Lab. de Mecanique des Fluides et d'Acoustique, URA CNRS 263, Ecole Centrale de Lyon, BP 163, 69131 Ecully Cedex, France
Incorporating random aspects in the numerical simulation of atmospheric sound propagation has led to a much better agreement between experiments and predictions. In particular, the turbulent scattering of sound into acoustic shadows had been demonstrated but some discrepancies still exist. In most of the numerical studies the fluctuations of the refractive index have been considered as scalar and characterized by a single length scale (Gaussian spectrum). However, sound propagation in the turbulent atmosphere is affected by quantities with different tensorial character and different scales. In this paper two possible ways of improving the simulations are investigated: by choosing a better spectral representation of the turbulence (von Karman instead of Gaussian form); by correctly taking into account the vectorial character of wind fluctuations. The turbulence is represented as a set of realizations of a random field generated by a limited number of scalar or vectorial random Fourier modes. Through each individual realization, the acoustic pressure field is computed with a wide-angle parabolic approximation. Ensemble averaging is then performed to obtain the statistical properties of the acoustic field: mean level, standard deviation of the fluctuations. Illustrations will be given for an upward refracting atmosphere, when a deep deterministic shadow zone is present.