6dF Data Release 2 The good, the bad and the ugly Will Saunders AAO 270405

1
6dF Data Release 2 The good, the badand the
uglyWill SaundersAAO 27/04/05
2
SDSS 100M 2MASS 40M 6dFGS 3M

• VPH Ghosts
• Fluxing issues
• Splicing issue
• (Wrong extensions)
• Fibre crosstalk
• Loose slit vanes
• Fringed fibres
• Caveat emptor

• Mostly 6dFGS data
• quality is good
• better and more
• consistent than
• 2dFGRS.
• But ergonomics very
• poor, and all depends
• on a giddy edifice of
• flaky hardware and
• software.

3
VPH Ghosts

• VPH gratings used in Littrow exhibit a zeroth
  order ghost image of the fibre slit
• Caused by dispersed light reflected off detector,
  back through camera optics, then undispersed by
  the grating in -1R mode
• Intensity few
• Is in ALL 6dF VPH data
• Most AAOmega gratings have slanted fringes to
  avoid this problem

4
Fluxing issues

• Fluxing is very crude.
• Single spectrophotometric standard star observed
  with each grating through a single fibre and
  reduced in normal way.
• Reduced spectrum used to calculate Transfer
  (Instrumental) Function for that setup.
• This Transfer Function applied to all data for
  that grating.
• No airmass or atmospheric transparency
  correction.
• Assumes average Transfer Function for 6dF
  constant for both field plates and for all time.
• No Atmospheric Dispersion Corrector in 6dF.
• Scattered light subtraction very crude gt fluxing
  at blue end of spectra especially uncertain

5
Splicing Issues

• Splicing uses overlap between red and blue
  spectra around 5577Å.
• Fits a continuum to each spectrum and matches
  levels.
• Mostly works remarkably well!
• No bad pixel mask in early data pedestal
  effect at ends of spectra due to differences in
  object/sky fibre coverage.
• Sometimes very little (or no!) overlap.
• Occasional drastic differences in red/blue S/N
  due to acquisition changes.

6
Fibre Crosstalk

• Camera optics have strong coma at corners of
  field.
• Strong features in a spectrum can contaminate
  neighbour(s).
• Ragged fibre slit wavelength calibration gt
  redshift of contaminated neighbour is different
  to contaminating spectrum!
• Usually caused by massively bright stars in sky
  (or parked) fibres in early data, or by
  Planetary Nebulae

7
Loose slit vane data

• Original slit vane screw mount replaced by a
  magnetic system at end of 2003.
• Possible to position slit vane not fully homed.
• Slit vane then jolted by shutter flag.
• Causes poor sky subtraction
• All data with shifts gt 1Å _at_ 5577Å held back from
  DR2 for rereduction.

8
Fringed Fibres

• Buttons often damaged during fibre positioning,
  ferrule bent up (Turkish coffee pot syndrome)
  but not broken.
• Fibre often fractured by this, forms perfect
  Fabry-Perot Interferometer.
• Usually few-20 cycles

9
Velocity precision

• runz now assigns individual redshift errors based
  on noise in cross-correlation function. Better
  than TonryDavis formula.
• But DR2 stills assigns errors based on quality
  flag 100/80/50km/s for Q3/4/5

• Cross-correlation done with latest
• ZCAT. 17,000 matches.
• Excluding ?vgt500km/s gives
• ltV6dF VZCATgt 22km/s
• ?2 1.21 looks reasonable.
• Outliers Q2 66
• (?vgt500km/s) Q3 14
• Q4 4
• Q5 4

10
Template errors?

• Same templates used as for 2dFGRS
• For 2dFGRS, systematic offsets (100km/s) found
  for each template in use.

• Some templates show offsets up to 50km/s with
  respect to ZCAT under investigation.
• ?2 looks reasonable, 1 for all templates.

11
Enjoy