1. Single Object Slitless Spectroscopy Simulations
Loïc Albert (with help from Étienne, David and Jason)
Science Team Meeting
UdeM, October

2. Simulation Objectives
Test different trace extraction scenarios to devise an optimal extraction method and calibration strategy.
Investigate our sensitivity to instrumental artifacts such as intra-pixel response, flat-fielding and wheel repeatability.
Provide simulated transit data sets to the community on our web page.

3. Main Difficulties
Want to limit digitization artifacts to <1 ppm therefore need PSF oversampling by x10 in simulations. ~1 day per order on Mac Pro using IDL.
Contamination by order 2 for λ>2.0 μm at a level of ~1% (10000 ppm). How to extract traces in this overlap region?
Wheel repeatability is ~0.15 degrees, perhaps quantized position (-0.15, 0.00,+0.15), perhaps a normal distribution? Impacts multi-epoch targets and calibrations (to a lesser extent).
Spatial axis tilted by a few degrees relative to the detector pixels (by design to oversample spectral lines). Can not simply assume constant wavelengths to be along detector rows.

4. 2nd and 1st orders cross contamination

5. Table interpolation (and extrapolation...)
Trace Position λ x y
0.7 30
0.8 40
200
...
2.5
220
1700
2.6
225
1900
JWST Cryo indices for
ZnS and ZnSe
CodeV
Table interpolation (and extrapolation...)

6. Monochromatic Wavelength PSFs

7. x10
oversampled

8. Stellar + Planetary Atmosphere Model
"Delta Function" Trace
Oversampled x10

9. "Delta Function"
Trace
Delta function trace convolution by PSF kernel for ~30 wavelengths + Interpolation for in-between wavelengths
Monochro-
matic PSF
80x80
256x2048
+ padding
...
30 monochromatic
convoluted traces
Interpolate traces between existing traces

10. Trace examples x10 oversampled
Order 1
Order 2

11. Zodiacal Light

12. Simulated background components
Modelled zodiacal light background
Modelled 1/f noise
Scattering on optics (10-3 and uniform, ref. Rohrbach simulations)
OTE thermal emission?

13. Spectrum extraction Baseline spectral extraction:
2D trace extracted and drizzled on a 1D, constant dispersion, oversampled spectrum. (See next slide)
Currently being implemented and tested as the baseline in our pipeline.
Other extraction ideas:
Traditionnal aperture extraction  λ-dependent transit modelling
1-D drizzle (like baseline) with mask to mitigate contamination  λ-dependent transit modelling
White-light transit modelling 2-D image subtraction of median  spectral extraction (any method)
Somehow feed information from order 1 to extract order 2 where traces overlap.

14. Spectrum extraction:
Drizzle the 2D trace to a 1D oversampled spectrum
Still have to define an extraction aperture.
Assumes a lambda+tilt vs. pixel calibration.

15. Background Subtraction
Spectral Traces
Stack Subtracted
Assumes the existence of a deep stack
Needs a "threshold" parameter to establish pixels to reject from fit (good pixel > threshold)
Currently use mean of un-masked pixel, column by column. Can think of more fancy fits and quantify them.
Masked
Corrected

16. Schedule Milestone Time Line
Handling limb darkening appropriately to generate realistic transit sequences
now-Oct 31
Generate a ~5hr simulated transit and make available on the web
Nov 7
Experiment extraction with 1D drizzle on time-series
Nov 1 and on
Test how wheel repeatability affects extracted planet spectra
Nov 7-15
JWST Transiting Workshop (Nov 16-18, STScI)
Nov 16-18
Experiment with a few other extraction methods
Till X-mas
Self wavelength calibration using the science target model
Spring 2016
Generate fake calibration data (A0 star for flat-fielding and trace profile)
Compare extracted planet spectra when including intra-pixel sensitivity
Could compare dependency on the non-linear coefficient maps.
Add red noise to the time-series and study