Separation of Multi-Mode Waveforms for Love Waves Using a Stacking-and-Stripping Technique

Abstract:

Higher-mode contamination is an inherent problem in Love wave tomography using fundamental mode data due to similar group velocities of different modes. It is essential to separate the fundamental mode waveform from the higher modes to improve the accuracy and the depth resolution of Love wave models. In this study, we have implemented several procedures to achieve this goal using synthetic and real waveform data. First, we identify multiple modes and estimate their phase velocities applying a frequency-waveform transform to filtered and windowed waveform data. The modes appear as separated maxima in a diagram of phase velocity versus group velocity. The second step is to separate waveforms of the identified modes by a phase-shifting and stacking approach. For each station, waveforms at all stations are shifted according to their epicentral distances using the fundamental-mode (mode 0) phase velocity and stacked to approximate the mode-0 waveform. The residual waveforms are obtained by stripping the stacked model 0 from the recorded waveforms. They are used to obtain the first higher-mode (mode-1) waveforms by repeatedly applying the stacking and stripping method. We validate the method using the synthetic waveforms containing mode-0 and mode-1 from the oceanic model. We also conduct experiments on earthquake data recorded by the Transportable Array stations in North America. Dispersion curves from the separated waveforms are much more coherent and precise than from the original data. This stacking and stripping technique can be applied to 2D seismic arrays to improve Love wave models.