Astrometrica reports two different SNRs for the same measurement. In astrometry reports it uses something that Herbert Raab calls “peak SNR”. This is the SNR in the single brightest pixel. In the photometry report it uses what we would think of as the normal SNR definition (i.e. the SNR in the photometric aperture). Both I think use a process similar to what we have been discussing on this thread but I only really care about the photometric version since that one (or actually the log of it) is included in ADES astrometry that gets sent to the MPC. I’m trying to find out how Astrometrica estimates the gain, G (e-/ADU) from the image since the photometric SNR it reports does not scale correctly.
All the discussion about estimating the sky standard deviation is very interesting but for bright sources the photon noise dominates and getting G wrong really messes things up.