03-05-2012, 03:27 AM
Ted King wrote: I can understand some concern about some models if the situation in question has a very large number of variables to model, where many assumptions have to be made about those variables and there may be some debate amongst experts about the assumptions.
This is true of all models. They can get as complex as you want them to. The MO for pretty much any computer modeling effort is to create a number of simple models with a minimal number of variables first, each model aimed at testing certain variables. You then calibrate and validate those models with observational data, typically by trying to simulate past climate data. You can often make fairly good predictions of certain phenomena with the simple models. However, to understand how complex systems work you need to combine a number of simple models to create a complex one. When you do that, you add up the uncertainties in each model, so more complex models have greater uncertainty. If you actually look at the various papers that release AGW model results, you will find that there is a lot of information regarding calibration and quantification of uncertainties. The results are then usually released with a range of final results, reflecting 1 or 2 sigma probabilities. Of course, what happens then is that the press gets hold of it and only publishes the worst (or best) case scenario, with no explanation of the range of possible solutions.
