Fusion of Multi-Resolution Seismic Tomography Maps Using Physics-Informed Probability Graphical Models
Description:
The resolution of velocity models obtained by tomography varies due to the inversion approach, ray coverage, etc. Fusing tomography models with different resolutions are desired in many cases, such as in updates of community models, or to enable more accurate ground motion simulations. Fusion involves the appropriate weighing of the internal velocity structure in each model. Simple combining the velocity model with edge interpolation may bias velocities along the edges of the models that are combined. Here, we propose an approach to fuse multi-resolution seismic tomography models with physics-informed probability graphical models (PIPGMs), which takes the physical information (ray-path density and gradient of the low-velocity zone) into consideration. We present the corresponding relation between subdomains with multiple resolutions, in terms of high-resolution (HR) and low-resolution (LR) component areas. By transferring the distribution information from the HR to the LR parts, the details in the LR areas can be enhanced by solving a maximum likelihood problem with prior knowledge from the HR models. To evaluate the efficacy of the PIPGM fusion method, we employ the fusion method on both synthetic checkerboard models and a fault zone structure imaged from the 2019 Ridgecrest, CA, earthquake sequence area. The Ridgecrest fault zone model consists of high-resolution Rayleigh velocities obtained from ambient noise tomography, which is embedded into the regional Southern California Earthquake Center Community Velocity Model (CVM) version S4.26-M01. The proposed method improves the combined models under multiple evaluation metrics, including travel time residual, compared to those obtained by multiple conventional methods. The proposed fusion method can merge any type of gridded multi-resolution velocity models and may provide a valuable tool for seismic tomography and hazard monitoring.
Session: Multi-scale Models for Seismic Hazard Analysis
Type: Oral
Date: 4/20/2023
Presentation Time: 02:15 PM (local time)
Presenting Author: Kim B. Olsen
Student Presenter: No
Invited Presentation:
Authors
Zheng Zhou Corresponding Author zhz039@ucsd.edu University of California, San Diego |
Peter Gerstoft pgerstoft@ucsd.edu University of California, San Diego |
Kim Olsen Presenting Author kbolsen@sdsu.edu San Diego State University |
|
|
|
|
|
|
Fusion of Multi-Resolution Seismic Tomography Maps Using Physics-Informed Probability Graphical Models
Category
Multi-scale Models for Seismic Hazard Analysis