Published online 25 October 2007 | Nature | doi:10.1038/news.2007.198


Climate sensitivity 'inherently unpredictable'

Knowing exactly how the planet will respond to greenhouse gases isn't necessary, experts say.

It's impossible to perfectly predict the planet's future temperature: but perhaps we don't need to.GETTY

Climate models might be improving but they will never be able to tell us exactly what to expect. That's the conclusion of experts from the University of Washington, Seattle, who have set out to prove that predicting the exact level of climate change is by its very nature an uncertain science.

Over the past 30 years, climate models have not appreciably narrowed down the precise relationship between greenhouse gases and the planet's temperature — despite huge advances in computing power, climate observations and the number of scientists studying the problem, say Gerard Roe and Marcia Baker. The researchers now argue that this is because the uncertainty simply cannot be reduced.

They and other climatologists are now calling on policy-makers to make decisive policies on avoiding dangerous climate change, even if we don't have perfect models. This means focusing on keeping the planet's temperature below a certain point (and being willing and able to adjust emissions targets to achieve that), rather than trying to work out far in advance the amount of greenhouse gases in the atmosphere that will produce that level of warming.

Sensitive subject

Most models predict that the world's 'climate sensitivity' — the increase in global mean surface temperature expected to result from a doubling of atmospheric carbon dioxide relative to preindustrial levels — is probably somewhere between 2 °C and 4.5 °C.

But this is what climate models are generally good at: describing the most probable, modest temperature responses. Working out the possible extreme responses — which are less likely — is much more difficult, Roe and Baker point out.

The graph of potential temperature responses to rising greenhouse-gas levels shows a cluster of probable outcomes at the bottom of the scale, but a 'long tail' of possible extreme outcomes.


Roe and Baker's analysis, described in this week's Science1, shows mathematically that this tail is unlikely to get any smaller, regardless of how much research is done. This comes down to the fact that, if extreme events are set in motion, the outcome is necessarily difficult to predict from today's data. Put simply, if average temperatures rise by 4 °C, the world will be so different from today's conditions that it will be hard to work out what might happen next.

"We have to accept that that's the fundamental nature of the beast," says Roe. As a result, he adds, "it's very hard to distinguish between [the probability of] 6 °C warming and 10 °C warming."

Off target

Politicians are therefore missing the point in attempting to draft policy targets based on a specific level of greenhouse gases in our skies, says Myles Allen, a climatologist at the University of Oxford, UK. "That's a daft policy," he says.

Many climate analysts point to an overall greenhouse-gas concentration equivalent to 450 parts per million of carbon dioxide as the limit at which levels should be stabilized. This is widely predicted to avoid the 2 °C warming thought to be the general threshold for dangerous climate effects. But again, that figure is very uncertain.

Allen says that politicians should focus on the temperature target, and adjust emissions targets accordingly as time passes, rather than committing to a firm emissions target now. "We need an explicit commitment to do what it takes to avoid 2 °C of warming," Allen argues. "You don't need to know the climate sensitivity to avoid dangerous climate change," he adds. 

  • References

    1. Roe, G. H. & Baker, M. B. Science 318, 629-632 (2007). | Article |
Commenting is now closed.