Intensive studies of some Seyfert 1 galaxies have shown that emission-line variations are clearly a response to continuum variations (e.g. NGC 5548: Clavel et al. 1991; NGC 3783: Reichert et al. 1994; Fairall 9: Rodríguez-Pascual et al. 1997). This gives strong support to the standard idea that emission-lines in active galactic nuclei (AGN) arise from photoionization of many small dense clouds moving around a central ionizing source (Netzer 1990). The nature of these clouds is still unclear, but an attractive possibility is that they might be the envelopes of bloated stars (Alexander 1997 and references therein).
To understand the structure and the kinematics of the emission-line region, it is of great interest to be able to decompose the emission-lines into several independent components. Most previous attempts decomposed the line profile into several Gaussian components (e.g. Wamsteker et al. 1990). Such decompositions do not ensure that the components are really varying independently.
An interesting way to overcome this problem is to decompose the emission-lines by a principal component analysis (PCA). This was first done by Mittaz et al. (1990), who applied a PCA to the IUE spectra of NGC 4151. They concluded that PCA has only a fairly limited application to the analysis of spectral variability, because they could not find the physical meaning of most components. The analysis we propose here is similar except that we consider only two components: the principal component and the rest component, which comprises all minor variations. Doing so, the physical interpretation of the components becomes much easier and PCA turns out to be a promising way to study emission-line variations and hence the structure and the kinematics of the line emitting region.
The great advantage of the PCA, as compared to reverberation mapping (Peterson 1993) or cross-correlation methods, is that it does not require very well-sampled observations. PCA allows us to study objects that have much too sparse observations to perform a meaningful cross-correlation between continuum and line fluxes. Our sample of 18 objects is to our knowledge the greatest on which line profile variations were studied in a consistent way. It gives us the opportunity to identify common behaviors among the AGN diversity and to look for possible trends with the object luminosity.
© European Southern Observatory (ESO) 1998
Online publication: December 16, 1997