zCOSMOS: A LARGE VLT/VIMOS REDSHIFT SURVEY COVERING 0 < z < 3 IN THE COSMOS FIELD
1
S. J. Lilly,
2
O. Le Fe
`
vre,
3
A. Renzini,
4
G. Zamorani,
5
M. Scodeggio,
6
T. Contini,
7
C. M. Carollo,
2
G. Hasinger,
8
J.-P. Kneib,
3
A. Iovino,
9
V. Le Brun,
3
C. Maier,
2
V. Mainieri,
8
M. Mignoli,
5
J. Silverman,
8
L. A. M. Tasca,
3
M. Bolzonella,
5
A. Bongiorno,
5
D. Bottini,
6
P. Capak,
10
K. Caputi,
2
A. Cimatti,
11
O. Cucciati,
9
E. Daddi,
12
R. Feldmann,
2
P. Franzetti,
6
B. Garilli,
6
L. Guzzo,
9
O. Ilbert,
5
P. Kampczyk,
2
K, Kovac,
2
F. Lamareille,
7
A. Leauthaud,
3
J.-F. Le Borgne,
7
H. J. McCracken,
13
C. Marinoni,
9
R. Pello,
7
E. Ricciardelli,
4
C. Scarlata,
2
D. Vergani,
6
D. B. Sanders,
14
E. Schinnerer,
15
N. Scoville,
10
Y. Taniguchi,
16
S. Arnouts,
3
H. Aussel,
13
S. Bardelli,
5
M. Brusa,
8
A. Cappi,
5
P. Ciliegi,
5
A. Finoguenov,
8
S. Foucaud,
17
R. Franceschini,
4
C. Halliday,
11
C. Impey,
18
C. Knobel,
2
A. Koekemoer,
19
J. Kurk,
11,15
D. Maccagni,
6
S. Maddox,
17
B. Marano,
20
G. Marconi,
21
B. Meneux,
6,9
B. Mobasher,
19
C. Moreau,
3
J. A. Peacock,
22
C. Porciani,
2
L. Pozzetti,
5
R. Scaramella,
23
D. Schiminovich,
24
P. Shopbell,
10
I. Smail,
25
D. Thompson,
10
L. Tresse,
3
G. Vettolani,
26
A. Zanichelli,
26
and E. Zucca
5
Received 2006 April 28; accepted 2006 November 21
ABSTRACT
zCOSMOS is a large-redshift survey that is being undertaken in the COSMOS field using 600 hr of observation
with the VIMOS spectrograph on the 8 m VLT. The survey is designed to characterize the environments of COSMOS
galaxies from the 100 kpc scales of galaxy groups up to the 100 Mpc scale of the cosmic web and to produce diag-
nostic information on galaxies and active galactic nuclei. The zCOSMOS survey consists of two parts: (1) zCOSMOS-
bright, a magnitude-limited I-band I
AB
< 22:5 sample of about 20,000 galaxies with 0:1 < z < 1:2 covering the whole
1.7 deg
2
COSMOS ACS field, for which the survey parameters at z 0:7 are designed to be directly comparable to
those of the 2dFGRS at z 0:1; and (2) zCOSMOS-deep, a survey of approximately 10,000 galaxies selected through
color-selection criteria to have 1:4 < z < 3:0, within the central 1 deg
2
. This paper describes the survey design and the
construction of the target catalogs and briefly outlines the observational program and the data pipeline. In the first
observing season, spectra of 1303 zCOSMOS-bright targets and 977 zCOSMOS-deep targets have been obtained.
These are briefly analyzed to demonstrate the characteristics that may be expected from zCOSMOS, and particularly
zCOSMOS-bright, when it is finally completed between 2008 and 2009. The power of combining spectroscopic and
photometric redshifts is demonstrated, especially in correctly identifying the emission line in single-line spectra and in
determining which of the less reliable spectroscopic redshifts are correct and which are incorrect. These techniques
bring the overall success rate in the zCOSMOS-bright so far to almost 90% and to above 97% in the 0:5 < z < 0:8
redshift range. Our zCOSMOS-deep spectra demonstrate the power of our selection techniques to isolate high-redshift
galaxies at 1:4 < z < 3 :0 and of VIMOS to measure their redshifts using ultraviolet absorption lines.
Subject headinggs: cosmology: observations — galaxies: active — galaxies: distances and redshifts —
galaxies: evolution — large-scale structure of universe — quasars: general — surveys
1. INTRODUCTION AND MOTIVATION
The overall scientific goals of the COSMOS survey (Scoville
et al. 2007a) are to understand the three-way physical interrelation-
ships between the cosmic evolution of galaxies, their central super-
massive black holes, and the larger scale environment in which they
reside. It is expected that the environment, from the 100 kpc scales
of groups up to the 100 Mpc scales of the cosmic web of filaments
and voids, must be playing a very large and possibly decisive role
in the evolution of galactic systems, yet rather little is known at
1
Based on observations undertaken at the European Southern Observatory
(ESO) Very Large Telescope (VLT) under Large Program 175.A-0839. Also based
on observations with the NASA / ESA Hubble Space Telescope, obtained at the Space
Telescop e Science Institute, operated by the Association of Universities for Research
in Astronomy, Inc. (AURA), under NASA contract NAS 5 Y26555, with the Subaru
Telescope, operated by the National Astronomical Observatory of Japan, with the
telescopes of the National Optic al As tronomy Observatory , operated by the Asso-
ciation of Universities for Research in Astronomy, Inc. (AURA), under cooperative
agreement with the National Science Foundation, and with the Canada-France-
Hawaii Telescope, operated by the National Research Council of Canada, the Centre
National de la Recherche Scientifique de France, and the University of Hawaii.
2
Institute of Astronomy, Department of Physics, Eidgeno¨ssische Technische
Hochschule, ETH Zurich, CH-8093, Switzerland.
3
Laboratoire d’Astrophysique de Marseille, France.
4
Dipartimento di Astronomia, Universita di Padova, Padova, Italy.
5
INAF Osservatorio Astronomico di Bologna, Bologna, Italy.
6
INAF-IASF Milano, Milan, Italy.
7
Laboratoire d’Astrophysique de l’Observatoire Midi-Pyre
´
ne
´
es, T oulouse, France.
8
Max Planck Institut f u
¨
r Extraterrestrische Physik, Garching, Germany.
9
INAF Osservatorio Astronomico di Brera, Milan, Italy.
10
California Institute of Technology, Pasadena, CA.
11
INAF Osservatorio Astrofisico di Arcetri, Florence, Italy.
12
National Optical Astronomy Observatories, Tucson, AZ; currently at
Laboratoire AIM, CEA / DSMYCNRS, Universite Paris Diderot, DAPNIA /SAp,
Orme des Merisiers, 91191 Gif-sur-Yvette, France.
13
Institut d’Astrophysique de Paris, UMR7095 CNRS, Universite
´
Pierre & Marie
Curie, 75014 Paris, France and Observatoire de Paris, LERMA, 75014 Paris, France.
14
Institute for Astronomy, University of Hawaii, Honololu, HI.
15
Max-Planck-Institut f u
¨
r Astronomie, Heidelberg, Germany.
16
Tohoku University, Tokyo, Japan.
17
Nottingham University, United Kingdom.
18
University of Arizona, Tucson, AZ.
19
Space Telescope Science Institute, Baltimore, MD.
20
Dipartimento di Astronomia, Universita‘ degli Studi di Bologna, Bologna, Italy.
21
European Southern Observatory, Garching, Germany.
22
University of Edinburgh, United Kingdom.
23
INAF, Rome, Italy.
24
Columbia University, New York, NY.
25
Institute for Computational Cosmology, Durham University, Durham,
United Kingdom.
26
IRA-INAF, Bologna, Italy.
70
The Astrophysical Journal Supplement Series, 172:70Y 85, 2007 September
# 2007. The American Astronomical Society. All rights reserved. Printed in U.S.A.
present about the environments of galaxies at high redshift. A re-
lated goal is to observe the dark matter distribution directly through
the gravitational weak shear of structures along the line of sight to
distant galaxies and to relate this to the distribution of galaxies.
The very impressive deep multiband photometry in the
COSMOS field ( T aniguchi et al. 2007; Capak et al. 2007) enables
the approximate redshifts of vast numbers of galaxies to be esti-
mated from their broadband spectral energy distributions ( Mobasher
et al. 2007; Feldmann et al. 2006). At moderate depths I
AB
23
and z 1:2 these photometric redshifts have a statistical accuracy
of
z
0:03(1 þ z). This is clearly adequate to identify the re-
gions of highest density (e.g., Scoville et al. 2006b) and to serve as
the basis of statistical studies of the galaxy distribution indepen-
dent of their environment (e.g., Sargent et al. 2007; Scarlata et al.
2007), but it is insufficient to delineate the cosmic web and com-
pletely inadequate to characterize the environments of galaxies on
the scale of galaxy groups—i.e., those environments in which most
galaxies actually reside and in which we may expect many of the
most important processes that may regulate the evolution and trans-
formation of galaxies to be operating. This is clearly illustrated in
Figure 1, where we construct mock surveys at z 0:7usingthe
COSMOS mock catalogs kindly provided by Kitzbichler et al.
(2006, private communication) using both spectroscopic and pho-
tometric redshifts. The power of spectroscopic redshifts in delin-
eating and characterizing the environments of galaxies therefore
motivates a major redshift survey of galaxies in the COSMOS field.
Specific science goals of zCOSMOS cover three broad cate-
gories. First, spectrosc opic redshifts allow us to generate maps
of the large-scale structure in the universe and to quan tify the
density field throu ghout the COSMOS volu me to z 3witha
precision impossible with photometrically estimated redshifts.
As well as enabling studies of the variation of galaxy properties
with local density, these density maps may be compared with
those produced by weak lensing shear maps (Rhodes et al. 2007)
and the hot-gas structures detected in X-rays ( Finoguenov et al.
2007). The density maps also allow us to determine where in
the large-scale structure X-ray and radio sources reside and, with
absorption-line studies of background quasars, to relate the dis-
tribution of gas to the large-scale distribution of galaxies. More
quantitatively, a major goal of zCOSMOS is to generate a cat-
alog of well-characterized groups to determine their number
density N() and to trace the development of galaxy properties
in groups with different physical characteristics, such as crossing
time and density, that are likely to be relevant for the evolution
of member galaxies. Many statistical measures of the galaxy dis-
tribution including the correlation function (r
p
, ) and the pair-
wise velocity dispersion can be determined as functions of galaxy
type.
Second, the spectra also provide important diagnostics on the
galaxies themselves, such as star-formation rates, active galactic
nucleus (AGN) classification, reddening by dust, stellar population
ages, and metallicities, as well as metallicities of emission-line gas,
fig. 1afig. 1b
Fig. 1.— (a) Simulation of structure revealed by a redshift survey of the characteristics of zCOSMOS-bright (see text for details) as generated from the mock catalogs of
Kitzbichler et al. (2006, private communication). Each panel (left to right, top to bottom) shows an increment of 0.003 in redshift starting at z ¼ 0:700 (top right). Solid
black dots show objects with spectroscopically determined redshifts. Larger shaded gray dots show all galaxies with I
AB
< 24 so as to show the underlying structure. The
depth of each panel in redshift space is approximately 5 times the redshift accuracy of the spectroscopic redshifts. (b) Same simulation as (a) showing galaxies inferred to lie
in the same volume (0:700 < z < 0:736) as defined with their photometric redshifts, which are assumed to have a 1 uncertainty of 0.03(1 þ z). In this case the depth of
this single panel is only 30% of the FWHM of the photometric redshift distribution leading not only to a smoothing of the structure in (a), but also the inclusion of galaxies
that in reality lie outside of this redshift range.
Fig. 1b
Fig. 1a
zCOSMOS 71
and the possibility, depending on the velocity resolution, of mea-
suring the internal dynamics of galaxies.
Finally, accurate and reliable spect roscopic redsh ifts re-
place and complement photometrically estimated redshifts. Spec-
troscopic redshifts provide a calibration of photometric-redshift
schemes that may then be applied to objects not observed spectros-
copically, including those fainter than the spectroscopic limit. By
eliminating catastrophic failures, and the sometimes complex red-
shift likelihood functions for individual objects, spectroscopic red-
shifts provide a secure determination of the various distribution
functions (properties) describing the galaxy population. Spec-
troscopic redshifts and spectral classification also provide con-
firmation of the identification of X-ray and radio sources.
The main goal of the spectroscopic survey zCOSMOS is thus
to characterize galactic environments throughout the COSMOS
volume out to redshifts of around z 3. At redshifts up to z 1
it is possible to design a survey that matches very closely the
parameters of the very large surveys of the local universe such as
the 2 Degree Field Galaxy Redshift Survey (2dFGRS; Colless
et al. 2001), allowing a precise quantitative comparison of struc-
tures, and the galaxies within them, over the last 50% of the life-
time of the universe. At higher redshifts, it is more difficult in
practical terms to s elect galaxies in a directly comparable way,
and it is also somewhat harder to measure the redshifts. In par-
ticular, some form of color preselection is required to isolate the
tail of high-redshift galaxies that appears at I
AB
> 23 (see, e.g.,
Le Fe
`
vre et al. 2005).
The VIMOS spectrograph (Le Fe
`
vre et al. 2003) on the 8 m
UT3 ‘‘Melipal’’ of the European Southern Observatory’s Very
large Telescope ( ESO VLT) offers a very high multiplexing gain,
making a large and densely sampled redshift survey of the large
COSMOS field practical. The zCOSMOS redshift survey has
been designed to efficiently utilize VIMOS by splitting the survey
into two parts. The first, ‘‘zCOSMOS-bright,’’ aims to produce a
redshift survey of approximately 20,000 I-bandYselected galaxies
at redshifts z 1 that is directly comparable to the 2dFGRS sam-
ple at z 0:1 in terms of the sampling rate and redshift measure-
ment success rate, the redshift velocity accuracy, and the range of
galaxy luminosities covered. Covering the approximately 1.7 deg
2
of the COSMOS field (essentially the full ACS-covered area)
thetransversedimensionatz 1 is 75 Mpc. The second part,
‘‘zCOSMOS-deep,’’ will observe about 10,000 galaxies se-
lected through well-defined color selection criteria to mostly lie
at 1: 5 < z < 3:0. Simply to keep the required amount of telescope
time manageable, the field of zCOSMOS-deep is restricted to the
central 1 deg
2
of the COSMOS field. However, at z 2 the sur-
vey subtends a transverse distance of 80 comoving Mpc, slightly
larger than the bright part of the survey at lower redshift. The
field centers are areas of zCOSMOS-bright and zCOSMOS-deep
and are given in Table 1.
zCOSMOS has been awarded about 600 hr of Service Mode ob-
serving on the ESO VLT, making it (Large Program 1 75.A-0839)
the largest single observing project undertaken so far on that fa-
cility. Observations started on 2005 April 1 and are expected to
take at least 3 years to complete. These first observations have al-
ready allowed us to assess the data quality and predict the ulti-
mate yield of the program.
zCOSMOS, like COSMOS generally, is undertaken in the
spirit of a Legacy program, with an emphasis on making the data
products of lasting and general usefulness to the broad community
of researchers. The purpose of this introductory paper is therefore
to explain the motivation for the detailed design of the observa-
tional program as it is currently being implemented at the VLT, to
describe the construction of the spectroscopic target catalogs and
to summarize the observational procedures, as well as the pipeline
used to reduce the spectra. We then present the results of various
checks undertaken on the first data obtained. These establish at
least a preliminary estimate of the reliability of the redshifts, and
their velocity accuracy, and allow us to anticipate the properties of
the final sample when the observing program is completed. We
show how the combination of spectroscopic measurements and
photometric redshift estimates can be used to verify the redshifts,
break the degeneracies caused by the (relatively few) single line
redshifts, and identify which of the less reliable spectroscopic red-
shifts are likely to be correct, further increasing the success rate.
Where necessary, a concordance cosmology with H
0
¼
70 km s
1
Mpc
1
,
0;m
¼ 0:3, and
0;
¼ 0:7isadopted.All
magnitudes are quoted in the AB system.
2. zCOSMOS SURVEY DESIGN
Given the scientific goals described above, the practical de-
sign of zCOSMOS is driven by the characteristics of the VIMOS
spectrograph. There are a number of trade-offs involving the bright-
ness and number density of the target population and the pattern of
telescope pointings, which af fect the required exposure time and the
success rate in determining redshifts, the total number of objects
that are observed and the sampling rate, which we define to be the
fraction of targets, selected according to some well-defined criteria,
that are actually observed spectroscopically.
The VIMOS spectrograph is a conventional multislit imaging
spectrograph that can observe simultaneously four quadrants,
each roughly 7 ; 8arcmin
2
, separated by a cross-shaped region 2
0
wide. The number of slits that may be placed in each mask de-
pends on the length of each spectrum and on the surface density
on the sky of the targets. While the total number of objects that
may be placed in the masks increases with increasing target density,
it is found that the sampling rate (i.e., the fraction of available tar-
gets for which spectra are obtained) decreases. In designing the
masks from an input catalog, some objects may be designated as
‘‘compulsory’’ targets of special interest, in which case they are
included in the mask design if at all possible. The majority of slits
are then assigned to ‘‘random’’ targets in the catalog, selected
so as to maximize the number of slits in each mask. Some o b-
jects may of course also be ‘‘forbidden’’ (e.g., if previously
TABLE 1
The zCOS MOS Field
Center (J2000.0)
Field R.A. Decl.
Total Field Size
( R.A. ; Decl.)
(deg)
Full Sampling Region
( R.A. ; Decl .)
(deg)
zCOSMOS-bright............. 10 00 28 02 13 37 1.30 ; 1.24 0.98 ; 0.95
zCOSMOS-deep............... 10 00 43 02 10 23 0.92 ; 0.91 0.60 ; 0.62
Note.—Units of right ascension are hours, minutes, and seconds, and units of declination are degrees, arcminutes,
and arcseconds.
LILLY ET AL.72 Vol. 172
observed) and not included. It is found statistically that the
addition of each compulsory target reduces the number of ran-
dom slits by two.
The quadrant design of VIMOS means that a large contiguous
area can be covered uniformly by stepping the field centers, or
‘‘pointings,’’ across the larger survey field by an amount in each
direction that is equal to the dimensions of the individual VIMOS
quadrants. We define the ‘‘coverage’’ as the number of opportu-
nities that a given point on the sky has to be included in a mask and
thus observed spectroscopically. The ‘‘sampling rate’’ is then the
fraction of real targets that are observed in the masks, which will
depend on the coverage and the local density of the targets. This
uniform pattern of pointings (see Fig. 2) produces a large con-
tiguous region of coverage equal to 4, surrounded by four edge
regions (of width the quadrant size plus the gap) with coverage
equal to 2, and four small corner regions with coverage equal to 1.
Repeating the pattern of pointings by designing more than one
mask for each pointing doubles these coverages. The multiple-
pass strategy, which is mandatory to achieve uniform coverage,
has the considerable benefit of producing a high coverage, sub-
stantially reducing the bias against near neighbors that is other-
wise inherent in slit spectrographs (see Fig. 4 below).
The primary science goal of zCOSMOS is to trace the large-
scale structure in the universe at high redshifts and to characterize
both linear and nonlinear density enhancements such as galaxy
groups. Ideally, the design goals were to ensure the following:
1. A high and uniform sampling rate across the field, with a
goal of 70%.
2. A high success rate in redshift determination, defined as the
fraction of objects actually observed that ultimately yield a reli-
able redshift, with a goal of 90%.
3. Velocity accuracies of order 100 km s
1
enabling dynam-
ical characterization of the environment down to low-mass scales.
4. A more or less contiguous redshift coverage in the COSMOS
field over 0 < z < 3, spanning 85% of cosmic time and containing
the peak in the global star formation rate and AGN activity.
Simulations with the mask design softw are (Bottini et a l.
2005) indicated that sampling rates of approximately 70% can
be achieved with VIMOS for a target density of about 20,000 deg
2
Fig. 2.—Pointing centers and total coverage of the two components of zCOSMOS compared with the mosaic of 600 COSMOS ACS images. The zCOSMOS survey
utilizes 90 pointings (with field centers indicated by the small circles) that are uniformly spaced in R.A. and declination so as to provide a constant sampling rate over a
large central contiguous area, with four edge regions with approximately half the central sampling rate and four corner regions with only a quarter of the central sampling
rate, as indicated by the regions bounded by straight lines in the figure. zCOSMOS-bright uses all 90 pointings with two mask designs at each position. This provides
coverage over almost the entire ACS mosaic and gives each target galaxy in the central region eight opportunities to be observed. zCOSMOS-deep uses a mosaic of 42
central pointings (crosses) with a correspondingly smaller coverage. The zCOSMOS pointings in the mosaic are designated in (x, y) starting at the lower right. The VIMOS
quadrant pattern is shown shaded for pointing (6, 5).
zCOSMOS 73No. 1, 2007
with the four-pass strategy described above for the LR-Blue grism
or with an eight-pass coverage (with two mask designs at each
pointing) for the longer spectra that are produced by the MR grism
and OS-Red filter. Taking into account these scientific and tech-
nical considerations, there is thus an optimal survey configu-
ration in two regimes, each having an input catalog with about
20,000 galax i es deg
2
.
2.1. zCOSMOS-Bri
ght
The brighter, lower redshift component of zCOSMOS has a
pure magnitude selec tion at I
AB
< 22:5 as used in the CFRS
(Lilly et al. 1995) and VVDS-wide surveys ( Le Fe
`
vre et al.
2005). This selection yields redshifts in the range 0:1 < z < 1:2.
The velocity accuracy below 100 km s
1
requires the R 600 MR
grism, necessitating an 8-pass sampling strategy and 1 hr integra-
tions to secure redshifts with a high success rate. The spectral range
is in the red (5550Y9650 8) to follow the strong spectral features
around 4000 8 to as high redshifts as possible. This observational
setup yields a sample that is directly comparable with the low-
redshift 2dFGRS at z 0:1 in terms of selection, in sampling and
success rates, and in velocity accuracy, as described in more detail
below.
2.2. zCOSMOS-Deep
In order to isolate galaxies at 1:5 < z < 2:5, from the much
larger number of low-luminosity galaxies at z < 1 (see Le Fe
`
vre
et al. 2005), some kind of color selection must be applied. The
use of well-defined color criteria for spectroscopic target selec-
tion is to be preferred over using the output of a photometric red-
shift scheme to ensure that the sample is uniquely and repeatably
defined.
At least two methods have been demonstrated to be effective
at isolating such galaxies: the BzK criteria of Daddi et al. (2004)
and the ultraviolet UGR ‘‘BX’’ and ‘‘BM’’ selection of Steidel
et al. (2004), which merges into the well-known U-dropout selec-
tion (Steidel et al. 1996) at higher redshifts.
The BzK criterion has the advantage of selecting both actively
star-forming as well as passively evolving galaxies in the range
1:5 < z < 2:5. Moreover, the star-forming BzK-selected galaxies
are on average more massive, are more dust-obscured, and have
higher star formation rates than UGR-selected galaxies ( Reddy
et al. 2005). To the limit K
AB
< 21:8, over 80% of them are de-
tected at 24 m with Spitzer MIPS in the GOODS field ( Daddi
et al. 2005; Reddy et al. 2005), and many qualify as ultraluminous
infrared galax ies ( ULIRGs). Howev er, their s urface den sity to
the limit of feasible optical spectroscopy and current K limits
(about 10
3
deg
2
; Kong et al. 2006), is too low to trace the LSS
with the desired accuracy, and for an optimal exploitation of the
multiplex of VIMOS. Thus, only combining the UG R and the
BzK criteria one can ensure a fairly complete inventory of star-
forming galaxies, trace the LSS to the required detail, and ful ly
exploit t he capabilities of the spectrograph. In the observations
carried out in 2005, the BzK selection has been limited to K
AB
<
21:85 (see below). In the meantime, considerably deeper K data in
theCOSMOSfieldhavebeenobtainedaswellasSpitzer 3.6 and
4.5 m data. These should allow a more optimized selection in
the future.
Galaxies in this redshift range are best observed with VIMOS
in the blue spectral region to pick up the stronger absorption fea-
tures in the range between 1200 and 1700 8. This effectively re-
quires that we apply an additional magnitude selection B
AB
25:0
to ensure an adequate signal in the continuum. This eliminates
most of the passively evolving galaxies at z 2. These galaxies
would be best observed with the LR-Red grism, but their surface
density is too low for an efficient use of VIMOS and will have to
be observed at other facilities. Clearly, the census of galaxies at
1:5 < z < 2:5 will not be complete until redshifts are secured
also for this kind of galaxies. These are likely to be the most mas-
sive in this redshift range, inhabiting the highest-density peaks in
the large-scale structure of the cosmic web. To complete the red-
shift coverage by including the passively evolving BzKs remains
a major goal of the broader COSMOS project.
Measurement of secure redshifts in this redshift range down to
B
AB
25 requires 4Y5hrofintegrationwiththeR 200 LR-Blue
grism, which gives a spectral range from 3600 to 6800 8. Com-
pared to zCOSMOS-bright, this setup extends the survey through
the relatively unexplored ‘‘redshift desert’’ to z 3, albeit with a
less straightforward selection function, somewhat lower velocity
accuracy, and, it is expected, a slightly lower success rate in mea-
suring secure redshifts.
3. CONSTRUCTION OF THE INPUT CATALOGS
3.1. The ‘‘zCOSMOS-Bri
ght’’ Galaxy Catalog
At the relatively bright magnitudes of zCOSMOS-bright, it is
possible to construct a catalog with selection criteria that are well
matched to those that have been used to construct the very large
spectroscopic surveys, e.g., 2dFGRS and the SDSS, in the local
universe, facilitating direct comparisons over more than half the
Hubble time.
There are complementary advantages and disadvantages to
using either HST o r ground-based images alone for the catalog
generation and the optimum strategy is to combine both approaches.
The primary input catalog was generated using SExtractor (Bertin
& Arnouts1996) applied to the COSMOS F814W HST ACS im-
ages sampled at 0.03
00
pixel
1
(Koekemoer et al. 2007; Leauthaud
et al. 2007) in a ‘‘hot and cold’’ two-pass process to firs t iden-
tify bright objects. This substantiall y reduced the tendency of
the HST-based catalog to ‘‘overresolve’’ extended galaxies into
multiple components. This initial S Extractor catalog was then
‘‘cleaned’’ by car rying out a de tailed comparison with one ex-
tracted from a stack of i
images obtained w ith MEGACAM on
the 3.6 m Canada-Fra nce-Hawaii telescope and processed at the
TERAPIX data reduction center in Paris. There is no significant
systematic photometric offset between these two catalogs at I
AB
22:5 but naturally there is some scatter, which causes objects
to cross th e boundary between the samples. Obje cts with mag-
nitudes within 0.3 mag were not considered significantly dis-
crepant, and primacy in these cases was given to the ACS-based
magnitudes.
All the more significant discrepancies were visually examined
over the entire field and resolved by eyeball inspection of the
images. These discrepancies had a number of origins that could
be easily dealt with:
1. Objects missing because they had been masked out of one
or other catalog due to bright stars. Fortunately, the diffraction
patterns in the two sets of images were rotated by about 10
, en-
abling the CFHT image to cover the extended diffraction spikes
in the HST images, making the final inaccessible area very small.
Objects missing in the ACS catalog were inserted with their
CFHT magnitudes.
2. Objects lying just outside of the jagged boundary of the
ACS imaging and in two small areas not observed by ACS. These
were also brought into the catalog with their CFHT magnitudes.
3. Multiple objects that had been blended together by CFHT.
Unless it was clear that in fact they represented a single galaxy
LILLY ET AL.74 Vol. 172
