Sea ice: the triumph of the William

This year's sea ice was unexciting. NOAA goes for "2015 Arctic sea ice fourth lowest on record" which is doubtless true.

Tamino has some helpful plots, so I've nicked one. I'm pleased that my 2014 comment "Hopefully that too [i.e., 2012] will look like an outlier in years to come" now looks quite true. From a statistical point of view, Tamino says "changepoint analysis can only confirm one rate change, but a smooth suggests more might be going on"; and I'm sure that's true from that view. From a physical viewpoint I'm less convinced, and would just take a simple LS fit to the whole (satellite) series. A continued long-term decline is the obvious prediction.

Speaking of which, we come to the topic of this post, my bet with CR, recorded here, formalised here, which is:

Crandles offers 3 separate bets on the [monthly average extent not area] average of [2012, 2013 and 2014] (to be above/below 4.294, I take the high side), of [2013, 2014 and 2015] (4.119, ditto) and of [2014, 2015 and 2016] (3.94, ditto). In the event of anything that clearly throws things out like a VEI6 volcanic eruption bets are voidable.

And the data shows:

year mo    data_type region extent   area   avg
2012  9      Goddard      N   3.63   2.15
2013  9      Goddard      N   5.35   3.74
2014  9      Goddard      N   5.28   3.70   4.75         (2012-14)
2015  9      NRTSI-G      N   4.63   3.31   5.08         (2013-15)
2016                          2 - X         3.97 - X/3   (2014-16)

(note: 2013 was also 5.35 extent in NRTSI, but 3.48 in area. Which shows what a good idea extent is). So I win (hurrah!) and for extra bonus points, as CR notes by email, sea ice extent has to be implausibly low - less that 2 - in 2016 for him to win; so for a £10 concession for early payment, we've settled the last one, too.

I believe (see Sea ice betting report wherein Neven unwisely converted his E50 win in 2012 into double-or-quits for 2013-15, that he now owes me... oh, yes, nothing. Because he won last time, so we're quits :-).

So what if anything does this mean? Or, in CR's terms, Were you just lucky or skillful? Most of my thinking (as I said before somewhere, I'm sure) was based on a Impact of instantaneous sea ice removal in a coupled general circulation model by D. Schröder (who did the work) and W. M. Connolley (who did something). Which showed - at least in a model - that the sea ice really isn't very sensitive to sudden loss; there is no tipping point; if you strip off the ice it recovers quickly, in a few years. To reduce the sea ice you need years of increasing forcing leading to less ice. Which, over the long term, we'll have I think. But it means I think that 2007 - and then 2012 - were less important than other people thought (see some quotes from 2008). However... it wouldn't like to over-emphasise that. Who knows what future years will bring?

More like this

Well done.

Re 'Most of my thinking was based on a Impact of instantaneous sea ice removal in a coupled general circulation model ... sea ice really isn’t very sensitive to sudden loss; there is no tipping point'

Agree there is no tipping point - the models pretty much agree on gompertz like shape I built into my projections. The models didn't seem to be very good at amount of ice or rate of decline.

It seems more like bet was about rate of decline than about sudden declines. Seems to have turned out that the long term rate of decline was a better guide than the short term rate despite there appearing to be evidence of acceleration particularly in PIOMAS data. There was a likelihood of deceleration in the rate again at some time but I thought I would be unlucky if that started soon.

Not very sensitive to sudden decline doesn't seem to say much about the rate of decline but maybe it correctly made you more dismissive of the 2007 outlier (bet made in 2011 so 2012 outlier did not cause winning value to be too low.)

You haven't said much to make me think skillful rather than lucky. However, not sure I want to bet again. Maybe that is partly because my views are now much more similar to yours.

While I could try to claim, I was unlucky deceleration started so soon, I am not sure I want to push that too much. For the moment it may look like fast rate of decline from 2000ish to 2008 then slowed down again. Maybe there is a physical explanation for this in terms of quantity of MYI over about 2m thick having reduced to as low a level as it can easily reach. Maybe this levelling off was predictable. William doesn't seem to be claiming this.

In a few years this brief acceleration and deceleration may well fade in importance and just a straight line through the data may well be appropriate as William indicates.

Anyway I lost, so people probably aren't too interested in what I think.

[PIOMASS is worth mentioning, and perhaps I should have. A few years ago PIOMAS was a pretty good support for the death spiral theory. But today? -W]

> But today?

Last 8 years of PIOMAS average Sept reaches zero about 2062 (think I used an estimate for Sept 15 volume before it was complete when I did that calc so maybe a little inaccurate) versus 2499 for NSIDC Sept extent.

Both are rather supportive of there still being sea ice in Summer in 50 years from 2011.

PIOMAS projected date not as strange as NSIDC projected date, but doubt that will make you a convert to using PIOMAS volume rather than extent data.

Yes not good for death spiral support. (but then I would suggest I never was a death spiral supporter preferring gompertz shaped extrapolations than rapid transition shapes)

[I didn't mean to appear to suggest that you were. I rather meant that those who had been using it to support "death spiral" had (thankfully) shut up a bit recently -W]

If you look at the difference between Cowtan & Way and HadCRUT4, or just directly at C&W Arctic average, there was an abrupt increase in Arctic surface temperatures preceding accelerated sea ice losses (volume and extent) from around 2005.

To me what happened looks like a straightforward response to rapidly increased "forcing" which was sustained over several years. The cause being some kind of increased net heat flux into the Arctic, plausibly related to the internal variability processes which saw cooling upper ocean Pacific temperatures over the same period.

>"If you look at [temperatures]"

Correlation is not necessarily causation, but in this case it seems probable that the major temperature changes are mainly an effect of thin ice but there is also a likely cause:

Look at whether the temp increase is in Spring and Summer or Autumn and Winter. Thinner ice and more open water is allowing more heat build up during Summer via albedo effect but this does not increase air temperatures much. The heat is then vented to atmosphere in Autumn, less ice allowing more warming of the atmosphere. This is the highly noticeable effect: less ice causes higher temperatures in Autumn. However, it does seem likely that higher temperatures also cause less ice, so disentangling temperature and ice thickness effects seems difficult.

"The cause being some kind of increased net heat flux"
well yes, I agree but is that saying much?

It is pretty obvious we have not passed a tipping point - YET. But that does not mean that there is not one. Moreover, when the summer ice goes to zero (an extent of < 1M sq km?) then it will happen in the same manner as 2007 and 2012. A new record low seems to folow from an abrupt fall and when zero is reached it will be a new record low. The question is will zero be a tipping point, and into what?

My full name is not conceit. It is to distiguish me from the winner of the Nobel Prize for physics - Prof. A. B. McDonald.

By Alastair Bain … (not verified) on 20 Oct 2015 #permalink

As a devotee of Pattern Recognition In Physics I utilized patterns back in May to predict this year an average September extent of 4.59 Mkm^2. NSIDC monthly extent for September came in at 4.63M km^2. Yes, I know, something about blind squirrels ....

Going forward, I'm already sensing that 2015 was 2006, 2009 all over again. I actually pay more attention to volume than area or extent. If the pattern holds through winter, then we might well see volume plummet below the 2SD band on PIOMAS again next year.

By Kevin O'Neill (not verified) on 20 Oct 2015 #permalink

crandles,

The abrupt temperature rise preceded accelerated ice losses. Seasonally, in terms of degree of departure from the previous several years, it mainly occurred in Winter (DJF). It could be that these elevated temperatures acted to reduce Winter sea ice growth more than usual, which made the Arctic particularly vulnerable in the melting season.

“The cause being some kind of increased net heat flux”
well yes, I agree but is that saying much?

Mainly, I find it interesting that this period of accelerated Arctic sea ice losses correlates well with the unusually sustained "La Nina" pattern in the Pacific.

I always felt you did not take the PIOMAS data seriously enough, and that a death spiral could have happened. So far you have been right. Its also interesting to go back to just after the 2012 minimum to see Tamino fitting a second order polynomial. An eyeball extrapolation would put this years extent at around 4m, so comfortably below 2007. Yet despite a remarkably warm July we were still above 2007.

Whats the bet that as soon as slow decline becomes a firm consensus in the sane parts of blog world we see a shocking single year drop?

[Don't misunderstand me. I don't think the year-to-year extent changes are going to become a smooth decline; there will remain year-to-year variability; large drops and "recoveries" will still happen -W]

By Michael Hauber (not verified) on 22 Oct 2015 #permalink

William, it's been a while.
Congrats on you bet wins against Crandles and Neven.

[Hello again, and thank you -W]

I am glad you won these bets, since if you would have lost them, the Arctic would have got to have been in imminent mortal decline.

That does not mean Arctic sea ice is not in dire straits. To the contrary, long term data leaves little doubt about where Arctic sea ice is heading :
http://blog.sme.sk/blog/1159/306419/kinnard.gif

Looking at the Kinnard et al graph, and the developments beyond 2000, it always kind of baffles me that some people still feel that terms like "death spiral" or "tipping point" are not appropriate for what is happening in our one and only Arctic Ocean.

[I don't think extrapolating through 2012 is a good idea. Nor, for examining the next few years or decade, is looking over that timescale very helpful I think. Nonetheless, as I've said in this post, the long term trend is clearly decline; the question is how fast -W]

Yet, I don't think any one of the people you bet with even mentioned "tipping point" let alone argued that there is one.

[Indeed, perhaps not. But the origin of all this "betting on sea ice" stuff was the tipping point folk, who were all over back then; and the death spirallers too. Which is why I mention it -W]

So I don't think that is a very good argument as to why you won these bets. And yes, your study (and many others) show that ice rebounces in winter if you take it away in summer, but these same models also grossly underestimate actual Arctic Sea Ice decline :
http://neven1.typepad.com/.a/6a0133f03a1e37970b017744cf5360970d-pi

[I agree they underestimate the decline. But were you to plot that to 2015 it wouldn't look so bad -W]

So neither the bets with Crandles and Neven, not the bet with me (which ends next year) was about "tipping points" or "death spirals".

These bets were all about HOW FAST the decline will happen.

And there, it seems to me that you (and all of us) got lucky with 2013 and 2014 being above the linear decline line.

Let us all hope that you are right and 2012 and 2007 were the outliers, and not 2013 and 2014.

Frankly I'm not convinced yet, but I'm sure you agree that next year will be a very interesting melting season, for many different reasons...:o)

[Yes, 2016 will be very interesting -W]

By Rob Dekker (not verified) on 25 Oct 2015 #permalink

I believe (see Sea ice betting report wherein Neven unwisely converted his E50 win in 2012 into double-or-quits for 2013-15, that he now owes me… oh, yes, nothing. Because he won last time, so we’re quits :-).

Those are the best bets, aren't they? We should both hang on to those 50 euros for when we meet one day, and then have drinks. Lots of them if we meet anywhere outside Scandinavian countries. ;-)

Maybe we should do another bet for the coming three years, but I like draws somehow ...

Most of my thinking (as I said before somewhere, I’m sure) was based on a Impact of instantaneous sea ice removal in a coupled general circulation model by D. Schröder (who did the work) and W. M. Connolley (who did something). Which showed – at least in a model – that the sea ice really isn’t very sensitive to sudden loss; there is no tipping point; if you strip off the ice it recovers quickly, in a few years.

I vaguely remember reading this paper once, should re-read it. If we apply this thinking to 2012, followed by 2013 and 2014: Did the sudden crash in 2012 cause the weather to be cold and cloudy in 2013 and 2014? And if so, how?

PaulS said :
Mainly, I find it interesting that this period of accelerated Arctic sea ice losses correlates well with the unusually sustained “La Nina” pattern in the Pacific.

I have not seen a study linking La Nina to Arctic ice melt.
In fact, if you look at ENSO patterns over the years
http://www.carbonbrief.org/media/321392/elninorecord_nasagiss.jpg
it seems to me that El Nino is a better predictor for Arctic sea ice melt than La Nina.
For example, the recent El Nino years of 2010, 2007 and 2005 coincide with extensive melt years in the Arctic.

By Rob Dekker (not verified) on 29 Oct 2015 #permalink

Rob Dekker,

That's now what I'm saying. "La Nina pattern" is just a phrase which has been commonly used to describe the unusual decadal scale Eastern Pacific SST cooling over about 2006-2013. I'm not referring to the inter-annual scale ENSO shifts. I guess I could have talked about "PDO" instead.

PaulS said
I’m not referring to the inter-annual scale ENSO shifts. I guess I could have talked about “PDO” instead.

Sure, you could have.
But then, there is not much correlation between Arctic sea ice and the PDO either, now is there ?
While Arctic sea ice is going down quite persistently
https://nsidc.org/data/seaice_index/images/n_plot_hires.png
the PDO is just doing its own thing :
http://blogs.discovermagazine.com/imageo/files/2015/04/pdoindex.gif
I have not seen any study linking the two.

By Rob Dekker (not verified) on 29 Oct 2015 #permalink