Sunday, February 12, 2012

Climate betting update - back to where we started, and I'm not worried

Several weeks ago I received my annual friendly email from David Evans, my betting partner/opponent over future temperature increases, with the 2011 temp data from GISS.  Bet details are here, summary is that I've bet $9,000 and David $6,000 for a series of decade-plus bets over whether temps will increase somewhat near the IPCC projected rate or at a much lower rate.

David mentioned that the five-year average we're betting on is, for 2009, back to where we started in 2007, which would be very bad for my side of the bet, but he also politely added that we've had some La Ninas to affect things.  I'm not too worried for my pocketbook - the five year average for 2009 overlaps too much with 2007 to expect a lot of movement.  The five year average for 2012, which we won't know until 2015, will be the first that has no overlap.  First bets pay out or are voided in 2020, so as David says, this betting will take a while.

One way to look at it is to assume a counterfactual that the bet started in 1989 and ended in 2009 - fair from my viewpoint of expecting past rises to continue/accelerate, but not what David expects.  My calculation is that I win all three of the low end bets, void one of the high end bets, lose the other two high end bets, and come out ahead $1000.

Not much other activity from me on the climate betting front, and I haven't heard much from others either besides William's sea ice, unless I've missed it.  I've been thinking that sea level rise might be less noisy than temps, allowing shorter bet periods and maybe using three year averages.  Maybe will look into that later.

Last year's betting update here.


Alastair said...

I think you will find when you look into sea level rise that sea level has actually fallen over the last three years!

Cheers, Alastair.

William M. Connolley said...

I've been tickling the Watties - see - and have had a nibble, but nothing firm yet. That would be only $1k.

J Bowers said...

"I think you will find when you look into sea level rise that sea level has actually fallen over the last three years!"

But not due to a reduction in thermal expansion or less ice melt. Scooped up from the oceans, dumped on land and making its way back to the oceans.

Elmer Fudd said...

What if the mathturbation has a seed of truth in it? If there are factors large enough to effect global temperatures but small enough to have escaped AR4 CMIP3 modeling? If so, you may find your own lack of doubt in the face of the old arts ... disturbing.

Anonymous said...

Alastair: I think if you look, you'll find that your 3 year figure has been out of date since the November update:

Perhaps you could get away with saying that sea level has not risen in two years. But I wouldn't be surprised if the next update shows a continue rise over the last couple of months...


a_ray_in_dilbert_space said...

Elmer Fudd,
And what if Martians have developed an invisibility cloak and a heat ray!!!???

Isn't it a pity there's all this evidence to inhibit our creativity.

Brian said...

Alastair, I was thinking of bets with a minimum time gap between reference points of five years instead of ten years as I've done for temps, and each reference point based on a three year average instead of a five year average.

Elmer said...

a_ray ... I understand that the models include the best physics they can within certain computational constraints (mainly calcluation time). The constraints force trade offs - spatial resolution, temporal resolution, physical parameter resoultion, physical equation complexity. Are you sure that nothing important is missing in the models? That they are essentially complete? If not, that they are at least good enough for govt work? Are you sure that the global temperature integrations are more supportive (of the current state) of the AOGC models rather than those curve fitters who include ~60yr sines? Is your certainty based on careful analysis or hubris?

word verification: wrecr

a_ray_in_dilbert_space said...

Given that all models are wrong, the question is which models are useful. And in science, useful means possessing predictive power. GCMs have a rather impressive record of being right--much better than the mathturbators.

But just to humor you, what if the models do not treat all important influences? Why would you conclude that the error must favor negative rather than positive feedback?

Indeed, the models are among the most important constraints limiting climate sensitivity on the high side. Without them, the upper 90% CL goes to 5.5 or perhaps even 6 degrees per doubling. What is more without the models, we are flying blind. Wouldn't that increase the urgency of drastic action rather than abate it?

I will never understand how those who argue against "alarmism" can at the same time contend the models suck. The models are the closest to friends you have in the actual arsenal of science.

Anonymous said...

Answers for Elmer:
Nothing important missing? Maybe
Essentially complete? No
Good enough for gov't work? Yes (for most purposes. No for adaptation work that requires confidence in small-scale, near-term-ish regional projections).
Better than curve fitters with 60yr sines: Yes, definitely.
Analysis or hubris? Experience with modeling, knowledge of physics, and understanding of statistics supports the conclusion that the curve fitting approach with 60 year sine waves is dumb.


Elmer said...

a_ray, appreciating a model means understanding its weaknesses as well as it's strength. I have not suggested that we abandon them or dismiss them. And what makes me consider that the errors (in the next few decades) might favor the negative is the evidence - the globally integrated temperature records. The divergence, while perhaps not yet statistically significant, is on the downside.

Why this should be of the utmost concern to those who wish to advocate policy based on modeling is that *IF* the ~60yr cycle is real and significant enough to influence the global temp and it peaked in mid-1940s and early-2000s, *THEN* there will be significant slowdown in _rate_ of the global temp increases. This will destroy all will to implement CO2 restricting policies. Then, 20-30 years from now, the cycle will push things up just that much faster. This timing could have very important political policy implications for adaptation or mitigation. Important enough to, IMHO, spend some time considering rather than simply dismissing.

dhogaza said...

" And what makes me consider that the errors (in the next few decades) might favor the negative is the evidence - the globally integrated temperature records. The divergence, while perhaps not yet statistically significant, is on the downside."

There's an obvious error - a clear failure to divine a fairly decent solar minimum during much of the last decade.

Of course, GCMs don't pretend to do solar modeling or such predictions, and indeed the solar experts are pretty much clueless in their ability to predict strength of solar cycles (or precise timing).

So I'd say this isn't evidence of GCMs being broken, but rather the predictiive abilities of solar scientists being broken. If they'd been able to forecast TSI for the last decade in advance, the slight drop in TSI could've been input as a model parameter would've been enhanced.

The recent paper removing known natural variability documented over at Tamino's place puts all this "divergence" crap in context.

Elmer said...

"If they'd been able to forecast TSI for the last decade in advance, the slight drop in TSI could've been input as a model parameter would've been enhanced."

It still can be input as a model parameter. Let's fire up GISS ModelE with and without actual TSI post CMIP3 and compare results!

Steve Bloom said...

Also, Elmer, while you were out hunting that Darned Model the ~60 year cycle thing has pretty much gone poof. Do pay more attention.

a_ray_in_dilbert_space said...

You act as if the temperature time series were the only data we have. It isn't. In fact, it is one of the more difficult datasets to interpret, because we do not have a good real-time answer for TSI + outgoing IR + tropospheric warming + shallow and deep ocean warming. Fortunately, we do have several independent datasets constraining climate sensitivity, and they all favor a range around 3 degrees per doubling.

Right now AMO is the denialist flavor of the month. Before that it was cosmic rays or "warming from the ice age". After that fails it will be something else. All of these are red herrings. If you look at the evidence in its entirety, there is only one way to make sense of it--the consensus model of Earth's climate. The GCMs are expressions of this model--some better than others. However, none of the alternatives has any verified predictions under its belt.

Models are comparative. You select the one that is best. Then you modify it as needed. I don't see a need for drastic modification coming any time soon.

Elmer said...

Poof? Down the rabbit hole? Hmmm

Imprint of the Atlantic Multidecadal Oscillation on Tree-Ring Widths in Northeastern Asia since 1568, Wang, 2011

Tracking the Atlantic Multidecadal Oscillation through the last 8,000 years Knudsen, 2011

Atlantic Multidecadal Oscillation and Northern Hemisphere’s climate variability Wyatt, 2011

Albeit ... Is the Atlantic Multidecadal Oscillation (AMO) a statistical
Vincze, 2011

"Poof" is right. I see no nail. I see no coffin. Just a blind determination to call the thing dead. Something not to be discussed openly in polite company. Nevertheless, I appreciate your well intentioned replies.

Jeffrey Davis said...


The AMO is not a heat source. It is a heat transport mechanism. If the temps associated with the AMO go up it's because something has added heat to the Atlantic Ocean.

Thinking that the AMO raises temps independently of some other heat source is akin to thinking that cars create people.

Elmer said...

JD ... the question on the table is not whether the AMO is a heat source. The question on the table is whether the AMO distributes heat in such a way that the short term global surface temperature predicted by GCMs is too high. If so, public confidence in GCMs will further erode and Brian may lose a bunch of carrots.

a_ray_in_dilbert_space said...

Fourier analysis tells us that if you take enough trigonometric functions and add them together, you can reproduce just about any curve over a finite interval that you want. Often it takes surprisingly few terms. AMO just happens to have the frequency they need to get close. It isn't physics. If Fun with Fourier.

Steve Bloom said...

Weally, Elmer? Try Tokinaga and Xie 2011(b). All the cool scientists know about it. Check out the FOD on that one, and bear in mind Hansen's remarks about aerosol forcing being much larger than previously thought. (Gavin promises a post on this very soon.)

This is not say that's it's proven that there is zero AMO signal historically. What is certain, though, is that the anthropogenic signal is far larger and will swamp any effort at quantifying something as insignificant as the postulated AMO.

EliRabett said...

Of xourse there is not AMO, Mike Mann invented it.

Steve Bloom said...

Iwonic, innit?

David B. Benson said...

Elmer Fudd --- Granger causality runs from the northern hemisphere temperature to the AMO; no surprise there.

Brian said...

I sure hope Elmer's right. It'll cost me $9,000 but win the planet a few extra decades. A worthwhile tradeoff, if just barely.

However, I think I'm coming home with carrots.

Robert said...

Important link: