There is a well known correlation between the mass and metallicity of star-forming galaxies. Because mass is correlated with luminosity, this relation is often exploited, when spectroscopy is not available, to estimate galaxy metallicities based on single band photometry.
We show that galaxy color is typically more effective than luminosity as a predictor of metallicity. Taken together, luminosity and a single optical color provide high precision photometric estimates of global galaxy metallicity.
Using Sloan Digital Sky Survey spectroscopy of 148,021 nearby galaxies, we derive "LZC relations," empirical relations between metallicity (in nine common strong line diagnostics), luminosity, and color (in three filter pairs). We show that these relations allow photometric metallicity estimates that are ∼ 40% more precise than those made based on luminosity alone, such that metallicity can be estimated to within 0.06 − 0.1 dex given multi-band imaging of a galaxy.
Not only is the statistical uncertainty in the metallicity estimate reduced by including information about galaxy color, but systematic biases are removed for populations dominated by high or low SFR environments. This new tool will lend more statistical power to studies of galaxy populations, such as supernova host environments, in ongoing and future wide field imaging surveys.
You can download a python module to automate the calculation of galaxy metallicities and associated uncertainties using the LZC relation at this git repository.
Below we diplay the luminosity-metallicity relation and LZC relations for a random subet of 10,000 SDSS galaxies in all gri filter combinations, along with trend lines built from the full 100,000 galaxy sample. A small subset of these are included in the paper. You can download the full version of Table 2 from the paper, containing a full description of all the LZC calibrations, in tex format here. We thank the MPA/JHU team for making available their catalog of properties of the SDSS star-forming galaxies.
For fully-photometric metallicity estimation, a redshift estimate must be obtained from photometry (photo-z) before the LZC relation can be applied, in order to calculate the K-corrected luminosity and color for the galaxy. We perform simple simulations using the SDSS photo-z estimates (Csabai et al. 2007) for galaxies in the MPA-JHU catalog. For each galaxy, we calculate metallicity using the LZC relations and assuming, in turn, the photometric and spectroscopic redshift estimates. We apply K-corrections using the analytic prescriptions of Chilingarian et al. (2010).
The results of these simulations are shown below, with the metallicitiy difference from photometric versus spectroscopic redshift estimates in dex versus the spectroscopic redshift. The blue errorbars show the median and standard deviation of points within redshift bins. The figure illustrates that for redshifts less than ~0.15, the uncertainty introduced by the photo-z is negligable compared to the intrinsic scatter in the LZC relations, but it dominates at higher redshifts.