Robust Reconstruction of Scattering Surfaces Using a Linear Microphone Array

The analysis of sound scattered by a rough surface and measured by multiple microphones positioned in the far field yields an estimate of the unknown scattering surface profile. Expanding from previous work, the approach used in this paper is based on an expansion and linearization of the Kirchhoff integral equation, and applies to a low density of receivers. Here, the original algorithm is modified in order to reduce the measurement bias, and extended to broadband signals to over-constrain the problem and improve its robustness. The improved method is rigorously assessed alongside the original algorithm and its small perturbation version, for a two-dimensional geometry and for scattering surfaces with a spatial power-function spectrum. The impact of the measurement setup and surface characteristics on the reconstruction uncertainty are evaluated by means of numerical simulations. Additional experimental data obtained for three known surface profiles reveal the impact of noise and measurement uncertainties. The optimal measurement configuration requires a trade-off between resolution (higher at high frequencies), and robustness (higher at low frequencies). This limit is overcome at least partially by the proposed multiple-frequency extension. The resulting measured uncertainties were close to the theoretical expectation of approximately 2% of the acoustic wavelength. © 2020 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/)