Session 1: The Big Picture: Patterns and Drivers of Antarctic Sea Ice Variability
Patterns of Antarctic sea ice variability: Observational insights, gaps, and challenges
One of the most variable but largest seasonal signals on Earth is the six-fold change in Antarctic sea ice extent that occurs every year in the Southern Ocean. Superimposed on this large seasonal transformation is a modest increase in Antarctic sea ice extent over the past four decades. However, this circumpolar average hides a mix of regional and seasonal sea ice extent trends, some of which are large but opposing, while others are modest and fall within the range of natural variability. While sea ice extent estimates are consistent between different satellite retrieval methods, less known or validated are changes in sea ice concentration, thickness, and snow cover. Although we know that regional and seasonal variability is caused by variable exposure to wind and wave events and heat and freshwater inputs, we lack the observations to properly assess their role in driving sea ice changes. Furthermore, differences in regional and seasonal exposure and sensitivity to large-scale climate variability create a variable mix of drivers and feedbacks operating at different time/space scales. Throughout this brief review, key gaps, challenges, and questions will be highlighted to stimulate discussion of strategies for improving assessments of Antarctic sea ice variability.
Understanding Antarctic climate change: How are we led and misled by models?
Observations of Southern Ocean climate variability indicate a consistent relationship between trends in atmospheric circulation, sea ice extent, and sea surface temperature. Despite an overall poor success by climate models to reproduce the climatology and recent trends in the surface climate, can we use climate models to identify a limited set of leading interactions that might be at play? The basic mechanisms of heat uptake and heat transport in the Southern Ocean that delay global warming are captured in climate models. But there is still vigorous debate about the role of recent wind anomalies, whether they are connected to tropical teleconnections, ozone depletion, or local natural variability. I will discuss why models yield mixed insights into the impacts of winds. I will also consider to what extent models may be simply missing key features, and where effort should be placed toward model improvements, such as coupled interactions with ocean and ice shelves; the influence of waves on sea ice; ocean mixing; the influence of small-scale ocean eddies, katabatic winds, or polynyas; and lastly the proper simulation of clouds and their impact on the surface energy budget.
Drivers and forcings of observed Southern Ocean sea ice change
Observations of Southern Ocean sea ice since 1979 show a spatially heterogeneous pattern of trends with a modest overall increase in areal extent, most intense in the austral autumn. By contrast, global coupled-model experiments indicate that the response to anthropogenic forcings is a spatially homogeneous decrease in ice cover, most intense in winter. Does this discrepancy indicate fundamental flaws in the climate models, or the entirely plausible scenario of a modest forced response being overwhelmed by the high internal variability of the system? To answer this question with confidence, it is necessary to consider the physical drivers of change, and to assess whether those drivers are adequately
represented in the models. In this presentation I will give a brief overview of atmosphere, ocean, and ice sheet processes that have been hypothesized to contribute to the observed patterns of change, and give a critical assessment of our current level of understanding. I will also outline the current challenges impeding a confident statement regarding the current and future response of Antarctic sea ice in an anthropogenically altered climate.
Session 2: The Observational Record: Insights and Gaps
Antarctic sea ice observations: Background
This short talk provides an assessment of various datasets available for analyzing large-scale change and variability in Antarctic sea ice, comprising both pack and fast ice and their snow covers, and what they show. In so doing, it highlights key uncertainties, gaps, and issues in the observational record to date. It also underlines the crucial need for improved understanding of complex interactive processes driving the weak overall increase in areal coverage since 1979 that represents the residual of opposing regional (and seasonal) trends. A brief introduction to the modern satellite passive-microwave record and a new coastal fast ice dataset is followed by an assessment of proxy and other data from the pre-1979 era that highlights the importance of placing recent overall and regional trends in longer-term context, and the dependence of trends on the end points chosen. Although widely used, sea ice extent alone is an incomplete/partial descriptor of sea ice state; also required is reliable information on sea ice and snowcover thickness and volume, with consideration of seasonality, different regional settings, and extreme events also being important factors.
Contribution of Antarctic sea ice thickness observations to understanding of sea ice volume change
Observational data of Antarctic sea ice thickness have been compiled from a wide variety of sources including in situ measurements, undersea sonar, and laser and radar altimetry sources. Each data set provides key information on the overall thickness and volume of Antarctic sea ice; however, there is much that needs to be done to use these records to understand and constrain observed variability and trends. To date, observational sources have shed light on the overall thickness and volume of the Antarctic sea ice pack from a climatological perspective, but uncertainties resulting from spatial sampling, instrument uncertainties, and temporal differences have precluded their use in assessing trends and variability in the volume. However, new advances in observations and promising methods to utilize the results with model records do provide the potential to utilize these records to understand change in the Antarctic sea ice system, which will be the focus of this presentation.
Ocean observations and Antarctic sea ice
The ocean is an integral component of the high-latitude air-sea-ice system, and various theories have been advanced concerning the role of the Southern Ocean in recent spatiotemporal changes in Antarctic sea ice. These include, but are not limited to, strengthening of the upper ocean stratification caused by enhanced freshwater inputs, wind-induced acceleration of the northward Ekman transport of cold surface waters, changes in the rate of upwelling of warm waters from the mid-layers of the Southern Ocean and subduction of waters back into the interior, and complex mixed-layer feedbacks that can sustain and promote atmospherically induced sea ice changes. While different models suggest that each of these can be important in different regions and on different time scales, testing the veracity of these models with observational data is challenging. Aside from the surface ocean (which is comparatively well
monitored by satellites) and a small number of relatively well-observed sites, the subsurface Southern Ocean is one of the biggest data deserts on the planet, and this is especially true for the ocean beneath the sea ice itself, especially in the winter months when processes critical to sea ice occur. Needed is a much more comprehensive ocean observing including a step-change in the capability for under-ice measurements, as was recently advocated by the Southern Ocean Observing System. This will enable collection of the data required to properly understand the role of the ocean in Antarctic sea ice change and to increase predictive skill concerning how it will change in the future.
Atmospheric observations over the Antarctic sea ice zone
The Antarctic sea ice zone is almost completely devoid of conventional meteorological observations. Thus, the atmospheric parameters of most relevance to studies of Antarctic sea ice behavior (winds, temperatures, precipitation, downward radiation fluxes, etc.) are most often derived from global atmospheric reanalyses that merge all available observations with short-term model forecasts. For the Southern Ocean, the most important data source is the wide variety of satellite observations. Unfortunately, the discontinuities between satellite missions, which even today emphasize operational needs rather than being of climate quality, lead to jumps and artificial trends in global reanalyses. There are also important shortcomings to the representation of forecast clouds in the atmospheric models used for reanalyses that negatively impact the forecast radiation fluxes. A survey of the performance of contemporary global atmospheric reanalyses for both the modern satellite era (post-1978) and for the centennial time scale will be presented focusing on those variables of most interest to sea ice investigations.
What can we learn from ice cores about Antarctic sea ice variability and trends prior to the observational era?
Overall Antarctic sea ice coverage appears to be more extensive today than in recent decades, in contrast both to many model predictions as well as to the recent dramatic decline of sea ice in the Arctic. One major challenge to understanding recent changes in Antarctic sea ice, and therefore to also improving predictions of future Antarctic sea ice trends under varying climate scenarios, is the extremely short period of direct observations, which are for the most part limited to the recent decades of the satellite era. And given the high spatial and temporal variability of Antarctic sea ice documented over recent decades, these direct observations may be insufficient to situate recent anomalies and trends into the proper temporal and climatological context. Thus, there is a need to develop new proxy records for reconstructing both longer-term and regionally resolved records of past sea ice conditions around Antarctica, not only for longer time periods (e.g., glacial-interglacial periods), but also critically for more recent centuries to decades. The most likely (and arguably the only possible) candidates for yielding such sea ice proxy information are the depositional records provided by marine sediment cores and ice cores, and indeed this is where many recent and current research efforts have been focused. A number of ice-core chemical constituents originate from aerosols sourced from the sea surface or the marginal sea ice zone, and thus their concentration in ice cores should be strongly influenced by sea ice behavior. The most notable (and best studied) of these to date are sea salts (Na and Cl), which can be sourced both from sea spray as well as the surface of newly formed sea ice, and methanesulfonic acid (MSA), a biogenic marker that is an oxidation product of dimethylsulfide (DMS), a gas characteristic of the near sea-ice zone. Efforts are under way to improve our ability to use variability in ice core–derived Na and MSA concentrations, as well newer glaciochemical proxies in development, to inform us about past sea ice spatial extent and/or its presence or absence through time. Yet many significant challenges remain in understanding the source, transport, and post-depositional behavior of all these marine-derived aerosols, as well as the identification
of characteristics of ice sheet environments that would be most suitable for reconstructing past regional sea ice behavior, all of which can be very site specific. In short, moving the utility of glaciochemical records from a qualitative understanding to the quantitative reconstructions needed by the sea ice research community, especially on interannual-to-decadal time scales, remains very much a work in progress. Nonetheless, despite the numerous challenges, ice core records are likely to remain one of the most promising avenues toward the reconstruction of past sea ice variability.
Session 3: Sea Ice Variability in Earth System Models
Modeling approaches to understanding observed Antarctic sea ice variability and trends
I will provide an overview of the different types of modeling experiments that are used to understand Antarctic sea ice variability and trends and discuss their strengths and limitations. At one end of the spectrum are free-running coupled climate models driven by historical radiative forcing estimates associated with variations in greenhouse gas and stratospheric ozone concentrations as well as anthropogenic and volcanic aerosols. Such simulations show a range of Antarctic sea ice behavior depending on how the model responds to the imposed radiative forcings and its particular chronology of internal (unforced) variability. At the other end of the spectrum are coupled ocean–sea ice models driven by the observed evolution of atmospheric conditions such as winds and air temperatures. Such simulations provide information on how the ocean–ice system responds to the imposed atmospheric conditions, but do not address the origin of the wind and air temperature variations. Intermediate in this hierarchy are coupled models in which sea surface temperatures (SSTs) in a particular region such as the tropical Pacific or North Atlantic are nudged to observations. Such “pacemaker” simulations test the role of remote SST changes on the behavior of Antarctic sea ice via oceanic and atmospheric pathways.
Using global climate models to understand recent Antarctic sea ice trends
After several decades of satellite observations, trends in Antarctic sea ice, at a fundamental level, remain poorly understood. General circulation models can be used to shed light on these surprising trends. Notably, such models provide key insights into the role of different anthropogenic forcings. The models also allow us to place the sea ice response caused by individual forcings in the broader context of the natural variability of the Antarctic climate system. Recent findings on these fronts will be reviewed and discussed.
Have the surface westerlies changed?
Models participating in Phase 5 of the Coupled Model Intercomparison Project (CMIP5) suggest that externally forced strengthening of the Southern Hemisphere surface westerlies may have impacted Antarctic sea ice coverage. However, confirming this in nature is made difficult by the lack of surface wind observations, large uncertainties in estimates from reanalyses, and internal variability. Across reanalyses from 1979 there is no consistent westerly wind trend except in the austral summer, where positive trends of varying magnitude (0.05 to 0.4 m/s per decade) are within the uncertainty range of the CMIP5 simulations. This is suggestive but not confirmation of externally forced westerly wind intensification in observations. Over the shorter period (post-1988) that satellite-based surface wind observations are available, there is a consistent pattern of surface wind change in the South Pacific in observations and reanalyses. However, this pattern of change is dissimilar from the CMIP5 average pattern and is likely a feature of internal variability rather than a signal of externally forced westerly wind intensification. In short,
while an externally forced westerly wind impact on Antarctic sea ice coverage is plausible, solid evidence has yet to emerge.
Antarctic sea ice variability and trends: The role of the ocean
The Southern Ocean stores and redistributes heat horizontally and vertically. This can damp or amplify any perturbation, for instance originating in the atmosphere, and provides a strong control on the sea ice variability. The associated mechanisms strongly depend on the mean state of the system; it is thus essential that models reproduce this mean state adequately. Furthermore, a correct initialization of the Southern Ocean is needed for decadal predictions and projections.
Session 4: Process and Attribution Studies of the Coupled System
Tropical Pacific teleconnections to the Southern Hemisphere high latitudes
In this talk, I will review the current understanding of tropical Pacific teleconnections to Antarctica, outlining how interannual and longer-term climate variability are influenced by such processes. On interannual time scales, both eastern Pacific (EP; canonical ENSO) and central Pacific (CP; El Niño-Modoki) SST variability drive pronounced perturbations to the Antarctic surface climate, but with marked seasonal dependence. During the austral cold season (JJAS), both the EP and CP force a similar Rossby wave response, driving anomalous sea ice variability in the West Antarctic region through impacting the Amundsen Sea Low. In the austral warm season (NDJF), by contrast, the EP is linked to zonally symmetric circulation structures that project onto the Southern Annular Mode, whereas only a weak teleconnection is observed from the CP. SST trends, specifically the observed warming in the central/western Pacific, have also been linked to contemporary circulation trends in the Amundsen Sea Low during austral winter and spring. The Rossby wave train mechanism has been proposed to explain such longer-term teleconnections, and as such, West Antarctic surface air temperature and sea ice trends have been further attributed to such processes. There is, however, debate on the extent to which Pacific teleconnections may explain high-latitude trends, and further work is required to clarify controversies in this regard.
Rossby and Kelvin waves link the Atlantic SST and Antarctic sea ice
During the past three decades, sea surface temperature (SST) over the north and tropical Atlantic exhibited a significant warming trend. Observational data and model simulations link this SST anomaly to the climate changes around Antarctica. These teleconnections critically depend on stationary Rossby waves and are thus sensitive to the background flow, particularly the subtropical and midlatitude jets. Using a hierarchy of climate models, we show that these jets reflect, guide, and focus the Rossby wave trains on western Antarctica, modifying the regional atmospheric circulation, and thus the Antarctic sea ice distribution. Further analyses show that the recently observed SST trends in different tropical ocean basins are linked with each other through atmospheric Rossby and Kelvin waves. These dynamics build multiple pathways from the tropical ocean basins to Antarctica, and thus further complicate the mechanisms of the Antarctic climate variability.
The role of tropical teleconnections in recent Antarctic climate change
This talk will review the role of tropical teleconnections in driving recent trends in the atmospheric circulation around Antarctica and discuss how this is important for understanding the trends in Antarctic
sea ice extent. In particular, a series of transient atmospheric model experiments is evaluated to show that changes in tropical sea surface temperatures have played the dominant role in the deepening of the Amundsen Sea Low during the satellite era (which coincides with the modern sea ice record). The impacts of stratospheric ozone depletion on the strengthening of the Southern Ocean westerlies are most evident with a time horizon extending back to about 1960. This knowledge of the drivers of recent circulation changes informs a new set of coupled model experiments in which the tropical Pacific sea surface temperatures are relaxed to the observed values, thereby constraining the model to the observed evolution of tropical climate (including the El Niño–Southern Oscillation and the Pacific Decadal Oscillation phenomena). While these simulations show improvements in simulating the recent atmospheric circulation changes over the Southern Ocean compared to their unconstrained counterparts, they also illustrate the need for improving the representation of physical processes contributing to the evolution of sea ice in coupled models.
Antarctic ocean–ice interactions
This talk focuses on the unique ocean–sea ice interactions whereby brine formation during sea ice growth drives upper ocean convection into the permanent pycnocline that releases heat into the mixed layer melting, or preventing from forming a large amount of sea ice. This keeps the sea ice thin (~60 cm, also making it easy to melt in the spring forcing a seasonal sea ice cover). Specifically, for every one unit of ice, the entrainment heat flux releases enough heat to melt 5-20 units of ice in both the open ocean Weddell gyre and continental shelf west Antarctic Peninsula regions. The presence of the cold halocline layer (CHL) in the Arctic prevents this feedback allowing that ice to grow extremely thick (meters) not as easy to melt completely in summer.
The Antarctic feedback is the result of a predominantly vertical system (halocline and thermocline coincide with depth) with a very warm deep layer (Upper Circumpolar Deep Water, UCDW) underlying the pycnocline. The interactions are well predicted from an analytic vertical model whose scaling laws using easily observed variables allow excellent estimates of the maximum ice thickness, magnitude of the feedback, ocean heat flux, and other characteristics of the temporal evolution of the system. UCDW has been warming dramatically the past several decades, reflecting the warming of global deep water. This is the water that melts the underside of the ice streams draining the west Antarctic Ice sheet into the Amundsen Sea Embayment (greatly accelerating their drainage into the ocean, raising global sea level). The warming ensures strong melt, even if the circulation beneath the ice margins slows down. Overall slight expansion of the sea ice extent around Antarctica, the consequence of expansion in some areas and contraction in others, is not understood yet, but the location of the Antarctic Circumpolar Current (ACC) limits the areal extent of polar water capable of forming ice, and the warming of the surface waters across the ACC limits the distance sea ice can be expanded to the north. So it is speculated here that subtle shifts in the ACC via changes in the westerlies may actually account for the statistical zonal sea ice expansion.
The two–time scale response of Antarctic sea surface temperature and sea ice cover to ozone depletion
In recent modeling work, we showed that the response of Antarctic sea surface temperature and sea ice cover to abrupt ozone depletion has two phases: a fast interannual (~1-5 y) adjustment in which the surface ocean cools and sea ice cover increases, followed by a slower decadal trend leading to a warming of the surface ocean and a reduction of sea ice cover. This result reconciles diverging views, found in the literature, on the relationship between ozone depletion and Antarctic sea ice changes. I will briefly describe the dynamics behind the two–time scale response and the sources of uncertainties in the modeled response. I will then discuss implications of the two–time scale response to evaluate the
contribution of ozone depletion (and recovery) to the observed decadal trends and future trends of the Antarctic sea ice.
The ocean’s role in Southern Ocean climate change: Warming delayed by circumpolar upwelling, and cooling driven by trends in SAM
The region of ocean surface south of the Antarctic Circumpolar Current (ACC) has warmed much more slowly than the global ocean since the 1950s, and has cooled since about 1980—coincident with an overall expansion of the Southern Ocean sea ice cover. Here we argue that ocean circulation plays a critical role in these observed changes.
Comprehensive (CMIP5) and idealized climate model simulations robustly simulate delayed sea surface warming around Antarctica in response to greenhouse gas forcing. Using models and observations, we show that the primary driver of this warming delay is the Southern Ocean’s background meridional overturning circulation, which upwells unmodified water to the surface from depth and advects surface waters northward. It is thus against a background of gradual, greenhouse gas–induced warming—rather than the rapid warming observed in the Arctic—that multidecadal Southern Ocean temperature and sea ice trends must be understood.
One proposed driver of the observed cooling and sea ice expansion in recent decades is the trend in the large-scale westerly winds, often characterized as a trend toward a positive phase of the Southern Annular Mode (SAM). Yet, targeted SAM perturbation experiments within comprehensive climate models generally find enhanced warming and sea ice loss. We summarize recent analyses of the Southern Ocean’s response to SAM changes within the preindustrial control simulations of CMIP5 models. Those models that simulate a Southern Ocean climatology that most closely matches observations robustly show cooling and sea ice expansion in response to trends in SAM. We argue that the CMIP5 models have been unable to replicate observed Southern Ocean changes due to a confluence of biases in their ocean climatologies and SAM trends that are too weak compared to observations.
The role of anthropogenic forcing and natural variability on Southern Hemisphere atmospheric circulation trends
This presentation will provide a brief overview of the roles of natural variability in comparison to anthropogenic forcing on atmospheric circulation trends near Antarctica, with direct implications for changes in sea ice parameters (season length, concentration, and extent). The primary focus will be on the Amundsen Sea Low (ASL), where the strongest regional sea ice changes have been observed. Notably, studies show the strongest influence from ozone depletion (leading) and greenhouse gases (a smaller contribution) in austral summer, both leading to a deepening of the ASL. Forced atmospheric circulation changes in other seasons are more model dependent, with results varying across studies. Tropical variability alone appears sufficient to capture a large portion of the variability (and even recent changes) in the non-summer ASL, and a recent study even suggests that ongoing sea ice trends are within the bounds of natural, unforced climate variability. Together these studies all suggest there is still some debate on the forcing on recent regional atmospheric circulation changes, and that in particular features of the ASL must be adequately simulated in order to accurately depict regional sea ice trends.
Dependence of Antarctic sea ice trends on zonally asymmetric atmospheric circulation changes and ocean stratification
In this presentation, I will illustrate how zonal asymmetries in the large-scale circulation around Antarctica drive variations and long-term trends in Antarctic sea ice and argue that this process is not fully
captured by coarse-resolution climate models. The observed large-scale circulation pattern is forcing an outflow of cold air from the two large ice-shelf regions on to the sea ice in the Ross and Weddell Seas. The sea ice strongly responds to the changes of this cold-air outflow with an increased northward transport of sea ice and an expansion of the ice edge when more cold air is leaving the continent, as observed over recent decades. Simulations with a global climate model show that this process is largely absent in the model due to rather zonally symmetric large-scale circulation changes in response to the anthropogenic forcing. I will argue that the underestimated zonal asymmetries in the atmospheric circulation are also causing a much weaker ocean surface stratification in the model. This results in a much higher sensitivity of the sea ice to changes in the subsurface heat fluxes, explaining why simulated sea ice changes are driven by a different mechanism than the changes observed in recent decades.
Seasonality of Antarctic sea ice trends
The Antarctic sea ice increase features opposing regional trends with a strong seasonal signal. These spatial and temporal patterns can offer significant insight. I will show trends in ice “intensification” (the rate of change of ice concentration), which can be viewed as trends in ice growth/melting or dynamics. The results suggest that the Antarctic sea ice trends originate in spring. The largest concentration trends, in autumn, appear to be caused by trends in intensification during spring. Autumn intensification trends directly oppose autumn concentration trends in most places, seemingly as a result of ice and ocean feedbacks. This causes a problem with the linkage between wind trends and ice trends, which agree in autumn, but do not agree in spring. This suggests that we need to understand more about how autumn trends affect the winter ice cover and spring melting.