New Algorithm Can Predict Major Earthquakes Months Ahead

A groundbreaking study from the University of Alaska Fairbanks indicates the possibility of predicting major earthquakes months in advance, leveraging machine learning to identify subtle seismic patterns that foretell significant quakes. This research, published in Nature Communications by UAF research assistant professor Társilo Girona and his colleague Kyriaki Drymoni from Ludwig-Maximilians-Universität, showcases how advanced statistical techniques can analyze earthquake catalogs to identify precursory seismic activity. This is preapred by SSP.

The study focused on two significant earthquakes: the 2018 Anchorage earthquake (magnitude 7.1) and the 2019 Ridgecrest, California earthquake sequence (magnitudes 6.4 to 7.1). The researchers’ algorithm indicated roughly three months of unusual low-magnitude seismic activity over 15% to 25% of Southcentral Alaska and Southern California before each quake, with most precursor activity featuring magnitudes below 1.5. Their findings revealed a sharp increase in the probability of a major quake occurring within 30 days, rising to about 80% three months prior, and escalating to 85% just days before the event.

Girona and Drymoni propose that increased pore fluid pressure within faults is the likely cause. This pressure alters faults' mechanical properties and causes uneven regional stress variations, triggering abnormal, low-magnitude seismicity. "Modern seismic networks generate vast datasets that, when coupled with machine learning, can reveal meaningful patterns before seismic events," Girona noted.

The potential for such predictive capabilities has profound implications for public safety and disaster preparedness, including timely evacuations, infrastructure reinforcement, and preparation of emergency services. However, Girona emphasizes the importance of perfecting and testing the algorithm further in real-world situations, ensuring it is tailored to specific regions’ historical seismic data to avoid false alarms or missed predictions.

As technological advancements in machine learning progress, this research could significantly enhance our ability to forecast major earthquakes, reducing economic losses and saving lives, while addressing the inherent ethical and practical challenges of earthquake forecasting.