Extended Sources in Close Proximity to Bright Stars

T. Jarrett, IPAC
(990310)

Bright Stars (K < 6th mag) are the source of a particular class of artifact seen in the extended source database. False galaxies associated with bright star "halos" or diffraction spikes or horizontal streaks are difficult to identify except that they are in close proximity to the progenitor.

Using the "release" database, I have gathered all of the bright stars, K < 8th mag, and compared their positions with those of the extended sources also found in the "release" database. The objective is to identify potential false galaxies and to determine empirical 'masking' radii to clean areas around bright stars.

I show both the extended sources that (in principle) should be part of the release "catalog" and all of the remaining extended sources in the DB (mostly faint sources). See more on the ext. src release DB and catalog jazz here.

Note: Bright stars are separated according to a K-band mag range. However, I do allow J and H mags to pull the bright star into the bin if it satisfies color range: Hmag within K_threshold+0.30 mag, Jmag within K_threshold+0.70 mag.

(-2 < K < 2) , 48 total stars

(2 < K < 3) , 537 total stars (3 < K < 4) , 6000 total stars (4 < K < 5) , 15500 total stars (5 < K < 6) , 19500 total stars (6 < K < 7) , 91869 total stars

The "deficit" of sources within 100 to 200" of bright stars is due to the incompleteness from bright star masking. There is no evidence of a significant "artifact" population due to bright stars. That is not to say that these artifacts do not exist (in fact, they most certainly do exist). Some of the sources within 200" of the bright stars may be false galaxies. The results of follow-up on some of these sources is given below.

Detailed Look at Some Sources Very Near Bright Stars

Most of the sources close to bright stars turn out to be real stars within the halos of bright stars. In turn, the bright stars are almost always on a coadd edge. This result is to be expected. Tracey's "DBMAPCOR" should be able to clean these spurious sources from the catalog.

Notes.

Results for One Night: 971116n, the "Sampler" Catalog

(-2 < K < 2) , 5 total stars (including a -1 or -2 mag boomer)

(2 < K < 3) , 22 total stars (3 < K < 4) , 23 total stars (4 < K < 5) , 72 total stars (5 < K < 6) , 184 total stars (6 < K < 7) , 427 total stars (7 < K < 8) , 866 total stars
(2 < K < 8) , >50,000 sources

Removal of Bright Star Artifacts

In principle removal of bright star artifacts is straight forward. Flag (or remove) sources within close proximity to bright stars. Proximity is this case means to: circular halo emission, diffraction spikes, horizontal stripes, glints and ghosts. For extended sources, the primary components to avoid are the bright star halo emission and diffraction spikes. It is the job of "DBMAP-CORR" to locate and flag sources associated with bright stars. However, as things stand (990317), this database module may not be ready in time for catalog generation (the 1st 2MASS incremental release). As such, I have written a mini-version of this module to be run on the extended source catalog (the reason this is possible is that the ext. src. cat is much smaller than the pt. src cat and can be downloaded to disk where I can manipulate it at will).

Bright stars on edges of coadds, are in particular, serious progenitors of artifacts. The horizontal stripes and persistence ghosts are ignored for simplicity. JCAT is used as the seed source.

Halo Proximity: halo radius is a function of bright star flux. See more on this in Bright Star Parameter Tuning and Automated Bright Star Masking .

Similarly, the diffraction spike radius is a function of the bright star flux. SEe the above links for information.

Both the halo and spike radii are modified when the source density is high; or put another way, the confusion noise affects the radii. The modification is performed in two ways: the bright star mag limit is throttled back with the confusion noise mag and the radii are descreased as a function of density. The prescription used is the same as that used in GALWORKS.