Color Optimization

Issue: Precise colors are vital toward galaxy classification, SED and photo-z estimation. To date, the most precised *colors* come from the gmags. The reason is that the aperture is relatively small (based on scaling the relatively tiny 2MASS XSC aperture). On the other hand, gmags can be wildly inaccurate because they do not have stars or foreground stuff removed before photometric measurement (i.e., they are simple and dumb apertures). Meanwhile more careful measurements, using proper source characterization has, by design, much larger apertures, reflecting the 1-sigma isophotes. However, this results in less precise colors (basically due to noise).

Nominal: W1 1-sigma isophotal aperture is the fiducial for W2, while W3 and W4 have their own isophotes.

Problem: W1 is bigger than W2, hence using the w1 aperture to measure the W2 flux results in extra noise and messy color.

Better: Instead, use the W2 1-sigma isophotal aperture as the fiducial for W1, and use W3 as the fiducial for W4 (only because W4 is so damned faint it's hard to get a useful isophote out of it).

Below is some basic analysis, using SGP (i.e., low confusion) measurements, that demonstrates how a smaller (matched) aperture results in more precise colors. The new proposed color is based on: W2 fiducial (for W1), and W3 fiducial (for W4)

To illustrate colors, we use the W1-W2 vs W2-W3 color-color plane (recall the bubble plot). We start with the "gmags" corrected to give roughly the correct flux (as measured using 1-sigma isophotes; calibrated from the Cluver & Jarrett analysis, and updated by Bilicki for W1 and W2).

As a figure of merit, I compute the mean and RMS for the W1-W2 color where 2.5 < W2-W3 < 2.8 mag (this is a nice compromise where resolved galaxies are located in the color plane). The mean color is not all that important, but rather the RMS indicates the "scatter" in the color. What you will notice is that the new (proposed) color gives about the same scatter as the gmags. We can do even better by using smaller apertures (as the last plot demonstrates), but that is dangerous, overweighting the "red" nuclear regions (in fact, notice how W2-W3 shifts redward, to the right).

gmags with corrections: w1corr = -0.358 mag; w2corr = -0.325; w3corr = -0.44 mag

Stats: < W1-W2 > = 0.034 +- 0.069

Current nominal, where W1 is the fiducial.

Stats: < W1-W2 > = 0.107 +- 0.078

New method, where W2 is the fiducial.

Stats: < W1-W2 > = 0.103 +- 0.069

Extreme (small aperture), using 50% of the nominal W2 aperture.

Stats: < W1-W2 > = 0.061 +- 0.064

Last update - [7 July 2013]
T.H. Jarrett (U Cape Town)