Photometric Repeatibility

updated (970525)

The updated results presented in this memo describe the photometric repeatibility of the eight Coma scans after the noted fixes to GALWORKS (see below). Both the preliminary and final (updated) plots are included for comparison purposes, however.

The preliminary results showed several galaxies with very poor photometric repeatibility. These 'outlier' points turned out to be either (1) galaxies in close proximity to large elliptical galaxies in the Coma core, (2) galaxies adversely affected by bright star blanking (GALWORKS did not handle this case properly). The latter problem was an easy fix to GALWORKS. The former problem required the addition of at least another parameter to the galworks output file. A good example of galaxy-galaxy contamination can be seen in the case study of A Detailed Look at the Coma Galaxy: 194.915329+27.953882 .

The changes to GALWORKS included the following:

With the new contamination parameter, we can mark separately those galaxies that may have contamination from nearby brighter galaxies. The method used in this study is to mark all galaxies located within 2.0 radii of the nearest known (re: NGC) galaxy. In this way, several galaxies in the core of Coma (coadd #137) fall within this category and are noted accordlingly in the updated plots below.

Galaxy photometry is not a straight forward process. A multitude of apertures (some fixed, some adaptive) may be used to integrate the flux of a galaxy. Thus, the uncertainties in the flux measurement are not only from the background noise (sky noise) but also from uncertainties in the shape and size of the integrating aperture itself. This memo presents the repeatibility results for a number of key photometric measures, including fixed circular apertures (radius = 10), both circular and elliptically shaped versions of: isophotal (20 mag per sq. arcsec), petrosian, and adaptive apertures. We must also keep in mind what the level-spec target:

For a description of the level-1 specifications for extended sources, see T. Chester's memo: Brief Summary of 2MASS Facts.

The following plots show the photometric repeatibility. The mean integrated flux is plotted versus the standard deviation in the population. So for example. the fixed circular aperture, radius = 10, the mean flux mag is J10, H10, and K10, versus the standard deviation, d(J10), d(H10) and d(K10), respectively.

Fixed Circular Aperture, Radius = 10

The white filled circles are the individual points and the green triangles are the mean of the binned distribution and the error bars are the one-sigma standard deviation in the distribution. The upper deviant points are ignored in the binned mean calculation. The magenta stars denote the predicted distribution uncertainty based on poisson statistics (including sky noise, source noise, and read noise).

Based on what we would predict for the scatter in the photometry (see magenta stars) the results (green triangles) are consistent with background limited photometry (i.e., the errors are dominated by the sky noise and at J, the read noise). It is interesting to note that we predict larger scatter for the faintest J mags, which suggests that we have either overestimated the read noise or the completeness is affecting the results (at the faint end, the completeness drops rapidly). The 3 or 4 very deviant points seen in the J results (and maybe in H) are probably the result of bright star blanking, which is more prevalent at J because the stars are brighter, or it could be the result of nearby galaxy masking (also a problem for Coma, since the density of galaxies is so high). These points require follow up to verify this supposition. Note: see the link near the bottom of this memo. I have looked at in detail one of these deviant galaxies -- the problem is due to contamination from a nearby giant elliptical galaxy.

Note that nearly all of the outliers are now gone and that the outliers that remain are in fact galaxies contaminated by the giant elliptical galaxies in the core of Coma (denoted by the blue filled circles).

Isophotal Photometry

The 20 mag per sq. arcsec isophot is fit with both a circular and elliptical aperture. Note that the 20 mag per sq. arcsec isophot corresponds to about 1-sigma of the sky background at K, 1.8-sigma at H, and 3.6-sigma at J; thus, it represents most of the flux at K, but only a fraction at H and J.

Circular Aperture

The large scatter associated with the brighter galaxies is due to the fact that the galaxies are bigger than the maximum window (about 70 pix) within which GALWORKS operates. The other very deviant point in the isophotal radii and the isophotal photometry are probably due to nearby source masking (either bright stars or bright galaxies). The sources should be followed up to confirm this hypothesis.

Note: see the link near the bottom of this memo. I have looked at in detail one of these deviant galaxies -- the problem is due to contamination from a nearby giant elliptical galaxy.

Note that there is a minimum with which the isophotal radius can have (about 5 arcsec). This limit is set in GALWORKS.

Nearly all of the outliers turn out to be contaminated galaxies (blue filled circles). There are still one or two outliers that need to be tracked down.

The 10% limits in the photometric repeatibility are about J = 15, H = 14.5, and K = 12.7. Thus the level-1 specification is achieved in this photometric measure.

Elliptical Aperture

Similar to the circular aperture case, the brightest galaxies have deviant values because they are too big to accurately determine the isophot radius. There are a few fainter galaxies with deviant point that require follow up analysis attention.

Note: see the link near the bottom of this memo. I have looked at in detail one of these deviant galaxies -- the problem is due to contamination from a nearby giant elliptical galaxy.

Nearly all of the outliers turn out to be contaminated galaxies (blue filled circles). There are still one or two outliers that need to be tracked down.

The 10% limits in the photometric repeatibility are about J = 15.1, H = 14.5, and K = 13.. Thus the level-1 specification is achieved in this photometric measure.

Petrosian Photometry

We define the petrosian radius to be the point where the ratio of the mean brightness in an aperture to the mean isophotal brightness becomes 5.0. GALWORKS determines this point by first fitting a modified exponential to the isophotal profile, then performing the ratio of mean surface brightness to the fitted isophotal brightness, and finally linearly interpolating to arrive at the petrosian radius. The reader may like to look at some single-channel detailed analysis of petrosian photometry to become familiar with this method; c.f. Circular and Fixed Elliptical Apertures: The Petrosian and Isophotal Photometry

Circular Aperture


Elliptical Aperture


The scatter in the petrosian radii is larger than that of the isophotal case, and correspondingly the photometry repeatibility is much worse than that of the isophotal case. This is to be expected given that the petrosian radius requires one additional non-trivial step compared to simple isophotal photometry.

Comparing the circular to ellipitical aperture cases, not surprisingly the ellipitical repeatibility has less scatter than the circular. Detailed follow up work is necessary here to confirm that the petrosian algorithm is being implemented at its optimum -- this is a job for Jessica R.

Adaptive Aperture Photometry -- Total Integrated Flux

An adaptive aperture is used to capture as much flux of the object as possible, which, for the fainter galaxies (and spirals), represents the total flux. The basic method is given in Adaptive Aperture Photometry

Circular Aperture

The magenta stars denote the predicted distribution uncertainty based on poisson statistics (including sky noise, source noise, and read noise).


Elliptical Aperture

The scatter is large for the brightest galaxies because, again, these galaxies are larger than the GALWORKS window. The other deviant point should be looked at to confirm the scatter in the radii and photometry (they are probably due to stellar contamination, or contamination from nearby galaxies, always a problem in Coma). Since the total flux requires a very large aperture (about twice as big as the isophotal aperture), the measure is at the mercy of background sky noise.

The elliptical apertures provide a more precise measurement compared to the circular case. Detailed follow up work is required here to verify the results.

Note: the following memo discusses one such galaxy. The galaxy is bright, around K = 12, and in close proximity to one of Coma's giant elliptical galaxies. The adaptive aperture photometry is indeed heavily influenced by the presence of the nearby galaxy, which we know has plenty of flux buried beneath the sky noise. See


The scatter in both the radial repeatibility and the photometric repeatibility have been significantly improved with the latest fixes to GALWORKS. Most of the outliers are indeed galaxies with contaminated profiles (blue filled circles).

Final Remarks

The good news is that the level-1 specs are achieved with the isophotal (20 mag per sq. arcsec) measure. The scatter in the photometric repeatibility is consistent with poisson statistics (based on the fixed circular aperture results). There are a few sources with very deviant results which require follow up work to determine their nature (e.g., are they next to a bright galaxy; are they near a bright star; etc). Also, detailed examination of the petrosian and adaptive aperture photometry is probably a good idea given the complexity of these measures. This work will be documented in a future html memo.

(970525) Note: Most of the outliers seen in the repeatibility plots have been confirmed as galaxies in close proximity to the giant elliptical galaxies near the center of Coma. Their radial profiles, surface brighness contours and integrated flux are probably contaminated by the host galaxy.