I wanted to briefly highlight a new paper by Nic Lewis and Peter Grünwald called [o]bjectively combining AR5 instrumental period and paleoclimate climate sensitivity evidence. You may want, however, to be cautious of the term objective. As the title indicates, they essentially combine instrumental and paleo estimates for equilibrium climate sensitivity (ECS). However, some care should be taken with this result, since instrumental estimates are essentially effective climate sensitivity, rather than equilibrium climate sensitivity, but I still think this is a really nice thing to have done.
Their headline result basically
provide[s] a 5–95% range for climate sensitivity of 1.1–4.05 K (median 1.87 K),
which is similar to the earlier Lewis and Curry results, but with a slightly higher median.
However, if one considers the figure below, there do seem to be number of slightly different results, depending on the method used. All seem to produce 5-95% ranges of about 1-4K, but the median seems to vary from just below 2K, to just above 2K.
Roughly speaking, this result seems broadly consistent with the IPCC range of 1.5-4.5K (although that might be a 17-83%, rather than a 5-95%, range), with a slight shift to lower values and a lower median than might be expected. A few general comments, though. Unless clouds happen to provide quite a strong negative forcing, there are robust physical arguments as to why ECS > 2K. There are also indications that the cloud feedback is probably positive. There are also reasons to be slightly cautious about some of these Bayesian estimates of climate sensitivity, which we discussed in some more detail in this post (in fact the priors shown in Figure 3 in the paper seem to highlight what we discussed about Nic Lewis’s priors).
However, I do think that this new paper is a nice extension of the work Nic Lewis has done before. What would be nice would be to maybe include more physics so as to exclude regions of parameter space that we regard virtually impossible (for example, ECS values below 1K). My suspicion is that doing so would shift the distribution to slightly higher values and would make the result even more consistent with the IPCC estimates.