IRS Spectral Extraction

1
IRS Spectral Extraction
Harry Teplitz the IRS IST Spitzer Science Center

• Spitzer Data Analysis Workshop
• March 24-25, 2005

2
Extraction Pipeline
3
Review of Pipeline Extraction

• Modules
• PROFILE - collapse orders in spectral direction
• RIDGE - find spatial peak and write out spectral
  trace
• EXTRACT - perform extraction along trace, using
  fractional pixels
• IRS_TUNE - apply flux calibration and tuning
  coefficients
• Caveats
• No post-BCD background subtraction
• Point-source extraction only at this time

4
Extraction Widths

• LOW RES point source extraction width scales with
  wavelength (diffraction limited spatial profile)
• Flux calibration is only supported for standard
  width at this time
• Smaller widths will also suffer increased effect
  from pixel under-sampling
• HIGH RES full slit extraction
• Widths are defined in psf_fov.tbl

5
Pseudo-Rectangles

• Spectrum is not rectilinear on the detector
• Orders traced as pseudo-rectangles
• Order definition provided in b?_wavsamp.tbl

6
SPICE

• SPICE allows the user to perform spectral
  extraction with the same tools used in the SSC
  Pipeline. Enables general users to run the
  offline pipeline to
• Extract after further reduction of 2D dispersed
  image
• Extract anywhere along the slit, or in non-target
  FOV
• SPICE provides visualization of the extraction
  window.
• SPICE is NOT designed to provide functionality
  beyond the current pipeline (i.e. no new fluxcal)
• SPICE code written by T. Hesselroth from DownLink

7
SPICE
8
SPICE
9
SPICE
10
SPICE
11
SPICE
12
Sky Subtraction

• dark subtraction removes some background
  signal, but is not optimized for particular sky
  locations or dates of observation
• Zodiacal background varies strongly as a function
  of sky position
• Some variation is also seen with time
• Residual sky emission can be a significant
  factor, especially for faint objects.
• Near-in-time sky can help to reduce the effect of
  detector artifacts (Rogue pixels), however
• Improvement may be more cosmetic than photometric
• Doesnt resolve flat-field issues with affected
  pixels
• Can make the problems harder to identify
• Consider comparing the results of different
  methods

13
SKY SUBTRACTION OPTIONS in LOW RES

• Difference the nods
• In staring mode, each observation is nodded along
  the slit by 1/3 of the slit length.
• Provides good separation for point sources in
  sparse fields
• Create a local sky from observations in the other
  sub-slit if available
• May provide higher S/N sky measurement
• Extract sky spectrum along the slit
• Use manual RIDGE location in SPICE
• For long/multiple AORs create a supersky
• Closer in time Supersky can greatly reduce noise
  contribution
• Time-varying signal (in dark current and/or rogue
  pixel response) suggests the use of more limited
  time window
• analagous to the running median imaging
  technique

14
SKY SUBTRACTION OPTIONS in HIGH RES

• Cant difference nods, because the PSF fills most
  of the slit
• Nods serve the function of pixel redundancy
• Some programs observe an off position in LH
• Possibly mitigate artifacts
• Subtract polynomial fit to sky estimates from
  e.g. SPOT

15
Bad and Rogue pixels

• Bad pixels and cosmic rays
• flagged in bmask.fits
• Arent used in extraction, so may result in
  negative spikes
• interpolate over them in 2D before extracting
• Rogue pixels
• Temporally unstable pixels, resulting from solar
  storm activity
• Not all flagged in pmask.fits
• Especially prevalent in LH
• 5 of pixels
• Should be flagged during post-BCD analysis
• Update bmask.fits and/or redefine mask in SPICE

16
SUMMARY I have my data what now?

• Background subtraction
• Make sky frame and subtract from BCD when
  possible
• Update BMASK
• Rogue pixels
• Consider interpolating BCD in 2D
• Extract spectrum using SPICE
• Do not change extraction without re-doing flux
  calibration
• Trim spectral orders
• Remove noisiest wavelengths at the edges
• Combine nod positions in 1D
• Tie absolute flux to photometric measurements