The Rest of the Cherries: 140 decades of Climate Models vs. Observations
by Roy W. Spencer, Ph. D.
Since one of the criticisms of our recent Remote Sensing paper was that we cherry-picked the climate models we chose to compare the satellite observations of climate variations to, here are all 140 10-year periods from all 14 climate models’ 20th Century runs we analyzed (click to see the full res. version):
As you can see, the observations of the Earth (in blue, CERES radiative energy budget versus HadCRUT3 surface temperature variations) are outside the range of climate model behavior, at least over the span of time lags we believe are most related to feedbacks, which in turn determine the sensitivity of the climate system to increasing greenhouse gas concentrations. (See Lindzen & Choi, 2011 for more about time lags).
Now, at ZERO time lag, there are a few decades from a few models (less than 10% of them) which exceed the satellite measurements. So, would you then say that the satellite measurements are “not inconsistent” with the models? I wouldn’t.
Especially since the IPCC’s best estimate of future warming (about 3 deg C.) from a doubling of atmospheric CO2 is almost exactly the AVERAGE response of ALL of the climate models. Note that the average of all 140 model decades (dashed black line in the above graph) is pretty darn far from the satellite data.
So, even with all of 140 cherries picked, we still see evidence there is something wrong with the IPCC models in general. And I believe the problem is they are too sensitive, and thus are predicting too much future global warming.
“So, even with all of 140 cherries picked, we still see evidence there is something wrong with the IPCC models in general. “
Ok but you could have avoided significant criticism by running this analysis in the first place. I don’t really understand why you didn’t and this post doesn’t explain why.
“And I believe the problem is they are too sensitive, and thus are predicting too much future global warming.”
According to your own paper the most sensitive and least sensitive models failed. It suggests the problem is not sensitivity but another factor of model competence. Based on Trenberth and Dessler it seems like ENSO simulation ability is a more important factor. What do you think?
Also: Do you have any suggestions for what’s missing from the models? Is it possible to produce a model which reproduces both ENSO and the CERES measurements but has a low climate sensitivity?
“Cherrymandering”, I love it!
Well now you’re just cherrypicking your reality. What about all the other alternative realities? /sarc off
“And I believe the problem is they are too sensitive, and thus are predicting too much future global warming.”
It comes down to where you put your stock. Seems the folks that put more stock in the models are missing something. Model output is not data. It should not take a pHD to understand that.
When models do not predict accurately – based on actual data – the model should be reevaluated. In a post-normal science world, they simply adjust the data to fit the models. This is utter nonsense.
As my father used to say about his second-hand suits: “It fits where it touches.”
Looks like a bunch of noise to me – with 14 models there are bound to be some that track observed outcomes reasonably well for a short period of time, and some that have been more diligently retrofitted to hindcast earlier decades accurately. It will be interesting to pick the three that have performed the best over the last 10 years and see whether they do well over the next 10. I would estimate that each one has about a 7.14% chance of being the most accurate over the next 10 years, or perhaps slightly less because they may have been disadvantaged by over-enthusiastic retrofitting.
no cherries here:
22 Sept: Wired: David Kravets: CIA Says Global-Warming Intelligence is ‘Classified’
Two years ago, the Central Intelligence Agency announced it was creating a center to analyze the geopolitical ramifications of “phenomena such as desertification, rising sea levels, population shifts and heightened competition for natural resources.”
But whatever work the Center on Climate Change and National Security has done remains secret.
In response to National Security Archive scholar Jeffrey Richelson’s Freedom of Information Act request, the CIA said all of its work is “classified.”
“We completed a thorough search for records responsive to your request and located material that we determined is currently and properly classified and must be denied in its entirety,” (.pdf) Susan Viscuso, the agency’s information and privacy coordinator, wrote Richelson…
Steven Aftergood, who directs the Federation of American Scientists Project on Government Secrecy, blasted the CIA’s response to Richelson.
The CIA’s position, he said, means all “the center’s work is classified and there is not even a single study, or a single passage in a single study, that could be released without damage to national security. That’s a familiar song, and it became tiresome long ago.”…
http://www.wired.com/threatlevel/2011/09/cia-classifies-global-warming-intelligence/?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+wired%2Findex+%28Wired%3A+Index+3+%28Top+Stories+2%29%29
@glacierman
“Model output is not data.”
Of course they’re not data. The only time people claim they are is when someone is just about to disprove the notion.
Models are purely a representation of what’s currently understood about climate. No individual is able to be an expert on more than a small fraction of the climate system and models allow scientists to work with representations of the understanding of that system.
To put it another way, if there were no models what would Dr Spencer test the CERES data against? How is he supposed to determine how well climate science in totality understands and predicts that data?
“When models do not predict accurately – based on actual data – the model should be reevaluated”
It would be pretty surprising if there was anyone who thought models don’t need a lot of work before they fully explain how climate works. The issue is to do with the applicability and usefulness of models. They may not be able to tell you if it’ll rain on Wednesday next week but they may well be able to tell you how Arctic sea ice will respond over decades to warming temperatures or how rainfall patterns will be affected over decades.
Dr. Spencer, you can beat them over the head with the obvious, they will refuse to see what you’re hitting them with. But, well done, and a nice response to the critics of your selection.
Thanks Dr Spencer.
I think your paper contained adequate explanation to make your point. Adding all this extra data would just have muddied the waters and would have gived the peer reviewers more junk to wade through causing more delays.
It’s good that you add the fringe data later when you are challenged instead of posting a link and saying “Look it up yourself” like some of the climate elite. You work hard.
Don’t forget that you CI are the outlines of the outer maxima and minima. Only a tiny bit of blue emerges from this mess.
Powerful work, Dr. Roy. Keep blasting them with the basics and watch them whimper. Most likely, they will give up the climate models for paleo-reconstructions. That gives them an advantage because paleo-reconstructions are even more fantastic than climate models and, therefore, safer from criticism based on actual observation of the real world.
sharper00 says:
September 22, 2011 at 1:40 pm
To put it another way, if there were no models what would Dr Spencer test the CERES data against? How is he supposed to determine how well climate science in totality understands and predicts that data?”
Well two things here…
1. He would do what he did do and match it vs the real data…
2. What was science before we had computer models?
Hate to break it to you but models are nothing more then the ramblings of the drunk guys at the bar. Sure when your drunk they make sense… sure to the drunks rambling it makes sense to him…Sure even drunks at the bar can make profound leaps of logic and thought…
However going to a bar to taking down the predictions/ramblings of every drunk and comparing it to real data will get into a range… and that avg range will be somewhat close to the data… However just like models its just a matter of having enough noise short term to find a “valid” prediction. Then by citing said valid predict from one drunk… and then another for another case you have global warming or really anything.
Sharperoo
ALL of the dire predictions of the temperature increase for our climate, and the calamatous consequences of a 2-6C warning, come from these models! But they have gotten nothing right over the past 12 years. No abnormal warming in the equatorial troposphere. No acceleration in sea level rise, which appears has begun to slow. No sighificant warming in the SST. No statistical temperature rise since ’98, despite GISS “modification” of historical data. The models are not producing valid projections over a period of 10 years, why should we believe they will miraculously be accurate after the end od this century? The AGW proponents can’t find the missing heat and have “theorized” it is in the deep oceans, not already out into space as shown by Spencer.
When the models projections fail to match observations, it is time to reanalyze the models!
Bill
Cherries seem to be the flavour of the month
Bob B. says- “Well now you’re just cherrypicking your reality. What about all the other alternative realities?”
Agreed. I strongly recommend standing on the shoulders of other climate giants who pioneered the wind speed temperature proxy for finding a mid-troposphere hot spot, and use measurements of low Earth orbit wind speeds as a proxy for TOA radiation flux…
As an engineer working in space (figuratively) I find this inability to determine models bogus when they fail to model reality accurately really sad. We use all sorts of sophisticated models in the business and exploration of space. How do you think we keep satellites in their positions, launch things the size of the Space Shuttle safely into orbit hundreds to thousands of miles up to rendezvous with a station or an orbital slot, or how we time our swing around planets to launch a spacecraft millions of miles towards its next encounter with incredible precision and timing?
These models reflect multiple gravitational forces and counter forces (like atmospheric drag, solar radiating, etc). They are accurate – they have to be. Miss in space (which is mind boggling vast) and you can lose your many $100’s of millions toy. These models’ accuracy degrades within 7 days to the point they have to be updated with real measurements in order to resync them to reality (the natural forces on satellites in any trajectory are so random and decoupled only re-measurement can bring the models back in sync with reality). Within a month they predictions are pretty much useless.
These models have been refined over 5 decades of space exploration – yet they cannot predict out a month. We constantly use RF links over hours of time to measure the position and velocity in order to re-tune the models to reality. Within a month we could lose the ability to know where a satellite is if it did not squawk back to us and give us a hint.
So when I see this nonsense about scattered temp measurements which are then smeared over massive geographical grids in an attempt to model and predict years and decades out something much more complex and dynamic than a mass traveling in free space I just shake my head in shame. Who are we kidding?
Honestly, what does it take to infuse a little humility and just realize this statistical flaying around with the temp record is a joke? With incredibly high accuracy all we have proven in 20 years of alarmist studies is we don’t have the knowledge or data or models to unravel the Earth’s climate.
When a model does not produce an accurate result it is wrong. When it says 2+2 = 5 it is wrong. Close does not count in math, it is a binary situation. randomly getting close is not correlation either. And correlation does not mean guaranteed causation.
I just find this sad. In any other industry this would not even be funny. You can twist the knobs all day long on climate stats and models, but in the end we do not have the data to tease out sub degree changes over decades. I know I am repeating myself, but even CRU admits their 1961 temperatures are only good to within 2-4°C. So if the annual data is that noisy, there is no way to tease out a global annual signal of 0.8°C. Especially when the samples are thinly spread, of unknown quality and timing, and span such a short geological time period.
OK off the soap box again.
sharper00 says:
September 22, 2011 at 1:00 pm
“According to your own paper the most sensitive and least sensitive models failed. It suggests the problem is not sensitivity but another factor of model competence. Based on Trenberth and Dessler it seems like ENSO simulation ability is a more important factor. What do you think?”
So, you think that selecting just the models that “do a good job” on ENSO would be a more reasonable test of the models? Why?
Once again, we see why models are garbage. When working with models, one has to refer to the model’s “ENSO simulation ability.” (That reference is accurate, because only the whole model can simulate ENSO, no part of it can.) If you had a physical theory, you could identify those particular physical hypotheses that specify the ENSO phenomena by implying it. Then you could do further testing on those particular hypotheses. But models are all or nothing. The whole model generates each simulation and, because no piece of computer code is “about physical phenomena,” (has its own cognitive content) the model has to stand or fall on its own. In other words, the task now is to rejigger the whole darn thing.
The answer to my question above should now be obvious. Each of the models has failed on some subset of the empirical evidence and so it matters not at all that some of them are good on ENSO. All of them are failures.
Now, it will be of interest to modelers that some models are better than others on ENSO. Fine, let them carefully investigate the models and rejigger toward ENSO similitude. But to claim that one, some or all of the models is a more accurate representation of Earth’s climate because it does a good job on ENSO is a misunderstanding so fundamental that it is akin to asserting: “Forest, I see no forest; there are only trees here.”
sharper00 says:
September 22, 2011 at 1:00 pm
[Y]ou could have avoided significant criticism by running this analysis in the first place. I don’t really understand why you didn’t and this post doesn’t explain why.
Nor did it intend to, I believe. Straw Man.
Sherper00;
Ok but you could have avoided significant criticism by running this analysis in the first place. I don’t really understand why you didn’t and this post doesn’t explain why.>>>
What remains unexplained? Dr Spencer published a paper that was focused on sensitivity and so compared to the 4 most sensitive, and the 4 least sensitive models from the IPCC suite. It was his critics that came up with this nonsense about evading the issues by excluding the other 14 models and not including the ones the came closest to modeling ENSO. A ridiculous argument made to look all the more ridiculous by Dr. Spencer showing the other 14 as well, and none of them match actual observations either. At best, they’re less wrong.,,,sometimes. But over all, they’re wrong, Wrong, WRONG!
So what is it that you are accusing Dr Spencer of hiding? Why must you muddy the fact that he’s responded to his critics with the very results they insisted he was avoiding comparing to, and now that he has and made them look just as foolish as the first 8 models did, you imply that he avoided these for some reason? What is to be gained by implying such a thing when the actual comparison that was complained about is now right in front of you, and shows nothing more than what the comparison to the first 8 models showed:
That they are wrong!
sharper00 says:
September 22, 2011 at 1:40 pm
“Models are purely a representation of what’s currently understood about climate. No individual is able to be an expert on more than a small fraction of the climate system and models allow scientists to work with representations of the understanding of that system.”
In the arena of scientific method, to claim that “A represents B” is to claim that A describes B. Models describe nothing. Models generate simulations which fail to match observed reality to some degree.
“To put it another way, if there were no models what would Dr Spencer test the CERES data against? How is he supposed to determine how well climate science in totality understands and predicts that data?”
Dr. Spencer is not testing the data. He is using the data to show that no model can generate a simulation of the data that tracks somewhere close to the actual data. In my terminology, he is once again demonstrating that models are garbage that cannot substitute for physical theory.
“It would be pretty surprising if there was anyone who thought models don’t need a lot of work before they fully explain how climate works.”
A model cannot explain anything. It is computer code. Computer code is not about anything, does not describe anything, and has no cognitive content of its own. In science, to explain something is to produce the physical hypotheses (having their own cognitive content) that describe the natural regularities which caused the phenomenon explained.
“The issue is to do with the applicability and usefulness of models. They may not be able to tell you if it’ll rain on Wednesday next week but they may well be able to tell you how Arctic sea ice will respond over decades to warming temperatures or how rainfall patterns will be affected over decades.”
In science, prediction and explanation go hand-in-hand. If A explains B then A can be used to predict B and vice-versa. If A cannot be used to predict B then A cannot be used to explain B. Neither prediction nor explanation are relative to time. If A cannot predict B today then A cannot predict B at any time. And that is a very good thing. Otherwise, charlatans would be able to say that their predictions cannot be tested today but will prove true when tested at a future date after the grant money has been paid.
sharper00:
Your two above posts are incorrect.
Your first at September 22, 2011 at 1:00 pm is directed to Roy Spencer and says;
“Ok but you could have avoided significant criticism by running this analysis in the first place. I don’t really understand why you didn’t and this post doesn’t explain why.”
There was no “significant criticism”: as his above article proves, there was only pointless and inappropriate carping.
And your comment responds to his statement of opinion saying;
“And I believe the problem is they are too sensitive, and thus are predicting too much future global warming.”
by asserting;
“According to your own paper the most sensitive and least sensitive models failed. It suggests the problem is not sensitivity but another factor of model competence. Based on Trenberth and Dessler it seems like ENSO simulation ability is a more important factor. What do you think?”
That is a silly question. It ignores his opinion it is claiming to answer. You may have a different opinion but that is for you to justify and not him.
Then at September 22, 2011 at 1:40 pm you assert to glacierman;
“To put it another way, if there were no models what would Dr Spencer test the CERES data against? How is he supposed to determine how well climate science in totality understands and predicts that data?”
That is a denial of the scientific method. Indeed, it is the reverse of science. In science models are tested against empirical data and the data cannot be tested against the models.
This is because empirical data are observations of reality but – as you admit – models are merely representations of understandings of reality. The data can show the understandings are flawed but the understandings cannot show reality is other than it is observed to be.
Please refrain from posting pseudoscience.
Richard
“It would be pretty surprising if there was anyone who thought models don’t need a lot of work before they fully explain how climate works.”
Politicians across the globe (and their allies in science/academia) are confident enough in the models’ catastrophic predictions to throw wrenches in the economic works in their futile efforts to ‘save the planet’. I don’t recall a preamble about how ‘these models need a lot of work’.
AJStrata says:
September 22, 2011 at 2:43 pm
“As an engineer working in space (figuratively) I find this inability to determine models bogus when they fail to model reality accurately really sad. We use all sorts of sophisticated models in the business and exploration of space. How do you think we keep satellites in their positions, launch things the size of the Space Shuttle safely into orbit hundreds to thousands of miles up to rendezvous with a station or an orbital slot, or how we time our swing around planets to launch a spacecraft millions of miles towards its next encounter with incredible precision and timing?”
Do you really think that what you use to launch and manage satellites is a model? Newton’s physical theory allowed him to demonstrate during his lifetime how to launch an object into orbit around Earth. The models you use are not substitutes for physical theory. They are ingeniously designed programs that have built into them the management tools that you need to launch and manage satellites. But those who put together your models used the physical theory that was created by Newton and improved over the years. And that is the big difference between what you do and what Warmista modelers do. Warmista modelers have no physical theory that can explain Earth’s climate. Sadly, they use their models as substitutes for physical theory and tell our government and our people that the models are theories. No model is a theory.
The proof is in the pudding. If you doubt what I have said then simply ask a Warmista modeler for the physical theory which is embodied in his model. He won’t be able to show it to you. All he can offer you is his model. Warmista have no physical theory of Earth’s climate.
AJStrata says:
September 22, 2011 at 2:43 pm
As an engineer working in space (figuratively) I find this inability to determine models bogus when they fail to model reality accurately really sad. We use all sorts of sophisticated models in the business and exploration of space. How do you think we keep satellites in their positions, launch things the size of the Space Shuttle safely into orbit hundreds to thousands of miles up to rendezvous with a station or an orbital slot, or how we time our swing around planets to launch a spacecraft millions of miles towards its next encounter with incredible precision and timing?
These models reflect multiple gravitational forces and counter forces (like atmospheric drag, solar radiating, etc). They are accurate – they have to be. Miss in space (which is mind boggling vast) and you can lose your many $100′s of millions toy…..”
And yet…..”climate scientists” don’t have to be accurate, seemingly. Even though the “toy” they and the Warmist movement are playing with, the global economy, is worth untold trillions and virtually everyone’s life depends on it. “Miss” on earth and……………………………………………………..