Models of various kinds are routinely used by scientists either as a first stage of work to prove a concept or because experimenting directly on the main system of interest is impractical, impossible or unethical. So, animal models are used to develop and test medicines: strains of mice particularly susceptible to a particular disease may be used to screen a promising candidate molecule for effectiveness before developing it further and progressing to clinical trials.
Plant scientists do something similar. If they want to show that a particular trait can be controlled by modifying certain genes, the first thing is often to demonstrate it in the simple brassica species Arabidopsis. This has a relatively simple genome and is easily transformed. If a transgenic trait works in this model, it may be worth progressing to a plant of commercial value.
Economists also use models. But the fact that the projections of the dismal scientists are so often wrong demonstrates a very important point: that models only show what would happen if a particular set of circumstances and input data prevails. Change something and the result is different.
The same caution has to be taken with the biological sciences. A new pharmaceutical may give very promising results in mice with a particular genetic makeup, but could be ineffective (or dangerous) in humans. Various failed attempts at gene therapy illustrate this all too clearly. Tweaking genes in Arabidopsis may be fine, but unexpected problems may come up with maize or soy.
In these cases, the models are a guide to what is likely to happen. For medical researchers and plant scientists, the reality can be tested more fully and the finding confirmed or found to be wanting. Governments cannot do real-life experiments on the basis of economic advice except by changing subsidies or other incentives, or changing tax rates. Central banks have the option of inducing widespread trends in domestic economies by raising or lowering interest rates. Citizens, however, would prefer not to be the subject of experiments, particularly if the advice to government turns out to have been wrong.
But there is one area of science in which the line between modelling and reality has become blurred. Climate scientists, who have such a big influence on energy policy and the economy more broadly, can do nothing more than observe, hypothesise and model; this is one area where experiments are impossible.
And, to make matters worse, here are two major complications. First, the mathematical models are based on an acknowledged incomplete understanding of the intricacies of the global climate. Clouds are a particular problem in that the different types in different parts of the atmosphere have to be treated differently. Clouds are highly variable, move around quickly and have quite different effects on flows of radiation depending on their type.
The second complication is the sheer scale of computing power needed. Global climate models (so-called General Circulation Models are currently in favour) divide the Earth’s atmosphere into cubes. Calculations have to be done over and over again for each cube to build up a picture of overall patterns. This is what some of the world’s most advanced supercomputers are being used for, but even these are only capable of producing a very ‘grainy’ image of circulatory changes.
This will change, of course, as Moore’s Law continues to operate, but even finer-grained pictures will still be of an incompletely-understood system. We know there are various long-term patterns, including the El Niño/Southern Oscillation (ENSO) the Pacific Decadal Oscillation (PDO) and the North Atlantic Oscillation (NAO) but we certainly do not know cause and effect. Neither can we predict the behaviour of the Jet Stream which can make so much difference to Western European summers.
Into this highly complex but flawed system are fed ranges of parameters and the output is a range of projections of future temperatures, rainfall patterns etc. But at the heart of this is a key assumption about the influence of the trace gas, carbon dioxide (still forming less than 0.05% of the atmosphere, despite a large percentage increase over the past century or so). Spectroscopic measurements of the CO2 molecule’s absorbance of certain frequencies of infra-red radiation lead to the generally accepted expectation that higher levels of carbon dioxide will raise average temperatures modestly.
This effect should be, to an extent, self-limiting, since the relationship between temperature and concentration is a logarithmic one; that is, a doubling from 275 to 550 parts per million would be expected to raise average temperatures by about 1°C. To increase temperatures by another degree would mean a further doubling of carbon dioxide concentration to 1100ppm, and so on. In other words, is this is the only long-term impact of more carbon dioxide, keeping below a 2° temperature rise, which has arbitrarily been set as the ‘danger’ level would be quite easy, since it is inconceivable that we would continue to burn fossil fuels for long enough to do this.
However, all the climate models used to project changes over the rest of the century and beyond are based on the assumption that there is in fact a positive feedback: additional CO2 raises temperatures, which increases the moisture content of the air and allows yet more CO2 to come out of the oceans. The net effect could be perhaps a three-fold increase in warming.
This positive feedback is included because this is believed by the modellers to be the only way to account for the pattern of temperature change occurring in the 20th Century. In practice, various fudge factors have had to be included for models to produce the actual pattern, including an assumption of atmospheric aerosols causing a plateau or slight fall from the 1940s to the mid ’70s.
Now, we find ourselves in another phase which the models did not reproduce: a lack of any further significant warming since 1998. There have been numerous attempts to explain this, generally involving the extra heat being stored in the deep ocean but, again, these are all hypotheses which have not been verified.
It is important to realise that pronouncements on the ‘safe’ level of carbon dioxide and the current highly expensive but ineffective policy on decarbonisation are based only on these flawed models. In the absence of an effective ‘no regrets’ policy response (actually, nuclear fits the bill but has not been given enough priority) the correct action is to work hard to increase the understanding of global climate and its drivers and develop effective options for dealing with change. To do otherwise runs the risk of taking the international community down a policy blind alley.
Martin Livermore
The Scientific Alliance
St John’s Innovation Centre
Cowley Road
Cambridge CB4 0WS
_______________________________________________________________