The combined sensitivity and positional accuracy of the FIRST catalog are unprecedented compared with any previous wide-area radio catalog. (The NVSS [[Condon et al. 1994]] catalog has a sensitivity times lower and a resolution that of FIRST.) As such, our source locations have an accuracy that match or exceed those of all currently available large astronomical catalogs. As an illustration of the utility of the FIRST catalog, we describe here comparisons we have undertaken with three of the largest existing catalogs at other wavelengths: the Guide Star Catalog ([Lasker et al. 1990]), the IRAS Faint Source Catalog ([Moshir et al. 1992]), and the ROSAT WGACAT of X-ray sources ([White, Giommi & Angelini 1994]).
The Guide Star Catalog (GSC), produced by the Space Telescope Science Institute from scans of the ``Quick-V'' Palomar plates taken between 1982 and 1984, contains 18 million optical objects brighter than magnitude 16. The astrometric accuracy of the catalog is roughly 1", although the centers of large galaxies are uncertain by a larger amount. All objects are classified as either stellar or nonstellar although, as we discuss below, the former category contains a significant number of galaxies.
We performed separate matches between the 30,861 objects classified as nonstellar and the 407,473 stellar objects which are found within the boundaries of our survey region. A total of 475 nonstellar objects are coincident with FIRST radio sources within 1", where is expected by chance. A number of matches in excess of the false rate is seen out to 10", although the false rate becomes outside of 5". These matches represent a mix of normal, nearby galaxies, starburst galaxies and relatively low-redshift AGN. This complete, flux-limited list of radio sources associated with galaxies in the local Universe is a useful sample in which to identify such objects. Note that we have simply matched the individual radio source positions to each GSC entry; no effort has yet been made to match the counterparts of classical double radio sources by searching between the radio lobes where the parent galaxy is to be expected. We pursue this issue in some detail in an upcoming publication ([McMahon et al. 1996]).
The match results for stellar objects yielded a surprising number of coincidences given the rarity of luminous stellar radio sources: 98 (182) ``stars'' are found to be coincident with FIRST radio sources within 1" (3"), with an expected false rate of only (40) objects. Closer examination of the matches demonstrates that most of these objects are misclassified galaxies. A search of the NED and SIMBAD database revealed that 100 (55%) of the 182 ``stellar'' matches are, in fact, cataloged galaxies. Another 60 (33%) objects have not been identified previously, although the radio and optical properties of these objects are not statistically different from those of the galaxies in this sample. At most 12 of these objects are certifiably stars, 9 found in the SIMBAD database and 3 observed by us as part of a spectroscopic follow-up of bright optical counterparts to FIRST sources ([Gregg et al. 1996]), although even the 7 stellar candidates within 1" are not significantly greater than the total false rate expected for the GSC of . While proper motions complicate the identification of FIRST radio stars, and some of the GSC sources which remain unidentified could be stellar radio sources, our basic results confirm the expected rarity of stars at flux densities above 1 mJy at 20 cm. A further discussion of the handful of true stellar radio sources discovered in our survey may be found in [Becker et al. 1996].
The IRAS Faint Source Catalog (FSC - [Moshir et al. 1992]) includes over 173,000 objects in the high latitude sky observed at 12, 25, 60, and 100 m as part of the IRAS Sky Survey in 1983. The sources represent a mix of stars and galaxies which are present in a ratio of 1:2 averaged over our range of Galactic latitudes. We have adopted the definition of [Meurs & Harmon 1988] in using the far infrared colors to identify the probable galaxies; given the results of the previous section we do not expect a significant match rate with the stars. The FSC source position uncertainties are described by error ellipses with a mean major axis of 23"; these are putative errors, although we find a significant number of real matches out to four times these values and beyond.
Of the IRAS sources within the FIRST survey area, 1814 have radio counterparts within the error ellipse, with only 65 expected by chance. All but 11 (0.6%) of these sources have galaxy colors, confirming the nature of the identifications. An additional matches in excess of the number expected by chance are found between 2 and 4 times the IRAS FSC error ellipse, although the false rate rises to 36%.
It is not surprising that 45% of all IRAS sources with galaxian far infrared colors appear in the FIRST survey. [Helou, Soifer & Rowan-Robinson 1985] have shown that a tight correlation exists between the fluxes of radio and far infrared emission in normal and starbursting galaxies. Given our peak flux density threshold of 1 mJy/beam and the median flux density ratio of for a radio flux-limited sample ([Hacking et al. 1989]), we would expect to detect ( ) of galaxies with 60 m flux densities above thresholds of mJy ( mJy). We fail to include many of these nearby, extended objects in our catalog because their radio surface brightnesses fall below our detection limit of 1 mJy/beam (including 0.25 mJy/beam CLEAN bias). We also suffer some secondary loss in sensitivity both as a consequence of resolving out the diffuse flux and as a result of the CLEAN bias (discussed above), which is more severe for extended sources. As we show in [BWH95], however, a large fraction of faint extended sources with peak flux densities below our threshold are present in our images at significance, implying that direct queries of the image database will allow us to detect a much larger fraction of FSC sources. Further exploration of the IRAS source component of the FIRST survey is left to a subsequent publication.
The Position Sensitive Proportional Counter (PSPC) on ROSAT conducted nearly 5000 pointings covering each in the 0.1-2.4 keV X-ray band during its four-year lifetime. The High Energy Science Archive Research Center (HEASARC) has produced a catalog of 68,907 sources derived from these images (the WGACAT; [White, Giommi & Angelini 1994]). The sources include a wide mix of astronomical objects in which coronally emitting stars and AGN dominate. Positional uncertainties are quoted as being 13" ( ) if the source was observed within the central 18' of the field and 50" if it fell outside this region. A total of 188 (164, 56) PSPC pointings with exposure times exceeding 500 s (2000 s, 10,000 s), covering a total of 520 , fall within the FIRST survey region.
Of the 2212 sources with the more accurate positions which fall within our radio coverage area, 311 are coincident to within 2 times the quoted X-ray uncertainty (nominally, 95% confidence error circles) with one or more radio sources; only 31 matches are expected by chance. Thirty-eight of these are double radio sources with separations of ", while an additional 5 are radio triples (only double or triple radio matches are expected by chance). The 3312 sources with larger X-ray positional uncertainties provide an additional real matches, meaning that 17% of the ROSAT sources in our survey area have FIRST radio counterparts. This sample represents one of the largest lists of radio/X-ray coincidences available to date (cf. [Brinkmann et al. 1995] and [Moran et al. 1996]), and extends the exploration of X-ray-selected radio sources a factor of 30 fainter in radio flux density than previous large-area surveys. In addition, unlike comparisons with previous radio catalogs, the radio data here provide a substantial improvement in positional accuracy, greatly simplifying the task of X-ray source optical identification.
In addition to the deep, but limited, coverage of the X-ray sky offered by the WGACAT, there exists a uniform, complete survey of the soft X-ray sky produced in the ROSAT All-Sky Survey (RASS) completed over five years ago. The scientific utility of a match between the RASS and the FIRST survey catalog is manifest, but the absence of a public version of this catalog has prevented us from carrying out the match. Perhaps this situation will be rectified in the near future.