Claim: A summer ice-free Arctic (called by some the “Blue Ocean Event”) will happen within the next few years and will cause an abrupt worsening of climate change and possible runaway feedbacks.
Reality: A summer ice-free Arctic will probably happen within the next few decades, but the exact year will depend on unpredictable natural variability. A summer ice-free Arctic would worsen regional warming and impacts, but would not cause a big or sudden increase in global temperatures.
This is the first post in a new climatetippingpoints.info series fact-checking claims that various climate tipping points have been crossed, and that sudden catastrophic warming is now inevitable. See the Introduction post for an overview.
Some commentators (e.g. 1,2,3,4,5,6) have become worried that the first summer instance of an ice-free Arctic – what some have termed a “Blue Ocean Event” – will happen very soon, and will lead to a rapid increase in global warming and extreme weather through positive feedbacks that could even end in ‘runaway’ warming. In this first post of climatetippingpoint.info‘s new Fact-Check series, we investigate whether these claims add up.
First let’s take a look at what the Arctic summer sea ice (ASSI) trends are in recent time. In our backgrounder on Arctic feedbacks, we explored how ASSI extent had declined in the last couple of decades, and how this is linked to the ice-albedo feedback. Here’s the data on ASSI extent until 2018:
It’s clear that ASSI is still in decline relative to pre-2007 norms, not only in extent but also in thickness and presence of multi-year ice (the most stable kind), and that this decline is almost entirely caused by human-driven warming. It’s also true that this is happening quicker than the IPCC (Intergovernmental Panel on Climate Change) originally projected:
People often share this figure to illustrate the glaring mismatch between IPCC model projections and actual observations. But that graph is 10 years old now, and since then models in the more recent IPCC report (AR5) has explored sea ice in some more detail. Although the models are still imperfect, some of them are now in much better agreement with the observed rate of decline (with both models and observations agreeing on ~4 to 5M sq. km for around 2020):
The new round of models (CMIP6) for the upcoming IPCC report (AR6) have reinforced this, with most models becoming effectively summer sea ice-free at around 2oC of warming and the chance of occasional ice-free summers growing above ~1.5oC (see figure below). Based on this, the first ASSI-free year could occur as early as the 2030s and is likely in the 2040s under any scenario, and will become permanent after around 2050 in high emission scenarios:
The decline in sea ice extent in turn has an impact on Arctic albedo, which as explained on climatetippingpoints.info before leads to an increase in regional warming and sea ice melt through an amplifying positive feedback process:
But what about the claim that ASSI will be gone by as soon as 2022? The source of this claim are the trends that can be fitted mathematically to the ASSI volume data, in what some have claimed is “the most important graph in the history of humanity“:
Firstly, it’s worth noting first that this graph is using sea ice volume rather than sea ice area, on the basis that total volume is more linked to climate forcing than extent alone and is declining faster than area, and that thicker ice is more stable. But for feedbacks on climate change we’re mostly interested in extent, as it’s the visible area that reflects back solar energy, and an effectively ice-free Arctic is defined by an extent of less than 1 million square kilometres of sea ice.
The graph above might look convincing enough at first. But one problem with fitting trends like this – and especially non-linear trends like the curved ones here – is that one or two especially low outlying points can drag the trend significantly down, but may well turn out to be anomalously bad years due to inter-year natural variability.
In this case, 2007 and 2012 were especially bad years, where natural variability amplified rather than countered the long-term warming trend. This tendency is shown by how re-fitting these same trends on this up-to-date data delayed the date for an ice-free Arctic by ~5 years when compared with a previous version of this very same graph. And unlike the first ASSI extent graph shown earlier, this graph only uses data from 1997 for it’s straight line rather than the whole dataset, making the straight line fit far less well than the other lines using the whole dataset.
But more importantly, fitted trends assume a predictable underlying pattern that is captured by the simple line you fit. But without understanding the actual underlying physical drivers of this pattern, they can give false confidence in projecting these trends in to the future.
The extent of Arctic sea ice is affected by many factors beyond surface air temperature that make it more difficult to predict, including weather events, cloud patterns, and fluctuating ocean currents, none of which is captured by fitting a simple line. And although the ice-albedo effect is strong, its impact will decline as total ice extent declines. Once sea ice extent is low, a further drop in extent will have less of an impact on regional warming compared to when the extent was higher. As a result, we would expect the decline to begin to slow down once lower extents are reached, and so for the summer ice-free Arctic date to be delayed.
This difficulty in predicting ASSI extent from year to year has led to many predictions of its total collapse that have not come to pass. Back in 2007 some thought that 2013 would first see an ice-free Arctic, a prediction which seemed to be supported by the record low extent in 2012. When 2013 came along and the ice was still above the 2012 record, some predicted 2015 or 2016 instead.
The well-known Arctic researcher Peter Wadhams stated in 2016 in his book Farewell to Ice that, having earlier expected to be gone by the mid-2010s, he now expected an ice-free summer Arctic by 2017 or 2018 – which again hasn’t happened. This repeated postponement has then been used by contrarians to wrongly claim that the wider climate science is flawed.
Now, we are seeing 2022 or later being predicted as the next big year to be worried about for sea ice. But the observed trend is overlain by so much natural variability that predicting a specific year becomes a guessing game that may both prompt excessive alarm and undermine confidence in climate science.
Based on what we’ve seen from both observations and the most recent IPCC report, an ice-free Arctic summer could happen within the next few decades and is likely by around mid-century under medium to high emissions scenarios, but the specific year will depend greatly on natural variability and how much more we emit in future.
What of the second part of the claim we started with, that an ice-free Arctic – the so-called ‘blue ocean event’ mentioned at the beginning of this post – would have catastrophic regional and global climate impacts that risk runaway warming? Here the story has also been overhyped.
Proponents of the ‘blue ocean event’ have proposed four elements for Arctic sea ice loss triggering a jump in temperatures, which has been estimated at 0.5–1.0oC, 0.6-0.9oC, or even 1.6oC above IPCC projections. The main proposed driver of this warming is the ice-albedo effect, meaning that an ice-free Arctic would absorb more heat (as the Arctic Ocean will be mostly blue, as referred to in the “Blue Ocean” name). This in itself true, but the magnitude of global warming because of this has been exaggerated by some.
It has been claimed that Arctic sea ice loss is already significantly boosting global radiative forcing (RF; the actual energy imbalance caused by human emissions since AD1750 that drives global warming, measured in Watts per square metre of the Earth’s surface). Many claims mix and match different units and scales, so for accuracy and understanding below we’ve found the appropriate RF value and then estimate 21st century temperature impact using a transient climate response of +~1.9oC per doubling of CO₂ [0.51oC per W/m²] (relevant over a few decades, versus a climate sensitivity of +~3oC per 2xCO₂ for a timescale of centuries). For example, the current total change in RF due to human activity is +2.3 W/m², which translates to a near-term warming of ~1.2oC.
It’s been suggested that sea ice loss in the Arctic is already having the same effect as one quarter of all the CO₂ released since 1979 as a result of the ice albedo effect. The source of this number calculated that the increase in RF from melting sea ice is equivalent to 25% of the warming caused by CO₂ alone from 1979 to 2011 or 2016 (based on an RF increase of ~0.2 W/m² from ASSI loss, which would lead to ~0.11oC of warming, and 0.8 W/m² from CO₂ – see below for more on this).
Other model-based estimates and emergent constraints for current forcing from ASSI loss tend to be a bit lower at ~0.1 W/m² from 1979-2007 (equal to +~0.05oC) or 0.11-0.15 W/m² per degree of warming, although the large drops in ASSI since 2007 likely accounts for some of the difference by 2016. There is also uncertainty over how much cloud feedbacks might interact with ASSI loss, with some arguing that negative cloud feedbacks elsewhere will counteract Arctic changes, while others predicting a probably negligible or maybe even a positive feedback from clouds.
For a totally ice-free Arctic summer it has been claimed that global temperatures would quickly go up by between 0.5oC and 1.0oC. This is linked to comments made by Peter Wadhams in his book Farewell to Ice, where he says that a summer ice-free Arctic would increase radiative forcing by 50% of total historic carbon emissions. Given that the RF from CO₂ emissions so far is ~+1.6 W/m², this would imply an RF increase of ~+0.8 W/m² (about +0.4oC in the near-term & ~0.6oC by 2100).
However, estimates in peer-reviewed scientific papers of the global RF increase for a consistently summer ice-free Arctic (which becomes likely by the time global warming reaches around +2oC) are much more modest at around +0.3 W/m², which would lead to near-term warming of about 0.15oC (and ~0.25oC by 2100).
The initial estimate for summer ice-free Arctic above is actually closer to the estimate for a year-round ice-free Arctic* (+RF ~0.6 to 0.7 W/m², or about +0.35-0.55oC, assuming negligible cloud feedbacks), which isn’t expected until at least the 22nd Century. There’s also no evidence that the increase in radiative forcing from albedo would come as an abrupt jump rather than scaling up gradually proportional with sea ice loss, with half this warming having already occurred.
[*A potential source of this mismatch is that the original source Wadhams cites for +25% of CO₂ warming so far – which is then doubled for future melt – used a lower figure for historic CO₂ RF of 0.8W/m² for its RF comparison. Half of this is 0.4W/m² & 0.2-0.3oC for ASSI loss, which fits with other literature estimates but is smaller than what is often publicly suggested. Some estimates appear to instead mistakenly use the higher CO₂ RF estimate of +1.6 W/m² from the IPCC (or even all of the +2.3 W/m² of RF so far) to calculate the 50% relative increase, but this doesn’t match the numbers used in original source for this comparison. To keep things consistent, we directly use the actual RF estimates (~0.2W/m² for ASSI loss so far & 0.3-0.4W/m² for total ASSI loss) from the original source that Wadhams and others cite.]
This means a summer ice-free Arctic – which we would expect within the next few decades unless aggressive emissions reductions are pursued in the next decade or so – will add ~0.15-0.2oC (of which around half has already happened) to global warming earlier than expected in earlier reports, and although small this makes it that much more difficult to limit warming beyond 2oC.
This additional warming will be mostly concentrated in the Arctic and most keenly felt regionally rather than globally, contributing to the already observed “Arctic amplification” of global warming. The Arctic has already warmed up over twice as much as the global average, which as well as helping to melt sea ice is also affecting ice on Greenland and the permafrost and Boreal forests surrounding the Arctic Ocean:
This amplified Arctic warming could in turn affect the Northern Polar Jet Stream, a “river” of fast-moving wind high up in the troposphere (the bottom layer of Earth’s atmosphere) which dictates much of the Northern Hemisphere’s broader weather.
A strong, stable jet stream relies on a large temperature difference between the equator and the pole. Now that the Arctic is warming up faster than the lower latitudes, that temperature difference is reducing and the jet stream is getting weaker. This may be making it easier for large air masses to ‘push’ the jet stream out of place, making it meander and sometimes split.
This means warm air masses can end up farther north than usual and vice versa- recent cold and warm weather extremes in North America and Europe might be linked to the Jet Stream weakening. Reduced ASSI might also be affecting the North Atlantic Oscillation, a multi-year air pressure pattern affecting weather in Europe and North America.
Another way that warming could lead to weather extremes in the mid-latitude Northern Hemisphere is the “polar vortex”. The stratospheric polar vortex is a region of air about 50 km high contained by another, polar, jet stream, known as the polar night jet (so-called as it only exists during the dark polar winter).
When the stratospheric polar vortex weakens it can trigger a sudden stratospheric warming and allow cold Arctic air to move south. Although these happen naturally over the last 37 years they have become more common, and some have suggested low sea ice in key areas might help trigger them.
If these links are significant, the fear is that these weather extremes could get worse in future and affect human health and crop yields in the Northern Hemisphere. But some proponents of the ‘blue ocean event’ scenario have recently argued that these weather extremes would have devastating impacts, with abrupt Arctic warming triggering the weakening or even loss of the Jet Stream, total crop failure, and societal collapse within only a few years.
However, the science has not yet been definitively settled on the extent to which Arctic sea ice and northern hemisphere weather extremes might be linked, and it’s too early to tell how strong the trend in Arctic-driven weather extremes currently is or will become. Claims of a radical Jet Stream weakening also rely on huge temperature jumps when summer sea ice is lost, which as discussed above is unlikely to be so dramatic.
The second proposed element of the ‘blue ocean event’ scenario is the ‘latent heat’ feedback. When heat (thermal energy) is applied to ice, the temperature of the ice increases. When the ice starts melting as energy continues to be applied, the ice’s temperature remains the same until it has converted to water. This energy that is absorbed to change the structure of the material rather than its temperature is called latent heat. Conversely, the freezing of water to ice releases heat.
This means that the net melting of Arctic sea ice slightly reduces global warming (as measured in surface air temperature), as the melting ice absorbs a bit of the ocean or atmosphere’s heat instead. Conversely, once sea ice has disappeared this would mean that less heat is used by melting ice, so more heat might remain in the ocean or atmosphere instead.
The proportion of anthropogenic heat that has gone into Arctic sea ice so far is only around a third of what has gone into the atmosphere (around 0.3oC of equivalent atmospheric warming) though, and tiny versus what’s gone into the ocean:
Some discussions of the latent heat feedback implicitly focus on the summer sea ice (sometimes interchanging it for all sea ice) and ignore that every winter the sea ice re-forms again, continuing the cycle of latent heat absorption and release even if summer sea ice reaches zero. This ice melt heat sink would only be entirely lost if all sea ice disappeared forever, but all of the trends we have been looking at are specifically for summer sea ice – no model or observations supports the total loss of winter sea ice this century, even after losing summer sea ice at some point.
This means that although there will be an increase* in the rate of heat accumulation in the ocean and atmosphere due to sea ice decline, it will be very small compared to the overall global heat balance [*the exact rate is hard to predict, as heat budget measurements have mostly focused on the dominant ocean].
It’s also important to note that this latent heat is not hidden away in the sea ice and so will not suddenly be released and cause and abrupt atmospheric temperature rise once the ice melts – the heat was absorbed by the melting ice to change its state and is effectively permanently stored in the liquid water.
On top of that, increasing evaporation from the warming oceans (including from the newly opened ice-free Arctic) also uses far more latent heat worldwide than just sea ice, further complicating how much latent heat fluxes will change both in the Arctic and globally in future.
Overall, with Arctic sea ice latent heat we have a globally-small heat sink getting smaller over time as the volume of sea ice melting each summer declines, relative to a far larger ocean heat sink and increases in the global latent heat flux due to evaporation from warmer waters. We do not have a sudden release of ‘hidden’ heat back into the atmosphere, as implied by some presentations of a catastrophic ‘blue ocean event’ scenario.
The final elements for the hypothesised catastrophic impacts of a ‘blue ocean event’ are the release of methane from both methane hydrates under the seabed and from permafrost on land – known by some as the “Arctic Methane Bomb”.
The issue of potential tipping points involving Arctic methane is a complex one which we explore in greater detail in this separate in-depth Fact-Check. But as a brief summary, the evidence for imminent and dramatic release of methane from the Arctic is very limited, and a longer-term, non-catastrophic leakage of methane is far more likely.
Further evidence for non-catastrophic impacts of an ice-free Arctic come from several thousand years ago in the early Holocene (shortly after the end of the last Ice Age glacial), a time of a relatively stable mild climate with no evidence of abrupt warming despite palaeo evidence of a period of relative regional warmth and ice-free summers in the Arctic Ocean.
Finally, an Arctic sea ice tipping point doesn’t mean that the Arctic becoming ice-free will cause a sudden and dramatic global warming event. A common misconception with tipping points is that they’re always abrupt, dramatic events, when in reality their most important feature is their self-perpetuation even if they then only happen gradually.
In this case, the tipping point would be when losing summer sea ice would become inevitable through elevated regional warming even if human emissions ceased. The latest IPCC reports suggest that for a low emissions scenario there’s still a chance that some summer sea ice could survive, although continued emissions make this unlikely. The latest report (AR6) also makes it clear that ASSI doesn’t appear to have a clear tipping threshold, unlike Arctic winter sea ice.
Together, observations, theory, modelling, and palaeorecords suggest that a summer ice-free Arctic is not yet imminent, and will not trigger a globally catastrophic “Blue Ocean Event” when it does.
The first ice-free summer in the Arctic will happen sooner than originally thought and likely sometime in the next few decades, but is hard to predict exactly when because of large natural variability on top of the human-driven warming trend.
And while losing the summer sea ice will drive significant regional warming and and may increase mid-latitude weather extremes, the global warming boost will be modest (estimated at +~0.15-0.2oC, of which around half has already happened) and won’t happen in one sudden jump.
As with many climate tipping points, an ice-free Arctic won’t be the trigger for an abrupt, dramatic event. It is rather an unwelcome extra amplification of climate change that will have an impact over many decades.
Climate tipping points will likely make climate change worse than expected and more difficult to counter, but claims of a near-inevitable chain reaction to runaway warming as a result of an ice-free Arctic are wide of the mark.
This post was written by Dr. David A. McKay, currently a Postdoctoral Researcher at Stockholm Resilience Centre (Stockholm University), where he is part of the Earth Resilience in the Anthropocene Project (funded by the European Research Council) and is researching non-linear climate-biosphere feedbacks. This post was written in his spare time with no funding support for this site, and was proofread and edited by Dr. Rachael Avery.
Updated: 26/4/19 with a link to a new paper which supports the summer/winter ice-free Arctic RF estimates and a recent paper with probabilities of an ice-free Summer relative to the level of global warming. 30/7/19 with updated references for radiative forcing from ASSI loss so far, additional support for an RF of ~0.7 W/m2 for total ASSI loss (but with large uncertainty from cloud feedbacks), and some figure rounding. 15/11/19 with a new emergent constraints study that supports current ASSI-free RF estimates and mid-century timeline, plus new subheading links to allow direct linking to sections; 12/6/20 clarified TCS value used here and a source provided; 6/7/20 with new CMIP6 sea ice projections; 24/8/20 with new note exploring the potential source of the mismatch between Prof. Wadham’s and other papers’ ASSI-loss warming estimates; 30/10/20 with clarification on long-term ASSI warming & estimate mismatches; 17/11/20 with new article on ASSI-jet stream link; 23/11/21 minor reformatting & updated ASSI decline chart.