Moderated Global Warming Discussion

[Westwall]

Actually, the criticism is mostly fair. There is scant overlap between his focus and the field of applied mathematical modeling of complex environmental systems. Grid size, the appropriate mathematical representation of physical and chemical systems, boundary conditions, translation of real data into model inputs, and on and on, are all important issues that a pure mathematician would have no more expertise on than a 7-11 clerk. "Mathematics" is much too broad a field to claim expertise in - or even working knowledge of.

Except when he has been proven correct as in the Gergis case. The models themselves are very poor. The Journal of Forecasting compared them (in a study published in 2011) with a Random Walk and the random walk was far more accurate than the models.

A perfect model gets a score of zero. A score above 1 did worse than uninformed guesses. The climate models got scores ranging from 2.4 to 3.7 as compared to straight statistical forecasts with no element of climatology which obtained results of .8 to 1, though a few scored as high as a 1.8....still significantly better than any climate model was able to accomplish.

Forgive me if I find reliance on the computer models to be silly but resluts like that are worse than Sylvia Brown gets and she's a charlatan as we all know.

....But she's more accurate than the computer models....and that is sad.

 

[bit_pattern]

Yes, I wonder how they arrived at that conclusion with no empirical evidence to support it.

What are you even talking about Care to support this claim with some evidence of your own?

 

[RandFan]

The warming of the present time (begun over 150 years ago) is nothing more than the regular cycles of warming and cooling that have been experienced since the end of the last ice age.

And Noah built a great big boat and took the Dinosaurs on a cruise. What good is science when you can just say anything you want? No need to make a case using evidence.

 

[bit_pattern]

And I would agree with you about the warming.. It's the cause of that warming where we differ. There is now ample evidence (over 100 peer reviewed papers) that the MWP and the RWP were global in extent and warmer than the current warming.

Right. So you should have no problem citing a few that you find the most convincing for critical evaluation by the forum then, will you? I dont know how long you've been following this thread for but I'm almost certain a anything you can pug up has already been discussed at length in the previous >6000 posts.

The warming of the present time (begun over 150 years ago) is nothing more than the regular cycles of warming and cooling that have been experienced since the end of the last ice age.

A bold assertion to make with no supporting evidence. Can you please outline what drives this 'natural' cycle, after all the climate doesn't change by magic, if there us some mysterious forcing causing a natural 'recovery' from an ice age you should be able to identify what it is. Without an alternative explanatory model you may as well be chasing ether or the god if the gaps.

 

[lomiller]

Yes, I wonder how they arrived at that conclusion with no empirical evidence to support it.

Yeah papers with no evidence to support them get published in Science, damn that American Association for the Advancement of Science and their shoddy publications...

 

[aleCcowaN]

Yes, I wonder how they arrived at that conclusion with no empirical evidence to support it.

Knock it off with your "empirical evidence" cliché, for Santa's sake! You have come to think that using that gives you some intellectual preponderance and in fact you don't have the faintest idea of what you are saying. What evidence do you have that there are no immortal humans? 8% of all the humans that have existed ever are still alive today (including Betty White). Isn't this "empirical evidence"?

Your attempted use of supposed empirical evidence included taking two extremely similar planets separated by 50 million years (time or light) and saying that as in one of them the set of species in a certain class is doing OK then the different set of species of the same class in the sister planet will do well if conditions change quickly and the second planet become almost a copy of the first one. That's not empirical evidence but not knowing how to aim the watering can at the daisies, so to speak.

 

[W.D.Clinger]

Except when he has been proven correct as in the Gergis case. The models themselves are very poor. The Journal of Forecasting compared them (in a study published in 2011) with a Random Walk and the random walk was far more accurate than the models.

FYI: The International Journal of Forecasting is a journal. It publishes research papers. It doesn't write them, and it doesn't do research itself.

A perfect model gets a score of zero. A score above 1 did worse than uninformed guesses. The climate models got scores ranging from 2.4 to 3.7 as compared to straight statistical forecasts with no element of climatology which obtained results of .8 to 1, though a few scored as high as a 1.8....still significantly better than any climate model was able to accomplish.

As summaries of technical papers go, that's pretty bad. (Yes, I know you plagiarized that summary from a denier site. See below.)

I doubt very much whether you've looked at the paper itself. It would surprise me if you've so much as looked at its abstract.

Here's what a proper citation looks like:

Robert Fildes and Nikolaos Kourentzes. Validation and forecasting accuracy in models of climate change. International Journal of Forecasting, Volume 27, Issue 4, October-December 2011, pages 968-995. http://dx.doi.org/10.1016/j.ijforecast.2011.03.008​

Here's the abstract:

Forecasting researchers, with few exceptions, have ignored the current major forecasting controversy: global warming and the role of climate modelling in resolving this challenging topic. In this paper, we take a forecaster’s perspective in reviewing established principles for validating the atmospheric-ocean general circulation models (AOGCMs) used in most climate forecasting, and in particular by the Intergovernmental Panel on Climate Change (IPCC). Such models should reproduce the behaviours characterising key model outputs, such as global and regional temperature changes. We develop various time series models and compare them with forecasts based on one well-established AOGCM from the UK Hadley Centre. Time series models perform strongly, and structural deficiencies in the AOGCM forecasts are identified using encompassing tests. Regional forecasts from various GCMs had even more deficiencies. We conclude that combining standard time series methods with the structure of AOGCMs may result in a higher forecasting accuracy. The methodology described here has implications for improving AOGCMs and for the effectiveness of environmental control policies which are focussed on carbon dioxide emissions alone. Critically, the forecast accuracy in decadal prediction has important consequences for environmental planning, so its improvement through this multiple modelling approach should be a priority.

That sounds like an interesting paper, but your summary of it was awful. If you want us to blame that summary's awfulness on the denier site you plagiarized, then you should link to or otherwise give credit/blame to the denier site.

If you hope to discuss the paper intelligently, you'll have to read it.

 

[lomiller]

There is now ample evidence (over 100 peer reviewed papers) that the MWP and the RWP were global in extent and warmer than the current warming.

Go ahead and show us these peer reviewed reconstructions supporting this. Please stick to reputable high profile journals this time.

There is good evidence for global changes in climate during the medieval period, but not global temperature increases. Every peer reconstruction that can show a statistically significant conclusion says the current temperatures are warmer than the MWP.

The Roman period is less clear because the error bars are bigger so most reconstructions can't conclude which is warmer, but most suggest current temperatures are warmer but can't say with high certainty. There are no reconstructions saying the Roman period was warmer.

 

[aleCcowaN]

Someone can also explain in a better English than mine about gas solubility decreasing with temperatures and why warm waters are so transparent and not the ideal place for fisheries. Opening either a Coke, beer or champaign will show what happens with CO2 in water when temperature is lower or higher. Certainly, having a bottle of warm champaign and removing the cork to see it foamy and spilled all over the place is not an example of what is going to happen to the seas in the -inescapable- event of more warming, but everyone can think of the seas absorbing CO2 in the actual conditions and deduct what is going to happen with that capacity if more CO2 drives to a warmer world with warmer oceans.

I was taught that the relative infertility of tropical oceans relative to sub-polar oceans related to bacteria.
Your champagne froths in response to a decrease in pressure. True, warm beers will froth more than cold ones.
Oceans will release dissolved gasses in response to warming from other causes (e.g., increase in insolation). This suggests to me that feedbacks are not "runaway" (one-directional) or the Earth would have seen the predicted warming crisis before.

More evidence of you not being able to manage a dynamic analysis nor a systemic one. You just made a pasticcio of static images and failed to see that I was implying a decreasing role of the oceans as carbon sink.

 

[lomiller]

FYI: The International Journal of Forecasting is a journal. It publishes research papers. It doesn't write them, and it doesn't do research itself.


As summaries of technical papers go, that's pretty bad. (Yes, I know you plagiarized that summary from a denier site. See below.)

I doubt very much whether you've looked at the paper itself. It would surprise me if you've so much as looked at its abstract.

Here's what a proper citation looks like:

Robert Fildes and Nikolaos Kourentzes. Validation and forecasting accuracy in models of climate change. International Journal of Forecasting, Volume 27, Issue 4, October-December 2011, pages 968-995. http://dx.doi.org/10.1016/j.ijforecast.2011.03.008​

Here's the abstract:




That sounds like an interesting paper, but your summary of it was awful. If you want us to blame that summary's awfulness on the denier site you plagiarized, then you should link to or otherwise give credit/blame to the denier site.

If you hope to discuss the paper intelligently, you'll have to read it.

Perhaps, but already there should be some red flags.

First this is a journal for statistical modeling. Climate models are physical models, which are very different beasts This difference opens the editors and reviewers to missing what would be very obvious errors.

Second it tests against regional predictions. No major climate model bills itself as being able to make accurate regional predictions, so what is the rational for this?

Third. They look at decade time scales. On these time scales the underlying physical behaves as a random walk with a too small to be detected trend. Predicting it with better accuracy than a random walk should not be expected so again they are testing something no one has suggested climate models ca do. Why?

 

[bit_pattern]

The Journal of Forecasting us NOT a reputable publication, it was founded by a climate science denier with a background in advertising & marketing

http://en.m.wikipedia.org/wiki/J._Scott_Armstrong

 

[W.D.Clinger]

Perhaps, but already there should be some red flags.

Agreed. Here's one:

The Journal of Forecasting us NOT a reputable publication, it was founded by s climate science denier with s background in advertising & marketing

http://en.m.wikipedia.org/wiki/J._Scott_Armstrong

Yes, Armstrong is quite a notorious denier, as is his co-author Willie Soon.

I don't recognize Fildes as a known denier, but his editorship of the journal did overlap with Armstrong's. You can make of that what you will.

First this is a journal for statistical modeling. Climate models are physical models, which are very different beasts This difference opens the editors and reviewers to missing what would be very obvious errors.

That's true.

It also works both ways. The statisticians are likely to miss errors that would be obvious to the physical scientists, and the physical scientists are likely to miss errors that would be obvious to the statisticians.

I happen to be interested in statistical criticisms of the physical models, even if some of those criticisms come from deniers. In my opinion, a few of those criticisms are valid, while many others are not. The validity and importance of statistical criticisms are routinely exaggerated by deniers, and almost as routinely denied by climate scientists. As you should recall, however, the NRC report agreed that climate research would be strengthened by collaborations with statisticians. That should not be controversial.

I haven't yet read the Fildes and Kourentzes paper myself, because it's behind a paywall, and I don't want to pay for it when I can download it for free on Monday.

In the meantime, I looked at their slides on "The role of decadal forecasting exercises in the validation of global circulation simulation models". These slides cite their journal paper, but they don't look anywhere near as bad as Westwall made their research sound. There are some red flags, however; one is that they cite Green and Armstrong (on slide 9) but never get around to citing the McShane/Wyner paper from Annals of Applied Statistics.

I also read two four-page comments on the paper that were published in the same issue, one by Patrick E McSharry and the other by Noel S Keenlyside. Both explain the interest in decadal predictions, and both offer measured praise for the Fildes/Kourentzes paper while offering constructive criticism. For example:

As a result of the challenges from skeptics, some climate scientists have become overly defensive and reluctant to engage in debate, which unfortunately undermines their position. Rather than communicate the many sources of uncertainty openly, there has been a temptation to play down some of the limitations of climate models. In addition, anyone questioning the adequacy of climate model risks been labelled a skeptic. It is hoped that the critical analysis undertaken by Fildes and Kourentzes will encourage researchers to participate in the climate change debate and ensure that each assumption is carefully tested.

Decision-makers thrive on certainty. The substantial levels of uncertainty surrounding global warming forecasts should not be used as a reason for inaction. While uncertainty about the rate of warming is high, there exists a scientific consensus, expressed by the IPCC and confirmed in the forecast comparison of Fildes and Kourentzes, that global temperatures will increase by 0.1–0.2 ^◦C per decade....

On the basis of those comments by McSharry and by Keenlyside, together with the abstract and slides by Fildes and Kourentzes, I'm fairly confident that Westwall misrepresented the paper.

 

[lomiller]

It also works both ways. The statisticians are likely to miss errors that would be obvious to the physical scientists, and the physical scientists are likely to miss errors that would be obvious to the statisticians.

If a climate scientist tried to write a paper on an actively researched topic in statistics, and then published it in a climate journal then yes the same type of red flags would certainly be raised. in fact any time someone attempts research outside their field and publishes it in a journal outside their field where people are unfamiliar with the active research it should raise red flags.

This is the situation I am referring to and it most certainly different from merely spotting errors in established technique or science. While I would certainly expect a statistician to be better at spotting misuse of established statistical techniques, I would certainly not expect them to be the only ones expert in using those techniques.

I happen to be interested in statistical criticisms of the physical models, even if some of those criticisms come from deniers.

TBH I'm not even sure what this means. Physical models are not extrapolated from existing data using statistical techniques the way statistical models are. What interesting statistical criticisms do you expect to find when statistics haven't been employed?

As you should recall, however, the NRC report agreed that climate research would be strengthened by collaborations with statisticians. That should not be controversial.

Neither is it relevant to this situation. Again, since physical models are not build on statistics, what would collaboration with statisticians statisticians bring to them?

In the meantime, I looked at their slides on "The role of decadal forecasting exercises in the validation of global circulation simulation models". These slides cite their journal paper, but they don't look anywhere near as bad as Westwall made their research sound. There are some red flags, however; one is that they cite Green and Armstrong (on slide 9) but never get around to citing the McShane/Wyner paper from Annals of Applied Statistics.

Again, the underlying physics system behaves as a random walk for these time scales so they are of no value in validating model results. Other than possibly ENSO or similar phenomenon you would not expect to make usable predictions on these time scales nor does anyone attempt to. Showing you can't is therefor a trivial result that little to the discussion.

 

[SezMe]

TBH I'm not even sure what this means. Physical models are not extrapolated from existing data using statistical techniques the way statistical models are. What interesting statistical criticisms do you expect to find when statistics haven't been employed?

I agree with your point that mathematical models are not usually amenable to statistical analysis. But there might be some cross-talk between the two disciplines if a mathematical model is using Monte Carlo methods to, for example, handle unknown variability in input parameters. That said, I don't think W.D.Clinger is commenting on that possibility.

 

[W.D.Clinger]

I'll post a real response on Monday evening, after I've had a chance to read the paper.

What interesting statistical criticisms do you expect to find when statistics haven't been employed?


Neither is it relevant to this situation. Again, since physical models are not build on statistics, what would collaboration with statisticians statisticians bring to them?

Are you really denying the relevance of statistics to physical sciences, and to climate models in particular?

If so, then you're arguing with these people:

Problems in climate research such as statistical climate reconstruction require sophisticated statistical approaches and a thorough understanding of the data used.

If you can't figure out who wrote that, I'll tell you Monday evening.

 

[aleCcowaN]

...

Are you really denying the relevance of statistics to physical sciences, and to climate models in particular?

If so, then you're arguing with these people:




If you can't figure out who wrote that, I'll tell you Monday evening.

The quote relates to proxy reconstructions, for instance, using pseudoproxy tests. Why on earth are you using it in a context of climate modelization?

 

[macdoc]

Perhaps Mr. Clinger needs some clarification from people who do this for a living.

# What is the difference between a physics-based model and a statistical model?

Models in statistics or in many colloquial uses of the term often imply a simple relationship that is fitted to some observations. A linear regression line through a change of temperature with time, or a sinusoidal fit to the seasonal cycle for instance. More complicated fits are also possible (neural nets for instance). These statistical models are very efficient at encapsulating existing information concisely and as long as things don’t change much, they can provide reasonable predictions of future behaviour. However, they aren’t much good for predictions if you know the underlying system is changing in ways that might possibly affect how your original variables will interact.

Physics-based models on the other hand, try to capture the real physical cause of any relationship, which hopefully are understood at a deeper level. Since those fundamentals are not likely to change in the future, the anticipation of a successful prediction is higher. A classic example is Newton’s Law of motion, F=ma, which can be used in multiple contexts to give highly accurate results completely independently of the data Newton himself had on hand.

Climate models are fundamentally physics-based, but some of the small scale physics is only known empirically (for instance, the increase of evaporation as the wind increases). Thus statistical fits to the observed data are included in the climate model formulation, but these are only used for process-level parameterisations, not for trends in time.

http://www.realclimate.org/index.php/archives/2008/11/faq-on-climate-models/

The entire article

FAQ on climate models

is a worthwhile read.

 

[lomiller]

I agree with your point that mathematical models are not usually amenable to statistical analysis. But there might be some cross-talk between the two disciplines if a mathematical model is using Monte Carlo methods to, for example, handle unknown variability in input parameters. That said, I don't think W.D.Clinger is commenting on that possibility.

Yes there may be some comparatively basic statistics used at some point in the process but these tools used by nearly everyone. Expecting climate scientists to consult with the statistical community on their Monte Carlo methods is a little out there and really a red herring.

We are discussing this because I mentioned this a s journal that looks at statistical modeling, not physical models so they are out of their sphere of expertise. What value are we expecting the statistical modeling community to bring to the table for a physical model?

Not to mention the fact that statistical models have a well deserved bad reputation. They work until they don't because they don't look at the underlying cause of the phenomenon they are observing.

 

[lenny]

Curry presents numerous anti-science topics on her blog.

can we have an example or two? something you find truly antiscience?

i pressed on some iffy points after her talk, she did not shy away. (or give gibberish/antiscience answers)

just to be clear: i am not defending her, but would like to see (a small piece of) the evidence behind your claim.

 

[Mikemcc]

can we have an example or two? something you find truly antiscience?

i pressed on some iffy points after her talk, she did not shy away. (or give gibberish/antiscience answers)

just to be clear: i am not defending her, but would like to see (a small piece of) the evidence behind your claim.

Curry's main approach is to harp on about uncertainity and how you can't trust other folks figures. Whenever she is challenged though she has been unable to put forwards any studies of her own that actually provides evidence for those claims.

http://scienceblogs.com/stoat/2010/09/21/attribution-errors/
http://julesandjames.blogspot.com/2010/10/she-who-refuses-to-do-arithmetic-is.html
http://thingsbreak.wordpress.com/2011/08/04/my-apologies-to-judith-curry/