quote:

Originally posted by John_M:

There are many kinds of modeling&simulation, and they are not the same. Physics-based modeling (such as climate models, airflow models, or automobile crash models) are so unlike JL's example that either:
(a) JL understands the difference and is being deliberately confusing in the example.
OR
(b) JL doesn't understand the difference and is over-generalizing from his sorts of models to these.

Giving the benefit of the doubt, I'll assume the latter.

JL appears to believe that the variables in climate models are simply parameterised with little or no application of actual physics, and that these parameters are tuned so that model runs match past real data.

quote:

JL said
The parameters have already been calculated from past data, so it ought to be a “good fit” to that data. That is no test, however, of how well it will perform in the future, because you can calculate parameters for variables which have no bearing on the subject (or random numbers), and they can still fit.

John_M and TrueSceptic:

Then we will have to agree to differ. The bottom line for me is that there has to be credible falsifiability, and at the moment AGW proponents have not convinced me that there is.

I have stated on numerous occasions what I would consider a reasonable test, and I cannot see that asking for some FUTURE numbers to compare with FUTURE observations is impossibly stringent.

Unless and until AGW proponents meet this perfectly reasonable request, I have to maintain a degree of scepticism - as any scientist should do.

quote:

Originally posted by JL(SFC58,AFCB):
John_M and TrueSceptic:

Then we will have to agree to differ. The bottom line for me is that there has to be credible falsifiability, and at the moment AGW proponents have not convinced me that there is.

I have stated on numerous occasions what I would consider a reasonable test, and I cannot see that asking for some FUTURE numbers to compare with FUTURE observations is impossibly stringent.

Unless and until AGW proponents meet this perfectly reasonable request, I have to maintain a degree of scepticism - as any scientist should do.

Yes but you cannot falsify a scenario as it isn't a prediction. However you can publish it in a so-called scientific document. Falsification is so last century, we're into post-normal science now...

quote:

Originally posted by JL(SFC58,AFCB):
John_M and TrueSceptic:

Then we will have to agree to differ. The bottom line for me is that there has to be credible falsifiability, and at the moment AGW proponents have not convinced me that there is.

I have stated on numerous occasions what I would consider a reasonable test, and I cannot see that asking for some FUTURE numbers to compare with FUTURE observations is impossibly stringent.

Unless and until AGW proponents meet this perfectly reasonable request, I have to maintain a degree of scepticism - as any scientist should do.

But do you agree that the models are not made in the way you thought?

Anyway, see if you can get some data from Hadley CRU and NASA GISS

quote:

Originally posted by TrueSceptic:
But do you agree that the models are not made in the way you thought?

Anyway, see if you can get some data from Hadley CRU and NASA GISS

No I don't - but I might not have expressed myself very well. I will just ask one question - how does the effect of cloud cover stopping sunshine reaching earth (and thereby reducing temperature) compare with the effect of cloud cover trapping heat beneath the cloud layer?

quote:

Originally posted by Lost in Kate Winslet:
Falsification is so last century, we're into post-normal science now...

Now who's being too sophisticated?!

quote:

Originally posted by JL(SFC58,AFCB):

quote:

Originally posted by TrueSceptic:
But do you agree that the models are not made in the way you thought?

Anyway, see if you can get some data from Hadley CRU and NASA GISS

No I don't - but I might not have expressed myself very well. I will just ask one question - how does the effect of cloud cover stopping sunshine reaching earth (and thereby reducing temperature) compare with the effect of cloud cover trapping heat beneath the cloud layer?

That question has nothing to do with your request for outputs from climate models. Why are you asking it now?

Perhaps Steve_M, John_M, or CobblyWorlds can answer it anyway.

quote:

Originally posted by Lost in Kate Winslet:
Yes but you cannot falsify a scenario as it isn't a prediction. However you can publish it in a so-called scientific document.

Two points here:

1 AGW theorists don't seem to have any difficulty in predicting big increases by the end of the 21st century though!

2 OK, there are different predictions for different levels of CO2 production. Shouldn't be tooooooo difficult to produce a table giving predicted outcomes year by year for different CO2 levels in the air. After all, they have managed to construct these wonderfully complex model.

quote:

Originally posted by TrueSceptic:
That question has nothing to do with your request for outputs from climate models. Why are you asking it now?

Perhaps Steve_M, John_M, or CobblyWorlds can answer it anyway.

The reason I ask for it now is to illustrate that the models ARE constructed from past data.

quote:

Originally posted by JL(SFC58,AFCB):

The reason I ask for it now is to illustrate that the models ARE constructed from past data.

Are you going to "validate" the outputs from the models or not?

JL,

I've finally got around to emailing Gavin Schmidt to ask for some goodness of fit stats regarding Hansen's model predictions from 1988. If you want the data yourself email him: gschmidt@giss.nasa.gov

I've also started a new thread to get the discussion going when I have some answers from Gavin.

Can I just summarise what I can see here so far in the debate. JL, you are making a legitimate request to see some model validation, fair dos. That is perfectly sensible. However, I do see a disingeneous element in your argumentation for a number of reasons. Principally, you seem to be seem to be suggesting that no attempt at validation has taken place. Let me empahsise, you seem to be suggesting this. I'll admit I could well be wrong in this assumption, but why else have you failed to search around in the literature to find any examples of such an effort?

This brings me to my next point, there has been a concerted effort to validate models by matching output to real data. One very obvious example which you are aware of is Hansen 2006 which is cited in this post on RealClimate which we have discussed:

http://www.realclimate.org/index.php/archives/2007/05/h...ns-1988-projections/

There are other examples too. Here's a few examples to get you started:

http://pubs.giss.nasa.gov/abstracts/2006/Hansen_etal_1.html

Climate response to volcanic forcing: Validation of climate sensitivity of a coupled atmosphere-ocean general circulation model

Yokohata T, Emori S, Nozawa T, Tsushima Y, Ogura T, Kimoto M

GEOPHYSICAL RESEARCH LETTERS 32 (21): Art. No. L21710 NOV 5 2005

----

Radiative fluxes in the ECHAM5 general circulation model

Wild M (Wild, Martin), Roeckner E (Roeckner, Erich)

JOURNAL OF CLIMATE 19 (16): 3792-3809 AUG 2006

---------

Combined observational and modeling based study of the aerosol indirect effect

Storelvmo T (Storelvmo, T.), Kristjansson JE (Kristjansson, J. E.), Myhre G (Myhre, G.), Johnsrud M (Johnsrud, M.), Stordal F (Stordal, F.)

ATMOSPHERIC CHEMISTRY AND PHYSICS 6: 3583-3601 SEP 4 2006

------

An evaluation of a general circulation model (GCM) and the NCEP-NCAR reanalysis data for winter precipitation in Greece

Tolika K, Maheras P, Flocas HA, Arseni-Papadimitrioua A

INTERNATIONAL JOURNAL OF CLIMATOLOGY 26 (7): 935-955 JUN 15 2006

-------

Climate/chemistry effects of the Pinatubo volcanic eruption simulated by the UIUC stratosphere/troposphere GCM with interactive photochemistry

Rozanov EV, Schlesinger ME, Andronova NG, Yang F, Malyshev SL, Zubov VA, Egorova TA, Li B

JOURNAL OF GEOPHYSICAL RESEARCH-ATMOSPHERES 107 (D21): Art. No. 4594 NOV 2002

-----

Really though, you need to do a search on ISI for GCM AND validation. There are 22 pages of results!

Another point, in any of the methodologies of these papers I find no mention of "we tuned our model to x". No, instead there actually seems to a be a legitimate attempt to do some real science and make honest comparisons between the model and the data.

quote:

The reason I ask for it now is to illustrate that the models ARE constructed from past data.

Could you elaborate on this? Models are mainly based on various formulae which describe physical phenomena eg. conservation of mass and energy, thermodynamics, Claussius-Clapeyron, radiative transfer and Navier-Stokes.

The parametrisations are validated using data, so in this they are entirely dependent on it, but to say that models ARE constructed from data is overstating this point. Your description suggests that they are in some way a statistical or empirical model. Without data there would be no validation of parameterisations and they would be total junk. If you do a ISI search for 'GCM' and 'validation' you will find a concerted effort to validate these parameterisations. Scientists seem to taking the paramerisations and testing them solely against data for whatever metric they are designed to describe.

Here's a better description: models are constructed from physical formulae and parameterisations. Parameterisations are attempts at describing the effects of small-scale physical phenomena at large scales, and these in turn are validated using data.

Maybe Steve_M can do better.

You got there before I did podbod.


Clouds - they remain uncertain, although there's a lot of ongoing research into them.

3 off the cuff reasons I'm not convinced clouds are a major problem for the theory of AGW:

1) The main external factor in clouds seems to be the cosmic ray flux issue - that does not remotely explain the warming of the last 30 years. There does not seem to be any trend in clouds that accounts for the warming of the last 30 years.

2) Without some external factor how do we know any observed changes in clouds are not down to CO2 driven climate change itself? That's not an unreasonable question due to changes in absolute humidity with warming, and Water Vapour Amplification.

3) Add 2 functions and graph them on a spreadsheet. One could be a random function that you could multiply by 5, e.g. y = 5*Rand(x). Then you could have a linear function with slope 0.1 e.g. y = 0.1*x. Even in that case where the random fluctuation is 50 times the trend magnitude you can still have an effect from the trend that can be extracted by averaging or trend fitting (over a long enough data set). So even if clouds are a big factor, if they don't have a trend, then a smaller trend due to another factor (CO2) can still cause observable effects with time - effects that the variation due to cloud cannot explain.

quote:

Maybe Steve_M can do better.

Nope, not really! Other than to say that exactly the same parametrisations are used in weather forecast models so they are proved against both weather and climatology.

To answer JL's cloud question directly, clouds have different effect depending on where and when they are and depending on their make-up. A big fluffy tropical anvil will reflect lots of sunshine and cool the earth. A layer of stratocumulous at night will keep the earth warm but some clouds aren't very good at reflecting away sunlight. Clouds at high latitudes might not have much effect because there is not much heat to keep in and not much sunlight to reflect. Clouds influenced by sulphates are brighter, and reflect more sunlight.

However, for observed warming to be caused by a change in clouds, there has to be a mechanism that caused the change. Lindzen has proposed one. Cosmic rays are another. Perhaps a misunderstanding of the aerosol effect is a third. But none of the mechanisms has much evidence behind it.

On the other hand, the radiative effects of CO2 are well understood, and we have plenty of evidence from prehistory that the climate feedbacks are more likely to be positive than negative.

Having looked at global temperature anomalies for individual models (eg page 763 of chapter 10 of the IPCC 4th AR), many do show quite a lot of annual to decadal variability. For example, 10+ year periods of flat or dropping surface temperatures appear in a number of the models superimposed on the upward trend. Without knowing the internals of the model, it is hard to know what is going on (it may be that the model experiences some sort of decadal oscillation that means more heat is going into the oceans), but it does illustrate the difficulty of making a detailed prediction for the next decade or so.

This is a good reason why a better test for current models is to compare against current climatology (in terms of climate and climate/weather variability), before attempting to fit to a cause of climate change (such as increasing CO2/solar or volcanoes).

quote:

Originally posted by Steve_M:
On the other hand, the radiative effects of CO2 are well understood, and we have plenty of evidence from prehistory that the climate feedbacks are more likely to be positive than negative.

Although these are limited in effect.

quote:

Although these are limited in effect.

Yes, they are limited to doubling or tripling the initial cause of warming (rather than, say, causing a "runaway" effect), suggesting a 2-3C rise for a doubling of CO2.

quote:

Well that is still quite speculative Steve. James Annan recent work in probabilistic parameter estimations suggests a much lower bound level of feedbacks.

They've looked at observational evidence from a number of different time frames and come up with a collective value of between 2.0 and 4.3C (90% confidence) with a best estimate of 3C.

So I'll stick with 2-3C even if it is an underestimate. I don't want to be alarmist .

Using multiple observationally-based constraints to estimate climate sensitivity. Annan & Hargreaves. GRL 2006.

quote:

Originally posted by Steve_M:

quote:

Well that is still quite speculative Steve. James Annan recent work in probabilistic parameter estimations suggests a much lower bound level of feedbacks.

They've looked at observational evidence from a number of different time frames and come up with a collective value of between 2.0 and 4.3C (90% confidence) with a best estimate of 3C.

So I'll stick with 2-3C even if it is an underestimate. I don't want to be alarmist .

Using multiple observationally-based constraints to estimate climate sensitivity. Annan & Hargreaves. GRL 2006.

A 2006 publication is out of date for James' thinking on this subject. He is still in the process of getting his most recent work published as there has been some hostility to his findings which show considerably less sensitively than he previously thought.

quote:

He is still in the process of getting his most recent work published

Unless I'm looking at the wrong paper, his more recent unpublished work appears to be focused on arguing against sensitivities above 6C rather than modifying his previous 2-4.3C value.