We have obtained CCD photometry for 9500, 12150, 13000
galaxies in the B, V, R (Johnson-Cousins) filters to limiting magnitudes of 24.5,
24.0, 23.5 respectively over an area of 0.4 deg
near the southern galactic
pole.
Automatic procedures were designed in order to reduce
the data in an homogeneous fashion. The main steps used in obtaining the
photometric catalogue are summarized as follows:
i) We have applied the standard pre-reduction techniques of bias subtraction,
flat-fielding by median filtering and cosmic-ray removal.
ii) Because different CCDs were used over the course of the programme, a large
set of colour equations and zero-points were measured in order to provide
reliable calibrations for the survey. We could then
transform all instrumental magnitudes into the Johnson-Cousins standard system.
iii) The survey data is a mosaic of 
CCD frames in each band which
have 
arcmin overlaps.
We could therefore perform an internal and global re-adjustment of the
zero-point of each CCD over the whole survey.
In this way, we reduce the zero-point field-to-field scatter from
mag to 
mag, in agreement with the photometric
uncertainty estimated using simulated images (
for 
and
at fainter magnitudes). We therefore reduce
the internal dispersion in our photometry to the intrinsic dispersion resulting
from the limitations of the photometric package.
iv) Accurate astrometry is performed for each CCD frame. The positional
accuracy is 
for objects with 
and increases to 
at
fainter
magnitudes, as estimated from objects located in the overlapping edges of
the CCDs.
v) Measurement of the object colours is performed by matching the astronomical coordinates in
each band with a tolerance of 
in position offset. The colour
completeness with respect to the B band (fraction of B objects with R and V
counterparts) is greater than 95% up to 
and drops to 60% at
24.5. The colour uncertainties are estimated as the corresponding
quadratic sum of the errors in the photometric magnitudes (
for 
and 
for 
).
vi) The star-galaxy separation is performed using a neural network and is
expected to be close to a 95% success rate for 
.
We use the R band
for the star/galaxy separation because it is our deeper band, it was
observed with the best seeing conditions, and it is used for the
spectroscopic selection. At 
, no separation is done due to the
small fraction of faint stars expected (
)
at the high galactic latitude
of the survey (
).
The first results from the large photometric catalogue show that:
i) The galaxy counts in apparent magnitude and their slopes in logarithmic scale
are in good agreement with previous CCD and photographic surveys and show an excess in all 3
bands with respect to the non-evolving models.
ii) The (B-V) and (B-R) median galaxy colours show a blueing shift of
mag from 
to 
. In contrast, the (V-R)
median colour is nearly constant up to 
.
iii) The galaxy counts are well fitted by a non-evolving model for 
.
using a 
derived from the new bright galaxy counts of
Bertin & Dennefeld (1997).
iv) Two peaks in the stellar distributions are present in good agreement with
the Robin & Crézé (1986) model of the galactic disk and halo.
The next step in the study of this photometric sample is
the analysis of the angular correlation function 
at faint magnitude
(Arnouts & de Lapparent 1997).
This work will provide information about the galaxy clustering
ever a wide range of scales and apparent magnitudes.
The high completeness rate in colour should also allow to
further characterize the properties of the faint blue galaxies.
Together with the photometric data, the redshift survey
of 
galaxies with 
will provide a multi-colour optical
luminosity function in the redshift range 
.
Altogether, these data
will allow us to address the issue of the evolution of galaxies
with 
. The photometric and redshift catalogues will also be
complemented by a spectral classification of all galaxies with 
(Galaz & de Lapparent 1996) and will thus provide an unique
database for studying the variations in the galaxy properties as a fonction
of environment and redshift.
Acknowledgements
We are grateful to the European Southern Observatory for the large amount of observing time and the corresponding logistic support which allowed to perform this observing programme in good conditions and to bring it to completion. We are grateful to Dr. P. Leisy for kindly providing us his cosmic removal algorithm, and to M. Fioc and Dr. B. Rocca-Volmerange for kindly providing us the non-evolving model to fit the differential galaxy number counts. We thank Dr. C. Willmer for fruitful discussions. We also thank the referee Dr. N. Metcalfe for his scientific comments and linguistic corrections.