Coupled afterslip and viscoelastic flow following the 2002 Denali Fault, Alaska earthquake

Models of postseismic deformation following the 2002 M 7.9 Denali Fault, Alaska earthquake provide insight into the rheologic structure of the Alaskan lithosphere and the physical processes activated following a large earthquake. We model coseismic GPS displacements and 4 yr of postseismic GPS position time-series with a coupled model of afterslip on the fault in the lithosphere and distributed viscous flow in the asthenosphere. Afterslip is assumed to be governed by a simplified version of a laboratory-derived rate-strengthening friction law that is characterized with a single parameter, sigma(a - b), where s is the effective normal stress on the fault and a - b is a dimensionless empirical parameter. Afterslip is driven by coseismic stress changes on the fault generated by the main shock. The lithosphere is modelled as an elastic plate overlying a linear, Maxwell, viscoelastic asthenosphere. We devise a scheme to simultaneously estimate the distributions of coseismic slip and afterslip, friction parameters, lithosphere thickness and asthenosphere viscosity. The postseismic GPS time-series are best reproduced with a 45-85 km thick elastic lithosphere overlying an asthenosphere of viscosity 0.6-1.5 x 10(19) Pa s. The 45-85 km elastic lithosphere thickness is greater than or equal to the average crustal thickness in the region suggesting that distributed postseismic flow occurs largely within the mantle while postseismic deformation in the crust is confined largely to slip on a discrete fault penetrating the entire crust and perhaps into the uppermost mantle. For typical laboratory values of a - b of order 10(-2) at temperatures corresponding to the inferred depth of afterslip, the estimated effective normal stress on the fault is similar to 50 MPa, which is about an order of magnitude lower than effective normal stresses at hydrostatic pore pressure. Previous studies showed that models with linear Newtonian rheology cannot reproduce the observed GPS time-series but that models incorporating non-linear or biviscous flow of the mantle do fit the data. We show that a model with afterslip governed by a non-linear fault zone rheology coupled to Newtonian mantle flow is sufficient to reproduce the GPS time-series.