[S-CG45] Science of slow-to-fast earthquakes

[SCG45-P34] DRESS: A Method for Detecting Repeating Earthquakes with a Single Station

*Xinrong Hou^1,2, Dawei Gao^1,2, Aitaro Kato², Zhenwei Guo¹ (1.Central South University, 2.University of Tokyo)

Keywords: Repeating Earthquake Identification, Waveform Similarity, Coda Wave

Repeating earthquakes (REs), recurrently rupturing the same fault patch, are critical for investigating fault weakening and healing processes. However, reliably identifying REs with limited data remains challenging. To address this issue, we propose a new method, named Detecting Repeating Earthquakes with a Single Station (DRESS). DRESS essentially adopts a hybrid approach by integrating waveform similarity and source physics, and operates in four stages:
1. Potential REs Identification: Potential REs are identified in existing massive catalogs using hierarchical clustering analysis or detected using cataloged events as templates to scan continuous waveform data.
2.Robust Waveform Similarity Filtering: A match-filtering with multi-segment cross-correlation (MFMC) method filters candidates with robust waveform similarity by considering the contributions from all segments of the waveforms.
3.Magnitude Consistency Check: Event pairs with magnitude differences exceeding +/-0.3 are excluded to ensure magnitude consistency.
4.Source Overlap Analysis: We use coda wave interferometry and coda spectral ratio analysis to precisely estimate inter-event distances (D) and corner frequency (f), respectively. An event pair is classified as REs if D/Rmax <= 0.8, where Rmax, the rupture radius of the larger event, is derived from f.
The effectiveness of DRESS has been validated using earthquake data recorded by a single station from Mogul. Compared to traditional methods, DRESS enhances the feasibility of REs identification in regions with sparse station coverage.