Stanford University researchers have published a study showing their AI model can predict earthquakes with magnitude 5.0 or greater up to 72 hours in advance with 78% accuracy.
Revolutionary Approach
The model analyzes seismic micro-tremors, ground deformation data from satellites, and electromagnetic signals that precede major earthquakes. It was trained on 40 years of global seismic data.
Validation
The system correctly predicted 7 of the last 9 significant earthquakes in the Pacific Ring of Fire during a blind test period. USGS is now evaluating the system for potential integration.
- 72-hour advance warning capability
- 78% accuracy for M5.0+ events
- Could save thousands of lives annually
