Last Updated Jan 2001

Copyright © 2001 The Association of Universities for Research in Astronomy, Inc. All Rights Reserved.

Astrometry accuracy:
From comparisons with external catalogs the overall astrometric accuracy is between 0.2 - 0.4 arc seconds. We have used techniques to reduce the plate-based distortion pattern found on Schmidt plates. However, the GSC 2.2 positional accuary depends on magnitude and positions on the plate, i.e. the accuaracy of the faint stars (F> 15) varies from 0.2 arc seconds near the plate center to 0.6 arc seconds near the plate edges.

Astrometric calibration method:
The astrometric calibrations of the GSC 2.2 are based on reference catalogs that provide local representations of a fundamental coordinate system. The brighter stars appearing in the reference catalogs, i.e., V<8.5, are often heavily overexposed on the Schmidt plates and thus not useful in calibrating the plates. The astrometric reference catalogs for the GSC 2.2 reduction are the ACT and the Tycho 2 (see referance catalogs for more details). Since there are more than 400 reference stars on each Schmidt plate, each plate is tied directly to the reference catalog and hence to the International Celestial Reference System (ICRS).

The GSC 2.2 astrometric calibration was a two-step process. The first step involved calibrating each plate by polynomial modeling against the ACT and TY2 reference catalogs. The second step involved removing plate-based position dependent systematics the polynomial modeling failed to remove.

In order to compensate for differential refraction across the plate the measured x and y pixel values are pre-corrected for refraction using the method described by Konig (ref). Initial estimates of the right ascension and declination for each measured image were determined by traditional global plate modeling. A quadratic model was adopted to represent the relationship between the x and y plate measures (pre-corrected for refraction) and the tangential coordinates (c, h). Because the Schmidt telescope has a curved focal plane, an equidistant projection, instead of a gnomonic projection, was used to transform the reference star's equatorial coordinates into the tangential coordinates.Only the objects flagged as reference stars were used in a least squares reduction to determine the coefficients of the model. These coefficients were then be used to map all the x,y measurements to the associated (c, h) values.

Once a significant number of plates had been reduced in this manner (approximately 100) the calculated c and h coordinates were compare to those determined by projecting the equatorial coordinates of the ACT reference stars onto the tangential plane. For each plate the residuals were accumulated in 4.4 mm by 4.4 mm bins. Adding and averaging the residuals from all the plates yields the systematic pattern the global plate model failed to remove. This plate-based systematic residual map, which we call an astrometric mask, was then used to correct the tangential coordinates of the all of the image on the plate. We determined a separate astrometric mask for each survey.