Spectral Classification of Galaxies Along the Hubble Sequence

Abstract

We develop a straightforward and quantitative two-step method for spectroscopically classifying galaxies from the low signal-to-noise (S/N) optical spectra typical of galaxy redshift surveys. First, using \chi^2-fitting of characteristic templates to the object spectrum, we determine the relative contributions of the old stellar component, the young stellar component, and various emission line spectra. Then, we classify the galaxy by comparing the relative strengths of the components with those of galaxies of known morphological type. In particular, we use the ratios of (1) the emission line to absorption line contribution, (2) the young to old stellar contribution, and (3) the oxygen to hydrogen emission line contribution. We calibrate and test the method using published morphological types for 32 galaxies from the long-slit spectroscopic survey of Kennicutt (1992) and for 304 galaxies from a fiber spectroscopic survey of nearby galaxy clusters. From an analysis of a sample of long-slit spectra of spiral galaxies in two galaxy clusters, we conclude that the majority of the galaxies observed in the fiber survey are sufficiently distant that their spectral classification is unaffected by aperture bias. Our spectral classification is consistent with the morphological classification to within one type (e.g. E to S0 or Sa to Sb) for \gtsim 80% of the galaxies. Disagreements between the spectral and morphological classifications of the remaining galaxies reflect a divergence in the correspondence between spectral and morphological types, rather than a problem with the data or method.