Of course if we want we already know enough to model the climate with relatively astounding accuracy — considering it really does not change much — and not just 100s of years out but even billions of years into the future.
Should we be forced to admit simple truths? For starters, the grid blocks that are used in the constructions of GCMs are too large to accurately simulate real-world climate conditions such as thunderstorms, hurricanes and other natural processes that transfer huge amounts of energy from the surface of the Earth to the stratosphere. Accordingly, various `parameters’ to account for what cannot be conceptualized and captured and, in any event, cannot be be quantified due to the limitations of the computing power availale on planet Earth. As a result, simple approximations of real-activity are used because the ability of GCMs to accurately represent actual, observable physical processes is impossible.
As these various parameters’ are selected simply to make the model agree with empirical data, the resulting models cannot be used to as evidence of reliability in capturing reality or forecasting the future. No one honestly argue that the use of such parameterization is proven or that the use of GCMs generally is justified because we cannot assume that a “model that has been ‘tuned’ to give a good representation of certain key observations will actually provide any predictive ability.
Such closely tuned models fail to satisfy a key requirement: they have not been formally evaluated to assure that the ‘tuning’ parameters used do not exceed the degrees of freedom based on GCM’s observational testing schemes.