The main field of research is: automated quantitative spectral classification
of stars with its applications to studies of Galactic stellar populations.
-
The current project is:
creating the quantitative spectral classification system based on
spectra of optimal resolution.
The project aims at developing an optimal system of quantitative spectral
classification of stars using a combination: Schmidt type telescope+
Objective prism+CCD detector.
Spectral classification based on low resolution objective prism spectra is a
powerful tool in astronomy providing the main physical parameters for
statistical samples of stars. The majority of the available classification
data has been obtained visually in the Morgan and Keenan (MK) system on
photographic emulsions (the wavelength range 3900--4900 A, about
2A resolution), many thousands of stars have been classified by now.
Some authors also developed a number of automated quantitative spectral classification
systems based on photographic emulsions (among others, Malyuto, Shvelidze, 1989;
Malyuto, Shvelidze, 1994, the criteria evaluation method was used).
Our more recent studies (1996-99) in spectral classification were based on
some available photoelectric spectral libraries (obtained with the use of
photoelectric scanners,
diod-array and CCD detector systems). Wider spectral regions
and much lower resolutions (tenths of A) can be used in classification now.
In the frames of the same criteria evaluation
approach in classification we analysed some additional criteria and found
possibility to combine a number of available photoelectrical spectral libraries
together (Malyuto, Oestreicher, Schmidt-Kaler, 1997a; Malyuto, Schmidt-Kaler,
1997b; Malyuto, Schmidt-Kaler, 1999).
Although the criteria evaluation approach is workable and have some advantages,
other recent approaches (the metric distance method, Artificial Neural Networks,
the multidimensional simplex routine and others) are more attractive because they can
exhaust the information contained in the spectrum.
Among them, we have selected the perturbation method (invented by Cayrel) which
promises a good classification
accuracy; it is based on direct comparison of synthetic spectra with
observations.
Our present idea is to simulate synthetic Kurucz spectra of different resolutions,
to convolve them with random numbers (for producing 'observed' spectra), to
classify the 'observed' stars by means of the
perturbation method and to estimate the classification accuracies at different
resolutions. The results may be used
to choose the resolution optimal in classification for galactic studies.
Under optimal resolution we mean the spectral resolution combining
reliable classification with good penetrating ability to fainter
stars to provide deep surveys of stellar populations in the Galaxy. We
should use as low spectral resolution as possible (to reach good penetrating
ability) but yet providing reliable classification. The first results are
presented in a submitted paper (Malyuto, Lazauskaite, Shvelidze, 2001).
By appling some concrete imstrumental characterisics, more realistic
spectra will be
simulated
at the next stage,
useful suggestions on optimized observing strategies can be
expected.
As one good example of future applications, the Abastumani Observatory has four
objective prisms (3 - 60A
resolutions), the expected classification accuracies may be
calculated with the
used approach, and
the optimal classification system (first of all,
the prism and the type of a planned
CCD-detector) may be reccommended. Future
projects based on the properly selected Schmidt-type
telescope+Objective prism+CCD detector
combinations
and using the optimized classification systems can provide the reliable
main physical parameters for statistical samples of stars in selected areas.
The results will be
important in studies of Galactic stellar populations.
Last updated: February 19, 2001.
V.Malyuto, valeri@aai.ee