Reliability of GCM's and related

[hunter]

@ steve,
"They can predict Atlantic storm activity". Do please show us this predictive ability.
And as for Pinatubo, since no one doubts that big volcanic events lower the world temps, I don't think a prediction about that qualifies for anything. That proves a climate model about as well as predicting phases of the moon.

[icefisher]

Apr 8, 2010 17:28:21 GMT steve said:

I'm stumped as to why you think an experiment involving greenhouses tells you about convection in models and budget diagrams. The Kiehl and Trenberth budget diagram is a bringing together of a number of empirical studies and experiments.

Empirical? Hardly by a long shot.

Note the following:

"b. Longwave radiation
We must rely on model calculations to determine
the surface radiative fluxes."

and a model parameter:

"Clouds are
assumed to exist in three layers and these layers are
assumed to be randomly overlapped."

Nothing empirical about those assumptions and they are almost certainly wrong as well as well it appears they are doing slab processes.

[steve]

Icefisher,

I'm still waiting for your budget diagram showing the movements of near surface heat that you claimed were wrong. You are now trying to change the subject by being picky.

The KT diagram is strongly empirically based. Identifying bits that come from models doesn't change this since anyone who knows anything would understand that a radiation model is based on empirical studies, and other aspects of the paper were directly related to empirical studies. But I'm not particularly interested in defending the KT diagram. I'm interested in your statement that the empirical studies of convection differ wildly from the budget diagrams as the empirical study you described is meaningless, and convection would only be shown in a budget diagram that looked at the budgets of different layers in the atmosphere.

So where is the budget diagram you were referring to?

[icefisher]

Apr 9, 2010 9:22:41 GMT steve said:

The KT diagram is strongly empirically based. Identifying bits that come from models doesn't change this since anyone who knows anything would understand that a radiation model is based on empirical studies, and other aspects of the paper were directly related to empirical studies.

LOL!

Empirical. and a model parameter:

"Clouds are
assumed to exist in three layers and these layers are
assumed to be randomly overlapped."

Clouds are the biggest driver in the climate system and there isn't even an attempt to model them. Its all based on slab atmosphere assumptions. . . .absolutely nothing empirical there.

But you insist on trying to pass that off as an empirical study. . . .that is just plain ignorant of what empirical science is and is not. By your definition there may not be a single SWAG that isn't an empirical study.

Read my lips!!! A model is not an empirical study anymore than a Revell plastic battleship model is a battleship. Never is, never was, and never will be. Hope I am not puncturing another hallucination of yours by revealing that!

The only way a model can be related to an empirical study is if the model is studied with observations. . . .and we are seeing the results of that!

[steve]

Icefisher,

Your argument seems to be to come up with some nonsense, and then when challenged you forget your nonsense and get picky about some irrelevant details of the challenge.

You produced what you thought was an empirical study and claimed it differed from an AGW budget diagram.

When I pointed out your empirical study was pointless and suggested that no such budget diagram existed you started to complain about the one budget diagram I did produce.

Of course models are based on physics and observations. Budget diagrams are simplified studies where the sun always shine - they are illustrative.

So where is the budget diagram you were referring to? Please answer with sentences that do not include the word "beef".

[icefisher]

Apr 9, 2010 13:47:32 GMT steve said:

Icefisher,

Your argument seems to be to come up with some nonsense, and then when challenged you forget your nonsense and get picky about some irrelevant details of the challenge.

You produced what you thought was an empirical study and claimed it differed from an AGW budget diagram.

When I pointed out your empirical study was pointless and suggested that no such budget diagram existed you started to complain about the one budget diagram I did produce.

Of course models are based on physics and observations. Budget diagrams are simplified studies where the sun always shine - they are illustrative.

So where is the budget diagram you were referring to? Please answer with sentences that do not include the word "beef".

Glad to answer Steve. But lets put it in context. I was replying to your post where you said: Obviously there is real internal variability that can be observed but that is hard to predict (ENSO, PDO etc). But while models can't necessarily predict when ENSO happens they can exhibit ENSO and PDO behaviour to make predictions of what happens to the earth's energy balance during their different phases. That's why models can be useful as tools for understanding.

Implicit in that statement are a number of implications 1) You are assuming all important variations are internal; 2) The models can be used to predict internal variability. 3) The models are properly reflecting energy balance and all we need to do is not be concerned by the timing of anomalies such as PDOs and ENSOs because we will learn more about that as time goes on.

Then you obfuscated with: 1)I'm stumped as to why you think an experiment involving greenhouses tells you about convection in models and budget diagrams. The Kiehl and Trenberth budget diagram is a bringing together of a number of empirical studies and experiments.

www.geo.utexas.edu/courses/387H/PAPERS/kiehl.pdf and 2) But I'm not particularly interested in defending the KT diagram.

Which of course raises the question about what you are defending above and why you offered the KT diagram.

Its just my opinion that to defend a model you first should empirically validate it. I am merely offering a variety of choices. Wait another 30 years (since clearly an 11 year change in direction doesn't cut it and then produce an empirically validated study or build some earthly experiments to demonstrate the model assumptions on energy balance transfers through the atmosphere.

And I say that because it seems implausible that IR would be nearly 300% stronger forcing than convection considering basic greenhouse measurements of convection at peak conditions are 2000% stronger than IR.

A more plausible balance would be much closer to full saturation and thus incapable of producing the kind of warming estimated by the models.

Your argument that I should produce a model to refute your untested model would probably be taken much better if you actually recognized the need to test these models before relying upon them.

[poitsplace]

Someone in another forum got SO CLOSE to understanding the problem for AGW created by powerful negative feedback of latent heat and convection. I pointed out how much convection and latent heat MUST increase if the temperature goes up several degrees...and their problem is that they didn't see where the energy would come from.

Funny and a little sad...because that's my whole point. There isn't anywhere for that much energy to come from.

[steve]

Glad to answer Steve. But lets put it in context. I was replying to your post where you said: Obviously there is real internal variability that can be observed but that is hard to predict (ENSO, PDO etc). But while models can't necessarily predict when ENSO happens they can exhibit ENSO and PDO behaviour to make predictions of what happens to the earth's energy balance during their different phases. That's why models can be useful as tools for understanding.

Implicit in that statement are a number of implications 1) You are assuming all important variations are internal; 2) The models can be used to predict internal variability. 3) The models are properly reflecting energy balance and all we need to do is not be concerned by the timing of anomalies such as PDOs and ENSOs because we will learn more about that as time goes on.

1) Not necessarily. If X causes Y causes Z, and X is an extremely subtle effect but Y and Z are big and potentially measurable changes, then finding the connection between Y and Z is interesting and useful even if the ultimate trigger is X. For example, the Milankovitch cycles are important external variability, but the actual amount of variability is small. The impact is the change in snow build up that has big impacts on albedo and big positive feedbacks which can be separately analysed as the initial change in snow build up could be caused by things other than Milankovitch cycles.

2) No. If Y happens, we may be able to predict Z. But we don't know when Y is going to happen if we don't know exactly what X is.

3) The question is whether models are *sufficiently* reflecting energy balances in response to PDO etc. If we think that there is an change leading to a positive energy balance that is unrelated to CO2, then models would be a good tool for exploring that change. Eg. if someone really does prove that cosmic rays cause clouds, then the phenomenon can be tested in models, and the model result can be compared (again) with observations.

PS. I brought up the KT diagram because you talked about budget diagrams and that was the only one I'm familiar with.

[poitsplace]

Another issue is the way models treat water vapor. Parametrization causes a separation between its convection/latent heat and absorption of outgoing IR. Absorption is necessarily included in the observed, moist adiabatic lapse rate...since water vapor carries the majority of its own energy with it. Absorption simply forces the water vapor to stay a vapor longer and stab higher into the troposphere.

About all we're left with is the extremely powerful, negative feedback for increases in latent heat/convection.

[icefisher]

Apr 11, 2010 17:20:01 GMT steve said:

3) The question is whether models are *sufficiently* reflecting energy balances in response to PDO etc. If we think that there is an change leading to a positive energy balance that is unrelated to CO2, then models would be a good tool for exploring that change. Eg. if someone really does prove that cosmic rays cause clouds, then the phenomenon can be tested in models, and the model result can be compared (again) with observations.

I think the models ought to be paid for out of campaign donations as that is about all they are worth at this point.

Its complete malarkey they lead to understanding when you don't have a model that actually tracks climate when you correct your parameters. Now there is undoubtedly portions of these GCMs that are useful but they aren't useful for climate prediction.

For a model to be useful when it goes off track you need to be able to point to why it went off track.

I have worked with models for the past 25 years. I understand fully when they are useful and when they are not. The government is just throwing money away on these and the only use they are being put to are political in nature.

Billions have been spent on models the public has no code for and as such the learning from those models are nil to the general public. . . .and I would expect that someone who is on the modeling gravy train to say these things are the most useful things in the world. . . . cause they buy everything they got.

[steve]

Apr 11, 2010 21:49:29 GMT icefisher said:

Apr 11, 2010 17:20:01 GMT steve said:

3) The question is whether models are *sufficiently* reflecting energy balances in response to PDO etc. If we think that there is an change leading to a positive energy balance that is unrelated to CO2, then models would be a good tool for exploring that change. Eg. if someone really does prove that cosmic rays cause clouds, then the phenomenon can be tested in models, and the model result can be compared (again) with observations.

I think the models ought to be paid for out of campaign donations as that is about all they are worth at this point.

Its complete malarkey they lead to understanding when you don't have a model that actually tracks climate when you correct your parameters. Now there is undoubtedly portions of these GCMs that are useful but they aren't useful for climate prediction.

So portions of GCMs are useful but they don't lead to understanding? What do you mean by those two contradictory points?

Separately, can't you see there is a difference between a general concept such as "climate prediction" that could mean predicting the equilibrium sensitivity, predicting the average rate of warming, predicting the change in frequency and strength of tropical storms, and the specific concept of "a model that actually tracks climate" which sounds more like a forecasting process.

For a model to be useful when it goes off track you need to be able to point to why it went off track.

In a general sense, we know that models go off track because they do not start from identical initial conditions and because they do not model all processes accurately (due to lack of resolution and due to lack of perfect knowledge). So yes, many of the weaknesses of models can be pointed to, and many of their strengths also. You are not the only person to have worked with models for 25 years, so you are not the only person who believes they understand "fully" when they are and are not useful.

It's not uncommon for people knowledgeable in one field to assume that people in another field are missing a trick when there are parallels involved between the two fields. Given that many businesses get into big trouble through inappropriate financial modelling I question the analogies being used.

I have worked with models for the past 25 years. I understand fully when they are useful and when they are not. The government is just throwing money away on these and the only use they are being put to are political in nature.

Billions have been spent on models the public has no code for and as such the learning from those models are nil to the general public. . . .and I would expect that someone who is on the modeling gravy train to say these things are the most useful things in the world. . . . cause they buy everything they got.

[steve]

Apr 11, 2010 18:37:18 GMT poitsplace said:

Another issue is the way models treat water vapor. Parametrization causes a separation between its convection/latent heat and absorption of outgoing IR. Absorption is necessarily included in the observed, moist adiabatic lapse rate...since water vapor carries the majority of its own energy with it. Absorption simply forces the water vapor to stay a vapor longer and stab higher into the troposphere.

Your description of the issue is a little unclear. Even if you take a very high resolution model with a time step measured in seconds, the convection process is still a lot slower than the radiation process. I don't see why the separation between the two parametrizations automatically gives rise to an issue.

It seems obvious that if there were an issue, it would be very apparent if you compared the model with the detailed observations.

About all we're left with is the extremely powerful, negative feedback for increases in latent heat/convection.

Again, CO2 is not the only thing that increases latent heat/convection. Models run the same code in summer and winter. Any inappropriate parametrization would be apparent from differences between the models and the weather in one or other of the climate domains.

[poitsplace]

Apr 12, 2010 7:36:48 GMT steve said:

Your description of the issue is a little unclear. Even if you take a very high resolution model with a time step measured in seconds, the convection process is still a lot slower than the radiation process. I don't see why the separation between the two parametrizations automatically gives rise to an issue.

It causes a problem because you cannot separate them. As the moist air rises and cools it gives off energy. If it absorbs energy along the way it doesn't condense and give off latent heat...it condenses slightly higher up. Its a different mode of energy transport but the same result. Its an inherent part of the observed lapse rate. The tropics don't lock heat in place as one would expect if GHG forcing worked with water vapor...they hemorrhage energy like crazy.

It seems obvious that if there were an issue, it would be very apparent if you compared the model with the detailed observations.

You mean like...they wouldn't be able to make short term projections? Yeah, something like that WOULD be obvious.

Again, CO2 is not the only thing that increases latent heat/convection. Models run the same code in summer and winter. Any inappropriate parametrization would be apparent from differences between the models and the weather in one or other of the climate domains.

CO2 does not directly increase latent heat at all.

[steve]

Apr 12, 2010 8:34:08 GMT poitsplace said:

Apr 12, 2010 7:36:48 GMT steve said:

Your description of the issue is a little unclear. Even if you take a very high resolution model with a time step measured in seconds, the convection process is still a lot slower than the radiation process. I don't see why the separation between the two parametrizations automatically gives rise to an issue.

It causes a problem because you cannot separate them. As the moist air rises and cools it gives off energy. If it absorbs energy along the way it doesn't condense and give off latent heat...it condenses slightly higher up. Its a different mode of energy transport but the same result. Its an inherent part of the observed lapse rate. The tropics don't lock heat in place as one would expect if GHG forcing worked with water vapor...they hemorrhage energy like crazy.

As moist air rises it cools adiabatically, it emits and absorbs radiation according to its temperature and constitution. I don't see how this is something that cannot inherently be modelled.

The increase in GHGs we are talking about (say a doubling of CO2) changes the radiation balance by roughly 2% - quite a small number. The phrase "lock heat in place" is painting a picture that I don't recognise.

You mean like...they wouldn't be able to make short term projections? Yeah, something like that WOULD be obvious.

Again, you are confusing the ability of models to say useful things about physical processes with their ability to perfectly reproduce the earth's climate state and weather. Thinking that sarcasm is a valid response is a reflection of your confusion about what we are discussing.

In terms of the discussion we are having, the models *do* make short term predictions about the response of the atmosphere to warming. The daily weather forecast is dependent on these particular processes of convection, latent heat etc.

Again, CO2 is not the only thing that increases latent heat/convection. Models run the same code in summer and winter. Any inappropriate parametrization would be apparent from differences between the models and the weather in one or other of the climate domains.

CO2 does not directly increase latent heat at all.

Well "directly" is your choice of word. All greenhouse gases contribute to the warmth of the planet. But in essence you seem now to be agreeing with me as the assumption must be that the amount of convection/latent heat transport on a given day is mostly dependent on temperature and not dependent on the details of the levels of CO2.

[icefisher]

Apr 12, 2010 9:30:15 GMT steve said:

The increase in GHGs we are talking about (say a doubling of CO2) changes the radiation balance by roughly 2% - quite a small number. The phrase "lock heat in place" is painting a picture that I don't recognise.

So if your modeling error on the first decade of use is 100% and you need better than +/-2% for the model to be useful. . . . .

Somebody maybe ought to go back to the drawing board and figure out why this model yawed over and pitchforked into the ground.