Ok, Robin. Yes, of course, the relationship between spectra has its significance, as does the simple comparison. With soft matlab, it is also straightforward to have them graphically. The problem arises when one wants to establish the relative weight between the pixel values (or relative intensities) of different bands or rather spectral frequencies.
Fortunately, polarisation is not detected. In that case I would abandon the study, I have no desire to tackle more advanced mathematical problems as well.
I spent the afternoon today in front of two spectra, zenith and 30° anti-solar. Tomorrow I will climb a peak of the Gran Sasso (2700 m) to take some spectra above the vapours of the valley floor.
British Astronomical Association
Supporting amateur astronomers since 1890