Data products: General - How good are the proper motions?

The proper motions are calculated during target selection. Target selection provides proper motions by matching each SDSS object against USNOA2.0. By the rules of target selection, each SDSS object must be processed independently of other SDSS objects, which forces us to a very simple matching algorithm. The closest object within 30 arcsec is taken as the match, and proper motions are calculated using that match. Compare the description of the astrometric quality and the astrometry algorithm

The accuracy of the matching (that is, how many SDSS objects were matched to the correct USNOA2.0 object) has been calculated by comparison with a more sophisticated matching algorithm. The following table summarizes the results for a single scan (scan 752):

Thus, 99.9% of all objects which matched within 1 arcsec were correctly matched. The number drops only slightly to 97.1% for objects matched between 1 and 2 arcsec. Thus, out to proper motions of 4-5 arcsec/century (depending on the epoch of the POSS plates), representing 96% of the stars, the contamination of false detections is only 2-3% for those objects with measurable proper motions (greater than about 1.5 arcsec/century). The contamination increases to about 35% for the fastest moving 1% of the objects.

The accuracy of the SDSS positions are better than 100 mas, those of USNOA2.0 about 150 mas, and the difference in epoch about 50 years. Thus, a proper motion of 1.44 arcsec/century is a 4 sigma detection.