AI Detects Hidden Earthquake Signals Near San Andreas Fault

Scientists have leveraged artificial intelligence to uncover subtle seismic signals along California's San Andreas Fault, a breakthrough that could refine our understanding of earthquake precursors. By analyzing extensive deformation data, researchers identified previously undetected "slow-slip events," which they believe may influence the timing and occurrence of low-frequency earthquakes (LFEs).

The ability to predict earthquakes remains one of seismology's greatest challenges. The U.S. Geological Survey officially states that accurate earthquake prediction is not currently feasible. However, this new research, published in Nature Communications, offers a glimmer of hope. The study focused on the San Andreas Fault near Parkfield, California, where dense networks of strainmeters collected nearly eight years of continuous deformation measurements between 2009 and 2016. These instruments capture minute shifts in the Earth's crust over durations ranging from seconds to weeks, filling a critical data gap between traditional seismometers and GPS sensors.

Zahra Zali, a geophysicist and seismologist and lead author of the new study, explained the significance of the AI's contribution. "We wanted to know if important slow displacement processes might be hidden in years of continuous deformation measurements," Zali stated. "Artificial intelligence enabled us to recognize their patterns, which would otherwise have gone unnoticed." The AI was trained on this vast dataset, which would be overwhelming for human analysis, to identify patterns indicative of these slow-slip events.

Understanding Aseismic Shifts

Tectonic faults can release built-up pressure either rapidly through seismic events or slowly through "aseismic" slips. These slow slips, which can persist from minutes to months, have historically been difficult to detect because they generate no significant seismic waves. "These events are difficult to identify using conventional methods because they are small and often hidden within complex background signals," Zali noted. This makes their potential connection to LFEs particularly intriguing, as LFEs themselves are often subtle and harder to attribute directly to large-scale tectonic stress buildup.

The research team, comprising scientists from Germany and the United States, found that these identified slow-slip events frequently coincided in time with LFEs occurring within approximately 6.2 miles (10 kilometers) and at depths under 12.4 miles (20 km). Patricia Martínez-Garzón, a professor of applied seismology at GFZ and coauthor of the study, elaborated on the findings: "Our results show that these 'earthquakes in slow motion' are not isolated phenomena, suggesting that slow sliding plays an important role in the development of stress conditions along active faults." This suggests a more interconnected system within fault lines than previously understood, where gradual movements can set the stage for more immediate seismic activity.

While the study identified a compelling link, it also highlighted a data disparity. The researchers had access to an estimated 500,000 LFEs during their study period but only identified 92 slow-slip events near Parkfield. This imbalance underscores the challenge of detecting these "quiet preambles" to seismic activity. Zali and her colleagues plan to expand their AI analysis to other fault lines to further validate the connection between slow slips and significant seismic events. "Many important faulting processes occur without causing damaging earthquakes," Zali said. "By detecting these hidden signals, we can gain a more complete picture of how faults behave between earthquakes and how stress is transmitted through the Earth's crust." The implications for future earthquake forecasting, while still distant, are significant, potentially moving seismology closer to a predictive capability.