3D double difference velocity tomography of the Middle America subduction zone beneath Nicaragua and Costa Rica

Waveform and arrival onset data collected on five amphibious arrays deployed along the Costa Rica-Nicaragua portion of the Middle America subduction zone are integrated to conduct high resolution velocity and location studies. Pick quality is evaluated using an automated arrival detection algorithm based on the wavelet transform and Akaike information criterion, resulting in revised pick weights for inversion studies. I explore the effect of new weighting and removal of poor data by relocating hypocenters through a minimum one dimensional velocity model and conducting double-difference local earthquake tomography (LET). Analysis of the hypocenter relocation and seismic velocity tomography results suggest that using the improved quality determinations improve sharpness in the velocity images.;Double-difference LET, utilizing catalog derived absolute and differential times and waveform cross-correlation derived differential times, is conducted with the quality-controlled dataset. Results show improved hypocentral locations of seismogenic zone earthquakes and compressional and shear velocity structure of the seismogenic zone. There is high variability in seismic structure along the length of the margin. I find that the up-dip limit of seismogenic zone microseismicity is variable but approximately corresponds to the 150°C isotherm. The downdip limit of interseismic microseismicity occurs near the continental Moho intersection with the subducting plate interface. Seismicity is sparser in Nicaragua and low velocities dominate the margin. Low Vp and high Vp/Vs in the oceanic mantle suggest serpentinization. Seismogenic zone seismicity resides in a low Vp and low Vp/Vs band that parallels the top of the high-velocity subducting slab. This low velocity band thin and weakens from north to south. This may reflect decreasing hydration or changes in overpressure along the plate interface. The addition of data from the TUCAN experiment provides additional raypath coverage along margin due to the broader aperture of the array. The results are compared to the coarse model and an ancillary method of calculating high-resolution Vp/Vs measurements. Using the highest resolution data along the Nicoya peninsula, we find that regions with subduction tremor and slow slip are associated with high Vp zones. These regions may be a better proxy signal for defining the limit of rupture during major earthquakes.