Reports from Observers

The zCOSMOS Redshift Survey

Simon Lilly (ETH, Zürich, Switzerland) that have been developed, using almost are the product of the gravitational growth and the zCOSMOS team* all of the most powerful observing facil- of initially tiny density fluctuations in the ities in the world. This next step is called distribution of dark matter in the Universe COSMOS and the ESO VLT will make a – fluctuations which likely arise from quan- The last ten years have seen the open- major enabling contribution to this pro- tum processes in the earliest moments ing up of dramatic new vistas of the fur- gramme through the zCOSMOS survey of the Big Bang, τ ~10–35 s. These densi- thest reaches of space and time – an being carried out with the VIMOS spec- ty fluctuations eventually collapse to make exploration in which the VLT has played trograph. gravitationally-bound dark matter struc- a major role. However, the work so far tures within which the baryonic material has been exploratory, and sampled only cools, concentrating at the bottom of the small and possibly unrepresentative vol- It is well known that the finite speed gravitational potential wells where it forms umes of the distant Universe. The next of light enables us to observe very distant the visible components of galaxies. step is to bring to bear on a single large objects as they were when the Universe area of sky the full range of techniques as a whole was much younger, and there- In many respects, the Λ-CDM paradigm by to directly observe the evolving prop- is strikingly successful, especially in de- erties of the galaxy population over cos- scribing large-scale structure. On galactic mic epoch. The most distant objects scales, current implementations of it face * Peter Capak2, Marcella Carollo1, Andrea Cimatti 3, Thierry Contini4, Emanuele Daddi 5, Luigi Guzzo 6, presently known lie at redshifts between some difficulties: for example, real galax- Gunther Hasinger 7, Jean-Paul Kneib8, Olivier six and seven (6 < z < 7) corresponding ies appear to have more angular momen- Le Fevre8, Steve Maddox 9, Henry McCracken10, to a “look-back” time of about 95 % of the tum than predicted in numerical Λ-CDM 11 12 Alvio Renzini , Marco Scodeggio , Gianni age of the Universe. Indeed, at the time simulations and the down-sizing trend Zamorani13, Stephane Arnouts8, Sandro Bardelli13, Micol Bolzonella13, Dario Bottini12, Angela that these objects emitted the light that we is in a sense opposite to that expected. Bongiorno13, Alberto Cappi13, Olga Cucciati12, now detect, the Universe was less than There is also no clear understanding Sebastien Foucaud9, Olivier Ilbert13, Angela Iovino12, one billion years old. of the links between galaxies and their 1 4 4 Pawel Kampczyk , Fabrice Lamareille , Roser Pello , nuclear supermassive black holes. These Vincent Le Brun8, Alexi Leauthaud8, Christian Maier1, Vincenzo Mainieri7, Christian Marinoni12, These observations have revealed a rich various shortcomings almost certain- Marco Mignoli13, Elena Ricciarelli13, Claudia phenomenology in the early Universe. ly reflect our poor understanding of how Scarlata1, John Silverman7, Lidia Tasca 8, Laurence As we look back in time, we see that the dark matter and baryons interact, of 8 12 13 Tresse , Daniela Vergani , Elena Zucca , Herve global star-formation rate was about the feed-back loops operating within the Aussel10, Marcella Brusa7, Alberto Franceschini 14, Paolo Franzetti12, Bianca Garilli12, Gianni Marconi 11, a factor of ten or more higher in the first baryonic material due to energy injection Baptiste Meneux8, Bahram Mobasher15, John third of the history of the Universe (at from star formation and active galac- Peacock16, David Sanders17, Roberto Scaramella18, z > 1) than it is now. It is clear that the tic nuclei and of the relative importance 19 20 Eva Schinnerer , David Schiminovich , Nicholas most violent star-bursting objects are en- in galaxies of internal dynamical evolu- Scoville 2, David Silva11, Ian Smail 21, Dario Maccagni12, Yoshi Taniguchi 22, Paolo Vettolani13, shrouded in dust and will make produc- tion and externally driven events such as and Alessandra Zanichelli13. tive targets of study in the future with mergers, in redistributing material within ALMA. Alongside these very active galax- them. Many of these current uncertainties 1 ETH Zürich, Switzerland ies, there are also examples of more are likely related to the environments 2 California Institute of Technology, Pasadena, USA 3 INAF – Osservatorio Astrofisico di Arcetri, Flo- passive galaxies which must have com- that a forming and evolving galaxy finds rence, Italy pleted their star formation quite early itself in. Except for the very richest en- 4 Laboratoire d’Astrophysique de l’Observatorie on. Consistent with our knowledge of the vironments (i.e. the rich clusters of galax- Midi-Pyrénées, Toulouse stellar content of galaxies today, we see ies), knowledge of the environments of 5 National Optical Astronomy Observatories, Tuc- son, USA in the high redshift Universe that high lev- distant galaxies is rather poor. One of the 6 Osservatorio di Brera, Milan, Italy els of star formation, and other signatures aims of zCOSMOS is to characterise 7 Max-Planck-Institut für Extraterrestrische Physik, of youthfulness, appear in progressive- these environments over a wide range of Garching, Germany ly more massive galaxies as we look back redshifts and thus to lead to a much bet- 8 Laboratoire d’Astrophysique de Marseille, France 9 Nottingham University, United Kingdom further in time, a phenomenon given the ter physical understanding of the forces 10 Institut d’Astrophysique de Paris, France rather confusing name of “down-sizing”. controlling the formation and evolution of 11 ESO galaxies through cosmic time. 12 CNR Istituto di Astrofisica Spaziale e In parallel, developments in cosmology Fisica Cosmica Milano, Italy 13 INAF – Osservatorio di Bologna, Italy and in particular the emergence of the Much of the progress in this field has 14 Università di Padova, Italy “concordance cosmology” (from observa- been driven by “Legacy” style pro- 15 Space Telescope Science Institute, Baltimore, tions of the microwave background, large- grammes, such as the Hubble Deep Fields Maryland scale structure in the present-day Uni- (HDF), and the GOODS project, in which 16 University of Edinburgh, United Kingdom 17 University of Hawaii, Honololu, USA verse and the Hubble diagram of distant the data have been archived and released 18 INAF Roma, Italy Type 1a supernovae) have given us for to the research community in a scienti- 19 Max-Planck-Institut für Astronomie, Heidelberg, the first time a theoretical paradigm for fically usable form. This allows a much Germany the formation of galaxies and other, larger larger community of astronomers, extend- 20 Columbia University, New York, USA 21 Durham University, United Kingdom scale, structures in the Universe – the ing well beyond the original team who 22 Tohoku University, Tokyo, Japan Λ-CDM model: Structures in the Universe acquire and first analyse the data, to use

42 The Messenger 121 – September 2005 the data to carry out their own research programmes. COSMOS and zCOSMOS 24.0 23.5 23.0 22.5 22.0 21.5 21.0 20.5 20.0 are both undertaken in this spirit, and the purpose of this Messenger article is to bring to the attention of potential users across the ESO community the features of the zCOSMOS programme, which has just started on the VLT in P75. GOODS

The global COSMOS project

The Cosmic Evolution Survey (COSMOS) was designed to bring to bear on a single very large field all of the tools and ob- servational techniques that have been de- veloped for the study of the distant Uni- verse. The COSMOS field (centred on HDFs 10 h 00m 29s +02° 12o 21ǥ) was chosen to be near the Celestial Equator so that it can be accessed from observatories in both hemispheres, e.g. the ESO VLT and ALMA as well as the large optical/infrared telescopes in Hawaii and Chile and the VLA radio telescope in New Mexico, USA.

The COSMOS project is built around a mosaic of 600 images taken with the Ad- 1.0 f 814–f 475 2.4 vanced Camera for Surveys (ACS) on the Hubble Space Telescope (HST). The Figure 1: The 1.7 deg2 COSMOS field compared with mosaic covers a contiguous area of large-scale structure in a ∆ z = 0.02 slice of the Uni- 2 verse at z ~ 1 (courtesy Andrew Benson). Dots repre- 1.7 deg and represents the largest single sent galaxies, colour-coded according to their ob- programme undertaken with the HST to served colour (lower scale) and by size according to date. The field spans a transverse dimen- apparent magnitude (upper scale). The distribution sion of 80 comoving Mpc at z ~ 1 and of galaxies in the cosmic web of filaments and voids is clearly seen. By comparison the much smaller fields 160 comoving Mpc at z ~ 3 and covers a of view of the HDFs and the GOODS surveys (on the 3 volume to z ~ 3 (about 50 million Mpc ) right) do not well sample the range of structures that is approaching that of the entire local and environments that are present in the Universe at Sloan Digital Sky Survey at redshifts any epoch. z < 0.1. The HST observations were com- pleted in June 2005. Despite being only ages from the CFHT and NOAO 4-m and near-infrared (150 nm – 8 µm) wave- single-orbit exposures, the broad F814W telescopes. The combined photometric bands. filter reaches to within 0.3 magnitudes catalogue contains well over 1 million of the well-known GOODS images even galaxies with photometrically estimated Deep imaging observations at other though the survey covers an area twenty redshifts and approximate spectral types. wavelengths, e.g. the X-ray and radio, re- times larger than the combined GOODS-N The field has also been observed with the veal the signatures of accretion onto and GOODS-S fields (see Figure 1). GALEX ultraviolet satellite to a depth black holes in active galactic nuclei (AGN) of AB ~ 26 at 150 and 225 nm. Observing and of energetic bursts of star formation Impressive as the HST images are, the time on the Spitzer observatory has been that are obscured by dust. A mosaic of real power of COSMOS stems from awarded to extend this photometric cov- deep X-ray images obtained with the ESA the addition of a wealth of other observa- erage into the mid-infrared 3–8 µm which XMM-Newton satellite have already yield- tions that are being amassed on this will substantially improve the photometric ed more than 1000 active galactic nuclei field by the truly global COSMOS consor- redshifts and the estimates of the stel- and about 100 X-ray selected groups tium. Most notably, the SuprimeCam on lar masses of the galaxies. Ultimately, we and clusters of galaxies, while deep VLA the Subaru 8-m has been used to ob- hope to have photometry and images images at 1.4 GHz reaching to about tain very deep BGVRIZ images of the from a suite of almost 30 intermediate- 50 µJy (5σ) will detect 4 000 radio sources. whole field with a limiting magnitude (5σ) and broad-band filters spanning the full Future observations planned at far-infra- of about AB ~ 26–27. These have been range of starlight in the ultraviolet, optical red and sub-millimetre wavelengths will supplemented by U-band and K-band im- complete the observational picture.

The Messenger 121 – September 2005 43 Reports from Observers Lilly S. et al., The zCOSMOS Redshift Survey

What distinguishes COSMOS from pre- Table1: Numbers of representative objects in the COSMOS survey. vious programmes such as the Hubble Deep Fields, GOODS and COMBO- COSMOS Inventory 17/GEMS, is its enormous area. This Category Selection Number Faint galaxies I < 27 1 million gives us: AB X-ray selected AGN I < 27 3 400 – unprecedentedly large samples of AB X-ray selected clusters S > 5 10 –16 c.g.s. 10 0 objects in the distant Universe, thereby X × Radio sources S > 50 µJy 4 000 ensuring statistical weight even for rare 1.4 classes of objects (see Table 1); Bright quasars B < 21 10 0 High z quarsars (z > 4) I < 25 50 – confidence that we are sampling truly AB representative volumes of the Universe ULIRGS … 3 000 Lyman break galaxies I < 25.5 10 000 at high redshift (mitigating the so- AB Passive galaxies (z ~ 3) K < 24 10 000 called cosmic variance problem asso- AB zCOSMOS 0.3 < z <1.2 I < 22.5 25 000 ciated with smaller surveys that effec- AB galaxies with redshift tively probe a one-dimensional beam zCOSMOS 1.3 < z < 2.5 BAB < 25 12 500 through the Universe); galaxies with redshift – the ability to place all objects in their environment, from small scale groups of galaxies up to the largest structures in the Universe.

Such a unique data set of course opens tra themselves yield important diagnostics requirements efficiently, the zCOSMOS other equally unique possibilities. For in- of the evolutionary state of individual programme is split into two components, stance, the HST images will allow the galaxies, including measures of the star- each requiring different VIMOS configura- distribution of dark matter to be mapped formation rate, dust extinction, the gas tions and exposure times. 13 down to structures of order 3 × 10 Mी, and stellar metallicities, and stellar popu- which may be compared with the distribu- lation parameters such as ages. The The “bright” sample of 25 000 COSMOS tion of luminous galaxies. In addition, the spectra can confirm the identifications of galaxies is selected to have IAB < 22.5. large number of quasars bright enough radio and X-ray sources through the The straight I-band selection yields for absorption line spectroscopy (Table 1) characteristic signatures of AGN or star- a sample of galaxies at 0.2 < z < 1.2, will enable us to map the distribution burst activity. Precise spectroscopic red- reaching 1.5 mag below L* at z ~ 0.7 of neutral gas in the intergalactic medium shifts can of course also be used to where it corresponds to selection in the and again, compare that with the large- improve and characterise the photometric rest-frame V-band. With a sampling scale structure defined by the galaxies. redshift schemes which can then be ap- rate of at least 70 % and a velocity accu- plied to every galaxy in the field. racy of at least 100 kms–1, enabling the 12.5 isolation of groups down to 3 × 10 Mी, The zCOSMOS redshift survey The VIMOS instrument on the VLT pro- the “bright” sample is designed to be vides ESO with a unique capability directly comparable to the very large zero- The COSMOS data sets mentioned above for undertaking such a survey and in P75 redshift samples (SDSS and 2dfGRS) consist of exquisite two-dimensional a Large Programme was awarded but at a look-back time of half the age of images of the sky at almost every imagi- 540 hours of observation time to carry out the Universe. The input target list is gene- nable wavelength. The crucial third the zCOSMOS redshift survey. This rated from the HST/ACS images. The dimension is added by knowledge of the programme is complementary to the other observations are made with the VIMOS redshifts of the sources. Some informa- large VIMOS programme, the VVDS MR grism in 1 hr exposures between tion on the redshifts may be derived from Survey carried out by the VIMOS Instru- 550 < λ < 960 nm at resolution R ~ 600. the broad-band colours of the objects, ment Team. About 160 galaxies can be observed so-called “photometric redshifts”, but the simultaneously. Successive VIMOS point- more secure and precise “spectroscopic The design of the zCOSMOS programme ings are stepped in Right Ascension redshifts” are required for many purposes: has been driven primarily by the desire to and Declination so that every galaxy in the The increased precision relative to the quantify the environments of galaxies and target sample has eight opportunities to best attainable photometric redshifts ena- AGN over a broad range of epochs. This be selected into a spectroscopic mask, bles the delineation of the cosmic web requires: A high sampling completeness ensuring a uniform statistical sampling a- of large-scale structure in the Universe, (~ 70 % of objects observed from a given cross the field without significant biases from small groups of galaxies up to the target sample); uniform sampling cover- against near neighbours, etc. largest filaments and voids. The measure- age across the whole field; a broadly con- ment of individual velocities of galaxies tiguous redshift coverage from very The extension to higher redshifts requires enables dynamical studies of these struc- low redshifts to redshifts z > 2.5 spanning a different strategy. We know from the tures, yielding masses, dynamical states 80 % of cosmic time; relatively high veloc- VVDS survey that simply selecting fainter and cosmological information. The spec- ity accuracy (100 kms–1). To achieve these galaxies results in a sample that is still

44 The Messenger 121 – September 2005 dominated by relatively low redshift galax- and the LR-Blue grism (370 < λ < 670 nm zCOSMOS schedule and ies, with only a small “tail” at higher red- at R ~ 200) should yield redshifts in data release plans shifts 1.5 < z < 4 emerging faintwards of 4.5 hr exposures for 12 500 such galax- AB ~ 23.5. In order to isolate this tail, the ies, with a similar sampling rate as for Execution of a programme of the size of zCOSMOS “faint” sample of 12 500 gal- the bright sample, and a velocity accuracy zCOSMOS on a single field places unique axies is selected using a combination of 300 kms–1. demands on ESO in the scheduling of the of proven colour-colour selection criteria, VLT: to be completed in a timely manner, specifically the (B-Z)/(Z-K) selection In both parts of the survey, radio source the UT3 must be used for this programme proven by the VLT K20 survey and a de- and X-ray source candidate identifications for essentially all the time that the field is velopment of the (U-G)/(G-R) selection are added to the masks either as random observable. Periodic data releases will be used by Charles Steidel and collaborators targets (which will be observed with made with a final comprehensive data set to isolate star-forming galaxies. These a roughly 70 % sampling rate) or, for high placed in the ESO Archive shortly follow- two selection criteria yield a sample of priority and urgently needed sources, as ing completion of the programme, thereby galaxies at 1.2 < z < 2.4 and BAB < 25.0. compulsory targets which are observed providing the general research community In order to keep the total programme size with close to 100 % efficiency early in the with a detailed census and sample of the manageable, the higher redshift part of programme. distant Universe. zCOSMOS is limited to the central 1 deg2 area, which nevertheless still yields a URL references comparable comoving transverse dimen- sion to that of the lower redshift com- http://www.astro.caltech.edu/~cosmos ponent. A four-pass strategy with VIMOS http://www.exp-astro.phys.ethz.ch/zCOSMOS

Figure 2: Simulation of the final zCOSMOS redshift survey showing the spatial distribution (in comoving space) of objects in the survey over the redshift in- terval 0 < z < 2.4, in which every dot represents a zCOSMOS galaxy with spectroscopically determined redshift. The red bars mark increments of 0.1 in red- shift z. The figure has been generated from the COSMOS mock catalogues produced from the Mil- lennium Run cosmological simulation (courtesy of Manfred Kitzbichler).

The Messenger 121 – September 2005 45