Chandra Deep Field-South Catalog

• The Chandra Deep Field-South Catalog is a K-band selected, multiwavelength survey integrating optical to radio data for studying high-redshift galaxies up to z ≈ 2.
• It employs robust photometric redshift techniques with EAZY and simulation-based calibration to achieve σ_z ≈ 0.036 and empirical completeness above 85% at faint limits.
• The catalog features precise astrometric calibration, cross-validated spectroscopic redshifts, and the InterRest utility for accurate restframe SED interpolation essential for galaxy evolution research.

The Chandra Deep Field-South (CDF-S) catalog, and its extension to the Extended Chandra Deep Field-South (ECDFS), are flagship multiwavelength datasets designed to enable precision studies of galaxy evolution, large-scale structure, and the cosmic history of star formation and black hole growth. The CDF-S/ECDFS region has been the focus of extraordinarily deep imaging and spectroscopic campaigns spanning the X-ray to the radio regime, producing a suite of homogeneous, well-calibrated public catalogs. Among these, the optical-to-near-infrared (OIR) "Chandra Deep Field-South Catalog" produced by the MUltiwavelength Survey by Yale-Chile (MUSYC) (0903.3051) provides a K-selected, ten-band photometric resource for approximately 0.25 deg², with robust ancillary data and direct links to spectroscopic and redshift catalogs. Integration of consistent reduction pipelines, detailed simulation-based completeness quantification, and cross-catalog validation have established these datasets as definitive tools for the paper of high-redshift galaxies out to z ≈ 2.

1. Catalog Design, Selection, and Photometric Construction

The MUSYC ECDFS catalog constitutes a K-band (2.2 µm)–selected multiwavelength photometric catalog covering the full 0.5° × 0.5° ECDFS area. The dataset is constructed from a combination of publicly available optical imaging (U, U₃₈, B, V, R, I, z′ from WFI and Mosaic II) and proprietary or newly acquired z′, J, K data, supplemented by H-band data for approximately 80% of the field (from ISPI and SofI). All bands are PSF-matched to a common 1.5″ FWHM prior to photometric extraction, but detection is performed on the higher-resolution 1.0″ FWHM K-band mosaic to enable superior deblending and source identification.

Photometry is measured in dual-image SExtractor mode using the K-band image for detection, ensuring sample uniformity and minimization of color biases. Aperture corrections, background oversubtraction corrections, and uncertainties are directly calibrated using simulations, with SExtractor's AUTO aperture used for total fluxes. For K < 22.0 (AB), the catalog achieves a point-source 5σ completeness and reliability, with empirical completeness above 85% even at the faintest modeled intervals (21.75 < K < 22.0). Simulated injection of R¹ᐟ⁴-law extended sources is used to calibrate completeness for galaxies versus the near-100% completeness for point sources at similar depths.

2. Photometric Redshift Derivation and Methods

Photometric redshifts (photo-z) for all catalog sources are computed with EAZY, a code utilizing a non-negative linear combination of six template SEDs derived via non-negative matrix factorization (NMF) of synthetic galaxy libraries (De Lucia & Blaizot semi-analytic models). An explicit luminous young, dusty template (t = 50 Myr, A_V = 2.75) is added to better account for highly obscured systems.

Key features of EAZY in this context:

• The calculation of redshift probability distributions, not mere best fits, with estimators including zₐ (maximum likelihood, no prior), zₚ (maximum-likelihood using a K-band prior p(z|K)), and two probability-weighted mean estimates (zₘ₁ and the recommended zₘ₂).
• The definition and use of the normalized median absolute deviation (NMAD) for photo-z error quantification:

$$\mathrm{NMAD} = 1.48 \times \mathrm{med}(|\Delta z/(1+z) - \mathrm{med}(\Delta z/(1+z))|)$$

yielding σ_z = 0.036 (1σ; Δz/(1+z) ≈ –0.025 median offset), verified against a large spectroscopic sample.

• Outlier fractions—defined as |Δz/(1+z)| > 5σ_z—are ≈5% overall, with higher rates among known X-ray sources, reflecting intrinsic SED diversity and (potentially) AGN contamination.

3. Spectroscopic Validation and Redshift Compilation

Cross-matching with published spectroscopic surveys (K20, VVDS, X-ray catalogs, FORS2/VIMOS, IMAGES, etc.) results in a compilation of 5,374 redshift measurements for 2,914 unique ECDFS catalog sources. After applying rigorous internal quality controls, 2,213 sources receive robust spectroscopic redshift flags: 1,966 galaxies (z > 0) and 247 stars. The catalog’s detailed appendix provides for each source the sample origin, Δz/(1+z) median and NMAD, and outlier statistics, enabling users to assess possible sample-specific biases or systematics.

This extensive spectroscopic anchoring enables not only validation of the EAZY photo-zs (confirming the above NMAD and outlier rates), but also enables robust identification of rare objects (e.g., extremely dusty or passive high-z galaxies) and supports studies of galaxy properties as a function of environment or redshift.

4. Restframe Photometry and the InterRest Utility

A distinctive feature of the MUSYC ECDFS catalog is the release of InterRest, an IDL utility for interpolating restframe SEDs from observed photometry. InterRest uses the EAZY template set to derive color–color relations at each redshift, allowing prediction of restframe fluxes (e.g., R restframe) from observed colors (e.g., V–I), by relating “rest–observed” colors to “observed–observed” colors. The method employs robust routines and is designed for SED-consistent, large-sample applications, with direct compatibility with EAZY input/output.

Validation tests using overlapping observed filters find systematic offsets in interpolated-minus-observed magnitudes of ≲0.02 mag and random scatter ~0.05 mag, indicating that error budgets for standard science uses (e.g., computation of stellar masses, restframe color gradients) are dominated by external factors rather than method error.

5. Catalog Validation: Completeness, Astrometry, and Cross-survey Consistency

Multiple internal and external validation pathways are implemented:

• Simulated injection of synthetic sources with R¹ᐟ⁴ profiles allows quantification of completeness: 100% for point sources at K ≲ 22 and ≥85% for extended sources at the faint limit.
• Corrections for missed flux in SExtractor AUTO apertures, and for background oversubtraction, yield point-source total fluxes with systematic errors ≲0.02 mag and random ~0.03 mag.
• Astrometric calibration—via comparison to SPM and 2MASS—yields relative source position accuracy of ~0.15″ and absolute accuracy better than 0.3″.
• Zeropoint calibration is confirmed to ≲0.05 mag via color–color diagrams for stellar sources, and further verified via star-SED fitting.
• Stellar and galaxy classification uses the Bz′K criterion:

$(z' - K) \leq 0.3(B - z') - 0.5$

enabling robust discrimination of stars and galaxies with little overlap.

Cross-comparisons with independent and deeper catalogs (e.g., FIREWORKS, COMBO-17, GOODS-MUSIC) demonstrate close agreement in both photometry and derived properties.

6. Ancillary Datasets and Scientific Applications

The MUSYC ECDFS catalog is inherently synergistic with a rich set of multiwavelength datasets available in the field. These include:

• X-ray imaging (Chandra Deep Field South)
• Deep Spitzer IRAC (SIMPLE survey), medium-band optical and NIR imaging (Subaru, NEWFIRM)
• Narrow-band datasets for Lyα emitter selection
• Deep radio imaging

Such integration enables precise SED fitting for galaxy evolution studies, improved constraints on stellar mass and star-formation histories at z < 2, and detailed analyses of environment—especially with the availability of massive spectroscopic and photometric redshift samples. For example, the robust restframe color coverage facilitates evolutionary studies of the massive galaxy population over cosmic time, exploration of the role of environment, and the calibration of photometric redshift algorithms for faint, high-redshift sources. The public release of InterRest further expands the catalog’s value for derivation of restframe properties critical for evolutionary synthesis models.

7. Summary Table: Key Catalog Properties

Attribute	Value/Specification	Note
Area	~0.25 deg² (0.5° × 0.5°)	ECDFS field, high-uniformity region
Detection/Selection	K-band selected, K(AB) ≤ 22.0	5σ point-source limit
Bands	U, U₃₈, B, V, R, I, z′, J, H, K	H for ~80% of area
Point-source Completeness	≳99% at K < 22, ≥85% for 21.75 < K < 22 (extended sources)	Simulation-validated
Photometric Redshift Method	EAZY, 10-band SEDs, NMF templates, K-prior	σ_z = 0.036, NMAD-defined
Secure Spectroscopic Redshifts	2213 (1966 galaxies, 247 stars)	Cross-catalog from multiple surveys
Restframe SEDs	InterRest algorithm, template-based color interpolation	Systematic error ≲0.02 mag
Astrometric Precision	~0.15″ relative, <0.3″ absolute	SPM, 2MASS cross-checks
Photometric Zeropoint	Agreement to ≲0.05 mag	Cross-verified with stars and deeper catalogs

This combination of deep K-selection, multiwavelength completeness, robust redshift information, and validated restframe photometry positions the MUSYC ECDFS catalog as a foundational tool for statistical studies of galaxy evolution, population demographics, and environmental dependence at intermediate to high redshift. The extensive calibration, public data release, and integration with other ECDFS datasets ensure maximal utility for the extragalactic community.