I. Selwyn Sacks

Strain Diffusion from Great Earthquakes

The motion of the plates in the uppermost Earth causes deformation that can lead to failure at the interaction boundaries. The deformation results in the slow buildup of stresses, but those stresses can be released rapidly and possibly catastrophically as earthquakes. Recent observations, initially from high-sensitivity borehole strainmeters, showed that these failures can also occur slowly on the same faults that have regular destructive earthquakes. The residual stress and the proximity to failure, therefore, cannot be estimated without knowledge of these slower events. The analyses of data from slow events and from volcanic eruptions are discussed further in the article above by Alan Linde.

Also important is another type of slow deformation of the Earth's surface due to strain diffusion from great earthquakes, volcanic eruptions, or spreading events at the separating plate boundaries. This type of deformation propagates across the surface at velocities that depend on the distance from the source. Near the source, the propagation velocity is some hundred kilometers per year, but at greater distances the velocities decrease to only a few kilometers per year. The parameters of the elastic crust and those of the viscoelastic underlying layers govern this highly dispersive propagation rate. Analyses of this slow deformation have allowed determination of the viscosity of the crust and uppermost mantle for Japan, California, and Iceland. Some of these deformations can be measured hundreds of kilometers from the source even after a century. An important consequence of this long-lasting disturbance is that the present-day strain field, even from highly precise Global Positioning Satellite (GPS) measurements, cannot be reliably estimated without allowing for the effects of past great earthquakes.

Another key effect of this strain diffusion has recently been recognized. The slowly propagating strain pulse is capable of unlocking favorably oriented, highly stressed faults. It does this by reducing the force pressing the two sides of the fault together, allowing the fault to slip in an earthquake. Former DTM fellow Fred Pollitz and I have shown that the highly destructive 1995 earthquake in Kobe, Japan, was likely triggered by strain diffusing from great earthquakes in 1944 and 1946. And the Tonankai earthquake of 1944, which occurred about 50 years earlier than expected, may itself have been triggered by strain from a great earthquake in 1891. Because of the orientation sensitivity, strain pulses are also capable of inhibiting earthquakes. Since in many active tectonic regions there are faults with many different orientations, it is possible to calculate the probabilities that for any particular fault the likelihood of failure has been enhanced or lessened. In a region in south central Japan, Pollitz and I found that during the period 1901-1969, all earthquakes occurred where the failure probability was enhanced by strain diffusion, mainly from the 1891 Nobi earthquake. No earthquakes occurred where the fault orientation was such that the strain pulse increased fault clamping even though many of these faults have been similarly active historically.

While the physics of strain diffusion is reasonably well understood, earthquake rupture at slow speeds is not. In particular the discovery by Alan Linde and me, that slow events can occur on the same faults that also fail rapidly at other times, is yet to be understood. It is obviously of considerable societal relevance to understand what conditions cause the different behavior. One of the most dramatic situations occurs off northeast Japan. Here there are great subduction events, as large as magnitude 8. However, most of the plate motion is released as slow, non-destructive events that, until recently, could not even be detected. The locked zone of these earthquakes is well to the east of Japan, so can best be studied using instruments on or below the seafloor.

During the summer of 1999 a DTM team -- Linde, Nelson McWhorter, Ben Pandit, Michael Acierno, and I -- spent two months on the drill ship of the international Ocean Drilling Program, the JOIDES Resolution, about 120 km off the coast of Japan. Two holes, about 50 km apart, were drilled and instrumented with strainmeters, tiltmeters, and seismometers, in collaboration with a Japanese team led by former DTM predoctoral fellow Kiyoshi Suyehiro. One hole is about 15 km above an area of active seismicity, which illuminates the subduction interface. The other is in a similar position, but the interface is seismically quiet. Since the plate motion here is about 10 cm per year, there is a strong expectation that these data will soon provide new insight into the complex behavior of this important subduction zone.