3D-HST WFC3 Photometric Catalogs

• 3D-HST WFC3-selected photometric catalogs are comprehensive multiwavelength datasets integrating HST imaging and extensive ancillary data to enable robust spectral energy distribution construction.
• They employ advanced detection strategies with dual-mode SExtractor, rigorous PSF homogenization, and flux scaling to achieve high completeness and precise photometry.
• Publicly released with detailed documentation and value-added physical parameters, these catalogs support in-depth analyses of photometric redshifts, stellar masses, and galaxy evolution.

The 3D-HST WFC3-selected photometric catalogs constitute a comprehensive, homogeneous set of multiwavelength photometric and derived data products for extragalactic fields targeted by the CANDELS and 3D-HST surveys. Built primarily on HST WFC3 near-infrared imaging and augmented by extensive ancillary ground-based and Spitzer/IRAC data, these catalogs enable detailed spectral energy distribution (SED) construction across 0.3–8 μm in five premier extragalactic survey fields: AEGIS, COSMOS, GOODS-North, GOODS-South, and UKIDSS UDS. The catalogs provide source identification, PSF-matched multi-band photometry, robust photometric redshifts, stellar population parameters, and are delivered in machine-readable formats with associated data products for public access and further analysis (Stefanon et al., 2017, Skelton et al., 2014).

1. Survey Fields, Data Content, and Catalog Scope

The 3D-HST WFC3-selected catalogs span five canonical CANDELS/3D-HST fields, each covering approximately 164–201 arcmin$^2$ of uniform depth after masking, totalling approximately 900 arcmin$^2$. The dataset integrates 147 distinct imaging datasets, including:

• HST/WFC3: F125W, F140W, F160W (primary detection bands; pivotal for near-infrared selection)
• HST/ACS: F435W, F606W, F775W, F814W, F850LP (core optical bands)
• Ground-based Optical/NIR: CFHTLS $u^* g'r'i'z'$, Subaru Suprime-Cam BRI/VRiz+medium, VLT/VIMOS, UltraVISTA, UKIDSS, NMBS, CFHT/WIRCam, VISTA, MOIRCS, ISAAC, TENIS
• Spitzer/IRAC: Channels 1–4 (3.6, 4.5, 5.8, 8.0 μm)
• Auxiliary data: Exposure/weight maps, empirical PSFs, segmentation images (Stefanon et al., 2017, Skelton et al., 2014)

Each field's catalog and the global master catalog contain photometry for tens of thousands of detected objects, and value-added tables for derived physical properties.

2. Object Detection and Source Extraction

Source detection is anchored to the WFC3 F160W band (or a noise-equalized combination of F125W/F140W/F160W) to optimize for high-redshift and faint galaxies, given the superior spatial resolution and sensitivity in these bands. Detection employs a two-stage SExtractor strategy in “hot” and “cold” modes to capture both compact and extended/bright sources:

• Cold mode: Prioritizes bright, extended sources using parameters such as DETECT_MINAREA=5 pixels, DETECT_THRESH=0.75σ, and deblending threshold DEBLEND_NTHRESH=16.
• Hot mode: Optimized for faint, compact objects (DETECT_MINAREA=10, DETECT_THRESH=0.70σ, DEBLEND_NTHRESH=64).
• Catalog merging: All cold-mode sources are included, while hot-mode detections are appended if they are outside cold Kron ellipses.
• Completeness: 90% point-source completeness is achieved at F160W = 26.62 AB (EGS field), with completeness versus magnitude described by a complementary error function:

$$\mathrm{completeness}(m) = 1 - \frac12 \mathrm{erfc}\left(\frac{m - m_\mathrm{lim}}{\sqrt{2}\,\sigma}\right)$$

where $m_\mathrm{lim}=26.62$, $\sigma\approx0.35$ (Stefanon et al., 2017).

This workflow ensures robust detection across object types and surface brightness, with high completeness confirmed via fake-source recovery experiments.

3. PSF Homogenization, Photometry, and Uncertainty Characterization

Photometric measurement across heterogeneous images requires rigorous PSF homogenization:

• Empirical PSFs are constructed for each band by stacking $\sim$50–200 bright, isolated stars, with neighbor masking and normalization.
• Convolution kernels are derived via deconvolution (e.g., IRAF/lucy, Fourier Hermite polynomials) to transform all images to the F160W PSF (FWHM ≈ $0.19''$–$0.20''$), with sub-1% matching accuracy within the color aperture (Stefanon et al., 2017, Skelton et al., 2014).
• Photometry: On PSF-matched HST imaging, SExtractor in dual-image mode measures:
  • FLUX_AUTO (Kron elliptical total flux, $\sim$94% of total)
  • FLUX_ISO (isophotal flux, maximized S/N for colors)
  • FLUX_APER (multiple circular apertures)
• Total flux calculation: For bands other than F160W:

$^2$0

This scheme preserves robust color information while providing consistent total-flux scaling (Stefanon et al., 2017).

Noise modeling and photometric errors are derived from local background RMS maps (HST bands) or by empty-aperture statistics (for lower-resolution or ground-based data). Flux uncertainties:

$^2$1

where $^2$2 is the extraction area and GAIN is the detector-electronics gain (Stefanon et al., 2017, Skelton et al., 2014).

4. Multiwavelength Catalog Assembly and Band Coverage

Each object’s multiwavelength SED is assembled by joining band-matched photometry using a unique identifier and enforcing a harmonized aperture correction across all filters. The catalogs span:

• UV–mid-IR: GALEX, optical (CFHT, Subaru), HST ACS/WFC3, NIR (WIRCam, NEWFIRM, UltraVISTA, UKIDSS), Spitzer/IRAC
• Photometry strategy:
  • High-resolution bands: Dual-image SExtractor on PSF-matched frames.
  • Low-resolution bands: TFIT or MOPHONGO template-fitting, using F160W-based segmentation as the prior, with neighbor subtraction.
• Non-detections: Fluxes and errors of −99 denote coverage gaps or bad pixels; $^2$3 upper limits can be constructed from local apertures (Stefanon et al., 2017).

Coverage variations, depth maps, and band-specific quality flags encode the spatially variable completeness and facilitate rigorous science analyses.

5. Photometric Redshifts and Stellar Population Parameters

Robust physical parameters are delivered as value-added products:

• Photometric redshifts ($^2$4):
  • Ten expert groups ran independent codes (EAzY, HyperZ, Le Phare, FAST, etc.) with varied template sets (BC03, MA05, PEGASE).
  • Training: 840 spectroscopic redshifts (DEEP2/DEEP3) in EGS, multiple spectroscopic sets in main fields.
  • Zeropoint calibrations minimize SED residuals (<0.05 mag typical shifts).
  • The consensus $^2$5 is taken as the median of individual $^2$6 PDFs; typical scatter $^2$7 (EGS), outlier fraction (|Δz|/(1+z)>0.15) ≈ 5% (Stefanon et al., 2017, Skelton et al., 2014).
• Stellar masses ($^2$8):
  • SEDs are fit at fixed $^2$9 by eight codes, using a range of SFH models (e.g., τ-models with Chabrier IMF).
  • Three groups apply nebular-line emission corrections.
  • Consensus $u^* g'r'i'z'$0 is the median of six τ-model fits (Chabrier IMF, BC03 templates).
  • 90% completeness thresholds for a passively evolving SSP: $u^* g'r'i'z'$1 (z=1), 10 (z=2), 11 (z=4) (Stefanon et al., 2017).
• Other parameters: SFR, age, $u^* g'r'i'z'$2 are reported as available; rest-frame colors (e.g., $u^* g'r'i'z'$3, $u^* g'r'i'z'$4) facilitate population studies (e.g., UVJ diagram bimodality) (Skelton et al., 2014).

6. Catalog Structure, Quality Flags, and Public Access

Each catalog is a multi-extension FITS binary or plain ASCII table containing core object identification, astrometry, photometry, photometric quality, and derived parameters. Key columns include:

• Object ID, RA/Dec (CFHTLS-aligned), IAU name
• For each filter: FLUX, FLUX_ERR (μJy), coverage/flag fields (−99 for non-detections)
• FLAGS (0=good, 1=low S/N, 2=star-spike, 3=both)
• Star/galaxy classifier (CLASS_STAR), AGNflag (X-ray), PhotFlag
• Matched spectroscopic redshifts (where available)
• Photo-$u^* g'r'i'z'$5 and stellar mass tables with per-method values, consensus metrics, confidence intervals (Stefanon et al., 2017)

Data products are publicly available at the MAST CANDELS portal, Vizier, and the CANDELS project website. Full documentation and README files accompany all releases, ensuring reproducibility and transparency (Stefanon et al., 2017, Skelton et al., 2014).

7. Validation, Completeness, and Scientific Impact

The catalogs have been subjected to rigorous validation:

• Completeness: Empirical tests using deeper data (e.g., HUDF) confirm 90% completeness at $u^* g'r'i'z'$6–$u^* g'r'i'z'$7 AB, field-dependent (Skelton et al., 2014).
• Number counts: Consistent galaxy surface densities ($u^* g'r'i'z'$880,000 deg$u^* g'r'i'z'$9 to $$\mathrm{completeness}(m) = 1 - \frac12 \mathrm{erfc}\left(\frac{m - m_\mathrm{lim}}{\sqrt{2}\,\sigma}\right)$$0 mag) across fields, within cosmic variance.
• Photometric accuracy: Total-flux comparisons with direct aperture photometry and GALFIT yield <0.05 mag median offsets.
• Photometric error reliability: Empty-aperture errors track the SED fit residuals to within 20–50%, subject to marginal underestimation.
• Physical parameter validation: Distribution of photometric redshifts and $$\mathrm{completeness}(m) = 1 - \frac12 \mathrm{erfc}\left(\frac{m - m_\mathrm{lim}}{\sqrt{2}\,\sigma}\right)$$1 reproduces known large-scale structures and expected evolutionary trends; rest-frame UVJ diagrams show clean quiescent/star-forming sequences out to $$\mathrm{completeness}(m) = 1 - \frac12 \mathrm{erfc}\left(\frac{m - m_\mathrm{lim}}{\sqrt{2}\,\sigma}\right)$$2 (Skelton et al., 2014).

The 3D-HST WFC3-selected catalogs, integrating homogeneous PSF-matched photometry, robust physical characterization, and set within the legacy CANDELS/3D-HST fields, are a cornerstone resource for extragalactic astrophysics, supporting grism spectroscopic studies, galaxy evolution analyses, and multiwavelength population characterizations (Stefanon et al., 2017, Skelton et al., 2014).