Stellar Classification

MIRA: Exploring the Universe
from the Central Coast

The page you are viewing is taken from an exhibit called MIRA: Exploring the Universe from the Central Coast. The exhibit ran from 1 July through 24 September 2000 at the Pacific Grove Museum of Natural History.

Temperature and Luminosity Although the Universe is filled with a nearly countless number of stars, almost all of them can be categorized into one of about 90 surface temperature classes, ranging from less that 1000º C to nearly 100,000º C. In addition, stars are classified into about eight size (more technically, surface gravity or luminosity) classes ranging from dwarf, like the Sun, to supergiant.

This is a spectrum of the Sun. The dark bands occur where color would have been, except that it was absorbed by a specific atom. These bands are like atomic fingerprints, allowing astronomers to determine the types of atoms in the atmosphere of the star.	The MK System The standard classification system, the Morgan-Keenan (MK) system was developed in the 1940s and, because photographic film was used, depended on a small region in the blue part of the stellar spectrum for classification. For the last 20 years, astronomers have relied on electronic detectors, which are more sensitive in the red regions of the spectrum.

A schematic drawing of an artificial neural network used to classify stellar spectra. The data representing the shape of the spectra is input on the left, major features are recognized in the middle layer of neurons, and the appropriate classification is presented on the right.	Tracings of near-infrared stellar spectra in the MIRA classification system for giant stars. Stars at the top of the diagram are about 30,000° C, those at the bottom about 3,000°

While color spectra are very attractive, they are difficult to study quantitatively; so, astronomers plot the brightness of the stars as a function of the color or wavelength of the light.

The MIRA Classification System In the 1990s, MIRA astronomers Drs. Wm. Bruce Weaver and Ana Torres-Dodgen extended the classification of stellar spectra into the red and near infrared parts of the spectrum to accommodate modern detector systems. In addition, they developed an artificial intelligence technique, called artificial neural networks, to classify these spectra. This was the first successful computerized stellar classification technique. The latest success of this system will be published this fall: the classification of the combined spectra of binary stars.

The combined spectrum (top) from the spectra of two single stars. The artificial neural network is able to deduce the temperatures, sizes, and relative brightnesses of the two stars from the top spectrum.

Return to MIRA home page	Return to MIRA exhibit welcome page
*mira@mira.org* © 2000 MIRA
Last updated February 22, 2001 by et.