Empirical validation and engineering model comparison of outdoor long-range 3D GPU FDTD acoustic simulations with turbulence

Abstract

We build upon previous work which introduced an acoustic GPU-accelerated 3D Finite-Difference Time-Domain solver, Empapu, equipped with a dynamic moving frame and turbulence modeled as a position-dependent refractive index for long-range outdoor sound propagation simulations. This physics-based model would serve as a benchmark to assess different engineering models, e.g., ISO 9613-2, CNOSSOS-EU and sonX, in the presence of uneven terrain with different ground properties and an inhomogeneous atmosphere. The solver now integrates an improved model of turbulence reproducing the kinetic spectral energy of velocity fluctuations, derived from wind velocity fields, with an eddy (quasiwavelet) approach, relevant in long-range propagations. This study validates our model by comparing level differences in third-octave band spectra at receiver points up to 300 m to analytical methods and measurements in a mostly flat topography comprised of asphalt and grassland. Further analysis is done by comparing results in different locations in Switzerland in the presence of uneven ground and barriers. This work will provide a physics-based basis for evaluating and comparing engineering models for outdoor sound propagation in lieu of measurements.