资源环境科技发展态势分析平台

Performance of earthquake early warning systems suffers from false alerts caused by local impulsive noise from natural or anthropogenic sources. To mitigate this problem, we train a generative adversarial network (GAN) to learn the characteristics of first-arrival earthquake P waves, using 300,000 waveforms recorded in southern California and Japan. We apply the GAN critic as an automatic feature extractor and train a Random Forest classifier with about 700,000 earthquake and noise waveforms. We show that the discriminator can recognize 99.2% of the earthquake P waves and 98.4% of the noise signals. This state-of-the-art performance is expected to reduce significantly the number of false triggers from local impulsive noise. Our study demonstrates that GANs can discover a compact and effective representation of seismic waves, which has the potential for wide applications in seismology.

Plain Language Summary Earthquake early warning systems are sometimes accidentally triggered by impulsive noise signals, rather than by real earthquake signals, which leads to false alerts. This may cause unnecessary economic loss and public concern. Here we use machine learning tools to determine if the waveforms are generated by earthquakes or local noise sources. We train the algorithms with about 700,000 waveforms recorded by southern California and Japan. We demonstrate that the trained machine learning discriminator can recognize 99.2% of the earthquakes and 98.4% of the noise. This discriminator can reduce a large number of false alerts and significantly improve the robustness of early warning systems.