2MASS Elliptical Galaxies

Photometry of Elliptical Galaxies

The following measurements are made with 2MASS "coadds" from the IRSA archive. The images do not have any background offsets or corrections. The galaxies are characterized using WISEPhot, developed for WISE large galaxy analysis, featuring simple local background estimation from a nearby annulus. Comparison is made with the 2MASS-XSC measurements, featuring non-linear background removal.

NGC 4187

2MASS J, H and Ks-band coadd images, and 3-color version.
blue ellipse: K-band 20 mag/sq. arcsec fiducial isophote;
green ellipse: half-light
yellow annulus: local background determination

Surface Brightness (coadd + WISEPhot)

	J-band	H-band	Ks-band
SB vs radius
SB vs Log radius

Photometry Comparison with the XSC

parameter	XSC	coadd+ WISEPhot
r_k20fe	30.5"	36.0"
j_m_k20fe	10.92	10.89 mag
h_m_k20fe	10.24	10.17 mag
k_m_k20fe	9.90	9.83 mag
r_ext	46"	78"
j_m_ext	10.79	10.75 mag
h_m_ext	10.09	10.03 mag
k_m_ext	10.77	10.70 mag

IC0522

2MASS J, H and Ks-band coadd images, and 3-color version.
blue ellipse: K-band 20 mag/sq. arcsec fiducial isophote;
green ellipse: half-light
yellow annulus: local background determination

Surface Brightness (coadd + WISEPhot)

	J-band	H-band	Ks-band
SB vs radius
SB vs Log radius

Photometry Comparison with the XSC

parameter	XSC	coadd+ WISEPhot
r_k20fe	24"	26"
j_m_k20fe	11.07	11.03 mag
h_m_k20fe	10.37	10.32 mag
k_m_k20fe	10.08	10.05 mag
r_ext	38"	64"
j_m_ext	10.95	10.77 mag
h_m_ext	10.28	10.10 mag
k_m_ext	9.97	9.96 mag

UGC 4551

2MASS J, H and Ks-band coadd images, and 3-color version.
blue ellipse: K-band 20 mag/sq. arcsec fiducial isophote;
green ellipse: half-light
yellow annulus: local background determination

Surface Brightness (coadd + WISEPhot)

	J-band	H-band	Ks-band
SB vs radius
SB vs Log radius

Photometry Comparison with the XSC

parameter	XSC	coadd+ WISEPhot
r_k20fe	27"	30"
j_m_k20fe	10.31	10.30 mag
h_m_k20fe	9.63	9.61 mag
k_m_k20fe	9.40	9.41 mag
r_ext	42"	74"
j_m_ext	10.08	10.16 mag
h_m_ext	9.43	9.44 mag
k_m_ext	9.33	9.34 mag

M 87

2MASS J, H and Ks-band coadd images, and 3-color version.
blue ellipse: K-band 20 mag/sq. arcsec fiducial isophote;
green ellipse: half-light
yellow annulus: local background determination

Surface Brightness (coadd + WISEPhot)

	J-band	H-band	Ks-band
SB vs radius
SB vs Log radius

Photometry Comparison with the XSC

parameter	XSC	coadd+ WISEPhot
r_k20fe	136"	163"
j_m_k20fe	6.83	6.73 mag
h_m_k20fe	6.16	6.04 mag
k_m_k20fe	5.90	5.78 mag
r_ext	233"	300"
j_m_ext	6.72	6.62 mag
h_m_ext	6.07	5.96 mag
k_m_ext	5.81	5.66 mag

NGC 1407

The following is analysis that investigated a possible bias in the XSC-LGA photometry. It turns out to be the background removal process, removing part of the "halo" of NGC 1407.

Click the following to get the 2MASS LGA Results:

2MASS Large Galaxy Atlas: NGC1407

and the Radial Profile Fit results.

For this case (N1407), we have confirmed that the outer halo of NGC1407 has likely been subtracted during the background removal process. This is (or may be) a problem that is isolated to the largest galaxies and the big R^1/4 things.

I've done some analysis of NGC 1407, both J and Ks images. I'll report both J and K since they both show the phenom.

It's not a bug or problem with the analysis routines. The SBs are being computed properly. But the issue, with N1407, is that the background removal (very likely) has the outer halo of N1407 in it. That is, the background is over-subtracted. It's not a simple offset, the LGA coadds are resampled and combined using a fit to the "background" that is non-linear (rolling 2-D cubic polynomial). The galaxy and stars are first masked before the background is determined. Here is an EXAMPLE of a coadd set with the background fit and removed.

But if the outer galaxy light is not masked or is present, it can figure in the solution. The upshot is as if an annulus was used that is too close to the galaxy (see last plot, below).

For the big R^1/4 galaxies, that scenario can certainly happen. There is a lot of light that extends beyond the usual fair seen in disky galaxies (as least for the NIR)

2MASS J-band Results
2MASS K-band Results
2MASS K-band with different annuli radii

So take a look at my first two plots: J-band results, and the K-band results. What I show is the mean radial surface brightness, computed for the 2MASS J-band and K-band coadds and for the LGA J and K-band images (where the polynomial background has been removed). I used IRAF's "ellipse" and the WISE Large Galaxy analysis tools (WISEphot) as two independent methods (of course the WISE stuff has 2MASS heritage). The published LGA result is in white.

What the plot shows is the "ellipse" (red) and WISEphot (yellow) results are about the same for the pristine coadds. Here a local background (offset) has been removed. In the case of WISEphot, I specify the annulus, which is big and well away from the galaxy light. For "ellipse", I just computed some spotty local background around the galaxy and averaged them, then subtracted from the image. Immediately you see that the SB is brighter for the "coadd" compared to the background-removed LGA image. This suggests the background has been over-estimated for the LGA. And remember this is not an offset, but a 2-D cubic (rolling looking thing).

Note: keep in mind, the true 2MASS backgrounds roll and yaw, waves of airglow that cross the images during the exposure period. There is no simple offset that represents the local background, this is particularly true in the H-band. So by estimating the background from some annulus, that is strictly an approximation and assumes a flat offset across the face of the galaxy.

Meanwhile, taking the LGA image and measuring the SB using "ellipse" (green) and with WISEphot (blue) you see they are consistent and close to the original LGA value. They show that it is the LGA image, not the measurements that is discrepant (in this case).

The last plot just shows the K-band SB as measured with different background annuli on the "coadd" image (which is pristine). They are all large annulus (thus weighting the outer part of the annulus), but somewhat closer in sequence. The SB drops down to the LGA value when the annulus starts to hug the outer isophote of the galaxy (as expected); indeed, simply demonstrating the importance (and sensitivity) of the background subtraction.

The smaller the galaxy the less effect (of over-subtraction) should happen, and likewise, the bigger the galaxy (perhaps also driven by the Sersic index) the more this effect can be in play. It depends on how well the initial masking was carried out (based on optical R25 radius) and the solution was behaved (e.g., airglow creates waves and non-linear undulations across the coadds, vexing to remove). For the largest ellipticals, a slight bias is likely to be there (e.g., M87 shows a bias at the 10% level).

Comment from John Lucey:

Excellent I think we now understand what the issue is. If a -ve halo has been introduced the mean SB count will start to increase again at larger radii. This is a check I've used in the past.

I checked a random galaxy (J_tot = 11.3) and I can confirm that the image in XSC cube is effectively identical to an extraction of the same area in the Atlas image. Hence as my pipeline uses these cubes apart from a few LGA galaxies our measurements for 6df are not affected.

For the record I use the listed j_m_ext, etc magnitudes in the reduction. I recall that as these agreed well with what I independently measured there seem no point in introducing another set of magnitudes. However I will rerun my pipeline using the 2MASS kron magnitudes to provide another cross check.

I've seen this -ve halo issue with Pan-Starrs, Megacam data and in SDSS before DR8. How this was fixed in DR8 as described in http://adsabs.harvard.edu/abs/2011AJ....142...31B

If you use SWARP, http://www.astromatic.net/software/swarp (which is a great bit of software) with the standard defaults to mosaic a large galaxy you introduce a very nice -ve halo. Fortunately this can be fixed by changing a few flags.

-john lucey