1. The scientific part (WGI) of the fifth IPCC assessment report (AR5), published in final form in January 2014, contains some really encouraging information.1 The best observational evidence indicates our climate is considerably less sensitive to greenhouse gases than climate scientists had previously thought. The clues and the relevant scientific papers are all mentioned in the full IPCC report. However, this important conclusion is not drawn in the full report—it is only mentioned as a possibility—and is ignored in the Summary for Policymakers (SPM).
2. Until AR5, for 30 years the scientific establishment’s best estimate and their uncertainty range for climate sensitivity had hardly changed. The best estimate for equilibrium climate sensitivity (ECS) started and ended at 3oC and the uncertainty range2 generally had a lower bound of 1.5oC and an upper bound of 4.5oC.3 However, several recent studies give best estimates of between 1.5oC and 2oC, substantially lower than most earlier studies indicated.
3. In the IPCC Fourth Assessment Report (AR4), the empirical estimates of climate sensitivity were largely based not only on data that has now been superseded, but also on an inappropriate statistical basis that biased them towards higher values, thus making the global warming problem appear ‘worse’. In AR5, many studies still use inappropriate data and/or statistical methodology. However, there is now a body of empirical estimates of climate sensitivity, prepared using sound methodology and appropriate data, that give substantially lower values—both of long-term warming and of transient warming towards the end of this century—than climate model simulations.
4. Since the last IPCC report was prepared greenhouse gas concentrations have continued to increase, yet global temperatures have not risen; more importantly, estimates of the cooling efficacy of aerosol pollution have been cut. This combination of factors is indicative of the climate system being less sensitive to greenhouse gases than previously appeared to be the case. But the new evidence about aerosol cooling is not reflected in the computer climate models.
5. Global climate models used to predict future climate change still generate model climate sensitivities in the range 2–4.5oC, averaging just over 3oC. Large parts of the IPCC reports are built around the computer model simulations. Almost all the projections of future climate change are based on them,4 and a complete chapter is devoted to model performance. Admitting in the IPCC report that the best observationally-based estimates5 of climate sensitivity are now only 1.5–2oC would imply that large parts of the AR5 report are out of line with the latest scientific evidence.
6. In our view, the IPCC WGI scientists were saddled with a dilemma. How should they deal with the discrepancy between climate sensitivity estimates based on models and sound observational estimates that are consistent with the new evidence about aerosol cooling? In conjunction with governments—who have the last say on the wording of the SPM—they appear to have decided to resolve this dilemma in the following way. First, they changed the ‘likely’ range for climate sensitivity slightly. It was 2–4.5oC in AR4 in 2007. They have now reduced the lower bound to 1.5oC, making the range 1.5–4.5oC. By doing this they went some way to reflect the new, lower estimates that have been published recently in the literature.
7. They also decided not to give a best estimate for climate sensitivity. The tradition of giving a best estimate for climate sensitivity goes all the way back to the Charney report in 1979, and all subsequent IPCC reports (except the third assessment report in 2001) gave one as well. In AR4 the best estimate was 3oC. At the time of approval of the SPM by governments in September 2013, the decision not to give a best estimate for climate sensitivity was mentioned only in a footnote in the SPM, citing ‘a lack of agreement on values across assessed lines of evidence and studies’. Only in the final report, published in January 2014, was a paragraph added in the Technical Summary giving slightly more explanation.
8. At a minimum, the SPM should have given a more informative explanation of the decision to widen the ECS ‘likely’ range and not give any best estimate for ECS. That could have taken the form of a straightforward statement that the best-quality observational evidence, based on improved estimates of the effects of aerosol pollution and the extended record of warming now available, points to a best estimate for ECS of 2oC or slightly less, while evidence from global climate models still suggests that it is about 3oC or slightly more. We—the authors of this report—were both expert reviewers of AR5 and in our review comments suggested that the IPCC should go further and give separate ranges for climate sensitivity based on models and on high quality observational studies.
9. In this report we suggest that the new observationally-based ‘likely’ range could be 1.25–3.0oC, with a best estimate of 1.75oC.6 If the IPCC had made that change—which would have been in line with the best quality scientific evidence available—it would have been picked up by all the major news outlets in the world as one of the major, if not the major, outcomes of the report. And rightly so.
10. In AR5 the IPCC felt even more certain (95% certain, compared to 90% in AR4) that humans have caused most (more than 50%) of the warming since 1950. The media treated this as the major conclusion of AR5, but it is in fact a relatively trivial finding. The high-quality observationally-based estimates for climate sensitivity discussed in this report assume that virtually all the measured warming (not just since 1950, but over the last 100â€“150 years) is due to humans. The far more important question now is how much warming is likely in the future under various scenarios.
11. Transient climate response (TCR), a measure of warming from a doubling of carbon dioxide (CO2) over a seventy-year period, reflects ocean heat uptake efficiency as well as climate sensitivity and is often seen as providing a better guide to warming over the twenty-first century than ECS.7 AR5 lowers the 10–90% range for TCR of 1–3oC established in AR4 to a ‘likely’ range of 1–2.5oC. In this report, we suggest that an observationally-based ‘likely’ range for TCR could reasonably be 1–2oC, with a best estimate of 1.35oC. The average TCR for global climate models is much higher, at just under 2oC.
12. These lower, observationally-based estimates for climate sensitivity and TCR suggest that considerably less warming and sea level rise is to be expected in the future than the model projections imply. Projected future warming based on the best observationally-based estimate of TCR is 40–50% lower than the IPCC’s model-based projected warming, and on the IPCC’s second highest emissions scenario cumulative warming would still be around the international target of 2oC in 2081–2100.
13. Our criticisms are directed at the IPCC as an organisation,8 on the constraints its process imposes, and on the excessive emphasis put on projections and other results derived from climate models. The scientists’ hands were largely tied; the scopes and even titles of the various chapters had already been determined. Even discriminating between models would have been awkward politically.
14. The purpose of the IPCC is ‘to assess on a comprehensive, objective, open and transparent basis the scientific, technical and socio-economic information relevant to understanding the scientific basis of risk of human-induced climate change’.9 We believe that, due largely to the constraints the climate model-orientated IPCC process imposed, the WGI report and the SPM failed to reflect satisfactorily such an assessment in the case of climate sensitivity and TCR, arguably the most important parameters in the climate discussion.
1The accepted final draft of the AR5 Working Group I report and the approved version of the Summary for Policymakers (SPM) were published in September 2013. Corrected final versions of the SPM and the full AR5 WGI report were released in January 2014.
2‘Likely’, defined as the central two-thirds probability in the last two IPCC reports; until then it was not
3The fourth IPCC assessment report, published in 2007, increased the lower bound to 2oC.
4Projected warming increases less than proportionally with ECS due to the moderating effect of heat uptake by the ocean. Projected warming in the models could conceivably be in line with observational evidence despite their ECS not being so. But it is not.
5Observationally-based methods do involve some limited use of models, but the ways they are used to help derive climate sensitivity estimates from observations differ greatly from the way global climate models are used to produce sensitivity estimates.
6This is based on giving precedence to high-quality estimates that use a long period of instrumental temperature data, in line with AR5’s appraisal of the different types of estimate, and discounting studies with identified substantial failings.
7However, sea level response depends more on the relationship between ECS and TCR than on TCR itself.
8The IPCC is not a research organisation, but its assessment report process significantly influences research carried out by climate scientists, in particular that involving simulations by climate models.