Author/s Kyle Clem1, Rob Massom2, Sharon Stammerjohn3, Phillip Reid4 Victoria University of Wellington, New Zealand Australian Antarctic Division, Australian Antarctic Program Partnership, and Australian Research Council Australian Centre of Excellence in Antarctic Science, Tasmania, Australia Institute of Arctic and Alpine Research, University of Colorado Boulder, USA Australian Bureau of Meteorology and Australian Antarctic Program Partnership, Tasmania, Australia Brief Overview Since reliable and continuous satellite records began in 1979, there have been strong regionally- and seasonally-varying patterns of change and variability in sea-ice extent around Antarctica (Fig. 1) – in contrast to a largely uniform loss of sea ice across the Arctic. Notably, the region west of the Antarctic Peninsula to the eastern Ross Sea has experienced significant sea ice loss in concert with a strengthening of the Amundsen Sea Low and increased warm northerly winds. On the western Peninsula, the sea ice loss and associated regional warming has led to dramatic and complex ecosystem change (see Antarctic Sea Ice #2) and has also been implicated in major ice-shelf disintegration events on the Peninsula (see Antarctic Sea Ice #1). Elsewhere, sea-ice coverage has expanded but with substantial interannual variability. The sum of these differing regional and seasonal contributions is a slight increasing trend in overall sea-ice coverage of 1.0 ± 0.5% per decade (or about 11,300 km2 per year) for 1979–2018 (Fig. 1a). Since 2012, sea ice in the Antarctic has undergone rapid and unanticipated swings in its net overall coverage, to first record high (2013-2015) then record low (2016-2022) coverage. Determining the drivers and effects of these abrupt shifts, together with differing seasonal and regional contributions, is a major focus of current research. This represents a substantial challenge given the complex processes involved in ice-ocean-atmosphere interactions and feedbacks. But better understanding of these interactions and feedbacks is crucial to improving the representation and simulation of sea-ice coverage and seasonality in climate and Earth System models. This is pivotal to improving near-term forecasting and long-term projections of Antarctic sea ice and the effects of sea ice change on the coupled climate-ocean-biosphere system in the coming decades. Sea-ice coverage is projected to significantly decrease in the Antarctic by the end of this century in response to anthropogenic climate change, but there is substantial uncertainty in current model estimates of areal loss. These estimates range from ~30% to 90% reductions in February (annual minimum, current average is ~2-4 million km2) to 15% to 50% reductions in September (annual maximum; current average is ~19-20 million km2). A major gap however is a lack of information on sea ice and snow thickness distributions, let alone whether these are changing. Antarctic sea ice loss is expected to have wide-ranging effects that include: warming of surface waters of the Southern Ocean, and subsequent changes in its seasonal freshening and stratification and thus its capacity to take up and store atmospheric heat and CO2; decreased production of Antarctic Bottom Water with effects on global thermohaline (overturning) ocean circulation; change in ocean-ice sheet interactions contributing to sea-level rise; potential changes in regional weather patterns (including across southern mid-latitudes); and changes in Southern Ocean species distributions, ecosystems and biogeochemical cycles and processes (see Antarctic Sea Ice #2). Detailed Overview Antarctic Sea Ice Trends and Variability Since 1979 Routine satellite passive-microwave measurements of global sea-ice concentration and extent, which began in 1979, have revealed a slight upward trend in overall (total) sea-ice extent (Fig. 1a) of 1.0 ± 0.5% per decade (or about 11,300 km2 per year) for 1979–2018 (Parkinson, 2019). This is in sharp contrast to the Arctic, where sea-ice extent has decreased precipitously (Maksym 2019; Serreze and Meier 2019) by -13.1 % per decade (or -82,700 km2 per year) over the period 1979-2020 (Perovich et al. 2020). The increasing linear trend in Antarctic sea-ice extent since 1979, and particularly from 1979 to 2014, is made up of distinctly differing regional (Comiso et al. 2017; Parkinson 2019) and seasonal (Holland 2014) contributions (Fig. 1b-c). Importantly, from both a physical and biological perspective (see Antarctic Sea Ice #2), the varying regional changes have occurred not only in sea-ice extent but also in the timings of sea-ice advance and retreat and resultant annual duration of coverage (Stammerjohn and Maksym 2017). This is again in contrast to the Arctic, where sea-ice loss and decreasing annual duration have been more spatially uniform across the Arctic Ocean (Stammerjohn et al. 2012; Maksym 2019). Mechanisms put forward to explain both the overall increase in total Antarctic sea-ice coverage and its pronounced regional and seasonal variability largely revolve around changes and variability in large-scale coupled oceanic and atmospheric circulation as they drive sea ice thermodynamic (formation and melt) and dynamic (wind-driven drift/transport and deformation) processes (Hobbs et al. 2016; Polar Research Board et al. 2017). Major atmospheric drivers proposed include: 1) a strengthening of the Amundsen Sea Low (a major low-pressure system located over the high-latitude South Pacific) e.g., Raphael et al. (2016); and 2) a strengthening of the Southern Ocean westerly winds e.g., Armour and Bitz (2015). Less well known and understood is the contribution of the Southern Ocean itself (Polar Research Board et al. 2017), and the possible role of feedbacks in the interactive ice-ocean-atmosphere system (Goosse et al. 2018). Regionally, Antarctic sea-ice extent increased in all sectors (by up to 2.3±1.2% per decade, for example, in the western Pacific Ocean/Ross Sea region) except for the Bellingshausen-Amundsen-eastern Ross seas (Fig. 1b) during 1979-2018 (Parkinson, 2019). There, SIE decreased by –2.5 ± 1.2% per decade (Parkinson 2019), and from 1979 to 2020 annual sea ice duration decreased by 2-3 months (Fig. 1c) due to later seasonal advance and earlier retreat (Stammerjohn et al. 2012; Stammerjohn and Maksym 2017). The changes in this region have been attributed to a regional increase in warm, northerly winds associated with strengthening of the Amundsen Sea Low, which enhance sea ice compaction and melt (e.g., Massom et al. 2008) while also inhibiting sea ice formation. By contrast, the most significant equatorward areal expansion is in the western Ross Sea where an increase in cold, southerly airflow on the opposite (western) side of the Amundsen Sea Low has led to both enhanced sea ice formation and northward drift (Holland and Kwok 2012) along with a 2-3 month increase in the duration of the sea-ice season (Stammerjohn and Maksym 2017). The strengthening of the Amundsen Sea Low and associated sea ice changes have been linked to natural multi-decadal climate variability (Turner et al. 2016), including a multi-decadal cooling of ocean temperatures in the eastern tropical Pacific (Meehl et al. 2016) and ocean warming in the tropical Atlantic (Li et al. 2014), both of which are tied to natural tropical variability and both of which strengthen the ASL remotely via large-scale atmospheric teleconnections (Li et al. 2021). The strengthening of the Amundsen Sea Low has also been linked to the anthropogenic depletion of stratospheric ozone over Antarctica (Turner et al. 2009; England et al. 2016), although the sensitivity of this relationship is less certain (Fogt and Zbacnik 2014). On year-to-year timescales, the location and strength of the Amundsen Sea Low central pressure are also sensitive to the phase of the El Niño-Southern Oscillation and the Southern Annular Mode (Clem et al. 2017; Lefebvre 2004). This makes attribution of Antarctic sea ice change and variability even more challenging. The overall increase in net sea-ice coverage has also been linked to strengthening of the belt of Southern Ocean westerly winds that encircle Antarctica (Armour and Bitz 2015), which is largely due to anthropogenic ozone depletion (Polvani et al. 2011) and greenhouse gas increases (Arblaster and Meehl 2006). Several mechanisms for this connection have been proposed, including wind-driven ocean-surface cooling (Thompson et al. 2011; Armour and Bitz 2015) and enhanced northward expansion of the ice edge by westerly wind-driven redistribution of the sea ice (Zhang 2014). However, this also leads to enhanced upwelling of warm deep waters along the Antarctic coast and within the seasonal sea ice zone, and it is predicted that this warming from below might eventually dominate, leading to sea-ice loss and surface ocean warming in coming decades (Ferreira et al. 2015). Other factors put forward for the net positive sea-ice trend include increased freshwater input to the Southern Ocean from melting Antarctic ice shelves (Bintanja et al. 2013) and increased precipitation/reduced evaporation (Liu and Curry 2010), and ice-ocean feedback processes (Goosse and Zunz 2014). These latter processes may have contributed to the observed Southern Ocean surface cooling since 1979 by increasing oceanic stratification and inhibiting vertical mixing that would otherwise bring warmer water from depth up to the surface (Aiken and England 2008), but intensification of the westerly winds appears to be the dominant driver of the ocean-surface cooling (Swart and Fyfe 2013). Looking to Antarctica’s dynamic coastal margins, analysis of satellite visible and thermal infrared imagery has similarly revealed distinct regionally-varying patterns in fast-ice distribution over the past two decades (Fraser et al. 2020). For the period 2000 through 2018, overall fast-ice extent displayed a marginally-significant decreasing trend of –1.9 ± 1.8 % (or −8820±8240 km2) per decade (albeit for an 18-year period only). Once again, the trend is not spatially uniform around Antarctica, but is comprised of varying regional contributions from eight distinct sectors, with increases in fast ice in East Antarctica and the southern Bellingshausen Sea but decreases elsewhere in West Antarctica. Particularly rapid changes in regional fast-ice distribution can occur not only due to changes in wind strength and direction (Massom et al. 2009), but also in response to major iceberg calving and grounding events. Fast ice is also susceptible to breakup by ocean swells, particularly in the absence of a protective pack-ice buffer (Crocker and Wadhams, 1989). Fast-ice change has implications for ice-shelf stability and sea-level rise (Massom et al. 2010, 2018), ecosystems (see Antarctic Sea Ice #2), and coastal sea ice production rates and associated water-mass modification (Tamura et al. 2012) because of the close association between fast ice and polynyas (Fraser et al. 2019). 2013-2022: High Variability in Antarctic Sea ice Coverage After 34 years (1979-2012) of relatively steady net Antarctic sea-ice increases, there was an unexpected shift to first record-high Antarctic sea ice in 2013 to mid-2015 (Reid and Massom 2015; Reid et al. 2016) to record lows from late 2016 to present (early 2022) (Parkinson 2019; Reid et al. in press) (Fig. 1a). The rapid decrease from 2016 exceeded rates observed in the Arctic, with Antarctic annual-mean sea-ice extent reaching its lowest value on record in 2017 – a monthly mean of 2.29 x 106 km2 in February (Parkinson 2019). Annual mean sea ice coverage decreased by fully 1.6 million km2 between 2014 and the spring of 2016 (Wang et al. 2019), with the rate of decrease in spring of 2016 being unprecedented for this season (Turner et al. 2017). The abrupt decrease in overall sea-ice coverage during 2016 has been attributed to various atmosphere and ocean anomalies. Firstly, record-strong low-pressure systems in multiple sectors of the Southern Ocean brought warm northerly airflow across much of the sea ice zone (Wang et al. 2019). Anomalously warm sea-surface temperatures (SSTs) – marking an abrupt reversal of the extensive surface ocean cooling observed during the previous decades – also emerged across the Southern Ocean in 2016 (Mazloff et al. 2017; Kusahara et al. 2018) and appear to have also played a role in the persistent sea ice decline since 2016 (Meehl et al. 2019; Haumann et al. 2020). The exceptional atmospheric circulation anomalies in 2016 have been linked to tropical variability (Wang et al. 2019; Meehl et al. 2019; Purich and England 2019) and an amplified state of the prominent circumpolar atmospheric circulation pattern called Zonal Wave 3 (three pairs of high-low pressure systems encircling Antarctica), which led to increase in warm northerly airflow across three sectors of the sea ice zone (Schlosser et al. 2018). The Weddell Sea, one of the few regions of the Southern Ocean where sea ice persists through the summer (Parkinson 2019), made the largest contribution to the persistent sea ice decline post-2016 (Turner et al. 2020). Sea-ice coverage there has decreased by over 1 million km2 after reaching a record-high summer extent in 2014/15 (Reid et al. 2016). The unprecedented loss of sea ice in the Weddell Sea is also linked to record-strong atmospheric low-pressure systems in 2016 and 2017, combined with warmer SSTs (Meehl et al. 2019). The loss of sea ice (and its high albedo) resulted in surface-ocean warming in the Weddell Sea that persisted through the 2019/20 summer and the Weddell Sea summer sea-ice extent has remained at near record lows through early 2022 (Reid et al. In press). Early and sustained retreat of sea ice in 2021/22 has led to new record low values of net sea-ice extent being observed, with daily values of net sea-ice extent dropping below 2 million km2 for the first time since satellite records began in 1979. It is obvious to ask if the recent and apparently ongoing (2016-2022) decline in sea-ice extent is a portent of the impact of climate change, along the lines of that seen in the Arctic, that are predicted by coupled climate model simulations (SROCC 2019). While further work is needed to fully understand the mechanisms associated with this recent era of rapid sea-ice decline, findings are emerging that suggest the decline in sea-ice extent may have resulted from a decades-long warming of sub-surface ocean temperatures (Eayrs et al. 2021). Over the past 40 years, global-mean temperatures increased by around 1°C relative to the mid-20th Century (https://climate.nasa.gov/vital-signs/global-temperature/). Over this relatively short period, we have learned important lessons from the observational record, including that: Natural climate variability, particularly in the tropics (e.g., multi-decadal SST variability), can drive significant multi-decadal trends in Antarctic sea ice, and notably in the South Pacific sector (the Bellingshausen, Amundsen and Ross seas) through strengthening or weakening the Amundsen Sea Low; Extreme weather events (such as record-strong storms) and abrupt changes in upper-ocean temperatures across the Southern Ocean (including marine heatwaves) can cause rapid and extreme changes in Antarctic sea ice that can persist for several years (at least), aided by ice-ocean albedo and heat content feedbacks (e.g., decreasing sea ice coverage leading to increased solar heating of the ocean which results in increasing loss of sea ice); and Anthropogenic climate change, involving ozone depletion and increasing atmospheric greenhouse-gas concentrations, can paradoxically result in multi-decadal increases in overall Antarctic sea ice coverage – through surface cooling and freshening of the Southern Ocean and increased northward sea ice drift and formation due to strengthened westerly winds. Figure 1. (Top) (a) From Parkinson (2019), Fig. 2c, the annual average total Antarctic sea ice coverage (extent) during 1979-2018 and its linear trend line (dashed) and trend value (top left insert). (Bottom) The linear trends/changes over 1979-2020 of (b) austral autumn (March-May) Antarctic sea ice concentration and (c) Antarctic sea ice season length (timing of annual advance minus timing of retreat). Black contours in (b-c) denote statistically significant trends at the p<0.01 confidence level. The locations of the Weddell (WS), Bellingshausen (BS), Amundsen (AS), and Ross (RS) Sea regions are inset in (b-c). The sea ice data used in (a-c) are sourced from passive-microwave data from the NASA Nimbus 7 and Department of Defense DMSP satellites. Parkinson (2019) derived sea ice extent in (a) from sea ice concentrations from the NASA Team 2 algorithm; sea ice concentration and season length in (b-c) are derived using the GSFC Bootstrap v3.1 algorithm. The sea ice concentration data can be downloaded from the National Snow and Ice Data Center (NSIDC) at the University of Colorado-Boulder (http://nsidc.org). Antarctic Sea Ice Prior to 1979 - The Longer-Term Context Interpreting and attributing observed sea-ice change and variability in the Antarctic since 1979 represents a considerable challenge. This is due (amongst other things) to the highly variable nature inherent to Antarctic sea ice and the shortness of the duration of the satellite passive-microwave record, which may not be representative of long-term trends and does not fully resolve multi-decadal/internal variability (Hobbs et al. 2016). Given these factors, proxy reconstructions of past sea-ice extent – extending back over the past century to thousands of years before present – are crucial to understanding the recent observational record in a historical context and providing insights into climate feedbacks within the Earth system at both short and long timescales (Crosta et al. 2022). In terms of the past hundred or so years, information from historical shipping and whaling records (e.g.,de la Mare 2009), ice-core records (Curran et al. 2003), and snapshot images from early meteorological satellites (Gallaher et al. 2014) indicate a decrease in Antarctic sea-ice extent since the 1950s in the Bellingshausen Sea west of the Antarctic Peninsula (Abram et al. 2010). In an effort to more widely cover circum-Antarctic and seasonal change, Fogt et al. (2022) use observation-based reconstruction ensembles of seasonal Antarctic SIE dating back to 1905. These show that: 1) over the 20th Century, the observational period since 1979 is the only time when all four seasons show significant increases in total sea-ice coverage; and 2) the observed increases contrast with statistically-significant decreases through much of the early and middle parts of the 20th Century. Looking further back in time, research within the PAGES (Past Global Changes) C-SIDE (Cycles of Sea ice Dynamics in the Earth System) program is focused on ice core- and marine sediment-based reconstructions of sea-ice changes through the last glacial-interglacial cycle (Crosta et al. 2022). This longer time frame includes the penultimate interglacial when Antarctica was about 2°C warmer than present – thereby providing a “process” analogue for future climate-warming scenarios. It also enables evaluation of the role of sea ice during both 1) the glacial onset (when the ocean sequestered carbon), and 2) deglaciation (when carbon was released to the atmosphere). Crosta et al. (2022) provide an in-depth review of what the proxy/palaeo records tell us about change and variability in Antarctic sea ice coverage over the past 130,000 years. Future projections Over the next 50 to 100 years, the distribution (extent, concentration and seasonality), properties and thickness of Antarctic sea ice and its snow cover (and regional patterns therein) will continue to be closely tied to natural multi-decadal climate processes operating in tandem with anthropogenic increases in global-mean temperature – including higher rates of warming in the polar regions (e.g., Stuecker et al. 2018). Increasing greenhouse-gas concentrations are expected to continue to strengthen the Southern Ocean westerly winds over the remainder of this century, which may offset their anticipated weakening from the ongoing recovery of the ozone hole (Kushner et al. 2001; Arblaster et al. 2011). More research is needed in understanding how upper-ocean temperatures (and therefore sea ice) will respond to continued strengthening of the westerly winds, as the cooling seen up to 2014 (Zhang et al. 2021) may have been an initial, short-term response only, while the long-term response may be upper-ocean warming, as seen in recent years (Sigmond and Fyfe 2014; Bitz and Polvani 2012; Smith et al. 2012; Haumann et al. 2014; Ferreira et al. 2015). It is anticipated that upper-ocean warming will contribute to long-term sea ice loss throughout the remainder of this century, as evidenced by its apparent role in the rapid and unprecedented retreat of Antarctic sea ice in 2016 (Meehl et al. 2019; Haumann et al. 2020). Anthropogenic increases in atmospheric temperatures – by at least +1.5°C (but possibly up to +4°C) by the end of this century under low (high) greenhouse-gas emission scenarios (SROCC 2019) – are also anticipated to cause reductions in both Antarctic sea-ice extent and volume by 2100. Climate models predict a decrease in sea-ice area that ranges (depending on the emission scenario) from 29% to 90% in February (minimum annual SIE) and from 15% to 50% in September (maximum annual SIE) (Roach et al. 2020). However, there is high uncertainty (low confidence) in current model projections of Antarctic sea ice (Meredith et al. 2019; IPCC 2021). This is due to a combination of the large multi-decadal trends that can be driven by natural variability as outlined above (Polvani and Smith 2013; Zunz et al. 2013; Mahlstein et al. 2013) as well as limitations in the ability of current climate models to simulate the observed changes in Antarctic sea ice including their seasonal and regional patterns (Hobbs et al. 2016). These limitations relate to: 1) the multiple external forcings involved i.e., greenhouse gases and ozone (Meredith et al. 2019); 2) coarse model resolution; and 3) deficiencies in simulating (and our understanding of) the complex ice-ocean-atmosphere dynamics, interactions and feedbacks that govern sea ice formation, drift and melt (Notz and Bitz 2017). Other key factors include snowfall and possible rainfall (Webster et al. 2018), ocean waves (and their role in both sea ice formation and breakup/melt; Kohout et al. 2014), and feedbacks (Goosse et al. 2018) including ice-ocean albedo (Nihashi and Cavalieri 2006) and those associated with sea ice biogeochemical processes and clouds (Vancoppenolle et al. 2013). Challenges Robust assessments of the nature and drivers of sea-ice change in the Antarctic are currently held back by a lack of both adequate observations and understanding of the complex ice-ocean-atmosphere-biosphere processes, interactions and feedbacks (Polar Research Board et al. 2017; Goosse et al. 2018) (see Antarctic Sea Ice #1). Addressing this critical gap needs a coordinated programme of integrated and sustained cross-disciplinary observations of the sea ice environment (coupled sea ice-ocean-atmosphere-ice sheet system) – as proposed by the Southern Ocean Observing System or SOOS (Newman et al. 2019) – together with dedicated cross-disciplinary experimental and process-based studies conducted from icebreakers and at different times of year. Autonomous technologies that operate on, above and below the ice and instrumented marine mammals and seabirds have a key role to play in advancing our understanding of the sea-ice environment, and also in bridging the scale gap between detailed in situ observations (on the floe scale) and large-scale and coarser resolution satellite observations (Smith et al. 2019). Calibration and validation of satellite-derived products is also critical e.g., sea ice and snow cover thickness (see below), given the crucial role of satellites in routinely and systematically measuring and monitoring Antarctic sea ice on the large-scale and over prolonged periods (Lubin and Massom 2006). While the large circumpolar-scale horizontal distribution of Antarctic sea ice has been continuously monitored from space since 1979, this is not the case for the thickness of the ice and the depth of its snow cover. Available observations (airborne, under-ice and surface) of Antarctic sea ice thickness and snow-cover depth are far too sparse to determine whether they are changing over space and time (Webster et al. 2018). This represents a critical knowledge gap and one that severely compromises model predictions and operational forecasting capability, as the climatic significance of Antarctic sea ice arises in large part from the volume of ice that freezes and melts each year, with snow depth playing a key role in this process (see Antarctic Sea Ice #1). New large-scale estimates of Antarctic sea ice thickness and snow depth are emerging from satellite altimeter datasets e.g., Kacimi and Kwok (2020), but definitive large-scale trends are yet to emerge and the satellite products require validation. Accurate and sustained large-scale information on precipitation rate and type over the Antarctic sea ice zone are also required (Webster et al. 2018). In parallel with observational challenges, there is a crucial need to improve representation of the Antarctic sea-ice environment in climate models, and also the simulation of patterns of sea ice distribution, change and variability observed over the satellite era (Polar Research Board et al. 2017). This is in turn a critical step to reducing current large uncertainty (low confidence) in model predictions of the nature and effects of sea-ice change over coming decades in response to climate warming (SROCC 2019). This includes extreme meteorological events, which can have an important and lasting impact on the coupled sea ice physical-ecological-biogeochemical system (e.g., Massom et al. 2006) and are likely to become more prevalent in a warming world (IPCC 2022). Improved regional coupled modelling at high resolution is also required to develop accurate operational sea ice forecasting capability (spanning synoptic/daily to seasonal scales) in support of safe and efficient shipping and operations in Antarctic waters. High-resolution modelling is also crucial to case-study investigations of processes, such as in narrow yet complex and immensely important Antarctic coastal environments, and attribution of change (e.g., Kusahara et al. 2018). Conclusion Total Antarctic sea ice coverage has until recently (2016) shown a general increasing trend that contrasts strongly with the decreasing trend in the Arctic Ocean. This full picture is, however, made up of strongly-varying regional and seasonal contributions – with contrasting regional trends in sea-ice extent and the timings of sea ice advance and retreat and resultant ice season duration around Antarctica. From 2013 through 2022, Antarctic sea-ice extent has undergone rapid and unexpected shifts, first towards an increase to record highs (2013 to mid-2015) then a rapid decrease to record lows (late 2016 to present). Many key individual drivers of Antarctic sea-ice change and variability are fairly well known, but interactions among different processes are not well known due to complex feedbacks involving not only the sea ice and its snow cover but also the ocean, atmosphere and ice sheet (as well as biogeochemical processes). Current critical knowledge gaps include the large-scale thickness distributions (and volume) of Antarctic sea ice and its snow cover, and whether these are changing. Model projections generally agree that over the next 50 to 100 years, Antarctic sea ice will decrease, with significant physical and biological implications for the polar region and beyond (e.g., for global climate, sea-level rise, and ecosystems). However, there is currently substantial uncertainty as to the magnitude, regionality, seasonality, and timing of future change in Antarctic sea ice coverage and properties. Placing observed sea-ice change (since 1979) in a longer-term (historical) context is crucial to both interpreting recent change and variability and improving our understanding of short- and long-term feedbacks, including their influence on the global climate system and ecosystems (see Antarctic Sea Ice #2).