Back to CATS manual [ CATS home ] [ Back to CATS list ]

IAU Symp. 179 New horizons from multi-wavelength sky surveys.
McLean et al. (eds) 1997 Printed in Netherlands. p.19-25.



National Radio Astronomy Observatory 580 Edgemont Road, Charlottesville, VA 88908 USA

1. Introduction

Radio surveys have an important role in astronomy, one that has changed with technology and scientific requirements. Most objects studied by radio astronomers today are the unexpected discoveries of early surveys. The survey "discovery" phase began with Jansky's detection of Galactic radio emission and Reber's 160 MHz survey showing that this emission is non-thermal. Surveys made just after World War II revealed strong discrete sources which were later identified with supernova remnants, radio galaxies, and quasars. Pulsars were discovered during a sky survey for scintillating sources. BL Lac objects appeared in early high-frequency surveys. The first gravitationally lensed quasar appeared in the extensive Jodrell Bank 960 MHz survey, and the first observation of gravitational radiation came from the binary pulsar serendipitously found in a pulsar survey.

2. Cosmological Evolution

The strong cosmological evolution evident in the first samples containing hundreds of extragalactic sources drove the second phase of radio surveying. Most radio sources in flux-limited samples are extragalactic, and evolution dominates their redshift distribution at all flux cutoffs So < 1 Jy. Consequently, radio samples are unlike the local samples of bright galaxies found at optical or infrared wavelengths:
(1) Nearby sources are rare. Less than 1% of the 5x104 northern-hemisphere radio sources with S >= 25 mJy at 4.85 GHz are associated with the 104 UGC galaxies larger than theta = 1', most of which are within about 100 Mpc (Condon et al. 1991). Similarly, there is < 1% overlap of this radio sample with the IRAS Faint Source Catalog, Version 2 (Moshir et al. 1992) galaxies (Condon et al. 1995). Unfortunately, disjoint samples discourage multiwavelength cooperation. Optical and near-infrared astronomers prefer the nearby galaxies which they can observe well, while radio astronomers favor luminous AGN at the edge of the universe.
(2) The majority of radio galaxies and quasars are found near their median redshift <z> =~ 0.8, nearly independent of So (Condon 1989). In the nearly Euclidean optical universe, faint objects are statistically more distant and have smaller angular sizes, but are intrinsically similar to bright ones. In the hollow-shell radio universe, reducing So does not yield a "deeper" sample; it adds sources with intrinsically lower luminosities. Characteristic flux densities and angular sizes exist because features in the radio luminosity and size functions are mapped directly onto flux-density and angular-size distributions. For example, there is a transition from classical radio galaxies and quasars to a mixture of starburst galaxies, Seyferts, and "normal" galaxies below S =~ 1 mJy at 1.4 GHz.

3. Extragalactic Astronomy

All-sky surveys continuum surveys capable of detecting 105 sources became practical with multichannel HEMT receivers in the 1980's. The NRAO 7-beam 4.85 GHz receiver was used on the Green Bank 91 m telescope to make the GB6 survey of about 75,000 sources (Gregory et al. 1996) stronger than 18 mJy in the northern hemisphere and on the Parkes 64 m to make the PMN surveys of the southern sky (Wright et al. 1996). Astronomical applications of such surveys include: (1) Obtaining the first statistically useful (N > 102) samples of nearby radio sources (e.g., Condon et al. 1991, 1995), (2) discovering intrinsically rare objects such as gravitational lenses, and (3) probing the large-scale structure of the universe at redshifts z~1.

4. New Large Continuum Surveys

Large aperture-synthesis surveys capable of detecting 10 sources are possible with modern computing power. Three are under way: the Westerbork Northern Sky Survey (WENSS), the VLA B-configuration Faint Images of the Radio Sky at Twenty Centimeters (FIRST) survey, and the NRAO VLA Sky Survey (NVSS) being made with the D and DnC-configurations. A fourth, covering delta < -30o, has been proposed (Large et al. 1994) for the Molonglo Observatory Synthesis Telescope. They all have significantly higher sensitivity, resolution, and position accuracy than existing single-dish surveys, and the WENSS and NVSS are the first source surveys sensitive to polarization as well as total intensity.
Surveys which can resolve many extragalactic sources face conflicting requirements for completeness and position accuracy. Surveys are not flux density(Jy) limited - they are surface-brightness (Jy beam-1 or K) limited.

condon figure N1
Figure 1. The cumulative fraction f(< phim) of faint sources with angular size phim is compared with the VLA B-, C-, and D-configuration resolutions is shown in the left, panel. The right panel indicates the completeness and reliability (for C =~ R and search radius rs = m sigmap ) for optical identifications of sources with rms position uncertainty sigmap with objects brighter than J = 22.5 near the Galactic pole (top), at |b| = 30o, |l| = 90o (middle), and in the Galactic bulge |b| < 30o, |l| < 45o (bottom).

For example, the average face-on disk surface brightness of spiral galaxies is =~ 1 K at 1.4 GHz, so this surface-brightness sensitivity is needed to detect complete samples of spiral galaxies above any flux limit So. Figure 1a shows the cumulative fractions of extragalactic sources versus angular size at 1.4 GHz flux densities S = 3 and 10 mJy. Surveys with resolution << 1' miss faint resolved sources. On the other hand, noise limits the rms position uncertainty sigmap of faint sources to sigmap >= sigma theta/(2SM), where sigma is the rms map noise, theta is the FWHM resolution, and SM is the source peak flux density. At the detection limit SM =~ 5sigma and sigmap =~ theta/10. The desire to make complete and reliable identifications with optically faint objects favors low sigma and hence low theta. For example, sigmap = 2.5" is needed to identify radio sources with J = 22.5 {the POSS II limit) objects at galactic latitude |b| = 30o at 90% completeness and reliability (Figure 1b), implying theta <= 25" is needed to identify sources at the survey limit. The WENSS and NVSS are optimized for completeness and photometric accuracy, while the FIRST survey favors high position accuracy.

4.1. WENSS

The WENSS (de Bruyn et al. 1994) will cover the Omega = 3.14 sr north of delta = +30o at 325 MHz and about one-third of this area at 610 MHz. The FWHM resolution of this survey is 55" x 55" cosec(delta) at 325 MHz and 30" x 30"cosec(delta) at 610 MHz. The 5sigma detection limit is about 15 mJy beam-1 at both frequencies, and about 3 x 105 sources are expected. The sky is being covered with a mosaic of fields separated by half the primary beamwidth. Every field is observed at 18 different hour angles in each of six array configurations over a period of six weeks. Thus the (u-v)-plane coverage and sensitivity to extended structures is excellent, and the data sample short-term variability at low frequencies. Finally, images in all four Stokes parameters (I, Q, U, and V) will yield polarization information for both discrete sources and diffuse Galactic emission.

Scientific goals of the WENSS include several spectral studies: (1) Selecting sources with extremely steep spectra, such as luminous radio galaxies at high redshifts, relic radio sources in clusters of galaxies, and pulsars. (2) Selecting a large sample of fiat-spectrum sources to be used in a search for gravitational lenses. (3) Selecting and studying compact steep-spectrum (CSS) sources. In addition, the WENSS should be effective at finding giant (> 1 Mpc) radio galaxies, low-brightness disk and halo emission from nearby spiral galaxies, and cross-identifying objects found in other wavelength regions.

4.2. FIRST

The 1.4 GHz FIRST is a high-resolution total-intensity survey of the north Galactic cap. It was designed to identify faint galaxies and quasars found by the Sloan Digital Sky Survey (Gunn 1995), detect significant numbers of faint starburst galaxies, and resolve many extended extragalactic sources. The 5.4" resolution ensures that the rms position errors are < 1" even at the survey sensitivity limit, 1 mJy beam-1 =~ 20 K. A catalog of fitted components from the first 1550 deg observed (= 75 sources deg-2) is now available on-line (Becker et al. 1995). Unlike the WENSS, both FIRST and NVSS cover each field with a single snapshot, so the (u, v) coverage is poor and the data must be processed carefully to maximize dynamic range.

4.3. NVSS

The 1.4 GHz NVSS is an "all-sky" survey, covering the Omega = 10.3 sr with delta >= -40o by a mosaic of 217,446 snapshot observations in the compact D and DnC configurations of the VLA. The principal data products will be (1) a set of 2326 4o x 4o "cubes" having three planes containing the Stokes I, Q, and U images with 4" FHWM resolution and (2) a catalog of almost 2 x 106 discrete sources brighter than 2.5 mJy beam-1 =~ 0.8 K (=~ 50 sources deg-2). The rms position errors are < 1" for strong sources, < 2.5 for the =~ 106sources stronger than 5 mJy, and ~= 5" for the faintest detectable sources. The principal scientific goal of the NVSS is to encourage multiwavelength research by providing complete and reliable samples of sources, especially nearby ones, for use by all astronomers. The NVSS should detect most bright galaxies {e.g., UGC galaxies), most sources in the IRAS Faint Source Catalog, and most classical radio galaxies and quasars (e.g., M87 at z = 2). To guarantee equal access, the NVSS team members have agreed to use only the electronically released results for their own research. Image cubes covering about half of the survey area are currently available on-line, along with a catalog of = 9 x 105 sources (Condon et al. 1996). The NVSS observations will be essentially complete on 1996 October 1.

5. Future Surveys

Imagine a phase space whose axes are survey parameters - sensitivity, frequency, polarization, time resolution, etc. ( cf. Harwit 1981). Astronomical phenomena are scattered throughout the phase space, some still awaiting discovery. Which regions remain to be explored, which will be scientifically interesting, and how can they be surveyed?


The first large-scale (survey volume ~= 107 Mpc3 >> 103 Mpc3 =~ galaxy correlation volume) spectroscopic survey for extragalactic H I clouds was recently proposed (Stavely-Smith et al. 1996). A 13-beam receiver on the Parkes 64 m telescope will cover the southern sky to a depth of 140h-1 Mpc with 14' angular resolution. The 5-sigma sensitivity of 20 mJy in one 14 km s-1 channel corresponds to an H I detection limit = 106 Msun(D/Mpc)2 for a galaxy with 200 km s-1 velocity width. This survey will rediscover known gas-rich galaxies (dwarf galaxies within about 30 Mpc, normal Sc galaxies within about 45 Mpc, and giant gas-rich galaxies such as Malin-1 out to 140 Mpc) and provide data on their H I mass function, space distribution, and group/cluster dynamics. It should discover new galaxies with low optical surface brightness and any isolated H I clouds. If numerous, they may contain a large portion of all H I in the universe. The longitude range 75o < l < 145o of the Galactic plane is being imaged in the H I with = 1' resolution and in the continuum at 408 and 1420 MHz (Normandeau et al. 1996). Single-dish data are used to fill the hole in the DRAO (u,v) plane and restore very extended structures.

5.2. Frequency nu; >> 5 GHz CONTINUUM SURVEYS

Continuum surveys now span 30 MHz to 5 GHz. Most known blazars were found in the higher-frequency surveys sensitive to compact sources with flat radio spectra. Surveys at higher frequencies have been considered in the hope that they might find a "new population" of sources with peaked or inverted spectra, but they are technically difficult. The beam solid angle of a telescope scales as nu-2 and system noise generally increases with frequency, so the the time needed to survey a fixed area of sky to a given flux limit rises very rapidly above 5 GHz. For example, the 100 m Green Bank Telescope (GBT) will be capable of making an all-sky survey at nu = 15 GHz with the resolution and sensitivity of the 1.4 GHz NVSS, but even with a 7-beam receiver it would take five years of continuous observing!

This region of phase space may also be empty. Chauvinistic radio astronomers see the IRAS lambda = 60 um survey as a \nu = 5000 GHz radio survey complete to S = 0.28 Jy over most of the sky. The hypothetical blazars with spectral peaks between 5 and 5000 GHz should appear in cross-identifications with radio sources stronger than 25 mJy at 5 GHz. Unfortunately, all IRAS blazars in the northern hemisphere are stronger than 250 mJy (or weaker than 25 mJy) at 5 GHz (Condon et al. 1995). Their absence in the decade above the radio flux-density limit suggests that few nonthermal sources peak at short cm wavelengths. The only significant new population of "radio" sources found by IRAS consists of dusty galaxies. Dusty galaxies at high redshifts remain the best candidates for surveys at the shortest radio wavelengths, in the lambda =~ 0.8 mm atmospheric window. At frequencies below the blackbody peak, the spectra of dusty galaxies and quasars are so steep (Snu =~nu3) that flux density is nearly independent of redshift above z =~ 1. Distant galaxies with luminosities near the knee of the evolved luminosity function pile up around S =~ 1 mJy, and the source counts near this critical flux density may exceed the Euclidean extrapolation by two orders of magnitude. Blain and Longair (1996) showed that the SCUBA multibeam bolometer mounted on the JCMT could survey about 0.1 deg2 to this level and detect up to 50 galaxies, enough to provide useful constraints on the early evolution of star-forming galaxies.


Strongly variable and transient radio sources are produced by X-ray binaries, black holes, gamma-sources, radio stars, novae, and other exotic objects in our Galaxy. Some have been discovered serendipitously and a few systematically (Gregory and Taylor 1986), but most remain unrecognized against the background of extragalactic radio sources. Known Galactic variables tend to be weak, have been or inverted radio spectra, variability time scales ranging from hours to days, and low duty cycles. A systematic survey of this population requires repeated observations with high sensitivity and resolution (to avoid Galactic and extragalactic confusion) at a short wavelength. A 7-beam lambda = 2 cm receiver could be built for the GBT and scanned parallel to the Galactic plane. At a scan rate of 10o min-1 it would make a Nyquist-sampled image covering |b| < 1o, 350o < l < 260o with 45" FWHM resolution and 2 mJy rms noise daily. Averaging images from N successive days would yield an image with N-1/2 lower noise for constant sources and differencing them would eventually detect all transient sources stronger than 10 mJy plus many fainter ones. This is a true "discovery" survey for the new century.


Becker, R.H., White, R.L., and Helfand, D.J. (1995) The FIRST Survey: Faint Images of the Radio Sky at Twenty Centimeters, ApJ, 450, 559 - 577

Blain, A.W., and Longair, M.S. (1996) Observing Strategies for Blank-field Surveys in the Submillimeter Waveband, MNRAS, 279, 847 - 858.

Condon, J.J. (1989) The 1.4 GHz Luminosity Function and Its Evolution, ApJ, 338, 13-23.

Condon, J.J., Anderson, E., and Broderick, J.J. (1995) Radio Identifications of Extragalactic IRAS Sources, A J, 109, 2318 - 2354.

Condon, J.J., Cotton, W.D,, Greisen, E.W., Yin, Q.F., Perley, R.A., Taylor, G.B., and Broderick, J.J. (1996) NRAO VLA Sky Survey,

Condon, J.J., Frayer, D.T., and Broderick, 3.J. (1991) UGC Galaxies Stronger Than 25mJy at 4.85 GHz, AJ, 101, 362 - 409.

de Bruyn, G., Miley, G., Tang, Y., Bremer, M., Brouw, W., Rengelink, R., Bremer, M., and Rottgering, H. (1994) The WENSS Project, Astron NFRA, August, 4-8 and

Gregory, P.C., Scott, W.K., Douglas, K., and Condon, J.J. (1996) The GB6 Catalog of Radio Sources, ApJS, 103, 427 - 432.

Gregory, P.C., and Taylor, A.R. (1986) Radio Patrol of the Northern Milky Way, A J, 92, 371-411.

Gunn, J.E. (1995) The Sloan Digital Sky Survey, BAAS, 28, 875

Harwit, M. (1981) Cosmic Discovery. Basic Books, New York.

Large, M.I., Campbell-Wilson, D., Cram, L.E., Davison, R.G., and Robertson, J.G. (1994) Increasing the Field Size of the Molonglo Observatory Synthesis Telescope, ProcASA, 11, 44 - 49.

Moshir, M., et al. (1992) Explanatory Supplement to the IRAS Faint Source Survey, Version 2 JPL D-10015 8/92, Jet Propulsion Laboratory, Pasadena.

Normadeau, M., Taylor, A.R., and Dewdney, P.E. (1996) The DRAO Galactic Plane Survey Pilot Project,

Wright, A.E., Griffith, M.R., Hunt, A.3., Troup, E., Burke, B.F., and Ekers, R.D. (1996) The Parkes-MIT-NRAO Surveys, ApJS, 103, 145 - 172.

WEB-version by S.Trushkin.