TimePicker 2017 – a fully automatic tool to extract P-wave arrivals for high-resolution unravelling structure and fabric of the lithosphere-asthenosphere system

Teleseismic body-wave tomography represents a powerful tool to study regional velocity structure of the upper mantle. Particularly, a need of retrieving anisotropic signal calls for processing of a huge amount of P-wave travel times (Munzarová et al., GJI 2018).  Therefore, automatic picking procedures are needed to supply tomography codes with a large amount of highly accurate absolute arrival times and/or travel-time residuals of body-wave propagation. We present and test a fully automated tool - TimePicker 2017 (Vecsey et al., 2021) for measuring P-wave arrival times on array recordings of passive experiments. The TimePicker 2017 is developed in the ObsPy/Python platform (Krischer et al., 2015) which combines picking, waveform cross-correlation and beamforming. The picker is based on two-step signal cross-correlations and allows us to measure absolute arrival times. Instead of a subjective selection of a reference trace, it cross-correlates all pairs of traces and forms a reference low-noise beam trace as a stack of the shifted traces at all stations. The picker cross-correlates all signals to the reference beam, automatically identifies outliers, and complements all picked absolute arrival times by their error estimates.

We applied the TimePicker 2017 on a set of seismograms from 1920 earthquakes from epicentral distances greater than 30° recorded at 240 temporary and permanent stations involved in the AlpArray experiments. We show uncertainties of measured P-wave arrivals, means and medians of uncertainties for both the complete dataset as well as for subset selected for tomography, and test effects of the standard selection of a reference trace vs. the low-noise beam trace as the reference trace in the TimePicker 2017.