2002A&A...382..378M

It was recently shown (Masciadri et al., 2000, SPIE Proc. 4006, 1136; Masciadri, 2001, ASP Conf. Ser., Astronomical Site Evalutation in the Visisble and Radio Range, Marrakech 13-17, November, 2000, in press), using a numerical simulation (Meso-Nh model), that considering horizontally uniform C_N² can have severe consequences. For example, the integration of the C_N² along lines of sight different from the zenith gives variations that can be larger than 0.50''. In this paper we try to validate this result following two approaches. (a) numerical approach: we compare measured and simulated C_N² profiles. The first ones were obtained with a Generalized Scidar (GS) and the latter are simulated with the Meso-Nh model at the same azimuth and elevation as those of the observations. Further measurements (DIMM, mast and balloons) are considered in order to support the comparison. (b) experimental approach: we compare C_N² profiles measured (almost simultaneously) by a GS along different lines of sight. The results of this study show that (1) the horizontal size of the turbulent layers can be finite, (2) the simulations and the measurements are well correlated and (3) for the first time, we show that the model can reproduce observed seeing values that vary as much as 0.50'' during the same night. This definitely shows that the numerical simulations are a useful tool in the context of the turbulence characterization for astronomical applications. Finally, we discuss the implications that a finite horizontal size of the turbulent layers could have on new adaptive optics techniques, particularly in applications to the extremely large size telescopes.