The earthquake cycle in subduction zones includes strain accumulation caused by subduction, sudden strain release by the earthquake rupture, and postseismic time-dependent deformation. GPS observations of postseismic transients after large subduction earthquakes provide a natural experiment to study rheology of the megathrust and of the sublithospheric mantle. Scientists still disagree on what mechanism prevails in the postseismic transients: frictional afterslip on the coseismic rupture, viscoelastic relaxation in the asthenosphere, or poroelastic rebound. In 2006-2007, a doublet of great earthquakes (Mw > 8) struck in the center of the Kuril subduction zone at the northwest Pacific Ocean: a thrust event at the subduction interface followed by an extensional event beneath the oceanward flank of the Kuril trench. It is probably the greatest doublet that has been observed in the era of satellite geodesy and broadband seismology. In 2006, the Kuril GPS Array was installed, which allowed to measure all components of the earthquake cycle. The data collected in 2006-2009 before and after the 2006-2007 earthquakes outlined a broad zone of postseismic deformation with initial horizontal velocities as fast as 100 mm/a, and a regional uplift. Most of the postseismic signal after the great Kuril doublet is caused by the viscoelastic relaxation of shear stresses in the weak asthenosphere. Viscoelastic relaxation was discriminated from other candidate mechanisms by the pattern of horizontal and vertical motions. We predict that the postseismic deformation will die out in about a decade after the earthquake doublet, so the preferred mechanism can be tested with observations continued for several years. Our initial results suggest large variations among subduction zones in the upper mantle viscosity, one of the most important parameters that governs the stress distribution.
The focus of work during this 1-year proposal period will be to: (1) Guarantee continuity of the postseismic time series, without which all future work would be handicapped. (2) Analyze and develop models of postseismic transient deformation for a period 4?5 years after the great 2006?2007 Kuril earthquakes, revisiting or testing several assumptions made in the work so far. (3) Compare and assess geodetic and seismologic models for the two events, and include the effect of the outer rise event in models. (4) Map the space-time distribution of frictional afterslip, and in particular study how stress changes from the second event affected ongoing afterslip. (5) Model the deformation caused by the eruption of Sarychev volcano (Matua Island in the central Kurils).
The project continues collaboration between the scientists of the US and Russia. The results of the project will have an impact on the region subject to disastrous earthquakes. This work and the methods we will develop are expected to lead to generally improved understanding of the earthquake potential of subduction zones and may thus ultimately lead to improved hazard estimation and mitigation elsewhere.
In 2006–2007, a doublet of great earthquakes (Mw > 8) struck in the center of the Kuril subduction zone, a thrust event followed by an extensional event. We used data from our Kuril GPS Array to study the postseismic deformation that followed this earthquake sequence, in order to learn about the mechanical properties of the Earth at this subduction zone. Postseismic deformation refers to transient motions that follow and are triggered by large earthquakes. The two generally most important postseismic physical processes are afterslip (post-earthquake slip on the fault surface) and viscoelastic relaxation of the mantle. Slip during the earthquake causes stress changes within the Earth, and threes stress changes drive further slip or flow of material at depth, which also produces observable surface displacements. Both physical processes can and often do occur, and it has been a big challenge to separate their effects based on surface observations so that we can better understand the mechanics of earthquakes. As part of the project, our Russian collaborators visited the remote stations of the Kuril GPS array to download data and replace batteries, so that they could continue to operate. We analyzed the GPS data to derive time series of positions, which we then used to assess the motion of each site. We assessed the postseismic transient motion by removing a model of steady plate motions and then considering a range of models for the normal deformation observed between earthquakes. This component reflects the build-up of stress that occurs over decades to centuries in advance of large earthquakes. Our observations of the Kuril GPS Array in 2006-2009 outline a broad zone of postseismic deformation with initial horizontal velocities to 100 mm/yr and postseismic uplift. For the time period starting in mid-2007, most of the postseismic signal after the great Kuril doublet is caused by the viscoelastic relaxation of shear stresses in the weak asthenosphere with the best-fitting Maxwell viscosity in the range of (5–10) × 10^17 Pa-sec, an order of magnitude smaller than was estimated for several subduction zones. We can determine this based primarily on the vertical motions. We can fit the postseismic displacements with a model of afterslip, but this model would predict postseismic subsidence, opposite to the observed uplift. Viscoelastic relaxation of the mantle, on the other hand, can explain both the horizontal and vertical motions. We predict that the postseismic deformation will die out in about a decade after the earthquake doublet, based on the estimated viscosity. Our results suggest large variations among subduction zones in the asthenospheric viscosity, one of the most important rheological parameters. We also analyzed the volcanic deformation signal from the eruption of Sarychev volcano on Matua Island in 2009. The GPS site on the island, MATC, was displaced a total of 3 cm horizontally toward the volcano and 3.2 cm downward over a 3-day period. We estimated a source model based on a Mogi source model, using both the ratio of horizontal to vertical deformation and a constrained grid search. The best fitting deflation source depth is 6.5 km below sea level. We then analyzed the longer time series to see if we could detect any further volcanic deformation. We expressed the horizontal components of the time series in terms of two orthogonal axes, one toward the volcano and one orthogonal to that. In this coordinate system, any volcanic deformation caused by the same source that caused the eruptive deformation should appear only on the toward the volcano component. We did not find any evidence for other volcanic sources in the remainder of the time series. This implies that the magma that erupted was in place prior to the establishment of the site, and that no significant magma recharge has occurred since the eruption.
