The fluid equations for AGCMs are discretised using either the finite difference method or the spectral method. For finite differences, a regular grid (i.e. with constant grid spacing) in latitude and longitude is most common.
So one of their approaches to solving Navier Stokes is to use finite difference methods, while the entire field of fluid dynamics moved away from that decades ago (because it is inherently unstable!). This is why all the climate models predict wildly different outcomes for the future: their results are only as good as their boundary conditions, which can only be measured in the present. To me, that's kindof a big deal.
However, the Proceedings of the National Academy of Sciences journal is a closed publisher and the article is locked behind a paywall, so I guess the vast majority of us will never know.
This has bothered me a lot lately. Bittorent to the rescue?
My computer can beat up your computer. - Karl Lehenbauer