« Oh, and BTW, the Passages are open | Main | IJIS SIE: 5 million km2 mark passed »


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

Seke Rob

MASIE had a meager net change... an assortment with the Arctic CB loosing a bit, the Greenland Sea gaining more. Here's the complete list, instead of a chart. Probably easier to poor over (all KmSq, space a thousand separators, bracket value change for day):

"Northern Hem. Tot. 4 872 569 (-16 889)"
"1) Beaufort Sea 606 425 (+4 181)"
"2) Chukchi Sea 184 045 (-10 034)"
"3) East Siberian Sea 532 439 (-13 565)"
"4) Laptev Sea 35 815 (-5 725)"
"5) Kara Sea 37 157 (-2 389)"
"6) Barents Sea 7 276 ( )"
"7) Greenland Sea 286 119 (+9 187)"
"8) Baffin Bay G.o.St Lawrence: 7 749 (-1 279)"
"9) Canadian Archipelago: 166 820 (+7 389)"
"10) Hudson Bay 22 434 (+1 701)"
"11) Central Arctic 2 985 183 (-6 354)"

Lost a little ground on JAXA, but with Paul K's observations in mind, it will come round again.

Seke Rob

Re: Christoffer Ladstein | August 31, 2011 at 17:22

Who cares for crime-fiction, when we have daily access to Nevens-realtime-"fiction":-).

... realtime "status-faction", guaranteed :D

Michael Fliss

Welcome to the Ice Museum

Record ice loss...a bit like being the winner in a nuclear war.

With apologies to Joni Mitchell and Big Yellow Taxi:

...They paved paradise
And put up a parking lot

They took all the ice
Put'em in an ice museum
And they charged the people
A dollar and a half just to see 'em

A Rambler

I still think that its better than 50/50 that the ice extent falls below the 2007 minimum -- Paul Klemencic

Then I suggest you get on Intrade. Last trade: 85%.

Minimum Arctic SIE for 2011 > 2007


UB prelim map is out and all I can say it an other 100K drop is coming tonight maybe more.

L. Hamilton

A new graphic showing the 1-day minimum extent, 1972 through 8/29/2011, based on Uni Bremen data (courtesy of Dr. Georg Heygster). I will update this when possible over the next few weeks.

Details regarding construction and validation of the Uni Bremen time series appear in a technical report by Milke and Heygster (2009):


Ned Ward

I still think that its better than 50/50 that the ice extent falls below the 2007 minimum

Does that mean a better than 50/50 chance that your "adjusted" 2011 IJIS extent falls below the "unadjusted" 2007 sea ice extent?

As I understand it, you are mentally subtracting some amount from the IJIS data. You suggest that that amount is "converging" on zero. What happens if your adjustment doesn't reach zero, or if it starts to increase again?

Do you subtract some amount of ice extent from the 2007 minimum? If so, how much?

Christoffer Ladstein

There's no need to get "Icy" in here...

In a democracy, let the "hard facts" speak for them selves. So in the meantime we just have to salute those providing more accurate measuringtools!

Christoffer Ladstein

Bravo, L. Hamilton, your graphs are as smooth and understandable as a novel by A. C. Clarke!
And now I have another reason to improve my german skills....

Ned Ward

That 19:07 may have come across as a bit hostile, when I really just want to know how Paul intends to compare the 2011 IJIS minimum to the 2007 IJIS minimum. Will he be subtracting some amount from both years, one year, or neither year?


@Frivolousz21 | August 31, 2011 at 18:51
UB prelim map is out and all I can say it an other 100K drop is coming tonight maybe more.
I don't see it - it looks like a bit of melt/dissipation on the fringes of the Beaufort and Ckukchi Sea, but the East Siberian Sea is "grayed out".

The rest of the Arctic seems like minor melt at the margins, with somewhat more open water showing up in Greenland Sea. It doesn't seem like 100,000 km^2 worth of Arctic Ocean being exposed.

For comparison, Banks Island, the large Canadian island on the southern side of the westernmost part of the Northwest Passage, is 70,028 km^2.

On the other hand, TOPAZ4 suggests that much of the ice in the East Siberian Sea is rather thin (large areas 17, 34, 51, 68, and 85 cm thick), so I expect a big part of that protrusion to melt away in the next 3 weeks - but probably not tonight. (And TOPAZ4 shows that protrusion in the Beaufort Sea is mainly about 2 to 3 meters thick ice - so most of it will survive the entire melt season)

L. Hamilton

Thanks, Christopher. Clarity is indeed my goal, in writing as well as in graphs. After freeze-up I'll post some different images, including a comparison of IJIS, NSIDC and UB time series; and a look at the seasonality of global ice. But for the moment those simple bar charts seem to capture the spirit of watching ice melt.

Regarding Paul K's writing, I for one read that with great interest. He goes out on a few limbs, but his posts reflect detailed study of different data sources, and serious effort to understand how they fit together. I hope that he will keep on doing this work, and sharing his insights for discussion.



I can't help you there. I hope you realize 1 line of Longitude is 60NM which is 112KM.

if you go into photoshop and overlay the graphs you can see large areas of 40-50km recede and then in the Beaufort areas of thin ice dissapeared.

On the East side ice receded slightly and ice in the Fram opened holes back up. And also receded.

We know ice in the ESB based on wind profiles will also go down.

I am sorry you can not see this. This is one of the larger drops. And looks quite a bit larger then yesterdays.


"So in the meantime we just have to salute those providing more accurate measuringtools!"

I've been saluting the CryoSat-2 team since 8 April 2010 -

but they've been pretty much ignoring me.


Friv - I think you meant Latitude

'I can't help you there. I hope you realize 1 line of Longitude is 60NM which is 112KM.'



your right, thank you. Regardless it shows how much weakening of the ice there was there and recede. Not surprising with record warm waters being pumped into super thin ice.


Regarding Paul K's writing, I for one read that with great interest. He goes out on a few limbs, but his posts reflect detailed study of different data sources, and serious effort to understand how they fit together. I hope that he will keep on doing this work, and sharing his insights for discussion.

I'll second that!

It's true that a large grey swath is hiding much of the edge of the ice pack, but in the protrusion in the Beaufort region you can see quite a bit of the ice go poof! That hole in the Laptev Bite is also getting bigger.


BTW, Larry, I of course stole your graph and found a place for it.

Kevin McKinney

Late to the party, catching that revision. Sorry, Neven. . .

But I did manage to hang onto my hat.


I'm dying to see if this keeps up, because if it does it could have a lot to say about the 'quite suddenly melting away' idea, even if it doesn't quite ALL melt away this year. There may even be some implications for curve fitters! ;-)


Yes, Kevin. SIE is dropping pretty hard at the moment, but we're not really experiencing 2007 weather conditions. The big question is of course: how long will it keep up? ECMWF weather forecasts (and AO Index forecast) are still not showing a sea level pressure distribution that compacts the ice like we saw in 2007 or last year even, leading up to the 10th of September. Which, I learned last year, is more or less the cut-off date.

Paul K, I had to think of you when I read this snippet by NSIDC's Julienne Stroeve over at the blog which shall not be named:

The 4.66 million sq-km value of today is the actual value, not averaged over 5-days which is what we show in our time-series plot that Anthony references in this blog.

Don't know if that's useful.

Ned Ward

Neven, just keep in mind that that "actual value" has a lot of noise in it. That's why they do the five-day averaging.

Lord Soth

Forgive me for being blind, but I can't seem to find actual daily numbers for NSIDC sea ice extent, 5 day averaged or raw, on their site.

Can somebody post the url for this info ?


@ Frivolousz21 and the foray over to WTF

If anyone here has been on WTF, they may have noticed the thread pertaining to the final Arctic forecast poll. For those unaware of what this means, the Arctic Research Consortium of the United States (ARCUS) has put together a poll from an eclectic variety of sources, including the people over at WTF. The respondents are all invited to submit their estimate (guess?) for the mean September Arctic SIE. For the purposes of said poll, and to introduce a de facto standard into the proceedings, the figures to be used will be those produced by the NSIDC.

Chris has mentioned it on this thread, and has pointed out that some of Mr Watts' fans are sticking with a September mean SIE that has already been undertaken by the 30th August Jaxa daily figure.

Although this might seem absurd at first glance, I feel that, in the interests of openness, I should deliberately play Devil's Advocate. (Actually, I don't think one can do that by accident!)

Looking at the Jaxa data, in 22.22 % of the occasions, the September average has ended up above the corresponding daily figure for the last day of August. (OK, that was spurious accuracy, it only happened twice - in 2002 & 2004.)

Also, although the Jaxa September average was above its NSIDC equivalent in each of the last 4 years, for the 5 years between 2002 and 2006, the NSIDC figure was the higher of the two.

Switching back to reality, although the Jaxa September means for 2002 & 2004 were higher than their 31st August equivalents, the difference was 50k sq km and 30k sq km respectively.

Regarding the 5 years (2002 - 2006) during which NSIDC produced a higher September mean SIE than Jaxa, the average difference was 46k sq km. Over the most recent 4 years, this "trend" has reversed, and the Jaxa September mean SIE has averaged about 110k sq km higher than its NSIDC equivalent.

In other words Chris, it's possible they could be right, but words like "straws" and "clutching" rather come to mind.


@ Lord Soth 21:19 31st Aug

Hi there LS,

Just seen an exchange between Mr Watts and Julienne Stroeve over on WTF on this very topic.

I'm afraid there was no real answer forthcoming.


Another Polar Animation here :


And by the way I still believe in the over-nite scenery, one morning you wake up and it is not there, I hope it will be in the end of season the year it happens otherwise we will get bored, and frustrated!

Regards Espen


Please, in the interest of civility, remember that that other blog is called Watt's Up With That, or WUWT.




Yes, either not name it, or name it as it is called. :-)

Kevin McKinney

"Yes, either not name it, or name it as it is called."

So "Voldemort" would be right out, then?


If I had to put my current line of thought on the role of these factors by percentage for different time periods here goes:

natural Variability
GHG's(including Co2)
Temperatures, including SST's


Natural variability(40%)

Natural variability(10%)

Since 2007 the role of AGW has likely went from 30-40% of the ice decline from 1990-2006. And Temperature feedback has gone up exponentially.

If this process can be shutdown by something else for a while...I can definitely see the ice stopping and gaining volume for a while. Until AGW catches up wit the ever increasing rise of GHG's.

On the flip side if this does not get impeded. We are about to see the ice dramatically drop to possibly near ice free conditions in less then 5 years.

The data on this is clear. The SSTs are crushing the ice from the sides and from below. Not AGW currently, even though natural variability and AGW were triggers for the SSTs to get out of control and decimate the ice.

it is likely since the early to mid 2000s that Ice volume was crippled.

If this continues, we will see the bottom drop out so to speak.

As in the ice as we can see in the laptev will melt out from below, not from the top or sides.

This is a complete admission that i estimated AGW's role in this as a much larger factor then it was and has been.

However SSTs are much more dangerous and deadly for ice in the short term.


Foxe Basin:

I have to give the sea ice in Foxe Basin praise for its stubbornness, but soon it is over.

Regards Espen

Kevin McKinney

Frivolous, how do you differentiate between "GHGs" and "temperature" for definitional purposes in your attribution? You seem to take the differentiation of the two for granted, but I have absolutely no idea how and why you are making it.

Can you clarify?


Temperature is mostly SST based which in the arctic is mostly from the absence of ice. Which might be started by GHGs and NV, but they did not warm the water that is killing the ice now necessarily directly.

We can blame the AGW for this but on a micro level I believe many factors come into play.

If that is true. A winter of massive ice growth say 16 mil km2 or higher then a summer with great cold conditions could slow the loss and reverse volume for a few years at best.

Kevin McKinney

So, "temperatures" mostly means SSTs--and I presume that "GHGs" then means mostly direct radiative forcing?

What is included under 'natural variability?' I mean, SSTs are subject to natural variability, too, as could be radiative forcings--but perhaps you're using the NV category to refer to basically anything 'unallocated' or not attributed? Kind of a catch-all?

Am I close to what you're thinking here?

Michael Stefan

Somewhat OT, but according to http://moe.met.fsu.edu/snow/ total NH snow cover (including land, sea ice, and Greenland) has set a new daily record low, surpassing the previous record set last year and prior to that, 2007; about 25% below the 1995-2009 average.

Also, statements to the effect of "GHGs account for 10% of current ice melt" ignore the fact that GHGs started the warming in the first place (ignoring natural variability, which has no significant trend and was on a long-term cooling trend until recently), even if feedbacks are becoming more significant, such as increased SSTs in ice-free areas.


Re Neven's "Besides, SIA usually follows SIE and the latter almost lost a century today."

How does that work? My reaction was, surely once an area is no longer counted for extent then it is less than 15% concentration and any further fall will not count as a future fall in area as it already isn't counted for area. If an area reduces it's concentration from 40% to 20% then area falls and extent will follow when it goes below 15%.

It is not that straightforward though. It is possible that 40% to 20% and 20% to 10% are slow but once 20% to 10% has occured then there are adjacent areas that rapidly fall from say 80% to 50%. This would give the impression of area following extent though it relates to different areas.

I haven't looked or tried any test so you may well be correct. Just wondered if a test of this had been done and how you viewed how it worked.

Steve Bloom

Bill, WUWT is frequently referred to much less politely, as it is a place where fantasy is substituted for science. If you want to talk about incivility, that's where it starts.

Steve Bloom

LS, rather than thrashing about for days and days as a certain ME PhD did here recently, write them and ask. They're very responsive.

Paul Klemencic

Wow, Neven you have a great site here. It took me over 40 minutes to read through the comments just in in the last five hours. Lots of info and lots of good insight. Where to begin?

First, Frivolousz21, I agree. This is a classic case of a positive feedback system, that initially was triggered by AGW, but now the ice melt is out of control!

And before anybody hammers me on that, I am using the terminology used by the Statistical Process Control (SPC) guys who use this technique in industry. When we see a system, and we see these types of systemic problems a lot, where a special cause (meaning the system or measurement system has changed), that in turn has resulted in a series of statistically unlikely outcomes (we use 3 sigma versus the mere 2 sigma used by most scientists), then we say "the system is out of control".

Looking at the NSIDC curve anyone can see the Arctic ice melt is out of control. One of the control tests used by SPC types, is to look for a sequence of measurements that are over 2 sigma out. One data point 3 sigma out, is only a 3 in a thousand event in a normal distribution system. Three out of four points in a row, 2 sigma out, is also a 3 sigma event (only 3 in a thousand). Or how about 4 out of 5 that are one sigma out- again a 3 sigma event. Even just nine a row that fall below the mean... 3 sigma event! (For anyone who cares, these are the old, antiquated, but famous,and accurate... "Western Electric Rules")

For Arctic ice melt, every data point in the summer is not only 1 sigma out, but 2 sigma out, and some seem to be 3 sigma out! The system fails all four Western Electric tests.

The Arctic ice system is completely "Out of Control!". That means anyone who comes along and tries to use normal distribution statistics on this system, is making a naive and ill-informed mistake! If the system continues to produce 3-sigma events, then why continue to use normal distribution statistics?

Statisticians who do this, are making a Type 1 error, thinking that the out of control data is still systemic variation ("the Arctic ice is recovering") and ignoring the data which point to systemic changes that is causing the results data (R-data) being observed. Anyone who has worked with out-of-control systems will tell you that you need to drop the R-data analysis, and go looking for P-data (system process data). Collect information on ice morbidity, sea water temperatures, changes in weather patterns, and most importantly, the observed heat transfer rates that is causing bottom ice melt.

In the static ice pack, the bottom ice melt heat transfer rate is about 50-100 watts per sq meter resulting in 1-2 cm per day loss of ice. One of the key reasons for this low heat transfer rate is the low temperature differential, likely less than 0.5 deg C due to the cooling effect of the melting ice on the relatively static layer of water under the static pack. The other reason is relatively low heat transfer coefficient, likely between 100-300 watts per sq meter per deg C.

When the pack can move, things jump in a hurry. Put the pack over a warm sea like the Kara this year, the temperature differential can rise to 2.0 deg C (initially the water is 4-6 deg C, but it cools off due to ice melt), and heat transfer rates hit 200-400 watts per sq meter, even assuming the same heat transfer coefficient.

But now break up the ice into floes, and put them out onto a well mixed sea of 2 deg C, and the heat transfer coefficient could jump to 1000 watts per sq meter per deg C; and the heat transfer rate hits 2000 watts per square meter, about 20-40 times the static pack heat transfer. Since this heat transfer rate is so high, I believed this would cool off the water, dropping the temperature relatively rapidly back to a temperature difference below 1 deg C.

But as we saw in the 'flash melt' event this August, heat transfer rates over 2000 watts per sq meter are possible, and can melt out 30-60 cm of ice in only twelve hours!

The ability of the Arctic waters to absorb enough solar radiation in the summer, to contain enough thermal energy to keep the temperature 2-5 deg C in the surrounding seas, and even 1-2 deg C right next to the pack is the key metric. With that much solar derived thermal energy in the water, the ice hasn't got much of a life left.

Bob Wallace

Very interesting Paul. (Thanks for all your contributions to this site.)

Now, let me ask you about this phrase...

"likely less than 0.5 deg C due to the cooling effect of the melting ice on the relatively static layer of water under the static pack"

The 0.5 degree C is an assumption or an actual measurement?

If assumption, does it include the "Boxall effect"?

"Year-old ice, however, remains fairly salty. And when it melts, it produces meltwater that's denser than the relatively fresh water from older ice.

As multi-year ice declines throughout the Arctic, more of the saltier meltwater from younger ice is mixing into the ocean. That colder, denser water sinks more quickly and forces less dense water from deeper in the ocean up to the surface.

Because fresh meltwater is colder than seawater, that means relatively warm water is being forced upwards. And that, said Boxall, may be part of the reason that sea ice is melting so much faster than anyone thought it would.

What we're seeing is that [fresh meltwater] being taken away from the surface and replaced by slightly warmer water," said Boxall. "The evidence is that the surface waters are [now] slightly warmer."



@ Kevin,

SSTS are natural variability. But only the 1st time. Once they get going they drive themselves. They become the cause. Even of NV and GHG allow the ice to melt to allow for SSTS to get out of control. Then it is on the SSTs. This year most of us will agree SSTs are 80-90 percent for this year.

GHGs and NV are a much smaller piece of it.

If the wind did it then NV would be much higher. Since that is only system specific where as SSTs come from multi forces then drive themselves.

Paul Klemencic

Ned, to answer your question about convergence of IJIS reported extent:

We have anecdotal evidence from Bremen (thank you very much, L. Hamilton) and evidence directly from the NSIDC (thank you Neven; and Julianne Stroeve from NSIDC... big thanks for going to WUWT).

Bremen info says ice extent hit 4.6 million roughly on August 29, and NSIDC says 4.66 million today (presumably August 30 data; but what is a "day" anyway? Depends on where you are; or when the last satellite pass that could be processed today was completed; or, well who knows?)

In any case my number for August 30 was 4.68 million. And I see about 200k vulnerable ice in the Beaufort region, about 150k in the Chukchi, and over 400k in the E. Siberian, and even 120k in the Laptev region/central Arctic Basin bordering the Laptev.
Thats a total of 870k, and if we lose only 60%, that brings the real extent at the minimum to 4.16 million, which is probably a good 50/50 point. We could lose less or more, especially if the E. Siberian ice, which has been amazingly stubborn hangs on; or on the bigger loss side, if the east side keeps receding, or we get the big flush on that side.

Now to the IJIS versus MASIE/NSIDC data, lets just review August data. On July 31 it seems that IJIS was reporting only 112k more than the date corrected MASIE extent. This was because the last week of July the ice melt really slowed down, and the slowly responding IJIS measurements finally caught up with NSIDC.

But on August 1st the ice melt took off again. MASIE shows a string of high melt days; 99k, 56k, 64k, two times 160k in two days, and 145k. The IJIS system responded slowly as usually, with 19k, 3k, and 23k before it finally picked up the increasing extent loss, and reported 95k, 110k, and 81k. But now IJIS was 466k above (and behind) the ice melt being observed by NSIDC(MASIE).

So when MASIE shows extent loss slowed down with a two days of gains totaling +55k , consistent with the maps, IJIS still showed losses with relatively high melts of 57k and 71k, but was still behind 284k.

So while MASIE reported widely observed small losses of 25k to 44k over the next four days, IJIS cranked out losses of 68k, 75k, then 56k, 77k, and 127k getting back to within 84k of NSIDC on August 13.

Then things reversed again and IJIS reported some weaker days, and so MASIE opened up a 240k lead again by August 22nd.

I don't have the last week of MASIE numbers yet, but I kept my observations of the Bremen map up to date. Using the latest reported MASIE extent of 4.87 million, and with four days of meager losses totaling 50k, followed by a big loss of 100k on August 29 (versus IJIS at 68k) and a smaller estimated loss of 40k or so (versus IJIS at 100k), I estimate that MASIE currently at 4.68 million, with IJIS only about 120k behind.

As we approach the minimum, unless we get a huge weather event in mid-September to flush out ice, I think the IJIS will close to within 50k at the minimum.

Since I expect 4.16 million to be the minimum MASIE number, I expect IJIS to bottom at around 4.21 million, to just beat the record low.

The key problem with the oversized IJIS extent reports the last four weeks, was the mis-leading forecasts and predictions based on those unrealistically high extent reports. For my methodology to be accurate, I needed accurate regional and overall Arctic extents.

Weather events could swing this 100k to the upside and over 200k to the downside.

Ned Ward

For Arctic ice melt, every data point in the summer is not only 1 sigma out, but 2 sigma out, and some seem to be 3 sigma out! The system fails all four Western Electric tests.

The Arctic ice system is completely "Out of Control!". That means anyone who comes along and tries to use normal distribution statistics on this system, is making a naive and ill-informed mistake! If the system continues to produce 3-sigma events, then why continue to use normal distribution statistics?

Paul, you're just engaging in arm-waving here. What is an "event" in this context -- every single day's sea ice extent measurement? Which events are 1, 2, or 3 "sigma out" from what, based on what model?

You aren't clear about exactly what you're criticizing, nor do you explain how you think something should have been done differently.

Let's take my model as an example. Back in early July I predicted that IJIS sea ice extent would decline according to a model based on the range of declines in other recent years (2002-2010). That model has been remarkably accurate. Right now, there's just slightly more ice out there than the model was predicting six weeks ago. It is off by much less than one sigma, and has never exceeded the two sigma envelope.

In early July, this model's best estimate for the daily minimum was about 4.5 million km2. Now, six weeks later, with nearly all of the ice loss behind us, the model predicts ... about 4.5 million km2.

So what exactly is the problem?

Ned Ward

Apparently Paul and I were each writing lengthy comments simultaneously. Thanks for answering my previous question, Paul. I think you're saying that you'll be comparing the actual reported IJIS minimum this Sept. to the actual reported IJIS minimum from 2007. That seems like the appropriate, consistent thing to do.

I guess we'll just have to agree to disagree on the other issues. Clearly, a lot of people here enjoy reading your comments and appreciate the (no doubt extensive) time and work you put into them. Hopefully this blog is capacious enough to provide habitat for widely differing approaches to understanding the changing nature of Arctic sea ice.

Paul Klemencic

Bob Wallace, I am using an educated guess. Static water systems typically exhibit heat transfer coefficients of about 500 W per sq meter deg C.

We know the melt rate under the static pack is about 1-2 cm per day, or about 39-78 W per sq meter. This is an observed heat transfer rate, which is equal to U A (delta T). The area (A) is likely more than one meter per meter of sea ice, so this increases the heat transfer somewhat. The heat transfer coefficient (U) is definitely less than in a clean fresh water system, and is the biggest unknown. I understand that there is static areas of slush, and bacteria growth under the ice that would cause much lower heat transfer film coefficients (due to a static layer of cooler water trapped against the ice). I used 200-300 W / sq km deg C, but it could be much lower. In that case the delta T would have to be higher than 0.5 deg C to get the observed heat transfer.

Also likely is that there are pockets of static water trapped under the static ice pack, especially the traditional thick ice. The water near the ice will cool down to close to freezing, as we know it eventually does in the fall. In ice melt in salt water systems the cooler water does sink to a certain level and have to build up before freezing begins in the fall, but this is a relatively slow process. I think the actual temperature differential in that last 10-20 cm of water next to the ice is only in tenths of a degree.

But lets bust that pack up and put in the open sea. The slush on the bottom is abraded, so the bacteria film is gone, leading to higher heat transfer coefficients; the water under the ice is replenished more rapidly leading to higher temperature differentials; and waves or ice edges provide more heat transfer area. All of these effects can increase the heat transfer rates by 2x to 4x each. So we could see 2000 watts per sq meter.

At that rate it should cool the water down and lower the temperature differential, and I think that usually happens. Check out the SSTs around the pack closely, and you will see a cooler band some distance back from the pack.

But now put the pack in motion. Push the pack 100 km over 2-4 deg C seas... big increase in heat transfer. These are the system changes we have been talking about for the last month.


SIE at 4,737,969(prelim) for the 31st. A drop of about 59 k. SIE is now about 30 k above the minimum reached on Sept 9 2008!


@Frivolousz21 | August 31, 2011 at 20:09
I am sorry you can not see this. This is one of the larger drops. And looks quite a bit larger then yesterdays.
@Frivolousz21 | August 31, 2011 at 18:51
UB prelim map is out and all I can say it an other 100K drop is coming tonight maybe more.
Well, maybe you expected more of a drop within the grayed-out East Siberian Sea due to winds.

A 58,906 km^2 drop is quite impressive this late in the melt season, but it's no 100,000 km^2. Still, 2011 is close to dropping below 2008 now: 09,09,2008,4707813 km^2 was the minimum for 2008.

As I said, Banks Island, the large Canadian island on the southern side of the westernmost part of the Northwest Passage, is 70,028 km^2. It helps to keep that area in mind when eyeballing the Bremen maps.


You must be great company to be around-

the UB final chart shows more than 60,000km2 loss.

You can Download PS5 demo and see for yourself.

Oh well. You were right, congrats.


That 70,000km2 was there yesterday too?

Steve Bloom

As noted above, two more days of the same would put 2011 into a virtual tie with 2007.

Greg Wellman

The last couple of days of the IJIS graph have me hearing Tom Petty.

(Free Falling...)

It's still probably less than a 50% chance to beat the 2007 daily low extent, but pretty close to. IJIS Area is dipping under the 2007 line. Uni Bremen on the other hand has 2011 extent lower than 2007 for today. It's all within minor weather fluctuations, so all we can really say is that the death spiral continues.


A 60,000 km^2 drop is quite impressive - for comparison, last year on this day there was a drop of 18,906 km^2.

I'm mainly interested in the rising ocean temperatures contributing to the thinning/disappearing sea ice (and the Argo system has little good data on the Arctic Ocean yet), the effects of all this new open ocean on the halocline and underwater currents, and the ice thickness maps - CryoSat-2 has been a disappointment this summer in their cautious data rollout. Given this lack of basic data, I haven't spent much time on the summer melt this year, but I do have my popcorn and Parmigiano-Reggiano ready to watch the final few weeks...

This paper has a nice systematic way of looking at the summer sea ice melt - I like how they distinguish bottom melting from ocean dynamics ∆h-botO, and bottom melting from local atmospheric heating of the ocean ∆h-botA, for instance:


Our analysis shows that top melt dominates total melt early in the summer, while bottom melt (and in particular, bottom melt due to ocean heat transport) dominates later in the summer as atmospheric heating declines. Bottom melt rates in summer 2007 were 34% higher relative to the previous 7 year average. The modeled partition of top versus bottom melt closely matches observed melt rates obtained by a drifting buoy. Bottom melting contributes about 2/3 of total volume melt but is geographically confined to the Marginal Ice Zone, while top melting contributes a lesser 1/3 of volume melt but occurs over a much broader area of the ice pack.

(This paper looks at the Pacific Sector - the Atlantic side probably has much more early melt due to warm Atlantic heat convection)

It's the bottom melt that is inexorably increasing year to year (since AGW dumps 90% of its heat into the oceans) and which will one summer make it all "disappear quite suddenly".



has Cryosat2 released any data outside of there jan/feb data?

will you update all of us on the latest?

Please and Thanks.

I try to find it, but it's hard to find it all together.

Michael Stefan


...For Arctic ice melt, every data point in the summer is not only 1 sigma out, but 2 sigma out, and some seem to be 3 sigma out...

The NSIDC graph shows that it is even more extreme than that; the shaded area is +/- 2 sigma and the current extent is about 2 1/2 times further away from the average than the 2 sigma band, making it closer to 5 sigma below average, which is supposed to be a 1 in 1744278 probability event (I know that you can't really use that probability value but it shows how anomalously low the ice is right now).

Bob Wallace

Cutting soot emissions from diesel engines and other sources could be the fastest, most economical way to slow global warming, a U.S. scientist says.

Mark Z. Jacobson says reducing soot could slow melting of sea ice in the arctic, considered by many as a tipping point for Earth's climate, a point of no return. That's because the ice, which reflects sunlight and heat back into space, would, as it melts, expose darker water that absorbs heat and makes global warming worse, he said at a meeting of the American Chemical Society in Denver.

Jacobson says his calculations show controlling soot could reduce warming above parts of the Arctic Circle by almost 3 degrees Fahrenheit within 15 years, virtually erasing all of the warming that has occurred in the region during the last 100 years, a society release reported Wednesday.

"No other measure could have such an immediate effect," Jacobson said. "Soot emissions are second only to carbon dioxide [CO2] in promoting global warming, but its effects have been underestimated in previous climate models.

"Soot emissions account for about 17 percent of global warming, more than greenhouse gases like methane.

"Soot's contribution, however, could be reduced by 90 percent in 5-10 years with aggressive national and international policies."

Decreasing soot could have a rapid effect, Jacobson explained, because unlike carbon dioxide, which remains in the atmosphere for years, soot disappears within a few weeks.

Read more: http://www.upi.com/Science_News/2011/08/31/Soots-role-in-climate-warming-examined/UPI-40061314831367/#ixzz1Wfupodvw

Mark is the Stanford professor who wrote the 2009 Scientific American article which laid out a plan to get ~100% of our electricity, transportation and heating energy from renewable sources and how we could get those systems in place in 20 years. (Given adequate political will.)



UB take another sharp drop.


With CT area at 3.12 a drop of only 9k it increasingly looks like we may be running out of time for an area record which looked virtually assured 5 days agos.

The concentration in the central egg shape look fairly high now so there is a chance that increases there reduce and bottom melt concentrated at ice edges mentioned above cause further drops. Nothing is a certainty yet.


Temperatures north of 80N have fallen sharply in recent days, much earlier than they did in 2007 and 2010, but very much like in 2008. The pictures from the Healy posted by Neven seem to confirm this.
In these conditions, I would expect that SIA would not be dropping much further. This being said, IJIS Jaxa still has SIA at the same level as 2007.


If JAXA extent changed in the same way as last year then 1 day min extent would fall to 4.22. Following any other year in JAXA data would not reach a record low but still last year might be becoming the new norm.

Rate of fall in last week averages 49.5k per day. 2008 had a fall of 56.7k per day for that week but all other years were less than this year rate of fall. Per Lucia this would also support a high further fall in extent.

Pete Dunkelberg

“It’s hard to predict the future of Arctic sea ice.”

William Crump observed earlier that the demise of multiyear ice creates new territory for first year ice to form in and occupy year after year.

Steele et al. find mostly bottom melt using this model:
"2. Methods
[5] Our main analysis tool is model output from a coupled sea‐ice‐ocean model of the arctic seas, the pan‐Arctic ice ocean
modeling and assimilation system (PIOMAS) of Zhang and Rothrock [2003]."

Does the model overestimate volume and thickness?

Note the investigations of Rampal et al.: http://web.mit.edu/~rampal/rampal_homepage/Publications.html.

Jacobson wants to clean up the transportion sector, and who doesn't? But as for cooling the Arctic, how much can that do vs other factors?

Perhaps after this year's minimum SIE there could be a thread just for analysis of papers. For now, it would nice to have definitive explanation of why Bremen and IJIS show current extent lower/higher than 2007 at this date.

Bob Wallace


"Jacobson wants to clean up the transportion sector, and who doesn't? But as for cooling the Arctic, how much can that do vs other factors?"

The paper quoted-

"Jacobson says his calculations show controlling soot could reduce warming above parts of the Arctic Circle by almost 3 degrees Fahrenheit within 15 years, virtually erasing all of the warming that has occurred in the region during the last 100 years...."

Ned Ward

I just looked back at my model predictions from early July. When I initially created this model, on July 10, its prediction for today, the first of September, was 4671181. That's not an even number of 12.5 km grid cells, though, so let's round that to 4671250.

That would be just 67K below this morning's prelim number for 31 August. A 67K drop in one day would be a lot ... but if today's final is revised downward, we'd be within striking distance.

Of course, getting the exact number seven weeks in advance would be purely chance. But wouldn't it be cool?

Keep your fingers crossed for an IJIS number of 4671250 tomorrow!


Paul Klemencic

Excellent job, Ned. What a great model!

What does your model predict for September 1, 2012?
How about September 1, 2016?


Great additions on the graphs Larry!

Seke Rob

Far too busy, but a quick peek and bash reveals that 2011 distanced itself a bit on the YTD anomaly average metric [1K], for the first 242 days of CT Area data. JAXA 2007 keeps sneaking up on the extent metric for same, but at 136,719 over 2011, that's been slowing.


Ain't not pretty.

NSIDC monthlies should be out later today or tomorrow (not yet asjust checked on their page).

Pete Dunkelberg

Bob Wallace, agreed, I had read that, but is it believable? "other factors" hint Rampal

Michael Stefan
Jacobson says his calculations show controlling soot could reduce warming above parts of the Arctic Circle by almost 3 degrees Fahrenheit within 15 years, virtually erasing all of the warming that has occurred in the region during the last 100 years

I read that as saying that GHGs have had NO warming effect whatsoever, which strikes me as very shortsighted, focusing on just one factor. Plus, why did the Arctic cool in the 1940s-50s when increased aerosols caused global temperatures to stagnate (generally following NH temperatures, note that SH temperatures didn't cool or stagnate, and the Arctic is rapidly warming over the past decade when mid-lat NH temperatures have stagnated, unlike the first time)? Sea ice also doesn't last long (especially now) so I'd think that soot wouldn't be able to build up much (except for Greenland); in fact, new ice that forms next winter will have had only months at the most to accumulate soot.

Ned Ward

This is frustrating. Once again I've posted a long comment that got eaten by Typepad. At least it doesn't seem to have shown up. I'll try to reconstruct it:

Paul writes: Excellent job, Ned. What a great model!

I assume you're being facetious, since as noted above, this is a very simple model that happens to have gotten lucky. If tomorrow's IJIS number actually did turn out to be 4671250, it would be mostly attributable to chance.

(If anybody at IJIS is reading this, remember that the number for tomorrow is 4-6-7-1-2-5-0...)

Back to Paul: What does your model predict for September 1, 2012? How about September 1, 2016?

I've never thought about running it beyond the end of this season, but in principle there's no reason one couldn't. However, the range of uncertainty becomes very large:

Prediction for 1 September 2012: 4.65 million km2 (95% confidence interval 3.4 to 5.9 million km2).

Yes, that's a joke, though the calculations are real. You need to understand that different kinds of questions require different kinds of models. If I wanted to make actual quantitative predictions of sea ice extent in future years (as opposed to future weeks), I'd want to do it differently.

Many people have come up with methods to predict future trends in sea ice; some of them are confident enough to bet US$10,000 on the outcome. I am not in that category.

If you want my qualitative interpretation of the likely future evolution of Arctic sea ice extent, here it is:

1. I expect the long-term decline in sea ice to continue, albeit with substantial interannual variability around that trend (as we have seen in the past three decades).

2. I expect the next few years (2012, 2013, 2014) to have very low summer ice extent, more similar to 2007-2011 than to the pre-2007 years.

3. Before too long (2018? 2028? who knows?) I expect the warming Arctic to lead to a point where any year with unfavorable weather conditions has no significant ice left in the Arctic ocean in mid-September (i.e., extent < 1 million km2). Even at that point, the ocean will continue to freeze over again in the winter.

4. As more time goes by, assuming we don't make serious reductions in CO2 emissions, I expect the Arctic to continue warming. Sea ice will disappear earlier in the summers, and form later in the fall. The Arctic Ocean will be more like the way Hudson Bay is today.

This will undoubtedly have large (and largely unpredictable) effects on northern hemisphere weather patterns. It will also impact marine biology and ecology, global shipping and resource extraction, and the rest of the planetary climate via the effects of an ice-free ocean on global biogeochemical cycles and energy fluxes.

I hope that's the kind of answer you were looking for.

Seke Rob

Ned, if using Firefox, suggest to get ahold of the Firefox addon "Open Textarea Cache window". Copies posts for longer, as you type, and even keeps the iterations in case you've rubbished something accidentally. (of course not suitable for those using the webbrowser to write to girlfriend 2.0 ;>)

Alternate solution is using "It's all text", then write a post in your fav html text editor, hit save and it gets automatically transferred to the post-text-area. ScribeFire does a good job too, but found that it would not transfer the formatting.

Sorry for the off-topic, but these have been often enough the savior of the day.

Seke Rob

NSIDC Monthly extent just out:

Slightly over 2007, where July 2011 was well below. Going to check how that monthly datafile number matches up with MASIE's past 31 day average (lest someone already did).

Seke Rob

MASIE is in for the day... the KmSq are very close to JAXA.

MASIE 4737839 (31st)
JAXA 4743750 (31st)
Difference -5911

Same as yesterday by region, net loss at top... hefty:

"Northern Hem. Tot. 4 737 839 (-134 730)"
"1) Beaufort Sea 576 837 (-29 588)"
"2) Chukchi Sea 183 457 (- 589)"
"3) East Siberian Sea 512 749 (-19 690)"
"4) Laptev Sea 28 127 (-7 688)"
"5) Kara Sea 31 121 (-6 036)"
"6) Barents Sea 5 593 (- 841)"
"7) Greenland Sea 274 953 (-11 166)"
"8) Baffin Bay G.o.St Lawrence: 4 147 (-3 602)"
"9) Canadian Archipelago: 134 095 (-32 724)"
"10) Hudson Bay 15 802 (-6 632)"
"11) Central Arctic 2 969 850 (-15 333)"

On extent for August, the comparative figures of the NSIDC monthly versus 31 day average of MASIE:

NSIDC: 5520000 (Area 3.31, including the blind spot of 0.31M KmSq)
MASIE: 5737066 (2 days no data provided)
Diff: 217066+ for MASIE

If in-fillling the missing days, their average works out at 5749046. Not very sciency, small difference.

Patrice Monroe Pustavrh

Ned, I do somehow agree with you, but I do expect more than one step to 1 mio sq km. But, yes, we will see some ups and downs, but general trend is clear. Down, to zero in the late summer, and less and less ice in the winter/spring.
A little bit of anger from my side: Did somebody mentioned world is not warming ? Oh yeah, it could be melting temperature of ice which is increasing, isn't it }:-> :D


Alaska Dispatch: Northern Sea Route setting Arctic commerce records

MSNBC: Second giant ice island set to break off Greenland glacier (maybe next year)

Seke Rob

Ned, I've got reservations with the sooth impact story [the share of the total] from past and recent readings. Anything in there on snow / days off? How much can there be on either, when most of it is FYI, and for snow the winter to summer melt on the NH is > 40 million KM sq. I'd be mighty interested in the Russian/Siberian 2010 and 2011 fire impacts. Think at one time one poster here shared a Terra image with a big smoke cloud in it.

Think I'll have a browse around over at the AGW Observer site if there is a category for this and ABC clouds effects.

L. Hamilton

Twemoran, thanks. I debated whether adding little numbers made the graphics too busy, but this is such a detail-focused crew here.

I hope to have more soon on the Uni Bremen time series, which adds 7 years to the NSIDC/CT records.

Seke Rob

Re: Seke Rob | September 01, 2011 at 19:14

Sorry, that was at Bob.

Larry, those numbers at top are right how I like it. Those that are poor readers of the Y-Axis scaling can immediately see the differences.

Paul Klemencic

Ned, that is actually a pretty good response.

I was going to make the point that if you had July 1st data for 2012, and it was identical to July 1st data for 2011, your model would make the same prediction as this year. Same for 2016.

What is missing of course, is whether the mechanisms for ice melt remain the same as 2011. Early season melt in June is driven by solar radiation, and even through July. But the critical months of August and September are driven by bottom melt, and by disbursement of the ice pack (it scatters), and by ice pack mobility (it moves and shifts over the warmer seas).

So the July 1st conditions could be the same, but the end of melt season minimum declines in 2012, and even more in 2016. dramatically as the last of the multiyear ice disappears and surrounding sea temperatures rise.

I am more interested to see just how fast the Arctic ice death spiral kicks in, than predicting a given year. Already this year, we saw some extraordinary events, that are not reassuring... Neven will likely discuss some of these in his End Zone series of posts.

I don't quite like the "canary in a coal mine" example used for the Arctic. The problem with the analogy, is that when the canary died, that event didn't make the mine more dangerous. But when the Arctic ice dies in the summer, this event should screw-up some systems on our planet.

So examining the ice melt to improve ability to predict when the Arctic death spiral opens up enough water to initiate larger changes in these other systems... then that is a rewarding endeavor.

Paul Klemencic

On a similar note: When I came to this site, I was unaware of the MASIE site. I love the site, and would love it more, if it was up to date.

On the Home Page, they say the site is underfunded and can't be maintained.
Is there any we could encourage funding this site?

Here is the message on the site:

NSIDC has received support to develop MASIE but not to maintain MASIE. We are actively seeking support to maintain the Web site and products over the long term. If you find MASIE helpful, please let us know with a quick message to NSIDC User Services at: nsidc@nsidc.org

If anyone has the time to write them a brief note, it would be nice.


2007 8 Goddard N 5.36 3.11
2008 8 PRELIM N 6.06 3.43
2009 8 NRTSI-G N 6.26 3.79
2010 8 NRTSI-G N 5.98 3.54
2011 8 NRTSI-G N 5.52 3.00

Record low area but 2nd for extent is no surprise.

Ned Ward

Today's revision was upwards (4743750, a drop of 53K).

That does not help my goal of having tomorrow's SIE be 4671250.

Oh well...

L. Hamilton

Crandles is right that August NSIDC area set a new record. For the actual values, note that we should add +.31 from 7/87 on, and +1.19 before that.

Lucia (The Blackboard)

I never read details about your model. You said it's simple. Does your model just use average ice loss and subtract that from the present extent?

Paul Klemencic

They updated the MASIE site spreadsheet, and it shows the NSIDC had an estimate of 4.74 million for... well, my analysis suggests this data is for August 25. The drop was 134k from August 24, so I missed seeing this big drop on the map.

Going to the regional reports, I can see why. On August 25, the biggest regional drop was in the Canadian Archipelago (34k), the Beaufort (30k) and the Greenland Sea kicked in 11k. I spotted and noted the Beaufort decline, but didn't check the Archipelago or the Greenland sea closely.

Revisiting the Bremen maps, you can see the loss of extent in these regions easily by a blink comparison from the 24th to 25th, especially in the archipelago and the Beaufort.

So as of August 25, the NSIDC had extent at 4.74 million sq km, much lower than my estimate of extent at the time (about 70K lower).

Clearly I have been too cautious and conservative in my estimate of extent over the last week. In the future, I will need to be more aggressive when estimating daily extent.

Ned Ward


Yes, it takes the average loss (or gain) of extent from Date 1 to Date 2, and uses the variance around that average to compute confidence intervals. Nothing fancy.

I kept a "static" version with the predictions from when I first created it (early July) and another "dynamic" version that updated automatically as each new day's SIE was posted.

IIRC you had a post about "toy models" once. Well, this is the batteryless, non-electronic, no-commercial-value version. It's the pickup sticks of toy models.


Next year I might try something a bit more sophisticated.


The Modis view of the landfast ice in Northeast Greenland has been obscurred for a couple of weeks now so I thought some of you might be interested in the ASAR radar images of the situation below the clouds
from DMI Center for Ocean and Ice;



L. Hamilton

Paul, the UB 1-day value for 8/25 was 4.69. Average for 8/23-8/27 was 4.75. Pretty close to NSIDC (4.74), compared w/IJIS 5.06.

Ned Ward

Paul, the UB 1-day value for 8/25 was 4.69. Average for 8/23-8/27 was 4.75. Pretty close to NSIDC (4.74), compared w/IJIS 5.06.

Or, for those who don't believe in the Time Lag, the actual MASIE figure for 25 August was 5.21, just slightly above the IJIS figure of 5.06.

Lucia (The Blackboard)

Nothing wrong with simple.


TZ - try one step further north and catch some really old stuff drifting away

BTW is there anything similar to the DMI site that covers the Canadian Archipeligo


A little bit of anger from my side: Did somebody mentioned world is not warming ? Oh yeah, it could be melting temperature of ice which is increasing, isn't it }:-> :D

Don't you mean 'decreasing'? ;-)
Actually if the proportion of MYI is decreasing then the melting temperature of sea-ice is decreasing!


Yes North and North East Greenland is the real action place these days, thousands of km2 shore/fast ice are on the loose right now, it is always fascinating to watch this part of the polar sea in the end of the melting seasons, especially now when the great Petermann is resting for a while, also a lot of small glaciers drying out in south east Greenland!
Regards Espen

Paul Klemencic

Ned and Lucia,

For every complex problem there is an answer that is clear, simple, and wrong. H. L. Mencken

Paul Klemencic

Ned, since Julianne Strove at the NSIDC commented at WUWT that the NSIDC sea ice extent was 4.66 million on August 30, we should have a very good confirmation of the MASIE delay in posting SIE on Monday (but since that is a holiday in USA) likely on Tuesday.

So we don't have long to wait before we know whether the simpler answer or the 'crazy' alternative is correct.

Lucia (The Blackboard)

For every complex problem there is are many answers that are confusing, muddled, convoluted and wrong.
L. M. Liljegren

Lucia (The Blackboard)

If Ned's forecast had less than pin-point accuracy, that will not make his simple method "wrong".


A useability question - does anyone know how to go from someones latest comment on the left of Nevens home page, to the latest comment at the end of the comments. At the moment I follow the link, scroll down to the end, click next, scroll down to the end, click next etc. bit of a pain - wonder if there is a quicker way?
ps great stuff again Neven, thank you.

Paul Klemencic

Lucia, You have managed to mangle and confuse an already complicated situation, by selecting the wrong answer to attribute my retort to, and then extrapolating your trend of thought into some imaginary world that you are creating in your mind.

Ned is choosing to believe that the MASIE spreadsheet of NSIDC data is correctly identifying the date that the NSIDC data was taken. I have made, I believe, an almost overwhelming case that MASIE is dating the data six days later than taken.

Only one of our theories can be correct.
And we will know that on Monday or Tuesday, when MASIE posts the data for September 5; I believe that data will actually be for August 30th, and the ice extent reported will be 4.66 million sq km.

Lucia (The Blackboard)

What makes you think you know what answer -- if any-- I attributed your retort about simplicity to?

I will admit that when retorts are addressed to me, I tend to expect they might be responding to something I posted. That said, I am aware that this is not always the case.

Thanks for pointing out that the comment you addressed to me had absolutely nothing to do with the conversation I was having with Ned.

I have made, I believe, an almost overwhelming case that MASIE is dating the data six days later than taken.
Well, I've read your various theories. Rest assured that I strongly suspected you are convinced you have presented an overwhelming case for your notion about the time lag.

Only one of our theories can be correct.
If you want to consider the full range of possibilities, you both could be wrong.

Lucia (The Blackboard)

does anyone know how to go from someones latest comment on the left of Nevens home page, to the latest comment at the end of the comments.
I've been wondering the same thing. I was hoping Neven could look into setting Typepad permits to see if he can get the side links to include the /comments/page/n/ etc stuff required to get us to the comment.

Paul Klemencic

Yes. Both could be wrong. But this statement is still correct:

Only one of our theories can be correct.


Ned perhaps you can clarify for me. From the documentation I read MASIE and IJIS have two entirely different data sources. Or more precisely MASIE uses more data sources than IJIS to create it's daily product. What am I missing? Thanks in advance your comments have always been most helpful.

The comments to this entry are closed.