Climate change scientists have models coming out of their ears. (Not the leggy variety, alas, but complex mathematical equations which try to project future temperature change.)
So how does a boffin decide which ones to use? Many scientists place their trust in models which accurately mimic past climate change, in the hope that they will continue to 'tell the truth' in the future.
But a new study published in the journal Geophysical Research Letters shows that this trust could be misplaced. The ability of a model to faithfully simulate past climate change is 'no guarantee of future skill', according to research by Catherine Reifen and Ralf Toumi at Imperial College .
Reifen and Toumi played around with the models of temperature change in the 20th century. To their surprise, they found that models which painted a faithful picture of climate change between 1900 and 1919 failed miserably between 1920 and 1939. The same applied between 1901-1920 and 1921-1940, and so on up until 1999.
Why's that? Simply put, models aren't very good at multi-tasking. A model may be very good at simulating the effect of El Nino on temperatures in the tropics, but isn't as good at everything else. But because El Nino and other 'strengths' like sea-ice come and go, no single model can consistently hit the nail on the head all the time.
That's why it's important that we include as many models as possible in our climate change projections, the study concludes. 'We do not know which feedbacks will dominate in the future', warn Reifen and Toumi, 'and the inclusion of the largest possible number of models could increase the range of predictions.'
So where does this leave us? You'll be glad to hear that the IPCC hasn't fallen into the trap flagged up by this report: it gives equal weighting to all models.
Less pleasing is its own admission of fallibility on the subject of models. 'What does the accuracy of a climate model's simulation of past or contemporary climate say about the accuracy of its projections of climate change?' mulls the IPCC in its Fourth Assessment. 'This question is just beginning to be addressed...'
Follow up: 'Unpredictable weather: why the climate is not a model citizen'