Can animals sense earthquakes or storms before they happen?

Seconds before shaking: early detection

The U.S. Geological Survey explains that some animals can feel the faster, smaller P waves before the larger S waves arrive and before people notice the shaking. The earthquake has already started at its source. This is a possible explanation for a reaction seconds before stronger shaking, not a prediction days ahead.

The agency distinguishes these responses from claims of much earlier warnings. Its assessment says that consistent, reliable animal behavior before seismic events, with an established explanatory mechanism, remains elusive. That does not prove every reported incident is invented or that detecting earlier signals is impossible.

What about animals becoming restless hours earlier?

A 2020 study continuously monitored cows, dogs and sheep around an Italian earthquake sequence. Researchers reported patterns of increased activity preceding earthquakes while the animals were inside a stable, but not on pasture. They proposed further monitoring as a possible route toward short-term forecasting.

This is a research claim worth testing, rather than an established alarm system. Findings from one farm and a particular earthquake sequence do not by themselves demonstrate reliable prediction at new locations. The setting-dependent result also matters: “farm animals sometimes showed a pattern under these conditions” is narrower than “animals know when an earthquake is coming.”

Storms: an intriguing bird observation

A paper published online in 2014 and in a 2015 issue of Current Biology tracked five male golden-winged warblers in Tennessee. They left breeding territories more than a day before a severe storm arrived and returned afterward.

The authors proposed distant, low-frequency sound as a likely cue, while retaining other possibilities. They did not experimentally establish what the birds sensed. This was one storm and a small tracked sample, so it does not show that all birds, pets or other animals forecast every storm.

Notice what “before” means here: before the storm reached the birds’ location. The weather system was already developing elsewhere. Detecting a distant event is different from predicting its formation from nothing.

What would turn a promising pattern into a forecast?

A useful test would define the animal behavior and warning window in advance, then record continuously. It must count false alarms—unusual behavior without a disaster—and missed events, when a disaster happens without the behavior.

Imagine a dog barking before one earthquake. That observation alone leaves unanswered how often the dog barks on ordinary days. A prediction method needs both kinds of records, plus testing on future events that were not used to choose its rules.

The evidence supports curiosity about animal sensing. It does not justify treating a pet’s behavior as a dependable substitute for official warnings. For the wider question of interpreting other creatures’ experiences, see animal minds.