2024-12-11 TSO CT Meeting notes

10min

• getting started rolling out the Rocky Worlds DDT program here at STScI, Hannah & Nestor are meeting with the program sub-group leads this week and will roll out info on the larger team later this week  
• IDL code for the TSO monitor is no longer being updated, we have been using the public JWST schedule to gather the info on recent TSO observations for now   

• lots of recent TSOs, including: Grand Tour observations of HAT-P-1b, TRAPPIST-1, TOI-125 (although part of the transit was missed due to TTVs), TOI-451, TOI-1468, LTT3780
• upcoming: TRAPPIST-1 observations taken with NIRSPEC yesterday 
• TTRB highlights: 
  • request to increase the time allocation for a WD-BD binary to increase the settling time following reviewer recommendation 
  • WASP-6 guide star failture (GS was a galaxy)
  • guide star failure for Claringbold program (pointing error)
  • NIRCam pre-imaging of TA star for LTT1445Ab COMPASS observations
    • this method was chosen over NIRSpec blind pointing due to outliers (recent TR on this here)

• the MIRI team has discussed changing the position of the slitless subarray, users use short (<6 groups) integrations which doesn't provide enough coverage in the phase solution - really need something like 8 groups per integration
  • NIRCam changes sub-arrays every other delivery so there is precedent for how to handle this 
• other options on the table: e.g., bigger frames, smaller frame time - Michael Regan will present at a future meeting all of the options discussed  
• alternating column depth in the time-series (cleaned up nicely by the EMI in the integrations), but this noise is still present at the light curve level - would the subarray change address this better than the EMI?
• how would the subarray change affect the background subtraction methods?
• the subarray change would also remove the higher frequency resonances that are present in MIRI data at the 390 Hz level  

• biases in ramp fitting can make a big difference when looking for 10s of ppm level signals
• currently the pipeline uses a 2-point difference method for the jump detection, ramp fitting uses OLS method with weighting up-the-ramp - but the weighting is discretized 
• goal of this work: extend discretized weighted to a continuous framework using a generalized covariance matrix from differenced groups to determine count rate, exclude individual reads to identify jumps in ramp
• 2 scenarios: default pipeline, likelihood fitting (experimental jump fitting) 
• default makes a difference in SOSS data as seen in 2D frames + at the WLC level (~200 ppm differences) and at the transmission spectrum level (~200 ppm differences)
• simulated data with different noise prescriptions
  
  • pristine (no noise): different scenarios have no difference 
  • poisson noise: no difference either, Gaussian scatter in noise
  • white noise: no meaningful effect either, with slight bias toward overestimating transit depth for fainter regions
  • 1/f noise: significant non-Gaussian scatter with ~25 ppm differences in the faintest regions (likelihood method performs better here)
  • all noise: scatter is still non-Gaussian, likelihood performs better by eye with ~25 ppm residuals in the faintest regions  
•  recommendation: likelihood method should be the default for SOSS
• jump/ramp fitting is driving my Poisson & 1/f noise 
• caveats: did not explore NIRISS backgrounds, CR effects, using simple subtraction apertures 

• sprint to check pipeline versions and affect of extract data, precisions, etc 
• goals: set of notebooks and supporting functions so anyone can run perform pipeline checks
• when? either last week of Feb or first week of Apr
• not necessarily a new in-house pipeline, would not supersede the default pipeline 
• will cover lots of modes! 
• NIRISS SOSS, NIRCam LW grism, NIRSpec prism & G395H, MIRI photometry & LRS