Earthquake forecasting tools powered by artificial intelligence can predict the risk of aftershocks within seconds of an initial tremor, according to a new study from the University of Edinburgh.
Researchers from the University of Edinburgh, the British Geological Survey and the University of Padua developed the AI-driven tools by training machine learning models on earthquake data from regions prone to seismic activity, including California, New Zealand, Italy, Japan and Greece.
The team analysed the AI models' ability to forecast the number of aftershocks within 24 hours following earthquakes of magnitude 4 or higher. They compared performance with the widely used Epidemic-Type Aftershock Sequence (ETAS) model, which is employed in Italy, New Zealand and the US.
While both model types showed similar forecasting accuracy, the ETAS model took up to several hours or days on a mid-range computer, whereas the AI models produced results in seconds.
PhD student Foteini Dervisi, of the University of Edinburgh's School of GeoSciences and the British Geological Survey, who led the study, said: 'This study shows that machine learning models can produce aftershock forecasts within seconds, showing comparable quality to that of ETAS forecasts. Their speed and low computational cost offer major benefits for operational use.'
The fast forecasts could assist authorities with decision-making about public safety measures and resource allocation in disaster-hit areas. The research was published in the journal Earth, Planets and Space and supported by the European Union's Horizon 2020 programme.
