Dr. Roy Spencer has in interesting explanation regarding CO2 as the only real driver of greenhouse warming.
Does CO2 Drive the Earth’s Climate System? Comments on the Latest NASA GISS Paper
There was a very clever paper published in Science this past week by Lacis, Schmidt, Rind, and Ruedy that uses the GISS climate model (ModelE) in an attempt to prove that carbon dioxide is the main driver of the climate system.
This paper admits that its goal is to counter the oft-quoted claim that water vapor is the main greenhouse gas in our atmosphere. (They provide a 1991 Lindzen reference as an example of that claim).
the Air Vent 
My metaphor from Andy Revkin’s DotEarth AGU thread in 2/08: The climate modelers are trying to keep their models on circular tracks on the ceiling.
==============
I don’t know why Roy Spencer is having a difficult time distinguishing between the state (clouds presently have a net cooling effect) versus incremental change (predicted positive feedback due to added CO2).
2-Reading the Lacis et al paper, it seems like they are ones who think that their model shows that the mean state is determined by the same factors involved in change: the feedback and forcing.
The truth is that an experiment like that of Lacis et al would be of little value even if their model was very good. The change in state from an atmosphere with CO2 and without (and back) is radically different from anything that has occurred or will occur in the recent history and probably distant future of the Earth, except perhaps for the effect of the Sun going Nova.
Andrew,
“The truth is that an experiment like that of Lacis et al..”
EXPERIMENT?
Et Tu Brute???
4-Hehe, yes, poor choice of words. By “experiment” I meant that this was a “test” of how the model behaves. Not an experiment in the typical scientific sense, more of a model exercise.
Ah but to them a computer model run is an experiment. The idea of experiments in the real world is outwith their experience from years of sitting in their air-conditioned offices fiddling at the keyboard.
The whole exercise struck me as a bit absurd. First, for making the assumption that the model predictions mean anything in terms of the Earth’s behavior, and second, for examining conditions which are bizarrely different from anything that we are talking about WTR CO2 increases. The real issue is (as always with the models) if assumptions like constant relative humidity with rising temperature are realistic or not. The lack of a measured tropical tropospheric hot spot and conflicting weather balloon humidity data sure suggest the models are on very shaky ground.
Will doubling CO2 lead to vast amplification due to water vapor? Maybe, I doubt it. But I am sure climate modelers don’t know for sure either.
(cross posted at Roy Spencer:)
“If they had forced the model with a water vapor change, it would have done the same thing.” [resulting in a large temperature response]
As Andy Lacis said in response, the expectation would be that the the climate would soon revert to its initial state, though I’m looking forward to see this tested (if it hasn’t yet been done?)
On the notion of the climate being able to change (e.g. warm) all by itself, without being ‘forced’: I can imagine two scenarios:
– The warming is balanced by cooling elsewhere (eg ocean, cryosphere, etc)
– The warming causes more energy (in the form of radiation) to leave the earth’ system, and as a consequence there will a negative energy imbalance at the top of the atmosphere. This will act to cool down the climate again.
Both scenario’s are highly unlikely to be playing a role right now: Warming (or its effects) is found in all segments of the climate system, and there is a positive energy balance at the TOA.
Why did Lacis et al not start by removing nitrogen, oxygen as second from the atmosphere? Because these two gasses do not influence climate on this planet by forcings?
Try to model that, you dumbwits.
After all, under these cicumstances, life as we know it would not exist. A virgin planet.
Geez guys, get over the use of the word “experiment” in context with a model run.
It is an experiment of sorts… a “numerical experiment.”
Suppose you have a detailed model that (you think) describes the physics.
And you have an analytic approximate formulation that you think describes how the detailed model behaves. And your colleague has a competing formulation.
How do you test which of the approximate formulations better describe the data, except by numerical experiment?
As the word is used here, it is precisely the same usage as one would use for a physical experiment (or a chemistry experiment, or a biology experiment, etc.)
If you don’t understand why a word is used in a particular context, it just may be because you are a science noob. 😉
Let me rephrase this slightly:
How do you test which of the approximate formulations better describe the model output, except by numerical experiment?
The ‘experiment’ tests the question, “Can we force simulated CO2 to drive our model?”
Yep, they can! 😯
Carrick,
Don’t mean to nitpick my friend, but you said:
“Suppose you have a detailed model that (you think) describes the physics.” OK
“And you have an analytic approximate formulation that you think describes how the detailed model behaves. And your colleague has a competing formulation.” OK
“How do you test which of the approximate formulations better describe the data, except by numerical experiment?” What?
What data? You said your formulation and your colleague’s formulation describe how the MODEL behaves. What is the data in this case? The predictions of the model, I assume. This exercise does not address the validity of the model, only the validity of your formulation of the model.
Am I missing something here?
RB wrote: “I don’t know why Roy Spencer is having a difficult time distinguishing between the state (clouds presently have a net cooling effect) versus incremental change (predicted positive feedback due to added CO2).”
First, why do you think that? Second, are you suggesting that the “net cooling effect” of clouds will become a “positive feedback” due to more CO2, or are you talking about some other feedback?
It is absurd to regard a model run as an experiment of the climate. A run of your model is just a test of your model/code.
When I was modeling physical inputs to predict specific real world events (with realtime life/death consequences), it was always clear to me that my models are just an approximation of the real world and not the real world. When you run your GCM models, you aren’t telling me anything about the real climate, you are just telling me how you have programmed your model.
In January (http://www.nature.com/climate/2010/1002/full/climate.2010.06.html), Kevin Trenberth had some interesting comments about GCMs (global climate models):
“So here is my prediction: the uncertainty in AR5’s climate predictions and projections will be much greater than in previous IPCC reports, primarily because of the factors noted above. This could present a major problem for public understanding of climate change. Is it not a reasonable expectation that as knowledge and understanding increase over time, uncertainty should decrease? But while our knowledge of certain factors does increase, so does our understanding of factors we previously did not account for or even recognize.”
The confidence intervals for GCMs are getting wider, not narrower. While this is a step forward in honesty, it is letting us know that even the GCM modelers are admitting that the science is not settled and that the GCMs are a work in progress and not an accurate model of the climate that can be used for experiments.
Steve Fitzpatrick:
See the follow up comment. What I meant to say was “model output” (I was using “data” in a more loose sense… as in result of a numerical experiment, but this can be confusing).
The purpose of analytic formulations is they provide insight into the underlying processes of the science. The more detailed models are meant to describe as exactly as possible the underlying physics, but at the cost of loss of all insight into why the assumed physics behaves the way it does.
In either case, you need physical measurement to validate theory. (To be clear, the theory is the same regardless of how you solve for a particular solution.)
What approximate analytic models do, through illumination of the underlying workings of the detailed model, is provide a method by which we may develop a testable phenomenology. It’s really pretty simple: We rarely if ever know the initial conditions and boundary conditions well enough to solve a detailed model and compare it directly to measurement (there are exceptions, but by and large this is the case). So what we look for instead are more processed results (amplitude of response to a tone) that exhibit predictable behavior as some independent quantity (e.g., frequency) gets varied.
What the more exact treatment of the physics does is allow us to validate our approximations of the underlying theory without having to first test with experimental measurement. After all, if the approximation fails to accurately capture the underlying theory, there is no surprise when it also fails to describe (presumably) future experimental measurements as well.
From the Trenberth paper cited by Steve Koch #15
‘…The timescale dictated by the IPCC process brings with it the risk of prematurely exposing problems with climate models as we learn how to develop them’.
How can any problem be exposed prematurely?
I naively thought that the sooner they were found the better.
Is this a small window into the way some of these people think?
This is what Professor Michael Kelly said in his input to the Oxburgh review of the CRU emails:
“I take real exception to having simulation runs described as experiments (without at least the qualification of ‘computer’ experiments). It does a disservice to centuries of real experimentation and allows simulations output to be considered as real data. This last is a very serious matter, as it can lead to the idea that real ‘real data’ might be wrong simply because it disagrees with the models! That is turning centuries of science on its head.”
See page 81 of http://www.whatdotheyknow.com/request/35907/response/94112/attach/4/David%20Hand%20s%20attachments%20from%20emails%20supplied.pdf
Expecting computer models to prove a scientific theory, instead of using physical observation, – is nothing short of intellectual and cultural decadence of the first order.
Models can be very useful for modelling well understood and well constrained engineering systems – but weather and climate – come on…
Comment from a Professional Computer Scientist…
18 Phillip Bratby. Might you please quote p 81? My version is not numbered.
At the risk of a snip for repeating a bit on Bishop Hill’s blog, the difference between traditional science and current climate science is stark. I attribute this to the older scientists being more intelligent, or through more intelligent selection criteria for promotion.
“Here is a snapshot of how it was done in the 1930s.
“James Chadwick was interned in Germany during WWI. In January 1932 he read a Curie paper mentioning a heavy particle. Immediately, he built apparatus resembling a discarded piece of plumbing and 3 weeks later announced the discovery of the neutron. This led to a publication – Chadwick, Sir James, “Possible Existence of a Neutron”, Nature, p. 312 (Feb. 27, 1932). A few evenings later he lectured to colleagues about this discovery of world importance, which led to a Nobel Prize in Physics in 1935. He had been elected a fellow of the Royal Society in 1927.
………………………………..
Why mention this?
1. Chadwick suffered personal discomfort and the Curies (Snr) death from their work.
2. Chadwick did his experiments before writing a note to Nature.
3. Chadwick told his colleagues of the work within weeks.
4. The Royal Society did not appear to influence Chadwick’s work.
5. There was no implication of impropriety levelled at Chadwick’s work.
6. The work was done at very low expense. Supercomputers did not exist.
7. His Nobel Prize has since been described as in the top 100 order of merit.
8. There was no serious challenge to the validity of his experiment or results.
9. Cockcroft and Walton built a rudimentary accelerator and split the atom on 14 April 1932, confirming Chadwick’s work 4 months after his idea.”
Now what’s this style of talk that says ‘the risk of prematurely exposing problems with climate models as we learn how to develop them’? Can one imagine Chadwick’s response to such words?
Anti-pagiarism note: The Chadwick story is told more fully by P.D. Smith, ‘Doomsday Men’, Allen Lane publ., 2007, whose guidance I have used with acknowledgement here.
15-“It is absurd to regard a model run as an experiment of the climate. A run of your model is just a test of your model/code.”
Oh, I completely agree. In fact, this is exactly what I said in addendum to my “experiment” statement (see comment 5)
Model runs are NOT experiments on the REAL system, they are experiments performed on the MODEL system. They tell you how the MODEL behaves, in the same way that an experiment on reality tells you how REALITY behaves. That is all I meant when using the word “experiment”, nothing more, nothing less.
Carrick,
never encountered solving a series of equations described as an experiment before unless it was testing whether the person or machine was capable.
An experiment is where you actually try something because you are at least doubtful that you CAN model it. Climate modelers have it backwards. They CLAIM they can model the climate reasonably, but, haven’t run the “experiments” yet to prove it!!
Carrick,
“As the word is used here, it is precisely the same usage as one would use for a physical experiment (or a chemistry experiment, or a biology experiment, etc.)”
We can screw up the formulation of a physical experiment, but, it will do what the laws of physics tells it to no matter what we THINK we are doing or not doing. There are still things to be learned from observing what happens as it is REALITY!
If we screw up the set up of a mathematical formulation there is no reality there to observe and learn anything from.
Carrick,
what i am trying to say is that you have to have a proven climate model before you can start adjusting things to find out what will happen under other conditions. Fiddling with something that only vaguely replicates something else gives no assurance of any usefulness as you do not know how the errors will affect the outcome as you do not know what the errors are or have any kind of quantification of them.
7
“assumptions like constant relative humidity with rising temperature are realistic or no”
Bit late, cos I’ve been talking about models on other threads. But GCM’s do not assume constant relative humidity or rising temperatures. Insofar as those are true, they are results, not assumptions.
#16, Carrick;
Which amount to saying that the models are suggestive speculations that could spark actual investigation resulting in a testable hypothesis. To quote G&T, “Never forget that climatology is not even a field, much less a science:
“Rather, the atmospheric greenhouse mechanism is a conjecture [= preliminary guess without evidence, which may lead to a hypothesis with pass-fail proposals, which may eventually qualify as a theory], …”