Monitoring shear-wave splitting at the CRL-NFO in near real-time

Characterization of Shear-wave Splitting (SwS) from local earthquake recordings can oftentimes prove challenging, as the fastshear-wave’s arrival may not be impulsive or its window may be contaminated by noise and secondary phases. Thus, accurately measuringSwS with automatic methods still requires heavy manual validation to avoid systematic issues. The difficulty in producing highly accurateresults may seem to render an automated workflow almost impossible to implement. Herein, we present a near real-time workflow for therobust measurement of SwS in the context of a Near-Fault Observatory (NFO). Our workflow is designed to tackle the issues of dataselection (as not all local recordings are suitable), filter selection and quality control. By syncretizing established methods (i.e., adaptivefiltering, rotation-correlation, eigenvalue optimization and cluster analysis) with a logic focused on near real-time application, we aim toprovide a scalable and efficient solution. The pipeline retrieves real-time information (waveforms, events and picks) from an earthquakemonitoring system, assesses their eligibility, analyzes SwS automatically and produces comprehensive quality control metrics to decidewhether the measurement is acceptable. Optimizations at the signal window level ensure the faster processing of, even single, event-station pairs. The workflow is capable of detecting and grading null measurements as well. Our workflow was tested and validated on real-time scenarios from the Corinth Rift Laboratory (CRL) NFO. We created a dataset comprised of waveforms that were analyzed with aconventional automatic SwS and were validated manually. Testing results indicated good accuracy of the automatic system, with a tradeoffbetween the measurement quality and volume of observations.