Should be able to take a 2D grid heightmap and use this for the weather simulation. By using a 2D heightmap, it should be possible to take advantage of numpy/scipy optimizations to dramatically speed up these steps.
Now just sampling these all from Perlin noise for the most part. The wind points all to the right and a little away from the equator, then perlin noise is used to rotate these wind vectors.
Should be able to take a 2D grid heightmap and use this for the weather simulation. By using a 2D heightmap, it should be possible to take advantage of numpy/scipy optimizations to dramatically speed up these steps.