Method of automated extracting dispersion curves based on time-frequency distribution of seismic data

Abstract

This paper discusses examples of testing a new implementation of the method of multi-channel surface wave analysis on synthetic data computed for elastic media with complex boundary geometry. The new implementation of the method includes developed algorithms for noise-resistant spectral analysis based on time-frequency domain filtering of seismograms and inversion of dispersion curves of phase velocities based on determination of ranges of possible transverse wave velocity models and application of artificial neural networks. Based on the results of synthetic data processing, the accuracy, lateral resolution limitations and applicability limits of the method under consideration are evaluated.