Uploaded image for project: 'Data Management'
  1. Data Management
  2. DM-34943

FinalizeCharacterizationTask fails to set the exposure psf prior to computing aperture corrections

    XMLWordPrintable

    Details

    • Type: Bug
    • Status: Done
    • Resolution: Done
    • Fix Version/s: None
    • Component/s: pipe_tasks
    • Labels:

      Description

      Metrics meeting says that adding the FinalizeCharacterizationTask makes stars bad.

      While the PSF looks fine, afaict, the aperture correction maps are all wonky.

      This is (at least partly) because the FinalizeCharacterizationTask fails to set the psf on the exposure prior to making the aperture correction maps, so it is using the wrong psf when doing measurement for the maps.

      I will test on rc2_subset because (fortunately) those metrics are showing the bad stellar locus measurements.

        Attachments

          Activity

          Hide
          erykoff Eli Rykoff added a comment -

          And I found the other problem. The aperture correction was being measured on the wrong psf fluxes and the configuration for the aperture correction was wrong. The default config uses all psf stars, but this has background problems and increases the scatter. Using just the brighter stars fixes this and is overridden here for characterizeImage: https://github.com/lsst/obs_subaru/blob/36659b9fbb10a093cc956a3b61bffb785cf3761d/config/charImage.py#L63-L74

          I am going to make this the overall default for finalizeCharacterization.

          Show
          erykoff Eli Rykoff added a comment - And I found the other problem. The aperture correction was being measured on the wrong psf fluxes and the configuration for the aperture correction was wrong. The default config uses all psf stars, but this has background problems and increases the scatter. Using just the brighter stars fixes this and is overridden here for characterizeImage: https://github.com/lsst/obs_subaru/blob/36659b9fbb10a093cc956a3b61bffb785cf3761d/config/charImage.py#L63-L74 I am going to make this the overall default for finalizeCharacterization.
          Hide
          erykoff Eli Rykoff added a comment -

          Good news! Running rc2_subset on my workstation on main gives wperp metric of 15.8 mmag (bad). Running on this ticket gives wperp metric of 6.96 mmag (good).

          Notably, not setting the psf correctly before doing the aperture correction was a bug but not the main problem because the psf does not change that much. Doing the aperture correction with all stars rather than bright stars was the biggest problem because the background issues create so much scatter/offsets in the calibration fluxes (aper12), and that messes everything up.

          Show
          erykoff Eli Rykoff added a comment - Good news! Running rc2_subset on my workstation on main gives wperp metric of 15.8 mmag (bad). Running on this ticket gives wperp metric of 6.96 mmag (good). Notably, not setting the psf correctly before doing the aperture correction was a bug but not the main problem because the psf does not change that much. Doing the aperture correction with all stars rather than bright stars was the biggest problem because the background issues create so much scatter/offsets in the calibration fluxes (aper12), and that messes everything up.
          Hide
          jbosch Jim Bosch added a comment -

          Looks good, marking myself as a review so I can mark it as reviewed.

          Show
          jbosch Jim Bosch added a comment - Looks good, marking myself as a review so I can mark it as reviewed.

            People

            Assignee:
            erykoff Eli Rykoff
            Reporter:
            erykoff Eli Rykoff
            Reviewers:
            Jim Bosch
            Watchers:
            Eli Rykoff, Jim Bosch, Yusra AlSayyad
            Votes:
            0 Vote for this issue
            Watchers:
            3 Start watching this issue

              Dates

              Created:
              Updated:
              Resolved:

                Jenkins

                No builds found.