Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.stsci.edu/jwst/instruments/nirspec/docarchive/JWST-STScI-001703.pdf Äàòà èçìåíåíèÿ: Unknown Äàòà èíäåêñèðîâàíèÿ: Tue Feb 5 20:39:49 2013 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: m 31

NIRSpec: from Ramps to Slopes
Bernard J. Rauscher for the NIRSpec Detector Subsystem Team NASA Goddard Space Flight Center
31 March 2009 STScI Calibration Summit 1


Topics
· Data Format · Pipeline processing · Detector artifacts and pipeline processes needed to correct them · IDL procedures

31 March 2009

STScI Calibration Summit

2


Data Format
· NIRSpec used JWST MULTIACCUM (AKA "MULTI-nxm") sampling
­ n up-the-ramp groups ­ m frames averaged per group ­ Baseline science integration is MULTI-22x4

·

Total noise averages down approximately as,

· ·

Where f is total photonic current including dark current Eqn. 1 neglects 1/f noise, which NIRSpec testing has shown to be significant with the SIDECAR ASIC. It therefore slightly underestimates the noise of JWST flight systems.
STScI Calibration Summit 3

31 March 2009


Set of Dark Ramps

Example: GSFC Pipeline Processing
MKMULTI REFSUB NOTE: REFSUB is frequently called by MKMULTI to do both boxes in one step

Slope Images Mean Image

ITERCOMBIN E Sigma Image

Pixel-toPixel Crosstalk

HOT_PIXEL_ CROSSTALK

IMHI2 Dark Rate Total Noise

IMHI 2

Other arc 31 M Analysis

h 2009

STScI Calibration Summit

4


Dark Ramps

Other Data Products Derived from Dark Ramps
POWSPEC_S CA, DINO SNOW BALL FFT Noise Analysis

Snowball Analysis

FIND_COSMIC_RA YS, CR_CROSSTALK

Pixel-toPixel Crosstalk

BIAS_DRIFT
31 March 2009 S T S cI C a

Bias Drift Outputs Correlations lCDtion Summit ibra S Noise

5


Slope Fitting Using MKMULTI
· MKMULTI computes integrated signal per pixel, · Pixel slopes computed using standard 2parameter least-squares fitting. For each pixel, · The standard equations (see e.g. "Numerical Recipes in C the Art of Scientific Programming") are,
31 March 2009 STScI Calibration Summit 6

(This IDL Procedure is included in the deliverable FPQLA Software)


Reference Pixel Correction using REFSUB (1)
(This IDL Procedure is included in the deliverable FPQLA · NIRSpec's HAWAII-2RG Software) lude 2 types of SCAs inc

reference pixel

Reference pixels in rows · Available on all outputs · These are always used!

0.0 e-/s/pixel +0.010

31 March 2009

STScI Calibration Summit

Reference pixels in columns · Available on outputs #0 and #3 only (2 of the 4 outputs) · These are sometimes used · Should be treated with caution because their use involves making assumptions about noise correlations across outputs

0.0 e-/s/pixel +0.010

7


Reference Pixel Correction using REFSUB (2): Using Reference Pixels in Rows
· In the IDL zero-offset array indexing convention, these are pixels,
[*,0:3] & [*,2044:2047].

· ·

Each output is treated independently For each output, the horizontal reference pixels are used in one of t"Simplys. wo wa e" "Reference Rows Only"

·For each output, we compute the mean [or median] of the reference pixels in rows 2044:2047 - If computing means, we use DJS_ITERSTAT to implement iterative sigma clipping - If computing medians, we use the IDL median function ·Therch 20a9 (or median) valueSisCa 31 Ma me 0 n S T cI subtracted from every pixel in

·For each output, we compute the mean [or median] of the reference pixels in rows; ("bottom") 0:3 & ("top") 2044:2047 - If computing means, we use DJS_ITERSTAT to implement iterative sigma clipping - If computing medians, we use the IDL median function -For each output and using the "top" and "bottom" reference pixel values, interpolate a reference pixel value for each row ·The model value for each row is subtracted from every pixel in the row.
8

libration Summit


Reference Pixel Correction using REFSUB (2): Using Reference Pixels in Columns
· Because only 2 of the 4 outputs actually have reference columns, the reference columns should be used only if there is significant (>~50%) output-to-output noise correlation
­ ­ For NIRSpec, it is not clear yet whether this condition will be satisfied For our "Pathfinder Detector Subsystem", this condition was not clearly satisfied

"Teledyne Imaging Sensors (TIS) Baseline"

·Needs to be done after applying a reference pixel correction using reference pixels in rows. ·For each row, subtract a rolling average of 4 reference rows on either side of the current row. ·Optionally, the two halves of the SCA can be treated independently.
31 March 2009 STScI Calibrati

"GSFC Approach" ·Testing shows this to be slightly (1%2%) more powerful than the TIS approach ·Each half of SCA treated independently ·All 4 reference columns on one half averaged together to form a mean reference column for that half ·Mean reference column is smoothed using either a Savitzky-Golay filter or Gaussian smoothing kernel. ·Smoothed mean reference value subtracted from every pixel in the on Summitrow 9 current


IDL Procedures
· The NIRSpec Detector Subsystem Team at GSFC works mostly in IDL · We have delivered a Focal Plane Quick Look Analysis (FPQLA) package to ESA · The FPQLA implements much of what we have discussed; slope fitting, reference pixel correction, etc. · We are willing to share this software on a best-effort basis with others in JWST · The FPQLA software runs on Red Hat Linux
31 March 2009 STScI Calibration Summit 10