Well, I'm entirely unqualified to assist but I won't let that stop me trying
. Sorting out the contribution of different types of light source to light pollution is a useful study but doing it spectroscopicly will be 'tricky'.
The first issue is to obtain a spectrum of skyglow. I can think of two approaches to make the measurement yourself. You either try to get a spectrum of a piece of empty sky or you get a spectrum of an object whose innate spectrum is known and then do spectral subtraction. This latter approach is what atmospheric gas monitoring stations do (at least at UoW), they measure the solar spectrum from the ground and then subtract the solar spectrum obtained from satellites. I'm not sure what 'standard candles' are available at night that have spectra measured well enough to be useful - maybe even the moon. Alternatively, is there satellite data with spectral information? Of course the spectrum of light emitted to space will differ from that reflected back off the atmosphere but if direct measurement is too difficult this may be a useful proxy.
Once you have your data you must interpret it and this will not be simple. Mixing models with two end-points are fairly straightforward but in this case there is 3 or more end points. The maths of extracting the contribution of different sources from a spectrum is known and used routinely in astronomy, chemistry and earth science. This math will determine the minimum number of lines or bands that must be measured. This must be known before you start gathering data, lest you wind up with data that is not fit for purpose. Hopefully there is an application that can be used to interpret your data because you'd sooner not write your own code! I'm not a programmer so if I had to design my own data analysis I would head for a spreadsheet - either that or head to the pub
.
