Today – after a long pause – a new Climate Dialogue has started about climate sensitivity and transient climate response. This of course is a highly relevant topic. Equilibrium climate sensitivity (ECS) and especially the transient climate response (TCR) largely determine how much warming we can expect this century.
Climate Dialogue is a platform for discussions between invited climate scientists on controversial climate topics. They represent a range of views. Discussions are strictly moderated and should be on topic. Last year we did four Climate Dialogues, about arctic sea ice, long-term persistence, regional climate models and the hot spot in the tropics.
We are very happy that a few key players in the debates surrounding climate sensitivity have agreed to participate in this Climate Dialogue. John Fasullo of NCAR is well-known for his publications with Kevin Trenberth about the role of the natural variability and the oceans in the recent hiatus. James Annan, who recently moved back to the UK after working for many years in Japan, has also published a lot about climate sensitivity and is also frequently blogging about it. Finally Nic Lewis is a relative newcomer in climate science. After a career in the financial sector he recently became a climate scientist, focusing on climate sensitivity. He published a few papers about climate sensitivity, one with a group of IPCC lead authors.
A Sensitive Matter
Some readers will remember that Nic Lewis and I published a GWPF report in March of this year with the title A Sensitive Matter: How the IPCC hid the good news on global warming. One of the main conclusions of our report was that observationally based estimates for ECS and TCR (based on the so-called instrumental period, say 1850-2014) are “superior” at the moment and are to be preferred over estimates from GCMs or paleorecords.
We claim that these “observational” estimates indicate best estimates for ECS that are close to the lower boundary of the likely range in several IPCC reports which is 1.5–4.5°C. We give an “observational” likely range of 1.25-3°C with a best estimate of 1.75°C. This is substantially lower than the average ECS value of the CMIP5 models which is more than 3°C.
IPCC gave a best estimate of 3°C in AR4 but none in AR5. This lack of a best estimate was only mentioned in a footnote in the SPM. Lewis and I wrote in our report that the discrepancy between the observationally based best estimate and the model based best estimate was likely the main reason why IPCC didn’t give a best estimate this time. So IPCC on the one hand says climate sensitivity will likely be anywhere between 1.5–4.5°C and no best value can be given. But when doing projecetions AR5 relies (also in the Working Group II and III reports) on climate models that have an average climate sensitivity of 3.2°C.
Discrepancy
So it is really important to find out where the discrepancy between the observationally based and model based estimates comes from. This will certainly be a crucial question in this Climate Dialogue.
Lewis in his guest blog repeated his likely range for ECS being lower than that of AR5:
The soundest observational evidence seems to point to a best estimate for ECS of about 1.7°C, with a ‘likely’ (17-83%) range of circa 1.2–3.0°C.
Annan agrees that the observational record points towards lower values for ECS and TCR. But nevertheless he comes up with slightly larger value for the best estimate than Lewis:
The recent transient warming (combined with ocean heat uptake and our knowledge of climate forcings) points towards a “moderate” value for the equilibrium sensitivity, and this is consistent with what we know from other analyses. Overall, I would find it hard to put a best estimate outside the range of 2-3°C.
Fasullo did not yet give his preferred likely range but he emphasized in his guest blog that lowering the lower boundary of the likely range in AR5 to 1.5°C (was 2°C in AR4) maybe was justified at the time but not any longer:
In short, I argue that although IPCC’s conservative and inclusive nature may have justified such a reduction at the time of their report, the evidence accumulated in recent years argues increasingly against such a change.
How Observational are the “observational” estimates?
A logical point of discussion is how “observational” the instrumental estimates for ECS and TCR really are. Both Annan and Fasullo are critical about this. Annan wrote:
Lewis’ four preferred observational studies are not as unimpeachable or model-free as he claims but based on a highly simplified model that imperfectly represents the climate system.
And Fasullo:
Uncertainty in observations and the need to disentangle the response of the system to CO2 from the convoluting influences of internal variability and responses to other forcings (aerosols, solar, etc) entails considerable uncertainty in ECS (Schwartz, 2012) and thus: 1) the use of a model is unavoidable, 2) it is a misnomer to present 20th Century instrumental approaches as being “observational estimates”.
Lewis’ reaction to these arguments was:
By contrast, in a properly designed observationally-based study, the best estimate for ECS is completely determined by the actual observations, as is normal in scientific experiments. To the extent that the model, simple or complex, used to relate those observations to ECS is inaccurate, or the observations themselves are, then so will the ECS estimate be. But, in any event, the ECS estimate will be far more closely related to observations than are GCM ECS values.
Hiatus
The hiatus will also play a role in this discussion. Fasullo suggested – referring to our report – that the hiatus was/is responsible for the relatively low estimates for climate sensitivity in the instrumental period:
The apparent slowdown of global surface warming has led some to conclude that evidence for lower climate sensitivity is “piling up” (11). Some have even argued that global warming has stopped.
Reference 11 is our GWPF report. Apparently Fasullo has not read our report very carefully because we make clear the hiatus has indeed some influence on estimates for TCR but not on ECS. Therefore in his first comment to Fasullo Lewis writes:
I believe John’s views of the impact of “hiatus” in global surface warming over the last circa 15 years on estimation of ECS are seriously mistaken. He starts by claiming that, based on simple models, the hypothesis that the hiatus argues for a reduction in the lower bound of the range for ECS was found sufficiently compelling that IPCC AR5 reduced the lower bound of its likely range for ECS. John cites Chapter 12 in that regard. But Box 12.2, which covers equilibrium climate sensitivity and transient climate response, does not even mention the slowdown in warming this century. It says (my emphasis):
Based on the combined evidence from observed climate change including the observed 20th century warming, climate models, feed¬back analysis and paleoclimate, ECS is likely in the range 1.5°C to 4.5°C with high confidence.
and goes on to say:
“The lower limit of the likely range of 1.5°C is less than the lower limit of 2°C in AR4. This change reflects the evidence from new studies of observed temperature change, using the extended records in atmosphere and ocean. These studies suggest a best fit to the observed surface and ocean warming for ECS values in the lower part of the likely range.”
Lewis later mentions the Aldrin and Otto paper that show higher estimates for ECS based on the most recent decade, not lower:
• Aldrin et al 2012 used data ending in 2007 for its main results ECS estimate but also presented an alternative estimate based on data ending in 2000. The median ECS estimate using data only up to 2000 was lower, not higher, than the main one using data to 2007. Moreover, their updated ECS estimate using data up to 2010, published in Figure 10.20b of AR5, had a higher median than that using data to 2007.
• The Otto et al 2013 median ECS estimate using 2000s data was the highest of all its ECS estimates; the ECS estimates using data from the 1970s, 1980s or 1990s were all lower.
There is a lot more to discover in the guest blogs and the first comments. As for obvious reasons I have a “conflict of interest” in this Dialogue I will not actively moderate the discussion. The moderation will be done by this year’s project leader Bart Strengers of PBL. Climate Dialogue is an initiative of two Dutch goverment-related climate institutes PBL Netherlands Environmental Assessment Agency and the Royal Netherlands Meteorological Institute (KNMI). Climate Dialogue is funded by the Dutch Ministry of Infrastructure and the Environment.