"So you are saying nobody can model the greenhouse effect at all, without imposing simplifications that destroy the utility of the model, even if they had enough past data to parameterize the model?"
From what I see from the shape of available historical data (ice core records, sediment records, paleogeological data),
climate fluctuations did occur on every possible time scale, from several years to hundreds thousand years, and the spectrum of fluctuations seems to be continuous. All the Milankovitch cycles seems to be just an artefact of limited period of observations, maybe with slight signature of synchronization with orbit precession. All this is a manifestation of a typical chaotic system. IMO, a correct climate model should be able to depict global glacial oscillations to some extent, but detailed deviations (local warmings and coolings, and turning points) are not predictable at all. A correct climate model should be able to hold the global climate trajectory within known historical limits (average surface temperatures from 12C to 22C) under a wide range of CO2 concentrations (up to 6000 ppm and maybe above), and for a wide configuration of continents and oceans. But immediate utility of the model will be very limited. More, interactions inside these models are very convoluted, so sometimes effects of changing a parameter may lead to orthogonal or even opposite effects, people should not be surprised with this, and should not reject half of trajectories as the climatprediction.net experiment is doing on the grounds that the trajectory does not meet their expectations. In short, things are very complicated in climate :-) :-) |
