DB_MAPCOR Point Source Results

Introduction

MAPCOR, working on one band of one scan at a time, can find bright star artifacts that are limited to the scan in which the star appears, namely persistence and filter and dichroic glints; DB_MAPCOR deals with the artifacts that are not limited to the same scan and thus require a database search. For 2MAPPS version 3, DB_MAPCOR searched for point sources photometrically contaminated by flux from other nearby stars (photometric confusion), artifacts caused by very bright stars (artifact confusion), point sources contaminated by diffraction spikes, and artifacts caused by diffraction spikes. (For more information on the search algorithms, see the  Artifact Identification Algorithms and Parameters page.)

Parent Source Lists

The parent source list was drawn from the working database point source table, selecting sources brighter than 10 mag in at least one band which passed the  catalog generation working point source database criteria. These are, essentially, reliable point sources found in the release scan set, located at least 10 arcsec from the scan edges and not already identified as artifacts by MAPCOR. Since the artifact identification parameters are different for the northern and southern telescopes, the parent list was split into two lists to search each hemisphere separately for artifacts and contamination. Note, though, that many parents were used for both hemispheres because many tiles were observed from both hemispheres; the northern hemisphere list includes all parents with Dec > -14.5°, and the southern hemisphere list includes all parents with Dec < 38.16°. The histograms of the magnitudes of these parent sources are below; they include each apparition of duplicate sources (sources found on more than one scan).

J Band Magnitudes	H Band Magnitudes	Ks Band Magnitudes
Northern Hemisphere
Southern Hemisphere

Because of time constraints, DB_MAPCOR was only run on the parent sources with J, H, or K_s magnitudes <= 5.0. This includes a total of 35720 parents for the northern hemisphere and 56413 for the southern hemisphere, not counting duplicate source apparitions. Since 23882 parents were used in both hemispheres, there are 68249 total distinct parents with some magnitude <= 5.0. However, DB_MAPCOR skips parents without valid detections in at least two bands. (For details, see  Single-Band Parents Skipped.) This removed 18 parents from consideration, 13 for the north and 16 for the south, leaving 35707 parents for the north and 56397 for the south. For reference, it took about three weeks of wall-clock time to run these sets of parents on two 4-processor machines, with about five jobs running on each machine at a time. Of these parents, 39992 caused new artifact flags in the J band, 35008 in the H band, and 49247 in the K_s band.

For analysis purposes, we are currently running DB_MAPCOR on the parent sources with J, H, or K_s magnitudes between 5.0 and 6.0. There are 69544 of these parents for the north and 102062 for the south, approximately twice the numbers of parents with magnitudes <= 5.0. However, the wall-clock runtime will be approximately the same for both sets, since the fainter parent sources require smaller search areas.

Possible Artifact Flag Values

DB_MAPCOR can update four 3-character artifact flags, but it will never lessen the severity of a flag already set by MAPCOR. Since artifacts already found by MAPCOR were excluded by the  catalog generation working point source database criteria, DB_MAPCOR didn't waste time searching through known artifacts.

One 3-character flag for each band, called j_prg_flg, h_prg_flg, and k_prg_flg, contains complete artifact flag information for each band in numerical codes. DB_MAPCOR can update these flags with the following values (``?'' means any single character, as in csh wildcards):

• ``?5?'' indicates that the source is bright enough to be accepted as a real source but is contaminated by a diffraction spike.
• ``?6?'' indicates that the source is likely part of a diffraction spike.
• ``??1'' indicates that the source is found within the photometric confusion radius of a parent source.
• ``??2'' indicates that the source is found within the artifact confusion radius of a parent source and is likely false.

After the [jhk]_prg_flg flags are set by both MAPCOR and DB_MAPCOR, the fourth 3-character flag, cc_flg, combines all the possible artifact and contamination values into a one-character code for each band. To set this flag, the [jhk]_prg_flg and persistence probability values for each band are checked in the order given below until a value is applied. This means that possible flag values further down this list will not be applied if a flag found above it is already set; in other words, the order indicates the flag hierarchy.

The cc_flg characters for each band can be:

• `P': Persistence probability ?_prob_pers > 0.5.
• `G': Source is positionally associated with a filter glint. (?_prg_flg = ``1??'')
• `C': Confusion Artifact. Source is likely a spurious detection of the confusion halo around a bright star. (?_prg_flg = ``??2'')
• `D': Diffraction Spike Artifact. Source is likely a detection of a diffraction spike from a nearby bright star. (?_prg_flg = ``?6?'')
• `p': Persistence probability 0.1 < ?_prob_pers <= 0.5.
• `c': Photometric Confusion. Source has biased photometry due to proximity to a bright star that causes an incorrect measurement of the background. (?_prg_flg = ``??1'')
• `d': Real source that is affected by proximity to a diffraction spike from a nearby star. (?_prg_flg = ``?5?'')
• `s': Source is likely a detection of or affected by a NICMOS3 ``stripe'' from a bright star at delta_y = 0, 256, -256 arcsec. (?_prg_flg = ``??1'')
• `b': Bandmerge confusion -- more than one bandmerge candidate. (cnf_flg for that band is >1 in either the first or second digit)
• `0': Source is unaffected by artifacts, or is not detected in that band. (?_prg_flg = ``000'' or ``fil'' or ``upl'')

The capital letters indicate that MAPCOR or DB_MAPCOR determined that the source was an artifact in that particular band; lower case letters indicate that the source is likely real but contaminated by an artifact in that band. Before the DB_MAPCOR processing, cc_flg values found in the catalog generation working point source database would have had only the lower case letters or '0' in all bands because of the selection criteria. DB_MAPCOR could only change a flag from a less-severe value, i. e. one lower in the list, to a value of 'd', 'c', 'D', or 'C'.

Artifact Flag Results

After DB_MAPCOR sets new artifact flag values, another program merges those values with those already found in the working tables so that no information is overwritten unnecessarily. The full lists of new artifact flag values set by DB_MAPCOR and of all of the values found in the catalog generation working point source database after this merge can be found in  this page. Summary tables of the new values set by DB_MAPCOR, affecting 527063 point sources, are below.
 

Summary tables of the post-merge values of all sources in the catalog generation working point source database are below.
 

Analysis Results

Analysis of the DB_MAPCOR results in the test fields by Raymond Tam can be found in  this page. The results indicate that some of the adjusted spike lengths in high source density regions were actually a bit too short, causing DB_MAPCOR to miss eight spike artifacts and one source contaminated by a spike in the test fields. On the other hand, 11 sources were flagged as confusion or spike artifacts by DB_MAPCOR that are most likely real sources. For this analysis, sources found around 17 very bright stars in the test field catalog generation DB were examined by eye.