In a Science paper last year (subscription required), Rahmstorf et al. pointed to 2001-2006 measurements of global temperature at the top end of the IPCC’s 2001 projections – and global sea level rise well beyond the range predicted in 2001 – as evidence that “the climate system, in particular sea level, may be responding more quickly to climate change than our current generation of models indicates.” Today in a letter to Nature Geoscience (subscription required), Roger Pielke, Jr, questions whether models from that 2001 generation improve on the predictive power of their forbears.

Pielke checks predictions from all four IPCC reports, dating back to 1990, against reality. Each report made a series of ‘if-then’ statements about the likely results of various emissions scenarios; in hindsight, Pielke can pick out which of these possible greenhouse experiments has actually been running on Earth since 1990 and compare the results to the IPCC’s shifting hypotheses.

Whereas the 2001 projections undershot the observed temperatures and sea levels, the 1990 projections overshot them, he concludes. Projections of temperature and sea level fell substantially between the 1990 and 1995 IPCC reports, when aerosols were added to models and carbon-cycle simulations were tweaked. But because they dropped too far, the adjusted post-1995 projections “are not obviously superior in capturing climate evolution”, says Pielke.

That’s not to say that 2001 models were no better than those a decade older. Including more information has made recent simulations more sophisticated – but so far it hasn’t much improved their ability to sketch out future climates, probably because important factors are still missing. Predictions from the two most recent reports do, however, seem to have crept toward the actual climate evolution, and additional rounds of of refinements may help the models to home in further.

Anna Barnett