Comparison Between 2MASS
Extended Source Coordinates and Catalogued Positions

T. Jarrett, IPAC
(991012)

As an external check of the 2MASS extended source coordinate positions we cross-identify galaxies with catalogued galaxies within the Nasa Extragalactic Database (NED). The 2MASS sample consists of extended sources in the Winter 1999 public data release. We find ~13,000 matches for northern declination galaxies and ~20,000 matches for the south, representing about 15 - 20% of the total number of 2MASS extended sources. We restrict coordinate comparison to only NED galaxies with reported accurate coordinate positions (uncertainty major axis < 2"), reducing the number of galaxy matches to a few thousand in total. These matches are almost always from high SNR sources (SNR > 12). Our preliminary results pointed to a systematic difference between 2MASS and NED galaxies. Further investigation (given below) suggests that two or three major catalogs in NED are the culprit.

The 2MASS extended source processor determinates the position for a galaxy in two different ways: the J-band peak-pixel position and the intensity-weighted (centroid) using the "super" J+H+Ks combined image. The latter position is the most precise based on repeatibility tests (and is consistent with the superior SNR and centroiding technique). The equatorial coordinates are then derived from the pixel coordinates using the position reconstructed parameters in the Atlas image from which the galaxy is detected.

The extended source equatorial position may be different from the point-source position (which is derived assuming the source is a point source), but there should be no statistical systematic difference. This is confirmed in the following plot. There is no evident systematic in either dra (ra difference) or ddec (declination difference). The RMS for the comparision between the J-band peak-pixel position and the point source position is ~0.5 arcsec, while the RMS for the "super" image centroid comparison is ~0.2 arcsec. We will use the "super" image centroid position for the NED vs. 2MASS comparison. The 2MASS point source astrometry has been shown to be accurate to 0.1-0.2" based on comparison with the Tycho astrometric catalogs (ref?); hence, the extended source positions should have a comparable absolute astrometric accuracy (~0.3" RMS for high SNR sources).

We will divide the 2MASS-NED matches between the northern and southern declinations to check for possible systematics in the 2MASS northern and southern data. The following plots show the results for 2MASS vs. NED where we have not attempted to delineate the NED catalogs (except in that we have used only sources with good positions).

The northern galaxies exhibit a systematic in the declination difference of order -0.24 arcsec, while the right ascension shows no systematic. For sourthern galaxies, there is both an ra and dec systematic, -0.2 and -0.4 arcsec, respectively. Since the NED sample consists of sources originating from different catalogs (and presumably, different astrometry) it is difficult to interpret these position systematics. Instead, we break down the NED sample into astrometrically uniform catalogs.

Here we see a clear dra,ddec systematic difference between 2MASS and the APMUKS (The APM Galaxy survey by Maddox, Sutherland, Efstathiou & Loveday, 1990, MNRAS, 243, 692) of -0.4" and -0.6", respectively.

Here both the northern and southern samples show a clear declination systematic of -0.4" between 2MASS and the NPM1G catalog (Lick Northern Proper Motion Program by Klemola, Jones, & Hanson 1987, AJ, 94, 501). There is no evident right ascension systematic for either northern or southern samples.

Similar to the 2MASS Extended Source Position vs. NPM comparision, 2MASS and the NGP (Odewahn & Aldering 1995, AJ, 110, 2009) shows a systematic -0.4" in declination comparison, suggesting some astrometric connection between NGP and NPM1G. And in fact, the NGP sample was astrometrically calibrated with NPM1G positons.

Comparing 2MASS to "Abell" identified galaxies, we see a systematic coordinate difference of -0.2 and -0.4" for dra and ddec, respectively. Since this sample is probably not astrometrically uniform, this comparison may not be all that enlightening.

Now there is no evident systematic difference in either the declination or right ascension comparison between 2MASS and FIRST (The FIRST Survey: Faint Images of the Radio Sky at Twenty Centimeters by Becker, White & Helfand, 1995, ApJ, 450, 559).

Conclusions:

Even though NED is composed of galaxies from a wide variety of surveys and projects, most of the positions are derived from just a few astrometric catalogs, including the APMUKS, NPM1G and FIRST (see references above). Consequently, a straight comparison between 2MASS positions and NED positions can reveal systematics in either data set. We do in fact see a clear systematic in the declination axis between 2MASS and NPM1G (both northern and southern declinations), and a bias in both ra and dec comparing 2MASS and APMUKS (southern galaxies). The most enlightening comparison comes from the FIRST sample of galaxies, claimed to be astrometrically accurate. Here we find no evidence for a bias in ra or dec. Hence, we believe the 2MASS coordinate positions are solid (giving a stronger weight to the radio positions of FIRST). Note that this is consistent with the absolute accuracy of the 2MASS point source astrometry. Major findings: