Labelling of artificial satellites in Rubin data

Hi all
In our ATLAS static transients survey we have increasing contamination from artificial satellites - especially ones at L2 (e.g. JWST and SpektrRG showed up recently). Can anyone let me know if there will be labels in the data (e.g. from the moving object pipeline) to indicate that a diff detection is an artificial satellite?

Hi @genghisken–our aim is to mask artificial satellites and not alert on them–see for algorithmic details.

1 Like

@ebellm For AP, are you planning to mask in the science image before subtraction, or in the diff afterward?

  1. If in the science image, how broadly will you propagate the mask? Will it be by a:
  • A convolution kernel size
  • a fixed (much smaller) pixel size
  • something variable on seeing?
  1. If one waits until the diff image is produced – which would be most analogous to the method in DMTN-197 and naively result in the least masked area, although the affected area in principle has the same decisions as in (1) – will there be checking to make sure the streaks didn’t affect stars used for PSF matching? Naively maybe this should only be a few percent at most so probably fine. Will there be some streak length continuous transition from satellite to NEO? Or, I suppose put more simply, what’s the minimum streak length for rejection?

@mwv It remains to be implemented and so this is provisional, but as a first pass we may mask the science image and propagate the mask by a small pixel boundary.

1 Like