from J BAA 110, 1, 2000 Feb

The use of modern catalogue data for variable star comparisons: a practical investigation

J Greaves

Visual and photometric V band light curves are compared. The visual data are recalibrated using the Johnson V magnitudes contained in the Tycho Catalogue1 and the effect of this substitution assessed. The exercise emphasises the fact that modern catalogue data is too contextual in its nature to be plucked from its source and forced into service on an arbitrary basis.

Introduction

Many machine readable star catalogues are now readily available to the owner of a personal computer equipped with modem2 and/or CD-ROM3,1 drive. The data are usually available in ASCII format and delineated as fixed length records with a columnar layout separated by spaces. These data are from various sources and are primarily intended for professional astronomers. They usually come complete with a description file explaining the catalogue's format and details of where the data came from. The origin and context of the data is an important consideration.

For example, is it a catalogue of primary data, directly derived through experimental work, or a secondary source based on analysis and calculations upon this primary data? Is it a compiled catalogue, derived from searches in the literature, or a combination of the latter alongside primary data: again are the data actual measurements or derived via calculation? With compiled data there is also the distinction as to whether or not the data in the literature and/or other source catalogues were critically assessed or checked in any way, or whether they were just copied directly. Such matters can be very important when it comes to the use of a catalogue.

Recently the Hipparcos and Tycho Catalogues1 have become available, constituting the results reduced directly from the Hipparcos Astrometric Satellite's measurements for approximately one hundred thousand and one million stars respectively. These catalogues contain an immense amount of data, often of independent nature, and provide a timely and useful data bank of positions and magnitudes.

However, people are not necessarily familiar with these catalogues, nor the fact that use of the photometric data comes with caveats. The mission was primarily an astrometric one, and photometric results are merely a bonus (albeit a significant one). The photometric data can be utilised, but only with due care and attention.

Of late much interest has been shown in the usefulness of Johnson V magnitudes from the Hipparcos and Tycho Catalogues for "upgrading" the values given for sequence stars in the BAA Variable Star Section's (BAAVSS) Binocular Programme. Utilising archived visual data from two stars upon that programme, alongside a re-reduction of this data from the original estimates via Tycho Vj values for the comparison stars, and also photoelectric photometry in the Johnson V band (PEP V), I will illustrate the problems and pitfalls that may befall any such "upgrading" if said due care and attention is not taken.

Catalogue Data

First comes the choice of catalogue. The Hipparcos and Tycho Catalogues1 are available on CD-ROM as part of CD-ROM no. 1 of Volume 17 of the publication. The Hippacos catalogue contains just over 100,000 stars, listing Johnson V magnitudes for the majority. However, these magnitudes are a mixed bunch. Some come from the literature, and are therefore a compilation of ground based measures and not connected to any measurements that the satellite may have made. Others are derived from Hipparcos measurements, but rely on Tycho measurements for their conversion from Hipparcos photometric magnitude (Hp) to Johnson V band (Vj) magnitudes.

The colour response for the Hipparcos experiment must be calibrated against B-Vj and V-IC (invariably from ground based measures, or even calculated from other criteria) as the transformation from Hp to Vj is not straightforward, and is very much dependent upon colour. The transformation of Hp to Vj for the stars with no ground based values for B-Vj can follow many different routes, as can some of the necessary subsidiary values (such as the V-IC, IC being in the Cousins' photometric system, which the Catalogue sometimes has to derive from transformation of Vj and IJ or V-IJ catalogue values).

The Tycho experiment, on the other hand, was a self contained and independent experiment. It consists of the data received from the star mapping instrumentation which the satellite used for orientation, i.e. to let itself know where it was pointing in space when the main Hipparcos experiment measurements were taken. In this respect it was very much a bonus mission. It had a better limiting magnitude than Hipparcos (pretty much complete to mag 9), and the Tycho V band was centred very near, and had a similar spectral response, to the Johnson V band.

Transformations from Tycho V to Johnson V are fairly straightforward (as outlined in the printed Volume 1 of the Catalogues) and are consistently applied. That is, all the Johnson V magnitudes listed in the Tycho Catalogue are derived from the Tycho V magnitudes as reduced from the direct measurements made by the spacecraft, and as such constitute a homogeneous whole. According to the introduction to the catalogues, well over half the stars have Johnson V magnitudes that have been shown to be good to at least +/-0.1 of a magnitude. This is also more or less true for the Johnson B-V values listed in the Catalogue.

The above, and the fact that the Tycho Catalogue contains data for over 1,000,000 stars, contributes strongly towards making the Tycho Catalogue the catalogue of choice for updating sequence star magnitudes in the binocular range. Also, due to its nature, any future re-reduction of the Tycho catalogue photometric data is liable to be a global re-reduction, further updating all the magnitudes with respect to each other, and not just a local update as would be the case with the heterogeneous Vj data in the Hipparcos catalogue.

Work is apparently underway at the United States Naval Observatory to convert the positional USNO A1.0 CD-ROMs' magnitude data into photometrically usable forms via a re-calibration against the Tycho photometric data; this could extend down to magnitudes of at least 15, if not all the way to the approximate plate limits of magnitude 19. This would give an internally consistent data bank of reference magnitudes stretching from the brightest to the faintest variables on most observing programmes, using whatever instruments, sufficient to fulfil any amateur variable star observer's dreams.

The Data

Kevin West4 made available his PEP V (Photoelectric photometry in the Johnson V band) for the stars UX Draconis and VY Ursae Majoris, details of analyses of these two stars have already been published.5 The BAAVSS archivist made available visual data from the BAAVSS computer archive6 contemporaneous with the PEP V data in the form of Julian Date, Estimate and Derived Magnitude. UX Dra is a semiregular variable of range 5.9 to 7.1 V, spectral type N0 (C7,3).7 VY UMa is an Lb variable of range 5.9 to 7.0 V, spectral type N0 (C6,3).7 These two stars are fairly typical of the usual BAAVSS Binocular Programme stars in that they are quite red 'long period' variables of low amplitude and slow time variation.

SAO7 identifications of the sequence stars for UX Dra and VY UMa as derived by Patrick Maloney were made available via the BAAVSS Director, Gary Poyner.8 Cross identifications were made to the HD [Henry Draper] catalogue9 using a machine readable cross identification file.10 These were cross referenced to the Tycho catalogue (which contains HD numbers as cross identifiers) in order to derive the Tycho listed Johnson V magnitudes for these stars. Some confusion existed with respect to large differences between some of the Tycho values and the nominal values given in the BAAVSS UX Dra sequence which entailed a direct re-identification of these stars using the Guide 6.011 software package and the original sequence chart.12

The PEP V data was utilised directly, after conversion from dV (the difference in magnitude between the variable and the comparison star) to true V via the Tycho listed Johnson V magnitude of the comparison star. Note that West uses a check star in his photometry,4 such that the dV data provided came from observing runs where the difference between the comparison star and a check star was never more than 0.02 V.

The PEP V comparison star for UX Dra was HD 178089, which was also BAAVSS sequence star B for this star. The PEP V data was reduced using the Tycho Vj of 6.56 for this star, as was the re-reduction of the BAAVSS visual estimates utilising Tycho Vj values. The BAAVSS reduction uses a magnitude of 6.5 for this star, the resultant discrepancy of 0.06 magnitudes can be considered negligible, especially in comparison to the distance between the PEP V and visual lightcurves. The PEP V comparison star for VY UMa was HD 92354, its Tycho derived Vj of 5.74 being used for reduction of both the PEP dV values to PEP V and the BAAVSS visual data Tycho re-calibration. This is also comparison star F in the 1983 BAAVSS sequence, but the value used in the BAAVSS reduction was unknown to the author at the time of analysis. These comparison stars act as a potential link between the photometric and visual data.

The visual data6 arrived already reduced to magnitude values via the 1982 and 1983 sequences of UX Dra and VY UMa respectively. Using the Tycho Johnson V values for the comparison stars the data was re-reduced via the estimates to a set of Tycho derived Johnson V magnitudes. As both visual lightcurves showed much scatter they were averaged to 20 day means to enable the practical investigation of the results.

Results

VY Ursae Majoris

The raw visual data is shown in figure 1, to illustrate the difficulty in using such data directly. Figure 2 shows the same data with the original magnitude reduction (to the BAAVSS 1983 sequence, henceforth simply 1983) and the Tycho derived Johnson V band reduction (henceforth Tycho) having been averaged to 20 day means.

Figure 1: BAAVSS visual data for VY UMA reduced to both the 1983
and Tycho sequences and plotted as raw visual observations.
Also included is Kevin West's photoelectric photometry in the
Johnson V band for VY UMa.
Figure 2: The lightcurves for the twenty day means of the VY UMa
visual data. The complete PEP V (not means) for VY UMa is also
shown, with circles denoting individual observations.
Although the twenty day means improve the situation with respect to the visual data, the re-reduction via Tycho has made little if any difference to the overall shape of the visual lightcurve, although it has brought it slightly closer in magnitude to the PEP V lightcurve. Only a rough overall similarity exists between the visual and PEP V lightcurves (where each individual observation is denoted by an open circle), the fine detail being quite different for the two classes of data.

It was decided to check if the Tycho reduction improved the situation in any way with respect to the analysis of the data. In figure 3 are plotted the Fourier transforms from a 50 to 500 days range (0.02 cycles per day to 0.002 cycles per day) for the raw PEP V data and the two individual twenty day means for the visual data. As can be seen, use of the Tycho as opposed to the 1983 sequence has made little difference in the shape of the transforms, both sharing quite distinct differences to that derived from the PEP V data.

Figure 3: Fouriergram of the individual data sets for VY UMa
(see text); cpd = cycles per day.
Figure 4: The lightcurves for the twenty day means of the BAAVSS
visual data for UX Dra (see text) and the complete PEP V for the
same period with individual PEP V observations denoted by circles.
Figure 5: Orbit (apparent)
diagram for OS363, sequence
star 6 on the 1982 Nov 7
BAAVSS chart for UX Dra.
An x marks the periastron
point. Note that the current
separation is barely 0.2.
UX Draconis

Here we have a quite different situation, and interestingly all but one of the sequence stars used in the 1982 BAAVSS chart has had its value derived visually. Figure 4 shows the raw PEP V and twenty day means light curves for the visual data. This time there have been two separate reductions with respect to Tycho. The grey line simply denoted Tycho represents a re-reduction made by using the Tycho derived Vj values for the sequence stars on the BAAVSS 1982 Nov 7 chart. As can be seen it differs quite significantly in structure from the 1982 sequence light curve as shown by the thick black line.

This highlights the problem of just simply taking a value from a modern catalogue and arbitrarily making use of it. It just so happens that sequence star 3 is the binary star S2571, consisting of a magnitude 7.6 and 8.3 pair separated by 11.3 at a position angle of 16deg.13 The Tycho experiment instrumentation was able to separate this binary and therefore it gives a derived Johnson V magnitude of 7.64 for this star. Binoculars, on the other hand, do not (usually) separate stars as close as this and therefore "see" these stars as a combined image of magnitude 7.1 (calculated using 7.64 and 8.3 for the magnitudes). The 1982 sequence lists star 3 as being of estimated visual magnitude 7.0.

With this knowledge, and making the relevant adjustments, we get the thin black line denoted Tycho+ in figure 4. This line follows the structure of the 1982 sequence line better than the thick grey line for the Tycho light curve. Deviations between Tycho and Tycho+ highlight the parts of the light curve where sequence star 3 was used: note for example that the grey line of Tycho is concomitant with the thin black line of Tycho+ after JD 2450450 (and is thus "hidden" on this plot) due to sequence star 3 not being used in the making of estimates here.

Unfortunately, whichever sequence is used, the shape and structure of the PEP V light curve is not well reproduced.

Other sequence stars for UX Dra also suffer from a vagaries, but these idiosyncrasies rarely effect the outcome of a re-reduction via Tycho values. Sequence star 6, for instance, is the binary star OS363 of 0.38" semimajor axis whose orbit14 is shown in figure 5: the secondary is currently approaching periastron when the separation of the pair will be about 0.17". In this instance the magnitude 7.6 and 7.8 component stars are just as unresolved by the Tycho experiment as they are by binoculars.

Figure 6: Fouriergram of the individual data sets for UX Dra
(see text); cpd = cycles per day.
Ironically in the case of UX Dra we have a star which is so barely regular that the strongest signal derived from a Fourier transformation of the visual data between the range of 50 and 500 days (0.02 and 0.002 cycles per day) is that of the annual alias, no matter which visual sequence is used, as shown in figure 6. The next strongest signal is at a period of 167.5, 164.3 or 165.2 days for the 1982, Tycho and Tycho+ data respectively. On the other hand the PEP V not only has no annual alias to speak of, but also gives but one good peak throughout the range, at 179.2 days. So, for this star little improvement was achieved, no matter which visual sequence was used.

Conclusion and Discussion

Although there are no doubt numerable instances where the BAAVSS Binocular Programme sequences could be usefully improved, the above two examples emphasise the need to be cautious during any such improvements. An (unpublished) derivation of Tycho Vj values for the larger part of the Programme's sequences undertaken by the author led to an earlier opinion that these values were rigorous and consistent enough to be capable of totally replacing the current values used, as long as all caveats regarding multiplicity and variability were adhered to.

The more practical investigation above leads to a different conclusion. In most of the cases re-reduction with Tycho as opposed to traditional sequence values led to little or no meaningful difference to the lightcurve. It definitely did not bring it any closer to the PEP V lightcurve, despite some commonality of comparison stars. In fact, in one instance, specifically that of sequence star 3 for UX Draconis, arbitrary replacement of the current value with the Tycho Vj value actually made the situation worse.

Therefore the most prudent procedure will be 'if it ain't broke, don't fix it'. However, there will be instances when sequences will be broke, and in need of fixing. In these instances the Vj values contained in the Tycho catalogue are the most homogeneous, consistent and sizeable set of values around, as well as coming from a well documented source that is also becoming widely available (e.g. reference 6). But these values can not be taken and used willy-nilly; their context has to be taken into consideration, such that any effects of multiplicity and/or variability upon the values are not ignored.

If usable values can be found and discrepancies still persist, the Tycho catalogue can be used further to check that the B-Vj of the sequence star is also suitable for the type of variable under examination. If not, then these may be instances where use of the particular sequence star in question should be discontinued.

The colour index "problem" has been investigated before,15 with the suggestion that the relationship v=V+0.16(B-V) is appropriate for the conversion of PEP V to visual magnitudes. As it is in fact the variable star that is being estimated, and its 'distance' (in terms of magnitude) from comparison stars that is being judged, the author feels that any such correction should be left until after the observations have been reduced, using the catalogued mean B-Vj of the variable star. There is a viewpoint that sequence stars should have their V magnitudes (if used) converted to visual ones for practical purposes, however the author stresses the opinion that it is the variable and not the sequence stars that is constantly being placed under examination via observation, and any and all such conversions should only be carried out upon the derived magnitudes for the variable. In any case, an archive of observational data should only contain primary data, and not data that has been pre-reduced via some convention and/or other calculation.

An example is shown in figure 7a, where the PEP V for VY UMa is transformed to equivalent visual magnitudes via its Tycho Catalogue listed B-Vj of +2.52 and plotted against twenty day means of the observations reduced via Tycho Vj values [the Tycho reduced data could just as readily be 'converted' to PEP V via V=v-0.16(B-Vj)]. The fact that B-V is also liable to vary slightly along with the magnitude is highlighted by a greater discrepancy at the brighter end of the range, but the adjustment does make it easier to compare the shape and structure of the Tycho reduced visual curve with that of the PEP V curve. Figure 7b shows the case for PEP V transformed to visual magnitudes against the BAAVSS 1983 sequence: it can be seen that there is a better fit to the PEP V 'standard' when the Tycho sequence is used.

Figure 7a: Twenty day means visual data for VY UMa as reduced via
Tycho Vj sequence values compared with the PEP V
transformed to visual via v=V+0.16(B-Vj(Tycho)).
Figure 7b: Twenty day means visual data for VY UMa as reduced via
BAAVSS 1983 sequence values compared with the PEP V transformed to
visual via v=V+0.16(B-Vj(Tycho)).
Note differing magnitude scale.
The same transformation was undertaken for the UX Dra data, using the Tycho catalogue listed B-Vj of +2.86. Figure 8a shows the visually transformed PEP V in comparison to the Tycho Vj sequence reduction as modified to account for the binarity of sequence star 3. Figure 8b shows it in comparison to the BAAVSS 1982 sequence (which, is primarily based upon visual estimates for the sequence stars). There is little to choose between the two.

Figure 8a: Twenty day means visual data for UX Dra as reduced via
Tycho Vj sequence values (including sequence star 3
adjustment for unresolved binarity) compared with the PEP V
transformed to visual via v=V+0.16(B-Vj(Tycho)).
Figure 8b: Twenty day means visual data for UX Dra as reduced via
BAAVSS 1982 sequence values compared with the PEP V transformed to
visual via v=V+0.16(B-Vj(Tycho)).
Note differing magnitude scale.

Acknowledgments

The author acknowledges the help of Kevin West without whose PEP V the assessments made in this paper would have been less meaningful, this PEP V acting as a referential standard in the investigation. The author is also grateful for information made available by members of the BAAVSS that were germane to the article. The referees' assistance in pointing out obvious gaffes and typographical errors is also greatly appreciated.

note added in press: This paper was written before the publication of the Tycho2 and USNO A2.0 catalogues. Neither essentially affect the conclusions drawn.

References

  1. The Hipparcos and Tycho Catalogue. ESA SP-1200, July 1997. Vol 1 Intro., Vol 17 data on CD
  2. CDS, Strasbourg, FRANCE: ftp://cdsarc.u-strasbg.fr/ and/or ADC, GSFC, NASA, Greenbelt, MD, USA: ftp://adc.gsfc.nasa.gov/
  3. Selected Astronomical Catalogues Volumes 1 to 4, ADC, GSFC, Greenbelt, MD, USA
  4. Kevin West, private communication
  5. Lloyd, C., West, K., IBVS 4335
  6. McAdam, D., BAAVSS Computer Archivist, private communication
  7. Smithsonian Astrophysical Observatory Star Catalog J2000.0, ADC SAC vol, op cit. (ref 3)
  8. Poyner, G, private communication
  9. Cannon, A. J., Pickering E. C., Henry Draper Catalogue, Harv. Ann. 91-99 (1918-1924). ADC SAC vol 1 op cit. (ref 3)
  10. Roman, N. G., SAO/J2000/HD/DM/GC Cross Index, ADC 1991. ADC SAC vol 1 op cit. (ref 3)
  11. Gray, B., Guide 6.0 CD-ROM Star Chart, Project Pluto 1997+, Bowdoinham, ME, USA
  12. Pickard, R D., private communication
  13. Worley, C. E., Douglas, G. G. Washington Double Star Catalogue 1996 (USNO),. ADC SAC vol 3 op cit. (ref 3)
  14. Worley, C. E., Heintz, W. D., 4th Catalogue of Orbits of Visual Binary Stars, Publ. USNO (2) 24, part VII. ADC SAC vol 1 op cit. (ref 3)
  15. Howarth, I. D., Bailey, J., J. Brit. Astron. Assoc., 90, 265-272, 1980