Comparison of 2D and 3D time-reverse imaging—A numerical case study

Time-reverse imaging has become an efficient tool to detect the origin of passively acquired seismic tremor signals. Practical experience has mainly been developed for 2D applications. Three component signals are reduced to two components and reverse propagated on the plane vertically below a station line. The data used for time-reverse imaging are the vertical and the horizontal particle displacement parallel to the line. Dropping the horizontal component perpendicular to the line causes partial loss of information on particle motion and directivity of the recorded waves. We present a comparison of 2D and 3D time-reverse imaging for a specific site with small cross-line gradients and investigate how closely 2D imaging approximates 3D imaging. Our large-scale synthetic survey with different S/N-ratios demonstrates how a subsurface source of tremor-like signals is imaged in different vertical planes. An imaging condition based on the energy density gives best results. We observe higher sensitivity to noise and stronger out-of-plane focusing for 2D than for 3D imaging. We suggest normalized visualization of multiple planes from 2D imaging in one 3D display as an approach to reliably locate sources. Comparison with examples of full 3D time-reverse imaging shows that normalized visualization of multiple 2D planes with a proper imaging condition can adequately approximate the result from full 3D imaging for the particular model considered in this study.