Everything Every Band All at Once I: A Global Morphology Catalog in Abell 2744 based on UNCOVER/MegaScience

Abstract: We present spectrally-resolved structural parameter measurements of 29,608 sources from the legacy lensing field of Abell 2744, quantifying global structures from observed $0.7 μm - 4.8 μm$ and spanning rest-frame UV to NIR at $R\sim15$. These measurements are made on imaging mosaics mainly from the UNCOVER/MegaScience survey, including 20 JWST NIRCam broad and medium bands. We perform single-component Sérsic fitting to these galaxies using \texttt{pysersic}, a Bayesian structural fitting tool, to infer their structural parameters and associated random uncertainties from the posterior distributions. Through various quality evaluation criteria, we infer robust structural parameters among $&gt; 90\%$ of the selected $\rm SNR&gt;10$ sources. For each galaxy with reliable sizes in at least two bands and a high-quality redshift, we fit its observed size as a function of wavelength and infer rest-frame UV, optical, and near-infrared sizes where applicable. By performing injection-recovery tests on simulated galaxy cutouts in selected bands, we establish that our structural parameter measurements achieve fractional error $&lt; 10 -20\%$ above $\rm SNR&gt;10$. With this paper, all raw structural measurements and fitted rest-frame sizes are quality-flagged, cataloged, and released to the community. Finally, we demonstrate that this catalog enables the structural study of galaxies over an unprecedentedly wide parameter space of redshift ($0.3<z<8$), stellar mass ($\rm 10^{7}\, M_{\odot}&lt;M_{*} &lt;10^{11.5}\, M_{\odot}$), and rest-frame optical size ($\rm 100 \,pc&lt;R_{e}&lt;10\,kpc$), after correcting for lensing magnification.

• The paper introduces a detailed morphological catalog of 29,608 galaxies in the Abell 2744 cluster using JWST/NIRCam imaging, achieving 90% + success rates at SNR > 10 and demonstrating a quantitative error of 20% or less for fluxes, half-light radii, and Sérsic indices.
• Sérsic profile fitting was performed with the Bayesian tool pysersic, ensuring robust modeling of neighboring sources and complex posterior distributions.
• The catalog covers a wide range of stellar masses, redshifts, and physical sizes and has been calibrated and standardized across multiple wavelength bands to avoid lens-magnification distortions, aside from the most highly magnified sources.

This paper presents a large, publicly released morphological catalog of 29,608 sources in the strong-lensing field of Abell 2744, constructed from the full 20-band JWST/NIRCam imaging suite of the UNCOVER and MegaScience programs (2603.00380). The work delivers single-component Sérsic fits performed with the Bayesian tool pysersic, quantified systematic uncertainties through injection–recovery tests, and derives rest-frame sizes at UV (0.2 µm), optical (0.5 µm), and near-infrared (1 µm) wavelengths. Together with a companion analysis of the mass–size relation, it establishes a standardized structural dataset spanning $0.3$10^7\,M_\odot<M_*<10^{11.5}\,M_\odot$, and rest-optical sizes of $100\,\mathrm{pc}<R_e<10\,\mathrm{kpc}$ after lensing correction.

Survey data and fitting methodology

The catalog is built on NIRCam mosaics from UNCOVER (GO-2561) and MegaScience (GO-4111), augmented by MAGNIF, ALT, GLASS, and other programs targeting Abell 2744 at $z=0.307$. The mosaics cover roughly 30 arcmin² across all filters to depths of 28–30 mag AB, drizzled to 0.02″/pixel at short wavelengths and 0.04″/pixel longward of F250M. Bright cluster galaxies and intra-cluster light are modeled and subtracted, and empirical PSFs constructed from stars in each band are used throughout.

All sources with $\mathrm{SNR}>10$ in any band are fitted with single Sérsic profiles using pysersic, which forward-models the PSF and infers full posterior distributions rather than the point estimates typical of $\chi^2$-minimization tools such as GALFIT. Neighboring sources within 2″ that are no more than 2 mag fainter than the primary are simultaneously modeled; others are masked. Priors are anchored on photometric-catalog centroids, fluxes, and flux radii, with log-uniform priors on the half-light radius and uniform priors on Sérsic index ($0.65$–$8$) and ellipticity ($0$–$0.9$). Two inference modes are used: stochastic variational inference with a multivariate normal approximation (SVI-MVN) for all targets, supplemented by NUTS-based MCMC sampling for the $10^7\,M_\odot<M_*<10^{11.5}\,M_\odot$0 population, where posteriors may be non-Gaussian. MCMC costs roughly three times more per fit but captures complex posterior shapes; about 60% of the released measurements derive from MCMC.

Quality flagging and catalog structure

Each fit is evaluated against seven criteria covering masked-pixel fraction, consistency between MAP estimates and posterior percentiles, boundary-pushing failure modes, MCMC convergence diagnostics ($10^7\,M_\odot<M_*<10^{11.5}\,M_\odot$1, effective sample size thresholds), reduced chi-squared per pixel, centroid agreement with the photometric catalog, and flux consistency within 2 magnitudes. Fits are assigned one of three grades: great quality (use_fit = 2), robust quality (use_fit = 1, passing all criteria but with elevated chi-squared due to asymmetries such as spiral arms while still yielding meaningful sizes), or failed. Success rates among SNR > 10 targets range around 90% per band for SVI-MVN fits.

Four products are released: a RAW catalog (~1000 columns) preserving every attempted fit and its diagnostics; a SUPER catalog containing only great or robust fits, preferring MCMC results where available; a SUPER_DUPER catalog restricted to great-quality fits; and a Rest Frame Size catalog. Fluxes recovered by profile fitting agree with aperture photometry to within systematic differences below $10^7\,M_\odot<M_*<10^{11.5}\,M_\odot$2, but Sérsic models systematically report higher fluxes at fainter magnitudes, likely reflecting differing assumptions about surface-brightness profiles in source outskirts. All image-plane measurements deliberately remain uncorrected for lensing magnification so users can adopt alternative lens models; physical sizes in the Rest Frame Size catalog alone are de-lensed via a $10^7\,M_\odot<M_*<10^{11.5}\,M_\odot$3 prescription valid for moderate magnifications ($10^7\,M_\odot<M_*<10^{11.5}\,M_\odot$4).

Quantified uncertainties from injection–recovery tests

Mock Sérsic galaxies injected into empty sky cutouts in F200W, F250M, and F444W quantify both systematic bias and random-error calibration. The key findings are:

Quantity	Systematic offset (SNR > 10)	Recommended uncertainty inflation
Flux	< 2%	factor of ~2
Half-light radius	< 2%	factor of ~2
Sérsic index	< 2%	factor of ~1.5
Ellipticity	negligible	none (posterior well calibrated)

Fractional errors remain below 10–20% for most instances above SNR ≈ 10, degrading sharply below SNR ≈ 6, where over 60% of mock fits fail quality criteria. The recovery tests also expose a positive covariance among flux, half-light radius, and Sérsic index—a known degeneracy of the Sérsic parameterization in which jointly inflated radius and index preserve central surface brightness while raising outer-profile flux. Because mock galaxies are ideal Sérsic profiles convolved with the same PSFs used in fitting, these tests do not capture errors from PSF model inaccuracies or intrinsic non-Sérsic morphology; real-source uncertainties can therefore exceed these estimates.

Rest-frame sizes and color gradients

For galaxies with robust sizes in at least two bands, $10^7\,M_\odot<M_*<10^{11.5}\,M_\odot$5 is modeled as a quadratic function of $10^7\,M_\odot<M_*<10^{11.5}\,M_\odot$6 using numpyro, with an outlier mixture following Hogg et al. and additional error terms accounting for the demonstrated underestimation of radius uncertainties and imperfect PSF models. This parameterization yields rest-frame sizes while minimizing extrapolation: rest-UV sizes for $10^7\,M_\odot<M_*<10^{11.5}\,M_\odot$7, rest-optical for $10^7\,M_\odot<M_*<10^{11.5}\,M_\odot$8, and rest-NIR for $10^7\,M_\odot<M_*<10^{11.5}\,M_\odot$9. The derived size ratios across rest-frame wavelengths reveal wide diversity in color gradient strength, with median logarithmic gradients near zero—consistent with heterogeneous stellar-population and dust structures across the population. A caveat noted by the authors is that broad-band and medium-band sizes can differ systematically owing to depth differences, and redshift uncertainty can dominate rest-frame size errors for strongly lensed or low-redshift sources with strong color gradients.

A first look at the mass–size relation

Combining rest-optical sizes with Prospector-$100\,\mathrm{pc}<R_e<10\,\mathrm{kpc}$0 stellar masses and magnifications from the Abell 2744 lens model demonstrates the catalog's reach. At $100\,\mathrm{pc}<R_e<10\,\mathrm{kpc}$1, the star-forming mass–size distribution agrees closely with prior UNCOVER-based results that enforced smooth wavelength dependence of Sérsic parameters—an independent validation supporting that assumption. At $100\,\mathrm{pc}<R_e<10\,\mathrm{kpc}$2, massive-galaxy distributions match HST-derived relations, but dwarf galaxies ($100\,\mathrm{pc}<R_e<10\,\mathrm{kpc}$3) appear systematically smaller than predicted by HST relations, plausibly because HST's ~0.06″ resolution tripled JWST's led to overestimated sizes or missed the smallest systems. The sample also contains roughly 1500 quiescent dwarfs at $100\,\mathrm{pc}<R_e<10\,\mathrm{kpc}$4 that are Abell 2744 cluster members down to $100\,\mathrm{pc}<R_e<10\,\mathrm{kpc}$5, enabling environmental studies of quiescent-dwarf structural evolution. The authors caution that apparent steepening of the dwarf mass–size slope may be an artifact of plotting only median stellar masses given ~0.22 dex mass uncertainties correlated with size and redshift uncertainties; rigorous quantification is deferred to the companion paper (2603.00380).

Limitations and open questions

Several limitations bound the interpretation of this catalog. The SNR > 10 selection and quality cuts introduce incompleteness: net completeness exceeds 80% above $100\,\mathrm{pc}<R_e<10\,\mathrm{kpc}$6 at $100\,\mathrm{pc}<R_e<10\,\mathrm{kpc}$7, $100\,\mathrm{pc}<R_e<10\,\mathrm{kpc}$8 at $100\,\mathrm{pc}<R_e<10\,\mathrm{kpc}$9, and $z=0.307$0 at $z=0.307$1, but drops to ~60% for $z=0.307$2 at low redshift—half attributable to bright asymmetries failing the chi-squared cut and half to suspected implementation issues in pysersic that remain unresolved. Recovery tests do not address crowded-field performance, PSF model errors, or non-Sérsic morphologies. The de-lensing prescription is invalid for high-magnification sources flagged by mu_flag. Dwarf quiescent classifications at $z=0.307$3 suffer from poor rest-frame coverage below 0.35 µm and possible contamination by misclassified star-forming dwarfs or lower-redshift cluster members. Finally, whether the observed dwarf mass–size behavior reflects genuine evolution or selection effects remains an open question requiring the correlated-uncertainty modeling of the companion analysis.

Conclusion

This paper provides a comprehensive, quality-controlled, multiwavelength structural catalog for nearly 30,000 sources in Abell 2744, with demonstrated fractional errors below 10–20% at SNR > 10, carefully characterized uncertainty calibration, and rest-frame size measurements spanning eight billion years of cosmic history. Its principal value lies in extending reliable galaxy-size measurements into the low-mass, small-size regime enabled by deep lensing-cluster imaging with JWST, and in providing the community a flexible foundation—raw posteriors, quality flags, and wavelength-interpolated sizes—for studies of galaxy growth, quenching, and color-gradient evolution across an unusually broad parameter space.