Differential image motion in astrometric observations with very large seeing-limited telescopes

We investigate how to quantitatively model the observed differential image motion (DIM) in relative astrometric observations. As a test bed we used differential astrometric observations from the FORS2 camera of the Very Large Telescope (VLT) obtained during 2010--2019 under several programs of observations of southern brown dwarfs . The measured image motion was compared to models that decompose atmospheric turbulence in frequency space and translate the vertical turbulence profile into DIM amplitude. This approach accounts for the spatial filtering by the telescope's entrance pupil and the observation parameters (field size, zenith angle, reference star brightness and distribution, and exposure time), and it aggregates that information into a newly defined metric integral term. We demonstrate excellent agreement (within 1) between the model parameters derived from the DIM variance and determined by the observations. For a 30 s exposure of a typical 1 field close to the Galactic plane, image motion limits astrometric precision to sim 60 mu as when sixth-order transformation polynomial is applicable. We confirm that the measured image motion variance is well described by Kolmogorov-type turbulence with exponent 11/3 dependence on the field size at effective altitudes of 16--18 km, where the best part of the DIM is generated. Extrapolation to observations with extremely large telescopes enables the estimation of the astrometric precision limit for seeing-limited observations of sim 5 mu as, which has a variety of exciting scientific applications.