A closer look
Our planet is constantly humming with tiny vibrations, and buried in that noise might be early whispers of future quakes. Seismologists are feeding years of ground motion data into machine learning systems to hunt for subtle patterns that precede major earthquakes, patterns too faint for humans to spot reliably. Some models have shown promising skill at forecasting elevated risk over days or weeks in certain regions, not down to the minute and street, but better than blind guesswork. Combined with satellite measurements of how land slowly warps along fault lines, this could one day give cities more time to brace, shut down gas lines, and move people away from the most fragile structures. We are nowhere near perfect prediction, but the idea that an algorithm could listen to the Earths nervous system is no longer fantasy.
