For the UCAC positions given here, no new pixel reductions were performed. Instead, the object detection results (x,y,mag) of the UCAC4 data were used in new “plate” reductions after matching with the TGAS stars. New frame-by-frame solutions were obtained with TGAS instead of Tycho-2 reference stars (as used in UCAC4). Thus these UCAC5 positions are automatically on the Gaia reference system. Systematic errors were re-visited and new field-distortion patterns generated and applied separately for the northern and southern operations. The pixel-phase error corrections were improved as well.
No changes in the systematic error correction model as function of magnitude were made. These were obtained from earlier dedicated reductions with 2MASS stars to allow complete coverage of the UCAC magnitude range. The TGAS data only reaches to about magnitude 11.5 and is thus not suitable for a re-evaluation of the systematic errors as a function of magnitude (the UCAC limiting magnitude is about 16.5).
Nevertheless, residuals of the TGAS solutions were plotted as a function of magnitude and systematic errors in the accessible magnitude range are small (order 5 mas or less) for the global averages of all data. Local sub-sets of the data show somewhat larger systematic errors in the order of up to 10 mas, as expected from previous UCAC4 results.
A new catalog of UCAC mean positions from these reductions with TGAS was constructed containing over 180 million entries covering all-sky. These positions have mean epochs around 1998-2000 in the south, and around 2001 to 2003 in the north. Formal position errors are about 8 mas around magnitude 11, 20 mas at magnitude 14, and 60 mas at magnitude 16 (see 2017arXiv170205621Z).
The Gaia DR1 catalog supplemented by NOMAD proper motions as described in Step 1 was propagated to epoch 2000, sorted by Dec and then matched with the TGAS reduced UCAC mean positions as described in Step 2. Again, a match was assumed if the position difference is less equal to 1 arcsec in each coordinate. This resulted in 107.7 million stars in common between the UCAC and Gaia data. New proper motions were calculated for all those stars based only on the 2 epoch positions with a mean epoch difference of about 14 years.
For several millions of stars up to about magnitude 14 the errors on these new proper motions are on the same level as the TGAS proper motions (1 to 2 mas/yr, see below). This set of UCAC - Gaia matched stars with new positions and proper motions is called UCAC5.
Formal errors on the new proper motions were calculated by root-square-sum of a) formal position error of UCAC, b) formal position error of TGAS (both at their respective mean epochs), and c) a systematic error contribution of 10 mas at the UCAC epoch data. This RSS value was then divided by the epoch difference to arrive at the proper motion errors given in the UCAC5 data files.
The catalog source can be obtained at http://vo.ari.uni-heidelberg.de/ucac5.