Anom-R4: Behavior at / near bright stars is characterized (esp. the range of CCDs rendered useless by scattered light and crosstalk from bright stars in the field) [DECam TO-9 for ghosts, TD-17 for crosstalk]. OBSTAC is tuned to avoid bright sources as needed.

We need to answer the question: what region in the vicinity of bright stars is spoiled by either the direct image or secondary images (ghosts).
See also;filename=des-star-counts.pdf;version=1

Above doc reports that ~30 tiles will have stars with m<3 in them, so a large fraction will have m<3 star within a few degrees. Brightest stars will be ~0 mag, 10x brighter or so. We want to place a mag 0 - 1 star in the array to check for how many CCDs are killed by too much stray light or crosstalk (i.e. should we just skip imaging that tile!). Can take a full-length exposure (100s) and a short one (10s) to simulate effect of mag ~3 star. Then we want to examine effect of such a bright star when it is off axis by:
  • 1/2 of field (1/2 degree)
  • Just off the DECam FOV (1 degree)
  • 2 degrees
  • 4 degrees
  • For each filter*: Reduced science frames of a field with one dominant star at mag=0 or 1 (ideally several magnitudes brighter than any other object in a radius of 2 deg around the center of the field). Then take 100s, 10s exposures with star near axis; and a series of 4 more exposures at above list of off-axis positions. Take during photometric conditions.


  • Eyeball analysis of all exposures to mark (a) how far from star the data is completely unusable from saturation or highly structured ghosts, and (b) how far from star is sky level significantly elevated above normal background.
  • Determine resultant avoidance radius vs magnitude for star.

Codes required:

Eyeball examination to suffice?


Use either an actual bright star catalog or expected bright star densities to calculate the masked area fraction if bright stars are not avoided.

Calculate the ghost surface brightness of a 6mag star. If this exceeds 25 mag arcsec^{-2}, TO.9 fails.

If masked area calculated as expected is above 5% of the survey area, we should connect with OBSTAC to make sure strategies of avoiding bright stars are in place.
If ghost brightness exceeds TO.9, there should be a two-fold approach: (1) check whether the additional masking of ghosts of bright stars hurts our area limit from the first test even for a star-avoidance strategy and (2) try to identify elements in the optical path that might cause the excess stray light.

Decide whether it is advantageous to skip or displace DES pointings that contain the brightest stars.


The area is no problem by more than an order of magnitude (PASSED).

The open question is the magnitude of ghosts. Steve, with the help of Vinu and the eye-ball squad, has analyzed several types of ghost images in Ghosting_Analyses.
The formal requirement (ghost surface brightness below 25 mag arcsec^{-2} for a 6th mag star) has to be checked independently for different reflections.
  • R Doradus, 1.5mag in I, has a ghost of 23.31 mag/arcsec^2 (double-reflection inside C5) (PASSED)
  • Mira has a ghost of 26 mag/arcsec^2 (reflection between the CCD in the focal plane and the
    backside of C5). Its R magnitude was 3.7 on JD 2456244 according to AAVSO, so this is safe (PASSED)
  • reflection between the focal plane and the front surface of C4 (yet undetermined)
  • spray at filter cage (yet undetermined)