Wavelength calibrators, flat fielding, darks

1
Wavelength calibrators, flat fielding, darks
D. Shupe the IRS IST Spitzer Science Center

• IRS Data Reduction Workshop
• 24/25 March 2005
• Pasadena, CA

2
Wavelength Calibration

• Calibrations built up primarily from P Cyg, HDE
  316285, NGCs 7027 and 6543, SMP 83, ? Cas, WR 6,
  primarily exhibiting usable lines of H I, He
  II, Ne II, III, V, S III, IV, Si II, Fe
  II, III.
• A wide range of sources is needed, as lines must
  be centroidable in 2D before flatfielding, to
  cover all orders with at least 3 lines (ideally),
  and statistically remove pointing effects.
• Some orders poorly covered, e.g., SH14, LH20.
• High-Z ULIRGS with lines red-shifted into these
  orders (e.g. Ne III) have helped tune the
  solutions, by 0.08-0.1 µm.
• Tuning of low-resolution wavelength calibration
  is in progress.

3
SH spectrum of NGC 6543
4
Wavelength calibration files

• The main wavelength calibration file is the
  wavsamp table.This file contains for each order,
  the location and central wavelength of
  pseudopixels that sample the 2-D
  spectrum.Columns include order, x-center,
  y-center, wavelength, position angle, and the
  pixel locations of four corners of the polygon.
• The wavsamp table is used for extracting 1-D
  spectra from 2-D spectral images. It is also
  used to map 1-D models of stars to 2-D for
  flat-fielding.
• Ancillary FITS files are provided with the
  wavsamp
• Wavsamp_wave.fits 2-D image of wavelength for
  each pixel
• Wavsamp_offset.fits 2-D image of spatial offset
  from slit center for each pixel, in arcseconds.
  Note that coordinates and orientation of the slit
  center are in the BCD header keywords RA_SLT,
  DEC_SLT, PA_SLT.
• SPICE uses these ancillary files to display
  wavelength and spatial offset for the cursor
  pixel location.

5
How accurate is the wavelength calibration?

• Statistics of (true observed) wavelengths of
  lines used to check wavelength calibration for
  low resolution
• SL2 ave -0.024 std. dev 0.015 microns 26
  measurements
• SL1 ave -0.048 std dev 0.009 microns 53
  measurements
• SLbonus ave -0.044 std dev 0.007 microns 9
  measurements
• LL2 ave 0.022 std dev 0.034 microns 44
  measurements
• LL1 ave -0.065 std dev 0.099 microns 32
  measurements
• LLbonus ave -0.137 std dev 0.035 microns 4
  measurements
• High resolution modules
• SH std dev 0.007 microns over all orders
• LH std dev 0.010 microns (doesnt include
  some orders)
• The wavelengths of an individual observation may
  be shifted by a pointing offset
• E.g. pointing shift of 0.5 pixel for a point
  source can shift wavelengths of that observation
  by 0.5 pixels
• Commanded pointing was re-worked for IRS Campaign
  16. Data taken in earlier campaigns may be
  affected by pointing drifts

6
Upcoming for S12.0

• The Cornell and SSC teams are working on an
  update to the low-res wavelength calibration
• Probable shifts
• SL1 shift all by 0.042 microns (0.68 pixels) to
  the blue
• SL2 change dispersion and shift, so 5.5 um is
  fixed and       7 microns is shifted 0.033
  microns (1.04 pixels) to       the blue.  This
  is a 2 change in dispersion. LL2 Change
  dispersion and shift so that 18.33 microns is
       unchanged, 19.5 microns shifts blue by
  0.027 microns      (0.29 pixels) and 14.5
  microns shifts red by 0.076 microns      (0.81
  pixels)
• LL1 Change dispersion and shift so that 32.4
  microns is      unchanged, 26 microns shifts
  blue by 0.137 microns      (0.77 pixels), and 36
  microns shifts red by 0.078 microns      (0.44
  pixels). 

7
Statistics after wavelength cal change for low-res

• Statistics of (true observed) wavelengths of
  lines used to check wavelength calibration
• SL2 ave -0.001 std. dev 0.006 microns 26
  measurements
• SL1 ave -0.006 std dev 0.009 microns 53
  measurements
• SLbonus ave -0.002 std dev 0.007 microns 9
  measurements
• LL2 ave 0.022 std dev 0.034 microns 44
  measurements
• LL1 ave -0.005 std dev 0.036 microns 32
  measurements
• LLbonus ave -0.011 std dev 0.028 microns 4
  measurements
• These are Preliminary, not Final numbers!

8
Creation of 2-D flats Step 1

• Flat-fielding calibration AORs step a star along
  the slit
• Step size is about 1/3 of a pixel
• 2-D spectra are summed to fill in the slit
• At right SH sum of flatfield observations of HR
  6688

9
Creation of 2-D Flats Step 2

• 1-D model spectrum from Decin et al. (2004, ApJS
  Special Issue) is the starting point
• Wavsamp table is used to map the one-dimensional
  spectrum into an ideal 2-D image
• At right SH example for HR 6688

10
Creation of 2-D Flats Step 3

• The proto-flat is created by dividing the
  summed spectra from Step 1 by the 2-D model flat
  from Step 2
• Bad or unreliable pixels are NaNd out (white in
  this image)
• At right proto-flat for SH for HR 6688
• Edge effects and possible imperfections in the
  sampling of the star, require one more step.

11
Creation of 2-D Flats Step 4

• Observations of high zodiacal light are used to
  fill the slit
• The zody observations are not well-suited by
  themselves for flat-fielding, due to unknown
  truth spectrum, and low SNR
• Using some judicious smoothing of the ratio of
  stellar and zody flats, the protoflat is tweaked
  to result in the final flat field, to make sure
  the illumination profile is correct. This is a
  very human-intensive, interactive process.
• At right High zody observation in SH.
• Note Not enough signal to do this for SL2!

12
LH flat requires two stars
For LH, two stars are combined to obtain
sufficient SNR
13
Current status and future plans

• How well-determined is the flat-field?
• Rms of flattened star / model is lt 3 for SL and
  SH, 5 for LL (S11)
• Spatial variations of zody from 15 to 85 of
  slit position
• SL1 4 standard deviation
• LL2 1.5 standard deviation
• SH 0.5 standard deviation
• LH 4 standard deviation
• A better test will be repeatability of sources
  observed at different slit positions (assuming
  pointing is steady)
• Future flat-fielding plans
• New flat-fielding data will be taken in the next
  two months for all modules
• Using upgraded telescope pointing
• More accurate stepping of star along slit
• Flat-fields will be re-evaluated for potential
  improvements.

14
Dark calibrations

• The IRS has no shutter. Dark calibrations are
  obtained at the North Ecliptic Pole every 12
  hrs, all integration times hit more than once in
  each campaign.
• Have also taken data in non-CVZ locations like
  HDF-North, for comparison.
• These are used to build super darks, and for
  monitoring.
• The SSC pipeline has subtracted darks from your
  data. The darks are 3-D cubes subtracted from
  every plane, before ramp fitting.
• Superdarks for SL, LL, SH
• Campaign darks for LH