ASTRODEEP-JWST Photometric Catalogs

• ASTRODEEP-JWST photometric catalogs are homogeneous HST+JWST datasets offering 16-band coverage to study high-redshift galaxies across multiple extragalactic deep fields.
• They employ robust methods including F356W+F444W detection, PSF-matched aperture photometry, and four independent photometric redshift estimates validated against 16,666 spectroscopic sources.
• These catalogs serve as a versatile resource for faint-galaxy analysis, strong-lens discovery, and color-based selection of dusty star-forming galaxies.

Searching arXiv for the core ASTRODEEP-JWST catalog paper and closely related catalog/use papers. ASTRODEEP-JWST photometric catalogs are homogeneous HST+JWST source catalogs designed to provide a comprehensive database for the study of galaxy populations in the high-redshift Universe. In the main release, data from eight JWST NIRCam observational programs in six extragalactic deep fields are assembled into a uniform 16-band resource spanning 0.44 to 4.44 $\mu$m, with photometry and photometric redshifts for approximately half a million sources over a total area of $\sim 0.2$ square degrees (Merlin et al., 2024). The catalogs inherit methodological elements from earlier JWST field releases in Abell 2744 and are intended for faint-galaxy and high-$z$ work, but they have also been used for object classification, strong-lens discovery, cosmological surface-brightness tests, and color-selected dusty-galaxy studies (Paris et al., 2023).

1. Definition, scope, and survey composition

The ASTRODEEP-JWST release presented in 2024 gathers data from eight JWST NIRCam observational programs targeting Abell 2744, EGS, COSMOS, UDS, GOODS North, and GOODS South. The contributing programs are GLASS-JWST, UNCOVER, DDT2756, GO3990, CEERS, JADES, NGDEEP, and PRIMER. The fields listed in the release are ABELL2744, CEERS, JADES-GN, JADES-GS, NGDEEP, PRIMER-COSMOS, and PRIMER-UDS, with a combined coverage described as approximately $0.2$ square degrees rather than a simple sum because of overlaps (Merlin et al., 2024).

The catalog contains 531,173 detected sources. The per-field source counts reported are 42,491 in ABELL2744, 82,547 in CEERS, 58,385 in JADES-GN, 73,638 in JADES-GS, 14,752 in NGDEEP, 123,094 in PRIMER-COSMOS, and 136,266 in PRIMER-UDS (Merlin et al., 2024).

A fixed set of 16 bands was adopted for uniform catalog production and subsequent photo-$z$ estimation. These are HST/ACS F435W, F606W, F775W, F814W; HST/WFC3 F105W, F125W, F140W, F160W; and JWST/NIRCam F090W, F115W, F150W, F200W, F277W, F356W, F410M, F444W. Band availability varies by field, but the release standardizes the photometric framework across the compilation (Merlin et al., 2024).

This multi-field compilation sits within a broader sequence of JWST catalog development. Earlier Abell 2744 releases had already established F444W-based, PSF-matched, faint-galaxy-oriented catalogs in a cluster environment, first for GLASS-JWST and then for the combined GLASS+UNCOVER+DDT stage-I release (Merlin et al., 2022); (Paris et al., 2023). A plausible implication is that ASTRODEEP-JWST generalizes those field-tested procedures from individual survey footprints to a cross-survey, cross-field cataloging framework.

2. Imaging basis, detection strategy, and photometric extraction

The catalogs are built from JWST NIRCam mosaics produced primarily by the program teams, with HST images re-projected onto the NIRCam grid for astrometric and photometric consistency. All data are scaled to $\mu$Jy units with AB zero-point 23.9, and astrometric alignment is refined to ensure sub-pixel registration (Merlin et al., 2024).

Source detection is performed with SExtractor v2.8.6 on a weighted stack of NIRCam F356W+F444W mosaics, a choice explicitly optimized for high-redshift source detection. This long-wavelength detection strategy is continuous with earlier Abell 2744 work: the 2022 GLASS-JWST extragalactic catalog also adopted F444W-based detection for faint high-$z$ galaxies, while the 2023 Abell 2744 stage-I release used an F444W-detected multi-band catalog including all NIRCam and available HST data (Merlin et al., 2024); (Merlin et al., 2022); (Paris et al., 2023).

Photometry is obtained through aperture measurements on PSF-matched images. All bands are smoothed to the F444W resolution for consistency, except HST WFC3 bands, which are matched to the F160W resolution. Colors are measured in matched apertures and then corrected to total fluxes using Kron-like apertures derived from a-phot, consistently referenced to the F356W+F444W detection stack (Merlin et al., 2024). The total-flux prescription is written as

$$f_{\mathrm{tot,band}} = c_{\mathrm{ap,det}} \times f_{\mathrm{ap,band}},$$

with

$$c_{\mathrm{ap,det}} = f_{\mathrm{tot,det}} / f_{\mathrm{ap,det}}.$$

Photometry is provided for nine apertures, from diameters of $0.2''$ up to $\sim 0.2$0, and an “optimal” aperture is assigned per source based on segment area. Two versions of the catalogs are released, with and without local background subtraction, and the local-subtraction version is recommended (Merlin et al., 2024). Empirical “über-PSFs” are constructed for each band by stacking isolated high-S/N stars across all fields, and growth curves are used to validate cross-band consistency (Merlin et al., 2024).

The methodological genealogy is visible in the precursor releases. The 2022 GLASS-JWST catalog used forced aperture photometry on PSF-matched images at the position of F444W-detected sources, with total-flux rescaling of matched-aperture measurements (Merlin et al., 2022). The 2023 Abell 2744 stage-I release extended this to a larger combined footprint and released five aperture-based catalog variants (Paris et al., 2023). This suggests that ASTRODEEP-JWST emphasizes procedural continuity and cross-field reproducibility over field-specific optimization.

3. Redshift estimation, validation, and quality control

ASTRODEEP-JWST provides four independent photometric redshift estimates per source. The codes used are zphot and EAzY, with three EAzY template sets: eazy_v1.3, LyaReduced, and Lya from Larson et al. 2023. Validation excludes AGN candidates identified through the literature and X-ray data (Merlin et al., 2024).

The spectroscopic benchmark comprises approximately 16,666 robust spectroscopic redshifts, drawn from NIRSpec, CANDELSz7, VANDELS, and several ground-based programs. Outliers are defined through

$\sim 0.2$1

and the reported global performance, using the median of three photo-$\sim 0.2$2 runs, is NMAD = 0.031, $\sim 0.2$3 for non-outliers, and outlier fraction = 6.3\% (Merlin et al., 2024). The normalized median absolute deviation is given as

$\sim 0.2$4

Quality control extends beyond redshift validation. Flags are assigned for missing coverage, contamination, blending, saturation, proximity to image border, point-like or stellar appearance, and cases where all HST fluxes are negative. Principal component analysis (PCA) is used for star/galaxy separation and for flagging spurious detections based on SExtractor morphological indicators, and sources with $\sim 0.2$5 are visually inspected to remove obvious spurious high-redshift candidates (Merlin et al., 2024).

A recurring misconception is that ASTRODEEP-JWST is primarily a photo-$\sim 0.2$6 product. In fact, the release is fundamentally a multiband photometric catalog with photo-$\sim 0.2$7 layers added on top. The inclusion of multiple aperture measurements, morphological diagnostics, flagging, and direct flux products indicates that the catalogs are intended to support custom downstream inference, not only the published redshift solutions (Merlin et al., 2024).

4. Catalog structure, released products, and public access

For each detected source, the released tables include a unique identifier, RA and Dec, pixel coordinates, SExtractor segmentation parameters, morphological information, total fluxes and errors in the 16 bands, a combined diagnostic quality or flag field, the “optimal” aperture size in the optimal-aperture catalogs, and four photometric redshift estimates plus redshift flags. The products are distributed as FITS tables with accompanying README material (Merlin et al., 2024).

The public access points are the Astrodeep website, the Rainbow Database, and CDS, with appendices and supplementary methodological material available on Zenodo (Merlin et al., 2024). The release is explicitly positioned as a community resource.

The structure of the main release should be distinguished from more specialized companion products in related surveys. In UNCOVER, for example, the public release of stellar population catalogs includes posterior moments, maximum-likelihood spectra, star-formation histories, and full posterior distributions obtained with Prospector and on-the-fly lensing magnification, whereas ASTRODEEP-JWST itself is centered on homogeneous multiband photometry and template-based photo-$\sim 0.2$8 estimation rather than joint stellar-population inference (Wang et al., 2023). That difference is substantive: ASTRODEEP-JWST is a survey-spanning photometric infrastructure, whereas UNCOVER’s stellar-population release is a field-specific inference product.

A useful comparison is provided by JADES. The 2023 JADES initial release supplied 23-band space-based photometric catalogs and EAZY-based photometric redshifts for approximately 47,500 sources, while the 2026 JADES DR5 catalogs expanded to 35 space-based imaging mosaics and introduced custom detection, deblending, circular and Kron photometry, curve-of-growth measurements, and a new pixel-level regression framework for uncertainty modeling (Rieke et al., 2023); (Robertson et al., 22 Jan 2026). ASTRODEEP-JWST therefore occupies an intermediate position: broader in field coverage than single-field releases, but less expansive in per-source band count and uncertainty modeling than the later JADES DR5 product.

5. Scientific applications and downstream use

The catalogs have been used as the basis for several distinct scientific workflows. In the Tolman-test study, 6,860 galaxies with reliably determined spectroscopic redshifts were selected from the ASTRODEEP-JWST photometric catalogue. That analysis used the catalog’s 16-filter photometry, aperture areas, and quality filtering to compute mean surface brightness through

$\sim 0.2$9

where $z$0 is the flux and $z$1 is the aperture area. The study reports that the mean surface brightness decreases with increasing distance, but with a trend departing from the standard expectation $z$2, finding an empirical exponent $z$3 in

$z$4

(Tsymbal et al., 30 Apr 2026). The catalog’s utility here lies in homogeneous photometry, supplied aperture information, and the availability of secure spectroscopic subsamples.

In strong-lens searches, the ASTRODEEP-JWST catalog was paired with AnomalyMatch and ESA Datalabs. After matching with available calibrated Level 3 data, about 470,732 ASTRODEEP sources were searched, yielding 58 unique gravitational lenses, of which 37 were previously uncatalogued and graded into 16 Grade A, 16 Grade B, and 26 Grade C systems (Dima et al., 5 May 2026). In this use case, the catalog provides source locations, photometric redshifts, and a uniform parent sample across multiple deep fields.

The catalogs have also served as a basis for color-selected populations. In GOODS-South, a study of dusty star-forming galaxies at $z$5 used ASTRODEEP-JWST sources with photometric redshifts in the range $z$6 and applied the NIRCam criteria

$z$7

obtaining 866 DSFGs from ASTRODEEP (Mitra et al., 10 Jun 2025). This illustrates a common pattern: the catalog provides the photometric substrate, while science-specific physical interpretation is imposed later through external selection functions or SED-fitting frameworks.

A plausible implication is that ASTRODEEP-JWST functions less as a single-purpose legacy catalog than as an interoperable source layer for multiple downstream pipelines, including cosmological tests, rare-object searches, and population-based selections.

6. Methodological context, comparisons, and limitations

ASTRODEEP-JWST should be understood relative to adjacent JWST catalog traditions. The Abell 2744 GLASS and stage-I releases established the use of augmented JWST pipeline reductions, custom mitigation of instrumental artifacts such as $z$8 noise, PSF matching to F444W, and F444W-centered detection for faint galaxies (Merlin et al., 2022); (Paris et al., 2023). UNCOVER adopted a different, cluster-optimized strategy with a noise-equalized F277W+F356W+F444W detection image, explicit bright-cluster-galaxy and intra-cluster-light subtraction, super-catalog aperture assignment, and per-source magnification estimates (Weaver et al., 2023). JADES, by contrast, developed custom signal-to-noise-based detection and deblending and, in DR5, a 35-band catalog with NIRCam, MIRI, ACS, and WFC3 coverage (Robertson et al., 22 Jan 2026).

These comparisons clarify both the strengths and the limits of ASTRODEEP-JWST. Its principal strength is homogeneity across multiple major extragalactic deep fields at fixed 16-band coverage, using consistent photometric and photo-$z$9 procedures (Merlin et al., 2024). Its main limitations are likewise explicit in the release: the catalogs are optimized for faint high-redshift galaxy studies; completeness and purity at the faint end reflect that priority; cluster-core regions in ABELL2744 require caution because of possible photometric contamination from residual intra-cluster light; and photometric redshifts are less reliable for low-S/N sources, especially for faint or extreme-redshift candidates (Merlin et al., 2024).

Another important contextual point emerges from later multiwavelength work. The ULTIMATE-deblending project argues that physical-property inference based solely on JWST/NIRCam and HST photometry can suffer systematic uncertainties because of limited wavelength coverage, and reports that adding deblended low-resolution ground-based photometry improves photometric-redshift accuracy by approximately 40% and reduces the outlier fraction by approximately 60% relative to HST+JWST-only photometry in the PRIMER fields (Sun et al., 5 Mar 2026). This does not invalidate ASTRODEEP-JWST; rather, it situates it as a high-quality HST+JWST baseline that can be extended when UV-to-MIR completeness is required.

A related misconception is that larger JWST-era catalogs automatically supersede ASTRODEEP-JWST in every use case. JADES DR5, for example, supersedes previous JADES releases within GOODS and introduces a richer photometric formalism (Robertson et al., 22 Jan 2026), but ASTRODEEP-JWST remains distinctive in compiling several major JWST programs and fields into one uniform catalog release (Merlin et al., 2024). The relevant comparison is therefore not simple replacement, but difference in scientific scope: field-specific depth and band richness versus cross-survey homogeneity.

7. Role in the JWST deep-field ecosystem

Within the JWST deep-field ecosystem, ASTRODEEP-JWST occupies the role of a broad, standardized extragalactic catalog layer spanning the principal early deep fields. Its design choices—F356W+F444W detection, PSF-matched aperture photometry, uniform 16-band coverage where available, multiple photo-$0.2$0 solutions, extensive validation against robust spectroscopy, and public release—make it especially suitable for comparative studies across surveys and for the construction of large parent samples (Merlin et al., 2024).

Its relationship to contemporaneous and subsequent efforts is complementary rather than redundant. UNCOVER extends the Abell 2744 program toward joint redshift and stellar-population posteriors under Prospector with on-the-fly lensing magnification (Wang et al., 2023). JADES defines an increasingly deep and methodologically specialized GOODS resource (Rieke et al., 2023); (Robertson et al., 22 Jan 2026). ULTIMATE-deblending expands toward UV-to-radio integration in PRIMER (Sun et al., 5 Mar 2026). ASTRODEEP-JWST, by contrast, is the principal homogeneous HST+NIRCam compilation across multiple flagship extragalactic fields.

For research practice, this means the catalogs are most naturally used as a validated, public entry point for large-sample selection, cross-field demographic work, and target definition for more specialized follow-up. That role is consistent with both the release philosophy and the demonstrated downstream uses in lens discovery, DSFG selection, and surface-brightness analyses (Dima et al., 5 May 2026); (Mitra et al., 10 Jun 2025); (Tsymbal et al., 30 Apr 2026).