ROSAT Status Report #113: Differences between WGACAT and ROSATSRC

[REQUEST@rosat.gsfc.nasa.gov]

		ROSAT Status Report # 113
			Dec 12 1994



     Comments on the WGACAT/ROSATSRC Comparison Report by MPE
                  N.E. White, L. Angelini and P. Giommi

In the report from MPE on a comparison of the wgacat and rosatsrc catalogs
(see below) they report several significant differences. Probably the most
critical is an offset in the positions between wgacat and simbad.
Investigation by us confirms there is a problem in the WGACAT positions,
with a 6 arc sec offset in the inner region and 15 arc sec in the outer.
These offsets are comparable to the quoted uncertainties in the ROSAT
boresight, which is one reason we had not noticed it before. If you have
been using the default 1 arc min search radius in browse, it is not likely
to be a problem. We have traced the source of the error and a corrected
catalog is being generated. We hope to have it released within the next 2-3
weeks. This will also contain additional sources from new fields that have
recently entered the archive. 

Here is a rule of thumb to correct the positions. Its not a simple scaling and
depends on whether the source is less than 7860,7860 in detector coords. If
the X pixel coord is less than 7860 then to the RA ADD 4 arc sec in the inner
region, and 10 arc sec in the outer (denoted by the source designation being
_in or _out). If the y coord is less than 7860 SUBTRACT from the dec 4 arc sec
and 10 arc sec if it is, respectively, in the inner or outer regions. In
addition to this, there is a overall offset of order 5 arc sec in the outer
region and 2 arc sec in the inner. These latter corrections should be applied
in the sense.

The other points mentioned are an excess of faint sources between 16 and 20
arc sec in wgacat. There is an overlap in the separate runs made for the inner
and outer regions, close to the inner rib. In the overlap region the source
detect for the outer region is more sensitive due to an overall lower
background. We would recommend that all _out sources within OFFSET < 19 arc
sec be treated with caution. The differences in the vignetting are much as
expected, since we used a simple approximation. On the other side, the
ROSATSRC count rates in the outer region are wrong because they did not
include the psf correction.


________________________________________________________________________________

			     MPE Report



Dear ROSAT observer,

following the report of a comparison of the two ROSAT PSPC source catalogs
WGACAT and ROSATSRC. The report and compressed postscript files of the figures
can be copied from the anonymous ftp account rosat_svc.mpe-garching.mpg.de
in the directory archive/sourcecat file wga_rosatsrc.doc.
---------------------------------------------------------------------------



 Differences in the two ROSAT catalogs of pointed PSPC observations

                Haberl F., Pietsch W., and Voges W.
          Max-Planck-Institut fuer extraterrestrische Physik

   We have compared the two catalogs of ROSAT sources (ROSATSRC, 
   Voges W., Gruber R., Haberl F., Kuerster M., Pietsch W., and Zimmermann U., 
   ROSAT NEWS No. 32; and WGACAT, White N.E., Giommi P., and Angelini L., 
   IAU Circular 6100) and find differences with respect to
      Source positions,
      Source distribution at different off-axis angles,
      Detector vignetting, and
      Count rates.
   
   This study was not done in depth yet, and should not be considered as 
   being complete, but the first results are so important in our view, that
   we want to bring it to the attention of the ROSAT community 
   at this early stage. We are continuing this work and will update this 
   information when available.

   A short summary of the results is given in the following:

Source positions

   The WGACAT and ROSATSRC catalogs have been correlated with SIMBAD
   stars which have a position accuracy of better than 1.5 arcsec.
   For the ROSATSRC catalog 68% of the matching sources are found within
   17 arcsec to the SIMBAD position (Fig. 1)

   In Fig. 1 one can even resolve the two peaks in declination separated 
   by about 20 arcsec. This is caused by the boresight correction which 
   was wrong by about 10 arcsec in the first processed revision 0 data.
   It also leads to a slight broadening of the distribution in right 
   ascension. The improved boresight values used in revision 1 moves the 
   peaks closer together (residual boresight error less than about 5 arcsec) 
   and the source catalog can be used to further improve the boresight 
   correction.

   Fig. 1a and 2a show the same correlation for sources in the inner part 
   of the detector (within 19 arcmin off-axis angle) and Fig. 1b and 2b 
   for the outer part.

   The distribution of distances to SIMBAD sources found for the WGACAT
   shows a systematic shift of about 5 arcsec in RA and about 15 arcsec 
   in DEC. Moreover the distribution is asymmetric (Fig. 2). From an 
   inspection of several images with plotted source positions (we used 
   the FITS data from the archive and standard EXSAS procedures for
   creating images and overlaying the source positions) from both 
   catalogs the shift between the source positions is visible by eye
   and is always in the same direction in the image (Fig. 3 and 4). The 
   shift is larger in the outer part of the image where the images for 
   the WGACAT have larger image binning. In the description of the WGACAT
   similar images and overlays of the same ROSAT pointing are presented. 
   There are no systematic offsets between sources in the image and the 
   overlay. This suggests an intrinsic problem connected to the relation 
   between ROSAT image pixel and sky position in the XIMAGE package.

Source distribution at different off-axis angles

   The number of sources found at different off-axis angles 
   is very different in the WGACAT and ROSATSRC catalogs.
   While in the inner part of the detector the maximum-likelihood method
   used in the ROSATSRC is more sensitive and finds more sources than the
   sliding window method in the WGACAT, the number of sources 
   in the WGACAT shows a strong peak between 16 and 20 arcmin (Fig. 5 and 6). 
   For the WGACAT the images were split into two parts; the inner part 
   inside 19 arcmin and the outer part outside 18 arcmin off-axis angle.
   The expected double identification due to the 1 arcmin overlap
   can not be responsible for the whole peak and the nature of the 
   additional sources is not clear. 

   The peak and a higher number of sources outside 22 arcmin, present 
   in the WGACAT, are very prominent for source count rates below 0.04 cts/s 
   (Fig. 7 and 8) while absent for count rates above 0.04 cts/s (Fig. 9 and 10).
   Also the distribution of sources which are common to both catalogs does not
   show the increased number of weak and spurious sources around the support 
   ring.
   (Fig. 11), indicating that a considerably higher number of weak sources 
   is contained in the WGACAT between about 16 and 40 arcmin off-axis angle. 
   (The absolute numbers in Fig. 11 are higher as they contain also 
   multiple matches.)

Detector vignetting

   The vignetting correction for the ROSATSRC sources is calculated per 
   photon and then averaged. In the inner part of the detector where
   ML was running, this is done also energy dependent.
   Fig. 12 shows the vignetting correction as function of off-axis angle.
   For the WGACAT the vignetting correction is determined from the position
   of the source only and therefore shows no scatter. The vignetting 
   correction in the WGACAT is energy independent and lies below 
   the values derived from the official effective area calibration file.
   The difference increases with off-axis angle and is about 2% at 10 
   arcmin up to about 10% at 50 arcmin, and is even larger for hard sources.

Count rates

   In the outer part of the detector the count rates in the ROSATSRC 
   catalog are not corrected for the point spread function and give only
   the counts inside the cut radius, corrected for vignetting. Therefore
   the count rates can be significantly too low and cases were found 
   with a factor of two discrepancy. Further investigations in this direction 
   are in progress.

Postscript files of figures:
Fig. 1:  mpe_simbad.ps.gz
Fig. 2:  wga_simbad.ps.gz
Fig. 1a: mpe_simbad_in.ps.gz
Fig. 2a: wga_simbad_in.ps.gz
Fig. 1b: mpe_simbad_out.ps.gz
Fig. 2b: wga_simbad_out.ps.gz
Fig. 3:  us200008_full.ps.gz
Fig. 4:  us200008_in.ps.gz
         The Pleiades Center pointing also shown in the WGA catalog 
         description;
         the crosses denote the ROSATSRC positions, 
         the circles the WGACAT positions from the inner part images and 
         the squares the WGACAT positions from the outer part images   
Fig. 5:  mpe_number_offaxis.ps.gz
Fig. 6:  wga_number_offaxis.ps.gz
Fig. 7:  mpe_weak_number_offaxis.ps.gz
Fig. 8:  wga_weak_number_offaxis.ps.gz
Fig. 9:  mpe_strong_number_offaxis.ps.gz
Fig.10:  wga_strong_number_offaxis.ps.gz
Fig.11:  wga_mpe_number_offaxis.ps.gz
Fig.12:  mpe_vign_offaxis.ps.gz