IntroductionIn the most recent issue of the Be Star Newsletter, we published a prediction of the circumstellar activity of the Be star Mu Cen (Rivinius et al. 1997), which in this regard is one of the most prolific stars of its kind. It not only exhibits low- and high-order line profile variability (Baade 1984) as most other Be stars, but also has apparently random and frequent line-emission outbursts (e.g. Baade et al. 1988;Hanuschik et al. 1993; Peters 1995). A link between both phenomena could not so far be established. But it now seems as if at least in Mu Cen part of this secret could be lifted. Photospheric lpvOur time series analysis, based on the overall radial velocities of mostly emission-free lines, revealed six coherent periods (Rivinius et al. 1998a, Rivinius et al. 1998b [Paper I], Rivinius et al. 1998c [Paper II]). These periods are sorted in two closely spaced groups at 0.28 and 0.5 days, respectively (Table 1). The trustworthiness of this analysis is intensively discussed in Paper II. A 2-dimensional time series analysis was performed on more than 20 spectral lines not only to confirm the multiperiodicity but to derive the properties of either variability. In each group, the periods are associated with identical surface variability patterns, whereas these patterns differ between the groups.
TABLE 1. Parameters for sine fits of the radial velocitiy variations in the line cores of He I 4121, 4168, 4438; these are NOT physical pulsation amplitudes. Constructing the ephemerisWorking on the datasets available in 1996 November, we noticed that the typical recurrence times of outbursts is on the same order of magnitude as the beat period of the two strongest variations of the 0.5-day group. A first check using only these two periods showed some, albeit weak, coincidence between outbursts and constructive interference of the two sinewaves by which we have approximated the measured radial velocity variations. By taking into account all periods of the 0.5-day group (Table 1) and developing an appropriate criterion for the determination of the beginning of an outburst in a way that is independent of the strength of the persistent disk emission, we could improve the correlation quite considerably (Fig. 1, panels b and c). As far as we can see, the 0.28-day group of periods does not significantly contribute to the outbursts. This indicates on the one hand that interference of variabilities with different surface patterns is much less effective. On the other hand, interferences within the 0.28-day group probably do not reach the amplitude threshold which apparently needs to be exceeded for an outburst.
FIGURE 1. The measured strength of the broad wings of the H-delta line (dots), the co-added amplitudes of the 0.5-day group of periods (line) and the calculated mass ejection (filled areas). The 1995, 1996 and 1997 data are the strengths of the broad emission wings measured in Hdelta. For 1987 the H-alpha equivalent width measured by Hanuschik et al. (1993) is shown. Line emission outbursts start when the co-added radial velocity amplitude is maximal. Note that the uncertainity of the periods can cause shifts of up to ten days for the 1987 calculation. However the temporal pattern of the outburst sequence is much more stable against errors of the periods. The vertical line in panel d marks the submission date of the prediction of the 1997 activity (Rivinius et al. 1997). In a sense, we have been extraordinarily lucky that only few combinations of variabilities are strong enough to trigger outbursts, namely P1+P2, P1+P3 and P2+P3, the latter beeing already so weak that it induces only minor activity. With a lower threshold, we might have lost the possibility to recognize individual events. Furthermore, the initially very weak emission from the disk has also helped a lot. In an experimental model, we calculated the amount of ejected mass by
The accuracy of the periods derived allowed us to perform an independent check to be performed with the data obtained by Hanuschik et al. (1993) back in 1987, using the same parameters as derived from the 1995 and 1996 data. The quite successful match between observations and model is shown in Panel a of Fig. 1. Also the outburst event reported by Peters (1995) between April 1 and 5, 1994 coincides with a positive amplitude interference. After reduction of our 1997 January data, the model turned out to still work well (Fig. 1, left half of panel d). But the ESO 50-cm telescope, with which our observations were mainly taken, was already earmarked for mothballing after 1997, March. Accordingly, further verification of the model from own observations would have become difficult. For this reason, and encouraged by the positive experience with the 1987 and 1997 datasets, we decided to submit our prediction of the activity and a request for help with the confirmation of the prediction already in February 1997, although the analysis of the data was still in an intermediate working stage. We are grateful to Michelle Thaller, Conny Aerts, and Martin Kurster for having provided us with some additional spectra. The results, also of our own observations in March (and 11 further nights with the ESO 1.5-m telescope in April) are shown on the right hand side of Panel d in Fig. 1. The coincidence of combined radial velocity amplitude maxima and outbursts was proven for each of the predicted events on MJD50515 and MJD50540 . Even the double structure of the latter event predicted by the preliminary model with sub-peaks on MJD50435 and MJD50447 was detected. Unfortunately our campaign ended just before an expected second major event. In spite of the success of predicting the times of increases of the line emission from the beating of the radial velocity curves, we will nevertheless have to revise our preliminary (and as of yet completely unphysical) model of mass ejection. Although the structure of the bursting pattern is now reproduced in quite some detail, the calculated times of the onset of the mass ejection are for larger outburts systematically too early (compared to the times when the line emission actually rises). Because the O-C values do not increase with time, this only shows that without a physical model the outburst mechanism cannot be properly understood. We wish to thank again Michelle Thaller, Conny Aerts and Martin Kurster and all Heros observers for their observational efforts to check our prediction.
We will keep the readers of the Newsletter informed by submitting the
abstracts of newly finished papers, starting with the abstract of
Paper I in this issue. Our publications are also available on the web
at URL
References
|