2 Galaxy catalogs

The generation of the EIS galaxy catalogs in patches C and D and their characteristics have been discussed in Paper VI. In that paper they were shown to be considerably more homogeneous than those derived from previous patches, with only small variations in depth. The 80% completeness limit was established to be $I\sim23.0$ and this has been chosen to be the imposed magnitude limit in the cluster search. As in previous papers, the odd and even catalogs extracted from single exposure images (see Papers I and II) were independently used to identify possible clusters of galaxies. This was done by applying the matched filter algorithm, described in Paper II, to six overlapping sections of approximately the same size covering each of the patches considered. For patch C the sections were chosen to avoid a small ($\sim\!\! 0.2$ square degree) shallow region mentioned in Paper VI. In order to guarantee a full overlap between the regions covered by the odd and even frames the edges of the patches were also trimmed, yielding an effective area of 5.3 and 5.5 square degrees for patches C and D, respectively.

The first set of candidate clusters derived from the even and odd frames consisted of over 150 objects in each patch. However, these included an unusually large number of unpaired highly significant detections. The visual inspection of all candidate clusters, together with the even and odd galaxy catalogs, showed that the observed asymmetries were due to the presence of spurious objects detected in the vicinity of bright, saturated stars. As pointed out in Paper VI, the reason for this is possibly an electronic problem of the old EMMI controller, when used in the dual-port readout mode. This problem affected the last three runs of EIS by producing faint light trails associated with saturated stars, when these are imaged in the lower-half part of the detector. Along the trail a number of spurious low surface brightness objects is identified by SExtractor, and these objects are therefore included in the even/odd galaxy catalogs. Their fraction is relatively small and they do not significantly affect the number counts or correlation function. However, they have a significant impact in the performance of the matched-filter algorithm, which identifies a large number of cluster candidates near bright stars. Since patches C and D are located at low galactic latitudes, where a large number of bright stars that produce saturated images is found, the frequency of the problem is large, affecting about 30% of the original detections.

Fortunately, the above problem can be partially overcome by taking advantage of the sampling strategy of the survey, whereby each position on the sky is sampled at least twice by different parts of the detector (see Paper I). Since the light trails are produced only when a bright star is imaged in the lower half of the detector, the spurious objects identified along the trail associated with any given star are present only in the even or in the odd catalog, but never in both. It is therefore possible to overcome the light-trail problem at the catalog level by using, instead of the odd and even catalogs, the catalog which only includes galaxies detected in both of these catalogs (hereafter, referred to as the paired catalog). By construction, this eliminates most spurious objects. The two disadvantages associated with this procedure are that only one catalog of candidate clusters can be produced, and that the galaxy sample is slightly shallower. To allow for a possible study of these effects, in this paper three lists of cluster candidates, corresponding to the even, odd and paired detections, are used. Of course, the above solution cannot be applied to samples extracted from the coadded images, which will therefore require some type of correction at the image level. Various alternatives are currently being considered.