A Comparison Between Seismogeodetic and Baseline Corrected Estimates of Ground Motion: Example Applications for the 2011 Mw9.0 Tohoku-oki Earthquake

Monday, 15 December 2014
Jessie K Saunders1, Diego Melgar1, Yehuda Bock2, Dominga Sanchez3, Brendan W Crowell4, Jennifer Susan Haase5, Jianghui Geng6 and Dara Goldberg1, (1)Scripps Institution of Oceanography, La Jolla, CA, United States, (2)UCSD/IGPP 0225, La Jolla, CA, United States, (3)Jacobs School of Engineering, University of California San Diego, La Jolla, CA, United States, (4)University of Washington, Seattle, WA, United States, (5)UCSD, La Jolla, CA, United States, (6)University of California San Diego, La Jolla, CA, United States
Automated baseline correction of strong motion data is of broad interest for regional earthquake source modeling and for rapid earthquake response. Strong motion data are typically high pass filtered before inverting for source properties, thereby eliminating any information on static deformation. High-rate (1 Hz) GPS data preserve the long period information, particularly the static offset, but the data are much noisier at higher frequencies. Through a Kalman filter algorithm we demonstrate that, when available, collocated GPS sensors provide sufficient information for suitable baseline corrections to the strong motion data; we call this the seismogeodetic solution. We discuss in detail the superior recovery of low frequency information using the seismogeodetic approach when compared to baseline corrected accelerometer waveforms. The GPS/accelerometer combination provides broadband motions from the Nyquist frequency of the accelerometer down to the static offset. Because Kalman filtering is a real time operation that does not require operator interaction, we conclude that this approach could provide input for broadband source modeling that could potentially be calculated shortly after large events. To illustrate the performance and usefulness of this algorithm, we conduct a kinematic slip inversion of the seismogeodetic data from collocated GPS/strong motion pairs in Japan for the 2011 Mw9 Tohoku-oki earthquake using three-component Kalman filtered displacements and velocities for a total of 288 waveforms. Such source models are key to understanding the earthquake rupture process and for guiding rapid response, especially for large events such as this one. We quantify the differences between the inversion results using seismogeodetic data compared to those using conventional approaches by assessing the fit to ground motions at independent stations in Japan. In addition, we compare tsunami inundation levels predicted by the different models with actual field surveys that were conducted after the event. We will summarize the benefits of GPS instruments that are collocated with strong motion accelerometers in regional monitoring networks.