« SEARCH September sea ice outlook: July report | Main | Animation 9: sea ice concentration »

Comments

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

Anu

Note that the sea ice area curve is not only holding the decline steady, but accelerating. It is well below 2009, and poised to be well below 2008 when that curve drops below 2009 in the next week.

http://ocean.dmi.dk/satellite/index.uk.php
(choose Geographical domain: Arctic Ocean)
The Beaufort Sea is now the warmest sea in the Arctic (up to 16° C), confronting a big chunk of sea ice and the Arctic Basin beyond. The Chukchi, East Siberian and Laptev Sea ice is all melting away, and will be the big areas of melt in the month to come:

ftp://ftp-projects.zmaw.de/seaice/NEAR_REAL_TIME/Arc_latest_large.png

I will be very surprised if sea ice extent for 2010 does not drop below 2009 again in the next week or two.

Jon Torrance

There's been some talk here and elsewhere about tracking the ratio of sea ice ice area to sea ice extent. I took a stab at that a couple of days ago using published JAXA extent numbers and not very precise measurements of area off the JAXA sea ice area chart and got the following results for July 26th:

Year Area/extent
2002 84.2%
2003 83.9%
2004 85.4%
2005 83.0%
2006 85.1%
2007 82.4%
2008 85.5%
2009 87.1%
2010 83.3%

I then tried the same thing with JAXA extent and, since they publish easily accessible numbers, Cryosphere Today area and got:

Year Area/extent
2002 72.6%
2003 72.5%
2004 71.0%
2005 70.5%
2006 72.6%
2007 68.9%
2008 69.3%
2009 73.3%
2010 68.2%

Caveat - I haven't triple-checked my work for errors in transcription. Assuming I didn't make any, it seems the two area measures don't always move perfectly in lockstep. Whether I made errors or not, clearly JAXA and CT area are greatly different things and I'd be interested if anyone can explain why and save me the effort of researching the question. I'd also be interested in hearing what thoughts the numbers give rise to.

Lord Soth

The sea ice area may end up lower than 2007, the ways things are going. The problem, is the media will report on the extent; not the area or the volume. That is unfortunate, as sea ice extent is becoming much less relevent, now that the sea ice pack is no longer a solid mass.

In the past when I mention that sea ice area, was a better measure of sea ice loss (than extent), I always got shot down by the experts, due to argument about melt ponds skewing the results. They sure dont't look like melt ponds this year!

As far as those adventuers who are trying to circumnavigate the arctic, wouldn't it been smarter to head west, as the islands around Severnaya Zemlya may not be clear of ice until September. Isn't this where Adrian "Wrong Way" Flanagan got stuck in 2007?

Neven

Lord Soth, I remember reading (I think in a comment on their blog) that it's better to start with the NEP because of winds blowing west to east. Besides the NWP usually doesn't open until September (or late August). If they make it through the NEP I think they have a very good chance of going through the NWP without hardly meeting any ice.

siili

Interesting development in todays preliminary Bremen map, maybe we'll see an increase in tomorrows Jaxa numbers due to the strong divergence at the atlantic side. The area should be another thing though. Also fun to see the fingers of thin ice starting to penetrate the 2.5m thick healthy recovering ice in the PIPS2 forecasts.

Jim Dowling

@: Lord Soth
The coriolis effect creates the prevailing westerly winds in the northern hemisphere. Hence you should go around the world eastwards (assuming the Greenwich longitudinal system).
The best place to start work have been somewhere like Petropavlovsk-kamchatsky, and do the NWP first.
However, as those guys are Norwegian, they left from Oslo. They'll probably have to wait a few weeks for the NEP to open. I reckon they'll make it anyway.

Kevin McKinney

Also, Kevin Oliver and Tony Lancashire, the British team who ended last season's rowing/sailing attempt on the NWP at Gjoa Haven, aim to finish up this season at Arctic Bay:

http://www.arcticmariner.org/

Good luck to 'em, I say!

Neven

The Bremen map looks interesting indeed. Lots and lots of yellow and if my eyes do not deceive me (am a bit colour-blind, says my wife) very small blue specks as well.

edit: This was the 1000th post of the Arctic Sea Ice blog. Allow me to congratulate myself! ;-)

Anu

@ siili | July 29, 2010 at 18:52

What do you mean, the light blue cloud of sea ice in the Greenland Sea ?
http://iup.physik.uni-bremen.de:8084/amsr/arctic_AMSRE_nic.png

http://www.geographicguide.net/pictures/arctic/arctic-map.gif

I wonder if that perpetual cloud of melting ice is where the sea ice that flows out of the Fram Strait dies - does anybody know how much sea ice is flowing out the Fram this summer ? Are there still buoys there, or someone analyzing the satellite image sequences ?

That giant floe that broke off a week or so ago (what was it, the size of Corsica ?) has broken up further, but not drifted south much. Perhaps farther out into the Greenland Sea, sea ice flows faster from the north.
http://rapidfire.sci.gsfc.nasa.gov/subsets/?subset=Arctic_r02c03.2010209.aqua.1km

Neven

Anu, here you can see the buoys. But I tend to call them 'the boys' because they look like sperm cells.

The boys are again pointing towards Fram Strait. Somehow I think ice transport through Fram hasn't been as high as it can be this melting season, but I'm not sure and don't have any numbers.

The poor little fellow that went in and out of Nares Strait is now back in again. What a delivery. ;-)

siili

There are clouds in the Kara sea as well that has been very reluctant to melt helped by cold water and cloudy skies, we'll know soon how the trimaran cope with that. But even clouds count in extent.

A few days back i asked if anybody knew what happened to the newly deployed IMBbuoys and when i looked at Perovich's site today, they where there with the guy dooing the Nares waltz, but only positions and no data. But in the July SEARCH update, one got mentioned citing melt data, so any more news about that?

Artful Dodger

@siili | July 29, 2010 at 21:39 "But even clouds count in extent."

Sea Ice Extent (and Area) are compiled from data obtained from the AMSR-E instrument aboard NASA's Aqua satellite.

It's passive microwave detector can accurately distinguish between precipitation rate, cloud water, water vapor, sea surface winds, sea surface temperature, sea ice concentration, snow water equivalent, and soil moisture.

So, Cloud does NOT count in extent.

Anu

ftp://ftp-projects.zmaw.de/seaice/NEAR_REAL_TIME/Arc_latest_large.png

Is this what thin ice looks like right before an enormous melt ? (Beaufort, Chukchi Seas)
Or, did the winds just shift and push some floes around ?

Film at 11...

Neven

Anu, you'll be pleased to hear I have just posted a blog post with animations from Uni Bremen, PIPS and OSI SAF sea ice concentration maps. Eat your heart out. :-)

Artful Dodger

Anu said: "The Chukchi, East Siberian and Laptev Sea ice is all melting away, and will be the big areas of melt in the month to come"

ftp://ftp-projects.zmaw.de/seaice/NEAR_REAL_TIME/Arc_latest_large.png


Indeed, the Uni-Hamburg graphic above is updated twice per day, and is the both the quickest updating and highest resolution Arctic Sea Ice map available. I base my animations on it, often cropping areas of interest. Plus, the images are archived on the same ftp sever. What's not to like!

The July 29 23:35 UTC update shows substantial loss of sea ice extent SE of the New Siberian Islands compared to 04:40 UTC, where spreading occurred over the last 2 days.

Since both Uni-Hamburg and IJIS obtain sea ice data from AMSR-E, I expect a robust decrease in sea ice extent reported tonight by IJIS. We shall see!

Neven

I won't jinx this one! Lord Soth, you shut up too! ;-)

I'll add an animation of Uni-Hamburg to the latest post.

Edit: It's up.

Artful Dodger

Hehe. But seriously have you seen this archive of Arctic SST anomalies ?

The maps summarize each month from June 2002 to present. The Jun-Nov 2007 sequence is especially interesting...


Anu

@Artful Dodger | July 30, 2010 at 03:01

I also said, right before that:
The Beaufort Sea is now the warmest sea in the Arctic (up to 16° C), confronting a big chunk of sea ice and the Arctic Basin beyond.
ftp://ftp-projects.zmaw.de/seaice/NEAR_REAL_TIME/Arc_latest_large.png

I think a huge melt from the Beaufort Sea inward is also coming in the month ahead...
The "Arctic Basin" seems to start at the latitude of the northernmost part of Banks Island, according to Cryosphere Today:
http://arctic.atmos.uiuc.edu/cryosphere/IMAGES/region.mask.gif

------------------------------------------------------------------------------------------------
Hey, there really was a film at 11...
@Neven, great minds think alike.

Neven

AD, that's one hell of a SST anomaly archive. I've asked Robert Grumbine if those OPHI SST daily images are archived anywhere, but apparently not for the time being, so I'm saving them every day and perhaps turn it into an animation next week or so.

Anu, I don't know about great, but our minds seem to be synchronous today. ;-)
It was siili's comment further up that made me have a closer look.

Artful Dodger

Anu: You haven't been telling porky pies... You're posts are so info-packed it's difficult to take in their full extent (cor blimey, mate)!

SSTs will be critical for the remainder of Summer 2010. The ultimate fate of the pack may be determined as Pacific heat confronts the last remaining Multi-year Arctic Sea Ice.

A pitched battle is shaping up. If warm water invades in force, the pack is toast.

Artful Dodger

Well today's decrease in extent is slightly higher than yesterday's (-51,562 before revision) but still well below the 5-day moving average: -60,031

Phil263

-51,562 for now, probably less tomorrow. It looks like we will still be above the 7,000,000 mark at the end of the month or not much under.

Evilreductionist.blogspot.com

I have to say that at this point Goddard's prediction looks in very good shape. A prediction of 4.7 to 4.8, which is where I was sitting, does not look too brilliant now. If Steve Goddard does get it right this year, I think that will boost the credibility of WUWT, which is not something that I think is good at this point, given that the ice is in a very bad way and it is weather keeping the extent from sliding this year.

Phil263

EvilReductionist

I have to say that at this point Goddard's prediction looks in very good shape

What is Goddard's prediction for this year?

Evilreductionist.blogspot.com

5.5

Phil263

@Evilreductionist

He might even be pleasantly surprised !

siili

I thought that the contect made it pretty clear that i was refering to Anu's "clouds of ICE" on the ground which gets counted in the extent in contrast to the fluffy things up in the sky which are not.

But as reality is not always as clear cut as we like it to, passive microwaves can be fooled by weather like rough sees beeing interpreted as ice.

Another thing has has shown up very clearly in the last couple of days is that clouds can mask disintegrating ice as a solid pack. If you compare microwaves and optical images from the same day you will notice that clouded bands correlate well with high concentration ice.

So this time of year, area is probably overerestimated compared to early in the melt where ponds give an underestimation of ice area.

Artful Dodger

Cool. Thanks for your comments, siili.

siili

Thanks as well Dodger. I found this article by the German team to be a good introduction on how to handle microwave data.

http://www.ifm.zmaw.de/fileadmin/files/images/Staff/Gunnar_Spreen/pdf/SpreenEtAl2008-JGR-AMSRE89GHzSeaIce.pdf

They only update their maps twice a day i think, so if one is impatient, the Danes are a bit quicker http://www.seaice.dk/iwicos/latest/ but unfortunately their site is a bit erratic, not always showing the latest maps.

Artful Dodger

Tom Hanks for the brass tacks, siili !

siili

Bremen's own extent graph is in for the day, and it looks like a small uptick, should give a new record for area/extent when Cryos area numbers arrive. They must use very similar reduction algorithms since their maps correlate so well.

Is there a daily updated list with both extent and area derived from AMSR-E, i have lost my link to Cryos hidden daily area table and don't remember if that included extent as well?


Artful Dodger

siili: This help?
http://arctic.atmos.uiuc.edu/cryosphere/timeseries.anom.1979-2008

Phil263


siili:

"Bremen's own extent graph is in for the day, and it looks like a small uptick"

Does this mean we'll get a positive change in SIE from Jaxa tomorrow?

siili

Thanks Dodger, that is the one! And it had both numbers, so just a small wait left for todays values.

What Jaxa does in detail is still a bit unclear to me, they do a two day moving average so that alone gives some delay to changes, but from their pictures it looks that they have a different reduction algorithm since dispersing ice seems to last longer??

There is also a large pedestal between Jaxa and Cryo, like yesterday Jaxa had 7.13 and Cryo 6.12 for the extent, so we just have to wait and see i suppose?

Artful Dodger

siili: When you do your analysis, see if there is a correlation between a spike in Area/Extent ratio and subsequent collapse in extent. Ideally, we'd have data per basin, but maybe there's some signal/noise in this data.

siili

Well, my data manipulating skills are a bit rusty, but from a quick look at area/extent for the last four years, there may be something to your hypothesis. 2007 and 2009 both have rather gradual drops whereas 2008 have a fast drop from about 0.85 to 0.70 between 26/7 and 12/8 something, which is in the right ballpark. So far 2010 have dropped from 0.83 to 0.77 between 18/7 and 28/7, still no point for yesterday.

The basin data must hide in the same catalog, we just have to find the links.

I'll continue my fumblings, but maybe others have done it right already?

FrankD

Sorry for going all OT, but I was checking the latest MODIS pix and noticed the incredible amount fire activity in Eastern Siberia:
http://rapidfire.sci.gsfc.nasa.gov/subsets/?subset=Arctic_r06c05.2010211.terra
http://rapidfire.sci.gsfc.nasa.gov/subsets/?subset=Arctic_r06c04.2010211.terra

Yesterdays scattered fires got fanned by strong southerlies and *bam*. Forecast is for rain, which should help, but that comes with more wind, and maybe lightning strikes.

Sorry. Ice. That's the white stuff. Back on topic....

Jon - I was one of the people talking, so thanks for that snapshot. Doesn't seem to be an obvious pattern from that. A lengthwise slice on contrasting years, say 2006 (low max, high minimum) and 2008 (high max, low min) might show something...I'll take a look over the weekend if no one gets there first.

siili

Please forget my previous post, i warned you that i was rusty. It turns out that what i thought was extent is only the historical mean, so what i looked at was just a variant of the usual anomaly plot, only using year/mean instead of year-mean. Sorry.

L. Hamilton

With this focus on area vs. extent, I thought it might be interesting to revisit the cycle plot idea, this time with both area and extent (NSIDC monthly means).

Winter ice area seemed more stable than extent, beginning its downward trend more recently. Summer ice area has been dropping precipitously in recent years.

http://img.photobucket.com/albums/v224/Chiloe/Climate/Cycle_Arctic_extent_4.png

Neven

Meanwhile Cryosphere Today sea ice area anomaly has dropped some more to -1.463 million square km. East Siberian Sea is losing sea ice area fast, the trend line is almost dropping vertically.

PIPS is still forecasting big arrows pointing towards Greenland and the Canadian Archipelago. As long as this is so, I don't think we'll see any (extent) century breaks. SLP over the Arctic still looks very jittery as well, with AO still positive.

The hole north of the Beaufort Sea is very pronounced.

Neven

BTW, Larry, I think I'll use that second graph (with area alongside extent) as well. Thanks for making it.

Jon Torrance

"Goddard's prediction looks in very good shape" and "He might even be pleasantly surprised !"? Let's not get carried away. Certainly this past month has demonstrated that weather can play games with extent, despite continuing decreases in area pretty much in line with past years, but I don't think anyone has shown a reason to believe that the weather is going to continue to spread the ice out throughout the rest of the melt season. Given that and the fact that, going by JAXA data, in only one of the last eight years would the remaining melt from July 29 to the minimum be small enough for Goddard's prediction not to be too high, I think the odds are still in favour of Goddard's prediction being too high. 4.7 to 4.8 may well end up too low but the remaining melt of either 2004, 2007 or 2008 would suffice to give us a minimum under 4.9 so Evilreductionist's prediction still looks better than Goddard's to me, even without taking into consideration the current low concentration of the ice pack.

Speaking of which, I updated my CT area/JAXA extent calculation for July 29th and got the following:

Year Area/extent
2003 72.2%
2004 71.2%
2005 72.2%
2006 72.0%
2007 68.0%
2008 69.9%
2009 73.4%
2010 64.9%

2002 is missing due to a gap in the JAXA extent data. I'll note that 2007 through 2009 are all within 1% of their July 26th value but 2010 concentration has dropped over 3% (I haven't found any explanation of how they produce CT area data yet but the result is certainly noisy, as Neven has noted in the past). Assuming the difference between JAXA extent and CT area can validly be interpreted as a measure of the amount of open water inside the boundary of the area with at least 15% concentration, then yesterday there were 609,691 sq kms more of open water inside the ice extent than on the same date in 2009, while the extent numbers for the two years on that date were nearly identical. Put another way, if 2010 ice was currently as concentrated on this measure as 2009 ice, the JAXA extent would be 6,273,877, i.e about 200,000 sq kms lower than 2007 extent on the same date. To me, that says that, all else being equal, we should expect a bigger remaining drop in extent this season than in 2009. The weather may, of course, cause all else not to be equal but I don't think anyone really knows what the weather is going to do through the middle of September so I'll stick with believing that the reasonable expectation is that Goddard's prediction will end up being to high.

Neven

Jon, that's some great reasoning. Insightful, original even, but simple enough, so that even I can follow what you're trying to say. The percentages you produce there are very indicative of what is happening with regards to the spreading of ice.

It is quite amazing really that just after we had the most adverse conditions for sea ice melt imaginable in the 4 weeks that matter the most, 2010 could still be on a path towards a minimum extent below 5 million square km.

Artful Dodger

Good analysis, Jon. I think you could estimate 2002 JAXA extent with linear interpolation (their numbers are just estimates of the underlying variable anyway). 2010 clearly has the lowest Area/Extent ratio during the AMSR-E era.

I was speculating idly yesterday about this year's Arctic weather. Wayne Davidson predicted 'Big Blue' over the North Pole with the waning El Nino, which seemed to drive the weather until about June 28. Since then it's been steady low pressure and cloud.

I wonder if all the open water in the Arctic has tipped the Polar weather to a Maritime Climate. This type of transition occurs in Northern Canada with a Continental Climate during Winter (after freeze up, the RH goes down and temps crash). Then in Summer after break up, humidity soars due to the abundance of lakes and the North switches to a Maritime Climate.

Jon, I wonder if there was an notifiable transition in the Arctic Sea Ice Area/Extent statistic in late June? Thanks again for your excellent analysis. Neven, should we post a curve plot?

Neven

AD, a curve plot is a graph, right? I'm not sure if I know how to make one.

I think a post on extent/area might be appropriate, especially after those concentration animations I have just made. With Tamino's recent post and some of the comments/graphs/calculations here there should be enough material.

With regards to Wayne Davidson. He updated his blog a few days ago with the following:

Great Asian Russian heat have disturbed planetary waves apparently forcing a near permanent low over the Arctic Ocean, as the last cool place in the Northern Hemisphere, is surrounded by very warm weather. I see this phenomenon as a natural effect from less cloudy land areas, causing prolonged slow moving anticyclones, fed by dryer hotter air, I have never seen these patterns before. This of course slowed down ice compressing over the Beaufort Gyre, having in effect caused a stall in what would have otherwise have been the greatest melting ever. The lack of persistent North Atlantic lows is another key feature making the UK unusually dry and warm. This is very interesting. If there is a persistent high over the North Atlantic, this would eventually accelerate the melt over the Atlantic side as opposed to the Pacific Arctic, making eventually the North Pole vulnerable to open water from 0 degrees meridian instead of 180. Given that ice will flow towards the Atlantic, a constant replenishment of ice gives the illusion of no melting, also making unlikely event that the Pole would be ice free. However, 2 months away from freezing onset, ice extent is already lower than many preceding years. But the Atlantic side melt is to watch closely. Not to say that gyrations towards a stronger La-Nina, ie even lesser cloud seeds, would make conditions better for a high anticyclone over the Beaufort Gyre area, but It is again certain that the Northwest and Northeast passages will be ice free.

Artful Dodger

Yup, a curve plot is a graph created with your spreadsheet.

Wayne sounds a little punch-drunk. ;^)

Nick Barnes

Make graphs using Google Charts.
http://code.google.com/apis/chart/index.html
It's very, very cool. And it lets you put your actual data in the IMG URL.

Neven

Can anyone tell me where I can get that CT area data? Is it a spreadsheet, like the IJIS extent data? In that case I might be able to pull something off. I'm not promising anything. And stop laughing.

Jon Torrance

Neven,

It's the link AD posted on July 30, 2010 at 11:26. I've already got it in a spreadsheet but the effort to import that into Excel is small enough you might as well do it yourself.

BTW, when I look at it over the course of the summer (my current chart runs from June 1 to September 30), the picture gets more complicated. Just as a for instance, as recently as July 21, the years from 2007 through 2010 were all in a virtual tie for concentration. You'll see for yourself soon but I'll no doubt say something else about it once I've pondered a bit. One thing that occurs to me is that perhaps it would make sense to do a two day moving average of the CT area data in order to match the period on which each extent number is based. Leaving aside the theoretical correctness, it would smooth out some of the noise in the CT area data.

Artful Dodger

Today's change in SIE is -76,563 before revision. So clearly we're seeing some effects of the 2-day averaging done by IJIS.

L. Hamilton

After reading Tamino I realized that my cycle plot upthread failed to adjust for the 1987 discontinuity in area measurements, which occurred due to a larger "hole" of unobserved Arctic in the earlier satellite records.

I went back and made the same correction Tamino describes (adding 1.19 to area estimates up through June 1987, and 0.31 after) and redrew my graph, as follows -- it has the same URL as before, so if anyone linked to that, no correction is needed:

http://img.photobucket.com/albums/v224/Chiloe/Climate/Cycle_Arctic_extent_4.png

With this correction, the area and extent squiggles match a lot better in the cycle plot. I wondered why area had earlier appeared to go up, then down. The answer is, that was an artifact.

L. Hamilton

Tamino offers his prediction of mean September ice extent, based on a simple quadratic model. Here's my own version of his graph:

http://img.photobucket.com/albums/v224/Chiloe/Climate/September_extent_predict.png


Applying a similar quadratic model to area yields a predicted September mean of 3.4:

http://img.photobucket.com/albums/v224/Chiloe/Climate/September_extent_predict.png

L. Hamilton

Sorry, that last graph was supposed to be area, not extent again:

http://img.photobucket.com/albums/v224/Chiloe/Climate/September_area_predict.png

Phil263

AD

Sizeable change in SIE today. We will definitely be under the 7,000,000 by the end of the month...

L Hamilton:

I showed your seasonal graph to a few "pseudo-skeptics" who are still in doubt that climate change is really happening in the arctic.. Clear evidence in my view. Great work !

Note that you second link in the previous comment shows extent not area.

Artful Dodger

Phil, I suspect the big drop in extent was actually yesterday, and reflected more in today's number due to 2-day averaging. As siili noticed yesterday, Bremen posted an uptick in SIE overnight, and now we can see IJIS Sea Ice Area graph also has a sidestep today. Be interesting tomorrow. It'd sure be nice to have a straightforward statement from each Organization as to how their reporting is done.

Artful Dodger

Larry, good job on the graph update. Can you post your regression equation please? I'd also love to see R-squared or StdErr. Is the vertical error bar around the 2010 plot a 95% confidence interval? It is quite large when visually compared to the distance of other data points to the regression curve. Oh, and thanks again!

siili

Ok, Cryo has reported for the day and at least i am a bit surprised that the anomaly is still dropping with result that the area/extent number is even a bit smaller with 63.9% extending the lead compared to previous years using the uncorrected Jaxa number.

Winds continue to blow in about the same direction, DMI is on the rise and just another wait untill Bremen reports todays image in about three hours.

Ice watching is as addictive as testcricket, and then you don't have to take a break after five days.

L. Hamilton

Just for Artful, here are the regressions (seen best in Courier font):

. describe year year0 year2 extent area2

storage display value
variable name type format label variable label
------------------------------------------------------------------------------------
year int %8.0g Year
year0 byte %9.0g year-1994
year2 int %9.0g year0^2
extent float %9.0g Mean September NSIDC extent
area2 float %9.0g Mean September NSIDC area -- corrected

. regress extent year0 year2

Source | SS df MS Number of obs = 31
-------------+------------------------------ F( 2, 28) = 38.03
Model | 17.1086288 2 8.55431441 Prob > F = 0.0000
Residual | 6.29755612 28 .224912719 R-squared = 0.7309
-------------+------------------------------ Adj R-squared = 0.7117
Total | 23.4061849 30 .780206164 Root MSE = .47425

------------------------------------------------------------------------------
extent | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
year0 | -.078625 .0095232 -8.26 0.000 -.0981323 -.0591177
year2 | -.0033518 .0011923 -2.81 0.009 -.005794 -.0009095
_cons | 6.898787 .1278778 53.95 0.000 6.636841 7.160732
------------------------------------------------------------------------------

. regress area2 year0 year2

Source | SS df MS Number of obs = 31
-------------+------------------------------ F( 2, 28) = 45.24
Model | 13.4330179 2 6.71650894 Prob > F = 0.0000
Residual | 4.15657556 28 .148449127 R-squared = 0.7637
-------------+------------------------------ Adj R-squared = 0.7468
Total | 17.5895934 30 .586319781 Root MSE = .38529

------------------------------------------------------------------------------
area2 | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
year0 | -.0711452 .0077368 -9.20 0.000 -.0869933 -.055297
year2 | -.0023586 .0009686 -2.43 0.022 -.0043427 -.0003744
_cons | 5.141265 .1038908 49.49 0.000 4.928454 5.354076
------------------------------------------------------------------------------

L. Hamilton

Hmm, that looks messy, the software seems to drop extra spaces. All the info is there, though. A couple of notes:

I regressed September mean extent and area on (year0 = year-1994) and (year2 = year0^2), in other words centered the X variable before squaring.

Tamino estimated his confidence bands from the residuals, but I was lazier and just drew mine as plus or minus twice the OLS standard errors of prediction. This gave me slightly wider bands than Tamino's.

You often see much narrower confidence bands with regressions, where people use standard errors for the conditional mean rather than (as I did, and Tamino did by a different method) standard errors for an individual forecast. Since we're focusing on one year, the narrower conditional-mean bands seem less appropriate.

L. Hamilton

One last techy note, then we take the dogs hiking. I often use toned-down colors in my graphs, but the bright blue and orange in those last few are an experiment suggested by Nick Cox. Reading a draft of a paper I'm working on, he commented that many of my color choices would be indistinguishable to someone with common forms of color blindness.

He recommended orange and blue as two colors that even most colorblind readers can distinguish. So I'm experimenting. Anyone have feedback?

Neven

Larry, my wife says I'm a bit colour blind, but that graph looks more than fine to me.

Jon, I'm having some trouble with the CT area data. First of all, I can't just copypaste the data. Second, the numbers aren't daily, right? They seem to divide every year in 1000 parts and then make 25-30 jumps. How do I know which area number to stock to its extent counterpart?

I've asked IJIS if there's any way for me to get a csv-file of sea ice area data, just like they have of extent data.

Artful Dodger

Hi Larry, call me 'Lodger', please! Thanks for posting R-sqrd, 0.747 shows a pretty strong signal. I especially liked how you set year0 to 1994, so your parabolic curve has its origin at approx. that point in time. Curious, did you attempt other fits for year zero? Notionally, 1994 sounds good but its always interesting to see what a regression spits out. I also agree with your call on the Error Bar, thanks for the explanation.

Jon Torrance

Neven,

If IJIS will give you their area data, that's even better. I'll be damned if I know why you couldn't simply copy and paste that table; that's all I did. As to the numbers, I'm pretty sure they are daily despite the unusual format for expressing the day of the year since they matched up to the daily numbers in recent editions of the chart at http://arctic.atmos.uiuc.edu/cryosphere/IMAGES/seaice.recent.arctic.png and there were the right number of rows per year for them to be daily going all the way back to the beginning of 2002. I just filled in more conventionally formatted dates for the last few days in the series and then used the handy Excel function where you left click on the lower right corner of a block of cells and drag the cursor up or down to extend whatever ascending or descending series was in those cells to fill in the dates back to the beginning of 2002.

Artful Dodger

The provisional change in IJIS Sea Ice Extent for July 31 is -92,812 km^2. Their Area Graph however shows a pronounced slowing in the rate of Sea Ice Area decline. We are probably seeing a situation where wind and current is compacting the sea ice, rather than an increase in daily melt.

Jon Torrance

Assuming AD is correct (and it's hard to see what other interpretation to put on it) I wonder whether the area melt has necessarily slowed or whether some areas that were below 15% concentration have been compacted back to above 15% concentration, effectively adding ice back to the area figure that had been excluded despite not having melted. That assumes IJIS tdefines sea ice area in the same way NSIDC does, with the only differences between them being the sensors they use and how they process the sensor data.

Not that the area melt in terms of the actual total amount of ocean surface covered by ice couldn't possibly have slowed. I may simply not be aware of good explanations for why it would have. Anyone got anything in the way of extreme cloudiness or lower sea surface temperatures or the like that would explain a slowdown in the rate of ice turning into water?

L. Hamilton

Curious, I played out the quadratic models to see where they lead. In case you're curious too, here they are:

EXTENT
http://img.photobucket.com/albums/v224/Chiloe/Climate/September_extent_predict2.png

AREA
http://img.photobucket.com/albums/v224/Chiloe/Climate/September_area_predict2.png

Anu

@ Artful Dodger | August 01, 2010 at 05:15

Yes, I noticed the same thing. Pretty impressive extent drop.
Notice too that 2010 is again less than 2009 in extent:
2010: The latest value : 6,915,938 km²
2009: The last value: 6,955,469 km²

That dip in sea ice area is interesting:
http://www.ijis.iarc.uaf.edu/seaice/extent/AMSRE_Sea_Ice_Area.png
2004, 2002, 2003, 2006 and 2009 all had similar little down-right-down dips in their curves at just about this day of the year - I wonder if it is related to some underlying physical cause, like an ocean current reversing at this time of year or something...
Curious, anyway.

I expect the sea ice area to resume its drop for another few days, at least. The first week of August seems to be when the sea ice area curve will begin to level out or not - July 31 is too early for that.

Yes, definitely a bit of compaction along the sea ice coasts - I downloaded
ftp://ftp-projects.zmaw.de/seaice/NEAR_REAL_TIME/Arc_latest_large.png
from today and yesterday, zoomed in equally on both in two different windows, and blink-compared them. Some parts of the Beaufort and East Siberian Seas looks like just some compaction on the coasts, but other parts of the Arctic are spreading out more (like the big sea ice pack nearest the Bering Strait).
It's a complicated mix of drift, melt, compaction and expansion.
Not sure what is happening with the sea ice area decline-slowing today...

Phil263

CT numbers as of the 30th still show a sharp drop in SIA (In fact, the graphs show a drop in almost all areas). Why the difference with IJIS?

Artful Dodger

Anu, one thing I have noticed recently wrt sea ice motion: The Central Basin pack is no longer responding as a single mass ie: rotating counter-clockwise about a central point in response to cyclonic (low pressure) winds.

Instead, the ice pack seems to be spreading from the center, in response to centrifugal force (look at the big bare patch centered at 179W 77N in your icemap).

If this is the case, one of the key physical properties responsible for the creation / retention of multi-year ice has fundamentally changed. Without the rotational inertia delivered by the mass of the entire central ice pack, there is reduced energy to create large pressure ridges, which thicken the ice.

Anu

@Phil263 | August 01, 2010 at 06:14

IJIS is showing data for 7/31 already - IJIS is a collaboration between IARC (International Arctic Research Center at the University of Fairbanks, Alaska) and JAXA (Japan Aerospace eXploration Agency), and I suspect that the researchers in Fairbanks keep some pretty late hours to collaborate with their Japanese colleagues:
http://www.iarc.uaf.edu/
http://www.jaxa.jp/index_e.html

Cryosphere Today is at University of Illinois at Urbana-Champaign.
Different time zones, and perhaps different time spent on doublechecking data and time-smoothing, lead to different release times of data. IARC-JAXA released 7/31 data at about 11 pm EST in the US on 7/31. Perhaps CT will just wait till some grad students come in Sunday morning before they post on their website...
http://arctic.atmos.uiuc.edu/cryosphere/

CT will probably show a stall in sea ice area decline for 7/31 too.
If not, I'll have to think about the time zones and procedures at different institutions more closely...

Artful Dodger

Hi Larry. Interesting graphs, with vastly better R-^2 fits than a linear regression. However, I'm not a big fan of efforts to model future SIE (or area) from past values, as one is predicting a variable from the variable (I don't play the stock market either). "Useful until something the underlying fundamentals change..."

I'd like to see an effort to predict Sea Ice (as the dependent variable) from cyclical physical properties such as Year + PDO + ADO + ENSO ( as the independent variables ).

Look at the work of D. Kelly O'Day in modeling Land and Ocean Temperature Anomaly (LOTA). The four ind. vars above, used in a multiple regression analysis, produce a remarkable fit.

The first thing I'd do is a Factor Analysis with every plausible variable, and look for the signal. If the results load on 'Year' start looking at CO2, Solar Insolation, Methane, etc... Any independent variable we have history for.

Anu

@Artful Dodger | August 01, 2010 at 06:25

Interesting idea, centrifugal force leading to ice pack spreading...
I don't think the archives at Uni Hamburg show anything like that being noticeable before.

I wonder if a giant chunk breaking off is more likely than a shattered-glass spread everywhere - maybe a chunk almost half the size of Greenland could break off by the Chukchi/East Siberian Seas, drift to more southerly latitudes (well, a few degrees, but maybe shatter further as it hits land), and melt faster as it enters much warmer waters:
ftp://ftp-projects.zmaw.de/seaice/NEAR_REAL_TIME/Arc_latest_large.png

But maybe this summer is too early for such radical breakup - those fast moving blue "holes" seem more like "The Wave" at a stadium, a phenomena of sea ice floes moving very little, but making it look like a big hole is moving within the ice pack. Maybe this is the end of the beginning, but not the beginning of the end.

The next month will be interesting - whatever condition the ice pack is in now, it will get much thinner around the edges. I'm not sure how far underneath the ice pack the warming waters go - I suppose it depends on winds, currents and seafloor topography.

I wouldn't be surprised if the ice surprises me in August :-)

Artful Dodger

Anu, I'm not sure there are any chucks that size left in the Chukchi/East Siberian Seas. The occasionally cloud-free peek we get with MODIS shows lots of ice floes. We just don't have a handle on the physical properties of the ice like strength in sheer or compression.

This is the data I'm hoping one of the in situ surveys planned for this summer will provide. David Barber's group is headed back to the Beaufort Sea on the Amundsen this Sep, but I don't know what is planned for Chukchi or E. Siberian Seas.

Wouldn't a particle / ray tracing animation package, programmed with the right variables, create a cool animation of the Central Basin ice pack? Any programmers in the forum?

siili

I think that the updates of crypsphere today is totally automatic and released on a fixed time, remember a few days ago, the updates stopped coming and it took a few days before anything happened.

So i expect it to look very similar to the Bremen map today as well.

The uptick in the area of the two day averaged Jaxa maps could be caused by Nevens blinking when more area is troubled by the pink clouds.

If this is the case, Cryos area might continue to drop today

Jim Dowling

Interesting August coming up, where I expect lots of ice loss in the Beaufort + Chukchi seas.
The massive wildfires in eastern siberia will be depositing a lot of soot on the ice pack - see
http://rapidfire.sci.gsfc.nasa.gov/subsets/?mosaic=Arctic.2010213.terra.4km

From Aug 6 for at least 5 days, it is expected that a large high pressure will sit over the pole:
http://weather.unisys.com/gfsx/9panel/gfsx_500p_9panel_nhem.html
That will cause in-situ melting, although the sun aint so strong. However, the soot from the fires should increase the albedo effect.
The high-pressure weather is not the di-pole anomoly setup. No warm air will be sucked up over the pole. In fact, the expected high pressure is the opposite of the typical low pressure over the pole at this time of year.

Neven

Jim, thanks for that. ECMWF is also showing some big highs in and around the Arctic starting about 5 days from now. I'm hoping this comes about as I'm dying to make some animations.

The Uni Bremen sea ice concentration map has blinked back to the state it was in two days ago, ie a lot of yellow and green.

Jon, I think something has gone wrong when importing the CT area data into a ods-spreadsheet, because the data I need in the third column, is divided over two columns in OpenOffice, which means I manually have to copy 2 rows from one column, then copy 12 rows from the other column, then copy 9 rows from the one column, etc. I'll just see if IJIS answers.

Artful Dodger

Artful Dodger said | July 28, 2010 at 06:13

"I'd like to track this statistic: Area/Extent. It would give us a sense of how 'spread out' the pack is at any point in time, an important predictor for melt.
---
There is a story emerging from the Area/Extent statistic. I just completed a preliminary analysis of ( CT Area) / ( IJIS Extent). Interesting for 2010, the ratio holds very flat at mostly 98% +/- 1% from March 5 until June 28. Then, the statistic begins a steady near linear decline to the present value of around 85% by the end of July, 2010.

I need to do some more checks on alignment between the two data sets, and create some graphs. However, this date range matches very well with the behavior we've seen in decline in rate of change for SIE in Summer 2010.

Here's a quick tease: If the Area of sea ice area present July 30, 2010 was compacted into the same ratio of Area/Extent as that date in 2007, then SIE for 2010 would be 6,428,125 km^2. This is exactly the same SIE value as 2007 so I've likely done something wrong! However I've checked it twice now and this Forum is about discussing the Science, so I want to get this out there now if people are trying to replicate the results.

I'll post on this topic again in a day or so after looking at my spreadsheet with fresh eyes. Wouldn't it be nice if we had a common files area like Yahoo Groups provides?

Lord Soth

Well 2010 pulled ahead of 2009 on July 31 with a preliminary melt of 92,812 sq. km.

There is still a slim possibility of a centruy melt if we got a favorable revision, later this morning.

2009 led 2010 for just 4 days. 2010 will now leave 2009 in the dust. With an average melt of 48,654, and no century melts, 2009 will not stand a chance of competing.

The new race will be with 2008. With 5 century melts and an average melt of 70,121; 2008 will stand a good chance of overtaking 2010. Things will be much clearer in seven days, as the average melt for the first seven days of august 2008 was an amazing 102,567 sq. km. per day.

FrankD

I've gone through the JAXA graph of area and compared it to their numbers of extent and posted an image of the results here:
http://img831.imageshack.us/img831/1870/areaoverextent.png
(Hope I haven't fubar'ed the link). It covers April through September only.

The percentages are much higher than Jon's due to difference is JAXA and CT's area figures (don't know why). I haven't really had a chance to think about it much, so no conclusions as yet. I was kind of hoping there'd be obvious troughs in the percentage just before big melts, but nothing leaps out at me at this point.

One for the brains trust to consider.

FrankD

Well I didn't fubar the link, just one of the formulae behind the April data. *sigh*

Correct graph can be found here:
http://img269.imageshack.us/img269/1870/areaoverextent.png

Neven

Frank, that graph looks very nice. Can you explain how you obtained those IJIS area numbers? It is funny that those numbers look so radically different from Jon's CTarea/IJISextent calculations:

Year Area/extent
2003 72.2%
2004 71.2%
2005 72.2%
2006 72.0%
2007 68.0%
2008 69.9%
2009 73.4%
2010 64.9%

Neven

Frank, one other thing I notice is that the 2010 trend line stops at July 8th, the point where things are getting interesting with regards to the recent diverging/spreading of the ice.

Jon Torrance

Still absent information, it occurs to me that the simplest explanation for the large discrepancy between CT sea ice area and IJIS sea ice area would be that CT uses a concentration threshold higher than 15%, maybe 30% like DMI does. Presumably, if that's it, a CT extent figure would be smaller too but they don't seem to publish one.

AD - your description of your preliminary CT area/IJIS extent results is radically divergent from what I got. I'm fairly certain I didn't do anything horribly wrong so I hope you see some mistake in your spreadsheet after a good night's sleep.

Neven

Jon, for area there are no concentration thresholds. I've explained some more in my latest blog post: area vs extent. I've used some of your numbers as well to show what you get when you divide area numbers by extent numbers.

Neven

I'll post on this topic again in a day or so after looking at my spreadsheet with fresh eyes. Wouldn't it be nice if we had a common files area like Yahoo Groups provides?

Maybe next year, AD! We'll have to make do with this ship until the melting season is over. ;-)

Jon Torrance

Neven,

I'd have to go find the link and then it would take you a while to reread NSIDC's lengthy explanation of their data but they are quite clear that any sea ice area number they produce excludes sea ice in cells that they determine to have a concentration below their 15% threshold.

Neven

Jon, I have finally managed to get the CT area and IJIS extent data into a spreadsheet and our numbers don't match up. As mentioned in the new Area vs Extent post I have uploaded my spreadsheet in Googledocs. Do you have a clue as to what might be causing the difference? Which column in the CT area data did you use?

I probably screwed up something with the dates. I'm very bad at these things when I take my time and this I did in a hurry (took me over 2 hours nonetheless).

Jon Torrance

Neven,

I think the only problem with your spreadsheet is that you assumed the last row of CT data was for August 1 rather than July 31st. Otherwise, it looks like you did everything the way I did it. Since I matched numbers to the CT area chart, which doesn't have any text indicating the date that I could see, I assumed the chart had a one day lag just like the ice concentration maps on the CT front page - which could be wrong, for all I know but seemed like the sensible assumption.

Neven

It seems sensible indeed. I will correct the spreadsheet tomorrow, am a bit fed up with graphs and spreadsheets at the moment.

Jon Torrance

On further checking, I take back my admission of possible error just above. Unless "2010.0000" doesn't represent January 1, 2010, then the last row in the table is definitely the one for July 31st.

Neven

It also explains the slight difference between our numbers. I'll just nudge all the rows one row upward and that should solve it, right? Thanks, Jon!

Verify your Comment

Previewing your Comment

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

Working...
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.

Working...

Post a comment

Your Information

(Name is required. Email address will not be displayed with the comment.)