Misrepresenting the climate sensitivity
Post date: Feb 24, 2014 7:30:57 PM
Introduction
The equilibrium climate sensitivity is an important concept to understand when discussing climate science. It is the equilibrium global temperature change in response to a change in global average radiative forcing, expressed in units of C/(W/m2). Because CO2 is well-established as the most important greenhouse gas that controls climate over geologic time, the climate sensitivity is often times expressed as C/(doubling of CO2 concentration). Equilibrium response in temperature takes hundreds of years to reach. Equilibrium climate sensitivity (hereafter equilibrium CS) should not be confused with effective climate sensitivity (hereafter effective CS) or transient climate response (hereafter TCR), which do not take into account warming yet to occur due to the internal inertia of the climate system. In complex models (such as coupled land-ocean models), CS is not a variable but can instead be calculated from simulations. In simpler models, CS is typically required as a parameter.
Because CS is an important concept to climate science, it is subject to many estimations using numerous independent techniques. It is also subject to some controversy, particularly in the media and from climate science skeptics. One particularly robust summary of these different estimates of CS came from Knutti and Hegerl (2008). From their Figure 3 (shown at right, click for larger version), CS estimates from the instrumental record, current mean climate state, general circulation models, the last millenium, volcanic eruptions, the last glacial maximum, other proxy data, and expert opinion were grouped and then combined into a best estimate. They found that the consensus range for CS is likely between 2.0C and 4.0C.
Different climate response terminology
Effective CS ignores the "warming in the pipeline" and represents only the warming that will have occurred at the exact point of measured CO2 concentration, and TCR refers specifically to the amount of warming that occurs specifically at the time of CO2 doubling. TCR is similar to effective CS in that "warming in the pipeline" is ignored. Values of effective CS and TCR should thus be lower than CS - sometimes substantially. The only exception to this being after hundreds of years when effective CS will become equivalent to CS. (Read more on climate sensitivity terminology at Skeptical Science)
The skeptical view of climate sensitivity
In recent years, some media accounts have suggested that there have been many new peer-reviewed papers suggesting a lower CS. Some of this confusion may also have come from changes to the IPCC-reported CS in the AR5 vs. the AR4.
One blog post in particular written by climate science skeptic Patrick Michaels and Paul Knappenberger for the climate science skeptic organization The Cato Institute showed a different summary of studies illustrating the equilibrium CS. In "Still Another Low Climate Sensitivity Estimate," the authors mostly discuss a new paper by Lewis (2013) that found a substantially lower equilibrium CS than the IPCC-reported values. As stated by the authors, "we welcome Lewis (2013) to the growing list of results published in the scientific literature since 2010 which find the climate sensitivity to be on the low side of the IPCC. God knows what the climategaters are emailing today." Then follows a figure summarizing some of the papers they selected, compared to the IPCC (see left, click for larger version).The authors also claim, "Based on the collection of results illustrated in our Figure, the future climate change projections about to be issued by the IPCC are off by an average of a whopping 70 percent." Yikes, that's quite a bit. But skeptical scientists must ask:
1. Have they actually represented each of these papers correctly?
2. Are each of the papers peer-reviewed?
3. Why didn't they include other estimates suggesting higher equilibrium CS, such as in the more robust summary by Knutti and Hegerl (2008)?
Summary Table
*Denotes that these values were subjectively taken from a chart rather than from text.
Correcting a misrepresentation of the climate sensitivity
After asking myself the above questions, I set out to see where Michaels and Knappenburger got their numbers. At the least, I wanted to see if there were any rebuttals and or comments to the listed papers that could provide context that the blog authors were leaving out.
IPCC AR4 and AR5
I already found the values provided by the IPCC reports AR4 and AR5 and provided them in a previous table.
Lewis (2013)
This was apparently based upon an earlier paper by Forest et al. (2006) which found an equilibrium CS range of 2.1-8.9C (most likely value of 2.9C). Lewis added 6 more years of data to their analysis, then also analyzed the data with a changed application of Bayes' Theorem. The modified Lewis approach yielded 1.0-3.0C (most likely value of 1.6C). More discussion on this can be found here.
Hargreaves et al. (2012)
The abstract provides an equilibrium CS estimate of 2.5C, but mentions many caveats.
Ring el al. (2012)
This paper seems to use a very simple method to estimate climate sensitivity via curve-fitting of empirical data. The paper seems to be estimating effective CS instead of equilibrium CS, and ignores warming yet to come from the current energy imbalance. The paper provides single value climate sensitivities for four different temperature datasets: 1.45C for GISTemp, 1.61C for HadCRUT, 1.99C for NOAA/NCDC, and 2.01C for JMA. No confidence interval information was provided, and it seems more appropriate to provide an average of these values rather than using a range from different datasets as if it were a confidence range. More discussion here.
van Hateran (2012)
This paper uses a very simple method to estimate climate sensitivity via curve-fitting of empirical data. The author provides four estimates. The first estimate assumes a very strong solar forcing and uses a millennium-scale response (not quote equilibrium CS, more like an effective CS), this yields 2.0C with a SE of 0.3C (~68% confidence interval). The second estimate also assumes a strong solar forcing, but just looks at the TCR (after about 70 years of 1% annual CO2 rise), this yields 1.5C with a SE of 0.2C. The third estimate uses a much smaller solar forcing with a millennium-scale response, this yields 2.5C with a SE of 0.4C. The fourth estimate is also a smaller solar forcing but just looks at the TCR, this yields a 1.9C with SE of 0.3C. The authors also acknowledge in their paper the mismatch between their results and the IPCC results, and assume that the difference between equilibrium CS and effective CS is partially responsible. More discussion on this paper can be found here.
To make these values match the other estimates that use a 90% confidence interval, I assumed 1.64xSE. These adjusted values are shown in parenthesis in the summary table.
Aldrin et al. (2012)
This paper reports concludes a mean climate sensitivity of 2.0C (90% range 1.2-3.5), but notes that this includes assumptions making this value not directly comparable to IPCC values. They also provided a several different estimates using various assumptions. Two scenarios included aerosols and clouds (more comparable to IPCC estimate), which yielded a CS estimates of 2.5C and 3.3C (ranges 1.3-4.9 and 1.5-7.3, read from a chart). For more elaboration, either read the paper or this blog post.
Annan & Hargreaves (2011)
The abstract suggests that the upper bound 95% confidence interval for a climate sensitivity is about 4.0C.
The original graphic from the Cato Institute alongside values I was able to verify from the referenced papers (below)
Conclusions
It is important to know the difference between the equilibrium CS, the effective CS, and the TCR. They are not the same values and you will reach incorrect conclusions when you compare them directly. It is also noted that simpler models can easily under-estimate the CS, especially when using only observed data (which ignores warming yet to come from the current energy imbalance). One notable case was the use of Aldrin et al. by Michaels in his chart. Michaels & Knappenberger used just one of the many scenarios provided by Aldrin. If Michaels & Knappenberger had used the scenarios that also included aerosols and clouds, the result would have been entirely different.
It seems that this is just one case of many where a limited amount of evidence is presented. This limited amount of evidence may also be mis-represented. Without the context of other studies, incorrect conclusions can be reached. When all the available evidence is taken into account, it is apparent that equilibrium climate sensitivity is likely closer to the IPCC estimates in the AR4 and AR5, contrary to the misleading chart produced by Michaels & Knappenberger.
Important Blog Note
Information presented in a blog post is accurate to the best of my knowledge but has not gone through a rigorous review process typical of articles in a peer reviewed journal. Technical information and or analysis provided is not necessarily intended to supersede conclusions made in the peer reviewed literature. As specified by the NOAA Scientific Integrity Policy, comments I post online through my Weather Underground account, personal website, or any social networking site such as Facebook reflects my personal and/or professional opinion, but may not reflect the official opinion of NOAA and the National Weather Service.