modelling atmospheric and solar influence

[k76]

Hi all,

I've enjoyed watching this site from the sidelines for quite some time now, thanks everyone. Trying to find out how our climate works is quite addictive, I think the only thing I've learnt over the last year is that we don't know anywhere near as much as I thought we knew.

Having done some research in another field in the past (hydrodynamics) I realize how difficult modelling nature is, and how big an influence even the tiniest parameter can have. But at least in hydrodynamics it is reasonably easy to check your results against experimental data, and you have a fair idea about what you are leaving out.

Now to my question: How well are climate scientists modelling the temperature difference between night and day? To me it seems like a good test of our understanding of how our atmosphere works: We know it's content and the output of the sun, so we should be able to model this without using external forcings, empirical factors etcetera. Are we?

[Maui]

I have tried to raise a discussion of a fundamental part of computer models (random numbers), but get no response. I believe that nature instead shows a Levy Flight distribution of events.

Most climate and weather models are not very good, perhaps because of the above flaw. I have had an interest in meteorology since taking a course at Cornell in 1980, and try to keep up with the state-of-the art.

Proof of Levy Flights was presented as a Letter in Nature #7194; 22 May 2008. This 80-year-old idea suggests that nature is prone to unexpected, high-energy events; I believe our recent discoveries concerning cosmic rays are further evidence.

[nautonnier]

Mar 23, 2009 17:08:26 GMT Maui said:

I have tried to raise a discussion of a fundamental part of computer models (random numbers), but get no response. I believe that nature instead shows a Levy Flight distribution of events.

Most climate and weather models are not very good, perhaps because of the above flaw. I have had an interest in meteorology since taking a course at Cornell in 1980, and try to keep up with the state-of-the art.

Proof of Levy Flights was presented as a Letter in Nature #7194; 22 May 2008. This 80-year-old idea suggests that nature is prone to unexpected, high-energy events; I believe our recent discoveries concerning cosmic rays are further evidence.

To join these two posts together - it may be difficult to separate an apparent 'Levy Flight' behavior from a chaotic system in which you are unaware of one of the variables or its effects or joint effects with other variables' states.

I really think that we are looking at a chaotic system where we don't know all the variables or their effects or their interrelationships. This makes the modeling of the system just luck. As long as the system stays relatively stable around an attractor it can appear _almost_ deterministic or at least its chaotic behavior has been bounded - by the attractor we do not recognize. We have been in such an apparently deterministic mode since computers allowed us to model the climate and the cosmos in detail. Now we may be moving toward another state due to some effect we have yet to identify and models of solar activity and climate are no longer as accurate as they had been.

I think the next few years is going to provide a salutary learning experience for some who like Lord Kelvin in 1900 thought all physics was discovered.

[Maui]

I am referring to specific algorithms in computer models. For example, the "state of the art" study of solar influence on volcanism used a "Monte Carlo" simulation. Is there any evidence to suggest that such a simulation mimics nature? There is now, after eighty years, evidence that nature is NOT random (or Brownian) in general.

NB--I have said that the Levy Flight study was by "non-linear physicists." In fact, I do not know how they would characterize themselves; their institution is the European Laboratory for Nonlinear Spectroscopy in Florence, Italy.