An Improved Algorithm for Real‐TimeS‐Wave Picking with Application to the (Augmented) ANZA Network in Southern California

We develop an automatic shear‐wave picking algorithm suitable for real‐time applications as well as with existing databases. The method can scan through packets of continuous waveforms and make picks without prior knowledge of whether earthquakes have occurred. This makes the algorithm suitable for detecting earthquakes at the same time. Expanding upon and improving the method of Ross and Ben‐Zion (2014a), the algorithm first uses polarization filters to remove P‐wave energy from the seismogram. Then, short‐term average/long‐term average and kurtosis detectors are applied to the data in tandem to lock in on the phase arrival. The method is tested by applying it to a full month of continuous waveform data recorded by a regional network at 123 stations and comparing the resulting automatic picks with 11,353 handmade picks. The automatic picks are found to be within 0.16 s of the analyst picks 75% of the time, and S picks are successful 92% of the time that a P‐wave pick is made. The algorithm is then applied to an entire year of continuous data and detects 11,197 earthquakes. The hypocenters of these earthquakes are, on average, improved by more than 1 km when compared with the regional network’s automated catalog