Paolo Giommi , ESA/ESRIN-ESIS
Lorella Angelini , NASA/Goddard Space Flight Center-HEASARC
We have taken all the data contained in the ROSAT public archive and generated a catalog of 45,000 point sources. This catalog has been made public through the HEASARC and ESIS online services. The detection technique uses a sliding box and has been optimized for point source detection. A K-S test for variability has been made and a number of variable sources identified. A hardness and softness ratio is also used to identify unusually soft or hard sources. Cross-correlations against the major optical, radio, x-ray and ir catalogs are presented.
WGACAT is a catalogue of point sources we have generated from all the ROSAT PSPC pointing observations from Feb 1991 to March 1994. These were the files available in the public archive at HEASARC as of September 1994. This catalog is an independent research effort aimed at releasing as quickly as possible a list of sources detected by ROSAT in its pointed phase to: 1) identify the detected sources, 2) ensure their timely observation by currently active X-ray missions e.g. ASCA, 3) to search for objects which show exceptional time variability and spectral properties and 4) to provide an independent check of the detection technique used in the official ROSAT project (SAS) processing.
In this paper we describe the initial results of this effort, and announce the public release of the catalog, which we call WGACAT. This first release contains of order 45,000 sources.
The source detect uses an optimized sliding cell detect algorithm in XIMAGE (first developed for the EXOSAT project). The total number of sequence processed to date is 2624 and includes all the USA and German/UK public archived data. As more data becomes available the catalog will be updated. WGACAT currently (Oct 30, 1994) contains in excess of 50,000 detections, with about 45,600 individual sources. An aitoff projection in galactic coordinates of the detected sources is shown in Figure 1. Images of a field centered on the Pleiades with the detected sources overlaid are shown in Figure 2 and Figure 3 for the inner and outer regions, respectively. Even though this is an extremely crowded region the XIMAGE detect did a good job resolving the sources.
The distribution of the signal to noise is shown in Figure 4 and the probability that a detection is due to a random fluctuation of the background in Figure Figure 5. The detection threshold was set to accept all sources with a signal to noise ratio greater than 2 and a Poisson probability less than 1.0E-04. The exposure distribution for the entire catalog from 100s up to 100,000s, with the vast majority of exposures in the range 5,000 to 30,000s (Figure 6).
The count rate distribution (Figure 7) ranges from 0.0005 up to about 100 ct/s, with the peak of the distribution around 0.008 ct/s. These counting rates have been corrected for the telescope vignetting at 1 keV, the point spread function and exposure. They have not been corrected for the detector ribs. The count rate distribution as a function of offaxis angle, shown in Figure 8, has a peak at offsets less than 2 arc minutes, which is where the target was placed. The detection threshold increases as the offset angle increases, because of the telescope vignetting and increase in the size of the point spread function.
Colour images that indicate the colour of each pixel (i.e. the average energy of all photons detected in that pixel) have been produced by calculating the mean of the PHA distribution at position x,y and rescaling it between 0 and 100. The maps are stored in GIF format and an example is shown in Figure 9 from an observation of M31. The colour image is on the right, with red indicating harder and blue softer sources. For comparison the intensity image is on the left, with the detected sources overlaid. The different spectral hardness is evident for each source. There are at least three very soft sources and two that are harder.
We have introduced timing images as a new tool to help identify variable sources. The method consists in comparing the time arrival distribution of the photons collected in each pixel with the corresponding distribution of the entire image using a Kolmogorov-Smirnov (KS) test. The result of the KS test is a chi2 value (with 2 d.o.f.) which is then used to assign an intensity value to each pixel. In this way pixels where the distribution of photon arrival times is not consistent with that of the entire image will be assigned high intensity values (high chi2 values). The image so constructed (time variability image) will visually show where strong time variations occurred. In Figure 10 a time variability image of the M31 field is shown. A highly variable source (no 34) is evident in the right hand image as a bright object. The lightcurve of this source is shown in Figure 11 and shows a transient outburst. A second outburst was found in other ROSAT data taken a year later. Inspection of Figure 9 shows that this source is relatively soft. Source no 34 is a luminous X-ray transient in M31, that was not known before. It was announced in the following IAU circular:
RX J0045.4+4154
N. E. White, Laboratory for High Energy Astrophysics, Goddard Space Flight Center (GSFC); P. Giommi, European Space Information System, European Space Agency; L. Angelini, GSFC; and S. Fantasia, University of Maryland, report the discovery of a recurrent supersoft x-ray transient in ROSAT archival data on M31: "The first outburst began on 1992 Feb. 2 and lasted at least four days (the end of that particular ROSAT observation sequence). The source had 'turned off' by the time of the next ROSAT observation of this region, which began on 1992 Aug. 10. A second outburst began on 1993 Jan. 7; the peak of the outburst was reached in about 4 days, after which a decline started. When the ROSAT observation ended 2.5 days after the peak, the source had not yet reached the minimum. Fitting of the source spectra on both occasions yields a blackbody temperature of 0.095 keV, consistent with the definition of a supersoft source. The hydrogen column density is 1.2 x 10E21 cmE-2 and the unabsorbed source luminosity in the band 0.1-4.0 keV is 3.4 x 10E38 erg/s (assuming the source is located in M31). The source position is R.A. = 0h45m28s.6, Decl. = +41o54'11".3 (equinox 2000.0; 90-percent confidence radius 13"). In the ROSAT archive there are four time periods in which this part of the sky was observed, each separated by six months (1991 July, 1992 Jan.-Feb. and Aug., 1993 Jan.). On two occasions each the source was in outburst (1992 Jan.-Feb. and 1993 Jan.) and in quiescence. If this source is not a foreground object, then it is the first recurrent x-ray transient to be found in M31. Further outbursts might be expected. If they are periodic, the next will occur in 1994 Nov. Further observations are suggested."
Circular No. 6064
A cross-correlation of WGACAT against all the major catalogs has allowed a preliminary identification of known optical, radio, X-ray and ir objects. The catalogs used include:
The results of these correlations are given in various parameters for each record including redshift, V Magnitude, IPC count rate, Hydrogen column density (from Stark et al.), radio flux, etc... We have also derived parameters from these including the X-ray to optical flux ratio, and the IPC to PSPC ratios.
Figure 12 gives the X-ray to Optical flux ratio distribution for various classes. The stars tend to have Fx/Fo less than 0.01, with the B stars having the lowest values. AGN and galaxies are at the other extreme with Fx/Fo between 0.01-10 and the clusters have similar values around 0.1-1. The CVs, X-ray Binaries and White Dwarfs have a broader distribution of Fx/Fo. The CVs are concentrated around 0.01-0.1 and the X-ray binaries at higher values above 1. The Fx/Fo of White dwarfs range over a wide range from 0.001 to 10, but are very sensitive to the assumed absorption.
These identifications represent a total of 5,000 objects, or about 10% of the catalog. They give a flavor of what can be expected when the other serendipitous sources are identified.
We are undertaking a comparison of the Einstein IPC and PSPC common detections, and Figure 13 shows the IPC against the PSPC count rates. A strong correlation is seen, but there are also sources that have exhibited strong variability. One example is the X-ray binary EXO 0748-676, which is a transient discovered by EXOSAT. It was not active during the Einstein era, but was still detected at 0.1 ct/s by the IPC. However, it was seen at 60 ct/s by the ROSAT PSPC.
There is also a trend for PSPC sources to have a systematically lower count rate than the corresponding IPC value. This probably reflects the lower PSPC upper energy band cutoff of 2.0 vs. 4.0 keV for the IPC. Sources that are absorbed will tend to have higher count rates in the IPC. This kind of comparison may provide a useful method to detect highly absorbed sources. A paper on the Einstein-EXOSAT-ROSAT comparison is in preparation (Giommi et al. 1995).
The redshift distribution of extragalactic objects found in the cross-correlation is given in Figure 14. The peak in the distribution is at a redshift of 1-2, with sources detected out to a redshift of 4. The ratio of the IPC and PSPC count rates is given as a function of redshift in Figure 15 (for extragalactic sources). The ratio shows a large scatter at low redshifts indicating variability over a factor of 10 or more. An example of dramatic variability is the type I Seyfert NGC 3516, which was detected with a PSPC count rate a factor of 80 higher than that of the IPC. This corresponds to approximately a factor 40-50 increase in flux. At higher redshifts the scatter becomes less evident, suggesting less variability and that low redshift low luminosity AGN are intrinsically more variable, as might be expected.
The Chi2 time variability statistic derived from the KS test is plotted against object class in Figure 16. Late type and pre-main sequence stars (1900-1999,2400-3000) tend to be highly variable, whereas extended galactic objects (3000-3999) are not. This confirms this test is working as expected. Notice that OB stars (2100-2399) are not variable, which perhaps supports the view that the coronal activity is different than in late type stars. CVs and X-ray binaries (1000-1699) are variable, as might be expected in accretion driven systems. Clusters (5000) are mostly non-variable, with the couple of exceptions probably due to misclassified objects. The various types of galaxies (6000-6500) are also quite calm, but many objects anonymously classified as galaxy (6700) are not. These may be AGN that need to be reclassified. The AGN (7000-7500) are also variable. Many variable objects are not yet classified and are sure to be a source of fruitful discoveries, like the transient discovered in M31.
The XIMAGE detect is very sensitive to finding point sources, but can also find spurious sources where there is extended emission. As part of the processing we generated a gif image with the detected sources overlaid, plus an individual thumbnail image of the source itself. These are used to make a quality check on the detect output (Figure 17). The obvious spurious sources caused by large extended structures have been flagged, and removed from the public catalog. There may still be subtle problems caused by low level diffuse emission. The gif images have been made available as part of the catalog, to allow a further quality check by interested users.
It is imperative that users of WGACAT be aware that it is their responsibility to check the quality of each detection. As a first step we have checked by eye the quality of each detection and written a flag from 0 to 10, in the parameter QFLAG. The visual inspection was of a gif file made for each field from ximage, with the detected sources overlayed and obvious fields with extended emission flagged. Users of this catalog should carefully check the QFLAG parameter (values >5 are good) for each source, and examine the gif images that have been provided.
By default BROWSE will load only the GOOD sample, which has QFLAG >= 5. To see all the entries use the command csam TOTAL to load the entire database. To check the quality of the detection, and to see the timing and color images use the browse command xv (from the xray account) or under the WWW click on the hyperlink to the gif products. In both cases an X display is required.
Alternately WGACAT can be accessed via the HEASARC WWW Browse interface under http://heasarc.gsfc.nasa.gov/StarTrax/Squery.pl?ROSAT by selecting WGACAT. The SAS Rev 0/1 catalog (ROSATSRC) can also be accessed from both the WWW Browse and the xray account.
An ASCII dump of the catalog can be copied from the HEASARC anonymous FTP (legacy.gsfc.nasa.gov under rosat/wgacat/files). Please read the README and other files in this directory, before proceeding to use the catalog.
The WGA catalog can be also accessed from the ESIS Catalogue Browser on Mosaic (by choosing WGA_CAT in the list of catalogs) and from the ESIS imaging software where the WGA sources can be directly overlayed on astronomical images.
Send mail to wgacat@athena.gsfc.nasa.gov if you have any questions regarding this catalog.
Figure 1 (above) shows an aitoff projection of all the detected sources. The sky distribution reflects the pointings, rather than that of any intrinsic population. It shows little bias towards the galactic place. The number of sources detected at low galactic latitudes (less than 20 degrees) is about 10,000, compared to 40,000 at high latitudes.
Figure 2. The inner region of the PSPC with the detected sources overlaid. This is from ror 200008, which is an observation of the Pleiades. WGACAT
Figure 3. The outer region of the PSPC for the same Pleiades field used for figure 2. The increase in the point spread function is evident and causes a blurring of the sources.
Figure 4. The signal to noise ratio distribution.
Figure 5. The random detection probability distribution.
Figure 6. A histogram of the exposure in seconds per detected source.
Figure 7. The count rate distribution (in counts/s).
Figure 8. The source count rate as a function of its off axis angle (in arc minutes).
Figure 9. The image on the right is a color image from an M31 observation. The image on the left shows the intensity image, with the detected sources overlaid. Blue is soft, and red is hard. The source detect only ran on the inner 19 arc min of the image.
Figure 10. The detect (left) and timing images (right) for the same M31 observation used in figure 9. Source number 34 has a high chi2 value and is variable. Its lightcurve is shown in Figure 11. The source detect only ran on the inner 19 arc min of the image.
Figure 11. The two outbursts of the ultrasoft X-ray recurrent supersoft transient in M31 RX J0045.4+4154 (source 34).
Figure 12. The X-ray to optical flux ratio distribution for various classes.
Figure 13. The Einstein IPC vs. ROSAT PSPC count rates.
Figure 14. The distribution of redshift for the detected extragalactic sources.
Figure 15. The IPC/PSPC count rate ratio vs. redshift.
Figure 16. The chi2 variability statistic verses different object classes. The class parameter has a numerical representation as follows:
Figure 17. The quality checking is done using mosaic. A series of thumbnail gifs are generated for all the sources in a particular image. Then each one is given a grade, based on its visual appearance. The thumbnail gifs are available as part of the catalog, so that users can reappraise the quality. This quality check is only applied to those fields where unusual source detect patterns are seen.