Updated NIRCam WFSS photom algorithm to account for varying dispersion

stscijgbot-jp commented 1 year ago

Issue JP-3238 was created on JIRA by Alicia Canipe:

NIRCam is working on addressing an issue with the interaction of the flight WFSS/TSGRISM flux cal files with the current design of the photom step, which is causing the flux calibration to be off by a factor of 10. The near-term fix will be to divide the PHOTMJSR values by 10. However, this triggered a discussion that dividing by 10 is not quite right – the dispersion varies across the detector, and is also wavelength dependent, so we can't just create a 2048x2048 array of dispersions and extract 2D boxes. Instead, the pipeline needs to know the wavelength at each pixel in order to find the dispersion. This is something that needs to be done individually for each source in the FOV. Forcing a pixel size/sr/bin size to the sensitivity will always be an approximation and introduces an error in the flux calibration. Because it's a small variation, it will probably not be noticed until the contamination and background subtraction are resolved, but we should begin investigating this issue to improve the quality of WFSS data products.

The photom step already creates the 2d array of wavelength values for each source. A 2d array of dispersion values would have to be constructed on the fly for each source and the pipeline would have to know the function for calculating dispersion.

stscijgbot-jp commented 1 year ago

Comment by Alicia Canipe on JIRA:

A note for tracking: this issue was raised after investigation of Help Desk ticket INC0189809.

stscijgbot-jp commented 1 year ago

Comment by Norbert Pirzkal on JIRA:

I do not mean to tell anyone how to implement this, but if the pipeline has a 2D image with the center wavelength of each pixels, the dispersion can be computed from that image down to a descent level of accuracy.

stscijgbot-jp commented 1 year ago

Comment by Alicia Canipe on JIRA:

Reassigning to Anton Koekemoer and Greg Sloan for CalWG input.

stscijgbot-jp commented 1 year ago

Comment by Anton Koekemoer on JIRA:

Adding Kevin Volk and André Martel for NIRISS – please examine whether or not this issue has NIRISS impacts, and edit the 'criticality' fields accordingly if you could please.

stscijgbot-jp commented 1 year ago

Comment by Kevin Volk on JIRA:

I am not able to understand why there is a factor of 10 problem in the NIRCam WFSS photometry calibration from the description in the ticket, so it is difficult to comment on whether there is some NIRISS equivalent. In the general sense for the SOSS mode we have a wavelength reference file that gives wavelengths pixel by pixel across the field and this allows for the (small) changes in dispersion over the array. In that mode the assumption is that all sources are at the same position so we do not need to worry about changes as a function of where the source is located in the direct images.

For the WFSS mode the dispersion does not appear to be a strong function of wavelength or position on the detector, although some small variations are measured. This does not appear to be an urgent issue, but I can take it to the WFSS sub-group within NIRISS for comments.

stscijgbot-jp commented 1 year ago

Comment by Norbert Pirzkal on JIRA:

The source of the factor of 10 in the spectroscopic calibration comes from the sensitivity being delivered as Mjy/sr/lamba while the pipeline assumes that it is in units of Mjy/sr/"pixel" where "pixel" is the wavelength range of a dispersed spectrum across a pixel. If NIRISS has assumed all along that field dependence and variation is not an issue and has calibrated things assuming that nothing varies on the detector then it should not be an issue (at least to within the error that this assumption introduces).

stscijgbot-jp commented 1 year ago

Comment by Anton Koekemoer on JIRA:

This is among the CalWG tasks prioritized in the quarterly development priorities meeting with WMO 2023-09-06

Howard Bushouse could you let us know please who from SCSB has been assigned to this, so that they can start working with Norbert Pirzkal Kevin Volk André Martel and others for the next steps in proceeding with this effort?