Sea ice: throwing in the towel?

No, not me, DC, who says I’m thinking of paying up – 2010 looks more and more like 2006, not 2007. And indeed the latest Jaxa stuff looks like good news for the good guys (that is me, in case you hadn’t realised):


And TS said:

Monckton is a blatant fraud who even lies about his own parliamentary status (or lack of status). Who cares what he has to say on some denialist blog? I would be interested to hear what you think is happening to the Arctic melt season. There was a lot of ghoulish pessimism a month ago, predictions of a huge melt this season and another record set for low had extent. What’s the real story on what’s happening, in your opinion?

Good question! Lets start off by getting rid of some disinformation, if I can, because really that is almost my only advantage. Look at the pic, and possibly at the hires version, and realise that there is almost no predictability of the September ice minimum from the early season activity. In particular, 2007 was totally unremarkable up until the beginning of July. Conversely, trying to predict from trends before July is very nearly worthless. Even now all I’d say is that there is no reason, from that pic, to expect anything very exciting from this year. C is looking for bids at intrade, where (2010 > 2009) is going for 45%. I’d still say it is 50-50 so that isn’t a very interesting offer. I’d take (2010 > 2007) like a shot, mind you.

That, of course, says very little about the various sat pix of actual ice concentration, and maybe the various ideas that the ice was looking a bit iffy. I don’t credit that much, either. Let me try to explain why. It depends on some modelling I did a while ago.

Impact of instantaneous sea ice removal in a coupled general circulation model (GRL, VOL. 34, L14502, 5 PP., 2007 doi:10.1029/2007GL030253 ) is a paper we did a few years ago, indeed one of the last things before I left science. David did most of it. We wanted to see how sensitive an ocean-atmosphere GCM was to initial conditions, specifically those to do with sea ice. So to begin with we did the simplest thing, which was to remove all the Arctic sea ice. We had planned to go on to do more sophisticated things, like take out 10 or 20% of the ice, but as it turned out even total removal of the ice made remarkably little difference, so we never did the more subtle tests. What happens is that the ice just grows back. It was at this point that I lost my belief in the “ah, but once the old ice goes away the new ice will crumble” stuff. Of course, this is a model not reality. And (slightly more subtlely) there is a wrinkle: of course when we remove the ice we’d left the sea “preconditioned” to grow more ice, because we didn’t touch the mixed layer state. So we tried again, this time warming the mixed layer to prevent sea ice reforming. And when we did that… well, still the ice grew back, but more slowly. So, absent evidence to the contrary, I don’t believe the “tipping point” stuff people talk about the Arctic ice.

Note that this says nothing about the long-term trend, which is of course downwards, as the world warms. But it does mean that my “prediction” for the year ahead is the long-term trendline plus a bound for noise (not forgetting to include “anomalous” years like 2007 and 2008 in the trend, because they are of course part of the natural variation, a mistake I made last year).

45 thoughts on “Sea ice: throwing in the towel?”

  1. But hadn’t you heard? “Model” is a dirty word.

    [:-). But it has, subsequently, had a good track record of prediction. Possibly better than it deserves (well, last year far better than it deserves, had I only used it right) -W]


  2. Betting on sea ice is a little like buying life insurance. You say “I bet I die.” The insurance company says “We bet you don’t.” Eventually you do die, but somehow they end up winning the bet.


  3. If you extrapolate the current 2010 trend line in
    with a line parallel to the 1979-2000 average, it will cross the 5 million sq km mark by the end of August. Granted, there is more variability in the August data, but I think you are forgetting “the return [in June] of the Arctic dipole anomaly, an atmospheric pressure pattern that contributed to the record sea ice loss in 2007.” My money is still on < 4.735 m sq km by mid September.


  4. Can you tell us more about how the model works? At what granularity?
    The sea ice this year, as in 2007, has been very strongly affected by weather. I don’t remember seeing the Beaufort Gyre reverse, and reverse again, in a short period (although sea ice experts have apparently been predicting this might happen, as the ice pack thins and breaks up), and this (and the related cessation of the Transpolar Drift) has had a huge effect on the melt rate: ice is melting in place rather than being exported to melt.

    Do the GCM sea ice models work at a granularity at which effects like that are visible?

    [It was just the standard HadCM3 model, which has an ocean grid of 1.25 x 1.25 degress, so about 100 km. The best mechanism for melting ice is to export it; if it is staying in place, I would have thought it less likely to melt, not more -W]


  5. The usual experiment to find hysteresis (if this is what you mean by a ‘tipping point’) would be to slowly scan the forcing over a wide range so that the system makes a transition from “lots of summer sea ice” to “almost no summer sea ice”, and then do it in the opposite direction, and see if the back transition happens at lower forcing than the forward transition.

    The numerical experiment you performed would only see a tipping point if you got lucky with the original parameters (i.e., close enough to the transition). So I’m not sure it is that convincing. But on the other hand, haven’t seen any positive evidence, either.


  6. Looking at the actual maps instead of the totals, it looks like the low values early this year were mostly due to an early thaw in Hudson Bay. While that sucks for the local polar bears, the effect on the rest of the ice pack is probably ameliorated by the canadian arctic islands. However, the guesses in this year’s contest are extremely alarmist, with half a dozen below 4 million km^2.

    Despite advertising on Wattsup, I only got a few entries for more ice than last year.


  7. You can’t really predict the summer minimum until August 15.

    By September 1 it’s pretty clear.
    Except in 2005.


  8. Yes, it’s much less likely to melt in place. That’s why the extent charts have flattened out in the last couple of weeks: because the Gyre has been in reverse and the Drift has been non-existent. There has been some very impressive in-place melting – including some huge areas of ~75% concentration north of 88N, judging by eyeball on MODIS – but almost no export. This may change in the next week or two, as the Gyre appears to have started back up. That was my point, although I see I was uselessly ambiguous.

    Despite consistently betting on the low side, I have always been a bit sceptical about the real alarmists. Geometry demands arctic sea ice: I was very surprised, a few years ago, to find out it was diminishing. And I have a notion, about which I wrote a comment on RC some years ago, that the obvious albedo feedback has a less-obvious counterpart: a dark surface should lose heat much more quickly than a white one, when exposed to the arctic winter night.

    [About that last point: I think yuo’re wrong: what counts is the IR albedo, not the visible albedo. In the IR, most things are black (aprat from aluminium, apparently) -W]


  9. Hmmmm …. definitions vary.

    “… The Earth radiation budget research community is very interested in the diminishing Arctic sea-ice extent, because of its disproportionate surface forcing on the absorbed solar radiation. By “surface forcing” of the sea ice in this study, we mean the surface fluxes with sea ice absent from the system, minus the same fluxes when sea ice is present, without any consideration for feedbacks. Note that this definition of forcing is different than the one in IPCC, in two ways. First, IPCC examines the flux difference at the tropopause, while we only look at the surface. Second, the IPCC forcing is the difference between the factor present and the factor absent, while ours is the other way round….
    … the radiative transfer equations are solved for 118 separate wavelengths for the UV-Vis-NIR part and for 10 bands for the IR part ….
    … If we run our model using a totally overcast sky we can produce a lower bound for the forcing by a meltdown of the Arctic sea ice….
    … In a sensitivity analysis of the sea-ice forcing with respect to sea-ice extent and cloud cover, the interplay of the three quantities became evident. However, we did not identify any strong non-linear behaviour.”

    Click to access acp-10-777-2010.pdf


  10. So are we saying that sea-ice can have a strong effect on the energy balance (i.e. less sea ice means more warming), but not vice versa: more Arctic warming does not feed back strongly to sea-ice (because geometry demands sea-ice, so you need a lot more warming before we become ice-free)? Whereas the albedo warming from sea-ice might have more pronounced effects on other systems, e.g. permafrost, land ice, weather, etc. Something like that?

    [Err, I don’t think so. There is a geometry effect: the ice is somewhat confined within the Arctic basin, which tends to be full. This is in contrast to the Antarctic. But as for the albedo, this is a real effect, in summer (though you have to qualify that somewhat as it is often cloudy) -W]


  11. IR albedo: Ah, yes, I remember now. I guess IR albedo is pretty important for the greenhouse effect too, as all the relevant radiation is IR. Must simplify things.


  12. Hank: very interesting paper.

    No arctic sea ice, if cloud cover is unchanged => 20 W/m2 annualised over the Arctic Ocean – a huge amount, sure to have dramatic effects of many sorts on local climate – which is 0.55 W/m2 annualised globally, which is also fairly immense (total net anthropogenic forcing at the moment is only around 1.5 W/m2).

    No arctic sea ice *in September*, if cloud cover is unchanged => 0.32 W/m2 annualised over the Arctic Ocean, which is 60 times smaller.

    So for really big effects, one needs open ocean in June and July, i.e. a much more rapid early thaw.


  13. “So for really big effects, one needs open ocean in June and July, i.e. a much more rapid early thaw.”
    Isn’t that what we were seeing this year, record open water during the SW intense June/July timeperiod? Because of its simplicity minimum extent recieves a lot of attention, but SW flux weighted time integral would be much more meaningful.


  14. People are sometimes surprised that in the IR range, the emissivity of white snow is pretty close to 1. As it is for many things.


  15. The tipping point will come when the summer melt is so great that the ice does not recover during the winter and so melts faster the following spring.

    Obviously we have not reached that tipping point yet, but the start of the recovery from the summer of 2007 was slow.

    If we had not had a prolonged freeze this spring, we might well have had a record melt, since the thinning ice seems to be melting faster.


  16. However, the guesses in this year’s contest are extremely alarmist, with half a dozen below 4 million km^2.

    At the time it wasn’t so alarmist, but you see what 3 weeks of adverse weather (for melting) can do. There is no doubt in my mind that if this summer (or one of the next) sees – or would have seen, to be precise – the same kind of weather conditions as in 2007 the minimum sea ice extent record would have been shattered. If the Beaufort Gyre hadn’t reversed and a strong high above the Canadian Archipelago would have stayed in place 2010 would by now still have been at least half a million square km in front of 2007. It’s simply a matter of waiting for the next perfect storm. In fact, with every passing year the storm can be less perfect for record breaking events.

    So you can’t blame people for ‘freaking out’ a bit after the May and June record melt, what with all the stories of rotten ice and ominous looking volume data.

    Hopefully weather conditions do reverse again the coming weeks, so we can see if the incredibly fast melt was only due to an early thaw in Hudson Bay. Besides, how many hundreds of thousands square km is Hudson Bay to have such an influence on total extent decline?


  17. “So you can’t blame people for ‘freaking out’ a bit after the May and June record melt, what with all the stories of rotten ice and ominous looking volume data.”

    Yes, I can blame them. Every single year we see that the wiggles within each season are just that. The wiggles can be inherently interesting to some people who follow physically what’s going on in detail, but the wiggles simply don’t extrapolate.

    Wake me up in 5 years, and we’ll see where we’ve come.


  18. You can’t expect people to stay über-rational when extent plummets like it did in May and June and for instance the Northwest Passage opens up in record time. Not after 2007, which most definitely wasn’t a wiggle.

    300mph winds or the Greenland ice sheet suddenly sliding into the Atlantic or the magnetic pole shifting due to the moving of the Greenland ice masses, now that’s alarmism.
    Expecting or fearing a new extent record isn’t what I would label alarmism. It’s perfectly normal. As soon as 2007 weather conditions strike again, there will be a new record minimum sea ice extent.

    Wake you up in 5 years? Why are you sleeping? AGW isn’t just a graph with wiggles that we can comfortably look at from our homes, you know. Besides, if I have understood correctly from what little I have read about systems science, when the wiggles go up and down in an increasingly erratic and extreme fashion, it’s a sign of the system shifting towards a new balance. It’s early to tell, but perhaps not too early.


  19. What about a tipping point for multi-year ice. Thinner ice means easier transport of multi-year ice, means thinner ice etc. In this case around 2000 may have been the start of a tipping point, and 2007 the end. Now we are in a new state of mostly seasonal ice with largely irrelevant multi-year ice. We do have some 2nd and 3rd year ice but that is just due to the amount of time first year ice from Siberia takes to cross the Arctic Basin and exit into the Atlantic.

    And now that the tipping point is over Arctic Ice will continue to reduce, but at a slower rate driven by further temperature rise, but no longer driven by reducing multi-year ice as it is about as low as it can reasonably go.

    Reasonable theory?


  20. Neven,
    I do expect people to stay rational. You invoke 2007 – was 2007 looking particularly weird in May?

    Why am I sleeping? Because I’m tired. And when a signal has a lot of interannual variability, I’m just not going to spend my time poring over the weekly stuff. I’m not saying you shouldn’t; it’s inherently interesting if you want to get into the dynamics of these things, and understand why it wiggles this way and that from week to week. More power to you. But I’m just not excited by it.


  21. As I’ve said before (and I’ll probably say it again, as a Sysphian punishment), you have to look at all the terms in the Arctic heat budget in order to get the full picture of what’s going on ‘oop North. Some of the terms are directly seasonal (summer warming v winter cooling), some episodic (ocean heat transport) and some highly variable (atmospheric) and of course they all interact. The big picture is the reduction in ice volume. Over the last 30 years PIOMAS shows a trend of -3,400 km3/decade, but loss in the last 10 years has been 10,000 km3. This suggests that the heat budget shows an excess of heat in over heat lost equivalent to the melting of 1000 km3 ice per year. Essentially, you can plot that trend down to some minimum amount (a rump of multi-year ice stuck to Greenland and the archipelago, perhaps). Doing that (as Maslowski did earlier this year) you get a seasonally ice-free Arctic ocean in the relatively near future. Even if the loss is “only” 340 km3 per year, it might happen within a decade. William’s model clearly wasn’t getting the overall heat budget right…

    After that happens, and assuming that the heat budget doesn’t markedly alter in favour of ice formation, then the energy excess will go into ocean warming, and even greater heat loss to the atmosphere over the autumn and early winter freeze-up – with direct impacts on NH winter climate. Prediction: lots of snowy winters like 09/10. (I’ve been meaning to do a post on this — will do soon.) You could also speculate that when the total accumulated heat exceeds the winter heat loss (=1.5m of ice over the whole Arctic Ocean?), then the AO will be substantially ice free in winter as well. Someone else can do the sums on that…

    [Whilst in principle I agree that the heat budget matters, I’m doubtful that the amount of energy required to melt 1m of ice is big enough to show up. Of course I haven’t done the numbers so I could be wrong… oh go on then, lets have a go. The latent heat of fusion ( is 333 kJ/kg. 1m of ice is 1000 kg (nearly) so we need 333 MJ/m2. Over a season of, say, 3*40 days that is ~3 MJ/day/m2 or 30 W/m2. Hmm, well that isn’t tiny. Maybe I’m wrong 🙂 -W]


  22. Couldn’t resist doing it myself…

    PIOMAS average annual volume curve shows 15,000 km3 difference between summer min and winter max, so if heat budget remains as per last ten years, then we get Arctic ice-free in winter 15 years after ice-free summer — say 2035. And then the NH warming will really let rip… 😦


  23. I do expect people to stay rational. You invoke 2007 – was 2007 looking particularly weird in May?

    No, it wasn’t, but 2010 was with an average daily melt rate of 68K (topping last year’s 54K, the highest met rate in the previous 5 years). June 2010 had an average daily melt rate of 74K, more than 10K more than 2007.

    So 2010 was looking pretty weird, and with the 2007 July and August melt rates in mind, it wasn’t that far-fetched at the end of June to expect/fear the worst. The weather has reversed dramatically since then. If it hadn’t and had stayed in constant positive mode (Arctic Dipole Anomaly as wells as Arctic Oscillation) like in 2007, we’d be heading for record territory at an alarming rate. Right now, it’s not so sure.

    But I’m just not excited by it.

    I understand and respect that, but for myself, like I wrote in the first blog post on the Arctic Sea Ice blog:

    So why is this important? First of all, an ice-free Arctic ocean has consequences for regional weather and the global climate as a whole.


    Furthermore, despite the ever-increasing evidence that the Earth’s atmosphere is warming, that it is mainly caused by human emissions of greenhouse gases and that this potentially could have serious consequences for the world’s ecosystems, economies and societies at large, there is still a large amount of controversy, mainly conjured up by contrarians, that undermines the public’s perception of the problem of Global Warming and thus delays meaningful and positive action that is needed to mitigate and adapt to the consequences of an atmosphere and oceans being charged with a large amount of extra energy. The – up till now – spectacular story of the Arctic sea ice melt might change this and move the debate forward towards solutions.


  24. W.,

    The GRL paper is pay walled. Does “even total removal of the ice made remarkably little difference” refer to global or sea surface temperatures or to the ability of the ice to recover?


  25. I was surprised the atmos-chem paper didn’t cite yours, actually. Different tribes, different customs?

    The ice volume chart ought to be updated about now. Maybe you could work up bets month by month on where the little dot and line will go next (or whether they’ll have to extend their chart).

    Paul, “even total removal of the ice made remarkably little difference …. the ice just grows back” refers to Arctic seasonal sea ice — remove it from the model; how does that affect the modeled next winter’s recovery.


  26. Thanks for this post, stoat, seems you’ve set something in motion, today’s IJIS melt figure was in excess of 100K km^2 for the first time in many days! 🙂


  27. Dhog, it’s too soon to cheer. NSIDC just has an update out:

    Through much of May and June, high pressure dominated the Beaufort Sea with low pressure over Siberia. Winds associated with this pattern, known as the dipole anomaly, helped speed up ice loss by pushing ice away from the coast and promoting melt.

    However, the dipole anomaly pattern broke down in early July. In the first half of July, cyclones (low pressure systems) generated over northern Eurasia tracked eastward along the Siberian coast and then into the central Arctic Ocean, where they tend to stall. This cyclone pattern is quite common in summer. The low-pressure cells have brought cooler and cloudier conditions over the Arctic Ocean. They have also promoted a cyclonic (anticlockwise) sea ice motion, which acts to spread the existing ice over a larger area. All of these factors likely contributed to the slower rate of ice loss over the past few weeks.

    In the last few days, high pressure has started to build again in the Beaufort Sea, but whether this will continue remains to be seen.


  28. Oops, sorry to spoil all the fun! 🙂

    According to a few weather forecasts some intense highs are projected to form over the Beaufort Sea by the end of the month!


  29. The ice volume chart was updated 7/17 (though the change didn’t appear for me ’til 7/20, I’m guessing they have several servers, as others also saw it only after a delay).


  30. A paper I found interesting.

    However, I expect William to ignore it as it doesn’t cite his paper.

    [I should certainly ignore anything not citing my fine paper.

    However (quick skim, I’m at work, and it is bumps week): this is at the opposite end of the spectrum from our work. What we did was largely experimental, this is very theoretical. Nothing wrong with that, as a pure piece of maths, but as applicaable to the climate system I’m very dubious it is relevant. Finding exciting bifurcations etc is generally a property of 1-D DE’s, and that is what they have, so that is what they find. The multitude of blurring frictional stabilising feedbacks that are present in the real world and our model are absent. However, I do hope to have tim eto read it in more detail -W]


  31. I was surprised by the “I am thinking of paying up” comment. The more sensible question when there is still some chsnce is ‘How much would you accept to settle the bet now?’

    If you want half the stake that would imply a 75% chance of you winning (ignoring risk aversity).

    I think I would hold on to the belief that there is more than a 25% chance of me winning and would not pay half to settle bet. But perhaps DC and you could agree terms?

    [I agree that ideally we’d establish a market in the bets. Indeed perhaps I’ll try to run one next year, since the intrade thing isn’t really satisfactory -W]


  32. Well if you are going to try a market, it may as well be thought out in order for it to be done properly. So worth throwing some comments at it and other comments would be welcome.

    I would like to see:

    1. A play money market like or even just get the contracts up on there. Open to anyone.
    2. A real money market – this requires consideration of who, if anyone, holds the money.
    3. A market open only to experts/researchers. The question arises of whether they would put the time/effort/money in? A possibility is to providing a play money market with a prize. I would offer a £50 prize to see this happen but it really wants someone to sponsor it and offer a larger prize. There must be some company doing something relevant like offering Arctic cruises who would offer something worthwile for some potential for some publicity.

    The purpose of the three markets would be to see which react first and which does best.

    What markets to list?

    If we want quite a few:
    NSIDC September average extent less than 5.5m km^2
    ” ” less than 5m km^2
    ” ” less than 4.5m km^2
    ” ” less than 4m km^2
    ” ” less than 3.5m km^2
    round((NSIDC September average extent – 3.5m)/2m,2)*100
    NSIDC September average area less than 4m km^2
    ” ” less than 3.5m km^2
    ” ” less than 3m km^2
    ” ” less than 2.5m km^2
    ” ” less than 2m km^2
    round(NSIDC September average area – 2m)/2m,2)*100

    It might be worth asking ideosphere if they would licence/make their code available for this sort of exercise without fee or very modest. It might already be free & open source for non commercial use.

    Further suggestions?


  33. Exciting bifurcations are also a function of the real world, as both Venus and the Neoproterozoic “glacial tills with cap carbonates” demonstrate.

    Your simulations didn’t cover the important range. Interesting, even very interesting, but not important. What you showed is that the Arctic sea ice is stable at pre-industrial levels of greenhouse gases, a conclusion that is also supported by recent geologic history, and by the evolutionary history of various plants and animals that live on, in or under the sea ice.

    “I do hope to have tim eto read it in more detail”

    I do want to know who Tim Eto is.


  34. >”Your simulations didn’t cover the important range. Interesting, even very interesting, but not important. What you showed is that the Arctic sea ice is stable at pre-industrial levels of greenhouse gases, a conclusion that is also supported by recent geologic history, and by the evolutionary history of various plants and animals that live on, in or under the sea ice.”

    I think that the paper showed more than that. It also showed that there is no big lag in the ice adjusting to new conditions.

    If the ice is diasppearing faster than the model predicts, then what CO2 conditions should be used to simulate the near future without summer ice to see the effects? Or is this being hopelessly alarmist?


  35. Thought we had more extent than 2009 today but Jaxa has adjusted down the figure.

    If the extent reductions follow 2009 reductions we would end up with 5.24m km^2 a win for WC
    If it follows 2008, we would get 4.41m – a loss for WC
    If it follows 2007, we would get 4.78m – buffer zone.
    So 1 all using last three years.
    If we go back another 3 years, it is 3 WC wins 2 losses and 1 buffer zone.

    Still enough uncertainty in it (and don’t forget ‘the past is not necessarily a guide to the future’ warning).


  36. Oh my, and I thought my blog was embarrassingly silly. At least I don’t go on camera talking about things I don’t know anything about.

    Thanks for the laughs, Rattus.


  37. According to the IJIS report for August 15, 2010 the Arctic ice extent has already fallen below 6m sq km and the trend line seems to be consistently below but paralleling 2008, so it will easily pass the 5.5m sq km before the end of the month. To quote crandles above #42, “If it follows 2008, we would get 4.41m – a loss for WC.”


Leave a Reply

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

You are commenting using your 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

%d bloggers like this: