A 3-D Tomographic Model of the P-wave Velocity Structure of the Central Cascadia Forearc

Graduation date: 2016Signals from the R/V Langseth’s tuned airgun array were recorded on an array of 33 EarthScope FlexArray seismometers deployed in the Oregon Coast Range from 43.5° to 45°N as part of the 2012 Ridge2Trench experiment to image the structure of the Juan de Fuca plate. This segment of the Juan de Fuca/North America plate boundary slipped in 2 moderate low-angle thrust earthquakes in 2004, and continuing seismicity has clustered around these events since that time. Previously acquired bathymetric, potential field, and seismic data indicate the presence of subducted seamounts in this region and hinted at large velocity variations of the overlying forearc crust immediately above the plate boundary. Approximately 18000 airgun shots along 4 profiles on the continental shelf and upper slope and a profile oblique to the coast that extends across the margin to the trench provided sources for this study. Data quality is excellent, with strong Pg and PmP arrivals observed at most stations for most shots. We present velocity models that merge the new data with constraints from onshore/offshore data acquired in 1989 and 1996 to improve the resolution and extend the velocity model for this segment of the Cascadia subduction zone, which appears to be a segment boundary in the paleoseismic record and currently exhibits anomalous locking behavior as indicated by geodetic data. Our approach to modeling involved starting with a 1D P-wave velocity model and using an inversion process to modify the model to beetter fit the observed first arrival times of the airgun shots. In a subsequent step, we used a 2D starting model that incorporated constraints from prior 2D experiments that extended farther seaward and landward of the 3D study region. Finally, a smoothed version of the model obtained from the 2D inversion was used as a starting model to see if the final model could be further refined. Our final preferred model used a 2D starting model and moderate smoothing. The RMS misfit of predicted travel times is reduced from 0.48s for the initial 2D model to 0.040s for the final preferred 3D model. This model includes interesting velocity structure including possible subducted seamounts on the subducting Juan de Fuca plate and a mafic dike related to Cape Perpetua volcanics within the upper plate. These results contribute to the broader goal of understanding the relationship between plate boundary behavior and crustal structure