Seismic modeling using the frozen Gaussian approximation

TL;DR

This paper introduces the frozen Gaussian approximation (FGA), a ray-based seismic wave modeling method that maintains fixed Gaussian widths during propagation, offering accurate wavefield simulations validated on complex models.

Contribution

The paper presents a novel FGA method for seismic modeling that improves accuracy by keeping Gaussian widths fixed and employs a fast FBI transform for initial data decomposition.

Findings

FGA accurately models seismic wavefields in complex models.

FGA outperforms traditional Gaussian-beam methods in accuracy.

Validation on Marmousi model confirms effectiveness.

Abstract

We adopt the frozen Gaussian approximation (FGA) for modeling seismic waves. The method belongs to the category of ray-based beam methods. It decomposes seismic wavefield into a set of Gaussian functions and propagates these Gaussian functions along appropriate ray paths. As opposed to the classic Gaussian-beam method, FGA keeps the Gaussians frozen (at a fixed width) during the propagation process and adjusts their amplitudes to produce an accurate approximation after summation. We perform the initial decomposition of seismic data using a fast version of the Fourier-Bros-Iagolnitzer (FBI) transform and propagate the frozen Gaussian beams numerically using ray tracing. A test using a smoothed Marmousi model confirms the validity of FGA for accurate modeling of seismic wavefields.