ForumSpringerAstron. Astrophys.
ForumWhats NewSearchOrders

Astron. Astrophys. 329, 291-314 (1998)

Previous Section Next Section Title Page Table of Contents

15. Summary and discussion

15.1. Calibration of SERTS-89

The large number of lines and high signal-to-noise of the SERTS-89 spectrum mean that we can bring together the many density insensitive ratios identified over the previous sections and try to answer the questions raised in several places about the instrument calibration. The main concern has been with the intensities of the lines found above 400 Å and we proceed to analyse this as follows.

For two lines at wavelengths [FORMULA] and [FORMULA] with observed intensities [FORMULA] and [FORMULA], we denote the observed ratio of the two lines as [FORMULA], while we give the CHIANTI ratio as [FORMULA].

In Fig. 1 we plot the values of [FORMULA] for all density insensitive line pairs given in the previous sections except those for which we have identified problems due to blending or atomic physics inaccuracies. For each ratio, we calculate a weighting W such that, if [FORMULA], then [FORMULA]. Graphically, different weightings are represented by the symbols:


The partitioning of W is chosen to provide equal numbers of ratios in each group.

[FIGURE] Fig. 1. Comparison of observations with theory for density insensitive ratios, using line intensities published by Thomas & Neupert. The shaded symbols indicate ratios with smaller error bars - see the text for more details.
[FIGURE] Fig. 2. Same comparison as above but with line intensities adjusted by a factor [FORMULA].

[FIGURE] Fig. 3. Plot of the function [FORMULA] used to derive Fig. 2 (not on the same vertical scale).

In Fig. 2 we display the same observations/theory comparison but with a new calibration scale used. This new scale is specified by the function [FORMULA], which is such that if, on the original scale, a particular observed ratio was [FORMULA] then on the new scale the value will be


Accordingly the original value of [FORMULA] is mapped to


The particular [FORMULA] function used to generate Fig. 2 is shown in Fig. 3, and has been chosen to increase the intensities of the lines found above 400 Å by a factor of up to 2.

The suggestion in the Ne VI, Mg VIII and Mg VII sections that the SERTS-89 calibration may need revision over the 430-450 Å region is demonstrated by a comparison of Figs. 1 and 2: the filled-in symbols - corresponding to smaller error bars - clearly fall into better alignment in Fig. 2. The major exceptions to this are the Fe XIV 444.24/334.17 and C IV 419.72/384.17 ratios. A similar calibration adjustment was considered by Bhatia & Thomas (1997) based on their analysis of Mg VIII. They point out, however, that C IV and Mg IX ratios presented by Keenan et al. (1993a,1994a) seem to confirm the published SERTS relative calibration, whereas Fe XIV lines studied by Brickhouse et al. (1995) imply that it may actually be too low at the long wavelength end (thus requiring [FORMULA]). We have discussed the latter work in Sect. 13.6, and feel that it merely reflects errors in the Fe XIV atomic data; while for Mg IX we have used different atomic data to Keenan et al. (1994a) and find that it now produces the same discrepancy as the other Mg and Ne ions. Overall we feel that the insensitive ratios looked at here strongly hint at a problem with the SERTS-89 calibration above around 430 Å, but definitive resolution of this issue will only be possible with further observations of this wavelength region.

Brickhouse et al. (1995) have also claimed that second order lines appear to be lower in intensity with regard to first order lines by around 50% in the SERTS-89 catalogue. Our study confirms that some second order lines of Fe IX, Fe X, Fe XII, Fe XIII and Fe XIV do indeed show evidence of being too weak relative to first order lines. However, we note the following: the Fe XI 352.67/188.21 ratio implies that the second order 188.21 Å line is too strong relative to the first order 352.67 Å line by a factor of around 2; for Fe IX the problem with the 171.07 Å line has been seen in many previous spectra; in Sect. 13.5 we noted that while the second order Fe XIII lines are too low relative to the first order lines above 300 Å, they are consistent with the first order line seen at 251.94 Å; there are many difficulties with the Fe XIV lines and while a stronger 211.32 Å intensity would alleviate some of these difficulties (e.g., the 274.20/211.32 ratio) it would not affect others (e.g., the 270.52/274.20 ratio). We also note that most of the second order lines are 3p-3d transitions in the iron ions, and the density insensitive ratios involve comparisons with 3s-3p transitions. Thus another possibility is that the observations may simply reflect atomic physics inaccuracies in comparing these two types of transition. Therefore, we do not feel that any significant modification of the SERTS-89 second-order calibration is called for on the basis of our present analysis.

15.2. Revisions to the SERTS-89 catalogue

The revisions to the SERTS-89 catalogue suggested by the previous analysis are displayed in Tables 23, 24 and 25. Table 23 gives identifications both for lines reported as unidentified in the catalogue and for newly-identified components to blended lines. Table 24 contains new spectral line fits obtained since original publication of the spectrum, and Table 25 gives lines which have been mis-identified in Thomas & Neupert but for which there are no other viable candidates in the CHIANTI database.


Table 23. New identifications for reported lines in the SERTS-89 catalogue. A `b' indicates that the identified transition is a blended component of a line already listed in the spectrum, `m' that the line had been mistakenly attributed to some other ion. Identifications first suggested by Brickhouse et al. (1995) are marked with `B', that by Dwivedi et al. (1997) with `D', and that by Keenan et al. (1996) with `K'.


Table 24. New spectral line fits not in the original SERTS-89 catalogue. Those marked `r' replace the previously published fit, `K' are from Keenan et al. (1996), `D' are from Dwivedi et al. (1997). The feature at 283.700Å is discussed in Sect. 13.4. Intensities are given on the old absolute calibration scale.


Table 25. Lines for which the theoretical estimate of the line intensity is far less than the observed value, indicating an incorrect identification but with no other CHIANTI candidates. Those marked with `D' have also been noted as erroneous by Dwivedi et al. (1997).

15.3. Conclusions

Although much of this paper has been spent discussing discrepancies between theory and observation, in fact such cases are relatively few considering the vast amount of detailed, quantitative information that is now available in both the CHIANTI database and the published SERTS-89 spectrum. For the great majority of lines there is good to excellent agreement between the SERTS-89 measurements and CHIANTI predictions. One important new result is the ability to quantify the effects of unavoidable blends, and thus make a number of additional spectral lines available for spectroscopic analyses. Several new useful lines, suggested by theory, have subsequently been found and measured in the data. Numerous new line identifications have been made, and many others have been tested and confirmed (or sometimes rejected). We have shown that the technique of selecting lines for spectroscopic diagnostics only after validation of branching/insensitive ratios leads to a substantial reduction in the scatter of derived densities, producing much more definitive measurements of solar plasma parameters. Where discrepancies do exist, they point out potential mis-identifications, additional blending, possible calibration problems, data reduction errors, or areas where theoretical calculations need to be improved. Some of these are correctable, even further strengthening the value of both CHIANTI and SERTS data sets. And in those cases, the confrontation of theory against measurements often provides the direct clues that lead to the eventual solution of the problem. While some inevitable limitations remain, and future improvements are suggested in several areas, the excellent overall consistency between theory and observation demonstrated in this paper provides considerable confidence in the present accuracy of the atomic data in the CHIANTI database, and in the high quality of the SERTS-89 spectral measurements.

The identification of the many density insensitive line ratios given here and their verification through the SERTS-89 spectrum will be useful for future EUV missions as a means of checking the calibration of the instruments. In a reciprocal manner these future missions may also reveal more information about the problems discussed here. As an example, the Coronal Diagnostic Spectrometer (CDS) on the recently launched Solar and Heliospheric Observatory contains two EUV spectrometers - the grazing incidence (GIS) and the normal incidence (NIS); both described in more detail in Harrison et al. (1995). Although neither spectrometer has better spectral resolution than SERTS, they both have the ability to observe the same lines in a huge range of different conditions, giving information on the composition of blends and checking the range of density sensitivity of line ratios. For both instruments, the use of density insensitive line ratios is underway in attempt to assess the quality of the laboratory calibration - a preliminary investigation of the NIS calibration can be found in Landi et al. (1997).

In this paper we have have been primarily interested in comparing the CHIANTI atomic data with the SERTS-89 spectrum and as a by-product have predicted electron densities for a number of different ions. A future paper will utilize this information to infer information about the solar atmosphere.

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 1998

Online publication: November 24, 1997