Sea ice round up

Its the traditional sea ice round up season. This year I have some solid wins (a bit of real money; Luna and Fred and Alastair I think) and a marginal loss (50 Quantloos to The Penguin).

Just to prove how debased I am, here is an excel spreadsheet graph:

seaice-2009-sep

My prediction was that the ice would “return to the trend”, which TP interpreted fairly enough (ah: I see I said it myself) as 1979-2006: 5.84, and the answer was 5.36. I see that if I’d chosen 1995-2006 I’d have been just about spot on (of course it is cheating to say that but I’ve drawn it on anyway). TP tells me what Excel won’t, which is that my SE is 0.5, so by that measure I’m right (and so is he).

To throw in a snark: various official bodies keep telling us that the sea ice is “well below climatology”. Doh! Of course it is. Absolutely no-one thinks it will return to climatology; as a “default” prediction that is completely useless, except to make this years ice “look good” (or bad, depending on how you view things).

Also, this make be a convenient place to link to:

* The inaugural 2007 bet
* The 2008 declaration of victory and invitation for the 2009 round
* http://www.ijis.iarc.uaf.edu/en/home/seaice_extent.htm which I like and can never find a link to.

Who else did what?

* The Buzz says that “trends for 2009 indicate that we may be on our way to a new record low” so Erin wins no points, and neither does Coby for repeating Arctic sea ice extent tracking below 2008 though he does correctly point up the variability.
* RC racked up an incredible 800+ responses to “Sea ice minimum forecasts”. They wisely make no predictions, but have some nice links.
* Serreze & Stroeve look a bit – ahem – incautious in their choice of words: Standing on the brink… Despite some recovery of the Arctic summer sea ice this year, the signs suggest the transition to a seasonally ice-free Arctic Ocean is underway. Though that reminds me: I have $333 on the Arctic not being ice-free by summer 2020 with Joe “I am a physicist” Romm (and E200 with PFD).
* The “pros” reported in June ARCUS were… a bit rubbish (sorry folks) with *no-one* guessing more than 5. July was a bit better, with M+U bravely guessing a bit more than 5; but August went bad again.

Lounge of the Lab Lemming ran a pool, and declared a result. Most of the guesses, I see were for well under the final value; unfortunately they didn’t come accompanied by a rationale.

Next years prediction is again the 1979-2006 trend line, +/- 0.5. Takers?

Comments

  1. #1 Nick Barnes
    2009/10/10

    I won 50 euros betting the area would be below 2005.

    [Against whom? -W]

  2. #2 Robert Grumbine
    2009/10/10

    The irony regarding your prediction being that if you’d taken all years, including the remarkable 2007 and 2008, then a straight line would have come out at 5.46 for this year and you’d have mopped the field. Going with 1979-2006 only says your prediction for next year is 5.78 million km^2. Using all 31 years to date says 5.37 for next year. So I’ll put 10 of my new quatloos with you 1:1 on the ice cover being below 5.575 million km^2 (as computed by NSIDC again, for September average). Me taking the ‘below’ again. I’ll probably have a few more wagers to make with you in a while after I do some actual work.

    [Lets make that a tentative wager then, subject to cancellation by either side before, let us say, March -W]

    One part of the additional work being that it bothers me that every prediction, every month, was below the observation. Perhaps not as much as it bothers you. But enough that I’ll probably be entering a much higher figure next year (in addition to one or two others), if only to bring some balance to the ensemble.

    [It does bother me that everyone made the same mistake. Perhaps there are lessons in herd mentality there -W]

    Unfortunately, as annoying as it is to you to hear people keep saying that the ice is not recovering to climatology, it’s the case in the US that a lot of noise is out there saying that the ice ‘is recovering’, or even ‘has recovered’. The more people deal with US media, the more, I expect, they’re harping on this point. Over here, at least, it does still need to be said.

    [Hmm. I still don't like the comparison to climatology, for the reasons I said. But if people really do need to be told this I suppose you'd better keep telling them :-) -W]

  3. #3 andrewt
    2009/10/10

    I won ~A$150 in an index bet against climatology from a colleague who bought into the “sea ice has recovered” claims on WUWT + the solar cycle/GCR stuff.

    The North-East Passage still seems to be open which must be encouraging for the Beluga people if they are looking to establish it as a regular route.

  4. #4 Keith Morrison
    2009/10/11

    The $333 bet is a little unspecific: much of the time up here ice doesn’t clear until late August/Early September, which means, seasonally, that it’s fall in my part of the world.

    Bit of wiggle room in that, no?

    Also, all the ice or enough that navigation is essentially free except for dodging the odd berg or floe?

    [I think we settled on something like only 10% left, but I forget the details -W]

  5. #5 TvO
    2009/10/11

    Dear Debased,

    Your slide into using inferior tools is regrettable maybe, but read the manual… in excel linest(known_ys, known_xs, true, true) will return an array function (ouch! advanced Excel) that includes all the stats including SE’s…)

    Cheers!

    [Read the manual? Who do you think I am? But I may, once I kick my son off the Big Computer -W]

  6. #6 crandles
    2009/10/11

    Hi,

    You are admirably sticking to your 1979-2006 line and I may want to bet against that. That line arrives at 5.78.

    However the obvious null forecast (which the penguin does not seem to want to admit) is a straight line through all satellite data which gives 5.37. So your +/- 0.5 seems too generous to you giving you 5.28 – 6.28 which appears to me to give you more than half of the obvious null forecast.

    So the question is: do you want to split the difference, so that you take the above side of 5.575 (average of 5.37 and 5.78)?

    If so, then I think you have yourself a even money £50 bet. (Alternatively if you want to stick to your 5.28 – 6.28 range then I would want 2:1 odds)

    [No thanks. I was only ever really interested in a fairly safe bet, against those who were determined to believe that seaice had begun its terminal melt down. There is a lot of year-to-year variability, and the +/- 0.5 doesn't really encompass this I'd say. I don't think we disagree enough to make a bet that we both consider profitable -W]

  7. #7 Robert Grumbine
    2009/10/11

    crandles:
    See my comment in @2.

  8. #8 crandles
    2009/10/11

    Robert,

    Sorry, yes I read your comment after I posted. If it needs saying to the US media then yes it does need to keep being repeated. However none of the people making predictions seem to need to be told, they are all building in declines at greater or occassionally lesser rates of decline.

    The obvious null forecast had an error of 0.1m Km^2 and the only people with lower errors were Hypocentre with an error of 0.05m Km^2 and Divalent with an error of 0.075m Km^2. These were 2 of 21 entrants to Lab Lemmings pool. 2 of 21 being better than the obvious null forecast could occur by chance in such a pool or it could have been down to skill. However if it was skill by those two, why did all the others fail to demonstrate skill? (It doesn’t seem likely to be down to knowledge with the winner being a seismologist.)

    This brings us back to JA point that there is a fundamental problem evaluating probabilistic estimates with a single data point.

  9. #9 Nick Barnes
    2009/10/11

    Against whom?
    Joeduck. He paid up very promptly and politely. It’s been a learning experience for both of us. See http://rankexploits.com/musings/2008/nothern-hemisphere-sea-ice-hows-it-going/

  10. #10 Gareth
    2009/10/11

    I was in two minds about allowing you to continue to forget our bet, which was double or quits on last year, but I’ll be honest instead. You win £40, by my reckoning, and I won’t be offering double or quits again… ;-) I conceded some time ago…

    [No, I won't gloat. You can keep the money till we work out something else to bet on :-) -W]

    My thoroughly amateur peeking around at satellite images suggests to me that there’s a lot of very thin mobile ice in places where there used to be thick multi-year ice, esp north of the Canadian archipelago. And the freeze-up seems to be slow. CT shows the biggest anomaly of the year at the time of writing (-2m km2), and IJIS also has this year below last (but it’s difficult to be sure with the thick lines on their graph). I’ll wait to see what Kwok et al have to say about ice volume before contemplating bets for next year…

    [Yes, I saw your post, and indeed the images. It will be interesting to see how this plays out -W]

  11. #11 dhogaza
    2009/10/11

    And the freeze-up seems to be slow. CT shows the biggest anomaly of the year at the time of writing (-2m km2), and IJIS also has this year below last (but it’s difficult to be sure with the thick lines on their graph).

    It’s picking up … IJIS shows a big steepening in the last couple of days, and NSIDC is starting to uptick.

    It’s going to be interesting to watch what happens over winter.

  12. #12 dhogaza
    2009/10/11

    The IJIS sea ice concentration graphic’s interesting.

    A lot of freezing north of Siberia while north of Alaska it’s been receeding a bit.

  13. #13 Michael Hauber
    2009/10/11

    Two things I’ve noticed on sea ice:

    The Arctic Oscillation had a rather extreme negative value during the recent melt season. How much would this have contributed to a higher summer ice value? However the AO was most negative early in summer when the ice was melting fast, and went postive late in summer when the melt slowed down dramatically. It is now highly positive and this seems to be the most dramatic swing in AO from one extreme to the other in the record.

    Observing GISS maps of temperature vs time and latitude it seems that waves of warming and cooling transmit from the equator to the Arctic over a 2-3 year period. The strong ENSO cooling from late 2007 to early 2009 would probably be due in the Arctic this year, and due to continue for another year or two before being followed by a warming pulse.

  14. #14 Gareth
    2009/10/12

    [No, I won't gloat. You can keep the money till we work out something else to bet on :-) -W]

    Given the current weakness of Sterling perhaps I should just pay up… ;-)

    Should you find your way to NZ, hospitality will be arranged.

    Cheers

  15. #15 Steve L
    2009/10/12

    Sorry for being slow: what does “below climatology” mean? Ditto for “return to climatology”, etc. Is this a reference to model predictions or ???

    [1979-200x average -W]

  16. #16 Nicolas nierenberg
    2009/10/13

    I forgot to gloat about my post back in June!

    “Anyway I will go with the naive model. This year’s minimum will be higher than 2008 and lower than 2005 as measured by this.”

    [Ah, but you put no money on it. Not that that would have helped: you asked for too big a range -W]

  17. #17 Alexander Ač
    2009/10/13

    Interesting,

    since 1979 there were never more than 3 consecutive years of increase in sea ice minimum. If that happen, it would be for the first time. So my bet it that the next year is lower than this year. Does this fits for the bet? :-)

    [I suspect taht we don't disagree enough -W]

  18. #18 crandles
    2009/10/13

    >”[Ah, but you put no money on it. Not that that would have helped: you asked for too big a range -W]”

    Is this making your views more apparent?

    5.57-4.68=0.89 is too big a range but your +/- 0.5 ie 1 is OK for you to ask for.

    Hmm a few possibilities remain to explain this:

    Maybe you are just demanding the odds be highly in your favour before you bet.

    Maybe 1m range is OK if you are taking at an extreme end of a range but this implies you don’t really believe your 5.78 figure.

    Maybe you do believe your 5.78 is the centre of the range but because you are aware that it isn’t at the centre of the consensus of everyone ranges you believe you can negotiate a good deal with you being allowed a large range.

    That you also won’t negotiate your range from 5.28 to 6.28 to above 5.36 doesn’t look good for your strength of belief in 5.78 but maybe this is just that you just want the odds to be heavily in your favour before you bet.

    [Certainly I want the odds in my favour. What would be the point otherwise? Quite how heavily in my favour is another matter. I don't think I've offered the +/- 0.5 as a bet range. All I said was that this was the SE that TP thinks I have -W]

  19. #19 Lab Lemming
    2009/10/13

    Is it worth running a pool on winter maximum extent, or is that somehow uninteresting?

    [Generally agreed to be less interesting: the Arctic is landlocked, which tends to bound the max -W]

    Aside from those of you who came over, I doubt there was much skill in my pool. I’m a geochemist who was looking for an amusing application for a Gaussian curve addition spreadsheet I wrote for a completely different purpose.

    [ :-) -W]

    So my only sea ice expertise is the ability to say that the 2010 extent will be less than that of 646,500,000 BC.

  20. #20 crandles
    2009/10/13

    >[I don't think I've offered the +/- 0.5 as a bet range. All I said was that this was the SE that TP thinks I have -W]

    “Next years prediction is again the 1979-2006 trend line, +/- 0.5. Takers?”

    That wasn’t an offer? Oh well, I suppose the takers is a separate sentence and can be interpreted as ‘an invitation to treat’ ie advert rather than offer. I doubt I would be the only one to misinterpret it as an offer though.

    [I'll grant you that was easy to misinterpret. It was more a general indication of intent to bet on something for next year; sorry. Having had a chance to think about this, I now think I'm wrong to be ignoring 2007, 8 in my trend calculation: they are exceptional, but so what, they are part of the natural variation, just like the few exceptionally high years. I'll put up a post in a bit with an actual offer to bet -W]

  21. #21 Luna_the_cat
    2009/10/13

    What the heck, I’m perfectly happy to put on another bet with you. Hopefully a more sensible one, though. I want my money back. ;) I want to give it more of a think before I settle on a probable figure or range, though.

    [I really need to find some wacko septic to bet against rather than sensible people. Judging by Romm;s fulminations against the freakonomics folks (
    http://climateprogress.org/2009/10/12/superfreakonomics-errors-levitt-caldeira-myhrvold/) which are, just for once, quite justified, I may have found someone -W]

  22. #22 Nicolas Nierenberg
    2009/10/13

    W,

    I think we agree that a bet on this type of thing is only interesting if there is a real qualitative difference of opinion. For example if someone wanted to bet on a new low next year I would do that. (Or a new high for that matter.)

    Alexander I would take a bet that next year’s minimum will be higher than this year, if for no other reason than the logic you used to arrive at your prediction.

  23. #23 crandles
    2009/10/15

    Professor Wadhams said: “The Catlin Arctic Survey data supports the new consensus view – based on seasonal variation of ice extent and thickness, changes in temperatures, winds and especially ice composition – that the Arctic will be ice-free in summer within about 20 years, and that much of the decrease will be happening within 10 years.

    http://news.bbc.co.uk/1/hi/sci/tech/8307272.stm

    Another bet target or is that not so bad? (Or have you been converted to accelerating ice decline?)

  24. #24 outeast
    2009/10/15

    William, what’s your PayPal email? I now owe you 30 sqids iirc.

  25. #25 Alastair McDonald
    2009/10/21

    Yes, and I owe you £100. Would you accept a donation to Greenpeace?

    Cheers, Alastair.

    [Sounds fair. You may take your pick or any combination of Greenpeace, FOE, WWF, Sustrans or The Woodland Trust -W]