T. Jarrett, IPAC
Bright stars present a number of problems to finding faint extended sources --- primarily because bright stars falsely generate fuzzy objects (e.g., persistence ghosts) and filamentary structures (diffraction spikes and other reflection stripes). Consequently, bright stars and their associated features must be "blanked" from the coadds early on in GALWORKS (i.e., before background determination). This memo briefly discusses how this procedure is accomplished and shows two examples, one very bright star and two moderately bright stars.
Bright stars -- in general we mean stars that saturate in read-2 -- have several "features" that must be blanked from the coadd before GALWORKS can compute the background solution and find galaxies. These features include:
diffraction spikes: the "spider" pattern is north-south, east-west
persistence ghosts: roughly circular flat-profiled hotspots demarking the location of the star as it is imaged at different times along the scan; the spots decay with time: the first persistence feature is the brightest of the set, approximately 6 mags fainter than the star itself, with subsequent persistence ghosts sequentially decreasing in brightness 0.5 mag.
horizontal stripes: three stripes can be seen, two located near the +- 3rd persistence ghost and one located along the star central position itself.
other ghosts: very bright stars show reflection ghosts located all over the place (see case study below); fortunately they are not too bright.
Blanking Operations:
rectangular corridors, located atop the diffraction spikes
3 horizontal corridors, located atop stripes
square boxes located atop persistence ghosts
A very bright star is located about 1 degree south of M51. It is approximately 2nd or 3rd mag, which means it is probably saturating in read-1. Thus, we do not have a very good estimate of its true mag. Nevertheless, we must be able to deal with this sort of case. The images below show the raw coadd and the 'cleansed' coadd. Here we have applied the blanking criteria according to the best estimate of the stars mag (in this case, we probably severely underestimate the total flux of this star).
The J-band image is particularly striking (that is, it is truely horrifying). For super bright stars, I may need to apply a big blanking square, perhaps 200 or more pixels in size.
Since this star is so bright, it has "persistence" features that extend into the adjacent coadd and beyond. Also there is a horizontal strip that is located approximately at the 3rd "persistence" feature which is also located on the adjacent coadd. The following images show the coadd directly adjacent (northward) of the coadd with the bright star in it. The cleansed images show the persistence ghosts removed and the horizontal stripes blanked.
There is a coadd a couple degrees north of M51 that has two bright stars:
