3D Fractal Analysis of Co-seismic Damage in the Nojima Fault Using X-Ray Tomography and SHPB Experiments

The origin of the intense damage found in active fault cores is still a matter of debate. We investigated the potential co-seismic contribution to this damage by studying the Nojima fault, which ruptured during the 1995 Kobe earthquake (Mw 6.9). Drilled just a year after the event, the Hirabayashi borehole offers a snapshot of the fault zone’s state shortly after a major rupture.

Working within the French ANR AlterAction, we analyzed drill core samples using X-ray computed tomography (CT) at a resolution of ~50 μm. Instead of relying on complex segmentation of fracture geometries, we applied a 3D fractal analysis to the spatial distribution of voids (empty space) versus the rock matrix. This method allowed us to quantify damage intensity and organization using the fractal dimension D. This metric, ranging from 2 (highly clustered voids) to 3 (homogeneous distribution), tracks the transition from localized fracture networks to diffuse pulverization and correlates well with fracture porosity.

We observed a damage zone extending roughly 70 m on either side of the fault core. While open fracture density generally spikes toward the core, it drops sharply in the immediate vicinity, likely due to rapid post-seismic healing. Our analysis shows D values near 2 in clustered zones, rising toward 3 where damage becomes volumetric. Interestingly, some samples display intense micro-fracturing but lack significant macroscopic deformation, resembling the "pulverized rock" seen at other active faults. This texture suggests high strain-rate loading occurred during the earthquake.

To test the dynamic origin of this damage, we ran Split Hopkinson Pressure Bar (SHPB) experiments on intact borehole samples to reproduce pulverization in the lab. We found a linear link between strain rate and absorbed energy. When combined with the CT data, this relationship helps distinguish two modes of propagation: diffuse pulverization (matching near-fault observations) and sparse, poorly connected networks. Crucially, the fractal dimensions of the experimental samples confirm these contrasting morphologies.

These results suggest that the intense damage in the Nojima fault core likely stems from co-seismic processes, marked by specific fractal patterns associated with high strain rates. We conclude that 3D fractal analysis of void space offers a robust tool, independent of geometry, for identifying the dynamic origins of fault zone damage.