Geoacoustic Model for the Inversion of Seismic Profiling Data

We present a geoacoustic model to predict sediment physical parameters from single-channel seismic profiling data. The model uses the concept of a simplified sediment structure, modeled as a binary grain-size sphere pack. The seismic/acoustic response is formulated using Biot’s poroelastic theory as the general framework that is extended by two viscoelastic models. These extensions describe the mechanisms that we consider to have the most significant influence on wave propagation through soft sediment. Viscoelastic response arising from local fluid flow in expandable clay minerals leads to frequency-dependent elastic moduli of the grain material. A heuristically modified Hertz-Mindlin/Walton based viscoelastic-contact model describes local fluid flow at the grain contacts, resulting in frequency-dependent elastic moduli of the sediment frame. Porosity, density and the structural Biot parameters (permeability, pore size, structure factor) follow from the binary grain-size sphere-pack model. The remaining input parameters to the geoacoustic model consist solely of the effective pressure, the mass fractions and the known mechanical properties of each mineral constituent. We will show an example of a successful application of this model for the inversion of single-channel seismic profiling data using a neural network inversion scheme.