Reliability Limit

 

CAUTION! The 2MASS Point and Extended Source Working Database tables contain extractions of real astrophysical sources as well as a large number of spurious extractions of noise, image artifacts and transients such as cosmic ray strikes and meteor trails. A reliability flag, rel, that indicates the estimated probability that an extraction is a valid detection of an astrophysical source is provided with each extraction. For extended sources, the reliability flag also indicates the confidence level that the extraction is resolved relative to the 2MASS point spread function.

Users must specify a Reliability Limit when querying the 2MASS Working Databases, that is based on the reliability flag. Only sources having reliabilities within the specified limit will be returned from the query.

The reliability flag is a single character flag (A-F), with "A" indicating the highest probability of reliability and "F" the lowest:

Reliability Scale:
relProbability of Reliability
A>90%
B80-90%
C70-80%
D50-70%
E20-50%
F<20%

The default value for the Reliability Limit is A. This will return only sources having rel='A' (reliability probability >90%) that satisfy any other query conditions. Selecting a reliability limit of A-B will return extractions having rel='A' or 'B' (reliability probability >80%), and so on. The table below summarizes the returned reliability flag range and reliability probability for each value of the Reliability Limit.

Reliability Limit:
Reliability Limitrel matchesReliability Range
A A >90%
A-B A or B >80%
A-C A or B or C >70%
A-D A or B or C or D >50%
A-E A or B or C or D or E >20%
A-F A or B or C or D or E or F all

 

CAVEAT EMPTOR:

Selecting any reliability limit other than A will return unreliable extractions in increasing numbers. In particular, extractions with rel='E' or 'F' are very likely to be spurious detections of noise or bright star artifacts. The user should only use these "unreliable" data with extreme caution.
 

The Reliability Flag

For point sources, the reliability flag is based on the signal-to-noise ratio of the extraction's flux measurements, frame-detection statistics, the chi-squared point source goodness-of-fit statistics, and artifact identification algorithms. The probability of reliability is based on statistics derived empirically for large ensembles of extractions, and may not be perfectly accurate for any individual source. Users are encouraged to read the full description of point source reliability scoring provided in Section A2.2.b of the 2MASS on-line Explanatory Supplement.

For extended sources, the reliability score is also derived using a probabilistic method based on the brightness, color and morphology of extractions that is calibrated using visual inspection of tens of thousands of objects. The reliability estimate is further modified using any a priori information, such as artifact association. Users are encouraged to read the detailed description of extended source reliability flagging in Section A2.2.b of the 2MASS on-line Explanatory Supplement.

Note that the reliability flag is not intended to encode the accuracy of the flux and/or position measurement of an extraction. Users should refer to the appropriate flux quality and flux and position uncertainty measurements for this.