[S-SS35_2PM1] Recent developments in microtremor surveys

[SSS35-04] Toward Earthquake Ground Motion Prediction using the Onshore-Offshore Ambient Seismic Field in Subduction Zones

*Loic VIENS¹, Hiroe MIYAKE¹, Kazuki KOKETSU¹ (1.Earthquake Research Institute, University of Tokyo)

Keywords:Ground motion Prediction, Ambient seismic field, Subduction zones, Low frequency, Deconvolution

Ground motion prediction is critical to evaluate the seismic hazard specially in high seismicity areas as Japan. A source of particular concern is the complex geological structures as sedimentary basins which can trap and amplify seismic waves. It has been proved by Prieto and Beroza (2008) that reliable phase and amplitude of the impulse response functions can be extracted by deconvolution of the ambient seismic field recorded by two on-land stations without any pre-processing. This approach has the great advantage to predict accurate ground motion of moderate earthquakes at periods longer than 4 s without the need of any external information about the velocity structure. However, this method allows only to recover relative, rather than absolute, amplitudes. To retrieve the corresponding Green's functions, impulse response amplitudes need to be calibrated using records of an earthquake which happened close to the "virtual" source. Moreover, as surface-to-surface Green's functions are extracted, some mismatches are observed between Green's functions and the earthquake records. This feature is due to the fact that depth and focal mechanism of the event are not taken into account. Despite of these disadvantages, accuracy of the predicted ground motion is high and such long-period ground motion investigation is critical to carried out seismic hazard assessment for high-rise buildings, bridges, or oil tank having long-period resonance. In this study, we use this technique in subduction zones to extract vertical-to-vertical component of the Green's functions between seismic stations located on the ocean bottom and on-land Hi-net stations. The target region is located in the Tokai/Tonankai areas where two submarine cable-based sea-bottom seismographic observation systems have been deployed by the Japan Meteorological Agency (JMA). We use one month of noisy data recorded in January 2013 to compute the Green's functions. The choice of these data is motivated by a strong signal-to-noise ratio of the causal part of the Green's functions during this period. We validate this approach by comparing computed Green's functions with offshore moderate earthquake (Mw ~5) records in the Nobi sedimentary basin where the Nagoya city is located. As megathrust earthquakes are expected in this area, we extrapolate our results to predict magnitude Mw ~6 or larger earthquake ground motions using the scaling law of seismic spectrum developed by Aki (1967). These results are finally compared to long-period ground motion prediction equations to evaluate their validity.