Redshift Measurement and Spectral Classification for eBOSS Galaxies with the Redmonster Software
We describe the redmonster automated redshift measurement and spectral classification software designed for the extended Baryon Oscillation Spectroscopic Survey (eBOSS) of the Sloan Digital Sky Survey IV (SDSS-IV). We describe the algorithms, the template standard and requirements, and the newly developed galaxy templates to be used on eBOSS spectra. We present results from testing on early data from eBOSS, where we have found a 90.5% automated redshift and spectral classification success rate for the luminous red galaxy sample (redshifts 0.6$\lesssim z \lesssim$1.0). The redmonster performance meets the eBOSS cosmology requirements for redshift classification and catastrophic failures, and represents a significant improvement over the previous pipeline. We describe the empirical processes used to determine the optimum number of additive polynomial terms in our models and an acceptable $Δχ_r^2$ threshold for declaring statistical confidence. Statistical errors on redshift measurement due to photon shot noise are assessed, and we find typical values of a few tens of km s$^{-1}$. An investigation of redshift differences in repeat observations scaled by error estimates yields a distribution with a Gaussian mean and standard deviation of $μ\sim$0.01 and $σ\sim$0.65, respectively, suggesting the reported statistical redshift uncertainties are over-estimated by $\sim$54%. We assess the effects of object magnitude, signal-to-noise ratio, fiber number, and fiber head location on the pipeline's redshift success rate. Finally, we describe directions of ongoing development.