Redshifts Derived from Galaxy Color

The color of a 'normal' galaxy is strongly dependent on the redshift of the object. As such, in order to derive the zero-redshift color of a galaxy, you need to apply a color correction to compensate for the "reddening" of the galaxy due to wavelength redshift (the "K correction"; also, galaxy evolution plays an increasing role in the color modification at high redshifts). Conversely, one may deduce the redshift of a galaxy by its colors and K-correction model. Question: how well can 2MASS estimate redshifts based upon measured colors?

2MASS will detect galaxies well out to z=0.1, with decreasing completeness thereafter. For z=0 to z=0.1, this translates to about 0.3 mag change in H-K color (J-H is much less sensitive to redshift for this z range). In order to differentiate redshift at the 0.01 to 0.02 level, this translates to better than 3% to 6% error measurement in H-K color. This is clearly only possible for the brightest (and therefore, nearest) sources -- should be good for Coma, Abell 262, and possibly Hercules. A more realistic measurement error for the color is between 10 and 15%, which tranlates to 0.03 to 0.05 in redshift resolution, which is adequate for the more distant (z > 0.05) clusters.

Data: several thousands galaxies are available for this pilot study, including galaxies from Coma, Hercules, 971110n, 971205n, and 980125n. All galaxies are found in relativly low stellar source density fields (to minimize effects due to confusion and reddening). Only a fraction of this sample has known redshifts (something like 900 galaxies in total). It is the latter subsample that defines the "truth" set. We assume that the redshifts found in NED are correct and for the most part have small measurement uncertainties (this is undoubtedly true, but there are always a few klinkers in the sample, although they should wash out in the statistical noise).

Procedure:

Match 2MASS galaxies with NED catalog(s) and separate out galaxies with known (measured) redshifts. We then cull out galaxies with poor J,H or Ks measurements (due to source confusion, bad pixels, edge effects, etc). Finally, we cull out galaxies with color measurement errors greater than some specified limit (e.g., 10%).

Given the J-H and H-Ks colors for a galaxy, we then deduce the redshift by comparing the colors with a K-correction model. The K-corrections correspond to SAB type galaxies with no evolution (from models Bruzual A., G., & Charlot, S. 1993, ApJ, 405, 538, and McLeod, B. A., & Rieke, M. J. 1995, ApJ, 454, 611; tables provided by Roc Cutri) and K-corrections corresponding to ellipticals with passive evolution (using models of Bruzual & Charlot; tables courtesy of Peter Eisenhardt and Adam Stanford). We do this comparison in two different ways:

(1) We find the point on the "spiral" K-correction track that is closest to the J-H and H-K color of the galaxy. This is referred to as the "spiral - project" solution. This point is orthogonal to the K-correction track.
(2) We find the point on the track that corresponds to the "intercept" of the track and a line defined by the galaxy color and the slope of the "giant" M-S track. In the near-IR, we are most sensitive to an old-population of stars, corresponding to roughly the giant track and, to a lesser extent, the red dwarf track (because they are intrinsically faint). Therefore, we expect a spread in color (at zero-redshift) that follows this track. As the results show, this method is indeed better than method (1). This method is referred to as "spiral - intercept solution".

The following plots show the results for data combined ("all") and for the individual clusters of Coma and Hercules. The combined data spans several nights, so systematics due to calibration will average down better than the data from Coma or Herc (because they were acquired on single nights only). We have also combined all known "spirals", "S0" and "elliptical" morph types of galaxies. Analysis of the individual types is forthcoming.

The plots show the known redshift vs. the redshift difference between known and what is deduced from the colors and K-correction track.

Colors are derived from apertures mags corresponding to fixed radius = 7 arcsec circular apertures -- unless when otherwise noted.

All Galaxy fields: coma,hercules,971110n,971205n,980125n

• Galaxies in JHK color-color Plane: delta_color <= 0.05
  
• Redshift Results for delta_color <= 0.05
  
  • Redshift Results for delta_color <= 0.05; K-fid Ellip Aper
  
  

  
  

• Galaxies in JHK color-color Plane: delta_color <= 0.10
  
• Redshift Results for delta_color <= 0.10
  
  • Redshift Results for delta_color <= 0.10; K-fid Ellip Aper
  
  

  
  

• Galaxies in JHK color-color Plane: delta_color <= 0.15
  
• Redshift Results for delta_color <= 0.15
  
  • Redshift Results for delta_color <= 0.15; K-fid Ellip Aper
  
  

  
  

• Galaxies in JHK color-color Plane: delta_color <= 0.20
  
• Redshift Results for delta_color <= 0.20
  
  • Redshift Results for delta_color <= 0.20; K-fid Ellip Aper
  
  

• Galaxies in JHK color-color Plane: delta_color <= 0.10
  
• Redshift Results for delta_color <= 0.10
  

  
  

• Galaxies in JHK color-color Plane: delta_color <= 0.15
  
• Redshift Results for delta_color <= 0.15
  

5 Dupe Fields of the Hercules Cluster

• Galaxies in JHK color-color Plane: delta_color <= 0.10
  
• Redshift Results for delta_color <= 0.10
  

  
  

• Galaxies in JHK color-color Plane: delta_color <= 0.15
  
• Redshift Results for delta_color <= 0.15
  

Comments

We are systematically deducing redshifts that are too high for galaxies that are nearby (z < 0.02). But we seem to do rather well, on average, for 0.2 < z < 0.1. The errors range from a few % to 30%.

The nearby sample (z < 0.02) is clearly not doing what we expect. The colors are too red (thus giving redshifts that are too large). This may be due to the fact that nearby galaxies are large (on the sky) and so we are really measuring only their inner cores (bulge component), which is known to be redder than the outer parts of galaxies. Our little circular aperture, r=7", is just too small to provide a realistic color for nearby large galaxies (conversely, a large aperture may work better except that 2MASS is not optimized to measure the fluxes of large galaxies; but, see results for ellip aperture colors). There are undoubtedly other systematics are work here (e.g., our nearby sample might be dominated by a singe morph type, e.g., ellipticals, due to small sample statistical fluctuations ). This needs more work.

All Galaxy fields: Low Redshift (z < 0.01)