Precision Spectroscopy for 1.9 Million Galaxies from SDSS-IV: Improved Spectral Measurements and Catalogs for eBOSS

TL;DR

This work delivers the eBOSS Data Analysis Pipeline (eBOSS-DAP), an adaptation of the MaNGA-DAP, to extract uniform emission-line fluxes, EWs, stellar and gas kinematics, continuum indices, and stellar-population weights from 1.9 million eBOSS galaxy spectra (z < 1.12). By leveraging the survey’s abundant duplicates, the authors quantify spectrophotometric calibration performance and deliver three public data products: a full emission-line catalog (27 lines) for 1.9 million spectra, a high-EW subset (78 lines) for ~167k spectra, and a SPIND catalog of spectral indices and template weights, along with per-spectrum FITS fits. The analysis demonstrates robust internal and external consistency (Gaussian vs integrated EWs, SDSS comparisons, repeat spectra), and uses the derived measurements to illustrate key scaling relations such as the mass–σ⋆, BPT classifications, SFR–M⋆, and the mass–metallicity relation across different eBOSS target classes. The catalog and software release enable wide-ranging studies in galaxy evolution and AGN activity at $z<1$, with explicit caveats on aperture and spectrophotometric limitations providing reliable context for subsequent analyses.

Abstract

The Sloan Digital Sky Survey IV DR17 Extended Baryon Oscillation Spectroscopic Survey (eBOSS) consists of 2,233,939 high-quality optical galaxy spectra obtained through 2" fibers, providing a rich spectroscopic resource for studying galaxy evolution across a broad redshift range. eBOSS was designed primarily for large-scale structure and BAO measurements and, as such, focused on galaxy redshifts, leaving much of the information contained in the spectra unexplored. In addition to the trove of spectra, the large number of repeat observations (197,521 duplicate spectra) enables evaluation of the survey's spectrophotometric quality. To unlock this potential, we introduce the eBOSS Data Analysis Pipeline (eBOSS-DAP), adapted from the MaNGA-DAP, which delivers uniform measurements of emission-line fluxes and equivalent widths, stellar and gas kinematics, continuum spectral indices, and stellar population fits. Using the eBOSS-DAP, we successfully analyze 1,899,553 high-quality galaxy spectra below a redshift of $z < 1.12$ to produce an extensive spectroscopic catalog for the eBOSS galaxy sample. We characterize the calibration performance, quantify the reliability of the derived measurements, and release a suite of data products that fully exploit the power of the eBOSS dataset. These catalogs open the door to a new generation of studies in galaxy evolution and cosmology.

Figures (26)

• Figure 1: Redshift distribution of the SDSS-I (light blue) and eBOSS (gray) galaxy samples. The eBOSS galaxies are further divided based on how they were targeted for spectroscopy. LRGs are Luminous Red Galaxies; these galaxies are further broken down by the survey from which they originate. ELGs are Emission Line Galaxies and take up a large portion of our sample at the highest redshifts. TDSS/SPIDERS are targeted based on variable photometry and X-ray data. Finally, the histogram labeled QSO represents objects that were initially targeted as quasars but, upon further analysis, were revealed to be compact blue galaxies. In total, the eBOSS galaxy sample peaks at z = 0.55 and contains 1,901,834 spectra. For comparison, the SDSS-I galaxy sample contains 987,729 spectra with two peaks at z = 0.1 and z = 0.35.
• Figure 2: Median signal-to-noise (S/N) per pixel distribution of the SDSS-I (light blue) and eBOSS (gray) galaxy spectra. The eBOSS galaxies are further divided based on how they were targeted for spectroscopy. In total, the eBOSS galaxy sample peaks at S/N = 1.88. For comparison, the SDSS-I galaxy sample has two peaks at S/N = 3.9 and S/N = 12.1.
• Figure 3: H$\beta$ emission line Equivalent Width vs. redshift for all six target categories and a 2D histogram of the full sample. The target contours begin at 50% of the data.
• Figure 4: A two-dimensional kernel density estimate plot of the fraction of galaxy light encompassed by the SDSS fiber for the different target categories as a function of redshift. The fraction of light in the fiber is computed from the ratio of the SDSS $r$-band flux measured from the spectrum over that measured from photometry. This shows that different targeting categories are more likely to be fully captured by the fiber than others. The outer contours enclose 90% of the sample.
• Figure 5: An estimate of the relative flux calibration error of the eBOSS spectra based on comparisons of repeat spectra of the same galaxies observed at two or more different epochs. Here we compare the sum the flux in a 200 Å window centered at the wavelength, LAMBDA_EFF, that the spectrophotometric calibration is optimized for. For the majority of galaxies, this is 5400 Å (Panel B), however most ELG targets have LAMBDA_EFF=7500 Å (Panel C) and most QSO targets classified as galaxies have LAMBDA_EFF = 4000 Å (Panel A). The orange lines show Gaussian fits; the 1-$\sigma$ spread of the data is listed in each plot. This represents the minimum uncertainty of a line flux measurement due to spectrophotometric calibration errors. These errors can be larger at wavelengths that are far from LAMBDA_EFF as shown in Figure \ref{['fig:Bowtie']}.