The CAFÉ seismic and magnetotelluric experiments: An investigation of fluid processes in the Cascadia subduction system using a sequential inversion of passive seismic and magnetotelluric data

Many of the natural hazards associated with subduction zones are intimately tied to the release of fluids bound within the subducting slab. Intraslab earthquakes appear to be associated with the reactivation of faults due to increased pore pressure brought on by fluid release from the slab. The periodicity of great megathrust earthquakes can be explained by episodic buildup and release of fluid pore pressure across plate boundaries down-dip of the locked zone. Most arc magmas are generally believed to be generated by the flux melting that occurs when fluids are released from the slab into the continental asthenosphere. Magnetotellurics (MT) is an attractive method to apply to subduction settings because MT is particularly sensitive to electrically conductive fluid phases and melt, and can therefore be used to constrain the location of fluids release during dehydration reactions in the slab. One drawback is that MT inversions generally place a strong emphasis on smoothness in order to minimize the generation of the structure that is not required by the data, but at the cost of not being able to resolve sharp transitions. In the presence of a-priori information regarding the location of such sharp transitions, it can be useful to incorporate that information into the MT inversion process. We present images generated first from the 2-D Generalized Radon Transform (GRT) migration of teleseismic data collected from 41 three-component broadband stations located roughly east to west across central Washington as part of the CAFÉ experiment. The GRT migration method is particularly adept at highlighting sharp velocity boundaries, making it an ideal candidate for combining with MT methods. We also present images generated from the non-linear conjugate gradient inversion of MT data collected at 60 broadband and 21 long-period stations along a line roughly collocated with the GRT array as part of the CAFÉ MT experiment. Finally, we present a third set of images generated from the inversion of the MT data incorporating the constraints suggested by the results of the GRT migration. By combining information from both datasets into a single set of images, we are able to improve resolution and reduce the inherent non-uniqueness of the data, which enables us to present convincing constraints for the fluid processes occurring beneath central Washington.