The observation that faults are nonplanar, rough rather than flat surfaces,poses a number of challenges in trying to understand how they move. First, modeling such surfaces is difficult numerically. Second, as slip accumulates on a single rough surface, normal stress builds up and the fault soon locks, unable to overcome friction. A standard approach has been to circumvent this lock-up by the method of ``backslipping'', whereby slip is prescribed, and the resulting heterogeneous loading stresses which would allow such prescribed slip to occur are then applied. Unfortunately, the resulting behaviors appear in many ways much closer to how planar faults behave, and thus some of the intrinsic impacts of the geometry appear to be canceled out by this purposefully constructed heterogeneous loading. This project seeks to develop a new approach to the problem, based on the observation that faults occur not as an individual fault surface, but as a system of fault surfaces in a fault zone. In particular, this work will examine fault strands, modeling faults not as one rough surface, but a set of rough surfaces. In this way, the proposal aims to find dynamically consistent fault zones which are both made of rough faults and when loaded uniformly do not lock up. To achieve this goal, take advantage of its anticipated success, and test its implications, three broad research elements would be carried out.
First, testing the hypothesis that a system of rough fault strands can slip under uniform loading without locking up. Continued development and application of a new numerical approach which does not require faults to align with an underlying mesh will open up new realms of study of potential geometry. Constructing and testing correlated sets of rough faults with this numerical approach would then occur. Demonstrating success would be a significant step in understanding and modeling the mechanics of faults. Second, applying the model fault zone to questions relevant to fault and earthquake mechanics. How is slip partitioned in the system? Are there a minimum number of active strands? Does partitioning change as the static coefficient of friction changes? Examining sequences of dynamic ruptures, questions of importance to earthquake mechanics could be asked. Is the distribution of sizes of events on rough faults more Gutenberg-Richter like than on planar faults? Do individual events tend to propagate down one strand, or jump from strand to strand? How often do ruptures break simultaneously multiple strands along-strike? How is slip along-strike related to fault roughness? Third, comparing the model system with geological and seismological observations. Two new measurements quantifying geological observables would be carried out. One measurement concerns quantifying correlations in bends and curvature between neighboring strands. A second measurement concerns measuring the distribution of lateral surface slip strains in large earthquakes. Preliminary work on this shows interesting potential relationships of typical values of surface slip strain with previous measurements by others of fault roughness. This measurement could provide fundamental constraints on the physics of earthquakes from direct geological observations.
Earthquakes pose a number of challenges to society in terms of hazards to public safety and property. Understanding the origins of earthquake behaviors will help us better understand these hazards. This project aims to tackle one of the fundamental questions in earthquake dynamics, and the dynamics of the fault systems on which earthquakes occur, namely, what is the role of the complicated geometry in the problem. Traditional ways of looking at the problem in terms of individual faults, or collections of faults, have run into problems dealing with how to accommodate accumulating slip on nonflat irregular geometries. The PI will examine the role that an observed property of faults, that they occur as fault strands-- multiple surfaces within faults--plays in allowing for slip to accumulate on the nonplanar structures.
The main thrust of this work is to seek a deeper understanding of the role that fault geometry plays in earthquake and fault mechanics. A big question this project sought ot address is how do non-flat faults manage to slip against eachother. Why do they not lock-up at the bumps and indentations? A hypothesis we sought to examine is the possibility that faults are not just single non-flat surfaces, but multiple ones, which can act together to allow motion to occur. We succeeded in developing new computational technologies to study this question. We also discovered a new field site where new data could be collected on natural small-scale fault systems, which will allow us to further study the multi-surface fault systems. Initial results support the basic hypothesis, and encouraged by these results work continues on this topic. With some support from this grant we also engaged in a study of how slip in earthquakes scale with earthquake size. We found new ways to characterize the scaling, and new kinds of data to bring into the problem. Results from this work are being applied to improve methods for estimating seismic hazard, and are being used in the latest seismic hazard maps now being developed (Uniform California Earthquake Rupture Forecast 3, Working Group on California Earthquake Probabilities, 2012). This has broad impacts on society, helping society better estimate and plan and design for the hazards our built environment faces.
