AI Maps 6 Hidden Core-Mantle Zones From 175,000 Seismic Signals

3 articles · Updated · Yahoo! Voices · Sep 1

Researchers at the Chinese Academy of Sciences used deep learning to detect about 174,929 faint PKP precursor signals and identify six previously undocumented scattering zones near Earth’s core-mantle boundary.
More than 2 million waveforms from roughly 5,000 magnitude-6-plus earthquakes over nearly 35 years fed the system, which found over 10 times as many signals as all previous catalogs combined.
The newly mapped zones—B1 through B6—sit beneath high-latitude Eurasia, Central Asia, the South Atlantic and other undersampled regions, suggesting broader belt-like heterogeneities rather than isolated anomalies.
At roughly 2,900 km depth, those signals point to sharp temperature or compositional differences where mantle rock meets the liquid outer core, though the data does not establish whether they came from slabs, melting or any Moon-forming impact debris.
The study, published in late August in Journal of Geophysical Research: Solid Earth, highlights how domain-specific AI can reopen old archives and refine models of Earth’s deep interior.

Sources

Yahoo! Voices5d ago

AI Discovers Six Hidden Zones Near Earth's Core from Seismic Data

indoneo.com5d ago

Deep learning found six hidden zones inside Earth's core

Instagram5d ago

Something strange is hiding deep inside Earth… and scientists found it. 🌍, Nearly 3,000 km below us, scientists detected six previously unknown structures near the boundary between Earth's mantle and ...

5 Sources

Are these newly mapped deep-Earth zones remnants of ancient tectonic plates, or something entirely unknown to science?

Could the hidden structures discovered by AI deep within our planet rewrite what we know about Earth's geological future?

What exactly is hiding 2,900 kilometers beneath us that eluded human detection for over three decades?

AI Maps Six Hidden Core-Mantle Boundary Zones: How Deep Learning Is Transforming Our Understanding of Earth's Interior

Overview

In August 2026, scientists used a deep-learning system to analyze 35 years of seismic data, uncovering nearly 175,000 faint signals from deep within the Earth. This breakthrough revealed six new, highly active zones at the core-mantle boundary, showing that these anomalies form continuous belts around the core. The findings connect the sinking of tectonic plates to changes in heat flow at the core-mantle boundary, which shapes the movement of liquid metal in the outer core and directly influences Earth's magnetic field. This AI-driven approach is transforming our understanding of how deep Earth processes drive surface volcanism and magnetic field behavior.

...