Rapid and automated reporting of earthquake focal mechanisms is crucial for understanding faulting geometry, stress changes, and aftershock patterns. This paper introduces Focal Mechanism Network (FMNet), a deep learning method trained on synthetic data to estimate focal mechanisms in real-time. FMNet successfully predicted focal mechanisms for four 2019 Ridgecrest earthquakes (Mw > 5.4), demonstrating its potential for application in regions with limited historical data. Prediction time is under 200 milliseconds on a single CPU.

Wenhuan Kuang, Congcong Yuan, Jie Zhang