Data‐Driven Clustering Reveals More Than 900 Small Magnitude Slow Earthquakes and Their Characteristics

Abstract

AbstractSmall magnitude slow earthquakes remain largely undetected in geodetic data due to noise levels. However, tremor and low‐frequency earthquakes (LFE) may manifest slowly slipping fault motion as a cluster of events, i.e., a slow earthquake. Here, we identify >900 slow earthquakes in southwest Japan via data‐driven clustering of tremor and LFE catalogs. We establish a more complete database for slow earthquakes in southwest Japan and demonstrate their characteristics and long‐term behavior. While sometimes sub‐episodes of well‐known episodic tremor and slip, the small slow earthquake clusters share similar scaling properties–energy, duration, and rupture area–with larger magnitude fast and slow earthquakes. The small slow earthquake clusters tend to rupture faster when migrating in their preferred rupture direction, but inter‐event rupture speeds are on average similar to those of larger slow earthquakes. This suggests that rupture speed does not necessarily control slow earthquake spatial extent or eventual magnitude.