Starting with #1911 , we think we're doing something somewhat sane with the aperture fluxes, though the aperture reference file is a bit problematic and RTB is improving it. RTB reported in their presentation (linked on our weekly tag up pages for 9/25, page 10, page 12) that the apertures we were computing didn't correspond to the sizes we intended and that we weren't using an appropriate cut off to separate stars and galaxies. We should be able to make a cut on a difference of aperture magnitudes along the lines of
https://github.com/spacetelescope/romancal/blob/main/romancal/source_catalog/aperture.py#L179-L180
that neatly separates stars and galaxies. We should start by making a plot of aper04_flux / aper02_flux as a function of magnitude for stars and verifying that the current cut, ee_fraction_04 / ee_fraction_02 * 1.1 works well for L2 files---i.e., that value is in the vicinity of the typical ratio for stars, but somewhat larger than it. We should then see how that compares for L3 files, where the ee_fraction_04 / ee_fraction_02 ratio is likely elevated due to the additional blurring introduced by drizzle.