Gone for good: Arctic Ocean ice free all year by the 2040s?

A few days ago I used a combination of Arctic sea ice volume data from the University of Washington’s PIOMAS model and NSIDC sea ice extent numbers to project that the Arctic Ocean would be effectively ice-free in late summer within ten years. The key to that exercise was the rate at which the volume of sea ice has been declining — 350 km3 per year over the last 30 years for the full dataset, 410 km3 per year over the last 20, and 740 km3 over the last decade at summer minimum. The rate of volume decline has obviously been increasing. Using those numbers to project ice extent in the future is one thing, but they also tell us something interesting about the overall Arctic heat budget — and we can use that to make a rough guess about when the Arctic will become ice-free year round. The answer is surprising…

The Arctic heat budget is the balance between the heat being shipped to the top of the planet by the summer sun, the atmosphere and ocean currents, and the heat lost over a sunless winter. The sea ice volume trends tell us that the polar heat budget over the last three decades has been dominated by excess incoming heat — enough to melt hundreds of cubic kilometres of ice every year. We don’t need to worry about the details at this point — we don’t need to look at all the items in the budget, positive and negative, but like a good CEO we can focus on the bottom line, and that’s what the volume decline represents.

Another aspect of the volume data tells us something else about the budget: the amount of heat lost over winter. If we take the full 30 year PIOMAS data, the average maximum ice volume is 28,600 km3, and the average minimum is 13,400 km3. In other words, in what used to be an average winter the Arctic lost enough heat to grow 15,200 km3 of new ice. Let’s be generous, and assume that in current circumstances (starting from a low summer minimum) winter ice growth is potentially around 20,000 km3.

Once the perennial ice has gone, the Arctic heat budget will not simply stop showing a surplus. Instead of melting ice, energy will go into heating the ocean and atmosphere. More heat will have to be lost each autumn before ice can start to reform, and the arrival of the first ice of winter will be delayed. This effect can already be seen in the data. Take a look at the area anomalies at Cryosphere Today — they are already greatest after the late summer minimum, reflecting the delayed formation of new ice (more on that here). The greater the delay, the smaller the volume of ice that will regrow over what’s left of winter. At some point in the future enough heat will have accumulated in the ocean to prevent winter ice formation, and the Arctic will have completed its transition to an ice-free state.

…it may be already too late to do anything to prevent the Arctic sea ice disappearing completely

How soon could that happen? The ice volume data provides a simple way to arrive at an estimate: divide the normal winter heat loss by the annual heat surplus. At 350 km3 of ice lost per year (the 30 year volume trend), it will take 57 years. If the Arctic is seasonally ice-free by 2020, then it will be ice-free year round by 2080. However the recent volume loss has been running at 740 km3 per year, and if that rate continues it might only take 27 years for the winter ice to disappear. If the summer ice is gone by 2016, as might be possible at that rate of decline, then the Arctic could be ice-free in winter as early as 2043. This is not only much faster than most researchers have been willing to countenance, but also falls within the 30 year climate commitment — the inevitable warming caused by current greenhouse gas levels, as the climate system comes back into thermal equilibrium. If my sums are right (please, prove me wrong!) then it may be already too late to do anything to prevent the Arctic sea ice disappearing completely.

This speculation is clearly a considerable over-simplification — not a back-of-the-envelope calculation so much as a topologically infeasible back-of-the-back-of-an-envelope scribble — but the basic logic seems sound to me. The mechanisms that are creating the Arctic’s energy surplus may not operate the same way in future — negative feedbacks may come into play (increased cloudiness, freshening of the surface layers of the ocean), but so will positive feedbacks such as the “albedo flip” (replacing ice with dark ocean will increase heat absorption from solar radiation). The balance between those factors will determine how long it will actually take to move to a permanently open ocean, but there are signs we are heading that way at some speed.

We have a good analogue in the geological record for what a warm Arctic might be like — the Paleocene Eocene Thermal Maximum 58.5 million years ago. A sudden warming episode, probably associated with the release of a large amount of methane from oceanic methane hydrate deposits, saw Arctic sea surface temperatures spike up to 24ºC, and there’s evidence of sub-tropical vegetation (specifically palms) growing on neighbouring land masses. Not good news for the Greenland ice sheet, or the huge amounts of methane hydrates trapped under an already melting permafrost cap on the seabed off Siberia…

When I wrote Hot Topic (over NZ summer 2006/7), I suggested that the Arctic might be ice free in summer during my lifetime and thought I was going out on a limb by saying that. The record retreat of the sea ice in 2007 was a worrying confirmation of my gut feeling. Now it seems possible that if I can hang on until I’m 89, I might see the Arctic lose all its sea ice. The complete and utter transformation of a planet in my lifetime. And still we do too little, too late…

[The Shins]

65 thoughts on “Gone for good: Arctic Ocean ice free all year by the 2040s?”

      1. Good find, AD. I dimly remember reading that when it came out. It provides justification for my view that once the ocean is seasonally ice-free, there’s nothing to stop it from moving to a fully ice-free state.

        As the ice-covered fraction of the year decreases in a warming climate, the stabilizing ice thermody- namic effects become less pronounced in the full annual cycle, and a bifurcation occurs when ice covers the Arctic Ocean during a sufficiently small fraction of the year to allow the ice–albedo feedback to dominate. Hence, when the Arctic warms beyond this point, the system supports only an ice-free solution.

        1. Indeed, as early as 1991, numerical simulations showed that “increasing the poleward flux by 30 W m−2 is enough to produce an arctic ocean which stores enough heat in summer to prevent any ice forming in winter.”

          The authors go on to say “The model predicts that today’s climate (1991) can support either a perennial ice pack or an ice free Arctic. However, a winter ice pack which melts completely each summer is not possible under any climate.” The equilibrium end-state will not be a seasonally sea ice-free Arctic, but a perennially sea ice free Arctic.

          1. Wow, I shall eat my hat then! Seemed so counter intuitive. Shows again that we ‘aint seen nothing yet’ when it comes the true extent of where all this is leading….

  1. Yes, I’m afraid that sounds logical, but isn’t there an effect to the thermohaline circulation too? If this weakens for the loss of ice, this could slow the transport of heat on the arctic.

  2. For what its worth, even the Baltic Sea, relatively warm in Summer with surface temps around 20Deg in places, freezes most winters in parts and in colder winters severely: http://www.thelocal.se/25364/20100305/

    I was once stuck in Denmark at about 55 Deg north for a week in the late 80ies as all ferry services were canceled due to thick ice in minus 20 Deg air conditions.

    My gut intuition is that an ice free polar winter is not going to happen. For this the air temperatures above the water must be kept high enough to avoid surface freezing. Hard to imagine during the 6 month polar night.

    I think where the argument here is vulnerable is not in the total heat accumulated or required to be lost per season but its distribution. For winter freezing only a relatively thin layer of water needs to get cold enough to allow floating ice to form. Waters maximum density is at about 4 Deg C. Colder water floats on top of that. So once the surface waters down to a reasonable depth are at or below 4 Deg C, the surface of that can then cool rapidly further, hiding warmer waters below. I think this is the point where the argument of an ice free winter arctic is lost.

    Once the first layer of floating ice is formed and the ocean heat content is insulated from the air above, the air temps will plummet far enough under the conditions to keep the ice frozen.


    1. A graph to see this further is here:

      This is a Jan 2010 graph at 85 North.

      What does this mean: In winter, at least these days, the water is anywhere well colder than 4 Deg. This means further that due to the density inversion below 4 Deg of water colder water will accumulate on the surface. No thermal mixing from the bottom up in winter!

      Hence I believe that freezing temps at the surface are always going to be reached during the arctic winter. One would have to keep the air temperature above that of freezing sea water to avoid that. I can’t see that happening.

      1. What Keith said… 😉

        The other point worth making is that while ice is there to melt (or form) the temperature of the water and air above it is always going to be close to freezing point. When the sea ice goes — even if only for a short period — the ocean can warm considerably. Note the comments and temps recorded by the Norwegian catamaran as it sailed through the Amundsen Gulf last month… That heat has to be lost before freezing can begin.

        And (if that wasn’t enough) warmer SSTs and open seas are leading to a marked increase in storminess in the Arctic Ocean, which has the effect of “stirring up” the surface layers.

        1. Thats correct. I am sure we will have a summer ice free north pole in our live times, possibly within 5 to 10 years.
          But I remain skeptical that it will ever be winter ice free as I maintain that the surface layer of the arctic is less dense than the lower layers despite being colder and thus will insulate itself conductivity from the rest and hence freeze over once the air temp goes below the freezing point. In a 6 month long polar night that will be pretty sure to happen.
          See my comment to Keith below.

    2. The temperature of maximum density for fresh water is 4oC, but for seawater it is below the freezing point because of the effects of the high salt concentration. Thus your density argument is not valid in this context. In other words, seawater, unlike freshwater, increases uniformly in density as temperature decreases.

      1. Hi Keith

        Not sure if I agree. According to my data even sea water at a salinity of 35 has its max density at 4Deg.

        Here is a calculator for sea water density where temp and salinity can be put in:

        Adding to the issue is the fact that the top layers of the arctic are less saline than the bulk below which means its even more buoyant. So I maintain that the cold top layer of the Arctic floats on top of the bulk and insulates it from convection from below.

        Together with the graph of temp/salinity with dept you can plot the density with depth:

        So I would think that explains why oceans (even those with considerable heat content like the Baltic) still freeze in winter.

        If you have any references for this to be otherwise please let me know as I would be keen to investigate further.

          1. You are correct and I am wrong! Thanks!
            I had indeed had overlooked the mg/l scale instead of the g/L scale for PSU… gross mistake of mine…. mea culpa…!

            So with finally the right data put into that calculator from the graph it insists that the max density of the sample graph I quoted is at 1 Deg or 200m depth or there about. That’s due to the lower salinity on top. Salinity however may indeed change with climate change. Depends on the run off from melt water etc and I would not risk a hypothesis where this will head.
            At present though it would seem (if that density calculator is right) that there is still an active density gradient keeping the coldest waters on top aiding freezing there.

            T PSU Density
            1 34.80 1027.9100 <– max at about 200m depth
            0 34.50 1027.7280
            -0.5 34.40 1027.6710
            -1 34.25 1027.5710
            -1.5 34.00 1027.3920
            -1.7 32.80 1026.4160

            But perhaps the calculator app is incorrect?

            So while I'd happily book a sailing trip across 90 North for the very near future in September, id still risk a bet that the pole will freeze during the polar night during my lifetime.
            But then again, if catastrophic feedbacks kick in, who knows…

            1. Hi Thomas: I haven’t checked the calculator yet, but I don’t like the look of the results. The temperature of maximum density and the freezing point coincide at a salinity of about 19, and at higher salinities seawater should freeze before reaching its theoretical maximum density.

              You might be interested in this paper which should enable you to set up a spreadsheet http://www.ocean-sci-discuss.net/6/153/2009/osd-6-153-2009-print.pdf

              The author, Frank Millero is the expert on this topic.

        1. Hi Thomas: Cute little calculator. But I can’t see how you got 4 degrees from it. When I put in a salinity of 33 (33,000 mg/L in the calculator) the density keeps increasing as T falls, even below zero. Are you sure that you had a correct salinity?

          I agree that Arctic water can be less dense because of reduced salinity, but it is still true that the temperature of maximum density is above zero only for very low salinities.

          Oops: Just noticed Peter Ellis’s post above. I agree Peter.

          Information about this is available in any textbook on marine chemistry. Try this link from UCSD http://sam.ucsd.edu/sio210/gifimages/dens.gif

          It nicely illustrates how density depends on T in both seawater and freshwater.

          1. Yes as Peter Ellis also pointed out, I had screwed it up on the first go as the calculator wants mg/L salinity not g/L so I was off by a factor 1000 on salinity…
            On my next shot I was more careful. See table above.
            Seems that the surface drop off of salinity in the Arctic is creating a slight inversion still, with colder less saline waters on top being supported through a lower density. But that hinges entirely on the salinity gradient. So a rise in surface salinity would wipe out the inversion and make it harder for the surface to freeze. With Greenland ice melting there should perhaps be more fresh water coming in rather than less.

            Frank Millero’s paper looks promising however he leaves the important temp range from -2 to 0 out.
            I don’t have the time to convert his work into a table or spreadsheet at the moment.
            But its an interesting topic. Fascinating how so many parameters play subtle roles in all this.

          2. Yep thanks. The salinity gradient in the Arctic makes it a bit more complex and freezing or not will partially depend on that I think as the top layer salinity gradient may create a density inversion isolating the top layer. But as Artful Doger above also quoted, there is quite a consensus that we might indeed end up with a perennially open Arctic ocean.
            Time to sell those skates….. 😉

  3. How far we have come. John Passmore in his 1974 book “Man’s Responsibility for Nature”:

    “We know at least this much, however. Men will need the biosphere. And it is sometimes suggested that our present level of industrial activity is so heating up the atmosphere that large parts of the earth’s surface will – as a result of the melting of polar ice – eventually be rendered uninhabitable. So, it is concluded, we ought at once, for the sake of posterity, to reduce the level of that activity. The Royal Commission on Environmental Pollution concluded that ‘such eventualities are not only remote: they are conjectural’. But this case serves as a sort of touchstone, an extreme example both in its uncertainty and in the disastrousness of the consequences it envisages, were they to eventuate……”

  4. Every prediction about the climate of the future has proved to be wrong and it is much worse than we could have thought. When Al Gore made his film and discussed Greenland he did not know about the record summer temperatures in France and later in Russia.
    16 countries with record high temperatures and none with a record low temperature. This is not good.
    Which ever way you look the major indicators are getting worse.

    1. 18 Bob, 18 countries this year with record high temps.

      I agree with you, the experts seem to be continually underestimating the rate of change in the natural environment. Consider scientific expectations only 4 years ago.

    2. Really. But on record means since when?

      And as we come out of the little ice age / Maunder Minimum it should be no surprise that we get some new maximums. But we still have plenty of maximums that were set a long time ago. We also have some record lows still being set (I presume thats global warming too though).


      Auckland did record a record in Feb 2009.

      “During the 3pm hour the temperature at the official weather station at Whenuapai Airbase reached 32.4 – the hottest ever recorded there.”

      That is climate change all right. When forest is replaced with tarmac the local climate will change.

      “Niwa Climate expert, Jim Salinger, says today’s high was the highest since their records began and equals a high recorded in the Auckland Domain 137 years ago, in 1872.”

      So not really a record then?


      It is all very well getting on your soap box and pointing out one or two record highs recorded in cities of many million people. But it would be preferable if you could show an actual statistical test of record temps. I suspect they are going up (the world is warming), but by not as much as you think.

      1. The record maps show that the ratio of hot record temperatures to record cold temps is running at about 2 to 1 consistently worldwide. In some places it’s running at about 5 to 1.

        Not just that, the number of degrees by which hot records exceed previous highs is very large in many areas.

        1. Link?

          As I state above, it should be expected that we have some record highs for a number of reasons. Especially in the world’s largest cities. But I think it is very easy to get on a blog and post a comment claiming things like “Not just that, the number of degrees by which hot records exceed previous highs is very large in many areas.” without giving references or detail on which areas are experiencing the highs. Are these parts of rural Canada or are they airports in large cities? (I fully expect some record highs in natural environments, but I just want you to put better context on your claims).

          1. As adelady has ably said elsewhere,

            “I really do not want to see what might happen if there were no record cold temperatures showing up anywhere in the world.”

            To paraphrase The Bard, “The Trend’s the thing”.

            The Yooper

          2. So R2 is trying to resurrect the old UHI myth, eh? One or two cities of many millions. You think?

            Here are the populations of the 18 places that set national record high temperatures this year:

            Belarus, Gorky, <20,000 (#1)
            Ukraine, Lukhansk, 445,000
            Cyprus, Lefconica, 1,500
            Finland, Joensuu, 73000
            Qatar, Doha, 998,000
            Russia, Yashkul, 8,000
            Sudan, Dongola, <700,000 (#2)
            Niger, Bilma, 2,500
            Saudi Arabia, Jeddah, 3.4m
            Chad, Faya, 14000
            Kuwait, Kuwait City, 2.38m
            Iraq, Basra, 3.8m
            Pakistan, Mohenju Daro, 0 (#3)
            Myanmar, Myinmu, 19,000
            Ascension Is, Georgetown, 940
            Solomon Is, Honiara, 78,000
            Colombia, Puerto Salgar, 10,000

            Yes R2, I'm sure it's the vast expanses of tarmac and air conditioners in Ascension Island that caused its record high…

            #1 Couldn’t find an actual figure, but the smallest city in Belarus I could find was 20,000, so must be smaller
            #2 700k is the figure for Northern province, of which Dongola is capital, so it is probably much smaller itself
            #3 This is an archaeological site in Pakistan, no actual population

          3. Also giving the lie to R2’s assertions that record temps are only being set at city airports is this map of the Russian heatwave. Note that the area in red is bigger than France and Germany combined.

            You know R2, you could do this research yourself if you were bothered. But no, you’d rather bury your head in the sand and pretend it’s not happening.

            1. And then there is the question of how much forest clearing and tarmac laying has really been happening at Whenuapai.

              My guess is not too much of the former in the last 100 years – dunno about the latter although Wikipedia seems to think not much has changed in the last 40 years (but then Wikepedia are part of the great global scam thanks to Wm Connelly).

          4. More of that UHI idiocy eh? Perhaps you should revisit some of the nonsense peddled by your denialist friends regarding the NIWA sites – featuring urbanised areas like Campbell Island! Bringing this up only underlines your profound ignorance, and the thousands of words you have wasted on this forum don’t disguise it. You have cleaerly learnt nothing. If you would simply shut up there might still be hope.

            1. Besides, the World’s Greatest Scientific Blogger, Anthony Watts, has already conclusively proved that UHI does NOT artificially inflate the temperature record.

              Of course, he thinks he has done the opposite, but that just shows how little he understands science.

            2. If you would simply shut up there might still be hope.

              There’s always the Red Button option

        2. And given that the rise in global temperature is well over 10 times faster than in the end of last glacial, the record high temperatures have the potential of still increasing in amount compared to cold records.

  5. This is the really scary picture:

    Drought over a vast area of land is productive today. And those who believe we can grow our tucker up North beware of the seasons! There is no substitute to the lush and fertile grounds of the tropical and subtropical lands with their long or all year round growing seasons. There is no way we can substitute for these in the Tundra….

    1. Joe, that ecoworld article is riddled with problems. First, it conflates criticism of media coverage with scientific understanding. It misses the fact that the Wilkins Ice Shelf has not actually yet collapsed. It ignores the recent Larson B collapse, which was holding back a number of large glaciers behind it. The glaciers calving into the sea in that region have sped up by very significant amounts (some by threefold) since the collapse of the shelf. It bases its claims on the non-warming of the Antarctic Ocean on a single data point, ignoring the longer term trend, which is clearly positive. Its claim that another significant indicator of polar temperature is the extent of floating sea ice is simply not true, as this is influenced by many factors, notably wind patterns, which have been altered by ozone depletion. But most significantly of all, it says absolutely nothing about land based ice, which is what really matters and which is declining overall (especially the West Antarctic Ice Sheet), based on a variety of measurements taken over a number of years.

      Did you read the link I posted? Can you answer any of the many peer-reviewed claims in it as I believe I’ve answered every (non peer-reviewed, unpublished) claim made in your article?

  6. So, loss of heat equivalent to 20,000 km^3 ice every year over winter divided by the heat equivalent of 350 km^3 excess melting each summer suggests just under 60 years before cooling isn’t sufficient enough to result in appreciable growth of ice.

    I agree that this ignores positive feedback regarding reduced albedo (dark water vs reflective ice). That’s very important. It’s good that you mention the freshening of the surface too, as a negative feedback. But also very important is this negative feedback: as you get closer to ice-free winter, the rate of heat loss will increase greatly. Hotter things lose heat more quickly, and loss of ice (an insulator that slows loss of heat from the Arctic in winter) will both exert effects. I think this will prevent us from seeing a perennially ice free Arctic within our lifetimes. Thank goodness.

    I have no qualifications in this field and the lack of quantitative argument in the above should make you question my conclusion even more than you normally would. I don’t mean to waste your time, but I found this post so provocative that I couldn’t resist replying. “And thanks for the trouble you took from her eyes/ I thought it was there for *good so I never tried.”

  7. Steve I prefer your view to mine. If you’re talking 60 years, how many el Ninos in that time, how many of those are strong el Ninos?

    Why focus on albedo change within the Arctic itself, I’d have thought heat transported in on the major oceans is a much bigger issue. A nice slug of much warmer than usual water from an el Nino heated Pacific into the Arctic at a vulnerable moment – looks like a good way to get rid of a fair volume of ice.

    1. Good point with respect to ocean currents — they will play a strong role in how this all plays out, but I haven’t read anything about how those currents are to change with respect to the Arctic. I think I’ve seen more about how winds will have more opportunity to stimulate mixing up there. Have you seen anything linking El Nino, specifically, to loss of sea ice? I haven’t, and I suspect that there’s too much distance and isolation between the tropical waters for there to be any direct connection. El Nino affects the California current, but that’s a long way away from the Arctic Ocean.

  8. Hi!
    Great calcs thanks. I’m surprised the folk at Cryosphere dont show the ice volume data from Washington on their charts, even tho’ its only updated every couple of weeks. I had done my own humble calculation on the data too, and 10 years or so looked like ice free summer.
    Keep up the great work!

  9. I agree with your post. I noticed the decline in the winter maximum this winter and calculated a loss of winter ice between 2050 and 2100. Your calculation is better though.

    In 2001, I had calculated the loss of Arctic summer ice by 2017 from the loss of 45% of ice volume from 1981 to 1998, taken from naval submarine observations of thickness. I have not seen any papers discuss what caused that ice loss. Unless that process was stopped, however, the loss of the remaining ice was going happen by 2017. A few years later, Maslowski published his forecast of 2020.

    The loss of Arctic ice, and the warming of the Arctic must drive Greenland into a hyper-melt mode, I believe. I would like to see more discussion about that. How fast will the Arctic Ocean warm? What temperatures will it reach? How will heat be transferred to the Greenland ice cap? If the Arctic is ice free year round, will Greenland be melting year round? What volume of water will melt each year, if melting is year round? How fast will the ocean level change on a annual, decade, and century time scale? How will bottom water temperatures change? Will they warm enough to drastically increase methane release? How much methane? If you have any thoughts on this, I would love to read them.

    1. Excellent questions. I really hope we get some answers to them sooner rather than later. I suspect we’re going to be hearing an awful lot more about the Arctic in the coming years since it seems to be the weakest link in a stable climate.

Leave a Reply