Rogue Pixels on IRS Arrays

1
Rogue Pixels on IRS Arrays
Jim Ingalls SIRTF Science Center

• IRS Data Workshop
• 18/19 August 2005
• Pasadena, CA

2
Outline

• 1 Definition of rogue pixels and how we detect
  them

2 Trends in rogue pixels and effects of
environment
3 Fixing rogue pixels
3
Rogues as pattern noise

• Rogue pixels manifest themselves as a changing
  pattern noise in all IRS arrays.
• They are most noticeable in LH and LL.

LH 60s Dark calibration ramp image, last
frame.
4
Rogue pixels vs. time
(DN/s)

• Rogues have elevated dark currents and they vary
  with time
• effect appears cumulative with time
• not cured by daily anneals

LH 60s Dark measurements, coadded per
campaign
5

Construction of Rogue masks
LH Dark slope image
1) Values of
Multiresolution wavelet transform (Starck
Murtagh 1998, PASP, 110, 193) allows us to
separate out the behavior at different spatial
scales.
The n0 component is mostly rogues and noise
n0
n1
n2
smooth image
n3
n4
6
Rogue Pixel Thresholding

• Defined by thresholding the n0 component.
• Threshold depends on the standard deviation of
  the original image, derived from the wavelet
  transform.

LH data from IRS-1 (December 2003)
IRS-1 LH Rogue Mask
1s 9 DN/s
7
Rogue Pixel Thresholding
LH (and to a lesser extent LL) appears to be
getting noisier with time. Most of this is rogue
pixels.
LH data from IRS-18 (February 2005)
IRS-18 LH Rogue Mask
1s 13 DN/s
8
Number vs. time
LH
9
Number vs. time
LL
10
Effects of space environment
(LH)
CR hits measured in pipeline, averaged per
campaign over all 4 arrays
11
Some arrays are better than others
12
LH Campaign-based darks
Keep in mind This only improves dark
currentbest for faint sources.
LH data
13
Campaign-based Rogue Masks

• All 4 modules
• FAQ
• Updated within 2 weeks of campaign end.
• campaigns.txt relates campaign to FITS keyword
  CAMPAIGN

http//ssc.spitzer.caltech.edu/irs/roguepixels/
14
Pixel Replacement

• Joint Cornell/SSC package
• In Beta-testing stage
• Will be released in Fall 2005, as a standalone
  IDL tool.

LH data
15
Pixel Replacement
LH data

• Input Rogue mask (in this case from Dark data).
  
• Interpolate directly along columns.

16
Pixel Replacement
LH data corrected

• Other Options
• Use flanking rows to estimate spatial behavior,
  normalize to good pixels in rogue row.
• Use wavelet decomposition to separate out rogues
• Rogue detection on the fly

17
Effectiveness vs. source brightness
LH data
18
The IRS IST Has Done (will do) the Following to
Help Fix Rogue Pixels
1 Implemented Campaign-dependent darks for LH.
(Done S11)
2 Made available rogue masks for general use.
(http//ssc.spitzer.caltech.edu/irs/roguepixels/)
3 Make available a rogue detection/ image
cleaning tool. (Plan Fall 2005)