IRAC: Detection of Saturated Stars

Saturated stars prove difficult to identify because their cores pixels are not behaving properly due to non-linear saturation. Pieces of the bright star may be identified, but the actual star center is difficult to identify with automated method.

Consider the example below:

IRAC.1.0007726080.0001.0000.bcd_fp.fits, a Ks=2.8 mag star, as seen in IRAC channel 1.

Note the deleterious artifacts from the bright star -- diffraction spikes, reflection ghosts, persistence ("latent") ghosts, column droop, and saturated pixels.

Line plot cutting across the nucleus of saturated star. Note the "crater" profile.

Methods

Three methods are used to identify *potentially* saturated stars. They are (1) visual inspection, (2) "blind" search for crater profiles, (3) use the 2MASS PSC to identify bright stars. The first method, visual inspection, is the most reliable method for the brightest stars, where the crater is easily identified. The expert user can identify the symmetric core of the star most reliabily (which is crucial to rectification of lost/saturated pixels). The second method uses an automated search that looks for reverse local maxima (i.e., craters) for the entire image in a blind or exhaustive fashion. The third method, using 2MASS, quickly identifies which stars might be saturated, narrowing the search space, then a walking centroid (or crater-finding) method can be used to zero-in on the nucleus.

Visual Inspection

It is very easy to identify bright, saturated stars (see example above), including the approximate location of the true center of the saturated star. This is probably the best method for dealing with very bright stars where the crater is large enough to defeat centroiding techniques. The downside is that it requires tedious user interaction, and the method is not condusive to moderately saturated stars (which may look normal to the "eye").

Automated Crater Finding

The idea is to identify pixels that have reversed their point-spread profile, from a positive maxima, to a positive or negative minima. The algorithm developed for IRAC stars with pixels that have a large positive value (i.e., the "wings" of saturated stars). It looks for craters or minima in close proximity to bright pixels. This technique tends to work best for sources that are moderately saturated, where clear craters are formed. For very bright stars, the technque can recognize craters, but it may find multiple craters per sources. Moreoever, the crater that it finds may be far from the actual center of the star (since non-linear pixels behave in unpredicable ways). For these cases, visual inspection is better at closing in on the true center of the star.

Using the 2MASS PSC

The 2MASS point source catalog is a handy way to identify bright stars. The Ks-band mag is closely related to IRAC channel 1 (the "L" band), and so the 2 um integrated flux should be a good metric for how bright the star at the short IRAC wavelengths (this is not necessarily true for channels 3 and 4). Using 2MASS, bright stars may be identified and checked for "crater" behavior. The down side to this method is that the astrometric position of the bright star might not be well determined for either 2MASS or IRAC --- for 2MASS the positions should be good to less than 1", while for IRAC the astrometry is comparable (or perhaps a bit worse).

A method has been devised that combines the 2MASS PSC identification of bright IRAC stars with a walking centroid that identifies the nucleus of the bright star. See a description of the tools to perform this task.

IRAC channel 4.

2MASS PSC sources shown with red circles. The blue circle denotes a "crater" that has been identified. Note how the 2MASS position is well offset from the true center of the star; nevertheless, the crater is found at approximately the correct position. See below.