Another one not quite VSX related. I am presently going over the classic Parenago catalogue of stars in the Orion Sword region. Much of the Orion literature uses these numbers to identify stars (especially pre-GAIA etc). There is a rough catalogue of the 3000 entries available in machine-readable form from VizieR, but it shows only approximate coordinates for the stars, and little else from the original 500-page monograph. Only maybe half of the stars have accurate coordinates in SIMBAD. I am working from a very good list made some 15 years ago by star-chart expert Mati Morel in Australia. Anyway, as part of the revision, I am having a look at LAMOST spectra for stars that appear there. (The LAMOST DR4 spectra are available through VizieR.) This shows that many of the spectral types in VSX (harvested from the literature) are wrong. VSX indeed copies the literature faithfully, but often the types were estimated in the blue part of the spectrum, traditionally from about 3800A to 5000A. However, the spectra of later-type stars, typically early-M variables of the T Tauri type, are veiled by what is assumed to be hot gas surrounding the stars (in the accretion disc etc) that ends up making the blue spectra look like an earlier class. George Herbig and others recognized this even 60 years ago, so this is not new.
The LAMOST spectra run from about 3700A out to 9000A, thus covering the entire visible and far-red region. Some of the spectra are faulty in various ways, but usually they are suitable for classification if the signal-to-noise is adequate. The veiling is much weaker in the far-red, and in the professional literature nowadays the classifying of late-type stars is done from the TiO bands in the far-red. An example is V500 Ori, which was classified in a 1983 paper by Merle Walker from a blue spectrum as K2:Ve (1983ApJ...271..642W), which is what VSX shows. The LAMOST spectrum in the blue is consistent with this. However, the star shows strong TiO bands in the far-red, so it is not a K star. My estimate is M2/3Ve (i.e. a somewhat uncertain temperature type between M2 and M3). There is emission in the Balmer lines and CaII H&K in the violet (the latter certainly not from the overlying nebulosity).
Among ordinary stars one would normally conclude that divergent spectral types in the blue-violet and far-red indicate a spectroscopic binary with hotter/cooler components. But I have so far seen only one such obvious to me in the Parenago list. This is V374 Ori, which VSX shows as GCVS variable-type 'IN', and only H-alpha emission reported in the spectrum (from Guillermo Haro's early H-alpha survey in the early 1950s). The star has been recently identified as a cv, and the LAMOST spectrum shows the unmistakable very broad Balmer lines of a white dwarf in the blue (with sharp emission cores), and the TiO bands of an M1 dwarf in the far-red. Wow! Interestingly, the GAIA parallax places the cv at the distance of the Orion Nebula (380 pc). The system cannot be as young as this implies, and must be unrelated but at a similar distance.
I'm getting close to the end of the work, which I'll supply to Sebastian and others --- hopefully to end up at VizieR so that it can become generally available.
Yes, Brian, please send us this file.
We are now able to easily import spectral types for all stars without that information in VSX because we added a spectral type column to our VizieR file, which is our main tool to do the updates.
In the case of revisions of stars that already have spectral types in VSX, it is more difficult, because unless someones does the work you are doing, checking objects individually or having specific knowledge about the quaility of the different sources, judging what is better and what is not becomes a non-trivial task.
So keeping in mind that this list has already been vetted by you and types have been improved, we will be happy to just modify the current types with these data.
And since you mentioned V0374 Ori I will check its classification right now!
Thanks and cheers,
The anomalous LAMOST spectrum of V374 Ori, which looked like a typical white-dwarf + M-dwarf binary, needed confirmation. After some skepticism from Sebastian (and his discussion with Taichi Kato), we found that the coarse spectrum available in GAIA DR3 was suitable. The GAIA spectra are accessible through VIzieR as item I/355/spectra, which displays the spectra using a link on each entry. This shows that there is _no_ hot component, but instead only a red star of type about M4 or M5 with H-alpha emission. So V374 Ori is a typical T Tauri-type star.
Going over two long lists of LAMOST DR2 spectra in the original Kepler field a few years ago, I noticed a number of cases where there the spectrum seemed to have no relation to the star color-index, distance etc. The suspicion was that there are occasional bookkeeping slips such that spectra are assigned wrong coordinates. I had thought these had been fixed in DR4, but evidently a few remain. So we think the anomalous V374 Ori spectrum results from some mix-up in the database. Caveat emptor!
Finding the GAIA spectra in VizieR was a pleasant surprise. They are very coarse, not high enough resolution to classify stars properly, but better than no spectrum(!), and of course there are zillions of them. As part of looking into V374 Ori, I ended up examining several dozen spectra for stars of all types just to see how things looked. The late-type stars with strong TiO bands are clear, and I could distinguish them from carbon stars. Early-type stars can be distinguished only approximately (like whole-letter class at best), though since there is good response in the violet there is some discrimination of B and A-type dwarfs versus supergiants from the Balmer decrement. And early-B stars show almost no decline in the violet. Also strong Balmer (and other?) emission is evident.
Does this last posting by Brian mean that, when updating the VSX, we should routinely check I/355/spectra (as well as LAMOST) when no Spectral Type is present in the VSX entry?
Can we assume that I/355/spectra will be listed by the VizieR search if it (the catalogue) contains an entry for the star of interest?
This is a policy question for Sebastian, but I think you'd want to be quite careful on this point. Do note that the spectral types are not given explicitly in the GAIA file. I am evaluating the spectra myself by comparing the traces with other spectra of well-known type from lists of MK standards and others. The LAMOST DR4 catalogue does show a spectral type, but often these result from some sort of automated fit that may not be very good. Again, I examine the linked spectra one-by-one. From my examination of (so far) a few hundred GAIA spectra, they seem to be useful only for types K5 and later (plus carbon types) without sophisticated analysis. For earlier-type stars there is simply not enough resolution to classify them very well. Briefly, this is because the Balmer lines and CaII H&K, which are the starting point of the classifications, are too smoothed-over to be useful. I am also discovering that when S/N is low the GAIA spectra can look rather bizarre with some cyclic hysteresis or something obviously instrumental in nature (not simply scattered data-points) --- not sure what's going on there, but it can be misleading.
Our policy is to take spectral types as they are published in papers, and not by looking at the spectra themselves. We would need to know how experienced the observer is to approve his/her determination. Not feasible, only in some individual cases.
But in fact, there are spectral types published in the table I/355/paramp and they meet the above criteria, so we are using them if there isn't anything else. They are just rough spectral types, no subtypes are given, so it is just to have an idea of what the star is and may help with some classifications. I've seen some issues with hotter stars but at least all variable red giants I checked are classified as M :)
We will make a massive update from this catalogue for stars with no spectral type in the future.
I am not taking too much time to use them in our current submissions and won't specifically ask for that to our VSX submitters but yes, you can check them and use them if you want.
The reference to provide is:
Creevey, O. L.; et al., 2022, Gaia Data Release 3: Astrophysical parameters inference system (Apsis) I -- methods and content overview