[lambda]

Hi, I am extremely new to CFD, and am considering using openFOAM to create a basic climate model that can be used to proceduraly generate worlds for games and such.

One of the largest difficulties that I can see is minimizing the computational requirements to the point that the system can make a prediction that "looks good enough" (e.g. models temperature, evaporation, and rainfall, while at least considering the effects of atmospheric circulation currents and terrain ) within a few minutes of compute time on consumer grade hardware.

With these restrictive computational constraints in mind, can one model convection over a flat, horizontal plane with a temperature differential, and flow obstructions without simulating a "full" third dimension?

I'm wondering if you couldn't have basically three two-dimensional "layers", simulating air-flow near ground level, air-flow at high altitude, and a intermediary to simulate gas exchange between the two. If there's been work on this already, even just a research paper to start from, I could use it, but I really don't have the expertise right now to implement this from scratch myself.

Secondly, modeling flow restrictions caused by terrain "properly" would require expensive precision. mountains might only be a mile or two tall, with Everest being 5.5ish, while the thickness of the atmosphere worth modeling could be as much as 10 miles. half-mile precision, by using twenty layers of atmospheric cells would be computationally expensive. Is there a cheaper way to simulate flow obstruction, and the forced rising of air caused by mountains and high terrain?

Finaly, what am I missing? Is there a forum for climate modeling that I don't know exists that might be helpful? is there a computationally cheap library for this exact sort of thing already around? Other general recommendations that you wish you knew when you started learning CFD?