SCEC Award Number 25260 View PDF
Proposal Category Collaborative Research Project (Multiple Investigators / Institutions)
Proposal Title Using both waveform coherence and complexity to cluster microearthquakes
Investigator(s)
Name Organization
Wenyuan Fan University of California, San Diego Peter Shearer University of California, San Diego
SCEC Milestones A2-3, A3-2, D1-1, D2-1 SCEC Groups Seismology, RC, FARM
Report Due Date 03/15/2026 Date Report Submitted 03/27/2026
Project Abstract
 Earthquake waveforms encode critical information about fault architecture, stress heterogeneity, and faultzone material properties. Traditional clustering of microearthquakes relies primarily on cross-correlation coefficients, which capture average waveform similarities but do not fully characterize the complexities of entire wavetrains. In this project, we developed a new earthquake clustering procedure that combines waveform coherence (cross-correlation) with an unsupervised machine learning method, the sequencing algorithm (Baron and Ménard, 2021), to cluster microearthquakes based on their full waveform characteristics. The sequencing algorithm orders a set of seismic records by maximizing the similarity between adjacent waveforms (e.g., Carr and Olugboji, 2024), capturing gradual changes in waveform complexity that conventional metrics overlook (Kim et al., 2020; Fang, 2024). This approach clusters events based not only on space and time but also on their entire wavetrains, providing a new means to detail earthquake source properties and fault-zone conditions. We have successfully developed and tested this method and applied it to the 2024 Mw 7.5 Noto Peninsula earthquake sequence in Japan, where a prolonged foreshock sequence offers a unique window into the preparatory processes of a major earthquake. Our results demonstrate that the method can effectively identify seismicity bursts and resolve high-resolution in-situ VP/VS ratios, yielding new insights into the temporal evolution of fault-zone stress, fluid conditions, and fracture networks prior to large earthquakes.
Intellectual Merit The central goal of this project was to develop a machine-learning-enabled clustering procedure that leverages both waveform coherence and complexity to group microearthquakes. A key advantage of the method is that the identified clusters yield high-quality differential arrival time measurements for both P- and S-waves, which can be directly used for resolving relative earthquake locations, estimating relative focal mechanisms, and inverting for in-situ VP/VS ratios. These capabilities make the clustering procedure a versatile tool that integrates with a range of existing earthquake analysis workflows.
Broader Impacts A manuscript presenting the full results of the Noto Peninsula application is currently in preparation. This work constitutes a core component of the Ph.D. thesis of Nicolas DeSalvio, a graduate student supported by this project. We plan to standardize the workflow and release the algorithm as an open-source software package, ensuring broad accessibility and encouraging community-driven applications.
Project Participants Nicolas DeSalvio, Wenyuan Fan, and Peter Shearer from SIO/UCSD.
Exemplary Figure Figure 3: In-situ VP/VS estimates near the Mw 7.5 Noto earthquake region, showing high anomalies prop-
agating from deep toward shallower depths. The results imply the presence of possible fluid pathways in the
complex fault network.
Linked Publications

Add missing publication or edit citation shown. Enter the SCEC project ID to link publication.