Optical Catalogs
The whole process of source detection, computation of their basic photometric parameters and star-galaxy classification has been performed using SExtractor (SEx; Bertin and Arnouts, 1996). To improve the final outcome of SEx we performed both a preliminar treatment of the images and a final interactive checking of the star-galaxy classification.
Preliminary image treatment
When a large galaxy is contaminated by a number of small projected sources, as it is the case of the brightest cluster galaxies (panel [a] of the above Figure shows the BCG in Abell 193), SEx erroneously assigns pixels of the large galaxy to the small projected objects (SEx segmentation map in panel [b]), producing a bad photometry of the large galaxy as well as of the objects projected onto it. A similar situation is also present around very bright stars with extended wings.
To minimize the effect on the photometry of such large halos, they
have been modeled (with IRAF-ELLIPSE) and removed, thus allowing
a better photometry 
of the small objects projected onto them. The
pixels of the projected small objects have been then substituted, in
the original image, by the corresponding pixels of the models, thus
allowing a better modeling of halos. The iterative process ends up with
a couple of background subtracted images, the first one containing all
the objects, but those which have been modeled; the other one just
containing the extended objects removed from the first image
(segmentation maps in panels [c] and [d]). The Figure
at the left illustrates the
improvement in the photometry of the modeled objects through the
differences of areas, colors and magnitudes before (B) and after (A)
applying the outlined procedure.
This procedure also improves the determination of the global
background map and increases the detection rate of the objects
projected onto the modeled halos.
Detection of sources
For each cluster the detection of sources, the determination of their positions and geometrical parameters as well as the star/galaxy classification were done on the V band imaging. The image with just extended halo objects and that with the remaining ones have been processed separately and the final catalog has been obtained by merging the two catalogs. For our research topics it is desirable to perform the galaxy photometry within a given surface brightness limit. Since our imaging comes from two different cameras with different pixel scales, in order to make the limiting surface brightness independent of the pixel size (for given observing conditions), we decided to keep fixed the detection threshold per square arcsec, rather than per pixel. In particular, the detection threshold was set to a signal to noise (S/N) level of 4.5 sbg/arcsec2 (σbg is the standard deviation of the background in ADUs), which corresponds to 1.5σbg/pixel for the WFC@INT and 1.07σbg/pixel for the WFI@ESO. Given the typical values of σbg found in our images, these limits translate into an average detection limit of mV~25.7 mag/arcsec2, about 0.6mag deeper than SDSS, at the same S/N level.
The catalogs have been produced running SEx in single-image mode, keeping only objects detected in both bands. This allowed us to reduce the number of spurious detections. Instead, in order to obtain precise color indexes, we performed a second run of SEx on the B band images using the dual-image mode, in which the V band images were used just for detection. This allowed us to sample in both bands the same circular region (usually 5kpc radius, except for small objects, for which 2kpc radius was used).
Star-Galaxy classification
The star-galaxy classification was done relying upon the stellarity index CLASS_STAR computed by SEx (0/1=galaxy/star). We decided to make a robust, bimodal classification, imposing the following criteria:
Stars: CLASS_STAR ≥ 0.8 && AXIAL_RATIO ≥ 0.7
Galaxies: CLASS_STAR ≤ 0.2
Objects not fitting these criteria were left in a third group of Unclassified
classification. The catalogs have been interactively checked for possible
star/galaxy misclassifications, using several plots of different combinations
of parameters:
In all these diagrams, stars populate a narrow and well defined region, while
galaxies are more spread throughout the plane. The fraction of misclassified
objects in the original SEx's catalogs turned out to be minimal, less than 1%
of misclassified stars and 0.6% of misclassified galaxies, up to V~22.
Therefore, we consider that, after interactive cleaning, the remaining
misclassifications should be even smaller. For fainter objects the previous
diagrams lose their utility since stars and galaxies populate almost the same
region.
The Catalogs
Three catalogs for each field have been produced, containing galaxies, stars
and objects of unknown classification. The structure of all of them is the same:
This Figure reports the fraction of objects in each group (Star/Galaxy/Unclassified) as a function of the magnitude. As expected, the fraction of unclassified objects increases at fainter magnitudes, the increase being steeper for V>22, which also corresponds with a change in the trend of the galaxy fraction. This fact suggests that most unclassified objects are actually galaxies.
As expected, the fraction of unclassified objects increases at fainter
magnitudes, the increase being steeper for V>22, which also corresponds
with a change in the trend of the galaxy fraction. This fact suggests that
most unclassified objects are actually galaxies.
The total number of records in the different catalogs for the 77 clusters
turns out to be:
Stars: 197,208 (including 8,893 saturated stars)
Galaxies: 400,048
Uncassified: 184,924
Photometric uncertainties
When running without weighted images, SEx computes just the
photon-noise errors of the magnitudes. Since these values are unrealistically
small, we performed extensive simulations in which synthetic stars and
galaxies with exponential and deVaucouleurs profiles were inserted in the
original images. 
This Figure below illustrates both the systematic errors and the uncertainties of SEx magnitudes as a function of the input magnitudes for stars, as well as for exponential and and deVaucouleurs profile galaxies. As already stated by Franceschini et al.(1998) and Benitez et al.(2004), for the latter ones, SEx produces total magnitudes that are fainter than the input ones. The black squares in the lower panel report the internal errors found when comparing the magnitudes from the two fields (Abell 780 and 970) we observed with both our instrumental set-ups (INT/ESO-2.2; see Fig.11 in Fasano et al. 2006; see also the Optical Photometry web page).
For magnitudes fainter than V~21.5mag the typical uncertainties can be approximated by the following expressions:
σV(stars) = 100.309V-7.62
σV(exp.) = 100.271V-6.67
σV(r1/4) = 100.259V-6.34
Completeness
The above mentioned simulations were originally intended also to check
the detection rate (completeness) and the success rate of the star/galaxy
classification. However, we realized that this second step couldn't be
achieved, since the tools commonly used to simulate stars and galaxies
don't work well at faint magnitudes. For that reason, simulations used
only to estimate the detection rate in each field.
The Figure at the left shows the average
detection rate run by run (dotted curves) and the total average detection
rate (continuous line). We can considered our global catalogs to be
90% complete for objects with V~21.7mag. The 50% detection level is reached
at V~23.2mag. In addition, we provide in the Table below the V band magnitudes at which the detection rate
drops 90%, 75% and 50% in each field, as well as the surface brightness
detection thresholds in the same band. 