Data-driven dispersive surface-wave prediction and mode separation using high-resolution dispersion estimation

Abstract

In seismic data processing, surface waves and body waves are usually processed differently. We propose a data-driven approach to predict and separate the dispersive surface waves in multichannel seismic data. We first estimate high-resolution surface-wave phase velocities from the recorded data based on the multi-channel nonlinear signal comparison (MNLSC) technique. This enables us to predict the surface waves at each receiver location. We then separate the predicted surface waves from the seismic data. The predicted surface waves can be further separated into the fundamental mode and higher modes if they are present. We have successfully applied our approach on two synthetic seismic gathers and one field seismic gather. These examples show that our new approach could effectively predict and separate dispersive surface waves from the data without damaging reflections.