Interplay between alteration and damage at the Nojima fault zone (Japan) revealed by borehole geophysics

Large active faults tend to differ from younger, fresh faults. They are the loci of the greatest earthquakes, but they often creep. The damage zone thickness tends to saturate with large displacement. Deep scientific drilling in large active faults systematically reveals differentiated fault gouge surrounded by a halo of fracturing and alteration. Despite its importance for fault evolution, the process of alteration in large active faults remains poorly understood.

After the devastating M_w 6.9 Nanbu-Kobe earthquake of 1995, a 750 m deep borehole was drilled into the Nojima fault, reaching the fault core at 625 m. The Hirabayashi borehole was drilled ~1 year after the earthquake and provided an extensive dataset on the structure of the fault where the earthquake originated. Continuous coring and borehole geophysics conducted within the borehole showed that the granodiorite protolith experienced several stages of alteration, including fault-related alteration that produced for example laumontite (Ca-rich zeolite) and smectite at T>150°C.

The ANR project AlterAction is revisiting this data with a multidisciplinary team to better understand the interplay between alteration and fault deformation, with 3D imaging of the core samples (see presentations by Romain Iaquinta and Maxime Jamet) and systematic petrophysical characterization.

This presentation will focus on the reanalysis of geophysical logs. Sonic velocities, electrical resistivities, and lithodensities progressively decrease when approaching the fault, starting to deviate from the protolith rock. In the granitoid rocks composing the borehole, variations in gamma-ray may reflect changes in the protolith rather than alteration. This fault zone starts at 370 m (255 m above the fault core) in the hanging wall, with a more pronounced decrease below 540m (85 m above the fault core, corresponding to a zone of thickness ~8 m, given the well trajectory, which is almost tangent to the fault dip). In the footwall, the strong decrease extends down to 680 m (55 m long zone), with lower velocities, resistivities and densities between 625 and 635m, below the fault core. Crossplotting the logging dataset shows the same trends, whether in the footwall or the hanging wall, regardless of the distance to the fault. This suggests that most of the fault zone is affected by the same interplay between alteration and damage. The fault core (623-626m) is singular owing to its relatively large sonic velocities, suggesting that sealing was strongly localized and effective in the fault core one year after the earthquake.