search

spacetelescope / jwst

Python library for science observations from the James Webb Space Telescope
https://jwst-pipeline.readthedocs.io/en/latest/
Other
557 stars 164 forks source link

Outlier detection's model readnoise is over-estimated, probably #923

Closed jdavies-st closed 3 years ago

jdavies-st commented 7 years ago

The current code in outlier_detection.flag_cr uses a detector's read noise to model the difference between the noise in a blotted image and that in the detector image. It gets this from the read noise reference file for that detector.

But the reference file gives the correlated double sampling (CDS) read noise for a detector. And JWST IR detectors have multiple reads which are combined to make up a final science exposure. This ISR from Bryan Hilbert on the NIRCam team shows that the read noise is lower if there are multiple reads for an exposure:

http://www.stsci.edu/hst/wfc3/documents/ISRs/WFC3-2008-25.pdf

Our current algorithm in flag_cr is based on that used in astrodrizzle, which traces its origins back to ocrrej for STIS, described in this ISR:

http://www.stsci.edu/hst/stis/documents/isrs/199822.pdf

It would be good to revisit this and updated the algorithm where needed.

jdavies-st commented 3 years ago

Fixed by https://github.com/spacetelescope/jwst/pull/5822

karllark commented 3 years ago

So, should this issue be closed? (Was checking my recent PR for another reason and saw this issue was referenced as being fixed, but not closed).

jdavies-st commented 3 years ago

Yes! Closing.