(This section is essentially the same as Section VI.1e.)
To evaluate the internal completeness and reliability, we examined the source data for more than 25 regions that were repeated 6 times in rapid succession. The data span a broad range of source density and were obtained on 5 separate nights in 1997 and 1998 from the Northern Facility, as described in http://spider.ipac.caltech.edu/staff/slw/cnr/cnr_results.html. S. Wheelock took these scan data and computed the number of times, N, a source could have been seen compared with the number of times, M, the source actually was seen (ideally, M=6). Wheelock trimmed the areas to ensure adequate geometric overlap. Her analysis of the individual scans can be found at the above URL. We accumulated the counts for all these scans into N/M bins, as shown in Tables 1-3 below. We used this information to generate the completeness and reliability (C&R) numbers given in the tables and in Figures 1, 2, and 3 (J, H, Ks, respectively) following Wheelock's prescription.
|Figure 1||Figure 2||Figure 3|
The Level 1 requirements on the Catalog are a differential completeness of 0.99 in the half-magnitude wide bin at the catalog limits (15.3-15.8 mag at J, 14.6-15.1 mag at H, and 13.8-14.3 mag at Ks). The reliability value, at the same magnitude limits, is meant to be 0.9995. It is clear from the tables that the 2MASS data handily meet the completeness requirement. However, the reliability reaches a plateau of about 0.995, which does not improve appreciably at brighter magnitudes. There are two reasons for this effect. First, these source counts have not had the benefit of the rigorous artifact removal that has been applied in the final Catalog. Thus, the reliability numbers are lower limits to the quality of the released Catalog. Second, examination of individual sources deemed unreliable by this type of analysis usually turns up marginally-resolved double stars or galaxies that come and go on different apparitions, depending on seeing and other effects. The reliability in the sense that there is one (or more) infrared sources at the quoted position is considerably closer to the stated requirement than indicated by the numbers in the tables. The best test of actual reliability will come from examination of deep co-added 2MASS images or comparison with external catalogs.
Wheelock is presently extracting the source data from all the scan overlap regions in this Catalog Release. This information, when it is available, will be representative of the completeness of the Catalog as a whole. However, the overlap data will be somewhat harder to analyze, since, with only two coverages (M=2, rather than M=6), it is harder to separate reliability and completeness problems.
|Table 1. J-Band C&R|
|Table 2. H-Band C&R|
|Table 3. Ks-Band C&R|
[Last Updated: 2000 Feb 2 by C.A. Beichman. Modified 2000 Jul 21 by S. Van Dyk.]