Baader BVRI Bessel Filter testing

     In principle there should be no serious problem with the transformations as long one includes several 'very red' stars in the standard-star set.  Expect the transformation to be non-linear, at least a quadratic if not a cubic.  An example is shown in J. Allyn Smith's Sloan standards paper:

https://ui.adsabs.harvard.edu/abs/2002AJ....123.2121S/abstract

...specifically look at Figure 15 at the very bottom of the paper.  This deals with a red leak in the Sloan u filter he used.  Expect something similar in transforming very red stars with an open-ended Ic filter (or any such filter), since in effect you're dealing with a red leak.  For the extreme cases in the plot, Smith specifies the stars --- there's your target list of test stars, by the way.  As has been pointed out, for the Baader I filter, which is basically open toward long wavelengths, the transformation will depend on how the sensitivity of one's detector falls off redward, which may itself depend on ambient temperature, possibly atmospheric water-vapor level as well because of a strong (and variable) H2O absorption band around 9500A.

     Arne argues that the red reference stars are likely variable, and that targets can be redder than any of the standards.  Well, yes if you don't pay attention.  The trick is to find either strongly reddened late-K giants or, especially, old dead (non-active) M dwarfs that have been calibrated by reliable observers.  Even if they are slightly variable in V, the variation in the V-I color-index (say) will be negligible, especially given the likely errors involved in one's own data.  So I disagree with the old lore that the situation is unworkable.  Plenty of the old M-dwarfs are not only Landolt standards but are also original primary UBV standards, and are simply not variable at the level we're talking about (a couple percent at most).  The standards in V-I color-index are defined out to V-I = 4.00, anchored by the M-dwarf star Wolf 359, a Landolt standard.  Unless you work with faint, strongly obscured stars, this will cover most cases you'll run into.  It does mean you want to observe that star whenever you have the opportunity (October through May) as a check on your system.

     An example is Wolf 358 = EE Leo (V-I ~2.8, only a couple degrees from Wolf 359), whose GCVS designation is based on the modest range of V magnitudes reported in the literature --- not accounting for likely transformation errors in the source data, where the red star is wagging up/down at the end of the transformation fit.  It is also a low-weight Landolt star.  From my observations this past spring Wolf 358 varied by 0.03 mag full amplitude in V with a period of about 65 days (i.e. the rotation period of the star).  Just a straight-line fit through the data, ignoring the variation, gives rms scatter only 0.010 mag.  Since the variation is from starspots, the range in the I filter would be something like 3x smaller, down in the millimag range.  So this star is plenty stable enough to use as a standard in the far-red.

     Similarly I have looked at the northern Landolt standard SA 41-204, which is a moderately-reddened early-M giant (Landolt V-I = 2.24).  On 28 nights (so far) from 2020 Aug to 2022 Jul, whatever variation there is is below 0.01 mag rms in V.  This was done differentially against three bright, fairly red non-Landolt stars, so there are potentially three more pretty good red stars to be calibrated in the standard system in this field.

     In case it helps, I have made a stripped-down list of red stars (mainly V-I > 2.0) from my working file of standard stars, copied here:

https://ftp.lowell.edu/pub/bas/starcats/red-std.txt

This includes coordinates (proper motions in milliarcsec/year for the M-dwarfs are important to find them 'tonight'), V, V-I, and the source.  Some bright and some faint stars, though remember they'll be ~ 3 mag brighter at I than in V.  Several of the stars are in the J. Allyn Smith plot.  Many are in fields that have blue and intermediate-color stars as well as the red ones.  Mostly equatorial or northern, but a few between -25 and -30 Dec that I am using for extinction determinations.

\Brian

I have always been puzzled as to why R aquared values are included in transformation coefficient results. It seems to me they add no useful information. The closer the coefficient is to one (steep plot, colour transforms such as Tbv) the closer the R squared is to one. The closer the coefficient is to zero (plot close to horizontal, filter transforms such as Tv_bv) the closer the R squared is to zero.

Roy