2023-08-09 TSO WG Meeting notes

Nestor Espinoza 

• Sarah Kendrew : upcoming ETC will have the MIRI spectra going "down" instead of to the side (as we see it on the detector). Brian Brooks checked and this is just a visualization-only feature.
  
  
• Sarah Kendrew will be the Deputy^2 for next week (Nestor and Nikolay are out).

Nestor Espinoza 

• Elena Manjavacas mentions a program failure due to TA failure — this was a coordinate mispecification problem by the user. Nestor Espinoza wonders what we could do to push harder on PI's to check their coordinates so they don't lose the time to this kind of errors. Sarah Kendrew mentions Greg Sloan has been working on this among others to provide APT updates. 
  
  
• So, for now, let's keep bringing this up during Instrument Reviews to PI's!

Everett Schlawin 

• Experiment to try to reduce 1/f noise for NIRCam time-series. The issue is that 1/f noise impacts all four amplifiers. 1/f noise can also be amplifier-specific. Biggest thing is the broadband lightcurve is significantly off from the read+photon noise precision (factor of ~4). 
  
  
• 1/f correction happens row-by-row. There is typically in NIRCam some background on the "right-hand" side of the spectrum, but not on the other amplifiers. Idea: move the source to the right — get background on the first amplifier too!
  
  
• Doing this implied "tricking" the NIRCam template. They had to pretend there was a "fake" star at the time of observation, including a tight aperture position angle special requirement — so the source could move. 
  
  
• Noise is significantly better. Around ~x1.5-2 noise improvement on the broadband precision. Still obvious systematics (e.g., ramp going down) — but does look better! Drop in noise as a function of wavelength too.
  
  
• Unsure if this can be generalized because of the visit-splitting distance.
  
  
• Perhaps can be documented as a strategy for Cycle 3? Nestor Espinoza will check with Mario Gennaro when he comes back from his time off.

Sarah Kendrew 

• Sarah Kendrew shares the problem: the "shadow region" problem. This is discussed in Bell+2022 (https://arxiv.org/abs/2301.06350). The 2D lightcurve shows the blue band as a transit, pale yellow band secondary eclipses — and x-axis is pixel positions. In some observations there is a discontinuity on the 2D lightcurve on the behavior of the detector response. Can be traced back to the location of the subarray on the detector — there's a portion of the subarray that doesn't get illuminated at all unless the prism is in place. This "shadow" region is in a useful wavelength region (10.5-11 um) — and is particularly hard to fit that out of the data in eg, phase curves:
  
  
• Working hypothesis is prior illumination of the array. This sets potential culprits: is this filter dependant? Filter wheel is not parked, so prior illumination might impact. Lenght of time spent idling prior to the TSO? TA filter? 
  
  
• Sarah Kendrew queried 25 programs to explore 2D lightcurves to identify which observations were afected; queried the engineering DB to check the above culprits. Overall, "small" number affected by this (6/25). Clear discontinuity from 150-230 pixels seen on the affected programs.
• First checked the TA filter check: no obvious pattern.
  
  
• Then, prior filters used: checked all filters used in the last 12 hours for MIRI. Looked at either if MIRI programs used filters prior to the TSO, and/or on which filter it stayed prior to the TSO. No obvious pattern either.
  
  
• Perhaps look at the uncal data and check the ramps? Is there differences in the pedestal that might help understand what is happening? Perhaps quantify the effect — this was visual inspection.
  
  
• Possible solutions: move subarray higher up the array, pushing the region to longer wavelengths?
  
  
• Everett Schlawin asks if there is anything related to Ngroups? Sarah Kendrew mentions that the number of groups are all around for affected programs.

Nestor Espinoza Everett Schlawin Michael Regan Brian Brooks Elena Manjavacas Nikolay Nikolov