cc: wg1-ar4-ch06@joss.ucar.edu date: Wed, 12 Jan 2005 08:08:36 -0500 from: David Rind subject: Re: [Wg1-ar4-ch06] Bullet points for the executive summary for to: Stefan Rahmstorf Stefan, I think we've decided to drop 'value laden' terms from the draft, so this discussion does not really concern IPCC Chapter 6; and in fact, I think Jonathan has agreed to put your bullet concerning 'success' into the document. This is now more a question of defining what the terms I've been referring to really signify. Parameterizing the dynamics means that everything the weather service does - not only the synoptic scale eddies, but all planetary waves - all atmospheric modes of variabililty (including the NAO, AO, PNA) - all the storms that produce precipitation, rain, snow, drought, patterns that influence paleoclimate observations and their regional distribution - all the cloud cover associated with these storms that affect radiation absorption - all of the elements of weather that influence climate are parameterized. The primary conservation law in the atmosphere is conservation of potential vorticity - by not solving the momentum equation, in effect that conservation is violated. And it is not just the eddies that are being parameterized - the mean fields, which depend on eddy convergences of heat and momentum can't be determined either. The primary advance in atmospheric dynamics over the last 50 years is the non-acceleration theorem, which includes the concept that one cannot determine what happens to any of the mean fields, like temperature, without calculating what happens to momentum; and momentum change with climate cannot be parameterized because it involves 1) unparameterizable things like frictional dissipation, which doesn't have any easy relationship to large scale energy fields ; and 2)nonlinear terms like momentum convergence. Peter Stone has spent half his life trying to come up with such parameterizations, and basically has given up. And it is not only extratropical dynamics - the ageostrophic circulations like the Hadley Cell and Walker Cells, and the hydrologic fields associated with them can't be calculated without solving the momentum equation. Attempts to parameterize the Hadley Cell always fail when compared with models that do calculate it (another example of Richard Lindzen's simplified view of the world). So this is not just 'a few other' parameterizations - it's a good part of the ball game. This is very different from saying that GCMs do not resolve small scale features, like convection, or eddies in the ocean, or turbulence in the boundary layer. Of course, at a certain level, every model includes parameterizations, but the real question is what is the order of the phenomena that is being parameterized, and when it is zeroth order, that is no longer what I would call a 'first principles' model. Throughout the draft in this chapter we have referred to the use of other types of models (i.e., EMICs) as valuable in suggesting linkages and concepts, and I think that is the case (some GCM people won't even go that far). The concepts associated with hysteresis that your papers in particular discuss are very important, and I think I've included that comment in some of the text I've written. So in no sense do I want to limit what models are applied to these complicated questions; we need all models on whatever scales and with whatever limitations, until the time when we come up with models that really can do everything. When I've suggested that such models cannot be used to 'prove' anything (words that are no longer in the draft) - hence where I draw the line in what I believe to be their usefulness - it is because of the limitations noted above. Note that climate change skeptics say the same things about GCMs using the same principles concerning GCM limitations you refer to below, and of course to some extent they are right - proof is a relative term in real world physics. But saying that doesn't mean the heuristic arguments that these skeptics use has the same value as real science, and it also doesn't mean that all models are created equal. EMICs give up a lot of validity in gaining the ability to run simulations with very large time steps - there are no free lunches. The fact that GCMs cannot run long enough to look at how past-climate influences, over thousands of years, affect things is a severe limitation when such effects are important, as in the case of ice sheets and perhaps the ocean circulation - thus in many paleoclimate applications. Let's not overlook the very severe limitations that EMICs take on themselves by avoiding that one. David At 12:40 PM +0100 1/12/05, Stefan Rahmstorf wrote: >A response to David's comment on bullet point 6.2.3: >this is not a bullet point that refers to any particular model, but >to all efforts of modeling paleoclimate. Personally, I think that >these efforts show encouraging successes. Now, if other people think >that the efforts in modeling paleoclimate over the past years were >mostly a failure, we may need to revise this bullet point. This >could turn out to be a discussion over wether the glass is half full >or half empty, though, which we may not resolve. >The reason I wrote about "encouraging successes" is that many people >have the impression that climate models are generally unable to >reproduce past climate changes, and I think that's not true, I think >we are starting to get some encouraging successes, while we are of >course still in an early stage of this field. > >On the different models there is indeed a disagreement between David >and myself, which partly reflects how in my perception the GCM >community is attempting to hold back the competition from other >models and retain an exclusive claim to the truth, so to speak. The >term "first principle models" used by David is a case in point, it >is a term chosen to make GCMs appear as "first principle", while >other models are put into a class below that. I don't think this >kind of language is helpful or can be justified in a scientific >sense. > >To give an example: arguably the most severe simplification in the >atmosphere model of CLIMBER2 is that synoptic eddies in the >atmosphere are not resolved, rather their effects on the large-scale >circulation are parameterised. To parameterise the effects of >synoptic eddies is standard practice in all ocean GCMs used in >coupled climate models. Hence, if using such a parameterisation of >synoptic eddies makes a model not "first principle" any more, then >none of the existing coupled GCMs is a "first principle" model >either. What David calls "first principle" models employ many >parameterisations that are not first principle - clouds, convection, >you name it. > >Clearly there is a whole spectrum of models, and they use different >ways of simulating and parameterising various mechanisms, which have >to be evaluated on a case by case basis, and some models are better >for some purposes than others. (Think of tsunami spreading models - >the best models for that will use the shallow water equations rather >than the primitive equations, and it would be silly to argue this is >somehow bad or not state-of-the-art because it is a "simplified" >model.) I would really object to any value-laden statements that >suggest GCMs are somehow fundamentally different from models that >parameterise or simplify one process or another. > >Cheers, Stefan > >-- >Stefan Rahmstorf >www.ozean-klima.de >www.realclimate.org _______________________________________________ Wg1-ar4-ch06 mailing list Wg1-ar4-ch06@joss.ucar.edu http://www.joss.ucar.edu/mailman/listinfo/wg1-ar4-ch06