Strong Upper‐Plate Heterogeneity at the Hikurangi Subduction Margin (North Island, New Zealand) Imaged by Adjoint Tomography

Abstract

In New Zealand, ray-based seismic tomography has produced detailed images of an active convergent plate boundary (Eberhart-

Evidence for Deeply Subducted Lower‐Plate Seamounts at the Hikurangi Subduction Margin: Implications for Seismic and Aseismic Behavior

“…We perform a zeroth moment resolution test to determine how robustly imaged these velocity structures are (Chow et al., 2022). Depth slices through the zeroth moment volume are shown in Figure A2, using a threshold value chosen to represent the lateral extent of sensitivity in our velocity model.…”

“…Green crosses show the locations of velocity anomalies below Pōrangahau and Māhia Peninsula. Pink circles and blue inverted triangles show earthquakes and receivers used to derive the velocity model (Chow et al., 2022). Thin black lines show active faults (Litchfield et al., 2014).…”

“…In a companion paper (Chow et al., 2022), we use adjoint tomography, an imaging technique that involves fitting short‐period ($$ > 4\hspace*{.5em}\mathrm{s}$$) earthquake‐generated seismic waveforms to corresponding synthetic waveforms, to refine a 3D velocity model of the North Island of New Zealand (Eberhart‐Phillips et al., 2020). Throughout the inversion, strong velocity anomalies in the forearc region are imaged at increasing resolution.…”

SASSIER22: Full‐Waveform Tomography of the Eastern Indonesian Region That Includes Topography, Bathymetry, and the Fluid Ocean

“…This has traditionally limited its usefulness, but ongoing access to more powerful computing resources has enabled its successful application in earthquake seismology across the scales (e.g., Blom et al, 2020;M. Chen et al, 2015;Chow et al, 2022;Fichtner et al, 2009;Gao et al, 2021;Lei et al, 2020;Rodgers et al, 2022;Tape et al, 2010).…”

“…This makes the method much more computationally expensive than ray‐based imaging methods; the computational cost of adjoint waveform tomography is independent from the number of receivers but scales linearly with the number of events used and to the fourth power of the highest frequency in 3‐D (e.g., Tromp, 2020). This has traditionally limited its usefulness, but ongoing access to more powerful computing resources has enabled its successful application in earthquake seismology across the scales (e.g., Blom et al., 2020; M. Chen et al., 2015; Chow et al., 2022; Fichtner et al., 2009; Gao et al., 2021; Lei et al., 2020; Rodgers et al., 2022; Tape et al., 2010).…”