What does the new IPCC report say about climate tipping points and feedbacks?

The Intergovernmental Panel on Climate Change (IPCC) released the first part of its sixth assessment report (AR6) in August. Much of the media attention focused on what the new report had to say about tipping points in particular, especially following an earlier leak of a draft of the report’s second part.

Now delegates are gathering in Glasgow for the ‘COP26‘ UN climate talks, where the aim is to ramp up global policy commitments to keep the Paris Agreement target of 1.5°C alive – a target seen as increasingly important for avoiding climate tipping points.

In this article we’ll focus in on what the new report has to say so far about climate tipping points and feedbacks, and what they mean for climate action.

The IPCC reports

Every six to seven years the IPCC compiles an assessment report which brings together and assesses all of the advances in climate science in that time on behalf of the UN’s member states. The recent release is the first of several parts (this one on physical science basis; parts two and three on impacts & adaptation and mitigation will be published in 2022), and is the biggest update from the IPCC since its one-off special reports on 1.5°C, land use, and the ocean & cryosphere in 2018 and 2019. The last big report (AR5) was in 2013-14, so AR6 brings together a powerful summary of recent climate science.

Its headline conclusions are simple: human-caused climate change is unequivocal, its size and rate is unprecedented, and it’s already causing weather and climate extremes across every inhabited region.

Maps showing how hot extremes are already increasing in almost every inhabited region (top), heavy rainfall is increasing in Eurasia, parts of North America, and Southern Africa (with limited data elsewhere), and droughts increasing in parts of Africa, Eurasia, Western North America, North-Eastern South America, and Southern Australia. [AR6 WG1 Fig. SPM.3]

For the future, it projects that the Paris Agreement limits of 1.5°C and 2°C will be passed in the 2030s and 2050s if emissions don’t immediately peak and rapidly decline over the next few decades. If emissions aren’t stopped we can expect even worse weather extremes, greater sea level rise, weakening natural carbon sinks, and a non-negligible risk of triggering low-probability high-impact events like tipping points.

Graphs showing future climate change projections based on very low (‘SSP1-1.9′, light blue lines), low (SSP1-2.6’, dark blue lines), medium (‘SSP2-4.5’, orange lines), high (‘SSP3-7.5’, light red lines), and very high (‘SSP5-8.5’, dark red lines) emission scenarios. Current policies are on a trajectory between SSP2-4.5 and SSP3-7.0, while the Paris Agreement aims for SSP1-2.6 or lower. [AR6 WG1 Fig. SPM.8]

Earlier leaks from the draft report triggered media stories about the IPCC making bolder warnings about climate tipping points. The stories mention that a dozen different climate tipping points have been identified by the IPCC, but these 12 tipping points have not yet been clarified as the leaks were from the report’s draft second part and only the first part has been released so far. However, the first part still features plenty of discussion of climate tipping points and feedbacks, which we’ll focus on next.

Carbon Cycle Feedbacks

Charts showing that natural carbon sinks are continuing to take up large amounts of human CO₂ emissions – for now. [AR6 WG1 FAQ5.1 Fig.1]

First of all, let’s look at feedbacks, in particular ones that could lead to extra carbon emissions from nature (known as ‘carbon cycle feedbacks’).

Key carbon cycle feedbacks include the weakening of the natural carbon sinks that currently take up around half of humanity’s emissions, and emissions from permafrost thawing in response to warming. These feedbacks can reduce the amount of carbon that can be emitted by humans before the 1.5-2°C Paris Agreement limits are reached – the ‘carbon budget‘ – and amplify the warming we’d expect for a given emission scenario.

Unlike what is sometimes claimed or implied, IPCC projections do in fact take account of carbon cycle feedbacks, albeit often in a simplified way. As explained in this Carbon Brief analysis, in order to assess remaining carbon budgets, in previous reports the IPCC took future climate projections of climate models without carbon cycle feedbacks and then added a best-estimate of these feedbacks back on (based on models that include these feedbacks).

For example, in the IPCC’s 2018 Special Report on 1.5°C, the carbon budget was arbitrarily reduced by 100 gigatonnes of CO₂ (GtCO₂, i.e. a billion tonnes of carbon dioxide) to account for unquantified feedbacks. This was a reasonable approximation, but a more recent analysis showed that a fuller representation of carbon cycle uncertainties expanded the range of possible warming for a given emission scenario, with this range skewing towards extra warming. If carbon cycle feedbacks end up at the high end then warming would be more than expected, which means there would be less carbon budget available for staying within 1.5°C.

In AR6 the IPCC now takes a more robust approach to account for the impact of carbon cycle feedbacks on the carbon budget. Joeri Rogelj, a lead scientist on the carbon budgeting process, explained that: “Taking into account not only permafrost thaw, but also a host of other biogeochemical and atmospheric feedbacks… remaining carbon budgets have to be reduced by 26 ± 97 GtCO₂ per °C of additional warming.” This is based on the figure below, with carbon cycle feedbacks shown in panel (b) having a negative (i.e. dampening) effect on climate change.

In combination with other improvements, the end result is that AR6 carbon budgets are fairly unchanged from 2018, indicating that the previous feedback simplification wasn’t too far off. Feedback estimates are also fairly unchanged between the previous two model generations (called CMIP5 & CMIP6 – hatched vs. solid bars in panel b below), with more recent land vegetation models tending to give slightly weaker negative feedbacks than before.

Chart showing the estimated strength of various ‘biogeochemical’ feedbacks (i.e. feedbacks via release of chemicals like carbon dioxide (CO₂), methane (CH), and nitrous oxide (NO) from the land and ocean in response to rising CO₂ and warming. In general, both the land and ocean tend to counteract increasing CO₂ (and so act as a ‘negative’ stabilising feedback, blue bars to left) but tend to slightly amplify warming (acting as a weaker ‘positive’ amplifying feedback, red bars to right). The overall effect are both the land and ocean acting as a net negative/stabilising feedback tending to counteract human-driven global warming. [AR6 WG1 Fig. 5.29 a & b]

As for the ‘core’ emission scenario and warming projections (shown in the future projection graphs earlier), AR6 still uses models without carbon cycle feedbacks but concludes that, although the combined effects of all feedbacks is to amplify the warming, this does not strongly affect the AR6 projections of warming this century.

Much of the carbon cycle feedback is determined by the response of natural carbon sinks to both CO₂ and to warming. The AR6 figures above (FAQ5.1) and below show how natural carbon sinks continue to take up a large proportion of the carbon released by humans, but the proportion shrinks with more global warming. So far natural carbon sinks have taken up over half (~54% in the 2010s) of anthropogenic emissions, with ~23% dissolving in to the ocean, causing acidification, and ~31% taken up by extra plant growth on land.

Chart showing the proportion of CO₂ emissions that are taken back up by “natural carbon sinks” (the ocean, causing acidification, and land, causing plant growth). The longer CO₂ emissions carry on in future for the less natural system will draw down for us, effectively acting as a positive feedback. [AR6 WG1 SPM.7]

If we stopped emissions now these carbon sinks would continue to draw down emissions for a while, reaching ~70% of human emissions in the Paris-compliant low emission ‘SSP1-1.9’ scenario (on the left). But with higher emissions, carbon sinks start to struggle – the ocean can dissolve less CO₂ at higher temperatures, and droughts and excess tropical warmth limit extra plant growth – so the proportion of human emissions they take up starts to fall, reaching only ~38-44% in the high emission scenarios on the right. This reduces the carbon budget available for a given warming level, and so acts like a positive/amplifying feedback.

Overall then, the picture on natural carbon sinks and feedbacks has remained similar in AR6, with updates mostly being some more constraints on likely future carbon sink uptake and to lay it out more clearly. But what about tipping points and the possibility of “low-impact high-probability” events?

Tipping Points

One of the key tables in AR6 WG1 (Table 4.10) listing potential tipping points, irreversible shifts, and abrupt events, and summarising the IPCC’s current conclusions on the evidence and likelihood for each. Note that not all of the above are considered to be tipping points by the IPCC. Current version is from the approved draft (and so has a watermark across it), and will be replaced when the final version is published.

Permafrost & methane

Permafrost soils contain a huge amount of carbon, and as they thaw with warming some of this carbon will degrade and be released as CO₂ and methane (CH₄). This makes permafrost both a key carbon cycle feedback and part of the cryosphere, but as it stores so much carbon and has the potential to become a tipping point it deserves special attention.

Maps showing where permafrost (permanently frozen soils) is storing carbon in the Arctic region (left), and where they may be vulnerable to abrupt thawing and therefore release of some of this carbon as extra CO₂ emissions as a result of global warming (right). [AR6 WG1 FAQ5.2 Fig.1]

In contrast to AR5, a few more Earth system models used by AR6 now include some (but not all) permafrost thaw processes, but as not all models include them AR6’s permafrost thaw projections are largely based on separate permafrost-enabled land surface models.

AR6 projects that for each degree of global warming humans cause, permafrost will likely emit around an additional 18 gigatonnes of carbon (GtC) by 2100 (with an uncertainty range from 3 to 41 GtC) in the form of CO₂, reaching a total release of ~240 GtC of CO₂ in the 21st century under a high emission scenario. For context, human CO₂ emissions are currently around 10 GtC a year so permafrost emissions represent an extra few years’ worth of current emissions per degree of warming, and ~24 years’ worth in an extreme case. Methane – which is a rarer but more powerful greenhouse gas than CO₂ – boosts the effect of these emissions by 3 GtC or so. However, abrupt thaw processes are currently poorly constrained and don’t feature in most models, and could boost emissions further.

Clathrates (also known as methane hydrates) are also mentioned, and are assigned a “very low” probability of tipping and a very likely small release of methane in the 21st century (a likely maximum of 20ppb). This is because it takes so long for excess heat from global warming to reach the seafloor and then travel through the heat-resistant sediment below, and clathrates themselves have self-stabilising mechanisms. Instead, clathrates are likely to act as a very long-term positive feedback on whatever level temperatures stabilise at, slowly releasing additional methane to the ocean and atmosphere. This is likely what occurred in the Earth’s past, with clathrates likely acting as a feedback maintaining the 5°C of warming during the ‘PETM’ event 55 million years ago.

Some potential climate-biosphere tipping points and their maximum impacts from the latest IPCC report [AR6 WG1 Table 5.6]. Image will be replaced by copy-edited version when available.

Forest Dieback

The Amazon rainforest generates up to a third of its rainfall recycling the moisture from rainfall in to more rainfall further inland, helping to sustain itself in areas that would otherwise be too dry to support a rainforest. This means that if rainforest downwind of these otherwise dry areas is lost (as a result of deforestation or climate change-induced drought or wildfires) then these drier areas can reach a tipping point and ‘die back’ to a grassy savannah state instead. However, only some of the models used by the IPCC show this as a dramatic instability, and so AR6 gives a low likelihood for Amazon dieback happening this century. If all tropical rainforests around the world were to die back, though, it’d release up to 200 GtC as CO₂, which would add an extra ~0.3°C of warming; previous estimates suggest a maximum release by the Amazon of 53-70 GtC, leading to ~0.1-0.2°C warming.

Boreal forests – the dense conifer forests around the edge of the Arctic, also known as the ‘Taiga’ – may also be vulnerable to dieback in places, with drought, climate-induced beetle infestations, and wildfire creating abrupt dieback tipping points in some models. This doesn’t happen in all models though, with some newer models being less sensitive to dieback processes (although these models still lack key ecosystem dynamics, and so may be underestimating this sensitivity). As a result AR6 gives boreal dieback a low likelihood in the 21st century as well, and gives a maximum carbon release of 27 GtC that’d likely be partly counterbalanced by boreal forest expansion in to the warming Arctic tundra.

Ice sheets & sea ice

Schematic showing how slowly ice sheets have grown and shrunk during the past ‘Ice Age’ glacial and drove large sea level changes (top), how ocean warming can drive the melting of ice sheets sitting below sea level and become self-sustaining (bottom left), and how atmosphere warming drives melting of ice sheets above sea level and also become self-sustaining (bottom right). [AR6 WG1 FAQ 9.1 Fig. 1]

Several parts of the Earth’s cryosphere – places dominated by ice, including ice sheets, sea ice, and permafrost – are described as featuring either tipping points, abrupt events, or irreversible changes in AR6, in keeping with other recent reports and analyses.

The West Antarctic ice sheet (WAIS) is listed by AR6 as a potential tipping element, with the possibility for abrupt and irreversible change. However, although some studies find a potential tipping point at 1.5-2°C and many studies find total WAIS loss committed at 2-3°C (pg.9-78), AR6 states in its summary that there is limited evidence for irreversible loss below 3°C (pg.9-9). This doesn’t so much mean the IPCC doesn’t think a WAIS tipping point exists, rather that there are simply not enough studies available to be sure if it could happen below 3°C. Irreversible loss is definitely expected beyond 3°C though, adding 3m+ to sea level rise over the next 2000 years and significantly extending the upper possible limit under high emissions. Subglacial basins in East Antarctica (including the Wilkes and Aurora basins) experience some loss below 3°C, and likely collapse at 3-5°C.

In contrast to the WAIS, the Greenland ice sheet (GrIS) is not described as a tipping element proper in AR6. This is based on a recent paper which didn’t find the same sharp transition to self-perpetuating ice loss as previous studies (pg.9-62). However, this paper did show that once around half the ice sheet was lost – likely somewhere between 2 and 3°C of warming – then full recovery became impossible even if temperatures returned to previous levels (something scientists call ‘hysteresis’), indicating some self-maintaining feedbacks are lost at this point. Some scientists might argue this is sufficient for GrIS to still be considered a tipping point, and that other studies did find a clearer threshold for self-sustaining ice loss, but either way it is clear that ~2°C+ destabilises the GrIS, and 3°C+ would cause irreversible GrIS loss and add 7m+ to sea level rise over the next 2000 years.

Graphs showing the long-term sea level rise commitments from different components (thermosteric = water expansion from warming). These indicate how the current trajectory of ~3°C of warming would result in up to between 3 and 20m of sea level rise from Antarctic ice sheet melt (including all of WAIS), at least half if not all of the Greenland ice sheet (max. 7m), and almost all glaciers (max. ~0.3m) [AR6 WG1 Fig. 9.30]

AR6 confirms that summer ice-free conditions will first periodically appear in the Arctic Ocean before 2050 in all emission scenarios, with ice-free years common beyond ~2°C of warming. But despite often being described as a tipping point case-study, AR6 confirms that Arctic summer sea ice doesn’t actually feature tipping dynamics itself (pg.9-48). This is because there’s no clear threshold beyond which further summer sea ice loss becomes self-sustaining, a key tipping point requirement. In contrast, Arctic winter sea ice often experiences a rapid (self-sustaining but reversible) collapse in models. Antarctic sea ice decline is also considered by AR6 as potentially abrupt, but is much more uncertain than the Arctic.

Graph showing the decline of Arctic sea ice in the summer at different levels of future warming in one model. Less than 2°C warming saves some summer sea ice, but reaching 3°C leaving the summer Arctic Ocean practically ice-free. [AR6 WG1 Fig. 4.5]

Ocean Circulation

The ‘Atlantic Meridional Overturning Circulation’ (AMOC) describes the northward movement of warm surface water up the Atlantic that then sinks off of Greenland (as a result of getting denser due to cooling and getting saltier as sea ice forms and takes up freshwater) and flows back down the deep Atlantic. Global warming, however, reduces sea ice formation and increases freshwater run-off from Greenland, and so interferes with this sinking. This will gradually weaken the AMOC, and there’s evidence from palaeoclimate records and some models that it could abruptly switch to a weaker mode. This would trigger severe cooling around the North Atlantic and disrupt the global monsoon, with potentially devastating consequences.

Schematic from AR6 WG1 (FAQ9.3, Figure 1) showing how both the wind-driven Gulf Stream and the thermohaline circulation-driven AMOC change with global warming from now (left) to future (right).

There is a mixed picture for the AMOC’s future in AR6, with confidence in historical decline actually reduced due to limited data and model mismatches. In keeping with recent reports, though, it judges that it is very likely that the AMOC will decline in the 21st century (albeit with low confidence on timings and magnitude) but that abrupt collapse is very unlikely before 2100 (and as likely as not by 2300). However, it’s noted that the models used for these projections tend towards over-stability and don’t feature the impact of run-off from the Greenland ice sheet. As a result, the IPCC now only has a medium confidence that the AMOC won’t collapse before 2100, and collapse remains possible in the longer term.

Graph showing the AMOC declining in models by about the same amount across all emission scenarios, but with no full collapse by 2100. [AR6 WG1 Fig. 4.6]

The IPCC have also included weakening of the Southern MOC (the equivalent around Antarctica to the northern AMOC) as a potential tipping point or irreversible/abrupt event. This is not typically included as a major tipping point in many analyses, but like with the AMOC a melt-induced shutdown would have global consequences. Models have struggled to capture the details of this ocean circulation, including lacking the impact of glacial meltwater on the nearby ocean, leading to lower confidence in these projections, but there have been some recent improvements. Current observations suggest the deep water mass it feeds (‘Antarctic Bottom Water’) has already started to shrink, and palaeo evidence suggests past meltwater pulses also weakened or even stopped this circulation. Based on this, the IPCC concludes with medium confidence that further weakening will continue through the 21st Century.

Other potentially irreversible or abrupt shifts

As well as potential tipping points, in AR6 the IPCC have also listed several potentially abrupt and/or irreversible shifts (in some cases in the same table, making it sometimes confusing as to whether they are considered to be tipping points by AR6 as well).

Another key table in AR6 WG1 (Cross-chapter Box 12.1, Table 1) listing potential tipping points, irreversible shifts, and abrupt events, and summarising the IPCC’s current conclusions on the evidence and likelihood for each. Note that not all of the above are considered to be tipping points. Current version is from the approved draft (and so has a watermark across it), and will be replaced when the final version is published.

For example, both ocean acidification and deoxygenation are listed as irreversible in the deep ocean as a result of the very slow turnover time for water in the ocean, meaning heat and CO₂ entering the ocean now will remain for centuries. They are also listed as potentially abrupt as a result of sudden slowdowns in ocean circulation (such as SMOC or AMOC shutdown) that’d stop as much CO₂ or heat being taken in to the deep ocean. Neither acidification nor heat storage have clear tipping points in themselves though, although the impact of acidification is non-linear and becomes severe beyond an ‘undersaturation’ threshold that can be reached in polar waters. There is also now robust evidence that coastal acidification has negative effects on calcifying organisms, including coral reefs.

Sea level rise and ocean heat content are also not usually considered to be tipping elements in themselves, but are listed by the IPCC as subject to potentially abrupt or irreversible change. Both are considered irreversible for centuries due to the slow nature of ocean warming and ice melt, and long-term sea level rise commitment is heavily dependent on ice sheet tipping points (especially beyond ~3°C). However, neither directly feature tipping points in themselves, and are mostly listed for their irreversibility.

Figure from AR6 WG1 (Fig. 8.27) showing how AMOC collapse affected rainfall around the world during the ‘Younger Dryas’ event around 12,000 years ago (left) and model projections for AMOC collapse in the future (right). AMOC collapse would also lead to warming in the North Atlantic region being partially cancelled out, while warming in the Southern Hemisphere would be marginally amplified.

Another system not usually considered as a tipping point but listed alongside by AR6 is the ‘Global Monsoon’. This describes the switching of circulation and wet/dry patterns in certain regions (Central America, Central South America, southern Africa, African Sahel, South Asia, and northern Australia) in response to the seasonal cycle. It’s expected that warming will lead to global monsoon regions getting bigger and wetter (although monsoon circulation itself weakens) while drier regions elsewhere get even drier. It’s also likely that this has already started within the last 40 years, but with a lot of noise confusing this signal. An abrupt or irreversible shift is not expected unless forced by AMOC collapse, which palaeorecords and models suggest would lead to drying in the African Sahel region, Central America, the northern Amazon, and Europe (see maps above).

The multiyear global shifts in weather driven by ‘El Niño’ and ‘La Niña’ (more fully the El Niño–Southern Oscillation, or ENSO) are expected to remain a strong feature of the Earth’s climate and become more intense with warming. In particular, increased rainfall variability and an eastward shift over the North Pacific and North America are considered very likely in AR6. However, no tipping behaviour or irreversibility is expected, in contrast to earlier worries that ENSO could tip to a ‘permanent El Niño’ state. Instead, a gradual intensification is expected with further warming, especially beyond 3°C of warming.


The latest IPCC report shows how far climate science has advanced in the previous ~8 years. The impact of carbon cycle feedbacks on the carbon budget and the decline of natural carbon sinks in response to warming is now much clearer, and the possibility of tipping points is discussed more directly. There are still some pretty big gaps in our knowledge on climate feedbacks and tipping points though, with AR6 holding off on giving estimates for likely global warming thresholds and impacts for many of the proposed tipping elements. And while the Earth system models used by the IPCC have improved since the last big reports they still struggle to include some key processes, making for example projections on future ice sheet loss or permafrost thaw more uncertain. Palaeoclimate evidence helps fill this gap though, indicating that tipping points were likely crossed during past episodes of climate change and could well be crossed again now. Even with these uncertainties it’s clear that further global warming only increases the risk of reaching tipping points and weakening natural carbon sinks, in particular beyond 2°C, making the case for keeping to the Paris Agreement target of 1.5°C even stronger.


This post was written by Dr. David A. McKay, a Climate-Biosphere Scientist working as a Freelance Research Consultant and Science Communicator at Georesilience Analytics. He’s currently working with Future Earth on the Earth Commission project, providing Earth system analysis and modelling support on setting safe and just limits for nutrients (focusing on ocean impacts) and for the climate (focusing on climate tipping points). This post and site are not externally funded, and the post was proofread and edited by Dr. Rachael Avery.

Featured Image: IPCC WG1 front page artwork – Changing by Alisa Singer (“As we witness our planet transforming around us we watch, listen, measure … respond.”) www.environmentalgraphiti.org – 2021.

1 Comment

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s