https://www.science.org/doi/10.1126/sciadv.aeb0166

Antarctic Sea Ice Loss and Southern Ocean Destratification: Complex Causes

Abstract

Antarctic sea ice extent has been decreasing since 2015, reaching a record minimum in 2023. To diagnose the causes of this decline, we analyzed an ocean–sea ice model based on observational data spanning 2013–2023 and identified three distinct stages of sea ice retreat. First, intensifying westerly winds increased upwelling of warm, saline Circumpolar Deep Water (CDW) into the Southern Ocean, preconditioning the region. Second, strong winds in 2015–2016 promoted mixing of CDW into the upper ocean, initiating sea ice loss, particularly in the southeastern sector. Third, continued mixing of CDW into surface waters and reduced equatorward export of freshwater from sea ice combined to maintain unprecedented low sea ice conditions. Sea ice loss in the southeastern sector was primarily driven by bottom-up processes through enhanced CDW fluxes, whereas sea ice loss in the southwestern sector was also influenced by longwave radiation flux anomalies. Our results suggest that under anthropogenic forcing, favorable upwelling conditions for CDW may persist, leading to a prolonged period of low sea ice extent in the Southern Ocean.

Introduction

Antarctic sea ice is a crucial component of the Earth's climate system, regulating the Southern Ocean's albedo (1), meridional overturning circulation (2), internal water mass transformations (3), ocean heat and carbon uptake (4), ocean heat content (5), and biological productivity (6). Antarctic sea ice exhibited a slight positive trend from 1979 to 2015 (7–9), characterized by large regional variations. During this period, sea ice expansion was likely associated with increased northward transport of sea ice driven by winds (10) and subsequent poleward freshwater export from melting sea ice, leading to surface freshening (11). However, since 2015, observed sea ice extent has experienced persistent negative anomalies, reaching record lows in both winter and summer seasons in 2023 (12). These negative SIE anomalies have been linked to warming temperatures within the top 100 meters of the water column (13) and increased surface salinity (14). This abrupt transition from record highs to record lows in Antarctic sea ice extent represents one of the most significant climate shifts currently underway in the Earth system, with the potential to accelerate global warming (1), disrupt heat and carbon sequestration pathways in the Southern Ocean (15), negatively impact ecosystems (16), and influence the stability of ice shelves that buttress vulnerable glaciers (17).

Several hypotheses have been proposed to explain the role of the ocean and atmosphere in regulating the recent decline in Antarctic SIE. Reclassifying these proposed mechanisms according to their temporal scales can provide valuable insights. At synoptic time scales, wind variability can immediately influence Ekman transport within the ocean. For example, abrupt weakening of westerly winds during the summers of 2016/17 and 2019/20 led to a reduction in northward Ekman transport of relatively cold and freshwater surface waters, resulting in surface warming and salinification in the outer Weddell Sea (18). This warming could have contributed to summer sea ice extent reductions and subsequent delays in sea ice growth. At seasonal time scales, intense polar cyclones may have been responsible for the formation and maintenance of offshore polynyas in the Weddell Sea during 2016 and 2017 (19–21). Additionally, a positive Zonal Wave-3 (ZW3) pattern (22) during 2016 was associated with enhanced poleward transport of warm subtropical air masses, leading to increased cloud cover and downward longwave radiation flux over the sea ice region (19, 23). The spatiotemporal trajectory of the ZW3 pattern in 2016 likely influenced sea ice concentration and drift, contributing to SIE reductions in the Weddell Sea, Amundsen Sea, Bellingshausen Sea, and western Ross Sea (23–26). Warm northerly wind currents were also associated with intensification of the Amundsen Sea Low (ASL) during 2016 and 2019 (21, 27).

At decadal to multidecadal time scales, the Southern Annular Mode (SAM) is the dominant climate variability mode in the high-latitude Southern Ocean (28). Positive SAM phases are associated with poleward shifts and intensification of westerly winds, modulating Ekman transport speeds. Therefore, a strengthened SAM is expected to have two opposing effects on upper ocean stratification: increased equatorward export of polar waters and sea ice, leading to surface freshening (11, 29), and enhanced Ekman pumping, resulting in salinification at higher latitudes (30). The SAM has exhibited a positive trend since the 1970s (28), which is linked to increased wind cyclonic circulation over the Southern Ocean, enhancing upwelling of warm deep waters through Ekman pumping (24, 31, 32). Furthermore, intensification of ASL and ZW3 patterns (33) has led to increasing zonal asymmetry in SAM, strengthening poleward flows of warm and moist subtropical air masses in specific regions of the Southern Ocean. Industrial-era climate model simulations suggest that SAM-ZW3 interactions have influenced regional variability in sea ice extent (34).

Future sea ice evolution is likely to be governed by a balance between competing mechanisms. For instance, increasing ocean and atmospheric heat content under anthropogenic forcing is expected to suppress sea ice growth. Conversely, increased precipitation (36) and meltwater input from glaciers (37) will lead to surface freshening (35), which can stratify the upper ocean and slow down deep convection cells (37, 38), thereby inhibiting vertical mixing of heat and potentially promoting sea ice growth. However, the Southern Ocean is currently experiencing an unexpected trend toward increased upper ocean salinity (14). This salinification weakens stratification (14) and reverses the stabilizing effect of freshwater, potentially enabling mixing of heat and salt from deeper CDW layers.

The potential role of intensifying westerly winds in determining the fate of Antarctic sea ice has been highlighted by idealized model studies. From this perspective, a two-hour timescale hypothesis (39) suggests that poleward intensification of westerly winds would induce immediate responses on a two-hour timescale. These immediate responses include increased northward Ekman transport of colder and fresher waters from high latitudes, leading to surface cooling and sea ice extent increases. Delayed responses involve enhanced upwelling of warm and saline CDW through Ekman pumping, resulting in warmer and saltier upper ocean conditions that contribute to sea ice extent reductions. However, observed oceanic responses are more complex and can be modulated by surface freshening (5) and mesoscale eddy activity (40, 41), which counteract the wind-induced changes in stratification and circulation. More realistic simulations suggest that SIE reductions can arise from favorable upwelling conditions driven by natural Southern Ocean variability (42) or historical forcing (43). Overall, while the observed pattern of gradual increase followed by a rapid decline in Antarctic sea ice extent since 2015 qualitatively aligns with the predictions of the two-hour timescale hypothesis (13), significant discrepancies remain.

Many potentially important processes have been proposed, but the mechanisms governing recent Antarctic sea ice evolution remain uncertain and are a focal point of active scientific discussion. Climate models generally struggle to represent observed variability and often simulate physically unrealistic scenarios (44). Here, we use eddy-permitting data assimilation sea ice–ocean state estimates, the Biogeochemical Southern Ocean State Estimate (SOSE) (45), to identify the drivers of Antarctic sea ice change between 2013 and 2023. This period encompasses a shift from record highs to record lows in sea ice extent (46). By constructing budgets of sea ice volume (SIV) and conserved upper ocean properties (e.g., heat and salinity), we can identify the key factors contributing to sea ice loss and assess their forcing mechanisms and the sequence of underlying causal events.

Our analysis reveals that recent Antarctic sea ice loss was a result of three interconnected driving stages. First, prior to mid-2015, SIE increased alongside a cold and freshened anomaly in the upper ocean, consistent with enhanced equatorward export of ice-origin freshwater stabilizing the surface layer. Second, after mid-2015, heat and salinity began accumulating in the upper ocean, initially as a result of the shoaling of warm and saline Circumpolar Deep Water (CDW). This response, qualitatively consistent with two-hour timescale hypotheses, was facilitated by favorable winds for upwelling and enhanced vertical mixing of heat and salinity driven by intensified westerlies. Third, after 2018, preceding sea ice changes altered surface freshwater fluxes, progressively becoming more important in maintaining increased salinity and weakened stratification in the upper ocean, ultimately promoting reduced persistence of Antarctic sea ice conditions.

Finally, we highlight substantial differences in sea ice evolution and its driving mechanisms between East Antarctica and West Antarctica. This zonal asymmetry stems from contrasting wind forcing and emphasizes the spatial complexity of coupled atmosphere–sea ice–ocean dynamics in the Southern Ocean. By identifying the dominant mechanisms within each region, we provide an integrated picture of Antarctic sea ice change.

Results

Southern Ocean Hydrographic Evolution Overview

In the subantarctic Southern Ocean, CDW resides beneath the weakly stratified surface waters during winter (Figs. S1 and S2) (47). Compared to surface waters, CDW is warmer, saltier, characterized by lower dissolved oxygen (DO) concentrations, and higher dissolved inorganic carbon (DIC) concentrations (48). In the eastern Antarctic continental shelf break region (E Ant; 60°W ~ 150°E), a warm anomaly developed below 100 m depth during 2013–2016, while the upper ocean (top 100 m) exhibited cooling and freshening anomalies. Subsequently, the upper ocean warmed and became saltier, with the lower ocean (200–500 m depth) cooling and slightly freshening after 2018 (Fig. 1, A and B). Upper ocean warming and salinification were accompanied by DO depletion (yellow contours in Fig. 1A) and DIC increase (black contours in Fig. 1B).

In the western Antarctic continental shelf break region (W Ant; 150°E ~ 60°W), the upper ocean exhibited cooling anomalies from 2013 to 2015, warming anomalies from 2016 to 2020, and then cooling anomalies again from 2021 to 2023 (Fig. 1C). The upper ocean showed DO depletion (yellow contours; Fig. 1C) from 2013 to 2019. The lower ocean exhibited DO enrichment (black contours; Fig. 1C) from 2017 to 2021. Salinity anomalies were positive in the upper ocean from 2013 to 2019 (Fig. 1D), and DIC enrichment was observed from 2015 to 2019 (black contours; Fig. 1D). After 2020, the upper ocean transitioned to a freshened state. The lower ocean shifted from colder and saltier anomalies before 2018 to warmer and fresher anomalies after 2018. This contrasts with the eastern Antarctic, where the lower ocean transitioned from warmer and saltier anomalies to colder and fresher anomalies.

To provide further context for these hydrographic changes, we include actual temperature and salinity profiles in Fig. S2. These show that the eastern Antarctic generally has a saltier upper ocean than the western Antarctic during winter. The internal thermocline fluctuates seasonally in both regions, shoaling during winter and deepening during summer. The eastern Antarctic also exhibits a shallower winter thermocline (50 m) compared to the western Antarctic (80 m).

In summary, from mid-2015 onward, the upper ocean over the eastern Antarctic continental shelf break underwent a clear transition from colder, fresher, DO-enriched, and DIC-depleted conditions to warmer, saltier, DO-depleted, and DIC-enriched conditions. These anomalous characteristics are consistent with CDW properties (48), suggesting an increase in CDW presence in the near-surface ocean, consistent with observational findings (32). We will delve into a detailed analysis of heat and salinity budgets in the following section to elucidate the associated mechanisms. In contrast, hydrographic evolution over the western Antarctic continental shelf break appears more complex. While there is a clear transition from saltier conditions (2013–2019) to fresher conditions (2020–2023) in the upper ocean, other parameters exhibit more intricate changes.

The SOSE-derived sea ice loss and associated upper ocean warming and salinification align with field hydrographic profiles (13) and satellite-based remote sensing surface salinity (14). Here, we demonstrate zonal asymmetry in upper ocean hydrographic property evolution. The underlying mechanisms driving these changes remain unexplored. In the following section, we will consider SIV, temperature, and salinity budgets to shed light on the dynamics governing Antarctic sea ice variability.

Antarctic SIE and Volume Anomalies

Satellite observations indicate sustained negative SIE anomalies in both eastern and western Antarctica since 2016 (Fig. 2A). Initial observations reveal a relatively stable sea ice regime characterized by zonal SIE anomaly patterns prior to 2008, with the western Antarctic (Ross Sea, Amundsen Sea, and Bellingshausen Sea) exhibiting predominantly positive anomalies and the eastern Antarctic (all other regions) showing predominantly negative anomalies. While SOSE SIE anomaly time series agree with satellite observations, they exhibit some positive bias in magnitude compared to observations during 2013–2016 in the eastern Antarctic and 2022–2023 in the western Antarctic. Furthermore, the temporal trend of SIE anomalies in SOSE generally aligns with those observed in satellite data. SIE anomalies primarily reflect anomalies at the equatorward edge of the sea ice pack; however, SOSE sea ice thickness anomalies indicate that sea ice thinning also occurred within the pack (Fig. S3). Both eastern and western Antarctica exhibit negative SIV anomalies after 2016, with western Antarctic continental shelf break sea ice showing recovery after 2021 (Fig. 2, B–E). This recovery period coincides with the positive bias in SOSE SIE. We will discuss the reasons for this in subsequent sections.

Budget analysis shows that SIV loss in the offshore region of the East Antarctic continental shelf (depth >3000m) was driven by net thermodynamic sea ice production reduction (i.e., melting relative to growth; Figure 2F and Figure S4F). This signal is consistent with previous research proposals suggesting that sea ice loss since 2015 was thermodynamically forced rather than mechanically driven by advection or divergence (49). However, due to uncertainties in sea ice thickness measurements, previous studies could not quantify the volume loss, which is now clarified by these results. Although sea ice advection and divergence (AD) partially offset SIP reduction, it was insufficient for complete compensation, resulting in negative SIV anomalies in East Antarctica (Figure 2B). Additional SIV contributions from the AD term stem from SIP increases in the East Antarctic continental shelf during 2015–2016, 2019–2020, and 2022–2023 (Figure 2G). This increase is likely due to increased sea ice export from the continental shelf, driven by SIP reduction in the offshore region (Figure 2G). Negative SIV anomalies have been observed in both the continental shelf and offshore regions of East Antarctica since 2020 (Figure 2, F and G).

Sea ice volume loss in the offshore region of the West Antarctic continental shelf is not as distinct and does not persist throughout the entire period since 2016 (Figure 2; West Antarctic offshore and West Antarctic shelf). The low SIV observed during 2016–2020 was driven by a combination of factors including low SIP (during 2013, 2015, 2017, and 2021–2022) and low AD (evident during 2018–2020; Figure 2H and Figure S4H). The SOSE SIV budget reveals longitudinal asymmetry in sea ice evolution, with East Antarctica showing relatively more persistent negative SIV anomalies since 2016 (Figure 2; East Antarctic offshore and East Antarctic shelf). The evolution of SIV anomalies is consistent with observed SIE anomalies, with losses more pronounced in East Antarctica.

The SIV budget emphasizes the important role of thermodynamics in sea ice decline in East Antarctica, suggesting that the mechanisms driving sea ice loss involve heat transfer to ice. Heat can be supplied from the atmosphere above or from warm CDW generally located below the thermocline. Sea ice loss was associated with DIC accumulation and DO depletion along with upper ocean warming and salinification. These features are consistent with increased CDW presence in the near-surface layer (50), but can also occur from enhanced atmospheric heat input. Detailed analysis of the processes controlling heat supply to the upper ocean in the following section is needed to distinguish between these possibilities.

In the large sub-Antarctic low pressure systems of the Weddell Sea and Ross Sea, the thermocline (and lower CDW layer) shoals to depths of 50–100m (32, 47). However, unless stratification is sufficiently weakened to promote CDW mixing into the mixed layer, this heat remains trapped below the thermocline (51, 52). In cold polar waters, stratification is primarily controlled by salinity (53). To investigate the processes driving stratification weakening and upward heat transfer to the mixed layer, we analyze upper ocean temperature and salinity budgets in the following two sections.

Southern Ocean Upper Ocean Warming

Shortwave and longwave fluxes are affected primarily by two factors: (i) sea ice, which alters surface albedo, and (ii) clouds, which reduce shortwave radiation transmission through the atmosphere but increase downward longwave radiation. The sub-Antarctic Southern Ocean is generally characterized by net heat loss through longwave radiation. Therefore, positive anomalies in downward longwave flux represent a reduction in this heat loss and effectively contribute to ocean warming.

Applying this framework to the East Antarctic offshore region, upper ocean warming is evident from 2015 to 2018 (Figure 3A). From late 2013 to mid-2015, surface fluxes ("surf"; excluding shortwave flux) show positive anomalies (warming trend) during the period of expanded sea ice extent. This is consistent with the insulating effect of sea ice, which suppresses heat loss to the atmosphere through longwave and sensible heat fluxes (2013–2014; red line in Figure 3C). From early 2016, surface fluxes show pronounced and persistent negative anomalies reflecting increased heat loss to the atmosphere due to reduced sea ice extent (red line in Figure 3C) (12).

Despite this transition, the primary driver of upper ocean warming beginning in mid-2015 is the vertical mixing term (Diff_v), which transitions from persistent negative anomalies to consistent positive anomalies through 2023 (blue line in Figure 3C). In contrast, shortwave flux (red dashed line in Figure 3C) only transitions to positive anomalies in mid-2016, after sea ice loss began in mid-2015. This timing suggests that shortwave flux did not initiate sea ice loss but subsequently promoted its enhancement. Although surface flux warmed the ocean from late 2013 to early 2015, this warming was offset by cooling from vertical mixing, resulting in little net temperature change. However, temperature begins to rise from mid-2015 (Figure 3A), initially driven by enhanced vertical mixing.

The West Antarctic offshore region also experiences upper ocean warming starting from 2015, peaking between 2017–2020, followed by a cooling period (Figure 3B). However, the mechanism driving this warming is markedly different from that in East Antarctica. In East Antarctica, vertical mixing plays an initial and sustained role, whereas in West Antarctica, vertical mixing contributes to warming only in specific years (2018, 2021, and 2022). Furthermore, unlike East Antarctica, shortwave flux in West Antarctica shows negative anomalies even during periods of anomalously reduced sea ice extent, such as the summers of 2016/17 and 2019/20. Simultaneously, the remaining surface flux components show positive anomalies. This pattern suggests warming in West Antarctica due to reduced heat loss through longwave radiation, which contrasts with East Antarctica. In East Antarctica, sea ice reduction is always consistent with increased shortwave flux and negative anomalies in other surface flux components.

To investigate surface heat flux components in more detail, we analyzed ERA5 for the East Antarctic and West Antarctic offshore regions. The results (Figure 4) are consistent with patterns observed in SOSE (Figure 3). In East Antarctica, periods of sea ice extent enhancement are generally associated with negative anomalies in shortwave flux: during periods of high SIE (positive anomalies), the time-mean anomalies are −0.3 W/m² for shortwave and 0.1 W/m² for longwave flux. Conversely, during periods of reduced SIE (negative anomalies), shortwave flux generally shows positive (warming) anomalies, while longwave flux generally shows negative anomalies, with corresponding means of 0.19 and −0.06 W/m² respectively (Figure 4A).

However, in West Antarctica, reduced SIE is not always associated with positive shortwave flux anomalies. During low-SIE periods such as the summers of 2016/17 and 2019/20, shortwave flux shows negative anomalies (cooling trend), while longwave flux shows positive anomalies (Figure 4C). The time-mean anomalies of shortwave and longwave flux during these summers are −0.46 and 0.14 W/m² respectively. This signal is related to increased cloud cover in the region (dotted blue line; Figure 4C) and is consistent with previous research (22, 33). These studies associate increased cloud cover related to poleward advection of warm and humid subtropical air driven by an enhanced ZW3 pattern.

Total heat flux in the East Antarctic shows brief positive anomalies from early 2013 to mid-2013, mid-2016, and from mid-2021 to 2023 (Figure 4B). In contrast, total heat flux in the West Antarctic shows longer periods of warm anomalies initially and from 2014 to late 2016, and again from mid-2018 to early 2020, and in 2023 (Figure 4D).

Warm anomalies are also evident in the net surface flux in 2023 (Figure 4D), resuming upper ocean warming in the West Antarctic (Figure 3D). In previous sea ice budget analyses, we pointed out discrepancies between SOSE and observations from 2022 and 2023, where SOSE failed to capture the observed sea ice decrease. However, the heat budget indicates that SOSE simulates a resumption of upper ocean warming during this period. This suggests that sea ice loss may follow subsequently. Rather than a fundamental inconsistency with observed trends, this highlights a possible delay in SOSE's sea ice response to upper ocean warming.

Changes in Salinity and Stratification in the Southern Ocean Upper Layer

Since vertical heat transport has been identified as a key driver of upper ocean warming, we now turn our attention to the processes controlling upper ocean stratification. In the subantarctic Southern Ocean, stratification is primarily set by salinity (53). Therefore, we begin by examining the salinity budget to identify the mechanisms governing stratification changes.

The East Antarctic upper ocean shows freshwater anomalies from 2013 to 2015 (Figure S7). Subsequently, the regional upper ocean becomes salinified (Figure 5A), accommodating the increase in upper ocean heat content. Part of the initial freshening may reflect increased equatorward sea ice export and melt (11), with subsequent salinification coinciding with a decrease in sea ice transport. In the following discussion, we show that upper ocean salinification is initiated by upward mixing of heat and salinity, followed by further contributions from reduced sea ice formation and associated equatorward export and melt.

However, in the West Antarctic, the increasing trends observed in the East Antarctic are absent. Instead, salinity on the West Antarctic shelf peaks in 2015 and 2016 and subsequently freshens, accompanied by enhanced stratification. Freshening and enhanced stratification are more pronounced over the West Antarctic shelf, while the outer shelf experiences slight freshening from 2019 to 2023 with associated enhanced stratification.

Salinity Budget of the East Antarctic Shelf Break

We now examine spatially averaged salinity and related budget terms to diagnose the causes of salinification over the East Antarctic and freshening over the West Antarctic. Salinification in the East Antarctic near-surface layer is associated with a decrease in upper ocean stratification (Δσ; defined as the potential density difference between 240 m and 0 m depth), reaching a minimum in 2023 (Figure 5A). The increase in upper ocean salinity is initially driven by the vertical advection term (grey line; Figure 5C and Figure S8C), which shows positive anomalies between 2013 and 2016. This term reflects the upwelling of saline waters, indicating a shoaling of the Circumpolar Deep Water (CDW) layer contributing to salinity increase in the upper 100 m. Vertical mixing also shows positive anomalies over the East Antarctic shelf break in 2015 and 2016 (blue line in Figure 5C). Horizontal advection also contributed to salinification over the East Antarctic shelf break during 2015–2016. Therefore, prior to 2016, both advective and diffusive terms played a role in enhancing vertical and horizontal transport of salinity. We will discuss the most likely drivers of this salinification pathway in the following section.

Both vertical diffusion and vertical advection contribute significantly to the salinity budget, but in the heat budget, vertical diffusion is dominant while vertical advection plays a smaller role. This difference arises because the vertical temperature gradient is approximately one to two orders of magnitude larger than the corresponding salinity gradient (Figure S10).

The surface flux term over the East Antarctic shelf break showed positive anomalies during 2013–2014, negative anomalies during 2015–2016, and positive anomalies thereafter. This pattern corresponds to SIP anomalies over the East Antarctic shelf break (Figure 2F), indicating that surface flux anomalies were driven by sea ice formation and transport variations. The positive anomaly in 2013–2014 reflects enhanced SIP, leading to increased salinity through salt expulsion. Conversely, the negative anomaly in 2015–2016 corresponds to reduced SIP, resulting in upper ocean freshening. Finally, the positive anomalies after 2016 are associated with a decrease in equatorward sea ice-derived freshwater transport (discussed further in the "Synthesis" section). Surface flux and SIP have a correlation of 0.68, with a 95% confidence interval of [0.64, 0.72] (square brackets denote the 95% confidence interval of the seasonally detrended and deseasonalized correlation throughout this paper). Upper ocean freshening resulted from both reduced salt expulsion and increased sea ice export from the East Antarctic shelf, which melts during summer and deposits freshwater into the outer shelf region.

Salinity Budget of the West Antarctic Shelf Break

Upper ocean salinity in the West Antarctic does not exhibit the increasing trend observed in the East Antarctic. Instead, salinity on the West Antarctic shelf peaks in 2015 and 2016 and subsequently freshens, accompanied by enhanced stratification. Freshening and enhanced stratification are more pronounced over the West Antarctic shelf, while the outer shelf experiences slight freshening from 2019 to 2023 with associated enhanced stratification.

West Antarctic freshening is explained by sustained negative anomalies in the vertical advection term (Figure 5D) over the West Antarctic shelf break during 2015–2021 and over the West Antarctic shelf during 2017–2019 (Figure S8D). Starting from 2020 and continuing through 2023, the vertical advection term shows signs of intensification in both the shelf and shelf break regions of the West Antarctic.

Surface Forcing on Salinity and Stratification Changes

To assess the drivers of upper ocean salinity and stratification changes underlying sea ice evolution, we re-examine the vertical and horizontal advection terms of the salinity budget and compare them with surface forcing from SOSE. Surface forcing for the ocean reflects the combined effects of wind and sea ice drift. Vertical salinity advection is expected to respond to the cyclonic circulation driven by surface stress anomalies, while upper ocean meridional salinity advection is expected to respond to meridional stress (54). This analysis is limited to the shelf break regions in both the East and West Antarctic. When calculating stress curls, the total (meridional and zonal) surface stress vector was considered, and positive (eastward) values were used only for calculating meridional stress anomalies. This approach is equivalent to using a spatial mask that selects eastward winds driving the proposed mechanism of sea ice loss (18).

As expected, during the period of enhanced circulation in surface stress (negative anomaly of ∇ × τ; Figure 6B), the vertical salinity advection term shows abnormally positive values (gray line in Figure 6A). Conversely, during the period of weakened circulation in surface stress (positive anomaly of ∇ × τ; Figure 6B), the vertical advection term shows negative anomalies. SOSE reveals zonal asymmetry in stress curl: East Antarctica experienced enhanced circulation from 2013 to 2017 and then weakened, while West Antarctica showed weak circulation from 2014 to 2017 and subsequently strengthened. This zonal asymmetry in surface stress curl corresponds to similar asymmetry in vertical advection between East Antarctica and West Antarctica from mid-2021 to 2023. Surface stress curl and vertical salinity advection in East Antarctica have a correlation of −0.26 [−0.33, −0.19], while in West Antarctica the correlation is −0.28 [−0.35, −0.21]. We point out here that Ekman pumping and vertical salinity advection are related through large-scale circulation, but their trends do not always coincide (Figure 6). Ekman pumping reflects the vertical velocity induced by the curl of surface wind stress, while vertical salinity advection depends on both this vertical velocity and the local vertical salinity gradient.

Zonal asymmetry is also found in the trends in zonal stress intensity above East and West Antarctica. Both regions show strengthening between 2014–2016 and during 2020–2023 (Figure 6B). However, meridional stress is generally stronger in East Antarctica between 2014–2016, while stronger in West Antarctica during 2020–2023. The horizontal advection terms above East and West Antarctica co-vary with temporal variations in zonal stress, suggesting that Ekman advection plays a role in driving horizontal salinity advection anomalies in the upper ocean. Horizontal advection of salinity and zonal meridional stress in East Antarctica are relatively strongly correlated at 0.62 [0.57, 0.67], while in West Antarctica they are relatively weakly correlated at 0.12 [0.05, 0.19]. Again, we point out that Ekman advection does not fully explain horizontal salinity advection. Other processes such as wind-driven baroclinic variability changes (55) and circulation changes (31) may also contribute.

To understand these results in the context of decadal changes in Southern Ocean climate, we analyze ERA5 wind fields (monthly averages at 10 m above sea level) for the offshore shelf regions of East and West Antarctica from 1970 to 2023 (Figure 6C). The zonal components of meridional winds are only considered in the 'WANT: u' and 'EANT: u' calculations. Wind curl and westerlies were calculated as anomalies relative to monthly averages, and we used a 10-year moving average to capture decadal variability. The subpolar Southern Ocean has experienced long-term strengthening of wind curl at both East and West Antarctica. Additionally, westerlies show a persistent positive trend, reaching the maximum magnitude during the 2010–2023 period. The long-term wind trend represents strengthening of processes driving upward transport of heat and salinity from the lower ocean.

Synthesis

The Antarctic SIE changes, upper ocean hydrology, and stratification changes documented in the previous section can be synthesized into three distinct periods: P1 (mid-2013–2014), P2 (2015–2017), and P3 (2018–2023). These periods were selected based on the temporal evolution in the East Antarctic offshore shelf, the major contributor to total Antarctic SIV changes (Figure 2B). P1 corresponds to the high SIV period. P2 marks the beginning of SIV loss. P3 captures the persistence of low SIV state with further decline.

During P1, the upper ocean shows freshwater anomalies over most of East Antarctica, while the West Antarctic region shows salinity anomalies (Figure 7; P1-Salinity). In P2, the magnitude of these anomalies generally decreases. P3 is characterized by salinity anomalies in East Antarctica and freshwater anomalies in West Antarctica. Salinity and freshwater anomalies are associated with decreased and enhanced upper ocean stratification, respectively.

During P1 (before 2015), the only budget term showing positive (salinification) anomalies throughout the subpolar Southern Ocean is the vertical advection term (Figure 7; P1-Vertical Advection). The surface flux term shows positive anomalies everywhere except in the Ross Sea, Amundsen Sea and Bellingshausen Sea offshore shelf regions, and the eastern Weddell Sea. Positive surface flux anomalies are consistent with increased salinity discharge due to enhanced SIP during this period.

P2 (2015–2017) is characterized by increases in upper ocean salinity in most of East Antarctica coinciding with sea ice loss. The vertical advection term continues to salinify the upper ocean over a wide region of East Antarctica during this period, consistent with the increased circulation of surface stress curl. Surface flux shows freshening tendency near the sea ice edge, consistent with increased sea ice import from the shelf and subsequent melting in the offshore region. Surface flux also shows salinifying tendency inside the sea ice pack in East Antarctica. Generally, Ekman advection (component of horizontal advection) transports colder and fresher water northward from higher latitudes. However, during P2, horizontal advection partially offsets surface flux by increasing salinity along the northern edge of the offshore shelf region while showing freshening tendency in more southern regions. This pattern occurs together with positive anomalies of zonal surface stress and strong negative anomalies of surface stress in West Antarctica during 2016. P2 is also characterized by increased vertical diffusion over much of the subpolar Southern Ocean (Figure 7; P2-Vertical Diffusion), which is likely to promote vertical mixing associated with surface stress strengthening.

▶ Source: https://www.science.org/doi/10.1126/sciadv.aeb0166