8.1. General variability of the chromospheric emission
A quantitative analysis of the H and H line profiles was presented in Sects. 4.1 and 4.2. At first glance the RV variations might suggest rotational modulation of active regions responsible for the chromospheric emission. This idea would be also supported by the sinusoidal variations of the star's radial velocity, RVphot (see Sect. 3.1). From Figs. 1, 3 and 4 it is seen that variations in radial velocity for the chromospheric Balmer lines are anticorrelated with variations in RVphot measured from photospheric lines. This would be expected to be observed if those variations were due to active regions consisting of photospheric spots with overlying plages. A spot approaching the receding limb would cause a blueshift in photospheric lines while, in contrast, the overlying plage would show itself as excess red-shifted emission in the chromospheric lines. The fact that a correlation is seen between the H and H lines, both showing anticorrelation with photospheric lines, argues strongly for active regions. However, there are some problems related to this interpretation. First, the RVphot modulation observed in the photospheric lines is weaker on August 4. It is possible that significant changes in the pattern of active regions are responsible for the different behaviour of RVphot in the second night, when strong flare activity has also been detected (see Sect. 7.2). Observational evidence in support of the association of strong flares with light curve changes have been found in active binary systems. Catalano & Frasca (1994) detected a large flare in the active RS CVn star HK Lac, and discussed its connection to the development of a new spot group in the stellar surface. More examples in other RS CVn systems can be found in Teriaca (1997). Also, it is well known that large flares are often triggered by the interaction between old and new magnetic structures in the solar case (Gaizauskas 1989).
Second, although there is a correlation between the RV variations seen in H and H, rotational modulation effects appear to be more firmly established in the case of H, since its RV curve has a larger amplitude and varies more symmetrically with respect to zero velocities. This is more difficult to explain in terms of active regions only.
Finally, the FWHM and EW of the Balmer lines do not show the same gradual trend seen in RV. In contrast, they are seen to undergo more irregular variations although, in general, the EW curves indicate that the chromosphere brightness becomes greater at 0.5, coinciding with two consecutive flare brightenings at the night of August 5.
It is clear that rotation of surface active regions cannot explain by itself the variability of the chromospheric emission. There are other important sources of variability. One is the strong absorption seen almost permanently in the red on the two nights but with higher contrast on August 4. This feature explains the permanent displacement of the H line towards the blue. In addition, because its strength is not uniform with phase (see Sect. 5), it may be also related to the RV modulation. In this case, however, better agreement would be expected between H and H observations. This and other aspects concerning the absorption features are discussed separately (see Sect. 8.3). Additional variations are due to transient flaring, particularly noticeable in EW and FWHM. However, these are seen as deviations from the general trend and would not cause any phased modulation. In general, flares are less well defined in the H EW and FWHM curves, which may explain some of the differences observed in the behaviour of this line with respect to that of H. Flaring is discussed in more detail below.
8.2. Flaring activity
The assignment of the observed emission transients to flares is supported by the emission seen at those phases when H is strongest, which is one of the invariable features in the solar flare spectrum during the flash phase (vestka 1976). Maximum flaring is displayed on the second night, when two consecutive flares were seen to occur in association with relevant phenomena we discuss next.
The strong enhancement of the blue wing of the H line during flares is explained in the context of flare dynamical models as an indicator of chromospheric evaporation. Excess emission at velocities between -150 and -200 km s-1 is observed as part of the flaring in the two nights, suggesting fast ejections of hot material. Considering initial velocities of that order the maximum height that is expected to be reached by the ejected gas if following a radial trajectory under gravity is about 33000 km, which is comparable to the typical loop heights in solar flares. However, the upflow would only last about 7 min. The fact that the extra emission in the blue is clearly visible at least during 2 hr in the second night suggests ejection is taking place continuously, in which case it could constitute a powerful supply of prominence material.
Fast and sporadic injections of chromospheric material are believed to contribute to the total mass of solar prominences. However, the mass required to maintain them cannot be supplied only by this source. At present, several models have been developed to study possible injection mechanisms that can account for prominence formation, involving macroscopic and microscopic flows of material. Among the most efficient macroscopic processes stands evaporation (Poland & Mariska, 1986), with corresponding upflows in the range of 100-500 km s-1, and ballistic injections of matter (An et al., 1988a, b; Wu et al. 1990), with initial velocities not exceeding 20 km s-1. However, loop prominences are known to be associated with energetic flares producing a large amount of fast particles. The prominence formation model due to Jefferies & Orrall (1965) consider that some of these fast particles travel up into the loop and remain stored in the magnetic field, thus providing the required mass and energy to form the prominence.
The behaviour of the H line during the decay of the second flare event observed at the night of August 5 differs remarkably from the standard. The distinct feature is a strong absorption seen at velocities v -50 km s-1 as explained in Sect. 7.2. A similar phenomenon is sometimes observed after the maximum of large solar flares, owing to cool dark loops that are seen in absorption against the flare background. The loops system appears to rise at a rate of 5-10 km s-1, as new loops form at larger heights while the old, lower ones, fade away. In the stellar case we do not have the spatial resolution to detect them. However, our result may provide indirect evidence for analogous forms of activity.
8.3. Nature of the H profile asymmetry
The fact that the observed H profile is very asymmetric, as well as narrower than expected given the calculated rotational velocity, has been proved to be due to a strong absorption at positive velocities. This would admit two possible interpretations.
First, it can be understood in terms of material falling down into the chromosphere under the effect of gravity, in a way reminiscing the coronal rain phenomenon.The same particle-acceleration mechanism used in Jefferies and Orrall's model was proposed by the authors to explain the origin of coronal rain from the ejected particles stored in the magnetic field of the inner corona. Additionally, coronal rain is in some cases the remnant of precedent flare loops in the Sun.
The other possible interpretation is as an intense and continuous downflow of material along magnetic loops, by analogy with which is typically observed in solar loop prominences (Tandberg-Hanssen, 1995). In fact, both phenomena, coronal rain and downflows in loop prominences, are normally very difficult to distinguish, even in the case of the Sun, where the spectral resolution allows to measure real velocities.
On a rapidly-rotating star like BD+, a strong downflow of cool absorbing material can be produced as a result of the interaction between the magnetic field and the stellar plasma in conditions of unstable mechanical equilibrium. This is explained in the context of prominence clouds formation and stability in late-type rapid rotators, according to a previous study of the mechanical forces acting on the neutral material that is tied to the magnetic field above the chromosphere (van den Oord et al. 1998). At the typical heights where the clouds are found the effective component of the magnetic field is the dipole component, and mechanical equilibrium on a field line is possible only in the equatorial plane. Under this assumption, the study demonstrates that for heights lower than a certain limit the neutral material suspended in the magnetic field is in unstable equilibrium and may fall back into the chromosphere along the field lines. Prominences may exist at larger heights in the equatorial plane, where stable equilibrium is possible. However, they would be impossible to detect in projection on the stellar disk due to the inclination of the stellar rotation axis, . In this case they would never result in absorption transients.
A downflow of cool material towards the stellar poles is not predicted by previous works on formation and stability of prominences in rapidly-rotating stars. Collier-Cameron (1988) developed a model to explain the formation of prominences as condensation in loops that extend beyond the co-rotation radius, where the effective gravity is directed towards the star. Ferreira & Jardine (1995) addressed the stability of filament-like structures in rapidly-rotating stars by considering a ring-like filament current around the stellar equator. In that scenario there is no connection between the filament field and the photosphere, in which case it would be difficult to argue for a downflow of material in any preferred direction.
Our results suggest that prominence phenomena may manifest in other ways than just as systematic variations in the fluxes. Therefore, the investigation of mass flows and asymmetric profiles may provide a way, perhaps unique, to detect prominence activity in other stars, specially in those where we have no spatial resolution, i.e. all non-rapid rotators. Evidence of asymmetries in the profiles of active late-type stars has increased considerably in the last few years. Byrne et al. (1995) has convincingly demonstrated a strong asymmetry in the H emission profile of the active RS CVn-type star II Peg (see also Byrne et al. 1997). Similar to the case of BD+, the H emission in II Peg is mainly seen in the blue. The same asymmetry seems to occur also in the central reversal and can be observed in other lines, like He I 10830 Å (Byrne et al. 1997). If this effect can be interpreted as due to mass infall and it is detected in a large number of stars, it may indicate that prominences are an important component of the stellar atmospheres.
8.4. Lithium and effects of spot activity
The Li I line is very strong and shows significant variations in equivalent width and profile shape. Small variations have been previously observed by J94 at lower spectral resolution. An explanation based on large spotted regions which move across the stellar surface was suggested by those authors. The existence of spots was supported by the light modulation and the observed sinusoidal variations of the stellar radial velocity. The higher spectral resolution of the present observations has allowed to unequivocally detect the effect of surface spots on the temperature-sensitive Li I line. Systematic profile changes were seen to correlate with variations in profile shape seen in other spot-sensitive lines (Ca I 6717 Å and Ca I 6439 Å). This also explains the apparent modulation of RVphot (see Sect. 3.1), as suggested in J94.
These kinds of effects are seen most clearly in the second night of observations. On August 4 the variability of the line profile is not so extreme and the amplitude of the RVphot variations is generally smaller. In addition, the Li I EW is systematically lower. Possible variations due to the overlying molecular absorption in the Li I spectral region have been suggested (see Sect. 6) but that would not explain the different behaviour of RVphot in both nights. The difference in phase coverage may be important and future observations are required to resolve this. Moreover, two consecutive flares were observed on August 5 (see Sect. 7.2) which are presumably connected to changes in spot configuration, as suggested in Sect. 8.1. If such was the case, variations in Li I EW could be a manifestation of stronger spot activity. Because the density of neutral lithium increases at lower temperatures, Li is expected to be enhanced in spotted regions. Therefore, Doppler imaging analysis of BD+ are very important to clarify this point.
According to previous results obtained by Pallavicini et al. (1993), no significant variations of the Li I line are to be expected in spotted stars. However, evidence of Li I variability, presumably associated to spot activity, has been reported in other stars (Fernández & Miranda 1998, Martín & Claret 1996). Results from future observations of BD+ promise valuable information in this respect.
© European Southern Observatory (ESO) 1999
Online publication: December 4, 1998