Resonant Fault Slip in Subduction Zones


arrived August 17, 2006 to Nature Magazine!

Utah State University's "Utah State Today" (17 Aug 2006)
The Deseret Morning News (21 Aug 2006)
New Scientist (8 Mar 2007)
Cache Valley Herald Journal (3 Apr 2007)
The Utah Statesman (16 Apr 2007)


Figure 1. Time series of GPS location in Guerrero, southern Mexico, show 1 to 10 cm reversals of steady-state motion at intervals of 12.0±0.3 months. These events indicate slow aseismic thrust slip occurring on the boundary between the Cocos and North American Plates.

Figure 2. Stress variations on the plate boundary interface in southern Mexico. (a). Time series of normal stress (blue; positive indicates fault compression) and shear stress (red; positive favors thrust slip) caused by hydrologic loading at the surface, calculated at a point beneath GPS site COYU. Grey bars denote periods of deep slow slip shown in Figure 1; peak slip occurs at the center of the bar. (b). Map view of peak-to-peak shear stress perturbation caused by hydrologic loading, projected from the plate interface to the surface. Inset shows plate geometry and strike-averaged topography versus distance from the trench; arrow indicates location of time series sampled in Figure 2a. (c). Rate of accumulation of tectonic shear stress caused by motion of the subducting Cocos plate relative to North America.

GPS measurements in various different tectonic environments have recorded fault movements similar to earthquakes but much slower, occurring over timescales of ~1 week to ~1 year (see for example the GPS time series in Figure 1). These "slow slip events" have been observed for example in Japan, Cascadia, Mexico, Alaska, Hawaii and New Zealand.

The phenomenon is poorly understood, but several observations hint at the processes underlying slow slip. Modelling of GPS data and estimates of associated tremor location indicate that slip focuses near the transition from unstable ("stick-slip") to stable friction at the deep limit of the seismogenic zone. Perhaps most intriguingly, slow slip is quasi-periodic at several locations, with recurrence varying from 6 to 25 months depending on which fault (or even fault segment) is examined.

Lowry [2006] demonstrates how periodic slow fault slip may be a resonant response to climate-driven stress perturbations. Fault slip in southern Mexico recurs annually, and as shown in Figure 2a, stress perturbations on the fault surface include an annual signal caused by changes in the weight of water and groundwater at the surface.

The annual stress variation is only of order a few hundred Pa (Figure 2b), which is a small fraction of the annual tectonic stress accumulation (Figure 2c), so this stress would not be expected to produce a robust slip response unless the fault's response to stress was somehow amplified at that period.

Physical relations for rate- and state-dependent friction actually predict an amplified or "resonant" response at periods of order months to years, depending on frictional properties (Figure 3). Fault slip resonance helps to explain why slip events are periodic, why periods differ from place to place and why slip focuses near the base of the seismogenic zone. It also explains why the time of peak slip differs from the time of peak stress by about three months in southern Mexico (Figure 2a), because a ±&pi/2 (=± 3 month) phase shift is predicted for resonant slip (Figure 3c). (Incidentally, it also helps to explain why event recurrence is not perfectly regular but instead somewhat variable: The phase relation has extremely steep slope near the critical period). Resonant slip should initiate slightly within the rupture zone of future great earthquakes, suggesting that slow slip may illuminate fault properties that control earthquake slip.

How slow slip resonance depends on rate-state-friction parameters. (a). Amplitude of slip velocity response to stress at period T for fault stiffness k (normalized to a critical values Tc=1 year and kc), for parameters shown in lower-left inset. Upper-right inset, velocity is amplified when (b−a) is small (i.e. at the frictional transition from stick-slip to stable sliding). Colored circles are detailed in b and c. (b). Slip amplitude near its maximum depends on k and T. (c). Phase of slip (relative to the peak stress) also depends on k and T.

This material is based upon work supported by the National Science Foundation under grant number 0207820. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation.

Back to Tony Lowry's homepage