Bishop Hill RealClimate on Lewis and Curry
Oct 6, 2014 Climate: sensitivity RealClimate has emerged from its latest bout of torpor to publish an article commenting on the recent Lewis and Curry climate sensitivity paper. It's written by Richard Millar, one of Myles Allen's post-docs, and the author seems to have adopted a much more businesslike tone than is normal at RC. Unfortunately, according to the first comment, which comes from Lewis himself, he hasn't got his facts right.
Richard Millar, you write
“They use the latest IPCC numbers for radiative forcing and global temperature changes, but not the latest IPCC ocean heat content data”.
The statement that Lewis and Curry (2014) does not use the latest IPCC ocean heat content data is simply untrue.
Section 3.2 of the paper explicitly states that it uses the climate system energy accumulation observational best estimates and uncertainty ranges shown in Box 3.1, Figure 1 of AR5, which extend to 2011, the final year of all the analyses carried out in the paper. The change in ocean heat content accounts for the bulk of the accumulation. Gregory Johnson is acknowledged in the paper for supplying the underlying data.
The bulk of Millar's case seems to be that the Lewis and Curry estimates are broadly similar to the 5-95% ranges of the IPCC's preferred estimates. However, as Nic points out in his comment, this obscures the fact that the energy budget estimates all have best estimates much nearer the 5% end than the 95% end.
Reader Comments (62)
Harold W, Nic Lewis
Papers such as Risbey et al 2014 and Shindell et al 2013 suggest that the mean TCR prediction from CMIP5 exceeds observed values for the last decade because the models ran with less negative aerosol forcing or and fewer La Nina years than actually occurred.
The model runs which most closely matched recent reality are those with the most negative forcing. This reflects the recent short term increase in negative forcing which is not representative of the long term trend. This implies that a heavy reliance on recent observational evidence will lead to underestimates of both TCR and ECS.
Harold W
I thought I'd best comment separately on RCP8.5 and its implications.
CO2 emissions continue to follow the RCP8.5 scenario, despite attempts to reduce the rate. Since the high waming rate peojections are mostly associated with RCP8.5 , it naturally leads to pessimism among warmists, especially those such as Hansen who believe ECS to be above 4.5.
Conversely, for lukewarmers to accept with equanimity the continuing RCP8.5 level of emissions, they must also believe in a low TCR and low ECS.
The tendency for the more pessimistic warmists to produce papers projecting high ECS and sceptics to project low ECS suggests that too much of the whole ECS projection process depends on subjective choices of starting assumptions. This makes me less willing to accept single papers, especially giving very high or very low estimates, and more inclined to accept the ensemble mean.
For the record, my own informal attempts to estimate ECS come in around 2.4, which makes me fairly conservative by warmest standards. :-)
Entropic Man - your reasoning is based on the quasi miraculous assumptions that observations be misleading right now and at the same time models would be good enough so that the larger negative forcings they need correspond to larger negative forcings in the real world.
In a sense you're assuming climate change and special present conditions. Little wonder you proceed to find climate change and special present conditions including a combination of feedbacks making the ECS far larger than CO2's alone.
Nic Lewis -
Thanks! For a single point of comparison, I chose 1.8 K as the median of 30 GCMs of WG1 Table 9.5, which is also the median of the 15 EMICs of Table 9.6. However, you make a good point that not all models were run against scenario RCP8.5, and the effective TCR of the RCP8.5 multi-model mean is most likely something like the average TCR of the 39 models which contributed at least one run. Unfortunately, not all such models have a TCR listed in Table 9.5. I'm presuming that your calculation of TCR ("close to 1.9 K") was arrived at by using ΔT and ΔF per equation 2 of Otto et al.(2013)? If so, was the comparison between the end of the 21st century and pre-industrial, or end-of-21st vs. recent times?
Jim Bouldin raised this question on his blog a few months ago, which started me thinking about the difference between TCR defined as the 70-year, 1% annual-increase scenario, and TCR defined by Otto et al. equation 2. You point out (in Lewis & Crok) that the projected rise from current temperature is about 1.7 times larger in CMIP5 than you calculate using a TCR of 1.35 K, but considering the rise from pre-industrial temperatures in Table 3, the warming ratio is closer to the ratio of TCRs. To what do you attribute the difference in ratios? [It's obvious arithmetically that (dT1+x)/(dT2+x)<dT1/dT2 for positive values. I was wondering whether this derives from a difference in "warming in the pipeline".]
Omnologos
There have long been sceptics complaining about the level of uncertainty in climate measurements. They then optimistically assume that reality is at the most benign end of the uncertainty range.
In practice, as measurement techniques improve, and the confidence limits shrink, the news tends to get worse rather than better.
.
Entropic man -
I note that your ECS estimate of 2.4 K/doubling qualifies you as a lukewarmer, per Steven Mosher's definition: "If offered an over/under bet on sensitivity of 3C, Lukewarmers will take the under."
Welcome to the club.
Harold:
...and Skeptics (such as myself) just snicker @3C over/under bets.
Entropic man -
"...those such as Hansen who believe ECS to be above 4.5."
As far as I can tell, Hansen has been fairly consistent recently in favoring an ECS of 3 K/doubling. E.g., here (p.10), where he refers to this paper. I don't think he ever stated that he thought ECS was above 4.5 K/doubling.
David Jay -
The term "skeptics" is claimed by many groups. Perhaps we need some new terms.
ZECSer - Zero ECS. Those who believe that the average temperature is not affected by pCO2.
MECSer - Minimal ECS; say, ECS < 1.2 K (the "no-feedback" value)
LECSer - Low ECS; 1.2 K < ECS < 3 K
HECSer - High ECS; ECS > 3K
Although it's not part of Mosher's brief definition, elsewhere he limits "lukewarmer" to those who believe in a positive ECS; hence "lukewarmer" = MECS + LECS (but not ZECS).
Entropic man
"as measurement techniques improve"
Would that be the repented often unjustified and poorly describes 'adjustments ' true by lucky chance they always work in favour of those whose career depends on a 'certain reality ' but I would not call them improved measurement techniques.
Meanwhile in space on the oceans improved measurements techniques are resulting in nothing more than a build of a 'need for more adjustments ' if the cause is to be kept on track .
Harold:
As a controls engineer, I cannot bring myself to accept the concept of a functioning, real-world system with net positive feedback.
So I would be a MECSer by your definitions, but way down at the low end (i.e. net negative feedback, not the least of which is T^4), not anywhere near 1.2C. I think the observational data sets that Lewis and Curry used have been adjusted and homogenized (dare I say molested???) beyond recognition and their results reflect the data sets used in the analysis.
hence, I snicker @ 3C over/under bets.
David Jay:
"As a controls engineer, I cannot bring myself to accept the concept of a functioning, real-world system with net positive feedback."
I'm with you on that one. However, the problem is an abuse of the terminology in climate science. The feedback model, to the extent that it's used in climate science, is that an increase in top-of-atmosphere forcing causes an increase in surface temperature. Increase in surface temperature causes an increase in outgoing thermal radiation (think Stefan-Boltzmann, the T^4 which you mentioned), reducing the TOA forcing. So there is a stable, negative-feedback loop. However, the earth is not an inert ball, and there are other mechanisms (clouds, land albedo, sea albedo, &c) which also enter into the picture. Some are "positive feedbacks" in which an increase in surface temperature induces an increase in TOA forcing. [One such example is reduction of ice cover.] The problem is that people have adopted the term "net positive feedback" if all the other mechanisms (besides the primary one described above) add up to a positive or negative value. What it really means to a control engineer, is that the feedback coefficient remains negative, but is somewhat less or more negative respectively, meaning that equilibrium is reached at a higher or lower temperature.
The most often quoted "no-feedback" value of sensitivity is 1.2 K/doubling. [Added later: I think I've seen 1.1 K as well. I'm going by memory here, with all the caveats that implies.] An ECS above this is called "net positive feedback"; hence my dividing line between MECS & LECS. Offhand, I can't recall any observational study which suggested that ECS was more likely below 1.2 K than above it.