MIT Scientists Develop AI Model to Optimize Underground CO2 Storage
“Our goal is to create a tool that allows us to map all underground structures - so that we can know exactly, for example, about the location of geothermal sources in Iceland. Our first developments showed that deep learning models can significantly bring the scientific community closer to this goal and give the opportunity to determine the location options for underground CO2 storage", said Laurent Demanne, professor of applied mathematics at MIT.
The AI model system, presented by a team of scientists, uses seismic activity data to accurately estimate low-frequency waves, which allows mapping underground structures with high accuracy. The revolutionary neural network underlying the solution monitors patterns in frequency data to establish the location of geothermal pockets. Thanks to this method, it's expected that in the future it will be possible to establish optimal locations for underground CO2 storage tanks.