Automatic microseismic event detection by band-limited phase-only correlation

Identification and detection of microseismic events is a significant issue in source locations and source mechanism analysis. The number of the records is notably large, especially in the case of some real-time monitoring, and while the majority of microseismic events are highly weak and sparse, automatic algorithms are indispensable. In this study, we introduce an effective method for the identification and detection of microseismic events by judging whether the P-wave phase exists in a local segment from a single three-component microseismic records. The new judging algorithm consists primarily of the following key steps: 1) transform the waveform time series into time-varying spectral representations using the S-transform; 2) calculate the similarity of the frequency content in the time-frequency domain using the phase-only correlation function; and 3) identify the P-phase by the combination analysis between any two components. The proposed algorithm is compared to a similar approach using the cross-correlation in the time domain between any two components and later tested with synthetic microseismic datasets and real field-recorded datasets. The results indicate that the proposed algorithm is able to distinguish similar and dissimilar waveforms, even for low signal noise ratio and emergent events, which is important for accurate and rapid selection of microseismic events from a large number of records. This method can be applied to other geophysical analyses based on the waveform data.