3D mapping of optical turbulence using an atmospheric numerical model

II. First results at Cerro Paranal

¹ U.M.R. 6525 Astrophysique, Université de Nice-Sophia Antipolis, Centre National de la Recherche Scientifique, Parc Valrose, 06108 Nice Cedex 2, France
² Centre National de Recherches Météorologiques Météo France, 42 Av. G. Coriolis, 31057 Toulouse, France

Send offprint request to: E. Masciadri

Received: 17 July 1998
Accepted: 4 February 1999

Abstract

The first statistical results of simulations of optical turbulence over Cerro Paranal by an atmospheric non-hydrostatic model (Meso-Nh) are presented. Measurements from the whole PARSCA93 campaign are analyzed. Simulations are compared to optical measurements obtained by a Scidar and a DIMM working at the same time during 14 nights 13-26 May 1993 over Paranal (Chile). An orographic model with a horizontal resolution of 500 m is implemented in Meso-Nh to study its sensitivity to the orographic effects. The model is initialized with radiosoundings profiles provided by Antofagasta station (70.43 W, 23.43 S) and ECMWF analyses extracted from the nearest grid point (70.31 W, 23.62 S) to this meteorological station. A detailed quality study of radiosounding and analyses is presented. No radiosoundings were provided by Antofagasta station during 4 nights and numerical instabilies were generated by the model during the 13 and 26 May 1993 nights. So, only 8 nights were actually studied. Two statistical methods are presented: Method A has a high temporal resolution, Method B has a high vertical spatial resolution. Method A compares integrated values (seeing) provided by simulations and measurements, Method B compares the profiles. Different statistical estimators are computed for both the methods. We demonstrate that the model can reproduce well the spatial distribution of the optical turbulence in both the high and low atmospheric regions but the unreliability of Antofagasta radiosoundings hampers the statistical results of the numerical technique. To better test the performance of the model, a comparison between the numerical and the forecasting-by-persistence techniques is presented. With a poor statistical sample (only 8 nights), no conclusive statements can be made about the performances of the two techniques. The model ability to discriminate between the best (ε = 0.38) and the worst (ε = 1.38) nights of the campaign is promising for the future implementation of the numerical technique for flexible scheduling of telescopes.