Test WFSS contam correction using latest NIRCam sims

[stscijgbot-jp]

Issue JP-2043 was created on JIRA by Howard Bushouse:

Nor Pirzkal has recently created some detailed NIRCam WFSS simulations, which include both direct and grism images. These should be used to test the latest wfss_contam prototype step for the calwebb_spec2 pipeline.

[stscijgbot-jp]

Comment by Howard Bushouse on JIRA:

Grabbed a subset of Nor's new simulated data, using just one instance of direct and grism images for a particular source field and filter+grism combo. Processed the direct images from the rate stage up through calwebb_image3, to created a combined direct image, as well as source catalog, and segmentation map. Used that as input to calwebb_spec2 processing of some of the associated grism images. This is for the NIRCam module A, GRISMR, with F322W2 filter.

Initial results showed a difference in the location of the simulated dispersed spectra in the grism image produced by wfss_contam compared to the locations in the grism images created by Nor (acting as the "real" observed spectra in this case, even though they too are simulated). The spectra in Nor's grism images showed a displacement of ~23 pix in the cross-dispersion direction relative to the location of corresponding sources in the direct images. The wfss_contam results did not show that displacement. Investigation revealed that this is due to the fact that the NIRCam WFSS team has derived new dispersion/trace solutions for all the grisms and filters over the past 1-2 years, such that the ref files in CRDS (used by wfss_contam) were out of date. Updating the CRDS ref files to use the latest calibration data from Nor's grism Conf files produces good agreement on the locations of spectra in our two sets of simulated data.

[stscijgbot-jp]

Comment by Howard Bushouse on JIRA:

Tests were also run using the new winclip C routine provided by Mihai Cara in the latest commits to #5508 The winclip function replaces the polyclip_c.c routine previously used in the wfss_contam code. Slight differences in results at the level of e-5 to e-7 were seen, but this is far below the level of significance for these operations.