Assessing the Utility of Strong Motion Data to Determine Static Ground Displacements During Great Megathrust Earthquakes: Tohoku and Iquique

Abstract:

Strong motion accelerometers can record large amplitude shaking on-scale in the near-field of large earthquake ruptures; however, numerical integration of such records to determine displacement is typically unstable due to baseline changes (i.e., distortions in the zero value) that occur during strong shaking. We use datasets from the 2011 M_w 9.0 Tohoku earthquake to assess whether a relatively simple empirical correction scheme (Boore et al., 2002) can return accurate displacement waveforms useful for constraining details of the fault slip. The coseismic deformation resulting from the Tohoku earthquake was recorded by the Kiban Kyoshin network (KiK-net) of strong motion instruments as well as by a dense network of high-rate (1 Hz) GPS instruments. After baseline correcting the KiK-net records and integrating to displacement, over 85% of the KiK-net borehole instrument waveforms and over 75% of the KiK-net surface instrument waveforms match collocated 1 Hz GPS displacement time series. Most of the records that do not match the GPS-derived displacements following the baseline correction have large, systematic drifts that can be automatically identified by examining the slopes in the first 5-10 seconds of the velocity time series.

We apply the same scheme to strong motion records from the 2014 M_w 8.2 Iquique earthquake. Close correspondence in both direction and amplitude between coseismic static offsets derived from the integrated strong motion time series and those predicted from a teleseismically-derived finite fault model, as well as displacement amplitudes consistent with InSAR-derived results, suggest that the correction scheme works successfully for the Iquique event. In the absence of GPS displacements, these strong motion-derived offsets provide constraints on the overall distribution of slip on the fault. In addition, the coseismic strong motion-derived displacement time series (50-100 s long) contain a near-field record of the temporal evolution of the rupture, supplementing teleseismic data and improving resolution of the location and timing of moment in finite fault models.