Brief Description of GSC 1.2

STScI ACDSD MAST CASB GSC I


Introduction

A collaboration between Space Telescope Science Institute and the Astronomisches Rechen-Institut has produced a new astrometric reduction of the Guide Star Catalog (GSC). This new version, GSC 1.2, has dramatically reduced the plate-based position-dependent and magnitude-dependent systematics present in GSC 1.1.

Return to top.

General Properties of GSC and GSC 1.1

The Guide Star Catalog (Lasker et al. 1990, Russell et al. 1990, Jenkner et al. 1990) contains approximately 19 million stars and other objects in the sixth to fifteenth magnitude range. The GSC is primarily based on an all-sky, single epoch collection of 6.4x6.4 degree Schmidt plates. A 1982 epoch ``Quick V'' (20 minute exposure) survey was obtained from the Palomar Observatory for centers at +6 degrees declination and north. For the south fields, 50-75 minute exposure plates, with material from the UK SERC J survey (epoch approx. 1975) and its equatorial extension (epoch approx. 1982) were used. These plates were reduced using a cubic polynomial. The reference catalogs used for this reduction and their properties are summarized in Table 1. The catalog created this was is called GSC 1.0.

           TABLE 1 - GSC 1.0/1.1 Reference Catalogues

  Catalog         GSC Plate centers      Precision at 1980
  AGK3                  dec > 2 deg         0.33 arcsec 
  SAOC             0  > dec > -60 deg       0.79 arcsec
  CPC             -65 > dec                 0.57 arcsec
At a later stage a number of improvements were introduced. A new version, GSC 1.1, was completed and distributed on the CD ROM dated August 1992. These improvements included corrections for spurious entries, for overexposed images for bright stars (V <7), and for different entries for the same object.

Over the last few years numerous studies have demonstrated that wide field-of-view Schmidt plates reduced solely with a traditional global plate model suffer from plate-based position-dependent systematics. These types of systematics can be detected from the differences in the overlapping area of a plate pair ( i.e. the differences obtained from the same GSC star located on more than one plate.) Figure 1 demonstrates the typical feature found for overlapping plate pairs in the GSC 1.1.

Another method to detect and quantify the systematic errors is to compare the GSC with an independent high-quality approximately coeval set of positions. For this comparison we used the Carlsberg Meridian Catalogue (CMC), the effects of proper motion for the approximately ten years difference between the epochs of the CMC and GSC were not considered. By binning and averaging the GSC-CMC residuals over a plate-based coordinate system, one obtains detailed maps (called masks , Taff, Lattanzi, and Bucciarelli 1990) of the scale and direction of the remaining residuals. Figure 2 is a mask constructed with the CAMC.

Both Figure 1 and Figure 2 demonstrate the strikingly clear swirl pattern of systematic errors remaining the GSC 1.1. These systematic errors were detected by Taff et al. 1990. Such studies along with others have shown that the GSC 1.1 has mean positional errors which are smallest at the plate center (for V <8, 0.40 arcsec, north; and 0.70 arcsec, south) and increase rapidly towards the edges, (1.0 arcsec, north; 1.2 arcsec, south). This global pattern of positional errors is a consequence of an insufficient plate model, physical deformation of the plate, and the characterists of Schmidt optics. In addition, the positional errors have a north-south asymmetry which is attributed to a combination of the different exposure depths and the less accurate catalogs used for the reduction in the south.

Besides the position-only dependent systematics, the GSC 1.1 also suffers from systematic errors which are a function of magnitude and radial distance from the plate center (Morrison, et al. 1996). The effect is small for radii under 2.7 degrees from the plate center, then rapidly increases. The average offset of the faint stars (15th mag ) versus the reference stars ( 10th mag) is 0.2 arcsec to 0.3 arcsec at a radius of 3.0 degrees and increases to 0.8 arcsec to 0.9 arcsec at the corners (radius of 4.2 degrees.) However, since the rapid radial dependence appears at 2.7 degrees which is also when the onset of vignetting by the Schmidt corrector occurs, it suggests there is a connection between vignetting and the magnitude effect. A more thorough study for the source of this effect remains to be performed.

Return to top.

GSC 1.2 Reference catalogs

The reference material for this new reduction is the Positions and Proper Motions Catalogue (PPM, Roeser et al. 1991, Bastian et al. 1993, and Roeser et al. 1994) and the Astrographic Catalogue (AC). Their properties are summarized in Table 2. Note the PPM positions have significantly better precision at the mean GSC plate epoch (1980) than those used to reduced GSC 1.0.

           TABLE 2 - GSC 1.2 Reference Catalogues

  Catalog         GSC Plate centers          Precision 
  PPM North              dec > -2.5 deg        0.23 arcsec (1980)
  PPM South              dec <-2.5 deg 0.09 arcsec (1980) AC all-sky 0.30 arcsec (1906) 

The AC was used to remove the mean systematics common to all the plates and its role in the new reduction will be described later in more detail.

Return to top.

New Reduction

Using the filter method (Roeser et al. 1995) GSC 1.1 was placed onto the PPM system. In this first process systematic errors originating from the coordinate system ( i.e. the use of catalogs not on the FK5 system) were reduced. A second correction to the positions removed the majority of the plate-based position-dependent systematics errors. This improvement was based on a mask built from the GSC/AC matches in the GSC plate-based system. Using these mean residual maps (similar to Figure 2) each GSC position was corrected by applying the residual corresponding to its location on the plate.

The correction for the radial magnitude effect was slightly more complicated to determine. The difficulty with trying to find magnitude-dependent terms on plates which cover a broad magnitude range (6-15th mag) has been the lack of astrometric reference catalogs covering the same range. Most reference catalogs have an approximate limiting magnitude of V=10th mag. Therefore, no magnitude dependent term for fainter stars could be reliably found by reducing the measurements based only on comparisons with reference stars. A dense all-sky reference catalog covering a broad magnitude range (preferably the same as the GSC) would be needed to quantify and remove the magnitude-dependent systematics (the magnitude range of the PPM is inadequate to support its usage for determining a magnitude term directly). The present lack of such a modern epoch catalog forced us to use a rather unconventional approach by using the Astrographic Catalogue.

The Astrographic Catalog contains 10 million measures of roughly 4 million stars with an accuracy between 0.2 arcsec to 0.4 arcsrc, depending on the observatory zone and a limiting magnitude of 12th mag (but there are many cases where it is as faint as 13th to 14th mag.) The difficulty with using this catalog is that it has a mean epoch of 1903 and contains no proper motions; therefore there are approximately 80 years difference between the epochs of the two catalogs and neither catalog contains proper motion information.

The poorer quality of the AC material, the zonal systematics, and some of the effects of proper motions are accounted for by stacking all the AC plates onto the GSC plate-based coordinate system and then determining the residuals in the positions as a function of magnitude and location on the plates. In our utilization of the AC, the remaining effects of proper motion from the difference in epochs are either canceled or subtracted out. As a consequence the GSC 1.2 positions at epoch are unaffected by any physical motions of the stars in the Galaxy.

Numerous tests we have performed on the magnitude effect have proven that the overwhelming part of it is radial. Note again that there is no physical motion of the stars conceivable that would result on an average radial motion on the mean of all the GSC plates spread over a hemisphere. Thus, as far as the magnitude effect is concerned, we are only interested in determining, for each GSC/AC match, the difference between the radial distance from the plate center calculated using the GSC position and its corresponding AC position. For a specific magnitude range, these radial differences are binned and averaged in thin rings centered on the center of the GSC plate-based coordinate system. The magnitude dependent systematics were removed from the GSC positions by corrections based on spline fits to the (GSC-AC) radial differences as a function of distance from the plate center and magnitude. (see Figure 3).

Return to top.

Final Results

To decide on the quality of our new reduction we have compared the GSC 1.2 with the PPM and the series of Carlsberg Meridian Catalogues (CMC). (Note, considering the rms error of the CMC and the effect unknown proper motions for these stars results in a CMC positional accuracy at the epoch of the GSC plates of 0.25 arcsec in the north and 0.37 arcsec in the south). The results from these comparisons are shown in Table 3. They are valid for the brighter end of the GSC (12 > V > 8) and are biased towards the brighter magnitudes in this interval. Because these bright stars are overexposed on GSC plates, this comparison is unfavorable for estimating the accuracy of the bulk of the GSC stars which are typically fainter than V = 12. For this important fainter part we can judge the astrometric quality from the standard deviation of stars located on more than one GSC plate (see Figure 5). This standard deviation is typically 0.35 arcsec, is independent of magnitude, and involves the individual rms error of two GSC positions. Considering this and the fact that the comparison with other reference stars contains the mean errors in the reference catalog and for the CMC the effect of unknown proper motion, (approx. 10 years) we can infer that the overall rms error of the GSC 1.2 is better than 0.3 arcsec.

           TABLE 3 - RMS Position Errors (arcsec)

  Version      PPM (V>8)         CAMC (V>8)
            ra.cos(dec)  dec    ra.cos(dec)  dec
  GSC1.1      0.65      0.53      0.57      0.54 
  GSC1.2      0.31      0.31      0.40      0.40

Finally we show an example of how successful our procedure was. By comparing a mask constructed from CMC/GSC residuals before (Figure 2) and after (Figure 4 ) the corrections are applied one can see that our new reduction has eliminated the plate-based systematics. This improvement is also apparent by comparing two overlapping GSC plates before (Figure 1) and after (Figure 5 ) the corrections were applied.

Return to top.

Conclusions

This new reduction of the GSC plate material has, on the average, eliminated the major systematic errors found in GSC 1.1. The astrometric data of GSC 1.2 are available via the web. In the future we will prepare a more complete error analysis, a journal article and the release of the catalog on suitable media.

A few caveats on GSC 1.2 are also appropriate. First, this is a reduction onto the system of the PPM, which clearly will need to be repeated when reference objects based on the HIPPARCOS catalog become available. Second, it must be remembered that astrometric header information in the ST ScI Digitized Sky Surveys is consistent with GSC 1.1, and there is no easy way to transfer it to the GSC 1.2 system. Finally, as GSC 1.2 has not been installed in the HST ground system, it must not be used for HST observation planning.

Return to top.

References

Bastian, U., Roeser, S., and Yagudin, L. I. 1993, PPM Star Catalogue, Vol. III and IV. Positions and Proper Motions of 197179 Stars South of -2.5 Degrees Declination. Astronomisches Rechen-Institut, Heidelberg. Spektrum Akademischer Verlag, Heidelberg, Berlin, New York

Jenkner, H., Lasker, B. M., Sturch, C.R., McLean, B.J., Shara, M. M. and Russell, J. L., 1990, AJ 99, 2081

Lasker, B. M., Sturch, C. S., McLean, B. J., Russell, J. L., Jenkner, H., and Shara, M. M. 1990, AJ, 99, 1019

Morrison, J. E., Roeser, S., Lasker, B. M., Smart, R. L., and Taff, L. G. 1996, AJ, 111, 1405.

Roeser, S. and Bastian, U. 1991, PPM Star Catalogue, Vol. I and II. Positions and Proper Motions of 181731 Stars North of -2.5 Degrees Declination. Astronomisches Rechen-Institut, Heidelberg. Spektrum Akademischer Verlag, Heidelberg, Berlin, New York

Roeser, S., Bastian, U., and Kuzmin, A. V. 1994, Astron. Ap. Suppl. Series, 105, 301

Roeser, S., Bastian, U., and Kuzmin, A. V. 1995, I.A.U. Colloquium 148, ASP Conf. Series, Vol 84, ed. J. M. Chapman et al.

Russell, J.L., Lasker, B. L., McLean, B. J., Sturch, C. R., and Jenkner, H. 1990, AJ, 99, 2059

Taff, L. G., Lattanzi, M. G., Bucciarelli, B., Gilmozzi, R., McLean, B. J., Jenkner, H., Laidler, V. G., Lasker, B. M., Shara, M. M., and Sturch, C. R. 1990, ApJ, 353, L45

Return to top.


CASB HomeGSC Home

Last Updated Jan 2001
Copyright © 2001 The Association of Universities for Research in Astronomy, Inc. All Rights Reserved.