Automated Stellar Atmospheric Physical Parameters Measurement Based On Template Matching

The majority knowledge about the stars is obtained through the research of stellar spectra. The physical parameters of stellar atmosphere, including the effective temperature, surface gravity and chemical abundance, are three main important factors leading to the differences of stellar spectra. At present, there are a variety of general algorithms which have been applied to extract physical parameters of stellar atmosphere in a relatively narrow parameters space, using moderate or low resolution spectra and photometric data.In this paper, automatic measurement methods based on template matching of stellar atmosphere parameters are mainly studied. The template libraries used in this paper include the theoretical template library and the measured template library. The template matching algorithms, including the K-nearest neighbor algorithm, the chi-square minimization algorithm and the cross-correlation algorithm, are applied to the automatic measurement of the physical parameters of stellar atmosphere. The experiments on different measured data show the effectiveness of these methods. The experiments also illustrate the influence of different normalization methods as well as the signal to noise ratio of spectrum on the measurement results. In order to reduce the complexity of template matching, a method using artificial neural network (ANN) to reduce the number of matching templates is proposed in this paper. In addition, the program can be deployed to parallel computing environment to further improve the efficiency. The program is implemented in the Linux environment ultimately.Description of the work in this paperThe main work of this paper is to study automatic measurement of the physical parameters of stellar atmosphere based on template matching. LAMOST has entered a pilot survey stage and is about to begin an official sky survey, which will produce a large number of spectra. The purpose of this paper is to process the one-dimensional stellar spectra and obtain the physical parameters of stellar atmosphere automatically using template matching algorithm.The works of this paper are as follows: 1. A method using artificial neural network (ANN) to reduce the number of matching templates is proposed to reduce the complexity of template matching, which also improves the efficiency of template matching and greatly reduces the matching time.2. The algorithms to measure the physical parameters of stellar atmosphere based on the template matching method are mainly studied, and the effectiveness of several template matching algorithms is demonstrated through the experiments on different measured data.3. To illustrate the influence of different normalization methods as well as the signal to noise ration of spectrum on the measurement results through experiments.4. Deploy the program to the parallel computing environment to further improve the efficiency.5. Use the Python language, combined with SciPy, NumPy, PyFITS and Matplotlib toolkits to implement the program of automatic measurement of the stellar atmosphere physical parameters based on template matching in the Linux environment.