TY - JOUR T1 - Body Wave Tomography of the Cascadia Subduction Zone and Juan de Fuca Plate System: Identifying Challenges and Solutions for Shore‐Crossing Data JF - Geochemistry Geophysics Geosystems Y1 - 2020 A1 - Bodmer, M. A1 - Toomey, D. R. A1 - VanderBeek, B. A1 - Hooft, E. E. A1 - Byrnes, J. S. AB - Recent seismic results from the Cascadia Initiative indicate that heterogeneity in the oceanic asthenosphere affects subduction dynamics. Accurate characterization of the oceanic upper mantle is thus necessary to fully understand subduction processes, including the behavioral segmentation of the megathrust. A key challenge is integrating onshore and offshore datasets, which span large variations in near‐surface features that teleseismic body wave tomography is ill‐posed to resolve. Here, we perform a series of P and S forward modeling predictions to better understand the relative contribution of elevation, crustal thickness, offshore sedimentation, and near‐surface velocity structure to teleseismic delay times. Crustal thickness and elevation variations dominate the signal, contributing ∼1 s of delay time difference for P‐waves (roughly double for S). We test several inversion strategies to account for near‐surface features, identifying potential artifacts and causes of imaging errors. Undamped station statics are found to absorb mantle structures and introduce low‐velocity artifacts beneath the forearc. Our preferred inversion strategy utilizes a three‐dimensional starting model (including elevation) of the upper 50 km and heavily damped station statics, which we find leads to better resolution of mantle structure, particularly at asthenospheric depths. These insights guide inversions of observed delay times from the Cascadia subduction zone and Juan de Fuca plate system. We present a new onshore‐offshore S model and an updated P model. Major features are common to both models, including localized subslab low‐velocity anomalies, along‐strike variations in slab structure, and offshore heterogeneity, while regional differences may reflect changes in Vp/Vs. VL - 21 UR - https://app.dimensions.ai/details/publication/pub.1132665170 ER -