Here’s a summary of the article:
Exploring Earth’s Inner Layers with Machine Learning
Current technological limitations make it impossible to drill deep into Earth’s layers. Instead, geologists analyze seismic activity caused by the dynamics of these layers to study the planet’s interior. A recent study from the Chinese Academy of Sciences utilized data from over two million earthquakes between 1990 and 2024, employing a machine learning algorithm to uncover structures located thousands of miles below the surface.
The researchers identified nearly 175,000 “PKP precursors,” seismic signals detectable before the main earthquake, enabling the mapping of Earth’s innermost structures. These precursors are believed to arise from small variations just above the meeting point of the solid mantle and liquid iron core, around 1,800 miles below the surface.
Traditionally a labor-intensive process, the new algorithm expedited the identification of PKP precursors, revealing a much more comprehensive global map of small-scale heterogeneities in the Earth’s interior. This map includes six previously undocumented areas with significant structural variations, suggesting that these structures may extend over large regions rather than isolated patches.
The findings hint that these structures could correlate with subduction zones, where material is dragged into the depths and altered by extreme conditions. This data may inform predictions about future seismic activity and deepen our understanding of Earth’s formation and tectonic plate development.
For further reading, refer to the full study in the Journal of Geophysical Research: Solid Earth and related articles on the topic.
Feel free to let me know if you need more details or specific information!
