Strain localization in sandstone-derived fault gouges under conditions relevant to earthquake nucleation
Data files
Aug 03, 2023 version files 9.87 GB
-
automatic_boundary_detection_code.ipynb
11.84 MB
-
Example_of_one_X-ray_CT_image_(from_s015).tif
383.56 KB
-
processed_data.rar
134.97 MB
-
raw_data.rar
9.73 GB
-
README.md
3.30 KB
Abstract
Constraining strain localization and the growth of shear fabrics within brittle fault zones at sub-seismic slip rates are important for understanding fault strength and frictional stability. We conducted direct shear experiments on simulated sandstone-derived fault gouges at an effective normal stress of 40 MPa, pore fluid pressure of 15 MPa, and temperature of 100°C. Using a passive strain marker and X-ray Computed Tomography (XCT), we analyzed the spatial deformation of the gouge samples obtained from the strain-hardening stage to strain-softening stage to steady-state at shearing velocities of 1, 30, and 1000 µm/s. We developed a machine-learning-based automatic boundary detection method to recognize the shear zone fabrics and quantify the slip partitioning between each fabric element. Our results show that R1 and Y (or boundary) shears are the two major shear zone fabrics. At velocities of 1 and 30 µm/s, the relative amount of slip on R1 shears is displacement dependent and increases to ~20% at the strain-softening stage and then decreases to ~10–18% at steady-state. This trend is absent at high velocity with an amount of ~18% through all investigated stages. At all velocities, the relative amount of slip on Y and boundary shears increases linearly with displacement to a total of more than 50% at steady-state. Our study provides constraints for the development of the active slip zone, which is an important input parameter for the heat budget for small-magnitude earthquakes with limited slip (mm-dm), such as those occurring in induced seismicity.
Methods
The raw data.zip is generated by conducting direct shear experiments on sandstone-derived fault gouges.
The processed data.zip is generated by executing automatic_boundary_detection_code in Jupyter Notebook on X-ray CT images in the raw data.
The automatic boundary detection code was developed based on the custom-designed machine-learning trainable segmentation tool from the scikit-image, an open-source image processing library (https://scikit-image.org/)