3.1. The observed line profile variations
Due to the varying resolution of our spectra with wavelength and reported similarity of all observed HeII profiles we choose the 4686 line as optimal for illustration of LPVs observed in the optical spectrum of HD 4004.
In Fig. 1 all profiles of HeII 4686 are presented that we obtained during both runs. All individual spectra are plotted so every night a number of profiles is given, according to Table 1. In September 1994 only three separate nights were used for observations due to weather conditions but the night of September 28, 1994 resulted in a series of 21 spectra obtained within time interval of 0.307 day. A shorter sequence was obtained on September 16-16 spectra within 0.225 day. In November 1995 two 3-day sequences of spectra were obtained: on 6-8 and 17-19. The night of November 17 resulted in longest run of observations - 0.436 day but due to changing weather conditions only 7 spectra were gathered. Similar situation took place in case of November 18 and 19 - see Table 1.
As shown in Fig. 1 sequences of profiles of HeII 4686 obtained within one night show slow but noticeable variability. This is best illustrated by the September 28, 1994 plots. The development of different shapes of the HeII profile is seen through November spectra, where the intensity of the red wing of the profile is rising during following nights from November 6 to 8. At the same time the blue wing shows less variability. Another, slightly less evident sequence is seen in November 17-19. There we see how the blue wing is getting stronger starting from an almost symmetric round topped profile on November 17 to a flat topped one two days later. The red wing in that period seems to be constant.
Interesting enough the profile variations we observe are not completely random. Some of them appear again even in our limited sample. For example the profile of November 6 looks very much like that of September 28, 1994. Also the profiles of November 8 and 19 are similar and show a mirror reflection of each other. Both sequences of profile changes of November 1995 suggest that there is some recurrence in profile variations of HeII 4686 but the possible period is certainly longer than 3 days.
We find that similar profiles variations of HeII were observed by Morel et al. (1999). Similar behavior of emission line profiles of helium and nitrogen in HD 50896 was also reported by Smith & Willis (1994).
More detailed information on hourly profile variations of HeII 4686 is provided in Fig. 2 and Fig. 3. Top panel of Fig. 2 presents all 16 profiles of that line observed during 5.5 hours of monitoring on September 16, 1994. The middle panel shows a montage of deviations present in individual profiles obtained by subtracting from every spectrum the first one taken that night (the spectra are artificially shifted upwards in the intensity scale by 0.1, as indicated by the arrow). A slow development of profile variations in time is clearly visible. In the bottom panel of Fig. 2 the Temporal Variance Spectrum, calculated for the September 16 spectra is presented. The dotted line is the contour of statistical significance of p=1.
In Fig. 3 the same data for the night of September 28, 1994 are shown. Due to longer monitoring (21 spectra in 8 hours) the development of profile variations is more evident. The TVS (bottom panel) confirms that the amplitude of variations is much larger in these data. Interesting enough our data resolve the observed profile variations into smaller components that appear and develop at fixed projected velocity. Our data show no migration of a profile deviation (change of projected velocity in time). The profile variations of HeII 4686 as observed on September 28, 1994 appear to be a result of varying intensity of 3 narrow components (300-700 km s-1) located at fixed positions relative to line center.
3.2. The observed spectrum and Temporal Variance Spectrum (TVS) analysis - general view
The mean spectra obtained within both observing runs as well as temporal variance spectra - TVS (Fullerton, Gies and Bolton 1996) are presented in Fig. 4. We can see a well developed emission line spectrum in the upper panels and TVS 1 (lower panels), which in case of all observed helium lines are very similar to that described by Niedzielski (2000) for HeII 5411. It shows that the variability in the case of HeII lines takes place mainly in the central part of observed profiles. Detectable profile variations are observed however already at projected velocities of over 1500 km s-1. The equivalent widths and line widths (FWHM) for both runs (mean values) are presented in Table 2.
Table 2. Measurements (mean values within an observing run) for the observed HeII lines.
It appears also that although the general shape of the spectrum is similar, all observed profiles of emission lines (averaged over a run) are different in both runs. However, all observed HeII lines show identical (on average) profiles within one run. This surprising finding was first reported by Niedzielski (1996). Morel et al. (1999) confirmed such correlation in the optical spectrum of HD 4004 through construction of Spearman rank-order correlation matrices for HeII 5411 with HeII 4686 and 4860. Morel et al. (1999a) found also the same behavior in the optical spectra of HD 191765. This property of helium lines is confirmed by the TVS plotted in lower panel of Fig. 4 where we can see that also the TVS for all HeII lines appear similar within one run.
In Table 3 we present our measurements of central intensities Ic (in continuum level units), amplitude of TVS and the ratio of . This ratio shows some scatter but it's value for all observed here HeII lines is very similar and contrary to McCandliss (1992) we report that all HeII vary with the same relative amplitude in HD 4004. Since according to models of WR envelopes (Hillier 1987) the helium lines observed here are effectively formed in distant layers of the envelope (see also Niedzielski 1994) we can also state that we do not see any development of the amplitude of the line profile disturbing agent outward in the envelope. On contrary our observations suggest that it is constant, at least in the zone of the envelope where the HeII lines are formed. It is true, however, that the amount of variability, or the amplitude of TVS is different during different observing runs. In our case the data of November 1995 show much larger (50) variations.
Table 3. Central intensities (Ic) and the amplitude of TVS for the observed emission lines in both runs.
Due to relatively low resolution of presented data we are not in position to discuss in detail (as in Niedzielski 2000) possible similarities or differences in the TVS shape of lines of other ions covered by our spectra. We are able to identify TVS variations above the p=1 level only for the NIII-V blend at around 4615 and the NV line at 4945. In case of the former one the amplitude of the TVS variations as presented in Table 3 is low, factor of 2 lower than that of HeII lines. It is however difficult to interpret because of problems with identification of the most important component of the blend and possible influence of the P Cygni absorption of NV at 4606. The former one, NV 4945 varies by approximately the same amount as HeII lines in September 1994 and a bit less in November 1995. The shape of the TVS for that line appears similar to helium lines in both runs but again, due to low resolution we are not able to compare them in detail.
3.3. The sub-peaks evolution
We have mentioned already that not only the mean emission line profiles but also the shape of TVS within a run is identical for all HeII lines. As presented in Fig. 4 the TVS for HeII lines reproduces exactly the same features what is best seen in case of 4686 and 5411 which are the strongest (when comparing details displayed on TVS of different lines one has to remember that in our prismatic spectra the resolution is decreasing with wavelength).
The measurements of projected velocities of individual sub-peaks (or valleys) present in TVS of lines under consideration are are presented in Table 4. We note an interesting property of those features, most clearly present in case of larger projected velocities. The projected velocity of a given, well identified sub-peak, present in all studied lines is usually highest in the strongest and widest lines and is decreasing in less intense, narrower or higher members of given series, lines. This is most readily seen when comparing 4686 and 4542, both recorded on our spectra with similar resolution but very different in widths.
Table 4. Projected velocities (in km s-1) of five strongest individual distinguished extrema (peaks or valley) in TVS over observed HeII lines. Their similarity within a run is noticeable. The peaks are not the same in both runs so any comparison between runs is not possible.
To study this effect in more detail we have measured the projected velocities of individual, well identified features which were present in "deformation patterns" of all HeII lines on the same spectrum. Such "deformation patterns," similar to that described in case of HD 191765 by Vreux et al. (1992) are obtained by subtracting the mean spectrum over a run from every spectrum (see also Fig. 2 and Fig. 3, middle panels). This way we have explored the fact that all HeII lines have very similar shape at the same time (on one spectrum) and therefore an individual sub-peak in deformation pattern can easily be identified in all profiles.
Our measurements confirm similarity in observed line profiles. The correlation between positions of individual features in different HeII lines is very high. For example in the case of 4686 and 4542 as observed in November 1995 this correlation reaches r=0.991. Similar situation is present in case of other line-pairs. However, projected velocities observed in different HeII lines are not identical. In Fig. 5 we plot the difference of the projected velocities of the same four sub-peaks identified in two helium lines 4542 and 4686 in November 1995 run as function of the projected velocity of the same sub-peak in 4542. One can easily note that a shift is present in the projected velocity, rising rapidly on both sides of the profile center (zero velocity). The projected velocity of sub-peak present in profile of 4686 (its absolute value) is always larger than in 4542.
The projected velocities do not scale in different HeII lines as the line widths (FWHM). The projected velocities measured in sub-peaks of 4542 are only 6 lower than in 4686 (best fit to data presented in Fig. 5) while the difference in FWHM suggest much larger difference of 20-25. This difference suggest that general line profiles and small profile deviations are not formed in the same conditions: the general wind law is different from that governing motion of the disturbing factor seen exposed on emission lines. This further suggests presence of a two-component wind in HD 4004.
© European Southern Observatory (ESO) 2000
Online publication: October 30, 19100