A geological fingerprint of low-viscosity fault fluids mobilized during an earthquake

[1] The absolute value of stress on a fault during slip is a critical unknown quantity in earthquake physics. One of the reasons for the uncertainty is a lack of geological constraints in real faults. Here we calculate the slip rate and stress on an ancient fault in a new way based on rocks preserved in an unusual exposure. The study area consists of a fault core on Kodiak Island that has a series of asymmetrical intrusions of ultrafinegrained fault rock into the surrounding cataclasite. The intrusive structures have ductile textures and emanate upward from a low-density layer. We interpret the intrusions as products of a gravitational (Rayleigh-Taylor) instability where the spacing between intrusions reflects the preferred wavelength of the flow. The spacing between intrusions is 1.4 ± 0.5 times the thickness of the layer. This low spacing-to-thickness ratio cannot be explained by a low Reynolds number flow but can be generated by one with moderate Reynolds numbers. Using a range of density contrasts and the geometry of the outcrop as constraints, we find that the distance between intrusions is best explained by moderately inertial flow with fluid velocities on the order of 10 cm/s. The angle that the intrusions are bent over implies that the horizontal slip velocity was comparable to the vertical rise velocity, and therefore, the fault was slipping at a speed of order 10 cm/s during emplacement. These slip velocities are typical of an earthquake or its immediate afterslip and thus require a coseismic origin. The Reynolds number of the buoyant flow requires a low viscous stress of at most 20 Pa during an earthquake.