Viscoelastic Postseismic Deformation Following the 2002 Mw7.9 Denali Fault Earthquake

Abstract:

The 2002 Mw7.9 Denali earthquake ruptured over 350 km with coseismic slip of up to ~10 m. Stress-driven relaxation of the upper mantle and afterslip of the fault cause postseismic crustal deformation that has been recorded at GPS stations. We analyze the time series of seven continuous and more than 150 campaign-mode GPS stations. We calculate the cumulative postseismic displacements in every three-year time window since 2003. The obtained postseismic displacements in a similar direction as the coseismic motions are up to ~15 cm in 2003 – 2006 and decrease with time to less than 5 cm in 2012 – 2015. The significant postseismic deformation provides a unique opportunity to better constrain the viscosity structure in the continental upper mantle in central Alaska and to test the contribution of afterslip following the earthquake. We have developed three-dimensional viscoelastic finite element models of the Denali earthquake to study these problems. The model includes an elastic lithosphere and a viscoelastic upper mantle. We assume that the upper mantle is characterized by a bi-viscous Burgers rheology. For simplicity, we assume that the transient Kelvin viscosity is one order of magnitude lower than that of the steady-state Maxwell viscosity. The stress-driven, time-dependent afterslip of the fault is modeled by a 2-km thick weak shear zone. Locked portions of the fault, that is, where no afterslip is allowed, are assumed to be outlined by the 5-m coseismic contours. Our preliminary model indicates that areas within ~100 km to the fault are controlled mainly by the afterslip, and the far-field is controlled mainly by the viscoelastic relaxation of the upper mantle. The modeled afterslip is up to more than 1 m in the first three years after the earthquake. The stead-state viscosity of the upper mantle between the Denali fault and the trench is determined to be on the order of 8 × 10¹⁸ Pa s. The viscosity of the upper mantle north of the fault has to be at least one order of magnitude higher in order to better fit the observed displacements in this area. A test model that includes the subduction Pacific slab and oceanic upper mantle produces negligible change in the surface deformation. A test model with the depth-variation in rock properties produces approximately the same results as that of the preferred two-layered model.