AI assisted forecasting of earthquake aftershock locations

Harvard and Google collaborated to apply deep learning to explain where earthquake aftershocks might occur.

They started with visual representation of the 1992 magnitude 7.3 southern California Landers earthquake where the multi-colored portion represents the initial quake and the red boxes represent aftershock locations.

They applied a neural net to analyze the relationships between static stress changes caused by the mainshocks and aftershock locations. The algorithm was able to identify useful patterns.

The end result was an improved model to forecast aftershock locations and while this system is still imprecise, it’s a motivating step forward. Machine learning-based forecasts may one day help deploy emergency services and inform evacuation plans for areas at risk of an aftershock

Nature – Deep learning of aftershock patterns following large earthquakes

Aftershocks are a response to changes in stress generated by large earthquakes and represent the most common observations of the triggering of earthquakes. The maximum magnitude of aftershocks and their temporal decay are well described by empirical laws (such as Bath’s law and Omori’s law), but explaining and forecasting the spatial distribution of aftershocks is more difficult. Coulomb failure stress change is perhaps the most widely used criterion to explain the spatial distributions of aftershocks but its applicability has been disputed. Here we use a deep-learning approach to identify a static-stress-based criterion that forecasts aftershock locations without prior assumptions about fault orientation. We show that a neural network trained on more than 131,000 mainshock–aftershock pairs can predict the locations of aftershocks in an independent test dataset of more than 30,000 mainshock–aftershock pairs more accurately (area under curve of 0.849) than can classic Coulomb failure stress change (area under curve of 0.583). We find that the learned aftershock pattern is physically interpretable: the maximum change in shear stress, the von Mises yield criterion (a scaled version of the second invariant of the deviatoric stress-change tensor) and the sum of the absolute values of the independent components of the stress-change tensor each explain more than 98% of the variance in the neural-network prediction. This machine-learning-driven insight provides improved forecasts of aftershock locations and identifies physical quantities that may control earthquake triggering during the most active part of the seismic cycle.