Aftershocks follow large earthquakes. But why? Embarrassingly, the mechanism by which earthquakes trigger other earthquakes is still an unresolved question in seismology. We have evidence that the ground shaking from seismic waves plays a key role in this process. For distant earthquakes, seismic waves clearly generate local seismicity (Brodsky et al., 2000; Brodsky and Prejean, 2005; Brodsky et al. 2006; Harrington and Brodsky, 2006; Miyazawa and Brodsky, 2008). There is some indication from the spatial distribution of aftershocks that the same process occurs near mainshocks (Felzer and Brodsky, 2006). Using a large dataset of distant and local triggering, we can show that triggering at all distances can be well predicted based on the amplitude of the seismic waves from previous earthquakes (Van der Elst and Brodsky, 2010). Solving the mystery of earthquake triggering provides constraints on the conditions for generally initiating earthquakes.The mutual triggering effect of earthquakes is also important in the study of human-generated earthquakes. Our group showed that accounting for aftershocks revealed a clearer connection between human and seismic activity at the Salton Sea Geothermal Field on the southern end of the San Andreas Fault (Brodsky & Lajoie, 2013).
Earthquakes can increase permeability in fractured rocks. In the farfield, such permeability increases are attributed to seismic waves and can last for months after the initial earthquake (Elkhoury et al., 2006). Laboratory studies suggest that unclogging of fractures by the transient flow driven by seismic waves is a viable mechanism (Elkhoury et al., 2011; Candela et al., 2014). Permeability enhancement by seismic waves could potentially be engineered and the experiments suggest the process will be most effective at a preferred frequency (Candela et al., 2015).We have recently observed similar processes inside active fault zones after major earthquakes. A borehole observatory in the fault that generated the M9.0 2011 Tohoku earthquake revealed a sequence of temperature pulses during the secondary aftershock sequence of an M7.3 aftershock (Fulton & Brodsky, 2016). The pulses are attributed to fluid advection by a flow through a zone of transiently increased permeability. Directly after the M7.3 earthquake, the fault zone was damaged and highly susceptible to further permeability enhancement, but ultimately heals within a month and becomes no longer as sensitive. Longer term healing was seen in the fault zone of the 2008 M7.9 Wenchuan earthquake (Xue et al., 2013). The competition between damage and healing (or clogging and unclogging) results in dynamically controlled permeability, storage and hydraulic diffusivity. Recent measurements of in situ fault zone architecture at the 1-10 meter scale suggest that active fault zones often have hydraulic diffusivities near 10-2 m2/s. This uniformity is true even within the damage zone of the San Andreas fault where permeability and storage increases balance each other to achieve this value of diffusivity over a ~400 m wide region (Xue et al., 2016). Fault zones may evolve to a preferred diffusivity in a dynamic equilibrium.
As a fault slips, a complex series of processes occur. The rocks around the fault fracture (Savage and Brodsky, 2011). Rocks on the fault are ground to a powder. Shear is focused into narrow slip surfaces and the shape of these surfaces is controlled by the undulations of the powder layer. Those undulations in turn accumulate stress that controls the location of future slip. Each of these parts of an earthquake leave some imprint on the geological record. We are trying to unravel the feedbacks involved in slip on natural faults by measuring the geometry and internal architecture of exposed fault zones (Sagy et al., 2007; Sagy and Brodsky, 2009; Brodsky et al., 2009; Brodsky et al., 2011).Based on these studies, our group recently introduced a new technique for measuring scale-dependent strength based on the roughness of wear surface. When faults slip, the shape of the surface left behind is governed by the strength of the rock. Bumps and protrusions that are too steep are sheared away and only a relatively smooth surface remains. Exactly how smooth may depend on the strength of the rock. We have measured the shape of nearly 30 faults that are exposed at the surface of the Earth and found that most are similarly smooth, but increase in roughness with decreasing scale (Brodsky et al., 2016). At small scales, steep protrusions can be sustained during slip indicating that the rock is stronger. Moreover, the strength is a predictable function of scale and may be associated with particular, identifiable failure processes. For instance, we observe a minimum scale of groove formation that implies a brittle-plastic transition as a function of scale (Candela & Brodsky, 2016). The topography seems to provide a straightforward, easily observable tools to measure roughness over a range of scale. We can now use topography to bridge the gap between laboratory experiments and natural earthquakes. Scale-dependent strength has long been recognized in rocks, but inferring the rock strength from the surface roughness is a new approach. The ease of measuring roughness of surfaces with modern instrumentation makes this approach a potentially powerful method.
Earthquakes occur when tectonic stresses overcome friction. Tectonic stresses are reasonably well-understood, but friction is not. The resisting stress of complex, multiphase faults moving at the high slip rates of an earthquake is the largest single unknown in physical models of earthquake rupture. My group is trying to constrain the processes and values of friction through a combination of laboratory and observational approaches. In the lab, we study the rheology of granular flows as analogs to fault gouge. We have discovered a new regime of granular flow where the acoustic waves generated by angular particles grinding against each other weakens the flow (Lu et al., 2007; Van der Elst et al., 2012).In the field, we are involved in efforts to quickly measure the heat dissipated by friction immediately after a major earthquake. After the 2011 Tohoku earthquake, we participated in the JFAST expedition and found that the effective coefficient of friction on the fault was ~0.1, which is much lower than conventionally thought (Fulton et al., 2013). Other lines of evidence for low friction include a narrow, clay-rich fault zone, laboratory experiments on the recovered gouge and the stability of the borehole (Chester et al., 2013; Ujiie et al, 2013; Lin et al., 2013). The technically ambitious project involved setting multiple engineering records by drilling into the fault where the water depth was ~7 km (Mori et al., 2014)
The power of seismological studies extend beyond earthquakes as we also probe volcanoes, landslides, glaciers and rivers (Brodsky et al. 2003, Harrington & Brodsky 2007, Hsu et al. 2011, Passarelli & Brodsky 2012, Prejean & Brodsky 2011, Roth et al. 2014, Walter et al. 2011). For instance, we showed that the duration of precursory earthquake swarms before volcanic eruptions vary with both magma viscosity and repose time. The correlation suggests magma mobility controls the duration of eruptive precursors (Passarelli & Brodsky 2012). The result has an immediate application in guiding evacuation duration during volcanic unrest while also uncovering new information about the mechanics of the system. It is the type of study that I find most satisfying as I endeavor to both understand and mitigate natural hazards.