4. The "quiescent" spectrum of II Peg
4.1. SWP line fluxes (1200-2000 Å)
As it was previously discussed (see e.g. Doyle et al. 1989b), active RS CVn stars like II Peg flare frequently. These flares are detected in IUE SWP spectra by dramatic but temporary increases in the flux of all emission lines but they are especially obvious in those of the CIV doublet (1548/51Å). Before discussing other possible types of variability we will study the evidence for flaring which are present in the data.
In Fig. 1 we show the sequence of measured CIV fluxes as a function of the Julian date. The two consecutive flux values recorded on spectra SWP45531 and SWP45532 (at the left of the graph) are obviously very much higher than the mean of the remaining values. In fact, their fluxes are 18.75 and 11.7 respectively above the mean value computed excluding these two fluxes. The probability of getting two deviations of this size in normally distributed values is vanishingly small. Accordingly we treat these two flux values as flares.
Without SWP45531 and SWP45532 the new mean CIV flux is 5.28 10-13erg cm-2s-1 with a standard deviation of 1.71 10-13erg cm-2s-1. With these new values two further spectra stand out as having anomalously large values, i.e. those of SWP45586 and SWP45618 (the two solid squares near the centre of Fig. 1), each 5.4 from the new mean excluding them. Again it is reasonable to suspect that they too are flares and we will treat them as such. We propose to treat those flux values outside of these individual flares (open squares in Fig. 1) as "quiescent" values.
There has been controversy over the years about the ability of IUE to detect weak features. Essentially this revolves around the question of "flat fielding" the complicated detector system and the repeatability of that process. Ayres (1990) has concluded that, taking all sources of error into account, the ability of IUE to detect weak, unresolved emission lines at a level of 3 with its SWP LORES camera in spectra of 50-100 min exposure is 6 10-14erg cm-2 s-1. He furthermore pointed out that co-adding separate spectra increases the signal-to-noise of the resulting spectrum by , where N is the number of spectra.
With this in mind we have formed an average SWP spectrum using all the SWP spectra except those positively identified above as flares, i.e. N=19. Note that, before preparing this mean all of the spectra were adjusted in wavelength so that the singlet HeII 1640Å line in each spectrum was centred on the laboratory wavelength. This was done to overcome any shifts in wavelength scale caused by imperfect centering of the star on the spectrograph entrance aperture, which in IUE has an appreciable inclination to the direction perpendicular to the dispersion. The result is given in Fig. 2.
Referring to the results of Ayres (1990) we expect a 3 detection limit in the mean spectrum of 6 10-14/, 1.4 10-14erg cm-2 s-1. The mean spectrum shown in Fig. 2 was examined for features exceeding this limit. All wavelengths at which we believe lines are present are listed in Table 1 and indicated on Fig. 2. The flux in each of these features was derived by fitting gaussians of FWHM equal to the resolution of the spectrograph. The central wavelengths of these fits were then compared with the atlas of the solar spectrum in the same region Brekke (1993). These proposed identifications and the measured fluxes are also given in Table 1.
Note, however, that there is a considerable amount of residual flux between the lines identified as discrete features in the tables. It seems unlikely that this could be due to photospheric continuum from the late-type primary of the system. There is no evidence of breaks in its distribution at the wavelengths of the Si I continua (at 1524Å and 1682Å), making an origin therein unlikely as suggested by Phillips et al. (1992) for the flaring state of II Peg. These may be due to the superposition of many weaker lines which, at this resolution, cannot be individually recognised.
How justified are we in treating these resulting fluxes as "quiescent"? The CIV doublet is the best exposed line in the quiescent spectra. Its mean flux is 4.80 10-13erg cm-2 s-1 with a standard deviation of 0.92 10-13erg cm-2 s-1. Within the measurements constituting this mean, however, there is still a factor of 2.2 variation between the extreme values (2.9 10-13erg cm-2 s-1 and 6.3 10-13erg cm s-1, respectively). In view of Ayres' conclusions, it seems unlikely that this "non-flare" spectrum is truly quiescent.
4.2. LWP line fluxes (2000-3000 Å)
As with the SWP spectra we show in Fig. 3 the variation of the essentially chromospheric MgII k line. One value (solid square) is clearly higher than all of the others. This spectrum (LWP23854) was bracketted in time by the two SWP spectra showing the large CIV flare. The slightly hotter Fe II lines near 2620-30Å whose time behaviour is shown in Fig. 4, are also clearly elevated at this time. Accordingly we consider it to be affected by a flare too.
A second point (that derived from LWP23864) is also high in MgII k. This spectrum was one of a pair of LW spectra taken on JD2448872 about 2.5 hr apart. The following spectrum, LWP23865, shows no evidence of enhanced flux. There was an intervening SWP spectrum (SWP45543) of 100 min duration which shows no clear evidence of flare activity in CIV. The slightly hotter Fe II 2625.7Å line is not elevated above the mean value. CIV flares tend, however, to be more short-lived than those in chromospheric lines. Furthermore, the MgII h line is close to the overall mean value. Therefore, there seems to be no good reason to omit this spectrum from determining the overall mean "quiescent" flux in the LWP lines.
4.3. Rotational modulation of line fluxes
One of the most striking characteristic of the RS CVn stars as a class is the rotational modulation of their visible light due to the presence of large scale cool starspots (see e.g. Byrne 1992a or Byrne 1992b). Whether chromospheric line emission undergoes a similar modulation is an unresolved issue (see e.g. Rodono et al. 1987; Andrews et al. 1988; Doyle 1988 or Doyle et al. 1989a). Our data set, providing regular monitoring of line fluxes ranging from the chromosphere to the mid-transition region, is well suited for examining this issue.
In Figs. 5 and 6 we show some of the main chromospheric and transition region line fluxes plotted against rotational phase. It is clear in Fig. 6 that there is no evidence of systematic rotational modulation in the transition region lines from CII 1335/6Å (log T) to CIV 1548/51Å (log T). The chromospheric lines in Fig. 5, on the other hand, show hints of such a modulation, albeit with considerable scatter about the mean trend. FeII 2625.7Å, in particular, shows evidence of a modulation with a peak-to-peak amplitude of 30% about the mean, the maximum being about phase, 0.4 and minimum at about 0.6. A quantitative assesment of the influence of rotation on the line fluxes can be obtained by fitting the data with constant and sinusoidal functions and comparing the values of the fits. This analysis shows how the decrease in the total error is only significant (above 5%) in the low transition region lines of C II (17%) and O I (26%) compared to a 48% in the Mg II h line, 5%in the Mg II k line and 44% in the Fe II lines.
4.4. Comparison with previous results
II Peg is one of the most intensively monitored RS CVn systems with IUE . Therefore we have the opportunity to compare our results with those of previous epochs. We have searched in the literature for previous determinations of the mean quiescent fluxes of the stronger lines in both the SWP and LWP ranges. These are summarized in Table 2 and Fig. 7.
Table 2. Mean quiescent line fluxes at Earth for some of the strongest lines in the ultraviolet spectrum of II Peg in 1992 compared to published values at previous epochs.
Note first of all that our present fluxes are comparable to those derived by previous authors. However, our values are among the lowest recorded in the transition region lines, CII 1335/6Å, CIV 1548/51Å and HeII 1640Å but second highest in the MgII 2796/803Å doublet.
As part of an investigation of the long-term variability of the ultraviolet emission of II Pegasi, the C IV flux for each image in the IUE archive was measured. All spectra, including those corresponding to the 1992 observing campaign, were obtained from the IUE Final Archive at http://archive.stsci.edu/iue and extracted using NEWSIPS Nichols & Linsky (1996) so the fluxes presented here do not necessarily coincide with the fluxes derived by the original observers. The results are presented in Fig. 8 where flares identified in the bibliography and poor signal to noise spectra have not been included. Despite the scatter of data points within each observing run, hints of some kind of modulation can be identified as a decrease in the mean fluxes between JD2445000 and JD2447000 and a non-monotonic increase thereafter. The most recent observations in the plot correspond to JD2450371 (October 1996) and were taken with the Goddard High Resolution Spectrograph on board HST .
© European Southern Observatory (ESO) 2000
Online publication: March 17, 2000