Appendix 1. Introduction to the 2MASS Extended Mission and Ancillary Data Products


4. Cautionary Notes

b. Survey Point Source "Reject" Table, 6x PSWDB and PSC, Calibration and LMC/SMC Calibration PSWDB

  1. Additional Information Available
  2. Using the Extended Mission Point Source Extraction Tables
  3. Unreliable Extractions - The Reliability Flag (rel)
  4. Multiple Source Detections
  5. Statistical Flux Overestimation For Faint Sources
  6. Scan Edge Effects

i. Please see Cautionary Notes for the 2MASS All-Sky Release Point Source Catalog (I.6.b)

This section contains brief summaries of the key limitations common to the Survey PSRT, 6x PSWDB/PSC, Calibration and LMC/SMC Calibration PSWDBs. It is not a complete collection of cautionary notes for each of the data sets. Please review the more complete descriptions for each product at:


ii. Using the Extended Mission Point Source Extraction Tables

The Survey PSRT and 6x and Calibration PSWDBs are not "Catalogs" in the same sense as the highly reliable, uniform All-Sky PSC. As a result, they are not suitable for statistical investigations of complete samples..

The 2MASS Survey PSRT and 6x and Calibration PSWDBs contain a complex mixture of measurements of real astrophysical sources, and spurious detections of faint, low SNR noise excursions, image artifacts and transient events such as cosmic ray strikes and meteor trails. Because of possible redundant source detections and the unreliable source extractions, the various Extended Mission point source tables do not provide valid statistical information about near infrared source populations such as those given by the the 2MASS All-Sky PSC. The exception to this is the 6x-PSC which was generated using criteria for reliability and uniformity similar to the All-Sky PSC, but over limited areas.

The Survey PSRT, 6x-PSWDB and Calibration PSWDB are resources for supplementing source information found in the All-Sky and 6x-PSC, or other external sources. For example, the Survey PSRT contains one or more independent measurements of 150,569,237 objects found in the All-Sky PSC that can be used to improve photometric and astrometric knowledge, or to search for photometric variability and/or proper motion. The Calibration PSWDB contains the individual epoch measurements of objects detected hundreds to thousands of times, enabling the generation of very high precision photometry and detection of very small motions. The detection statistics of sources observed multiple times found in the the Survey PSRT and 6x and Calibration PSWDBs also provide an additional source reliability information. A convenient way to utilize the multi-epoch information contained in the Extended Mission point source tables is via the 2MASS Survey, 6x and Calibration Merged Point Source Information Tables. In these tables, the independent measurements of objects detected more than once in each data set have been combined to provide improved photometry, astrometry and detection statistics.

The Extended Mission point source tables also contain real detections of faint sources that may have been excluded from the All-Sky and 6x PSCs because they fall below the SNR threshold for reliability. These faint source extractions must be used cautiously, though, because they are subject to flux overestimation biases. Faint PSWDB extractions also have a much lower statistical reliability than brighter objects because of confusion with the large number of spurious detections of the faint noise excursions. For example, there is a 1-1.5% probability that a search made within 2" of a random position of the sky will return a match to an unreliable, low SNR noise extraction in the 2MASS Survey PSRT. The probability of a random association with a noise detection increases linearly with the number of times a position was scanned, so searches in high coverage areas such as the equatorial poles in the main survey, and the calibration fields are very likely to return spurious associations.

iii. Unreliable Extractions - The Reliability Flag (rel)

The 2MASS Survey PSRT and 6x and Calibration PSWDBs contain large numbers of unreliable "source" extractions.

Source detection limits were set intentionally to relatively low (SNR>3) levels during all 2MASS pipeline data reduction to insure completeness of the All-Sky and 6x Catalogs. This resulted in the detection and measurement of real astronomical objects as well as low SNR noise excursions, artifacts caused by bright stars such as diffraction spikes, latent images and ghosts, and transient events including residual meteor streaks, cosmic rays or hot pixels. All of these "source" detections were loaded into the Survey, 6x and Calibration PSWDBs. The All-Sky and 6x PSC were drawn from the superset of detections in the respective PSWDBs by selecting those detections that satisfy various reliability criteria (see V.3 and A6.3.c).

To help discriminate between real source detections and the many spurious extractions that are contained in the Survey PSRT and 6x and Calibration PSWDB, each database entry has been assigned a reliability flag value, rel, that is related to the probability that it is a detection of an astrophysical object at the time of the 2MASS observation. The reliability flag is a single character in the range "A" to "F", with "A" representing sources with the highest reliability and "F" the lowest. Appendix 5 contains a description of the algorithm used to assign the reliability flag values and limitations of the estimator. Table 1 contains a breakdown of the Survey PSRT and 6x and Calibration PSWDBs according to reliability flag value, and the corresponding probability of reliability associated with each value of rel. The distributions in Table 1 show that a significant fraction of entries in each database are not reliable detections of astronomical sources. The Survey PSRT has the poorest fractional reliability, containing nearly 60% unreliable extractions.

Select Survey PSRT or 6x and Calibration PSWDB sources having a reliability flag value of rel="A" to minimize the number of spurious extractions. Caution should be exercised when using any source with a lower probability of reliability. However, even the rel="A" sources in the PSRT and PSWDBs have not received the same degree of scrutiny as the highly reliable All-Sky PSC. When in doubt about the reliability of any source extraction, we strongly recommend examining the image of the source using the Survey, 6x or Calibration Image Atlas.

Table 1 - Breakdown of Extended Mission Point Source Table Entries by Estimated Reliability (rel)
rel ValueProbability of ReliabilityNumber
Survey PSRT6x PSWDBCal PSWDBLMC/SMC
Cal PSWDB
A>90%24389280715604449905154647221609
B80-90%3730412513907276462533537858
C70-80%253030485706393416630411962
D50-70%322272085831224453285551256
E30-50%72557964100793810963858954695
F<30%4327037454866827756522503259310


Unreliable "Reject" Table and WDB entries generally exhibit one or more of the characteristics described in Table 2. Users should scrutinize such sources carefully. However, possession of any one or more of these attributes does not necessarily guarantee that a detection is unreliable, only that it merits examination. We strongly advise examining the images of any Survey PSRT, 6x or Calibration PSWDB entry whose reliability may be in question.

Table 2 - Common Characteristics of Unreliable Extractions in the Extended Mission Point Source Tables
ConditionSource Attribute
Single band detection rd_flg="0" in two bands
Contamination or confusion flagging cc_flg != "000"
Large value of profile-fit photometry 2 value rd_flg="2" and [jhk]_psfchi >> 1
Low frame-detection count for SNR>7 sources ndet[1,3,5] 1 and [jhk]_snr >7
READ1 detection that is fainter than the READ2 saturation limits (rd_flg[1]="1" and j_m>11) or (rd_flg[2]="1" and h_m>10.5) or (rd_flg[3]="1" and k_m>10)
Source is fragment of an extended sourcegal_contam="2"
Source is associated with a solar system objectmp_flg="1"


iv. Multiple Source Detections

The 2MASS Survey PSRT and 6x and Calibration PSWDBs may contain multiple, independent measurements of objects that fall in regions observed multiple times during 2MASS survey, 6x and calibration observations. Because of this "over-completeness," the Extended Mission point source tables are not suitable for studies based on source count statistics. Use the 2MASS All-Sky PSC and 6x-PSC for this type of inquiry.

Because the Survey, 6x and Calibration PSWDBs contain every source detection extracted from all 2MASS observations made under photometric conditions, they may contain more than one independent measurement of objects in regions that were scanned multiple times. The number of duplicate source detections within each data set varies considerably across the sky, depending on the observation depth-of-coverage, and it is different between the data sets. See A2.2, A3.2 and A4.2 for descriptions of the coverage of the survey, 6x and calibration observations, respectively.

The Survey, 6x and Calibration Merged Point Source Information Table provides combined brightness and position measurements for multiply-detected point sources in the respective observations, and statistics on source brightness and position distributions that are useful for identifying variables and moving objects. Source confirmation statistics from the Merged Source Information Tables are also included in the Survey PSRT and 6x and Calibration PSWDBs: the spos and sdet columns give the number of scans that covered the position of a source and the number of unique scans in which the source was detected, respectively.

v. Statistical Flux Overestimation of Faint Sources

Flux measurements of faint source extractions with SNR<8-9 in the 2MASS Survey PSRT and 6x and Calibration PSWDBs are overestimated because of statistical measurement bias. This bias is ~5% at SNR~7 and increases rapidly with decreasing SNR.

Photometry of faint source extractions in the Extended Mission point source tables is affected by statistical flux overestimation bias. This bias, described in V.3.a.ii, results naturally when measuring sources with non-zero uncertainty at true flux densities near system sensitivity limits. Sources will preferentially be detected when noise drives up the apparent flux rather than driving it down. Therefore, sources detected near the sensitivity limit will have, on average, a measured brightness higher than their true brightness, or equivalently a higher SNR than their true value. Such sources will also have measurement errors that do not accurately represent their true SNR. The closer a measured brightness is to the detection limit, the larger the amplitude of the statistical overestimation.

An illustration of the characteristic flux overestimation bias in the 2MASS 7.8 s READ2-READ1 exposures is shown in Figure 1. In the top panels of this figure, differences between aperture photometry extracted from the very deep, Combined Calibration Field Images of the 90067 calibration field and the average calibration scan photometry for the field from the Calibration Merged Point Source Information Table are shown plotted as a function of average magnitude from the Merged Information Table. The combined calibration images have an effective sensitivity approximately four magnitudes fainter than the single scans and are therefore are free from bias at the flux limits of the individual scans. The bottom panels of Figure 1 show the RMS (root variance) of the merged calibration scan photometry plotted as a function of average magnitude. The RMS of the merged calibration field photometry is representative of the sensitivity from individual survey scans, averaged over all observing conditions.

The upturn in the main locus of points in the top panels of Figure 1 shows that statistical flux overestimation sets in at J>16.3, H>15.2 and Ks>14.7 mag. Reading down to the bottom panel, these magnitudes corresponds to SNR~8-9. By SNR=7, the flux overestimation is nearly ~0.05 mag, and rises steeply exceeding 0.2 mags at SNR=5. This behavior agrees well with the bias predicted from simulations in V.3.a.ii. The scattered points that fall below that main locus in the upper panels are confused sources, such as close multiple stars, whose brightness is typically overestimated by the simple aperture photometry performed on the Combined Calibration Field images. A flattening of the RMS vs. magnitude curves below SNR~7 that is also a manifestation of incompleteness and flux overestimation can be seen in the bottom panels of Figure 1. At the lower SNR levels, sources are not detected when noise fluctuations drive the flux down. This effectively truncates the faint end of the measured flux distribution leading to an artificially narrow distribution and hence an artificially low flux uncertainty and biased flux estimate.

Figure 1 - Illustration of statistical flux overestimation as a function of magnitude and SNR in the 90067 calibration field. (top panels) Difference in J, H and Ks magnitudes measured on deep, combined calibration scan images and the average magnitudes from single calibration scans plotted as a function of average magnitude. (bottom panels) RMS (root variance) of merged (average) calibration scan magnitudes plotted as a function of average magnitude. Red horizontal lines indicate SNR=5,7 and 10 levels.


vi. Scan Edge Effects

Point source extractions that fall within 10" of scan edges may be affected by: 1) incomplete band coverage because of small detector array misalignments; 2) biased photometry due to source or or sky reference apertures intersecting detector edges; or 3) degraded astrometric accuracy because of residual field distortion.

The 2MASS Survey PSRT and 6x and Calibration PSWDBs contain source detections that extend to within approximately 4" of the individual detector array edges. The completeness, centroiding and photometric measurements of objects close to the detector edges can be flawed for the reasons described below. In order to meet Catalog photometric and astrometric accuracy requirements, sources within 10" of scan edges were excluded from the All-Sky PSC (and 6x-PSC). There are usually alternate detections of sources in adjacent scans that are farther from their scan edges, so completeness in the All-Sky PSC is not impacted. This may not be the case for the Extended Mission data sets for which scans do not have overlapping neighbors, such as the 6x and calibration field observations. Users of the Survey PSRT and 6x and Calibration PSWDBs should be wary of extractions within 10" of scan edges. The east/west and north/south scan edge distances for each point source extraction are given by the dist_edge_ew and dist_edge_ns parameters in each record in the Extended Mission point source tables.

Incomplete Band Coverage

The three arrays in each 2MASS camera were aligned to within ~10" of a common center (see Figure 8 in III.1.b). However, sources that fall at the very edge of the focal plane may not illuminate all three detectors. This can produce a source extraction that is artificially missing detections in one or two bands

Corrupted Photometry

Profile-fit and aperture photometric measurements of objects within 5-10" of detector edges may be corrupted or biased because the sky reference annulus or source measurement kernel or aperture intersects the array edge. This can result in either over- or under-estimated fluxes, or outright measurement failures. Measurements near detector edges often have elevated photometric uncertainties ([jhk]_cmsig and [jhk]_msig) and/or profile-fit chi-squared values ([jhk]_psfchi), and sometimes have failed measurements signified by their rd_flg values (rd_flg="9").

Degraded Astrometric Accuracy

Comparison between the positions of 2MASS All-Sky PSC sources and their counterparts in the UCAC show that the mean radial astrometric offsets increase approximately 15% (18-19 mas) for objects close to the cross-scan edges of the 2MASS focal plane because of residual field distortion (See Figures 15-17 in VI.6.a) This slight increase also means that the quoted 2MASS astrometric uncertainties may be slightly underestimated for such sources.

[Last Updated: 2008 February 12; by R. Cutri]


Previous page. Next page.
Return to Explanatory Supplement TOC Page.