Hockeysticks, the tragedy of the commons and sustainability of climate science.
Hans von Storch – Director of Institute of Coastal Research of the GKSS Research Centre in Geesthacht. Professor at the Meteorological Institute of the University of Hamburg, Germany
Panelists: Warren Washington, Caspar Amman and Doug Nychka, NCAR, and Roger Pielke Jr. (CIRES)
Location: Mesa Lab Main Seminar Room Time 3:00pm
Abstract: The "hockey stick", elevated to icon-status by the IPCC, plays a crucial role in debate regarding climate change. Yet the methods used to develop it have not been completely explicated. We have tested the method in the artificial laboratory of the output of a global climate model, and found it to significantly underestimate both low-frequency variability and associated uncertainties. Our work focuses on multi-century simulations with two global climate models to generate a realistic mix of natural and externally (greenhouse gases, solar output, volcanic load) forced climate variations. Such simulations are then used to examine the performance of empirically based methods to reconstruct historical climate. This is done by deriving "pseudo proxies" from the model output, which provide incomplete and spatially limited evidence about the global distribution of a variable.
Our simulation study was published in "Science" but received less response than expected – almost no open response, a bit in the media; but many colleagues indicated privately that such a publication would damage the good case of a climate protection policy.
In this talk the methodical critique of the hockey stick methods will be presented, followed by a personal discussion about the problem of post-normal climate science operating in a highly politicized environment. The presentation will be followed by a panel discussion on the science of the hockey stick in the context of high-profile political issues.
Maybe someone will ask Ammann about the R2 of his reconstruction, whether his emulation method is almost identical to mine, whether he’s replicated Mann’s Preisendorfer selections (or any of the other questions in my list) and how he’s coming along with his GRL submission. Someone should ask von Storch whether he can get a high RE combined with a negligible R2 in his simulations and what the effect would be of pseudoproxies with a nonclimatic trend (emulating bristlecones) ?
Climate Audit 
19 Comments
Maybe someone can ask Roger whether he would defy a request from a Congressional committee to provide information that the US taxpayers have paid for, risking a subpoena and loss of further funding for his research…as he suggested that Mann do.
John A.- Hard to say not being in Mann’s shoes what I would do (he obviously knows stuff about his own work that I do not); it is of course easy to armchair quarterback. But I generally wouldn’t give advice contrary to what I’d do myself. The proper place for Barton to go if he wants answers is NSF (as he has done).
Re #2
Roger,
Isn’t it a really risky strategy to ignore a clear request from a Congressional committee which can subpoena if they get no reply? Who are members of the majority party and could prevent you ever doing publicly funded research?
Don’t you think that there might be quite a lot at stake?
Also you’ll notice that von Storch said that the "Hockey Stick" was
With so much riding on it, why shouldn’t the methodology, code, source data, be investigated in a proper due diligence?
John-
No, I don’t think it is risky. This is after all still the USA. A respectful reply declining to provide the materials would put the ball back into Barton’s court. He could subpoena, which would carry with it a lot of political risk (recall the discussion of this when hearings were called on steroids and the baseball players), and then the scientists would face another decision (I would not advise ignoring a subpoena). I really do not think that Barton or anyone on the Hill would have an ability to directly affect the research funding for an individual. To the contrary, such circumstances might make it more difficult for an agency to deny funding, as the individual could always claim a political bias when his proposal was turned down. Of course, politicians can make life hell for NSF and affect NSF funding (there is a lot of experience with policy makers tinkering with agencies and their budgets for small-seeming reasons). But this too would carry political risks that seem quite out of proportion with the issue. If Barton wants to contribute substantively to this issue, he should focus his efforts on NSF if indeed paleoscientists are violating NSF’s archival policies. Congress oversees agencies not state-employed university scientists. NSF is responsible for activities carried out with NSF funds. There are better and worse ways of dealing with this issue.
And I see no problem calling for exercising the diligence that you have suggested. Over at my blog I have recommended to MBH that they release everything that you guys are asking for. At this point it would be good for the community for them to do so.
The use of psuedoproxies demonstrates how weak the hockey stick hypothesis really is.
Even if Barton is on a McCarthyite fishing trip, wouldn’t the esteemed Dr. Mann jump at the chance to humiliate McIntyre and McKitrick? This would put it to bed once and for all, and we could start discussing ‘action’.
Link for those looking for the source
Jim, I agree. If Mann is correct in his assertions about his paleoclimate reconstruction, surely releasing his code is such a huge potential victory for him personally and for the IPCC, that one has to wonder why he won’t release. It is going too far to take the obvious inference, but one has to wonder at his tactics. It all seems, well, kind of juvenile, somewhat amplified by noobs like Dano. I can understand the initial distaste, but Steve (& Ross) do seem to have uncovered some genuine and significant issues. Science is big enough to deal with the controversy, it remains to be seen if the scientists are.
I doubt, Roger, that your recommendation went down very well at realclimate ?
Given the fact that no GCM is capable of actually predicting actual future climates, one wonders what climatologists mean by, “such a publication would damage the good case of a climate protection policy.” There can be no case for a “climate protection policy” when it’s not known what climate will do without anthropogenic inputs, nor, really, what it will do with anthropogenic inputs, either. There is a tacit assumption here that the recent climate past is the measure of the (non-impacted) climate future, but that assumption is certainly not true.
I’d like to know why scientists believe that it is possible to project climate for hundred years in advance (especially temperature), when they can’t even project a weather report more than 3 days ahead?
I mean the further in the future they give us temperature reports, the less definite and detailed they are. Is this true for climate models as well? Do they get weaker the further you go into the future?
Re#10 Well, what the climate models are projecting are averages, not actual temperatures. As such they can theoretically be reasonably accurate even though they don’t even attempt to predict the temperature, etc. on a given day. The problem is that the models have all sorts of assumptions built into them and you have to either spend tons of time sorting through the code of those models which are available on-line or it’s like pulling nails to find out just what the assumptions are.
There hasn’t been much discussion of the subject on this board, but while we’re asking for due diligence and all that for papers which are produced, there also ought to be made a push for access and clarity of the assumptions made in climate models. I understand why scientists are loathe to do so, since technical writing is one of the most difficult things to do in a way which can be understood by the intelligent layman. But if it’s not done, we lose any way for the individual in a society to make informed decisions.
This is what drew me to the Global Warming debate in the first place. I’d seen lots of claims of one sort or another and was trying to find a place where all the data and ‘stuff’ was available for someone who knew science but wasn’t a climate scientist to find out what he needed to find to know what side of the debate to choose. What I found was that there is no such place and that the scientists on the ‘concensus’ side had no interest in providing it. There ARE plenty of sites where warmers are willing to feed pablum to the rank beginner, but not one where the nitty gritty is available for someone who’s willing to spend some time on studying the issues but can’t work at it full-time.
re: #10
Weather and Climate are two different things. Weather is specific wrt time and place over very short periods of time and Climate are general conditions prevailing over longer periods of time. It follows then that Weather prediction and Climate prediction are two different things though both rely heavily on similar or the same computer models of the atmosphere and the oceans. For short periods of time it is possible to model the atmosphere at a very high resolution which results in better predictions but as soon as you begin to run the models for longer periods of time (decades and centuries) the resolution becomes more coarse out of computing necessity which may result in predictions of future climate that are accurate or innacurate depending on your starting assumptions and model tuning from past climate reconstructions. The question here for this site is whether past reconstructions of temperature are significant inputs to climate models to result in significant errors in future predictions that influence policy decisions.
Steve,
What statistical measures are your referring to as R2 and RE? I am guessing one is the RMS error of a curve fit. Thanks for any response. I enjoy your site…
Dave, could you make brain surgery something the layman could do? Or accountancy? Or building bridges? Or building skyscraper? Of course not. There is no way round it, you either have to do the study (and, like you, I haven’t done much) or you either go the way you choose which is ‘My limited knowledge tells me I know better than the expert’ or the way I choose ‘I think the experts are probably right and I’m not going to go outside my envelope of learning’.
Now, you’ll probably say ‘Oh, but a brain surgeon could explain what he is doing, like how and operation might effect me’. OK, and that is what both brain surgeons and the IPCC and places like RealClimate have done. The REAL problem is that you wont accept what the experts tell you. Well, then it seems to me you have but one option, become an expert 🙂
Sigh! I’m not asking to be able to become a brain surgeon. I’m asking to be able to decide BASED ON HIS METHODS whether a proposed operation would actually solve a brain problem. And if you think the IPCC or RealClimate allow you to do that, perhaps you should go see that brain surgeon. The IPCC merely references tons of previous work. If you try actually working backword you quickly find the information you need is only available at research libraries, which a typical observer won’t have access to or time to visit. Something like RealClimate might work, but the people who run it don’t have time to perform the backwards looks which would be needed. If you want to try a sample experiment in what I’m talking about, let me know.
I just had two surgeries done for athletic injuries. And I researched them and rejected some docs, procedures. You have to get involved. Doesn’t mean you know more than the expert. But yes, you do have to second guess the expert. I learned that in the Navy. Heck, that is part of critical thinking.
The trouble with trusting experts is, who says they are experts?
Why, they do themselves! In the words of Mandy Rice Davis, ‘They would say that, wouldn’t they?’
Some people may check their brains in when they go to consult (sorry, sit worshipfully at the feet of) a surgeon, accountant, or self-proclaimed climatologist. but these of us who are interested in knowing the truth and live in the real world do not.
Churchill put it this way: ‘Experts should be on tap, not on top.’
Steven Hales, I appreciate your post. Unfortunately, it contains several errors. In it you say:
Actually, there are three different things. One is observations. I used to have to fill out the forms in the Army. They would say things like “Weather Station Baker, 3:30 PM, 3/8 cirrostratus, bases at 1000′, temperature 62°, relative humidity 36%”. These, not weather, are specific wrt time and place.
Weather is what the forecasters give us, like “Forecast for Tuesday, partly cloudy with a chance of showers over the San Francisco region”. It is the forecast average of the observations over a short space and time span. Weather can also be the actual average, as in “yesterday it was partly cloudy, with winds of 5 to 15 knots”.
Climate is the longer term average of weather in time, but not necessarily in space, as in “The climate in San Francisco is Mediterranean, with cool summers and wet, generally warm winters”.
Thus, both weather and climate are averages of observations, and the difference is only that the length of the time average of climate (generally thought of as >30 years) is longer than that of weather (generally days to years, as in “The weather has been cool the last few years at my house”.
Because both weather and climate are averages of instantaneous observations, with different time spans, I’ve never been able to believe that while weather models cannot forecast very far into the future, climate models supposedly can. I simply don’t see how.
You go on to say:
The phrase “out of computing necessity” conceals several assumptions. The main one is that if we had a big enough computer, we could successfully model the climate. However, the climate is a forced system, and some of the forcings (solar, cosmic rays, cosmic dust) originate outside the earth and are not necessarily predictable.
Even if that problem could be solved, another problem is that predicting future climate may take a computer the size of the earth … that is to say, it may be that the earth itself is the smallest computer capable of predicting the earth’s climate.
Also, you say that the prediction may be “accurate or inaccurate” depending in part on “model tuning from past climate reconstructions”. However, the fact that a model can be successfully tuned to model past climate is no indication of how well it will perform in the future. If it were, then stock market models would work … and they don’t.
Your conclusion is:
Reconstructions of past climate to date have been highly dubious, and fraught with error. The first question is not whether they are “significant inputs to climate models”, but whether they’re worth a bucket of warm spit …
w.
Willis,
To reinforce your last statment #18, there are many posts on this thread which question whether the “signal” in proxy reconstructions is primarily temperature. Many people assume that all the proxy reconstructions are linearly related to past temperatures, however this appears to be true only for some proxies and only over some temperature ranges.