The behaviour of the H
and H
emission line profiles
observed at the ESO in 1991 and 1992 was analysed in Paper I.
These results allowed to suggest the model of a gaseous
circumstellar envelope around HD163296 containing the region
of a non-stationary stellar wind and a more stable remote shell.
However, new observations performed at the LNA in 1992-1995
have shown that this model is too simplified.
Episodic appearance of the variable blue absorptions travelling
towards zero velocity (Figs. 1 (click here)-3 (click here)) gives evidence of a
boundary layer, arising from the interaction between an active wind and
an outer shell.
"Standing waves'' on the residual H profiles observed within one night
(Fig. 4 (click here)) are likely to be a common type of the rapid
variability in the Herbig Ae/Be stars. Beskrovnaya et al.
(1995) proposed an interpretation of this phenomenon on the basis
of numerical modeling of an envelope containing
a jet-like inhomogeneity (Pogodin 1990).
It has been found that the rotation of a long-lived jet around the star
results in the local intensity changes in the line
profile without noticeable positional shift on the timescale of a few hours.
This is connected with the fact that near the line of sight
the jet is moving along the surfaces of equal radial velocities.
If the lifetime of a gaseous jet exceeds its spin period, the following peculiarities in the line profile behaviour are expected:
Observational data available allow to perform the diagnostics
of circumstellar inhomogeneities on the basis of the first two
criteria.
The changes of the red and blue H emission wings in counter phase were detected
in two data sets: April, 1992 and April, 1995 (Figs. 1 (click here) and
3 (click here)). Figure 8 (click here)a presents the dependence of the red (
) and
blue (
) edges of the H emission line for all observing seasons.
The anti-correlation character of these variations is clearly seen.
It is remarkable,
that during 3 nights in July, 1992, this anti-correlation
is also noticeable, but with steeper slope, while in July, 1991
(5 nights) it was not observed at all (Fig. 8 (click here)b).
At the same time, the position of the central peak (
)
of the H and H emission lines detected in July, 1991 is very variable
(Fig. 9 (click here)). The observing series not being sufficiently long, it is
difficult to determine clearly the character of the variability.
It could be a single episode but cyclic sinusoidal-like
variations with a period of about 4-5 days cannot be discarded either.
In July, 1992 such a phenomenon was not observed.
Figure 7: a) Correlation between the radial velocities
of the blue () and red () limits of the H emission.
Only mean data for 1991 and 1992 seasons
are given for spectra obtained at the ESO.
b) The same as a), but for the individual spectra obtained
at the ESO in 1991 (filled triangles) and in 1992
(open circles)
Figure 8: Positions of emission peaks of H\
(open circles) and H (filled diamonds) for the data
obtained at the ESO in July, 1991 and July, 1992
Figure 9: Theoretical changes of the Q, U polarization
parameters calculated for a model of a rotating magnetized gaseous
condensation (
, B=120G, 
)
a) for the V band, and b) the I band
All these results make it possible to assume an episodic appearance
of azimuthal inhomogeneities in the gaseous envelope of HD163296.
The existence of inhomogeneities in the outer envelope of HD163296 is confirmed
by the results of polarimetric observations in July 1995.
Sinusoidal component of variability of the 
Stokes parameters
in V and R passbands can be caused by the orbital motion
of a dust or gas condensation scattering non-selectively the stellar light.
Similar dusty inhomogeneities in the form of planetesimals were
considered by Grinin et al. (1994).
If the trajectory of such a feature at the time of the observations
passed in the immediate vicinity of the star then the observed
pattern of the Stokes parameter variations could arise in the form of
a sinusoid.
However, the model involving a dust condensation cannot explain the non-sinusoidal character of the Q parameter variations in the I passband. Nevertheless, the observed phenomenon can be interpreted assuming that a rotating inhomogeneity arises in the gaseous media in the presence of magnetic field. In particular, such a feature might originate from the interaction of an inhomogeneous stellar wind with an outer shell.
Magnetic fields in the circumstellar environment of young stars are presently one of the most popular hypotheses for the explanation of kinematical peculiarities in their gaseous envelopes (Appenzeller 1994).
In order to test the influence of circumstellar magnetic fields of different configurations on the parameters of linear polarization observed from a star with azimuthally inhomogeneous envelope, Beskrovnaya & Pogodin (1997) calculated the Stokes parameters Q and U within a model of a rotating magnetized gaseous condensation for different values of the orbit inclination (i) and the magnetic field strength (B). The calculations were performed with allowance for the single electron scattering and the Faraday rotation of the polarization plane in the presence of magnetic field (Gnedin & Silant'ev 1984).
Figure 7 (click here) presents the results of calculations for the following
model parameters: the orbital period 
,
, and B=120G with the lines of force along the radius-vector.
Comparison of the theoretical curves with observed variations
(Fig. 6 (click here)) demonstrate that this model can explain phase shifts on
the Q, U-curves as well as the disappearance of the second harmonic of the
Q-parameter variations in the I-band, where the influence of magnetic
field is stronger.
The positional shifts of the blue absorption features
observed in the H profiles during consecutive nights
make it possible to reconstruct the dependence of
these variations on the wind velocity V (Fig. 10 (click here)).
This figure also presents theoretical tracks of the Keplerian
wind deceleration, calculated with the use of the following expression:
where G is the gravitational constant, 
is the stellar mass,
for hyperbolic motion (
is the terminal velocity),
and 
for elliptical motion
(
is the apoastron distance).
Figure 10: Observational deceleration of moving blueshifted
absorption features on Hprofile in comparison with
theoretical tracks of Keplerian deceleration for the following
parameters: a)
, b)
,
c)
, d)a=0, e)
,
and f)
As can be seen in Fig. 10 (click here), the observed values of 
follow a line which is less steep than the theoretical
tracks. This might be connected with:
, ranging from
0 up to 150
.
Taking into account that the width of the DNaI emission line was
about 
during all observing seasons, we can suspect
a kinematical stratification of the stellar wind near
HD163296. It seems probable that there exists
a region of the wind with low initial velocity which cannot leave the
system and is likely to support the outer shell. Interaction
of such a shell with the faster wind can result in spatial
inhomogeneities of the envelope, as well as in the loss
of angular momentum by some portion of the circumstellar gas
rotating around the star, leading to its subsequent infall onto the stellar
surface. This hypothesis can explain the appearance of local
features with positive radial velocities in the H, He I 5876
and DNaI line profiles.
Reconstructing a complete picture of the cyclic phenomena in the circumstellar envelope of HD163296 requires long continuous series of spectral and polarimetric observations, which are planned for our forthcoming investigation.
Acknowledgements
The research described in this paper was made possible in part by the ESO C&EE grant A-05-21 for the acquisition of the necessary computer equipment. The work of the Russian authors was sponsored by the INTAS under grant 93-2478-Ext. The Brazilian authors acknowledge CNPq and FAPEMIG for financial support.