The increasing volume of seismic data from long-term continuous monitoring motivates the development of algorithms based on convolutional neural network (CNN) for faster and more reliable phase detection and picking. However, many less studied regions lack a significant amount of labeled events needed for traditional CNN approaches. In this paper, we present a CNN-based Phase-Identification Classifier (CPIC) designed for phase detection and picking on small to medium sized training datasets. When trained on 30,146 labeled phases and applied to one-month of continuous recordings during the aftershock sequences of the 2008 M W 7.9 Wenchuan Earthquake in Sichuan, China, CPIC detects 97.5% of the manually picked phases in the standard catalog and predicts their arrival times with a five-times improvement over the ObsPy AR picker. In addition, unlike other CNN-based approaches that require millions of training samples, when the off-line training set size of CPIC is reduced to only a few thousand training samples the accuracy stays above 95%. The online implementation of CPIC takes less than 12 hours to pick arrivals in 31-day recordings on 14 stations. In addition to the catalog phases manually picked by analysts, CPIC finds more phases for existing events and new events missed in the catalog. Among those additional detections, some are confirmed by a matched filter method while others require further investigation. Finally, when tested on a small dataset from a different region (Oklahoma, US), CPIC achieves 97% accuracy after fine tuning only the fully connected layer of the model. This result suggests that the CPIC developed in this study can be used to identify and pick P/S arrivals in other regions with no or minimum labeled phases.

We apply a waveform matching technique to obtain a detailed earthquake catalog around the rupture zone of the 5 September 2012 moment magnitude 7.6 Nicoya earthquake, with emphasis on its aftershock sequence. Starting from a preliminary catalog, we relocate ~7900 events using TomoDD to better quantify their spatiotemporal behavior. Relocated aftershocks are mostly clustered in two groups. The first is immediately above the major coseismic slip patch, partially overlapping with shallow afterslip. The second one is 50 km SE to the main shock nucleation point and near the terminus of coseismic rupture, in a zone that exhibited little resolvable afterslip. Using the relocated events as templates, we scan through the continuous recording from 29 June 2012 to 30 December 2012, detecting approximately 17 times more than template events. We find 190 aftershocks in the first half hour following the main shock, mostly along the plate interface. Later events become more scattered in location, showing moderate expansion in both along‐trench and downdip directions. From the detected catalog we identify 53 repeating aftershock clusters with mean cross‐correlation values larger than 0.9, and indistinguishably intracluster event locations, suggesting slip on the same fault patch. Most repeating clusters occurred within the first major aftershock group. Very few repeating clusters were found in the aftershock grouping along the southern edge of the Peninsula, which is not associated with substantial afterslip. Our observations suggest that loading from nearby afterslip along the plate interface drives spatiotemporal evolution of aftershocks just above the main shock rupture patch, while aftershocks in the SE group are to the SE of the observed updip afterslip and poorly constrained.

Foreshocks can provide valuable information about possible nucleation process of a mainshock. However, their physical mechanisms are still under debate. In this study, we present a comprehensive analysis of the earthquake sequence preceding the 2010 M w 7.2 El Mayor-Cucapah mainshock, including waveform detection of missing smaller events, relative relocation, and source parameter analysis. Based on a template matching method, we find a tenfold increase in the number of earthquakes than reported in the Southern California Seismic Network catalog. The entire sequence exhibits nearly continuous episodes of foreshocks that can be loosely separated into two active clusters. Relocated foreshocks show several seismicity streaks at depth, with a consistently active cluster at depths between 14 and 16 km where the mainshock was nucleated. Stress drop measurements from a spectral ratio approach based on empirical Green's functions show a range between 3.8 and 41.7 MPa with a median of 13.0 MPa and no clear temporal variations. The relocation results, together with the source patches estimated from earthquake corner frequencies, revealed a migration front toward the mainshock hypocenter within last 8 hr and a chain of active burst immediately 6 min prior to the mainshock. Our results support combined effects of aseismic slip and cascading failure on the evolution of foreshocks. Plain Language Summary The 2010 M w 7.2 El Mayor-Cucapah (EMC) earthquake was preceded by a prominent sequence of foreshocks starting~21 days before the mainshock. Several methods based on the similarities of waveforms are applied to obtain spatiotemporal evolution of foreshocks. Ten times more events are found from a template matching method when compared to the SCSN catalog. The refined relative locations reveal two main active clusters in time, as well as two spatial patches with a shallower one to the north of the mainshock epicenter. The depth distribution indicates several linear lines of seismicity, with a consistently active cluster at depths of 14-16 km where mainshock started. An active cluster of foreshocks occurred in the last 6 min. They likely altered the stress state near the hypocenter and ultimately triggered the mainshock. Our analysis indicates that both aseismic slip and cascade triggering processes occurred and contributed to the eventual triggering of the EMC mainshock.