intlibber -
"a climate is an economy"?
No.
Re Armstrong. Here is his selection treee for forecasting methods:
http://www.forecastingprinciples.com/se ... _tree.html
Notice the right-hand branches. He thinks that in a deterministic system with small changes extrapolation/NN models etc should be used instead of causal physical theories. Why? Because he is not a physicist and not concerned with forecasting of physical systems.
He claims that extrapolation (not causal physical theories) is right way to forecast climate change. Thus according to his own methodology he expects only small changes. Given that there has been a LARGE change in one of the forcing inputs to climate (CO2) this seems a questionable assumption and certainly cannot be proved by arguments conditional on its assumption as his are!
Global Warming Concensus Broken
The problem with the current method is that because of the large grid squares it is not physical.He claims that extrapolation (not causal physical theories) is right way to forecast climate change.
If it was physical, because of Navier-Stokes, it would be chaotic.
Which would you prefer?
Engineering is the art of making what you want from what you can get at a profit.
Because the evidence is we get better forecasts that way. Besides, he's probably getting there from the "good knowledge of relationships = NO" branch.He thinks that in a deterministic system with small changes extrapolation/NN models etc should be used instead of causal physical theories. Why?
Shrug. That CO2 is a forcing input is an AGW assumption. In any case, we're certainly not expecting large changes in climate (e.g., temperature doubling along with CO2), nor are we expecting a change in how CO2 affects temp.Given that there has been a LARGE change in one of the forcing inputs to climate (CO2)
Simon -
There are many aspects of the simulation which obey physical laws independent of the chaos. And the chaos itself is a property of a deterministic (modulo QM) physical system which can in principle be modelled with great accuracy. That does not mean we can get around the chaos - but it does mean that emergent properties can be sufficient for accurate modelling. The analogy with thermodynamics is not stupid.
Whether GCMs now have this right enough? I cannot myself judge without having done a very thorough LS. But I do trust the peer-review process to move knowledge towards what is true, and it has been going for a long time on GCMs.
We know that micro-scale GCMs with fine grids are run to determine parameters for global GCMs. (It is the same GCM, run at different scales, but with fewer parameters at the micro-scale). This is a validation process that can detect whether or not the big-cell-level approximations are valid or not. If not, the literature can be challenged.
It is exciting to think that we contimue to get more sophisticated and accurate GCMs - I would have more respect for the anti-AGW advocates if they looked at the details of the GCMs and argues against there reliability, publishing arguments which could be sustained or refuted.
Best wishes, Tom
There are many aspects of the simulation which obey physical laws independent of the chaos. And the chaos itself is a property of a deterministic (modulo QM) physical system which can in principle be modelled with great accuracy. That does not mean we can get around the chaos - but it does mean that emergent properties can be sufficient for accurate modelling. The analogy with thermodynamics is not stupid.
Whether GCMs now have this right enough? I cannot myself judge without having done a very thorough LS. But I do trust the peer-review process to move knowledge towards what is true, and it has been going for a long time on GCMs.
We know that micro-scale GCMs with fine grids are run to determine parameters for global GCMs. (It is the same GCM, run at different scales, but with fewer parameters at the micro-scale). This is a validation process that can detect whether or not the big-cell-level approximations are valid or not. If not, the literature can be challenged.
It is exciting to think that we contimue to get more sophisticated and accurate GCMs - I would have more respect for the anti-AGW advocates if they looked at the details of the GCMs and argues against there reliability, publishing arguments which could be sustained or refuted.
Best wishes, Tom
I think we all do. Steady State gave way to the Big Bang, and the infinitely hot oven gave way to quantums of energy. The problem is that AGW is all mixed up with politics now, and that corrupts the scientific process.But I do trust the peer-review process to move knowledge towards what is true
Sure, and AGW skeptics would have more respect for AGW advocates if they didn't seem to hold a nonfalsifiable belief in manmade gasses driving climate (remember in the 1970s, Hansen was writing code to prove global cooling was caused by manmade gasses), and a hostility towards the scientific process of gradually falsifying our way towards truth.I would have more respect for the anti-AGW advocates if they looked at the details of the GCMs and argues against there reliability, publishing arguments which could be sustained or refuted.
Here's the main problem with AGW advocates: too often, they are zealots and politicians using science to further their aims, not scientists. They make demands, based on their conclusions, that everyone knows will have little effect on the outcome. Kyoto was ridiculous; leaving out China (the leading CO2 producer) and India and Africa makes the agreement worthless. It's essentially a religious act, a sacrifice to Gaia.
Honestly, AGW advocates are their own worst enemies. When James Hansen compares the coal industry to Auschwitz, it's hard to take him seriously as a scientist. All the pictures of "lost" polar bears don't help either.
Actually, it is. A climate is a complex, chaotic heat engine with sources and sinks, it is a system for handling heat energy in a fluid dynamical way.tomclarke wrote:intlibber -
"a climate is an economy"?
No.
Every calorie is a penny in the worlds climate economy.
As Dave says, the idea that CO2 is a forcing input is an unproven and questionable assumption of the AGW Hockey Team. The problem is that CO2 lags behind temperature, so if one is to believe CO2 is a forcing input, then you must also believe in the violation of causality and that the earth's climate functions via time travel.
Re Armstrong. Here is his selection treee for forecasting methods:
http://www.forecastingprinciples.com/se ... _tree.html
Notice the right-hand branches. He thinks that in a deterministic system with small changes extrapolation/NN models etc should be used instead of causal physical theories. Why? Because he is not a physicist and not concerned with forecasting of physical systems.
He claims that extrapolation (not causal physical theories) is right way to forecast climate change. Thus according to his own methodology he expects only small changes. Given that there has been a LARGE change in one of the forcing inputs to climate (CO2) this seems a questionable assumption and certainly cannot be proved by arguments conditional on its assumption as his are!
Tom,
You are confusing deterministic with calculable. Chaotic systems are deterministic (i.e. if you could set you the EXACT same conditions every time you would get the same results). The problem is that you can't set up the exact same conditions except in a computer run. Why? Measurement accuracy.
Let us look at a simple problem F=ma. If I calculate it to two significant digits or 10 the results agree with each other.
OTOH if I calculate a chaotic system with 2 significant digits the results will be quite different if I use 10. And different from 20. Or 11.
You see the difficulties are fundamental to the problem. Just because you use F=ma in parts of the calculation does not make the whole calculation deterministic. And that assumes you are working with a fine enough mesh to actually calculate the problem. Which the climate models assuredly do not do.
So what do they do? Parameterize. i.e. at such and such a place with 75% RH and temps of x then y% of the area will have clouds of albedo z. Provided the average wind in the block is below q.
Now making those kinds of assumptions may be historical. They are not physical.
You are confusing deterministic with calculable. Chaotic systems are deterministic (i.e. if you could set you the EXACT same conditions every time you would get the same results). The problem is that you can't set up the exact same conditions except in a computer run. Why? Measurement accuracy.
Let us look at a simple problem F=ma. If I calculate it to two significant digits or 10 the results agree with each other.
OTOH if I calculate a chaotic system with 2 significant digits the results will be quite different if I use 10. And different from 20. Or 11.
You see the difficulties are fundamental to the problem. Just because you use F=ma in parts of the calculation does not make the whole calculation deterministic. And that assumes you are working with a fine enough mesh to actually calculate the problem. Which the climate models assuredly do not do.
So what do they do? Parameterize. i.e. at such and such a place with 75% RH and temps of x then y% of the area will have clouds of albedo z. Provided the average wind in the block is below q.
Now making those kinds of assumptions may be historical. They are not physical.
Engineering is the art of making what you want from what you can get at a profit.
OK -
Simon - I have not, since I was 15 years old and writing (very bad) 3 body motion simulations in a decimal (!) machine code on a schools computer, confused deterministic with calculable. Hence the [modulo chaos] reservation.
You are right that any chaotic system (weather) will be incalculable over long enough timescales. In fact strictly also non-deterministic since QM uncertainty is amplified. But practically there is no difference - it is chaotic.
My point is a subtler one than this. Systems which are chaotic may nevertheless have invariants, and deterministic behaviour when time-averaged over a longer time than the chaotic timescales.
For example a radiator may result in chaotic airflow, but nonetheless there will be a deterministic relationship after averaging between temperature and heat loss. True - calculating precisely that relationship requires detailed simulation of the chaos, and is not analytically tractable.
So the GCMs can run with small grid-size to work out average transport parameters that depend on fine-structure chaos, then plug those parameters into a larger grid-size simulation. Large-structure chaos is still modelled explicitly by the simulation. This process will introduce errors which are amplified by chaos - but that is irrelevant. The chaos will never be long-term forecastable anyway. Providing the errors do not systematically skew the average heat transport they will not matter.
Your point about accuracy needs care. Any errors will be amplified and lead to different evolution - but not necessarily alter the long-term averaged results that are needed. I have all your reservations about tweaking GCMs to fit the last 100 years climate history, and overfitting. But my understanding is that they are constructed in a much more sophisticated way than that, and the above argumnet shows that it is in principle possible to calculate what is needed from the physics.
Chaos affects the weather system at all levels - up to decadal for ocean currents. So we should not expect any accurate long-range weather forecasts other than very long-term averages. But the GW issues, which are about determining the average affect of additional CO2 etc, can still safely be calculated through multiple runs with very slightly different initial conditions. There remain quite large known fluctuations (PDO etc). It would be a mistake to ignore deterministic long-term trends just because of these.
Thw process of calculating GW is not simple. I get frustrated at all the uninformed blog comment because it seems to me that few people have the interest to go on asking questions and working out for themselves what is really going on. When you claim GCMs are not safe I cannot of course answer that without very carefully reading all the literature. But you do not equally have any clear argument against the validity of GCMs.
Best wishes, Tom
Simon - I have not, since I was 15 years old and writing (very bad) 3 body motion simulations in a decimal (!) machine code on a schools computer, confused deterministic with calculable. Hence the [modulo chaos] reservation.
You are right that any chaotic system (weather) will be incalculable over long enough timescales. In fact strictly also non-deterministic since QM uncertainty is amplified. But practically there is no difference - it is chaotic.
My point is a subtler one than this. Systems which are chaotic may nevertheless have invariants, and deterministic behaviour when time-averaged over a longer time than the chaotic timescales.
For example a radiator may result in chaotic airflow, but nonetheless there will be a deterministic relationship after averaging between temperature and heat loss. True - calculating precisely that relationship requires detailed simulation of the chaos, and is not analytically tractable.
So the GCMs can run with small grid-size to work out average transport parameters that depend on fine-structure chaos, then plug those parameters into a larger grid-size simulation. Large-structure chaos is still modelled explicitly by the simulation. This process will introduce errors which are amplified by chaos - but that is irrelevant. The chaos will never be long-term forecastable anyway. Providing the errors do not systematically skew the average heat transport they will not matter.
Your point about accuracy needs care. Any errors will be amplified and lead to different evolution - but not necessarily alter the long-term averaged results that are needed. I have all your reservations about tweaking GCMs to fit the last 100 years climate history, and overfitting. But my understanding is that they are constructed in a much more sophisticated way than that, and the above argumnet shows that it is in principle possible to calculate what is needed from the physics.
Chaos affects the weather system at all levels - up to decadal for ocean currents. So we should not expect any accurate long-range weather forecasts other than very long-term averages. But the GW issues, which are about determining the average affect of additional CO2 etc, can still safely be calculated through multiple runs with very slightly different initial conditions. There remain quite large known fluctuations (PDO etc). It would be a mistake to ignore deterministic long-term trends just because of these.
Thw process of calculating GW is not simple. I get frustrated at all the uninformed blog comment because it seems to me that few people have the interest to go on asking questions and working out for themselves what is really going on. When you claim GCMs are not safe I cannot of course answer that without very carefully reading all the literature. But you do not equally have any clear argument against the validity of GCMs.
Best wishes, Tom
Here is a good (because accurate) but rather simple guide to weather, climate & chaos
Here is a great discussion on climate & chaos. The original article is not so informative, but if you read the long discussion afterwards there are many subtle & interesting issues drawn out. Specifically the way in which chaotic systems can jump from one attractor to another and how to know when this is happening in the climate.
Roger Pielke Snr (often seeming to be on the anti-AGW side) is making the point that - in his view - time-averaged global temperature is not a particularly helpful statistic for policymakers. Others are arguing that it is.
My real issue with the GCM discussion is that I would be happier if there were a clearer detailed discussion of how GCM skill is evaluated, and what are the quantitative checks to ensure no aspect of the parametrisation is overfitted. Both these issues are very complex so perhaps difficult to make accessible on the blogosphere. But worth trying.
Best wishes, Tom
Here is a great discussion on climate & chaos. The original article is not so informative, but if you read the long discussion afterwards there are many subtle & interesting issues drawn out. Specifically the way in which chaotic systems can jump from one attractor to another and how to know when this is happening in the climate.
Roger Pielke Snr (often seeming to be on the anti-AGW side) is making the point that - in his view - time-averaged global temperature is not a particularly helpful statistic for policymakers. Others are arguing that it is.
My real issue with the GCM discussion is that I would be happier if there were a clearer detailed discussion of how GCM skill is evaluated, and what are the quantitative checks to ensure no aspect of the parametrisation is overfitted. Both these issues are very complex so perhaps difficult to make accessible on the blogosphere. But worth trying.
Best wishes, Tom
Provided there is no jump in the phase space - i.e. climate sticks with the current strange attractor.My point is a subtler one than this. Systems which are chaotic may nevertheless have invariants, and deterministic behaviour when time-averaged over a longer time than the chaotic timescales.
In any case that brings us to apportioning cause and effect. Due to the chaotic nature of the system (or as the climate guys prefer - internal variability) local chaos may seem to amplify a given cause when in fact it is correlation not causation.
So we have the PDO (internal chaos) and the temp rise it "caused" is accounted for as CO2. Which may not be entirely correct. ;-)
This is a well known pitfall in climate studies and yet it is not given its proper weight. IMO.
In fact there was a paper a while back showing how internal variability could give rise to seeming positive feedback where the feedback was actually negative. Roy Spenser or Roger Pielke I think. Or maybe Lindzen. ICRC.
Engineering is the art of making what you want from what you can get at a profit.
Here's a study suggesting we're going to fall into another Ice Age over the next 20,000 years.
What's particularly infuriating about the AGW crowd is that they don't really know whether what they're doing could actually help doom Mankind to something far worse than what they fear, but they insist on spending huge amounts of resources on it anyway. The worst-case global warming scenario just requires lots of people to move inland; you can still grow crops when it's warm. An Ice Age would most likely end human civilization, with billions dying of starvation along the way.
What's particularly infuriating about the AGW crowd is that they don't really know whether what they're doing could actually help doom Mankind to something far worse than what they fear, but they insist on spending huge amounts of resources on it anyway. The worst-case global warming scenario just requires lots of people to move inland; you can still grow crops when it's warm. An Ice Age would most likely end human civilization, with billions dying of starvation along the way.
Well, anybody telling you the ice caps will collapse is simply lying to you. They've been stable through warmer and higher CO2 periods than today. Antarctica has been stable for 22 million years. The period from 8000 bc to 5000 BC was significantly warmer than today, and the ice caps were fine. Additionally, the Sahara was much higher in rainfall as a result of the warmer weather, with savannahs and forests and rivers across it. The Western US was also much wetter. Anybody saying we'll be worse off with global warming is simply not paying attention to the prehistoric record.TallDave wrote:Here's a study suggesting we're going to fall into another Ice Age over the next 20,000 years.
What's particularly infuriating about the AGW crowd is that they don't really know whether what they're doing could actually help doom Mankind to something far worse than what they fear, but they insist on spending huge amounts of resources on it anyway. The worst-case global warming scenario just requires lots of people to move inland; you can still grow crops when it's warm. An Ice Age would most likely end human civilization, with billions dying of starvation along the way.
-
alexjrgreen
- Posts: 815
- Joined: Thu Nov 13, 2008 4:03 pm
- Location: UK
You might want to make a distinction between the East Antarctic ice sheet, which is stable, and West Antarctic ice sheet, which has collapsed several times in the last 75,000 years.IntLibber wrote:Well, anybody telling you the ice caps will collapse is simply lying to you. They've been stable through warmer and higher CO2 periods than today. Antarctica has been stable for 22 million years. The period from 8000 bc to 5000 BC was significantly warmer than today, and the ice caps were fine. Additionally, the Sahara was much higher in rainfall as a result of the warmer weather, with savannahs and forests and rivers across it.
http://www.usatoday.com/tech/news/2009- ... ming_N.htm
http://www.springerlink.com/content/6800156543x9126j/
http://neptune.gsfc.nasa.gov/wais/pastm ... /Stone.pdf
http://www.abc.net.au/science/articles/ ... 283071.htm
The Sahara had rivers when the mountains in Africa were covered in glacial ice. As that melted, the flow rerouted around 10,500 BC to create the River Nile instead. Since then the Sahara has got gradually drier and almost all of the glaciers have now gone...
Ars artis est celare artem.
Most of that is sensible, until...
Once again, this appears to be ideology leading with science following behind offering rationales.
This triggers the bullshit detector big-time. Even assuming everything AGW proponents say is true, if the WAIS is going to collapse in the next 20 years nothing we do between now and then regarding fossil fuels will make the slightest difference.Dr Bradley Opdyke, a paleoceanographer from the Australia National University (ANU) believes the West Antarctic Ice Sheet (WAIS) could partially collapse within 20 years, resulting in a dramatic jump in sea levels.
...
McCulloch believes the collapse of the WAIS can be averted if the world community becomes "open minded" in its search for alternatives to fossil fuels.
Once again, this appears to be ideology leading with science following behind offering rationales.
Expeditions to the katabatic canyons to core the ancient glacier ice buried under spallated rock have proven that this disasturbationist theory of frequent and sudden WAIS collapse is simply false. That glacial ice simply would not be there, of that age, if WAIS collapse happened repeatedly.alexjrgreen wrote:You might want to make a distinction between the East Antarctic ice sheet, which is stable, and West Antarctic ice sheet, which has collapsed several times in the last 75,000 years.IntLibber wrote:Well, anybody telling you the ice caps will collapse is simply lying to you. They've been stable through warmer and higher CO2 periods than today. Antarctica has been stable for 22 million years. The period from 8000 bc to 5000 BC was significantly warmer than today, and the ice caps were fine. Additionally, the Sahara was much higher in rainfall as a result of the warmer weather, with savannahs and forests and rivers across it.
What kept the glaciers in africa there? Precipitation. There was a lot more precipitation than just icemelt, a glacier wont keep a river fed for thousands of years without precipitation.http://www.usatoday.com/tech/news/2009- ... ming_N.htm
http://www.springerlink.com/content/6800156543x9126j/
http://neptune.gsfc.nasa.gov/wais/pastm ... /Stone.pdf
http://www.abc.net.au/science/articles/ ... 283071.htm
The Sahara had rivers when the mountains in Africa were covered in glacial ice. As that melted, the flow rerouted around 10,500 BC to create the River Nile instead. Since then the Sahara has got gradually drier and almost all of the glaciers have now gone...