« NSIDC Daily Extent Sea Ice Index - More Information - Duplication | Main | Peeking through the clouds 4 »


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


Note that this is a measure of number of "ice days" of melting (scaled by area): probably reasonably correlated to actual cm3 of melt, but not precisely.


Aaron Lewis

In the old days, water vapor (latent heat) from the south tended to condense on the sea ice.

Now the open ocean water tends to add to the latent heat in the air flowing over GIS (and other sea ice.)

Local temperature and local insolation are no longer good indicators of melt potential. We need to look at total available energy including latent heat and potential energy from the fact that some of that ice is kilometers above sea level.

Chris Reynolds

So this is what it's like when the Arctic still has significant (albeit reduced) sea ice cover.

William Hughes-Games

Could this melting be the first sign of a Walker cell developing between rising moist air over the open part of the Arctic ocean (and Atlantic, for that matter) and the descending katabatic (density) winds pouring down the slope of the Greenland Ice Sheet. The air that pours down Greenland has to come from somewhere.


William, thanks for that piece you wrote 4 years back. It was very informative and educative.

Espen Olsen

Area crossing 3 million line at 2.986 and anomaly at 2.271!

Rob Dekker

Neven, thanks for this post. Greenland seems especially hard hit this year, and may even be unprecedented in a 1000 year history :

But the 'skeptic' in me does not see any more info here than what you already posted before.
For example, I understand that this post originates here :
where they refer to the "melting index" :

The melting index is computed from passive microwave satellite measurements and it can be seen as a measure of the ‘strength’ of the melting season: the higher the index the more melting occurred.

OK. But where is this "melting index" defined ? And it this quantified in any way ?
Which scientific paper defines this melting index ?


>"OK. But where is this "melting index" defined ? And it this quantified in any way ?
Which scientific paper defines this melting index ?"

Um in the article above. I guess it isn't yet published.

"To quantify the changes, he calculated the duration and extent of melting throughout the season across the whole ice sheet, using data collected by microwave satellite sensors.*

This ‘cumulative melting index’ can be seen as a measure of the ‘strength’ of the melting season: the higher the index, the more melting has occurred. (The index is defined as the number of days when melting occurs multiplied by the total area subject to melting.)"

>" does not see any more info here than what you already posted before."

Lost on that. This seems to me to originate 15 Aug and therefore Neven is quick off the mark. This is not the same as widespread melting for a few day blogged about before.


Sorry - too quick to react and post. That definition isn't sufficient as negative numbers shouldn't come out of that formula. Perhaps that graph just shows annual anomalies in the index? Maybe normalised?


Some googling resulted in this:

Melting Index (MI)= Area * number of melting days

Standardized Melting Index (SMI) is calculated from MI by subtracting the mean for the period 1979 - 2012 and dividing by the standard deviation for the same period.

In other words: if the SMI over 2012 eventually ends at 2.0, we have a 2-sigma event.

Philippe Terrier

I see differences between the new "SMI index" figure


and this old one published in 2011:


2007 was the second highest melting years, but not in the new figure...

Change in methods ?


The same story told in maps and plots here:



Thanks Wipneus. 9 out of first 10 years are negative and 9 out of last 10 years are positive. About a 1 in 4970 chance if distribution is just random. So I think that passes a 95% chance there is a trend.

So should we conclude that this year is only slightly worse than should be expected and most subsequent years will be worse than this?

At what point should you start calculating standard deviations as residual from trend rather than anomaly from average?


Not to get off topic, but is that a CT area record I see this morning.? Also 4 weeks early. As they say here is Newfoundland, "she's gone, b'ys. She's gone."



Lots of differences: 1987 is the lowest in the old, unremarkable near zero in the new figure.


You assume uncorrelated data for the 1 in 4970 chance calculation (nitpick).

Nobody will prevent you from calculating trends. It is the conclusions that you might draw from that, e.g. did you consider cycles (AMO,AO)?


I know someone has answered this before, but what is the time lag on CT ? One day? Or is it an average?


Rob Dekker, I haven't had time to look into this properly (too busy with some other stuff), but I thought it'd be a good starter for the 7-course meal of records that is to come. I'm sure more info will appear in the coming weeks.


NLPatents, no CT record as of yet. The latest value of 2.986 million km2 is for August 14th.


With yesterday's drop in CT SIA of 111k km2, 2012 area has dropped to 2.986 million km2. Some stats:

  • That's only the third time in the record area has been below 3 million km2 (2007 and 2011 were the others, of course).
  • It only took 138 days this year for area to fall below 3 million from this year's maximum. 2011 accomplished that feat in 169 days--one month longer--while 2007 needed 178 days, nearly six weeks more. Perhaps more astonishing, that 138 days is less time than it took most years that fell below 5 million km2 to reach that point from their individual maximums.
  • 2012 SIA fell below 3 million km2 nine days ahead of 2007, and ten days ahead of 2011. 2012 area is currently 405k km2 ahead of 2011 on the same day, and 356k km2 ahead of 2007.
  • As others have noted, an area record is just 81,585 km2 away. Even if area were to see an additional decrease this year equivalent to only the lowest post-6191 drop on record--1997's 194k--2012 would still end up with just 2.793 million km2. At the other end of the scale, 1984's post-6191 drop of 1.113 million km2 would render a very far-fetched 1.874 million km2.
  • 2012 SIA has now been in first place for the past 47 consecutive days--since June 30--and 63 out of the last 68.
  • 2012's SIA anomaly has been below -2 million km2 for 14 consecutive days. That's 20% of the 70 days on record with an anomaly greater than -2 million. (The greatest anomaly so far this year ranked #20 on that list of 70; the 19 largest negative anomalies were all in October of 2007.)

Thanks, Jim. That'll come in handy when the record is broken.

Seke Rob

Update projection and "still to melt": http://bit.ly/CTNHM2 with 94% of prior years minima already surpassed and 2-3 weeks, and who knows more days of melt reduction, limbo, and redistributions to go [compact/spread as the wind blows]. Projection worst case: 2.55M Km^2... Dire Straits, no not Mark Knopler and co but cold commerce: http://www.cpreview.org/2012/05/dire-straits/


>"Projection worst case: 2.55M Km^2"

I have largest decline from day .6191 in a leap year (.6165 in non leap year) as 1.15 to give a worst case projection of 1.83. I consider this very unlikely.

The best case projection of 2.986-0.196=2.79

Isn't 2.55 a 'near' best case projection (where near means one sixth of possible outcomes would be more extreme)?

I think the forecast weather will throw out that one sixth probability except strictly in the sense intended that if the area is equally likely of follow any one of the past 33 years paths with equal probability and not anything else then ....

I think the forecast weather is making a value nearer the best case 2.79 more likely than the worst case 1.83.

Jim Williams

Rob, that is if Russia's plans are not stopped by burning peat and a bubbling sea of methane.

Espen Olsen

Jøkelbugt North East Greenland:

A very interesting situation is unfolding in the Joekelbay. A massive fast piece of ice (5 - 6.ooo km2) is stuck in the sea of the coast +/- 50 km. When blinking the 2 links below (images from Modis) you will see that this "piece" did not move an inch over the last month, it proves that something is holding it back, and it can not only be Tobias Ø (Island found 1993)?

Open the images in 2 different tabs in your browser! And then blink them!




>NLPatents, no CT record as of yet. The latest value of 2.986 million km2 is for August 14th."

I thought we had had it confirmed that .0027 is Jan 1 which seems to me to make
2012.6191 -2.2705851 2.9863231
August 13.

I could be wrong, I have thought I had got it sussed at least 3 times then seemed to find I was wrong.

This makes the interactive chart a bit weird with its horizontal year axis running from 31 Dec to 30 Dec. But it does label 2.986 as day 226 which would be 13 April in a leap year.

So what is the source to trust? Are the maps two days later than the latest available number? Is this to allow a 5 day average?


Should take a couple of years for this to start showing up on the sea level rise. But it reminds me of what Stephen Schnieder used to say, that we know there are tipping points but we just dont know where they are.

We may have already crossed a tipping point but it could be decades before it becomes obvious and when it does, it will be far too late.

Chris Reynolds


I use the part after the decimal point to multiply 365, then add 1, because the entire timeseries starts with 1979.000. As does each year, so that must be 1/1/xxxx

That said, day 2 of the series is 1979.0054 which is day 3! Still I overlook this anomaly, my calculation gives me day 365 correctly, at the end of each year.

So I work out 2.9863231, the figure released today noon (UK time), as relating to day 227, which is yesterday - 15/8/12.

Have I dropped a clanger somewhere?
PS, the graphics seem to update almost a day later.

Chris Reynolds

Oh I should clarify.

This year I'm working the dates out as 2011. This is because every year in the CT series ends .9973, or day 365. So the CT series seems to drop a value during the leap year.

That also makes it tie up neatly with PIOMAS which also ignores leap years.


Day 227 is August 14th in a leap year, right? Does CT count the leap year day? My god, I can't believe we're having this dicussion again. This has driven me insane multiple times.

You can do whatever you like, I'm not changing my spreadsheets again. I have 2986323 km2 for August 14th and that's that.

Chris Reynolds

And I'm not changing MY spreadsheets, so there. Blrgh! ;)

Dave Leaton

To add to Jim's post, and for what it's worth, we're also 164972.5 km2 from setting the melt season CT area loss record.

2008 10.8871598
2010 10.739908
2012 10.7221873
2009 10.4284585
2007 10.3977587
2002 10.2877698
1993 10.2820811

And just in case you're curious, the biggest freeze season gainers in the CTA record:

2007 10.9712765
2008 10.8495002
2011 10.8037708
2009 10.38744
2000 10.3575464
2002 10.3043513
2010 10.0722902


And 2012 has now equalled 2010's record of 46 century breaks.

That's according to my CT SIA spreadsheet. :-p


Not to wade deep into the off-topic weeds here--again--but the problem is the inconsistent way CT has handled leap years over the course of the record.

  • In 1980, a day .1631 was added between .1616 and .1644
  • In 1984, a day .1658 was shoved in between .1644 and .1671
  • In 1988, 1992, 1996, 2000, and 2004, there were two consecutive days marked .1644
  • In 2008, there were two consecutive days marked .4438
  • This year, the values for days two days numbered .1644 were combined into one. Those were then broken apart, then the values for two days numbered .1671 were instead combined.
  • Non-leap year wise, 2007 was (for some reason, though I suspect that some day earlier in the year was omitted) given two consecutive days numbered .9808, while 2009 was given two consecutive days numbered .4274

The bottom line, obviously, is that with the leap days being accounted for near ice maximum when the least daily change usually occurs, it seems fair to me to simply ignore any daily loss/gain during that period, and focus instead on the total ice coverage, which will be accurate despite what CT has done with the calendar. And graph-wise, I tend to just ignore the leap days as well, since every bit of the true ice loss I track occurs subsequent to leap day, anyway.

(FWIW, unless there's more calendar manipulation by CT, December 31 will be day .9973, the 365th day in the CT record [because of the aforementioned combining of two days numbered .1671].)

We have to strive for accuracy and precision, of course. But if our data sources are themselves a bit idiosyncratic, we have to accomodate and compensate for those idiosyncracies as best and honestly as we can. That's my goal, anyway... ;-)

Neven wrote: And 2012 has now equalled 2010's record of 46 century breaks.

That's according to my CT SIA spreadsheet. :-p

That would be correct. ;-) 46 century breaks this year, and 7 double century breaks. By comparison, 2011 had, respectively, 39 and 3, while 2007 saw 41 and 6.

Seke Rob
46 century breaks this year, and 7 double century breaks.
*and*, or *to include* (Just like dotting the eyes at times ;>)

Funny is that last time Neven quoted the century count, he had one more than I had, *and* I made no change to my sheet [LoL]


I've just looked at the area records again. Around this time of year, 1979 (and 1996) went below 6m, 2004 went below 5m, 2009 went below 4m and 2012 is going below 3m. You can see the melt speed up in the PIOMAS volume record, but it also clearly shows in area now.

James Benison


I believe that ice is grounded on the Belgica Bank. That area was just mapped in 2004.



>"This year I'm working the dates out as 2011. This is because every year in the CT series ends .9973, or day 365. So the CT series seems to drop a value during the leap year."

There are 366 values for 2008 because one digital date is repeated
2008.4438 -0.9495078 9.2806721 10.2301798
2008.4438 -0.8914949 9.2806721 10.1721668

now that might just be a repetition of the 9.2806721 compensating for a missing 29 Feb.

However for 2004 there is
2004.1644 -0.4795871 13.6285448 14.1081324
2004.1644 -0.4727035 13.6381149 14.1108189

That seems like different data and according to my interpretation .1644 is both 29th Feb and 1 March.

So I think there are 366 lines of data for each leap year. Also repetition at .1644 looks good for my interpretation :o) (so I'm not changing my spreadsheet either :-P )

Dave Leaton

One more small piece of esoterica that might have rhetorical value: CT Area has been below 50% of the 1979 daily numbers for the last week. It had never dropped below 50% before this season.

Fairfax Climate Watch

here's a question for those mathematically inclined.

Given the U.S. lower 48 was 3.3 F above average in July, how much extra heat energy does that 3.3 F anomaly translate to? And, how much Arctic sea ice would that extra energy melt if transferred with 100% efficiency?

Frankd 1977

Between 140E and 105E at 85N (just west of Severnaya Zemlya) are two large masses that seem to be seperating from the CAB pack. The current ice flow is moving counter clockwise around the south of the SZ islands towards the Kara Sea. What do you think the chances are that it changes direction and puches them in the the Laptev Sea?

Artful Dodger

crandles wrote "But it does label 2.986 as day 226 which would be 13 April in a leap year."

Look at the labeling on the first (leftmost) day in the interactive chart. You'll see it's labeled 'Day 0'. That's why you have to add 1 to CT days to match with MODIS usage, which has Jan 1st as Day 1.

Chris Reynolds wrote "because the entire timeseries starts with 1979.000"

uh, no. Look again at the data file. The first line is:

 1979.0027   0.5883108  12.9048309  12.3165197

This is the data for Jan 1st, 1979 and the start of the time series.

Neven's CT date usage agrees with Larry Hamilton's. It would be helpful if we added a FAQ section to this blog over the Winter, I think.


Rob Dekker

Neven, crandles, sorry for coming in a bit harsh in my comment.
My point was that I did not find the source article (greenlandmelting.com) in the post, and when I found it, the article itself was not very well sourced. So naturally a little 'yellow' light went off inside of me.

Thanks to Wipneus the issue got cleared up.
In fact, the definition of the Melting Index (Area * number of melting days) makes a lot of sense, and Standardized Melting Index (SMI) definition even more. Wipneus writes :

SMI is calculated from MI by subtracting the mean for the period 1979 - 2012 and dividing by the standard deviation for the same period.

This is actually a very helpful definition which puts the Goliath 2012 Greenland melting season into very clear perspective.

And, to complete my apology to Neven : Yes, this post DOES add information to your previous posts, since neither the albedo-change article nor the unprecedented 97%-of-Greenland melting present the crucial info about sea-level-rise and Greenland water run-off on exactly how much area overall melted for exactly how many days.

So, Neven, this post is not just presenting another record in Greenland, but in fact a very important and concerning one.


Hi all,

Over the page on Daily Graphs, Arthropolis is showing a temp of 13C for Alert. This is about as hot as I can recall ever seeing.

Simultaneously, the DMI temp plot for latitudes above 80N is now dipping below freezing.

Any explanations?



Quite a lot of non ice covered land near Alert. That is going to give a higher temp than average of north of 80N most of which is ice covered with a little water only slightly above freezing point of about -1.5C. Land with little ice cover is only small fraction of the area.

Kevin McKinney

Historic data for Alert:


13 C there is warm, but far short of record warmth, FWIW.

crandles has the right of it; the northern tip of Ellesmere is just a tiny fraction of the total area N of 80.



Alert is probably more affected by low albedo than NP - more likely with higher temperature in Alert than NP due to lower albedo in the neighborhood is my guess. NP will probably not have higher summer temperature until it has lower albedo i close vicinity.

Artful Dodger

Hi idunno,

The DMI temp plot is also the 24 hour average temperature. Alert's average temp for the day is what you should compare. Here's Environment Canada's 24-hour history page for Alert, Nunavut:


I find it even more remarkable that winds currently are SSW 51 km/h gust 63 km/h. That'll move some sea ice!

Lord Soth, what would it take to get a webcam up there at Alert?


Many of the curves I've seen extrapolate more or less smoothly to zero ice extent or zero ice area or zero ice volume. But, isn't there a threshold thickness below which the ice cannot maintain its integrity under the combined forces of ocean and atmosphere dynamics, and essentially collapses. And, if so, how far are we away from that point?


>"more or less smoothly"

most would be smooth with curves of different shapes. However second graph in Wipneus' outlook


Do you extrapolate with straight line through origin, straight line not through origin or a curve.

If you believe in "a threshold thickness below which the ice cannot maintain its integrity" then perhaps the trend switches from one straight line to the other creating a non-smooth point.

I think I am more inclined towards a curve gently bending one way then back the other way down toward origin turning increasingly sharply. That is based on not really believing in one sharp point ie break down in integrity will happen at different thicknesses for different sized pieces of ice so it doesn't just happen all at once. However that is pretty much just wild speculation without any evidence to support it rather than a confident prediction.


Today a 4,5 km2 calving appeared on Kangerlussuaq Glacier, SW Greenland.
the calving front was more or less stable during the last three years. When I put MODIS r02c02 under my CAD overlay day 230, the front was still intact.
This underlines the dynamic change all around and over the GIS this season.

Artful Dodger

NASA has launched a new GRACE website for the 10th Anniversary of the mission.

No news on the Summer 2012 Greenland melt, yet...


Artful Dodger

Wow, Chris B!

Sea level has returned to the trend-line established before the dramatic floods of Summer 2011. Do you have numerical figures in mm/yr for sea level rise this year?


Chris Biscan


You can get them below in text files.

There is a huge jump in July it seems.

the other main sources haven't updated yet.

Artful Dodger


It looks like sea levels have more than recovered the temporary decline of Summer 2011, increasing about 13 mm above the dip a year ago. Now also over 4 mm above the linear trend line for the whole data series which begins in 1993.

"Unglaublich aber wahr, aber unmöglich weil wir leugnen."

Wouldn't want to live in the low Countries (no matter how beautiful ;^) ...



Morning Lodger.
I've got a personal refuge on 430 m + in France... Should be safe yeah(if drought isn't gonna get me..)
No kidding, Chris B came up with the first signs of SLR that 'should' appear.
Spread 700 km3 melt of the GIS over 350Mkm2 oceans and its clear the count is on. El Nino is going top contribute too.
The first step, 15 cm in 10 years, south tip Greenland falling is in the make. This gets serious when it hits 4000 km3 a year (by 2016?). Should be visible by then on MODIS in general retreat of the sheet margin.

Artful Dodger

No man is an island, Werther ;^)

You think 4000 km^3 Greenland melt by 2016? What would that make the time for the doubling?

Last time I looked, people were thinking a possible range of 6-10 years for the doubling, awaiting data. I think Hansen bases his 2100 SLR predictions on a 10 year rate.



Lodger, actually I didn't figure this out thinking in terms of doubling. As I studied the geographics of southern Greenland late 2006, this was what seemed to fit natural processes. It's a step-thing; melt concentrated on the south tip first, phase 1 up to 2050, steepest curve in the '30-s, then a slowing while forcing builds and the whole GIS gets involved.
I hold this to fit well into forecasts by Rahmstorf and Church.


Oh...forgot to mention that my timing sucks... Had expected this all to happen a little earlier.
I hope that flaw continues and provides adaption time!

Artful Dodger

Hi Werther,

I agree about the geographic progression of melt in Greenland. Clearly we're seeing it already. Luckily we have good people like Jason Box and colleagues watching the melt, so we shouldn't be blind-sided.

I've read elsewhere about kinematic constraints to acceleration of Greenland glacier flow. Basically the long, narrow and shallow fjords work to slow rapid advance of glaciers.

Antarctica, especially the WAIS, is another matter entirely.


Artful Dodger

For you Data-hounds and Chart-mongers: (h/t Chris Biscan)

Global mean sea level data in .csv format from TOPEX, Jason-1, and Jason-2 satellites, with Seasonal signals removed


Artful Dodger


Greenland does indeed look like a hot-spot for SLR. (click the image to go to the NOAA webpage)


Great sites through the links again, Lodger.
Could be fresh water = less dense = larger volume.
Same effect for Hudson, Bothnian Gulf in Scandinavia.
They remark that in the data isostatic rebound through ice loss isn't incorporated.
Indonesia: coriolis force and warm upwelling...
Pity there's no data for the Arctic. Would the fresh water bulge in Beaufort have shown up?

Chris Biscan


Form that page, there is links on the left side of the page to the 4 major sea level trend algorithmic sources. most updated through 2012.4518

the one posted here with the big spike goes further to 2012.6500 or so.

John Christensen

An article has been posted on the record melting event in Greenland last summer:


The combination of an extended blocking high during most of the summer and a negative NAO created the conditions for this level of melting according to the article.

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