The color catalog described here has been constructed from objects detected in the 150 s exposures using multicolor data presented in Paper III. Color catalogs extracted from the co-added images will only be available in the final release.
As discussed in Paper III, the generation of a color catalog for generic use is a complex task given the range of possibilities in its definition, which depends to a large extent on the science goals. Since the intention of the present paper is primarily to understand some general characteristics of the color catalog and evaluate its usefulness for the primary science goals of the survey, the following prescription has been adopted. Starting from the unique catalogs covering the whole patch extracted from the best available images in each passband (see Paper III), a color catalog is created by associating the objects detected in the different passbands (see Paper I). The final product is a unique entry for each object, listing the photometric parameters and flags for each detection in each passband, and information about the best seeing frames in all bands. This information allows limits on the color for objects not detected in some of the passbands to be calculated. At this point all detections are considered, regardless of the band in which they were detected and the nature of the object (star/galaxy).
It has been found, for instance, that this catalog contains objects
detected in the blue passbands without counterparts in V and
I. Visual inspection of some of these cases, which represent roughly
2% of the total number of objects for , shows that the
unusual colors observed come mostly from objects detected at the
border of individual frames or along diffraction spikes of bright
stars. Others are due to ghost images present in the B and V
images (see Paper III) and to merged images that are de-blended in
some passbands and not in others, depending on the seeing. Even
though this is the most general color catalog, it is not the most
convenient catalog to use. This is especially the case for extended
objects for which are contaminated by the problems mentioned above and
for which one would like to have the object centroid and measure
magnitudes with the same aperture in all bands. Another problem is
that the star/galaxy classification may vary from band to band, thus
implying that the object classification may not be unique.
To avoid some of these ambiguities, this paper considers only
point-sources as defined in the I-band where image quality is more
homogeneus than in other bands and it was preferred to the V-band
(where image quality is slightly better Figs. 1
and 3 of Paper III),
for consistency with the other patches and because the interest is
mainly red objects. The selected sample includes only objects
detected in the I-band brighter than I=23.0 and with SExtractor
stellarity index .
Note that star/galaxy classification is only reliable for magnitudes
brighter than and colors were computed using the
mag_auto estimator introduced in Paper I. The mean error in the
colors are less than 0.15 mag for sources brighter than the
star-galaxy separation limit, and about 0.4 mag near the limiting
magnitude of the sample.
Since the primary goal of this paper is to illustrate the possible use
of EIS data for different science goals and to evaluate the size of
interesting samples for follow-up work, the criteria adopted have been
in general conservative. In particular, in the selection of targets
discussed below only reliable I-band detections, with the error
() on the I-band magnitude
(
) are considered. Furthermore, to avoid regions of
very poor seeing and low transparency, some regions were discarded,
based on the distributions of seeing and limiting isophothes shown in
Paper III, leading to a sample covering a total area of 1.27 square
degrees (e.g.,. 4). In particular, it was
excluded the region with
,
whith B and I-band seeing
, and the region
,
with limiting
isophote in I-band
(very poor transparency).
To illustrate the general properties of the sample the color-magnitude
diagrams V versus (B-V) and V versus (V-I) are shown in
Fig. 1 for 3233 point-sources with
in
the I passband. The color distribution for sources brighter
than V=21 has already been discussed in Paper III and compared with
model predictions. This plot is presented in V-band to make the
comparison with other works easier
(e.g.,
cite[Reid & Majewski 1993]RM93). Note
that at faint magnitudes () incompleteness in the stellar
sample sets in (see Paper III). In the figure blue,
,and red,
concentrations are clearly seen for
. This reflects the typical bimodal color distribution observed
for faint stars at high-galactic latitude, with the blue peak arising
from the turnoff of the main sequence for low-metalicity halo stars
and the red peak from disk stars. Note that the blue peak is
well-defined in the magnitude range 18.5 < V < 21.5. For fainter
magnitudes it fades away partly due to the incompleteness of the
catalog and partly because at this magnitude limit one is approaching
the outer parts of the halo.
From the figure one sees that for there is a large number
of objects bluer than the concentration associated with the halo
stars, and for
there is a population of very red objects
with (V-I)>3.5. Both are examples of interesting populations, the
identification of which is better explored using the color-color
diagram as discussed in the following section.
Copyright The European Southern Observatory (ESO)