Arctic Ice Extent Discrepancy: NSIDC versus Cryosphere Today

There is a good running Sea Ice discussion at CA: http://www.climateaudit.org/?p=3336
I also like this link: http://www.ijis.iarc.uaf.edu/en/home/seaice_extent.htm

There is another website /www.ijis.iarc.uaf.edu/en/home/seaice_extent.htm from Japan which gives ice coverages. This curve looks similar to NSIDC curve. However, for the last few days, the difference to the 2007 curve appears to be a bit larger.
Keep in mind that NSIDC gives an area with at least 15 % ‘sea ice’, while the http://arctic.atmos.uiuc.edu/cryosphere/ website probably only shows areas of full ice coverage. Consequently, the UIUC ice area should be considerably smaller than the (presently) 6.07 million km2 of IJIS and a very similar number of NSIDC. Maybe, you could check this from your pixel data.

See detailed comparative graphs by
Aaron Wells’ Sea Ice Stretch Run #2 Comment 612

The final revision is in for yesterday’s data. Here are my graphs for yesterday’s ice extent data, based upon IARC/JAXA Ice Extent data.
Daily Arctic Sea Ice Extent Reduction Day 227
Arctic Sea Ice Extent Day 227
2008 Arctic Ice Extent Differences From Previous Years Day 227

These look much closer to the Cyrosphere comparison above.

Ah, El Reg’s Mr. Goddard! Who is he, one may ask, as a “Steven Goddard” doesn’t even exist on the interweb before April 2008.
Yet the Register chose him to continously enlighten us with his views on climate change without criticism. Is he perhaps having some connection to IT professional John Atkinson, who was allowed to post a similarly incoherent article on the Register? Is it just a coincidence that someone using the moniker “John A.” pops up regularly on the blog of Stephen McIntyre, who works in IT as well and is known for making powerful oil interests heard?

I have had problems in the past with NSIDC. I was emailing back and forth last February with an NSIDC expert on a technical item (the age of an in-situe formed ice shelf that ‘suddenly’ broke free, like they ever slowly break off, sheesh). I contended that it was a relevant problem of interest since if there was no ice there during the MWP and it formed during the LIA, what’s the problem.
Eventually the NSIDC expert sent me to RealClimate.org. Go figure. He and his lunchroom buddies have already imbibed in green fool-aide.

Sorry Anthony – blame that one on a senior moment

Mr. Conrad –
Regardless of who Goddard is or is not – can you refute the substance of his report? To say that he never existed before 2007 on the internet is absurd. Did you personally look through every Google hit?

Mr Conrad,
Implying that anyone who writes articles that debunk global warming is funded by “powerful oil interests” is kind of silly. I believe the ratio of nutbar pro-warming funding to oil interest funding is 1000 to 1.
Al Gore himself has a 300 million budget for just this year to propagandize. (Admittedly most of that goes to fund the energy budget of his home and private jets and yacht).
Tens of thousands of scientists and journalists rely on AGW funding just to pay their salary. They are bought and paid for.

Joseph Conrad –
Interesting suggestion that Steven Goddard and John A. may be the same person. If true, the Goddard penname is likely a mischevious tweak of GISS. So what? Please address the facts of the article. What information that is presented as fact is untrue?
Calling people big oil shills just doesn’t work anymore, if it ever did. It resonates in the AGW echo chamber, but the public isn’t buying it. The public view Exxon as a company that provides a product that makes their life easier, not as one hell-bent on destroying Mother Earth. Public opinion polls showing greater than 65% approval for more domestic drilling make this point crystal clear.
Making your case to the masses, persuading them to elect like-minded leaders, and implementing coices involving sacrifice is hard work. Ad homs are so much easier, but not very effective.

Alan Millar (11:43:57)
I’ve copied and saved every daily graph from NSIDC since July 31, except for Aug 1. (Also, the Aug 9 graph wasn’t updated; it actually went from Aug 8 to Aug 10}. The axes on the graph change on Aug 5, but it still shows regular ‘adjustments’ to the line’s slope every 2-3 days, sometimes up, sometimes down. The biggest change happened Aug 1, when several days’ worth of an uptick were changed downward. So far, this year’s line is generally running parallel to last year’s line.
So far I’m up to 14 daily .bmp files totalling 34MB.
Anthony – You’ve got my e-mail if you want me to send you these images for use here.

Anthony
Well, I learn something new everyday. I turned those 14 .bmp files with 34MB total size into 14 .gif files of a bit over 320KB total size. Then my email automatically combined them into a self-extracting zip file. Hope you can make something out of them. Watching the battles of upticks/downticks is interesting.

counters (15:19:01) wrote : “This is ridiculous. There isn’t anything scientific about this piece at all. Let me know when you’ve got an analysis that goes a tiny bit further than counting pixels on pictures… perhaps a mathematical analysis of recent years’ trends and their deviations from the mean ones, or an analysis of the methodology used by NSIDC to calculate their data? I’m not interestied in how many pixels are a different color on two pictures.”
counters, your knowledge of pixels seems to match you knowledge of climate change. You have a lot of bluster, but few facts. I use pixels day in and day out and believe me, you can obtain an accurate representation of a given area through a “head” count.
Jack Koenig, Editor
The Mysterious Climate Project
http://www.climateclinic.com

One of my professional activities is doing DNA sequencing using electron microscope images. Pixels of various brightnesses are counted and analyzed, to determine the location and type of chemicals present.
When dealing with micro or macro imagery, pixel colors and brightnesses are normally the only tool you have to work with. How else would you take satellite imagery and turn it into sea ice information?
Some of the vocal critics might want to slow down their mouths, and do a bit of thinking (for a change.)

Jeff C,
I noticed that too. You would think there is a full moon the way they all came out on this one.

Counters
I am a skeptic but, I do enjoy someone spicing up the discussion here. Sorry to see you knocked silly on the pixel comment. It is Friday and you might be a bit tired.
My question is I don’t see a comparison with the NIC IMS product. http://www.natice.noaa.gov/ims/
It shows the ice extend being much greater than the cryosphere pictures. Especially in the area around Eastern Siberia across to Canada. It shows much more ice. I have been looking at those pictures and comparing them. For example, compare the ice around Wrangel Island north of Eastern Siberia. In the latest NIC picture it is completely surrounded by ice. In Cryosphere it is ice free.
I assume they use different thresholds.

How does the location of an IP address guarantee that the poster actually is at that location? What happens when someone is logged in remotely? The Register happens to have an office in Edinburgh.

Jeff –
“The University of Illinois webpage cited in the article look has graphs of the sea ice area and sea ice area anomalies and see that the long-term trend is towards smaller sea ice extent year-round. And that’s what’s really important.”
And I would add The South pole sea ice trend is toward a larger sea ice extent/area. Why isn’t that what’s really important?
Either way, it’s beside the point. Goddard’s article states that he sees a discrepancy between what is reported and what he observes. I commend you for checking into it yourself and more of us should be doing the same. Perhaps the difference lies in the definition of extent vs. area or different data sources. However the info from the Cryrosphere today does seem at odds with the NSIDC chart.
This is an article in an online journal that raises a question, not some “peer-reviewed” paper claining to settle the science. There are many smart people that are regular commenters here. Perhaps we can all do some auditing ourselves and figure it out.

NSIDC is reporting sea ice EXTENT, not sea ice AREA, so in fact it’s quite possible that their 10 percent figure is correct.
Jeff’s explanation seems highly plausible. Note this graph, from Cryosphere Today, which shows current area at c.4 million km2 –
http://arctic.atmos.uiuc.edu/cryosphere/IMAGES/current.365.jpg
By contrast, NSIDC’s graph of extent shows in excess of 6 million km2:
http://nsidc.org/images/arcticseaicenews/20080811_Figure2.png
So, Mr Goddard’s comparison is based on a basic error. Can we look forward to you clarifying this on your site, Mr Goddard?

The answer is the same reason all the TV weathermaps now show yellow for 60 degree temperatures — The map now looks all ” reddish and hot” to the naive viewer. If they had a disclaimer about the scale change, who would care — But they don’t. It’s clearly designed with a single intent in mind, try and convince people it’s getting hotter. Why they would want do this, I have no idea.
The ‘icy truth’ lies in the data processing software, I see no reason why the science group would not openly provide source code, and access to the raw pixel data. They could at least tell us what the change log looks like and explain any changes to the code. That would be good “open” science, instead of wasting people’s time trying to figure it out.
Something which is designed to fix just these types of problems … The U.S. has a law, Data Quality Act of 2001, this law requires federal agencies to ensure the integrity of the information they use and distribute. It also allows outside parties to petition to force the correction of information they believe is wrong.
The Data Quality Act (DQA) passed through the United States Congress in Section 515 of the Consolidated Appropriations Act, 2001 (Pub.L. 106-554). The Government Accountability Office calls it the Information Quality Act, while others call it the Data Quality Act.
This public law also applies to all Federal grantees, and all Federally funded groups.

counters
‘If someone has serious concern with the methodology used to produce a major, then the appropriate channels of communication to be used are the scientific ones.’
And I’m sure you were the first one to complain to IPCC about HOW new& improved “hockey stick” made it into their report.

I overlaid the iceextent line from 08/11 onto the earlier chart from 07/14 and found the line has been changed from even the stating point at the beginning of May. I have been following the extent all summer and have seen discussion re: the slope change last month, but more than that has been changed. I may be off a pixel or two, but believe this is pretty accurate:
http://i33.tinypic.com/2rgyzvm.jpg

Phil – thanks for clarifying the projection used in the data. I assumed that the projection used standard k-space, i.e. the sphere tangent to the projection plane at the pole. Now that you bring it up, I do recall the 70 deg comment from the Sea Ice thread at CA.
Your comments may be inconvenient, but they are always thoughtful. It is appreciated.

Can some define what Extent is?
Area is an easy concept for me, Extent seems like it can have many defintions.

Why is there so much attention on the Arctic ice?
For the same reason there was so much attention paid to hurricanes a few years ago and global surface temps before that. The warmist spin machine is desperate to find weather events which can be trumpteted as evidence of catastrophic anthropogenic global warming.
If artic ice rebounds, the COGW will seize on something else.

John McDonald,
Well the NSIDC chart of “Sea Ice Extent” shows “Area of ocean with at least 15% sea ice”.
http://nsidc.org/data/seaice_index/images/daily_images/N_timeseries.png
So extent is a measure of two dimension area.
I don’t know how that would differ from the “atmos” pics of “sea ice concentrations” except they can go lower than 15%. However, I’ve not seen that.
http://arctic.atmos.uiuc.edu/cryosphere/IMAGES/ARCHIVE/20070812.jpg

Steven Goddard,
You say:
The NSIDC 10% figure was taken from their graph issued with the press release, which showed about 6.3 vs 5.7 last year.
So you have eyeballed a graph and think that your guestimate of what it shows will do, when you could use the actual figures, as I have quoted to you? These figures were given in the very same press release from which that graph comes! To repeat:

Arctic sea ice extent on August 10 was 6.54 million square kilometers (2.52 million square miles), a decline of 1 million square kilometers (390,000 square miles) since the beginning of the month. Extent is now within 780,000 square kilometers (300,000 square miles) of last year’s value on the same date and is 1.50 million square kilometers (580,000 square miles) below the 1979 to 2000 average.
http://nsidc.org/arcticseaicenews/index.html
That is 13.5% greater than last year, not 10% as you claimed. This is basic arithmetic! You are significantly in error, and should have the integrity to say so (and given this blog’s proper concern for accuracy, that is an expectation that I suggest should be widespread here).
As for your comment here –

The point of the article is that some of the NSIDC data seems inconsistent with other sources. Some posters here are trying to use NSIDC data to prove NSIDC data – which is obviously a flawed approach.
I presume by ‘other sources’ that you mean the Cryosphere Today site? What, pray, is the source of their data, other than the NSIDC? It is actually you who is trying to use one representation of NSIDC data to question another! The data is the same in both cases!.
Anyone can make a mistake, Mr Goddard, but not everyone has the capacity to admit it, it seems.

rjb,
How can I best put this? Gavin Menzies’ book has been the object of considerable contempt from professional historians. Here’s a basic Wiki article on the subject –
http://en.wikipedia.org/wiki/1421_hypothesis
Of course, the ‘consensus’ of historians might be wrong and Menzies might be right, so you’re welcome to make your own judgment. Lord Monckton has also referenced Menzies in support of his arguments (http://www.telegraph.co.uk/news/uknews/1533290/Climate-chaos-Don%27t-believe-it.html), so I guess you have company if you judge it to be plausible.

Ed,
You are confusing the calculation of ice area with extent. “Extent” treats all regions greater than 15% concentration equally – just as I did.
Phil and Steven Talbot,
If the UIUC maps are correct, there is nothing in my extremely straightforward calculations which could be broken. You seem to be (rather obtusely) attempting to make an argument that the UIUC maps do not correctly map the satellite data on to the polar projection they use. Polar map projections have minimum distortion at the pole, where the land is perpendicular to the viewer.
If you wish to argue that the UIUC maps are incorrect, that might be an interesting topic of discussion. However, you have yet to make any attempt to analyze the methodology I used – which (obviously) assumes the correctness of the maps.
According to the UIUC maps, there has been almost no change in the amount of Arctic ice since August 8. Once again, if you believe those maps are incorrect – please state your case – and I suggest that you take it up with Phil Chapman over at CT.

My Pixel for pixel, calculations for 2005, 2007 and 2008 16´th august:
(ice area 2007 16 aug = 100% )
2007: 100%
2008: 130,3%
2005: 132,7%
I thus get 30,3 % more ice in 2008 than 2007, very close to Mr Goddards 30 % in the article. I can only agree.
Notice also how close 2008 and 2005 are. We are as good as back to 2005 of ice extend.
I then measured the colour nuances, and the area*concentration with 2007 as 100% is :
2007: 100%
2008: 125,0%
2005: 123,6%
So yes, if we incalculate concentrations, 2008 hass not 30 % larger ice extent but ”only” 25% larger ice extend than 2007.
BUT, notice that doing so reveals that there is now more ice in 2008 than there where in 2005…
Another thing, i see here and there in this debate the argument that “NSIDC measures from 15% ice extend”. The thing is, is changes nothing realy. If you compare 2007 with 2008 you use SAME method on both pictures, so if you on BOTH pictures set the limit to 15% this cannot realy change outcome when comparing. Unless of course that one year had extremely much 15% ice. But thats certainly not likely. Check the cryosphere photos and you will see that the big areas are never with low concentration. Such low cincentrations areas of ice mekt away very quickly.
Summa whether you use 15% or 25% changes nothing as long as you use same method on the 2 figures you compare.
Thanks Anthony and Goddard for pointing out this example of distorted data.
Its realy appreaciated by many many many peoble, thankyou.
Cryosphere opereates with concentrations in their graphs, but NSIDC does not, and therefore has no obvious excuse for their underestimation of this years ice extend. It just does not look good as so many many other things in this strange debate.

Phil wrote: “Also the Arctic melt has not ‘nearly stopped’, in fact it’s continuing even faster than last year as shown by Aaron Wells graph from CA: http://img352.imageshack.us/img352/4760/seaiceextentlossratebi9.jpg”
Although you claim the the above, I could not find the link anywhere on climateaudit.org that led to the above. Could you provide a definitive URL on CA that leads to the above?
Thanks!
Jack Koenig, Editor
The Mysterious Climate Project
http://www.climateclinic.com

Steven Goddard,
However, you have yet to make any attempt to analyze the methodology I used – which (obviously) assumes the correctness of the maps.
Sigh. I have demonstrated that your premise of 10% above 2007 (on the basis of the graph) was incorrect, and that the correct figure was 13.5%. Sorry to repeat myself, but you seem to be ignoring this incontrovertible fact.
If your name were Hansen or Mann you would be eviscerated here for making such an error, and for avoiding recognition of it. As it is, you’ve received plenty of messages of ‘support’, since your article seems to fuel the notion of data being cooked. Such is the nature of ‘scepticism’ when one gets a proper look at it, I guess.
But I give up. I didn’t really imagine that you would want to respond to ‘review’ in the way that we (quite rightly) expect scientists to do.

Ed:
“Goddard “error” (i’ll use this euphamism for what he really did) is obvious even to the layman. He totally ignores the scale on the image that shows the various colors represent different levels of sea ice coverage. Red is 60% while purple is 100%. Goddard treats them equally but it is clear from inspection the current year has more area with less % coverage – so goddard’s primitive calculation is bound to be way off.
Its pretty funny how you guys can question the NSDIC’s numbers based on this obviously bogus analysis.”
Perhaps you should take a look at the NSIDC graph header: “Area of ocean with at least 15% sea ice”. Why should Goddard not treat the colors equally?
And why is his a bogus analysis? The Cryosphere maps are or should be based on real data, and that data should match data that other analyses are based. My understanding is that Steven has attempted to pull data out of the Cryosphere map and compare that data to the NSIDC’s, and I see nothing in principle wrong about that concept.

But I have to break my promise already, since I realise I haven’t yet commented on the ‘leap year trick’.
1)Goddard’s premise is that the graph shows 10% greater for the same date. This is false, as I have demonstrated: it is actually 13.5% greater.
2) He then compares this with the ‘map’ renditions which he offsets on the grounds that it’s a leap year. Er, hang on a minute, wasn’t it a leap year when you were looking at the graph???????
3) So, he already i)has a false premise and ii)is not comparing the same days, the difference being in favour of the discrepancy he wants to display!
Frankly, how gullible does one have to be to be classed as a true ‘sceptic’?

Steven,
Are you sure both Cryosphere maps and NSIDC charts use the same data?
Here’s the closest I could come to data sources
http://nsidc.org/data/collections.html
http://arctic.atmos.uiuc.edu/SEAICE/arctic.historical.seaice.doc.txt
Looks like several satellites and other methods are available.

I’m enjoying the irony that GW skeptics have been claiming that the small sea ice extent in 2007 had more to due with an unusual wind regime in the Arctic than with atmospheric warming and now they’re claiming that the greater amount of sea ice in 2008 vs. 2007 proves that the atmosphere isn’t warming.

Jeff — “I’m enjoying the irony that GW skeptics have been claiming that the small sea ice extent in 2007 had more to due with an unusual wind regime in the Arctic than with atmospheric warming and now they’re claiming that the greater amount of sea ice in 2008 vs. 2007 proves that the atmosphere isn’t warming.”
Say, have a nice steaming cup of strawman with that enjoyment — on me!
What people here are responding to is the contention that 2008 was going to OBVIOUSLY be worse than 2007 because all of that 2008 ice was just “first year ice” and not the reliable crusty thick multi-year stuff.
The fact that 2008’s extent isn’t significantly BELOW the 2007 number because of the aforementioned claimed problem of lacking multi-year ice is simply remarkable. So much for the predictions of a worse year. The AGW alarmist’s 2008 predictions weren’t merely just wrong, they were in a class of abysmal prognostication skill that ranks up there with psychics. Witch doctors with chicken guts or orangutans with darts could have fared better.
Oh, and it wasn’t the skeptics claiming things about wind and current. They got that from this outfit called NASA. You may have heard of them.

Here is an update from the Canadian Weather Service showing new ice rapidly forming in the Northwest Passage.
http://awberrimilla.blogspot.com/2008/08/yesterdays-new-ice.html
The Australian sailors attempting to navigate it are having a very difficult time with the ice , snow and cold, and are stuck in a snowstorm with strong northerly winds. Hopefully they won’t become the next victims of Arctic warming media hype.
Jeff,
Your logic is again flawed. NSIDC predicted that Arctic Ice extent would >12% lower than last year. Even using their highly questionable measurements, they are off by at least 25%. The point of the article is that the experts were wrong. The North Pole did not melt and the Arctic is not ice free.
Steve Talbot,
As far as the 10% vs. 13% goes, it sounds like you are making the argument that the NSIDC graph does not accurately represent their published data, showing less ice than there actually is. Regardless, it is irrelevant as the maps now (August 16) show 40% greater ice extent than 2007. The point is that there is a large disconnect between the maps and the graph. I don’t see any significant difference between the NSIDC maps and the UIUC maps, so the mystery remains unsolved.
All of the expert predictions of Arctic ice collapse in 2008 were dead wrong. The NSIDC maps and the NASA satellite images clearly show that there is a lot more ice than last year. If people can’t see that, then it must be because they don’t want to.

Patrick Henry
“The only thing which can increase sea ice area is cold temperatures.”
Wrong. Winds and currents play a role. If you can find a data set showing that temperatures around (as opposed to over the interior) of Antarctica are cooling, please post it. Every one that I’ve seen has shown warming.

From the UIUC data you can add up all polar sea areas covered by ice: I get 3.3 million km2 for August 16, 2008. The IJIS value (15 % ice coverage and more) of 08/16 is 5.91 million km2, a change of 0.07 million km2 from 08/15. The NSIDC value is very similar. The corresponding numbers for 2007 are 0.7 to 0.8 million km2 lower, both for IJIJ, NSIDC and for UIUC (the latter is estimated ‘by visual inspection’).
In Germany, the low 2007 ice coverage was seen by the AGW protagonists as the tipping point, beyond which any skepticism was no longer allowed or should even be forbidden.

Steven Talbot,
I’m keeping a running count of 2008 vs. 2007 from the full resolution (850×850) CT maps. August 15, 2008 shows 41% greater extent than August 16, 2007.
An interesting statistic is that according to CT, Arctic ice area bottomed out by August 17 last year – which means that ongoing losses in extent after that date were probably due to compaction, rather than melting.

Steven Goddard
“You keep describing the NSIDC map projection (ad nauseum,) which is not the one that CT uses in their maps and not the map set that I used. Rather, I used the CT maps which are higher resolution than the NSIDC maps, have more detail, and are archived every day. No matter how many times you describe the NSIDC projection, it doesn’t have any impact on this discussion.”
The CT maps might contain more pixels than the NSIDC maps but that doesn’t change the fundamental resolution of the data. It just creates an image that is larger on a computer monitor.

Glenn,
You didn’t answer the question, and it was you who insisted on the data originating from the same source.
As I have indicated, I was naive enough to accept that Goddard’s statement was accurate. This confused me until you made clear that it could not be. Yes, I was insisting the data was the same source, since that was what Goddard had stated. My argument was proceeding from his stated premises.
Not sure whether the above quote from Goddard is correctly interpreted as a claim that the Cryosphere map is derived from NSIDC.
Huh? The statement was “these maps (derived from NSIDC data)”. That seems a pretty clear process of interpretation to me, Glenn!
Another mistake of yours is the attempt made (02:37:07) to confuse sea ice area with arctic ocean area with a minimum sea ice concentration. Both
the NSIDC graph and the Cryosphere maps depict ocean area, and is what Goddard has compared.
Yes, I accept that, though it wasn’t an “attempt to confuse”, but a mistake. Please note that I accept when I have made a mistake, unlike some others.
Another is your criticism of Goddard’s 10% figure, which is much closer than your 13.5%, upon analysis of the NSIDC graph itself that Goddard used…..You used some quote from an NSIDC webpage to make your calcs. So you again used a different source to refute a claim based on a different source.
No, that is wrong – it is the same source, the 11th August press release, to which Goddard refers. The graph was not released independently but as part of that press release, which includes the actual figures (in preference to either your or Goddard’s eyeballing). It seems pretty obvious to me that the graph does not have the resolution to compute percentage differences accurately and that one should refer to the figures which were given.

Yet another is the big deal you made of a leap day, and it isn’t a big deal at all. IMO it is quite fair and reasonable to compare “same day as last year” as 365 days ago, not 366 or 3 or 498.
You seem to have missed my point entirely. That may be reasonable in itself, but Goddard only applied this principle in comparing maps. With the graph he compares the same calendar date. Therefore, the offset was in favour of showing a greater differential. (Please note that I have accepted a mistake above. This is a mistake, or a failure to understand the point, on your part).
It appears that Goddard has a point, whether the data source is the same or not. And I don’t need to count pixels, just use a pencil as a straightedge on the monitor looking at the Cryosphere maps. It is most obvious just eyeballing that quite a bit more area is covered than a half mil from the same time last year. And his 30% more probably isn’t far off if at all.
It’s certainly possible that there is a point to be made about how effective these maps are as a visual representation of sea ice extent. Not very, I think. But the map is not a data source but rather a representation of a data source (just as a graph is not a data source). Goddard stated “The problem is that this graph does not appear to be correct. Other data sources show Arctic ice having made a nice recovery this summer…”. Three points here: 1) there are no grounds to presume that the graph is “not correct” rather than that the maps are “not correct” and 2) the NISDC maps exhibit the same visual coverage as the Cryosphere ones, so the insinuation that “other sources” show the graph to be “not correct” is unsound and 3) Goddard ignores the actual data sources, which were available.
The article allows the inference that someone is “cooking the data” in a biased fashion. IMV, that was its intent. IMV also, that is unjustified.

I just found this site. Daily SO2 images from around the globe.
http://satepsanone.nesdis.noaa.gov/pub/OMI/OMISO2/index.html

Resolution of the image has nothing to do with the monitor. Most Image files contain a sequence of bytes representing the location, color, transparency and brightness of the pixel. The CT image files have a resolution of 850×850 which is 722,500 pixels. The calculations are done by software analyzing each of the pixels. You don’t need any monitor to do this analysis.
I don’t believe that anyone is doing anything intentionally wrong with the data. There is an inconsistency between the CT images and the NSIDC published extent numbers. There also is an inconsistency between the CT images and the CT area measurements – which are likely derived from someone else’s extent numbers. I’m keen to find out what the source of the discrepancies are. From comparing with satellite images, my sense is that the CT images are golden.
My beef with NSIDC is that they have a habit of taking spectacular stories to the press, which get misquoted and propagated in high places. They should take a more conservative approach to dealing with the press. They should also be more conservative in their predictions. Scientists should be careful to avoid the appearance of pursuing an agenda, as it potentially affects their objectivity.
Suppose that 2008 had of set another record melt. Why couldn’t they wait until September to discuss it, rather than making high profile predictions to the press in May? It is not normal behaviour for scientists to be so impatient.

Steven Talbot,
“As I have indicated, I was naive enough to accept that Goddard’s statement was accurate. This confused me until you made clear that it could not be. Yes, I was insisting the data was the same source, since that was what Goddard had stated. My argument was proceeding from his stated premises.”
If you are referring to the data sources being the same, I did not make it clear that it could not be. I asked you for your evidence, since you had employed the assumption in a reasoned argument against Goddard’s claim.
This issue is still unresolved, but:
“Huh? The statement was “these maps (derived from NSIDC data)”. That seems a pretty clear process of interpretation to me, Glenn!”
is a quote mine. To use your partial quote from a previous post:
“A comparison of these maps (derived from NSIDC data) below shows…”
Again, it just isn’t clear whether Goddard meant that the comparison was derived from, or employed, NSIDC numbers, or whether he is claiming that the maps themselves are derived from NSIDC data. Nor should it be an issue of whether the same source of data was used for both.
“That may be reasonable in itself, but Goddard only applied this principle in comparing maps.”
I didn’t miss the point you claim is made here. The problem is that the maps are not the graph, and he was comparing the maps when he included a leap day. He could not have with the graph. You may be right that this favored a greater difference, but not a substantial one. Nit picking.
“Just as a graph is not a data source”
Now that is a hoot. Yes, Steven, graphs are sources of data. They aren’t satellite instruments, but representations of data allow data to be drawn from them. Were it not so, we should be thinking that we might as well be looking at cartoon reruns of Popeye. Data can reasonably be taken from the NSIDC graph, which is what I did, and resolution has little to do with quality here with respect to hundreds of thousands of square kilometers of ocean.
As to your claim about NSDIC data, your source reports for Aug 10, not Aug 11: “Arctic sea ice extent on August 10”, and if you look at Goddard’s archived graph of Aug 11, this year’s line drops a bit at the last day. That in itself makes your data from the NSDIC site and Goddard’s different by some amount. So aren’t you as guilty here as you claim Goddard is with respect to the leap day inclusion?

This blog has put up two very different pictures of Antarctica temperature anomalies, both NASA origin it seems. Connolley believes the high-cooling one is the work of a “PR droid” (naturally) but I have other thoughts I put here.
More discrepancies in polar data? Any thoughts?

I now have my new-and-improved sea ice area calculation. I did 2 things differently:
1) I limited the area to the “Arctic” by only looking for grid cells north of 63 degrees.
2) I assumed that the “pole hole” in the NSIDC data has more that 15 percent sea ice (a rather safe assumption)
I found 8467 grid cells with > 15 percent concentration on Aug. 12, 2007 and 9649 on Aug. 11, 2008. Since NSIDC also supplies a file with the area of each grid cell, I converted these to areas and came up with an area of 5523918 km^2 on Aug. 12, 2007 and 6281339 km^2 on Aug. 11, 2008.
I’m still at about a 14 percent difference.
REPLY: What was the rationale for 63 degrees N?

So many different measurements leaves me with little confidence in any. I don’t want to accept this as good science. Global temps are shown down to
the thousanth of a degree, yet different sources vary tremendously more, almost as much as the claimed increase for the past 30 years.
You might have a point if they were all purporting to be measuring in equivalent ways, but I am sure you are aware that they are not. The significant consideration is trend, and concern should be for the consistency of measurement within each set. Satellites, for example, exclude measurement of polar regions, the Himalayas and parts of the Andes, as I expect you know.
I see absolutely no responsibility shown by the AGW crowd.
Whereas I see that you’re happy to accept eyeballing a graph over calculating from raw data, accepting a pixel count of a mapped representation without assessing projection and disregarding the possible impact of spurious pixels, so long as it is supportive of the anti-AGW crowd?

Steven Goddard
“I’m keeping a running count of 2008 vs. 2007 from the full resolution (850×850) CT maps. August 15, 2008 shows 41% greater extent than August 16, 2007.”
Using the actual data, I get a difference of about 13.6 percent.

Steven Talbot,
Hopefully you are aware that alphanumeric characters are only representations of data as well, and not the true source. Recording, reproducing and representing data can all suffer from error. Sheesh, even the data source can and and often does contain error. That doesn’t mean that a graph isn’t a source of the data used to create it. Again if that were so, graphs would not be used. Some representations are better than others, of course. Your complaint seems to be rooted in the belief that these graphs and maps are error ridden. If so, then it would be immoral to publicize them. Your claim that graphs aren’t data sources is not believable, since you realize that they are representations of data.
As to the graph and your quote from the NSIDC page, all I know is that archiving is done when it is done, that the graph itself is said to be updated daily online, and when I read “On Aug 10th…” I take it that the event described is represented as having occured on that date. The graph URL is dated Aug 11 and is Goddard’s reference, and the data you used to falsify his conclusions was said to be from Aug 10, and resided on a separate webpage. Clearly you used different data.

Steven Talbot,
“You might have a point if they were all purporting to be measuring in equivalent ways, but I am sure you are aware that they are not. The significant consideration is trend, and concern should be for the consistency of measurement within each set. Satellites, for example, exclude measurement of polar regions, the Himalayas and parts of the Andes, as I expect you know.”
The significant consideration is accuracy, my friend. You are right that there are different methodologies for measuring, and comparisons should only be made with data measuring the same thing(s). Yet different methodologies should generate the same data, since the reality is that there is only one Earth and one “global temperature”. If you feel comfortable will all this inconsistency, I think you are being just as irresponsible as the experts who generate it. If I hire two people to come into my house and make termperature measurements, I expect they will both arrive at the same result, or both be considered wrong and their data discarded. I’m puculiar about *measurements* that way, and science is or should be especially so.
Me: “I see absolutely no responsibility shown by the AGW crowd.”
“Whereas I see that you’re happy to accept eyeballing a graph over calculating from raw data, accepting a pixel count of a mapped representation without assessing projection and disregarding the possible impact of spurious pixels, so long as it is supportive of the anti-AGW crowd?”
Excuse me, but I’ve offered no conclusions beyond the maps being inconsistent with the graph. This can not be seen as being supportive of anything other than consistency. Regardless, it appears that the Arctic melt is largely if not exclusively a result of events not related to “global warming”, so for me this issue has nothing directly to do with being supportive of the “anti-AGW crowd” or critical of the “AGW crowd”.
But making claims about what I “happily” haven’t done or have disregarded, in the absence of evidence of that, isn’t quite honest. It’s a tactic I expect from an AGWer. Are you one, or rather do you accept the “projections” made by the various AGWers that spell doom and gloom if we don’t all stop drivin that hot rod Lincoln?

Glenn
“I’ve offered no conclusions beyond the maps being inconsistent with the graph.”.
Ignoring the data is a tactic that I expect from an [snip- you don’t get to use that word here]. NSIDC’s graph is consistent with the data. I’ve posted the location of the data, so you can do your own count. NSIDC’s graph is also consistent with the “recent ice area” on the CT website.
The only “evidence” to support the “conclusion” that NSIDC’s graph is inconsistent with the maps is someone’s claim to have counted pixels on re-projected images created using an unspecified algorithm (actually, there were likely multiple interpolations. NSIDC sea ice data is stored in polar stereographic grids. I’m guessing that the data first had to be interpolated to a lat-lon grid and then interpolated again into the image space).
Personally, I tend to believe objective data rather than someone’s guesstimate. But to each his own. Some people prefer to believe that what they believe is true. Then again, when a friend showed me the maps I guesstimated that there was about a 15 percent difference. Maybe my eyes are just better calibrated than most peoples’.

Charles
‘At this point, our simulation model, Project 6, says that #3 is most likely, and sometime next year. Depending upon harvests and the US foreign policy among other factors.’
And, of course, your model incorporates as many of the forecasting principals as possible as laid out in the Principals of Forecasting hand book. If not your model is nothing more than a reflection of your guestimation.

If this story gets 1% the coverage the original alarmist story got I will be shocked.

Dipole,
Remember Goddard’s comparison is for a specific time, Aug 11, not the 17th.

Jeff says “Ignoring the data is a tactic that I expect from an AGW denier”
Just eyeballing these two maps
http://igloo.atmos.uiuc.edu/cgi-bin/test/print.sh?fm=08&fd=11&fy=2007&sm=08&sd=11&sy=2008
should convince most anyone without an agenda that the right side is quite a bit more than 10% more than the left side map. The 10% comes from data taken from this graph
http://nsidc.org/images/arcticseaicenews/20080811_Figure2.png
It needn’t be rocket science to see the inconsistency, and no need to ignore any data.

Ignore the above post _ I was looking at the wrong picture

Glenn,
You say:
“The significant consideration is accuracy, my friend.”
I agree that is significant, it goes without saying, which is why I do not think eyeballing is a good enough basis upon which to draw your conclusion of “the maps being inconsistent with the graph” (my eyeballing is inconsistent with yours, by the way).
If you feel comfortable will all this inconsistency, I think you are being just as irresponsible as the experts who generate it.
That’s pretty close to a straw man, Glenn. The inconsistency I feel most concerned about is that between the satellite records, since they are purportedly measuring the same things, AFAIAA, and they also show greater divergence than between other types of record.
Regardless, it appears that the Arctic melt is largely if not exclusively a result of events not related to “global warming”
That’s an interesting, and very confident, claim. Would you be able to explain to me how you quantify “largely if not exclusively”? I am well aware that wind patterns will influence the particular extent in a given season, for example, but do you have evidence to establish that the long-term trend is unrelated to warming?

But making claims about what I “happily” haven’t done or have disregarded, in the absence of evidence of that, isn’t quite honest. It’s a tactic I expect from an AGWer. Are you one, or rather do you accept the “projections” made by the various AGWers that spell doom and gloom if we don’t all stop drivin that hot rod Lincoln?
I am not a member of any church, Glenn. At present I am persuaded that the risks of AGW are high. I look forward to whatever new evidence, either way, may be forthcoming. I don’t count your eyeballing of graphs and maps to be useful evidence of anything much.

Evan Jones
“What does make a difference is whether or not NSIDC is accurately reporting the data.”
And I have shown that the NSIDC esimates are reasonable [snip – thats ad hom]

“Just eyeballing these two maps
http://igloo.atmos.uiuc.edu/cgi-bin/test/print.sh?fm=08&fd=11&fy=2007&sm=08&sd=11&sy=2008
should convince most anyone without an agenda that the right side is quite a bit more than 10% more than the left side map. The 10% comes from data taken from this graph”
Nobody who doesn’t have an agenda would refuse to look at the actual data.
You don’t seem to be able to comprehend that I actually COUNTED the number of grid cells with sea ice on each day and therefore I actually KNOW what the difference in areas is. My 14 percent figure is not something that “looks right” to me, it is the actual difference. There’s no other way to explain this. I’m trying to make this as simple as possible.

…the disappearing sea ice meme…
It looks like we’re heading towards the second lowest summer coverage on record, contending with 2005….
…no problem, then – that must mean the world is cooling. eh?

Someone better warn Gordon Pugh not to launch his kayak.
Pugh said, “Over recent decades, we have lost more than half the Arctic summer sea ice cover; decades ahead of predictions, showing climate change has been hugely underestimated. We must insist that our leaders take urgent action to halt climate change.”

By the way, Steven Goddard, have you not figured out where the Berrimilla is (whose blog you link to as an “eyewitness account” of the NW Passage supposedly closing up)? Today it’s been moving through Lancaster Sound, heading into Baffin Bay. Lancaster Sound, from which the report you link to was posted, is here:
http://en.wikipedia.org/wiki/Lancaster_Sound
Check that out against the NSIDC map you’ve linked to and you will see that it is recorded as clear blue, and is a very long way from the more marginal areas of the NW Passage!
Do some basic research, I’d suggest!

the north west passage has been open many times in history. It’s a non event , ho hum time for bed.

Steven Talbot,
Had you actually read the Bermilla blogs, you could have saved yourself quite a bit of embarrasment. The Northwest Passage is a complete route over the top of Canada which the Bermilla is currently undertaking. They are taking the shortest and fastest route across the NW Passage. What part of They got out just in time. Resolute is freezing in behind them is it that you don’t understand?
Friday’s new ice from the Canadian Ice Service
http://awberrimilla.blogspot.com/2008/08/yesterdays-new-ice.html
The purple is the new ice from the evening 13th and morning 14th, and the first I’ve seen this year.
Sunday’s new ice from the Canadian Ice Service
http://awberrimilla.blogspot.com/2008/08/lancaster-map-again.html
Purple is new ice. ‘E’,’L’ &’G’ also has new ice
Today’s NSIDC map shows the southern leg of the NW passage now blocked, which they came through a few days ago.
http://nsidc.org/data/seaice_index/images/daily_images/N_daily_extent_hires.png
If you have ever watched a body of water freeze, you would know that it starts out as thin ice and later gets thicker. Given that it is mid-August, someone of at least moderate intelligence might be clever enough to realize that the Arctic summer has just ended and the freeze is just beginining. They described the new ice as being The worst kind to a small boat. Tricky concept for someone who likes to shoot their mouth off before they think.

Anthony –
Your explanation of methodology is basically accurate. First, the software removes all non-ice pixels from the UIUC image. It then removes any pixels outside the radius of the earth. Then it converts all remaining pixels to a single color. Finally, it counts all the pixels of that color. Using a single color to represent all >15% concentrations is the same technique which NSIDC uses, and makes the generated videos easier to visualize.
I generated a graph showing the daily percentage gain in ice extent during August vs. last year, and was surprised to see that the trend has a positive slope and is thus increasing. The average over the month is 30% gain.
http://spreadsheets.google.com/pub?key=pj0h2MODqj3jAfVXaaKHbCQ&oid=5&output=image
This graph uses only conservative, unadjusted pixel counts – which tend to underestimate the actual gain. I will try to keep it updated through the remainder of the season.
REPLY: Steve, perhaps to add strength to your argument, you can run the same method on the maps at the URL in the previous comment? Replication is always good.

Steven Goddard: I believe you have counted the pixels correctly, even if you didn’t eyeball the ‘10% difference’ correctly (more like 13.5-14%).
But according to the NSIDC’s website, sea ice extent is not simply derived from a pixel count, but from the number of pixels multiplied by their ice concentration:
“The values for ice area are obtained by summing the concentration of ice within each pixel over the entire ice extent. For example, if a pixel’s area was 600 km² and its ice concentration was 75%, then the ice area for that pixel would be 450 km²”
http://nsidc.org/data/seaice_index/derivation.html
Have you done this? And may I also draw your attention to this sea ice extent graph from Japanese researchers – it looks very similar to that of the NSIDC, no?
http://www.ijis.iarc.uaf.edu/en/home/seaice_extent.htm

Anthony,
Good suggestion. The Bremen maps, NSIDC maps and NSIDC graphs are in close agreement. I’m not disagreeing with their internal consistency. The NSIDC/Bremen maps and area calculations use an adjusted projection (an attempt at equal area,) which involves a fair amount of math to generate.
By contrast, the UIUC maps use a natural, unadjusted projection – essentially what an astronaut flying over the North Pole on a cloudless day would see. The published UIUC maps also have much higher resolution of the Arctic Basin than either NSIDC or Bremen maps.
I can measure the UIUC maps directly, and make reliable estimates of relative areas from map to map. There is a radial distortion in these maps, but as I have explained before, this distortion causes my estimates of larger ice areas to be conservative – on the low side.
Because the Bremen/NSIDC maps are intentionally distorted using a more complex algorithm, there is no simple way to determine why they appear to disagree with the UIUC maps. I will assert that if the UIUC maps are correct, and if I counted the pixels properly, there is the possibility of a mathematical error in the NSIDC calculations. There is an inconsistency.
Either NSIDC is in error, UIUC is in error, or I have measured/interpreted the UIUC maps incorrectly. Like I said, no one has made any attempt to demonstrate any one of those three possibilities. I am fully aware that my numbers are inconsistent with NSIDC. That is exactly the point!
Do we trust our own eyes? The area growth on August 11vs. last year was clearly greater than the 10/12/14% (whatever) which the August 11 NSIDC graph indicated.

If someone wants to prove me wrong, they need to demonstrate that either the UIUC maps are incorrect, or that I am not interpreting them correctly. I would greatly appreciate it if people would stop parroting that my numbers are inconsistent with NSIDC. I know that, and that was one of my primary reasons for writing the article.

phil,
You are again confusing “area” and “extent.” Read the paragraph you quoted from NSIDC more closely. “Area” is derived from both “extent” and “concentration.”
The NSIDC map for today again shows that all routes through the Northwest Passage are blocked. This was not true a few days ago. Obviously this has to be due to new ice, as described in the Bermilla blog.
They got out just in time. Resolute is freezing in behind them
What is it they “got out” of? What is freezing “behind them?” The answers seem quite obvious, que no?

Steven Goddard
“f someone wants to prove me wrong, they need to demonstrate that either the UIUC maps are incorrect, or that I am not interpreting them correctly.”
You’re the one who claimed that NSIDC was fudging data. It’s up to you to prove that NSIDC’s data are wrong. This is just a standard [snip – stop with the ad-homs, last warning] tactic: burden of proof shifting.

Steven Goddard
“The NSIDC/Bremen maps and area calculations use an adjusted projection (an attempt at equal area,) which involves a fair amount of math to generate.”
The maps are not an attempt at equal area projections. They are Polar Stereographic projections, which have been used in meteorology for decades.
“The Bremen maps, NSIDC maps and NSIDC graphs are in close agreement.”
As are the NCEP maps. In fact, everything is in close agreement EXCEPT your alleged calculations.
http://polar.ncep.noaa.gov/seaice/hires/nh.xml
“Because the Bremen/NSIDC maps are intentionally distorted using a more complex algorithm, there is no simple way to determine why they appear to disagree with the UIUC maps.”
It’s actually easy to do get a qualitative idea of the difference. And what you find is that there isn’t any real difference.

Patrick Henry,
You appear to be someone who likes to call others stupid but who is not prepared to look at presented evidence yourself –

The NSIDC map for today again shows that all routes through the Northwest Passage are blocked. This was not true a few days ago. Obviously this has to be due to new ice, as described in the Bermilla blog.
The NSDIC shows low resolution cells. The far higher-res Bremen image, to which I have already linked you, shows the Amundsen route as open:
http://www.iup.uni-bremen.de:8084/amsr/arctic_AMSRE_visual.png
Your comment please?
“They got out just in time. Resolute is freezing in behind them”
What is it they “got out” of? What is freezing “behind them?” The answers seem quite obvious, que no?
This comment is written by the back-up, who is sitting somewhere warm misinterpreting an ice map, as has been carefully explained to you! To use your own words, “someone of at least moderate intelligence might be clever enough to realize” what is being said, but I guess it’s a “tricky concept for someone who likes to shoot their mouth off before they think.”

Incidentally, Patrick Henry, to help you commenting on the difference between the Bremen image and the NSIDC image, see this from the BSIDC FAQ:
Why do other sources indicate there is no ice where passive-microwave data shows ice?

The passive-microwave sensor records ice in 25-by-25-kilometer (16-by-16-mile) areas, which is lower resolution than other types of satellite sensors. This means that the ice edge could be off by as much as 25 to 50 kilometers (16 to 31 miles) in passive-microwave data compared to higher-resolution satellite systems. We define ice extent as anywhere with at least 15% ice. Regions with 15% ice will look quite open in higher-resolution satellite data, but will still count as ice in the passive-microwave extent fields.
Other reasons that passive-microwave data may show ice where none actually exists on the ground include signal variation along coastlines between land and water, the shift in albedo of actively melting ice, and atmospheric interference from rain or high winds over the ice-free ocean. In the daily extent data images, gaps (shown in dark gray in the extent map) are replaced with values interpolated from surrounding days, but temporary spurious results may occur.
http://nsidc.org/arcticseaicenews/faq.html#data_sea_ice_index
Your comments?
I recommend looking at the NSIDC FAQ, as Mark Serreze has suggested. Doing some basic research is usually a more profitable process than leaping to a conclusion that suits you.

Dr. Meier,
To speed the discussion up, a bit more on the type of explanation I am looking for wrt the ice extent dispute. Something like this –
1. “Pixel counting UIUC maps doesn’t work, because the map is distorted in a way which makes the area around 80N latitude too large. The distortion can be defined mathematically as follows ………”
or
2. “Pixel counting UIUC maps doesn’t work because the resolution per pixel is too low. The grid cell size is smaller than one pixel, so a region of four contiguous pixels may appear to represent more extent than what is measured on the higher resolution grid. This can be defined mathematically as follows……….. The reason we believe our grid size to be the correct one, rather than the one pixel grid you are using in the article, is because …….”
Thanks again. It is very late and I need some sleep. I look forward to seeing your response.

Dr. Meier, this is apples and oranges.
The chart you refer to appears to be sea ice concentration. It does not depict ocean area with sea ice as does the NSIDC chart and the UIUC maps under consideration, and the numbers are off by about two million sqKm.
The UIUC maps appear to show a recognizable and significant difference in percentage between the same time last year and now that does not correlate at all well with the NSIDC graph. That is my concern, and what I understand Mr. Goddard’s concern to be as well. Simply put (hint) the NSIDC chart tells me there is about 10% more ice than last year, and the UIUC maps tell me that there is about 30% more ice than last year.

Earlier I complained I was having some trouble matching the UIUC map colours to the included scale. Closer examination showed that the 100% ice cover on the map isn’t well represented on the scale. But using the sum of absolute differences of RGB values as a metric and a threshhold of 50 I can get what looks like a reasonable match with the ice extent on the UIUC maps. I would have to play around a bit more before being totally happy with this.
But with this threshhold I get a 30.5% increase in pixel count for 20080811 over 20070812, in support of Steven Goddard’s original claim.

No, Smokey, the ball is not in the NSIDC court. Rather, the match is over. Steven Goddard refuses to look at the data, and those scientists who have looked at the data, have determined that Goddard’s measurement of 30% increase in sea ice extent, from August 11, 2007 to August 11, 2008, is incorrect.
Dr. Meiers made the point quite clear:
“8. An image is simply a way to convey data; it is not data itself. Therefor it is not proper to do analysis on the image. You need to use the data.
9. The gridded data, when analyzed, must account for the projection in terms of the area of the grid cells. You have to sum the ice, weighted by the correct area for each grid cell. NSIDC uses a polar stereographic projection with a true latitude of 70 N. Other than at 70 N there will be distortion that needs to be corrected for, as NSIDC does.
10. NSIDC freely distributes all the data, tools to work with the data, and the grid cell area files. So anyone can do their own analysis.”
People on this thread did the analysis and confirm Mr. Goddard’s conclusion is incorrect…
Why doesn’t Mr. Goddard use the actual data? What does he have against the actual reported data?
Yesterday, I already suggested to Mr. Anthony Watts that he needs to publish a correction, but he embargoed my comment. I will attempt to make the comment again.
REPLY: PaulK Your “request for correction” was one that you wrote with specifics about language and placement. That’s not how it works. It’s not your blog, you don’t have the priviledge to dictate terms, and you don’t get to tell me what to write. The story is written by owned by Mr. Goddard, and since then he has offered to print a correction if his questions are satisfied, I’m going to respect his right to choose what he wants to do and when.
Right now we have some interesting questions in play, and I want to see how they are answered.

Glenn
“The UIUC maps appear to show a recognizable and significant difference in percentage between the same time last year and now that does not correlate at all well with the NSIDC graph.”
The alleged difference in the UIUC maps doesn’t correlate well with the UIUC graph either, as Walt Meier pointed out. We again get back to the issue that everything is in agreement except for Mr. Goddard’s numbers.

Patrick Henry… In some cases, you are correct, but not in this case. The image Goddard is using, was generated from a database, not directly recorded.
The key mistake Goddard and dipole are making, if I understand Steven Talbot and Jeff correctly, is that there is no way to show the surface of the globe on a flat screen as a continuous image. So the image people have been eyeballing, MUST have some errors in it. So where are the errors?
Here I will attempt an explanation… (and please correct me if I get this wrong, Steve or Jeff). It is very difficult to illustrate a curved surface on a flat piece of paper, or a flat computer screen. So “projections” are used for maps, drawings or illustrations depicting the surface of the earth. Steven Goddard assumed that the projection used on Cryosphere showed the area close to the North Pole as accurate to scale, and then that this projection would not represent areas correctly at lower latitudes.
But in reality the scientists at NSIDC were one or two steps ahead of Mr. Goddard. They used a projection accurate at 70 deg N, not at the pole (90N), as Mr. Goddard assumed. The scientists used this projection, because historically, the latitudes around 70 deg N, is roughly mid-way between summer mins and winter maxs in ice extent. (No one expected the massive ice melts nearer the Pole , that we saw last year, or this year, for another 50-100 years.) The scientists picked a projection that worked best for measuring sea ice extent, in a non-warming world. Where is 70N? It is about halfway down the West Coast of Greenland.
https://www.cia.gov/library/publications/the-world-factbook/geos/gl.html
If Cryosphere (UIUC) uses the same projection in the image Mr. Goddard is working with, that the NSIDC uses (and they should), then the image that you and I and Mr. Goddard are looking at, is not accurate at latitudes higher than 70N, and the higher the latitude, the bigger the inaccuracy. This isn’t because someone made an error; this is because you can’t show a three dimensional object in two dimensions without some error.
Imagine you are looking at the North Pole, looking down at a globe, and you are trying to draw that image on a flat sheet of paper (making a projection). If the area shown on the flat image is accurate down around the mid-west coast of Greenland, then a square inch of area on the globe at that latitude will occupy a square inch on the flat image. But the globe has more surface area (square inches) than the flat image. So the area, and the number of square inches above the 70N latitude, on the flat image, will be LESS than the actual area on the globe. So the flat image you and I see, shows less area in the Arctic Basin, than the actual area on the globe. If that area is covered by ice, and included in the ice extent, then Goddard ended up measuring fewer pixels in the Arctic Basin, than he should, because he is using a flat image to measure his data. And the guys eyeballing it, are making the same mistake. Since the YOY differences in sea extent are at lower latitudes, then the difference in areas there seem bigger, and then this difference is divided by the total extent of the sea ice to get the percentage change.
Lets try it with some approximate numbers: By getting a lower count of pixels at the high latitude, Mr. Goddard is dividing the difference of around 750k sq KM by less than the actual 5500k sq KM (which would calculate to 13.6% increase), he instead divides by a lower number of pixels equivalent to 2500 sq KM and gets a 30% increase. But his analysis is wrong, as are the eyeball estimates. The actual NSIDC data show that the YOY extent was about 13.5% different on August 11.
OK, that is how I understand the mistake. The key issue here, is that taking measurements from a flat image representation of our planet, will always end up in error, UNLESS you adjust for the projection error. Mr. Goddard should have known that, and he should have heeded Steven Talbot and Jeff, when they pointed out the projection error.
REPLY: That is a good explanation, and I appreciate the detail. Another question that needs to be answered is how the satellite derived data is “photographed”. As I understand it, it isn’t a single snapshot from above the pole, but a series of successive orbit slices that are patched together. Thus, how do we get from a curved rectangular view “slice” from the satellite persepctive to the raw data and to the imagery? What sort of projections errors or uninetentional distortions are introduced and are they relevant?
While there may be errors in the analysis presented here, how would the public know which presentation from which group better reflects the reality of sea ice? Problems like this one we are discussing arise when data presentations for public cconsumption, perhaps uninetntionally, show differences that can’t be easily resolved by the layman.

Paul K (00:05:31)
The key mistake Goddard and dipole are making…
Originally I was just counting pixels. If I claimed further significance for the results it was unintentional. Where’s the ‘mistake’?
You are claiming that the interpretation of pixel count as ice extent is invalid because of spacial distortion. Steven Goddard described the projection used on the UIUC maps, and in my last post I incorporated this into my calculation. It didn’t seem to materially affect the outcome.
If you, or anyone else, can demonstrate I’m using the wrong projection I can have another go.

Has UIUC been emailed to see what they say on the topic? I get the feeling NSIDC are a bit unhappy with this story being broke before the author got in touch with them first so may not be that forthcoming? Or am I reading it wrong?
It does beg the question though that if the data is right, and having read all this I still think it is, then why are the “popular” graphs at Cryosphere seemingly so far out of kilter? A nice visiual should enable you to compare one year to the next without having to count pixels or apply algorithms or actually just junk it and go back to the data, because then it is useless.
Maybe these are questions more for UIUC than NSIDC, so I would like to know if anyone has had any input from them?
Talking of images, the direct NW passage is either nearly open or pretty well closed depending on what you look at.
http://iup.physik.uni-bremen.de:8084/amsr/arctic_AMSRE_nic.png
http://www.seaice.dk/iwicos/latest/amsr.n.ice.20080819.gif
http://ice-glaces.ec.gc.ca/Ice_Can/CMMBCTCA.gif
http://manati.orbit.nesdis.noaa.gov/ice_image21/D08232.NHEIMSK.GIF
There’s not too much consistency there to me to be honest. Are all these just pretty to look at in the main?

I think the distance might be 10000km; not 10000 miles. I did an estimate based on the latitudes of regions I can make out just on the perimeter. 10000km would mean the perimeter is roughly latitude 23.

Patrick Henry (22:54:25) :
Images certainly are data. They contain location, intensity and color information.

Images are data, but they’re not necessarily good data, and they do not contain location and intensity information. They contain an x,y location in image space and values for red, green, and blue. If you know the projection (and we don’t know the projection of the Cryosphere Today images) you can calculate location, but it’s not intrinsically in the image. Similarly, the RGB values don’t translate directly to intensity: they merely represent the relative strengths of three different sub-pixels on a computer monitor. Additionally, the Cryosphere Today images use lossy compression (even the PNGs contain compression artifacts) so they can’t be used for precise quantitative analysis. (Contrary to an assertion earlier in this thread, JPEG noise can not be removed.)
REPLY: I’ve been working with computer graphics since 1983, and my company produces computer graphics as seen on the side panel to the right. Thus I’m qualified to speak to the concerns of image formats and compression since I have used them both since their inception and produce them daily.
PNG is a 32 lossless compression image format, there are no artifacts, just like with the 8 bit GIF compression, there are no artifacts. The run length encoding schemes simply packetize pixels that are the same into tokens. It is totally lossless and without any compression or noise artifacts to convert the packets of RLE data to pixels upon display.
http://en.wikipedia.org/wiki/Portable_Network_Graphics
So there is no error of any kind in reading PNG pixels and translating it’s RGB component back to the numeric value of the original data as defined by the color key.
Unlike PNG, JPEG is not perfect, and much depends on the original compression setting, known as the Q factor. See the sample photographs in this article to see what I mean: http://en.wikipedia.org/wiki/JPEG
JPEG’s work because humans can resolve detail better than they can resolve color. NTSC and PAL color television use the same trick. They preserve the detail (luminance) while mostly compressing the color (chrominance). The weighting equation assigns the most compression to reds and blues, while preserving green, which we resolve better.
As for removing JPEG noise from it’s lossy scheme. I’ve done it and have a program that does that that is commercially available. Granted, depending on the compression factors applied to luminance and chrominance originally there may not be a 100% perfect restoration, but noise can be removed. Most of the error is usually in color space. JPEG does not appreciably alter the shape of the photo or graphic, thus in this exercise, while there may be a margin of error for recovery, Nonetheless for a comparison of two images to determine differences in pixels, it can be done with a small margin of error. I estimate about 2% or less for these images. To prove that to yourself, create a couple of images that have varying graphical color elements of known pixel area, then save as PNG and as JPEG, then run a pixel count on each. You may find that while the colors count varies in JPEG, the luminance count varies little, and perhaps not at all depending on the compression setting used.
But the Cryosphere today images, such as these here: http://igloo.atmos.uiuc.edu/cgi-bin/test/print.sh?fm=08&fd=15&fy=2007&sm=08&sd=15&sy=2008
are PNG and totally lossless. So I would expect accurate pixel counts between them.
That doesn’t address the larger questions about earth’s spherical geometry and what distortions may exist, but the approach of pixel counting/comparing two images can be expected to give repeatable results with a margin of error for JPEG and exact results for PNG – Anthony

PaulK,
I have described in some detail the distortion in the UIUC projection, but obviously not enough. In that projection, there is nothing magical about 70 degrees or any other latitude – other than 90 degrees. The oft quoted 70 degree number relates to the projection being used by NSIDC, not UIUC.
Starting with the ideal case of a viewpoint at infinity (no parallax) – at the pole, the image is undistorted. As you move away from the pole, the distortion increases. At the equator, the distortion would be infinite. In this case, the distortion would be = 1 / sine(latitude)
Looking at the UIUC maps, the actual viewpoint is from a height less than infinity, so the latitude where the distortion becomes infinite is at a higher latitude than 0 – maybe ten or twenty degrees. That isn’t hugely important because there is no ice anywhere near that latitude.
In that type of projection, a smaller circle centered around the pole (i.e 2007) is less distorted than a larger circle (i.e. 2008.) If the calculations were corrected to adjust for this, the larger circle (2008) would be adjusted upwards more than the smaller circle (2007.) This would have the effect of making the percentage gain in 2008 larger than what I reported. I did not make any adjustments to the raw pixel count, and that is why I described the methodology as conservative. The actual 2008 percentage gain is somewhat larger than I reported.
If there is a flaw in the methodology, it is not at that point. It would have to be tied to an inaccuracy in the UIUC map or my understanding of that map. This may well be the case, and I am keen to understand where the error is.
If Walt or Mark can explain to me where the error is, I will most certainly retract and apologize. There appears to be a discrepancy between the UIUC maps and the NSIDC calculations. I’m certain that the NSIDC calculations were done carefully and diligently.

Anthony,
Thanks for your follow up on the image color issue.
In the case of “ice extent” calculations, the image color precision is not important. All pixels >= 15% are treated equally in extent calculations.
If the calculation was for “ice area” using the images would likely introduce more error, and jpeg would probably not be the best format. I haven’t and wouldn’t attempt to do a calculation of sea ice area, using the images.

I had another look at the distortion factor as noted at (01:28:52). Now I get +5% for 20080811 and +4% for 20070812 for a 2007 to 2008 increase of 31.6%.
Still for view height of 10000 miles.
I would still not be betting on the accuracy of my calculation, but for these images and projection it seems no big deal anyway.

Looking back at earlier posts, a few things caught my eye which I might be able to clarify:
1) The north pole issue: Back in June, there was some coverage about the possibility of the North Pole being ice free by the end of this summer. This was based on recognition that the area around the north pole was covered by firstyear ice that tends to be rather thin. Thin ice is the most vulnerable to melting our in summer. I gave it a 50/50 chance. Looks like I’ll lose my own bet and Santa Claus will be safe for another year. There was indeed some coverage a some years back of an open north pole (and I was interviewed). This opening, however, was pretty clearly causes by unusual winds breaking up this ice, and not from melting out.
2) The uptiicks/downticks in our updated time series of ice extent. The time series, as presented on our “Sea Ice News and Analysis Site” is based on 5-day averages. This is done to smooth out short-term “blips” that can occur from noise in the data (basically weather effects and surface melt effect that contaminate the passive microwave retreivals). Speciicially, for a given day N,
the 5-day average is: (N-2 + N-1 + N + N+1 + N+2)/5 . You run into a problem for the current day (e.g., August 20 as I write this) as we don’t yet have data for N+1 and N+2. Similarly, for August 19 (yesterday), we don’t yet have data for day N+2. So, for the current day and the day before, we do a modified 5-day mean by projecting values forward by 1 or 2 day based on the slope over the past few days. This 5 averaging procedure is why the slope for the last few days or our graph shows these somewhat puzzling upticks/downticks. It turns out we actually changed procedures slightly right around August 1. Before then, the five day average for the current day was (N-2 + N-1 + N + N + N)/5, which really gave too much weight to the current day. Nothing nefarious about making the change – it was a decision based on simply trying to improve presentation of the results.
3) Antarctic Sea Ice: There seems to be a great deal of misconception about why sea ice in Antarctica is not declining (and indeed does show an increase) and why Antarctica (except for the peninsula) is not strongly warming like the Arctic. One has to realize that a much slower and subdued response of the Antarctic to greeenhouse gas loading is something that was projected by even our earlier climate models over 25 years ago. There’s a number of reasons for this involving the much different nature of the ocean and atmospheric circulation, but the fact that it is the Arctic which is changing rapidly and not the Antarctic is no surprise. Indeed it’s exactly what we expected. The recent sea ice ingrease in Antarctica turns out be be a weak trend in a noisy time series, largely drivent by changes in atmospheric circulation. Please go to the “frequently asked questions” on our “sea ice analysis and news” site where these issue ae discussed.
4) I see no point in further discussion about malfeasance on the part on NSIDC. We stand by our data and results. My colleague Walt Meier stated the issues quite clearly. Some do not like the story that our data tell. I don’t like the story either, but I’ll continue to tell it as I see it.

Pioneer stock like me eyeball weather. If we don’t, our crops don’t get put up and either freeze, rot or dry up in the field. So eyeballing the two Arctic images of last year and this year, this same date, we have lots more ice up there. Nuf said. Cold winter coming and soon. Don’t bother with a 3rd cutting. Irrigate like hell so what is left can be used to extend feeding till pastures run out or get snowed out and we have to start using expensive bailed hay. Plant winter wheat again now and hope to gawd it don’t freeze when it comes up like it did this year. Spring wheat won’t get in the ground in time to harvest and besides, you don’t get much for spring wheat. Winter wheat always sells higher. Buy heating fuel or cut wood now. And buy/cut at least enough to last 9 months.

“But the Cryosphere today images, such as these here: http://igloo.atmos.uiuc.edu/cgi-bin/test/print.sh?fm=08&fd=15&fy=2007&sm=08&sd=15&sy=2008
are PNG and totally lossless. So I would expect accurate pixel counts between them.”
Have you looked at those images closely? Although the image pairs are distributed as PNGs, at some point the imagery has been through lossy compression. The errors are most obvious along the ice margins, visible as a speckled pattern within 8 by 8 pixel blocks. Try selecting a single color in the color bar with the magic wand tool in Photoshop, or a similar image processing application: it doesn’t select a single horizontal line, which you would expect in a clean image.
REPLY: Simmon, good catch. the web page presentation compresses the image size, but if you do a right click and pull up the URL for the image pair that was generated, you get the full size image which is 1709 x 856 pixels.
I did that and copied it into my imaging program and zoomed in to look for JPEG artifacts, and indeed they are there. So yes at some point they did use JPEG.
This begs the question as to why would they take JPEG images, then save them as PNG? It defeats the purpose of lossless compression! It makes me wonder if UIUC is truly cognizant of what they were doing graphically.
The new color scheme images, such as this one
http://arctic.atmos.uiuc.edu/cryosphere/NEWIMAGES/arctic.seaice.some.000.png
Don’t appear to have been JPEG compressed. I don’t see the edge artifacts common to JPEG compression.