Astrometric precision of observations at VLT/FORS2

Abstract

In this paper we test the astrometric precision of VLT/FORS2 observations using a serie of CCD frames taken in Galactic bulge area. A special reduction method based on symmetrization of reference fields was used to reduce the atmospheric image motion. Positional precision of unsaturated R=16 mag star images at 17 sec exposure and 0.55 arcsec seeing was found to be equal to 300 microarcsec. The total error of observations was decomposed into components. It was shown that astrometric error depends largely on the photon centroiding error of the target (250 microarcsec for 16 mag stars) while the image motion is much less (110 microarcsec). At galactic latitudes to about 20 degrees, precision for a serie of frames with a 10 min total exposure is estimated to be 30-50 microarcsec for 14-16 mag stars providing the images are not overexposed and the filter R "special" is used. Error estimates for fields with smaller sky star density are given. We conclude that astrometric observations with large telescopes, under optimal reduction, are never atmospheric limited. The bias caused by differential chromatic refraction and residual chromatism of LADC is considered and expressions valid for correcting color effects in the measured positions are given.