By Andy May
Christian Freuer has translated this post to German here.
In part one we discussed various estimates of climate sensitivity (ECS, TCR, and observation-based values) and what they mean, especially those reported in the latest IPCC report, AR6. In part 2 we discussed the uncertainty in estimating cloud feedback to surface warming, and cloud feedback’s relationship with ECS. In this part we compare the values from various sources to one another.
AR4, AR5, and AR6 define preindustrial as before 1750 or when the CO2 atmospheric concentration was about 280 ppm. This is just after the worst part of the Little Ice Age. AR6 estimates the total anthropogenic forcing from 1750 to 2019 to be 2.72 W/m2, a 19% increase over AR5’s estimate (AR6, p 926). AR6 also changes the estimate of ECS, both ECS estimates are compared to other estimates in Table 1.
In AR6, the IPCC states that:
“… the best estimate of ECS is 3°C, the likely range is 2.5 to 4°C and the very likely range is 2 to 5°C. It is virtually certain that ECS is larger than 1.5°C.”(AR6, p 926)
They are virtually certain that ECS is greater than 1.5°C/2xCO2. Yet, the peer-reviewed literature contains numerous lower estimates of climate sensitivity to CO2, based on observations, as admitted in AR6 on page 1007. Six lower estimates are listed, in bold, in Table 1. The IPCC does not independently estimate ECS, they gather peer-reviewed estimates made by others and use their best judgement to derive a most likely value and a range of possible values. They appear to have ignored many peer-reviewed observation-based lower estimates of climate sensitivity. Many estimates, far too numerous to list here, show possible values below 1.5°C/2xCO2.
One reason they give for their new ECS higher range and estimate is they believe the “feedback parameter increases as temperature increases.” Feedbacks on top of feedbacks. Thus, they have created an endlessly changing model framework for their calculation, making an already untestable hypothesis even more untestable. When building a computer model, it is never a good idea to use the primary target calculation, in this case surface temperature, drive the model structure or the target feedbacks. This is the computer equivalent of circular reasoning.
The background climate state does change and there is no doubt that feedbacks will have a different effect when the climate state changes. However, AR6 focusses on the temperature-dependence of feedbacks without showing how the climate state changes when temperature changes. Javier Vinós has shown that climate state changes are possibly related to changes in solar activity and major ocean internal oscillations. Thus, it is possible that a changing climate state causes the feedback and temperature changes, and not the other way around. AR6 may have confused cause and effect.
ECS is an artificial model construct with little meaning outside the climate model world. An instantaneous or nearly instantaneous CO2 doubling is unlikely to occur, and it would take hundreds, perhaps thousands, of years for the full ECS temperature response to work through the climate system. It is extremely unlikely that other factors affecting climate would stay in equilibrium that long.
To make matters worse, the models used to calculate ECS are not consistent. Some calculations use a full atmosphere-ocean model and some use observed ocean temperatures. Some simple models construct an energy balance based only upon surface temperature, these are called zero-dimensional models, other simple models add additional zones, or complexities. It is widely recognized that ECS is unreal and as a result some have redefined it as “effective climate sensitivity” as previously discussed in part 1. But this is still unreal, untestable, and not scientific, as defined by Karl Popper. Further it only affects humanity 150, or more, years in the future, a meaningless time frame to consider today.
The CO2 climate sensitivity estimates listed in bold in the bottom six rows of Table 1, are not directly comparable to the IPCC model-based estimates, because they are based on real world observations. These six estimates use data collected over periods of less than 100 years and the CO2 increases occurred over time.
Nicola Scafetta offers a more comprehensive look at the AR6 model ECS estimates. Scafetta shows that AR6 ECS calculations from models range from 1.83 to 5.67°C/2xCO2. He found that all the models with an ECS above 3°C/2xCO2 run very hot relative to observations and should be discounted. Scafetta found that the models that had excess warming (over observations) of less than 0.2°C in 50% or more of their grid cells, were those with an ECS less than 2°C/2xCO2. Further, these are the only models that can be considered statistically valid. Scafetta and many other climate researchers have shown that an ECS between one and two °C/2xCO2 fits observations best, higher values are not supported by observations.
As already mentioned, AR6 relies very heavily on the flawed analysis of Sherwood, et al. The AR6 estimate of ECS, shown in Table 1, is like Sherwood’s, which is about 3.2°C (5-95% range 2.3 – 4.7°C). Using the same data as Sherwood, but using a more objective set of criteria, and fixing some errors in Sherwood’s statistical techniques, Nic Lewis lowers Sherwood’s estimate of climate sensitivity to 2.2°C, from 3.2°C, and finds that values below 2°C have a 36% probability, higher than the probability of climate sensitivity exceeding 2.5°C.
TCR (the Transient Climate Response) is the short-term—roughly 70 years—change in temperature due to a sustained 1%/year increase in CO2 to the point where the CO2 concentration doubles. While TCR is still an artificial construct, it plays out in 70 years and can be checked and potentially falsified. It is both more relevant and scientific. In this discussion, we will ignore the unreal and untestable ECS, whether the “E” stands for equilibrium or effective. Table 2 compares various estimates of TCR to our empirical, observation-based estimates of climate sensitivity in the real world.
The IPCC values of TCR in Table 2 are closer to the measured estimates shown in bold, but still too high. AR6 has this to say about their estimate of TCR:
“… the best estimate of Transient Climate Response (TCR) is 1.8°C, the likely range is 1.4 to 2.2°C and the very likely range is 1.2 to 2.4 °C.”(AR6, p 927).
AR6 on estimates based upon the historical record:
“Global energy budget constraints indicate a best estimate (median) value of TCR of 1.9°C … and very likely in the range 1.3°C to 2.7°C (high confidence).”(AR6, p 999)
Their overall assessment is a little smaller than their estimate from the historical record, but higher than the observation-based estimates we cite in Tables 1 and 2. Clearly, they are cherry picking the data they use. To set the lower bound of their “very likely” range above the six or seven observation-based estimates in Tables 1 and 2 is disingenuous.
AR6 do discuss Nic Lewis and Judith Curry’s 2018 paper, which has a lower bound below 1°C/2xCO2, and similar estimates by Ragnhild Skeie and colleagues, and Alexander Otto and colleagues. Christy and McNider’s 2017 estimate of TCR is completely ignored. AR6 dismisses these lower estimates because the studies necessarily assume radiative feedbacks will remain constant as CO2 causes the atmosphere to warm, at least with respect to ECS. The assumption of constant radiative feedback has a smaller effect on observation-based estimates of TCR. This refers to the IPCC speculative positive feedbacks to feedbacks idea introduced in AR6 as discussed above. They have high confidence that the feedbacks will increase as temperature rises, which will cause additional warming, this confidence comes primarily from model studies. Obviously, observation-based studies must assume that the feedbacks are constant over the period studied. AR6 assumes that climate state changes are a result of temperature changes, that is they are a temperature feedback, and ignores the very real possibility that the temperature changes are due to climate state changes.
Positive feedbacks to feedbacks
The IPCC AR6 models do not predict historical SST warming very well. Depending upon the area, sometimes the models overpredict warming and sometimes they underpredict it. It seems their logic is that the models cannot be wrong, so they assume the temperature feedback values must be changing. They try and explain their logic on pages 989 to 997. Their explanation reminds us of this passage from Karl Popper’s book,
“The Marxist theory of history, in spite of the serious efforts of some of its founders and followers, ultimately adopted [a] soothsaying practice. In some of its earlier formulations (for example in Marx’s analysis of the character of the ‘coming social revolution’) their predictions were testable, and in fact falsified. Yet instead of accepting the refutations the followers of Marx re-interpreted both the theory and the evidence in order to make them agree. In this way they rescued the theory from refutation; but they did so at the price of adopting a device which made it irrefutable. They thus gave a ‘conventionalist twist’ to the theory; and by this stratagem they destroyed its much-advertised claim to scientific status.”(Popper, 1962, p. 37).
The detailed description (section 18.104.22.168, page 989) of their positive feedback to feedbacks idea, is based upon comparisons of observed ocean warming versus modeled ocean warming. Quite simply they do not match; as their figure 7.14 on page 990 shows. Their “spatial pattern” analysis of modeled SSTs to observed SSTs, is supported by “multiple generations of climate models” and little else. They call upon the poorly understood net cloud feedback set of adjustable model parameters and use them to explain why the models are not properly predicting Pacific SSTs. Richard Seager and his colleagues have this to say about this idea:
“The tropical Pacific Ocean response to rising GHGs impacts all of the world’s population. State-of-the-art climate models predict that rising GHGs reduce the west-to-east warm-to-cool sea surface temperature gradient across the equatorial Pacific. In nature, however, the gradient has strengthened in recent decades as GHG concentrations have risen sharply. This stark discrepancy between models and observations has troubled the climate research community for two decades. … The failure of state-of-the-art models to capture the correct response introduces critical error into their projections of climate change in the many regions sensitive to tropical Pacific sea-surface-temperatures.”Seager, et al., 2019
Ross McKitrick, in his comments on the AR6 second order draft (SOD) Chapter 7, notes that the IPCC base their conjectures about “feedbacks on feedbacks” and a higher ECS on their ability to predict tropical climate accurately. Yet, as he and John Christy explain in their 2018 and 2020 papers, every run of every CMIP5 model over-predicts warming in the 200 hPa to 300 hPa (10-12 km) layer in the tropical troposphere and the differences are statistically significant in most cases. When observations are significantly different than the model results, the simplest explanation is that the models are wrong, not that the feedbacks are changing with increasing temperatures.
Summary and Conclusions
AR6 Chapter 7, “The Earth’s energy budget, climate feedbacks, and climate sensitivity,” was the source for most of the material in the first three parts of this series. It exudes a certain desperation, the reader is inundated with the phrases “high confidence,” “virtually certain,” and “very likely” ad nauseam. They are used less to describe and more to persuade.
When the IPCC discovered they were overestimating warming in the eastern Pacific and in the Southern Ocean, they did not conclude the obvious, that their models were wrong. Instead, they created an elaborate scenario, based on “patterns” of ocean surface warming that hypothesized that their CO2-caused warming feedbacks were subject to positive (warming) feedbacks themselves! Using a key model output, in this case surface temperature, to compute a critical feedback, that in turn is used to compute the same output, makes the model unstable and unreliable.
We have previously emphasized the importance of recognizing that climate change is not a global thing, it varies regionally, and particularly by latitude (see figure 3 here). CO2 is a well-mixed gas and has a nearly constant atmospheric concentration around the world and vertically through the atmosphere. As a result, if CO2 were a significant influence on climate, it might cause climate change globally. Presumably, this is why the IPCC focusses on global changes.
AR6 acknowledges that climate changes regionally, yet they do not acknowledge that this is evidence that their models and assumptions are wrong. Natural climate change is local, mainly by latitude, they seem to have decided that their hypothesized feedbacks are changing at different rates, in regional patterns, and call it the “pattern effect.” Isn’t it more logical to just acknowledge that most of climate change is natural, and that is why the models are not reproducing what we observe?
Finally, the IPCC, as well as many worldwide government agencies, are recommending that we curtail fossil fuel burning to limit warming to 1.5°C above what they call the preindustrial period. This period ends in 1750, the end of the coldest century (~1650-~1750) since the last glacial period, at least in the extra-tropical Northern Hemisphere. Human civilization has never seen colder temperatures. Very few people would want to return to the miserable climate of that time. Our modern climate is better and the additional CO2 we enjoy today has greatly improved agricultural productivity.
The IPCC has failed to measure the impact of CO2 and other GHGs on climate or global warming, that is, measure the climate sensitivity to CO2. Many researchers have used measurements to estimate climate sensitivity, but when those estimates are below what the IPCC wants, they simply ignore them.
The AR6 methodology, like the Sherwood, et al. methodology, was subjective in what estimates were included. In fact, AR6 specifically excludes many valid estimates of climate sensitivity, without explaining why, from page 1007 in Chapter 7:
“History has seen a multitude of studies (e.g., Svensmark, 1998; Lindzen et al., 2001; Schwartz, 2007) mostly implying lower ECS than the range assessed as very likely here.”AR6, p 1007
The “multitude” of estimates is simply ignored, without explanation. The explanation given is that much higher estimates based on paleoclimate studies are also ignored, although, the higher estimates were: “… shown to be overestimated due to a lack of accounting for orbital forcing and long-term ice-sheet feedbacks (Schmidt et al., 2017b).”
AR6, stepped away from the past practice of directly calculating ECS and TCR from model output. Instead, they used measurements, such as those by Lewis and Curry, in combination with several complex model-based calculations to constrain the values of ECS and TCR to an expected range. The methodology as explained in AR6 and in Sherwood, et al. was set up so that model-derived estimates swamped the instrument-based estimates, especially at the low end, allowing them to dial in the output they wanted.
In part 4, the final part of this series we examine how modern observations of CO2 and global average temperature are used to compute climate sensitivity and then how the computation is converted into a pseudo-ECS. Once the conversion is done, what does it mean? Look for part 4 tomorrow.
Download the bibliography here.
Including: Lindzen, R., & Choi, Y.-S. (2009, August 26). On the determination of climate feedbacks from ERBE data. Geophysical Research Letters, 36(16), Lindzen, R., & Choi, Y.-S. (2011, August 28). On the Observational Determination of Climate Sensitivity and Implications. Asia-Pacific Journal of Atmospheric Sciences, 47(377)., Idso, S. (1998). CO2-induced global warming: a skeptic’s view of potential climate change. Climate Research, 10(1), 69-82, Newell, R., & Dopplick, T. (1979). Questions Concerning the Possible Influence of Anthropogenic CO2 on Atmospheric Temperature. J. Applied Meterology, 18, 822-825., and (Lewis & Curry, The impact of recent forcing and ocean heat uptake data on estimates of climate sensitivity, 2018), among many others. ↑
AR6, pp. 981 and Figure 7.11 ↑
(Vinós, Climate of the Past, Present and Future, A Scientific Debate, 2022, pp. 184-187) ↑
AR6, page 980. ↑
(Vinós, Climate of the Past, Present and Future, A Scientific Debate, 2022, p. 189) ↑
Sherwood, S. C., Webb, M. J., Annan, J. D., Armour, K. C., J., P. M., Hargreaves, C., . . . Knutti, R. (2020, July 22). An Assessment of Earth’s Climate Sensitivity Using Multiple Lines of Evidence. Reviews of Geophysics, 58. doi:https://doi.org/10.1029/2019RG000678 ↑
Bates, J. R. (2016). Estimating climate sensitivity using two-zone energy balance models. Earth and Space Science, 3(5), 207-225. ↑
(Sherwood, et al., 2020). ↑
Popper, K. R. (1962). Conjectures and Refutations, The Growth of Scientific Knowledge. New York: Basic Books. Pages 35-37. ↑
Scafetta, N. (2021). Testing the CMIP6 GCM Simulations versus Surface Temperature Records from 1980–1990 to 2011–2021: High ECS Is Not Supported. Climate, 9(161) ↑
Lewis, N. (2022). Objectively combining climate sensitivity evidence. Climate Dynamics. ↑
Lewis, N., & Curry, J. (2018, April 23). The impact of recent forcing and ocean heat uptake data on estimates of climate sensitivity. Journal of Climate. ↑
Skeie, R. B., Berntsen, T., Aldrin, M., Holden, M., & Myhre, G. (2018). Climate sensitivity estimates – sensitivity to radiative forcing time series and observational data. Earth System Dynamics, 9, 879-894. ↑
Otto, A., Otto, F. B., Church, J., Hegerl, G., Forster, P. M., Gillett, N. P., . . . Stevens, B. (2013, May 19). Energy budget constraints on climate response. Nature Geoscience, 415-416. ↑
Christy, J., & McNider, R. (2017). Satellite Bulk Tropospheric Temperatures as a Metric for Climate Sensitivity. Asia-Pac. J. Atmos. Sci., 53(4). ↑
AR6, p 996 ↑
AR6, p 990. ↑
AR6, p 990. ↑
Seager, R., Cane, M. H., Lee, D.-E., Abernathey, R., & Zhang, H. (2019, June 24). Strengthening tropical Pacific zonal sea surface temperature gradient consistent with rising greenhouse gases. Nature Climate Change, 517-522. ↑
(McKitrick & Christy, 2018) and (McKitrick & Christy, 2020) ↑
(Vinós, Climate of the Past, Present and Future, A Scientific Debate, 2022, pp. 155-161) ↑
AR6, page 990 ↑
AR6, page 990, see AR6 figure 7.14 for a comparison of model results to observations. ↑
IPCC. (2018). Global Warming of 1.5 degrees C. (Masson-Delmotte, V., P. Zhai, H.-O. Pörtner, D. Roberts, J. Skea, . . . a. T. Waterfield, Eds.) Geneva: World Meteorological Organization. ↑
Idso, C. (2013). The Positve Externalities of Carbon Dioxide: Estimating the Monetary Benefits of Rising Atmospheric CO2 Concentrations on Global Food production. Center for the study of Carbon Dioxide and Global Change. ↑
AR6, p 1007 ↑
10 thoughts on “The Mysterious AR6 ECS, Part 3, What is Climate Sensitivity?”
I’m anticipating part 4 will set a record for four articles in a row by the same author that I recommend on my climate science and energy blog, with all four designated to be among my Top Four articles of the day. This is the best series of articles I’ve read in years, and I read 20 climate and energy articles every morning of the year. Mainly energy articles these days, because Nut Zero is a panic reaction to the guessed ECS range (which does not get enough attention these days).
One observation abot “ECS observations”
There are no observations of ECS or TCR outside of a controlled experiment in a laboratory.
What most people mean by observations is a worst case assumption. The assumption being that the measured global warming is entirely caused by manmade CO2 emissions. In reality, a change in the average temperature is the net result of ALL global, regional and local climate change variables. There is no way to isolate the effect of CO2 emissions alone. So an easy worst case assumption is to blame all the warming on CO2.
It appears TCR replaced the older TCS term
Is there any meaningful difference?
It seems the year used to be 1750, and then 1850,
and sometimes just “pre-industrila;”.
The important point, rarely mentioned is that estimates
of the average temperature are nearly worthless pre-1900.
The only number we hear is a very rough estimate of the average
Northern Hemisphere temperature, based mainly on weather stations
in the US and Europe, and some very haphazard ocean measurements
in NH shipping lanes. That number is likely to be is far from an accurate
global average temperature, and could be off by +/- 1 degree C, in either direction
The effect of CO2 is not global.
CO2 has no warming effect in most of Antarctica
that has a permanant temperature inversion, and during
frequent temporary winter temperature inversions in the Arctic.
Curiously, all Arctic warming is in the wointer,
I was thrilled that you did not declare your own personal
ECS estimate (unless I missed it). We already have too many
ECS estimates (I call them guesses) and too many long term
climate predictions. The ECS guess is the house of cards
“foundation” for climate scaremongering.
Bingham Farms, Michigan
Over 30,000 page views in the last three months
“It appears TCR replaced the older TCS term
Is there any meaningful difference?”
I’ve seen TCS used, but I think it is an error and the authors that use it mean TCR. TCR was introduced in TAR on page 78 and has been used in every IPCC report since. TAR says this:
“Used for the first time in this IPCC report is the Transient Climate Response (TCR). The TCR is defined as the globally averaged surface air temperature change, at the time of doubling of CO2 , in a 1%/yr CO2 -increase experiment. This rate of CO2 increase is assumed to represent the radiative forcing from all greenhouse gases. The TCR combines elements of model sensitivity and factors that affect response (e.g., ocean heat uptake). The range of the TCR for current AOGCMs is 1.1 to 3.1°C. “
It seems the year used to be 1750, and then 1850,
and sometimes just “pre-industrial”.”
Confusing to be sure, here is how AR6 explains it on page 192 in chapter 1.
“Although there is some evidence for human influence on climate before 1750 (e.g., Ruddiman and Thomson, 2001; Koch et al., 2019), the magnitude of the effect is still disputed (Section 22.214.171.124; e.g., Joos et al., 2004; J. Beck et al., 2018), and most studies analyse the human influence on climate over the industrial period. Historically, the widespread use of coal-powered machinery started the Industrial Revolution in Britain in the late 18th century (Ashton, 1997), but the global effects were small for several decades. In line with this, previous IPCC assessment reports considered changes in radiative forcing relative to 1750, and temperature changes were often reported relative to the ‘late 19th century’. The AR5 and SR1.5 made the specific pragmatic choice to approximate pre-industrial global temperatures by using the average of the 1850–1900 period, when permanent surface observing networks emerged that provide sufficiently accurate and continuous measurements on a near-global scale (Sections 1.3.1 and 126.96.36.199), and because model simulations of the historical period used 1850 as their start date. For the same reasons, to ensure continuity with previous assessments, and because of larger uncertainties and lower confidence in climatic changes before 1850 than after, AR6 makes the same choice to approximate pre-industrial global temperatures by using the average of the 1850–1900 period.”
So, they define preindustrial as before 1750, but use temperatures since 1850, as if there were no difference.
Only a government bureaucrat could produce that word salad.
Where can I buy an IPCC gibberish decoder ring?
“I was thrilled that you did not declare your own personal
ECS estimate (unless I missed it). We already have too many
ECS estimates (I call them guesses) and too many long term
climate predictions. The ECS guess is the house of cards
“foundation” for climate scaremongering.”
Thanks for that. I don’t believe the ECS idea has any real meaning. However, “climate sensitivity” to CO2 does. All I can say is observations provide a maximum climate sensitivity of about 0.7 to about 1.4, well below all IPCC estimates.
Lot of good questions, so one at a time.
“What most people mean by observations is a worst case assumption.”
This is the subject of part 4, already up.