One of the criticisms leveled at lighting codes that require lights be below a certain Correlated Color Temperature (CCT), e.g., 3,000K, is that CCT does not directly indicate how much blue light an LED emits. Enter the spectral G-index. This parameter provides a single number that characterized how much short-wavelength light is emitted by a source:
Seems like a reasonable figure of merit for nighttime lighting.
I think a weighted average based on the amount of scattering expected at each wavelength would provide a better indication of a light's effect on sky-glow, but the G-index seems like a step in the right direction.