UNIONS-VLASS Catalog Overview

• The UNIONS–VLASS Catalog is a comprehensive dataset combining optical UNIONS and VLASS radio imaging to identify 146,212 radio galaxies over ~4,200 deg².
• It employs a 1.5'' positional cross-match yielding ~92.7% completeness and ~7–10% false associations, balancing clean one-to-one matches.
• Incorporating multifrequency radio data and photometric redshifts, it enables detailed studies of radio-loud AGN, high-redshift galaxies, and jet evolution.

The UNIONS–VLASS Catalog is the optical identification catalog produced by cross-matching the Ultraviolet Near Infrared Optical Northern Survey (UNIONS) with the Very Large Array Sky Survey (VLASS) Epoch 2 catalog, and is presented by the UNVEIL project as a catalog of approximately 146,000 radio galaxies extending to redshifts of approximately $z\sim 5$ (Zhong et al., 11 Jul 2025). It combines wide-area optical imaging, radio continuum selection at $3$ GHz, photometric redshifts, radio loudness estimates, and ancillary cross-matches to FIRST and LoTSS. In the published implementation, the cross-match yields 146,212 radio galaxies down to $r=24.5$ mag over an area of approximately $4{,}200\ \mathrm{deg^2}$, with explicit utility for studies of radio-loud AGN, high-redshift radio galaxies, radio spectral populations, and forthcoming Euclid wide-field analyses (Zhong et al., 11 Jul 2025).

1. Survey basis and observational parameter space

The catalog is built from the overlap between UNIONS optical imaging and VLASS radio imaging. UNIONS is described as covering declination $>-40^\circ$ over both Northern and Southern Galactic Caps, ultimately reaching approximately $5{,}000\ \mathrm{deg^2}$, while the internal “UNIONS5000” release used for the cross-match covers approximately $4{,}200\ \mathrm{deg^2}$. The optical data comprise complementary $u$, $g$, $r$, $3$0, and $3$1 imaging from CFIS, WHIGS, Pan-STARRS, and WISHES, with median $3$2 depths in $3$3 apertures of approximately $3$4 mag for $3$5, respectively. The $3$6-band reaches $3$7 at $3$8 or $3$9 at $r=24.5$0, and the median image seeing in $r=24.5$1 is approximately $r=24.5$2 (Zhong et al., 11 Jul 2025).

The radio input catalog is VLASS Epoch 2 quick-look imaging at central frequency $r=24.5$3 GHz with bandwidth $r=24.5$4–$r=24.5$5 GHz and survey area approximately $r=24.5$6 for declination $r=24.5$7. The angular resolution is approximately $r=24.5$8, the single-epoch rms is approximately $r=24.5$9 at $4{,}200\ \mathrm{deg^2}$0, and the VLASS2 “clean” catalog after astrometric and quality cuts contains $4{,}200\ \mathrm{deg^2}$1 sources in the overlap region (Zhong et al., 11 Jul 2025).

The resulting parameter space is defined by a relatively deep optical survey matched to a high-resolution, all-sky radio survey. A plausible implication is that this combination is particularly effective for identifying compact radio-selected counterparts with usable optical photometry, while also leaving some incompleteness for the most extended radio morphologies, as discussed in the caveats below.

2. Cross-matching methodology and catalog construction

The published cross-match uses a positional criterion

$4{,}200\ \mathrm{deg^2}$2

chosen to balance completeness and false associations in the context of a median UNIONS galaxy density of approximately $4{,}200\ \mathrm{deg^2}$3 (Zhong et al., 11 Jul 2025). At this threshold, the completeness is reported as approximately $4{,}200\ \mathrm{deg^2}$4, while false associations are approximately $4{,}200\ \mathrm{deg^2}$5. The estimated false-match probability is described more generally as approximately $4{,}200\ \mathrm{deg^2}$6–$4{,}200\ \mathrm{deg^2}$7 at $4{,}200\ \mathrm{deg^2}$8, and the catalog “cleanness” versus “completeness” curves intersect near $4{,}200\ \mathrm{deg^2}$9 for the median UNIONS source density (Zhong et al., 11 Jul 2025).

Before matching, the VLASS2 catalog is pre-filtered using the following criteria: $>-40^\circ$0; $>-40^\circ$1; $>-40^\circ$2; $>-40^\circ$3; sidelobe probability $>-40^\circ$4; and nearest-neighbor distance $>-40^\circ$5 to avoid complex multi-component sources (Zhong et al., 11 Jul 2025). These cuts define a clean sample optimized for one-to-one positional associations between optical and radio detections.

The final matched sample contains 146,212 UNIONS–VLASS sources, described as “radio galaxies,” corresponding to a number density of approximately $>-40^\circ$6 over approximately $>-40^\circ$7 (Zhong et al., 11 Jul 2025). The $>-40^\circ$8-band magnitude distribution peaks at $>-40^\circ$9–$5{,}000\ \mathrm{deg^2}$0 mag, and the UNIONS $5{,}000\ \mathrm{deg^2}$1 completeness limit of approximately $5{,}000\ \mathrm{deg^2}$2 mag yields a matched fraction $5{,}000\ \mathrm{deg^2}$3. Source morphology is quantified through $5{,}000\ \mathrm{deg^2}$4 versus signal-to-noise envelopes, giving approximately $5{,}000\ \mathrm{deg^2}$5 extended sources with solid angles larger than the beam (Zhong et al., 11 Jul 2025).

A common misconception would be to treat the catalog as a morphology-complete census of all VLASS radio galaxies in the overlap area. That interpretation is not supported by the selection function: the $5{,}000\ \mathrm{deg^2}$6 filter explicitly removes complex multi-component systems, especially bright extended radio galaxies with large hotspot separations (Zhong et al., 11 Jul 2025).

3. Redshift information and high-redshift populations

Photometric redshifts are available for 63,019 matched sources, using validity flags based on 5-band $5{,}000\ \mathrm{deg^2}$7 or 4-band $5{,}000\ \mathrm{deg^2}$8 fits (Zhong et al., 11 Jul 2025). Within that subset, 8,692 sources satisfy $5{,}000\ \mathrm{deg^2}$9, and 1,171 satisfy $4{,}200\ \mathrm{deg^2}$0. In addition, a $4{,}200\ \mathrm{deg^2}$1-dropout selection designed to identify sources at approximately $4{,}200\ \mathrm{deg^2}$2–$4{,}200\ \mathrm{deg^2}$3 produces 477 candidates after $4{,}200\ \mathrm{deg^2}$4-band non-detection and $4{,}200\ \mathrm{deg^2}$5, implying an optimistic sample of more than 200 sources at $4{,}200\ \mathrm{deg^2}$6 (Zhong et al., 11 Jul 2025).

Spectroscopic redshifts are available for approximately 2,175 sources via SDSS DR16. For photometric-redshift performance, the catalog reports an overall bias of approximately $4{,}200\ \mathrm{deg^2}$7 and $4{,}200\ \mathrm{deg^2}$8 for $4{,}200\ \mathrm{deg^2}$9 in UNIONS5000 $u$0. For radio-selected QSOs, the scatter is substantially worse, with $u$1 and an outlier fraction of approximately $u$2 at $u$3. The abstract-level summary further states a photometric accuracy with $u$4 bias approximately $u$5 for $u$6, $u$7, and catastrophic outliers of approximately $u$8 at $u$9 (Zhong et al., 11 Jul 2025).

These numbers set the interpretive regime for the catalog’s high-redshift content. The low- to intermediate-redshift photometric-redshift performance is adequate for broad population statistics, but the high-redshift AGN-dominated regime is explicitly uncertain. The paper also notes a potential underestimation of $g$0 for AGN-dominated SEDs because of the lack of AGN templates (Zhong et al., 11 Jul 2025). This suggests that the catalog’s $g$1 and especially $g$2 populations are valuable as candidate samples, but should not be treated as spectroscopically finalized without follow-up.

4. Radio luminosity, radio loudness, and AGN classification

The catalog derives rest-frame radio luminosities using

$g$3

where $g$4 is the luminosity distance, $g$5 is the observed integrated flux density, and $g$6 is the spectral index defined through $g$7 (Zhong et al., 11 Jul 2025). The spectral index is estimated between 3 GHz and 1.4 GHz when FIRST is available, otherwise $g$8 is assumed; analogous estimates are used between 144 MHz and 3 GHz or 1.4 GHz (Zhong et al., 11 Jul 2025). The luminosity threshold adopted for radio-loud AGN is the commonly used criterion

$g$9

Using valid photometric redshifts, the catalog identifies approximately 49,000 radio-loud AGNs by the luminosity criterion $r$0, and all radio galaxies at $r$1 are reported as radio-loud under that definition (Zhong et al., 11 Jul 2025). The catalog also provides an observed radio-loudness measure

$r$2

with $r$3 derived from the $r$4-band flux density without $r$5-correction. An equivalent expression is

$r$6

and sources are classified as radio-quiet for $r$7 and radio-loud for $r$8 (Zhong et al., 11 Jul 2025).

Under the observed radio-loudness definition, 138,266 of the 146,212 clean $r$9-detected sources are radio-loud AGNs, corresponding to $3$00, while 7,946, or $3$01, are radio-quiet AGNs (Zhong et al., 11 Jul 2025). The consistency between the observed radio-loudness criterion and the luminosity criterion is asymmetric: approximately $3$02 overlap from the luminosity-based sample, but approximately $3$03 from the $3$04-based sample (Zhong et al., 11 Jul 2025).

This dual definition is important because the catalog is not using a single operational notion of “radio-loud.” The luminosity threshold isolates intrinsically powerful radio emitters among sources with valid redshifts, whereas $3$05 can be applied more broadly but is explicitly based on observed quantities without $3$06-correction. A plausible implication is that the catalog is optimized both for intrinsic radio-power studies and for practical wide-area selection in the absence of complete redshift coverage.

5. Multifrequency radio characterization

The catalog is extended through cross-matching with FIRST at $3$07 GHz and LoTSS DR2 at $3$08 MHz. The FIRST cross-match yields 79,638 UNIONS–VLASS sources within $3$09, and after sidelobe filtering there are approximately 76,000 clean matches. The LoTSS DR2 cross-match yields 101,671 matches, and the triple-matched VLASS–FIRST–LoTSS sample contains 64,672 sources (Zhong et al., 11 Jul 2025).

For 30,091 sources with all three radio bands, the paper applies a spectral-index “color-color” classification. Quadrant 1 is designated upturned, with excess at low frequency suggesting transients or recent outbursts. Quadrant 2 is inverted, with $3$10 and synchrotron self-absorption dominance. Quadrant 3 is a power-law population with $3$11, described as typical synchrotron emission. Quadrant 4 is peaked or GPS, showing a turnover at GHz frequencies and associated with young CSS sources (Zhong et al., 11 Jul 2025).

Projected linear sizes are estimated from VLASS deconvolved FWHM. Within this framework, ultra-steep-spectrum sources are defined by $3$12 and are described as evolved sources with aged jets. Compact-steep-spectrum sources are defined by $3$13 and projected size $3$14 kpc, corresponding to young jets (Zhong et al., 11 Jul 2025). The abstract-level summary states that the multifrequency radio dataset reveals sources with steep spectra characteristic of aged populations, peaked spectra characteristic of young populations, and upturned spectra potentially associated with transient sources (Zhong et al., 11 Jul 2025).

This multifrequency structure makes the catalog more than a positional cross-identification list. It functions as a spectral-population resource spanning aged, young, and potentially transient radio-source classes. A plausible implication is that it is suitable for work that requires distinguishing long-lived jet populations from compact young objects and flare-like radio phenomena without requiring immediate bespoke cross-survey assembly.

6. Catalog format, limitations, and scientific role

The released catalog contains approximately 146,212 rows and includes UNIONS positions ($3$15, $3$16), $3$17 magnitudes and errors, photometric-redshift quantities such as $3$18 and flags, VLASS component names, $3$19, $3$20, flux errors, beam information, astrometric errors, sidelobe probability, and $3$21 (Zhong et al., 11 Jul 2025). It also includes FIRST and LoTSS match separations, FIRST and LoTSS fluxes and errors, in-band radio magnitudes such as $3$22 and $3$23, spectral indices $3$24 and $3$25 with errors, rest-frame luminosities $3$26 and $3$27, observed radio loudness $3$28, and flags for extendedness, multi-optical matches, quality, and photometric-redshift reliability. Distribution is in electronic FITS and CSV tables via the UNIONS data release site and CDS (Zhong et al., 11 Jul 2025).

The paper identifies several caveats. In radio flux terms, VLASS has a $3$29 completeness of approximately $3$30 mJy, corresponding to $3$31 AB mag, while FIRST has a $3$32 completeness of approximately $3$33 mJy, corresponding to $3$34 AB mag (Zhong et al., 11 Jul 2025). In optical terms, the matched fraction declines from $3$35 at $3$36 to $3$37 at $3$38, which the paper attributes to Malmquist bias and CMB inverse-Compton suppression (Zhong et al., 11 Jul 2025). The VLASS quick-look catalog also suffers CLEAN-bias, which underestimates peak fluxes by approximately $3$39 mJy (Zhong et al., 11 Jul 2025). Most importantly for source demographics, the $3$40 cut biases the sample toward compact sources such as CSS and PS objects by removing bright extended radio galaxies, including systems with large FR II hotspot separations (Zhong et al., 11 Jul 2025).

Within those limitations, the scientific role of the catalog is explicitly framed in terms of Euclid synergy and high-redshift radio-source studies. The UNIONS $3$41 data are intended to provide optical counterparts and photometric redshifts for Euclid wide fields in VIS and NIR, to calibrate PSF chromaticity, and to supply photo-$3$42 values for radio-selected AGN (Zhong et al., 11 Jul 2025). The paper also notes a PFS filler program targeting approximately 3,000 UNIONS–VLASS radio galaxies for redshifts and spectroscopic diagnostics. Enumerated use-cases include statistics of high-redshift radio galaxies at $3$43 and $3$44 $3$45-dropouts, studies of RLAGN feedback and host-galaxy co-evolution, multifrequency SED fitting to discriminate jet- versus starburst-dominated radio emission, transient searches using upturned-spectrum and inverted sources, and large-scale-structure analyses using radio galaxies as tracers of massive halos at high redshift (Zhong et al., 11 Jul 2025).

Taken together, the catalog is presented as a foundational dataset for multiwavelength studies of AGN physics, jet evolution, and galaxy co-evolution up to $3$46 (Zhong et al., 11 Jul 2025). That conclusion follows directly from its combination of wide-area optical identification, radio-loudness characterization, partial redshift information, and multifrequency radio spectral typing, while the documented selection biases delimit the regimes in which statistical interpretation is most robust.