Point Source Contamination From Galaxies





Introduction

 Since one has to look past the stars of our own galaxy to see those objects beyond, stars frequently appear projected onto the visible disks of galaxies. As photometry of these stars may be contaminated by the non-uniform background of the underlying galaxy, it is desirable to flag these objects. In addition, pieces of galaxies may have been identified as point sources, especially for the larger, nearby galaxies, and these sources require flagging. Finally, since the point source photometry of extended objects may not be reliable, the objects corresponding to the cores of galaxies need to be flagged. To take care of all of these issues, we have assigned a galaxy contamination flag ("gal_contam") to every point source. This flag takes the following values:

0 = source not contaminated
1 = source is identified as the nucleus of the galaxy itself
2 = source is contaminated by the galaxy or may be part of the galaxy

Identifying the Sources

Tom Jarrett supplied the following lists of extended objects from the V3 Extended Source Catalog having R20 > 10 arcsec. Most of them are galaxies.

North
South
Large Galaxies (LGA) [181533 objects total]

The goal was to perform elliptical cone searches with parameters varying according to the size, shape and orientation of each object. The elliptical parameters are tabulated in Tom's tables, where the semimajor axes are taken to be 10% greater than the "r_k20fe" radius. The job was first done by running a long series of ISISQL2.0 calls on the V3 Point Source Catalog, a process which took 22 days to complete. It was then discovered that an undetermined number of objects had been skipped, probably owing to blocktimes making the database unavailable. The job therefore had to be repeated, and this time it was run using IRSA's tbl_compare utility, which took a mere 5 hours.

Flagging the Sources

Those objects where the counter number from the point source catalog matched the point source key from the extended source catalog were assigned gal_contam = 1, since these are the point sources uniquely identified with the cores of the extended objects. The rest were flagged with gal_contam = 2. At the same time, only those objects with the flag dirty=1 were kept. Duplicates were removed, and a gal_contam flag of 1 was given precedence over a flag of 2. Cases like these occur if two galaxies are in close proximity and a search around one finds the other.

Final Tally of Sources
507988 with gal_contam not 0
166264 with gal_contam = 1
341724 with gal_contam = 2

Comments

A few comments must be made about these results. Note that there are 166264 point sources given gal_contam=1 versus the original input list of galaxies containing 181533 objects. There are five reasons for this discrepancy:

1. None of the 531 LGA galaxies have corresponding point source IDs (pt_src keys are all null), so the objects found around these positions are all given gal_contam flags of 2. There are no objects with a flag of 1.

2. Similarly, there are 153 galaxies NOT from the LGA which curiously have pt_src keys of null. As above, there are no gal_contam=1 objects here. These are a set of low surface brightness galaxies that were forced to be found in GALWORKS.

3. There are 12 galaxies (non-LGA) for which the point source position and extd source positions are far enough separated that the elliptical cone search around the galaxy position did not pick up the point source. So again, there are no gal_contam=1 objects here, It would be a simple matter to add them to the table, but this has not been done.

4. There is a complex and extremely red extended source in the galactic plane at 308.17184 +38.76953 and 308.17291 +38.77092 (cntrs = 751965 and 751964) which share the same point source counter (306319293). This point source is midway between the two extended sources which are near the two brightest cores. Therefore gal_contam = 1 is ambiguous for the point source, but is the best that can be done.

5. The remaining 14572 sources have dirty=0 and therefore are not part of this flag setting procedure.

Examples

In order to check on this operation, images of several galaxies were overlaid with sources as follows
green: dirty = 1
red: dirty = 0
plus: gal_contam = 0
box/plus: gal_contam = 1 or 2

The green ellipse shows the outline of the cone search done on this object.

Large Galaxies From the LGA
 Values of r_k20fe are given in arcsec.
M65, 211.6"
NGC891, 225.4"
NGC253, 630.2"
M101, 236.3"
M83, 312.4"
NGC4302, 140.9"
NGC7469, 31.8"
NGC6744, 217.2"

Smaller Galaxies
 Here are some smaller galaxies, selected at random according to their size. Note that 283885 is in a crowded field in Orion and does not have a reliable size/shape measurement. It appears to be highly reddened. Listed here are the counter number in the XSC, RA and Dec, and r_k20fe values in arcsec. In the images, only the objects within the ellipses are overlaid.
135615, 60.46391, 14.23385, 10.0"
630629, 273.05978, 40.63829, 15.0"
283885, 93.37497, 17.93652, 20.0"
2454192, 229.71378, 6.43597, 20.0"
1759643, 91.94141, 32.41842, 25.0"
1696827, 203.85240, 1.41046, 30.0"
1751728, 183.26273, 7.03879, 40.0"
1738661, 184.14067, 7.46212, 60.0"

In almost every case, the flagging appears to be appropriate, and even slightly conservative: objects are flagged as contaminated out until the galaxy merges with the sky background, or slightly after this point. The exceptions are with M101 and NGC 891. In both of these cases, there are small portions of the galaxy outside the ellipse which are clearly contaminating some good point sources. This shows that the derived semimajor axis (and therefore r_k20fe) is too small in some cases.

These plots furthermore verify that the elliptical cone search routine in ISISQL2.0 is working perfectly. Without exception, the green ellipses separate all objects found in the cone searches from other field objects.


B. Nelson - IPAC
Last updated 28 August 2002