I read this paper and is very good work, but I think maybe I find one big problem that can make all their results not so reliable.
You see, they use MARVEL procedure to get energy levels from experimental lines, but their network have many “floating components” …..it mean some transitions not connected to main network, so they cannot get absolute energies for those states. To fix this, they use 44 lines from CDSD-2019 database to connect these floating parts. But CDSD-2019 is theoretical line list, not experimental!
This is problem because MARVEL is supposed to be only from experimental data, no? If they put theoretical lines in the network, then the energy levels they calculate become not pure empirical…..they mix with theory. And if CDSD lines have errors, then all the energies connected through these lines will be wrong also.
I am thinking…..why they not mention how they validate these CDSD lines? Maybe the uncertainty they give to these lines (0.0005 cm⁻¹) is too optimistic for theoretical data? And if these connecting lines are wrong, the error will spread in whole network, no?
Maybe I missing something…..but if they rely on theoretical data to build their empirical energies, then how we can trust the final energies is truly empirical? This seem like big compromise to their method.
What you think about this? Is this acceptable in MARVEL analysis or it make the whole dataset not pure?