Researchers at Harvard University has teamed up with Google to develop an Artificial Intelligence (AI) that forecasts the precise location of the earthquake aftershocks. The Earth has been plagued by natural disasters like Tsunamis, Earthquakes, Volcanoes leading to tremendous destruction in the last decade. Coming to earthquakes, they usually take place in a sequence of events; the main shock followed by a series of aftershocks. Even though the effects of aftershocks are minimal when compared to main shocks, nevertheless they too occupy a prominent position. It is these aftershocks that hamper the recovery effects.
Although the timing and size of aftershocks can be predicted using empirical laws, it is not possible to predict the exact location of the events. For this purpose, the Harvard researches have teamed up with Google AI experts to see whether artificial intelligence can help them forecast the aftershock locations. And the results are encouraging! They fed the deep learning algorithms with earthquake information in 118 locations from across the world.
After that, they applied a neural net to analyse the relationship between static stress changes caused by mainshocks and aftershocks. The system was able to generate useful patterns. Although the system seems imprecise but is a major step forward. It may come in handy to deploy emergency services in areas at risk of aftershock and mitigate their harmful effects. Efforts are also in place to predict the magnitude of an earthquake using AI.
Also, the research helped to identify the physical quantities that are critical in earthquake generation. Going forward this would help to find new physical theories to better understand the natural phenomenon. The research is led by Phoebe DeVries, a post-doctoral student at Harward, Brendan Meade, professor of Earth and planetary sciences at Harvard and others. The entire research study is published in the journal Nature.