Automatic event detection and picking of P, S seismic phases for earthquake early warning and application for the 2008 Wenchuan earthquake

We proposed a set of high-precision combination algorithms to detect and pick seismic phases for the earthquake early warning (EEW). First by the polarisation analysis of the three-component seismograms, we developed two filters to separate P and S waves for each record through a sliding time window. Then based on the short term average/long term average characteristic function on the polarised traces, an amplification coefficient, in terms of δ, was introduced to be multiplied by the original ratio to sensitively reflect the changes of signals' amplitude and frequency, where a better detection of the phase arrival was achieved. Next according to the preliminary detections, we used the Akaike information criteria function combined with the higher order statistics to refine the signal and lock on the arrival time with a higher degree of accuracy. We tested our techniques to the main-shock and aftershocks of the Ms 8.0 Wenchuan earthquake, where hundreds of three-component acceleration records over magnitudes of Ms 5.0 were treated. In comparison to the analyst picks, the results of the proposed detection algorithms were shown to perform well and can be applied for the early warning of impending earthquakes occurred with diverse focal mechanisms, complicated propagation process and site effects.