Closed stscijgbot-jp closed 6 months ago
Comment by Melanie Clarke on JIRA:
PR for the change is here:
I expect no changes to regression tests with default settings. I am testing the optional calibration parameter with draft code to generate new bias files: https://github.com/spacetelescope/nirspec_reffiles
IRS2 biases generated with the jwst pipeline and the new preserve_refpix option are comparable to those produced by independent commissioning code, when the additional change from JP-3437 is used to subtract the mean reference offset.
Comment by Howard Bushouse on JIRA:
David Law Does this need any kind of JPCT review before it gets accepted for inclusion in the pipeline?
If the regression test is clean and it passes the usual code review I think it's ok to move ahead as this is an optional off-by-default calibration addition. Looks like we need a corresponding doc update.
I'll add it to the agenda for next Wednesday just to make sure everyone is aware of it.
Comment by Melanie Clarke on JIRA:
Thanks David Law - I have just pushed an update to the refpix docs for the new argument.
Comment by Howard Bushouse on JIRA:
Fixed by #8255
Comment by Melanie Clarke on JIRA:
Closing: this fix is working as intended.
Issue JP-3529 was created on JIRA by Melanie Clarke:
In order to generate new bias reference files for IRS2 from dark uncal files, we need an option in the reference pixel correction step to preserve the interleaved reference pixels in the output.
Generating clean bias files requires applying a reference pixel correction to the raw data, but since bias correction is applied before reference pixel correction, the interleaved reference pixels must still be present in the bias file for IRS2 data.
We'd like to add an option to process the interleaved pixels alongside the normal pixels and leave them in the file output from the refpix step. This would be off by default, and should only be used for calibration or diagnostic reductions.