First Get the Science Right

The Normalised Impulse Response curves of CO2 concentration. Dashed curve: model prediction. Full curve: estimated from observations. The curves show how long CO2 will remain in the atmosphere.

by John Reid

My paper The Statistical Testing of Climate Models is presently under review with Proceedings of the National Academy of Sciences of the United States of America (PNAS).

Concern with possible effects of increasing atmospheric CO2 concentrations on climate began in earnest in the 1970s. Coincidentally, this was at a time when digital computers were rapidly becoming more powerful so that computer modelling of the ocean-atmosphere system of the entire planet was possible and presented an obvious research stratagem. The atmosphere and ocean are divided up conceptually into perhaps a million or so rectangular boxes which interact with adjacent boxes. The laws of physics are then applied to the material in each box to create an exact replica of what happens in the real world.

Except that it can never be an exact replica!

One physical law cannot be properly incorporated into such models. It is the Second Law of Thermodynamics which deals with probabilities, turbulence, heat flow and the direction of time.  There are other aspects of the physical world which climate models do not handle well such as cloud formation. There are numerous fudges and ad hoc “rules of thumb” included in the models to account for such things. Despite these shortcomings, numerical modelling has become the sine qua non of climate science to the extent that the number of papers describing model output far exceeds the number concerned with real world observations.

Things could have turned out differently. The 1970s was also a time of great development in statistics. Some statistical methods developed then are well suited to the environmental sciences because, unlike numerical models, they take into account the inherent randomness of the physical world. It is a “bottom up” approach which starts from the data, unlike modelling which is “top down” and starts with the modeller’s opinions about how things ought to behave. Climate modelling resembles Ptolemaic astronomy according to which planetary orbits must always be circles (or circles on circles …). The Ptolemaic, top-down model was finally overthrown by the bottom-up approach of Galileo, Brahe and Kepler, culminating in elliptical orbits and the gravitational insights of Newton.

The t-test probablity, P, that the Endogenous Sensitivity of CO2 Concentration will have a given value. The vertical dashed lines are the 95 percent confidence limits.

It is high time that modern statistical methods were brought to bear on climate. My paper does just that and reaches the conclusion that the current concern that carbon dioxide emissions will persist in the atmosphere indefinitely is unwarranted. My analysis clearly demonstrates that C02 will dissipate more quickly than current projections. This has huge ramifications for climate policy.

This is not an ivory tower squabble; it has enormous consequences. People are right to worry about recent exponential growth in fossil fuel consumption and population. Such bubbles have been seen before in animal populations and in stock markets. They are always followed by collapse with catastrophic consequences for the animals or investors concerned.

World leaders at COP26 are grappling with the wrong issue. It is not the global ecology which is threatened so much as the global economy. Fossil fuel reserves are finite. Peak Oil is inevitable and not far off. Energy Starvation is the real issue, not Emissions Reduction. We have had a foretaste of slowed economic activity during the COVID lock-downs. It was very small compared with what we face in the near future, and had no measurable effect on CO2 concentration.

Economic catastrophe is not inevitable. As a species we have the ability to understand our interactions with the natural world and their consequences and, hopefully, the ability to do something about them. The decay side of the peak oil curve is likely to be slower than the onset side, which means we may have time to develop new fuels such as wood-derived gasoline substitutes and safe, cheap nuclear power such as Hydrogen-Boron fusion.

But we must take a hard-nosed engineering approach and base our policies on real science rather than propaganda and political loyalties. Above all we must not conflate the coming energy crisis with other causes, however noble they may be.

A PDF pre-print of the paper can be downloaded here: Reid2021

The paper was rejected by PNAS on 16 Nov 2021, viz.:

Because we receive more than 18,000 submissions every year, incoming manuscripts undergo an initial evaluation by a member of the Editorial Board, who is also a member of the National Academy of Sciences, to determine whether the potential novelty, impact, and relevance in the broad scientific community merit further detailed technical review. In your case, our assessment is that your manuscript does not meet one or more of the principal aims of our journal and on this basis we expect that the likelihood that detailed review will lead to publication is low. This decision is necessarily subjective and does not reflect an evaluation of the technical quality of your work or of its appropriateness for a more specialized audience.

Editorial Board Remarks to Author:

After consulting with several experts, we have decided that this manuscript is not suitable for the broad readership of PNAS and should instead be submitted to a more specialized journal.