--- title: 'Quaia: Gaia-unWISE Quasar Catalog' url: https://www.emergentmind.com/topics/quaia-quasar-catalog type: topic --- # Quaia: Gaia-unWISE Quasar Catalog Quaia, the Gaia-unWISE Quasar Catalog, is an all-sky quasar catalog constructed by combining Gaia quasar candidates with unWISE infrared data and SDSS-based redshift calibration. Introduced as a catalog useful for cosmological and astrophysical quasar studies, it draws on 6,649,162 Gaia quasar candidates with redshift estimates from Gaia’s low-resolution BP/RP spectra and, in its initial release, contains 1,295,502 quasars with $G<20.5$ and 755,850 candidates in a cleaner $G<20.0$ sample, together with rigorous selection-function models. It was presented as sampling the largest comoving volume of any existing spectroscopic quasar sample [2306.17749]. ## 1. Origin, scope, and nomenclature Quaia was designed around a specific limitation of the Gaia quasar-candidate sample. The parent Gaia sample is highly homogeneous and complete, but has low purity, and even among bright confirmed quasars with $G<20.0$, 18% have discrepant Gaia redshift estimates satisfying $|\Delta z/(1+z)|>0.2$ relative to the Sloan Digital Sky Survey (SDSS). Quaia addresses this by combining Gaia candidates with unWISE infrared data, applying purification cuts, and replacing the raw Gaia redshifts with SDSS-trained estimates [2306.17749]. The catalog title emphasizes an “all-sky spectroscopic quasar sample,” while later methodological descriptions often refer to the redshifts as “spectrophotometric.” The two usages reflect different stages of the pipeline: Gaia contributes low-resolution BP/RP spectral information, while downstream analyses describe the final Quaia redshift estimator as a machine-learning combination of Gaia and unWISE observables trained on SDSS spectroscopic labels [2306.17749][2402.05761]. Quaia is also “all-sky” in the sense of parent-survey reach rather than literal use of the full celestial sphere in every analysis. Subsequent cosmological applications consistently impose angular masks derived from the catalog selection function, so the effective sky fraction depends on the science case and masking threshold rather than on the nominal survey footprint alone [2306.17748]. ## 2. Construction from Gaia and unWISE The initial Quaia construction begins from Gaia quasar candidates and augments them with unWISE infrared data based on the Wide-field Infrared Survey Explorer survey. In the main catalog paper, proper-motion cuts and Gaia and unWISE color cuts are applied, reducing the number of contaminants by approximately $4\times$ [2306.17749]. A later clustering analysis summarizes the construction in more operational terms. In that description, the parent sample is the Gaia DR3 Quasar Candidate Sample, with unWISE full-sky coadds in W1 and W2 added through positional cross-match. Proper-motion filtering rejects sources inconsistent with zero motion beyond $3\sigma$, combinations of $(G-W1)$, $(BP-RP)$, and $W1-W2$ are used to suppress stellar contaminants, and astrometric excess noise together with RUWE flags are used to reject blended or poorly measured sources. After these decontamination steps, known contaminants are reduced by $\gtrsim4\times$ while losing only $\sim1.2\%$ of SDSS spectroscopic quasars [2511.17413]. This construction gives Quaia an intermediate methodological position between purely photometric AGN compilations and conventional slit-spectroscopic quasar surveys. The catalog is not assembled from a uniform ground-based spectroscopic campaign; instead, it is built from Gaia low-resolution spectra, infrared colors, astrometry, and supervised recalibration against SDSS spectroscopy [2306.17749]. ## 3. Redshift estimation and catalog statistics The defining technical change relative to the raw Gaia candidate list is the replacement of Gaia redshifts with a $k$-nearest neighbors model trained on SDSS redshifts. In the initial paper, this improves the $G<20.0$ sample to only 6% catastrophic errors with $|\Delta z/(1+z)|>0.2$ and 10% with $|\Delta z/(1+z)|>0.1$, corresponding to reductions of approximately $3\times$ and $2\times$, respectively, relative to the Gaia redshifts [2306.17749]. A companion cosmology analysis gives the estimator in more explicit form: redshifts are obtained with a K-nearest-neighbours regressor using Gaia $G$, BP, RP magnitudes, unWISE W1 and W2 photometry, and Gaia’s spectrophotometric $z_{\rm BP}$. The output $z_{\rm Quaia}$ is the median of the $K=27$ neighbours, and the quoted uncertainty $\sigma_z$ is their standard deviation. Validation against SDSS DR16Q reports that 62% of quasars satisfy $|\Delta z|/(1+z)<0.01$, and 83% satisfy $|\Delta z|/(1+z)<0.1$ [2306.17748]. The catalog’s principal size measures are stable across the subsequent literature. The full sample contains 1,295,502 quasars with $G<20.5$, while the cleaner subset contains 755,850 candidates with $G<20.0$ [2306.17749]. Later analyses describe the redshift range as $0\lesssim z\lesssim 4$, peaking at $z\sim1.5$, with the overall $dN/dz$ obtained either from direct calibration against SDSS spectroscopy or by stacking Gaussian per-object redshift PDFs [2306.17748][2601.16948]. For applications focused on cosmic-web reconstruction at intermediate-to-high redshift, a derived subset retains only sources with $0.80.52$, leaving 708,483 quasars over 24,372 deg$^2$ [2509.17696]. ## 4. Selection function, masking, and effective volume A central feature of Quaia is the provision of explicit selection-function models. In one cosmological implementation, the sky is divided into HEALPix pixels at $\mathrm{Nside}=64$, and a non-parametric Gaussian-process model is fit to observed counts versus four systematics templates: Milky Way extinction, stellar density, Large and Small Magellanic Cloud components, and the Gaia depth map $M_{10}$. This yields a pixel-by-pixel selection function $w(\hat n)$; pixels with $w<0.5$ are masked, and linear deprojection of the same templates is applied to the final overdensity maps [2306.17748]. The projected overdensity field used in later analyses is typically written as $$ \delta_g(\hat n)=\frac{N(\hat n)}{\langle N\rangle\,w(\hat n)}-1, $$ where $N(\hat n)$ is the pixel count and $w(\hat n)$ is the angular selection function [2402.05761]. A frequent misconception is that “all-sky” implies a single immutable footprint. In practice, different analyses adopt different effective masks. One study reports effective footprints of 74% of the sky for the low-redshift bin and 67% for the high-redshift bin after masking high stellar-density and high-extinction regions; another states that 67% of the sample passes the final mask; and ARF-based analyses describe a $\sim60\%$ footprint after imposing $S(\mathrm{pix})>0.5$ [2306.17748][2402.05761][2601.16948]. The statistical reach implied by this footprint is large. A later clustering paper quantifies the effective volume of the $G<20.5$ sample as $V_{\rm eff}\approx6.5\,h^{-3}\,\mathrm{Gpc}^3$, almost twice that of SDSS DR16Q or eBOSS at similar depths [2511.17413]. | Redshift bin | $N$ | Mean redshift | |---|---:|---:| | $0 \leq z < 1$ | 327,195 | 0.69 | | $1 \leq z < 2$ | 648,095 | 1.48 | | $2 \leq z < 3$ | 283,958 | 2.40 | | $3 \leq z < 4$ | 35,466 | 3.28 | These counts, reported for the full $G<20.5$ catalog, show that Quaia is most densely populated at intermediate redshift while still retaining a substantial high-$z$ tail. For analyses of the highest-redshift bin, an additional $|b|>30^\circ$ cut has been used to mitigate dust-modelling uncertainties near the Galactic plane [2511.17413]. ## 5. Cosmological uses and systematics debates Quaia was introduced with the explicit claim that its large effective volume makes it highly competitive for cosmological large-scale-structure analyses, and subsequent work has borne out that assessment. Using two redshift bins centered at $z=1.0$ and $z=2.1$, together with Planck CMB lensing and a low-$z$ BAO prior, one study obtained $\sigma_8=0.766\pm0.034$ and $\Omega_m=0.343^{+0.017}_{-0.019}$, with a 21.7$\sigma$ detection of the Quaia–lensing cross-correlation [2306.17748]. A later three-bin analysis extended the growth-history reconstruction to $\bar z_i=[0.69,1.59,2.72]$ and reported $\sigma_8(z)=\{0.65\pm0.10,\,0.39\pm0.06,\,0.22\pm0.06\}$, including one of the highest-redshift measurements of $\sigma_8$, namely $\sigma_8(z=2.72)=0.22\pm0.06$. That work also found quasar-bias values similar to other quasar samples, though with a less steep evolution at high redshifts [2402.05761]. Quaia has also been used to constrain primordial non-Gaussianity through angular redshift fluctuations (ARF), a two-dimensional observable based on angular deviations of the average redshift inside a tomographic shell. Combining Quaia quasar angular density and ARF with CMB lensing yielded $f_{\rm NL}=-3\pm14$ at 68% confidence level, described as the second tightest constraint on $f_{\rm NL}$ using large-scale-structure two-point statistics to date and the best measurement achieved using two-point projected summary statistics [2601.16948]. Clustering analyses with Quaia further support its role as a uniformly selected, all-sky quasar tracer over $0\leq z\leq4$. One such study measured $r_0=6.8\pm0.2\,h^{-1}\mathrm{Mpc}$ at $0\leq z<1$, $8.0\pm0.2\,h^{-1}\mathrm{Mpc}$ at $1\leq z<2$, $10.8\pm0.2\,h^{-1}\mathrm{Mpc}$ at $2\leq z<3$, and $13.9\pm1.2\,h^{-1}\mathrm{Mpc}$ at $3\leq z<4$, with slopes consistent with $\gamma\approx2$. The same analysis inferred characteristic minimum halo masses of quasar hosts near $\log_{10}(M_{\mathrm{min}}/M_\odot)\approx12.8$ across all redshifts and duty cycles rising from approximately 2% to approximately 7% with increasing redshift, corresponding to integrated quasar lifetimes of $t_{\rm QSO}\sim10^8$ years [2511.17413]. Systematics discussions in the Quaia literature have focused especially on CMB-lensing foreground contamination. One analysis argued that the high-redshift sample’s lower preferred $\sigma_8$ could plausibly be driven by contamination of the temperature-based CMB lensing map by high-redshift extragalactic foregrounds, while a later study found evidence of contamination in the $z\sim1.7$ cross-correlations but concluded that its impact on $\sigma_8(z)$ and $b(z)$ was negligible [2306.17748][2402.05761]. ## 6. Extensions, public products, and derived catalogues Quaia has developed into a platform for derivative data products. The original catalog is publicly available through Zenodo at `https://zenodo.org/records/10403370`, and later work provides selection-function maps, random catalogs, and analysis-specific products built on the same underlying sample [2306.17749]. For covariance estimation and forward modelling, 100 full-sky quasar mock catalogs with smooth redshift evolution from $z=0$ to $z\sim4$ were generated specifically for Quaia. These mocks incorporate spectrophotometric redshift uncertainties, the angular selection function, and the observed redshift number counts, and were validated against full-sky maps, $\sigma_z(z)$, $n(z)$, angular power spectra, covariance matrices, and angular two-point correlation functions. The published summary reports excellent agreement between the mocks and the data, with percent-level differences only in the lowest-$\ell$ power-spectrum point [2509.15890]. Quaia has also been used to construct high-redshift cosmic-web catalogs. Using 708,483 quasars in the interval $0.8