[S-CG40] Science of slow-to-fast earthquakes

[SCG40-P16] Source Imaging of Subduction Zone Tremors Using Time Reversal Imaging

*Amarjeet Kumar¹, Rie Nakata^2,3, Yusuke Yamashita⁴ (1.Indian Institute of Science Education and Research Kolkata, 2.Earthquake Research Institute, University of Tokyo, Tokyo, Japan, 3.Lawrence Berkeley National Laboratory, Berkeley, USA, 4. Disaster Prevention Research Institute, Kyoto University, Japan)

Keywords:Tremors, Time Reversal Imaging, subduction zone

Tremors are weak long-duration slow earthquakes that indicate fluid movement and stress condition changes along the plate boundary, in the volcanic chambers, in hydrocarbon and renewal energy production and potentially a precursor to larger fast earthquakes. Tremors migrate spatiotemporally and understanding their evolution requires reliable estimates of the source location of tremors. However, due to their weak signal, observed seismic records lack clear P and S wave arrivals, making them extremely difficult to locate with classic hypocenter determination methods. Alternative methods have been proposed, but they use only a part of the waveforms. We propose to utilize time reversal imaging and to exploit the full waveform information with no a priori assumption about source or observed signals.

We investigate the feasibility of the time reversal imaging method by using a synthetic dataset created for the Nankai Trough subduction zone. We backpropagate the generated 3 component tremor time series from multiple receiver locations simultaneously back into the medium. The reversed signals focus at the tremor source location due to the constructive interference of the back-propagated wavefields from all individual receivers. Given that tremors are long-duration signals with multiple peaks, the reversed signals appear and disappear at the source location. Hence, individual time snapshots may be inappropriate to characterize the interference pattern. We propose to stack the reversed signals in the time domain using a time-window to enhance the focusing effect at the source location.

We generate a synthetic tremor source and conduct simulations of forward and backward wave propagation using the 2D finite difference method to solve the elastic wave equation. A 2D section of the GO_3D_OBS model is used as an elastic parameter model. The receivers are positioned at the seafloor, and the sources are located at the subduction boundary interface. We test both isotropic and double-couple tremor sources, and place sources at various locations along the plate boundary. Additionally, we test adding noise to the receiver signal, increasing station spacing, and using inaccurate velocity structures. We demonstrate that we can accurately image sources up to S/N ratio of 0.2 and sensor intervals of 20 km, and the bulk velocity shift causes systematic depth errors in the tremor location. We are now extending our simulations to 3D cases and also applying our methodology to real tremor data from the Hyuga-nada region in the subduction zone.