Can machine learning be used to predict earthquakes?
A new study uses machine learning to successfully predict earthquakes, but so far in lab models only.
Dr. Qiunya Liu, from the Departments of Physics and Earth Sciences at the University of Toronto, says the Cambridge University study using machine learning to help predict earthquakes is a very promising technique for use outside the lab in the future. But for now, predicting real earthquakes is still very challenging for a number of reasons. One is that high frequencies from small tremors very deep in the earth are difficult to monitor. Also there a number of earthquake rupture causes to consider, including volcanoes, landslides and the sliping of faults. As well it is difficult to monitor an entire fault, which can be hundreds of kilometres in length.