Please refer to IV.6 for a detailed description of the astrometric calibration procedure used for primary 2MASS survey data. Changes were made to this procedure to account for the smaller area of many of the 6x scans, to overcome difficulties introduced by the longer Read_2 exposure times relative to Read_1, and to correct known problems in final survey processing.
i. 2MASS Reference Catalog Used for 6x Position Reconstruction (2MREF)
The most significant difference between astrometric calibration of the 6x and main survey data is that the 2MASS All-Sky Point Source Catalog was used as the primary astrometric reference for source positions rather than the Tycho-2 Catalog.
Using the Tycho-2 Catalog as the reference for 6x processing would have presented two problems. The first is that the Tycho-2 density is frequently too low to provide reliable reconstruction for the short 1° 6x scans. This is similar to the problem faced for the 1° calibration scans during main survey processing. Because absolute accuracy was not required for the calibration scan data, the USNO-A2.0 Catalog was used as the astrometric reference which resulted in systematic offsets between calibration scan and survey scan astrometry (A4.1.c). The second basic problem in using Tycho-2 for 6x data reduction is that the brightness overlap between the Read_1 and Read_2-Read_1 exposure regime is significantly reduced. Tycho-2 stars tend to be bright and most saturate the 1.3 s Read_2-Read_1 exposures. The positions of stars in the 2-3 magnitudes of non-saturated overlap between Read_1 and Read_2-Read_1 were used to tie the reference frame together for the main survey.
To address these challenges, a subset of the 2MASS All-Sky Point Source Catalog was selected to serve as the astrometric reference catalog for 6x processing. This PSC subset is referred to as the 2MASS Reference (2MREF) catalog, and was chosen to provide the best astrometric quality while retaining a high reference star density. The 2MREF was selected from the 2MASS All-Sky Catalog using the following criteria:
These criteria select reliable, non-contaminated, unresolved sources that have relatively high signal-to-noise ratios in the non-saturated Read_2-Read_1 brightness regime. The resulting reference catalog has an overall average density of ~2600 stars/deg2, with an average internal accuracy of ~70 mas in each axis.
It was assumed that the 6x frame positions were known to within a few arcseconds and matching to the reference catalog could be done based on a priori positions. Thus, no scan-to-scan corrections were made to the a priori positions, as was done for the main survey data. The high density of the 2MREF pulls the frame solutions very nicely back into agreement with the All-Sky PSC.
ii. Read_1 Distortion Problem Correction
Although distortion was corrected for both Read_1 and Read_2-Read_1 sources during survey pipeline processing, due to a coding error the Read_1 source positions were sent to band-merging before the correction was applied. This error results in a small position bias between PSC sources in the two broad brightness regimes that varies with cross-scan position and between observatories. Since Read_1 and Read_2-Read_1 data were lumped together, the Read_1 problems were masked by the far larger set of Read_2-Read_1 data points, and residual differences were attributed to the small sample size. This coding error was corrected for 6x data reduction, and monitoring was done using very large samples and analyzing separately Read_1 and Read_2-Read_1 data.
iii. Tycho-2 Associations Expanded
Tycho-2 stars with lesser quality astrometry were not used during data reduction for the main survey. These included Tycho-2 catalog entries with the "pflag" set to "P" or "X" and/or the "posflg" set to "D" or "P". Because optical association was integrated into the position reconstruction process within 2MAPPS, the excluded Tycho-2 stars had to be matched during a post-pipeline a processing step via a database comparison between the All-Sky PSC and the entire Tycho-2 catalog.
For 6x processing the full Tycho-2 catalog has been used for the optical association process. The previously excluded, poorer quality entries are used for associations only, and are not included in statistics used to verify 6x reconstruction quality. A subset of Tycho-2 catalog sources are components of double stars that have the same position and different magnitudes. Because only one optical association was made with each 6x extraction, some Tycho-2 catalog entries will not be associated with 2MASS sources.
iv. Adjustment of Positional Uncertainties
In order to obtain absolute positional uncertainties, it is necessary to account for errors in the frame positions generated during the first stages of position reconstruction. During survey processing, these frame uncertainties were available as a result of the scan-to-scan position comparison process done offline between the preliminary and the final data reduction. The uncertainties were added in quadrature to the relative positional uncertainties arising from bandmerging to provide the final positional uncertainties. This is not a practical approach for 6x processing because many of the 6x scans are not bounded by adjacent scans. However, the same end can be achieved through comparisons with the high-density 2MREF.
To do this, each 6x scan is broken down into half-degree segments, and statistics on the 6x-2MREF position differences within each segment are computed. The high density of 2MREF sources available in each segment insures that the 6x reconstruction is tied strongly to 2MREF reference frame with a negligible systematic difference in any segment. Thus, the rms of the 6x-2MREF differences reflects the random errors of the two catalogs. But we also know that the 2MREF contains systematic errors and that these systematic errors determine the uncertainty of the reconstructed 6x frames. Given the contribution of random error to the quoted 2MREF uncertainty, the systematic 2MREF error for the segment is backed out and then combined with relative 6x positional uncertainties from bandmerging, to give the final 6x positional uncertainties.
[Last Updated: 2007 January 7; by H. McCallon and R. Cutri]