This page archives Slack comments from the splinter session on ramp fitting of the Improving JWST Data Products Workshop (IJDPW).
James Davis
NGDEEP
https://www.stsci.edu/jwst/phase2-public/2079.pdf
Alicia Canipe
Description of the issue they reported: “(1) The measured depths from the final mosaics in NGDEEP are generally 0.1-0.5
magnitude shallower than what is predicted by the ETC. (2) Comparing with the MIRI Deep
Survey, which has a similar strategy in a partially overlapping field, NGDEEP clearly is not
deeper though it has 50-70% more exposure time in some filters.”
Dan Coe
Created a channel for those interested in working on this: #topic_ngdeep_depth
Alicia Canipe
My notes based on Mike’s notes (lemme know if I should fix anything):Investigate:
- There are issues with the DEEP8 readout pattern — high overhead, CRs affect a lot of data. New MEDIUM-DEEP readout pattern could help (JSOCINT-722).
- For cases with jumps and only 1 or 2 groups, we should compare with and without the correction (? not sure if this is what Mike meant).
- NIRISS occasionally sees issues with the first sample in the ramp. Kevin will send Mike and Eddie data to investigate.
- NIRCam has 1/f noise issues for subarrays with no reference pixels. We could consider an IRS^2 mode for NRC. Super/substripe can also do this (JSOCINT-733, in progress). Superstripe could be baseline for taking better care of 1/f.
- Issues with the nonlinearity fit and whether high order polynomials are the best option. Discussed using mean parameters over the array to improve the correction (would require a single saturation value for all pixels). Legendre polynomials are better than ordinary polynomials and scaling to saturation is more physical. Strong NIRCam pixel source follower current causes a kink in the fit — discussed other options, e.g., a running mean and a spline fit.
Improvements:
- Improve nonlinearity uncertainty. We currently have 1-2% errors in our linearity, but we can get order of magnitude better.
- Jumps within groups are a problem. Tim Brandt has code for handling jumps between or within groups.
- Using the first group in the default pipeline would be useful. Tim has code for handling single groups better than our pipeline (would be especially useful for NIRCam ZEROFRAME). Consider using the reset frame to get more dynamic range.
- Update jump step to use all integrations in the range, rather than working within integrations.
- Saturation in groups is not flagged. Internal discussion about an algorithm for flagging saturation in grouped data is in JP-3318 (Optimal Saturation Flagging). Corresponding github issue here: https://github.com/spacetelescope/jwst/issues/7765
- Consider implementing high-efficiency readout modes for JWST (JSOCINT-764)
Bryan Hilbert
See JP-3318 for a proposed algorithm for flagging saturation in grouped data.
Alicia Canipe
Mikes notes:
Uncertanity in nonlinearity -> improve nonlinearity correction
Functional form of the fit of the nonlinearity
NIRCam pixel source follower current is high
Using mean parameters over the array to improve (require a single saturation value for all pixels)
Legendre polynomials are better, scaling to saturation is more physical
Jumps within groups are a problem
Jumps 1 or 2 groups compare with without
Tim can handle jumps between or within
Jump step using all ints range
Saturation in groups is not flagged
Tim's code can handle the NIRCam 0th frame
One idea is using the reset frame to get more dynamic range
Would replace 0th frame in NIRCam or not
Kevin sometimes sees lst sample in ramp is sometimes off --> to Eddie
Mario, use one amp as a way to make an easier
Eddie showed how substrip changes
NIRCam could use IRS2
Deep8 problems, high overhead, CRs affet a lot of data
Medium Deep NIRCam readout mode - 15 frames per group already in JSOC Int - read 8 drop 7 or read 2 drop 13 (in progress)
