ASTRODEEP-JWST Catalogue Overview

• ASTRODEEP-JWST Catalogue is a homogeneous photometric and redshift dataset that integrates 16 bands from JWST NIRCam and HST imaging across six extragalactic deep-field regions.
• It employs an aggressive detection strategy and PSF-matched photometry to maximize depth and completeness for faint, high-redshift sources, enabling robust statistical analyses.
• The catalogue underpins diverse scientific explorations—from galaxy size–redshift evolution to gravitational lens and extreme-redshift candidate searches—with a consistent multiwavelength framework.

The ASTRODEEP-JWST Catalogue is a homogeneous set of NIRCam–HST photometric and photometric-redshift catalogues built from eight JWST NIRCam observational programmes over six major extragalactic deep-field regions, complemented by HST archival imaging. It contains 531,173 detected sources over a total area of $\sim 0.2$ deg$^2$, provides photometry in 16 bands spanning $\sim 0.44$ to $4.44\,\mu\mathrm{m}$, and reports four independent photometric-redshift estimates per object. Its stated purpose is to provide a comprehensive, homogeneous database for studies of galaxy populations, especially in the high-redshift Universe (Merlin et al., 2024).

1. Survey definition and lineage

ASTRODEEP-JWST extends the catalogue philosophy developed in the pre-JWST ASTRODEEP programme, especially the Hubble Frontier Fields releases, which established a standardized multiwavelength pipeline, consistent detection and PSF-matching procedures, and a uniform data-product structure across fields (Bradac et al., 2019). In the JWST release, that philosophy is transferred from the HST+VLT+Spitzer regime to HST+JWST/NIRCam imaging (Merlin et al., 2024).

The catalogue is organized into seven field products corresponding to six sky regions, because PRIMER is split into COSMOS and UDS. The field coverage reported for the release is:

Field product	JWST programme(s)	Area (arcmin$^2$)
ABELL2744	GLASS-JWST, UNCOVER, DDT 2756, GO 3990	45.7
CEERS	CEERS	94.6
JADES-GN	JADES, FRESCO	83.0
JADES-GS	JADES, FRESCO	84.5
NGDEEP	NGDEEP	9.5
PRIMER-COSMOS	PRIMER	141.8
PRIMER-UDS	PRIMER	251.2

The fixed 16-band framework combines HST/ACS, HST/WFC3-IR, and JWST/NIRCam imaging. The HST bands are F435W, F606W, F775W, F814W, F105W, F125W, F140W, and F160W; the JWST bands are F090W, F115W, F150W, F200W, F277W, F356W, F410M, and F444W (Merlin et al., 2024). Not every band is available in every field, but the common filter list defines a uniform spectral-energy-distribution sampling strategy across the release.

2. Detection and photometric architecture

Source detection is performed only on JWST/NIRCam imaging. The catalogue uses a weighted stack of the F356W and F444W mosaics as the detection image, smoothed with a Gaussian filter of FWHM $=0.14''$, and processed with an aggressive SExtractor configuration intended to maximize completeness for faint sources. The release explicitly notes that this strategy prioritizes depth and completeness for faint objects and therefore tolerates additional spurious detections and complicated blends (Merlin et al., 2024).

All scientific photometry is measured on PSF-matched images with a-phot. Empirical PSFs are built by stacking bright isolated stars, and the release adopts cross-field “über-PSFs” rather than field-specific PSFs because they yield more stable growth curves and improved photometric-redshift performance. All JWST bands are convolved to the F444W resolution, while HST/WFC3 bands are convolved to F160W (Merlin et al., 2024). This enforces consistent aperture colors across the full HST+JWST set.

Fluxes are measured in fixed circular apertures with diameters 0.20, 0.28, 0.33, 0.50, 0.66, 0.70, 1.32, 2.65, and 5.30 arcsec. Total fluxes are obtained by scaling aperture colors to a Kron-like total flux measured on the F356W+F444W detection stack. The catalogue defines

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

with the errors propagated as

$e_{tot,band} = c_{ap,det}\times e_{ap,band}.$

An additional correction based on the F444W PSF curve of growth accounts for light beyond the Kron radius, which is reported to be large for JWST bands, typically at the 10–20% level (Merlin et al., 2024).

For practical use, the release provides an “optimal aperture” or optap catalogue. The chosen aperture is based on the detection-image isophotal area: an effective radius $r_{\rm eff}=\sqrt{\mathrm{ISOAREA\_IMAGE}/\pi}$ is computed, and the smallest aperture whose radius exceeds that effective radius is selected (Merlin et al., 2024). This creates one preferred total-flux estimate per source and band, while preserving the full family of fixed-aperture catalogues.

Noise handling is empirical rather than purely weight-map driven. RMS maps are rescaled by injecting fake PSF point sources into exposure-time regions and comparing flux dispersions with RMS-based uncertainties. Limiting magnitudes are then computed pixel by pixel from the scaled RMS map as

$$\mathrm{depth}_{5\sigma,i} = -2.5\log\left(\frac{5\times \sqrt{A}\times \mathrm{RMS}_i}{fr_{0.2}}\right)+\mathrm{ZP},$$

where $^2$0 is the 0.2-arcsec aperture area in pixels, $^2$1 is the encircled-energy fraction in 0.2 arcsec, and $^2$2 (Merlin et al., 2024).

3. Catalogue products, flags, and photometric-redshift layer

For each field, the release provides several photometric catalogues—one for each aperture size, plus the optap catalogue—and a corresponding photometric-redshift catalogue based on the optap photometry with local background subtraction (Merlin et al., 2024). Each photometric table contains source identifiers, sky coordinates, detection-image quantities, a-phot structural quantities, total fluxes and uncertainties in all available bands, and a diagnostic flag. The photo-$^2$3 tables append spectroscopic redshifts when available and list four independent photometric-redshift estimates.

The four redshift estimates are produced with zphot and three EAzY runs: one with the standard eazy_v1.3 templates and two with Larson et al. template configurations designed for extreme high-$^2$4 SEDs (Merlin et al., 2024). The release often uses a median photo-$^2$5 from the independent runs for global statistics, while keeping the individual estimates in the catalogue.

Photometric-redshift performance is assessed against a final spectroscopic calibration sample of 16,666 sources assembled from JWST/NIRSpec, HST grism, MUSE, and extensive ground-based spectroscopy. The quantity

$^2$6

is used to evaluate performance, with outliers defined by

$^2$7

For the global median photo-$^2$8, the release reports $^2$9, a standard deviation of $\sim 0.44$0, and an outlier fraction of $\sim 0.44$1 (Merlin et al., 2024). The catalogue also reports $\sim 0.44$2-derived percentiles, including z025, z160, z500, z840, and z975, which permit explicit propagation of redshift uncertainty.

The flag system is additive and encodes missing HST or JWST coverage, contamination by neighbours or bad pixels, blending, saturation, proximity to image borders, point-like classification, spurious detections, and a small number of field-specific pathologies such as all-negative HST fluxes in part of Abell 2744 (Merlin et al., 2024). The release explicitly does not provide stellar masses, SFRs, or other physical parameters; those are deferred to future work or to user-supplied SED fitting (Merlin et al., 2024). This makes the catalogue fundamentally a photometric and redshift layer rather than a full stellar-population product.

4. Scientific exploitation

The catalogue has already been used as a uniform observational basis for multiple classes of analysis. In size–redshift work, a sample of 6,860 galaxies with reliable spectroscopic redshifts and 319,771 galaxies with photometric redshifts was extracted from ASTRODEEP-JWST to study angular and linear size evolution out to $\sim 0.44$3 photometric redshift and $\sim 0.44$4 spectroscopic redshift (Raikov et al., 25 Jul 2025). In Tolman surface-brightness analyses, the same catalogue supplied 6,860 galaxies with reliable spectroscopic redshifts and multi-band aperture photometry, enabling rest-frame and observed-frame surface-brightness measurements over a very broad redshift range (Tsymbal et al., 30 Apr 2026). These applications rely on the catalogue’s homogeneity, common mosaic system, and consistent aperture definitions rather than on any single field.

The catalogue has also been used for extreme-redshift candidate searches. A dedicated $\sim 0.44$5–30 study applied customized Lyman-break selections to the ASTRODEEP-JWST multi-band catalogues of CEERS, Abell 2744, JADES, NGDEEP, and PRIMER over $\sim 0.44$6 sq. deg, identifying nine candidates at $\sim 0.44$7 and none for the $\sim 0.44$8 color selection. That work found that all candidates display multimodal redshift probability distributions across different SED-fitting codes, with alternative solutions corresponding to dusty or quiescent lower-redshift populations, and concluded that comparable-depth F150W and F200W imaging is critical for reducing contamination (Castellano et al., 8 Apr 2025).

In GOODS-South, the catalogue has been used as the observed JWST sample in dusty-galaxy studies. Starting from the ASTRODEEP-JWST GOODS-S catalogue of 73,638 sources and restricting to 55,028 objects with $\sim 0.44$9, a NIRCam color selection based on $4.44\,\mu\mathrm{m}$0 and $4.44\,\mu\mathrm{m}$1 produced 866 JWST-selected dusty star-forming galaxy candidates, whose stellar masses and SFRs were then compared with proto-spheroid simulations (Mitra et al., 10 Jun 2025).

ASTRODEEP-JWST has also functioned as the parent source list for machine-learning searches for rare objects. In a semi-supervised gravitational-lens search, the ASTRODEEP source list was combined with COSMOS-Web and trawled with AnomalyMatch, contributing to a final sample of 58 unique strong-lens candidates, of which 37 were previously uncatalogued (Dima et al., 5 May 2026). This demonstrates that the catalogue is not only a resource for classical SED-based science but also a substrate for morphology-driven searches across large JWST imaging sets.

5. Limitations and interpretive cautions

Several limitations are intrinsic to the release design. First, the detection configuration is deliberately optimized for faint, high-$4.44\,\mu\mathrm{m}$2 completeness, so the catalogue acknowledges that more spurious detections and complicated blends are accepted than in brighter-source-oriented catalogues (Merlin et al., 2024). This is particularly relevant for the highest-redshift candidates, where the absence of blue-band detections can make residual artifacts appear astrophysically plausible.

Second, photometric-redshift robustness degrades strongly at low signal-to-noise. For a random subset of 20,000 PRIMER-COSMOS sources, the release shows that the standard deviation among three photo-$4.44\,\mu\mathrm{m}$3 estimates increases significantly when the signal-to-noise ratio in red bands falls below $4.44\,\mu\mathrm{m}$4 and when blue-band signal-to-noise is below $4.44\,\mu\mathrm{m}$5 (Merlin et al., 2024). The visual vetting of the high-$4.44\,\mu\mathrm{m}$6 tail illustrates the same point: from 7,123 initial $4.44\,\mu\mathrm{m}$7 objects in the zphot runs, manual inspection and spurious-source removal left 3,068, but only 798 had a standard deviation $4.44\,\mu\mathrm{m}$8 among the three reported photo-$4.44\,\mu\mathrm{m}$9 values (Merlin et al., 2024). A common misconception is therefore that the raw $^2$0 counts can be treated as a statistically clean sample; the release explicitly advises against that interpretation.

Third, band coverage is not uniform. Some fields lack specific HST or JWST filters by design, and the flag system must be interpreted whenever the Lyman break or Balmer-break placement depends on a missing band (Merlin et al., 2024). In Abell 2744, an additional field-specific caveat applies: unlike the earlier ASTRODEEP Frontier Fields products, the JWST release did not yet implement the dedicated intra-cluster-light and bright-galaxy subtraction developed for the HFF-era cluster catalogues, so sources near the cluster core are more susceptible to residual ICL, over-subtraction artifacts, and local-background errors (Merlin et al., 2024, Bradac et al., 2019).

Finally, the catalogue is intentionally conservative in scope. It provides calibrated, homogeneous photometry and photo-$^2$1s, but not a release of derived stellar masses, SFRs, or SFHs. For many science cases that omission is advantageous, because it leaves modelling choices—IMF, SFH family, nebular prescription, dust law, priors—to the downstream analysis. For others, it means ASTRODEEP-JWST should be understood as a base layer rather than a complete inference product.

6. Related catalogues and subsequent developments

Subsequent JWST-era catalogues can be read as extensions or complements to the ASTRODEEP-JWST model rather than replacements for it. In GOODS-N and GOODS-S, the JADES DR5 stellar-population catalogue provides homogeneous Bayesian stellar-population inference for roughly 500,000 sources using Prospector, flexible non-parametric SFHs, nebular emission, dust attenuation, metallicities, and AGN/dust emission, and is explicitly described as a JWST-era analogue or successor to an ASTRODEEP-style catalogue for those fields (Duan et al., 20 May 2026).

In the PRIMER fields, ULTIMATE-deblending is presented explicitly as an ASTRODEEP-style JWST catalogue built on the same prior-based deblending philosophy, but extended to 50 bands from CFHT $^2$2 to JWST/MIRI F1800W across 627.1 arcmin$^2$3. Its main methodological result is that adding deblended low-resolution UV-to-MIR photometry improves photometric-redshift accuracy by $^2$4 and reduces the outlier fraction by $^2$5 relative to photometry using only HST and JWST bands (Sun et al., 5 Mar 2026). This suggests that ASTRODEEP-JWST is best viewed as a homogeneous, cross-field HST+JWST baseline to which deeper wavelength coverage or more elaborate Bayesian inference can be added.

Within that broader landscape, ASTRODEEP-JWST occupies a distinct position. It is the first large, homogeneous, public NIRCam–HST photometric release spanning the principal early JWST deep fields in a single framework; it preserves the ASTRODEEP emphasis on catalogue uniformity and explicit validation; and it has already served as a common observational substrate for studies ranging from high-$^2$6 galaxy selection to cosmological tests and rare-object searches (Merlin et al., 2024).