Imaging Low-Frequency Earthquakes with Geometric-Mean Reverse Time Migration

Abstract:

Time reversal is a powerful tool to image directly both the location and mechanism of sources. This technique assumes seismic velocities in the medium and propagates time-reversed observations of ground motion from each receiver location. Assuming an accurate velocity model and adequate array aperture, the waves will focus at the source location. Although multiple sensors are used simultaneously to estimate the source parameters, we can only image temporally compact sources due to a technical limitation of back projection. In this study, we propose a new approach for passive seismic migration that contains crosscorrelation within the time-reversal scheme. We first individually extrapolate wavefields at each receiver, and then crosscorrelate these wavefields (as a product in the frequency domain: Geometric-mean RTM, GmRTM). Because of the correlation, we can accumulate the energy of sources along the time axis in the image domain and enhance the source signals when the source has extended duration. As a test of this technique, we apply our RTM to synthetic earthquake waveforms and low-frequency earthquakes in Mexico. Results in Guerrero are compared with tectonic tremor locations determined with an independent technique, namely the Tremor Energy and Polarization (TREP) method. We successfully improve the SNR of the source image compared with conventional time-reversal imaging.