Sure, we could use an standard atmosphere and somehow determine the amount of light reflected back down from all the surface lighting shining up. If we know the level of the surface lighting within some radius, especially by looking down on it, we could model how much we might expect to be driven back down to our eye.
We might have to measure the lighting and the albedo of the surface we see from above, as I believe the LP map does, or just a simple measure of total radiance as seen from space. I agree, it's more straightforward to just measure the light that is raining back down on us. But, surely surface radiance can be modeled to that amount of light reflected back down by standard or average atmospheric properties.
I like the map above, even if it (or all models) might not be completely accurate. I especially like the radiance measure, I am thinking it might be as useful looking down on all those lights as measuring the sky itself with a meter, if we have one. The meter is more accurate, surely, and more real time. But, I don't have one.
"...places where there is little light pollution and dark skies can have greater radiance than places with a great deal of light pollution."
I have to ponder that. I can't get my head around why a dark sky would have greater radiance than a light polluted sky given the idea we're looking down on all of that light, much of which is driven through the sky as radiance from the surface. That light is traveling into space where we observe it, some of it is redirected toward the ground. In a dark sky, this is also happening, but at a much lower radiance because there are few things radiating at visual frequencies.
Once we get past all that, we still have the problem of converting units of surface area in cm^2 looking down to units of magnitude per arcsecond^2 looking up. And whether the units of either measure are even compatible. (I just went in over my head.
)