Abstract:

Numerical simulation of seismic wave fields is crucial for seismic exploration, seismic data processing, and the study of Earth’s structures. The wave equation numerical simulation method takes into account the dynamic and geometric characteristics of seismic wave propagation, providing a solid theoretical basis for studying the mechanism of seismic wave propagation and interpreting complex geological structures. It is currently one of the most widely used methods for simulating seismic wave fields. This article surveys five wave equation-based numerical simulation methods: The finite difference method is easy to understand, but suffers from numerical dispersion issues; The pseudo spectral method offers high accuracy but low efficiency; The finite element method is suitable for complex models but requires substantial computational resources; The spectral element method is appropriate for high-precision problems but demands significant computational memory; And the deep learning method based on physics-informed neural networks demonstrates strong adaptability, though it comes with high training costs. The theoretical foundations, applicable conditions, and latest advancements of these five numerical simulation methods are described respectively. In the future, seismic wave field numerical simulation methods should integrate cutting-edge technologies such as deep learning, optimize boundary conditions to simulate actual boundary reflections, and enhance the precision and efficiency of simulations.

Key words: wavefield simulation, finite difference method, pseudo spectral method, finite element method, spectral element method, physics-informed neural network