Forum Springer Astron. Astrophys.
Forum Whats New Search Orders

Astron. Astrophys. 332, 459-478 (1998)

Previous Section Next Section Title Page Table of Contents

1. Introduction

The classification of the galaxies in a 3-D galaxy map provides invaluable information for studying the formation and evolution of galaxies in relation to the large-scale structure. With these goals in mind, we have performed a spectral classification for the ESO-Sculptor Faint Galaxy Redshift Survey (ESS, hereafter; de Lapparent et al. 1993). The photometric catalogue of the ESS is based on CCD imaging and provides the B, V and R(Johnson-Cousins) photometry of [FORMULA] galaxies (Arnouts et al. 1997). The spectroscopic catalogue provides the flux-calibrated spectra of a complete sub-sample of [FORMULA] 700 galaxies with [FORMULA] obtained by multi-slit spectroscopy (Bellanger et al. 1995). The ESS allows for the first time to map in detail the large-scale clustering at [FORMULA] (Bellanger & de Lapparent 1995).

Morphological classification (Hubble 1936, de Vaucouleurs & de Vaucouleurs 1961, Sandage 1975), is based on the recognition of image patterns and it naturally started with the investigation of the nearest galaxies. For non-local galaxies, producing an objective morphological classification in the same classification system as for local galaxies is extremely difficult and would require a very complex taxonomy (Ripley 1993). The major limitation is the angular resolution given by ground-based telescopes. Only a rough classification can be made, for example by fitting de Vaucouleurs or exponential profiles or using the relationship between the central concentration index and the mean surface brightness (Doi et al. 1993). Moreover, this method can only be applied up to modest redshifts ([FORMULA]). Recently, with the refurbished Hubble Space Telescope (HST), can one see the detailed morphology of galaxies up to [FORMULA] (or I [FORMULA]) (Abraham et al. 1994) and derive an acceptable morphological description up to [FORMULA] (van den Bergh 1997). Galaxies at these very high redshifts present a wide variety of morphologies, when compared to the nearby galaxies. However, when high redshift galaxies ([FORMULA]) are observed through visual photometric filters (e.g., the HST filters), the morphology is delineated by the redshifted blue or the UV emission due to young stars or by star-forming regions, making the objects appear of later morphological types than they really are. This effect could partially explain the high rate of distorted galaxies in the Hubble Deep Field (HDF) (van den Bergh 1997). In summary, the existing morphological classifications are severely dependent on the image spatial resolution, on the photometric filter, and as a result, on the redshift of the objects.

In contrast to the qualitative approach of the morphological classification, the principal physical characteristics of galaxies can be efficiently quantified by their spectral energy distributions (SED). For a given galaxy, the SED measures the relative contribution of the most representative stellar populations and constrains the gas content and average metallicity. It is therefore sensible to classify galaxies along a spectral sequence rather than a morphological sequence. Morgan & Mayall (1957) have shown that indeed there is a fundamental relationship between spectra of galaxies and their morphologies: three different populations which in general constitute every galaxy -the gas and the young and old stars (Bershady 1993, 1995)- contribute both to delineate the main morphological features (bulge, spirals arms, etc...), and the spectral features (the continuum shape, the emission lines and absorption bands). The spectral classification has several advantages over the morphological classification. The spectral range covered by low resolution spectroscopy (R [FORMULA] 500) is wider than the standard filters, and thus allows to define a common interval for objects describing a wide range of redshifts. Furthermore, spectra are easier to handle than 2-D images when a large amount of data is processed.

In this paper we perform the spectral classification of the galaxies in the ESS, using the Principal Component Analysis (PCA). The PCA technique has been applied to many problems of variate nature, from social to biological sciences. In astronomy, it is frequently used for compressing data to extract the variables which are truly correlated (Bijaoui 1974; Faber 1973; Efstathiou & Fall 1984). The PCA has already been used to study inherent relationships between some selected features or quantities calculated from the spectra of Seyfert galaxies (equivalent widths and line ratios) and their photometric magnitudes (Dultzin-Hacyan & Ruano 1996), and on QSO spectra (Francis et al. 1992). In a recent study, Connolly et al. (1995) have tested the PCA using the spectral and morphological templates of Kinney et al. (1996), to show how the spectral properties and the Hubble sequence are related. Using Kennicutt spectra (Kennicutt 1992a), Folkes et al. (1996) and Sodre & Cuevas (1997), show the correlation between spectral properties and morphological type for normal galaxies.

Here we further test the PCA technique as a tool to achieve a reliable spectral classification for a new sample of distant galaxies. The PCA method is shown to be a powerful tool for measuring both, the systematic and non-systematic spectral properties of a galaxy sample. We also study the behavior of the PCA with respect to the data noise level.

The paper is organized as follows. In Sect. 2 we describe the ESO-Sculptor (ESS) spectroscopic data. In Sect. 3 a brief overview of the PCA technique and its application to the spectral classification are given, as well as the classification procedure using the [FORMULA] test. In Sect. 4 we apply the PCA to a sample of normal Kennicutt galaxies, and illustrate some specific features of the method. In Sect. 5 the PCA is applied to the ESS. The analysis and the spectral classification are described in Sect. 6, along with the visual classification for the brightest galaxies. In Sect. 7 we compare our main results with those of other studies. The conclusions and prospects are summarized in Sect. 8.

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 1998

Online publication: March 23, 1998