AUTOMATED-ANALYSIS OF SLITLESS SPECTRA - STARS

We report on the stellar spectral classification that can be achieved with very low-dispersion spectroscopy. We apply several methods to slitless spectra taken from the automated grens plates analysis project undertaken at Laval University. Broad-band photometry and template matching have been used to probe the global spectral information, while intermediate-band photometry has been applied to examine local features. Multivariate analysis (MVA) has also been carried out as an objective tool to complement the study. B and V photoelectric data and the Gunn & Stryker library of stellar spectra have been used as reference material to calibrate and/or compare the data. It appears that an accuracy in B - V of approximately 0.1 mag at B approximately 19 for stars bluer than B - V = 0.8 and approximately 0.2 mag for redder stars can be reached by probing the large-scale spectral distribution. DDO-like intermediate-band results show that grens data behave similarly to the Gunn & Stryker library of standard stars for many indices and so the classification philosophy behind such a system can be applied to grens data. Results from MVA are coherent with the methods of classification mentioned above and give a powerful and objective means of cataloguing spectra of the same spectral shape. Rare objects can be extracted from the database by many of the methods explored.