Are Anomalous Stresses in Upper Cook Inlet Basin Linked to the Geometry of the Underlying Subducting Slab? Static and Time-Dependent Stress Models of the 1964 Great Alaska Earthquake

The current stress field of upper Cook Inlet basin is unusual in that the maximum horizontal stress is oriented ~45° counterclockwise from, rather than parallel to, the motion vector of the subducting Pacific Plate. A 3-dimensional, elastic dislocation model of the 1964, Mw 9.2, great Alaska earthquake demonstrates that sharp changes in geometry of the subduction interface may strongly influence the stress field in the upper plate and may account for the anomalous orientation of the principal stresses. The model accurately represents the current view of the 170,000 km2 event as rupturing across a subducted transform boundary that is characterized by complex, rapid changes in slab geometry. Static stress transfer from the 1964 event into the overlying North American plate altered Coulomb stresses on the Lake Clark-Castle Mountain fault system and on several blind, oblique thrust faults that core anticlines of the upper Cook Inlet petroleum province. Each of these faults presents a significant seismic hazard to the greater Anchorage area and to regional petroleum infrastructure and production. Modeled, static Coulomb stress changes caused by the 1964 event suggest a localized decrease in fault stability of the Castle Mountain fault and decreased stability of most east-dipping, upper Cook Inlet thrust faults. Notably, the local region of decreased fault stability along the Castle Mountain fault coincides with rapid changes in the geometry of the underlying subducting slab; models that do not account for changes in slab geometry tend to show increased stability along the length of the fault. The zone of decreased stability correlates with the western segment of the Castle Mountain fault, the only known upper plate fault in the greater Anchorage with unequivocal Holocene surface rupture. A time-dependent rheological, visco-elastic model of the 1964 event suggests that in regions where the subduction interface has not relocked, the regional stress field will evolve for decades in response to post-seismic viscous flow. The model uses a Maxwell rheology and incorporates a one-dimensional earth stratification with a viscoelastic lower crust sandwiched between the brittle upper crust and the ductile upper mantle. Model results show a rotation of principal stresses from a reverse to a strike-slip faulting regime along the right lateral Castle Mountain fault at 40 to 60 years post-rupture and a gradual shift from a tensile to a reverse stress regime near the Cook Inlet thrust faults that peaks at approximately 60 years post-rupture. The modeled rotations may suggest a time-dependent link between megathrust earthquakes and rupture of the Castle Mountain and upper Cook Inlet faults; the rotations may also suggest that the 1964 event may adversely affect the long-term permeability of petroleum producing anticlines.