miniJPAS Data Release Overview

• miniJPAS is a public survey data set featuring 60-band imaging and photometric redshift catalogs, providing quasi-spectroscopic optical SED sampling.
• It employs rigorous PSF-corrected photometry and iterative calibration strategies to ensure homogenous data quality and precise redshift estimates.
• The release supports a multi-layer catalogue structure with value-added data for galaxy, star, and quasar classification as well as structure detection.

miniJPAS Data Release is the public survey product of miniJPAS, the $\sim 1\,\mathrm{deg}^2$ pathfinder of J-PAS in the AEGIS/EGS field, obtained with the JST/T250 at the Observatorio Astrofísico de Javalambre and designed to reproduce the J-PAS 56-filter strategy on a small area. It combines calibrated imaging, 60-band photometry, photometric-redshift catalogues, morphology and classification layers, and the identifiers needed to construct higher-level products such as stellar, quasar, galaxy-group, cluster, and spatially resolved galaxy catalogues. Because miniJPAS uses the same filter system, calibration philosophy, and much of the same analysis machinery as J-PAS, it serves simultaneously as a science data set and as a validation platform for future wide-area releases (Hernán-Caballero et al., 2021, Rodríguez-Martín, 2024, Pérez-Ràfols et al., 15 Jul 2025).

1. Survey definition and observational architecture

miniJPAS is a pathfinder survey for J-PAS and observes the AEGIS field in four pointings, with an effective area after masks of $\approx 0.895\,\mathrm{deg}^2$. The observations were obtained with the Pathfinder camera on the JST/T250, using a single $9.2k\times 9.2k$ CCD with pixel scale $0.23''\,\mathrm{pix}^{-1}$ and an effective field of view of $0.27\,\mathrm{deg}^2$ per pointing. Across the literature considered here, miniJPAS is consistently described as a proof-of-concept implementation of the J-PAS observing model on a much smaller footprint (Hernán-Caballero et al., 2021, Yuan et al., 2022, Rodríguez-Martín, 2024).

The survey uses the full J-PAS filter system: 54 narrow-band filters spanning roughly $3780$–$9100\,\mathrm{\AA}$, plus two broader edge filters, typically described as $u_{\rm JAVA}$ or J0348 on the blue side and J1007 on the red side. In addition, miniJPAS photometric catalogues include four SDSS-like broad bands. The result is a 60-band data model per source, with the narrow-band subset providing a quasi-spectroscopic sampling of the optical SED at an effective resolution $R\sim 60$. This “photo-spectrum” or “J-spectrum” concept is central to essentially all downstream miniJPAS analyses, from galaxy photo-$z$ estimation to quasar classification and spatially resolved stellar population work (Delgado et al., 2021, Hernán-Caballero et al., 2021).

Detection is performed in the $\approx 0.895\,\mathrm{deg}^2$0 band with SExtractor in dual mode, so the $\approx 0.895\,\mathrm{deg}^2$1-band image defines the source footprints and the same apertures are propagated to the remaining filters. This design makes the release intrinsically multi-band and aperture-consistent, which is why miniJPAS could immediately support template fitting, machine-learning classification, and pseudo-spectroscopic analyses without requiring band-by-band source matching (Baqui et al., 2020, Hernán-Caballero et al., 2021).

2. Public data products and catalogue structure

The starting point of the public release is an $\approx 0.895\,\mathrm{deg}^2$2-selected catalogue of 64,293 detected objects. The released products include coadded science images, PSF models, dual-mode source catalogues, multiple photometric measurements, per-band flags, and value-added tables derived from the same base photometry. Several papers explicitly refer to the miniJPAS Public Data Release and its archive endpoints, including $\approx 0.895\,\mathrm{deg}^2$3 and $\approx 0.895\,\mathrm{deg}^2$4 (Hernán-Caballero et al., 2021, Yuan et al., 2022).

The core source tables include multi-aperture magnitudes and fluxes. In the data model used by subsequent analyses, catalogues such as MagABDualObj and FLambdaSingleObj provide MAG_AUTO, MAG_PETRO, MAG_APER, and MAG_PSFCOR, together with shape parameters such as A_WORLD, B_WORLD, ellipticity, and CLASS_STAR, plus quality layers such as FLAGS and MASK_FLAGS. PSFEx models are available and are used in later work for PSF homogenization and aperture-consistent spatially resolved photometry (Rodríguez-Martín, 2024, Rodríguez-Martín et al., 15 Sep 2025).

The release is best understood as a layered system rather than a single table.

Layer	Representative products	Main contents
Core imaging/photometry	MagABDualObj, FLambdaSingleObj	60-band magnitudes and fluxes, apertures, shapes, flags
Photo-$\approx 0.895\,\mathrm{deg}^2$5	PhotoZLephare_updated	PHOTOZ, Z_ML, ODDS, confidence intervals
Value-added classification/structure	minijpas.StarGalClass, quasar and cluster catalogues	star/galaxy probabilities, quasar scores, memberships, mass proxies

This structure matters because most miniJPAS science papers do not remeasure the basic photometry from scratch. Instead, they join on source identifiers, quality flags, morphology, photo-$\approx 0.895\,\mathrm{deg}^2$6, and external value-added layers. The DR therefore functions as a stable relational substrate for higher-level catalogues rather than only as a set of images (Baqui et al., 2020, Rodríguez-Martín et al., 15 Sep 2025).

3. Calibration strategy and photometric-redshift framework

A defining feature of the miniJPAS Data Release is that the photo-$\approx 0.895\,\mathrm{deg}^2$7 products are not an ancillary add-on but one of the core scientific outputs. The catalogue described in "The miniJPAS survey: the photometric redshift catalogue" (Hernán-Caballero et al., 2021) is built from PSF-corrected forced-aperture photometry, with systematic photometric offsets removed through iterative fitting to stellar population synthesis models, and with photo-$\approx 0.895\,\mathrm{deg}^2$8 computed through jphotoz, a Python wrapper around a customised version of LePhare.

For colour work, miniJPAS uses PSF-corrected small-aperture photometry rather than total magnitudes. Systematic offsets are estimated by fitting model SEDs to spectroscopic galaxies and iteratively applying band-dependent magnitude shifts until the residual corrections fall below $\approx 0.895\,\mathrm{deg}^2$9 mag. The photo-$9.2k\times 9.2k$0 engine then samples redshift on a fine grid, $9.2k\times 9.2k$1 with $9.2k\times 9.2k$2, using a template set optimised for the J-PAS filter system and a prior derived from VVDS spectroscopy (Hernán-Caballero et al., 2021).

The release exposes both point estimates and PDF-derived summary statistics. The main published quantities are PHOTOZ (LePhare Z_BEST), Z_ML, Z_BEST68_LOW, Z_BEST68_HIGH, PHOTOZ_ERR, and ODDS. The odds parameter is defined as

$9.2k\times 9.2k$3

and the release literature argues that it is more informative than the formal $9.2k\times 9.2k$4-based uncertainty because the latter underestimates true photo-$9.2k\times 9.2k$5 errors, especially for high-S/N objects (Hernán-Caballero et al., 2021).

For the $9.2k\times 9.2k$6 galaxy population, miniJPAS yields about $9.2k\times 9.2k$7 galaxies per square degree with valid photo-$9.2k\times 9.2k$8 estimates, of which about $9.2k\times 9.2k$9 are expected to satisfy $0.23''\,\mathrm{pix}^{-1}$0. Without an odds cut, the inferred population-level performance is $0.23''\,\mathrm{pix}^{-1}$1 with outlier fraction $0.23''\,\mathrm{pix}^{-1}$2 for $0.23''\,\mathrm{pix}^{-1}$3. With $0.23''\,\mathrm{pix}^{-1}$4, the survey reaches $0.23''\,\mathrm{pix}^{-1}$5, $0.23''\,\mathrm{pix}^{-1}$6, and $0.23''\,\mathrm{pix}^{-1}$7, which is the regime used to motivate J-PAS BAO applications (Hernán-Caballero et al., 2021).

Beyond galaxy photo-$0.23''\,\mathrm{pix}^{-1}$8, later studies refine calibration for specialized uses. Stellar-locus analyses identify spatial residuals attributable to flat-fielding or illumination effects, with amplitudes up to a few percent in the blue filters and $0.23''\,\mathrm{pix}^{-1}$9 in the red filters, and show that a 2D polynomial correction can reduce the typical locus scatter to $0.27\,\mathrm{deg}^2$0 mag. This indicates that the public DR is scientifically usable as released, but also that mmag-level stellar work benefits from post-release recalibration layers (Yuan et al., 2022).

4. Value-added catalogues and the miniJPAS analysis ecosystem

The DR rapidly became a platform for value-added catalogues. One of the earliest layers is the machine-learning star-galaxy classifier of Baqui et al., released as minijpas.StarGalClass, with ert_prob_star and rf_prob_star as probabilistic outputs. On the full $0.27\,\mathrm{deg}^2$1 range, the best pure-photometric model reaches $0.27\,\mathrm{deg}^2$2, while the morphology-plus-photometry ERT model reaches $0.27\,\mathrm{deg}^2$3, outperforming traditional classifiers at faint magnitudes and supplying the morphological pre-classification used by later quasar work (Baqui et al., 2020).

The quasar programme built several catalogue layers on top of the DR. The most comprehensive release-level product is the combined quasar catalogue of the fifth paper in the series, which applies a stacking meta-classifier to miniJPAS photometry and publishes two catalogues: a point-like quasar catalogue with 784 entries and an all-sources catalogue with 11,487 entries, together with class confidences and redshift estimates. Earlier in the same series, the SQUEzE-based catalogue published 301 point-like candidates and 1,049 candidates when extended sources were also included, illustrating how the same underlying DR can support multiple classification paradigms and selection functions (Pérez-Ràfols et al., 15 Jul 2025, Pérez-Ràfols et al., 2023).

Cluster and group detection generated another major catalogue family. AMICO identifies 94 clusters in miniJPAS for $0.27\,\mathrm{deg}^2$4 at SNR $0.27\,\mathrm{deg}^2$5, while the PZWav analysis detects 574 candidates over the same redshift range and retains 221 after a richness-based cleaning. A cross-comparison with VT shows 43 common clusters with cluster-centre offsets of $0.27\,\mathrm{deg}^2$6 kpc and redshift differences below $0.27\,\mathrm{deg}^2$7. These catalogues include probabilistic memberships and optical mass tracers such as richness, optical luminosity, and stellar mass, extending miniJPAS from a photometric DR to a structure-finding testbed (Doubrawa et al., 2023).

A further layer is methodological rather than tabular. Py2DJPAS is a pipeline that interfaces directly with the DR by querying catalogues and downloading stamps, science images, PSF models, zero points, and photo-$0.27\,\mathrm{deg}^2$8 information, then producing PSF-homogenized regional photometry for resolved galaxies. Its existence shows that the miniJPAS DR was released with sufficient metadata, calibration products, and coordinate consistency to support automated IFU-like workflows (Rodríguez-Martín et al., 15 Sep 2025).

5. Scientific capabilities demonstrated by the release

The most direct demonstration of DR scientific scope is the galaxy-population study based on SED fitting of a complete flux-limited sample with $0.27\,\mathrm{deg}^2$9 and $3780$0. Several fitting codes recover consistent stellar masses, ages, extinctions, and colours from the 56-filter J-spectra, and for galaxies with $3780$1 the reported precision is $3780$2 mag in rest-frame $3780$3, $3780$4 dex in $3780$5, $3780$6 mag in $3780$7, and $3780$8 dex in mass-weighted age. The same study derives empirical stellar-mass completeness limits of $3780$9, $9100\,\mathrm{\AA}$0, and $9100\,\mathrm{\AA}$1 at $9100\,\mathrm{\AA}$2, $9100\,\mathrm{\AA}$3, and $9100\,\mathrm{\AA}$4, respectively (Delgado et al., 2021).

Stellar work shows a complementary use of the same release. In the 56 J-PAS filters plus 4 SDSS-like bands, VOSA fitting yields stellar effective temperatures with precision $9100\,\mathrm{\AA}$5 K, while metallicity-dependent stellar loci achieve a typical $9100\,\mathrm{\AA}$6 precision of $9100\,\mathrm{\AA}$7 dex. An XGBoost dwarf/giant classifier trained on miniJPAS photometry identifies J0520 and J0510 as the most informative bands, consistent with their sensitivity to the Mg triplet and surface gravity. This establishes the DR as a precision stellar-photometry data set rather than only a cosmological pilot (Yuan et al., 2022).

White-dwarf analyses exploit the same low-resolution photo-spectra at the hot, blue end of parameter space. For a blue point-like sample with $9100\,\mathrm{\AA}$8, Bayesian fits to miniJPAS photo-spectra yield $9100\,\mathrm{\AA}$9 uncertainty in $u_{\rm JAVA}$0, classify H-dominated versus He-dominated atmospheres with 99% confidence, and detect calcium absorption and polluting metals down to $u_{\rm JAVA}$1 for sources with $u_{\rm JAVA}$2 K. In the same framework, hot white dwarfs are cleanly separated from blue QSOs using optical photometry alone (López-Sanjuan et al., 2022).

Environmental and resolved-galaxy studies further broaden the release’s scope. Using AMICO groups, BaySeAGal SED fitting, and miniJPAS photometry, one analysis finds that the fraction of red and quiescent galaxies in groups is always higher than in the field and that the quenched fraction excess rises from a few percent to higher than 60% over $u_{\rm JAVA}$3 to $u_{\rm JAVA}$4, with fading time scales $u_{\rm JAVA}$5 Gyr (Delgado et al., 2022). Spatially resolved work on 51 galaxies, built directly on DR images and catalogues, shows that denser and redder regions are older, more metal rich, and have lower sSFR, while bluer and less dense regions show stronger emission lines and higher sSFR. Comparisons with MaNGA for galaxy 2470-10239 find excellent agreement in J-spectra and stellar-mass surface-density profiles within 1 HLR, while miniJPAS extends the analysis to 4 HLR at $u_{\rm JAVA}$6 (Rodríguez-Martín, 2024, Rodríguez-Martín et al., 15 Sep 2025).

The quasar programme demonstrates still another operating mode of the DR. The combined algorithm paper reports $u_{\rm JAVA}$7 for high-$u_{\rm JAVA}$8 quasars and $u_{\rm JAVA}$9 for low-$R\sim 60$0 quasars on balanced mocks down to $R\sim 60$1, while real-data validation against DESI EDR gives $R\sim 60$2 for high-$R\sim 60$3 and $R\sim 60$4 for low-$R\sim 60$5 point-like quasars at the same limit. The same work finds mock-based quasar redshift uncertainty $R\sim 60$6, with real matched quasars reaching $R\sim 60$7 in the available DESI cross-match (Pérez-Ràfols et al., 15 Jul 2025).

6. Limitations, caveats, and relation to full J-PAS

The central limitation of the miniJPAS Data Release is scale. Its $R\sim 60$8 footprint implies strong cosmic variance, a small number of massive structures, and limited spectroscopic overlap for some specialised tasks. Several studies therefore stress that miniJPAS is a pathfinder rather than a statistically complete substitute for J-PAS, and that validation samples can be too small for full redshift-, magnitude-, or SED-dependent characterization (Hernán-Caballero et al., 2021, Rodríguez-Martín, 2024, Pérez-Ràfols et al., 15 Jul 2025).

A second limitation is that formal uncertainty estimates are not always sufficient. In the photo-$R\sim 60$9 catalogue, $z$0-based intervals underestimate the true redshift uncertainty, and the release literature explicitly recommends the ODDS parameter as the better predictor of catastrophic-error probability. In stellar applications, sub-percent flat-fielding or illumination residuals survive at the level of a few percent in blue filters unless locus-based corrections are applied. In quasar work, classifier performance depends strongly on the realism of the mocks and on morphology assumptions; the all-sources quasar catalogue is explicitly described as being provided “at the user’s risk” because the classifier was trained only on point-like objects (Hernán-Caballero et al., 2021, Yuan et al., 2022, Pérez-Ràfols et al., 15 Jul 2025).

The release also has conservative calibration choices that are science-enabling but not yet ultimate. For example, the absolute zero-point uncertainty is sometimes carried as $z$1 mag in all bands in spatially resolved work, and several analyses note that future J-PAS releases should reduce such systematics as calibration and training sets improve (Rodríguez-Martín, 2024, Rodríguez-Martín et al., 15 Sep 2025). A plausible implication is that miniJPAS should be read less as a finished terminal product than as a stable public baseline on which progressively more specialised calibration and inference layers can be built.

Within that pathfinder role, however, the release is unusually complete. It already demonstrates the operational coupling of calibrated 56-filter photometry, PSF-aware apertures, photo-$z$2 PDFs, probabilistic classification, structure finding, and physically interpreted SED fitting. In that sense, miniJPAS Data Release is both the first public J-PAS-like survey product and the prototype of how future J-PAS data releases are expected to function scientifically and technically across galaxies, stars, quasars, and large-scale structure (Delgado et al., 2021, Hernán-Caballero et al., 2021, Doubrawa et al., 2023).