If mean a static shadow map or "lightmap" that WED generates, it's not really one big file, but a number of files. The coordinates are generated automatically by the compiler. The UVs are flattened and sized accordingly to polygon size in the level, and selected resolution from the compiler..the texture coords are assigned to the second set of texture coords on the vertices. Then lightmaps are calculated and rendered using I believe a depth-based approach (similar in theory to how the real-time shadow maps in my Sphere Engine are calculated). The bitmaps themselves are just normal 24-bit integer graphics files that are stored internally in the wmp file...they are normally in color..Then they are applied in the engine using simple multitextureing and a multiply2X blend mode.

I hope this helps. Is there some reason you want to know about this?