Abstract
The ALHAMBRA-Survey is a project to image a large area, 4 square degrees,
with 20 contiguous, equal width, medium band filters covering from
3500 Å to 9700 Å, plus the standard JHKs near-infrared bands.
The photometric system in the optical was optimized to get (for a fixed
amount of total observing time) the maximum number of objects with accurate
SED classification and redshift and to be sensitive to relatively faint
emission features in the spectrum. Thus the ALHAMBRA-Survey is expected to
produce accurate enough photometric redshifts to track the cosmic evolution,
i. e., the change with z of the content and properties of the Universe,
a kind of Cosmic Tomography.
The observations will be carried out with the Calar Alto 3.5m telescope using
the new wide field cameras in the optical, LAICA, and in the NIR, OMEGA-2000.
We intend to reach, for a total of 100 ks integration time per pointing, the
limit AB ≥ 25 mag (for an unresolved object, S/N=5) in all the optical
filters from the bluest to 8300 Å, and from AB = 24.7 to 23.4 for the
remainder. The expected limit in the NIR, for a total of 15ks exposure time
per pointing, is Ks = 20 mag, H = 21 mag, J = 22 mag, at S/N=10.
The deep, homogeneous and contiguous spectral coverage will result in several
hundred thousand objects with accurate SED identification and z-values. This
accuracy will allow us to study, among others, the large scale structure
evolution with z, the identification of clusters of galaxies (with
membership assessment for a fraction of the galaxies), the identification of
families of objects, and other detailed studies, without the need for any
further follow-up. Indeed, it will provide exciting targets for 10m class
telescopes, the GTC in particular. Given its area and spectral coverage and
its depth, apart from those main goals, the ALHAMBRA-Survey will also produce
valuable data for galactic studies.
1 Introduction: Global scientific aim and opportunity
One of the main topics in Cosmology is Cosmic Evolution. The central
issue is to disentangle genuine cosmic evolution from physical variance at a
given redshift and the details of the metric, what has been a permanent
challenge for Physical Cosmology. To approach the question of Cosmic
Evolution meaningfully it is therefore necessary to sample large physical
volumes even at low redshifts, to capture not only representative average
properties but also their variance.
From an observational point of view this implies a combination of wide area
and depth, with a continuous spectral coverage to avoid having complex
selection functions depending on z and on the nature of the objects.
The quest for precision implies a good enough spectral resolution and
photometric accuracy as well. Up to now, the largest surveys were photometric,
ensuring a complete spectral coverage with broad-band filters. The resulting
precision in z obtained via photometric redshift techniques (~ 0.1 in
∆z/(1+z), at best) and in Spectral Energy Distribution (SED)
determination are correspondingly rough. Moreover, large area surveys like
the SDSS are correspondingly shallow, whereas deeper surveys have sampled the
distant and/or faint Universe in rather small areas. At the other extreme in
spectral resolution, spectroscopic surveys cannot go as deep as the
photometric surveys, reaching only I ≈ 24 with the use of large
telescopes. The covered fields are necessarily small and cannot cope with
the complex variety of objects in the Universe.
Obviously the spectrophotometric surveys will always run behind the
photometric ones. Thus, the problem at any given moment is to find the
optimal filter combination to produce the deepest and more accurate possible
photometric survey. The first proposal to use intermediate band filters to
cover the optical range, as a compromise between resolution field and depth,
was that of Hickson et al (1994). Later, the project CADIS was defined and
carried on with the 2.2m telescope in Calar Alto, using a combination of
broad and narrow band filters. Its primary goal was to find emission line
galaxies, even if it was exploited in many other fields (see the
CADIS
web page ). Later, on the experience gained by CADIS, the project
COMBO-17 was defined for the MPIA 2.2m telescope in La Silla. It also uses a
combination of broad (UBVRI) and narrow band filters to find accurate
z-values and SEDs for tens of thousand objects (see
COMBO web page
for all the details).
From the existing experience, we have designed a project to get the optimum
compromise between large area and depth, good spectral resolution and
coverage, to produce the optimum output in terms of redshift accuracy. This
is the Advanced Large Homogeneous Area
Medium Band Redshift Astronomical,
ALHAMBRA-Survey, a photometric survey primarily intended for cosmic
evolution studies. As we explain in the following, we intend to use
specifically designed intermediate band filters to continuously cover the
optical spectral range, plus the standard J, H, Ks NIR filters. The filter
system in the optical has been purposely designed to produce accurate enough
redshift and SED for hundreds thousand objects. Aiming at capturing the
cosmic variance at even relatively low z-values, we also decided to cover a
large area to very faint limits. We propose to cover a large-area with 20
contiguous, equal width, medium band optical filters from 3500 Å to
9700 Å, plus the three standard broad band, JHKs, in the NIR. Thus,
the ALHAMBRA-Survey is placed halfway in between the traditional imaging and
spectroscopic surveys.
It will make possible the study of many different astronomical problems in a
self-contained way. By design, the ALHAMBRA-Survey will provide precise
(∆z < 0.015(1+z)) photometric redshifts and SED classification for
≥ 300,000 galaxies and AGNs, allowing for any kind of analysis
regarding populations, structures and evolution. Thanks to the unbiased nature
of this survey (i.e. neither designed to detect a given class of objects nor
to be precise in fixed windows only), important problems other than Cosmic
Evolution can be addressed. These include the study of stellar populations in
the galactic halo, the search for very cold stars and blue stragglers, and the possible detection of debris from galactic satellites in the Milky Way halo.
Moreover, the large surveyed area and the ability to finely discriminate
between different spectral energy distributions will permit the serendipitous
detection of objects that could be classified as exotic or rare.
This broad category includes very high redshift galaxies ( ≈2500 objects
at z>5, with ∆z <0.01, expected from scaled HDF observations) and QSOs.
The main strategic goal of the ALHAMBRA-Survey is to provide the community
with a set of data appropriate for the systematic study of Cosmic Evolution.
The hypothesis of homogeneity and isotropy implies the existence of maximally
symmetric subspaces and the existence of a 1-to-1 relation between redshift
and time. This is a model-independent prediction, prior to any consideration
about the value of the cosmological parameters. Precisely, we intend to
materialize a foliation of the space-time, producing narrow slices in
the z-direction whereas the spatial sections are large enough to be
cosmologically representative, what could be called Cosmic Tomography.
This is a very demanding project indeed. The opportunity to undertake it was
prompted by the new possibilities open to the Spanish community in the Calar
Alto Observatory, now shared on equal basis by Germany and Spain. The Calar
Alto telescopes, in particular the 3.5m, equipped with the new instruments
OMEGA-2000 and LAICA, are specially well suited for such a work.
Indeed, a project like the ALHAMBRA-survey requires to be allocated an
important fraction of the observing time if it is going to be accomplished
in a reasonable period of time. Therefore the support of a large fraction of
the Spanish astronomical community was requested and eventually obtained,
and our survey was finally allocated Guaranteed Spanish Time.
2 The project implementation
The idea to use photometric information to determine the redshift of faint
sources was first proposed by Baum (1962), and later re-launched by
Loh & Spillar (1986) and Koo (1986) as a poor person machine to get
redshifts. Later, as we said before, Hickson et al (1994) discussed the
possibility to continuously cover the whole optical spectral range with
medium band filters. No discussion was done however in this work on the
number and kind of filters to optimize the output.
Taking together the constraints in coverage, resolution and depth imposed by
our scientific goals we have designed a system that optimizes the output in
terms of number of objects with accurate enough z and SED. The
ALHAMBRA-Survey is a multi-narrowband survey with complete spectral coverage
in the optical range. We have defined the filter system to have a complete,
homogeneous spectral coverage in the optical domain, and added the NIR
standard filters to complement the information about the detected objects and
to improve the z and SED determination, in the case of relatively
large photometric errors (see below) or for particular classes of objects. It
has also been adapted to the instrumental capabilities available now in Calar
Alto.
2.1 The ALHAMBRA-Survey optical filter system
The project was designed having in mind all the subtleties of the techniques
to get photometric redshifts, to be able to use them in the most advantageous
way (see Wolf, Meisenheimer and Röser, 2000, for an analysis of systems
including broad and narrow band filters). Our goal was to optimize the number
and width of the of filters to get, for a fixed total amount of observing
time, accurate SED and z determination for the largest possible number
of objects, and to be sensitive to relatively faint emission lines. Given the
performance of the instruments to be used, the total exposure time in the
optical domain was fixed to 100 ksec. Since we will use the standard filters
in the NIR, we concentrate in the following on the characterization of the
optical filter set.
We have analyzed four filter sets: constant or logarithmic (∆λ
∝ (1+λ)) width, with either minimal or 50% overlap. In all cases
the filters continuously cover the whole optical interval from 3500 Å
to 9700 Å, with almost square transmission. To test the efficiency of
the different systems we generated a mock catalogue of 13,000 galaxies upon
the magnitude, redshift and spectral type distribution of the galaxies in the
Hubble Deep Field (Fernández-Soto et al, 1999). Since the accuracy of
the input photometric redshifts is ≈ 0.06(1+z), we perturbed them by a
similar, randomly distributed amount to produce a more realistic redshift
distribution. We then generated magnitudes in each of the filter systems
above with realistic photometric noise added with the estimated performance
of the site + 3.5m telescope + LAICA cameras. The exposure times were
distributed among the different filters trying to reach constant S/N per
filter, but with two constraints: the minimal exposure time per filter is,
for practical reasons, at least 2.500 s, and we do not expose more than
twice this time in a homogeneous exposure distribution, to avoid spending
all our time on the less efficient filter/detector combinations.
The photometric redshifts were calculated using the BPZ software
(Benítez 2000, with the templates as in Benítez et al. 2004).
The main result of the simulations is illustrated in Figure 1, where it
appears that the minimally overlapping filter systems, either of constant or
logarithmic width, are the best performer for any fixed number of filters.
We also find that the improvement in the number-weighted precision of the
survey is slow after nf = 15 filters. Therefore, from that
point of view, the conclusion would be to use 15 minimally overlapping
filters, of 410 Å if they were of constant width, to cover the whole
spectral range.
3 Field selection
3.1 The total covered area
It is well known that astronomical objects are clustered on the sky on
different scales. The clustering signature contains a wealth of information
about the structure formation process. A survey wanting to describe and
understand the clustering needs to probe as many scales as possible, until
the homogeneity scale is reached. In particular, searching contiguous areas
is important to cover smoothly the smallest scales where the signal is
stronger and to obtain an optimally-shaped window function.
Measuring a population of a certain volume density is a Poissonian process
with an associated variance. One would obtain different densities of the same
population when measuring in different places. The variance in those measures
is dictated by the volume density of the population under study, the volume
searched and the clustering of the population. In order to beat down this
sample (or cosmic) variance one needs to sample independent volumes. So there
should be a balance between probing contiguous area and independent areas. On
the technical side, the geometry of LAICA, the 3.5m Calar Alto telescope
instrument with which the ALHAMBRA survey is devised to be carried out in
the optical, imposes a minimum contiguous area patch of 1° x 0.25°.
The relative error in any counting statistical measure will scale as
3.2 The selected fields
To select the fields to be covered we have taken into account evident
criteria like low extinction, no (or few) known bright sources, high
galactic latitude, and, very important for future work, significant overlap
with other surveys and/or other wavelengths. The selected fields are listed
in the Table 1 below.
Field name | RA(J2000) | DEC(J2000) | 100 µm | E(B-V) | l | b |
---|---|---|---|---|---|---|
ALHAMBRA-1 | 00 29 46.0 | +05 25 28 | 0.72 | 0.017 | 113 | -57 |
ALHAMBRA-2 | 01 30 16.0 | +04 15 40 | 0.80 | 0.022 | 140 | -57 |
ALHAMBRA-3/SDSS | 09 16 20 | +46 02 20 | 0.015 | 174 | +44 | |
ALHAMBRA-4/COSMOS | 10 00 28.6 | +02 12 21 | 0.90 | 0.018 | 236 | +42 |
ALHAMBRA-5/HDF-N | 12 35 00.0 | +61 57 00 | 0.60 | 0.011 | 125 | +55 |
ALHAMBRA-6/GROTH | 14 16 38.0 | +52 25 05 | 0.35 | 0.007 | 95 | +60 | ALHAMBRA-7/ELAIS-N1 | 16 12 10.0 | +54 30 00 | 0.27 | 0.005 | 84 | +45 | ALHAMBRA-8/SDSS | 23 45 50.0 | +15 34 50 | 1.17 | 0.027 | 99 | -44 |
4 Global expectations
4.1 Sensitivity considerations. Instrument performances. Calibration Strategy
Taking into account the average extinction in Calar Alto and the performance
of the telescope and cameras, we have calculated the exposure time per filter
to reach the proposed limit. The results are plotted in Figure 3. They were
calculated for AM = 1.3, FWHM = 1.2". The average exposure time per filter
amounts to 5000s, for the reasons discussed before. In the bluest filters the
exposure time is fixed by the need to get a minimum number, actually 5, of
exposures to correct for cosmic and transitory artifacts. In the reddest
filters the exposure time is limited to fit within the total exposure time
allowed for a given pointing.
4.2 The number of objects with accurate SED and z determination
In Figure 4 it can be seen that we can obtain highly accurate, Odds=1,
∆ z/(1+z) ≈ 0.015 redshifts for ≈ 90% of galaxies with
IAB < 23.5, a total over 300,000. If we relax the selection
criteria to Odds > 0.99, we would then reach 90\% completeness at
$AB = 24, with a photo-z accuracy of ∆ z/(1+z) ≈ 0.03
(more than 500,000 galaxies). The results have been obtained for
the simulations described before. Let us point out that this is a minimum
since we intend to analyze the implementation and to use new and more
detailed and specific templates than those used in the simulations, that
could improve the quality of the fittings and the final results.
5 The first observing run
Last August we had the first ALHAMBRA observing run, with OMEGA-2000.
Despite the problems encountered with the telescope control system, then in
the process of being overhauled, and the fact that we were the first visitor
users of the camera, we could obtain high quality data in J,H,Ks for the
equivalent of 2/3 square degrees. The data are not yet fully processed as we
are developing a whole pipeline to handle all the data flux we expect. We
experienced observing conditions that yielded a wide range of FWHM in the
resulting images. Eighty-five percent of the on-source integration time met
our requirement of image quality of FWHM ≤ 1.4''. With the observed
standard stars we estimate zero points for the system that are slightly
better than those measured with ISPI, a similar instrument mounted at the
Blanco 4m telescope at CTIO. Just comparing the count level in one of the
Ks images we obtained with the magnitudes of stars in common with 2MASS in
the field, we estimate that we reach a magnitude limit (preliminary value,
without correction for extinction) of ~19.8 at 5σ. This is somewhat
less deep than expected but we cannot make any clear statement before the
data are fully reduced and calibrated.
6 Final Considerations and Conclusions
The ALHAMBRA survey aims at filling a yet empty niche in astronomical
surveys, halfway between relatively shallow, wide-area spectroscopic
surveys, and deep, narrow-area photometric surveys. We intend to observe a
large area (a minimum of 4 square degrees) divided in eight separate
sub-fields using a specially designed set of 20 mid-band, minimally
overlapping filters covering the whole visible range from 3700 Å to
9700 Å, plus the standard JHKs near infrared filters.
The survey has been designed having in mind the use of photometric redshift
techniques as the basic analysis tool. We have carried on detailed
simulations based on available deep catalogues, and estimate that we can
measure high-quality redshifts and accurate spectral types for more than
300,000 galaxies with ∆ z/(1+z) ≈ 0.015, and for more than half
a million galaxies down to IAB ≈ 24, with redshift
accuracy ∆ z/(1+z) ≈ 0.03. Approximately 2000 of these galaxies
will be at z > 5.
The main objective of our survey is the study of cosmological evolution,
under the many facets it can offer. We will study the evolution of the large
scale structure, the evolution of the populations of different cosmic
objects, and the processes leading to galaxy formation, evolution, and
differentiation. The unbiased nature of the survey will also allow for the
study of many different kinds of objects, ranging from emission-line galaxies
to the diverse types of AGNs, and stars in our own Galaxy.
At the present stage of the project we have started the data collection, with
the first run having taken place in August 2004, and next runs happening in
December 2004, February, and May 2005. The Core Team, constituted by the
authors of the present article, has already designed the data analysis
routines, and the first version of the data analysis pipeline will be
operative by mid 2005. We intend to complete the survey data acquisition
after six semesters (approximately July 2007) and offer the survey products
to the community two years after the acquisition of the last data.
The project involves the effort of many astronomers. The Core Team members
have the charge of the implementation, observations, data reduction and
analysis till producing the final Catalogue. It is backed by the Extended
Team, made by 51 astronomers form essentially all the institutions in Spain
and some others, that help in defining specific tasks or implementations.
They have access to the data on equal foot as the CT members and will play
a fundamental role to extract the maximum scientific output from the
ALHAMBRA-Survey.
References
Baum, W.A., 1962, IAU Symp. 15, 390
Benítez, N., 2000, ApJ, 536, 571
Benítez, N. et al, 2004, ApJS, 150, 1
Fernández-Soto, A., Lanzetta, K.M., Yahil, A., 1999, ApJ, 513, 34
Fukugita, M., Ichikawa, T., Gunn, J.E., Doi, M., Shimasaku, K., Schneider,
D.P., 1996, AJ, 111, 1748
Hickson, P., Gibson, B.K., Callaghan, K.A S., 1994,MNRAS, 267, 911
Hopp, U. \& Fernández, M., 2002, Calar Alto News Letter
Koo, D.C., 1999, ASP Conf. Ser. 191, 3
Loh, E.D., Spillar, E.J., 1986, ApJ, 303, 154
Massey, P., Gronwall, C., 1990, ApJ, 358, 344
Oke, J.B., 1990, AJ, 99, 1621
Oke, J.B., Gunn, J.E., 1983, ApJ, 266, 713
Smith, J.A. et al, 2002, AJ, 123, 2121
Stone, R.P.S., 1996, ApJS, 107, 423
Terlevich, R.J., Melnick, J., Masegosa, J., Moles, M., Capetti, H., 1991,
A&AS, 91, 285
Wolf, C., Maisenheimer, K,, R\"oser, H.-J.,2001, AA, 365, 660