Automatic seismic phase picking based on unsupervised machine-learning classification and content information analysis

Accurate identification and picking of P- S-wave arrivals is important in earthquake exploration seismology. Often, existing algorithms are lacking automation, multiphase classification picking, as well performance accuracy. We have developed a new fully automated four-step workflow for efficient arrival times on microseismic data sets. First, time intervals with possible waveform recordings identified using the fuzzy c-means clustering algorithm. Second, these classified corresponding to P-, S-, or unidentified waves polarization attributes waveforms contained within. Third, unidentified-waves picked Akaike information criterion picker intervals. Fourth, P S based moveouts. The application synthetic real sets indicates that it yields accurate picks high low signal-to-noise ratio waveforms.