Many things came together this year for the FARM group and valuable progress was made in several areas. The connections between the studies and progress in these different areas are particularly notable. They highlight the fact that the SCEC collaboration leads to more rapid advances than would be the case without SCEC.

Larger imageProgressively magnified views of the ultracataclasite layer in the core of the Punchbowl Fault. a) Block of ultracataclasite containing a portion of the continuous, relatively planar principal slip surface mapped in outcrop exposures and interpreted as the site of most recent fault displacement; b) Cross polarized light image of a petrographic section across the slip surface showing that the slip surface is distinct in texture and approximately 1 mm thick; c) Ultracataclasite matrix in plane polarized light. Scale bar 200 μm; d) Bright-field TEM image of ultracataclasite layer showing crystalline nature of nanoscale particles. Scale bar 100 nm.

Experimental studies of weakening at high slip speeds have been made by David Goldsby and Terry Tullis of Brown University. They find two high speed weakening mechanism, one due to the formation of silica gel that acts as a lubricating layer and one due to local or “flash” melting at asperity contacts. Chemical analysis of the thin gel layer has shown that it does indeed contain hydrogen as expected for a gel. It behaves as a thixotropic material, becoming weak only at high deformation rates. David and an undergraduate student from the University of Puerto Rico, Carla Roig Silva, find that the amount of weakening at a given slip velocity increases as the SiO2 content of the rock increases, further supporting the gel-weakening hypothesis. The flash melting mechanism occurs at higher sliding velocities, and requires much less displacement for weakening, than does the gel weakening. The weakening agrees well with theoretical predictions for flash melting made by Jim Rice of Harvard University and Nick Beeler of the USGS (Figure III.5).

Although natural fault cores are meters thick, detailed microstructural studies by Judi Chester and others demonstrate that the zone of active shear during a slip event is even more localized, on the order of a few mm thick or less. This characteristic of natural fault zones is compatible with mechanisms of fault weakening activated in laboratory studies at high slip speeds that require extreme localization of slip, such as flash heating, lubrication by formation of silica gel, and thermal pressurization.

Theoretical analysis of weakening due to thermal pressurization due to shearing on a surface or in a think layer has been studied by Jim Rice, transferring the important field observations into theoretical analysis that can be used in dynamic rupture models. Nadia Lapusta of California Institute of Technology and Jim Rice have developed numerical models of earthquake cycles and dynamic rupture that use rate and state friction and include strong dynamic weakening such as are seen with flash melting or thermal pore fluid pressurization. Their dynamic rupture models show that most of the slip can occur with a very low dynamic stress, can involve a low static stress drop, and rupture can propagate with a tectonic stress that is much smaller than the static strength, as long as the rupture initiates at some location where the tectonic stress and the static strength become equal. This behavior could be the solution to the “low stress” or “heat flow paradox” on faults that slip primarily via earthquakes. Because the stress difference need not be large between the initial and the dynamic values of stress on the fault, the mechanism should not produce accelerations that would exceed those observed for earthquakes.

Exploratory friction experiments have been made by Vikas Prakash of Case Western Reserve University using an experimental apparatus, the torsional Kolsky bar, that shows promise for collecting high speed friction data in a range of slip speeds and normal stresses that are similar to those occurring earthquakes. Furthermore it involves slip displacements up to 10 mm, much higher than the slip attainable by the pressure-shear impact friction experiments Vikas has previously explored as part of the SCEC program. Not only have the initial friction experiments using the torsional Kolsky bar shown its promise for more investigation, they have shown friction values on a novaculite of about 0.2, a value that is similar to those seen in the experiments of Goldsby and Tullis.

In summary the results of the FARM investigators working on field studies, laboratory experiments, and theoretical modeling are showing consistent results. The progress in each area has benefited from better communication between the research scientists that has been fostered by SCEC. The collective results suggest that fault slip is localized, that frictional resistance at high slip speeds can be quite low, and that dynamic rupture models with such resistance can be useful in simulating earthquake sources. This should allow creating more rigorously physics-based earthquake scenarios than are presently in use. Although much work still needs to be done, these results are laying out a path to follow in order to attain one important SCEC goal, that of going from a physics-based understanding of the processes that occur on faults during the earthquake cycle to realistic and accurate models of ground motions that can lead to implementation of better building design.