1.2.2.4 Permafrost

Permafrost is defined as ground frozen for at least two consecutive years (Van Everdingen, 2005) (Figure 1.2.11). Permafrost underlies about 14 million sq km (15 per cent of the land surface area) in the Northern Hemisphere (Obu, 2021), mainly in Russia, Canada, the US (Alaska), and China (Tibetan Plateau). In addition, there is about 2.5 million sq km of relict permafrost in the Arctic shelf seafloor (Overduin et al., 2019), which was submerged by rising sea levels at the end of the ice age.

Permafrost landscapes are complex. They commonly exhibit an active layer, which is the uppermost layer of soil or ground that thaws during the warmer months of the year and freezes again during colder months (Figure 1.2.12). Permafrost is further characterised by factors such as variable topography, ground ice presence, vegetation dynamics, and soil climatic conditions. For example, the presence of hills, valleys and slopes affects the distribution and characteristics of continental permafrost at different spatial and temporal scales. The interaction and feedback between these factors contribute to the complexity of permafrost environments and suggest a variety of potential responses of the permafrost domain to climatic changes.

Vast amounts of organic carbon and ground ice that accumulated during past cold climates in northern high latitudes are still preserved in permafrost today. The frozen conditions in permafrost soils prevent the microbial decomposition of organic material accumulated in the past during relatively warm summers. Currently, it is estimated that the upper three metres of permafrost soils contain about 1,035 ±150 GtC (Hugelius et al., 2014) or about 50 per cent more than today’s atmosphere (Figure 1.2.13). Subsea permafrost stores additional organic matter estimated at between 560 (Sayedi et al., 2020) and 2,822 (1,518-4,982) GtC (Miesner et al., 2023). Further, permafrost also contains or caps large quantities of frozen methane and other gases. Such deposits are known as permafrost-associated gas hydrates and a conservative estimate suggested that about 20 GtC are currently locked in permafrost-associated gas hydrates (Ruppel, 2015).

Over the last four decades, the Arctic warmed almost four times faster than the rest of the globe (Rantanen et al., 2022). Ongoing climate change causes thawing of permafrost soils (Schuur et al., 2015, 2022; McGuire et al., 2018), which leads to the subsidence, erosion and potential collapse of the previously frozen ground in regions of diverse permafrost landforms. The degradation of organic matter and the dissociation of permafrost-associated gas hydrates are linked to the release of carbon dioxide (CO₂) and methane (CH₄) into the atmosphere as a consequence of permafrost thaw. This carbon loss is irreversible over several centuries. These permafrost carbon emissions contribute to a positive climate feedback in which GHG emissions lead to additional warming, which, in turn, releases more GHG. This is called the permafrost carbon-climate feedback (Koven et al., 2011, Schuur et al., 2015, 2022, Canadell et al., 2021).

Current-generation climate models suggest a net positive impact of the permafrost carbon-climate feedback on global climate with estimates of additional warming of 0.05-0.7°C by 2100 (Schaefer et al., 2014, Burke et al., 2018, Kleinen and Brovkin, 2018, Nitzbon et al., 2023) based on low- to high-emissions scenarios, respectively. Methane emissions from permafrost could temporarily contribute up to 50 per cent of the permafrost-induced radiative forcing due to its higher warming potential (Walter Anthony et al., 2016, Turetsky et al., 2020, Miner et al., 2022). Overall, however, Canadell et al., (2021) summarise that “thawing terrestrial permafrost will lead to carbon release (high confidence), but there is low confidence in the timing, magnitude and relative roles of CO₂ and CH₄” of the permafrost carbon-climate feedback.

In addition, permafrost thaw impacts society in the permafrost region through changes at the land surface, e.g. wetting or drying of landscapes, ground subsidence due to melted ice, damaged infrastructure (roads, buildings, pipelines), and ecosystem changes such as ocean acidification or eutrophication (Hjort et al., 2018, 2022; Miner et al., 2021, Langer et al., 2023) (see Chapter 2.2 for societal impacts).

Evidence for tipping dynamics

Permafrost thaw is commonly denoted as gradual or abrupt. On land, gradual thaw occurs wherever the upper layer of thawed soil (active layer) gets successively deeper every year. Based on current projections, there is a high level of confidence that continued warming will result in ongoing, gradual declines in the volume of near-surface permafrost. It is anticipated that for every additional 1°C of warming, there will be a 25 per cent reduction in the global volume of perennially frozen ground found near the surface (Arias et al., 2021), which happens over the course of years to decades. The associated decomposition of permafrost carbon takes place on longer timescales, from centuries to millennia. 

These models also suggest that the amount of carbon released from gradual thaw is roughly proportional to the amount of global warming in low- to high-emission scenarios, with the best estimate being 18 (3-41) GtC per degree of global warming (Canadell et al., 2021; Burke et al., 2017, 2018). Permafrost carbon release represents a relatively higher contribution to the remaining carbon budget for low-emission scenarios (Gasser et al., 2018; Kleinen and Brovkin, 2018), specifically when the permafrost carbon-climate feedback is taken into account in the carbon budget estimates (Canadell et al., 2021). 

Abrupt or rapid thaw occurs where excess or massive ice is present in the ground and leads to the development of ‘thermokarst’. When the ice melts and drains away, the land surface subsides. This leads to the development of characteristic landforms such as thaw lakes, thaw slumps, or eroding gullies and valleys (Figure 1.2.14). Their development is reinforced by increased heat conductivity of water and the decreasing stability of water body edges that further increases their size. Thus, these processes can permanently transform permafrost landscapes. Environments in which these processes are expected to occur are estimated to cover about 20 per cent of the present Arctic permafrost region (Olefeldt et al., 2016). 

Thermokarst processes can occur in response to local disturbances or across regions experiencing rapid warming or extreme events, and positive/amplifying feedbacks can drive rapid permafrost loss (Nitzbon et al., 2020). Further, it is estimated that carbon emissions related to abrupt thaw processes could contribute an additional 40 per cent of emissions from newly formed features such as thaw slumps and thermokarst lake and wetland formation, which may double the radiative forcing from circumpolar permafrost-soil carbon fluxes (Turetsky et al., 2020; Walther Anthony et al., 2018). However, these processes are dependent on local environmental conditions that are unevenly distributed across the permafrost region (Olefeldt et al., 2016). Thus, despite the rapid nonlinear response at local-to-regional scale, the permafrost thaw and carbon emissions from thermokarst processes are likely to aggregate to a near linear response globally (Nitzbon et al., 2023). 

The loss of ground ice and the ecosystem changes are irreversible, with many local implications on topography and hydrology, including subsidence, drying or wetting, and changes in the microbial communities. In this context, microbial heat production was hypothesised as a possible self-reinforcing feedback on permafrost thaw (Khvorostyanov et al., 2008, Hollesen et al., 2015), but a consequential abrupt release of permafrost carbon through this ‘compost bomb’ mechanism (Clarke et al., 2021) is assessed to be unlikely. It would require organic carbon of very high quality and large quantity as well as comparably low ice contents, but such environmental preconditions are not prevailing over vast areas of the permafrost region. Accordingly, large-scale modelling studies found this effect to be of minor (Koven et al., 2011) or negligible (de Vrese et al., 2021) relevance to future projections of permafrost region carbon emissions.

While nonlinearity of the permafrost response to warming is exemplified in rapid thaw on local-to-regional scales, it is uncertain how these changes propagate to a larger scale. Some studies argue that an interaction of local feedbacks could lead to a quasi-linear response on a global scale (Schuur et al., 2015, Chadburn et al., 2017, Hugelius et al., 2020, Nitzbon et al., 2023), while others found multiple stable states in the permafrost system with potential nonlinear response on a large scale (de Vrese and Brovkin, 2021). 

For the permafrost carbon-climate feedback to have large-scale tipping behaviour, it must be strong enough to cause self-sustaining permafrost loss beyond a certain warming threshold at either a global or subcontinental scale. Current AR6-based estimates yield a small positive amplification factor, indicating that the permafrost carbon-climate feedback is too small to be self-perpetuating on a global scale (Nitzbon et al., 2023). However, for future projections, both ‘offline’ permafrost models and Earth system models do not capture large-scale abrupt thawing throughout the Arctic.

Important processes such as interactions between fire, vegetation, permafrost, and carbon, as well as the potential for sudden releases through thermokarst phenomena, are currently not consistently considered (Natali et al., 2021). As a result, existing projections of permafrost thaw under various temperature thresholds are likely to be underestimates, indicating that the actual thaw potential may be greater than currently predicted.

Since the flooding of the Arctic shelf after the last ice-age, the ocean floor has been exposed to relatively slow warming with small seasonal changes. Therefore subsea permafrost is thawing at a slow but continuous rate, leading to carbon emissions of 0.048 (0.025-0.085) Gt/yr (Miesner et al., 2023), an order of magnitude smaller than terrestrial permafrost carbon emissions (Figure 1.2.15). The disappearance of sea ice that has an insulating effect on ocean water temperature or major circulation changes in the Arctic Ocean may accelerate gradual thaw of subsea permafrost (Wilkenskjeld et al., 2022). However, this degradation process happens too slowly to support abrupt methane release (Reagan and Moridis, 2007; O’Connor et al., 2010). In addition, permafrost-associated gas hydrates within and below subsea permafrost are stabilised by the temperature and pressure conditions created by the permafrost. Permafrost thus acts as a lid on these GHG reservoirs and warming is expected to take centuries to penetrate them (Dmitrenko et al., 2011; Marín‐Moreno et al., 2013). Some of these hydrates are relict deposits that are not necessarily stable under current conditions, but are self-preserving. 

Subsea permafrost thaw only shows a delayed and dampened response to climate warming. In addition, microbial degradation rates are slow and strong methane sinks in both sediment and ocean likely limit net GHG emissions (James et al., 2016, Ruppel and Kessler, 2016). Another important aspect is the long timescale of permafrost thaw. Instantaneous changes in GHG emissions are quasi-linear, but committed changes on a centennial-to-millennial timescale could be nonlinear – as, for example, when a large area with frozen carbon storages is simultaneously affected by a strong warming. An example from palaeoclimate is a stepwise increase in atmospheric CO₂ concentration in response to an abrupt warming at about 14,700 years ago, plausibly explained by the permafrost thaw (Köhler et al., 2014).

Assessment and knowledge gaps

Accounting for its potential nonlinear response to warming, permafrost was considered a tipping system in numerous previous assessments (Armstrong Mckay et al., 2022, Fabbri et al., 2021, Yumashev et al., 2019, Schellnhuber et al., 2016, Steffen et al., 2018, IPCC AR6, Hamburg Climate Future Outlook). However, the aggregation of nonlinear or rapid local-to-regional permafrost degradation as a result of global warming results in a quasi-linear transient response of global permafrost extent on decadal to centennial timescales (Burke et al., 2020). The resulting permafrost carbon-climate feedback is likely positive, but current climate conditions do not support its self-sustenance, hence permafrost thaw is not expected to cause runaway global warming. 

We conclude that permafrost exerts localised tipping points, which, however, do not aggregate to a large-scale tipping point at a global temperature threshold on decadal to centennial timescales. Similarly, subsea permafrost thaw happens relatively slowly, resulting in carbon emissions a magnitude smaller than from terrestrial permafrost. According to the strength of the available evidence, we have medium confidence in these assessments of both land and subsea permafrost. 

The communication of a specific tipping threshold for permafrost could give a false sense of a temperature ‘safe zone’ at which permafrost is less vulnerable.

The effects of permafrost degradation are already seen today with implications for ecosystems and societies, where committed changes will continue to be relevant for centuries.

Given the current modelling limitations, improvements in modelling permafrost dynamics will improve the confidence of evaluating permafrost stability, carbon loss, response linearity, and their impact on global climate.