An empirical algorithm for broadband photometric redshifts of quasars from the sloan digital sky survey

We present an empirical algorithm for obtaining photometric redshifts of quasars using five-band Sloan Digital Sky Survey (SDSS) photometry. Our algorithm generates an empirical model of the quasar color-redshift relation, compares the colors of a quasar candidate with this model, and calculates possible photometric redshifts. Using the 3814 quasars of the SDSS Early Data Release (EDR) Quasar Catalog to generate a median color-redshift relation as a function of redshift we find that, for this same sample, 83% of our predicted redshifts are correct to within |Δz| < 0.3. The algorithm also determines the probability that the redshift is correct, allowing for even more robust photometric redshift determination for smaller, more restricted samples. We apply this technique to a set of 8740 quasar candidates selected by the final version of the SDSS quasar-selection algorithm. The photometric redshifts assigned to nonquasars are restricted to a few well-defined values. In addition, 90% of the objects with spectra that have photometric redshifts between 0.8 and 2.2 are quasars with accurate (|Δz| < 0.3) photometric redshifts. Many of these quasars lie in a single region of color space; judicious application of color-cuts can effectively select quasars with accurate photometric redshifts from the SDSS database-without reference to the SDSS quasar selection algorithm. When the SDSS is complete, this technique will allow the determination of photometric redshifts for ∼10 ⁶ faint SDSS quasar candidates, enabling advances in our knowledge of the quasar luminosity function, gravitational lensing of quasars, and correlations among quasars and between galaxies.