Using the Vertical Component of the Surface Velocity Field to Map the Locked Zone at Cascadia Subduction Zone

Abstract:

At present, our understanding of the locked zone at Cascadia subduction zone is based on thermal modeling and elastic modeling of horizontal GPS velocities. The thermal model by Hyndman and Wang (1995) provided a first-order assessment of where the subduction thrust might be cold enough for stick-slip behavior. The alternative approach by McCaffrey et al. (2007) is to use a Green's function that relates horizontal surface velocities, as recorded by GPS, to interseismic elastic deformation. The thermal modeling approach is limited by a lack of information about the amount of frictional heating occurring on the thrust (Molnar and England, 1990). The GPS approach is limited in that the horizontal velocity component is fairly insensitive to the structure of the locked zone. The vertical velocity component is much more useful for this purpose. We are fortunate in that vertical velocities can now be measured by GPS to a precision of about 0.2 mm/a. The dislocation model predicts that vertical velocities should range up to about 20 percent of the subduction velocity, which means maximum values of ~7 mm/a. The locked zone is generally entirely offshore at Cascadia, except for the Olympic Peninsula region, where the underlying Juan De Fuca plate has an anomalously low dip. Previous thermal and GPS modeling, as well as tide gauge data and episodic tremors indicate the locked zone there extends about 50 to 75 km onland. This situation provides an opportunity to directly study the locked zone. With that objective in mind, we have constructed a full 3D geodynamic model of the Cascadia subduction zone. At present, the model provides a full representation of the interseismic elastic deformation due to variations of slip on the subduction thrust. The model has been benchmarked against the Savage (2D) and Okada (3D) analytical solutions. This model has an important advantage over traditional dislocation modeling in that we include temperature-sensitive viscosity for the upper and lower plates, and also use realistic constitutive models to represent the locked zone. Another important advantage is that the 3D model provides a full representation of the interseismic deformation, which is important for interpreting GPS data.