Full-waveform inversion in three-dimensional PML-truncated elastic media

TL;DR

This paper presents a robust 3D full-waveform inversion method for elastic media using PMLs and a hybrid finite element approach, enabling accurate subsurface imaging with efficient computation.

Contribution

It introduces a PDE-constrained optimization framework for 3D elastic media inversion with PMLs and a hybrid finite element scheme for improved efficiency and robustness.

Findings

Successful reconstruction of Lame parameters for smooth profiles.

Accurate material gradient verification demonstrated.

Efficient large-scale computation achieved with explicit Runge-Kutta.

Abstract

We are concerned with high-fidelity subsurface imaging of the soil, which commonly arises in geotechnical site characterization and geophysical explorations. Specifically, we attempt to image the spatial distribution of the Lame parameters in semi-infinite, three-dimensional, arbitrarily heterogeneous formations, using surficial measurements of the soil's response to probing elastic waves. We use the complete waveform response of the medium to derive the inverse problem, by using a partial-differential-equation (PDE)-constrained optimization approach, directly in the time-domain, to minimize the misfit between the observed response of the medium at select measurement locations, and a computed response corresponding to a trial distribution of the Lame parameters. We discuss strategies that lend algorithmic robustness to our proposed inversion scheme. To limit the computational domain to the size of interest, we employ perfectly-matched-layers (PMLs). In order to resolve the forward problem, we use a recently developed hybrid finite element approach, where a displacement-stress formulation for the PML is coupled to a standard displacement-only formulation for the interior domain, thus leading to a computationally cost-efficient scheme. Time-integration is accomplished by using an explicit Runge-Kutta scheme, which is well-suited for large-scale problems on parallel computers. We verify the accuracy of the material gradients obtained via our proposed scheme, and report numerical results demonstrating successful reconstruction of the two Lame parameters for both smooth and sharp profiles.