S. Zoonematkermani and D. J. Helfand
Columbia Astrophysics Laboratory, Columbia University
R. H. Becker
Department of Physics, University of California at Davis
and
Institute of Geophysics and Planetary Physics
Lawrence Livermore National Laboratory
R.L. White
Space Telescope Science Institute
and
R.A. Perley
NRAO, Socorro, NM, USA
Despite the very large gain in sensitivity and angular resolution it can provide, the NRAO Very Large Array has not, until recently, been employed to survey the Galaxy. We present here the first of several Galactic plane surveys which are in progress using this instrument. In sect. 2, we describe the observations and their analysis. The following sections present the catalogue of 1992 sources we have derived from these data, and discuss the completeness and integrity of the survey. In IV we provide annotation to the catalogue based on a comparison with previous radio surveys, the IRAS point source catalogue, and other compilations of astronomical objects. Future papers (Paper II - Becker et al. 1990, Paper III - White et al. 1990) will report on source identifications derived from radio, optical, and infrared followup work.
For use in calculating source surface densities, a survey coverage-sensitivity map with 2' resolution has been constructed; a portion of this map is displayed in Figure 1. In addition to accounting for changes in sensitivity due to vignetting of the primary beam, the map incorporates the higher sensitivity thresholds applicable to ~6% of the fields in which sidelobes from a particularly bright source or poor data quality increases the map rms (e.g., at 19.5 + 0.0 in the portion of the map shown in the figure); these "bad" fields along with their field-center detection thresholds are listed in Table 3.
As is apparent from Figure 1, our sensitivity is far from uniform; the highest sensitivity, achieved on axis, is quoted as 5 times the field rms or 10 mJy, whichever is greater, while the lightest level in the grayscale plot includes regions with sensitivities ranging from 30 to 60 mJy. In particular, the overlap regions at l = 340.25d + n(0.5d), b =+-0.25d (n = 1,2,3...) contain 20 to 40 arcmin2 regions (depending on t) with a threshold > 50 mJy. In addition, at these same values of l with b =+-0.75d there are small unobserved areas within |b|<0.8d, our adopted latitude limit; the exact shape of the missing areas is a function of galactic longitude. In comparing our source list with other catalogues, it is important to remain cognizant of these low sensitivity areas. A histogram of the area surveyed as a function of sensitivity is shown in Figure 2; the solid curve representes the coverage which would have been obtained for this set of pointings if all fields had a 10 mJy central threshold, while the dashed line provides the actual coverage of our survey. Of the 220 square degrees observed, 33% of the area is covered to a peak flux limit of 15 mJy; the survey is 75~o complete to 25 mJy, and 95% complete at a 50 mJy threshold.
Each image was visually inspected for sources which, upon identification, were fit with a two-dimensional Gaussian (with a linear background term subtracted) to determine peak and integrated flux densities, sizes, and source positions (using the AIPS algorithm JMFIT). For sources consisting of two maxima separated by < 15", a simultaneous double Gaussian fit was used. For those extended sources which had obviously non-Gaussian brightness distributions, a flux density was determined by summing the intensity within a polygon enclosing the source (the lack of background subtraction renders these flux densities particularly uncertain); these sources are annotated with a + in the catalogue (see Table 4). For roughly a dozen other sources where the Gaussian fit clearl.y misses surrounding (and presumably related) extended emission, we provide an integrated flux density in the comments column of the catalogue. Finally, in cases where we have detected several adjacent discrete sources which appear to be peaks of an extended region of emission, we highlight the related sources with asterisks.
The antenna spacings in the B-array determine the angular resolution of the images and also limit the sensitivity of the observations to extended structure. The shortest antenna spacing of ~500 wavelengths means that emission on scales larger than ~20"-30" will be invisible or highly attenuated in the images. Thus, some of the catalogued sources represent substructure in a much larger source, and many of the large, bright HII region complexes which dominate lower-resolution maps of the plane have been completely missed. As discussed below and in Paper II, the flux densities of sources larger than 10" in diameter are very ill-determined.
The arrangement of survey field centers resulted in significant overlap in adjacent images. Hence a substantial number of sources were observed two or more times, and this redundancy in the program allowed us to determine the accuracy of the derived source parameters. In Figure 5a we show a histogram of the difference in measured position for all 255 multiply observed sources (MOS). (In the following discussion of survey integrity derived from the MOS sample, we consider only the 215 sources with declinations >~-30d; below this declination limit the beam becomes very elongated, reducing the accuracy of positions, sizes and flux densities. For these southerly sources, error estimates should be doubled.) Over 65% of the positions for sources with sizes < 3" agree to better than 1" or 1/5 of the synthesized beam; for extended sources, the accuracy is only slightly lower. By definition, the MOS all lie at the periphery of the images where the accuracy is lowest; thus, the 90% confidence error circle of 3" derived from this sample is an upper limit to the true uncertainty in source positions. For all MOS, positions quoted in the catalogue are those from the observation in which the source is closest to the field center. We have also checked the accuracy of the catalogue positions as part of a followup study of selected sources using the VLA at 6 cm, 3.5 cm, and 2 cm. In Figure 5b we show the angular offset between the catalogue position and that measured in followup observations, as a function of radial distance of the source from the closest survey field center (the followup observation always has the source at the field center); again, with a few exceptions, a positional accuracy of <~3" is indicated.
The FWHM of the synthesized beam of the VLA in the B-configuration is 3.9" at the zenith for a full-synthesis observation. This degrades by a factor of ~1.3 for snapshot observations, and for sources at the southern declination limit, the north-south elongation reaches 15" (although the A/B configuration was used for most of these fields). In investigating the various components which describe the galactic radio source population, it will be useful to know which of the sources in the catalogue are point-like and which are definitely resolved by our observations. Owing to statistical and systematic uncertainties in measuring the properties of faint sources, our division of sources into "point-like" and "extended" should be a function of source intensity. We find, however, that independent of flux density, > 95% of all sources with measured sizes > 3" have S_i/S_p > 1.3, whereas fewer than 5% of all sources smaller than this have a flux ratio exceeding this value (Figure 6a). Since we will frequently only consider sources with fluxes > 20 mJy where the dividing line is even more marked, we adopt hereafter a division between point and extended sources of 3" (Figure 6b).
Finally, we may use the MOS to examine the quality of our flux determinations. In Figure 7, we plot the fractional flux density differences (S_close - S_far)/S_close as a function of off-axis angle, where S_close is the integrated flux density measured for the observation in which the source is closer to the field center. For the unresolved sources, a typical fractional flux density error of approximately +-30% appears to be appropriate. The extended sources, however, clearly represent a more serious problem; this is discussed in some detail in Paper II.
In Table 6 we list the source catalogues with which we have compared our galactic plane survey sources; the catalogues in the first half of the Table exist in machine-readable form and we have performed a direct comparison with Table 4, adopting a 3" error circle radius for our sources and the quoted uncertainties for the other catalogues. For the IRAS point source catalogue, a more sophisticated matching algorithm which ascribes a probability of true association to each potential match has been developed and is described in detail in Paper III; we mark in Table 4 all objects whose chance association probability is < 25%, although more than three quarters of these sources have a chance rate of < 5%. The expected false coincidence rate in all comparisons has been determined by repeating the matching algorithm with all radio source positions offset by 10' in each of the four cardinal directions and averaging the numbers of false hits obtained. Lists of stellar counterparts, planetary nebulae, and X-ray source coincidences are compiled in Table 7.
The catalogues of radio sources in the lower half of Table 6 are not readily available in machine-readable form and, given the lower spatial resolution of the observations contained therein, are not directly comparable to individual sources in our catalogue. To search for potential associations (e.g., instances where we have detected the highest surface brightness spot(s) of an extended supernova remnant or HII region complex), we have constructed maps of our source positions scaled to the contour maps presented in each of the first three references. When one (or more) of our source(s) falls within the contours of a catalogued object, the association is noted. The recombination line survey source lists are drawn from these same catalogues, and if at least one line has been detected, an annotation is included. Sources falling within the contours of the known supernova remnants within our survey area are also noted. The large factor by which we underestimate the fluxes of SNRs and the fact that most of the remnants are not detected at all is a reminder of our insensitivity to extended emission. Thus, no spectral information is derivable from a comparison of our 20 cm flux densities and the flux densities at higher frequencies listed in the single-dish catalogues. The Texas interferometer survey does, however, have a resolution comparable to our observations and a comparison of the two catalogues can provide useful spectral information as we discuss in Paper II.
Based on such radio spectra, on the coincidence of a small-diameter (< 3") radio source with an IRAS point source, and on a variety of other criteria, we have undertaken followup observations of 77 catalogue sources with the VLA at 20 cm, 6 cm, 3.5 cm and 2 cm wavelengths. The implications for source identifications are presented in Paper II; here we include in Table 8 the flux densities derived from these data for 44 of the sources with followup observations. Data on the remaining objects, selected to search for young supernova remnants, are given in Paper II. Finally, we have undertaken optical imaging and spectroscopic followup observations using the 1-m and 3-m telescopes of the Lick Observatory. Details of these observations will be presented in a forthcoming paper (Becker et al. 1990b); the existence of optical observations is indicated by an LO in the Comments column. The information in Table 4 is available in machine readable form and will be supplied to any interested users on request. An analysis of the survey will provide, among other things, a complete flux-limited sample of compact HII regions, a limit on the number of young SNR in the galactic plane, a significant number of new planetary nebulae and radio stars, and most likely, several unexpected discoveries. When coupled with the results of ongoing VLA plane surveys we are conducting at 6 cm and 90 cm, these data provide a qualitatively new view of the radio source population of the Milky Way.
----+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8 Frequency Telescope Resolution S_min Coverage No. of Reference (MHz) (arcmin) (Jy) l/deg b/deg Sources ----+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8 408 Molongo Cross 3. ~1. 195 - 55 +-3.0 ... Green 1974 408 DRAO Interferometer 3.5 .003* 65 - 180 +-3 in prog. Higgs 1989; Green 1989 1400 Greenbank 400 ft 10. 1. 12 - 55 +-4.0 356 Altenhoff et al. 1970 1400 Effelsberg 100 m 9. .3 93- 162 +-4.0 236 Kallas and Reich 1980 1400 DRAO Interferometer 1. .001* sampled in prog. Jonces and Higgs 1990 2700 Greenbank 140 ft 11. 1. 345 - 75 +-2.0 356 Altenhoff et al. 1970 2700 Parkes 64 m 8.2 .2 190 - 61 +-2.0 ~890 Day et al. 1972 2700 Effelsberg 100 m 4.3 ~.1 357.4- 76 +-1.5 1212 Reich et al. 1984 4875 Effelsberg 100 m 2.6 ~.4 357.5 - 60 +-1. 1186 Altenhoff et al. 1978 5000 Ft. Davis 85 ft 10.8 4. 335 - 55 +-3.0 356 Altenhoff et aL 1970 5000 Parkes 64 m 4.4 ~.5 190 - 40 +-2.0 915 Haynes et al. 1978 10050 Nobeyama 45 m 2.7 ~.5 356 - 56 +-1.5 144 Handa et al.