Time-Dependent Variations of Slow Slip Events in Lower Cook Inlet of the Alaska-Aleutian Subduction Zone

We identfied a series of abrupt changes in GPS site motions observed in Lower Cook Inlet of the Alaska-Aleutian subduction zone in late 2004, early 2010 and late 2011. The site motions from 1995-2004, 2004-2010 and post-2011, however, appear to be steady. To first order, the deformation rates for 1995-2004 and 2010-2011 are similar to each other, as are 2004-2010 and post-2011. This pattrn of toggling between two deformation patterns is due to the time-dependent slip variations on the Alaska-Aleutian subduction plate interface. It is possible that a deeper part of the subduction interface had been suddenly locked between 2004 and 2010, causing additional interseismic strain accumulation, and we test this hypothesis.

We model time-dependent variatins in this seismogenic zone using the software TDEFNODE to estimate the slip rate deficit distribution on the Alaska-Aleutian subduction plate interface along with block rotation of upper plate blocks. To do so, we first divided the GPS time series into four time periods: before 2004, 2004 to 2010, 2010-2011, and post-2011. We removed the ongoing postseismic deformation due to the 1964 earthquake fom these GPS velocity fields. We constructed an upper plate block model including three main blocks, Southern Alaska (SOAK), the Peninsula block and the Bering Plate to model the contributions from the secular tectonic motions of the upper plate. We modeled the subduction fault surface based on the Slab1.0 model for the subduction zone, with the slab extended to the east as Slab1.0 terminates the slab under Prince William Sound. We found 50~60 mm/yr slip rate deficit in the shallow part of the seismogenic zone, to depths of ~30km which we identify as the main asperity that ruptured in 1964 earthquake. The shape of the locked region under Lower Cook Inlet differs for each time period, with the locked region being wider in 2004-2010 and post-2011. In the near future we will use TDEFNODE to model the GPS time series directly to estimate the time-dependent slilp deficit, and identify the likely range of temporal variations in the width of the seismogenic zone.