3. Commonalities and different appearances of outbursts
Fig. 3 illustrates that the variability of the equivalent width of , Pa15, and Fe ii 5169 is very complex. Similarly to the photometric variability of Be stars (Stefl et al., 1995; Stefl & Balona, 1996, e.g.), it can be understood as composed of variations on 2 or 3 different time scales. Long-term season-to-season variations of the H equivalent width become easily apparent from the comparison of the mean emission level in the individual sub-panels of Fig. 3 and are assumed to be connected with changes of the total disk mass. The outburst-like variations, which in Fig. 3 appear as numerous upward spikes and are discussed in more detail below, are at the short end of the time scales. In addition, there may be links to trends on a time scale of months.
There are indications that the Boller-Chivens (Sect. 5.2) and some of the CAT (Sect. 5.3) observations were obtained during outbursts. Only in our systematic long-term monitoring data from HEROS can we unambiguously recognize and follow the entire development of the emission outbursts. In the period 1995-1997, we detected 11 outbursts of different strengths. In Table 3 their dates and strengths are listed.
Table 3. Times and relative amplitudes of line emission outburst detected in the HEROS observations in the period 1995 - 1997. MJD refers to the beginning of the outbursts as derived from the time of increase of the strength of the H emission wings and is accurate to within a day (see Sect. 3). The relative change of the (total) equivalent width of H and Pa15 is expressed as a percentage of the value at the beginning of the respective outburst. The typical errors are 1-4 percentage points for the H values and 10-20 for Pa15. The range of these errors reflects mainly the gradual long-term increase of emission. The usage of indicates that the actual beginning of the burst was not covered by our data, but must have been shortly before
Unlike previous high-resolution spectrographs, HEROS provides us with simultaneous data on spectral lines formed in different parts of the photosphere and the circumstellar disk during all phases of the variability. By following in detail the evolution in individual outbursts and spectral lines we were able to define general phases of an emission outburst. However, the same approach - although logical - is not suitable for a compact presentation. It is beyond the scope of this paper to include a full description of the spectral variations in the interval covered by our data. Thus we first give a generalized picture of an outburst in this section and only then describe the typical evolution of the spectral lines in its individual phases in Sect. 4. This way, rather than commenting on individual events we attempt to abstract and describe typical states and patterns which may ultimately form the basis for the physical understanding of the observations.
Fig. 4 represents the schematic picture of µ Cen's line-emission outbursts. It needs to be understood that this scheme is not representative of any particular outburst, although a few outbursts with such a course of the H equivalent width were observed in 1995. Rather, Fig. 4 is typical in the sense that it combines all main phases of the outburst as they appear in equivalent width, ratio, emission wing strength (as measured by fitting a Gaussian to the higher Balmer lines, for which the contribution of the central emission peak is less disturbing, for more details see Rivinius et al., 1998c), and peak separation at different levels of the quiescent emission.
The four primary phases of an outburst cycle are (i) periods of (relative) quiescence, (ii) sudden, short (5-13 days) drops in strength of all circumstellar emission lines (termed 'precursor' phase below), (iii) subsequent rapid increase (within 2-15 days) of the emission strength, which henceforth will be called outburst for brevity, and (iv) the transition phase which after the outburst eventually evolves into relative quiescence (10-35 days) and for which we have chosen the term relaxation. The details of the main phases are provided in Sect. 4.
Already from this coarse description it is clear that a single scalar quantity such as the often used equivalent width or also the ratio cannot adequately describe the temporal run of an outburst. Only based on the variability of the entire line profile is this possible. Particularly the initial phases of an outburst can look significantly different if monitored in different spectral lines or at different levels of the quiescent emission and also depending on whether recorded in equivalent widths or ratios. For instance, at low emission levels, the equivalent width is not a sensitive indicator of the onset of an outburst. So long as the underlying absorption profile is not filled in, the decrease of the emission peak height can be compensated by the enhanced emission wings. Therefore the total equivalent width may remain about constant, whereas the outburst can already be well recognized in the ratio. This happened during the outbursts observed in 1995. Similarly, exclusive usage of the ratio cannot overcome this problem either, since the peak height increases only after the extended emission wings have appeared.
The problem of a suitable scalar quantity concerns not only the full description of the outburst, but also the choice of the criterion which defines the moment when the outbursts starts. Our analysis shows that the emission in the wings of higher Balmer lines, mainly H , is a reliable indicator at all quiescent emission levels covered by our monitoring. We measured the strengths of the wing emission by fitting a Gaussian to the complete line profile. Due to emission in the line core, the derived equivalent widths are systematically wrong in their absolute value, but their gradient is high just at the beginning of the outburst (see Fig. 4). The method is not influenced by a missing precursor phase in equivalent width and/or variations and was also used in order to determine the outburst dates in Table 3.
In many actual outbursts some of the phases listed above may appear missing. The apparent lack of a precursor phase in equivalent width for weak mean emission was already mentioned. A corresponding threshold in may be , -2. A corresponding limit can be presumed for the Paschen lines but it would fall in phases of relative quiescence not covered by our observations. The precursor phase is getting more conspicuous with increasing average strength of the emission. Then, the initial drop in emission strength can reach as much as 30% as observed in 1997, when , and . By contrast, the outburst phase appears more conspicuous when the mean emission strength is weak. This was the case in 1995, whereas after the additional increase of the emission strength in 1996 and 1997 the outburst phases were mostly missing in and still weak in the Paschen lines.
The above also suggests how the appearance of line emission outbursts will evolve further if the emission at quiescence from the disk continues to strengthen. It would be interesting to test this prediction observationally.
© European Southern Observatory (ESO) 1998
Online publication: April 15, 1998