Astron. Astrophys. 346, L69-L72 (1999)

4. Results and discussion

Fig. 3 shows the averaged spectra for each wavelength region. We identified the Si IV lines using the SUMER line lists given in Curdt et al. (1997) and Feldman et al. (1997), and fitted the data using Gaussian profiles, and obtained the central wavelengths of the lines and the fluxes (with 1 uncertainties) using the STARLINK DIPSO routine. Table 1 gives these observed values for each of the Si IV lines, as well as the derived line fluxes. By correlation with well-calibrated secondary standard reference spectra (Curdt 1999, in preparation), an uncertainty of 0.01 Å is assumed for the observed wavelengths. The 1 uncertainties given for the line fluxes are from the Gaussian fits. The radiometric calibration is estimated to introduce a further 22% uncertainty in the ratios to and a further 5% in to (Wilhelm et al. 1997). This estimate assumes that part of the systematic error cancels out in the ratios, leaving a 10% uncertainty at 800 Å and 1100 Å, and a 20% uncertainty at 1400 Å.

Fig. 3. Observed Si IV spectra, when intensities have been averaged over the full raster for each wavelength region. Arrows indicate the Si IV lines at 815.05, 818.13, 1122.49, 1128.33, 1393.76 and 1402.77 Å.

Table 1. Results of the Gaussian fits to the observed Si IV lines. The observed central wavelengths of the lines are given in the first column, each of which has an observed uncertainty of 0.01 Å by correlation with well-calibrated secondary standard reference spectra (Curdt 1999, in preparation). The line fluxes are given in the second column, with 1 error bars imposed by the Gaussian fits.

The observed wavelengths in Table 1 agree with the theoretical predictions to within 0.1 Å. The 818 and 1403 Å lines were well-defined and well-represented by single Gaussian fits. A faint Ni II line present in the blue wing of the 1394 Å line was separated using multi-Gaussian fits. The 1128 Å Si IV line lies on the red wing of the Fe III - transition at 1127.75 Å, and on the blue wing of the - Fe III transition at 1128.39 Å. We were able to obtain reasonable fits to these three lines, and saw no obvious evidence for resolvable blending in the residuals.

The 1122Å line lies within a C I multiplet between 1121.91 and 1123.09 Å (Curdt 1999). The Si IV line is blended with C I and Fe III transitions at 1122.52 Å, which could not be resolved. Multi-Gaussian fits were used to fit the lines, and we saw no evidence of further resolvable blending. Finally, a reasonable Gaussian fit was obtained to the 815 Å line, despite its weakness in our spectra.

Using the line fluxes in Table 1, observed values of the -sensitive ratios to were derived, which are summarised in Table 2, along with the electron temperatures derived from Fig. 2. In Table 3 we list the observed values of the line ratios , and , which are insensitive to both and .

Table 2. Observed logarithmic line ratios to (with associated 1 error bars) and derived values of . The upper limits to and are larger than the temperature range covered by our line ratio calculations, so only lower limits of are given for these ratios.

Table 3. The - and -insensitive ratios. The theoretical predictions for these ratios () are given in the second column, where the errors represent the small changes in these ratios over the and range of the solar transition region. The observed ratio values () are given with 1 error bars.

It can be seen from Table 3 that the observed value of is in good agreement with the theoretical prediction. The theoretical value falls within 3 of the observed value, which corresponds to an error of only 8% in the ratio. One can also see that pairs of ratios with the same line as the numerator, and the 1394 and 1403 Å lines as the denominators, produce very similar values of . It thus seems highly probable that the 1394 and 1403 Å lines are unblended, at least at the level of a few percent.

The observed ratio is 70% greater than the theoretical value, implying that the 815 Å line is blended by approximately 40%, although this figure could be as low as 20% to within the uncertainty. The and ratios produce values of in good agreement with the theoretical predictions for the temperature of maximum ionization fraction, K, and it therefore seems very likely that the 818 Å line is unblended to within its uncertainty (10% level).

The upper limits to the values given by and are higher than theory predicts. A blending of about 30% in the 815 Å line intensity (in agreement with that found from ) would decrease the derived value of by approximately 0.2 dex, bringing it into good agreement with the temperature found from and .

The observed ratio is too high, suggesting that the 1122 Å line is blended by approximately 55%. The values of and are also higher than theory suggests they should be. Correspondingly, the electron temperatures derived from and are much larger than expected, with a lower limit of compared to . This agrees with the observations of Keenan & Doyle (1988) and Feldman & Doschek (1977).

The and ratios give values of that are somewhat high. Adopting , this suggests that the 1128 Å line is blended by approximately 45%. This may be due to the presence of cold lines, which could not be ruled out as a source of blending by Doschek et al. (1997), who observed that their values of consistently gave slightly high values of .

Schmahl & Orral (1979) have argued that there is significant absorption in the Lyman continuum for and derive a minimum neutral hydrogen column density of 310¹⁷ atoms cm^-2. If such a column density was applicable to the quiet Sun, the ratios - corrected for the hydrogen absorption would indicate temperatures of = 5.1-5.2. The agreement between the and ratios (which involve the 818 Å line) and the theoretical predictions for the temperature of maximum ionization fraction seems to rule out their hypothesis that Lyman absorption has a significant effect on line intensities at wavelengths below 912 Å.

The above results provide observational support for the accuracy of the Si IV diagnostic calculations used in this work. We have included, for the first time, analysis of the Si IV emission lines around 800 Å. Ratios involving the 1394, 1403 and 818 Å Si IV lines are found to give good agreement with theoretical predictions for the temperature of the Si IV emitting region of the quiet solar plasma, and are recommended for use as -diagnostics at instrumental wavelength resolutions of 0.04 Å or smaller. The 1122, 1128 and 815 Å lines are considered to be blended, and their use should be treated with caution.

© European Southern Observatory (ESO) 1999

Online publication: June 17, 1999
helpdesk.link@springer.de