Many models for the acoustical properties of rigid-porous media require knowledge of parameter values that are not available for outdoor ground surfaces. The relationship used between tortuosity and porosity for stacked spheres results in five characteristic impedance models that require not more than two adjustable parameters. These models and hard-backed-layer versions are considered further through numerical fitting of 42 short range level difference spectra measured over various ground surfaces. For all but eight sites, slit-pore, phenomenological and variable porosity models yield lower fitting errors than those given by the widely used one-parameter semi-empirical model. Data for 12 of 26 grassland sites and for three beech wood sites are fitted better by hard-backed-layer models. Parameter values obtained by fitting slit-pore and phenomenological models to data for relatively low flow resistivity grounds, such as forest floors, porous asphalt, and gravel, are consistent with values that have been obtained non-acoustically. Three impedance models yield reasonable fits to a narrow band excess attenuation spectrum measured at short range over railway ballast but, if extended reaction is taken into account, the hard-backed-layer version of the slit-pore model gives the most reasonable parameter values.

The computational tools available for prediction of sound propagation through the atmosphere have increased dramatically during the past decade. The numerical techniques include analytical solutions for selected index of refraction profiles, ray trace techniques which include interaction with a complex impedance boundary, a Gaussian beam ray trace algorithm, and more sophisticated approximate solutions to the full wave equation; the fast field program (FFP) and the parabolic equation (PE) solutions. This large array of computational approaches raises questions concerning under what conditions the various approaches are reliable and concerns about possible errors in specific implementations. This paper presents comparisons of predictions from the several models assuming a complex impedance ground and four atmospheric conditions. For the cases studied, it was found that the FFP and PE algorithms agree to within numerical accuracy over the full range of conditions and agree with the analytical solutions where available. Comparisons to ray solutions define regimes where ray approaches can be used. There is no attempt to compare calculated transmission losses to measurements.

No abstract

An efficient numerical method of deducing ground-surface impedance from measurements of short-range propagation from a point source is proposed. The method involves finding the root of a complex equation and uses the fact that the classical approximation for the spherical wave-reflection coefficient near grazing incidence is an analytic function of the impedance. Examples of fitting measured excess-attenuation spectra are given. Methods for ensuring the uniqueness of the resulting impedance fits are discussed. Subsequent deductions of parameters in a two-parameter impedance model are also considered and exemplified.

Analytical and numerical approaches have been made to the problems of (a) propagation through a doubly periodic array of elastic shells in air, (b) scattering by a single elastic shell in air, and (c) scattering by a finite periodic array of elastic shells in air. Using the Rayleigh identity and the Kirchhoff-Love approximations, a relationship is found between the elastic material parameters and the size of the bandgap below the first Bragg frequency, which results from the axisymmetric resonance of the shells in an array. Predictions and laboratory data confirm that use of a suitably "soft" non-vulcanized rubber results in substantial insertion loss peaks related to the resonances of the shells. Inclusion of viscoelasticity is found to improve the correspondence between predictions and data. In addition the possible influences of inhomogeneity due to the manufacturing of the elastic shells (i.e., the effects of gluing sheet edges together) and of departures from circular cylindrical cross-sections are considered by means of numerical methods.