The reliability flag, "rel", indicates the confidence level that the source is truly extended. It is derived using a probabilistic method that employs the visual inspection "truth" table. The method is described here, and the results are given here.
The reliability is scored from 0 to 1, where 0.0 indicates a source is not extended or is bogus, and a score 1.0 indicates a source is almost certainly extended and real. For convenience, this score is converted to a letter grade, ranging from "A" (good source) to "F" (bad source).
rel = "A" Source has a high probability of being extended. The probability score ranges from 90 to 100% reliability.
rel = "B" Source has a moderate probability of being extended. The probability score ranges from 80 to 90% reliability.
rel = "C" Source has a moderate probability of being extended. The probability score ranges from 70 to 80% reliability.
rel = "D" Source has a poor probability of being extended. The probability score ranges from 50 to 70% reliability.
rel = "E" Source has a poor probability of being extended. The probability score ranges from 20 to 50% reliability.
rel = "F" Source is either bogus (artifact) or point-like. The probability score ranges from 0.0 to 20% reliability.
Sources that are within close promimity to large galaxies are marked as "chaff" and given a reliability of "F". This image shows M31/M32/M110 with "chaff" marked in green; the red point at the center indicates a source that is within 4" of the nominal position of the LGA galaxies.
A = 126,804
B = 65,189
C = 4,670
D = 12,227
E = 6,212
F = 31,884
An exceptional case of poor reliability is "galaxy chaff", or pieces of bright extended sources. Sources identified as "chaff" are given the cc_flag value of "z". Here is a link to the details. Most of the time the reliability of these sources is worthy of a grade "E" or "F".
But these sources may be real, discrete extended sources, in some cases (e.g., globular clusters around big galaxies), while their photometry is not of high quality (due to the background gradient from the large extended source). What is the reliability under these circumstances?
Three proposals: (1) assign a grade of "F" to all chaff sources (cc_flg = "z"); (2) assign a special grade of "U" to these sources, recognizing that the reliability is largely unknown; (3) leave as is (use the nominal quantitative reliability score).
Update: May 28, 2004
Distribution:
[Last Updated: 2004 Aug 3; by Tom Jarrett]
