Researchers have identified six massive, never-before-seen structures located deep within the Earth, near the boundary between the mantle and the liquid outer core. These formations, described as 'thermochemical piles,' are distinct from the surrounding mantle material and were detected using seismic waves from earthquakes.
These structures lie approximately 1,800 miles (2,900 km) beneath the Earth's surface. Scientists study these deep layers indirectly by analyzing how seismic waves, generated by powerful earthquakes, travel through the planet. The newly discovered formations subtly influence the paths of these waves, making them detectable.
The research, published in JGR Solid Earth, suggests these structures could be remnants of material from Earth's early history, potentially including fragments of the continental crust or even parts of Theia, a protoplanet believed to have collided with Earth billions of years ago to form the Moon. Under the extreme pressure and high temperatures at the mantle boundary, such materials would transform, partially melt, or undergo mineralogical changes.
The boundary between the mantle and the outer core is a region of intense activity, marked by a significant temperature difference of about 1,000°C (1,800°F). This thermal gradient drives convection currents in the mantle, which in turn influence volcanic activity on the surface. The density differences at this boundary also cause seismic waves to slow down, providing geologists with a way to 'image' this deep layer.
To identify these structures, scientists focused on a specific type of seismic wave known as PKP precursors. These are faint waves that arrive just before stronger seismic waves and are scattered by variations, or heterogeneities, at the mantle boundary. Previously, manually identifying these weak signals was a labor-intensive and inefficient process.
The research team utilized an artificial intelligence (AI) model to analyze over two million seismic recordings from 5,000 earthquakes. The AI was trained on human-identified PKP precursors and subsequently detected 174,929 high-quality signals, significantly more than all previous studies combined. This AI-driven analysis provided an unprecedented view of the mantle boundary, revealing large, continuous belts of previously unmapped structures. Six specific areas were identified as likely hosts for significant heterogeneities, marking priority targets for future exploration.
While the exact composition and formation process of these thermochemical piles remain uncertain, their distinct nature from the surrounding mantle is clear. The growing capabilities of AI in analyzing vast seismic datasets are expected to provide a clearer picture of Earth's deep interior in the future. This refined understanding of the lowermost mantle's structure offers crucial constraints for comprehending the geodynamic processes shaping our planet.