1. NIRSpec pipeline concept Guido De Marchi, Tracy Beck, Torsten Böker
Instrument characteristics
multi-object spectrograph --> every detector pixel sees every wavelength
reflective optics (incl. dispersive elements) --> large, variable slit curvature
used on a diffraction-limited telescope --> PSF varies with l
wide wavelength range ( mm) --> chromatic slit losses
off-axis telescope and wide field of view --> significant distortion

2. NIRSpec pipeline concept
Active MSA area
Mounting frame
Fixed slits and IFU aperture
Detector array
Direction of dispersion

3. NIRSpec pipeline concept Guido De Marchi, Tracy Beck, Torsten Böker
Instrument characteristics
multi-object spectrograph --> every detector pixel sees every wavelength
reflective optics (incl. dispersive elements) --> large, variable slit curvature
used on a diffraction-limited telescope --> PSF varies with l
wide wavelength range ( mm) --> chromatic slit losses
off-axis telescope and wide field of view --> significant distortion
Headaches.....

4. A first outline of the NIRSpec pipeline
RAW DATA
BIAS & DARK
SUBTRACTION
PIXEL-TO-PIXEL
DQE CORR.
LOCATE
EXTR. WINDOW
THROUGHPUT CORRECTION
(incl. L-flat, blaze function, transmission of optics, & “default” chromatic slit loss)
MSA CONFIG. &
DISTORTION MAP
GW TELEMETRY
GEOMETRIC
DISTORTION
(spatial)
FINAL l
CALIBRATION
(dispersion solution)
“DELTA” CHROMATIC
SLIT LOSS CORR.
ABSOLUTE
FLUX CALIBRATION
EXTRACT
1-D SPECTRUM
SUBTRACT
BACKGROUND
FINAL 1D
SPECTRUM
FLATFIELD
REF. CUBE
GRATING
EQUATION
PHOTFLAM
KEYWORD
BIAS/DARK
REF. FRAMES
“P-FLAT”
REF. FRAME
LINEARITY CORR.
LINEARITY
(Assuming no l-dependence)
One window for every open shutter….

5. Defining the extraction windows
MSA mask
Red: object, green: background
Each spectrum has “extraction box” on FPA
Extraction boxes overlap, possible “spill-over”

6. An outline of the NIRSpec pipeline
RAW DATA
BIAS & DARK
SUBTRACTION
PIXEL-TO-PIXEL
DQE CORR.
LOCATE
EXTR. WINDOW
THROUGHPUT CORRECTION
(incl. L-flat, blaze function, transmission of optics, & “default” chromatic slit loss)
MSA CONFIG. &
DISTORTION MAP
GW TELEMETRY
GEOMETRIC
DISTORTION
(spatial)
FINAL l
CALIBRATION
(dispersion solution)
“DELTA” CHROMATIC
SLIT LOSS CORR.
ABSOLUTE
FLUX CALIBRATION
EXTRACT
1-D SPECTRUM
SUBTRACT
BACKGROUND
FINAL 1D
SPECTRUM
FLATFIELD
REF. CUBE
GRATING
EQUATION
PHOTFLAM
KEYWORD
BIAS/DARK
REF. FRAMES
“P-FLAT”
REF. FRAME
LINEARITY CORR.
LINEARITY
(Assuming no l-dependence)
One window for every open shutter….

7. Operations on the extraction windows
Same as those carried out for traditional ground-based MOS
flat-field (uniformity of detector response): f(l), f(x,y)
tracing the spectrum and rectifying it: f(l), f(x,y)
wavelength calibration: f(l), f(x,y)
flux calibration (throughput of optics and gratings): f(l), f(x,y)
absolute flux calibration
But wait…
Because of NIRSpec’s design (e.g. MSA always in the way), some of the steps above must be carried out simultaneously, require calibration measurements that are intertwined

8. Throughput correction of
“Flat-fielding” NIRSpec spectra
Throughput correction of ^
Need a “throughput” data cube
(for each filter/grating combination)
Output (assuming a source with flat spectrum)
l --->
x --->
y --->
l --->
l --->
Goal: to correct for the total instrumental throughput variations, both as a function of wavelength (e.g. optics transmission, blaze function) and field angle (e.g. DQE, vignetting).

9. Contributions to the “Throughput correction”
reflectivity of all mirrors: f(l), f(x,y)
transmission curves of filters: f(l)
blaze function of gratings: f(l), f(x,y)
large-scale response of detector (L-flat): f(l), f(x,y)
All contributions are measured at component level, and built into a physical/optical instrument model. Once NIRSpec is assembled, they cannot be measured individually. However, once a shutter has been specified, all of them are - in principle - deterministic, and can be accurately modeled. Using the instrument model, all these effects will be corrected simultaneously.
But wait……

10. The bummer: chromatic slit loss
Fixed slit size, but variable PSF width…….
1 mm
3 mm
5 mm
…… causes “flaring” and intensity gradient:
A “default” correction for e.g. a perfectly centered point source can be included in throughput correction.
The user needs to optimize this
correction later…..
l --->

11. An outline of the NIRSpec pipeline
RAW DATA
BIAS & DARK
SUBTRACTION
PIXEL-TO-PIXEL
DQE CORR.
LOCATE
EXTR. WINDOW
THROUGHPUT CORRECTION
(incl. L-flat, blaze function, transmission of optics, & “default” chromatic slit loss)
MSA CONFIG. &
DISTORTION MAP
GW TELEMETRY
GEOMETRIC
DISTORTION
(spatial)
FINAL l
CALIBRATION
(dispersion solution)
“DELTA” CHROMATIC
SLIT LOSS CORR.
ABSOLUTE
FLUX CALIBRATION
EXTRACT
1-D SPECTRUM
SUBTRACT
BACKGROUND
FINAL 1D
SPECTRUM
FLATFIELD
REF. CUBE
GRATING
EQUATION
PHOTFLAM
KEYWORD
BIAS/DARK
REF. FRAMES
“P-FLAT”
REF. FRAME
LINEARITY CORR.
LINEARITY
(Assuming no l-dependence)
One window for every open shutter….
(output is re-sampled grid)

12. Rectifying NIRSpec Spectra
Slit is curved (function of field angle)
Lines of constant l spread over multiple pixel columns l1 l2 li
Use “Drizzle” technique as possible approach for coordinate transform
Need rebinning before final spectrum is extracted l1 l2 li

13. “Delta” correction for chromatic slit loss
- depends on source shape and position within shutter
must be user-controlled l1 l2 li l1 l2 li
Collapse to 1-d spectrum
- depends on source extent and background subtraction
must be user-controlled
For quick-look analysis, pipeline subtracts TBD “default” background

14. An outline of the NIRSpec pipeline
RAW DATA
BIAS & DARK
SUBTRACTION
PIXEL-TO-PIXEL
DQE CORR.
LOCATE
EXTR. WINDOW
THROUGHPUT CORRECTION
(incl. L-flat, blaze function, transmission of optics, & “default” chromatic slit loss)
MSA CONFIG. &
DISTORTION MAP
GW TELEMETRY
GEOMETRIC
DISTORTION
(spatial)
FINAL l
CALIBRATION
(dispersion solution)
“DELTA” CHROMATIC
SLIT LOSS CORR.
ABSOLUTE
FLUX CALIBRATION
EXTRACT
1-D SPECTRUM
SUBTRACT
BACKGROUND
FINAL 1D
SPECTRUM
FLATFIELD
REF. CUBE
GRATING
EQUATION
PHOTFLAM
KEYWORD
BIAS/DARK
REF. FRAMES
“P-FLAT”
REF. FRAME
LINEARITY CORR.
LINEARITY
“CALWEBB”
(Assuming no l-dependence)
(one window for every open shutter…. )
(output is re-sampled grid)
“CALNIRSpecA”
(one spectrum for every open shutter)
“CALNIRSpecB”
(defined by user or average)
(defined by user)
(erg/cm2/s/Å, sampled within variable slit aperture)