← Back to Sessions

Overset Grid Finite Difference Method for Seismic Wave Simulation in the Presence of the Complex Free Surface Topography

Session: Numerical Modeling of Earthquake Motion, Rupture Dynamics, Seismic Noise, Wave Propagation and Inverse Problems I

Type: Oral

Date: 4/23/2021

Presentation Time: 10:15 AM Pacific

Description: 

Efficient seismic wave simulation is an important tool for strong ground motion prediction and inverting high resolution structures by waveform inversion. Many numerical methods, e.g., the spectral-element method (SEM), discontinuous Galerkin (DG) method, Curvilinear-grid finite-difference method (CG-FDM), have been developed for seismic wave simulation considering surface topography, however, for large-scale problems, the high-quality and efficient grid generation is not a trivial task.

In this work, we propose an overset grid finite-difference method with a collocated grid for the simulation of seismic waves in the presence of steep topography. The complex geological model is discretized into simpler subdomains, and the Cartesian grid is used as the background grid to discretize the whole model, while a few curvilinear grid layers are employed to fit the irregular topography at the free surface. The Cartesian grid and curvilinear grid overlap arbitrarily without requiring point-matched connectivity within the overlapping zone. For the data exchange between different grid blocks, we employ sixth-order explicit Lagrangian interpolation.

Then, some numerical tests are performed. A homogeneous full-space model and a layered velocity model with surface topography verify the proposed overset grid is sufficiently accurate and stable for numerical simulation of seismic wave propagation with steep topography. A rough topographic model that is discretized with a vertical deformed grid and overset grid illustrates that the overset grid can increase the maximum allowable time step by a factor of nearly 2-4 times. And the foothills model is utilized to show the proposed overset grid method can be applied to realistic complex models.

In summary, the overset grid can not only reduce the grid generation difficulty but also improve the computational efficiency. We can obtain stable and accurate results by employing an overset grid even in realistic geological models.

Presenting Author: Nan Zang

Student Presenter: Yes

Authors