« ASI 2014 update 3: here comes the Sun (again) | Main | ASI 2014 update 4: high times »


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

James Sutherland

I was thinking of submitting under the name The High Arctic Wizard and His Dice Of Spectacular Predictions. They'd have to add Magic as an outlook method though. This would be the control group. If your predictions fare poorer than the Arctic Wizard's then your prediction method must be very poor indeed!


Neven, I don't think citing Mr Wrong Bastardi , who believes CO2 is a trace gas of no significance, does any good at all but propagating bad science. It is rather more useful to concentrate on success, for those who get it right often cited less, by coincidence, are proponents of the scientific method. I rather see this like we have been hijacked by contrarians made popular like Don Quixote. But usually in the real world, like at current World Cup, the team that looses out in disgrace does not get more attention then the finalists, most will remember the winner rather the teams who exited the contest earlier. So how come we keep on harping on the loosing contrarian team with attention deserved to winners? , Again and again? Would it be rather more useful to explore the reasons why one person or group got it right?

As far as 2013 was concerned, the lessons learned was that no compaction of sea ice causes a feedback loop favouring cyclonic penetrations or persistence over the Arctic ocean, these cyclones fuelled by thinner ice or more open water (rather than less) brought a sun shield (clouds) and winds cancelling the Arctic Basin Gyre.

Now this lesson,completely gets lost in the lore of contrarians, who despite dimly trying, don't have a clue about the meaning of a single season. Rather they call it "cooling", because of some misinterpreted reason they take from who knows where.

The predictions for biggest melt then were wrong, because most, including myself who despite seeing the cyclonic season coming in April, did not appreciate the importance of counterclockwise winds enough.

Instead of saying this guy or that model got it right, it is rather much more useful in explaining how , if wrong likewise. But why bother with a guy not accepting very basic scientific precepts? Unless he or she gets it always right, which sadly for contrarians, almost never happens, when it does, on the rare occasion they are heroes after missing the goal 1000 times, they score a fluke, so what?


So you're admitting there's excessive alarmism in the ASIF


Wayne, I fully agree with what you're saying, but at the same time one of my interests connected to Arctic sea ice is the perception of what is happening with the sea ice and the consequences thereof. For me fake skeptics represent that part of the collective consciousness that keeps it where it is, in denial of reality. There might come a point when Arctic sea ice loss keeps continuing, where they can't wriggle out any longer in a credible way (not that the ones profiting from it, won't try). That could be a sign that the collective consciousness is changing.

It has nothing to do with the science of predicting the minimum whatsoever.

So you're admitting there's excessive alarmism in the ASIF

Well, there was last year. People (including myself) kept voting low, even in August, when it was becoming clear that all that potential, the huge dispersal and holes all over the ice pack, wouldn't come to fruition and get expressed in extent numbers.

At the same time it's understandable that alarmedness about mid- to long-term trends in Arctic sea ice loss - given the extremely rapid loss so far - leads to alarmism regarding the short term trend. Paradoxically, people who are worried about Arctic sea ice loss, also hope it will come faster as a wake-up call (I wrote a piece about that back in 2010: To melt or not to melt). Combine that with how difficult it is to make sense of what is going on and remember enough of recent years as a reference, and you get those low votes.

But again, absolutely no one expected last year's massive outlier. Not even the community over at WUWT. And in 2012 it was the other way around. Practically, no one, including me, was expecting that record smashing crash. Which also explains people bracing themselves for the worst one year later.

The jury is still out on this year. It looked like a repeat of 2013, though less extreme, but now it doesn't. Who knows what it'll look like in 2 weeks. I still don't think the record will be broken, but it's the Arctic, so mustn't rule it out either (yet).


Neven, as I said in the last post, why is it conventional wisdom that 2013 was weaksauce from the beginning of the melt season?

Looking at the CT comparison, 2013 was keeping up well until it stalled out for 9 days.


I agree Rlkittiwake, but I said: "People (including myself) kept voting low, even in August".

Last year I learned that extent records don't get broken when there are persistent cyclones during the start of the melting season.

Chris Reynolds


I'd say so.

Then again I have the fifth lowest prediction in the June SIPN, in my defence I worked out the method and then came up with the numbers.

I've learnt to avoid using terms like 'crash' in blog posts. And have noticed that blog posts where I suggest the ice will take longer to reach a seasonally sea ice free state get fewer views than ones where I suggest the transition could be rapid. Still not made my mind up on that, but I think I'm tending to later next decade.



The June SIPN predictions together with uncertainty ranges are plotted on this graphic.

The prediction that Bastardi is using is the 'Wang' prediction for June SIPN. Updates are given here:

The method is discussed in the PDF at the bottom of the SIPN page here:

Essentially they're using NCEP/NCAR as a forecasting model. Having read the WUWT page in which the forecast was hurriedly arranged, given the impact of that (one) piece of information of the WUWT readers. Wang and WUWT can be considered closely related (at least).

Predicted September extent is plotted here:
See L2 under "Sea ice extent and concentration" at the NOAA page linked to above.

Initial conditions are stated to be for 13 to 22 June. I would suggest that the weather component of the model does not predict the current excellent melt weather to persist. GFS and ECMWF show dominant highs over the Arctic Ocean for the next 8 days. I expect significant losses for at least that period.


A note on my prediction...

Prior to the 18 June when the mode shifted from low pressure to high pressure I had been very pessimistic about the chances of my prediction (Reynolds SIPN) being borne out. Now, despite the slow start to the season in June, it seems that there is a good chance that the current Arctic highs may persist, with ridging over Greenland starting leading to a summer with weather like the 2007 to 2012 summer average (SLP), a pre-requisite for my prediction. I am now moderately confident (~30%) that 2014 September extent will be in the upper half of my range (i.e. 4.06 to 4.62 million kmsq).

Chris Reynolds


2013 was characterised by a stall in volume loss in May leading to an increase in volume entering the melt season proper after the solstice (21 June).

Without this spring volume loss, prospects for large loss of area/extent was limited. This was then compounded by the weather during 2013.

Here's average SLP (Sea Level Pressure) anomaly (difference from average) for June July August 2013.
This was almost exactly opposite to the average anomaly from 2007 to 2012.

That was the weather impact, conditions were almost exactly opposed to those that favoured ice loss in the years 2007 to 2012.

Within each summer from 2007 to 2012 the short term SLP anomaly would not necessarily look like the average shown above. But just to look at what happened after 18 June this year, here's before 18 June average SLP anomaly.
And here's the average SLP anomaly afterwards.

As an aside...

Looking at that last plot above, it seems to me that what is missing is low pressure around the south of Greenland. If this emerges this year it will be formed as the atmosphere over Greenland ridges anomalously and steers the jetstream to the UK breaking down the blocking high seen westward of the UK since the 18 June. However within the Arctic in the 6 days since 18 June 2014 NSIDC Extent has declined almost as much (-0.707M kmsq) as in the 14 days preceding 17 June 2014 (-0.719M kmsq).


To play Devil's Advocate, the systems up there are so chaotic that it seems like there's the danger of falling into the "supersitious pigeon" hole, where tenuous causations become doctrine.

Which isn't to say that the weather up there is tenuous, it's just that there's still several more months of it before we bottom out.

Speaking for myself, I'm skeptical because I've been surprised every single year since 2007. ;)


The issue with any prediction is that August is highly variable and dependent on weather, whereas melt to end of July is comparatively consistent.
Based on the PIOMAS data, the loss of average thickness to Aug 1st ranged from 1.77cm - 1.85 cm over the past three years. 2013 was 1.82. After August 1 the loss was 52cm 2012, 26 cm 2011, and 18cm 2013.
I'll be happy if my Aug 1st prediction is close (~ 6.000 Mkm2)


Well projected Chris, but it should be you that is more known for sea ice predictions at least more than Bastardi! And so many of my co-writers here, you are light years ahead of deniers, may pleasant light shine on you to educate an eager audience.

buoy 2014b is just about to float I would say. Incredible amount of water on top and no real visible ice cold layer stratum left. Near by beaufort sea opened amazingly rapidly http://eh2r.blogspot.ca.

Buoy data is very interesting, 2014b has more water on top less ice in bottom, not keeping up with sensor estimates in my opinion, I'd say it has 90 CM of ice from 160 in no time at all , it took a few days only.

Chris Biscan

That Watts blog prediction might out do this blog from last year.

The ice in a bad way currently.

Worse then most realize.

Rob Dekker

With all the speculations about the extreme predictions (high and low) in this SIO report, we may almost forget about the bulk and the median, where most projections hang out.

But either way, let's take this report with a grain of salt.

Remember that this report is based on May data, and the variability of Arctic sea ice between May and September is about 500 k km^2 standard deviation. That means the 95 % confidence interval is +/- 1 million km^2.

And then recall that according to Chris Reynolds brilliant graph :
most predictions don't even obtain less than than 500 k km^2 standard deviation.

That is not to say that we can't make any predictions. It just means it is hard, if you are using data available in May.

The problem is not just the unpredictability of weather (in summer) to come.

The problem is that in order to make a solid prediction for September, based on physical parameters in May, you FIRST need to explain the PHYSICAL parameters of why there is a down trend in the first place before your prediction can be better than just linear extrapolation of the trend.

And that is where global warming and albedo feedback comes in.

We know that we emit CO2 at a rate that increases the concentration in the atmosphere by 2 ppm/year. At the current level (400 ppm) and a generally accepted 3.7 Watt/m^2, this means that each square meter of our planet retains 27 mW/m^2/year. For the entire Northern Hemisphere, that is 6.75 TW added each year (see also the Hiroshima counter on this web site). That's 67.5 TW/decade for CO2 alone.

Now put that against the impact of 1 million km^2 of snow or ice lost. Conservatively, if 100 W/m^2 makes it through the clouds on the edge where snow or ice melts in spring, then the albedo difference may cause half of that to be absorbed by the land or ocean. That is 50 W/m^2, or 50 TW per 1 million km^2.

So a 1 million km^2 snow or ice anomaly causes our planet to warm about the same as as the CO2 forcing added over a decade.

Now, look at the snow cover anomalies from May (which is kind of the average over spring) :

and notice that over the past 2 decades, some 2 million km^2 of snow were lost.

That means that over the past two decades, snow cover alone DOUBLED the GHG effect of CO2 over the Northern Hemisphere in spring.

That's albedo feedback at work.

For predictions between May and September, now that we know that snow cover has such a profound effect on the Northern Hemisphere heat uptake, and if heat (as in forcing) is the cause of the multi-decadal down trend of Arctic Sea Ice, then surely we should be able to see a correlation between spring land snow cover in spring, and Arctic sea ice extent in September.

And we do.

which is the basis for my prediction, but more importantly, I hope that we all understand that the Sept minimum is not just a toss of the dice, or a wild guess that depends only on the weather in summer, or an opinion based on political beliefs.

Our planet and the Arctic comply with the laws of physics. Its our job to find out what matters for Arctic sea ice decline, and what does not.


In my case I really low balled the guess. 3.25-3.5. I have no education per say in it, but here is my reasoning. The best ice needs time to squeeze out air and salt. It also needs cold. The average winter temps in the Arctic are going up, in fact, other them May it was a very warm winter by Arctic standards. Not only that, it is my conclusion that most if not all ice that has stayed around until winter since 2007 has been routinely been broken up into smaller pieces. Although every winter they are frozen back together, salt and air have infiltrated and not allowed to get pressed out because it needs more then 1 winter to do that. So in conclusion, although you may start the melt season with big single sheets of MYI, it IMO can not really be put into the category of MYI, because it is full of air and salt. What you really have is very thick FYI. Combine that with the fact it has been very warm in the winter, you end up with very wet FYI. This means IMO the weather in May really is not as important as it was before 2012 (remeber '12 up until GAC2012 really was not conducive to a big melt up until that storm weather wise), the big factor is weather at the end. Case in point, last year the melt earlier in the year was going beating 2012 then July/August the winds really did nothing other then move it around and fragment it some more, it remained cool and cloudy for the most part. Now the SSTA were on the high side in Aug. This was not long enough to cause the ice to melt out entirely, but I do think it cased farther deterioration to the ice quality.
Putting this all together, IMO Aug weather this year will be the decider as to how much ice we have left. That is a roll of the dice.
By then we could possibly be seeing the effects of El Nino starting which could mean warmer then normal (above new norms) with the polar jet stream pulling south. That brings in the second roll of the dice. The jet stream is far weaker then in the past and therefore far more erratic. This behaviour really causes difficulties in forecasting because it is primarily the jet that defines weather boundaries.
Not really scientific, but my reasoning.


Took a stab at measuring how much ice is left with Buoy 2014b.
http://eh2r.blogspot.ca The key is variation in temperature, as the horizon rises or drops with temperature differences between ice and air.

Rob Dekker

I find it interesting that from all the 28 projections in this SIO June report, which has a median projection of 4.7 million km^2, that the only projection that obtained any significant media exposure is the 5.4 million km^2 projection from the Reading University Schroder et al team,
which appears to be one of the highest predictions in the SIO report.

And there seems to be some chest pounding going on there too :
with remarks like "Professor Feltham says his team's system has been shown in a recent analysis to be of ‘unprecedented skill' compared to all other methods.".

I have nothing bad to say about their methods (using melting ponds as a predictor) but even their own graph in Nature :

shows that by the end of May, melting ponds are really not that significant yet (let alone be a good predictor for things to come).

And thus, one would expect their high 5.4 prediction for Sept extent would be accompanied by a very high uncertainty.
But its not.

The team claims "500 k km^2" standard deviation for their prediction, which seems remarkably low for a May data prediction, especially since according to their own graph June and July are the big 'melting pond' months. Not May.

So I am skeptical of the uncertainty in their widely publicized prediction.

And I'm disappointed that the media singled our one prediction from the 28 in the SIO report.

Jim Hunt

Wayne - 2013B is a "seasonal" ice mass balance buoy, as is 2013C. Here's a paper detailing a variety of ways in which they aren't necessarily measuring what you think they're measuring:


This early in the season I'd be inclined to go along with the bottom sounder estimate of 153 cm ice thickness.

Rob - I'm continually disappointed by the news from the Arctic that the mass media singles out for publication! Regarding your physics/albedo discussion, what used to be called "the pack" isn't packed together the way it used to be any more. What do you suppose the average albedo of this area of sea ice will be over the next couple of months?


Remember there's no such thing as bad publicity. Hopefully, now that they have reported an initial projection, they will revisit the SIPN projections at the end of the melt season. Much better they report it twice than not at all.
At least they didn't report the WUWT projection.


Well, Rob, these are results published in Nature. They deserve some extra credit for that. Their mediatic prediction for 2014, not sure.

To put an example if Hansen and others had published their results in blogs, we *all* probably be talking about AGW as a conspiracy theory fabricated by loons

This will all the respect to this blog and your model in particular that surprised me so much how well it pointed in the right direction last year

Colorado Bob

The Heat is on in Greenland: Support the Dark Snow Project
By: Dr. Jeff Masters , 2:50 PM GMT on June 25, 2014

The heat is on in Greenland, where the high temperature on Tuesday hit an unusually warm 67°F at Kangerlussuaq (Sønder Strømfjord) in southwestern Greenland. It's been a hot June at Kangerlussuaq, where the temperature peaked at 73°F on June 15. That's not far below the all-time hottest temperature ever recorded in Greenland of 78.6°F, set just last year on July 30 at nearby Maniitsoq Mittarfia, as documented at wunderground's extremes page. The unusual warmth this year melted nearly 40% of the Greenland Ice Sheet in mid-June, according to data from the National Snow and Ice Data Center--far above the usual 15% figure. The warm June temperatures could be setting the stage for a big Greenland melt season this summer, and scientists with the Dark Snow Project are on the ice, 48 miles east Kangerlussuaq, conducting a two-month field experiment on the causes and implications of Greenland ice melt.


Bill Fothergill

@ Rob Dekker from yesterday,

Hi Rob,
I'm afraid your maths and/or physics went a little bit walkabout there. It can be all to easy to say something other that what one actually means when discussing power (measured in watts) and energy (measured in joules).

When I was doing this stuff at University a few million years ago, the evil bastards were not content to have us work in just the m.k.s. system. It was almost compulsory to mix in c.g.s. and f.p.s. (imperial). However much one might have hated it at the time, it did mean that one learned to get the units sorted out.

If the Earth System is out of radiative balance - as it is present due to the build up of gases which are active in the Infra Red - then the system will, over time, accumulate energy. (Politicians might accumulate power, but systems accumulate energy.) One of the many manifestations of this energy imbalance will be an eventual rise in temperatures, hence pushing the system back towards radiative balance. This is characterised in the Stefan Boltzman relationship (emitted energy being proportional to the 4th power of the temperature of the emitter) and is the big negative feedback that stops our atmosphere from going the way of Venus.

(NOTE: There is a huge hysteresis effect present in the Earth System, so returning to radiative balance most certainly does not infer a return to the status quo.)

Using your forcing value of 3.7 watts/sq metre, one can perform the following calculation...

(NB I've followed the Excel convention of showing 10 to the power of, say, 4 as E+4)

Approximate area of the planet ~ 5.11 E+14 sq metres

Number of seconds in a year ~ 3.16 E+7 (3600 x 24 x 365.25)

Energy accumulation in one year ~ 6 E+22 joules

In other words, that's 60 billion terajoules added to the planetary pot each year. In slightly different units, that's equal to about 1.7 E+16 kilowatt-hours, or 17 million terawatt-hours.

I hope that clarified, rather than confused.


Buoy 2014B is a drain hole. With T17 going from -1.38 to -1.07 C in 3 hours, confirmed by lesser changes in higher thermistors, clearly fresh water is trickling down very near the thermistor chain, most likely down the side of the buoy. The thermistors appear to be functioning correctly, but most likely are not representative of the floe it is embedded in.

I concur with there being approximately 1m of solid ice left along the thermistor chain, but this does not include water, wet snow or permeable rotten ice, and would not contradict the sounder readings even if they were along exactly the same line, which they're not.


Fairly good idea Blaine, but there seems to be some diurnal thermal variance coherence, which shouldn't happen by top of ice water flowing downwards on the thermistor string, the temperatures should be more in line with surface water. Can't be confirmed without a person there, hopefully by helicopter or back up kayak.

Chris Reynolds


Really, I wouldn't want that kind of publicity.


My CT Area prediction for 2014 is:

3.0 to 3.3 million kmsq CT Area for the 2014 daily minimum.

I was tempted to extend that downwards by 0.1M kmsq giving 2.9 to 3.3. But in the end I decided to avoid such fiddling.

Using a bit of simple maths that transforms that prediction into NSIDC Extent for September, the equivalent NISDC September Extent prediction is:

4.0 to 4.7 million kmsq NSIDC Monthly average sea ice extent for September 2014.

More detail here:


Chris, being known for having made good predictions consistently is part of this science. Somehow those who always flunk are propped up just as much.

It reads as a very competent projection, right or wrong the attempt is based on science, not by words based on a made up fake reputation.

Colorado Bob

Studying Arctic ice from cradle to grave

When spring 2015 approaches, sun spilling the landscape will find a group of scientists adrift at sea on RV Lance – once a top-of-the-line seal hunting boat, now turned research vessel. On board, an international collection of researchers will watch up-close as the arctic wakes, with instruments tuned not only to wildlife but to the most important creature of them all – the sea ice.

“The ice itself has changed. It’s a new system, a different system of ice,” says Harald Steen of the Norwegian Polar Institute to BarentsObserver. “Through the winter we’ll study the ice, and when the sun returns we’ll be there to study the awakening of the Arctic ecosystem.”

Steen is the project leader for the Norwegian Young Sea Ice Cruise (N-ICE2015). Scientists from the Norwegian Polar Institute, University of Bergen, Norwegian Meteorological Institute, Colorado State University and a dozen other research organizations will be on board RV Lance studying Arctic ice dynamics. The research vessel will go to sea late December 2014 and will freeze into the ice north of Nordaustlandet, Svalbard and passively drift with the ice through the winter. The RV Lance was chosen for its ability to withstand the crushing power of the winter ice.

Chuck Simmons

I'm going with the unskilled statistical prediction of 4.0. 2014 has been following the 2012, 2011, and 2007 curves fairly well.

Rob Dekker

Regarding Schroder et al my criticism of their 5.4 prediction, seattlerocks said :

Rob, these are results published in Nature. They deserve some extra credit for that.

You are absolutely right.

I purchased their article from Nature, and read it, and found out that I was confused, incorrect (and incoherent) with my comments about the science in Schroder et al 2014.

So, I would like to withdraw my somewhat chippy comments about their prediction.

In fact, I think they did a great job, and even though I think they are biased high this year, their prediction (of 5.4) should be taken seriously.

I think their method (using only melting pond data in May) is biased too high this year and I expect they will adjust their projection downward in the July report, more in line with other predictions which may go up.

Rob Dekker

Bill Fothergill, I'm sorry that I created some confusion with my choice of words.

Here is the deal. A doubling of CO2 will cause a 3.7 W/m^2 forcing (that is a log 2 curve). At a rate of 2 ppm/year, at the current 400 ppm concentration, that adds 2/400 * 3.7/ln(2) = 27 mW/m^2/year of radiative forcing. That's 67.5 TW/decade added forcing for CO2 for the Northern Hemisphere alone.

A 1 million km^2 loss of snow causes some 50 TW added forcing (and that may very well be an underestimate). And the Northern Hemisphere lost some 2 million km^2 in spring over the past two decades.

So the radiative FORCING from loss of snow in spring in the Northern Hemisphere over the past two decades is in the same range as the increase in CO2 GHGs over the same period.

So if we assume that Arctic sea ice decline is caused by CO2 GHG emissions, then we should be able to see the signature of spring snow cover reflected in the Arctic sea ice decline.

And we do.

There is a strong correlation between April and May snow cover and September ice extent minimum, enough to make a skilled prediction under ARCUS SIO, which for example explains much of the difference between 2012 and 2013 :

The point is that snow cover variability tells us how sensitive the Arctic (Sept sea ice minimum) is to radiative forcing, and the albedo feedback during spring.

Bill Fothergill

Hi Rob D

I think see what you're getting at now, but I still think there's a little mistake in your calculations.

You give the following formula...
2/400 * 3.7/ln(2) = 27 mW/m^2/year
to derive the Delta in radiative forcing.

Just to make sure I'm not misreading your numerators and denominators, I think you're saying that ...

Delta RF = (2ppm/400ppm) * (3.7/ln(2)) = 27 milliwatts/sq metre

and that this is happening each year.

However, that means that if the rate of CO2 increment was instead 4 ppm, the Delta RF would be exactly doubled. That is, of course, not the case.

To get the Delta RF (over any time period), I think you should be using the following...

Delta RF = 5.35 * ln{(New CO2 conc)/(Original CO2 conc)}

For example, if one starts with 400ppm as the baseline, the increments 400 to 402, 402 to 404, 404 to 406, etc., each give a slightly lower Delta in the Radiative Forcing.

Putting in the numbers...

400ppm to 402ppm gives a Delta RF of 26.68 milliwatts/sq m

402ppm to 404ppm gives a Delta RF of 26.55 milliwatts/sq m

Putting these two together...

400ppm to 404ppm gives a Delta RF of 53.23 milliwatts/sq m

If one doubles the CO2 conc, say from 400 to 800 ppm, the formula reduces to...

Delta RF = 5.35 * ln(2) watts/sq m = 3.708 watts/sq m

With this formula, any doubling (eg 280ppm to 560ppm, or 400ppm to 800ppm) would produce the same Delta in radiative forcing. (To be perfectly honest, I'm actually more than a bit sceptical about that being much more than a reasonable approximation!)

If memory serves (fat chance) I think the baseline figure used in the IPCC AR4 was the pre-industrial value of 280 or 275 or something like that.

However, setting aside the nit-picking, I wholeheartedly agree that additional RF due to factors such as albedo change due to loss of snow/ice is a major contributor to so called Arctic amplification.

Does that make sense, or am I, as my wife usually describes it, talking bollocks?


@ Rob & Bill: a few points I ponder about in your formula. 1) our current temp forcing levels are based according to CO2 levels 30 years ago based on what I have read. 2) the albedo impacts I presume are more immediate. 3) is rising CO2 levels = rising temps or could it possibly be geometric or exponential 3) could we not have a CO2 burst when ocean CO2 levels reach saturation and regurgitate CO2? Do we have any idea when that could occur?
This means we are at 347.55ppm affects now not the current 401.85ppm. That comes in 2044. Scary thought that is.

Hans Gunnstaddar

"This means we are at 347.55ppm affects now not the current 401.85ppm. That comes in 2044. Scary thought that is."

At an average annual increase of 2ppm CO2, in the year 2044, (30 years hence) the total will be approx. 462 ppm, and we'll need to wait until 2074 to find out the resulting effects of that thermal inertia. Quite scary & very toasty.

Chris Reynolds

Error in message of 26 June.

Actually the error is in my spreadsheet. Steven picked it up over at the forum, thanks go to him.

I'd not changed an equation in three cells to my transformation of the CT Area prediction in NSIDC September extent was wrong.

The prediction of September extent should have been 4.6 to 5.0 million kmsq for September 2014 sea ice extent. This is above the range of my June SIPN submission. It would include the years 2008, 2010, and 2011.

I'm withdrawing that extent prediction until I see June's full area/extent and volume data in early July.


Now we'll see where rubber hits the pavement. I expected sea ice next to CA Archipelago to be tough to melt. Therefore the "cliff" effect would be happening more like 2007. By all present indications, it will be as estimated, a massive melt is in progress.
The only thing I didn't foresee is Kara sea being slow to melt, but it is much similar to 2007.

The buoy data offers certain glimpses of the physics of the melting process. A cloudy day appears to transfer a whole lot less heat downwards the ice column. While the melting continues in the bottom. Also not really a surprise, a cloudy day has a temperature diurnal effect on the surface. This was observed optically hundreds of times. It is quite possible that these days of high sun (where cloudless), a lot of thermal energy is dedicated to warming sea ice columns, rather than the air, "in summer sea ice is a heat sink" as written by O-buoy site. So the appearance of a moderate melt at present is very misleading.

One must not forget the players from the South, the 2 biggest ones: the super hot North Pacific temperature anomaly twinned with burgeoning El-Nino (sometimes stalling), create more cloud seeds but also extra upper air heat, making the creation of clouds less favourable.

Bill Fothergill


I'll try to address your points in sequence...

Point 1) Reference date for Radiative Forcing values.

It doesn't really matter which start point is used, as long as everyone using the numbers is singing from the same hymn sheet. As I said, the IPCC tend to use a pre industrial start time. Take the following example from AR5...


If one scrolls down to Box 1.1, Figure 1, the RF has a zero value just before 1800, and has risen to the current values mentioned by Rob. (I was trying to post a copy of the relevant diagram, but am too stupid to do so.)

Point 2) Albedo response

I'm somewhat nervous about attempting to speak for someone else, but I'm pretty sure Rob and I completely agree that reduction in albedo will have far greater immediate significance than the "marginal" increase in RF. (Although the delta RF will grow, and grow, and grow....)

Point 3) Temperatures and CO2 levels.

As Rob correctly observed, CO2 concentration levels and Radiative Forcing are considered to be related via a logarithmic function. The contentious issue is then of the Climate Sensitivity to the relevant forcing. Current thinking seems to imply a relatively linear relationship between the cumulative forcing and temperature rise.

As a thicko engineer, I don't know enough about Climate Sensitivity calculations to go out on a limb. (I don't like recommending Wikipedia articles on principle, but you might want to look at their offering on Climate Sensitivity as a starting point for further reading.)

The second Point 3) CO2 Release

Yep. The oceans will eventually start to outgas CO2 as the solubility decreases with rising temperature. (I've never read anything about how that will play out as regards the acidification problem. Has anyone seen such an article?) However, it's not only the oceans that need to be considered. There is a concern that we might reach a turning point as regards land sinks, and that some of these could end up as sources instead. Again, that's not my bailiwick, so I don't have any strong opinions.

On top of that, there are, of course, the various clathrates in the deep ocean and the permafrost (tempafrost?) to consider.

Thank God I'm an old bastard.

Chris Reynolds


I think you need to define what you mean by the 'cliff effect'.

CT Area anomalies show a 'cliff' in 2007 and 2010 to 2013.

However the anomalies are relative to the average seasonal cycle, and now the average seasonal cycle losses are increasing, making the formation of an anomaly cliff all the harder. I don't think 2014 will see a cliff.

Bill Fothergill


In an earlier post, I quoted numbers down to hundredths of a milliwatt/sq metre. The purpose for that level of extremely spurious accuracy was simply to illustrate the gradual drop off in incremental Radiative Forcing due to the logarithmic relationship between RF and the concentration of IR active gases.

I am not for one moment suggesting that the numbers are that precise in the real world.

Sorry for not making that clear from the start.


@Bill I know you understand about same song sheet, I was just pointing it because I tend to google queries and quite often it appears they are saying current CO2=current temps, and rarely do I see stated that in fact it is (current yr - 30)get CO2 level=current temp, and that RF can either be immediately impacting or a delay (such as desertification has both immediate and delay amplification).
I do know that most of us understand this I just wanted to point it out as I see so little of this explained. You get that in political discussions, they definitely do not understand the song sheet most of us are sing from.


Each year has its own cliff Chris

2014 for instance has a split surface sea ice climate given structure. One sector, Pole to Russia which has had less cooling due to cloudy winter spring(unlike 2013). The other sector Pole to North America had more cooling due to Continental winter so frosty American deniers did many cart wheels on their way to their idling driveway Hummers. The two have incompatible climate conditions, one side now cloudy, Pole to Russia, the other sunny Pole to NA. The heat sink effect by sun from clear air takes time to weaken sea ice, from what I learned recently from buoys, sea ice column temperatures reaches a point where ice becomes tragically weak, temperatures in the columns on most buoys are now all close to disintegration . While the other side of Pole, thinner from cloudy winter sea ice is protected by a vanishing cloud shield, this thinner ice still melts though. The net effect is a later cliff in sea ice area anomaly.

I captured some very interesting cloudy vs sunny effect on sea ice http://eh2r.blogspot.ca/2014/06/anatomy-of-doomed-sea-ice.html . There is no doubt that sun melts more ice at this time of the year.


@wayne Could it be that sunny at this time of year works really well because other melting forces have weakened ice conditions in such a way that sun can do optimum damage?

Gerhard Trausner

My tip
4.7 mill. km ²


Once the extent reaches 7.750 MKm2 (about mid July) the ice is going to disappear as if you placed it in a hot kettle.
My prediction is:
Not in lowest 3 1%
Not in lowest 2 5%
Not lowest 30%
Less than 3 M Km2 30%.
In each of the past 3 years we have seen at least 2 M of average melt, that is sufficient to place 2014 adjacent to 2007.
An extra 10 cm places us below 2012. 2012 saw an extra 35 cm.
Breathe deeply until mid July, because after that it is shock and awe.


nice to see a confident prediction, but I'll believe it when I see it :-/


LRC, sunny and mainly clear is always devastating at this time of summer. The only thing missing is strong compaction, which makes things even worse. But the thinner ice of North Pole to Russia sector is the most vulnerable being thinnest. What I look for is old leads with even thinner ice, and strong buoy displacements. There should be a moderate Cyclone hitting Beaufort sea area soon, lets see what kind of structural damage the ice in that area went through lately…

Bill Fothergill

@ DavidR

Remembering all the fun I used to have ensuring that logic circuits utilizing the NOT function actually did what they were meant to do, can I rephrase what I think you just said?

Using your figures, but doing a quick mental Venn diagram to remove the naughty NOTs, gives the following probabilities for where the 2014 extent will end up...

Probability of ending in lowest position = 70%

Probability of being 2nd lowest = 25%

Probability of being 3rd lowest = 4%

Probability of any other result = 1%

Is that what you meant to say? If so, it's certainly going out on a bit of a limb!

As regards mid July, I know Chris Biscan has always thought that's really the make or break time. I also did a little post last year talking about the weakness of any correlation prior to about that time. (By which time, of course, we are almost governed by autocorrelation.)

Chris Reynolds


I disagree.

The CT Area June anomaly cliff is a feature of 2007, 2011, 2012 and 2013 that has led to lower summer CT Area (even in 2013 when June drops lowered end of June area substantially.

NSIDC Extent anomalies show strong drops throughout June and early July in many recent years, these play a role in recent record lows.

In both cases these drops in anomaly are definitely not part of the long term behaviour but are clearly a new behaviour (driven by increased open water formation efficiency with thinner ice - IMO). I can post graphs if needed.


Since posting my CT Area prediction above (and the corrected NSIDC Extent prediction based on that), I've had nagging doubts about it. Given the April/May ice state from PIOMAS, and the behaviour since 18 June I think that method will be wrong this year. Now on top of >100k average daily losses since 18 June, we have a 250k loss in NSIDC Extent for the most recent data.

The following graphic shows the years from 2007 to 2013 relative to 2014.
Put simply each year's trace is that year's extent minus the extent for 2014.

Note that 2014 is rapidly closing the gap on 2010. However 2010 is not at issue with regards the minimum. In 2010 a large export of MYI slowed losses in the remainder of the melt season. 2014 could easily challenge 2011 and 2007, I think there's a good chance of a second place behind 2012.


@Bill Fothergill,
Yes that's my prediction based on the April PIOMAS volume and NSIDC extent averages. According to PIOMAS there was a massive amount of ice that was in the average thickness range of 1.6 - 2m at the start of the season. All of this will melt out this season and maybe a lot more.
I expect that to start showing up in the extent loss figures when extent drops to 7.750M Km2.
Currently we appear to have lost the first 1.25m of the thickness distribution.

Bill Fothergill

@ David,

Thanks for the clarification. It appears that some whole scale losses might just be starting to appear already. JAXA is now showing 2014 virtually identical to 2012 and 2011.

Let's see what happens around mid July - or around 7.75M km2

Cheers Bill F

Rob Dekker

Neven, sorry for posting in the wrong thread yesterday. Here is where this info should go.

A quick summary :
My submission ("Dekker" in the list) to SIO June report uses Northern Hemisphere snow cover in spring, as a predictor for how much ice will melt between now and September.

Using May data (for the June report) I ended up with 4.6 million km^2, which was just below the median.

The nice thing about using snow cover as a variable is that the standard deviation of the prediction is lower than most other entries.

Now, June numbers are in (for Rutgers' snow cover, as well as NSIDC area and extent).

Snow cover came in quite high at 3.6 million km^2, which is higher than it has been since 2009:

The formula that gives the best correlation between ice loss (area in June to extent in September) includes June, May, April and March snow cover, as well as a factor (June_extent - June_area) which I think represents the amount of melting ponds and polynia during June.

If I plug in the (snow cover, ice extent and ice area) numbers for the past 18 years, then this is the resulting prediction (using June data) versus the actual Sept extent :

June 14 Sea Ice forecast photo June-14_zps7336859b.jpg

You may have to click the graph to see the whole thing.

Also the prediction (of 4.7 million km^2) for this year is included in this graph.

And I will submit that prediction to SIPN for the July report.

The really important part about this method (of using mostly snow cover in spring as a predictor)is the standard deviation (319 k km^2), which is way better than the standard deviation on a linear trend (550 k km^2), and as far as I can see the smallest standard deviation of any of the methods presented in the SIO report.

To put it in simple words, 4.7 with a SD of 0.319, mean that there is a virtually no chance that 2014 will turn out to be breaking the 2012 record, and more importantly, there is only a 2.5 % chance that 2014 will go above 2013's 5.3 million km^2.

I'm real curious which July report SIO predictions will be above 5.3 or below 4.1, since these appear to have a probability of realization of only 2.5 % each.


Thanks, Rob. Very interesting, as usual.

Rob Dekker

The significance of the accuracy (sigma 319 k km^2) of using snow cover in spring as a predictor for Sept ice extent is that maybe summer weather is not as important as we thought it would be, or, even more interesting, that maybe summer weather may be biased by the amount of energy that spring snow cover inserted into the Northern Hemisphere atmospheric system.

Either way, you guys can totally kill me if Sept 2014 shows ice extent below 4.1 or above 5.3.


Rob, You can be taken to task now :) . Your forecast is a model based stance which does not include dynamical meteorology. As was with 2013 , this may work most times but not always. As we know, the Canadian Arctic Archipelago has more snow now than usual . Which is more part of a larger well known presence of High pressure systems being more persistent since the Spring. These Highs gave more insolation elsewhere likely contributing to current lows in extent at IARC-JAXA, #3 lowest at this time. The idea that one may take one indicator and make a projection out of it lacks the holistic approach which has likely a far better chance of success. I rather use what indicators may provide and integrate it with larger influences. So I perfectly new that sea ice in the High Arctic Archipelago would last strongest since March because of more than one indictor combining synergistically. A cold zone in the CAA in spring invites the presence of Highs as long as thicker ice and larger snow extent, it didn't mean what it implied, once knowing the greater impact of more sunshine.


Lets not get too extreme here, my prediction is based on the thickness distribution in April and it puts the likelihood of an extent above 4.3 at about 1%, with the likelihood of a record at about 70%. Interesting to see who is closer but at this stage I think the weather is on my side.

Artful Dodger

Hi Rob. That's a valuable correlation you've uncovered between Spring on-shore snow extent and September sea ice extent. It appears to be part of the larger Arctic-wide albedo flip phenomenon which is driving loss of MYI and indeed Arctic amplification itself.

We need to remember that Summer weather is not just an independent variable in the Arctic equation, but also a feed-back effect due in-part to the loss of highly reflective, high-albedo surface area.

Coupled with other positive feedbacks (ie: methane release from thawing permafrost), these large-scale changes to the Arctic surface heat budget can CAUSE the weather that we treat as a random variable.

When playing dice, the trick is to know when the dice are loaded. That's all we need to decide.


Artful Dodger

Oh, and Happy 4th of July to all my American cousins! ;^)

L. Hamilton

So, should we do an ASIB poll as a contribution to the SEARCH Sea Ice Outlook this month? Or leave well enough alone?

Rob Dekker

lodger, nice to see your post. That was exactly what I was thinking.

Regarding snow cover and Arctic amplification, here is a great post from Tamino that puts snow cover loss in perspective of AGW forcing

In this case the net change is about 1150 TW. If spread over the entire surface of the earth, and if the difference in TOA albedo between snow/ice-covered and uncovered regions is 0.2, this accounts for a total climate forcing of about 0.45 W/m^2.


Considering that AGW forcing increased about 0.93 W/m^2 over the same period, the albedo effect is significant.

But here is the kicker : most of that heat is generated during the spring/summer months, and virtually nothing during the fall/winter. So spring/summer forcing is probably 2x what he reported.

And since over a few months heat does not travel all around the globe, but likely stays within the Northern Hemisphere, we may have another 2x factor on our hands.

So spring/summer forcing due to albedo effect may be 4x his annualized, globalized number, or in the range of 4*0.45= 1.8 W/m^2.

Put that against AGW forcing over the same period (since 1980) of about 0.93 W/m^2 and you see that albedo amplification in spring/summer due to snow/ice loss quantified as being about double the CO2 GHG effect.

These numbers are no small potatoes.

If the GCM don't get snow cover right, then they WILL underestimate the RATE at which Arctic sea ice decline disappears, easily by a factor of 2.


I put a thread up on the forum to discuss SIPN predictions, which close for July on the 9th. Just before the poll closes here.


My prediction based on Chris Reynolds's PIOMAS data remains at 3.2M Km2, with 90% confidence that the final extent will be a record. Anything above 2007 seems out of the question.

Artful Dodger

Hi Rob,

An R of 0.94 (R-Squared of 0.8836) is impressive: this simple model shows that 94% of the variability in September SIE is explained by May snow cover.

So then, what are the correlations with all other months? In particular, I'd be interested in Feb snow cover, as this may show a high negative correlation to Sep SIE. The high albedo of snow also means good insulation of the ground during the coldest months. More snow on land, less freeze of permafrost and more water to runoff in Spring.

Apologies if you've mentioned this previously, do have a link to the numerical data for NH snow extent? I'd like to muck about with the data a bit. :^)

Land change is an important topic for the Arctic. Have you seen this paper, included in IPCC4, WG1, chp 4 :^)

Lemke, P., J. Ren, R.B. Alley, I. Allison, J. Carrasco, G. Flato, Y. Fujii, G. Kaser, P. Mote, R.H. Thomas and T. Zhang, 2007: Observations: Changes in Snow, Ice and Frozen Ground. In: Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change [Solomon, S., D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M. Tignor and H.L. Miller (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.

Lemke figure 4-2-l

How about including Arctic river outflow? I'd think there would be some auto-correlation with snow extent, since the underlying variable (heat) is captured by melt rate. Maybe there's a way to tease out heat flux?

arctic river basins

This map of Arctic outflowing river basins shows starkly how important Greenland will be to the future heat budget of the Arctic. 2012 was MOST notable for the widespread (and ominous) melt of the Greenland ice cap (a first since at least 1898), whereas SIE loss was more of the same...


Rob Dekker

lodger, thanks.
For starters, this correlation is not just with May snow cover. It includes March, April, May and June snow cover, as well as a factor of (June-extent minus June-area) which I believe represents melting ponds and polynia in June. I did not find any correlation between snow cover further back (such as Feb) and Sept ice extent. Correlation starts in March, and grows until June.

Allow me a day or two to write down the details (which I will submit to SIPN for the July report), since I only have a few minutes right now.

And thank you for the link to Lemke et al. I'll read it.

Verify your Comment

Previewing your Comment

This is only a preview. Your comment has not yet been posted.

Your comment could not be posted. Error type:
Your comment has been posted. Post another comment

The letters and numbers you entered did not match the image. Please try again.

As a final step before posting your comment, enter the letters and numbers you see in the image below. This prevents automated programs from posting comments.

Having trouble reading this image? View an alternate.


Post a comment