A paper presenting an updated assessment of climate tipping points was published yesterday in the journal Science. Here, the paper’s lead author (and publisher of climatetippingpoints.info) Dr. David Armstrong McKay explains the paper’s findings and the implications for climate action.
Over the past two decades climate tipping points have become a major global concern.
Research on climate tipping points was catalysed by the publication of the first major assessment of them in 2008, which found that nine parts of the climate system (named “tipping elements”) could feature such points beyond which non-linear change occurs, and provided a definition.
Since then there have been major advances in climate science, with improved climate models, more observations, and new records of ancient climate change helping to better understand how and when tipping points could occur in the climate system. New tipping elements (e.g. permafrost) have been proposed, and other proposals (e.g. monsoons) have been questioned.
This improved understanding has led to estimates of tipping point warming thresholds dropping over the years. In the 2008 assessment, most of the identified tipping elements had tipping points around 3-5°C, but the IPCC’s most recent report stated that climate tipping point risks emerge above 1°C, become high around 2°C, and reach very high around 2.5-4°C. That means many tipping points are probably closer than we thought, and could start to be an issue even at today’s warming of 1.1-1.2°C.
Recent observations also suggest that some tipping elements might already be being destabilised. The grounding lines of some glaciers in West Antarctica are already close to where retreat could become unstoppable, while potential “early warning signals” have been detected in the Amazon, parts of the Greenland ice sheet, and possibly the Atlantic overturning circulation as well.
Despite these advances, there are still some issues around our understanding of climate tipping points. Definitions of what count as a tipping point vary across scientific papers and in the media, leading to threshold-free feedbacks, abruptly-forced events, and arbitrary thresholds sometimes being labelled as tipping points despite lacking tipping dynamics.
Earth system models have improved a lot in the past couple of decades, but many still lack processes and resolution necessary to fully incorporate tipping dynamics. For example, the IPCC stated in AR6 that models are biased towards the Atlantic Meridional Overturning Circulation (AMOC) being overly stable, and models have also tended to underestimate the observed decline in the tropical carbon sink.
And while there are potential signs of destabilisation in some tipping elements, observational data are often too short to be able to robustly assess the trend. The AMOC has for example declined by ~15% over the last 50 years, but this cannot yet be clearly differentiated from natural variability.
Finally, much discussion of climate tipping points still relies on the 2008 assessment, but there have been major advances from improved models, observations, and palaeorecords in our understanding of the different tipping elements since then.
Reassessing Climate Tipping Points
Based on this, we decided a reassessment of climate tipping elements and their potential tipping points would be timely. In the paper published yesterday in Science we reviewed over 220 papers published since 2008 covering the many tipping elements proposed in the literature, searching for evidence of tipping dynamics and extracting estimates for the thresholds, timescales, and impacts of potential tipping points. From this we categorised the proposed tipping elements according to their dynamics, synthesised estimates for their thresholds, timescales, and impacts, and assigned confidence levels in our assessments.
We also updated the climate tipping points definition set out in 2008 so that each candidate tipping element could be clearly categorised by its dynamics. In particular, we focus on self-sustaining change beyond a threshold as they key characteristic of tipping dynamics:
A good metaphor for a tipping point is a seesaw – if a ball was being pushed up a seesaw but the pushing stopped before it reached the pivot the ball would roll back to its original place, but after passing the pivot it would carry on rolling until it reaches a new state even if the pushing stopped:
Tipping points are often abrupt and/or irreversible – both of which have been used by the IPCC as proxies for potential tipping dynamics (e.g. AR6 WG1 Table 4.10) – but our definition here doesn’t specify abruptness or irreversibility as necessary conditions. This is because in subsystems with slow timescales (such as ice sheets), self-perpetuating change (such as ice sheet collapse) can happen over much longer timescales than the duration of the driver, and in some special cases self-perpetuating change can also occur across what are defined mathematically as non-catastrophic thresholds.
We also now differentiate between global ‘core’ and regional ‘impact’ tipping elements. We define global core elements as being ‘sub-continental’ in size (more than a million square kilometres in our definition) and substantially affecting the Earth system’s current state (e.g. having ice sheets at both poles, or it tipping amplifying global warming by more than 0.1°C). Regional impact elements may be more spatially distributed, but localised tipping still occurs nearly simultaneously across a sub-continental area and either affect the well-being of 100s of millions of people or are greatly valued in themselves. We also categorised some candidate elements as threshold-free feedbacks (i.e. with no clear threshold tipping occurs beyond), uncertain, or unlikely.
Meet the Tipping Elements
We categorised 16 candidates as climate tipping elements (compared to nine in 2008), nine of which as global ‘core’ tipping elements and seven as regional ‘impact’ tipping elements.
There are several new entries on the list, and some removals as well. East Antarctic subglacial basins have been separated out from the rest of the land-based East Antarctic ice sheet, as ice in the subglacial basins sits below sea level and so has the same vulnerability to warm water-driven grounding line retreat as the West Antarctic ice sheet. Labrador-Irminger Sea convection and the North Atlantic subpolar gyre has also been separated out from the AMOC, as while they have similar drivers and impacts, more recent modelling indicates that convection in the Labrador and Irminger Seas can collapse separately and earlier than the AMOC.
Removals notably include the Arctic summer sea ice on the basis of recent research showing no clear threshold beyond which sea ice loss becomes self-sustaining (however, there is some evidence for this occurring with Arctic winter sea ice loss at much higher warming levels, and we also separate out Barents Sea ice). We also do not categorise the Indian summer monsoon as a certain enough climate tipping element as it is projected to incrementally strengthen with warming (although aerosols could still cause it to collapse); and the previously-proposed shift to a permanent El Niño state has not been supported by more recent evidence.
A key result from the assessment is our synthesis estimates for the global warming thresholds climate tipping points would likely be triggered by based on current evidence:
This indicates that at current levels of global warming (around 1.1-1.2°C), five climate tipping points are already possible in our assessment (‘possible’ here meaning above the minimum but below the central threshold estimate). These include Greenland and West Antarctic ice sheet collapse, tropical coral reef die-off, widespread abrupt permafrost thaw, and Labrador-Irminger Sea convection collapse. These are not yet likely, but we cannot rule out that they could still be tipped even if warming stabilised at current levels.
At 1.5°C – the more ambitious of the Paris Agreement aims, and the minimum warming level possible under the most ambitious emission reduction scenarios – four of these possible tipping points become likely (‘likely’ being above the central estimate) in our assessment. Another five tipping points become possible by 1.5°C, including AMOC collapse, Barents Sea ice collapse, mountain glaciers loss, boreal forest southern dieback, and boreal forest northern expansion. Labrador-Irminger Sea convection collapse and Barents Sea ice loss also move from possible to likely further moving through the Paris range of 1.5-<2°C.
While the promises made at the Glasgow COP26 climate conference could potentially limit warming to just within the Paris range, current policies are projected to be putting the world on a trajectory to around 2.6°C (with an uncertainty range of 1.9-3.7°C). This would make seven tipping points likely (adding mountain glaciers) and six possible (adding East Antarctic subglacial basins collapse, Amazon rainforest collapse, and Sahel greening). And if warming ended up at the upper end of the current policy uncertainty range – as a result of climate sensitivity or carbon cycle feedbacks being on the higher side of current best estimates – then ten tipping points would be likely (adding East Antarctic subglacial basins collapse, Amazon rainforest collapse, and Sahel greening) and four possible (adding permafrost collapse).
When comparing our threshold estimates with the IPCC’s future emission scenarios, in our assessment most but not all climate tipping points are avoidable on the lowest trajectories (SSP1-1.9 & SSP1-2.6), and some climate tipping points will likely be passed in the coming couple of decades:
We also estimate the likely timescales and impacts of each global core and regional impact tipping element. Of the tipping points that are already possible in our assessment, Greenland and West Antarctic ice sheet collapse would occur over hundreds to thousands of years, and would additionally take a few decades above the tipping threshold for tipping to be triggered, allowing some potential ‘overshoot’ time. Conversely, coral reef die-off and Labrador-Irminger Sea convection collapse could occur over only a decade or so:
The tipping points that become likely within the Paris range of 1.5-<2°C would lock in major impacts. If warming stabilised above their thresholds for more than a few decades, the Greenland and West Antarctic ice sheets would gradually raise sea levels by more than 10 metres over hundreds to thousands of years, reshaping the Earth’s coasts and displacing major cities. Widespread abrupt permafrost thaw would amplify permafrost emissions by ~50% and disrupt northern landscapes, including infrastructure and Arctic Indigenous peoples’ way of life. Coral reef die-off would decimate ecosystems and damage the livelihoods of hundreds of millions of people who rely on them for fisheries and coastal protection. And convection collapse in the Labrador & Irminger Seas would cause major cooling around the North Atlantic, disrupt weather patterns in Europe and North America, and shift tropical monsoons.
In the 2-3°C likely with current policies, committed mountain glacier loss also becomes likely in our assessment, which would threaten the water supplies of millions around the world. Sahel greening also becomes likely, which might seem positive but would still disrupt the livelihoods of the many people living there and neighbouring regions might receive less rainfall as well. And while Amazon rainforest dieback isn’t likely until ~3.5°C in our analysis, this is for climate-drivers only and so could occur much sooner as a result of deforestation. Amazon dieback would add ~0.1°C to global warming and cause far more warming regionally, disrupt rainfall across South America, and irreparably damage one of Earth’s greatest biodiversity hotspots.
Beyond 3°C, AMOC collapse would cause major cooling around the North Atlantic region, cause some temporary global cooling, and disrupt weather patterns in Europe and North America and monsoons around the world. Boreal forest tipping points include both southern dieback and northern expansion, which would disrupt boreal and tundra biodiversity and human communities and have complex and somewhat counterbalancing climate impacts via both carbon and albedo. Permafrost collapse in carbon-rich deposits like in the Yedoma regions could add around 0.2-0.4°C to global temperatures. It’s worth noting though that this amplifies temperatures by 5-10% over several decades. That makes permafrost a strong but not a runaway feedback on global warming, which we should nonetheless try to limit as much as possible.
At very high warming levels, Arctic winter sea ice and East Antarctic ice sheet collapse would cause substantial warming feedbacks as well as massive (50+ metre) sea level rise from the latter, but are thankfully very unlikely from even on the current highest possible warming trajectory.
Our estimates suggest then that Earth may have left a ‘safe’ climate state beyond 1°C, as five climate tipping points become possible beyond this point in our assessment. At 1.5°C, which average warming will pass sometime in the 2030s, four of these tipping points become likely and five more become possible in our assessment, with likelihoods of these increasing through the Paris range of 1.5-<2°C. Current policies leading to ~2.6°C warming are clearly unsafe because they would likely trigger multiple climate tipping points and make many more possible.
These results provide strong scientific support for rapid emission cuts in line with the Paris Agreement target of 1.5°C, which would minimise the likelihood of triggering climate tipping points. However, several tipping points are still possible or even likely at this level, some of which could lock in substantial sea level rise and ecological disruption for hundreds of millions of people, likely requiring substantial adaptation programmes.
Even if we hit some climate tipping points in the Paris Agreement range though, our estimates of the global impacts of these tipping points indicate that most at this level don’t substantially amplify global warming in the shorter term, and so won’t trigger a ‘runaway’ climate change scenario. This means that further warming and the likelihood of passing further tipping points will still be determined by how rapidly we cut emissions now, and climate tipping points don’t make it a ‘game over’ situation.
Our study is a first attempt at an updated systematic assessment of climate tipping points, but many of our estimates still have high uncertainties and varying confidence levels. It should therefore be followed by wider community projects (such as expert elicitations, CMIP6 model scanning, and the new “TIPMIP” Tipping Points Model Inter-comparison Project) to improve our understanding and monitoring of climate tipping points.
Copies of the paper, including a free referral link to the final published version, are available here.
Thanks to all my co-authors (including Arie Staal of Utrecht University, Jesse Abrams and Tim Lenton of the University of Exeter’s Global Systems Institute, Ingo Fetzer and Sarah Cornell of Stockholm Resilience Centre, and Ricarda Winkelmann, Sina Loriani, Boris Sakschewski, and Johan Rockström of the Potsdam Institute), and for funding and support from the European research Council-funded Earth Resilience in the Anthropocene project and the Earth Commission.
This post was written by Dr. David Armstrong McKay, a GSI Visiting Fellow at University of Exeter and Associated Researcher at Stockholm Resilience Centre, and currently working as a Research Consultant with the Earth Commission. A draft was proof-read by Dr. Rachael Avery.
Post update log: 4/11/22 updated link to paper copies & corrected AR6 RFC5 risk level statements.