Pristine-Gaia Synthetic Catalogue (PGS)

• Pristine-Gaia Synthetic Catalogue is an all-sky photometric-metallicity resource that synthesizes CaHK measurements from Gaia XP spectra using the Pristine survey’s metallicity model.
• It infers metallicities for over 30 million FGK stars through refined calibration procedures, iterative extinction treatment, and robust quality cuts.
• PGS enhances target selection for Galactic archaeology by efficiently identifying very metal-poor stars and mapping chemical structures within the Milky Way.

Searching arXiv for the core Pristine-Gaia synthetic catalogue papers and closely related context. The Pristine-Gaia-Synthetic catalogue, commonly abbreviated PGS, is an all-sky photometric-metallicity catalogue derived by combining Gaia DR3 BP/RP spectro-photometry with the Pristine survey’s metallicity-sensitive CaHK methodology. Its defining operation is the synthesis of a Pristine-like narrow-band magnitude around the Ca II H and K lines from Gaia XP spectra, followed by inference of photometric metallicities for FGK stars through the Pristine metallicity model. In the Pristine DR1 framework, the synthetic product provides $CaHK_\mathrm{syn}$ for 219.2 million Gaia DR3 sources with BP/RP coefficients and, after quality cuts, an all-sky catalogue of 31,605,960 reliable photometric metallicities; a closely related follow-up paper characterizes the resulting resource as providing reliable metallicities for about 30 million FGK stars (Martin et al., 2023, Viswanathan et al., 2024).

1. Definition and survey context

PGS emerged from the convergence of two survey assets with complementary strengths. Gaia DR3 supplies all-sky low-resolution BP/RP spectro-photometry, while Pristine supplies a narrow-band CaHK framework centered on the calcium H and K lines near 3933.7 and 3968.5 Å, where metal-poor FGK stars exhibit diagnostically weaker absorption. The catalogue therefore functions as the “Gaia XP version” of the Pristine metallicity programme: Gaia spectra are converted into synthetic CaHK magnitudes, and those synthetic measurements are then mapped into $[\mathrm{Fe/H}]$ through the Pristine calibration (Martin et al., 2023, Viswanathan et al., 2024).

This construction gives PGS a specific position among Gaia-derived resources. It is all-sky, unlike the direct Pristine footprint of more than 6,500 deg$^2$, but it is also shallower, because Gaia XP spectra impose a brightness limit and the blue BP region is comparatively noisy. In practical terms, the catalogue is especially suited to the exploration of the metal-poor Milky Way, with the all-sky synthetic product being effective down to about $[\mathrm{Fe/H}] \sim -3.5$ and particularly useful in the regime $[\mathrm{Fe/H}]<-1.0$ (Martin et al., 2023).

A common misconception is that “synthetic” here denotes a Galaxy simulation or a forward model of the sky. In the PGS context, “synthetic” refers instead to synthetic CaHK photometry inferred from real Gaia DR3 XP spectra. The catalogue is therefore observationally anchored rather than a pre-launch or cosmological mock.

2. Construction from Gaia XP spectra and the Pristine metallicity model

The core observable in PGS is the synthetic narrow-band magnitude $CaHK_\mathrm{syn}$. It is computed by integrating Gaia DR3 BP/RP spectro-photometry through the transmission curve of the CFHT/MegaCam Pristine CaHK filter, using GaiaXPy on the BP/RP coefficient representation. These synthetic magnitudes are then combined with Gaia broadband photometry $G$, $G_{BP}$, and $G_{RP}$ to infer photometric metallicity (Martin et al., 2023).

The metallicity method is an updated Pristine algorithm tied to Gaia broadband photometry only, rather than SDSS as in earlier Pristine work. It is restricted to FGK stars in the color interval

$0.5 < (G_\mathrm{BP}-G_\mathrm{RP})_0 < 1.5,$

roughly corresponding to $[\mathrm{Fe/H}]$0 K. The model is built in color–color space, with $[\mathrm{Fe/H}]$1 acting as a temperature proxy and the CaHK-based color encoding the metallicity-sensitive Ca H and K absorption. The training set is drawn primarily from SDSS/SEGUE and APOGEE, supplemented with known very metal-poor stars from Boötes I, LAMOST, PASTEL, and Pristine follow-up targets. At the low-metallicity boundary, the model is anchored by synthetic spectra from MARCS/Turbospectrum, including a $[\mathrm{Fe/H}]$2 grid and a no-metals grid; cells near or above the no-metals line are assigned $[\mathrm{Fe/H}]$3, and the model does not return values below $[\mathrm{Fe/H}]$4 (Martin et al., 2023).

Extinction treatment is iterative. Because extinction coefficients depend on temperature, metallicity, and reddening, the catalogue iterates between extinction and metallicity, using synthetic-spectra-based extinction coefficients and starting from Schlegel et al. reddening values. The paper expresses the calibrated Pristine magnitude through

$[\mathrm{Fe/H}]$5

where $[\mathrm{Fe/H}]$6 is an image-level zero-point and $[\mathrm{Fe/H}]$7 is a run-specific field-of-view correction. This absolute recalibration is solved with PhotCalib over approximately 11,500 Pristine images from 85 MegaCam runs (Martin et al., 2023).

The catalogue also introduces a variability model to suppress spurious metallicities from variable stars and propagates uncertainties through 100 Monte Carlo realizations. For each star it stores the median and percentile bounds of the metallicity PDF as well as the fraction of Monte Carlo samples remaining inside the valid model grid.

3. Catalogue products, parameters, and quality control

The released data products are structurally layered rather than monolithic. The synthetic Gaia-only component underlies both calibration and science use, while a second branch uses the recalibrated deeper Pristine imaging where available.

Product	Content	Scale
Full-sky $[\mathrm{Fe/H}]$8 catalogue	Synthetic CaHK magnitudes and uncertainties for Gaia DR3 BP/RP sources	219.2 million sources
Pristine-$[\mathrm{Fe/H}]$9 metallicity catalogue	All-sky Gaia-only photometric metallicities	31,605,960 reliable metallicities
Pristine DR1 metallicity catalogue	Recalibrated Pristine CaHK plus Gaia colours	3,783,243 reliable metallicities

For the overlap between Pristine and Gaia DR3 BP/RP, the recalibrated DR1 product contains 6,311,676 stars. The merged tables provide Gaia identifiers and positions, synthetic CaHK fluxes and uncertainties, de-reddened $^2$0, $^2$1, $^2$2, variability probability $^2$3, Monte Carlo confidence bounds on metallicity, $^2$4, and Gaia quality indicators including RUWE and corrected flux excess $^2$5 (Martin et al., 2023).

In later usage papers the synthetic metallicity field is explicitly denoted $^2$6 or operationally as FeH_CaHKsyn, with uncertainty information carried by FeH_CaHKsyn_84th, FeH_CaHKsyn_16th, and mcfrac_CaHKsyn. Quality-filtered science samples typically impose conservative conditions such as $^2$7, $^2$8 dex or photometric metallicity uncertainty $^2$9 dex, RUWE $[\mathrm{Fe/H}] \sim -3.5$0, $[\mathrm{Fe/H}] \sim -3.5$1, $[\mathrm{Fe/H}] \sim -3.5$2, and Monte Carlo grid fraction $[\mathrm{Fe/H}] \sim -3.5$3 (Viswanathan et al., 2024, Kuzma et al., 8 Jul 2025, Viswanathan et al., 2024).

The distinction between the all-sky synthetic branch and the direct Pristine branch is methodologically important. PDR1 is deeper and typically more precise because it uses actual CaHK imaging from CFHT/MegaCam; PGS is all-sky but more exposed to magnitude-dependent and metallicity-dependent selection effects inherited from Gaia XP availability and CaHK uncertainty cuts (Viswanathan et al., 2024).

4. Precision, validation, and known failure modes

Validation proceeds at two levels: internal comparison to spectroscopic training and external confirmation through dedicated follow-up. In the Pristine DR1 release paper, the photometric metallicities show good agreement with spectroscopy for training-sample giants. At the bright end, $[\mathrm{Fe/H}] \sim -3.5$4, both the synthetic and direct-Pristine branches achieve average offsets around 0.11–0.15 dex. At fainter magnitudes, the direct-Pristine metallicities remain comparatively stable, whereas the Gaia-only $[\mathrm{Fe/H}] \sim -3.5$5 metallicities degrade as the synthetic CaHK photometry becomes noisy; the scatter rises to about 0.33 dex for the synthetic catalogue, compared with 0.14 dex for Pristine (Martin et al., 2023).

A more direct empirical assessment comes from the first low-to-medium-resolution spectroscopic follow-up of bright, distant, very and extremely metal-poor RGB stars selected from the Pristine-Gaia synthetic catalogue. In that analysis, 76% of the observed stars have spectroscopic $[\mathrm{Fe/H}] \sim -3.5$6, only 3% are outliers with $[\mathrm{Fe/H}] \sim -3.5$7, the success rate is 77% for $[\mathrm{Fe/H}] \sim -3.5$8, and 38% for $[\mathrm{Fe/H}] \sim -3.5$9. For the photometric–spectroscopic comparison, the median offset is approximately $[\mathrm{Fe/H}]<-1.0$0 dex and the scatter is $[\mathrm{Fe/H}]<-1.0$1 dex; stricter cuts reduce the scatter to about 0.35 dex but remove about 20% of the VMP stars, which the authors judge not to be an advantageous trade (Viswanathan et al., 2024).

The catalogue’s published caveats are substantial and should be treated as intrinsic to the data model rather than as secondary warnings. Variable stars can contaminate photometric metallicity estimates; the recommended cut $[\mathrm{Fe/H}]<-1.0$2 removes about 20% of the Gaia BP/RP sample. Stars near the $[\mathrm{Fe/H}]<-1.0$3 model boundary or with $[\mathrm{Fe/H}]<-1.0$4 require caution. Reliability is strongest for $[\mathrm{Fe/H}]<-1.0$5, while $[\mathrm{Fe/H}]<-1.0$6 begins to show systematic offsets. A further limitation concerns carbon-enhanced metal-poor stars: for cool stars with $[\mathrm{Fe/H}]<-1.0$7 K and $[\mathrm{Fe/H}]<-1.0$8, the mean metallicity bias can be about 0.70 dex, because carbon absorption in the CaHK band mimics stronger calcium absorption and therefore a higher metallicity (Martin et al., 2023).

5. Scientific applications in Galactic archaeology

The principal scientific role of PGS is high-yield preselection and cartography of the metal-poor Milky Way. In the Pristine DR1 release, the combined photometric metallicity catalogues contain more than 2 million stars with $[\mathrm{Fe/H}]<-1.0$9, more than 200,000 with $CaHK_\mathrm{syn}$0, and about 8,000 with $CaHK_\mathrm{syn}$1. This scale turns the catalogue into a practical discovery engine for very and extremely metal-poor stars (Martin et al., 2023).

The first spectroscopic follow-up paper uses PGS to identify bright ($CaHK_\mathrm{syn}$2) RGB candidates out to 35 kpc, then shows that the resulting sample is dynamically rich as well as chemically extreme. The analysed stars are associated with known accretion structures including Gaia-Enceladus-Sausage, LMS-1/Wukong, Thamnos, the Helmi streams, Sagittarius, and Sequoia. The same study finds a 4$CaHK_\mathrm{syn}$3 overdensity in the prograde low-$CaHK_\mathrm{syn}$4 region relative to its retrograde counterpart and reports three new members of the C-19 stream, one approximately $CaHK_\mathrm{syn}$5 from the main body. A stated implication is that planned WEAVE follow-up of Pristine-selected candidates could produce 10,000–20,000 homogeneously analysed EMP stars (Viswanathan et al., 2024).

A complementary use appears in outer-halo mapping. By selecting RGB stars from PGS and deriving photometric distances with BaSTI $CaHK_\mathrm{syn}$6-enhanced isochrones, the outer-halo study publishes a PGS giants catalogue of 2,420,898 stars, of which 1,706,006 have 6D phase-space data, reaching to approximately 70 kpc. The paper emphasizes that PGS is not the preferred sample for an unbiased halo MDF as a function of distance, but it is the preferred sample for an all-sky view of the very metal-poor outer halo. Using this property, the study tentatively associates 41 stars with a very metal-poor stellar counterpart to the Magellanic Stream / Pisces Plume system (Viswanathan et al., 2024).

6. Use in cluster peripheries and relation to other Gaia synthetic catalogues

PGS has also been used as the core dataset for chemically informed searches around Milky Way globular clusters. In that application, the catalogue is treated as a cross-match between Pristine and Gaia delivering all-sky synthetic, de-reddened photometry, $CaHK_\mathrm{syn}$7, $CaHK_\mathrm{syn}$8, $CaHK_\mathrm{syn}$9, and proper motions. The addition of a CaHK-based photometric metallicity dimension enables searches in a combined space of proper motion, CMD, and CaHK-sensitive color–color information. For 30 globular clusters examined within 5 degrees of the cluster centres, the study finds extended structure in 22, with 4 showing diffuse envelope-like morphologies and the remainder showing tidal-tail-like structures; six of these detections are described as tentative new cases (Kuzma et al., 8 Jul 2025).

The catalogue’s role in that work illustrates what distinguishes PGS from other Gaia “synthetic catalogues.” GUMS, the Gaia Universe Model Snapshot, is a pre-instrumental, idealized synthetic Gaia catalogue generated within the Gaia simulator and explicitly excludes detection efficiency, instrumental response, and observational errors; it represents what Gaia could potentially observe rather than a catalogue built from real observations (Robin et al., 2012). GDR2mock is a Gaia DR2 mock stellar catalog generated with Galaxia from a Besançon Galactic model, designed to reproduce the Gaia DR2 data model and query behaviour, with noise-free values and scaled nominal uncertainties (Rybizki et al., 2018). The Ananke DR3 surveys are survey-like mocks based on FIRE-2 cosmological simulations that match Gaia DR3 selection, photometry, astrometry, radial velocities, extinction, and observational uncertainties while also retaining simulation truth values (Nguyen et al., 2023).

PGS is categorically different from all three. It is neither a pristine Universe-model snapshot, nor a data-model emulator, nor a cosmological mock survey. It is a data-derived synthetic photometric-metallicity catalogue whose synthetic element is the reconstruction of a Pristine-like CaHK measurement from Gaia XP spectra. This difference has methodological consequences. PGS does not provide simulation truth or a generative Milky Way model, but it does provide all-sky chemical discrimination in a regime where Gaia astrometry alone is often insufficient, especially for faint cluster outskirts, very metal-poor halo searches, and preselection of rare spectroscopic targets.

7. Scope, limitations, and enduring significance

The enduring significance of PGS lies in its resolution of a longstanding observational trade-off between sky coverage and metallicity sensitivity. By turning Gaia DR3 XP spectra into synthetic CaHK measurements, the catalogue exports the Pristine metallicity methodology to the full Gaia sky, albeit at shallower depth than direct Pristine imaging. This gives the Milky Way metal-poor population a homogeneous all-sky representation that is immediately usable for target selection, structure finding, and chemically tagged dynamical studies (Martin et al., 2023).

Its limitations are equally structural. PGS is constrained by the availability and signal-to-noise of Gaia XP spectra, especially in the blue; it is shallower than the direct Pristine catalogue, generally reaching only to about $G$0 in the outer-halo RGB application; and its selection function introduces a distance–metallicity selection bias that makes it less suitable than PDR1 for reconstructing an unbiased halo metallicity distribution versus distance (Viswanathan et al., 2024). In cluster work, its depth means that it mostly probes the red giant branch, while the lower-mass stars dominating tidal tails often remain too faint to recover cleanly (Kuzma et al., 8 Jul 2025).

Within those limits, PGS has become a central infrastructure product for low-metallicity Galactic astronomy. It supports the efficient discovery of VMP and EMP stars, extends Pristine-style metallicity inference beyond the Pristine footprint, enables chemically informed membership classification in contaminated fields, and provides a practical interface between Gaia DR3 spectro-photometry and the observational programme of Galactic archaeology.