Detection of dispersive signals using multitaper dual-frequency coherence

Abstract

We demonstrate the use of 'dual-frequency' coherence in detecting and characterizing dispersive waves. Using a multitaper method, we calculate the coherence between different frequencies of one or multiple signals. We test the algorithm both on a variety of synthetic signals and on broad-band seismic data. Dispersive waves such as seismic surface waves are easily identified, and we show that the method is robust in the presence of noise. Phase relationships between different frequencies can be extracted, allowing reconstruction of the original phase function. 'Dual-frequency' coherence is useful in identifying overtones and frequency shifts between signals, features that are undetectable by standard coherence measures. We construct a filter to extract only the coherent frequencies from a waveform, and show that it significantly increases the signal-to-noise ratio for dispersive waveforms.