A new paper from Hausfather and colleagues (incl. me) has just been published with the most comprehensive assessment of climate model projections since the 1970s. Bottom line? Once you correct for small errors in the projected forcings, they did remarkably well.
Climate models are a core part of our understanding of our future climate. They also have been frequently attacked by those dismissive of climate change, who argue that since climate models are inevitably approximations they have no predictive power, or indeed, that they aren’t even scientific.
In an upcoming paper in Geophysical Research Letters, Zeke Hausfather, Henri Drake, Tristan Abbott and I took a look at how well climate models have actually been able to accurately project warming in the years after they were published. This is an extension of the comparisons we have been making on RealClimate for many years, but with a broader scope and a deeper analysis. We gathered all the climate models published between 1970 and the mid-2000s that gave projections of both future warming and future concentrations of CO2 and other climate forcings – from Manabe (1970) and Mitchell (1970) through to CMIP3 in IPCC 2007.
We found that climate models – even those published back in the 1970s – did remarkably well, with 14 out of the 17 projections statistically indistinguishable from what actually occurred.
We evaluated these models both on how well modeled warming compared with observed warming after models were published, and how well the relationship between warming and CO2 (and other climate forcings) in models compares to observations (the implied transient climate response) (see Figure). The second approach is important because even if an old model had gotten all the physics right, the future projected warming would be off if they assumed we would have 450 ppm CO2 in 2020 (which some did!). Future emissions depend on human societal behavior, not physical systems, and we can usefully distinguish evaluation of climate models physics from paths of future concentrations.
However, it is not totally obvious how one should correct for the forcing assumptions because of subtle issues related to the different efficacy of different forcings and, of course, the remaining uncertainty in the real value of the actual forcings (driven predominantly by the aerosol component). For forcing projections that were close to linear, this didn’t make that much difference, but for scenarios that weren’t (notably scenario C in Hansen et al (1988)), the correction does not work well.
There are a few other results that stand out, notably the (infamous?) low sensitivity result in Rasool and Schneider (1971), which was mainly due to a lack of stratospheric adjustment and water vapor short wave absorption in their formulation. This was noted by Schneider (1975) and the calculation redone by Schneider and Thompson (1981) which turned out to be far more accurate. On the other hand, only Mitchell (1970) appears to have substantially overestimated the TCR – even while he predicted the temperature rise quite accurately (due to a compensation between a too large sensitivity and an underestimate of the forcings). [Amusing aside, both Manabe’s and Mitchell’s 1970 projections appeared in a special volume on the Global Effects of Environmental Pollution, reporting on an 1968 AAAS workshop and edited by (the now-notorious) S. Fred Singer before he went off the deep end].
It’s worth noting that this comparison includes two kinds of climate model – those published prior to 1988 which are energy balance models of varying complexity, and those published afterwards which are true GCMs and include atmospheric (and eventually, ocean) dynamics. Of the early models, the work of Sawyer (1972) stands out as being the most accurate in terms of both temperature trends and forcings, though this must be considered somewhat fortuitous.
The fact that both classes of climate model did so well in projecting future warming should increase our confidence that current climate models are getting things right for mostly the right reasons. While there are still real uncertainties in future warming associated with climate sensitivity, we can confidently state that the rate of surface warming we are experiencing today is pretty much what past climate models projected it would be.
Gosh, maybe we know something about climate after all!
Note: all the data and code for this study are available here.
References
- Z. Hausfather, H.F. Drake, T. Abbott, and G.A. Schmidt, "Evaluating the Performance of Past Climate Model Projections", Geophysical Research Letters, vol. 47, 2020. http://dx.doi.org/10.1029/2019GL085378
- S.I. Rasool, and S.H. Schneider, "Atmospheric Carbon Dioxide and Aerosols: Effects of Large Increases on Global Climate", Science, vol. 173, pp. 138-141, 1971. http://dx.doi.org/10.1126/science.173.3992.138
- S.H. Schneider, "On the Carbon Dioxide–Climate Confusion", Journal of the Atmospheric Sciences, vol. 32, pp. 2060-2066, 1975. http://dx.doi.org/10.1175/1520-0469(1975)032<2060:OTCDC>2.0.CO;2
- S.H. Schneider, and S.L. Thompson, "Atmospheric CO2 and climate: Importance of the transient response", Journal of Geophysical Research: Oceans, vol. 86, pp. 3135-3147, 1981. http://dx.doi.org/10.1029/JC086iC04p03135
- "Global Effects of Environmental Pollution", 1970. http://dx.doi.org/10.1007/978-94-010-3290-2
- J.S. SAWYER, "Man-made Carbon Dioxide and the “Greenhouse” Effect", Nature, vol. 239, pp. 23-26, 1972. http://dx.doi.org/10.1038/239023a0
Where’s the most famous climate model of the ’80’s, TTAPS, 1983, and Thompson & Schneider’s 1986 critique of it ?
Nuclear winter reappraised. Foreign Affairs, 64. Summer 1986.
https://www.foreignaffairs.com/articles/1986-06-01/nuclear-winter-reappraised
Dr. Roger Pielke Jr. offers misguided comments in response to Dr. Hausfather’s discussion of the paper on Twitter. The paper serves as a nice antidote to Dr. Pielke’s distortion of the accuracy of the IPCC’s projections, among others. He basically didn’t correct for differences in projected vs. observed forcings, as one would need to do if one wanted to assess shorter-term sensitivity (as a ratio of temperature change vs. change in forcing). It’s also ironic that Dr. Gavin Schmidt is a co-author on this paper, since it vindicates the claims he made, and which Dr. Pielke mocked.
Sources on this below from Pielke, for the curious:
“Like the IPCC in 1990, the Hansen 1988 forecast overshot the mark. This is no crime, but a useful data point in trying to understand the predictive capabilities of climate models (not so good as yet, see Rahmstorf et. al 2007 for evidence that the 2001 IPCC has missed the mark). Gavin Schmidt’s claims that the 1988 Hansen prediction was right, and so too was the 1990 IPCC, so too was the 1995 IPCC, so too was the 2001 IPCC, so too was the 2007 IPCC are laughable on their face, as these predictions are not consistent with one another.
It is amazing how resistant some people (especially some modelers) are to forecast verification. Some of this is obviously for political reasons, some for pride, but the exercise that Steve has conducted here is fair and of great value.”
https://climateaudit.org/2008/01/16/thoughts-on-hansen-et-al-1988/#comment-132772
“Climate predictions and observations”
https://sciencepolicy.colorado.edu/admin/publication_files/resource-2592-2008.07.pdf
http://rogerpielkejr.blogspot.com/2013/09/global-temperature-trends-and-ipcc.html
Thanks for that article.
I assume the years under the names are 1) publishing year, 2) last year of the predicting interval. Correct?
BTW: good that all data and code are now on github.
It would be fine if the paleoclimate-data also would be available there.
I already had in mind to collect the data and put them on github. The problem was – even with the new collection of data, you once mentioned to me with url – that I could not read all the data.
Some of the formats were not readable.
I think some were doc-files others zip-files.
I would like to have the good old ASCII-based files back, so that reading the data is no problem.
Maybe you could do that, Gavin? (Put the p<leo-climate data ASCII-based on github?)
If it's there, I would try to make a good graph from it, as far as my knowledge on the data would be enough to interpret it. Some of the files I could read, were full of comments, which seem to presuppose/require some domain knowledge, which I don't have.
Taking climate model evaluation to the next level
Nature Climate Change volume 9, pages102–110 (2019)
https://doi.org/10.1038/s41558-018-0355-y
Earth system models are complex and represent a large number of processes, resulting in a persistent spread across climate projections for a given future scenario. Owing to different model performances against observations and the lack of independence among models, there is now evidence that giving equal weight to each available model projection is suboptimal.
Comparing model results to observations provides insight into the quality of model simulations and the way in which various processes are represented. Comparisons with observations can reveal short-comings in individual models and systematic errors in a large multi-model ensemble7’20. An example of a systematic error is the excessive simulated band of precipitation in the tropical Pacific south of the Equator, a feature not present in observations. Taken together with the usually correctly simulated climatological intertropical conver-gence zone (ITCZ) precipitation maximum that stretches across the tropical Pacific north of the Equator, this systematic splitting of tropical Pacific rainfall into two discrete branches is commonly referred to as the double ITCZ. Other examples of systematic errors include a dry Amazon bias, a warm bias in the eastern parts of tropical ocean basins, differences in the magnitude and frequency of El Nino and La Nina events, biases in sea surface temperatures (SSTs) in the Southern Ocean, a warm and dry bias of land surfaces during summer, and differences in the position of the Southern Hemisphere atmospheric jet.
He’s using rigged data. So his assessment is worthless.
Why not include AR5? (I can’t access the paper at the moment, sorry.)
Hey, Russell, try here:
Robock, A.; Oman, L.; Stenchikov, G. L. (2007). “Nuclear winter revisited with a modern climate model and current nuclear arsenals: Still catastrophic consequences” (PDF). Journal of Geophysical Research. 112 (D13): n/a. Bibcode:2007JGRD..11213107R. doi:10.1029/2006JD008235
Just like old times!
Just for context here’s my 2008 comment at Climate Audit that was selectively quoted above.
I’m happy to see that Gavin et al have confirmed that which has been long obvious, the Hansen forecasts ran hot.
“Roger Pielke. Jr.
Posted Jan 17, 2008 at 8:04 PM | Permalink
It is completely irrelevant which scenario Jim Hansen advertised as his favorite in 1988. Completely. I do not believe that it is possible to accurately predict energy use, emissions, policy developments, or technological innovations, among other things relevant to future forcings. So there is not point in faulting Hansen for not accurately predicting any of these things.
What he did in 1988 was entirely appropriate — try to map out the range of future possibilities based on some manner of bounding the uncertainties. With hindsight we are able to go back to that 1988 forecast and identify which scenario most approximated reality as use that as the basis for evaluating the performance of the forecast. It is quite interesting that Hansen’s forecasts in 1988 did not bound the possibilities as the observed record lies outside his realization space.
The A/B CO2 part of the projection was pretty good, and the other forcing agents less so. But the good news is that many of these played much less of a role in the overall forcing. So either A or B are fair comparisons against the observed record. C is obviously unrealistic.
The differences between A and B are largely irrelevant out to 2007, and it seems fairly clear that the actual temperature record under performed the 1988 prediction. People who try to make this about Pat Michaels and Scenario A are desperately trying to change the subject. But so too are those trying to make this about Hansen and Scenario A. It should be about forecast verification.
Like the IPCC in 1990, the Hansen 1988 forecast overshot the mark. This is no crime, but a useful data point in trying to understand the predictive capabilities of climate models (not so good as yet, see Rahmstorf et. al 2007 for evidence that the 2001 IPCC has missed the mark). Gavin Schmidt’s claims that the 1988 Hansen prediction was right, and so too was the 1990 IPCC, so too was the 1995 IPCC, so too was the 2001 IPCC, so too was the 2007 IPCC are laughable on their face, as these predictions are not consistent with one another.
It is amazing how resistant some people (especially some modelers) are to forecast verification. Some of this is obviously for political reasons, some for pride, but the exercise that Steve has conducted here is fair and of great value.‘
Some quotations from across the spectrum to keep modelers humble: https://www.masterresource.org/north-gerald-texas-am/climate-models-north-today/
Roger Pielke Jr No.8 It’s almost entirely ENSO. The Trade Winds increased 30%. That’s a lot.
Regarding Rasool 1971, I understand that this paper was also based on the assumption of extremely high future aerosol levels that never materialized. Were aerosols adjusted to match observations when this paper was evaluated?
Where does the fantasy that they did remarkebly well come from?
Over 2/3 of the predictions predicted warmer temperatures. Some were out by a whopping 50%.
7 Hey, Barton
That’s a decade old rehash by the original posse- a state-of-the-art Los Alamos modeling effort was reported in JGR last year:
Journal of Geophysical Research: Atmospheres, Volume 123, Issue 5, pp. 2752-2772
Climate Impact of a Regional Nuclear Weapons Exchange: An Improved Assessment Based On Detailed Source Calculations
Reisner, D’Angelo, Koo et al
DOI 10.1002/2017JD027331
The takeaway is :
” while our thorough simulations of the firestorm produce about 3.7 × 109 kg of black carbon, we find that the vast majority of the black carbon never reaches an altitude above weather systems (approximately 12 km).
Therefore, our Earth system model simulations conducted with model-informed atmospheric distributions of black carbon produce significantly lower global climatic impacts than assessed in prior studies, as the carbon at lower altitudes is more quickly removed from the atmosphere. In addition, our model ensembles indicate that statistically significant effects on global surface temperatures are limited to the first 5 years and are much smaller in magnitude than those shown in earlier works. None of the simulations produced a nuclear winter effect.
We find that the effects on global surface temperatures are not uniform and are concentrated primarily around the highest arctic latitudes, dramatically reducing the global impact on human health and agriculture compared with that reported by earlier studies. ”
https://vvattsupwiththat.blogspot.com/2019/10/there-they-go-again.html
Since Mr Pielke Jr seems to be around, I would like to ask him
1) Who has put forth hindcasts and forecasts that are superior to the ones you are so down on?
2) Absent large (and of course unpredictable) volcanic eruptions, do you expect in the next few decades any great deviation from the rate of global surface temperature warming measured over the last 4 decades?
@ #8
Dr. Pielke, there are false claims in your post. For example:
1) Hansen et al.’s 1988 forecasts bounded reality, since observed warming was between scenarios B and C, as per the paper. That makes sense, since observed forcings were less than scenario B, but more than scenario C, due to the Montreal Protocol limiting CFC emissions, factors contributing to limited methane emissions [ex: collapse of the Soviet Union), etc.
2) Your claim that the “forecasts ran hot” does not make sense. If you mean for total observed warming, then scenario C didn’t run hot. And we already know why scenarios A and B showed greater warming than observed: their projected forcings were greater than observed forcings. That’s not a flaw in Hansen et al.’s model physics, and you yourself admit “there is not [sic] point in faulting Hansen for not accurately predicting any of these things”. If you meant the “forecasts ran hot” on the more meaningful metric of implied TCR [as warming per unit of forcing increase], then the paper already rebuts that claim for scenarios A and B.
More generally, you and a lot of other people seem to be treating model-based climate projections unfairly (using special pleading, a.k.a. a double-standard). An analogy closer to my own field illustrates that point:
Smoking’s health risks are dose-dependent, with risk increasing as one smokes more (DOIs: 10.1161/CIRCULATIONAHA.109.904235 , 10.1136/tc.2005.011932 , 10.1001/jamainternmed.2016.7511). Suppose a doctor offers projections on smoking based on epidemiological models, saying that if you smoke, then your risk of cancer increases by various estimated amounts. They offer projections for no smoking, smoking 1 pack per day, and smoking 2 packs per day.
It would be absurd to say those projections are wrong, or that the epidemiological models were wrong, simply because patients smoked 1.5 packs per day, instead of the smoking frequencies the doctor listed in their projections. The point of the doctor’s projections was not to precisely predict exactly how much the patients would smoke. Instead the doctor offered conditional projections which patients could then use to help inform their decisions on smoking. It is then up to the patients to decide how much they will smoke, if they decide to smoke at all.
Similarly, Hansen et al. need not accurately predict subsequent greenhouse gas levels or emissions with great precision, in order for them to offer projections that people could then use to inform their decision on topics such as greenhouse-gas-emitting industries. Moreover, one can test the accuracy Hansen et al.’s model by comparing their predicted ratio of warming vs. increased forcing (implied TCR) to the observed ratio, just like one can evaluate the accuracy of the doctor’s epidemiological models by comparing their predicted ratio of cancer risk vs. smoking frequency to the observed ratio, as per Hausfather et al.’s paper.
It’s getting really frustrating seeing people treat climate science unfairly, in a way they wouldn’t dare treat other scientific fields.
Armando,
Your first sentence is this:
Earth system models are complex and represent a large number of processes, resulting in a persistent spread across climate projections for a given future scenario.
In other words, “These models can’t be trusted”.
As for the rest of your post, it’s just Mumbo Jumbo.
Another message from the paper is that greenhouse gas forcing has so thoroughly dominated natural variability in the past 50 years, that even energy balance models of the 1970s with no real ocean circulation and rudimentary cloud feedbacks were able to capture the GMST trend. All that despite the presence in that period of several record breaking El Ninos and a pronounced change in the phase of the PDO.
I’m quite sure that the calculations in the paper are correct. The outcome to: between 1970 and 2000’s the “early models” do not diverge too much from the observations. However, the question remains: Do they correctly replicate the TCR of the real world? During timespans of about 30…40 years the (multi) decadal internal variability is more or less muted. It would be interesting to see the sensitivities from about 1940 on, because the observed TCR in this timespan is about 1.3 as is was stated in Otto et al (2013) and Lewis/Curry (2014,2018) and to my knowledge nobody contradicted those values deduced from observations using a sufficient long timespan.
@Bubba, read the paper.
https://doi.org/10.1038/s41558-018-0355-y
That’s a regional conflict, Russell, not a global nuclear war.
B 16: Your first sentence is this:
Earth system models are complex and represent a large number of processes, resulting in a persistent spread across climate projections for a given future scenario.
In other words, “These models can’t be trusted”.
BPL: Thank you for that perfect example of a non sequitur.
Are events in the real world making these discussions rather esoteric?
Isn’t it time to say
Now that we know we are in real trouble, it’s about time to shame those that have hindered. As it’s election time in the UK can I mention Jacob Rees Mogg on climate?.
Also to ask for support for my petition Green number plates for clean cars, red number oplates for dirty ones ?
@Bubba @BPL
The text is from the paper.
Maybe if you look who wrote it (about 30 climatescientists) you will read it.
Earth system models are complex and represent a large number of processes, resulting in a persistent spread across climate projections for a given future scenario. Owing to different model performances against observations and the lack of independence among models, there is now evidence that giving equal weight to each available model projection is suboptimal.
In other words, “complex” and “untrustworthy” are not synonymes.
For the discussion on models and on facts and fact-checking:
The Mathematics of Climate Change
Rupert Klein, FU Berlin: How Math helps structuring climate discussions
Truth Under Siege: Climate Knowledge in an Age of Transparency, Skepticism, and Science Denial
#21 I am wondering is B 16 for real or is this an attempt to reinforce stereotypes?
Bubba:
If your pseudonym signifies your cultural identity:
then your comment suggests a bubba can’t tell the difference between science and superstition. That may be true (I blame the American school system), but are you actually proud of it? Humble ignorance is excusable, and easily corrected, you know; assertive ignorance, OTOH, not so much of either.
It’s not my job to educate you, thankfully, so I’ll leave you with this: “All models are wrong, but some are useful.” (George Box)
BEF: Roger Pielke Jr No.8 It’s almost entirely ENSO. The Trade Winds increased 30%. That’s a lot.
AB: For sure, and I’m talking through my hat here, but isn’t a model’s not predicting a 30% increase in trade winds a sign of significant error?
Bubba @16 “As for the rest of your (Armandos) post, it’s just Mumbo Jumbo.”
No its just too complicated for a twit like you to understand.
20 Give it a rest- BPL – I cited Robock & Stechnikov 2007 in my 2013 Nature piece as fair game for model intercomparison, because:
A.their paper recycled aerosol lifetime , fuel loading, and other parameter assumptions from the TTAPS paper era
B. The S in TTAPS told Foreign Affairs readers that optical depth and apocalyptic results rivaling the TTAPS 5,000 megaton “baseline exchange ” could ensue from a 100 megaton ” pure tactical war , in Europe, say.”
Which, as you just said Barton, is :
” a regional conflict, Russell, not a global nuclear war.”
@18, frankclimate
You said: “During timespans of about 30…40 years the (multi) decadal internal variability is more or less muted. It would be interesting to see the sensitivities from about 1940 on, because the observed TCR in this timespan is about 1.3 as is was stated in Otto et al (2013) and Lewis/Curry (2014,2018) and to my knowledge nobody contradicted those values deduced from observations using a sufficient long timespan.”
There are other estimates rebutting your point, especially with respect to Lewis+Curry’s repeatedly debunked work. For example, see below (the first listed paper is particularly relevant since its authors discussed it on RealClimate, it shares a co-author with Otto et al. 2013, and I’m pretty sure you should be aware of the paper already):
“A limited role for unforced internal variability in twentieth-century warming”
Figure 1 of: “Beyond equilibrium climate sensitivity”
“Econometric estimates of Earth’s transient climate sensitivity”
“A multicointegration model of global climate change”
“Estimating the transient climate response from observed warming”
#28 Al Bundy,
“I have a warrant for your arrest.”
You are actually asking a good question, but that’s because so far, we haven’t established a context:
What exactly are models supposed to “predict about the future”?
I would ask those who are critical of the work to answer that question as a condition for receiving a response. That’s a question that has to be answered with respect to precision, accuracy, and which metrics are being predicted/observed.
Gavin’s explanation with respect to projections depending on forcings or inputs is clear, as is AS at #15. However, I think it would be helpful to be more specific about what the product of the model offers, and how inputs are categorized.
This is a case where I do think lists might be useful.
31: I know the papers you mentioned, however I can’t find any “debunking” of L/C (14,18) instead they debate, if one underestimates the TCR of the real earth if one uses the EBM-approach. This is not the issue of the H.(2019) paper. There they use in princip the same EBM approach as it did Otto et al (2013) etc. ( they state this explicitly in the text) and come to TRC-values from obs. which are near 1.8. This is the result of the shorter time span they use. If one would estimate the better constrained TCR-values of Otto et al etc. from obs. one would find some overestimation of the modelled warming.
I continue to be shocked by what is accepted as “science” by the leading climatologists in this ongoing tragic moment in human history.
“While climate models should be evaluated based on the accuracy of model physics
formulations, climate modelers cannot be expected to accurately project future emissions
and associated changes in external forcings, which depend on human behavior,
technological change, and economic and population growth. Climate modellers often bypass
the task of deterministically predicting future emissions by instead projecting a range of
forcing trajectories representative of several plausible futures bracketed by marginally plausible
extremes.”
So the authors start by saying it’s impossible to be certain of important aspects, but when we look back at the models that guessed the right parameters the models were close.
That is not how science works nor does it lend an ounce of credibility to current expectations that range wildly from RCP 2.6 to RCP 8.5. Saying models are accurate and then at the same time accepting the range of predictions from RCP 2.6 to RCP 8.5 is not science. It is guessing, and don’t take my word for it. Here is what the MIT Center for Global Change has to say:
“Due to computational and conceptual limitations, however, most climate models do not incorporate realistic descriptions of the coupling between ocean circulation and climate and none include a realistic description of the biogeochemical cycle of CO2. If the future oceanic CO2 uptake rates cannot be predicted, then certainly the future levels of atmospheric CO2 cannot be predicted, let alone how they will affect global climate.”
Kris, #12–
Gee, I dunno. Maybe from the fact that 2/3 of the “predictions” were within the observational uncertainty?
Here’s a suggestion. If climate models are so good at forecasting then let us use them to model economic systems (with similar stochastic variables). I wonder why economic models fail to forecast stock prices even a day in advance but climate models have such precision forecasting ability with input variables that are similar in behavior to those in climate models?
[Response: Climate depends on physics. The economy not so much. -gavin]
Gordon, #11, asks:
Gordon, my understanding is that that’s the essential point they discussed in paragraph 5 of the post, to wit:
@33, frankclimate
Actually, H.(2019) had a TCR estimate of ~1.6, and covered a period of 1850 to about the mid-to-late 2010s. That’s clearly shown in figure 5 of the paper and further discussed in section 4 on page 4903:
“A limited role for unforced internal variability in twentieth-century warming”
https://journals.ametsoc.org/doi/pdf/10.1175/JCLI-D-18-0555.1
That is longer than the post-1940 period you asked for in your comment 18. So no, their higher TCR is not due to them using a shorter time period.
On your debunking point: I didn’t cite those papers in my comment 31 as examples debunking Lewis+Curry’s work. I cited them as examples of higher TCR estimates over longer time periods. Also, when X correctly points out an error in Y’s reasoning that invalidates Y’s conclusion or argument, then X has debunked Y. Y debating that point is compatible with Y being debunked. Similarly, Spencer + Christy in the 1990s and 2000s debated whether the troposphere warmed by a statistically significant amount, even though multiple research groups debunked their conclusion that it hadn’t.
So the mere existence of debate does not prevent one from one from figuring out who is clearly right and who is clearly wrong, allowing one to figure out who’s been debunked. In that respect, there’s a large literature debunking the approach used by Lewis+Curry, regardless of whether Lewis+Curry choose to continue debating the issue. Citing all the literature would make my response, too long, but I’ll include DOIs for a few recent examples (10.1007/s00382-019-04825-x ; 10.1029/2018GL080714 ; 10.1029/2018JD028481 ; 10.1029/2018EF000889).
Joe Peck:
Dude, what makes you think you know how science works? Ever hear of the Dunning-Kruger effect? Your argument is solely from ignorance, which means you haven’t an ounce of credibility yourself. You didn’t link to your quote on cgcs.mit.edu, but you’re plainly reading too much into it. “Most”, “realistic” and “cannot be predicted” are binary judgments on scalar attributes, evincing failure to recognize the relativity of wrong. Why do you think the author is more credible than Gavin is?
Most scientifically meta-literate people, assuredly including the authors of RC along with many of its regular commenters, understand that “all models are wrong, but some are useful” (G. Box). A realistic climate simulation, for example, would require an environmental chamber as big as the Earth, proceed in real time and leave you feeling windblown! Coupled GCMs are demonstrably useful, not for predicting particular outcomes with certainty, but for projecting more likely outcomes when parameters vary randomly within quantified uncertainty limits. IOW, your misinformed notion of “how science works” isn’t science’s problem.
Not only that Frank, but there is some new work by Lewis and Mauritian concluding that EBM observationally based ECS estimates are not biased low.
https://www.giss.nasa.gov/meetings/cfmip2019/s2/5_nicholas_lewis_c.pdf
Mal Adptive: You didn’t link to your quote on cgcs.mit.edu, but you’re plainly reading too much into it. “Most”, “realistic” and “cannot be predicted” are binary judgments on scalar attributes, evincing failure to recognize the relativity of wrong.
AB: I don’t care that I’m heterosexual because sapiosexual trumps all. Marry me, Mal.
R 30: Give it a rest- BPL
BPL: You give it a rest. You’ve got this permanent bee in your bonnet that TTAPS was a political study by partisans and is completely unreliable, and this somehow proves Carl Sagan was a terrible leftist, or whatever the hell your point is. You’re the one who keeps bringing it up. If you don’t want to hear more about it, shut the hell up. Nobody cares about your political prejudices about a man who’s been dead for 25 years.
abhay abhyankar: “Here’s a suggestion. If climate models are so good at forecasting then let us use them to model economic systems (with similar stochastic variables).”
Actually, if you have a 401K or other managed fund, you are doing just that. However, it is clear that you aren’t bright enough to understand your own analogy. Picking stock prices is akin to predicting the weather on a daily to weekly basis. Climate models are more akin to the models that look for trends–how will the aggregate of stock prices compare to bonds, and what mix will be most likely to yield the best return for a certain acceptable risk.
So, it appears that you understand neither climate nor finance.
Joe Peck,
Actually, this is precisely how science works. Models make predictions, and then you look at how those predictions did and how they bear on the RELEVANT assumptions of the model. Climate science has an exceptionally good track record in this regard. The models have very good predictive power insofar as the most relevant aspects.
What this means is that the influence of CO2 is so significant that details such as completely realistic ocean-atmosphere coupling, cloud physics, etc. are of only secondary importance. Tamino has shown repeatedly that a relatively simple two-box model contains sufficient detail to capture the relevnant physics.
#39, Mal–
Yes, Joe Peck’s comment at #34 is a pretty epic fail. Apparently he thinks that for climate science to have any validity, or even to be counted as “science”, the practitioners not only have to model the physics and meteorology accurately, they also are responsible for advancing sociology and political science to predictive heights never before seen.
He apparently also thinks that the concluding MIT quote somehow supports his contention–which it does not, other than that both are clearly intended to cast doubt. I suspect that it was taken well out of context, and in an unsuccessful attempt to verify that, I used some of the quote as a search term. The exercise did turn up a couple of interesting papers on the carbon cycle, and particularly the marine side of it.
https://eos.org/research-spotlights/can-we-predict-the-future-of-ocean-carbon-dioxide-uptake
https://phys.org/news/2015-06-insight-future-ocean-carbon-uptake.html
@Joe Peck, #34:
“I continue to be shocked by what is accepted as “science” by the leading climatologists in this ongoing tragic moment in human history.”
If the climate warnings from scientists are just a hoax, as some people state, I wonder why then ‘in this ongoing tragic moment in human history’ comes up in your comment.
Afraid of the socialism-zombie that pops up everyhwere, and that will eat your limbs?
Hmhhh, sounds like alarmism, what some AGW-deniers forecast, when problem aware people just talk about possible solutions to the CO2-problem.
There is another important opinion piece by Palmer and Stevens both of whom have impressed me in the past that I think breaks the stonewall of silence about how much climate models are missing. They want a big program to make better resolved simulations possible. That’s fine as far as it goes, but no one knows if this will really resolve the issues or not since there is weak theoretical justification and recent work on LES shows some disturbing issues even when everything is resolved.
https://www.pnas.org/content/pnas/early/2019/11/26/1906691116.full.pdf
It also appears that at least some climate models can be pretty unskillful even on global average temperature anomaly.
https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2019MS001829?af=R
The problem here is that we don’t need climate models to simulate global average temperature anomaly. Energy balance models can be calibrated to do a fantastic job. We need climate models to tell us the pattern of changes. They are not skillful for this.
I would really ask what the purpose of this post is. Everyone with modeling experience knows that if you tune a model with many parameters you can produce something that is skillful on the outputs used in tuning. That’s due to cancellation of numerical and physics errors. It’s unremarkable. In the case of climate models if you get the TOA radiative balance right and the ocean heat uptake right, then by conservation of energy you will come close on average temperature anomaly. This insight completely explains the situation we are in.
Al Bundy:
Er – yeah, thanks for the compliment, but no 8^}.
David Young said:
Speak for yourself. Your climate models may not be skillful (if you’ve even attempted to develop one), but mine are quite useful for deconstructing patterns. Most of these involve the oceanic dipoles, such as ENSO, which due to their intensity end up governing much of the natural variability in climate.
Yet, because they are standing wave dipoles they typically show zero trend, and so for trend modeling the radiative transfer models are equally skillful at showing the rising trend in temperature.
Climate models didn’t seem to work so well in 2014 for NOAA. It seems that whatever down to earth methods that “The Old Farmer’s Almanac” used are more accurate than NOAA’s multimillion dollar computer generated climate models.
“NOAA: Another warm winter likely for western U.S., South may see colder weather
Repeat of last year’s extremely cold, snowy winter east of Rockies unlikely”
October 16, 2014
(Credit: NOAA)
Below average temperatures are favored in parts of the south-central and southeastern United States, while above-average temperatures are most likely in the western U.S., Alaska, Hawaii and New England, according to the U.S. Winter Outlook, issued today by NOAA’s Climate Prediction Center.
http://www.noaanews.noaa.gov/stories2014/20141016_winteroutlook.html
How did this work out for NOAA? It looks like not well and the people of New England would more than likely agree.
Here is what Farmers’ Almanac is saying about this current winter.
“The Old Farmer’s Almanac’s long-range weather predictions for 2014–2015 are available—and another teeth-chatteringly cold winter is on its way across the United States!
“Colder is just almost too familiar a term,” Editor Janice Stillman said. “Think of it as a refriger-nation.”
With its traditionally 80 percent–accurate weather forecasts, The Old Farmer’s Almanacpredicts that this winter will be another arctic blast with above-normal snowfall throughout much of the nation. The extreme weather will continue into Summer 2015, which is expected to be predominantly hot and dry.”
http://www.almanac.com/content/winter-weather-predictions-old-farmers-almanac