American scientists from Ohio State University have developed a neural network to detect natural hydrogen reserves underground. The research was published in the official gazette Web site American Geophysical Union (AGU).
The location of hydrogen deposits can be determined by the characteristic egg-shaped depressions in the ground. These depressions are visible when observed from the air, but are often hidden by vegetation or located in agricultural lands.
A scientific team trained artificial intelligence (AI) to detect semicircular depressions from satellite images. The team also used remote sensing data and information about geomorphic and spectral patterns to improve the accuracy of hydrogen deposit detection.
Scientists noted that natural hydrogen reserves can be found in different geological formations and places different from oil and gas deposits. The process of discovering hydrogen reserves can be accelerated by using machine learning algorithms.
According to the authors of the discovery, natural hydrogen reserves could be integrated into the global energy system in just a few years.
Russian scientists before to create How to train AI 20 times.