« Animation 9: sea ice concentration | Main | Area vs Extent »


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

Artful Dodger

Neven, thanks for your tireless efforts. It is a pleasure to read your blog!


Thanks a lot for the nice words, AD. It isn't easy, being extremely busy with work and not being an English native speaker. But I'm really happy with the way the blog has turned out.

I've just put up a new post on area vs extent. I hope you guys can help me sorting out percentages and graphs, as it is very relevant and interesting at the moment. Especially as we'll have a few more days of ice being pushed away from the centre (like AD noted based on the PIPS ice displacement forecasts) and might then see a switch towards compaction.

I will try and copy CT area data later today and come up with a graph of my own, but I won't mind if someone else does it! Even better would be IJIS providing me with area data.

Jim Dowling

Wow Neven, surprised you're not a native speaker, your english is that good. Nice blog, btw.


Thanks, Jim, but if you look closely you'll see I keep using the same kinds of words and sentence structures. But hey, apparently it's enough to bluff myself through this melting season. :-)


I'll echo AD's thanks - it really has been a pleasure following this blog's evolution.

One thing I'm curious about, if anyone has any thoughts:

Up here High Pressure usually leads to hot days and cold nights. In the Summer the former is dominant, but in the other seasons it is the other way round - from (approximately) September to April we can expect daily average temperatures to be lower than normal during a high pressure event.

Clearly the Arctic has been cloudy for most of July - and this had an impact. If it continued to be cloudy during the next couple of months, would this still have an impact? I suspect not, but this is just a "gut feeling" - so any pointers would be appreciated.


Just a typo, as Jim is implying more melting, but I think Jim's comments about soot increasing the albedo effect should be decreasing. i.e. high albedo is high reflectivity. Soot increases absorbtion, decreases reflection/albedo - I think!.

Very interesting blog and contributions - a bit compulsive - I kind of get the sense of watching a car crash about to happen. I just hope that the more evidence for what is happening then the more likelyhood that denial interests will be challenged.


Artful Dodger

... or how about surfing a hurricane party? ;^)

Artful Dodger

Century Break for Aug 1: SIE change is -101,562 km^2 (preliminary).

Meanwhile, the IJIS Area graph has resumed its downward track.



I am really enjoying this forum. What happens after September 21st? Melt season in the arctic is over but the melt season will be on for us downunder! Even though Antarctica is nowhere as exciting as the Nort Pole , I think there is a need for a lot more resources about what's happening to the sea ice there.
AD are you interested?

Artful Dodger

Phil/Neven, I'm willing to BABYSIT so we can keep this blog open! how many kids is you have again, Nevi?


@ Artful Dodger | August 02, 2010 at 05:17
I was on another thread, didn't see your post...

Yes, quite the extent drop today.
The new race might be with 2008, but 2010 enters this race with a 280,469 sq km lead:

The IJIS area graph stalled in its decline for July 30 and 31 - I guess August 1 is halfway towards resuming its old slope - the next day or so will tell.


Phil, I'm not giving it much thought at the moment, trying to enjoy the here and now. If the blog would get continued, it would need to be upgraded for various things such as making use of the sidebar, plugins for widgets etc, etc. For this I would like to share the blog with 1 or 2 other persons. It's become easier, but it's still a lot of work. Well worth it though. But stopping something at its climax also has its charm, I think. :-)

We'll discuss it come September.

The new race might be with 2008, but 2010 enters this race with a 280,469 sq km lead:

Anu, make that a 383,125 square km lead. Or did I do something wrong with that effing leap year again?

Lord Soth

Re: Phil263: South Pole Melt

The dynamics of the South Pole sea ice melt is totally different and mostly irrelevent.

Its basically land fast ice stuck to a very cold continental mass. The year to year variance in South Pole ice melt it totally related to local weather conditions in Antarctica.

But someday between tomorrow and the next 10000 years, the West Antarctic Ice Sheet will begin to fracture and drift away, and Antarctica will look like the following:


We can go from now to what that picture looks like in a hundred years, if the WAIS fractures.

The most interest in Antarctica is the Pine Island Glacier; as this is the weak underbelly of the WAIS.

Artful Dodger

Aug 1 is officially a Century Break: -102,500 km^2 change in SIE (revision is -938 km^2).


What a surprise. CT area anomaly has gone up 120K, though.

Artful Dodger

Neven, that CT's data is currently timestamped '2010.5808'. That works out to day 212, or 08/01/2010 00:00:00 (the first moment of the day on Aug 1st).

The new IJIS data is for the end of day on Aug 1st, so I hope you're allowing for this potential 'off-by-one' error when you did your AREA/EXTENT charts. Cheers!


This should be ok then, since Jaxa is a two day average.

I remember an old discussion if there was an delay in the CT numbers, but judging from this and the previous CT increase ,20-21 July, they are well correlated to especially large pink clouds. When they disperse we get the area decrease back with a vengence like the 22/7.

Jon Torrance


Really, let's not ask the impossible of Neven. Considering the IJIS data is a 2 day moving average and therefore presumably based on 48 hours of observations, I don't think it matters whether the CT data is really a snapshot of the first second of each day or something a bit fuzzier. The large scale trends will come out just fine.


Not only did the CT area anomaly go up 120K, the CT area went up about 55K. Noise or evidence, as I suggested a day or three ago, that some areas have been compacted from under whatever concentration threshold CT uses to above it? No easy way to tell that I can see - anyone feel like doing Goddard-style pixel counting on the CT (or U. Bremen) concentration maps?

Jon Torrance

Or siili's idea - remind me, what weather data visualisation does the description "especially large pink clouds" apply to? If it's provided sufficiently close to real time to let us predict an uptick in the CT area data, that would provide decently strong evidence that siili is on to something.


Take yesterdays images as an exampel, where we have a large increase in concentration south of Wrangel island cleraly visible on both Bremen and CT maps.
The large area with very dispersed ice, completely goes up in smoke ;-)

Also there is a bad stitching to the strip towards Bering strait, another indication of intefering weather.. Compare this to the same area in the 3-6-7 MODIS image you will find my pink cloud.


Pink clouds... You guys are drinking too much beer. :-p

Really, let's not ask the impossible of Neven.

Let's not even ask the possible of Neven. You haven been forcefully nudging my learning curve enough as it is. Give me a week to catch my breath. ;-)

L. Hamilton

Graph du jour:


After earlier posting several black-box quadratic models that naively fit trends in ice extent and area, I got curious to see whether substantive predictors could fare any better. Some first impressions, from experiments with RSS and GISS temperature indexes:
- Temperature anomaly indexes generally have the expected relationship to ice extent and area, across all seasons.
- Extent appears somewhat more responsive to temperature than area is.
- The high and mid-north latitude RSS indexes (lower troposphere) do not seem to predict Arctic ice extent much better than the general northern hemisphere RSS index.
- GISTEMP northern hemisphere fits somewhat better than RSS northern hemisphere.
- Extent-temperature correlations are highest for the month of June. Including May temperatures does not much improve prediction of June ice extent.

All this led to the graph above.

Artful Dodger

Larry, you've definitely located a signal in there, with R^2 = 0.75

GISTEMP is where really 'where the rubber meets the road' for sea ice, but can we predict it? Else, it just becomes a co-linear variable with SIE.

Now if we had more independent variables with good correlation you could do a multiple regression...

L. Hamilton

Neven, one constraint on modeling ambition is that we start out with only 30-some degrees of freedom (the satellite era) for annual or seasonal measures of Arctic ice. That makes it tough to resolve separate effects from multiple, correlated predictors. A worthwhile challenge, though.


@ Neven | August 02, 2010 at 10:47

Anu, make that a 383,125 square km lead. Or did I do something wrong with that effing leap year again?

The calendar year is shorter than a solar year by about 6 hours - the leap year day is to make the calendar dates line up with astronomical events like the summer solstice.

In 2008, they stuck in Feb 29 to line up the dates again - August 1 2008 should be pretty well aligned with the summer solstice. Now, in 2010, we are out of sync by about 12 hours.
Close enough - I'm not going to worry about leap year off-by-one-half-day problems. (I guess someday the data will be timely and precise enough that we can compare the ice at 726 hours from summer solstice in year 2026 to 726 hours from summer solstice in year 2032, but that day is not yet here)

IJIS did correct down today - so, the lead has increased to 281,407 sq km.

Grab your pink clouds, we're off to the races...


We should probably do a forensic investigation on the teethmarks of that big bite somebody took out of the Beaufort Sea today, looks like a couple of Banks Islands at least.

Gas Glo

>"GISTEMP is where really 'where the rubber meets the road' for sea ice, but can we predict it? Else, it just becomes a co-linear variable with SIE."

Multiple linear regression of temp with time and various appropriate lags behind an ENSO index? I think MEI is likely to work best.

Would it be better to work on north polar plus northern extratropics regions rather than northern hemisphere or just north polar region? UAH provides such splits for satellite t2lt data though I would suspect that SST may be better than satellite temps.

L. Hamilton

Gas Glo, as background before drawing my graph upthread, I tried out each of the RSS latitude splits. I expected their high-north temps would best predict sea ice, but they don't. Below are some example correlations, *June only,* between NSIDC extent and area (adjusted) and NH GIS anomaly ("nhgis"), NH RSS anomaly ("nhemi"), and each of the other RSS latitude-band anomalies.

Although RSS high-north ("northern" here) anomalies correlate r = -.71 with ice extent and r = -78 with ice area, these offer no improvement over the RSS northern-hemisphere ("nhemi") correlations (-.74 and -.77).

And GISTEMP June northern-hemisphere anomaly ("nhgis") correlates more strongly with June ice extent or area (r = -.87 or -.88) than do any of the satellite-based RSS measures.

-------------| extent area
extent | 1.0000
area2 | 0.9361 1.0000
nhgis | -0.8678 -0.8766
nhemi | -0.7387 -0.7666
northern | -0.7136 -0.7796
globe | -0.6731 -0.6668
tropics | -0.4578 -0.4337
normid | -0.7350 -0.7780
soumid | -0.2885 -0.2204
southern | 0.0064 0.0096
tusa48 | -0.3588 -0.3177
shemi | -0.4352 -0.3796

Gas Glo

Thanks Larry.

Neven, I had a go at starting a public spreadsheet for doing graphs and calculations:


But that just isn't workable because it is far too unresponsive.

Jon Torrance

Offhand reactions to Larry's recent work:

- Score one for GISTEMP and extrapolating temperature over long distances between observing stations.
- It shouldn't be surprising that the average surface temperature of the whole northern hemisphere would influence the arctic sea ice even more than air temperatures just above the ice since surely it's a measure that would tend to move in tandem with the heat content of the water at and just below the ocean surface.


@ siili | August 02, 2010 at 18:55

We should probably do a forensic investigation on the teethmarks of that big bite somebody took out of the Beaufort Sea today, looks like a couple of Banks Islands at least.

What map were you looking at ? (can you give the URL ?)
I didn't see it till hours later (and didn't have time to ask till now...)



For some on the ground observations the blog of the Norwegians who want to circumnavigate the Arctic this season is becoming increasingly interesting:

Even if it is a detour, we decided to go through the Yugorsky Strait, because there is still a fair bit of loose, drifting ice on the east side of Kara Strait, which is north of Vaygach Island. Actually, the nuclear icebreaker “Vaigach” is working its way there to escort vessels in and out of that strait.

Yugorsky Shar, as it’s also called, is a narrower strait – and as far as we know the strait itself it is free of ice. We are now at the western end. We expect to meet loose drift-ice when we enter Kara Sea on the eastern side of this strait, so that will be exciting.


Newbie Hello:

Nice to find a group as fascinated with the daily "squiggly lines" as I am.

To satisfy a curiosity I made a graph (stretched way out) to make the daily changes almost a straight line (2880x640) about 40Kb. If you squint and not too color blind it's at


Welcome, Jack Taylor! Nice graph.

L. Hamilton

Anu, that melting multi-year ice along Greenland's NE coast has been dramatic to watch this year. The NSIDC report May 4, before much melting occurred, forecast what would happen:

"Ice extent for April 2010 was the largest for that month in the past decade. At the same time, changing wind patterns have caused older, thicker ice to move south along Greenland’s east coast, where it will likely melt during the summer."

Artful Dodger

Aug 2 SIE change is -88,437. Expect to see a larger drop tomorrow due to 2-day averaging at IJIS.


A preliminary melt of 88,437 square km. Not bad.


Anybody knows what happened with the Uni Bremen site? I haven't been able to access for a while....


The 2010 - 2007 difference has dropped from 607,000 of 7/29 to 514,000 on 8/2, so I think the race is still on, and it's a marathon at that.

The key is daily changes. With 50+ melt days left, 2010 only needs to average 10,000 km^2 more than 2007 to exceed 2007.

I'm tracking the 2010 and 2007 daily losses here:

2007 had 4 straight -100,000 km^2, then the 2007 loss rates drops off. The race is not over yet.,


Phil, this happened a few weeks ago as well. Uni Bremen was offline for quite a while then. Perhaps the blinking has something to do with it, but it's probably hardware/server issues.


Kelly/chartsgraphs, I could be wrong but if atmospheric conditions don't switch to some highs dominating the Arctic real soon it will be difficult enough for 2010 to end up under 5 million square km.


The new PIOMAS (2010-07-31) chart is out - anybody care to analyze it to see what the average ice pack thickness is right now ?



The ice thickness is about 4.4 [1000 km**3]- very low indeed. This is a bit rough, as I estimate the daily average from the graph. If anyone knows where the actual data is, that helps. In any case we're dealing with model estimates.


Ok, what do we have: Volume anomaly about -10
Average value, assuming dots are month centred about 16.5
Area from Cryo 4.4
that should give (16.5-10)/4.4=1.5 m mean thickness

Zhangs July update of the model still also looks in good shape with a very good fidelity to the real world, if anything to much ice on the Siberian side. He should have a August update out in a few days.

My bitemarks was from Bremens preliminary map from yesterday afternoon, but todays final Hamburg version is very similar.

Seaice.dk is also up and running again, but it looks that they have mixed up their timestamps on their images and are a few days behind.


That was volume, average thickness is 4.4/6.7, about 0.657 meter



I think that that is volume, and I think that that is too low. Based on my reading of the graph, I think that the volume is 6.2 thousand cubic kilometres - which is still disastrously low.

Thickness would thus seem to be around 1.4 metres.


Clearly the average for July end is about 14.4 (maybe 14.5). Subtract 10 and you get 4.4. The area to use is probably 6.9 (31 Jul), you get 4.4/6.9, about 0.638.

Artful Dodger

Gili1, are you using Sea Ice Extent in your computation? As you know, Extent measures can include up to 85% open ocean (thickness=0) which will skew the average thickness estimate.

A more appropriate measure is Sea Ice Area, since this measure directly reflects sea ice concentration. The CT Sea Ice Area for July 31 was 4,469,000 km^2


This is beginning to get a bit tiresome. I thought that everybody agreed, and please read the first paragraph of the PIOMAS page to find that the dots mark middle of the month.

Then to compare volume with extent, you measure a lot of water with not much do do with the remaining ice, apart from melting it.


I guess it's best to ask the PIOMAS guys for their number, as it really matters if you interpret the daily average graph value as start-of-month or month average (mid-month).

Artful Dodger

siilii, this is a different issue to PIOMAS volume. The issue is which 2-D measure is appropriate, volume or extent. Perhaps if you are becoming tired, you could wait before jumping in on a question directed to Gili1?


Dodger, Gili, Evil and everybody else i might have offended. Please forgive my rudeness.
I will instead continue to read the hilarious comments on the last update from the dark sides version of this blog.


Dodger, you're right, should use the area number, not extent.

Artful Dodger

Please, siili don't be deterred. We love your imput here and this place'd be poorer without you. I know I have to shake my head daily ;^)

Cheers, Mate!


Me deterred? You must be joking mate! I like to wear my ignorance like a badge of honour, since i am generally interested to learn more about how this world works, and maybe i get a bit impatient sometimes. You and many others at this place are good teachers, and i sincerley would like to get answers to the questions i ask, like links to the data of this years IMBs, iceage maps and yes the pink clouds.

On the funny side, look at the quadrupole in the preasure pattern
http://www.meteo.uni-koeln.de/meteo.php?show=En_We_We some mighty winds blowing there. If they havn't updated yet, is there an archive there as well?

Artful Dodger

Yeah, I saw that. Cool, wot? I haven't seen a quadrupole since the 'Nineties, with a Supercomputer simulation of magnetohydrodynamics for propulsion in the Ionosphere... Now I can hear all my Aussie friends saying 'Pig's Arse!' HAH! s'truth


Just one more arctic observation, http://iabp.apl.washington.edu/maps_daily_tracknsidc.html ,looks like there are going to be quite a few new swimming buoys in the not to distant future. Maybe they can go salmon like the one in Norway and boldly explore mighty rivers?


My contribution to the "annoyingly wrong calculation thread":

I read the 31 July PIOMAS average volume (half-way between the July and August points) as 16150 km^3.
The PIOMAS anomaly (31 July on the lastest graph) sits at ~-10100 km^3.
That leaves a net volume of 6050 km^3
My estimate of area from the IJIS area graph (area should definitely be used, but I'm not so happy with CT figures) for 31 July is 5582000 km^2

Which yields an average thickness of 1.08 metres (insert caveats about computing while cooking - again!).


Lovely, this thread is shaping up to an interesting, enlightening discussion. And one sure gets hungry from all talk about food.

But the difference between CT and Jaxa area is really huge, and it would be nice to understand what is causing it.
The hole at the pole is one thing, but that should only be 0.31M(km^2) right? So that gives CT at 4.41+0.31=4.72
Then we have Jaxas two days averaging, but that alone should only give us half a days decline of say 0.1M(km^2)?
So what is the root of the remaining difference of about 0.8M(km^2)? Pink ice? Off to the kitchen to feed the hungry clan!

Peter Ellis

I don't think it has anything to do with the hole at the pole - IJIS and CT annual *maxima* are pretty close to one another, so however they're handling the pole hole, they both must be doing the same thing, which is what matters.

Peter Ellis

Looking at the maps, the CT concentration map and the Bremen one are not all that similar: look especially at the region from 135-160 degrees West, 75-80 degrees North. Bremen has that quite a lot more concentrated than CT. If IJIS bases its data on something closer to the Bremen map, that might explain the differences.

It would be good to have a side-by-side comparison of Bremen and CT concentration maps, appropriately rotated, perhaps compare with MODIS images. Looking at the latest Arctic mosaic, I think CT looks more likely to be correct.

Note also that the IJIS area graph stalled at the start of July - isn't that about when the Arctic started clouding over? I think they may be getting misled by clouds.

Artful Dodger

I think the sensor IJIS uses sees right through Cloud. Do both groups use the AMSR-E sensor?

There may also be a different cutoff threshold for sea ice concentration ie: one group may use 15% while the other uses 30%. I know this sounds funny because we are talking about Sea Ice Area, but as BFraser said | August 02, 2010 at 01:25 in the "Area vs Extent" thread:

"they apply a "filter" of 15% to their area computation, so the 10% cell in your example would count as 0, rather than 6.25 km^2."

So if one group uses 30%, their Area numbers will be lower than a group which uses a 15% threshold.

Can anybody shed the light?


They use the same radarfeed from AMSR-E, look at the bad sticking south of Wrangel, definitely the same strips. I also think that disregarding the obvious differences in colourscheme CT http://arctic.atmos.uiuc.edu/cryosphere/NEWIMAGES/arctic.seaice.bandw.000.png
and Bremen http://www.iup.uni-bremen.de:8084/amsredata/asi_daygrid_swath/l1a/n6250/2010/aug/asi-n6250-20100801-v5_visual.png are very similar.

Bremen has also started to produce it's own extent graph, and that is quite different from Jaxa, so i think that of the trio JAXA is the odd man out.

Maybe they can be persuaded to start producing an area map as well, or even better releasing their measurments in numerical form.

Artful Dodger

UniBremen has posted a steep drop in SIE for Aug 3 on their graph. By eyeball, the slope looks like the steepest 1-day drop of the year.


May one be bold and expect something similar from CT in their update, coming out in about two hours together with their new image for 2/8?

I also like CT's combination of ice and optical background, the phytoplankton just can't wait to follow the ice.

Artful Dodger

CT's tabular data (incl. graphs but not maps) is currently on Day 212 (end of day, July 31). So their next graph/table update should be end of Aug 1.

The UniBremen and Uni-Hamburg maps first showed the 'Big-Beaufort-Breakdown' on the afternoon of Aug 2.

So no, I don't think the next CT update will show the 'BBB'. But it should be on the 2nd next update.


Ok, Dodger is the master! Only a small drop in area, but the image shows the bite, so is this confirmation that the numbers are one day behind the images as the timestamp suggest?

Artful Dodger

Please, call be 'Lodger'! The Aug 2 SIE revision is just 156 km^2.

L. Hamilton

One more note regarding my "June is more predictable than September" discovery (which probably was discovered long ago by others):

GISTEMP annual zonal anomalies for 24N-90N, or for 64N-90N, both predict September mean ice area (or ice extent, or area & extent in other months). But the annual anomalies, like the monthly anomalies I graphed earlier, work best for predicting June ice area or extent.

In this scatterplot matrix, you can see more random scatter for September than for June, although both months trend down with rising temperatures as we'd expect.

And for either month, the 24N-90N temperatures prove to be *slightly* better predictors than 64N-90N.



Bremen appears to have taken a break again, but the Danes are up and running and their preliminary image of today looks interesting: http://www.seaice.dk/iwicos/latest/amsr.n.comb.20100803.gif ,they use a rather agressive colourscheme, so handle with care. They also update their newest map with an about bihourly cadense untill it is filled which should happen in about two updates.

L. Hamilton

Around 80% of the variance in June ice extent is explained (statistically) by March, April, May and June NH tempertature: R-square equals .81, or .78 adjusted for degrees of freedom.

September ice extent is not as well predicted from June through September NH temperature. R-square equals .63, or .57 adjusted for degrees of freedom.

Similar results occur with area for these months.


L Hamilton

Just checking to see if you took the non stationary aspect of the data series into account. I'm still trying to understand spurious correlations of time series, so I may be over reacting,

This post provides some background on problems with ordinary least squares analysis of time series.


Kelly O'Day


Jaxa figures show an SIE loss of -79,844 (preliminary) for August 3, but I also note that the SIA curve has taken a sharp swerve to the right.

L. Hamilton

Kelly, the original time series (e.g., June extent) are autocorrelated, but residuals from the models are not -- they test as white noise. ARMAX versions of the regressions show nonsignificant autoregressive or moving average terms, so OLS seemed reasonable for my exploratory purposes here.


Steady declines in the JAXA extents as reported before with losses of about 80 100 90 and 80 k(km^2) the last four days, wheras CT area is more jumpy with 70 -60 50 and ? where the - sign marks an increase in area and the ? that no area has been reported yet.

So, any brave pixelcounters (AD?) ready to give an estimate of what ? will be today?

Still no Bremen map, but Hamburg looks like this ftp://ftp-projects.zmaw.de/seaice/NEAR_REAL_TIME/201008/Arc/20100803.png

Over at the dark mirror a discussion is brewing between NSIDC and the other side about what is causing the last days changes. Maybe we should invite them over here as well for some professional input?


A perhaps enlightening excerpt from the other side gives a comparison between NSIDCs raw one day numbers (usually the use a sort of 5 day moving average in their graph) and JAXAs two day average. I may be a day off in the matching, but i think this is the correct ordering.

NSIDC JAXA Difference
29/7 7.16
7.11 7.01 +0.11
30/7 7.06
6.97 6.92 +0.05
31/7 6.88
6.82 6.82 0.00
1/8 6.77
6.66 6.73 -0.07
2/8 6.56


Ok, sorry about the formating, but the numbers give: date, raw number NSIDC, average for NSIDC (combined with next day), JAXA ,difference like 29/7 7.16 7.11(7.06) 7.01 +0.11

Artful Dodger

Is anyone looking at winds in the Beaufort Sea? Over the past day, there have been fresh Westerlies at 16 knots (30 kph) extending from 180W to 125W long. between 78N and 81N lat.

This is very close to the area where Uni-Hamburg recently shows sea ice concentrations drop to between 0-50 per cent over a ~420,000 km^2 area.

Mechanical mixing of the ocean surface layer with thin sea ice is enough to cause the ice to melt. In deep open Ocean, 16 knot winds produce 6–9 ft (2–3 m) waves of some length, and many white horses (breaking waves).

With the amount of open water in this 1000 km wide area, and the wind blowing down the its long axis, I wonder if we are seeing the loss of sea ice due to wave action?

Does anyone have access to satellite wave height data for the Beaufort Sea/Central Basin?


Good point, at least the details of the ice-edge correlate well between 2-3 so i don't think that we are "seeing" the weather, like the long thin area of "ice" next to the Kuril islands at the top of the image. To bad there are so much clouds in the area so visible images are hard to interpret correctly and also the microwaves are affected causing the blinking in iceconcentration.

Artful Dodger

siili : I think the 'blinking' is caused by the sea ice concentration teetering around the 15% cutoff line.

Did you notice in the Uni-Hamburg images the large lead in the Beaufort Sea extending to 135W 75N? It was marked 'open water' in the Aug 2 map, and 'low concentration' on the Aug 3 map? It will melt out for good in the next few days.

I predict that the 420,000 km^2 area from 180w-135w, 75n-79n is going to collapse in the next 1-2 weeks. I also think all the sea ice South of 79N in the Western Arctic is at risk this season (this is an ~1.25M km^2 area). Only the small arc E of 135W lat. (W of Prince Patrick Is.) looks like it might have a chance to survive.


What a curious old melt season. June goes balastic, July sopophoric and now August seems to be starting with a bit of a kick.


AD, there are two types of blinking, yours are very clearly in effect in the way you say, the one i have been talking about is the "high end" blinking between 100% and lower concentration which appears to be caused by "pink clouds" which e.g. can be seen between 1-2/8 images south of Wrangel Island and many other places.

There are so many changes going on so all i can dare to predict is that it is going to be eventfull times ahead, and Zhangs model is looking more and more conservative.

And Bremen is up again, i think their images are easier to look at. Their extent graphs in breaking up after tuching 2007.

There is very little old ice left, even the tounge sticking out above Banks Island is starting to break up, after that who knows?

Artful Dodger

silli: I agree that the 'sea ice' near the Kuril Islands (south of Kamchatka, Russia) are artifacts in the Uni-Hamburg maps. The NOAA SST NE map indicates water temps there are 6-8C, so it's unlikely any fastice has set sail and survived the Pacific. Who was talking about ice floes as AC carriers? Could be a surprise attack from Shangri La ;^)


http://en.wikipedia.org/wiki/Project_Habakkuk ,but i find the true story about ice-island T-3 and Russian station NP-22 even more exiting.

Any thoughts on my previous ramblings?


You flatterer!

But have you noticed that todays MODIS actually have some good holes through the clouds, showing ice streatching almost all the way from Wrangel to Alaska, not showing on Bremen, don't tell the other side! At least part of it is faintly seen on todays
http://www.seaice.dk/iwicos/latest/amsr.n.comb.20100804.gif and even better on http://www.seaice.dk/iwicos/latest/amsr.n.ice.20100804.gif

If you know the difference, please educate me, my master.

Artful Dodger

Yeah, I peeked through the clouds in MODIS today, and like Aristotle, saw what I expected to see. I guess I'll wait for the Fall... David Barber/Amundsen + CryoSat-2 should be the final word on this melt season.


Good call, if there is much left to study that is. I saw the Chinese are up there as well. Your attempt looking into the CT crystallball, was not that much better than mine,
so let's just observe the fall and ramble on.


In my new role as passive observer i noticed that floe outside Alaska, if it keeps up that pace it should make the circle a lot faster than those Norse cheeters using sails.

L. Hamilton

So un-smoothed NSIDC dailies show century breaks one after another the past few days, even 216k from 8/1 to 8/2. A much faster decline than JAXA, which explains why NSIDC (unlike JAXA) shows 2010 graphically closing the gap with 2007.

year, month, day, jaxa, jaxadif, nsidc, nsidcdif
2010, 7, 29, 7086719, -45312, 7160520, .
2010, 7, 30, 7008750, -77969, 7059950, -100570
2010, 7, 31, 6922031, -86719, 6881230, -178720
2010, 8, 1, 6819531, -102500, 6772950 ,-108280
2010, 8, 2, 6731250, -88281, 6556750, -216200

Peter Ellis

There are two possible reasons for the difference between IJIS and NSIDC: the data source (IJIS uses the AMSR-E satellite while NSIDC uses the SSM / I satellite) and the algorithm used to convert satellite readings into ice extent (i.e. filtering out land contamination, surface choppiness, clouds, melt ponds etc.

The ROOS site also uses SSM / I data, and they too show a rapid decline in the last couple of weeks, so this at least looks to be a difference in the readings between the two satellites.

On the other hand, ROOS has 2010 well above 2007, which I assume is due to different data processing since they use the same source. Similarly, IJIS has 2010 well above 2007, while Bremen has them almost touching, despite the fact that IJIS and Bremen both use AMSR-E data.


Thanks for the help with the formating, but if you take into account that JAXA is a two day average, and even more that they use different detectors, JAXA AMSRE-E and NSIDC the older SSM/I, the likeness is rather good i think, especially in the very fast changing arctic ice theese days.

Lord Soth

Wow, I havent looked at the NSIDC for about two weeks. At the current rate, the 2010 line will converge with the 2007 line.

The one thing I have learned, is that there is quite a bit of variations between the varous graphs, from data that only comes from a few sources.


If I've understood correctly, what we alarmists need, is SSM/I data converted by the Uni Bremen algorithm. Just kidding... ;-)

My apologies for the slowdown in posts and comments. That area/extent thing put a big burden on me, as I have been determined to produce my own graphs in OpenOffice and learning new things take time, one commodity I don't have much of, as it is the busy season in my line of work as well. I will post the new graphs some time later today and a new SIE update tomorrow.

Nick Barnes

Is any of the code for any of the charts published?

Artful Dodger

Well, we have are first serial spammer...
Do not follow the link to runescape.


Thanks, Lodger. Deleted. It was the third serial spammer by the way.

Artful Dodger

Well, I hear it's a 1200 year cycle ;^)
Feel free to delete my comment as well...
On with the Melt.

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