5. Investigation of variable components in the H profile
We first address the issue of the large asymmetry of the mean observed emission profiles (see Sect. 3). The expected FWHM of a spectral line in the H region from a star rotating at 69 km s-1 in the absence of limb-darkening is at least 2.61 Å. This is a reasonable approximation for a chromospheric line like H since the bulk of the observed flux arises over a thin layer of uniform emissivity with depth. The observed mean FWHM is 2.3 Å, substantially less. The H line, which is formed at a deeper layer in the chromosphere, shows a similar asymmetry. This would suggest the effect is caused by a circumstellar source rather than by surface inhomogeneities. Otherwise more disagreement would be expected between asymmetries in H and H, since the limb-darkening coefficients are different in both lines. The time-averaged H profiles should have a FWHM at least as big as the Doppler width, so it seems reasonable to assume that they are absorbed in the red over a substantial part of the star. On the other hand, if the absorption was localized on the stellar surface it should move on the profile with rotation. The analysis presented in Sect. 4, however, suggests its effect is not exactly uniform with phase.
In order to investigate possible variations of this asymmetry we examined in detail the H profiles by comparison to a reference spectrum. We first made an average of several H line spectra taken on the first night, selected so that they were not affected by significant variability. Their phases are marked in Fig. 3 (left panel ) as interval G. Next the blue wing of the average spectrum was reflected to the red with respect to the H rest wavelength. The resulting profile has a FWHM of 2.6 Å, which agrees well with the value predicted from the vsini. It seems reasonable therefore to assume that such a profile would correspond to that of an uniform chromosphere for this star. We used it as a reference to investigate the nature of the variations seen in Figs. 3 and 4.
The individual spectra were divided by the reference template and the resulting ratio profiles stored in a 2-dimensional array for each night. Fig. 5 shows the results as grey-scale images, which will be referred to as dynamic ratio spectra . Each line of an image corresponds to a single ratio profile with time running from bottom to top.
As can be seen in Fig. 5 (left ), the general appearance of the H profiles on August 4 is dominated by a strong absorption in the red at all phases, being especially dense during 0.0-0.4, i.e. intervals A-F in Fig. 3. This effect is slightly attenuated at the end of the night (interval G in Fig. 3). The general trend of the variations is complicated by the sudden appearance of variable emission in the blue wing with velocities of the order of -200 km s-1 (0-0.2 and 0.6). Its behaviour will be discussed later in more detail.
Evidence for red absorption was also found on August 5 (Fig. 5, right ) but its appearance is more variable with phase. The dominant feature is, however, a strong emission transient that affects mainly the blue wing and the centre of the profile during the first half of the night. This event starts as a dramatic enhancement of the blue wing at 0.5, extending out to -220 km s-1. Its effects are visible in Fig. 3 as a large increase in both EW and FWHM (interval B), with velocities that are significantly blueshifted. Emission moves progressively to the red as the blue wing is seen to fade between phases 0.6 0.8. A dark feature develops at a velocity close to the maximum Doppler blueshift associated to rotational velocity (0.78) and moves redward until it merges with the absorption feature in the red (0.95).
© European Southern Observatory (ESO) 1999
Online publication: December 4, 1998