Microroughness Variation with Inferred Changes in Fault-Normal Stress Along a Pseudotachylyte-Bearing Fault

Monday, 15 December 2014
Phillip G Resor1, Katherine Shervais1, Giulio Di Toro2 and William A Griffith3, (1)Wesleyan University, Middletown, CT, United States, (2)University of Padua, Padua, Italy, (3)University of Texas Arlington, Arlington, TX, United States
Experimental simulations of earthquake-like conditions reveal a significant decrease in effective steady-state friction during dynamic slip. Unlike static friction, the effective dynamic friction coefficient has a clear dependence on the applied normal stress, typical of lubrication. In the case of melt generation it has been hypothesized that the steady state represents a balance between melt generation, melt extrusion, and fault surface microtopography. We have undertaken an investigation of a pseudotachylyte-bearing fault zone to determine if similar relationships between fault-normal stress and fault zone geometry result from natural earthquake slip.

The study area is a single fault strand with ~60 cm of slip within the Gole Larghe Fault Zone, Italy. At the outcrop scale, this strand is distinctly wavy with contractional and extensional bends as well as relatively straight (neutral) sections. In order to quantify the pseudotachylyte geometry and microroughness of fault surfaces we collected cm-scale cores from locations along the fault that we infer to have experienced different fault-normal stress during slip. The internal geometry of the cores was imaged with ~0.03 mm resolution using high-resolution x-ray computed tomography (CT). The pseudotachylyte-bearing fault zone was then segmented from the CT volume by combining manual interpretation with gradient-based edge detection. Finally, the microroughness of the fault zone surfaces (contact between pseudotachylyte and wall rock) was quantified using Fourier spectral analysis.

Both the geometry and microroughness of the fault zone vary with inferred fault-normal stress. Regions of low stress have thicker pseudotachylyte (up to 5.080 mm) and are rougher at short wavelengths (0.01 mm). Regions of high stress are thinner (as low as 0.256 mm) and are smoother at short wavelengths. At wavelengths near the grain-scale (1 mm) spectral power is poorly correlated with stress, indicating that the scaling of roughness (beta value) also varies with fault normal stress. We interpret these patterns of roughness as evidence of competing processes of roughening and smoothing associated with melting during earthquake slip.