The DEM analysis requires the use of density insensitive lines for determining the Correction Function, so that the measurements of are not dependent on .For this reason, we have used only density-insensitive or weakly density-sensitive lines. For the latter, theoretical intensities have been calculated assuming cm-3 in active Sun and cm-3 in quiet Sun; these values were determined from the data using the Fe XIII 203.80/202.04 line ratio.
The line selection for the DEM study has been carried out with the additional requirement that line ghosts should not play a major role in the determination of the observed intensity of the candidate lines. For this reason all the weak lines which are likely to be blended with ghosts generated by strong lines have been rejected, and strong lines producing ghosts have been omitted from this study. There are a few exceptions to the latter requirement, for lines whose ghosts could be traced back and corrected and lines whose ghosts are too weak to alter significantly () the total intensity of the lines.
The DEM study allows the direct comparison of the intensities of lines emitted by different ions formed at similar and observed in different GIS channels. The values of the DEM provided by such lines are expected to be identical within experimental uncertainties, thus allowing a first check on the relative intensity calibration of the different GIS channels.
The lines used in the DEM analysis are listed in Table 1, indicated by "dem" flag and ordered in temperature. The lines flagged with "Ne" are density-sensitive and can be used for estimating electron densities.
As Table 1 shows, several cool lines are present (mostly in the GIS 4 channel) and may be used in both quiet Sun and active region spectra, to constrain the DEM for temperatures as low as 104.6. Since opacity may play an important role in determining the observed intensities of some of the coolest lines used in the present study (N II, N III - Brooks et al. 1998b), the low temperature tail of the DEM curve should always be treated with caution. Lines from N, O, Ne, Na, Mg, Al, Si, S and Fe were used for this study. This allowed a check on element abundances, in the sense that lines of different elements but emitted at the same temperature should have the same DEM values, if the relative abundances applied were correct.
The presence of the hot Fe XV line at 284.160 Å ensures that the quiet Sun DEM can be studied for temperatures as high as K.
The resulting quiet Sun DEM curve is displayed in Fig. 6 (top). The theoretical and experimental intensities are reported in Tables 2 and 3.
|Figure 6: Top) DEM for the quiet Sun region: the mean is 3.6; bottom) DEM for the active Sun region: the mean is 2.8|
The O, N and Ne lines, observed in the GIS 3 and GIS 4 channels, are in relatively good agreement between themselves, with the only exception of the cool N II and O II lines, which may be affected by some opacity problem. The Ne IV multiplet agrees within 15% with N IV and O V lines formed at similar , as well as Ne V and O V, despite some difference in their . All this suggests that both the relative GIS 3 - GIS 4 intensity calibration and the abundances of these three elements are approximately correct.
Some problems arise between the aforementioned N, O and Ne lines and Mg lines. Mg VI and Mg VII lines values are very close to those of Ne V, VI and VII, but the DEM value they provide always disagree, the Mg theoretical intensities being overestimated by a factor relatively to the Neon values. Since the Mg VI, VII and Ne VI, VII lines are observed in the GIS 3 spectral range, any intercalibration problem is not expected. Moreover, the slight density-dependence of the Mg VI and VII lines is not able to account for the discrepancy. For these reasons, and remembering that the O, N and Ne abundances are in agreement, the abundances of these three element were increased by a factor 3 in the DEM analysis, bringing relative element abundances closer to photospheric values. It is worth noting that N, O and Ne have high First Ionization Potential (FIP) while all the other elements used for the quiet Sun DEM study are low-FIP elements. The disagreement found between the Feldman (1992) coronal abundances and those required by the present dataset suggests a much reduced FIP effect (see Haisch et al. 1996) for this quiet Sun observation.
Ne VIII deserves some comment. The lines of this ion, observed in GIS 4, are formed at the same as Mg VIII, and unlike the other Ne ions they are in relatively good agreement with the Mg lines with the Feldman (1992) abundances. The correction factor for the Ne, N and O elements required by the other Ne ions causes the Ne VIII lines to disagree with the other Mg lines by a similar amount; the cause for this peculiar behaviour is still not understood. Ghosts are unlikely to be the cause of this discrepancy since the Ne VIII ghosts can be traced back and corrected; possible intensity calibration problems between GIS 4 and the other channels are ruled out by the O, N and Ne lines. It is important to note that the same problem with Ne VIII is found in the active Sun DEM analysis, where no need to change the Feldman (1992) abundances is found (see Sect. 6.1.3).
Mg VIII and Mg IX lines have similar to Fe IX, Fe X, Si VII and Al X lines and the transitions from several ions of these elements are in good agreement. Since these ions are observed in the GIS 1, GIS 2 and GIS 3 channels, their agreement both confirms that the relative intensity calibration of these channels have no problems and that the adopted relative abundances of the Mg, Al, Si and Fe elements should be approximately correct.
It is also important to note that the Mg VIII intercombination transition at 782.34 Å is in agreement with the other Mg VIII lines and this further suggests that the GIS 2, GIS 3 and GIS 4 relative intensity calibration is approximately correct.
In the present active region spectrum no evidence is found for a need to change the adopted element abundances. Most of the lines used for the active region DEM determination show agreement to better than 20%, with only a few exceptions due to blending and uncertainties in the fitting of the background.
|Figure 7: L-functions for the Mg VII lines observed in the CDS GIS-2 and GIS-3 detectors. All the lines meet at cm-3|
As in the quiet Sun spectrum, some more serious problems seem to arise with the GIS 4 Ne VIII lines whose is around 106 K. In spite of the agreement between Ne and Mg abundances the Ne VIII GIS 4 doublet observed at 770.4 Å and 779.5 Å never agree with the Mg VIII and Mg IX transitions, whose is very similar to the Ne VIII values. The theoretical intensity of the Neon lines seem to be overestimated by a factor 2.5 relatively to those of the Mg lines - a factor very similar to the value found in the quiet Sun spectrum. These Ne VIII transitions have been corrected for ghosting. Ne VIII is observed in GIS 4, while the other Mg lines are observed in GIS 3, and this could suggest some problem in the relative calibration of the two channels. Nevertheless, the other lines observed in GIS 3 and GIS 4 agree within the errors (e.g. the Mg IX 749.55 Å line with the other GIS 3 Mg IX lines), indicating that no gross intercalibration correction is required between these two channels. Therefore the case of the Ne VIII active region lines is isolated, and does not provide any definitive conclusion about the GIS 4 intensity calibration relative to the other channels.
The Fe XV, XVI and S XIV lines, observed in GIS 1, GIS 2 and GIS 3, are in good agreement. This is a further indication that no correction is required to the GIS 1, GIS 2 and GIS 3 relative intensity calibration.
The Mg VII active region L-functions are displayed in Fig. 7, showing that all the L-functions meet for cm-3, in agreement with the density value obtained with the Fe XIII lines. As the observed lines come from the GIS 2 and GIS 3 detectors, this is an indication that their relative calibration needs no correction. The Mg VIII quiet Sun L-functions are displayed in Fig. 8, indicating that the electron density is between 108 and 1012 cm-3, which is consistent with the value used in the DEM analysis. An overall agreement is found between the lines, showing that all the displayed L-functions are equal within the experimental uncertainties in the 108 and 1012 cm-3 density range.
For each of these ions, the L-functions of most of the observed lines are in agreement among themselves better than 20%. This means that no gross correction is required by the GIS neither between different detectors nor as a function of wavelength within each of the detectors. Disagreements between the few remaining lines can be explained by atomic physics problem or blending. No systematic trend was observed in these discrepancies which could point to some problem in the relative calibration of the four GIS detectors; moreover problems with these lines have already been reported in literature (Young et al. 1998 and references therein) and the similarity of our results with those of Young et al. (1998) is in turn a further indication that the GIS intensity calibration is approximately correct.
Only the GIS 3 and GIS 4 channels presented second order spectral lines, with the GIS 3 second order bandpass partially overlapping GIS 1 (196 - 246 Å), and with GIS 4 overlapping a small part of the GIS 2 channel (329 - 392 Å). Since most of the second order lines visible in GIS 3 and GIS 4 belong to ions formed at high temperatures, the second order intensity calibration has been assessed using the active region spectrum, since in the quiet Sun very few second order lines are visible. It is important to note, however, that whenever observing hot plasma, the contribution of second order lines in the two channels is non-negligible. In the pre-flight calibration (Bromage et al. 1996) no estimates of second order efficiencies were obtained.
Second order lines are sometimes either blended or affected by the ghosting problem and therefore only few of them can be used for intensity calibration studies with some confidence. The list of the lines used for second order calibration is given in Table 4. Some of the GIS 3 and GIS 4 second order lines are observed also in GIS 1 and GIS 2 respectively and this allows a direct comparison of line intensities, and the measurement of the second to first order relative calibration is straightforward. In the remainder of the GIS 3 and GIS 4 spectral ranges measurements are made using the L-function method of line analysis described in Sect. 2.1.
|Figure 9: Top) GIS 3 second order intensity calibration correction factor. Bottom) GIS 4 second order intensity calibration correction factor (see text)|
In the present work we are concerned with the measurement of the correction factor C so that the true intensity of second order lines can be obtained from the calibrated intensity (calculated using the first order efficiencies) as
The resulting correction curve has been fitted with a best fit second order polynomial:
and can be applied to second order lines between 400 Å and 465 Å. This GIS 3 second order calibration curve is displayed in Fig. 9 (top).
The resulting GIS 4 second order calibration curve is displayed in Fig. 9 (bottom). The correction curve has been fitted with a best fit second order polynomial:
and can be applied to second order lines between 665 Å and 740 Å.
Copyright The European Southern Observatory (ESO)