Figure 2: Sky distribution. Closed symbols are new detections, open
symbols the previously known OH/IR stars re-detected in this survey. The
offsets refer to Sgr A* (
in Table 2). Note that
the overall concentration of the new detections toward the center is
partly due to the "survey beam response"
(Fig. 1 (click here)); the concentration in
the very center is however real
Figure 3: Velocity distribution. Symbols as in
Fig. 2 (click here). The dotted
lines outline the limited coverage of the monitor; the solid lines the
velocity interval covered by LWHM
Figures 2 (click here) and 3 (click here) summarise our results, where it should be noted that the
stronger (previously known) sources can be detected further out from the
pointing center because of the drop in sensitivity. Tables 2
and 3
(also available electronically through CDS) list all detections within
the surveyed area, also when it concerns a previously known source.
Figures 4 (click here) and 5 (click here) (only available electronically through CDS)
display the 1612 MHz OH maser spectra of all sources. Single peak or
suspected double peak detections with a peak flux exceeding 8
(36 mJy before correction in the ATCA data), are also listed. However,
we do not list any of the obvious single peaks in the region where
confusion with the extended continuum emission occurs. These detections
are probably not of stellar origin. This means we have excluded all
single peak detections within the box defined by 
, even when the peak flux exceeds
36 mJy. Actually, some of the double peak detections in this region
might be debatable for being stellar sources. Such cases are indicated
in Table 3. In case of a failure detecting the second peak, we usually
took the velocity of the second peak from our VLA/ATCA data, or else
from LWHM.
For each detection, at the red and blue shifted peaks, fluxes and velocities were determined. The position, together with the formal errors were measured with the AIPS fitting program IMFIT, in the channel with the highest peak flux. For the VLA data, J2000 positions and Galactic coordinates (l,b) were calculated from the B1950 positions and then truncated according to the IAU convention. The Galactic coordinates from the ATCA data were calculated after the inverse transformation from J2000 to B1950. All positional and kinematic data of the detections are given in Table 2. Table 3 lists corresponding physical data to each entry in Table 2. Where appropriate, we first list the VLA ("a", or "b") and secondly the ATCA ("c") data. If seen in both the low and high resolution VLA image cubes, we used the high resolution ("b") result for the position (Table 2) and the low resolution ("a") result for the velocity and flux information (Tables 2 and 3 and Fig. 4 (click here)); the other result was used as consistency check in such cases.
As the fluxes measured for the VLA data are averages over 17 observations and depend on the resolution, we show VLA and ATCA spectra separately in Fig. 4 (click here) and in Fig. 5 (click here). There are considerable difficulties when reaching such low noise levels in the GC area; residuals of the extended OH absorption, continuum subtraction, and very strong maser sources cause poor baselines in both the VLA and ATCA data. Therefore, at the position of the star in the image cube and avoiding the region of the stellar maser emission, a linear baseline was fitted to, and subtracted from the spectrum.
Figure 4: Multi-epoch averaged spectra from the VLA survey
Figure 4: Continued
Figure 4: Continued
Figure 4: Continued
Figure 4: Continued
Figure 5: Spectra from the ATCA survey
Figure 5: Continued
Figure 5: Continued
Figure 5: Continued
Figure 5: Continued
Figure 5: Continued
Table 2: OH/IR stars in the Galactic center: positional and kinematic data
In Table 2 for each entry we list in order: the survey code, the given
name according to its Galactic coordinates, the measured R.A. and
Declination in J2000 with the maximum error in either R.A. or
Declination, the blue and red shifted velocities, the derived stellar
velocity and the Galactic coordinate offsets with respect to Sgr A*. The
stellar velocity is taken as the mean of the velocities of both
intensity maxima, whereas the Galactic coordinate offsets (i.e. for a
"flat sky", or 
) is defined with respect to Sgr A*:\
Table 3: OH/IR stars in the Galactic center: physical data
In Table 3 we repeat the survey code and source name, give the primary beam attenuation factor, the peak flux density and (spatially and spectrally) integrated flux together with the estimated relative integrated flux error for the blue shifted side, as well as for the red shifted side of the stellar velocity. We also list the the shell expansion velocity, and the OH maser luminosity (for an assumed isotropic radiation field and a distance of 8 kpc to the GC; Reid 1993). If the source appears to be extended, we determined an approximate deconvolved elliptical Gaussian for the source (the major axis, minor axis and position angle). The angular broadening of sources is probably caused by instrumental effects, time averaging of the visibilities or the extended background, but we cannot exclude that an extreme case of interstellar scattering of individual sources also plays a role (Van Langevelde & Diamond 1991; Van Langevelde et al. 1992b; Frail et al.\ 1994). The primary beam attenuation factor was calculated for both the VLA and ATCA with a polynomial, given internally in AIPS. The expansion velocity is half of the velocity separation between the maxima at both sides; however, it is not always the full extent of the feature. The blue and red shifted integrated fluxes are calculated by integrating flux densities over the channels from the stellar velocity to the first negative flux density outside the maximum.
Both VLA and ATCA data sets show positional offsets when compared with the positions measured by LWHM and vLJGHW. The small difference between the LWHM and vLJGHW data is due to using phase calibrators with different positional accuracies (B1730-130 and B1748-253, respectively); the LWHM and vLJGHW positions however are consistent with each other. The internal alignment of the VLA data introduced a systematic positional shift, as did the self-cal iteration of the ATCA data. Hence, there is a significant offset of a few arc-seconds between the positions measured in our VLA and ATCA image cubes. Our ATCA positions however are roughly consistent with the LWHM and vLJGHW data. We therefore attribute the systematic difference between our VLA and ATCA positions to the self-cal iteration performed to align the monitor data. Furthermore, small errors are
introduced by transforming the coordinates to (and from) epoch J2000. We stress that our absolute positions are not expected to be accurate at the one arc-second level. By measuring the offsets with respect to Sgr A* (assuming no measurable relative proper motion of the OH/IR stars with respect to Sgr A* during the monitor), however we have taken out the relative differences between our VLA and ATCA observations, and from there they can be linked to other data sets.
The positional error quoted, for a given source in a data set, is the
larger of the formal errors in R.A. and Declination as derived by IMFIT
in the channel with the highest peak flux density of the source. The
measured positions in individual channels are consistent with the
position of the source. Flux density errors are also taken from IMFIT.
This relative error can be used to derive the error in the OH maser
luminosity. Because the spectral features are unresolved, we can only
quote the velocity resolution as estimate of the error in velocity: 1.14
(or 2.27) 
for the VLA and
for
the ATCA observations.
Similarly, the flux densities from the VLA data set depend on the spectral resolution used and on the effect of averaging flux densities over several epochs. OH/IR star OH flux densities and luminosities are variable, up to a factor two (e.g. Harvey et al. 1974; Herman & Habing 1985; Van Langevelde et al. 1990). The measured ATCA fluxes are therefore snapshots of the flux variability; the VLA fluxes approximate the average fluxes better, because of averaging out the amplitude variations over the monitor period. Estimating the variability effect would require detailed knowledge of the OH period and amplitude distributions and is therefore not attempted.
