High resolution observations of the RGB sample were made with the NRAO's Very
Large Array
(VLA) between
October 1992 and September 1995. The observations were recorded with the
two standard 50 MHz bandwidth IFs at an effective frequency of 4.885 GHz.
Table 1 (click here) summarizes the observing parameters including the epoch,
array configuration, average exposure time per source and beam size. On
October 3, 1992, data were collected while the VLA was in a hybrid A/D
configuration. We were able to obtain flux densities and positions for
these sources only by using the antennas in the low-resolution D-like
configuration which yielded insufficient positional accuracy for unambiguous
optical identification. For completeness we list these sources separately
but do not consider them part of our well-defined sample. They are excluded
from further analysis. The region of the sky covered by the D-configuration
observations is approximately defined by 0
< 
< 15
, 0
< 
< 40
and 15
< 
< 16
, 0
< 
< 15
,
although high resolution observations for several objects in this region
were obtained.
Except for the September 1995 experiment which used 3C 48, absolute flux calibration was set using 3C 286 and the flux scale of Baars et al. (1977) as modified in the 15APR92 version of the Astronomical Image Processing System (AIPS). Phase calibrators were observed every few hours during each of the experiments.
Data reduction consisted of making tapered 
CLEANed images and using only the first clean component to phase
self-calibrate the data (equivalent to using a point source model at the
location of the strongest radio source). A second untapered map, centered at
the location of the peak on the first map, was made and CLEANed. The rms
noise was measured in a region excluding all sources on the final map. The
position and flux density of all sources whose signal-to-noise ratio exceeded
5 were recorded.
In Table 2 we present the 1861 RGB sources for which radio components were
detected. We present only a sample page here; a full copy of the table is
available from the CDS via anonymous ftp to cdsarc.
[0]u-strasbg.fr
(130.79.128.5), via the WWW at http://cdsweb.
[0]u-strasbg.fr/
[0]Abstract.html or at ftp://ftp.
[0]astro.
[0]psu.
[0]edu
[0]/pub/
[0]edf
[0]/rgb_tab2.html
and rgb_tab3.html, or by contacting the authors. The columns in Table 2 give
the source name, J2000 radio position, observation code (defined in
Table 1 (click here)), signal-to-noise ratio, corrected 5GHz core VLA flux
density (
), total 5GHz Green Bank flux density taken
from the GB96 catalog or from the reanalysis of the GB
survey images (
), and error of the total flux density if
the source appeared in GB96. We refer to individual
sources using the catalog prefix ``RGB J'' (RASS-Green Bank catalog, J2000
epoch positions) and append ``A'', ``B'', ``C'', etc. to denote multiple
radio sources found on a particular field.
In Table 3 we show a typical page of similar information for the 436 sources detected only at low resolution which have been excluded from further analysis. Table 7 (click here) lists the 83 fields for which no source with a signal-to-noise ratio greater than 5 was detected. Many of these are faint sources cataloged by Neumann et al. (1994) but not in GB96 which used the stricter criterion for source existence and are probably spurious. The columns list the source name, J2000 Green Bank position, observation code, total GB 5GHz flux density, and error in the total flux density if the source appeared in GB96. In addition, two RGB sources (RGB J0425+179, RGB J1303+488) were not observed with the VLA, but are part of the complete RGB sample.
The last column in the tables indicates the presence of a note which
indicates: (1) the source may be spurious or related to a diffuse Galactic
object (e.g. a supernova remnant); (2) the core radio flux density is from an
observation other than this paper; or (3) the source is more than 3
from its GB position (Sect. 3.2).
Instrumental effects degrade the measured flux density for sources far from
the field center. While time average smearing is insignificant for our
observations, both bandwidth smearing (chromatic aberration) and primary beam
degradation are significant for many sources in the RGB catalog. The
corrected flux density, S, is given by:
where 
is the flux density from the final map. The bandwidth
smearing correction term, B, and the primary beam correction term,
P, are given by (Condon et al. 1996):
Here 
is the bandwidth (50MHz), 
is the observing frequency
(4.885GHz), 
is the angular distance from the field center, 
is the restoring beam size (Table 1 (click here)), 
(
in arcminutes), 
, 
, 
, 
and 
.
While the formal uncertainties for our reported flux densities and positions can be defined as a quadratic sum of the squares of several error terms (e.g. Condon et al. 1982; Kollgaard et al. 1994), we found these formal uncertainties underestimated the true uncertainties in the reported source parameters. The biggest sources of error in the RGB catalog are instead due to instrumental and technical effects intrinsic to our snapshot mode. In order to assess the reliability in our measured flux densities and positions, we observed 20 RGB sources at more than one epoch after the main survey was completed and used the same data reduction procedure to obtain core flux densities and positions.
Figure 1: Distribution of VLA - Green Bank radio positions
These repeated observations show the reported positions for sources in
Table 2
are accurate to 
while those in Table 3 are accurate to
. The core flux densities of the sources observed at
multiple epochs varied significantly, however, with the source intensity
varying by as much 80% between epochs separated by as little as 10 days.
While some of this variability may be intrinsic to the sources, we
believe much of it is due to instrumental causes such as different VLA
resolution, the lack of phase calibrators near individual sources, very short
observation times, and consequently the small number of visibilities used to
image large fields. Our tests show the reported flux densities of sources
fainter than 
20 mJy are generally accurate to 
50% and the brighter
sources accurate to 
20%.
Figure 2: Log - Log diagram of the core VLA radio flux density (mJy) vs. Green
Bank flux density (mJy) at 5GHz
