2024-07-24 TSO CT Meeting notes

• DDT for Exoplanets is happening! https://www.stsci.edu/contents/news/jwst/2024/stsci-initiates-a-concerted-search-for-atmospheres-around-m-dwarf-exoplanets

Nestor Espinoza 

Aarynn Carter 

• Aarynn Carter has been looking at the jump detection and ramp fitting for TSOs. This was motivated by examples on MIRI Coronography which has given very nice results. This tested Tim Brandt's new optimal ramp-fitting algorithm.
  
  
• So, how are TSOs impacted by ramp fitting/outlier rejection? Took NIRISS/SOSS WASP-39b data as an example to try this out given the good measurements on orbital parameters, lack of strong correlated noise, etc. (see Carter & May et al., 2024).
  
  
• Three scenarios: (a) default pipeline, (b) TSO-based jump detection in transitspectroscopy, and then default ramp-fitting from the pipeline and (c) Brandt's likelihood-based methods (hosted here!).
  
  
• There's already some changes in the white-light lightcurve — big differences in scatter between the likelihood and the JWST pipeline.
  
  
• Differences on the transit spectrum is small, but correlated. It seems those are larger when you have less flux — also seems a chunk of the differences have to do with 1/f noise. 
  
  
• In general, it seems the likelihood method estimates a larger flux in regions between the orders. In particular, both the flux and the standard deviation is different in the 2D frames around when the background picks up. Lots of compounding effects! Hard to quantify which of the methods is right. Decided thus to create simulations with both "noise" and "noiseless" reductions to test the methods against.
  
  
• Studied two cases in the simulations: (a) 'noiseless' — ramps with only Poisson noise and (b) 'noise' — with 1/f noise. Checked results with (a) only for now. 
  
  
• Noted that the background shows highest residuals and also higher standard deviation ---- perhaps this is a simulation effect/problem? Future work: (1) how this pans out with 1/f noise, (2) what happens when you inject a transit, (3) inject a different transit depth — objective is to write a TR that provides a quantitative answer to which algorithm can impact at what signal level.

Feedback:

• Everett Schlawin — with the simulated data, you subtract different pipelines against — what if you subtract the input rates? Which ones is right?
  
  
• Also Everett Schlawin — are there any knobs to turn on the likelihood method to use it? Aarynn Carter notes that Tim had some thoughts on real data. For example, he believes the 1/f correction should never be done in the rates but on the groups. Nestor Espinoza notes work by Michael Radica that touches on this (note: is not straightforward). He believes real thing would be to do ramp-fitting accounting for 1/f likelihood — has student cracking this up.
  
  
• Loic Albert notes they have a SOSS simulator that could be useful for this. Nestor Espinoza they could join forces.

• Nestor Espinoza 

• NIRCam — working on flux/wavelength calibration for DHS. ETC finishing up results. Wrapping up APT JDox pages for DHS.
  
  
• NIRISS — Aarynn currently implementing PASTASOSS on the pipeline. Ongoing work. Goal to have this in before the next pipeline update. Idea is to generate a single reference file that would replace the wavemap and spectrace ones — so there is a calculation on those maps with PWCPOS. This would be changing in theory ATOCA as well as this does transformations to the data. Loic Albert interested in knowing the plan on this.
  
  
• NIRSpec —  currently Leonardo is the TSO person in NIRSpec.