What Are We Inverting For?

Abstract

The goal of geoacoustic inversion is to estimate environmental characteristics from measured acoustic field values, with the aid of a physically realistic computational acoustic model. As modeled fields can be insensitive to variations in some parameters (or coordinated variations in multiple parameters), precise and unique inversions can be difficult to achieve. However, if the results of the inversion are only required for sonar performance prediction (for one example), it is only the resulting acoustic field in the water that matters, often at long range and within a restricted range of frequency. In this context, perhaps a precise description of the seabed structure is not necessary, and it might be sufficient to imagine a simpler seabed model having the same acoustic effect on the underwater sound field within the range-frequency domain of interest. An “effective seabed” model could be built upon the observation that the near-grazing acoustic reflection at the seabed can be described by one complex-valued quantity: the surface impedance. Analytic expressions are derived for an ideal fluid/solid case showing exactly how the five geoacoustic properties reduce to two real parameters: the real part and the imaginary part of the surface impedance at grazing incidence, which are connected with the loss and phase shift of the reflected wave, respectively. It is also shown that several inversion algorithms applied to a typical benchmark problem show variation in the inverted geoacoustic parameters of the seabed model, while all agree on the imaginary part of the impedance (phase shift).