« Nares Express is ready to leave | Main | Ice pack in full »


Feed You can follow this conversation by subscribing to the comment feed for this post.



I've moved up a notch since last time. My approach is primarily heuristic, however, I use statistics and trend analysis to establish upper and lower limits as to what could be realistic projections.

In looking at the past decade, years that had an up-tick, did not have dramatically big increases. Whereas, years like 2007 and 2012 had dramatic losses from the previous year. Therefore, I think an increase of more than 10-15% is unrealistic.
On the lower limit, I can not envision two successive years of dramatic losses like last years.

During the February/March fracturing event, I thought for certain that the stage was set for massive losses this year. Then the slow start of melting and low DMI temps began to indicate that we may not break any records this year. PAC-2013 has certainly done a great deal of damage to the integrity of the ice cap, and now the meteorological conditions are ideal for rapid melting.

Every year is a new science experiment in the arctic. The lessons we learn this year will be valuable in helping to understand what is happening in the few remaining years of summer ice.


My gut tells me to move back from 3.75 to 3.95 million km2, right between 2007 and 2012. But I find it extremely difficult to make a choice. I really have almost no clue as to how this melting season is going to turn out.

Artful Dodger

3.9 M km^2
+/- 0.8 M for 95% Confidence Interval.
Method: linear least squares fit from 2006 to 2012 NSIDC Sep SIE.

Chris Alemany


I'm basing this simply on intuition and nothing really concrete aside from the assumption that there will be a recovery after last years big drop.



.4 higher than my initial estimate.

The poor condition of the ice doesn't seem as important as I thought. Looking at the numbers, even 3.1 seems way too low. Nevertheless, I wouldn't be surprised by something much lower, given the right weather conditions.

Jai Mitchell

I am going to wait at least one more week before revising my estimations lower. . .

Hans Gunnstaddar

I'm zealously holding on to the idea my first two identical predictions of 4.35 can still deliver the goods in September. So I'm sticking with that number, which if it comes to pass would put it just above 2007. Nothing like unwavering consistency to create the illusion of predictive accuracy.



It's what I had, and it's what I'll keep. For better or for worse, for warmer or for colder.


Average September Extent: 4.9 million

This is much lower than my June prediction on the basis of the recent catch up in the melt and also the fact that the ice on the East coast of Greenalnd has almost completely disappeared. However if you are looking at SIA (CT) and SIE(IJIS)numbers, we are still hugging 2008 and 2010 trends, so I would expect to see the minimum around the same figure (4.7 in 2008, 4.9 in 2010).

Glenn Tamblyn

3.25 which is higher than my previous estimate.

Beaufort/Chukchi lagging so much although the high over there now may reverse that rapidly in which case we can catch up to 2012 within a few weeks.

CAB thinning has to have an impact that we will see later

The FRAM Express has been derailed lately. If that restarts, with the East Greenland Sea/Fram so empty right now, transport out that way becomes easier when the wind patterns are conducive.

Around mid August the previous years had started to shallow out but 2012 just kept dropping. If we see the same process this year then we could still see a small number. CAB thinning is the thing most likely to cause this.


.75 m^2K
I view this as very conservative, looking at this graph, which to my mind is tracking the mean more than predictive, looking at the residuals shows despite the numbers/records we've had 2 years of reverting to the mean/recovery!
Personally I prefer the green line in this graph, since it looks more realistic for the beginning of the curve, and I would like to see the residuals for that line, which would show a more pronounced recovery still.
Again ignoring the numbers, but eyeballing all the graphs on ASIG, seems to indicate entering new territory around D230, so lets say Aug. 21 for a new record, based on the 4 common elements of the melt season graphs, which are the initial curve around peak ice, the following 2 slopes which change around the solstice and the slowdown/recovery curve from early sept. to the equinox. Once established the 2 slopes tend to persist albeit with deviation, and this years post solstice slope is particularly steep, across the graphs, indicating, to me at least, the possibility of entering September in record territory.
I think we have a new regime in the disguise of the old.


Average - under 2 million sq km for the month.
Minimum - under 1 million sq km

Based on the horrid condition of the ice.

If you look at the cryosphere today regions, ALL of them are irrelevant and will be 0, except the central arctic basin. If you look at the ice condition in the central arctic basin, it is terrible. It is extremely fractured, blue and thin compared to history. The ice has pulled off Ellesmere at least once this year along the entire shore, and a couple of times for much of it.

Once the ice shatters, like ice cubes in a warm punch bowl, it will melt quickly as the normally stagnant insulating layer under the ice mixes with deeper more saline waters as wave and wind take over.

There is a chance this won't be that year. I rank that less than 50%. If it isn't, next year is likely the big precursor to the first ice free arctic summer.

But the vagaries of polar weather (any weather anywhere for that matter) is such that any prediction from 0 to 4 million sq km has a likelihood if being correct.

Looking at the drift of the Borneo web cams, it looks like they will go for a swim in the next month, probably early in the next month.



I'm revising to 3.25 (up a tick).

Chukchi, Laptev and the CAB on the side of Svalbard are moving quite fast at the moment. Overall ice integrity still seems to be a mess.

I've also moved my predicted chance for near zero ice from 10% to 5% (lots of ground to catch up). I still think there's a high (60%) chance we'll see a record in area, extent, or volume.

Ac A




It's what I had, and it's what I'll keep. For better or for worse, for warmer or for colder.

You may now kiss the PositiveAO. ;-)

Charles Longway

As expressed by Paul Beckworth, “Zero, Nothing, Nada” for all the ice north of the North Pole, that is the Russian side. On the American side a few ice cubes are left in the Beaufort, Baffin and Greenland, perhaps 0.3. This leaves the rest in the CAB and CAA.
CAB 3.0 - The dragon returns and flash melts the Western side of the Central Basin with a cyclone. The Laptev bite extends to the pole for the first time. The media and public do not notice. Parents fail to inform children that Santa was forced to relocate to Greenland during the off season.
CAA 0.3 – Ice is exported from the CAB through the Queen Elizabeth Islands into the CAA. This export through the CAA opens this year, giving assurance of a complete melt to come before 2020. My expectation is that something will happen this year that has not happened before. The heat is in the water and air. Nature will find a way to deliver it to the arctic.


3.25 +/- .25

I'm still sticking with this. If I'm off significantly, I think it will be because I am too high.

I'm generally in agreement that the Siberian side will melt out extensively first, but, the Chukchi melt is extensive, and there has been significant "draw back" from the coast along the Beaufort in the last few days. I think that there's a 50/50 chance of 90N being ice free at the end of the season.


4.18 on September 14 -- I still think CAB will hold up just fine with cooler than average temperatures and little additional bottom melt. Additional in ES and Beaufort/CAA.


I'm hoping to do an ensemble model for August with some automated variable selection, but I'd like to add some measure of the variations in pressure over time - I saw in a previous post something about the arctic oscillation index ... could somebody tell me how to get that data for use in modeling?


3.3m for me.

That's up from two previous 2.8m estimates and just a reflection on the reality of weather.

I've been watching the Artic Sea Ice Graphs and temperature just hasn't risen above average in the arctic for a couple of months - that's a dampener on what could have been a massive record year IMHO, with the same or similar net conditions as 2012.

Still very low in context and we could get surprises especially if the Beaufort Gyre kicks in to action. North East Greenland http://ocean.dmi.dk/arctic/image_container.php is a mess with land fast ice slewing off and plenty of open waters to melt out any MYI that comes its way.


4.2m again.

PIOMAS volume seems to be lagging a long way behind the last couple of years, meaning a big catch-up would be needed to get close to last year. But with all that first year ice, there should still be a big melt to come... so I'm still betting it'll reach 2nd place, if only barely. (BTW, I really don't know all that much about what I'm talking about).

John Christensen

4.25M KM^2, because:

- Better geographical distribution of increase in winter ice volume gain based on increase in the CAB, ESS, Laptev, Beaufort and Kara in winter of 2012/13 rather than in Bering, Okhotsk and Baffin as we saw in the winter of 2011/12

- Heated surface water is much more effective at melting ice than heated air or sun radiation, and CAB, Kara, Beaufort and ESS all have better ice cover (CT SIA) than most recent years at this time still causing less opportunity for heating top water layers

- DMI 80N temps staying low also indicating reduced water heating

- Solstice has been passed 3.5 weeks ago causing radition-based water heating to be slowly retreating

- Seemingly reduced ice transport via Fram, since SST along Greenland east coast is about normal

- An optimistic inclination to favor weather vs AGW factors, believing 90% of what we are looking at is just weather..

Kevin McKinney


I'm revising upwards in light of the prolonged 'slow start.' The basis is roughly the same: 2012 extent, 'fudged.' But I've flipped the fudge factor of 5% from lower than 2012--based on the possibility of thinner ice resulting in a 'regime change'--to 5% higher, based on conditions throughout the melt season to date.

I still wouldn't be that surprised to see 3.4, mind you--but the odds seem tilted more the other way by 2013 weather.


At least one more week for me to watch and see how this current trend goes. If we start breaking records on extent and area loss, then I'll see whether my estimation needs to change. If not then I'll up it slightly.

I notice the CT area plot is showing that we're ~0.5m away from 2011 which means we gained ~0.5m in the last week on 2011.

It's still not looking much, if anything, like a bounce to me.

Ned Ward

4.4 (same as last month).

I have a simple "weighted mean of models" method that predicts the NSIDC Sept monthly minimum based on (1) the most recent IJIS extent, (2) the most recent PIOMAS volume, and (3) the year. (All of the models are fit based on the 2002-2012 data). I take the mean of these three predictions, weighting each model's prediction based on its standard error. The results this month are basically the same as last month.

It sure looks like there are a lot of commenters here expect to see a remarkably low Sept minimum. I see that from this month's SEARCH report, we (Neven's ASI blog) are at the extreme low end of the predictions. Unless there's a pretty radical change in ice behavior soon, it seems possible that WUWT's June prediction (4.8) will be more accurate than our August one. That would be a bit embarrassing.


Ned the only thing right about WUWT , the only value that they give to the world, is that they are always wrong, you can go to the betting shop place a wager against anything they predict and get rich.

The only way for WUWT prediction to be right is for the laws of physics to be altered, an act of God. :)


2.2 (same as before).

Reasons more or less as before but I will confess to an irrational influence: Here in York it's hot enough to give me a headache.


Wayne: "you can go to the betting shop place a wager against anything they predict and get rich."

Usually, but beware the liar who sometimes comes up with a truth even if its accidental. ;-)

Artful Dodger

Recall that not a single prediction submitted to the 2012 SIO was below the actual September NSIDC SIE extent. I believe that's because SIE is a poor proxy for the true state of the ice pack. The people who predict zero SIE are not wrong, they're just early:

Prof. Peter Wadhams said in Sep 2012:

"This collapse, I predicted would occur in 2015-16 at which time the summer Arctic (August to September) would become ice-free. The final collapse towards that state is now happening and will probably be complete by those dates."

It's very possible however that a prediction of 6.0 is wrong, in the sense that it may never happen again (say, in the next 1,000 years?)

Even with the most optimistic linear model of sea ice decline, there's a best a 50/50 chance of ever seen that level again due to natural variability. Big volcano? Better come soon. But how likely is that?


Ned Ward

Well, it seems to me that if you're specifying a date in a prediction, then the date matters. Saying "Sea ice will be X in 2013" means you're predicting it will be X in 2013, not 2015 or 2020. Early is wrong, late is wrong, too high is wrong, too low is wrong.

Of course, some predictions may be more wrong than others.

As for the fact that 2012's extent was much lower than all the SEARCH predictions, I wouldn't put much weight on that. If SEARCH had been operating in 2007/2008, it's reasonable to assume that 2007 would have been lower than all predictions. But 2008 was not lower than 2007. Likewise, there's no reason to assume that 2013's extent must be lower than 2012's.

Jai Mitchell


in your 1000 year estimation you do not consider future cryosphere response to the potential uncertainty of an AMOC system slowdown and its unknown effect on future september sea ice extent levels. I have a personal pet theory that this dynamic is responsible for much of the paleoclimate stability in the geologic record (and generates much of the uncertainty in interglacial CO2 sensitivity studies).

Paul Beckwith

Essentially zero. I have posted a smattering of my reasons in previous predictions on this site and these still hold. I expect that as the sea ice extent decreases and the SST in these fringes increases there will be sufficient evaporation along with the large temperature gradients to set up a massive basin wide cyclone similar but larger than the one in August, 2012. This event, if persisting for a few weeks would basically clear the basin of sea ice. A second similar cyclone would finish off the job. These events could be called "The Great Arctic Flush".

Unless there's a pretty radical change in ice behavior soon, it seems possible that WUWT's June prediction (4.8) will be more accurate than our August one. That would be a bit embarrassing.

In the last 21 days alone, IJIS Extent has dropped almost 2.4 Million Sq Km. Since much of the remaining ice is very vulnerable, I find it inconceivable to lose less than 4.0 Million Sq Km in the next 6-8 weeks. 2007 is fair game and 2012 is threatened.


I'm sticking with my original .9 million Sq. Km with at least one day being ice free. One good cyclone will IMO cause the Belmont Lake Effect AKA the Great Arctic Flush. We'll see soon enough. Regardless of the exact timing, (in geologic terms a few years is an eye blink) the fact that we are witnessing the death spiral of the Arctic Ice is both depressing and exhilarating at the same time. It sure makes MY head hurt.

Hans Gunnstaddar

Although I've predicted a slight rebound (4.35) already posted above in this thread, I commend you Paul on your steadfastness of sticking with your initial essentially zero prediction. Even if there is a slight drop from 2012 the alarm bells should ring loud worldwide (although it will probably be ignored), but essentially zero certainly should gain great attention if not actual action.

I'm curious about your opinion as to what happens once zero comes to pass (in whatever year that may occur). May I ask what do you expect in the following years? Do you see a steady state of freeze and thaw of FYI in Sept., or an earlier date each year for the flush, or increasing methane releases, or...?


Good maths OldLeatherneck! 200 K daily losses are easily achievable as well, is a matter of winds pushing the ice wherever open water is. It is also conceivable that the ice will be so scattered as to appear to be a slower melt, like now, explaining this to the public remains essential, because the contrarians will not hesitate in misinforming.

Glenn Tamblyn

Looking at both the extent and area graphs from various sources, if this year were to continue on its current trajectory it is likely to meet the 2012 curve at around day 220.

2 things suggest to me that this is possible.

The thinning reported by Hycom in the CAB seems to be allowing the pack to compact towards the pole, allowing scope for extent/area reduction just due to compaction.

And if the high pressure system over the Chukchi/Beaufort persists, it can reduce that area significantly.

The significant difference in 2012 was that around mid August, when the curves in other years had started to shallow out, 2012 kept on going down. If 2013 can catch 2012 by then and then the same phenomenon occurs again after that, this year could easily reach similar values to last year.


@ Artful Dodger: Saw a program about Yellowstone Super Volcano. based on historical record way over do for it to blow. Mind you as it would virtually kill off man kind that would not be my 1st choice. It would definitely take care of our heating problems. Them of course there is always the dropping of an asteroid into the Gulf of Mexico. That could become a greater reality if a government desperate enough for a solution had the resources to do so. Think I watch too many of the scientific variety of programs for my health.
As for prediction I am going to wait till near the end of July to see what happens with the ice till then. Believe (BTW must declare myself as 0 on the ice knowledge front) the ice quality is extremely poor, even MYI, very spread out and could easily succumb to what some call a flash melt. There has been enough surprise bad weather patterns going on this year that there is still plenty of time for the 'perfect' conditions to take effect that could set up a flash melt.

Jai Mitchell


because you can't hindcast the future on a new planet. Comparisons with temperature and extent values/declines in 2009 (most similar temp) 2011 and 2012 (most similar current slope) show that current melt is much worse than 2009 even at lower ambient air temperatures. It also shows that 2012 experienced extreme late term melt, also independent of ambient air temperatures. 2012 experienced a continuation of melt rate beyond all previously studied seasons. This indicates a melt "momentum" that requires signficant atmospheric braking to slow, then halt (apparently by day 250 or so). In this new environment, it appears that current melt and dissasociation momentum will continue according to 2012 and surpass 2012's record low extent to reach 2.4 by day 265, or so.

of course I am just guessing. . .

Oyvind Johnsen

Since last year was an extreme drop, I expect some kind of regression to the mean downward trend. (Which we don't know, of course...) 3.9-4.0 has been my guess since the start of the season.


I'm sticking with a cautious 4.5 million km^2.

No reason other than my gut.


Pure heuristics 3.6


NSIDC extent data are interesting. If we average the last 31 days of extent loss (.1035mkm^2 average) and extend it to the end of August on Average. We get 3.35mkm^2.

If we do the same for the last 14 days (.1299mkm^2 average) and extend it till the end of August, we come up with 2.09mkm^2.

On 12 July we hit .2408mkm^2 one day loss of extent. Granted this is outlying continental areas, but it's still NOT hot.

Bets on not making 2012 or exceeding it???


Don't see any reason to change my 4.5-4.6 prediction. If anything, it may be too low and we end up something in the higher 4s.


3.4 million km2. Laptev bite promises to be special this year, but some first year ice may stay around it.


I will keep my prediction at 3.8 million km2. I agree with D, the only thing that is really catching my eye right now is the big Laptev bite. This may change later, but 3.8 seems like a good prediction to me.

Lord Soth

I am totally amazed how much ice is melting for this cool crappy arctic summer. I have revised my numbers from (lucky to break 5 million km^2 from last month) to 4.25 million km^2.

Six years ago it took a combination of perfect weather to get a record arctic melt, that stood for 5 years until 2012. This year will probably tie the 2007 record, with one of the coolest arctic summers in a good while. Which says a lot for the condition of the ice.


It does indeed.

I just compared my figures above with 2012.

1 - 14 July average extent loss

2013 .1299mkm^2
2012 .1043mkm^2

15 June to 14 July average extent loss

2013 .1035mkm^2
2012 .0899mkm^2

The next figures will be interesting to track

2012 average loss of extent for July


2012 average loss of extent for the first 14 days of August


2012 average loss of extent for August


Those, I'm guessing, will be the figures to compare. It may have been slow to start, but it's moving much faster than 2012 at this point in time. They must either converge in the end or 2013 has to slow rapidly.

They must either converge in the end or 2013 has to slow rapidly.
2013 will slow rapidly, because we will reach the end of the melt season. I think 2013 is going to run out of time.

Looking at Jim's graph here, it looks to me as though 2013 is roughly 1 million square km behind 2012, and 2012 lost roughly 3 million square km from now until the end of the melt season.

So, in order to catch up with 2012, 2013 would have to lose ice at a rate one-third faster than 2012. Your comparisons fall a little bit short of that.

I'm not saying it is impossible - if the cloud clears and a dipole anomaly gets itself set-up ice melt and export could be very dramatic - but I reckon it is strongly odds-against at this point.


I've given up being surprised at what the cryopshere in the Arctic can do. Either slowing and resetting itself or melting rapidly without apparent cause.

At this point observation is all...

But, no, I don't expect a sudden and very rapid melt of that degree unless something in the weather changes. However if something in the weather does change, the scope for melt twice as fast as 2012 is there. Because Extent is 15%, or more, concentration and there must be huge areas out there which claim extent from almost no volume. Thus are susceptible to sudden melt or compaction.

Which is why I say that being confident that it won't happen is not a very strong place to stand.


I should have checked the sheet before I replied...

2012 7 14 7.87031
2013 7 14 8.32997

which is a difference of 0.45966 (Jim's graph is area, not NSIDC extent.)

Or 1/6th more melting required in 2013 to catch up. Right now, that is happening.

End of July will reveal the trend. Although I do expect it to slow a touch from now to the end of the month.

Kevin McKinney

" Although I do expect it to slow a touch from now to the end of the month."

Yes, it normally does as insolation decreases. The shock in 2012 was how little it slowed, and for how long.

This year? Well... we shall see.

Climate Changes

2.82 MKm2 and NP ice free.

Same as my last 2 predictions for me. In spite of this years lower Arctic mean temperatures the melt goes on as 'normal'. If extent decline carries on as it is at the moment it will join 2012 near mid August.

Nightvid Cole

3.05 M km^2

large areas of low concentration in the middle of the icepack spell serious bottom melt later in the season. This should more than overcome the 'slow start'.

Aaron Lewis

1 M km^2

Paul Beckwith

Once we reach zero in year t0 (likely for < 1 month duration the first time) then I would expect longer duration zeros in subsequent years until we reach an ice free state year-round. I would expect that year t0 + 1 or t0 + 2 would be ice free for about 3 months, year t0 + 4 for about 6 months and year t0 + 7 or t0 + 8 to be ice free year-round. Clearly this will lead to an extremely different planet. Of course, the main feedbacks that drive us there are the Arctic albedo collapse (sea ice + snow cover) and the higher GHG levels (both CO2 and CH4) in the Arctic from terrestrial and marine permafrost thaws. Given the ramp-up in global extreme weather events and crop yield reductions that are already occurring today I think the near future will get very unpleasant for people. Thus my plug in blogs and in AMEG over the last few years for the "anthropogenic Arctic volcano" to slow the melt and buy a few years to slash anthropogenic fossil fuel emissions before the Arctic sources explode and dwarf human emissions.

Susan Anderson

2.9 plus/minus 1.0

same as before, and I too am influenced by local heat dome, silly me; nights are even hotter relatively than days (US northeast). Seems like things are still slow and Greenland turning but that's my story and I'm sticking to it. Strictly amateur, and that ice is going to have to get some skates to catch up.


Does anyone know what impact the increased shipping, drilling, harbour investment, fishing and other activity as the ice melts is expected to have on the rate at which what ice is left declines? Because such activity should probably start rising to a peak now... and I have to wonder if this particular positive feedback mechanism may not have been one of the reasons last year's melt went on as long as it did.


Based on melt-favouring weather, that central 'hole' melting out, and some compaction at the end of the season.

Surprises may be on the up-side, but I will be very surprised if the minimum is not lower than 2007.

I would consider this a low-skill prediction - I have learned much from this blog and other sources over the years since 2007, but by no means am an expert.

John Christensen

@Paul (and sorry about having a post on this thread not directly related to the Sept SIE minimum):

I do find your numbers to be hard to imagine, strictly from a scientific perspective, even if my understanding is quite limited, and many factors may evolve in unpredictable ways in the coming years.

But 8 years between ice-free Arctic in the summer and year-round ice-free Arctic is quite radical.

It would probably require both a 15-20C summertime Arctic SST increase, as well as probably a 20-25C increase in average winter air temperatures in the Arctic, which is hard to imagine without significant orbital change, or some catastrophic event that would drastically alter our atmosphere and likely make the Earth uninhabitable for humans.

As we saw last fall and early winter, there was probably even a positive correlation between lack of sea ice/high Arctic SST and the record high NH snow cover in December 2012.

Since ice-free Arctic summer will probably incur in the next three to ten years (or maybe a few years later), I do not see how other factors could all change in such an extreme way, and do not recall having seen any climate model being able to show such a catastrophic scenario.


"But 8 years between ice-free Arctic in the summer and year-round ice-free Arctic is quite radical."

Hmm, I don't think it's radical. Honestly, after watching the ice for over 10 years ( as a scientist but not in climate science...so enough to follow the graphs, physics, maths and images...it's my hobby :)

When it happens it will be quick. Once it happens there are no negative feedbacks - unless a volcano erupts.

The air circulation patterns are pushing hot air from all continental land masses into the arctic. As I type I'm watching a hugely long lasting powerful system push ice through the Fram.

This year I'm taking screenshots of nearly every arctic image I can find. Why? Because this year is very different, even if it doesn't look like it yet.

John Christensen

Observing the period of 1979-2013, CT has winter SIA maximum going from initially 14-15 million km2 to 13-14 million km2 in a period of 34 years.

A reduction of about 7% in 34 years.

As the last two winters have shown, even strong summer melt does not prevent a large thin layer of FYI to reappear in the winter season.

PIOMAS volume has gone from 32 kkm3 to 21 kkm3, about 34% reduction in 34 years.

As you know, the difference between PIOMAS and CT SIA development is due to the transition from a MYI pack to a FYI pack.

What graphs, physics, or math are you using to say that FYI will no longer appear in winter, perhaps only 15-20 years from now??



I completely agree. I wouldn't expect to see zero ice in winter at any point in the next hundred years - months of sunless winter are sure to freeze at least a thin layer of the ocean surface, even if the amount formed continues to slowly fall, and melts away earlier each year. Even if arctic temperatures went up 10 degrees celsius, they'd still be regularly going below -30 degrees. So why would the ice stop forming?

k eotw

4.24 +- 0.007 millions km2


Unless the arctic somehow starts seeing sunshine in the winter, there is zero chance of an ice free winter any time in the next couple hundred years. Maybe a huge asteroid will smash us and alter the tilt of the earth. If so, we got bigger problems than an ice free arctic.

Kevin McKinney

"What graphs, physics, or math are you using to say that FYI will no longer appear in winter, perhaps only 15-20 years from now??"

"Unless the arctic somehow starts seeing sunshine in the winter, there is zero chance of an ice free winter any time in the next couple hundred years."

I know, it's shocking to think that the Arctic could be annually ice-free, and perhaps even in the lifetime of some of us. But it is supported in the literature.

I had much the same unbelieving reaction to this idea when first I heard it. It was a couple of years ago, and the proponent was Gareth Renowdon (hope I'm remembering the spelling right, but fear I'm not) of Hot Topic, a very good climate blog in New Zealand. The latest post as of writing is here (and includes a hat tip to ASI, coincidentally):


Gareth pointed out that the sea ice volume loss is a proxy for the Arctic energy balance. Therefore, the question arises, what happens once we hit an ice-free minimum? Does the energy balance shift radically as a result?

Or does the energy balance continue positive (i.e., more energy in than out over the course of the year), with the result being more warming rather than more melt? If it's the latter, then we should see a progression toward 'ice-free' winters as well.

To be fair, we do know of a couple of negative feedbacks: the primary one is radiative; as the Arctic warms, it radiates better. Good old Planck! But there is also the fact that ice and snow help isolate the ocean from the cold atmosphere. With much less ice, one would expect much greater radiative fluxes.

But there's something else. The larger the body of water, the harder it is to freeze. In processing Gareth's suggestion, I considered the Great Lakes as an analogy. Even when I was a child (and living in the Great Lakes region), it was common for Lake Superior not to freeze over completely. Since then, it has become considerably more common, as you might imagine. The point is, though, that smaller bodies of water freeze over when Superior doesn't--even ones much farther south, which don't often see the fairly intense cold conditions that Superior gets in winter.

Why? Well, thermal mass, to start with. If mixing continues, then freeze is considerably inhibited because cooled surface water is being mixed with warmer water from below. That's one reason why cold, still nights are so good for ice formation. Superior has lots of water to mix, hence freezing can take a long, long time--longer than the freezing season lasts, increasingly.

Another reason is that wave action never quite stops on really big water: you can get a 'millpond calm' on a mill pond, or even on a lake a mile or two across. You essentially never get it in the middle of Lake Superior, much less the Arctic Ocean. As ice declines, wave action increases.

Finally, there's the water vapor feedback. As ice declines, specific humidity increases over the Arctic basin. That's a powerful feedback, and one that seems to be in operation now, according to some research. (No, I don't have a cite on that.)

So--the question is complicated. But luckily, there is a new paper on this--Winton, 2013:


I haven't read it in detail, just skimmed. But basically, it finds some support in modeling experiments for non-linear behavior:

Two of the climate models have Arctic Ocean simulations that become annually sea ice–free under the stronger CO2 increase to quadrupling forcing. Both of these runs show increases in polar amplification at polar temperatures above −5°C, and one exhibits heat budget changes that are consistent with the small ice cap instability of simple energy balance models.

But basically, Arctic behavior is described as 'very linear'--in most simulations there's no marked albedo 'tipping point.'

Of course, the modeling has been under predicting melt for quite a while, though decreasingly so, so that's something to bear in mind in interpreting this study.

It is clear, though, that an annually ice-free Arctic is possible under SRES scenarios. You can't take the winter ice for granted, apparently. If Winton, 2013 is right, we could pretty much take winter ice for granted for the "next 15-20 years"--but not in 2100.

John Christensen


Thanks, fully agree on all points. It's the idea of 'summer ice-free + 8 years' that is hard to imagine. The span between ice-free summer and ice-free winter will be much longer.

Hans Gunnstaddar

KM wrote: "You essentially never get it in the middle of Lake Superior, much less the Arctic Ocean. As ice declines, wave action increases."

I recall an incident that occurred just a few years ago in which natives to the Arctic pulled their kayak up on bar of land just barely above the water line to rest and eat for a while. Based on teachings from their elders and experience they expected the water to remain still and smooth as it always had. So it caught them by surprise when small waves caused the kayak to slide off the bar. The father and son swam for the boat but couldn't catch it and both drowned. The others were later rescued.

So evidently increasing wave action is occurring in the Arctic due to reduced ice volume/extent and in support of your point about Lake Superior. It is a strange idea the arctic could be ice free in the winter, but I suppose with higher water temps and humidity, it is possible by 2100 in light of how fast things are changing.

Kevin McKinney

It is hard to imagine, John.

According to Winton 2013, it is probably true that "The span between ice-free summer and ice-free winter will be much longer," since much higher forcings are needed, and there's a good chance that the approach to perennially ice-free will be linear, not sudden.

Kevin McKinney

"I recall an incident that occurred just a few years ago in which natives to the Arctic pulled their kayak up on bar of land just barely above the water line..."

Yes, I recall reading about that tragedy, too. Climate change up close and personal...

Coastal erosion is less obviously personal, but its increase provides another illustration of the point. (And also affects lives in a deep, long-lasting way--especially when whole communities must move.)

Martin Gisser

3.6 +/-0.8
Up quite a bit from my last guess. Pure guesswork from holding a piece of paper to the latest NSIDC extent graph and remembering the February/March fracturing: The current slope will quite probably continue into August and catch up with 2012. Methinks more than guesswork is a waste of brain.


Very difficult to decide, so I stick with my last prediction of 2.8 million km^2. Looks pretty low when looking at all the extent, area and volume graphs but I still have the impression that they don't show us the state of the ice. Volume is probably overestimated because of the holes in the center of the ice pack and subsequent thinning therefore couldn't be modeled correctly. What is really different to all other years, is the Atlantic side where the ice is pulling back and might go beyond the pole. There will also be significant melting from the Siberian side and in the Chukchi Sea. This leaves the broad side of the ice pack exposed and may delay refreeze. A lot of the remaining multiyear ice is beeing pushed into the Beaufort Sea where it has the longest time to melt such that a lot of it will melt as well. But still, my prediction might be quite a bit off to either side, I'm really not sure anymore...

Jai Mitchell


Excellent paper, thanks! The problem with part of the assumptions in it is when it says,

"Holland and Bitz [2003] have shown that the ocean also transports more heat into the Arctic, even as the heat transport is being reduced at lower latitudes in association with the weakened
meridional overturning circulation."

now, I didn't actually read Holland and bitz but I can tell you that they projected a reduction in AMOC based on increases in temperatures due to current arctic (circa 2003) models that held a very slow rate of decline in both sea ice and AMOC intensity. Therefore, their heat transport analysis indicated much greater warming of the oceans and a greater rate of heat transport to the arctic over the longer period of time that the AMOC was declining.

instead of the reality we face today with a (30%?) decline if the AMOC in the last 10 years and a collapse of arctic sea ice.

what then do we really see? yes, an increase of mixing a very large increase in atmospheric heat transport from mid latitudes to the polar

. (by the way, have any of you seen THIS lately? http://www.ssd.noaa.gov/goes/west/nepac/flash-wv.html -- yes those are all cut off lows.)

--this is basically a collapse of the polar cell,

an increase of vapor transport to the arctic, an increase in winter-longwave emission to space with increases in snowfall and albedo on land during early spring periods.

It is exactly this dynamic (increased atmospheric northern latitude precipitation) that allowed the Greenland ice sheets to maintain volume during the Eemian.

It is also this dynamic that will continue to cause AMOC declines, increased mid latitude sea surface temperatures and the kind of changes in weather patterns (jet stream blocking, extra-tropical mega hurricanes, summer heat waves and atmospheric rivers)

long story short, no we won't have ice free winters anytime soon in the arctic (I am betting) not until we reach +5C from pre-industrial globally averaged temps, and maybe even MIS-31 temps (+8 C) probably sometime around 2150 if we don't do something major between now and 2030. and hopefully not ever if we DO do something globally significant to reduce greenhouse gas production.

Ned Ward

Any prediction that Sept average extent will be below 2012 implies a fairly radical break with the pattern of previous years.

Starting with today's extent figure, here is the Sept monthly mean that would occur if we followed each previous year's trend:

2006... 5.8
2011... 5.4
2003... 5.3
2005... 5.1
2010... 5.1
2009... 5.0
2002... 5.0
2004... 4.9
2007... 4.8
2008... 4.4
2012... 4.1

So my prediction (4.4) assumes that the next two months will match the second-fastest melt rate of the past decade.

Predictions below 4.1 assume that the next two months will have more ice loss than any previous year. For example:

05% faster than 2012: 3.9
10% faster than 2012: 3.7
15% faster than 2012: 3.5
20% faster than 2012: 3.4
30% faster than 2012: 3.0

Note also that it's not necessary to go back very far to find years that would give Sept mean extents greater than 5.0 when starting from today's extent.

In fact, three of the past four years would yield Sept mean extents of 5.0 or greater, using their trend from today's extent.

Based on the numbers posted so far in this thread, the average prediction from this blog's commenters apparently assumes that the next two months will have a 20% faster ice loss than the same period in any previous year. For comparison, over the past two months 2012 has ranked fourth.

It could happen. But it seems unlikely to me, this year.


Does anyone know how much warmer an ice free ocean (3 months) will get?

Craig Merry

There is way too much weak ice (proportions of age and zones) for it to survive bad conditions. There's a lot of time for the ice to melt out. One good melt and it'll blow past 2012.

1.5 +-1


The extent loss rate is already at almost 2012 + 30% for the last 14 days or so.

The question is not if but whether it can be maintained. Or worse, exceeded. In the current weather conditions it would be another paradigm shift.

Time will tell.

Vergent Bill

ASIB has just been cited as a source and quoted in a Washington Post article.


The losses to come in the Arctic Basin are unprecedented. The area can be best seen here;

Credit danp and Lance-Modis


Vergent Bill


The second link didn't come through. You may get a security warning, or you may have to click on a shield on the right of the address line.


Ned Ward

NeilT writes: "The extent loss rate is already at almost 2012 + 30% for the last 14 days or so."

Well, it's one thing to have a high rate for two weeks. It's another to sustain that for two months.

Again, it could happen. More often, though, the weather changes sooner or later to a different pattern. I guess we'll just keep watching and see what happens.

Paul Beckwith

I wonder how many people will revise their forecasts down in the next week or two. Both the GFSx and ECMWF are showing that persistent cyclones are on the way with nice amplitudes in about 6 days. The Navy sea ice motion forecast is showing that there will be some action even within the next few days.

For those that think a year round open water Arctic ocean is 100 or 200 years away I offer the following wager. If there is any sea ice there at any time during the year in 2100 I pay you $10,000. If the ice vanishes year round before 2100 then I collect the $10,000 from you at the time of this occurrence:) Which would be in a decade or two. Wager null and void upon the death of either party.

People forget that the Arctic stayed ice free quite happily year round in the past when the continents were in the same locations as today. The Arctic region supported lush forests and critters that required temperatures to remain above freezing. This flora and fauna adapted quite happily with months of total darkness and total lightness. Also forgotten, is that there have been many abrupt changes in the past, on time scales of a decade or two. Paleoclimates by Cronin is a good reference book on some of the basics in this field.

Paul Beckwith

From the U.S. Navy sea ice thickness plots, notice how the black and red thick ice just north of Canada has no cohesive strength at all. This became abundantly clear with the fracturing events in March and also with the ice divergence in the red region from small persistent cyclones. This thick ice consists of fractured ice that has simply ridged up in these regions. The fractured chunks may be comprised of broken MYI but the aggregate has no strength so will behave completely differently from previous years.


3.45. I had bet 3.25 before, but i believe the slow start will limit the loss of extent. This would still be a record, i believe the late season melt will be strong. I do not think the area (cryosphere today) will break a record, i predict it will be tide or slightly more than last year.


Wet, very wet.



@Paul Beckwith

"People forget that the Arctic stayed ice free quite happily year round in the past when the continents were in the same locations as today. The Arctic region supported lush forests and critters that required temperatures to remain above freezing. This flora and fauna adapted quite happily with months of total darkness and total lightness."

Are you implying that an ice free Arctic ocean will have no impact on civilization? Our civilization depends on agriculture which, unlike a natural ecosystem, is very fragile and susceptible to extreme short term changes.

Paul Beckwith

Of course not. I am not implying any such thing; such wrenching changes will devastate populations and civilization and are already underway. My comment was geared to the people who have a hard time thinking of an ice free Arctic.

Given the extremely weak state of the ice back in March, I think that the extent drop was delayed a lot this year due to the much greater melting occurring within the ice from the greatly increased surface area of all the fractured ice chunks. This used up all the available heat (latent) and kept the surface temperatures (sensible heat)lower than normal and subsequently the atmospheric temperatures lower.

In other words, the sea ice caused the cooler Arctic temperatures, not the other way around.


Paul Beckwith Said: In other words, the sea ice caused the cooler Arctic temperatures, not the other way around.

Ah, excellent. Nice to see someone with more knowledge provide support for a tentative hypothesis I was forming regarding exactly those temperatures. I find it interesting that for a significant period of the time when temperatures were low, in many areas they were hovering right around -1 to -2C. If the temps were being driven by radiation or some other mechanism, I would expect the ice to provide insulation against heat transfer from water to atmosphere. If there was sufficient exposed water, or thin ice, we might just see exactly this sort of temperature range. Anyone who's lived on a coast and experienced sea breezes would agree that water tends to affect the temperature of atmosphere far more than the other way around. Similarly (in coastal Massachusetts) in winter, we could depend on the water keeping us about 10 degrees F warmer than the interior during cold snaps.


As Paul has started to point out, there seem to be many that have forgotten that there was NO red left at the the end of last year. I am no expert, but based on what happened to the rest of the ice by the time of maximum I do not think that that band of ice now was created by temp. I believe that was mainly compacted ice. Not only that I do not think that last winters weather would have made it stable thick ice.
Based on that premise, that band of thick 'MYI' is just waiting for the right conditions to start moving. Is there time. Sure, there is still at least 6 weeks. Get some wind and rain there and that could become very thin very fast.
Also. remember that cloud can also mean rain. Rain can melt out ice faster then sun. Not only that, rain can melt through the ice by creating small channels from top to bottom that satellites can not see.
I believe that what we are really witnessing in the last few years is that once the ice is weak enough, there are very many more ways of melting ice then sun and high temps to melt ice. As one who lives in a zone where you can go from -30C to over 30C during an entire year, once temps get close to 0C you always have to be wary of ice under foot and over head now mater what the weather is, because that ice has become unstable and can move and/or disappear on you endangering you. This can occur very fast even at night.
I will not give my uneducated prediction until almost deadline time because I do not trust the shape that ice is in, and I do not believe that there is quality volume out there there can hold out from even a mild weather melt attack which other then the PAC very early in the melt season has not happened yet. Get some above average temps with some wind and rain and I am convinced that ice will flash melt all over the Arctic. The quality of the ice this year is very very poor IMHO.


For such a cool and cloudy year, 2013 continues to lose extent faster than 2012, on average, over the month.

I very clearly remember 2006 and the expectations. Then the weather came over, it cooled down and it all stopped for a while. Diverging greatly from 2005.

That has happened this year in a way, but the stop was earlier and the pickup in extent loss earlier too.

This is a different ice pack and the energy in the ocean is a different value. I don't think we can make comparisons with previous years. We can only observe and record.

I don't think it's possible to call the state of the ice, in September, right now. It is far too dependent on the next 2 - 3 weeks.


The circulation in the arctic is still going on. The latest sat polar orbiting visible shows the system spinning away merrily. It is just about to get a kick from a jet stream loop ( I'd roughly guess about 4-6hrs from now )

It will be very interesting to see where it sits once that boost of air hits. Coastal areas are very warm so I imagine the inflow to warm up the system fairly quickly.

I love this site for the vast array of satellite data.


The latest image from CT shows the swirl of the storm affecting the ice! The hook of red into the north pole is a sweeping arm from the circulating storm.

I'd watch that red area over the next day or so - it will continue to curve into the south of the pole and through the Fram. That will push the mushy ice into warm ocean. There will be a dramatic drop in the next 3-5 days.



I hate to say it, I know you guys think I'm an oddball, but the next few days will bring in the storm so many people here know will break what MYI is left into oblivion


@ Kate | July 19, 2013 at 13:51

I am unable to remember one oddball who warned me of the GAC-2012.

I like what you see. Could the weather produce a GAC-2013 during the next 2 -to- 3 weeks?

A crude animation, starting 13JUL2013

from IARC-JAXA "Arctic Sea Ice Monitor"

Of note to me is on 13-JULY the area between Novaya Zemlya and Svalbard 'seemed normal' then on 14-JULY I read 996 millibars on the Sea Level Pressure "Overlay" with "LOW" moving toward the NP on the following days, then turns toward the Laptev Sea.

If it's able to pick up warm (hot) air from the Siberian area could the low-go-lower?

Let the the ones who warned us last year, 2-3 weeks prior to the GAC-2012, cast the first oddball comment.

We don't know "exactly" what is going to happen up there so any clue you find talk to me about.


Everyone, could we keep this comment section restricted to predictions + rationales?

There will be a new ASI update on Sunday, and there's plenty of room on the Arctic Sea Ice Forum.

Sorry for my late suggestion. I'm busy and I don't get notified of new comments on blog posts I didn't publish.

Patrice Monroe Pustavrh

On my gut feeling, I'd put this year somewhere between 3.2 to 4.8 mio km2, with center on 4.0 mio km2.
Rationale: there has been indeed slow start, but ice looks weak at some points and last year we've seen quite rapid melting in august.



My last month's guess was 3.8; my rationale is mostly heuristic.

Initially I thought that with volume decreasing every year, at some point each successive year could easily yield a new record, culminating with an ice-free (<1m) September in 2016 +/- 3 years.

On the other hand, I also supposed that right before the end, the area and extent could indeed rise, like a drink where the few remaining ice cubes have been replaced by a slushee.

What I didn't expect was to see PIOMAS volume actually rise this year - so last month I figured all bets are off, and there could indeed be a slight recovery this year - hence, 3.8

However, one thing I am still struck by, is that it seems to me that everyone who has said it would be nearly impossible to break last year's record, either because of the early cool period or statistical methods, is comparing, contrasting or correlating this year to past years...

I realize we do not have the advantage of future years' numbers with which to compare, but I do recall a good argument in recent years that there was a significant shift in various graphs between 2005 and 2010 (IIRC), and that 2007 was more likely to be a dragon-king event rather than a black swan. From my admittedly purely gut perspective, are not 2011 and 2012 more likely to be confirmations of this, and not more black swans?

If so, I would think that 2013 will more likely correlate to the melt mechanisms of the future (or even less predictably, a present in transition) rather than the past.

I realize this is not at all helpful in the quest for an actual reason for a particular prediction or expectation based on the only data numbers we have available, but the call was put out for all input, and this one is mine.

One last thing - if GAC2012 melted out an area last year that was going to melt out anyway - just earlier - could PAC2013 be its converse? i.e., it only delayed the melt which was inevitable?

I still wonder about the PIOMAS volume, though... I suppose only time will tell.



The cyclone is definitely intensifying as I write this. I follow a few sites to get a very visual idea of what's happening. I guess this is why I'm sometimes a bit vague ( I know nothing of the maths! )

Heatwave air from the US is now in the North Atlantic, curving up along the east coast of Greenland - no matter what this is very warm dry air and it going anywhere near the arctic is bad.

A "GAC-2013" is happening now, yes. I would say so. I see no negative feedback to the system that's forming atm.

I use these links to give me an idea of what's happening. Sorry if some don't come out right. And sorry if this is not what you wanted. I do predict a big year.






Worldview( link not working atm )




Sorry Neven, just saw your post.


3.3 +-0.6M Km2

Up from 3.1 previously.

100% intuition.
To be honest think that the value could be anywhere in the range provided. There is at least 1M km2 of flat, bad quality thin ice. Whether this melts or not depends on factors which cannot be predicted or even measured. That is, the metric of minimum sea ice extent is now chaotic.

Kevin O'Neill

2.9 Mkm^2 (-0.5, +1.0)

Consistent not because this number seems inherently correct, but because I can't think of any good reason to change it. With this completely new melt pattern comparison to previous years is difficult.

The comments to this entry are closed.