My query to Norway Research Council (@forskningsradet) on research finding weak warming impact from greenhouse gas buildup:
Magne Aldrin1,2,*, Marit Holden1, Peter Guttorp1,3, Ragnhild Bieltvedt Skeie4,Gunnar Myhre4, Terje Koren Berntsen4,5
Thanks for a prompt reply.
Reto Knutti of the Swiss Institute for Atmospheric and Climate Science, offered this valuable context:
As you said, Aldrin et al. (DOI: 10.1002/env.2140) is published, whereas Skeie et al. is not yet as far as I know. But here are some thoughts that are largely independent of this paper to put these types of studies in context.
If you look at the Fig. 3a in our review (http://www.iac.ethz.ch/people/knuttir/papers/knutti08natgeo.pdf, red lines at the top) you see that many previous estimates based on the observed warming/ocean heat uptake had a tendency to peak at values below 3°C (that review is from 2008). The Norwegian study is just another one of these studies looking at the global energy budget. The first ones go back more than a decade, so the idea is hardly new. The idea is always the same: if you assume a distribution for the observed warming, the ocean heat uptake, and the radiative forcing, then you can derive a distribution for climate sensitivity.
What is obvious is that including the data of the past few years pushes the estimates of climate sensitivity downward, because there was little warming over the past decade despite a larger greenhouse gas forcing. Also in some datasets the ocean warming in the top 700m is rather small, with very small uncertainties (Levitus GRL 2012), pushing the sensitivity down further. However, in my view one should be careful in over interpreting these results for several reasons:
a) the uncertainties in the assumed radiative forcings are still very large. Recently, Solomon et al. Science (2010, 2011) raised questions about the stratospheric water vapor and aerosol, and just days ago there was another paper arguing for a larger effect of black carbon (http://www.agu.org/news/press/pr_archives/2013/2013-01.shtml, a massive 280 pages…).
b) Results are sensitive to the data used, as shown by Libardoni and Forest DOI: 10.1029/2011GL049431 and others, and particularly sensitive to how the last decade of data is treated. Very different methods (detection attribution optimal fingerprint) have also shown that the last decade makes a difference (Gillett et al. 2011, doi:10.1029/2011GL050226).
c) The uncertainties in the ocean heat uptake may be underestimated by Levitus, and there are additional uncertainties regarding the role of deep ocean heat uptake (Meehl et al. 2011 Nature Climate Change).
Even though we have many of these studies (and I am responsible for a couple of them) I’m getting more and more nervous about them, because they are so sensitive to the climate model, the prior distributions, the forcing, the ocean data, the error model, etc. The reason for this, to a large extent, is that the data constraint is weak, so the outcome (posterior) is dominated by what you put in (prior).
It is important to note that the IPCC assessment of climate sensitivity is based on many lines of evidence (see Fig. 3 in our 2008 review for an overview). The observed energy budget is just one of them. The latest paleoclimate synthesis (Rohling et al. Nature 2012) supports the “likely 2-4.5°C”, and all GCMs have sensitivities in the range 2-5°C, the mean in CMIP5 is above 3°C, and once you start evaluating models with observations that tends to get pushed upward (Fasullo and Trenberth, Science 2012).
Finally, note that the effect of the last few years of data is smaller on the transient climate response than on climate sensitivity. It’s the transient climate response (TCR) that determines the 21st century warming and peak warming.