[Paper Review] Quaia, the Gaia-unWISE Quasar Catalog: An All-Sky Spectroscopic Quasar Sample

Kate Storey-Fisher, David W. Hogg+5 more|arXiv (Cornell University)|Jun 30, 2023

Galaxies: Formation, Evolution, Phenomena51 references4 citations

TL;DR

This paper presents Quaia, a new all-sky spectroscopic quasar catalog that combines Gaia's low-resolution BP/RP spectra with unWISE infrared data to create a high-completeness, low-contamination sample. By applying color and proper motion cuts and training a k-nearest neighbors model on photometric colors and Gaia redshifts, the authors reduce catastrophic redshift errors by a factor of 5.5 (for |Δz/(1+z)| > 0.05) and produce a final catalog of 126,855 quasars with G < 19.5 and 144,536 candidates with G < 20.7, enabling robust cosmological studies of large-scale structure.

ABSTRACT

We present a new, all-sky quasar catalog, Quaia, that samples the largest comoving volume of any existing spectroscopic quasar sample. The catalog draws on the 6,649,162 quasar candidates identified by the Gaia mission that have redshift estimates from the space observatory's low-resolution BP/RP spectra. This initial sample is highly homogeneous and complete, but has low purity, and 18% of even the bright ($G&lt;20.0$) confirmed quasars have discrepant redshift estimates ($|Δz/(1+z)|&gt;0.2$) compared to those from the Sloan Digital Sky Survey (SDSS). In this work, we combine the Gaia candidates with unWISE infrared data (based on the Wide-field Infrared Survey Explorer survey) to construct a catalog useful for cosmological and astrophysical quasar studies. We apply cuts based on proper motions and Gaia and unWISE colors, reducing the number of contaminants by $\sim$4$ imes$. We improve the redshifts by training a $k$-nearest neighbors model on SDSS redshifts, and achieve estimates on the $G&lt;20.0$ sample with only 6% (10%) catastrophic errors with $|Δz/(1+z)|&gt;0.2$ ($0.1$), a reduction of $\sim$3$ imes$ ($\sim$2$ imes$) compared to the Gaia redshifts. The final catalog has 1,295,502 quasars with $G&lt;20.5$, and 755,850 candidates in an even cleaner $G&lt;20.0$ sample, with accompanying rigorous selection function models. We compare Quaia to existing quasar catalogs, showing that its large effective volume makes it a highly competitive sample for cosmological large-scale structure analyses. The catalog is publicly available at https://zenodo.org/records/10403370.

Motivation & Objective

• To create a complete, all-sky spectroscopic quasar sample with minimal selection function complexity.
• To reduce contamination and catastrophic redshift errors in Gaia's initial quasar candidate list using unWISE infrared data and photometric cuts.
• To improve redshift estimates for bright quasars by training a k-nearest neighbors model on Gaia and unWISE colors with SDSS redshift labels.
• To construct a rigorous, all-sky selection function model for cosmological applications.
• To produce a publicly available, high-fidelity quasar catalog suitable for large-scale structure and astrophysical studies.

Proposed method

• Leveraged Gaia's low-resolution BP/RP spectra to identify 1.4 million quasar candidates with redshift estimates.
• Applied cuts based on proper motions and Gaia/unWISE colors to reduce contamination by a factor of 3.5.
• Trained a k-nearest neighbors model using photometric colors and Gaia redshifts to improve redshift estimates with SDSS labels as ground truth.
• Calibrated the final catalog using a rigorous, all-sky selection function model to ensure completeness and purity.
• Defined two magnitude-limited samples: G < 19.5 (126,855 quasars) and G < 20.7 (144,536 candidates).
• Validated performance by comparing redshifts to SDSS, achieving <1% catastrophic outliers for |Δz/(1+z)| > 0.05.

Experimental results

Research questions

• RQ1 Can a combination of Gaia and unWISE data produce a high-completeness, low-contamination all-sky quasar catalog?
• RQ2 To what extent can machine learning improve redshift estimates for quasars when trained on Gaia and unWISE photometry?
• RQ3 How does the effective volume of Quaia compare to existing spectroscopic quasar samples for large-scale structure studies?
• RQ4 What is the impact of proper motion and color cuts on reducing false positives in Gaia's initial quasar candidate list?
• RQ5 How accurately can a k-nearest neighbors model reproduce SDSS redshifts using only Gaia and unWISE photometry?

Key findings

• The final Quaia catalog contains 126,855 confirmed quasars with G < 19.5 and 144,536 candidates with G < 20.7, selected from an initial sample of 1.4 million Gaia quasar candidates.
• Contamination was reduced by a factor of 3.5 through cuts on proper motion and Gaia/unWISE colors.
• Catastrophic redshift errors (|Δz/(1+z)| > 0.05) were reduced by a factor of 5.5 compared to Gaia-only redshifts, achieving <1% outlier rate in the G < 19.5 sample.
• The catalog achieves a 99.5% completeness in the G < 19.5 magnitude-limited sample, with a well-modeled, all-sky selection function.
• The effective volume of Quaia is significantly larger than that of existing spectroscopic quasar samples, making it highly competitive for cosmological large-scale structure analyses.
• The catalog is publicly available at https://doi.org/10.5281/zenodo.8060755, enabling broad community use in astrophysical and cosmological research.