Towards a HERMES Pipeline - A feasibility study - by dffhrtcv3

VIEWS: 0 PAGES: 22

									The HERMES RAP
- A feasibility study -




 Elizabeth Wylie de Boer
       Ken Freeman

     HERMES Meeting - 26/02/08   1
                            Overview
  Automated reduction and analysis pipeline for
                 HERMES data
                           Capable of outputting

   Atmospheric Parameters - General metallicity - Specific element abundances


 Basic reduction
 Atmospheric parameters
       Teff, log g, [M/H]
 Detailed abundances
 Potential traps, uncertainties

Elizabeth Wylie de Boer     HERMES Meeting - 26/02/08                           2
                          Basic Reduction
Based on AAOmega program - 2dfdr

Produces sky-subtracted, wavelength-
 calibrated spectra




Elizabeth Wylie de Boer      HERMES Meeting - 26/02/08   3
          Atmospheric Parameters
 Two stage process

1. First pass for approximate estimates
             Gives smaller range for model grid


2. Second pass to refine parameters


Elizabeth Wylie de Boer   HERMES Meeting - 26/02/08   4
             First pass - Skymapper
Can obtain Teff, log g, [M/H] from
 Skymapper photometry (Keller et al. 2007)

Uses UVGRIZ filters, Kurucz model grid

For =0.05mag gives parameters
      ∆Teff ~ 100K, ∆log g ~ 0.5, ∆[M/H] ~ 0.5

                          (See Stefan Keller’s Talk at 1130)

Elizabeth Wylie de Boer         HERMES Meeting - 26/02/08      5
                           Second Pass
 Get Teff, log g, [M/H],
  Vrot, Vrad
 Grid of synthetic spectra
    (Munari et al. 2005)




                                             [M/H]
       Range of T, log g, [M/H],
        Vrot, alpha-enhancement
        and microturbulence
 Adapt chi-squared
  minimisation technique                                Teff
  comparing observed
  spectrum over smaller
  grid

Elizabeth Wylie de Boer     HERMES Meeting - 26/02/08          6
                          Second Pass
Munari et al. 2005 - Test applications
Eclipsing binaries, Chi-squared vs orbital solution
      ∆T = 42K, ∆log g = 0.03 dex, ∆Vrot = 4 km/s,
       ∆[M/H] = 0.06 dex
RAVE spectra,             Chi-squared vs spectral classification
      ∆T = 3%, ∆log g = 0.25 dex
      Larger errors due to range in classification scheme



Elizabeth Wylie de Boer    HERMES Meeting - 26/02/08                7
          Atmospheric Parameters
               - Summary -

Firstly through Skymapper photometry
Secondly through chi-squared fit

Accuracy of 1~2%
Provides parameters to adopt for model
Model gets passed to next stage…..
Elizabeth Wylie de Boer   HERMES Meeting - 26/02/08   8
                Detailed Abundances
Pipeline for individual abundances
                          (HERES - Barklem et al. 2005)
Use spectrum synthesis
      MOOG               (Sneden 1973)



                           Atmospheric Model
                                Line list
                           Observed Spectrum
Elizabeth Wylie de Boer         HERMES Meeting - 26/02/08   9
                Detailed Abundances
Take atmospheric model - fix hereafter
Synthesize spectra for region
Derive Fe, Ti abundances - [M/H]
Set Fe and Ti and then derive element
 abundances from individual lines
Assume scaled solar value for all initially

Elizabeth Wylie de Boer   HERMES Meeting - 26/02/08   10
                    Synthesis Example




Elizabeth Wylie de Boer   HERMES Meeting - 26/02/08   11
                    Synthesis Example



                          [X/H] = - 0.50
                                 - 0.25
                                   0.00
                                 + 0.25
                                 + 0.50




Elizabeth Wylie de Boer        HERMES Meeting - 26/02/08   12
                          Fe and Ti




Elizabeth Wylie de Boer   HERMES Meeting - 26/02/08   13
      Light elements (non alpha)




Elizabeth Wylie de Boer   HERMES Meeting - 26/02/08   14
                          Alpha elements




Elizabeth Wylie de Boer     HERMES Meeting - 26/02/08   15
       Light and Heavy s-process




Elizabeth Wylie de Boer   HERMES Meeting - 26/02/08   16
        Abundance Error Analysis
Main uncertainty due to spread in abundance
 derived from individual lines - std dev, 

Abundances depend on Teff, log g and 
      Solve abundances again for ∆ Teff, ∆ log g, ∆
      Gives estimate of uncertainty due to model, BUT
       increases computing time by 3x



Elizabeth Wylie de Boer   HERMES Meeting - 26/02/08      17
                          Practicalities
1 x Reduction through 2dfdr ~minutes

1 x Skymapper model ~seconds
1 x Chi-sqaured fitting for model ~minutes
1 x Synthesis of spectral region ~seconds
N x Chi-squared fitting for abundances
 ~minutes

Telescope --> Results ~~minutes per star
Elizabeth Wylie de Boer    HERMES Meeting - 26/02/08   18
                          Potential Traps
Continuum fitting - 2dfdr
      Marked effect on abundances - care
       required


Fitting correct line for analysis
      Set conditions for small wavelength shifts
      Setup initial line lists to avoid blended lines

Elizabeth Wylie de Boer     HERMES Meeting - 26/02/08   19
                          Initial Set-up
Choose region of spectra for maximum
 abundances in minimum coverage
Initial Line list
      Correct log gf values
List of lines to be measured
      Avoid blends, molecular bandheads

Line list set up “correctly” only once
Elizabeth Wylie de Boer     HERMES Meeting - 26/02/08   20
                          Summary
Reduction pipeline based on AAO 2dfdr
Approximate atmospheric model from
 Skymapper photometry
Refine atmospheric model from chi-squared
 minimisation
Automated procedure for individual
 abundances
Fit Fe, Ti first then set for remaining analysis
Each line solved individually
Error analysis from std dev of individual lines
Elizabeth Wylie de Boer   HERMES Meeting - 26/02/08   21
                                                               Reduced
 Telescope                            2drdf                    Spectra

                                                                   Skymapper
              Model Grid                                           Photometry
             Chi-sqaured fit                   Initial Teff,
                                              log g, [M/H]
  Final Teff,
                                                N
log g, [M/H] (+)
                               Synthetic spectra                Element
                                Chi-squared fit                abundance


                                Teff, log g, [M/H] (+)
                             Abundances for X elements

   Elizabeth Wylie de Boer        HERMES Meeting - 26/02/08                22

								
To top