Rallying to the Chief Scientist's Call


Professor Ian Chubb, Australia's Chief Scientist and a director of the CSIRO, is urging scientists to contribute to the climate change debate. "The majority of scientists should be out there," he said recently, "explaining to the public - why they do science, how they do science and how they accumulate scientific evidence."

I became a scientist because I wanted to understand how things work. For me, the only reliable way to gain this understanding is the scientific method. Without it, we are left with prophecy, divination, fortune-telling and group-think.

Science is primarily about evidence, and the rigorous testing of hypotheses against observations.

According to the Anthropogenic Global Warming (AGW) hypothesis, an observed increase in atmospheric carbon dioxide concentration was the primary cause of the apparent global temperature increase of about 0.8 degrees Celsius during the 20th century. Yet there is little observational evidence supporting it, nor would a small temperature variation over one hundred years be an unusual or "extreme" event.

The truth is AGW remains only an hypothesis; not a proven fact about the Earth's climate. Support for it comes, we are told, from climate modelling. But can a grab-bag of hypotheses expressed in numerical form be described as genuine empirical evidence? And do the models stand up well when rigorously tested against observation?

As a researcher with CSIRO with a PhD in Upper Atmosphere Physics, I was involved in developing numerical fluid dynamical models similar to those used by climate modellers today.

The CSIRO website statement - "Climate models, which are based on the laws of physics and thoroughly tested, provide credible quantitative estimates of future climate change" - is misleading. The "laws of physics" actually play only a minor part in climate model design. Furthermore, the models have no real predictive power. Their "projections" rarely match observations. If they have been "thoroughly tested", the public should be told of the outcomes - especially the frequency of failure. As it is the modelers only show graphs which support their hypothesis.

Why the Chief Scientist, whose research expertise is in neuroscience, is so keen to defend them - and the AGW hypothesis they claim to "prove" - is a mystery, at least to me.

Prof Chubb also expressed concern about the safety of climate change scientists, some of whom apparently received anonymous death threats. If true, I share his concern. Calls for them to be "thrown in jail" were, he said, "reminiscent of a time long past." A little jail time, however, would not be an unreasonable penalty for fraud - be it corporate, criminal or scientific.

John Reid




1. The "laws of physics" referred to relate to the conservation of mass and energy and Newton's Second Law as encapsulated in the Navier-Stokes equations. Numerical models based on these equations alone are generally unstable. They have to be "tweaked" by assigning unrealistically large values to internal frictional terms. Furthermore a large number of ad hoc rules-rules-of-thumb known as "parameterizations" need to be included to account for wind stress, bottom friction, precipitation, cloud formation and so on. Many such parameterizations are only loosely based on experiment and can scarcely be described as "the laws of physics". Even the massive transference of energy during tropical thunderstorms occurs on a spatial scale too small to be resolved by the models and must be parameterised.

  2. The site also includes graphs designed to illustrate how well its model predictions of global quantities match the observations over time. What really matters, the acid test, is how well the models match the spatial distribution of such quantities. This detailed matching of prediction against observation is known as "hindcasting". Hindcast testing is rarely published by climate modellers. When it is, we see just how poor the models really are. Predictions rarely match observations. Hindcasts generally fail abysmally. Areas predicted to be warmer by the models are cooler in reality and vice-versa. The models don't work. Maybe the models are indeed "thoroughly tested". The CSIRO web pages fail to mention that they generally fail the tests.

When I was with CSIRO we developed models for the offshore oil and gas industry. Any models this bad would have been summarily rejected by our client.
  3. Why then do the model predictions of average quantities look so good? The answer lies in the parameterizations. Parameters are adjusted to make the model fit the averaged data. One such parameter controls "water vapour feedback". This has to be made positive in order for the models to simulate even the modest warming which occurred in the last 3 decades of the 20th century. Some experimenters dispute the validity of this assumption and claim that satellite radiometer data shows water vapour feedback to be negative. The editor of the journal "Remote Sensing" had to resign recently after publishing an experimental paper supporting this view. It remains a contentious issue.