6. Discussion and conclusions
Several studies using SUMER data have been conducted with the aim of determining the Doppler shift pattern of solar emission lines (e.g., Brekke et al. 1997; Warren et al. 1997; Chae et al. 1998; Brynildsen et al. 1998; Peter 1998; Hassler et al. 1999; Peter & Judge 1999; Stucki et al. 1999). In many cases, the Ne VIII (770) line played an important role in these studies. However, it became increasingly clear that a consistent interpretation of the results met with difficulties under the assumption of a vacuum wavelength of the Ne VIII line at 770.409 Å. This motivated our attempt to determine an accurate rest wavelength for this line. We find = (770.428 0.003) Å, where the uncertainty estimate is given at the 1 level. The wavelength uncertainty translates into a velocity uncertainty of 1.2 km s-1. Note, however, that the LOS speeds relative to the baseline in the corona can be determined with higher precision. Our value is in excellent agreement with the results of Peter (1998) who used a SUMER full Sun Ne VIII image to estimate = (770.42 0.01) Å, as well as Peter & Judge (1999) who analysed SUMER spectra observed during a 360o roll manoeuvre of SOHO and obtained = (770.428 0.007) Å. The new value is also consistent with all rest wavelengths reported in the literature for Ne VIII , if the uncertainty assessment of Sect. 2 is taken into account. This can be seen from Table 2 where the rest wavelengths available are compiled with their uncertainty margins.
Table 2. Comparison of the rest wavelength determinations for the transition Ne VIII ( - )
Assuming the new rest wavelength, a net blue shift for Ne VIII would prevail for the quiet Sun in the reports of Brekke et al. (1997) and Chae et al. (1998). In general, all LOS speeds deduced from this line would be shifted by an increment of 7.4 km s-1 towards the outflow direction. At this stage, we cannot comment on other lines, more than what has been stated in relation to Figs. 4 and 5. The data obtained for this study will allow us to deduce additional information on the C IV lines in a future publication. But the Mg X line, in particular, will also have to be re-assessed before the issue of the average Doppler shifts of solar emission lines can be settled. In this context, it should be mentioned that Peter & Judge (1999) came to the conclusion that the rest wavelength of Mg X (625) also must be offset by 18 mÅ or 8.6 km s-1, and Stucki et al. (1999) find blue shifts for Ne VIII and Fe XII in coronal holes of -4 km s-1 and -5 km s-1, respectively. For the Fe XII (1349) line, a blue shift of -6 km s-1 at disk centre with respect to the limb was reported by Sandlin et al. (1977).
With a mean projection angle of 70o in the coronal hole and the assumption of an outflow direction according to the direction seen in Ne VIII near the limb (cf., Fig. 2 of Wilhelm et al. 1998), which is off the radial direction, we get an average outflow speed of 9 km s-1 from the LOS velocity in Table 1.
The coronal hole profile of Ne VIII exhibits a significant shoulder on the red slope (cf., Fig. 7). Even after deducting the Si I contribution, a deviation from a Gaussian shape is evident in Fig. 9b. With a multi-Gauss fit, we have found that the Ne VIII line is centred at 770.414 Å. Note that the coronal hole data set used in this figure is not identical with that of Fig. 8 and, consequently, the small difference in the corresponding wavelengths of + 2 mÅ does not constitute a discrepancy. The red-wing line is positioned at 1541.040 Å (in first order). The only line observed in solar spectra near this position is the H2 line at 1541.063 Å (Sandlin et al. 1986). We are not concerned with the line identification, but have to demonstrate that such a blend does not interfere with our arguments and the centroid method employed in Sect. 5.1. To this end, we have determined the wavelength of Ne VIII in the coronal hole before and after deduction of the Si I and H2 (?) contributions and find differences of less than our uncertainty margin of 3 mÅ. In any case, we would estimate the Ne VIII line position in the coronal hole to be too "red" and, consequently, our coronal outflow speed might be underestimated by 1 km s-1. It should also be noted that this complication is not present in the quiet Sun regions and, in particular, in the off-limb corona, where the Ne VIII (770) rest wavelength is obtained (see Fig. 9a and c).
It would be highly desirable if laboratory measurements and theoretical calculations could be conducted to establish state-of-the-art rest wavelengths for Ne VIII and other lines in spectra of highly ionized species, such as Mg X and Fe XII , for comparison with solar wavelength determinations.
© European Southern Observatory (ESO) 1999
Online publication: May 6, 1999