Spectroscopic Target Selection in the Sloan Digital Sky Survey: The Main Galaxy Sample

TL;DR

The paper presents a uniform, Petrosian-based target selection algorithm for the SDSS main galaxy spectroscopic sample, detailing how Petrosian quantities are defined and used to construct a flux-limited, extinction-corrected catalog with robust, distance-independent flux measurements. By selecting $r_P\leq 17.77$ and ${\mu}_{50}\leq 24.5$ in the $r$ band, the authors achieve a mean density of about 92 galaxies per square degree and a median redshift near $z\approx 0.1$, while maintaining high completeness and low stellar contamination. They validate the method with extensive tests on star-galaxy separation, spectroscopic quality, completeness, and reproducibility, and quantify fiber-collision losses and sky-subtraction effects. The resulting, well-characterized selection function supports precise measurements of galaxy clustering and properties, enabling robust cosmological analyses from the SDSS imaging and spectroscopy. The work provides a practical, reproducible framework for constructing uniform galaxy redshift samples in large surveys.

Abstract

We describe the algorithm that selects the main sample of galaxies for spectroscopy in the Sloan Digital Sky Survey from the photometric data obtained by the imaging survey. Galaxy photometric properties are measured using the Petrosian magnitude system, which measures flux in apertures determined by the shape of the surface brightness profile. The metric aperture used is essentially independent of cosmological surface brightness dimming, foreground extinction, sky brightness, and the galaxy central surface brightness. The main galaxy sample consists of galaxies with r-band Petrosian magnitude r < 17.77 and r-band Petrosian half-light surface brightness < 24.5 magnitudes per square arcsec. These cuts select about 90 galaxy targets per square degree, with a median redshift of 0.104. We carry out a number of tests to show that (a) our star-galaxy separation criterion is effective at eliminating nearly all stellar contamination while removing almost no genuine galaxies, (b) the fraction of galaxies eliminated by our surface brightness cut is very small (0.1%), (c) the completeness of the sample is high, exceeding 99%, and (d) the reproducibility of target selection based on repeated imaging scans is consistent with the expected random photometric errors. (abridged)

Figures (12)

• Figure 1: Illustration of the Petrosian aperture procedure for a de Vaucouleurs profile (top) and an exponential profile (bottom), assuming an axis ratio of one and negligible seeing. In each panel, the dashed curve shows the curve of growth (fraction of total light within radius $\theta$), and the solid curve shows the Petrosian ratio ${\cal R}(\theta)$. The dotted curve shows the logarithmic surface brightness profile, using the right-hand axis scale. The central arrow marks the Petrosian radius at ${\cal R}(\theta)=0.2$. Outer and inner arrows represent the radius of the Petrosian aperture ($f_2 {\theta_P}$) and the Petrosian half-light radius ${\theta_{50}}$, respectively. All radii are scaled to the true half-light radius $\theta_e$, which is 1.678 scale lengths for the exponential profile.
• Figure 2: Effect of axis ratio on the Petrosian flux, measured using our circular aperture definitions of Petrosian quantities. The dashed line shows the fraction of the total flux within the Petrosian aperture for inclined exponential disks as a function of axis ratio. The solid line shows the same quantity for a de Vaucouleurs law galaxy. The dotted line represents a galaxy with an inclined (exponential) disk and a circular (de Vaucouleurs law) bulge, assuming a 1:1 bulge-to-disk ratio and a bulge half-light radius that is half that of the disk.
• Figure 3: The fraction of the total light within the Petrosian aperture $f_2{\theta_P}$ as a function of the value of $f_2$, for $f_1=1/5$ (solid lines), $1/6$ (dotted lines), and $1/4$ (dashed lines). Lower curves correspond to de Vaucouleurs law profiles and upper curves to exponential profiles. Filled circles mark our adopted values $f_1=1/5$, $f_2=2$. These calculations simulate idealized circular galaxies, and seeing is assumed to be negligible.
• Figure 4: $R_P$, the Petrosian radius ${\theta_P}$ multiplied by the angular diameter distance (assuming $\Omega_m=0.3$ and $\Omega_\Lambda=0.7$) as a function of redshift $z$ for several small ranges of absolute magnitude in the main galaxy sample; in each panel, we show a linear regression of $R_p$ along $z$ as the solid line.
• Figure 5: Schematic flow diagram of the main galaxy target selection algorithm. All quantities are measured in the $r$ band and are corrected for foreground extinction. See the text for a full description of all quantities referred to in this figure.