Fast approximate simulation of seismic waves with deep learning

TL;DR

This paper introduces a deep learning approach for seismic wave simulation that significantly reduces computation time, enabling near real-time applications and improving seismic inversion processes.

Contribution

The authors develop a neural network model inspired by WaveNet that directly predicts seismic responses, offering a faster alternative to traditional finite-difference methods.

Findings

Achieves an order of magnitude faster simulation (0.1 s vs 1 s)

Successfully trained on 50,000 synthetic examples

Potential extension to arbitrary velocity models

Abstract

We simulate the response of acoustic seismic waves in horizontally layered media using a deep neural network. In contrast to traditional finite-difference modelling techniques our network is able to directly approximate the recorded seismic response at multiple receiver locations in a single inference step, without needing to iteratively model the seismic wavefield through time. This results in an order of magnitude reduction in simulation time from the order of 1 s for FD modelling to the order of 0.1 s using our approach. Such a speed improvement could lead to real-time seismic simulation applications and benefit seismic inversion algorithms based on forward modelling, such as full waveform inversion. Our proof of concept deep neural network is trained using 50,000 synthetic examples of seismic waves propagating through different 2D horizontally layered velocity models. We discuss how our approach could be extended to arbitrary velocity models. Our deep neural network design is inspired by the WaveNet architecture used for speech synthesis. We also investigate using deep neural networks for simulating the full seismic wavefield and for carrying out seismic inversion directly.