2010ApJ...713..970A

We present a set of low-resolution empirical spectral energy distribution (SED) templates for active galactic nuclei (AGNs) and galaxies in the wavelength range from 0.03 µm to 30 µm based on the multi-wavelength photometric observations of the NOAO Deep-Wide Field Survey Boötes field and the spectroscopic observations of the AGN and Galaxy Evolution Survey. Our training sample is comprised of 14,448 galaxies in the redshift range 0 ≲ z ≲ 1 and 5347 likely AGNs in the range 0 ≲ z ≲ 5.58. The galaxy templates correspond to the SED templates presented in 2008 by Assef et al. extended into the UV and mid-IR by the addition of FUV and NUV GALEX and MIPS 24 µm data for the field. We use our templates to determine photometric redshifts for galaxies and AGNs. While they are relatively accurate for galaxies (σ_ z_/(1 + z) = 0.04, with 5% outlier rejection), their accuracies for AGNs are a strong function of the luminosity ratio between the AGN and galaxy components. Somewhat surprisingly, the relative luminosities of the AGN and its host are well determined even when the photometric redshift is significantly in error. We also use our templates to study the mid-IR AGN selection criteria developed by Stern et al. in 2005 and Lacy et al. in 2004. We find that the Stern et al. criterion suffers from significant incompleteness when there is a strong host galaxy component and at z =~ 4.5, when the broad Hα emission line is redshifted into the [3.6] band, but that it is little contaminated by low- and intermediate-redshift galaxies. The Lacy et al. criterion is not affected by incompleteness at z =~ 4.5 and is somewhat less affected by strong galaxy host components, but is heavily contaminated by low-redshift star-forming galaxies. Finally, we use our templates to predict the color-color distribution of sources in the upcoming Wide-Field Infrared Survey Explorer (WISE) mission and define a color criterion to select AGNs analogous to those developed for IRAC photometry. We estimate that in between 640,000 and 1,700,000 AGNs will be identified by these criteria, but without additional information, WISE-selected quasars will have serious completeness problems for z ≳ 3.4.