Poster - Splinter Exoplanets

Friday, 17 September 2021, 16:15   (virtual Exo)

Least-squares deconvolution of HARPS spectra

Cassian Herbon
University of Göttingen - Master Student

For exoplanet research, the study of the host star has always been essential. From the HARPS satellite, a large number of stellar spectra were acquired in recent years. These spectra can be analyzed using a least-squares deconvolution modeling approach. The width and intensity of the absorption lines observed in these spectra depend not only on the chemical composition of the host star, but also on other effects, such as the star’s rotation rate, thermal broadening due to the star’s size and temperature, as well as effects due to the Earth’s atmosphere, the observing instrument itself and also noise. In theory, absorption lines occur at well-defined wavelengths as a single delta peak, but in practice the lines in the observations are broadened significantly. This can mathematically be described as a so called convolution of the absorption lines with a broadening kernel which combines all the effects described above. The reverse process is called deconvolution, so the broadening kernel, as well as the unbroadened absorptions lines can be calculated from the observed spectra. In this work a python-based least squares model was developed. A sample of 14 stellar HARPS spectra (four K stars, four G stars, four F stars and two M stars) from the ESO Science Archive were analyzed. The model replicates the observed spectra, using a reference absorption line list from VALD and a kernel calculated using the least-squares deconvolution in an iterative process. From the kernel the rotation profile of the star can be estimated, while the absorption line depths show the elemental composition of the star. For one of the spectra in the sample, it was found that it is made up of a double star system, so using this least-square deconvolution approach systems with multiple stars can be identified.