This approach makes expensive direct beam computations even when solar zenith angle is < 0, but masks those results out in layers where the direct beam does not reach. This will have practical impacts only in columns where the top of the column is in sunlight but the bottom not (or if users aren't removing columns entirely below the horizon).
Adopt suggestions by @peterukk in #215
This approach makes expensive direct beam computations even when solar zenith angle is < 0, but masks those results out in layers where the direct beam does not reach. This will have practical impacts only in columns where the top of the column is in sunlight but the bottom not (or if users aren't removing columns entirely below the horizon).