Separation-PINN: Seismic P- and S-waves mode separation with physics informed neural networks

View at publisher Summarise with VIRAT
Authors: Xinru Mu, Shijun Cheng, Tariq A. Alkhalifah
Cited by: 0
References: 16

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Accurate separation of P- and S-waves is crucial for elastic reverse time migration, as it helps prevent interference between different wave modes and usually ensures high-precision imaging results. Traditional P- and S-waves separation methods rely on the Christoffel equation to calculate the polarization direction of P- and S-waves in the wavenumber domain, and this process is numerically expensive. Even machine learning was utilized to improve the efficiency of the separation process, but that required a supervised learning approach with labeled data, often not available for field data. Therefore, we develop a wavefield separation method based on a physics informed neural network (PINN). It is an unsupervised machine learning approach applicable to label-less data. In addition, the trained PINN model provides a mesh-free numerical solution, which captures wavefield features at multiple scales. Numerical simulation tests show that the proposed PINN-based wavefield separation scheme can accurately obtain separated P- and S-waves.

Fields of Study & Subjects

Seismic Imaging and Inversion Techniques Model Reduction and Neural Networks Seismic Waves and Analysis

Publication details

Journal / source
Second International Meeting for Applied Geoscience & Energy
Volume
1
Pages
2532–2536
Publication date
10 February 2026
Language
EN