VI. Analysis of the 2MASS Second Incremental Release Catalogs

7. Comparison to External Catalogs and Models

d. Point Source Astrometric Performance

The Level 1 Requirement for point source astrometric accuracy is better than 0.5´´ with respect to the reference frame. It appears that the Second Incremental Release is substantially better than this, in that the available astrometric tests indicate that the error is of the order of 0.15´´ or better.

The design of the astrometric reconstruction algorithms take advantage of two key recent advances in astrometry. First, the adoption of USNO's ACT catalog provides a deep, dense set of positional standards whose errors are known to be less than or equal to 0.030´´. Adoption of this catalog avoids intermediate astrometric catalogs and the potential for the introduction of errors. Second, each 2MASS scan is separately tied to the reference catalog. This means that should a scan have a significant error, that error does not propagate into adjacent scans.

Three separate astrometric tests were applied to the Second Incremental Release Catalog. The overall conclusions demonstrate the astrometric quality.

• The Second Incremental Release and Tycho-2 were correlated, and the residuals for the ~1.0 million ACT stars were computed. The observed total dispersion (quadrature sum of the errors in the Tycho-2 and SIR positions) was found to be 0.110´´ in each coordinate. It is believed that this error is smaller than the true value because these stars were used (in a least squares sense) in the astrometric reduction, and the associated correction to the number of free parameters cannot be evaluated easily.

• The Second Incremental Release and Tycho-2 were correlated, and the residuals for the ~1.5 million stars new to Tycho-2 were computed. The observed total dispersion was found to be 0.180´´ in each coordinate. The increase in dispersion is believed to be attributable to the increased astrometric error in the Tycho-2 catalog arising from the fainter nature (about 1.5 magnitudes) of these stars.

• The Second Incremental Release and the internal release of USNO's UCAC catalog were correlated, and the residuals for the ~10 million stars in common were computed. The observed total residual was found to be about 0.120´´. UCAC covers the declination range of -90° to -5°, so the entire Second Incremental Release could not be tested, but the same reduction algorithms were used for the entire Second Incremental Release Catalog. UCAC stars go as faint as R=16 mag, so this test covers substantially different stars than are involved in the Tycho-2 tests. ACT goes to about V=11 mag, and the new to Tycho-2 stars go to about V=12.5 mag.

There is the possibility that an isolated scan may have poorer astrometry, and future studies of the stars in the scan-to-scan overlap regions can test for such errors and remove them. However, it is clear that the astrometric accuracy of the typical Second Incremental Release position is about a factor of 3 better than the Level 1 Requirement.

[Last Updated: 2000 January 24 by D. Monet. Modified 2000 Aug 29 by S. Van Dyk.]