There are usually quite a bit fewer than 100 altitude slices. There may be as few as 10: http://www.ipcc-data.org/guidelines/pages/gcm_guide.html http://www.ipc

There are usually quite a bit fewer than 100 altitude slices. There may be as few as 10: http://www.ipcc-data.org/guidelines/pages/gcm_guide.html http://www.ipcc-data.org/guidelines/pages/gcm_guide.html Simulations are typically done in one day time-steps or less. The thing is, there is a lot of physics going on in each cell, and in some respects reducing the cell size makes things more problematic rather than less. For example, the effects of a supercell thunderstorm can dominate the atmosphere over tens of kilometers. With a 100 km (60 mile) scale you can plausibly average over those effects within a cell, as you can reasonably claim that most such "local" phenomena will be happening well away from cell boundaries. So as you reduce the cell size, you start having to think about the internal details of local disturbances that you were previously averaging away and hoping to catch in your overall parameterizations, and the model complexity goes up tremendously, likely without becoming a whole lot more accurate. Even at these large scales the amount of physics that goes on within a single cell is huge, and getting all the heat and mass balances right is fantastically difficult, which is where a lot of the computational power goes. So it isn't really a question of halving the grid size while keeping the physics the same, because you are going to have to substantially revise your model to accommodate the new scale. I'm sure this work is ongoing, but it's extremely complex and difficult.