Bustos Fierro & Calderón (2019) introduced a two-stage, purely astrometric method for extracting globular-cluster members from Gaia DR2 and acknowledged five open issues: uncorrected Gaia catalogue incompleteness in crowded cluster cores, a human-operated validation step, the exclusion of parallaxes from the clustering, elevated contamination in high-background fields with overlapping proper motions (NGC 6139, NGC 6712), and unweighted mean-proper-motion estimators that ignore per-star errors and covariances. Subsequent literature shows that the Gaia EDR3/DR3 generation resolved the estimator and parallax strands — Vasiliev & Baumgardt (2021) perform membership with a Gaussian mixture in the joint parallax–proper-motion space, deconvolved for the full per-star error covariance, and derive error- and systematics-aware mean motions for 170 clusters — and that automated pipelines such as BALRoGO superseded the manual validation step; a crowding-aware empirical model of the Gaia DR3 selection function is now public. The surviving open core is the original problem's hardest regime: complete, decontaminated membership for globular clusters in crowded cores and dense bulge-direction fields, where Gaia's own catalogue is incomplete, its proper-motion precision degrades with stellar density, and current practice substitutes Hubble or near-infrared data rather than correcting Gaia-based membership. The problem asks for Gaia-based membership in these regimes with a validated completeness correction and quantified contamination, checked against independent HST/near-infrared memberships and artificial-star tests.

Background

Globular clusters are dense, gravitationally bound assemblies of 10410^4–10610^6 old stars orbiting the Milky Way. Establishing which stars in a telescope field belong to a given cluster — cluster membership — is the substrate of nearly all cluster science: mean proper motions (PMs, the angular sky-plane velocity components μα∗=μαcos⁡δ\mu_{\alpha*}=\mu_\alpha\cos\delta and μδ\mu_\delta), which combined with distances yield cluster orbits; internal kinematics; and luminosity and mass functions. The European Space Agency's Gaia mission provides all-sky astrometry — celestial positions, PMs, and parallaxes (the apparent annual angular displacement that encodes distance) — for over a billion stars. Data Release 2 (DR2, 2018) measured five astrometric parameters for most sources brighter than G≈21G\approx21; Early Data Release 3 (EDR3, 2020; astrometry carried unchanged into DR3, 2022) reduced the statistical PM and parallax uncertainties by roughly a factor of two and improved systematics.

Membership from astrometry exploits the fact that cluster stars share a common PM and are spatially concentrated, whereas field stars (mostly Galactic disc and bulge) spread widely in both. Standard diagnostics are the vector point diagram (VPD: the scatter plot of the two PM components) and the colour–magnitude diagram (CMD: brightness versus colour, on which cluster stellar sequences are tight). Two metrics quantify extraction quality: completeness, the fraction of true members recovered, and contamination, the fraction of extracted 'members' that are actually field stars.

Two regimes resist Gaia-based extraction. First, crowding: Gaia's source detection and astrometric solutions degrade in dense fields; in DR2 the completeness in the centres of globular clusters fell below 75% at G∼20G\sim20, with almost all stars fainter than G=19G=19 missing in the centres of some clusters (Arenou et al. 2018), so many cluster stars lack measured PMs altogether. Second, high-background fields toward the Galactic bulge, where the stellar surface density is extreme and the field PM distribution overlaps the cluster's.

Bustos Fierro & Calderón (2019) developed a two-stage, purely astrometric membership method on Gaia DR2: (1) BIRCH (a hierarchical clustering algorithm) in the four-dimensional space of gnomonic (tangent-plane) sky coordinates and PMs, followed by iterative Gaussian sigma-clipping (repeatedly discarding outliers that deviate by more than a chosen multiple of the sample standard deviation); (2) analysis of the projected spatial distribution of stars in PM slices around the cluster mean PM. A human operator inspects the spatial distribution, VPD and CMD of the extracted stars — the only non-automated step. Applied to eight clusters, the method extracted 37–90% of expected members (fractions that are upper limits, since DR2 catalogue incompleteness was not accounted for) with 1–7% overall contamination — except NGC 6139 (37%) and NGC 6712 (18%), the clusters in the densest bulge-direction fields, where the cluster and field PM distributions overlap (field densities 51 and 180 stars per square arcmin). Mean PMs were computed as unweighted averages of member PMs. The authors acknowledged five open issues: uncorrected Gaia incompleteness in crowded cores; the human validation step; the exclusion of parallaxes, whose large DR2 uncertainties degraded clustering, with error-aware integration unaddressed; elevated contamination in high-background, overlapping-PM fields; and a mean-PM estimator ignoring per-star errors and covariances.

The EDR3/DR3 era largely answered the estimator, parallax, and automation issues. Vasiliev & Baumgardt (2021) modelled the joint distribution of parallax and PM for 170 clusters with a Gaussian mixture deconvolved from per-star measurement errors — using Extreme Deconvolution (a density-estimation technique that accounts for individual error covariances, here the full 3×33\times3 astrometric covariance from the catalogue) and Markov Chain Monte Carlo sampling (a standard method for propagating uncertainties through a model) — combined with a Plummer-model (spherical) spatial density profile; cluster mean PMs and parallaxes then follow with a full treatment of spatially correlated systematic errors, and the parallax dimension has the intrinsic cluster dispersion fixed to zero. The fully automated Bayesian code BALRoGO (Vitral 2021), which fits the cluster surface density and models field-star PMs with a heavy-tailed Pearson VII distribution, derived centres, radii and bulk PMs for over one hundred clusters without operator inspection. An empirical, crowding- and scanning-law-aware model of the Gaia DR3 selection function — the probability that a star of given magnitude and position enters the catalogue — is public (Cantat-Gaudin et al. 2023, the gaiasf package), calibrated on deep ground-based imaging and tested against Hubble Space Telescope (HST) observations of globular clusters.

What EDR3/DR3 did not fix is the crowded core and the high-background bulge regime. Vasiliev & Baumgardt (2021) found that in the central 1–2 arcmin of some clusters virtually no stars pass astrometric quality cuts, that formal uncertainties are underestimated by 10–20% in dense central regions, and that published composite quality filters fail precisely there; the clusters with the largest PM disagreements with earlier work are those in the densest fields. Alonso Luna et al. (2023) showed that Gaia DR3 PMs toward the bulge require a stellar-density-dependent inflation of their uncertainties to agree with HST. For the most crowded and obscured clusters the operational answer is no longer Gaia alone: the Hubble Missing Globular Cluster Survey combines HST and Gaia positions to measure PMs for stars for which Gaia could measure only positions, not proper motions (Rosignoli et al. 2026), with photometric completeness quantified by artificial-star tests (simulated stars inserted into the images to measure the detection probability; Libralato et al. 2026).

Problem Statement

The surviving open core of the problem posed by the acknowledged limitations of Bustos Fierro & Calderón (2019) — after the EDR3-era resolution of the estimator, parallax-integration and automation strands — is the crowded-core and high-background regime:

Determine membership of Milky Way globular clusters from Gaia astrometry in the regimes where Gaia's own catalogue currently fails or degrades — the crowded cores, where source detection and astrometric quality collapse and a large fraction of stars lack measured proper motions, and the dense bulge-direction fields with overlapping cluster/field proper-motion distributions exemplified by NGC 6139 and NGC 6712 — such that:

(1) Completeness is corrected rather than merely bounded: the fraction of true members recovered is quantified through a validated, crowding- and magnitude-dependent treatment of Gaia catalogue incompleteness (including stars without measured proper motions), so that recovered member counts, luminosity functions, and cluster mean proper motions are unbiased estimates over the whole cluster rather than upper limits.

(2) Member–field discrimination remains reliable where the field is dense and the proper-motion distributions overlap: contamination is quantified, and membership probabilities, together with error- and covariance-aware estimates of cluster mean proper motions (and parallaxes where the data admit them), reach the standard that current probabilistic methods achieve in uncrowded fields.

These two facets serve a single objective — unbiased, complete Gaia-based membership across the globular-cluster system — and must be demonstrated jointly on a representative set of high-density clusters in the crowding and high-background regime — NGC 6139 and NGC 6712, whose fields have stellar surface densities of 51 and 180 stars per square arcmin, are examples of this regime rather than the required list, and any representative set of such clusters suffices to resolve the task — rather than only mildly crowded cases, since correcting completeness without controlling contamination (or vice versa) does not yield usable membership. The membership determination itself is Gaia-based: external data (Hubble Space Telescope or near-infrared astrometry, artificial-star experiments) may be used to calibrate and validate the completeness and contamination claims, which is the field's accepted validation convention, whereas substituting external catalogues for Gaia membership cluster by cluster is the current workaround that the problem asks to go beyond. Where the innermost core of a given cluster admits no Gaia-based recovery at all, that limit must be stated and justified explicitly for the clusters and radii concerned, alongside the completeness-corrected membership obtained wherever the Gaia astrometry does carry information.

References

  1. Iván H. Bustos Fierro & J. H. Calderón (2019). Extraction of globular clusters members with Gaia DR2 astrometry. Monthly Notices of the Royal Astronomical Society 488, 3024. DOI: 10.1093/mnras/stz1879 (arXiv:1907.02612). https://doi.org/10.1093/mnras/stz1879
  2. Eugene Vasiliev (2019). Proper motions and dynamics of the Milky Way globular cluster system from Gaia DR2. Monthly Notices of the Royal Astronomical Society 484, 2832–2850. DOI: 10.1093/mnras/stz171. https://doi.org/10.1093/mnras/stz171
  3. Eugene Vasiliev & Holger Baumgardt (2021). Gaia EDR3 view on galactic globular clusters. Monthly Notices of the Royal Astronomical Society 505, 5978–6002. DOI: 10.1093/mnras/stab1475 (arXiv:2102.09568). https://doi.org/10.1093/mnras/stab1475
  4. Eduardo Vitral (2021). BALRoGO: Bayesian Astrometric Likelihood Recovery of Galactic Objects – Global properties of over one hundred globular clusters with Gaia EDR3. Monthly Notices of the Royal Astronomical Society 504, 1355–1369. DOI: 10.1093/mnras/stab947 (arXiv:2102.04841). https://doi.org/10.1093/mnras/stab947
  5. C. Fabricius, X. Luri, F. Arenou, C. Babusiaux, A. Helmi, T. Muraveva, C. Reylé, F. Spoto, A. Vallenari, T. Antoja, E. Balbinot, C. Barache, N. Bauchet, A. Bragaglia, D. Busonero, T. Cantat-Gaudin, J. M. Carrasco, S. Diakité, M. Fabrizio, F. Figueras, A. Garcia-Gutierrez, A. Garofalo, C. Jordi, P. Kervella, S. Khanna, N. Leclerc, E. Licata, S. Lambert, P. M. Marrese, A. Masip, P. Ramos, N. Robichon, A. C. Robin, M. Romero-Gómez, S. Rubele, M. Weiler (2021). Gaia Early Data Release 3: Catalogue validation. Astronomy & Astrophysics 649, A5. DOI: 10.1051/0004-6361/202039834. https://doi.org/10.1051/0004-6361/202039834
  6. Tristan Cantat-Gaudin, Morgan Fouesneau, Hans-Walter Rix, Anthony G. A. Brown, Alfred Castro-Ginard, Zuzanna Kostrzewa-Rutkowska, Ronald Drimmel, David W. Hogg, Andrew R. Casey, Shourya Khanna, Semyeong Oh, Adrian M. Price-Whelan, Vasily Belokurov, Andrew K. Saydjari, G. Green (2023). An empirical model of the Gaia DR3 selection function. Astronomy & Astrophysics 669, A55. DOI: 10.1051/0004-6361/202244784 (arXiv:2208.09335). https://doi.org/10.1051/0004-6361/202244784
  7. Alonso Luna, Tommaso Marchetti, Marina Rejkuba, Dante Minniti (2023). Astrometry in crowded fields towards the Galactic bulge. Astronomy & Astrophysics 677, A185. DOI: 10.1051/0004-6361/202346257 (arXiv:2307.13719). https://doi.org/10.1051/0004-6361/202346257
  8. M. A. Yepez, A. Arellano Ferro, I. Bustos Fierro, A. Luna (2023). The variable stars in the field of NGC 6139: a critical approach to their reddening and membership. Monthly Notices of the Royal Astronomical Society 524, 1503–1514. DOI: 10.1093/mnras/stad1831 (arXiv:2306.08710). https://doi.org/10.1093/mnras/stad1831
  9. Emily L. Hunt & Sabine Reffert (2023). Improving the open cluster census. II. An all-sky cluster catalogue with Gaia DR3. Astronomy & Astrophysics 673, A114. DOI: 10.1051/0004-6361/202346285 (arXiv:2303.13424). https://doi.org/10.1051/0004-6361/202346285
  10. L. Rosignoli, M. Libralato, R. Pascale, D. Massari, E. Dalessandro, E. Ceccarelli, H. Baumgardt, M. Bellazzini, A. Bellini, F. Aguado-Agelet, S. Cassisi, M. Monelli, A. Mucciarelli, E. Pancino, M. Salaris, E. Dodd, F. R. Ferraro, B. Lanzoni (2026). The Hubble Missing Globular Cluster Survey. II. Survey membership tools and kinematic analysis of NGC 6749. Astronomy & Astrophysics 707, A258. DOI: 10.1051/0004-6361/202558333 (arXiv:2512.01530). https://doi.org/10.1051/0004-6361/202558333
  11. M. Libralato, A. Bellini, D. Massari, M. Bellazzini, F. Aguado-Agelet, S. Cassisi, E. Ceccarelli, E. Dalessandro, E. Dodd, F. R. Ferraro, C. Gallart, B. Lanzoni, M. Monelli, A. Mucciarelli, E. Pancino, R. Pascale, L. Rosignoli, M. Salaris, S. Saracino, C. Zerbinati (2026). The Hubble Missing Globular Cluster Survey. III. Astro-photometric catalogs, artificial-star tests, and improved absolute proper motions. arXiv:2603.25848 (accepted, Astronomy & Astrophysics 709, A140). https://arxiv.org/abs/2603.25848
  12. F. Arenou, X. Luri, C. Babusiaux, C. Fabricius, A. Helmi, T. Muraveva, A. C. Robin, F. Spoto, A. Vallenari, T. Antoja, T. Cantat-Gaudin, C. Jordi, N. Leclerc, C. Reylé, M. Romero-Gómez, I.-C. Shih, S. Soria, C. Barache, D. Bossini, A. Bragaglia, M. A. Breddels, M. Fabrizio, S. Lambert, P. M. Marrese, D. Massari, A. Moitinho, N. Robichon, L. Ruiz-Dern, R. Sordo, J. Veljanoski, L. Eyer, G. Jasniewicz, E. Pancino, C. Soubiran, A. Spagna, P. Tanga, C. Turon, C. Zurbach (2018). Gaia Data Release 2: Catalogue validation. Astronomy & Astrophysics 616, A17. DOI: 10.1051/0004-6361/201833234. https://doi.org/10.1051/0004-6361/201833234