Hot models

SCIENCE & TECHNOLOGY

Climate change

Jan 27th 2005
From The Economist print edition

How to model the climate on the cheap

THIS month saw the start of a period when climate change will be in the news a lot. On January 17th Chris Landsea, a hurricanes specialist, withdrew from the Intergovernmental Panel on Climate Change, accusing it of “having become politicised”. On January 25th a report by the International Climate Change Taskforce called for drastic actions to cut carbon dioxide (CO2) emissions. On February 1st a conference on the subject will be held at Britain's Hadley climate centre. And on February 16th the Kyoto protocol comes into force. You have been warned.

As part of the fun, there was also the publication of an intriguing paper in Nature. The research was intriguing not so much for what it said (things may get hotter than expected), but for how it arrived at that conclusion. For it was one of the first serious scientific studies to exploit the idea of distributed computing.

Modelling the climate requires a huge amount of computing power, which is one reason the models have not been tested as thoroughly as they might have been. So David Stainforth of Oxford University and his colleagues decided to employ the idle capacity of 95,000 private desk-top computers to do it for very little money. Willing computer owners registered at a website (climateprediction.net), downloaded the appropriate software and then used their machines as normal. This enabled Dr Stainforth to test what is known as the Met Office Unified Model with 2,000 different sets of starting parameters.

Dr Stainforth wanted to explore the consequences of doubling CO2 levels from their pre-industrial values (at the moment they are about 1.6 times what they were at the beginning of the industrial revolution). What he observed, depending on the values of the parameters (such things as average cloud cover) was a range of increases between 2°C and 11°C, which is far greater than the current consensus of 2-3°C.

That is not as alarming as it sounds since, by his own admission, the model Dr Stainforth has been testing is crude. It dumps all its CO2 into the atmosphere in one go, instead of leaking it in over the years. And its description of the interaction between atmosphere and ocean is far too simple. But it does point the way towards a better way of doing the modelling business. And a cheaper one, too.

Copyright © 2005 The Economist Newspaper and The Economist Group. All rights reserved.