Spectral energy distributions and luminosities of galaxies and active galactic nuclei in the Spitzer Wide-area Infrared Extragalactic (SWIRE) legacy survey

We discuss optical associations, spectral energy distributions (SEDs), and photometric redshifts for Spitzer Wide-Area Infrared Extragalactic (SWIRE) Survey sources in the European Large-Area ISO Survey (ELAIS) N1 area and the Lockman Validation Field (VF). The band-merged Infrared Array Camera (IRAC) (3.6,4.5,5.8, and 8.0 μm) and Multiband Imaging Photometer for Spitzer (24, 70, and 160 μm) data have been associated with optical UgriZ data from the Isaac Newton Telescope Wide Field Survey in ELAIS N1 and with our own optical Ugri data in Lockman-VF. Criteria for eliminating spurious infrared sources and for carrying out star-quasar-galaxy separation are discussed, and statistics of the identification rate are given. Thirty-two percent of sources in the ELAIS N1 field are found to be optically blank (to r = 23.5) and 16% in Lockman-VF (to r = 25). The SEDs of selected ELAIS sources in N1 detected by SWIRE, most with spectroscopic redshifts, are modeled in terms of a simple set of galaxy and quasar templates in the optical and near-infrared (NIR), and with a set of dust emission templates (cirrus, M82 starburst, Arp 220 starburst, and active galactic nucleus [AGN] dust torus) in the mid-infrared. The optical data, together with the IRAC 3.6 and 4.5 μm data, have been used to determine photometric redshifts. For galaxies with known spectroscopic redshifts, there is a notable improvement in the photometric redshift when the IRAC data are used, with a reduction in the rms scatter from 10% in (1 + z) to 7%. Although further spectroscopic data are needed to confirm this result, the prospect of determining good photometric redshifts for much of the SWIRE survey, expected to yield over 2 million extragalactic objects, is excellent. Some modifications to the optical templates were required in the previously uninvestigated wavelength region 2-5 μm. The photometric redshifts are used to derive the 3.6 and 24 μm redshift distribution and to compare this with the predictions of models. For those sources with a clear mid-infrared excess, relative to the galaxy starlight model used for the optical and NIR, the mid- and far-infrared data are modeled in terms of the same dust emission templates (cirrus, M82, Arp 220, and AGN dust torus). The proportions found of each template type are cirrus, 31%; M82, 29%; Arp 220,10%; and AGN dust tori, 29%. The distribution of the different infrared SED types in the L IR/L opt versus L IR plane, where L IR and L opt are the infrared and optical bolometric luminosities, respectively, is discussed. There is an interesting population of luminous cool cirrus galaxies with L IR > L opt, implying a substantial dust optical depth. Galaxies with Arp 220-like SEDs, of which there are a surprising preponderance compared with preexisting source count models, tend to have high ratios of infrared to optical bolometric luminosity, consistent with having very high extinction. There is also a high proportion of galaxies whose mid-infrared SEDs are fitted by an AGN dust torus template (29%). Of these only 8% of these are type 1 AGNs according to the optical-NIR template fitting, whereas 25% are fitted with galaxy templates in the optical-NIR and have L IR > L opt and so have to be type 2 AGN. The remainder have L IR < L opt and so can be Seyfert galaxies, in which the optical AGN fails to be detected against the light of the host galaxy. The implied dust covering factor, ≥75%, is much higher than that inferred for bright optically selected quasars.