NH017-01
Moment Magnitude Determination for Local Tsunami Warnings Using Seismogeodesy
Moment Magnitude Determination for Local Tsunami Warnings Using Seismogeodesy
Wednesday, 9 December 2020: 10:35
Virtual
Abstract:
A Tsunami from a local earthquake is expected to arrive onshore within minutes of the causative earthquake’s origin time. For that reason, it is crucial to reliably estimate the size of the earthquake as soon as possible and warn the closest populations. The Pacific Tsunami Warning Center (PTWC) uses the Mwp method (Tsuboi et al., 1995; Tsuboi et al., 1999) as the basis for issuing these initial Tsunami messages, nominally within 5 minutes of earthquake origin time. Mwp performs well over the 6<Mw<8.5 range, with three important caveats; first, for large (Mw>8-8.5) earthquakes, Mwp may saturate due to the seismic sensors limited long period sensitivity. Second, for complex earthquakes with several source time function peaks that are well separated in time, Mwp will underestimate the magnitude if the first peak is not the largest (2001 Mw8.4 Peru earthquake, for example). Third, and perhaps the greatest challenge for a Tsunami warning system, is the identification of slow rupture velocity earthquakes or "Tsunami Earthquakes" (Kanamori, 1972). Mwp underestimates the magnitude of these events because of their relatively slow moment release over a longer rupture time. The combination of GNSS with strong-motion data (Bock et al., 2011) yields both broadband velocity and displacement waveforms that do not clip, and are sensitive to the entire spectrum of ground motions from the Nyquist frequency of the accelerometer (typically 50Hz) to the static offset of the GNSS displacement. We use the velocity time series to define the appropriate coseismic time window for the displacement record. We theoretically consider these displacements as an approximate source time function, and integrate them to obtain moment as a function of time at each station. By integrating from the P-wave arrival time to the end of the coseismic time window we minimize the effect of long period noise. In order to demonstrate our approach, we collect data from several earthquakes in the 7.1<Mw<9.1 range, and obtain reliable moment magnitude estimates within minutes of rupture initiation that are based on theory instead of empirical relationships.