Sea ice update

I've often referred to cassandras and polyannas. The polynya thing was just a happy accident.

[You've missed my subtle dig at your spelling. Its "Pollyanna" :-) -W]

What a nice Scrabble word! I hope someone somewhere has had the chance to play "polynya" for the bonus and survive a challenge as well.

Was hoping you'd have a new graph, though.

It's not Polyanna!? Thank Cthulhu! No longer will I be afflicted by nightmares filled with endless crowds of unreasonably optimistic young female clones...

Actually the oddest thing about the article was the claim that it was an "exclusive." The content had been out in other media for at least a week.

The Sunset Times also has an article in a similar vein:

http://www.timesonline.co.uk/tol/news/environment/article4232375.ece

Ice area anomaly today is below anything before 2005.

http://arctic.atmos.uiuc.edu/cryosphere/IMAGES/current.anom.jpg

I'm not going to offer 2:1 odds (since I don't think the chances of a new sea ice record are better than even) but we can increase our current EUR 40 bet to EUR 60 if you like.

[OK, taken, based on todays NSIDC pic! -W]

I suggest that you read the actual story as printed in The Independent. It clearly states "Scientists warn that this summer there may be....no ice at the North Pole". Indescribly accurate, is it not? I can send you a pdf of the front page if you have trouble with believing me.

[No offense meant, I-O-H is just the traditional term, comes from Private Eye, doesn't it? Newspapers usually have tense problems: "Exclusive: No ice at the North Pole" strongly implies present tense, to me -W]

"comes from Private Eye, doesn't it?"

I thought PE calls/ed it the "Indescribablyboring", whereas IOH is JA's version (at least I think I first saw it on his blog), based on some AGW coverage?

[Well according to google, the source is me :-). But it could easily be JA -W]

Well that's one meaning of the word "traditional". ;)

McIntyre says you are favored to win your bets. have a look at his post

[Don't be a tease, provide the link... -W]

I'm on William's side of this bet (no new record) and would also be happy to take bets on this at climateaudit.

[Hey, you can't steal my money, unless you're offering better odds? -W]

I'm unimpressed by McIntyre's analysis (and even moreso by his tone, but that's a different matter). He seems to put an inexplicable amount of weight on the idea that, if 2008 were going to exceed 2007, it would have to show rapid melting during the exact same week when there was fastest melting in 2007.

Why? No reason given.

In fact, looking at McIntyre's figure (http://www.climateaudit.org/wp-content/uploads/2008/07/seaice35.gif), I see several cases where "high-melt" weeks occur on different Julian dates in different years. I think McIntyre's exuberance must have gotten the better of him on this one.

IMHO, this year's minimum will probably be close to last year's, but I remain agnostic as to which side it will fall on. It is amusing to watch the CA commenters leap to the conclusion that Steve's just proven something significant, and start planning the funeral for "sea ice alarmism".

J, take a chill pill. I just thought William would be interested.

I got my number in at Lucia a long while back.

http://rankexploits.com/musings/2008/test-post/

I picked 6.5.

last years record doesnt get beat. If I'm wrong I have to write a haiku or two i suppose.

[7.4 is indistinguishable from 7.7. I'd take the low side, if either is interested in putting their money where their mouth is. For 6.5 I'd be inclined to take the low side too, though with less certainty. Mind you, we'd have to sort out exactly whats being used, as Lucias figure appears curiously incompatible with the cryosphere today's -W]

I never suggested that I'd "proven" anything. I said that I was just guessing and explaining the basis for the guess. Time will tell - my take on the data is that 2008 is already too many games behind 2007 to win the championship, but I'm not saying that this is more than a guess.

William, not trying to cut into your book; you were betting when there was a lot more risk than I believe there to be at present. Only if there was some overflow or if you wanted to lock in some profits by laying off part of your bet at better odds. 2008 fell behind another 100,000 sq km today so it's going to be pretty hard for 2008 to gain in the backstretch.

From the NSIDC's latest report - as blogged here:

June sea ice extents in 2008 and 2007 are essentially identical, and near the lowest values for June ever recorded by satellite for the Arctic.

Then look at the graphic showing the onset of melting.

Start writing that cheque, William... ;-)

[I'm quietly confident. Notice that the NSIDC page contradicts itself - fig 3 shows june '08 lower than '07; fig 2 shows not -W]

If this year doesn't set a new record, and next year does, would you bet against 2010 setting a first consecutive record? If it depends on some factors, what would they be?

I'm quietly confident. Notice that the NSIDC page contradicts itself - fig 3 shows june '08 lower than '07; fig 2 shows not

I believe they are showing two different figures. According to the text, the 2007 extent for June averaged at 0.05 million square k less than the average for June last year. This is what Figure 3 shows - though it is only stated in the figure description. Figure 2 shows the extent at any point in time along the line, thus the current extent is higher, but the average was lower, probably due to 2008 line being below the 2007 line when the extent was larger (earlier in the month).

[Don't believe this. You can integrate f2 in your head easily enough, 2008 is clearly above 2007 on average -W]

From NSIDC:

Last year, open water quickly developed along the coasts of the Chukchi and Laptev seas. This year, an unusually large polynya has opened in the Beaufort Sea, and there is significantly less sea ice in Hudson's Bay and Baffin Bay.

Does this make the melt more or less dependant on wind patterns than last year, or not much difference? I'd have thought the Hudson Bay stuff was fairly indifferent to winds, but not so sure on the other stuff. Anyone?

Don't believe this.

From NSIDC, which seems to have been added (or I just missed it first time round):

Figure 2 indicates that on a daily basis, sea ice extent appears slightly higher than 2007 for most of the month. This apparent contradiction arises because of the monthly averaging calculation and because some days may have areas of missing data. To be included as an ice-covered region in the monthly average, the average concentration for that region must exceed 15 percent. So if the concentration is 15 percent for 29 days, but less than 15 percent for 1 day, it will not be included in the average ice extent for the month. Also, since ice extent decreases during June, if there is slightly more missing data in the early part of the month the monthly average could slightly underestimate the sea ice extent.

June sea ice extents in 2008 and 2007 are essentially identical, and near the lowest values for June ever recorded by satellite for the Arctic.

They also linked to this page:
http://nsidc.org/data/seaice_index/derivation.html#monthlymeanconcent

[I didn't see that first time, either. I bet they just stuck that in. I did email them to ask, and got that para back as a reply -W]

Just to say, that para: "June sea ice...the Arctic" should be italicised as it's part of the quote.

This year, an unusually large polynya has opened in the Beaufort Sea, and there is significantly less sea ice in Hudson's Bay and Baffin Bay.

Of course, Hudson's Bay and Baffin Bay are basically ice-free every summer, and the Beaufort Sea nearly so. So the fact that ice is melting faster than normal there seems not especially meaningful to me. The big question is what happens in the Arctic basin.

W.

"thats what happens when you rely on spellcheckers to proof read your articles"

What do you use to spell check words like "that's"? ;)

My bets are based on the sea ice data set Atmoz proposed when he suggested this bet. It's the summer column at this specific page at The Cryosphere Today. I believe if you visit Atmoz's blog, you'll see his plot of the summer ice extent is quite similar to mine. Both our plots are quite similar to the treen summer trace on the graph posted at The Cryosphere Today

I'm afraid I'm going to have to turn down your offer to bet for several reasons. When I bet, I only bet home-baked cookies. To avoid excess postage and stale cookies, I limit bets to people in the US. There are additional conditions, but I don't think it's necessary to describe them here.

For this particular wager, Atmoz and I discussed the time frame and terms. We closed bets in late April. This was done to ensure everyone had similar amounts of data when placing bets. So, I'm afraid you are too late.

However, if you posted a specific numerical prediction for this metric sometime in April, I'd be happy to add it to batch of predictions. This will permit the curious to see how you did compared to others who bet.

The winner will be the person who comes closest to the value that will be posted on this specific page at The Cryosphere Today. The prize is bragging rights.

"there is significantly less sea ice in Hudson's Bay and Baffin Bay"

"seems not especially meaningful to me"

Hmm. NSIDC does not seem accurate here - well I probably should say not up to date rather than indicating inaccuracy.

See:

Comparison images

Given that pair of images are we going to see William and Steve fight it out to see who will offer highest odds? ;o)

Lucia, can I sell my bragging rights to William? Or better yet
If he cant get closer than me he has to write us a nice haiku.
he should do that anyway, so maybe not a good idea.

or bring us a shubbery.

What appears to be missing is that 2008 is going to have very low Arctic sea ice coverage. It will be a close thing if it beats the 2007 collapse, but it will be close, and that bodes ill from man made global climate changes.

[Latest pics are looking good for the good guys, ie your carrots are in danger. However, this bet is all a matter of weather, and says nothing about the longer term trends. Though if we do get through this year will more than last, the multi-year ice might start to rebuild, and all the worry about lots of single year ice my look overdone. If not, then... -W.

You are whistling past the ice cave there Stoatee. Given increasing global warming, a rebound from La Nina and the solar cycle picking up, Eli sees little reason for thinking there will be a gain of multi year ice in the Arctic, so, instead of doubling down on this year, how about doubling down on next.

[Wanna start the book on next years ice, wabbit :-)? -W]

proÄ ne?

http://www.iup.uni-bremen.de:8084/amsr/arctic_AMSRE_nic.png

[There is still time for an upset, but if you look at http://arctic.atmos.uiuc.edu/cryosphere/IMAGES/current.365.jpg you'll see (with my eyes) the anomaly hovering around -1M, rather than last years 2M -W]

Yeah, been there before, seen that. Frankly there is a serious difference between the three groups tracking all this.

"So let me clarify: my "prediction" was based purely on my reading of the statistics of the time series to-date: a record is rarely followed by another."

A record is rarely followed by another record in two years time either. So is anyone offering bets for 2009?

[Could be. Are you offering to take the low side on even odds? -W]

OK but just a small amount for a bit of fun. Still makes sense to clarify what we are betting on.

IIRC you were using the 5 day running mean extent figure from NSIDC

http://nsidc.org/news/press/2007_seaiceminimum/20070810_index.html
"the value of 4.13 million square kilometers (1.59 million square miles) five-day running mean extent, observed on September 16, which appears to be the 2007 minimum."

'appears' doesn't sound definitive so I better also quote
"This year, the five-day running minimum occurred on September 16, 2007" from
http://nsidc.org/news/press/2007_seaiceminimum/20071001_pressrelease.ht…

So if the minimum 5 day running mean extent in 2009 is less than 4.13 million square kilometers you pay me £10 and if over 4.13 million square kilometers I pay you £10.

If you want to extend it, I'll also offer
if the minimum 5 day running mean extent in 2010 is less than 4.13 million square kilometers you pay me £10 and if over 4.13 million square kilometers I pay you £12.

Eli will take some of that action with a minor change

So if the minimum 5 day running mean extent in 2009 is less than 4.13 million square kilometers you pay me £10 and if over 4.13 million square kilometers I pay you $10.

Seems only fair, you get to bet against the ice and the Brun

If Eli thinks you might accept that bet then I think I better withdraw my offer and replace with me paying £8 for 2009 and £10 for 2010.

As you say the 2009 betting pool hasn't started, I'll withdraw my offer in case there is important new information causing you to start your betting pool.

I think your even odds only betting is rather limiting - you cannot see how the odds change over time. I would much prefer to see it listed on intrade. How about several of us suggesting it to intrade?

[Sure. How do you do that? Or I could offer various options -W]

At https://www.intrade.com/jsp/intrade/contractSearch/
there is a link for "Suggest Contract" which simply opens an email to markets@intrade.com with a subject of Contract Suggestion.

Would you always bet against two consecutive records (barring losing a previous bet, of course)? Or is there a threshold below which you'd stop making that bet? Just curious, really.

[I made the bet on the basis of the statistics, which said to me that the odds were in my favour. That would only fail to be true if we've entered a radically different regime where the yearly falls were greater than the interannual variability. People seem to have decided that, on the basis of one year, 2007. I don't think thats plausible -W]

Okay thanks.

"That would only fail to be true if we've entered a radically different regime where the yearly falls were greater than the interannual variability."

I think that means that there is a computable "interannual variability" in the time series, and that you'd need a number of (consecutive?) falls (from the trendline?) to pass some statistical threshold where you then think that two records in a row is better than 50/50? Is that right?

[Nope, totally wrong :-)

I mean that the seaice extent, like lots of other climate variables, can be usefully considered as a trend plus natural variation. Over time, the variation sums to 0 (of course) but the trend doesn't. Oe, more formally, E_i = C + i*A +B_i, where A is the trend, i is the year regarded as an integer, and B_i is the noise. B_i is numerically larger than A, so in the year following a record minimum, there is little reason to expect another record min, because "A*i" hasn't changed by much. Only if A grows very large (which means a regime shift, because in this simple model A is constant) do you expect repeated records. Of course, you can put in different models for B_i, which is the noise. If you assume very high year-to-year autocorrelation of the noise, then you can expect repeated records. But as I recall the record doesn't support that (and neight did our models) -W]

Okay that makes sense, many thanks.

I'll have a look, but I don't suppose the numbers for A & B are currently known?

According to tamino recently the trend has accelerated in the last ten year, but we'd need to see another acceleration to get to that state?

2007 may have been the start of an accelerated trend, or it may have been a blip. But we'll need a few more years of data to know?

These sorts of explanations would make good blog posts.

>"2007 may have been the start of an accelerated trend, or it may have been a blip."

William and I agree 2007 was probably mainly a blip. However I disagree with your statement quoted above prefering something like:

Somewhere between 2000 and 2003 may have been the start of an accelerated trend. The rate from 2003 to 2006 shows a 4.5 fold increase over 1979 to 2000 trend. Alternatively, 2000 to 2007 may just be a period below the trend - it is longer than other periods below the trend but so what? The record since 1979 is short. 2003 to 2006 is awfully short to announce a new trend is apparent. 2007 looked like further acceleration of the 4.5 fold increase but 2008 is making that 4.5 fold increase look too large. We'll need a few more years of data to know.

The other thing about the trend line is that you might expect an accelerating rate then a reducing rate.

For the Joe Romm bet of down to 10% by 2020 we would need a further 1.7 fold increase on top of the 4.5 fold increase.

A mere personal opinion - If all the 4.5 fold increase is genuine rather than any being down to noise then a further 1.7 fold increase may not be an outrageous expectation but some of it may well be down to noise.

My statement about accelerated trend was based (from memory) on this:

"Not only has arctic sea ice been in decline for decades, the decline is accelerating. The anomalies don't just show a decline, they show a statistically significant greater rate of decline now than for the first two decades of satellite obserations. This result isn't just due to the dramatic 2007 dip; even if we eliminate data after 2007, using only the data up to the end of 2006, the recent increase in the decline rate is still statistically significant."

http://tamino.wordpress.com/2008/07/17/arctic-ice-update/

seeing the recent dip in sea ice extent, is it 'game on' for the bet for this year? Or are you still confident there's no chance for a new record this year?