Rupture Propagation Dynamics in Branch Fault Systems: A Case Study of the San Andreas–Garlock Fault Junction Applying a Machine Learning Approach
Description:
Determining whether rupture propagates in a branch fault system and in what direction remains a complex geophysical challenge. This process depends on understanding the evolution of critical parameters, such as stress, strength, and friction, across different fault segments and evaluating their influence on rupture behavior. In this study, we apply a machine learning framework to analyze rupture propagation dynamics in a branch fault system, using synthetic earthquake catalog generated by the Rate-State Earthquake Simulator (RSQSim). Focusing on the San Andreas–Garlock fault junction in southern California, we incorporate realistic fault geometries and input parameters derived from the Uniform California Earthquake Rupture Forecast, Version 3 (UCERF3), to build a comprehensive earthquake catalog.
Our dataset includes approximately 9,800 simulated rupture events, which are used to train machine learning models employing gradient boosting and random forest algorithms. These models identify critical parameters influencing rupture propagation and accurately predict the direction of rupture with high precision and recall. Our findings reveal that the nucleation conditions at the fault branch are the most significant factors governing rupture direction. This study highlights the potential of integrating physics-based simulations with machine learning to advance our understanding of complex fault systems and their rupture dynamics, providing a novel methodology for exploring earthquake processes.
Session: Predictability of Seismic and Aseismic Slip: From Basic Science to Operational Forecasts [Poster]
Type: Poster
Date: 4/16/2025
Presentation Time: 08:00 AM (local time)
Presenting Author: Abhijit
Student Presenter: No
Invited Presentation:
Poster Number: 42
Authors
Shankho Niyogi sniyo001@ucr.edu University of California, Riverside |
Abhijit Ghosh Presenting Author Corresponding Author aghosh@ucr.edu University of California, Riverside |
Evan Marschall emars009@ucr.edu University of California, Riverside |
Roby Douilly robyd@ucr.edu University of California, Riverside |
David Oglesby doglesby@ucr.edu University of California, Riverside |
|
|
|
|
Rupture Propagation Dynamics in Branch Fault Systems: A Case Study of the San Andreas–Garlock Fault Junction Applying a Machine Learning Approach
Category
Predictability of Seismic and Aseismic Slip: From Basic Science to Operational Forecasts