Researchers develop AI method to predict earthquakes months in advance.

A team of researchers has taken a crucial step in earthquake prediction by developing a machine learning-based method that could allow for the identification of low-level tectonic agitation months in advance. This innovative approach, presented in a study published in the prestigious scientific journal Nature Communications, focuses on analyzing large volumes of seismic data. Led by Társilo Girona, an assistant research professor at the Geophysical Institute of the University of Alaska Fairbanks, and Kyriaki Drymoni, a geologist at Ludwig Maximilians University in Munich, the team has employed advanced algorithms to detect patterns that might indicate the imminent occurrence of a major earthquake. The study is based on the premise that, before a significant quake occurs, there are precursory signals in the form of lower magnitude seismic activity. According to the researchers, this phenomenon primarily manifests in earthquakes with magnitudes below 1.5. By applying their algorithm to the earthquakes in Anchorage in 2018 and Ridgecrest in 2019, the team was able to identify abnormal seismic activity in approximately 15% to 25% of the affected areas about three months before the main events. This finding is not only significant but also offers a new perspective on how earthquakes can be understood and anticipated. Girona explains that the use of advanced statistical techniques, particularly machine learning, has shown its potential in identifying precursors to major earthquakes. The algorithm developed by the team seeks signals of unusual seismic activity in large datasets, representing a breakthrough in the forecasting of these natural phenomena. The ability to predict a seismic event can be crucial for the safety of communities in at-risk areas. In the case of Anchorage, the researchers observed that the probability of a significant earthquake rose to 80% in the 30 days leading up to the event on November 30. Just before the earthquake, this figure reached 85%. Similarly, in Ridgecrest, the probability increase began about 40 days before the first quake in the sequence, demonstrating the effectiveness of the method in early prediction. One of the theories suggested by the researchers is that the precursory activity could be related to variations in interstitial fluid pressure within tectonic faults. Drymoni argues that these variations control the abnormal and low-magnitude precursory seismicity, adding a new dimension to the understanding of earthquake mechanics and their prediction. However, the team has also emphasized the need for caution. Girona warns that this method should not be applied to new regions without prior training on historical seismicity data from the area. This is essential to ensure the accuracy of predictions, as each region has unique tectonic characteristics that can influence seismic activity. Moreover, the researchers have highlighted the ethical challenges posed by earthquake prediction. The possibility of issuing false alerts or making inaccurate predictions can generate panic in communities and undermine trust in scientific institutions. Therefore, it is crucial that any advancements in this field are handled responsibly and communicated effectively to the public. Accurate earthquake forecasting has the potential to save lives, especially in areas where seismic activity is frequent. However, as Girona concludes, it also raises important ethical and practical questions that must be addressed as this research progresses. The combination of machine learning and seismic sciences could herald a new era in the prediction of natural disasters, offering hope and security to many vulnerable communities. This advancement not only represents a milestone in seismic research but could also change how cities and governments prepare to face earthquakes, redesigning risk prevention and mitigation policies. Over time, this approach could be integrated into early warning systems, providing communities with more time to react and prepare in the face of an impending major quake.