Google's New AI Can Predict Earthquakes

To this day, natural phenomena are still very unpredictable. The Earth has continually been producing unforeseen dramatic activity we call ‘natural disasters’. It's astonishing that we've come such a long way in terms of technological advancements but still fail to predict tsunamis and earthquakes which cause massive destruction. The timing and size of these aftershocks are fairly well understood but forecasting the location accuratelyis a real challenge. In pursuitof predicting the location of earthquakes and their aftershocks more accurately, Harvard has teamed up with machine learning experts at Google to unravel the mystery behind earthquakes in order to mitigate their effects.

A 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

A 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

A database with information on 118 major earthquakes from around the world was fed into Google’s machine learning software which puts itthrough a neural network to analyse and identify any useful patterns and relationships between different factors, such as the relationship between static stress changes caused by the mainshock’s aftershock location. In the end it has resulted in a model that can finally forecast the location of these aftershocks and could be used to make evacuation plans more effective for areas at risk, saving countless lives. while still imprecise, it’s a promising step forward.

GoogleGavin Aren