Climate Statistics Prof rips climate model claims: Certain Trade War over Uncertain Models — a response to Professor William Nordhaus,
June 8, 2015: Certain Trade War over Uncertain Models — a response to Professor William Nordhaus, sent to the New York Review of Books.
June 5, 2015
Letter to the Editor, New York Review of Books:
William Nordhaus must really be convinced that a catastrophe will unfold if, as the UN’s Intergovernmental Panel on Climate Change predicts, industrial emissions of carbon dioxide raise the concentration in the atmosphere of this generally beneficial warming gas from four hundredths of one percent to eight hundredths of one percent over the next 100 years (A New Solution: The Climate Club, June 4, 2015).
After all, Professor Nordhaus is willing to start a global trade war by forming a “Climate Club” of governments who agree to place heavy tariffs on the products of governments who refuse to join them in sharply restricting their emissions. (One hopes that the Club would exempt Africa from this trade war, since life expectancy there hovers in the mid-50’s, and at present only carbon-based energy can deliver electricity to the 75 percent of homes that currently lack it.)
What is the basis for Professor Nordhaus’ conviction? Like the UN, President Obama, and members of the Democratic Party as a rule, it’s the models: “Scientists are increasingly confident that the basic results of climate modeling are accurate. Climate models calculate that past emissions have contributed to warming of almost one degree centigrade over the last century, with rapid continued warming projected over the present century and beyond.” Professor Nordhaus claims that there have been “disruptions” of the climate to date due to this tiny warming, and that there are “potential dangers” to come.
As someone who has helped students in math modeling and statistics classes sort through these sorts of hypotheses over the past decade, I cannot share Professor Nordhaus’ certainty on either the causes or the effects of the warming to date and the predicted warming to come. The models do not, in fact, attribute the recorded 1.2 degree Fahrenheit (or .7 degree Celsius) rise in temperature since around 1880 entirely to human-based emissions of warming gasses. (And remember, this is a very rough estimate of the rise, since methods of producing a “global mean temperature” have varied widely in global coverage and accuracy over this period.)
Fully half of the rise came before human-based emissions were large enough to have much effect. The computer models attributes that warming, from 1880 to 1940, to the very difficult-to-measure variables of solar power and decadal ocean oscillations that they choose to exclude as causes in more recent years. The IPCC’s finding is only that it is “extremely likely” that “most” of the half a degree rise from 1960 to 2000 (when a 15-year “pause” ensued) was due to human-based sources.
Even that claim is highly suspect. The “time-step” computer models of the climate that predict conditions in the near future, and then use those predictions to estimate conditions in the next near-future, have not, in fact, gotten fundamentally more convincing since the 1950’s. There was a reason that Cold War professor John von Neumann gave up trying to use climate models to prepare a U.S. “climate war” that would cause drought and starvation in the Soviet Union. Models of our complex and chaotic climate system simply don’t make useful predictions after a few days’ time.
The climate models have gotten more complex, for sure, with thousands of estimated parameters for warming potential, vorticity, circulation patterns, absorption of heat, pressure, energy, and momentum by various layers or atmosphere, land, ocean, and sea-ice. But they are still like models of the stock market, based not just on theoretical causations, but on correlations in conditions that will never be repeated. This makes prediction, well, unpredictable. A climate model runs on many imperfect assumptions, including crucially the precise warming potential of carbon dioxide. It must be constantly “tuned” to control its predictions, since slight changes in any of the thousands of assumptions and parameters can cause it to explode upwards or nose-dive to zero in future time periods.
The modelers themselves call their future estimates “scenarios” and not “predictions.” They agree that their models’ workings are partially based on ideal theory and partially based on “opaque” equations, since they must play with the levels of parameters representing the unknown effect of the interaction of numerous variables until they get a nice “back-fit” with previous temperature records. What the models are not based on is a sound knowledge of what in the world causes our global temperature to move the way it does.
Every 100,000 years the earth experiences a swing of 20 degrees Fahrenheit down and up. Atmospheric physicists agree that this swing is very likely caused, at some remove, by one of the “Milankovitch cycles,” because the earth’s orbit oscillates between a nearly-perfect circle to a five percent oval and back during this period. As Oxford physicist Fred Taylor has written, however, the amount of solar power reaching the earth is not significantly affected by this change in orbit, and nobody has identified either the initial mechanism or the feedback mechanisms by which the temperature responds. Given the difficulty of identifying the source of a 20-degree oscillation, a little humility is in order about our knowledge of the cause of a fairly typical change of one degree in one hundred years.
Caleb Stewart Rossiter
Adjunct professor, Department of Mathematics and Statistics, and
School of International Service, American University, Washington, DC