Overview
- The Nature Communications study, published June 9, 2026, produced the first catalogue of short‑duration slow slip events beneath the Parkfield section of the San Andreas Fault.
- Researchers used a deep‑learning autoencoder and unsupervised clustering to scan years of continuous borehole strainmeter data and isolate dozens of hours‑to‑days long aseismic slip episodes that conventional methods missed.
- The detected slips occur at shallow depth, show right‑lateral motion consistent with the San Andreas Fault, and follow the same seismic moment–duration scaling observed for ordinary earthquakes.
- Low‑frequency earthquake activity rose systematically after the identified slow slips, indicating these silent movements can change local stress conditions though the study does not show they directly trigger large quakes.
- Because Parkfield is a densely instrumented natural laboratory, the team says applying the AI workflow to other fault networks could reveal more hidden slip and improve understanding of how silent faults influence earthquake processes.