Generative Adversarial Networks for Model Order Reduction in Seismic Full-Waveform Inversion

TL;DR

This paper introduces a GAN-based approach to seismic full-waveform inversion, reducing model parameters and enhancing the plausibility of inverted seismic wave speed models, demonstrated on a 2D SEAM model section.

Contribution

It presents a novel integration of GANs into seismic inversion to produce more realistic models with fewer parameters, improving inversion results.

Findings

GAN produces more plausible seismic models

Reduces number of model parameters in inversion

Improves inversion results on SEAM model

Abstract

I train a Generative Adversarial Network to produce realistic seismic wave speed models. I integrate the generator network into seismic Full-Waveform Inversion to reduce the number of model parameters and restrict the inverted models to only those that are plausible. Applying the method to a 2D section of the SEAM model, I demonstrate that it can produce more plausible results than conventional Full-Waveform Inversion.