1.2.2.1 Ice sheets

Most of Earth’s freshwater is stored in the ice sheets of Greenland and Antarctica (Figure 1.2.3). These represent by far the largest potential sources for sea level rise under ongoing and future warming: if the Greenland Ice Sheet (GrIS) were to melt entirely, global sea levels would rise by about 7 metres (Morlighem et al., 2017), for the Antarctic Ice Sheet, the total sea level rise potential is 58 metres (Fretwell et al., 2013; Morlighem et al., 2020). Even if only part of these masses were to undergo abrupt ice loss or tipping behaviour, this would have far-reaching consequences for coastal communities, infrastructure and ecosystems worldwide (Fox-Kemper et al., 2021).

The ice sheets have been losing mass at an accelerating rate: from an average of about 105 gigatonnes (Gt – i.e. one billion tonnes) per year between 1992 and 1996 to around 372 Gt per year between 2016 and 2020 (Otosaka et al., 2023) (Figure 1.2.4). The Greenland ice sheet is (still) the major player, with an average mass loss rate of 169±9 Gt per year between 1992 and 2020, similar to the mass lost from glaciers outside of Greenland and Antarctica (Fox-Kemer et al., 2021; Hugonnet et al., 2021). Over the same period, ice losses in Antarctica were predominantly occurring in West Antarctica (The IMBIE team, 2018; Otosaka et al., 2023). 

The long-term stability of the ice sheets depends on a complex interplay of amplifying (including self-sustaining) and damping feedbacks (e.g. Fyke et al., 2018). Based on multiple lines of evidence from modelling studies, observations and palaeo evidence, the ice sheets or parts thereof are considered ‘global core’ climate tipping systems (Armstrong McKay et al., 2022). In the following, we describe the underlying mechanisms, critical thresholds, timescales and potential for (ir)reversibility. Since the ice loss is dominated by different processes, we differentiate between the GrIS, the West Antarctic Ice Sheet (WAIS), the marine basins of East Antarctica and non-marine parts of East Antarctica.

Greenland Ice Sheet

The GrIS is a land-based continental ice sheet, with an area of 1.71 million square kilometres. At its margins, ice flows to the sea through marine-terminating outlet glaciers. The currently observed mass loss predominantly occurs through enhanced surface melting and iceberg calving (breaking at the edges) (King et al., 2020; Shepherd et al., 2020). Interactions with the atmosphere play an important role for the overall stability of the ice sheet. Several amplifying and damping feedbacks between the ice sheet and atmosphere are active in a warming climate, and these are associated with different timescales. On short timescales, a warmer climate will, on average, produce more precipitation via the added moisture-carrying capacity of the air. This mitigates some of the mass losses, since it increases accumulation (snow fall) as the climate warms. Atmospheric circulation and wind patterns will also change in response to a changing ice sheet geometry, but the effect on the overall mass balance (i.e. the balance between snow inputs and meltwater/calving outputs) of the ice sheet is not well understood. 

Evidence for tipping dynamics

Associated with surface melting is a self-amplifying feedback, the melt-elevation feedback (Oerlemans, 1981), in which substantial melt can cause parts of the ice sheet surface to sink to lower elevations, exposing the surface to warmer air masses which in turn can lead to further melt (Figure 1.2.5). This effect is compounded by the melt-albedo feedback: as snowpack melts to bare ice, surface albedo (level of reflection) reduces, leading to increased absorption of solar radiation. This in turn leads to further melting and a further albedo reduction (e.g., Box et al., 2012). Glacier algae growing on bare ice can lower albedo further, a process known as the biological albedo feedback (Cook et al., 2020). Both ice sheet modelling and palaeoclimate data indicate that a tipping point can occur when the melt-elevation feedback gets strong enough to support self-accelerating mass loss (Huybrechts, 1994; Robinson et al., 2012; Ridley et al., 2010; Levermann and Winkelmann, 2016).

On longer timescales (over the course of centuries to millennia), isostatic rebound can also act as a negative/damping feedback on ice sheet retreat (glacial isostatic adjustment (GIA); Whitehouse et al., 2019): a decrease in ice load leads to a slow rebound of the bedrock underneath – as the ice surface is thus lifted to higher elevations with colder surrounding air masses, this can lead to a reduction in surface melt, or even to net accumulation at the surface.

Current estimates for a critical threshold for the GrIS range from 0.8°C to 3°C of warming relative to pre-industrial levels, with a best estimate of about 1.5°C (Robinson et al., 2012; van Breedam et al., 2020; Noël et al., 2021; Höning et al., 2023). This is supported by palaeorecords which indicate that GrIS had at least partially retreated during the MIS-5 interglacial, and likely collapsed during MIS-11, which was 1-2°C warmer than pre-industrial (Christ et al., 2021). At lower warming levels, simulations with a coupled ice sheet atmosphere model indicate that additional atmospheric dynamic changes in precipitation patterns can restabilise the ice sheet, but above 2°C warming, positive/amplifying feedbacks leading to loss of the majority of the ice sheet cannot be overcome (Gregory et al., 2020).

While the respective warming threshold could be reached within the coming decades (Fox-Kemper et al., 2021; Tebaldi et al., 2021), the response times of the ice sheet are such that the ice loss and resulting sea level rise would unfold over several millennia (Robinson et al., 2012; van Breedam et al., 2020). The timescales of ice sheet decline depend on the magnitude of warming beyond this threshold, where stronger warming leads to a faster ice sheet decay (Robinson et al., 2012). Several studies further indicate a strong hysteresis of the GrIS, meaning that substantial ice loss is likely irreversible on multi-millennial timescales (Robinson et al., 2012; Höning et al., 2023).

Slow-onset tipping processes such as ice sheet collapse might also be able to withstand a short period of temperature overshoot if the overshoot time is short compared to the effective timescale of the tipping system (Ritchie et al., 2021). For ice sheets this overshoot time could be in the order of decades to centuries (Ritchie et al., 2021; Bochow et al.,2023), which might for example theoretically allow global warming to overshoot a tipping threshold of 1.5°C and return below it by 2100 without triggering ice sheet collapse (Armstrong McKay et al., 2022). However, such overshoot times are very uncertain, and given the distinct challenges of reducing global temperatures over short time horizons, this possibility should not be relied upon in policy.

Assessment and knowledge gaps

Given the broad evidence base, we have high confidence that the GrIS is a tipping system. This is in line with previous assessments (Fox-Kemper et al., 2021; Armstrong McKay et al., 2022). 

West Antarctic Ice Sheet (WAIS)

Since temperatures in Antarctica are generally lower than in Greenland (being centred over the South Pole) and the surface is generally brighter, there is overall less surface melt (Broeke et al., 2023). Recent observations show melt occurrences on ice shelves along the coastline of Antarctica, with most intense melting occuring on the Antarctic Peninsula (Trusel et al., 2013; Jakobs et al., 2020; Lenaerts et al., 2016; Stokes et al., 2019). In contrast to Greenland, however, the currently observed mass loss, especially in the WAIS, is dominated by ocean-induced melting at the underside of the floating ice shelves (e.g., Otosaka et al., 2023; Millilo et al., 2022; Paolo et al., 2015; Adusumili et al., 2020). 

Large parts of the WAIS are grounded below sea level (so-called marine basins), surrounded by floating ice shelves, and where these ice shelves are in contact with warmer ocean waters, melting at their base occurs. While the direct contribution to sea level rise of this ice shelf melting is negligible, it plays an important indirect role for the overall mass balance. Due to the thinning of the ice shelves, the buttressing (i.e. the backstress imparted to the grounded ice) is reduced, causing the movement of grounded ice upstream to accelerate, which in turn can lead to substantial sea level rise (Scambos et al., 2004; Rignot et al., 2004; Reese et al., 2018). Substantial ocean warming and ice shelf basal melting is committed in the Amundsen Sea over the 21^st Century, which will likely accelerate the retreat of several key WAIS outlet glaciers including the Thwaites and Pine Island glaciers (Naughten et al. 2023).

Evidence for tipping dynamics

Different amplifying feedbacks can lead to self-sustained ice loss from the WAIS once a critical threshold is passed (Figure 1.2.5). One of the key feedbacks is the marine ice sheet instability (MISI – Figure 1.2.6, top) (Weertman, 1974; Schoof, 2007; Mengel and Levermann, 2014; Feldmann and Levermann, 2015; Garbe et al., 2020), which can occur where the grounding – the separation line between the grounded ice sheet and floating ice shelves – sits on retrograde bedrock slopes. If the grounding line retreats into regions of greater ice thickness, for instance due to enhanced sub-shelf melting, this increases the flux across the grounding line, leading to further retreat. Such self-sustained retreat may be stabilised by the buttressing effect of ice shelves (Gudmundsson et al., 2012; Pegler, 2018; Haseloff and Sergienko, 2018).

The MISI has been suggested to have driven the collapse of WAIS during previous interglacials (Pollard et al., 2015; DeConto and Pollard, 2016; Sutter et al., 2016; Turney et al., 2020; Thomas et al., 2020; Weber et al., 2021). There is also palaeoclimate evidence for a collapse of WAIS and around 20m higher sea level (implying substantial Antarctic Ice Sheet loss) during ~2-3^oC warmer periods of the Pliocene, (Naish et al., 2009; Grant et al., 2019; DeConto et al., 2021). It has further been suggested that this instability might already be underway in the Amundsen Sea Embayment, including at the Thwaites and Pine Island glaciers (Rignot et al., 2014; Joughin et al., 2014; Favier et al., 2014; Turner et al., 2017; De Rydt et al., 2021). 

While a recent intercomparison study using three different ice sheet models (Hill et al., 2023) concluded that the current observed retreat of grounding lines in West Antarctica is not yet driven by this instability, mounting evidence from modelling studies (e.g., Reese et al., 2023; Seroussi et al., 2017; Arthern and Williams 2017; Golledge et al., 2021; Garbe et al., 2020) suggests that, unless the current warming trend is reversed to colder conditions in the near future, parts of the WAIS such as the Amundsen basin would be committed to long-term irreversible grounding-line retreat driven by MISI. The loss of the Amundsen basin alone would raise global sea levels by roughly 1.2 metres, (Morlighem et al., 2020). Additional large-scale ice sheet changes in West Antarctica could be triggered in the coming decades in response to projected warming. Due to the long response time of the ice sheet, the respective mass loss would unfold and sea level thus keep rising for centuries to millennia (Golledge et al., 2015; Winkelmann et al., 2015). 

Another proposed destabilising feedback mechanism is known as marine ice cliff instability (MICI – Figure 1.2.6, bottom) (Bassis and Walker, 2012; Bassis and Jacobs, 2013; Pollard et al., 2015; DeConto and Pollard 2016). The MICI hypothesis proposes that tall marine-terminating ice cliffs, which could result from ice shelf collapse, for example, are inherently unstable and could rapidly collapse, potentially associated with a self-reinforcing and irreversible inland ice retreat on both retrograde and prograde sloping marine beds. Such retreat would proceed until water depths shallow or the ice cliff is buttressed (DeConto and Pollard, 2016). The critical height of the ice cliff resulting in its failure depends on the ice properties and the extent of crevassing, but is currently poorly constrained (Bassis and Walker, 2012). In addition, processes potentially mitigating or slowing the self-sustained ice retreat due to MICI such as mélange buttressing or the speed of the preceding ice shelf disintegration introduce additional uncertainties (Clerc et al., 2019; Edwards et al., 2019; Robel and Banwell 2019; Schlemm et al., 2022; Pollard et al., 2018). Low confidence has been assigned to this process in the latest IPCC assessment (Fox-Kemper et al., 2021), partially because it has not yet been observed (Needell and Holschuh, 2023). 

Assessment and knowledge gaps

Based on these different lines of evidence, there is high confidence that the WAIS is a tipping system, with the potential for widespread, and at least partly irreversible ice loss. Recent estimates of the respective global warming levels at which such tipping dynamics are triggered range from 1°C to 3°C of warming compared to pre-industrial levels (Garbe et al., 2020; Golledge et al., 2017; Reese et al., 2023). This means that the complete decline of the WAIS could be triggered by warming projected under higher-emission scenarios for this century (Chambers et al., 2022; Golledge et al., 2015). 

Due to the complexity of interacting processes with the other parts of the climate system and their lack of representation in fully coupled (Earth system) models, it remains a challenging task to reduce the respective uncertainty range and project the resulting ice loss in the near future. For example, the potential effect of ocean stratification or solid-Earth feedbacks on grounding line migration is currently not well-constrained (e.d., Kachuk et al., 2020; Larour et al., 2019; Gomez et al., 2020; Coulon et al., 2021; Golledge et al., 2019). Given the high vulnerability of the WAIS and the far-reaching consequences of its potential collapse, it is important to narrow down the critical thresholds, and in particular the timing of the onset of potential large-scale retreat.

Marine basins East Antarctica

The East Antarctic marine basins include the Wilkes, Aurora and Recovery Basins, and 19.2 metres of sea level equivalent (Fretwell et al., 2013). They have been proposed as ‘global core’ climate tipping systems, due to the potential for instabilities in the marine ice sheet and ice cliff (Garbe et al., 2020; Armstrong McKay et al., 2022). The processes affecting the marine basins of East Antarctica are thus similar to those described above for the WAIS.   

Evidence for tipping dynamics

Outlet glaciers in the Aurora subglacial basin, for instance Totten and Denman glaciers, already experience acceleration, retreat and mass loss at present (e.g., Rignot et al., 2019; Shepherd et al., 2019; Rintoul et al., 2016; Li et al., 2015, 2016; Miles et al., 2021; Shen et al., 2018). There is limited evidence for change in Recovery and Wilkes basins in current observations (e.g., Gardner et al., 2018). However, palaeorecords and models suggest the ice margin may have undergone substantial retreat deep inland of Wilkes subglacial basin during Pleistocene interglacials (Blackburn et al., 2020; Wilson et al., 2018; Iizuka et al., 2023) and in warm periods of the Pliocene (Cook et al., 2013; DeConto et al., 2021; Blasco et al., 2023 [in review]) with global mean atmospheric warming of at least 1-2°C above pre-industrial, as suggested by palaeorecords (Blackburn et al., 2020). Other work has suggested that ice sheet retreat in the Wilkes subglacial basin remained relatively limited during the Last Interglacial, when Southern Ocean sea surface temperatures were about 1-2°C and Antarctic surface air temperatures were at least 2°C above pre-industrial averages (Capron et al., 2017; Hoffman et al., 2017; Chandler and Langebroek, 2021), placing an upper sea-level contribution from the Wilkes basin during that period at 0.4-0.8 m (Sutter et al., 2021). 

Recent model simulations show that the risk of substantial sub-shelf melt-induced or calving-induced ice loss and the associated timescales vary strongly for the individual subglacial basins (Garbe et al., 2020): A drainage of the Recovery basin may be driven by oceanic warming of 1-3°C (Golledge et al., 2017), while self-sustained grounding-line retreat in the Wilkes basin is initiated in models when exceeding an atmospheric warming of 2-4°C above present-day levels (Garbe et al., 2020; Golledge et al., 2017). The decay of the drainage basin may occur over a time period of centuries to tens of thousands  of years, as indicated in palaeorecords (Bertram et al., 2018) and model experiments (Mengel and Levermann, 2014), depending on the warming trajectory (DeConto and Pollard, 2016). Modelling studies suggest that ice loss from the Aurora subglacial basin is triggered when sustaining stronger warming of about 5-8°C above present-day levels (Garbe et al., 2020; Golledge et al., 2017; Winkelmann et al., 2015; Bulthuis et al., 2020; Van Breedam et al., 2020; Golledge et al., 2015). Palaeo evidence and models suggest that, once triggered, ice loss from these marine basins can only be reversed if the climate were to cool far below pre-industrial levels, leading to hysteresis behaviour (Garbe et al., 2020; Mengel and Levermann, 2014). 

Assessment and knowledge gaps

Being characterised by self-sustained dynamics as well as abrupt and irreversible changes beyond a warming threshold in various studies, we identify the marine basins of East Antarctica as parts of the cryosphere exhibiting tipping behaviour with high confidence, in line with previous assessments (Armstrong McKay et al., 2022). Further work is needed to better constrain existing estimates of critical thresholds and timescales from available ice sheet modelling and palaeoclimate data for individual subglacial basins – for example, by improving the treatment of sub-shelf melt and taking into account model and parametric uncertainty. 

Non-marine East Antarctica

In East Antarctica, a major part of the ice sheet initially built up at the Eocene-Oligocene transition is grounded above sea level (DeConto and Pollard, 2003; Liu et al., 2009; Morlighem et al., 2020; Hutchinson et al., 2021). At present, observations still indicate mass gain in this terrestrial part of the Antarctic Ice Sheet (for instance, in Dronning-Maud Land) though mass balance estimates are associated with high uncertainties (Otosaka et al., 2023; Schröder et al., 2019). As such, West Antarctic ice loss over the past decades was balanced to some extent by mass accumulation in East Antarctica (Medley and Thomas, 2019).  

Evidence for tipping dynamics

Long-term model assessments suggest that large-scale ice loss from terrestrial regions of East Antarctica may be induced for global mean atmospheric warming of 6°C or higher above pre-industrial levels (Garbe et al., 2020) until East Antarctica potentially becomes completely ice-free. Given the wide range of warming projected in the recent sixth phase of the Coupled Model Intercomparison Project (CMIP6), exceedance of respective critical forcing levels cannot be excluded beyond the end of this century under high emissions (e.g. SSP5-8.5 and SSP3-7.0 in the 22nd century; IPCC AR6 WG1 Ch4) in combination with a high climate sensitivity (Tebaldi et al., 2021). The disintegration of the land-based portions of the East Antarctic Ice Sheet may eventually raise global mean sea level by ~34 m (Fretwell et al., 2013), but unfolding over multi-millennial timescales (~10,000 years or longer) according to modelling studies (Winkelmann et al., 2015; Clark et al., 2016). 

Here, the melt-elevation feedback (similar to the GrIS) propels self-sustained mass loss by enhancing surface melt once the respective tipping point is crossed. It also gives rise to pronounced hysteresis behaviour with distinct stable ice sheet configurations within a range of climatic boundary conditions (Garbe et al., 2020; Pollard and DeConto, 2005; Huybrechts 1994). A strong cooling is consequently required for regrowth of the terrestrial East Antarctic Ice Sheet, and sustained cooling to at least pre-industrial temperature levels to recover its present-day volume and extents (Garbe et al., 2020). Due to this hysteresis, large land-based portions of the East Antarctic Ice Sheet persisted for more than 8 million years (Shakun et al., 2018) through the warm intervals of the early to mid-Miocene, 23-14 million years ago (Gasson et al., 2016; Levy et al., 2016). 

Assessment and knowledge gaps

Self-amplifying feedback mechanisms (such as the melt-elevation feedback) can occur in East Antarctica, contributing to abrupt and irreversible ice sheet changes with a substantial impact through sea level rise beyond a critical threshold. There are few modelling studies on multi-millennial timescales covering the warming range that may be relevant for the potential nonlinear response of the terrestrial ice sheet in East Antarctica. Thus, there is medium confidence in the assessment of the non-marine East Antarctic Ice Sheet as a cryospheric tipping system. Reducing uncertainties in temperature thresholds and timescales of collapse requires multi-model ensembles and better representation of ice surface processes, as well as the inclusion of interaction with the rest of the climate system. Additionally, more research on how climate forcing varies regionally and interacts with regional processes and feedbacks would help better constrain the drivers and timescale of tipping.