Uploaded image for project: 'Data Management'
  1. Data Management
  2. DM-20163

Investigate using WCS Jacobian BoundedField in fgcmcal fit

    Details

    • Type: Story
    • Status: Done
    • Resolution: Done
    • Fix Version/s: None
    • Component/s: None
    • Labels:
      None
    • Story Points:
      8
    • Epic Link:
    • Sprint:
      DRP F19-5, DRP F19-6 (Nov)
    • Team:
      Data Release Production

      Description

      The long term ideas for decomposing the photometric fits into the wcs jacobian term and the illumination correction are described in https://community.lsst.org/t/photometric-calibration-and-astrometric-jacobians/3232. Over the near term, fgcmcal has been ignoring the WCS jacobian and simply constraining the combined calibration term, as this is what can be handled in the PhotoCalib objects. However, in the case of the tract-based fgcmcal (DM-16704), this might not work as well because there is not as many star observations (compared to a global fit) to constrain the relevant spatial variation terms. This ticket investigates, in conjunction with DM-20161, whether performance can be improved in tract-based fgcmcal by distinguishing these terms. If so, some functionality of combining a WCS jacobian BoundedField (DM-15431) with a Chebyshev BoundedField will need to be added to afw. If this is not necessary, we can wait for the longer term solution when calexp images will be fluence images thus taking the wcs jacobian out of the equation when it comes to photometric calibration.

        Attachments

          Issue Links

            Activity

            Hide
            erykoff Eli Rykoff added a comment -

            Tests show that this is not too difficult to do, and should help the fits (as well as being the right thing to do). However, it came up that in order to actually make use of it we need to be able to have bounded field classes that are products of other bounded field classes (DM-22069), and the PixelAreaBoundedField (replacing PixelScaleBoundedField) needs to be persistable (DM-22070).

            Show
            erykoff Eli Rykoff added a comment - Tests show that this is not too difficult to do, and should help the fits (as well as being the right thing to do). However, it came up that in order to actually make use of it we need to be able to have bounded field classes that are products of other bounded field classes ( DM-22069 ), and the PixelAreaBoundedField (replacing PixelScaleBoundedField ) needs to be persistable ( DM-22070 ).
            Hide
            jmeyers314 Joshua Meyers added a comment -

            LGTM.  I left some minor comments on GitHub.

             

            Show
            jmeyers314 Joshua Meyers added a comment - LGTM.  I left some minor comments on GitHub.  

              People

              • Assignee:
                erykoff Eli Rykoff
                Reporter:
                erykoff Eli Rykoff
                Reviewers:
                Joshua Meyers
                Watchers:
                Eli Rykoff, Joshua Meyers
              • Votes:
                0 Vote for this issue
                Watchers:
                2 Start watching this issue

                Dates

                • Created:
                  Updated:
                  Resolved:

                  Summary Panel