Gaia RVS automated stellar parameterization for the study of Galactic Populations

1. Gaia RVS automated stellar parameterization for the study of Galactic Populations M. Manteiga1, D. Ordóñez-Blanco2, J.C. Dafonte2, B. Arcay2, and A. Ulla3 1 Depto. Ciencias de la Navegación y de la Tierra, Universidade da Coruña (UDC) 2 Depto. Tecnologías de la Información y de las Comunicaciones, UDC 3 Depto. de Física Aplicada, Área de Astronomía y Astrofísica, Universidade de Vigo. manteiga@udc.es 1-ABSTRACT One of the oncoming European Space Agency missions is GAIA, a spacecraft that will be launched by the end of 2011 and will carry out a stereoscopic census of the Galaxy, measuring astrometry and radial velocities with unprecedent accuracy for a significant 1% of the objects in the Milky Way. One of its instruments is RVS, a radial velocity spectrograph that will provide spectra in the near IR CaII triplet region with an expected SNR between 100 and 20 for F-G-K stars with visual magnitude between 8 and 15. During these years previous to the mission launch, being part of the Gaia scientific team, we have been testing several analysis techniques in order to be prepared to extract all possible astrophysical information from RVS stellar spectra. Figure 2. RVS spectra of F3 giant (V=16 mag) S/N = 7 (single measurement) S/N = 130 (summed over mission) (credits: D. Katz). Figure 1. Schematic view of the RVS spectrograph 2- METHODOLOGY: A combination of data processing in the frequency domain (Fourier analysis and Wavelet multilevel decomposition) and connectionist systems (Artificial Neural Networks) have proven to be a good approach to derive the fundamental stellar parameters, Teff, logg, [Fe/H] and [α/Fe], on the basis of RVS synthetic spectra blurred with noise at different SNR. Signal processing techniques allowed us an estimation of the SNR, which in turn is found to be essential as the optimal algorithm for parameterization is very dependent on SNR. The derivation of the stellar parameters is performed by the use of ANNs trained with the error backpropagation algorithm [1]. 3- RESULTS: Figures 3, 4 and 5 show the accuracies obtained in the derivation of the main stellar parameters from synthetic (3, 4) and real spectra (5) Figure 4. Quartile 80 errors in thederived parameters for different stellar population tracers (synthetic spectra). Figure 3. Derived stellar parameters absolute errors obtained from synthetic spectra with added noise. The past history and origin of the different Galactic stellar populations are manifested in different chemical abundance patterns: Thin disk  traced by metal rich cool dwarfs Thick disk intermediated metallicity cool giants Halo  very metal poor dwarfs Open issues: extragalactic origin of some disc dynamical entities with differences in abundance patterns? reliable models of the Milky Way formation and evolution STELLAR POPULATIONS IN THE MILKY WAY Figure 4. Parameter derivation in MILES [2] library of stars. Bibliography [1] Ordóñez, D., Dafonte, C., Manteiga, M., Arcay, B. 2009, ESWA doi:10.1016/j.eswa.2009.07.038 [2] Sánchez-Blázquez, P., Peletier, R.F. et al, 2006, MNRAS 371, 703.