Non-Markov behavior of acoustic phase variance in the atmospheric boundary layer

The Markov approximation is widely used in wave propagation in random media. This approximation is valid if the propagation path length is greater than the scale of the medium inhomogeneities affecting a particular statistical moment of a wave field and the moment changes insignificantly over this scale. These conditions might be violated for the variance of the phase fluctuations and other statistical moments of acoustic signals that have propagated through atmospheric turbulence: the scale of the largest eddies can be hundreds of meters, and fluctuations in the acoustic refractive index are relatively strong. In the current article, the phase variance of a spherical sound wave in statistically inhomogeneous turbulence is formulated without the Markov approximation. For propagation ranges smaller than the scale of the largest eddies, the phase variance without the Markov approximation is significantly smaller than when this approximation is employed. As the range increases, the difference between the two results tends toward a constant value (a ‘memory’ effect), which might be significant in many applications. The phase variance without the Markov approximation agrees better with the experimental data on sound propagation through the atmosphere, while the variance calculated with this approximation significantly overpredicts the data.