Vector acoustic full waveform inversion: taking advantage of de-aliasing and receiver ghosts

TL;DR

This paper introduces a vector acoustic full waveform inversion (VAFWI) method that leverages vector data, such as pressure and particle velocity, to improve subsurface imaging by reducing artifacts and effectively handling receiver ghosts, especially under undersampled conditions.

Contribution

The paper develops a VAFWI approach that connects the adjoint operator with inverse wavefield extrapolation, demonstrating its advantages over conventional FWI in aliasing and ghost handling.

Findings

VAFWI's adjoint acts as inverse wavefield extrapolation under certain conditions.

VAFWI reduces imaging artifacts caused by aliasing in undersampled data.

Proper handling of receiver ghosts enhances subsurface image quality.

Abstract

Data acquisition is changing to incorporate more components of the recorded wavefield. An example of this is so-called vector data in which both the pressure and particle velocity are recorded in marine data. We present a vector acoustic full waveform inversion (VAFWI) method. We demonstrate the connection of the VAFWI adjoint operator with inverse wavefield extrapolation and show that under the assumption of a plane wave propagating towards an infinite flat recording surface at normal incidence, the VAFWI adjoint is equivalent to inverse extrapolation of the normal derivative of the recorded field. Thus, unlike the conventional FWI adjoint, the VAFWI adjoint is an inverse wavefield extrapolation operator. If these assumptions are violated, this equivalence relation is no longer true and becomes an approximation. We argue that this has implications in the handling of receiver ghosts by the two inversion methods: in the VAFWI adjoint, the receiver ghosts interact constructively with the back-propagated reflected field. We show numerically that the de-aliasing property of vector data extends to VAFWI, which results in fewer artefacts in VAFWI images compared to conventional FWI when data are spatially undersampled and aliased. Additional information about the subsurface contained in properly handled receiver ghosts can be utilized to achieve further VAFWI image improvement.