Polynyas are inherently linked to sea ice cover, which influences Earth's albedo and modifies the exchange of heat, freshwater, momentum and gases between the atmosphere and ocean. The formation of open-ocean polynyas is influenced by regional scale sea ice conditions, particularly in the Ross, Amundsen, and Weddell seas .

Since 1979, Antarctic sea ice concentration (SIC) has been closely monitored by satellites. Between 1979 and 2014, an increasing trend was observed in the Antarctic sea ice extent (SIE: the total area of sea ice coverage with a SIC greater than 15 %) , with varying regional trends. However, a record minimum SIE was observed on 21 February 2023, about 1.05 × 10⁶ km² below the 1981–2010 average , and 136 000 km² below the previous record low reached on 25 February 2022 . The record minimum follows a sharp decline in SIE and SIC in 2016 . The sustained negative sea ice area anomalies since 2016 indicate a possible transition to a negative trend , akin to what is already evident in the Arctic, and consistent with global warming.

Processes involved in circumpolar sea ice loss also favor conditions that enable polynya formation. The Weddell Sea has notably experienced a significant decline in SIE since 2016 . This decline is attributed to factors such as a positive El Niño-Southern Oscillation (ENSO), an enhanced Zonal Wave Number 3 pattern , and the unprecedented occurrence of severe storms and atmospheric rivers . The reduction was also accompanied by the reopening of the polynya near the Maud Rise in 2016 and 2017 (Fig. b), with each of these events being significantly larger than any since the 1970s extreme events (Fig. ). The sea ice loss can cause a robust weakening of the tropospheric westerly jet and favor the negative phase of the Southern Annular Mode (SAM) . The further decline in sea ice in 2022 and 2023, especially in the Bellingshausen Sea region, highlights the impact of regional warming and atmospheric and oceanic drivers on SIE .

In addition to SIC, sea ice thickness is also used to identify polynyas, especially coastal polynyas. However, unlike SIE, there is an absence of comprehensive, long-term records of sea ice thickness across Antarctica to detect polynyas and explore the sea ice processes surrounding WSP and MRP occurrences in particular. Despite the technical challenges involved in measuring sea ice thickness and volume, recent satellite observations have identified instances of notable thinning of sea ice up to three months ahead of the opening of Maud Rise polynyas in 2016–2017 .

Numerical models augment satellite observations as a valuable tool for analyzing Antarctic sea ice variability. The ensemble mean among historical simulations from CMIP6 models is close to Antarctic SIE observations, especially since most CMIP6 models can simulate multi-year ice in the Weddell Sea . However, these global climate models overestimate Antarctic sea ice loss and variability over 1979–2018 . Moreover, many CMIP6 models simulate unrealistically frequent open-ocean polynyas in the Southern Ocean , although there has been slight improvement over the previous CMIP phase.

The most dramatic observation of an open-ocean polynya occurred during the winters of 1974 to 1976 when a large opening with a maximum spatial extent of 350 000 km² was detected from a passive microwave satellite within the ice pack of the eastern Weddell Sea (Fig. ). It initially appeared near the Maud Rise seamount and persisted throughout the three consecutive winters . summarized that the WSP occupied the northeastern flank of Maud Rise in 1974 and extended into the northwest flanks in 1975, while the polynya quickly drifted southwestward into the Weddell Sea in 1976 . Studies have shown that MRPs can be a precursor for WSP occurrence . MRPs occur more frequently but irregularly and the largest and most recent openings occurred in the winter seasons of 2016 and 2017. The opening in July–August 2016 lasted for 21 d and reached an open-water area of 33 000 km², while the opening in September 2017 lasted 1.5 months and expanded to 50 000 km² . Sea ice properties of the 2016 and 2017 MRPs were observed by the Advanced Microwave Scanning Radiometer 2 (AMSR2) satellite , while oceanic measurements within the 2017 polynya were limited to observations from one Argo float . Unlike the WSP in the 1970s, which propagated westward from Maud Rise at an average velocity of 0.013 m s⁻¹ deduced from satellite images , recent MRPs remained in the vicinity of Maud Rise. Furthermore, whereas the observed WSP existed from the beginning of the freezing season , MRPs typically occur later in the season, near the end of winter, just before the retreat of the sea ice edge.

Although the specific impact of a large open-ocean polynya on paleoclimate records (e.g., ice cores) is not well known, some of these records likely have a signal related to open-ocean polynya formation. For example, the WSP of the mid-1970s is estimated to have caused an annual mean warming of almost 1 °C in coastal regions, and an additional snow accumulation averaged over the sector of about 10 Gt yr⁻¹ on the continent, at least in the sector between roughly 50° W and 0° E . , therefore, used ice core records to reconstruct the occurrence of polynyas. They found that large open-ocean polynyas, such as those observed in the 1970s, are rare events, occurring at most a few times per century, and typically lasting only a few years without clear periodicity. However, it is important to note that these results are limited by the proxies used, such as oxygen isotopes, which are influenced by precipitation in specific locations and are unlikely to capture the occurrence of smaller polynyas. Although these findings remain somewhat ambiguous , they underscore the potential for further reconstructions of past polynya activity using other paleoclimate proxies.

Open-ocean polynya formation is thought to be associated with large-scale atmospheric changes, although it remains unclear whether these changes are a cause or a consequence of polynya events. In particular, the SAM was in its negative phase before the observed WSP event and has been linked to open-ocean polynya occurrence . Reconstructions of the SAM over the last millennium might thus indicate a relatively high probability of polynya occurrence during 1350–1400 and 1600–1650. Similarly, long-term variations in the rate of AABW formation could be partly linked to WSP frequency , and it has been suggested that decreased open-ocean polynya activity might have slowed AABW production over the 20th century .

The Weddell Sea is characterized by a seasonal pycnocline, formed by surface warming and sea ice melting in summer, and a permanent pycnocline that ensures vertical stability of the water column . From April to June, the seasonal pycnocline disappears due to heat loss to the cold atmosphere and brine rejection (Fig. ), and the permanent pycnocline shoals and weak . The permanent pycnocline separates cold and fresh surface waters from warmer and saltier deep waters, often referred to as Warm Deep Water (WDW). During winter, the salinity difference between surface waters and WDW (i.e., the halocline) controls the resistance to deep convection across the Weddell Sea .

The winter halocline, or permanent pycnocline, is key to polynya formation and maintenance. To generate and maintain a polynya, the subsurface heat of WDW must be continuously brought to the surface by vertical mixing . This upward heat transport can erode the pycnocline , which is facilitated by both surface forcing and internal circulation processes. In examining polynya mechanisms, we differentiate between immediate short-term triggers (days to weeks before a polynya) versus long-term processes (active over months to years before a polynya). The long-term processes gradually erode the stratification of the water column, creating conditions that are favorable for polynya formation, referred to as “preconditioning”. Drawing from previous polynya studies based on observations, process-oriented models and Earth System Models (ESMs), we will discuss the roles of WDW inflow, flow-topography interactions, and surface forcing in the formation of WSPs and MRPs.

The properties of WDW within the Weddell Sea are influenced by water mass transformation within the region , variations in the inflowing water masses , and changes in the inflow's strength . The WDW around Maud Rise plays a crucial role in preconditioning a polynya, with its properties significantly modulated by the eastern limb of the Weddell Gyre, which traverses the Maud Rise . The Weddell Gyre variability consequently affects the stratification and the amount of heat contained in the subsurface ocean (Fig. ). This variability occurs on both seasonal and interannual timescales. Seasonally, the gyre weakens during winter as expanding sea ice diminishes ocean surface stresses . On interannual timescales, variations in the Weddell Sea Low, a region of low pressure located east of Maud Rise, influence the cyclonicity of winds in the eastern Weddell Sea (Fig. ). A deeper Weddell Sea Low enhances Weddell Gyre circulation and directly impacts the eastern limb of the gyre . This gyre spinup enables the shoaling of the pycnocline, bringing the warm and saline WDW layer closer to the mixed layer. This proximity facilitates greater mixing of salt into the mixed layer, eroding the stratification in the years preceding a polynya event .

The impingement of the Weddell Gyre on Maud Rise creates a local dynamical system, where a Taylor Cap is situated atop the Maud Rise and a warm-water Halo wraps around the northern half of Maud Rise . The warm-water Halo is distinguished by a relatively warm subsurface layer with maximum temperature exceeding 1 °C, lying below a notably shallow surface mixed layer. In contrast, the Taylor Cap exhibits relatively cold subsurface temperature and a deep mixed layer, accompanied by the least stable stratification in Maud Rise . A circulation is observed in the Halo, characterized by a strong surface-intensified jet with westward velocities exceeding 0.1 m s⁻¹, accompanied by an eastward retroflection current to the south of this jet . Due to its isolation from the Gyre's flow, the Taylor Cap functions as a temporary local reservoir of heat and salt.

Large scale changes in the Weddell Gyre circulation also affect stratification locally around Maud Rise. These processes encompass flow-topography interactions, as well as surface and internal forcing. For example, flow-topography interactions generate eddies on the Maud Rise flank, which favor upwelling of warm subsurface water into the mixed layer, providing the necessary heat to melt the sea ice . Additionally, the local shedding of eddies (Fig. ) can lead to a divergent Ekman stress that forces the sea ice apart and moves water masses from the flanks to the center of the seamount. Because the flushing timescale over the dynamically isolated seamount far exceeds the flushing timescales over the more rapidly flowing flanks, this allows for the buildup of heat and salt anomalies over the Taylor Cap, as observed in the years 2015 through 2018 . However, persistent enhanced vertical mixing and weak stratification in the subsurface may prevent the buildup of the heat reservoir via continuous, gradual heat flow toward the surface, potentially inhibiting MRP formation . The surface forcing, such as storms (Fig. ), can generate turbulence at the base of the mixed layer, enhancing vertical mixing . In terms of internal forcing, observations reveal intrusions between the Taylor Cap and northern flank of Maud Rise, characterized by water mass interleaving . These intrusions suggest lateral and vertical mixing along and across sloping isopycnals, transporting the flank WDW to the mixed layer base and subsurface layer over Maud Rise. Within this area, the structure of cold surface water overlying WDW, combined with great pressure and temperature gradients along the isopycnals, facilitates thermobaric effects, leading to convective plumes penetrating the warm deep layer. These plumes, in turn, cause the warm deep water to ascend, eroding the mixed layer and destabilizing the surface layer, potentially triggering abrupt deep convection . Moreover, these intrusions create conditions conducive to both types of double-diffusive instability , further promoting the mixing of water masses.

With the onset of deep convection, heat from the subsurface layer is transferred to the mixed layer, leading to the melting of sea ice. Melting acts to restratify the water column, cutting off vertical mixing. This negative feedback hinders the formation of polynyas, even though the water column is very weakly stratified in most years over the Maud Rise. To sustain convection, an additional input of salt to the surface is necessary. For example, the 2016 and 2017 MRPs were initiated on the northern flank as Ekman advection transported additional salinity from the Taylor Cap to the northern flank between 2015 and 2018 . Deep convection events from preceding years also result in preconditioning by decreasing the vertical stability of the water column, making it more vulnerable to subsequent triggers .

Some of the observed processes were confirmed by the various modeling approaches of the scientific community. Typically, ESMs, ocean general circulation models (OGCMs) or reanalysis models are employed to investigate the polynya and its formation processes. Each modeling approach has its own set of limitations that should be borne in mind when interpreting the results. The scales of polynyas generated in ESMs generally far exceed those observed, yet ESMs still inform about theoretical ways of generating open-ocean polynyas . Some global OGCMs are able to represent the WSP on reasonable scales , but MRPs are hardly reproduced and atmospheric feedbacks are absent in such models.

The likelihood of a polynya opening at Maud Rise is determined by the “preconditioning” that the water column can undergo on various timescales ranging up to multiple decades. Coupled ocean-atmosphere models indicate a periodic buildup of heat below the pycnocline in the Southern Ocean on multidecadal timescales preceding polynyas . Once preconditioned, the polynya and the associated deep convection can be triggered and sustained by the advection of surface salinity anomalies .

In ESMs, the formation of the larger WSPs is often preceded by the occurrence of MRPs and is favored by preconditioning on multidecadal timescales. WSPs that far outgrow MRPs require a spin-up of the Weddell Gyre . The opening of the WSP is maintained by the accumulated heat in the Weddell Gyre . Besides the heat accumulation, a negative SAM favors the formation of WSPs by conferring a saltier surface ocean through a drier atmosphere . A negative SAM results in a saltier surface in the Weddell Sea, and under certain circumstances, the smaller MRP can be advected westward by the Weddell Gyre to form a WSP.

Building on the similarities between observational and modeling studies, a roadmap to polynya formation can be created. The first stage is a weakening of the local stratification at Maud Rise via long-term preconditioning processes. Then, the onset of near-term mixing processes leads to catastrophic deep mixing bringing the heat from depth to the surface (Fig. ), resulting in the opening of the sea ice pack. Occasionally a brief opening is observed in the first winter, as for example in 2016, which leaves the water column weakly stratified due to the associated deep mixing. In the following winter, reduced stratification increases the likelihood of another polynya opening, allowing processes such as the advection of surface salt anomalies to easily trigger deep convection. The re-occurrence of the polynya is determined by the remaining heat at depth, the SAM, and whether the weak stratification is replaced by a stronger one via the advection of more stratified water masses by the Weddell Gyre.

Polynya formation is a coupled ocean-atmosphere process, with many studies suggesting various atmospheric triggers. Large-scale atmospheric variability, such as the SAM and Zonal Wave patterns, plays a foundational role in preconditioning the system, while localized processes such as cyclone activity and atmospheric rivers often act as immediate triggers.

Sustained open-ocean polynyas are associated with deep convective mixing, which brings old deep waters into contact with the surface while injecting young ventilated waters into the ocean interior. Both effects have major implications for marine biogeochemistry: upwelling of deep waters brings remineralized carbon and nutrients to the surface, while downwelling of surface waters brings oxygen and preformed nutrients to the deep ocean. Notably, deep convective mixing does not have the same impact on ocean ventilation as gravity currents on the continental slope. Slope convection carries dense shelf waters down the continental slope to the abyssal ocean, but does not directly expose old deep waters to the surface. This distinction means that shelf and open-ocean modes of AABW formation do not have the same repercussions on marine biogeochemistry. Deep convective mixing such as that observed in the mid-1970s also transfers substantial amounts of heat from the deep ocean to the atmosphere , affecting global deep ocean temperatures , regional stratification and sea ice coverage , all of which will ultimately impact air-sea gas exchange, nutrient distributions and marine ecosystems.

In situ observations collected in the aftermath of the 1970s WSP event revealed chimneys of anomalously high dissolved oxygen concentrations in the region reaching depths of 4000 m , with concentrations >5.6 mL L⁻¹ c.f. 5.0–5.4 mL L⁻¹ in surrounding waters . These observations unambiguously reveal the transfer of oxygen into deep layers of the subpolar Southern Ocean, including mid-depth layers that are customarily deprived of oxygen. Unfortunately, in situ observations from the 1970s do not allow to quantify the impact of Southern Ocean oxygenation at larger scale, nor do they provide information on nutrient and carbon fluxes associated with the WSP. Recently, Argo floats equipped with oxygen sensors profiled in the Maud Rise region; the measurements reveal high oxygen anomalies to about 1000 m depth (e.g., ±10 µmol kg⁻¹ at ca. 650 m) following the opening of MRP in 2017 , confirming the effectiveness of polynya events in ventilating deep waters.

Polynya impacts on ocean-atmosphere carbon exchange have been explored using a coupled climate model that simulates strong, intermittent deep convection events in the Weddell Sea . These model experiments show that each polynya event leads to substantial outgassing of natural remineralized carbon. Convection also leads to cooling of WDW which means this water can hold more preformed carbon. Thereby, ingassing of anthropogenic CO₂ can also occur and can dominate over outgassing if atmospheric partial pressure of CO₂ (pCO₂) is sufficiently high . Depending on whether the outgassing of remineralized carbon or the subduction of preformed carbon dominates, the concentration of dissolved inorganic carbon in WDW will either decrease or increase. The change in dissolved inorganic carbon in WDW induced by the polynya will then change the lateral transport of carbon out of the Weddell Gyre, which normally is the component that leads to the gyre acting as a net sink of CO₂ . Under a high-end anthropogenic climate change scenario, the model used by simulates a cessation of deep convection events, around the end of the 20th century . This change in circulation and mixing tends to increase natural carbon storage (reduced outgassing of remineralized carbon) but also to reduce anthropogenic carbon uptake (reduced ingassing) over the 21st century, resulting in net weakening of cumulative ocean carbon uptake by 2100 . Hence, the impact of WSP events on ocean-atmosphere carbon partitioning depends on a subtle balance that evolves over time with atmospheric pCO₂ and polynya frequency. In the model used by , the cessation of Weddell Sea convection contributes respectively 5 % (-4.3 ± 1.9 PgC) and 18 % (-10.1 ± 3.9 PgC) of the global climate-induced reduction in total and anthropogenic ocean CO₂ uptake, even though it only constitutes 1 % of the global ocean area.

Polynya-induced modifications to nutrient distributions (Fig. ) can further alter the carbon cycle and its response to future change by affecting primary production and export. Phytoplankton growth in the Weddell Sea, as in the wider Southern Ocean, is primarily limited by the availability of light and iron . Manganese as a micronutrient has also been shown to play a co-limiting role in the Weddell Sea . Deep winter mixing supplies iron and other nutrients to the surface , thus spring blooms are more likely to be limited by light, while iron limitation sets in later in the growing season . Around the Maud Rise, chlorophyll (Chl-a) levels are ∼ 2–4 times higher than in the surrounding areas, with mean annual maximum levels (2002–2018) exceeding 2.7 mg m⁻³ , likely due to topographically-induced mixing . Here, iron-limited spring blooms have been reported , although argue that this is likely not the case during a polynya event with ongoing deep convection and associated iron supply. argue that the increased iron supply can sustain higher primary productivity in the region in polynya years c.f. non-polynya years. This could, in turn, offset some of the increase in surface pCO₂ caused by upwelling of waters over-saturated in remineralized carbon.

Due to the rare occurrence of open-ocean polynya in this region, observational studies of their effect on ocean productivity are scarce. Phytoplankton bloom intensity, duration and timing in the MRP area were observed with autonomous biogeochemical floats for four years starting late 2014 , thus covering the polynya years 2016–2017. While the bloom intensity in terms of column inventory of chlorophyll was not increased in polynya years, the onset of the bloom happened earlier in the year due to increased light availability. Satellite data confirm that this was associated with unprecedented surface levels of Chl-a for this time of year, reaching up to 4.7 mg m⁻³ during 2017 . Meanwhile, show that an early onset of the bloom leads to a prolonged growing season, allowing for a distinct iron-limited autumn season bloom that can only occur when the growing season is sufficiently long. further demonstrate that a longer growing season allows for increased annual net community production by up to 100 % over Maud Rise, and thus suggest that the open-ocean polynya likely leads to enhanced carbon export to the ocean interior. However, they emphasize that similar behavior can be triggered by other processes causing early sea ice retreat in the region and is not unique to the polynya. By the same mechanisms, open-ocean polynyas are suggested to have contributed to biological carbon sequestration in the glacial ocean , and polynyas are suggested to have been a key environment for survival of phytoplankton species in glacial Antarctic waters .

The processes described in Sect.  and the many links between them indicate that the presence of polynyas significantly influences air-ice-ocean interactions. A clear understanding of these processes and interactions is crucial for determining the role of high latitudes in global climate regulation, particularly regarding deep ocean ventilation. For a long time, this has remained one of the major challenges facing contemporary polar research .

provided an initial comprehensive overview of polynya dynamics, ranging from observational insights to modeling approaches. The study further delved into their interactions with atmospheric and oceanic boundary layers, and their contributions to the surface heat, freshwater, and ice budgets. This included detailed analyses of in situ and remote sensing data on polynya locations and seasonality, as well as the atmospheric and oceanic conditions that foster their formation and maintenance. Studies reviewed in this work pointed towards an interplay between anomalous wind fields, heat and freshwater fluxes, sea ice divergence, and ocean flow interaction with local topography at Maud Rise in forming and sustaining the WSP . Despite the efforts of and the numerous studies cited in their review, questions remained about the role of polynyas in shaping dense water production and air-sea heat fluxes that contribute to global ocean and atmospheric circulation patterns. Future climate change effects on polynyas, and thus on any atmosphere-ice-ocean processes influenced by their presence, were also unknown (see Sect. ).

Recent studies using coupled models to observe the two-way interactions of the polynyas show that there may be links to distant climate, such as the Interdecadal Pacific Oscillation (IPO) . However, these models have polynyas larger than those observed, which may lead to a greater heat flux to the atmosphere (Fig. ), and therefore a stronger response . In their study of the atmospheric response to polynya, noted that polynya occurrences not only affect the regional radiative heat balance, but also the local precipitation and cloud formation, albeit to a limited extent. On the other hand, climatic responses are not necessarily achieved solely through an atmospheric mechanism. Deep convective mixing and subsurface ocean cooling may trigger important changes in ocean circulation and hydrography. Some studies suggest an increase in AABW formation by open-ocean polynyas like the WSP of 1974-1976 . Models and observations also indicate that the WSP causes transient cooling and freshening of the deep Southern Ocean and, with some lag, of the global ocean abyss . A minor component of the observed abyssal warming since the 1990s may be linked to the recovery from the mid-1970s WSP events . The proposed mechanism is that the deep convective event associated with the WSPs injected an anomalously cold, dense water mass into the abyssal Southern Ocean. As this water gradually mixed, upwelled, and equilibrated over subsequent decades, it could have contributed to a delayed warming signal, consistent with deep ocean adjustment timescales in multi-century coupled model experiments .

Observational data on polynyas, particularly for the 1974–1976 WSPs, are notably scarce. Recent technical advances have significantly improved our ability to observe interactions within the polynya, especially with the wealth of data available on the 2016–2017 MRPs. For example, data from tagged marine seals and high-frequency profiling floats from the Argo program or the SOCCOM (Southern Ocean Carbon and Climate Observations and Modeling) program provide valuable insights into vertical water stratification, as well as the structure of ocean fronts and cross-front mixing . In addition, other observations have enabled detailed investigations into sea ice thickness – a key indicator of air-ice-ocean interactions – within polynyas using satellite technology. Notably, techniques such as L-band microwave radiometry from the Soil Moisture Ocean Salinity (SMOS) and Soil Moisture Active Passive (SMAP) satellites allow for thickness measurements of thin sea ice in these regions . The detailed observations acquired during the 2016–2017 MRP events will help to ground-truth the next generation of fully coupled atmosphere–ice–ocean general circulation models. highlighted the critical role of fully coupled atmosphere–ice–ocean general circulation models in more accurately simulating the formation processes of open-ocean polynyas. The advent of the latest high-resolution ESMs has opened new avenues for studying air-ice-ocean interactions within polynyas, as demonstrated by recent studies . For instance, employed statistical causal inference techniques within a high-resolution, fully coupled climate model to investigate how the MRP influences or reacts to climate variability. Their research established a clear link between polynya heat loss and various indicators of atmosphere–ocean variability in the high-latitude climate system.

Both the WSP and MRP events illustrate the link between deep convection and interannual high-latitude climate variability. These events are influenced by both natural climate variability and human-induced trends, which affect ocean, ice, and atmospheric conditions that in turn control the formation of polynyas and deep convection in the Southern Ocean. For example, under preindustrial conditions, several climate model simulations exhibit natural oscillations in deep convection occurring from interdecadal to centennial timescales . Observed variations in WSP and MRP occurrences since the mid 1970s could be linked to internally generated or externally forced climate variability . As an example, the 2016 MRP event has been linked to an anomalous deepening of the Amundsen Sea Low , itself caused by the combination of a strongly negative Southern Annular Mode and La Niña-like sea surface temperature anomalies in the tropics . Deep convection in the open Southern Ocean could also drive internal variability in the ice-ocean-atmosphere system at high southern latitudes . Indeed, it may be an important mechanism regulating the Southern Ocean heat balance on interannual to centennial timescales .

Feedback mechanisms and atmosphere-ocean-ice variability in the polynyas are evident in the preindustrial control simulation of the Energy Exascale Earth System Model (E3SMv0-HR). According to , during polynya years, upward turbulent heat fluxes shift toward the poles, leading to increased ocean heat loss (Fig. ) from the Weddell Sea, a reduction in the meridional temperature gradient, and an intensification of the cyclonic wind circulation in the Weddell Sea. This indicates that open-ocean polynyas in E3SMv0-HR can both influence and respond to atmospheric variability in the region. Another pre-industrial simulation, described by and based on the high-resolution (0.1° for the ocean and sea ice components) configuration of the Community Earth System Model version 1.3 (CESM1.3), showed the plausibility of multidecadal climate variability of WSPs, with occurrences approximately every 40–50 years. However, these WSP features are notably absent in low-resolution CESM simulations. also found that multidecadal variability of WSPs is not only closely correlated with an IPO-like pattern in sea surface temperatures but also precedes the IPO by about 4 years. This suggests interactions between Southern Ocean sea ice and tropical Pacific climate variability . While high-resolution climate simulations offer significant advantages, they are computationally demanding and energy consuming, making it challenging to link equilibrium climate sensitivity and polynyas from these simulations. Some challenges including the model resolution are further discussed in Sect. .

Interestingly, in NOAA's Geophysical Fluid Dynamics Laboratory (GFDL) CM2G coupled climate model, found that when WSPs are forced regularly, it takes 18–25 years for Antarctic Bottom Water (AABW) volume transport anomalies to reach the South Atlantic, consistent with findings from the Norwegian Earth System Model (NorESM-OC1.2) under normal year forcing . In a 2000-year preindustrial control simulation from CM2G, it was shown that WSP recovery could account for 10 % ± 8 % of the recent warming in the abyssal Southern Ocean , with the strongest warming signals observed in the South Atlantic and Southern Ocean basins, diminishing with distance from the Weddell Sea .

utilized a coupled ocean–atmosphere model which simulates multidecadal variability in Southern Ocean deep convection resembling the observed 1970s WSP. They demonstrated that the model can broadly reproduce observed trends in sea surface temperature and SIC over 1979–2012 when the model is initialized in an active phase of open-ocean convection. They suggested that the natural multidecadal variability of Southern Ocean deep convection could influence the detectability of surface temperature and sea ice responses to global warming. Echoing the findings of , analyzed 41 CMIP6 models and further discovered that those with strong Southern Ocean deep convection in their pre-industrial state tend to show greater warming at ocean depths but less surface warming under greenhouse gas forcing. This leads to a more negative short-wave cloud feedback. The analysis indicates that the feedback from marine boundary layer clouds and deep ocean warming are interconnected processes within a coupled system, which together account for a significant portion of the variation in effective climate sensitivity across different models .

Despite substantial advances in observations, modeling, and theoretical understanding, open-ocean polynyas, particularly in the Weddell Sea, remain among the least understood components of the Southern Ocean climate system. Limitations persist across observational records, model parameterizations, and process representations, hindering our ability to predict their occurrence and assess their climate relevance.

The identification and characterization of polynyas in models and observations are themselves subject to uncertainty. In particular, the findings underscore how polynya detection is highly sensitive to the choice of sea ice metrics (SIC vs. SIT), and the temporal scale of analysis. Winter polynyas, for instance, may be masked in SIC fields due to rapid refreezing, whereas SIT thresholds can better capture low-ice regions that remain important for ocean convection. These differences complicate direct comparisons between satellite observations and model output, often leading to apparent inconsistencies in polynya detection and frequency. A coordinated use of both metrics – SIC and SIT – alongside improved model diagnostics will be essential for more robust future assessments.

Observational uncertainty is further exacerbated by the rarity and remoteness of large WSP events. While the 2016–2017 MRPs are relatively well-documented via reanalysis, satellite, and Argo float data, the 1970s WSP events remain sparsely observed, with little to no direct in situ information on vertical structure, mixing, or biogeochemical fluxes. This limits our ability to quantify key processes such as brine rejection, deep water formation, and heat flux partitioning, particularly during polynya onset and decay phases.

On the modeling side, CMIP-class models tend to produce either overly persistent or unrealistically large polynyas, or fail to simulate them at all. Differences stem from a wide range of factors including vertical mixing schemes, representation of eddies and dense water overflows, sea ice initial conditions, and parameterizations of air–sea–ice coupling . For example, coarse-resolution models often underestimate the restratification effect of mesoscale eddies, while high-resolution models that include eddy-resolving capability better maintain upper-ocean stratification and can produce more realistic, intermittent polynya events . The sensitivity of WSP behavior to these factors has been confirmed by inter-model comparisons and perturbation experiments .

Furthermore, the potential biogeochemical impacts of WSPs remain underexplored. Few models include detailed representations of carbon or nutrient cycling within polynyas, and observational coverage of relevant fluxes (e.g., iron, dissolved inorganic carbon, alkalinity) is limited. As a result, the broader role of WSPs in Southern Ocean biogeochemistry and global carbon storage remains speculative, particularly under changing climate conditions.

To clarify and compare key differences in formation, magnitude, and impacts across WSP and MRP studies, we provide a synthesis in Tables  and . The tables highlight observational versus modeled results across key metrics including formation mechanisms, ice production, heat fluxes, and broader climate feedback.

The past frequency of open-ocean polynya events in the Weddell Sea is largely unknown . Forecasting the future frequency and magnitude of such events is thus a major challenge. Nonetheless, it is possible to outline the likely future changes in environmental conditions that favor or hinder polynya formation. Naturally, these changes depend on the future trajectory of greenhouse gas concentrations and the ensuing level of global warming. In addition, the transient century-scale response may differ from the longer term response . In the following discussion, we focus on climate projections over the 21st to 23rd centuries under high-end emissions scenarios.

The first key environmental factor to consider is the surface freshwater input at southern high latitudes. Climate models such as those participating in CMIP5 and CMIP6 indicate that precipitation over the high-latitude Southern Ocean has increased over recent decades and will likely continue to increase in the future . Additionally, future mass losses of the Antarctic ice sheet could deliver substantial additional freshwater to the Southern Ocean surface . Hence, it may be expected that salinity stratification in the high-latitude Southern Ocean will intensify and thereby impede future open-ocean convection events . CMIP5 models that simulate open-ocean convection in the Southern Ocean provide support for this scenario: they predict a decline in convection over the 2000–2300 period under strong greenhouse forcing . This decline occurs in all such CMIP5 projections despite their exclusion of ice sheet mass losses, and despite simulated changes in heat fluxes, sea ice and winds.

Other lines of reasoning could overturn this projected decline. First, it is possible that CMIP5 models overestimate sensitivity to freshwater fluxes due to biases in ocean circulation and mixing . For example, the weakness of Antarctic slope currents in such coarse models may lead to overestimated freshwater transport from Antarctic margins to the open Southern Ocean . Second, ESMs could underestimate the influence of future wind changes, due to poor representation of relevant processes such as mesoscale eddies or wind-driven mixing. As the SAM trends towards more positive under greenhouse forcing, atmospheric storms over the Weddell Sea may become stronger or more frequent, potentially contributing to destabilize the water column in the Maud Rise region . In addition, a positive SAM will intensify negative wind stress curl and the gyre within the Weddell Sea , activating mesoscale eddies near Maud Rise and potentially promoting polynya events . Third, future changes in temperature stratification could overcome changes in salinity stratification . For example, subsurface WDW in the Weddell Sea may warm in response to increasing meridional heat transport within an intensifying Weddell Gyre . Rapid build-up of the subsurface heat reservoir could ultimately favor stronger or more frequent convection events .

Several modeling studies point to a potential future increase of open-ocean polynya events. By simulating surface wind stress, Antarctic meltwater, and their combined forcing in a pre-industrial control simulation with a global climate model and an ESM, found the diminished sea ice cover, in combination with wind stress perturbation, enhances heat loss at high latitudes through enhanced upwelling of WDW and triggering of polynya events in the Weddell Sea. Similarly, showed that combining positive momentum, heat, and freshwater flux perturbations in the Australian Community Climate and Earth System Simulator Ocean Model (ACCESS-OM2) leads to significant sea ice reduction, expansion of the Ross and Weddell polynyas, and increased associated deep convection. These polynyas cease when the heat reservoir is depleted. A mixed scenario featuring a near-term increase in polynya activity followed by a decline is also a possibility .

In essence, the future formation of open-ocean polynyas in the Weddell Sea or the wider Southern Ocean can be understood as a competition between opposing forces. On the one hand, there is the stabilizing influence of relatively fresh surface waters, influenced by surface freshening and thermodynamic effects linked to positive SAM and possibly increased ice sheet melting . On the other hand, there are destabilizing factors such as the frequency of polar cyclones, the strength of the Weddell Sea gyre, and meridional heat transport, which can promote and trigger convection . Because ocean stratification in this region is controlled by a subtle balance of competing processes that are difficult to accurately model, the future of MRPs and WSPs in a warmer world remains highly uncertain.