Following the earthquakes in Turkey, a team of scientists from Ariel University led by Yuval Reuveni published a paper about a new method for predicting earthquakes 48 hours in advance. The method uses machine learning, specifically support vector machine (SVM) algorithms, and GPS ionospheric total electron content (TEC) estimates. The team has been researching the relationship between TEC and geodynamic activity for years.

SVM algorithms analyze data to identify patterns and relationships and can predict the occurrence of specific events. By using SVM to analyze TEC data from the US Geological Survey, Israeli scientists were able to accurately predict seismic activity’s location, size, and depth worldwide. The SVM algorithm correctly predicted seismic activity with up to 83% success, including 85.7% accuracy for true negative predictions and 80% accuracy for true positive predictions within 48 hours.

Reuveni stated that there is a clear correlation between ionospheric TEC values and geodynamic activity, and this machine-learning approach could accurately predict earthquakes to some degree. Although it is not yet a foolproof method, it is a promising step forward in efforts to better anticipate seismic activity. The findings, published in the Remote Sensing scientific journal, could have significant implications for earthquake prediction and early warning systems. The method uses GPS receivers to estimate TEC data, which is a cost-effective and non-invasive way to monitor geodynamic activity in real time. This could provide valuable time for people to prepare for earthquakes and reduce the damage caused by seismic events.

For more information please contact Ms. Veshala Gajaraj, Trade Officer at




Source: Israel21c