What does Climate Model output look like?

Lots of FUD about climate models get thrown around in the Climate Wars, but what is it they are really doing anyway?

The contrarians would have us believe they just take in a bunch of contrived parameters and spit out the worst case possible scenario for global average temperature increase. But the truth is the kinds of models the IPCC report on are very complex and nuanced.

Since I'm no expert and pictures are worth thousands of words, I would like to offer a few beautiful video realizations of GCM output.

The first is from the National Center for Atmospheric Research (NCAR) CCSM climate model and shows a single month of global weather in 25 seconds. Aside from being a beautiful thing to watch, I find it very fascinating. (h/t MT)

(source - and a larger image area)

Notice the incredible detail of weather patterns, the circumpolar wind current in the southern hemisphere, the flowing storm fronts, the tropical cyclone forming out of the Indian Ocean near the end, the daily pulsation of tropical precipitation over middle Africa and the Amazon rain forest. These things are not painted in after watching satellite feeds, they arise from the physics that is programmed in. I don't claim that all the resolution and nuance you see is right out of the numbers, I am sure that graphic "dots" were connected to smooth the coarseness of the grids the model work out of. But the general correctness of the large wind patterns and storm tracks means they are doing something very right.

The next one takes much longer and is output from Japan's Earth Simulator running the Hadley Centre's HadGEM1 model. Unfortunately it is too long for YouTube's 10 minute limit (I tried). You can find and view it here. (I wonder if this is the project James Annan is involved in.)

I prefer the other from a purely visual standpoint, but with this one you can watch the arctic sea ice grow and retreat.

The last is the Nonhydrostatic ICosahedral Atmospheric Model (NICAM), a cloud resolving model and shows a one week period.

(Source and description here

Imagine how much care and work has gone into building these virtual worlds. Now imagine the ridiculous notion of building these while artificially ensuring that no matter what else, they will scare the world into ceding control of our lives to the UN. There is no line of code in there that says IF CO2 > 450 THEN Fry(world).

More like this

Those are great videos, thanks for digging those up.

The level of confusion over how GCMs actually work just astound me at times. For instance, from WUWT we have today:

Nothing in current climatology considers irregular flows of energy between sea and air and between air and space underpinned by another level of irregularity in solar input. The models currently work backwards from meteorolgical observations rather than forward from measurements of net energy flow in and out of the different sections (oceans and air) of the system.

What does this person think the "C" in "GCM" means, if not "coupling" in the sense of modeling "net energy flow in and out of the different sections (oceans and air) of the system", with the physics in the model being driven by observation and theoretical work to explain the observations?

In my early days as a climate change advocate, I heard a lot of yelling about computer models, so I assumed they were just "okay" and that the other threads of evidence were more reliable.

But then I saw this video - http://www.youtube.com/watch?v=D6Un69RMNSw

It's like James Hansen had a time machine or something.

I'd love you guys to come check out my blog about climate change in the scientific world compared to the popular press. Link on my name, or go to ClimateSight.org

Thanks.

@2: The original acronym expansion of GCM is "General Circulation Model". The earliest had no radiation and little or no energy flow (condensation, for instance, was absent). They served to get us started with understanding the dynamics (circulation) of the atmosphere.

Then, for decades, and to the present if you look to the right authors (probably), GCM = atmospheric model. i.e., add radiative inputs, condensation/evaporation, and such to the atmospheric circulation model. Ocean was a swamp; it supplied water to the atmosphere, and was a heat source/sink, but the water never moved. The only degree of oceanography involved was to decide whether this swamp was 25 or 50 (or whatever) meters thick.

The quote you respond to, on the other hand, could not have been written by anybody even vaguely aware of how climate modeling is done, or has been at any time in at least the last 40 years.

Watch out, the top video features weather. Soon there will be a "Models don't predict weather 100% right therefore don't predict long term trends right!!!11!!1" post.

By Nils Ross (not verified) on 16 Jul 2009 #permalink

The original acronym expansion of GCM is "General Circulation Model".

Oops, thanks. I was confusing that with the HadCM[n] - where the C does stand for "Coupled".

Ocean was a swamp; it supplied water to the atmosphere, and was a heat source/sink, but the water never moved.

From what little I've read of HadCM4 (and I guess 3) the ocean model used is a dynamic, gridded one nowadays, but since the ocean changes relatively slowly they use observed local ocean temperature data rather than the dynamic model when using the atmospheric half for weather forecasting in the UK. I had a great link to HadCM documentation about a month ago but lost it, wish I hadn't.

Reading the GISS Model E documentation, they have several models that can be used.

One is the simplified kind you mention above, with the actual surface temps modified periodically during the model run from files of observed data.

On the other hand the GISS Dynamic ocean model is a complex, gridded dynamic model. I find the way they get around the fact that some straits connecting oceans are too small to be resolved by the grid system interesting:

The model contains up to 12 variable depth subgrid scale straits which contect ocean grid boxes, which would not be connected at the resolution used. In particular, the Straits of Gibraltar, Hormuz, and Nares straits are so modelled

I assume the coupled model is run with one ocean model or another based on what research questions are being explored, as clearly the dynamic model will run much more slowly than the "change the temp array via datafile" one.

Allow us to politely disagree with the straw man in this article at TheChillingEffect.org. We think science is our best bet but these models are not working -- and it's not necessarily that they have to have been corrupted from malice on the part of modelers.

Allow us to politely disagree with the straw man in this article at TheChillingEffect.org. We think science is our best bet but these models are not working

Well, yes, but the paper you cite shows the opposite of a "chilling effect". Rather, it suggests that climate sensitivy might be *underestimated* by models by a factor of two. That it might be 5-6C rather than the current best estimate, 3C.

Mounting evidence that CO2-forced warming is WORSE than we currently believe is, through a form of mental gymnastics that is incomprehensible to knowledgable people, twisted into an argument that CO2-forced warming doesn't exist at all.

On your site you ask

Could those dire predictions of global warming be wrong?

The answer, if the paper you've fallen head-over-heels in love with is correct, is "yes, they're wrong. It's going to be twice as bad as we think".

Ta-ta and cheers, old boy.

I'd agree you can't infer too much from the models.

This is a bit more mundanely due to the chaotic behavior of Navier-Stokes. But also the nature of the solvers used for numeric simulation which are themselves chaotically-inclined, starting with Newton-difference routines, which are about as simple and straight forward as one can get. And when they aren't chaotic they are accumulating error like an SOB over the long haul!

There are no other areas of science or engineering where these same numerical simulation algorithms can miraculously go long periods of time into the future without going wrong - they can't here either. The trick used to avoid error is to not extrapolate in time very far.

When I hear "the simulation proves X" I get nervous because no one should never be that comfortable with the output of a computer simulation. I know simulations in 0D, 1D, 2D and 3D well enough to have seen them badly misused by naively generalizing the result.

The fidelity of the model is so utterly critical to the result - it always needs to be minimally physics-based and proven to have all the critical effects. Numerical simulation codes are excellent error amplifiers: you can take a small error or mistake and multiply it up far faster than a hand calculation ever could. Things like omitting cloud albedo, which has incredible sensitive effect, and then making politically charged claims and predictions are unconscionable wrong.

In engineering only fools believe the simulation is truth first, second or third time around - you can kill people by doing that. Yes, there are more than enough engineers who are such fools.

Here we're dealing with physical time constants that are beyond human intuitive comprehension - this makes using intuitive a dangerous thing if not tempered by extreme rigor. Politics is utterly without rigor and is based on a good deal of intuition. This is what makes politicization of Global Warming very, very dangerous. People get sloppy on both sides of the political argument.

And yes. I've generally not seen evidence in the arguments that most climatologist understand the numerical simulation limitations they are using.

Things like omitting cloud albedo...

Bzzzt. Wrong.

If you don't understand what you're criticizing, those who do understand ain't going to pay much attention to you.

Just sayin'

Ctrl+F: "Convection"

No results found.

Climate models fail!

By Anonymous (not verified) on 18 Jul 2009 #permalink

Ctrl+F: "Convection"

No results found.

Climate models fail!

Bzzzt

dhogaza, you write "The answer, if the paper you've fallen head-over-heels in love with is correct, is "yes, they're wrong. It's going to be twice as bad as we think"

Actually, we haven't fallen in love with any particular paper. We do prefer a more sensible take on the world than throwing away scarce economic resources for the unlikely event of catastrophic global warming.

You are certainly entitled to your beliefs, though it seems difficult to support the notion that climate models are correct precisely because they've actually *underestimated* the negative impact. Incorrect is incorrect. Accepting a fault as evidence of success is a phenomenon seen in cults, not in solid science.

You are certainly entitled to your beliefs, though it seems difficult to support the notion that climate models are correct precisely because they've actually *underestimated* the negative impact.

No one argues that they're correct - the old saw about being useful holds, here.

Accepting a fault as evidence of success is a phenomenon seen in cults, not in solid science.

1. This is one paper, which may or may not hold up to scrutiny.

2. If it's right, then this:

We do prefer a more sensible take on the world than throwing away scarce economic resources for the unlikely event of catastrophic global warming.

is an incredibly stupid thing to say, because it means that things are going to be much worse than is currently thought.

Incorrect is incorrect.

Actually, the authors of this paper are arguing that the models are *incomplete*. Not "incorrect" in the sense that the physics being modeled are wrong, or the model implementation wrong.

Specifically, they're suggesting the models are incomplete because they don't include the impact of increased methane in the atmosphere due to the melting of clathrates from permafrost.

This is an *additive* effect to what's already being modeled. That means the number for CO2 sensitivity can only go one direction if the authors are correct - UP.

Your argument that a paper stating "things are worse than we believe" suggests "we should do nothing" is breathtaking.

As to why models don't include the impacts of increased methane from the melting of clathrates frozen in permafrost ...

The models are *conservative*. They model what's known, rather than speculation, and thus far the proposal that permafrost melting will lead to a surge in atmospheric methane has been speculative.

This paper you cite adds evidence that the speculation might be correct.

Atmospheric methane recently increased for the first time in (IIRC) a decade, also evidence that the speculation might be correct.

There are reports of direct measurements in Siberia that would be evidence that the speculation might be correct.

When or if the speculation becomes accepted as real, and can be quantified in relationship to temperature in the arctic, *then* the models will take this into account.

Not before.

The fact that modelers are conservative as to what they include isn't a "fault". it's the sound, scientific way to do models.

Regardless, there are two possible outcomes:

1. permafrost melt won't progress rapidly enough to cause methane to be added to the atmosphere in large amounts in the time frame we're currently mostly interested in (now until 2100). In this case, the model outputs of CO2 sensitivity in the 2C-4.5C range are right.

2. permafrost melting will be more aggressive than thought. In this cause, additional methane will be added to the atmosphere and model outputs of CO2 sensitivity will rise accordingly.

Either way, the current accepted figure for CO2 remains a LOWER BOUND.

"No one argues that they're correct..." and then quibble with "incorrect is incorrect"? Interesting. Thanks for helping me make my point.

Incomplete does not have to be, but can be, incorrect. However, incomplete in one area means there's no reason to suspect it is fully complete in another. To assume that there is only one problem with models is to assume a greater degree of scientific wherewithal than we currently enjoy, but to each his own assumptions.

The logic of refusing to act based on troubled models -- whether they are incorrect to the upside or downside -- is fairly apparent to those who prefer common sense and rational economic behavior.

Those demanding rush to action on climate change rarely seek rational policy and frequently prefer immediate action of any kind, even if it causes greater long-term suffering. Many feel the need to "do something" even if something is worse than nothing. Indeed, even those most accredited scientists investing in anthropogenic global warming recognize that catastrophe is a possible, but not guaranteed outcome. The economic consequences of rash policy, though, have a negative guaranteed outcome.

Judging by the way your converse with commenters here and elsewhere on the net, I don't suppose I could convince you of much. And that's cool. I do think your behavior helps convince more people to listen to me, though, and for that I thank you.

"No one argues that they're correct..." and then quibble with "incorrect is incorrect"? Interesting. Thanks for helping me make my point.

No model is "correct" in any absolute sense, that's a given. Yet, many are useful, as the old saw goes.

The primitive model used to evaluate "gun" designs for fission weapons during the Manhattan project were extremely primitive by today's standards - the "computers" who executed them were *people* pulling handles on mechanical calculators.

Yet, the model results were so robust that

1. They didn't bother testing the uranium "gun" design. Only one - "little boy" was built, and dropped on Hiroshima. From model results, the physicists knew that the only thing that would keep it from blowing up would be a mechanical failure of some sort. Not the modeled physics.

2. They didn't bother testing any plutonium "gun" design. Model results made it clear that a plutonium "gun" design could not be made to work.

"Incorrect" - in the sense of incomplete, just as climate models are today - but *useful*.

Same was true of the somewhat more sophisticated models used to design the first hydrogen bombs. By then (1950ish) there were early computers available. Modeling results showed that Teller's first design could not possibly work. By the time of the Mike shot, modeling had convinced researchers the bomb would work. Instrumentation was used to gather experimental results to fine-tune the models - fusion being rather hard to do in the lab, modeling totally led the way until real data from a real hydrogen bomb could be gathered.

Now, modeling wasn't the only thing going on, there were experiments to establish basic parameters regarding the physics, etc. Same is true with climate models, though, there's constant physics work going on which leads to improvements in the models.

Incomplete does not have to be, but can be, incorrect.

Handwaving on your part won't convince anyone, though. If you think the models implement physics incorrectly, or that the physics in the models are incorrect, the source to GISS Model E is online and you're free to *prove* your point rather than make handwaving assertions.

However, incomplete in one area means there's no reason to suspect it is fully complete in another.">However, incomplete in one area means there's no reason to suspect it is fully complete in another.

Modelers are very clear on the fact that modeling of clouds is incomplete. Regardless, given that the source to GISS Model E is online, there's no need to speculate as to where it is, or is not, complete. It's right there for you to look at.

I do think your behavior helps convince more people to listen to me, though, and for that I thank you.

People who are convinced by handwaving arguments typically have already made up their minds.

Fact is, thus far you've shown no understanding of how models work, what they might get right, what they might get wrong.

Your argument boils down to: OK, here's a paper that suggests models are wrong by a factor of 2, that climate sensitivity to CO2 is far higher than computed. Therefore, society should act as though CO2 is not a problem.

Dumb. Just ... dumb.

Thousands of lives depend upon the accuracy of models daily. Cruise ships, fishing boats, NASA launches, the Air Force and Navy and others depend on these models to report accurately. To dismiss models completely is a naive and ignorant choice.

Not all models are made the same, however. Most use real weather observations to correct their forecasts as time progresses, some don't and are purely theoretical and statistical. Then, there are boundary conditions. What happens when the location of interest is at the edge of the model? In most models that I've seen, the boundary region produces inaccurate predictions.

Finally, there's the issue of resolution. You can't use a global model to predict the weather in your city. You can't use a 48hr forecast to predict at exactly what time it will start raining.

The idea behind forecasts and models is to look for trends and apply those patterns to your real experiences. If you've noticed that the lake in your town is getting lower year after year, and the climate models tell you it's going to continue to get warmer, you can expect that your lake will eventually dry up. Hundreds of lakes in Minnesota have disappeared in the past 20 years, and I expect many more to go away in the next 20 years.

Paul,

I concur on the usefulness of models and the importance of examining real-world events. That's why models have been problematic for global warming, which has taken a decade-long hiatus.

"Dumb. Just ... dumb." Interesting way to convince others you're correct. I assume you're a hit at parties.

I've not claimed to be an expert on models, but I have addressed the logic used in the post above (which you have meticulously failed to address), the costs of just "doing something" (which you have generally failed to address), and your overzealous defense of inadequate models.

Think models are great but can be improved? That's a logical position. Your seemingly endless love affair with imperfection -- which, if used to drive policy will be costly -- is less logical.

You may invest as much as you like the validity of models. You may convince yourself of your righteousness. But data -- incomplete or incorrect -- without the application of real-world decisions on cost/benefit are essentially meaningless at best or misguided at worst.

That's why models have been problematic for global warming, which has taken a decade-long hiatus.

Not really

(you meant to say ELEVEN years, you're supposed to cherry-pick to include the super El Niño year of 1998)

Your seemingly endless love affair with imperfection -- which, if used to drive policy will be costly -- is less logical.

All models are imperfect. To say that one can't do anything based on models unless they're perfect is to say that models can't be used, ever.

You've heard of the nuclear test ban treaty, right? Since we can't test weapons any more ... guess what's used?

Why pick on climate models? I know! I know! Because you don't like the political implications that you fear will follow once our political leadership has figured out that climate scientists aren't really fraudulent green-helicopter flying commies after all.

But data -- incomplete or incorrect -- without the application of real-world decisions on cost/benefit are essentially meaningless at best or misguided at worst.

Ahhh ... the classic denialist goal-post move. Now it's "cost/benefit analysis" ... now where did I say such analysis is a bad idea? Or mention it at all?

I should state right upfront that I don't particularly enjoy wading in on climate change debates, since I am not exactly qualified to debate the issues, but there does seem to be a large and quite frankly scary misrepresentation of the science behind these models. Having sat through enough seminars and lectures on the issue (I'm right around the corner from the Center for Atmospheric Science at Cambridge), the thing almost everyone fails to mention is the HUGE error margin on these models. And I do mean huge. As in, if you select slightly different parameters, the earth enters an ice age.

I realise this debate is important, and the results are potentially catastrophic, but the climate change debate is quite possibly the worst poster child for proper science I can think of. We are dealing with a massively complex system, which we can't perform controlled experiments on, and our best predictions are based around simulations with unimaginable numbers of interrelated variables. Even in situations where we can perform experiments, the system is governed by simple equations and there is abundant data to generalize from, our simulations are poor. I speak primarily as a chemist; numerical modelling of chemical behaviour is a comparatively simple problem, yet we still pretty much keep tweaking the parameters of the simulation until it gives us what has happened in the past. I'll admit you have to publish papers somehow, but it doesn't make for great science.

My problem throughout this is not that the models are right or wrong, but that they are misrepresented. Real science deals with uncertainty. You investigate a problem, build a model, test it, report where it succeeded, and most importantly where it went wrong. Science is built on falsifiability, not confirmation. You give an indication as to the likelihood that the models are correct, but over-representing the power of the model is bordering on misconduct. Sadly, people who should know better bow to political pressure to make a judgement one way or the other. The IPCC (in earlier reports at least), instead of making a nuanced judgement and report, stating that while global warming is the most likely outcome of the model, the conclusions are not solid, instead adopts a political position of almost certainty. Admittedly recently it's been a bit more subtle, but even so, you don't really hear about that on the news - "climate change a possibility" is not something a journalist wants to hear.

The correct way to test these models is to make a prediction, and then test them. Testing on previous data doesn't count - you can just fiddle the parameters until they fit. Note that this fiddling is often unconscious; you make your model, and if it doesn't fit, you leave it and try another. If it does fit, you have no need to tweak it. Richard Feynmann famously said that the easiest person to fool is yourself; once you've made sure that you're doing good science, not fooling other people is a simple matter of honesty. Unfortunately, too few people check themselves in this way, and these tests take time. They're hard, and you don't publish many papers while you're waiting for results. Consequently, the debate is shaped by those less honest members of the academic community, who publish anyway. As I said, hardly a poster child for good science. The worst thing is, of course, if global warming doesn't occur, one side will say it was because it wasn't real, and the other will say it's because we took steps to avert it. We won't have actually learned anything.

By Massive_hair (not verified) on 18 Jul 2009 #permalink

My problem throughout this is not that the models are right or wrong, but that they are misrepresented. Real science deals with uncertainty

You mean like this? "Climate sensitivity to a doubling of CO2 is in the range 2C-4.5C"?

over-representing the power of the model is bordering on misconduct.

Consequently, the debate is shaped by those less honest members of the academic community, who publish anyway. As I said, hardly a poster child for good science.

Oh oh ... another accusation of science fraud. Ad hom on the field at large.

Tch tch.

I saw exactly these same words on another site recently ... hmmm ...

The worst thing is, of course, if global warming doesn't occur

It already *has* occurred. And *is* occurring. By the end of '09, 9 of the ten warmest years on record will have been in the 21st century. How people label this as "evidence of global cooling" is beyond me.

The correct way to test these models is to make a prediction, and then test them.

Which, of course, has been done. For someone who claims to know so much, you seem to know little.

Testing on previous data doesn't count - you can just fiddle the parameters until they fit. Note that this fiddling is often unconscious; you make your model, and if it doesn't fit, you leave it and try another. If it does fit, you have no need to tweak it.

Someone else who doesn't understand how GCMs work. They're not statistical models juggled to fit past data.

massive_hair,

The correct way to test these models is to make a prediction, and then test them.

This goes on all the time using various things, eg large volcanic eruptions or some previously unobserved aspects of the climate system like radiation at the TOA (top of the atmosphere). Read more about models and tested predictions here and here.

If however you mean we need make a thirty year projection and then wait, this is not a realistic option.

Testing on previous data doesn't count

What is this, some kind of collegiate sporting event? There is no reason to exclude using hindcasts to test models and every reason to do so. The past is an excellent source of test data involving all kinds of very different climates and climate changes.

Unfortunately the rest of your comment is a confusion of journalism criticisms and actual research with no acknowledgement of the difference that it can not be untangled enough to be answered. Your statement "everyone fails to mention is the HUGE error margin on these models. And I do mean huge. As in, if you select slightly different parameters, the earth enters an ice age." is both wrong and confused. No papers are published without uncertaintiies. And your assertion about entering an ice age, if true, would imply great sensitivity, not great uncertainty.

You should fix the initial problem you identified ("I am not exactly qualified to debate the issues") by asking questions rather than opining at such great length!

How come no one will address the basic logical error in this post, which is that criticism of climate models -- even if you think that the critic is "dumb" -- must be wrong since so much work goes into models? A ton of work can go into building bridges, and unfortunately sometimes they have structural problems.

Personally, I would have just chocked it up to a hasty blog post -- everyone does it and there's no shame in wishing to take a mulligan.

But the defensiveness shown in the comments is telling. Does the average scientist become irate if he thinks a neanderthal misunderstands gravity? If one is so convinced that their position is scientifically valid, there is no reason for such hostility. It is usually when one's religious beliefs are under attack that such reactions are seen.

... and still no one has addressed the simple logical error of this post. Coby and dhogaza may be in good supply of confidence -- and, perhaps even a good supply of knowledge -- but appear lacking in grace and logic.

How come no one will address the basic logical error in this post

A second moving of the goalposts.

even if you think that the critic is "dumb"

I don't think you're dumb, indeed, when defeated on a point, moving the goalposts is smart.

Because it's your only hope of maintaining any semblance of credibility.

Boy, didn't take long to break you down from your triumph claims of victory into being a goal-post moving troll, did it?

Does the average scientist become irate if he thinks a neanderthal misunderstands gravity?

Glad you identify with the neanderthals. But ... no, the average scientist gets irate when the neanderthal tells him that the *scientist* misunderstands gravity. And when, even after repeated references to the success of ballistic tables to guide accurate cannon shots that destroys the neanderthal's village, the neanderthal *still* says the scientist doesn't understand gravity. While digging his own wife and children from under rubble.

Thanks for the analogy, it's entirely fitting.

How come no one will address the basic logical error in this post, which is that criticism of climate models -- even if you think that the critic is "dumb" -- must be wrong since so much work goes into models?

This statement is, of course, a lie.

Coby didn't say criticism of climate models is wrong because so much work goes into them.

He says this:

There is no line of code in there that says IF CO2 > 450 THEN Fry(world).

In other words the typical, widely-stated, accepted-by-denialist shits, complaint that the climates "assume" global warming is total bullshit. In other words, that ignorant, dumb criticism based on having no idea as to how they work is bullshit.

He's right. He could've worded it better so liars like you couldn't intentionally "misunderstand" his statement, but it's not his fault that you choose to lie.

Oh, and it might be nice if model critics spent even 0.001% of the time understanding models vs. the time spent building them. Because criticizing them based on total misunderstanding of how they work is totally unscientific.

Indeed, one might suspect it's ideological ... and I suspect that, of you.

Coby,

Re your post #26, Could I ask you one simple question - can you put a figure on how certain you are that the recent global warming is due to anthropogenic CO2?

... and still no one has addressed the simple logical error of this post. Coby and dhogaza may be in good supply of confidence -- and, perhaps even a good supply of knowledge -- but appear lacking in grace and logic.

Good supply of knowledge? Yes
Good supply of logic? Yes

The only one you might be anything near right on is grace. And, hell, having to answer the same denialist bollocks over and over again, and having them repeat the same argument as if no one had said anything, is incredibly frustrating.

If one is so convinced that their position is scientifically valid, there is no reason for such hostility. It is usually when one's religious beliefs are under attack that such reactions are seen.

Actually, this type of reaction is fairly typical when it comes to dealing with denialists. You see the same hostility directed to evolution deniers, AIDs denialists, holocaust denialists, and justifiably so.

Richard, the IPCC is 90% certain. I think they are being cautious and conservative and are afraid to overstate the case. I am personally 99% certain that the recent global warming of between .1 and .2 oC per decade is primarily caused by anthropogenic factors, foremost of which is CO2 emissions. You should prefer their conclusion over mine, as I am not an expert.

Bret, perhaps you are unaware of the very vapid criticisms of climate models that are out there, especially the continuous insinuations that they are simplistic contrivances designed solely to show increased CO2 will cause severe warming. This post is offered to address those arguments only. I put a little bit more thought into defending model results on the two pages I linked to in comment 26.

Unsophisticated arguments don't require sophisticated debunking.

Notwithstanding the above, yes of course you are right that the fact they are complex does not mean they are above criticism.

Can you acknowledge one or two of dhogaza's points rather than changing the subject? I am specifically interested in whether or not you accept that a paper saying there is evidence the models are underestimating the climate's response to CO2 by half (your offering) is not a compelling reason to conclude we should just continue as we are til the models get better.

Coby,

You say "the IPCC is 90% certain. I think they are being cautious and conservative and are afraid to overstate the case. I am personally 99% certain that the recent global warming of between .1 and .2 oC per decade is primarily caused by anthropogenic factors, foremost of which is CO2 emissions. You should prefer their conclusion over mine, as I am not an expert."

Why should I accept either of these conclusions? Neither of these opinions are founded on science. The IPCC's conclusion is purely political, and yours purely based on faith.

Can you show me in the scientific or technical analysis of the IPCC HOW THEY HAVE ARRIVED AT THIS FIGURE?

The technical report of the IPCC 2001 has this to say when it examines the question of "WHETHER A HUMAN INFLUENCE ON CLIMATE CHANGE TO DATE CAN BE IDENTIFIED.":

"The SAR concluded that âTHE BALANCE OF EVIDENCE SUGGESTS THAT THERE IS A DISCERNIBLE HUMAN INFLUENCE ON GLOBAL CLIMATEâ. It noted that THE DETECTION AND ATTRIBUTION OF ANTHROPOGENIC CLIMATE CHANGE SIGNALS WILL BE ACCOMPLISHED THROUGH A GRADUAL ACCUMULATION OF EVIDENCE. THE SAR ALSO NOTED UNCERTAINTIES IN A NUMBER OF FACTORS, INCLUDING INTERNAL VARIABILITY AND THE MAGNITUDE AND PATTERNS OF FORCING AND RESPONSE, WHICH PREVENTED THEM FROM DRAWING A STRONGER CONCLUSION."

The most important thing in my mind in the above is the uncertainties that the IPCC admits in THE RESPONSE to these forcings, which I believe from the evidence is to mitigate and offset the effects, in other words a negative feedback, as the climate has always behaved in the past.

When there are a large number of uncertainties, which there are in climate models, as you are probably aware the uncertainties are multiplied with each other so that the end result is even less uncertainty and not more.

Nothing in the technical analysis suggests a 90% certainty and most importantly nowhere have they shown a working on how they have arrived at that certainty.

nowhere have they shown a working on how they have arrived at that certainty

Am at a loss as to how to answer this, the entire report is how they have arrived at that conclusion! It is a ridiculous question.

You should read it, start with the summary for policy makers (large PDF) and you might want to focus on the section about attribution(large PDF).

When there are a large number of uncertainties, which there are in climate models, as you are probably aware the uncertainties are multiplied with each other so that the end result is even less uncertainty and not more.

As an unqualified general statement this is false. Multiple, independent uncertainties can in fact work to constrain each other. Consider multiple independent measurements of the same property as a simple example. If your uncertainty on each measurement is 10%, 100 measurements will get you a much more constrained range of values, not less. Likewise, if you have a number of uncertain factors, but you know they must add up to 100%, you can actually reduce some of the individual uncertainties.

The IPCC's conclusion is purely political, and yours purely based on faith.

The IPCC is a political body, but its assessment is a scientific one and has been reviewed by national academies and scientific bodies all over the world. With the exception of a more lukewarm statement from the American Association of Petroleum Geologists, all of this various scientific bodies concur with the IPCC's conclusion.

You are not obligated to accept their or my conclusion, but both are based on the scientific evidence. Considering all the evidence I have compiled on this site, agree with it or not, your dismissal of my opinion as purely faith based is much more a reflection of you than it is of me.

Coby -
"Am at a loss as to how to answer this, the entire report is how they have arrived at that conclusion!"

I will accept that you are at a loss as to how to answer the question. But to tell me to read the whole report is a cop out.

Certainty or uncertainty is a mathematical computation. Where exactly have they worked this out? Or even claimed to have worked this out? You have given explanations of various questions on your website. Give one on this also, or admit frankly that there is none.

Richard,

Try following Coby's link. That's where I found this, which explains how the IPCC defines its probability estimates.

The IPCC in part calculate certainty quantitatively where possible and estimate it with expert opinion elsewhere.

You say:

"The IPCC's conclusion is purely political, and yours purely based on faith."

That's an unfounded slur. At least, it seems to be, given that you provide no supporting evidence.

You also provide a quote referencing the TAR which discusses the previous IPCC report, the SAR, and conclude that this says something about our current state of knowledge. There are two things wrong with this:

1. The reference you quote in the TAR to the SAR speaks nothing of the TAR's conclusions.

2. Why are you quoting the TAR when we've got 4AR? Which, btw, states the following with regard to attribution of greenhouse gas forcing:

"Most of the observed increase in global average
temperatures since the mid-20th century is very
likely due to the observed increase in anthropogenic
greenhouse gas concentrations."

If you disagree with this certainty estimate perhaps you should explain why? It's not clear to me why anyone would think this statement is false.

Paul H,

Thank you for your comments.

When I said "The IPCC's conclusion is purely political, and yours purely based on faith.", I did not mean it as a slur on Coby and I stand by the statement that the IPCC's conclusion is purely political.

Their conclusions are always political vetted line by line by the political bosses of the IPCC.

But this is not the only reason why I say this.

In the link you sent me, (which are Guidance Notes for the Lead Authors of the IPCC), on making expert judgements they have advised: âBe prepared to make expert judgments AND EXPLAIN THOSE BY PROVIDING A TRACEABLE ACCOUNT OF THE STEPS USED TO ARRIVE AT ESTIMATES OF UNCERTAINTY OR CONFIDENCE FOR KEY FINDINGSâ

No where have they explained their key finding that global average temperatures have "very likely" increased due to increase in anthropogenic greenhouse gasses, much less provided a traceable account of the steps by which they have reached this conclusion.

As for Coby's belief - I think it is mistaken, and hence ultimately founded on faith rather than science.

The document I linked to expresses how the IPCC scientists use evidence to arrive at conclusions of differing certainty. If you read the IPCC chapter on attribution the evidence is laid out in a manner which enables you to see how they reach the conclusion we are discussing.

e.g. the document says "Be prepared to make expert judgments and explain those by providing a traceable account of the steps used
to arrive at estimates of uncertainty or confidence for key findings â e.g. an agreed hierarchy of information,
standards of evidence applied, approaches to combining or reconciling multiple lines of evidence, and
explanation of critical factors.
"

It's clear to me, at least, that chapter 9 carries this out when attributing late 20th C warming to GHG increases and other sources of RF. For instance, they indicate that the consistency of conclusion from multiple independent lines of evidence creates a robust overall conclusion with regard to GHG radiative forcing attribution i.e. in the IPCC's language, a series of conclusions derived from independent sources, indicating the same overall conclusion, with more limited certainty (only likely), add up to robust conclusion with high certainty. The chances of all of these lines of evidence being flawed is very low. If you only read part of it, read the summary 9.7, which is a great example of this method in action. Now you know that this statement is derived from the certainty placed on the interpretation of other lines of evidence why don't you look at how those conclusions are reached. It's explained in detail in this chapter.

As a matter of fact:

1) The 19-page AR4 IPCC SPM is a political document, since it has to be agreed on, sentence-by-sentence by all governments, including for example, the USA, Russia, and Saudi Arabia.

That tends to water down the conclusions. Actually, it's pretty amazing that anything gets through, especially when one talks to the scientists who participate.

2) Of course, that is why the *scientists* write not only the full reports, but for each a Technical Summary, which for WG I is 73 pages, or one can go to the full report, which is ~900 pages, or the peer-reviewed references, which are vast. They even do a pretty good job of introductory material (FAQ's, boxes). Even the TS alone ought to answer most rational people's questions.

3) Coby is quite right.

By John Mashey (not verified) on 21 Jul 2009 #permalink

It seems to have been forgotten that chaos theory got its (recent) start through the problem of climate modeling. Tiny disturbances can have huge effects.

In addition, if the models are run backward, do they "predict" the past well over the last, say, 100 years?

By Peregrine (not verified) on 02 Jan 2010 #permalink

It seems to have been forgotten that chaos theory got its (recent) start through the problem of climate modeling.

Sigh ... the confusion of weather with climate extends even to this?

"An early pioneer of the theory was Edward Lorenz whose interest in chaos came about accidentally through his work on weather prediction in 1961.[37] Lorenz was using a simple digital computer, a Royal McBee LGP-30, to run his weather simulation. He wanted to see a sequence of data again and to save time he started the simulation in the middle of its course. He was able to do this by entering a printout of the data corresponding to conditions in the middle of his simulation which he had calculated last time."

See? Weather prediction, not climate modeling.

We can make many, many climate prediction like ... this January in Portland, Oregon will be on average colder than next July. Even though I can't predict the weather for each day in either month.

In addition, if the models are run backward, do they "predict" the past well over the last, say, 100 years?

Someone doesn't understand how models work. You don't run them backwards.

However some models do a pretty good job of hindcasting not only in the recent past but various paleoclimate scenarios, too.