ehl logo Acoustics & Environmental hydroacoustics Laboratory

Laboratories of Image, Signal processing and Acoustics - Ecole polytechnique de Bruxelles / Faculté des Sciences

Feature-oriented acoustic tomography

The applicability of acoustic tomography is highly dependent on the reliability of the sound-speed field parameterization. In environments where specific oceanic structures are present (front, eddies), the temperature (sound speed) field can be described by adequate mathematical expressions to construct a two-dimensional range-dependent field with a reduced dimension of the parameter space, the so-called feature model. In such situation, the acoustic propagation is mainly affected by these structures and mode coupling is usually significant in the frequency range of interest (200 Hz-2000 Hz).

Features models were initially developed for oceanic data compression and assimilation purposes. Feature modelling is also used to study the complex interactions that can exist between oceanographic processes. Acoustically, they can enhance sonar performance predictions with respect to range-independent models.

In the feature-oriented acoustic tomography (FOAT) approach, the principal characteristics of the temperature field are estimated by global optimization approach or tracked sequentially with a nonlinear Kalman filter, involving a feature model for the parameterization of the field. Since inhomogeneities or small-scale variability of the temperature field are not reproduced by the feature model parameterization, broadband acoustic data assimilation helps increasing the robustness of the tracking.