The application of machine learning to predictions of optical turbulence in the surface layer at Baikal Astrophysical Observatory

Abstract

ABSTRACT In this study, we apply machine learning to predict optical turbulence in the surface layer at the Baikal Astrophysical Observatory. Advance knowledge of optical turbulence is important for maximizing the efficiency of adaptive optics systems, telescope operations, and the scheduling of the planned observations. Typically, optical turbulence is characterized by the structure constant of the refractive index of air $C_n^2$. The Monin–Obukhov similarity theory (MOST) provides a scientific basis for estimating the structure constant of the refractive index from meteorological variables in the surface layer. However, the MOST becomes unreliable for stable atmospheric conditions, which occurred for more periods regardless of the time of day at the Baikal Astrophysical Observatory. We propose the application of a neural network based on the group method of data handling (GMDH), one of the earliest deep-learning techniques, to predict the surface-layer refractive-index structure constant. The magnitudes of the predicted values of the structure constant of the refractive index and measurements are in agreement. Correlation coefficients ranging from 0.79–0.91 for a stably stratified atmosphere have been obtained. The explicit analytical expression is an advantage of the proposed approach, in contrast to many other machine-learning techniques that have a black-box model.