The sea ice post!

People have been asking for the sea ice post, so here it is. I've been putting it off, obviously. The $10k bet comes due this year but its looking like a bad year; and secondly I don't really have all that much to say. Just to remind you, the terms are

If both NSIDC and IARC-JAXA September 2016 monthly average sea ice extent report are above 4.80 million km^2, RD pays WMC US$ 10,000. If both are below 3.10 million km^2, WMC pays RD US$ 10,000. In all other cases the bet is null and void.

You should also read the argument that Rob presented in that post. That bet was arranged in 2011. 5 years is not long for climatological changes, unless they're huge, and the bet was a compromise. Now, let's look at the current ice:


That's from That's still at an all-time low for the time of year, but only just, which is better than its been for the last few months. And of course its been, globally, ridiculously hot for the first few months of this year - even if its been a cold spring in the UK. The other thing to remember is that winter or early spring isn't a terribly good predictor of summer minimum. Nonetheless I find it hard to believe, at this moment, that I'm likely to win - ice going below 4.8 seems really rather likely - so the question becomes whether I'm likely to lose, which in a sense is a moral victory for Rob already.

This is a good point to mention that Rob and I were betting with each other because the blowhard nutters wouldn't put up. In his case, the denialists (and I had a go, too); in my case, people like Wadhams who is still predicting essentially no Arctic sea ice this year. Or possibly next year. At any rate, real soon now, for reason that are never very clearly explained.

Note: I'm completely hopeless at remembering where I had, so forgive me if I've forgotten some stuff. CR threw in the towel last year, so I'm clear with him.

More like this

I think the bet will likely be null and void.

2016 Arctic SIE is currently 3.1% (-0.33 Mkm2) below 2015 (JAXA data), and has been in the -2% range or more (in absolute value) since mid-April.

2015's JAXA's Sept average was 4.51 Mkm2. So the trend puts 2016 JAXA Sept average about 4.4 Mkm2. Of course, it all depends on the weather up there.

By David Appell (not verified) on 13 Jun 2016 #permalink

Melting seems to have ground almost to a halt in June, normally a big melt month. If the cool weather persists for the rest of the month, you might have a chance of winning - though I would have thought the opposite two weeks ago.

It back to the low edge as normal on most (all?) measures. Area has moved slightly above 2012 according to Wipneus's shadow Cryosphere Today measure. Weather seems set to continue to be good for slow sea ice loss for next week so I agree it is looking much more likely for you to avoid having to pay out.

These polls seem to usually expect low ice:,1572.50.html
so while median is around 3.1 mark for daily minimum, that should probably be adjusted to your bet probably having no pay result.

Last year NSIDC monthly Sept was 4.63. Ice looks worse this year so far, but last year we had an amazingly hot July and current weather is pretty mediocre. While I think a final result nearer to 4 is likely, I don't think a value as high as 4.8 can be ruled out yet. Maybe even slightly better chance than below 3.1.

By Michael Hauber (not verified) on 14 Jun 2016 #permalink

BIST is back!
"As of June 14, 2016, NSIDC has completed the transition to the Defense Meteorological Satellite Program (DMSP) F-18 satellite for sea ice data. Sea Ice Index updates have also resumed."…

Extent doesn't say much about compactness or for that matter thickness. The difference might just be how compact the ice pack is at the minimum. Not just the temperature determines the minimum sea ice extent.

By Phil Hays (not verified) on 14 Jun 2016 #permalink

Yeah, if I were asked odds, I'd say 80% for a draw, and 10% each for either payoff case. So I suggest you will want to have your finances ready, but probably won't need it.

By Omega Centauri (not verified) on 14 Jun 2016 #permalink

I would think someone must have computed correlations between the September minimum and lag N month values, but maybe not. Anyway it would be simple enough to do.

By Raymond Arritt (not verified) on 15 Jun 2016 #permalink

For those who like this sort of thing, the DSCOVR imagery gives you the best imagery of the Arctic these weeks around the Summer solstice, because: geometry

Imagery updates are quite variable, some days have more than one image per hour, some days only a few images; priority for bandwidth is given to whatever the USAF is doing with the satellite.

By Hank Roberts (not verified) on 16 Jun 2016 #permalink

Recovery! I see it!

By Craig Thomas (not verified) on 17 Jun 2016 #permalink

Thank you Russell, that was beautiful.

My initial reaction was hope they get a lot of "investors" in their fund, but I can't really wish that even on the gullible and greedy who appear to be their target audience.

By Robert Day (not verified) on 17 Jun 2016 #permalink

Some kind of prize ought to be awarded to Russell for finding that one.

Talk about betting on climate ....

By Hank Roberts (not verified) on 18 Jun 2016 #permalink

I hope our host looked at the link Russell provided, considering his prior interest in David Evans' mysterious ideas.

I'm not so sure this is an 'own goal' yet. First, remember that grifters gotta grift. Evans probably jumped the gun since he appears to have started CFFM in November, 2015 -- he should have waited until Jan or Feb 2016 so he could have used the peak temperatures from El Nino as his starting point. No matter, he can finesse that.

We should note that: Since CFFM was founded global temperatures have ALREADY STARTED DROPPING!!!!

We are witnessing the beginning of the next ice age. Is *YOUR* investment fund situated to take advantage of it???? Don't wait until it's too late.

As I said, grifters gotta grift.

By Kevin O'Neill (not verified) on 18 Jun 2016 #permalink

Is it possible to short a hedge fund? Some derivative offering here could potentially suck up a lot of our good friends' money :)

[My assumption (I haven't looked closely yet) is that its all vapourware; or that he may be sincere but no-one will be mad enough to take him up. So I doubt there is anything there to short -W]

By Arthur Smith (not verified) on 18 Jun 2016 #permalink

Well, don't be gloomy.
Based on current melt pond coverage: " simulations suggest sea-ice extent for the end of this coming summer will be 4.5 (+/- 0.5) million square km."

[Nice story, thanks. I used to work with David; and saw him again just recently. But seasonal forecasting is difficult, so I shan't relax yet -W]

By Phil Hays (not verified) on 19 Jun 2016 #permalink

>"My assumption (I haven’t looked closely yet) is that its all vapourware"

$41,630 of $375k
Raised by 192 people in 7 months

I make it 37 donations totalling $9405 of that $41630 is in last 3 days

$375k is not much of a hedge fund

[Yup, found that independently now, blog entry in progress. Indeed $375k is less a minnow and more a diatom; but I suppose $41k is something, though not a lot -W]


$41,630 is enough to start prospecting for ice mines on Cayman Brac, if a yacht is provided.

Hi William, thanks for your post.
There are about 3 months to go before the result of our bet will be known, but I'm glad you put up this sea ice post in anticipation of the results.

It may be interesting to note how this bet came about.
After a great deal of prior discussion, William put up this graph :

We then discussed the various tracks (red, green and dotted) and considering the variability in summer weather, I proposed a void between our points of view :

So here is the basic idea :

If by 2016, sea ice extent is above the green line, you win $10,000. If it is below the red line, I win $10,000.
If it is in between, the bet is null and void.

The rationale for this structure of bet is simple : You (and IPCC projections) hold the opinion that sea ice extent will recover from 2007-like fluke minima, based on modeling of ice extent (Tsietche et al and your own publications). I hold the opinion that sea ice extent will reduce exponentially based on observations, albedo effect and ice volume reductions.
If the truth is in the middle, we call it even, but if either you or me is explicitly correct, we win big time.

What do you think ? Is this a framework that you can agree too ?

[OK, I agree, it is about time we stopped fencing and came to and agreement, or not. I’m happy with the terms you offer, provided only that by “If by 2016” you mean “If in 2016”, so it reads: “If in 2016, sea ice extent is above the green line, WMC wins $10,000. If it is below the red line, RD wins $10,000. If it is in between, the bet is null and void.” I think we have to interpret it like that to make sense; otherwise, some years could be above green and others below red.

My interpretation is that I think the ice is likely to be somewhat under the green, though not by any great margin; so I believe the chances of me winning are less than 50%, but let us say greater than 25%. I regard the chances of the ice being below the red line (3.1 in 2016) as “very unlikely” (less than 10%, let us say) so the bet is reasonable, from my viewpoint. I regard the most likely outcome to be void -W]

Ref : comment # 8.

Now, September 2016 is coming up, and it appears that we are heading for the 'void' area between 3.1 and 4.8 M km^2.

Specifically, based on May numbers for "land snow cover", "ice extent" and "ice area", with the method explained at Nevens' Sea Ice Blog here :
and at SIPN here :
where the Sept '15 projection based on June '15 data was spot on (4.6 M km^2).

This method suggests that for Sept '16 we are heading for 3.8 M km^2 with a standard deviation of 460 k km^2, based on May data.

Based on this method, at this point it looks be that our 2016 bet will end up in the 'void' area, with William's 4.8 number is about 2 SD's off (2.5% chance of winning) and Rob's 3.1 number is about 1.5 SD's off (some 5% chance of winning). There is at this point thus a more than 90% chance of ending up in the 'void'.
These are ballpark numbers, so take them with a grain of salt.

Of course, if we end up in the 'void' area, it is somewhat disappointing. It is much more exciting if either William or me would win, so that one or the other argument appears to carry more weight.

But contrary to William, I DO have a lot to say about sea ice.
Not just that the entire "void" area of our bet is quite a bit below the IPCC projections for this year, but also that ice volume still declines faster than ice "extent".

Let me leave it at that for now.

[Welcome. It's hard to know what to say at this point, either about this year or the longer term trend. For the year, well, we'll get an answer within a few months at least. For the longer term trend, it seems likely that this year is a warm outlier, but that may be wrong too -W]

By Rob Dekker (not verified) on 21 Jun 2016 #permalink

William and Rob, if I may presume, you both rock! It will be interesting to see if circumstances push one way or t'other (I lean towards surprises based on degraded ice conditions, but perhaps not enough to push out of the void category), but as honorable men, you show up the whole unskeptical "skeptic" enterprise as a jumble of contradictions with no foundation unwilling to put up or shut up.

By Susan Anderson (not verified) on 21 Jun 2016 #permalink

Hmm. Evans says "Because of the numerous mismatches with observations, the computer models and the basic model would in science, normally be called into question."
I wonder if y'all could con, er, convince him to take a bet like this, but model based? - if the ice minimum is 1 sigma greater than model predictions, he wins 10k, if it finishes lower than 1 sigma below, he pays 10k, but wins by showing that the models are inaccurate. e.g. "you are here 2016" is lower (warmer) or higher (cooler as he predicts) than theIPCC models

By Brian Dodge (not verified) on 21 Jun 2016 #permalink

William said : "It’s hard to know what to say at this point, either about this year or the longer term trend. "

I find that disappointing.
There is still this nagging problem that our models continue to underestimate Arctic sea ice decline.
That issue was blatantly apparent with CMIP3 models, exemplified in graphs like the one Brian posted above.

But the problem is still there in CMIP5 AR5 models.
IPCC AR5 even admits that (section of the WG1 report) : " the magnitude of the CMIP5 multi-model
mean trend in September Arctic sea ice extent over the satellite
era is more consistent with, but still underestimates, the observed one (see also Massonnet et al., 2012; Stroeve et al., 2012; Wang and Overland, 2012; Overland and Wang, 2013)"

I have previously suggested that the underestimate of the ice extent decline in GCMs is rooted in the decline in land snow cover, which is underestimated by a factor of 2 or 3 in CMIP5 models (Brutel-Vuilmet et al 2013), thereby significantly underestimating albedo feedbacks during the melting season.

Tamino did a rough calculation and concluded that a whopping 880 TW of warming is added into the melting season because of snow and ice decline.

Since that number is underestimated in GCMs, it makes a BIG difference for the Arctic, and can very well explain why GCMs underestimate Arctic sea ice decline.

Isn't that something we should talk about ?

[There's a disparity there, as you say the IPCC says; but what to do about it isn't clear. How badly wrong are the models? There's so much "noise" aka interannual(and longer term) variation in the obs, that it is hard to know. Which is part of what this bet was about; when it was made, anticipating a "death spiral" wasn't that implausible; I think it is less so now -W]

By Rob Dekker (not verified) on 24 Jun 2016 #permalink

William said :
How badly wrong are the models?
Mmm. Let's see.
Here are the IPCC CMIP5 model results :
We are interested in summer Arctic sea ice extent, which is figure (b).
There we see that our bet range (3.1 - 4.8) is well below the projections for this year 2016.
And we see that the mean of the models projects the observations NOW in 2040 or so.

By Rob Dekker (not verified) on 25 Jun 2016 #permalink

I'm sorry. That IPCC projection of 2040 for the ice extent that we observe already over the past couple of years is for RCP8.5.
For RCP6.0 we were not supposed to reach current levels of Arctic summer sea ice extent until about 2060.

That's how bad the models are, William.

So when I tell you that much of the difference (between observation and models) can be explained by the underestimate in the models of land snow cover, is that of any interest ?

P.S. Sorry to hear about Brexit. I think the UK made a big mistake.

By Rob Dekker (not verified) on 25 Jun 2016 #permalink

SIPN June report out.
Lowest prediction 3.4 so none under 3.10
1 prediction of 4.8 and 4 over 4.8
Certainly most think void bet but perhaps indicates you have a better chance of winning than losing?

[Things have got better over the last couple of weeks I think -W]

Yes, thinks have gotten better in the Arctic.
SIPN projects 4.3 as the median.
With 0.56 standard deviation.
SIPN suggests that you have something like a 15 % probability of winning, while I have a 1 % of winning.
Overall, the most likely outcome is a 'void'.
Not to mention that my albedo (physics-based) projection is lower that 4.3, at 3.8. But June numbers will refine that.

[We could set up a market in our bets. That might be interesting. Suppose we take your numbers, above. Then your side winning is "worth" $10k * 1% = $100. My side winning is worth $1500. Your position has a net value of -$1400, mine +$1400. So if I or you (were we to do this I don't think it would be necessary for us both to need to agree to do it) were to sell "shares" in the bet, lets say 10% shares, a 10% share of my side ought to be worth $140; and people ought to be prepared to accept a 10% share of your side for a fixed $140 from you.

I think that's right. So, just for the fun of it, would anybody be interested, at such rates? It would give me a chance to lock in profit and mitigate loss -W]

By Rob Dekker (not verified) on 01 Jul 2016 #permalink

That is creative thinking, William.

For starters, here is the SIPN June report (based on May data) :

If anyone believes that this year's September sea ice extent will go above 4.8 M km^2 (remember, still quite far below IPCC projections for this year) that make William an offer to buy shares in this bet for a bargain price. SIPN puts his odds at 15%, and it seems that William is interested in selling at that rate.

You may want to give Anthony Watts a call and see if his crowd is interested. Not that I think they are, since they never put their money where their mouth is. That's how this bet between you and me came about after all.

Me personally, if you check the "Dekker" entry in that SIPN report, I believe we are heading for 3.8, and thus that the odds are still in my favor (5% against you 2.5%). So I would entertain offers for my shares in the range of 5%. And since I think you (and SIPN) are overestimating, I am interested in a 'shorting' your shares. Are you interested in that ?

We would need to make a note that in case of "void" results, everyone gets their money back, or else nobody would be interested...

By Rob Dekker (not verified) on 01 Jul 2016 #permalink

Sorry. I did not think that through.
Seems that you are selling shares 15:1 in your favor.
So a fair price for your shares would be 93 cents on the dollar.
And I am selling shares 2:1 in my favor.
So a fair price for my shares would be 50 cents on the dollar.
With 'void' being 'void' so everyone gets their money back.
Is that what you are proposing ?

By Rob Dekker (not verified) on 02 Jul 2016 #permalink

William wasn't proposing " ‘void’ being ‘void’ so everyone gets their money back" he wants to lock in a $140 profit in the event of a void bet. Person X buying a 10% share of the bet would pay $140 for Williams side (or receive $140 for your side) in any event and may win or lose $1000.

Seems William thinks either void outcome is more than 84% likely or he has more than 1% chance of losing. 1% seems a bit low to me so not surprised William is offering this and I am not interested in taking that.

[I'll try to think through what I actually meant, and what probabilities i actually think might make sense, and do a post on that, just for the fun -W]

Latest piomas volume is around 13.176 k Km^3 for day 182 compared with previous lowest 3 being
2010 182 13.258
2011 182 12.665
2012 182 12.078

So volume appears well above 2012 even if area and extent are at similar levels to 2012.

Extent is back into first.
8,598,133 km2(July 4, 2016)

By Phil Hays (not verified) on 05 Jul 2016 #permalink

I just posted the June data results of my SIPN September projection analysis at Neven's :…

There has been some speculation that 2016 will beat all records, and some speculation that it will not make the top-3.

I don't want to disappoint anyone, but regression analysis suggests that it will be neither of the two.

NSIDC's June 2016 "extent" and "area" numbers came in, as well as Rutger's snow lab's numbers for land snow in June.
Using these numbers in a formula that estimates how "white" the Arctic was in June
Regressed over past extent, I obtain between 4.0 and 4.1 M km^2 as the most likely outcome of the September sea ice extent this year.

More importantly, the standard deviation over this prediction is 340 k km^2, which is pretty darn 'tight', which means that we can say something about the probability of 2016's ranking.
Based on these numbers :

There is a 92% chance that 2016 will end up in the top-3 (only 2 out of 24 years showed a larger difference between 2011 and this 2016 projection).

There is 66% chance that 2016 will be second place (after 2007 but shy of 2012).

There is an 8 percent chance of 2016 beating the 2012 record in September.


Specifically for our bet :
There is a less than 4% chance that William will win this bet (not a single year over the past 24 showed more than 750 k km^2 (4.8 - 4.05) upside on the residual in the prediction), and

There is a less than 4% chance that I will win this bet (not a single year over the past 24 showed more than 950 k km^2 downside on residual in the prediction).

William, I'm looking forward to your post, but remember it looks like there is at least a 92% chance of this bet ending in 'void'.

[Its going to be hard to sell futures in the bet when there's so little chance of a win for either side. Although I notice your estimate of our chances have shifted since a week or two ago. Perhaps we were both too cautious; and it would have been more exciting if we'd been prepared to go more towards the middle -W]

By Rob Dekker (not verified) on 06 Jul 2016 #permalink

Taking a step back, it is kind of surprising that this best effort (based on snow and ice albedo effect) projection is smack in the middle of our projections from 5 years ago.

I feel we can both learn something here.

By Rob Dekker (not verified) on 06 Jul 2016 #permalink

As this bet is in US dollars, it keeps getting larger in pounds, thanks to Brexit. GBPUSD=x is 0.7752 this morning vs about 0.6109 when it was made.
I hope you are currency hedged.

By Phil Hays (not verified) on 06 Jul 2016 #permalink

William said " Although I notice your estimate of our chances have shifted since a week or two ago. "

Yes. That is mostly because the new estimate (4.05) is based on June data, while the previous estimate (3.8) was based on May data. 4.05 is a bit closer to your 4.8 projection than it is to my 3.1 projection, so you have the advantage at this point. By how much in absolute sense, that is very much debatable, especially since the uncertainty in the projection is reduced.

William said : "Perhaps we were both too cautious; and it would have been more exciting if we’d been prepared to go more towards the middle"

Yes, it would have been more exciting, especially with the amount of money at stake.
But I think we did the right thing (leaving the size of the void as we did).
After all the (weather) uncertainty in projecting sea ice extent several years in advance is something like 700 k km^2 standard deviation. By leaving a 1.7 M km^2 void, we essentially reduced the odds of winning to a high probability of our individual positions to be correct.

Your position was that sea ice extent would recover to IPCC GCM projection trends.
My position was that sea ice extent would follow ice volume decline and thus accellerate its decline.

Neither one of us appeared to be correct.

By Rob Dekker (not verified) on 07 Jul 2016 #permalink

Currently 2016 is in third place, at 7,908,219 km2 (July 12).

By Phil Hays (not verified) on 13 Jul 2016 #permalink

SIPN July report (based on June data) is out.
Lowest 3.6 median 4.33 highest 5.2
6 of 36 at or above 4.8, none below 3.1

Large proportion of forecasts agreeing on close to 4.3 - not sure whether this suggests ensemble is improving.

Don't think this or July data changes void outcome being the most likely outcome.

Currently 2016 is in third place,

By Phil Hays (not verified) on 02 Aug 2016 #permalink

Currently 2016 is in third place, 6,201,034
km2 (Aug 2)

By Phil Hays (not verified) on 02 Aug 2016 #permalink

ADS( formerly IJIS) 2006 declined unusually little - only 1.2m km^2 from 3rd Aug to minimum. Current 6.08 -1.2 = 4.88. So for you to win the extent would have to decline as little as 2006. Seems very unlikely due to lower volume after 2007 and lower area in season to date likely meaning more heat in water which will very likely cause more bottom melt.

There would have to be record breaking large decline from 3rd Aug for you to lose. Not going to call it as void yet but maybe soon that will become possible.

Bottom, bottom, where is the bottom?

By Phil Hays (not verified) on 08 Aug 2016 #permalink

5.71 so you need a fall of 0.9m km^2 or less. Lowest decline since 2000 is 2006 with fall of 0.94.…

In Alaskan waters, near Glacier Bay National Park.

Northwest passage looks open on ADS and Bremen. Not clear on NSIDC, still pixels set near shore, but NSIDC has pixels set where there is clearly no ice. Terra image might be the best look:…

There looks to be some scattered ice north of King William Island. Weeks of melting to go, probably not an issue at all.

By Phil Hays (not verified) on 10 Aug 2016 #permalink