Bad seeing conditions at my observation site contribute substantially to line broadening and the contribution of coma to the line shape is probably rather small, so that the lines should be approximated by a Gaussian model (however, this has to be proved by a line shape analysis; some discussions on this topic can be found here: https://www.cloudyni...2nd-derivative/).
Finally I have found "Fityk", that is a quite comfortable peak fitting program https://fityk.nieto.pl/
One big advantage: It can read .dat-files created by RSpec or VSpec. Below is a SA-200 spectrum of NGC 7662 (blue snowball) that was recorded in the converging beam setup:
The emission lines can be, indeed, fitted using a Gaussian model:
Of course, the fit isn't perfect, but the overlapping [OIII] lines are deconvoluted satisfactory (the theorectical line ratio of the non-resolved [OIII] lines is about 1:3, which compares quite well with the result of the deconvolution).
The constraints of this approach are well known:
-the deconvolution is a so-called ill-posed problem, which means that many different functions solve a convolution equation within bounds of experimental data
-deconvolution delivers thus only meaningful results when the actual number of componets of the unresolved peak is known a priori and the fwhm of the lines due to the response function of the system is known (which can be determined from the peak shape of a resolved peak)
-experimental spectral data can thus only be verfified, but the extraction of novel information is limited
Conclusion: It could be shown for my setup that the broadening of emission lines in a star analyser spectrum, that was recorded in a converging beam, is dominated by the seeing rather than by the spectral coma. Deconvolution is of limited value but a nice exercise for rainy days and cloudy nights.
Edited by mwr, 01 February 2020 - 05:28 AM.