend. Although this analysis supports the long-term increase in mean Southern Ocean surface temperature found by earlier analyses, it suggests that while the Atlantic and Indian Basins continue to warm, there has been significant cooling in the South Pacific as a result of the most recent patterns in the AAO and SO.