Group A, Poster #249, San Andreas Fault System (SAFS)

Predicting Off-fault Deformation Using Experiment-Trained Convolutional Neural Networks on Faults of Different Maturity in Southern California

Christ F. Ramos Sanchez, Michele L. Cooke, Hanna M. Elston, Laainam Chaipornkaew, Sarah Visage, Pauline Souloumiac, & Akshay Sharma
Poster Image: 

Poster Presentation

2022 SCEC Annual Meeting, Poster #249, SCEC Contribution #12080 VIEW PDF
Surface offsets might not represent slip at seismogenic depths because earthquakes produce shallow distributed off-fault deformation. Scaled physical experiments simulate off-fault deformation processes and provide direct observations of deformation partitioning during fault evolution. Using experimental fault maps, we train and test a 2D Convolutional Neural Network (CNN) that can predict off-fault deformation of strike-slip fault trace maps along the southern San Andreas fault system. The CNN predicts kinematic efficiency (KE), the ratio of incremental strike-slip accommodated along the faults to the total incremental displacement, of strike-slip faults. Kinematic efficient faults are indi...cative of mature and through-going fault surfaces that require less work to maintain deformation resulting in small off-fault deformation and limited shallow slip deficit. On the other hand, immature strike-slip faults with segmented and complex trace geometry produce greater off-fault deformation and larger shallow slip deficit. We train the CNN on strike-slip experiments in both kaolin and sand with different loading rates and basal conditions to simulate a wide range of conditions that control evolution of fault geometry and off-fault deformation. Whereas wet kaolin has more localized slip surfaces and higher kinematic efficiency, the dilatancy of sand produces larger off-fault deformation and lower KE. We compare how the CNN trained on faults that grow in one rheology and boundary condition perform in predicting KE of unseen fault maps with different conditions in order to assess the predictive power of the CNN. All tests are combined for training the CNN that incorporates expected variations in the crust and tested on crustal fault maps of different maturity in southern California. Applying the clay, sand, and clay+sand trained CNN models to fault maps of the southern San Andreas fault system predict off-fault deformation that overlap independent estimates. The mature San Andreas fault at Mecca Hills have greater KE than the immature Ridgecrest rupture. The northwest portion of the Ridgecrest rupture has almost zero KE where the fault is the most immature. Here, we see parallel left-lateral faults that are similar to features that develop prior to the onset of fault in the wet kaolin. The similarities of the fault evolution and the match of off-fault deformation predictions suggest that scaled physical experiments provide valuable insights into crustal faulting.
SHOW MORE

Citation: Ramos Sanchez, C. F., Cooke, M. L., Elston, H. M., Chaipornkaew, L., Visage, S., Souloumiac, P., & Sharma , A. (2022, 09). Predicting Off-fault Deformation Using Experiment-Trained Convolutional Neural Networks on Faults of Different Maturity in Southern California . Poster Presentation at 2022 SCEC Annual Meeting.
Relevant SCEC Themes:
Stress and Deformation Over Time