Currently, seismologists can predict when aftershocks will strike and how strong they will be but there is less certainty about where. Until now that is.
A group of researchers from Harvard University and Google have trained a type of artificial intelligence known as a ‘deep learning’ programme with data about tens of thousands of earthquakes and aftershocks to see if predictions could be improved upon. And they were successful.
“The previous baseline for aftershock forecasting has a precision of around three percent across the testing data set, said Phoebe DeVries, co-author of the study published in the journal Nature on Thursday. “Our neural network approach has a precision of around six percent.”
Testing the System
More than 131,000 mainshock-aftershock pairs were used to train the network and then scientists selected unrelated 30,000 pairs for a test. While this AI approach was found to be more reliable at detecting the locations of aftershocks than a widely-used model scientists caution that it’s too early to put the system into operation.
That’s because they focused on only one set of changes caused by earthquakes that can affect where aftershocks occur. Nonetheless, the research is a very promising start to improving on our current ability to make predictions.
To read more about the research and how this application of AI could one day save countless lives, click here.