|
Post by icefisher on Jun 20, 2010 22:31:38 GMT
That is true, Trenberth takes the value from climate model results. Well since a climate model has to model the atmosphere that means it has to have values for all the various components of the budget diagram to compute an imbalance. So you said on one hand the imbalance is not calculated by the budget diagram but one has to then ask if the model it was calculated from did it use Trenberth's values. Or did Trenberth just take the result and ad hoc plant it in there and subtract it randomly from whatever value he had to have subtracted if from. I would ask which value that was but it doesn't seem very worthwhile because it could have been removed from any value in the diagram without anybody noticing.
|
|
|
Post by socold on Jun 20, 2010 23:34:52 GMT
I think the models generate the values for the components themselves rather than them being input in. The initial state is going to be input but the model will then change the components until it's own world settles into equilibrium.
As the imbalance is supposedly caused by reduction of energy into space I guess he reduced that one by 0.9 to make it fit. Id have to read the paper in more detail to see what he did specifically.
|
|
|
Post by icefisher on Jun 21, 2010 0:09:24 GMT
I think the models generate the values for the components themselves rather than them being input in. The initial state is going to be input but the model will then change the components until it's own world settles into equilibrium.
As the imbalance is supposedly caused by reduction of energy into space I guess he reduced that one by 0.9 to make it fit. Id have to read the paper in more detail to see what he did specifically.
Thats fascinating Socold. So how did they model the initial state if they don't know what the initial budget is? Did they just rely on assumptions of CO2 driving everything in the atmosphere so a budget wouldn't matter?
|
|
|
Post by sigurdur on Jun 21, 2010 1:29:30 GMT
From Trenbeth's paper we can see why he is upset he can't find the heat. He used assumptions and guesses to values in his paper. The line of thought in the paper exposes more than anything else. You just guess......
This is actually proving very interesting the more I think of it. No wonder the GCM's are so totally screwed up at predictive values.
|
|
|
Post by socold on Jun 21, 2010 18:41:12 GMT
I think the models generate the values for the components themselves rather than them being input in. The initial state is going to be input but the model will then change the components until it's own world settles into equilibrium.
As the imbalance is supposedly caused by reduction of energy into space I guess he reduced that one by 0.9 to make it fit. Id have to read the paper in more detail to see what he did specifically.
Thats fascinating Socold. So how did they model the initial state if they don't know what the initial budget is? Did they just rely on assumptions of CO2 driving everything in the atmosphere so a budget wouldn't matter? They don't need to model the initial state. They should be able to set the initial state roughly and let the model sort out the rest. If the initial state they set is too warm the model will cool down to whatever the equillibrium level is, if it is too cold it will warm up to equillibrium. The model will find it's own equillibrium and there's no need to feed that into the model.
|
|
|
Post by icefisher on Jun 21, 2010 18:57:16 GMT
They don't need to model the initial state. They should be able to set the initial state roughly and let the model sort out the rest. If the initial state they set is too warm the model will cool down to whatever the equillibrium level is, if it is too cold it will warm up to equillibrium. The model will find it's own equillibrium and there's no need to feed that into the model.
So you are telling us the models are smarter than the guys that wrote them. When do you think they will roll this out to disprove the paper by Spencer?
|
|
|
Post by socold on Jun 21, 2010 20:16:17 GMT
Say we have a model of a 1000 watt bulb pointed at a pot of water on a stove, we want to start the model with the the water temperature at equilibrium. Problem is we don't know what that equilibrium temperature is. No problem, that's one of the questions we can use the model for. Do one or more test runs and set the initial water temperature in the model to 10C, 50C or even -40C and let the model either warm or cool the water through it's thermodynamic codes. After a sufficient amount of model-world time the modeled temperature of the water will have settled to the equilibrium temperature. You can then restart the model starting with that equilibrium temperature as the baseline.
|
|
|
Post by sigurdur on Jun 21, 2010 20:32:03 GMT
Socold: Doing so would demand that you know the physical metrics of the 1,000 watt bulb heating the pot of water. Since we don't know all the metrics, equilibrium can never be achieved.
It is that simple.
|
|
|
Post by nautonnier on Jun 21, 2010 20:45:05 GMT
Say we have a model of a 1000 watt bulb pointed at a pot of water on a stove, we want to start the model with the the water temperature at equilibrium. Problem is we don't know what that equilibrium temperature is. No problem, that's one of the questions we can use the model for. Do one or more test runs and set the initial water temperature in the model to 10C, 50C or even -40C and let the model either warm or cool the water through it's thermodynamic codes. After a sufficient amount of model-world time the modeled temperature of the water will have settled to the equilibrium temperature. You can then restart the model starting with that equilibrium temperature as the baseline. The problem is that the system that you are describing in your example is a simple deterministic system. The system that Trenberth was attempting to describe is a chaotic system made up of inter-reacting nP subsystems that are also chaotic. Chaotic systems do not have an 'equilibrium' or more correctly they can have multiple different 'equilibria' known as attractors. The one thing that is certain about any modeling of chaotic systems is that you MUST have the correct start parameters for EVERY variable as precise as possible and even then the model will move away from actuality disappointingly rapidly. This is basic Lorenz chaos theory.
|
|
|
Post by sigurdur on Jun 21, 2010 20:49:23 GMT
Precisely Nautonnier: You don't know when a sheet of paper will come between the bulb and the pot. And you don't know the density of that paper nor how long it will come between the bulb and the pot.
The variables, unless taken in a slab occurrance, are so chaotic that any output of the model thinking it is in equilibrium is a false reading.
|
|
|
Post by icefisher on Jun 21, 2010 21:10:00 GMT
Socold: Doing so would demand that you know the physical metrics of the 1,000 watt bulb heating the pot of water. Since we don't know all the metrics, equilibrium can never be achieved.
It is that simple. Everybody has an reason to be. Socold's is apparently to imitate a large mouth bass (Micropterus salmoides floridanus) sucking in a wooden mop handle end bristling with three 1/0 treble hooks tied to a string.
|
|
|
Post by socold on Jun 22, 2010 20:30:40 GMT
The problem is that the system that you are describing in your example is a simple deterministic system. The system that Trenberth was attempting to describe is a chaotic system made up of inter-reacting nP subsystems that are also chaotic. Chaotic systems do not have an 'equilibrium' or more correctly they can have multiple different 'equilibria' known as attractors. The one thing that is certain about any modeling of chaotic systems is that you MUST have the correct start parameters for EVERY variable as precise as possible and even then the model will move away from actuality disappointingly rapidly. This is basic Lorenz chaos theory. They can have multiple attractors, but you can see if it that the case and to what extent it is the case by starting the model with slightly different initial conditions. You only need to set the initial conditions as accurately as the range covered by the local equilibrium.
|
|
|
Post by socold on Jun 22, 2010 20:34:15 GMT
Socold: Doing so would demand that you know the physical metrics of the 1,000 watt bulb heating the pot of water. Since we don't know all the metrics, equilibrium can never be achieved. It is that simple. Surely I just need to leave the pot of water for an hour and then come back and measure what temperature it is.
|
|
|
Post by sigurdur on Jun 22, 2010 21:33:36 GMT
But during that hour, was a piece of paper dropped between the pot and the light? Or was there a temp power outage? My point is, there are things you wouldn't know happened. Just as in climate, there are things that we don't know that could severely upset that potential equilibrium. Remember the old phrase....assume? and the results? In climate models there are a lot of assumptions. I rest my case.
|
|
|
Post by nautonnier on Jun 24, 2010 16:31:01 GMT
The problem is that the system that you are describing in your example is a simple deterministic system. The system that Trenberth was attempting to describe is a chaotic system made up of inter-reacting nP subsystems that are also chaotic. Chaotic systems do not have an 'equilibrium' or more correctly they can have multiple different 'equilibria' known as attractors. The one thing that is certain about any modeling of chaotic systems is that you MUST have the correct start parameters for EVERY variable as precise as possible and even then the model will move away from actuality disappointingly rapidly. This is basic Lorenz chaos theory. They can have multiple attractors, but you can see if it that the case and to what extent it is the case by starting the model with slightly different initial conditions. You only need to set the initial conditions as accurately as the range covered by the local equilibrium. But you do not know what the attractors are and which part of the levy-flight of changes in reactions and feedbacks will trigger move toward an attractor. What you are effectively saying is that you will build a partial and deterministic model of a hugely chaotic system of chaotic systems then claim - with no validation whatsoever, that the output is what will happen in (your lifetime plus a few years) time. That is not science - it is casting the runes And you wish the entire global economy to be wrecked based on that?
|
|