Full waveform inversion and modeling

In addition to the strategy of incorporating multiple datasets as many as possible, utilizing the full information of seismic waveform becomes an appealing alternative recently. Benefited by the advances of numerical modeling techniques (e.g. SEM) and the application of adjoint theory in seismology. The full waveform tomography that is experimented successfully in south California (Chen et al., 2007; Tape et al., 2009) shows the potential for the new generation of tomography technique, in which the 3D off-path effects can also be considered more realistically. Current travel-time tomographic models of Taiwan region can simulate waveform up to 5-sec period (Huang et al., 2014). However, this achieved band has not been high enough for community applications (e.g. hazard mitigation), and more data (i.e. waveform information) and the 3D off-path effect have to be considered for higher frequency band.  The high-density seismic network (station spacing ~5 km) and abundant earthquakes (~17,000/yr) in Taiwan is promising for the test of 3D full waveform tomography, but also very challenged by the great tectonic complexity in such small-scale region of Taiwan.

Progress on AGU2014 [poster]

A practice demo of the event kernel for an deep earthquake southeast offshore