Constraints on The Northern Cascadia Subduction Zone Structure From 3D Shear-wave Tomographic Velocities

A 3-D first-arrival travel-time tomographic inversion was employed to construct an S-wave minimum structure velocity model for SW British Columbia and NW Washington State to image the Juan de Fuca slab position, and to constrain the structure of the forearc crust along a 300 km north-south stretch of the Northern Cascadia subduction zone. Approximately 28,000 S-wave travel time picks recorded at 91 stations for 2,500 earthquakes were used in the inversion. The velocity model was parameterized in the forward and inverse directions with a node and cell spacing of (2 × 2 ×2) km and (4 × 4 × 2) km, respectively. Initial 1-D S-wave velocity model was constructed from the P-wave velocity model, obtained from a previous tomographic inversion of P-wave first arrival travel-times from earthquake recordings, employing a Vp/Vs ratio of 1.75. The RMS travel-time misfit for the initial and final S-wave travel-time data was 727 ms and 282 ms, respectively. The S wave velocity model along with a previously constructed P-wave velocity model were used to constrain the structure of the forearc crust, and the position of the subducting Juan de Fuca slab. In the forearc crustal section, the Eocene volcanic Crescent Terrane rocks are mapped with high shear-wave velocities in the mid crust and many earthquakes hypocenters fall with in this region. The Olympic Core rocks are imaged with low shear-wave velocity, devoid of earthquakes, and are inferred to underthrust the Crescent Terrane rocks down to approximately 35 km depth. In the Cascadia subduction zone to the west of the volcanic front, there is no P-wave reflection signature of the forearc Moho in deep seismic sections suggesting a small impedance contrast or a gradient boundary. The S-wave velocity shows a good signature of the forearc Moho as a velocity gradient at approximately 35 km depth. The junction of the forearc crust, forearc mantle and the subducting slab is clearly imaged in the S-wave velocity model along the length of the margin. A better estimate of the slab position is necessary to define the location of the locked zone along the subduction interface for better earthquake hazard estimation. The slab position inferred from this study on vertical cross-sections of S-wave velocity model is consistent with the position of the plate mapped earlier using P-wave tomographic velocities. The velocity model does not show the inversion of high shear velocity lower crust over lower velocity upper mantle inferred in a recent study. Also the plate position beneath south-western British Columbia is of much debate and the uncertainty in the plate position between various studies is of the order of ~10 km. Our interpretation of the regional S-wave and P-wave velocity models indicates that the slab is deeper than the position inferred in recent studies, consistent with earlier interpretations.