Moderated Global Warming Discussion

[Mikemcc]

The Munich Re report is a major new event. Many casual deniers (those who simply can't believe it's true for ideological reasons rather than any other reasonable doubt) may possibly be swayed by economic rather than scientific evidence.

 

[lomiller]

Was it news to you that climate models generally do better than statistical models?

It looks to me as though you're trusting what lomiller and the denier blogs said instead of reading the research literature for yourself.

Don't try to link me to denier blog sites.


My objections are not a function of whether I agree or disagree with the contents of the paper. Publication in minion off discipline journals is VERY suspicions tactic and should raise red flags regardless of whether you agree with the paper or not.

As it turns out my initial suspicion was right on the mark. We can be pretty sure at this point the papers authors lacked the requisite understanding of the subject because it they clearly didn't understand why these decadal scale predictions would be run nor did they contribute anything of substance to either climate modeling of forecasting.

 

[W.D.Clinger]

Was it news to you that climate models generally do better than statistical models?

It looks to me as though you're trusting what lomiller and the denier blogs said instead of reading the research literature for yourself.

Don't try to link me to denier blog sites.

I am not suggesting your claims had anything to do with ideology or science.

Although Stephen McIntyre and the denier blogs want us to believe that purely statistical models (and specifically random walks) predict decadal climate changes at least as well as models based on climate science, that belief is contradicted by two of the journal articles I have discussed in some detail within this thread.

Despite that contrary evidence, you (and CapelDodger) appear to have accepted that aspect of McIntyre's argument. Here are two examples, with my highlighting:

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?

Again, the underlying physics system behaves as a random walk for these time scales so they are of no value in validating model results.

I've been trying to bring a relevant fact to your attention: Two different journal papers have tested that belief and found that random walks do not really predict climate changes as well as science-based models, even on decadal time scales.

 

[lomiller]

I've been trying to bring a relevant fact to your attention: Two different journal papers have tested that belief and found that random walks do not really predict climate changes as well as science-based models, even on decadal time scales.

You are still not getting it. What the people at DePreSys have shown is that if you know initial conditions of slower changing phenomenon like ENSO, climate models can make some types of predictions with some skill.


While interesting it has no direct barring on the real function of the climate models because they are used and measured in a completely different way.

The paper you are supporting does not add to or contribute to this finding in any meaningful way. The DePreSys team already did suitable skill calculations.



The main expected benefit of this type of work is better representation of ENSO and other such phenomenons in climate models, but it isn't certain that being able to predict the state of ENSO 2 years in the future will improve the predictions for the distribution of ENSO 50 years from now, which is what these models are designed to do.

The authors of you paper seem to completely miss the distinction and seem to think these decadal scale predictions are somehow a metric that impacts on the ability to make longer term predictions. I've even shown you a published response for a climate modeler doing just this type of decadal modeling pointing this out.

My original complaint, was that this sort of error and misunderstanding would not be caught by people with no experience with physical modelling publishing in an off discipline journal. I've subsequently shown you that the problems are indeed present, yet you still don't seem to understand that this tactic isn't just a recipe for bad papers it's something people looking to publish bad papers do deliberately. You should be highly suspicions of any paper appearing in off discipline journals.

 

[W.D.Clinger]

You are still not getting it. What the people at DePreSys have shown is that if you know initial conditions of slower changing phenomenon like ENSO, climate models can make some types of predictions with some skill.

I understand your argument, but I don't buy it. You're arguing that discussion of DePreSys has no scientific value unless the discussion is restricted to the specific purpose you aver.

I've been pointing out that your argument isn't valid. In particular, the Fildes and Kourentzes paper refutes a specific claim that you have made repeatedly in this thread. That refutation of your claim has scientific value.

The false claim you have been repeating in this thread has been popularized by denier blogs, where it is often cited as scientific evidence against AGW. Some of those blogs have gone so far as to cite the Fildes and Kourentzes paper as support for your claim. That's amusing because the paper actually demolishes your claim.

Whether the paper's scientific value approaches its amusement value is questionable, but I'm content to point out that the paper did make a scientific contribution by refuting your claim.

 

[Malcolm Kirkpatrick]

Trying to pull an improvised primer...your deeply mistaken...your manipulation skills are noted...it is all you got...you are caught saying some bull...your distorted temperatures...you reply every message as if the original was speaking of something else -feigning the fool-and try to start some silly debate...such crappy ramblings...

...the smoke screen of choice...

Nope. The AGW faithful standards of argument around here haven't improved.

You can stir words as much as you want, but you failed to provide valid arguments. Clearly, no matter how important empirical evidence is, your examples remain dead wrong.[/QUOTE]

 

[aleCcowaN]

Nope. The AGW faithful standards of argument around here haven't improved.

You can stir words as much as you want, but you failed to provide valid arguments. Clearly, no matter how important empirical evidence is, your examples remain dead wrong.

Malcolm, you should have answered as Westwall persona. Be careful and consistent and at least don't mix avatars. About your arguments, we lost hope long time ago.

Again, I'm correcting here your mistakes using the post editor:

Nope. The AGW faithful standards of argument around here haven't improved.

You can stir words as much as you want, but you failed to provide valid arguments. Clearly, no matter how important empirical evidence is, your examples remain dead wrong.

Malcolm, you know it's pretty obvious to everyone here that your call for "quality" on the debate came after you a)failed to provide any valid argumentation and felt in rhetorical loops, b) you got several infractions after breaking several forum rules and throwing rhetorical tantrums.

I hope that once tomorrow is ended you may come back developing your Westwall-like style further.

 

[lomiller]

I understand your argument

I really don’t think you do.

The work of DePreSys is of moderate interest to climate modlers in their constant quest to find ways to make their models better. As a topic of general interest on climate change than no it’s not of any direct value to the science.

In particular, the Fildes and Kourentzes paper refutes a specific claim that you have made repeatedly in this thread. That refutation of your claim has scientific value.

I’ll certainly concede that that statement isn’t generically true, though in the context of climate predictions and how climate modeling it remains valid. In the context of weather predictions (which is essentially what DePreSys is doing) DePreSys performed all the requisite skill testing so the Fildes and Kourentzes paper is producing anything novel or new.


At this point it may be worthwhile discussing climate modeling vs weather modeling, and why the work DePreSys is doing is much closer to the latter. This is a distinction Fildes and Kourentzes most definitely missed and demonstrates why we should have misgivings about people publishing outside their specialty in an off-discipline journal. Naturaly the scepticism should remain even if NO errors were present, this just isn’t a trustworthy mecanism

Weather models start from known existing conditions and try to predict a specific realization of the system. Because the system is chaotic, however, the difficulty of predicting the specific state of a realization becomes exponentially more difficult the farther out you try and look. For a normal weather model a few days is about as far as you can get before it starts to break down. DePreSys is tracking different, slower changing phenomenon like ENSO, NAO, Arctic Oscillation etc, which is why they can have some success but they are still trying to predict the state of a specific realization.

Deniers typically try to present climate models as being similar, and thus subject to the same issues with time horizons, and Fildes and Kourentzes makes this same error. How climate models are used in practice, however, is somewhat different.

Rather than trying to predict the state of a specific realization they try to generate different realizations of the same physical system. While the system remains chaotic different realizations of the same system share the same attractor and, barring non-linearity, will track that attractor if something changes it. The distribution of the different realizations around this attractor will essentially be random, but over time a trend can still be discernable over this noise.

Of the highest importance for a climate model then, isn’t the ability to make point in time predictions but to get the attractor and the distribution of realizations around that attractor right. While this is generally accomplished by making the physical model more accurate and resolving to a smaller grid cost vs benefit will end up favouring some optimizations over others for climate modeling and others for weather modeling.

While it’s encouraging to see climate models can predict things like El Nino with skill over a 1-4 year period this is not, nor will ever be the basis for trusting that climate models are getting it right over longer periods. This is a point that was completely missed by Fildes and Kourentzes.

 

[macdoc]

Nope. The AGW faithful standards of argument around here haven't improved.

You can stir words as much as you want, but you failed to provide valid arguments. Clearly, no matter how important empirical evidence is, your examples remain dead wrong.

[/QUOTE]

How's that falsification coming there MK?? Significantly silent on that front...after all - that is the heart of the matter despite your skirting of the issue.

The fossil fuel scientists called it "irrefutable" when reporting to Exxon et al back in 1995.
Yet you have the hubris to think you know better.
•••

Would be nice to get off the dead end nattering about the reality ...since that is ALL the deniers are after.
and move the discussion on to the policy choices facing individuals, regions and nations regarding both coping and mitigation.

some unlikely allies in the process....

The Pentagon's drive for green energy represents a tremendous opportunity. If the military meets its targets, it could transform the energy landscape to everyone's benefit. When it comes to creating markets for new technologies, the Pentagon's procurement machine has no equal. If it decides to pump money into green energy, the economics suddenly look more favourable. They don't call it the military-industrial complex for nothing.

http://www.newscientist.com/article...itary-is-a-useful-ally-on-climate-change.html

 

[macdoc]

While it’s encouraging to see climate models can predict things like El Nino

can you reference me on that - I have a hard time seeing how ocean circulation is predictable given the ongoing changes.....
I can see climate models predicting the impact of El Nino - but the emergence???

 

[lomiller]

can you reference me on that - I have a hard time seeing how ocean circulation is predictable given the ongoing changes.....
I can see climate models predicting the impact of El Nino - but the emergence???

Keep in mind the context here is 1-4 years with known initial conditions and that ENSO is just an example of the things than make predictions on this scale possible.

http://www.metoffice.gov.uk/research/modelling-systems/unified-model/climate-models/depresys

Some of this natural variability is potentially predictable months or even years in advance because it is related to relatively slow processes in the ocean, such as El Niño, fluctuations in the thermohaline circulation, and large-scale anomalies of ocean heat content. Decadal forecasts therefore attempt to predict natural variability in addition to externally forced changes. This is achieved by starting a climate model from the current observed state of the climate system, as well as specifying changes in anthropogenic sources of greenhouse gases and aerosol concentrations and projected changes in solar irradiance and volcanic aerosol.

 

[Reality Check]

Nope. The AGW faithful standards of argument around here haven't improved.

Let us see if your climate change denier (see I can make dumb assumptions too ) standards of argument have improved:
Malcolm Kirkpatrick:

Originally Posted by Reality Check
Malcolm Kirkpatrick: Species of Foraminifera could not adapt to temperature changes that took thousands of years.
What do you think will happen to Foraminifera when similar temperature changes take place over a couple of centuries or even decades?

Revised question from 10th September 2012

(originally What do you think will happen to Foraminifera when CO2 changes over decades?)

57 days and counting.

It is fairly ignorant to call people who have read the evidence for AGW "the faithful" when the scientific evidence is overwhelming that global warming is happening and that we are the dominant driver. This is nothing to do with faith.

 

[macdoc]

Some of this natural variability is potentially predictable months or even years in advance because it is related to relatively slow processes in the ocean, such as El Niño, fluctuations in the thermohaline circulation, and large-scale anomalies of ocean heat content. Decadal forecasts therefore attempt to predict natural variability in addition to externally forced changes. This is achieved by starting a climate model from the current observed state of the climate system, as well as specifying changes in anthropogenic sources of greenhouse gases and aerosol concentrations and projected changes in solar irradiance and volcanic aerosol.

my reading of this is not that they are predicting an ENSO events ( like perhaps the back to back La Nina a couple years back ) but they can take into account Enso events just as they can take into account a volcanic event but cannot predict one.

 

[W.D.Clinger]

I understand your argument

I really don’t think you do.

It's possible that I don't, but you continue to offer evidence that I do.

In particular, the Fildes and Kourentzes paper refutes a specific claim that you have made repeatedly in this thread. That refutation of your claim has scientific value.

I’ll certainly concede that that statement isn’t generically true,

Thank you. Now we need only get the denier blogs to concede the point.

Deniers typically try to present climate models as being similar, and thus subject to the same issues with time horizons, and Fildes and Kourentzes makes this same error.

You realize, I hope, that DePreSys uses HadCM3.

While it’s encouraging to see climate models can predict things like El Nino with skill over a 1-4 year period this is not, nor will ever be the basis for trusting that climate models are getting it right over longer periods. This is a point that was completely missed by Fildes and Kourentzes.

You write as though you were unaware that Fildes and Kourentzes found DePreSys to be clearly superior to random walks over a ten-year period as well.

lenny asked whether you had even read the Fildes and Kourentzes paper. Your avoidance of that question has not escaped notice.

Keep in mind the context here is 1-4 years with known initial conditions and that ENSO is just an example of the things than make predictions on this scale possible.

Once again, you write as though you are unaware that Fildes and Kourentzes also considered a ten-year interval and obtained a similar result.

 

[lomiller]

You realize, I hope, that DePreSys uses HadCM3.

Umm yes, and deniers try to claim all climate models, including HadCM have the same issues with modeling chaotic behaviour as weather models do. In this case DePreSys is essentially using HadCM as a weather model to look at longer scale phenomenon as detailed in my post above.

What Fildes and Kourentzes failed to realize is that HadCM is not being used in its normal operation so it’s not a suitable test for how well the model performs.

You write as though you were unaware that Fildes and Kourentzes found DePreSys to be clearly superior to random walks over a ten-year period as well.

Irrelevant as DePreSys only passed tests for skill in the first 1-4 years. If neither method generates a skilful prediction it doesn’t really matter which is “better”.

In fact it outright doesn’t matter which is better since there is no compelling case for either being particularly useful outside of model tuning. If you could, make skilful predictions of ENSO 2 years out maybe there would be some use in relating that to regional droughts or weather patterns, but thus far at least this doesn’t appear to be the case.

 

[macdoc]

I'd be mighty impressed with a consistent 2 year prediction of ENSO events.
They are critical for both South America and Australia and important for North America.

I really do fail to see how that would be possible. Seems 6 months trending is the situation now based on observation.

 

[W.D.Clinger]

You write as though you were unaware that Fildes and Kourentzes found DePreSys to be clearly superior to random walks over a ten-year period as well.

Irrelevant as DePreSys only passed tests for skill in the first 1-4 years.

Hmm.

Should I believe you? Or should I believe the guys who published this in Science:

....We present a new modeling system that predicts both internal variability and externally forced changes and hence forecasts surface temperature with substantially improved skill throughout a decade, both globally and in many regions....

....We examined the potential skill of decadal predictions using the newly developed Decadal Climate Prediction System (DePreSys), based on the Hadley Centre Coupled Model, version 3 (HadCM3) (17), a dynamical global climate model (GCM)....

....For 5-year means, the RMSE was reduced by 38% (a 61% reduction in E), from 0.106°C to 0.066°C; and for 9-year means, the RMSE was reduced by 49% (a 74% reduction in E), from 0.090°C to 0.046°C....

....We find that the increased skill of DePreSys over NoAssim is consistent with an improved ability to predict El Niño for the first 15 to 18 months, but not at longer lead times....

 

[macdoc]

Well we are 5 years on and I wonder just how well they predicted the ENSO episodes during that period.

"Our system predicts that internal variability will partially offset the anthropogenic global warming signal for the next few years," the authors wrote. "However, climate will continue to warm, with at least half of the years after 2009 predicted to exceed the warmest year currently on record."
Read more at http://news.mongabay.com/2007/0809-climate_model.html#JISdFds8DDWSc58X.99

I'd be very curious as to how they are coming up with the ENSO predictions.

This would seem to indicate - 5 years on - that there is little progress.

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 117, D20111, 17 PP., 2012
doi:10.1029/2012JD018004

Identifying the causes of the poor decadal climate prediction skill over the North Pacific

Key Points
The decadal climate prediction skill is particularly low in the North Pacific
Two major warmings around 1963 and 1968 are missed by the forecast systems
Their failure is most likely due to their ocean stratification biases

http://www.agu.org/pubs/crossref/2012/2012JD018004.shtml

 

[W.D.Clinger]

Well we are 5 years on and I wonder just how well they predicted the ENSO episodes during that period.


I'd be very curious as to how they are coming up with the ENSO predictions.

Smith et al. never said they were able to predict ENSO more than 18 months out, and their ability to predict ENSO at that time scale was very limited.

It was lomiller who attributed the predictive skill of DePreSys entirely to its ability to predict ENSO. He seems to have been mistaken about that, just as he seems to have been mistaken in claiming that DePreSys had no predictive skill beyond four years.

(In looking for a source that agrees with lomiller about the latter, I ran across a blog entry written by Roger Pielke, Sr. Although I was suitably impressed by Dr Pielke's exclamation mark, I did not run across any supporting evidence when I followed the links Pielke provided in support of his claim. Perhaps I did not dig deeply enough.)

Smith et al. attribute the predictive skill at decadal time scales to other factors, such as:

....we conclude that the improvement of DePreSys over NoAssim in predicting T_s on interannual-to-decadal time scales results mainly from initializing upper ocean heat content.

Although that could be misread as a reference to ENSO, they explicitly disclaim any improvement in predicting ENSO beyond 15 to 18 months, and I quoted that disclaimer in my previous post.

In the following quotation, H is "global annual mean ocean heat content in the upper 113 m":

We therefore conclude that the increased predictive skill of DePreSys over NoAssim at forecast periods longer than 15 months results mainly from initializing the low-frequency variability of H, thereby removing errors of H from the NoAssim initial conditions (SOM text).

For more details, read the paper and its supporting online material (SOM). Note also that a followup paper has been accepted by Geophysical Research Letters.

 

[macdoc]

My sense of this is that rather than predictive of ENSO it takes the events into consideration as it does with volcanoes ( which are inherently unpredictable ) and shows the outcomes more accurately.

I'd like to see hindcasting results from DePreSys as any model which allows for a tighter range of outcomes for specific regions is very valuable.

It reminds me of how NOAA shifted their hurricane model a few years back with superb results in both hind casting and forecasting of seasonal frequency.

These only become useful when there is some confidence that the resulting forecasts have a decent degree of validity.