Published online 11 August 2004 | Nature | doi:10.1038/news040809-9


Climate predictions gain surer footing

Range-based model raises global warming estimate by a degree.

Climatologists predict an increase in the number of extreme weather events as temperatures rise.Climatologists predict an increase in the number of extreme weather events as temperatures rise.© Punchstock

Researchers who have devised a new approach to calculating global warming say they have reduced our uncertainty about the extent of warming to expect over the next 100 years.

The method, which predicts a temperature rise of at least 2.4°C over the next century, is not dependent on guessing the values of unknown factors. This should put climate modelling on a more solid footing and give policy makers a more rational basis for making decisions about preventing climate change and dealing with its consequences, says study leader James Murphy from the Hadley Centre for Climate Prediction and Research in Exeter.

“It is like having hundreds of different planets to study climate on”

Bruce Wielicki
NASA Langley Research Center

As carbon dioxide levels continue to rise, more heat becomes trapped within our atmosphere, causing the Earth's temperature to increase. But current models of global warming disagree on how sensitive the climate is to carbon dioxide, making it impossible to say exactly how much temperature will rise.

One big problem is a lack of precise knowledge about many of the factors that are plugged in to the models, such as the ability of clouds to reflect heat.

In the past, researchers have dealt with these unknowns by making an informed guess at the correct values to feed into their model. Instead of doing this, Murphy's team asked experts to provide a range of likely values for each unknown.

They then ran their model many times over, using different values for each unknown, to produce a range of predictions for the future. As a quality control test, they used each version of the model to predict current climate, and weeded out those versions that failed.

The main advantage of this approach is that the range of temperatures it predicts is not dependent on having guessed the unknown components correctly.

Bruce Wielicki, a climatologist at the NASA Langley Research Center likens it to having hundreds of different planets to study climate on. "An attack along these lines is indeed the future key to rigorous uncertainty in climate predictions," he says.

Near certainty?

The results, presented in Nature1, suggest that if carbon dioxide concentrations double over the next hundred years - as many believe they will - the planet will warm by between 2.4 and 5.4°C.

A previous estimate released by the Intergovernmental Panel on Climate Change (IPCC), predicted a 1.4 to 5.8°C range.

The new climate modelling approach takes into account unknowns in the planet's atmosphere.The new climate modelling approach takes into account unknowns in the planet's atmosphere.© NASA

Although it may not seem like much, the one-degree increase at the lower end of the scale is significant, says climatologist Richard Allan from Reading University. It could correspond to a significant rise in sea level and an increase in extreme weather events, both of which will now need to be planned for as a near certainty.

It also effectively dismisses the argument of sceptics who use the current uncertainty to argue for a 'wait and see approach' to reducing carbon dioxide emissions.

"What we've really done is just the first step," says Murphy. His team would now like to test all possible combinations of unknowns, but that would mean running over 4 million models, and they don't have the computing power to do that yet.

To achieve this, Murphy and his colleagues plan to take advantage of a project called, which distributes climate-modelling software that anyone can run on their computer when it's not busy. 

NASA Langley Research Center

  • References

    1. Murphy, et al. Nature, 430. 768 - 772 (2004). | Article |