More on Climate Predictions

No pun intended.

Scott Armstrong has again asserted on the JDM list that global warming forecasts are merely unscientific opinions (ignoring my prior objections to the claim). My response follows (a bit enhanced here, e.g., providing links).


Today would be an auspicious day to declare the death of climate science, but I’m afraid the announcement would be premature.

JDM researchers might be interested in the forecasts of global warming as they are based on unaided subjective forecasts (unaided by forecasting principles) entered into complex computer models.

This seems to say that climate scientists first form an opinion about the temperature in 2100, or perhaps about climate sensitivity to 2x CO2, then tweak their models to reproduce the desired result. This is a misperception about models and modeling. First, in a complex physical model, there is no direct way for opinions that represent outcomes (like climate sensitivity) to be “entered in.” Outcomes emerge from the specification and calibration process. In a complex, nonlinear, stochastic model it is rather difficult to get a desired behavior, particularly when the model must conform to data. Climate models are not just replicating the time series of global temperature; they first must replicate geographic and seasonal patterns of temperature and precipitation, vertical structure of the atmosphere, etc. With a model that takes hours or weeks to execute, it’s simply not practical to bend the results to reflect preconceived notions. Second, not all models are big and complex. Low order energy balance models can be fully estimated from data, and still yield nonzero climate sensitivity.

I presume that the backing for the statement above is to be found in Green and Armstrong (2007), on which I have already commented here and on the JDM list.

G&A 2007 says:

Reid Bryson, the world’s most cited climatologist, wrote in a 1993 article that a model is “nothing more than a formal statement of how the modeler believes that the part of the world of his concern actually works” (p. 798-790). Based on the explanations of climate models that we have seen, we concur. While advocates of complex climate models claim that they are based on “well established laws of physics”, there is clearly much more to the models than the laws of physics otherwise they would all produce the same output, which patently they do not. And there would be no need for confidence estimates for model forecasts, which there most certainly are. Climate models are, in effect, mathematical ways for the experts to express their opinions.”

Notice that Bryson says how things work, not what will happen. Again, except in trivial models, the behavior that emerges from the operational specification of how things work may not be easy to manipulate, making models a difficult vehicle for opinions. It is obvious that models reflect beliefs; how could it be otherwise? There is no magic procedure or list of principles that leads to an assumption-free representation of a system. The question is to what extent those beliefs have been tested, and models make it easier to perform tests. When people refer to “well established laws of physics” they refer to beliefs that have been tested very extensively, like the freezing point of water or the laws of thermodynamics. Where climate models differ is on the description of not-so-well-tested beliefs, like phenomena below the resolution of the model. For those, verification is partly direct (is there any science at the micro level to back up the assumptions?) and partly indirect (does the assumption yield plausible macro behavior in the model?). In both cases, there may be uncertainty, but it does not follow that the macro results are wrong, or that the modelers fabricated the inputs to yield a desired result. Nor does it follow that models with perfect physics would not need confidence estimates; uncertainty also arises from things like error in measuring the initial state of the system. So, yes, a model expresses an opinion, but not in the usual sense of “polar bears are cute.” Models express opinions in a manner that is at least replicable, shareable, and testable, even if wrong.

G&A 2007 goes on to assert that:

To our knowledge, there is no empirical evidence to suggest that presenting opinions in mathematical terms rather than in words will contribute to forecast accuracy.”

This is obviously a ridiculous statement in many situations. Are we to believe that there is nothing to be gained by translating the statement “earth goes around the sun” into a mathematical model involving G*m1*m2/r^2, and plugging in the gravitational constant, earth mass, etc.? Looking at the procedures by which a model was created may raise important questions, but it’s neither necessary nor sufficient to show that the model yields bad predictions. We also need to understand the problem domain and the models, and ultimately there is no substitute for testing.

Even the developers say that these models do not provide forecasts — they are only scenarios.

Often this is true, but it doesn’t imply that models have no predictive power. What climate scientists mean is that they are producing contingent predictions of climate response to greenhouse gas and natural forcings, but they are not forecasting the inputs to the process (e.g., carbon emissions and volcanic eruptions). Failure to forecast the inputs is not necessarily debilitating because it is clear that atmospheric concentrations of greenhouse gases go up in any scenario not involving deep emissions cuts.

Gore will stick to the story line that this is a scientific approach. Yet, as noted in extensive research, especially by Phil Tetlock, it is false.

Has Tetlock studied climate modeling or climate scientists? I am only aware of his research on political forecasting. His CV, while stellar, does not appear to cover the natural sciences. What evidence allows us to generalize from politics to other problem domains?

This holds even if Gore’s assumptions about climate science were true — and many of them have been shown to be false — such as the hypothesis the CO2 causes global warming (published evidence by Willie Soon and others suggests that the causality runs the other way).

Presumably this refers to Soon’s 2007 Physical Geography article. Soon does not claim or prove that CO2 has no warming effect. Instead he argues that, in the paleoclimate record, orbital forcing and other feedbacks are sufficient to explain temperature and ice changes without greenhouse gases as an amplifier. Certainly many climate scientists would agree with the contention that orbital and other factors played a major role in glaciations, though I suspect many would also disagree with Soon’s conclusion that non-GHG forcings are sufficient by themselves. Correct or not, Soon’s conclusion does not say much about warming effects of anthropogenic GHGs under current conditions, where we can expect no significant orbital changes over our lifetimes. In fact, his citations (figure 6, for example) show that models with mainstream climate sensitivity are in fact consistent with paleo events. (One might also argue that, if climate models are useless, then most of Soon’s paper is also meaningless).

Personally, I find the Soon article to be a useful illustration of the hazards of model-free inference about complex phenomena. It considers over a hundred citations and dozens of competing feedback mechanisms and data streams, without an integrating framework that allows the reader to determine where assumptions conflict or coincide. In any case, it is quite silly to argue about which way causality runs when the variables concerned are in a feedback loop, as temperature and greenhouse gases certainly are. Borrowing a phrase, the fact that chickens hatch from eggs does not preclude them from also laying them.

It is also interesting that Gore refuses to debate the issues. He has also refused to engage in a test of his claims; the following announcement was sent out on March 28:

Al Gore misses the March 26 deadline for the Global Warming Challenge.
The extended due date for the Global Warming Challenge passed with no word from Mr. Gore. Although he and Professor Armstrong have had a number of communications, Mr. Gore offered no response to the key question:
“When and under what conditions would you be willing to engage in a scientific test of your forecasts?”

I’m not really surprised. If someone offered me a sucker bet with adversarial fanfare, I wouldn’t be inclined to pursue the conversation. The Global Warming Challenge is indeed a sucker bet, with terms slanted to favor the naive forecast. It focuses on temperature at just 10 specific stations over only 10 years, thus exploiting the facts that (a) GCMs do not have local resolution (their grids are typically several degrees) (b) GCMs, unlike weather models, do not have infrastructure for realtime updating of forcings and initial conditions (c) ten stations is a pathetically small sample, and thus a low signal-to-noise ratio is expected under any circumstances (d) the decadal trend in global temperature is small compared to natural variability.

Validation of forecasting methods is a key issue in climate change because, although we know that climate varies, we have been unable to locate a single scientific forecast that supports global warming. If Mr. Gore or anyone else is aware of such a forecast, they should reveal the source to the scientific community. Claims that science supports global warming forecasts have, to date, failed to provide sources.

It seems to me that the failure is due either to lack of trying or narrow framing of the term “scientific forecast” or both. By my count, G&A 2007 only cites the primary peer-reviewed climate literature 5 times out of 41; it cites a single skeptical book 4 times. It neglects the early climate literature, including the Sawyer article cited here by George Christopoulos.

In contrast to the billions of dollars being spent to support global warming research and policies (and the $300 million on advertising), skeptics on the issue are nearly all working with little or no funding.

Given that the Heartland Institute just funded a high-profile skeptics’ conference with hundreds of attendees, the AEI last year offered $10,000 bounties for articles critical of the IPCC, and the CEI is running national TV ad campaigns, I find it a little hard to believe that a skeptic with a credible argument would have to go begging.

A number of skeptics have been fired from government position for failure to toe the party line on warming.

Perhaps you could list some names to substantiate this claim?

Those of you interested in the study of persuasion will note that the global warmers have adhered to many of the principles in Pratkanis, Anthony R. & Elliot Aronson (2000), The Age of Propaganda: The Everyday Use and Abuse of Persuasion. New York: W. H. Freeman.

Ironically this very sentence uses one of the principles: labeling colleagues who hold an opposing view as “global warmers,” implying agenda-driven science and unity of thought. I’m not aware of compelling examples of propaganda from climate scientists, though certainly there are exaggerated claims in the press, so I would appreciate a few references. Propagandistic excesses in the pursuit of climate skepticism are widely known, though again I would hesitate to call them science even when they occur under a pseudoscientific veil. Take, for example, the “Oregon Petition,” a one-sided pseudo-survey with a cover letter misleadingly designed to look like a PNAS communication. The letter author, Fred Seitz, was scientific advisor to RJ Reynolds’ medical research program, and a senior scientist at the Marshall Institute along with Soon (cited above). Then there’s the Great Global Warming Swindle, which distorted data and was disavowed by several scientists interviewed in it. Or take Patrick Michaels’ erasure of the most plausible trajectory from a 1988 article by Jim Hansen, in order to make it appear that model results were incorrect. If these aren’t propaganda, I don’t know what is.

Certainly there is an important research agenda here. Science and modeling are social processes, and have been known to go astray. Ordinarily, science is self-correcting as anomalies accumulate around theories and models with low explanatory power. When the stakes are high, it would be useful to ensure that anomalies are attended to in order to speed the correction, and that decisions are made with proper appreciation of uncertainties. Failures of that process in the past or in non-science disciplines do not necessarily indict climate science. To really understand the situation requires dialog with climate scientists and examination of actual models and results, not polemics against Al Gore based on anecdotes and a shallow reading of synthesis documents. In fact, such external attacks may be counterproductive if they induce threat-rigidity.