Using the similarity Matrix Profile to detect and study foreshocks and aftershocks of the 2019 Ridgecrest sequence

The July 2019 Ridgecrest earthquakes - the July 4 M6.4 and July 5 M7.1 events -- occurred 34 hours apart on a conjugate fault system. Multiple nearby seismic stations monitored this region continuously during the mainshocks and aftershocks, making it a promising case for studying the physics of interactions between foreshocks, the mainshocks and aftershocks, as well as specific questions arising from the sequence, e.g. why the M7.1 source fault did not rupture dynamically in the M6.4 event, and why it took 34 hours for the second mainshock to nucleate and rupture.

Here we apply our new and advanced similarity search method, the similarity Matrix Profile (MP, Zhu et al., 2016) to study aftershocks and foreshocks of the full Ridgecrest sequence. The MP approach is similar to template matching, but does not require any templates, as all sub-windows in the continuous stream of seismic data are compared with the rest of the data. Thus, the MP method might find groups of events that were not detected by the catalog, and therefore could not be detected with the template matching method.

Using 35 days of data from station SRT, starting 14 days before the M6.4 event, we detect ~40 high correlation coefficient (CC; 0.95 or greater) events per hour following the M6.4 event. These highly similar events could be very closely separated or even repeating events (same fault patch ruptures multiple times). The rate of high CC events drops substantially to ~25 per hour immediately after the M7.1 event, and then slowly decreases (i.e. with Omori decay) afterward. This suggests the M7.1 earthquake might have reduced stresses in some regions where the M6.4 event caused increases. We aim to apply our method to all stations in this region for the ongoing aftershock sequence, to detect as many events as possible, minimizing false positive detections, and find out if there are groups of events missed by template matching. We will then locate those highly similar events along with any other uncataloged events (with lower CCs) either by cross-referencing them to the catalog or using double-difference relocation with respect to catalog events, and use them to test possible clustering of foreshocks near the M7.1 hypocenter and quantify Gutenburg-Richter and Omori-Utsu relationships for this sequence.