There is a new paper in GRL, Does the Last Glacial Maximum constrain climate sensitivity? (subs) by Crucifix. Now the assumption of linearity in sensitivity was part of JA/JH’s work on constraining climate sensitivity, this may be a partial challenge (note that JH is thanked in the paper, so presumably they know about this stuff).
Figure 1 shows that whilst the LGM response is about the same in all the models, the 2*CO2 response varies widely. Figure 2 breaks it down by region; for Antarctica there is a nice linear relation (OK, between only 4 models (ahem, updae: a reader points out that the relation is actually inverse; I got confused by the y scale…)); for Greenland its a blob; for the tropical oceans its rather like fig 1. Crucifix concludes Therefore, climate sensitivity cannot be easily estimated from the Last Glacial Maximum global temperature.
If the climate model results were indeed physically accurate realisations of LGM and 2*CO2, then that would indeed be true. But of course they can’t all be – at best, only one can be, since they disagree. So this seems to be a case of taking the climate models too seriously? The conclusion quoted above would be true if you added “from GCMs” into it. Clearly this shows that in the models, different processes operate to give different sensitivities for the two cases. Which does indeed suggest that the same could well be true in reality. It ends up: careful model-data comparisons on the details of the spatial distribution of changes in temperature and precipitation at the LGM are needed to identify the ”best models”, that is, those that reliably predict the response of climate dynamics to a given forcing. And to warm RP Sr’s heart: Global temperature is not sufficient.