A GMBCG Galaxy Cluster Catalog of 55,424 Rich Clusters from SDSS DR7

We present a large catalog of optically selected galaxy clusters from the application of a new Gaussian Mixture Brightest Cluster Galaxy (GMBCG) algorithm to SDSS Data Release 7 data. The algorithm detects clusters by identifying the red sequence plus Brightest Cluster Galaxy (BCG) feature, which is unique for galaxy clusters and does not exist among field galaxies. Red sequence clustering in color space is detected using an Error Corrected Gaussian Mixture Model. We run GMBCG on 8240 square degrees of photometric data from SDSS DR7 to assemble the largest ever optical galaxy cluster catalog, consisting of over 55,000 rich clusters across the redshift range from 0.1 < z < 0.55. We present Monte Carlo tests of completeness and purity and perform cross-matching with X-ray clusters and with the maxBCG sample at low redshift. These tests indicate high completeness and purity across the full redshift range for clusters with 15 or more members.

💡 Research Summary

The paper introduces a novel optical cluster‑finding algorithm called GMBCG (Gaussian Mixture Brightest Cluster Galaxy) and applies it to the Sloan Digital Sky Survey Data Release 7 (SDSS DR7) to produce a catalog of 55,424 rich galaxy clusters spanning the redshift interval 0.1 < z < 0.55. The method builds on the well‑established red‑sequence technique but adds two decisive improvements. First, it explicitly searches for the Brightest Cluster Galaxy (BCG) and requires that a candidate BCG and a red‑sequence population be mutually consistent in color and spatial location. Because BCGs are typically located near the gravitational centre and have a distinctive colour‑magnitude relation, this dual criterion dramatically reduces contamination from field galaxies, especially at higher redshift where the red‑sequence becomes less pronounced. Second, the algorithm models the colour distribution of galaxies around each BCG with an Error‑Corrected Gaussian Mixture Model (ECGMM). Unlike a standard Gaussian mixture, ECGMM incorporates the individual photometric colour uncertainties as weights, yielding unbiased estimates of the red‑sequence mean colour, scatter, and the fraction of red‑sequence members.

The pipeline proceeds in four steps. (1) Candidate BCGs are selected using absolute r‑band magnitude (M_r < ‑22.5) and colour cuts that roughly follow the expected red‑sequence locus. (2) For each candidate BCG, galaxies within a projected radius of ~0.5 Mpc are gathered, and a preliminary colour filter (g‑r, r‑i) is applied to remove obvious outliers. (3) The colour distribution of the remaining galaxies is fitted with a one‑ or two‑component ECGMM; the red‑sequence component is identified by its higher mean colour and smaller dispersion. (4) A cluster is confirmed if the BCG’s colour matches the red‑sequence mean within the photometric errors and if the number of red‑sequence members (richness) exceeds a threshold of 15. Richness is defined as the count of red‑sequence galaxies after background subtraction, and it serves as a proxy for cluster mass.

The authors processed 8,240 deg² of SDSS imaging, using the five‑band (u,g,r,i,z) model magnitudes corrected for Galactic extinction, K‑correction, and photometric errors. Red‑sequence colour–redshift relations are derived in 0.01 redshift slices, allowing the algorithm to adapt to the gradual evolution of galaxy colours with redshift.

Performance is evaluated through two complementary tests. Monte‑Carlo simulations inject synthetic clusters of varying richness and redshift into the real photometric catalog; the recovery rate for clusters with ≥ 15 members exceeds 90 % completeness and 95 % purity. Cross‑matching with external X‑ray selected cluster samples (ROSAT, Chandra) yields an 85 % match fraction, confirming that the optical detections correspond to genuine massive halos. In the low‑redshift regime (0.1 < z < 0.3) the GMBCG catalog overlaps with the well‑known maxBCG catalog at the 93 % level, demonstrating consistency with previous work while offering a larger sky coverage and higher richness threshold.

The results show that GMBCG maintains high purity across the full redshift range and achieves completeness comparable to, or slightly better than, existing red‑sequence methods. The algorithm remains robust up to z ≈ 0.5, where the red‑sequence colour scatter widens and the number of detectable members declines; completeness drops to ~70 % in this regime, indicating room for improvement. Limitations also arise from the depth of SDSS imaging, which can bias richness estimates for distant or low‑mass clusters.

Future directions outlined by the authors include (i) incorporating near‑infrared data (e.g., WISE, UKIDSS) to extend reliable red‑sequence detection to higher redshifts, (ii) replacing the Gaussian mixture with more flexible machine‑learning models (e.g., deep neural networks) to capture non‑Gaussian colour distributions, and (iii) integrating spectroscopic redshifts where available to refine cluster mass proxies and improve calibration of the richness–mass relation.

In summary, the GMBCG algorithm provides an efficient, scalable, and statistically robust framework for constructing large optical cluster catalogs. The resulting 55,424‑cluster catalog constitutes the largest homogeneous optical cluster sample to date, offering a valuable resource for cosmological analyses (e.g., cluster abundance, large‑scale structure studies) and for investigations of galaxy evolution within dense environments.