rhorn@infinet.UUCP (Rob Horn) (08/08/87)
References: [This is a continuation of debate on nuclear winter from sci.astro. It no langer has anything to do with astronomy, so I am moving it here.] When I read the original Science article (TTAPS) I was surprised by two unusual omissions. It is normal in climatology to publish all the underlying equations of state that were used, the quantizations used, the approximations used to deal with sub-scale phenomena, and the boundary conditions used. This is either by inclusion in the article, or by reference to a prior publication. TTAPS did neither. It referred instead to an article in preparation. This is very non-traditional in climatology. You never publish the results before the model. I excused this on the grounds that Science is not a normal meteorology journal, and the scheduling must have been mismatched. The second oddity was the complete omission of references to prior art concerning sensitivity studies on the type of model used. In particular, since prior art indicated some astonishingly severe sensitivity to certain approximations, it was odd that these approximations were never mentioned. I ascribed this to political bias, and figured that the model article would cover them. The excuse could be that without the model description these other analyses would not be in context. However, it has now been four years and the companion article has never been published. I know people who have submitted routine articles since then and had them published already. I now suspect that their model is severely flawed. TTAPS describes a one-dimensional climate model, and sketches very briefly the addition of opaque material into the model at unspecified altitudes and opacities. A one-dimensional model assumes that the entire earth's atmosphere can be simulated by taking a vertical bar of air (typically 1m^2 for convenience) and analyzing its behaviour. All of the interesting issues like oceans, winds, artic vs tropic variations, etc. are parameterized by magic approximations. These models are popular with people who lack the computer time to handle 2 or 3 dimensional simulations. All of the professionals switched to 3-D as soon as CRAY's were readily available. During the 70's some serious sensitivity analyses were performed on one-dimensional models that cast them in very poor repute. The one that is most applicable to TTAPS was an analysis of the importance of the land-sea approximation. A model that assumed purely land behaviour at the bottom boundary condition was calibrated using mean solar radiation. Then the sensitivity was analyzed by running analyses at plus and minus one standard error in solar radiation. This error is the measurement error due to undetected cirrus clouds and molecular absorbtion at the various desert sites that measured solar intensity. This tiny variation (less than 1%) was enough to swing the model between ice age and tropical rain-forest. The conclusion was that proper air-sea and air-land interface approximation was crucial. TTAPS never revealed what approximation they used. There was a similar sensitivity to parameterization of winds. The TTAPS model approximation for north-south air motion was never revealed. Of somewhat lesser importance (but shown crucial to accuracy for 5-day forecasts) is the parameterization for rainfall and clouds. These too have never been revealed by TTAPS. The reason that there have been few direct complaints is that extreme sensitivity does not mean that nuke winter is wrong. The professional response has been to say ``Maybe. Give us the time and money to perform proper simulations''. The very first of these are now being published. I don't expect professional complaints until there is real defendable modeling to support professional statements. They know that they would be crucified if they debunked nuke winter for over-sensitivity, and then the accurate models confirmed nuke winter. It is also very difficult to criticize a model without knowing more about its details. For example, a recent article analyzed smoke plume behaviour, based on pre-TTAPS measurements of major fires, to determine what level in the atmosphere smoke would reach. The indications are that TTAPS overestimated soot deposition into the stratosphere. Unfortunately, TTAPS has never published the actual estimates that they used, so you cannot compare their assumptions with the results of this analysis. -- Rob Horn UUCP: ...harvard!adelie!infinet!rhorn Snail: Infinet, 40 High St., North Andover, MA (Note: harvard!infinet path is in maps but not working yet)