All about models (no, not those models!)

As I have explained before,  the notion that global warming will be catastrophic for humanity unless we stop emissions of carbon dioxide rests in part on observations (though not recent ones) and mostly on the output of global circulation models (GCMs), giant collections of simultaneous equations which attempt to provide an accurate ‘model’ of the earth’s climate. The models rely on supercomputers, because there are so many variables and so many possibilities in employing them.

When the AGW scare began the warnings about what we were facing rested on the fact that temperatures at the time seemed to be going up in similar fashion to carbon dioxide accumulations. The models projected climate possibilities for the future given certain assumptions about what might happen (business as usual, reduced emissions, no more emissions, and so on). The possibilities were always dire, though on the face of it, you would expect that a warmer planet would benefit a lot of people, a lot of regions and a lot of nations, if not everyone.

The projections became predictions, if a government minister or an AGW believer were speaking, and the predictions led to knee-jerk reactions from governments over possibilities like a dramatic rise in sea levels, about which I have also written before. The output of models began to be called ‘evidence’ (as in ‘the evidence from the models’), though models don’t provide evidence at all. In vain did sceptics point out that models need to be verified against reality before you can take much notion of their output, or that  climate, as a chaotic non-linear phenomenon, probably can’t be modelled accurately anyway. That the models were thought to work was a necessary basis for carbon taxes and similar policies, and it was convenient for the policymakers to gloss over the awkward problems associated with the models.

The most awkward of them so far is that the models did not predict the pause in warming, now 15 or more years long, and it is their failure to do so that has produced such interest in the forthcoming IPCC Fifth Assessment Report (AR5), which will have to say something about it! Incidentally, what we will see in  a few days is not the report of the working group on the science, but the Summary for Policy Makers (SPM), which is a political rendition of the science. My guess is that the SPM will gloss over it, and pretend that nothing much has happened.

Back in argument-land, however, a lot has been happening. Papers keep coming out that worry over the failure. Bjorn Lomborg points out the arithmetic in a recent post on WUWT.  Another recent one, by a group of Canadian researchers, was published in Nature Climate Change, which is not in any sense a sceptical bible. ‘Recent observed global warming,’ the authors say, ‘is significantly less than that simulated by climate models. This difference might be explained by some combination of errors in external forcing, model response and internal climate variability.’

The observed global temperature anomaly over the twenty years from 1993 to 2012 rose by 0.14 plus or minus about 0.06 degrees Celsius, which was a good deal slower than a set of models that are part of a large model inter-comparison study, even though these models are said to simulate natural variability well. The models said the earth should have warmed by about 0.3 degrees C. If you take only the last fifteen years, the failure of the models is even more striking, the simulated warming being four times the observed warming. The authors say that ‘the current generation of climate models (when run as a group, with the CMIP5 prescribed forcings) do not reproduce the observed global warming over the past 20 years, or the slowdown in global warming over the past fifteen years’.

Why don’t they? What is wrong? The authors didn’t set out to answer that question, but offer a number of possibilities, whose cumulative message is that there is really a lot we don’t know about how the earth’s climate actually works. Oh, the authors do offer, in less than a sentence, another possibility: ‘the transient climate sensitivity of the CMIP5 models could be on average too high’.

Oh dear, that’s an unforgivable thing to say. No one disagrees with the notion that a doubling of carbon dioxide in the atmosphere would produce an average increase in temperature of about 1.1 degree C. But it is IPCC dogma that climate sensitivity is the real problem for us humans: water vapour and clouds multiply the ordinary warming effect of carbon dioxide by three, four or even five  times (or much more, for those who delight in warning us about the seas boiling…). As it happens, there are half a dozen recent papers that suggest that climate sensitivity is really quite low.

And only a little while ago came an acceptance by the US National Academy of Sciences, in a report about the need to sustain advances in climate modelling, that climate models aren’t yet ready for the use some would like to make of them:

Overall, climate modeling has made enormous progress in the past several decades, but meeting the information needs of users will require further advances in the coming decades… 

The fundamental science of greenhouse gas-induced climate change is simple and compelling. However, genuine and important uncertainties remain (e.g., the response of clouds, ecosystems, and the polar regions) and need to be considered in developing scientifically based strategies for societal response to climate change.

Yes. Some people have been saying the same for many years.