It is well known that the completeness and the accuracy of the atomic and molecular line data are of paramount importance for the reliability of any synthetic spectrum (see, for instance, Bell et al. 1994 and Kurucz 1995). To improve the reliability of the parameters governing the strength and the shape of the spectral lines, we compared the observed solar spectrum (Kurucz 1991) with a synthetic one. Empirical adjustments to line data were determined by applying a trial and error procedure as in Peterson et al. (1993). Our corrections, although small with respect to the uncertainties attributed to the line data parameters (Kurucz 1995), significantly improved the ability of the synthetic spectrum to mimic the observed one.
The model used for the comparison has the following atmospheric parameters: K, log g=4.43770, and element abundance from Anders & Grevesse (1989). For the micro- and macro turbulent velocities, the values 1 and 1.5km s-1, respectively, were adopted from Thévenin (1989). The synthetic spectrum was computed at the spectral resolution of , so as to match the resolution of the observed spectrum.
The atomic and molecular line data were extracted from the updated line list compiled by Kurucz. The working line list includes more than 45 000 lines in the 4850-5400Å spectral interval. Molecular features included in the computations are CN, C2, MgH, SiH and CH and account for the majority of the lines. Out of the working line list, about 1 500 lines were carefully investigated and about 2000 modifications on their parameters (log gf values and van der Waals damping constants) were applied. The improvement achieved after the modifications is testified by the decrease in the rms of the differences between the computed and the observed spectra (both normalized to the continuum and rebinned to the constant step Å): by using the original line list, the rms value was 0.071 while is 0.049 after our modifications.
The distribution of the residuals (computed - observed residual intensities) is represented in Figs. 1 (click here)a and b: the distribution obtained after the corrections clearly shows an higher concentration in the central bin (Fig. 1 (click here)a: 18 903 instead of 16 157 data points over a total of 27 252). Figure 1 (click here)b shows that the negative wing of the distribution is significantly reduced due to the removal of strong predicted lines which do not correspond to observed features; unfortunately, the improvement in the positive wing is less significant due to the impossibility of solving the problem of unidentified lines (see, for instance, the line at Å in Fig. 2 (click here)).
Figures 2 (click here) and 3 (click here) illustrate in more detail the kind of modifications we performed by showing the region around a strong magnesium feature at 5172.684Å and that one containing the iron feature at 5328.038Å, respectively. In both figures, the thick solid line corresponds to the observed solar central intensity spectrum, the dotted line indicates the synthetic spectrum computed without modifying the line parameters and the thin solid line shows the spectrum computed after the modifications. The better agreement with the observations is the result of modifying the van der Waals damping constants and/or the log gf values in the line data of Kurucz (1993b). Moreover, for some lines, small adjustments in wavelength were also required (see Fig. 3 (click here)). For example, in order to match the observed profile with the theoretical one, the of the central MgI line in Fig. 2 (click here) was corrected by subtracting 0.18 from the original value of -7.12. For the case of the iron feature at 5328Å (Fig. 3 (click here)), its in Kurucz's list (-7.87) was increased by +0.30.
With the updated line list we computed the grid of high resolution synthetic spectra described in the next section.
Figure 1: a and b) Histogram of the residuals (computed - observed) for the synthetic spectra computed before (thin line) and after (thick line) adjusting the line parameters: a) the distribution in absolute numbers evidentiates the enhancement of the central bin, b) the logarithmic scale enhances the differences in the wings of the distribution
Figure 2: Magnesium feature in the interval 5171-5174 Å. The thick solid line corresponds to the observed spectrum and the thin solid and dotted lines indicate the theoretical solar intensity spectra computed with and without adjusting the line parameters, respectively
Figure 3: Iron and Chromium features in the region 5327-5331 Å (symbols as in Fig. 2 (click here))