Dogs suddenly begin barking. Birds abandon their nests. Livestock become restless, while snakes and rodents appear to leave their hiding places. Then, hours or days later, an earthquake strikes.
Stories like these have circulated for centuries, raising a fascinating question: can animals detect natural disasters before humans realise something is wrong?
There are scientific reasons to think some animals can detect environmental signals that humans miss. Their hearing, sense of smell, sensitivity to vibration and ability to notice subtle changes in their surroundings can give them an advantage in particular circumstances.
But the popular idea that animals can reliably predict earthquakes, tsunamis or volcanic eruptions days in advance is much harder to establish. Researchers have documented unusual animal behaviour before some disasters, yet consistent, repeatable evidence that would support a dependable forecasting system remains limited.
The real explanation may be more interesting than the idea of a mysterious sixth sense. Animals can detect signals we overlook—but recognising a signal is not the same as predicting a catastrophe.
The Earthquake Warning That Animals May Detect First
Earthquakes are among the most frequently discussed examples of animal behaviour preceding a disaster. Reports describe dogs whining, farm animals becoming agitated and wild creatures leaving their usual habitats before significant tremors.
One explanation is straightforward: animals may detect the earliest physical waves from an earthquake before people notice the stronger shaking.
When an earthquake begins, it generates different types of seismic waves. Primary waves, or P-waves, generally travel faster than the more damaging secondary waves, or S-waves. Sensitive animals may detect weak vibrations from the first waves before the stronger shaking arrives.
This could give an animal a brief head start—sometimes seconds—without requiring it to predict an earthquake before it begins.
The US Geological Survey explains this distinction in its research on animals and earthquake prediction. Detecting the first waves of an earthquake that has already started is fundamentally different from predicting where and when an earthquake will occur.
Claims of behaviour changing hours or days beforehand are more difficult to explain. Scientists have considered possible environmental signals, including subtle ground movements, changes in groundwater and unusual sounds. However, there is no established combination of signals that animals consistently detect before damaging earthquakes.
Why Animals May Notice Environmental Changes We Miss
Human senses are well suited to our everyday environment, but they have limits. Other species experience the world differently.
Hearing: Dogs can detect higher-frequency sounds than humans, while some animals can hear or produce very low-frequency sounds. Such abilities may help them notice environmental noises associated with storms, geological activity or distant movement.
Smell: Many animals have highly developed olfactory systems. Changes in air composition, smoke, disturbed soil or other environmental odours may alert them to a developing hazard.
Vibration: Animals that spend time close to the ground may detect vibrations through their feet or bodies. This could help explain reactions to approaching movement or the first seismic waves of an earthquake.
Vision: Some species are sensitive to light levels and visual changes that humans may overlook. Birds, for example, use environmental cues to navigate and adjust their behaviour.
These abilities are well established in biology, but their relevance depends on the particular disaster. A dog detecting smoke from a distant fire is not the same as a dog sensing an earthquake several days before it happens.
Researchers must identify the specific signal, establish that it occurs before the event and demonstrate that the animal’s response is consistent enough to be useful.
What About Animals That Behave Strangely Days Before an Earthquake?
This is where the evidence becomes more controversial.
A review published in Applied Sciences in May 2024 examined unusual animal behaviour as a possible candidate for earthquake prediction. It discussed reports of changes in animal activity before earthquakes and considered possible mechanisms, including environmental and electromagnetic signals.
The review illustrates that the subject remains an area of investigation. It does not establish that animals can reliably forecast earthquakes.
Read the 2024 review on unusual animal behaviour and earthquake prediction.
A central problem is that unusual behaviour is common even when no disaster follows. Dogs bark at unfamiliar sounds, birds respond to predators, livestock react to weather and wild animals change their routines for many reasons.
If an animal behaves strangely before an earthquake, the connection may be meaningful—or coincidental. Without records of its normal behaviour, researchers cannot reliably distinguish the two.
Another problem is retrospective reporting. After a major earthquake, people naturally search their memories for possible warning signs. An unusual event that seemed unimportant at the time may suddenly appear significant.
Scientists therefore need continuous observations collected before, during and after earthquakes, including periods when no earthquake occurs. Otherwise, they risk counting apparent successes while overlooking all the false alarms.
Can Animals Predict Tsunamis and Storms?
The answer depends on the type of disaster.
Some animals can respond to environmental changes associated with approaching storms. Birds may alter their activity as wind, air pressure and weather conditions change. Other species may seek shelter when conditions become dangerous.
These reactions can be useful survival adaptations. However, they do not necessarily mean animals can forecast the precise location, timing or severity of a storm.
Tsunamis present a different challenge. Underwater earthquakes can generate tsunamis, but the wave may travel across the ocean before reaching a coastline. Animals living near shorelines might respond to local environmental changes, yet there is no established evidence that animals can reliably warn communities of an approaching tsunami across the ocean.
Stories about wildlife escaping before the 2004 Indian Ocean tsunami became widely known. Some accounts described animals moving inland or seeking higher ground. Such reports are intriguing, but they do not establish a dependable warning mechanism.
Tsunami warnings rely on seismic monitoring, ocean-bottom pressure sensors, tide gauges and other scientific instruments. These systems provide measurable information about an event and the movement of the sea.
Animal behaviour may offer clues worth studying, but it should not replace official warnings or established emergency procedures.
The Famous Haicheng Earthquake Prediction
One historical event frequently cited in discussions of animal behaviour is the 1975 Haicheng earthquake in China.
Before the earthquake, authorities issued warnings amid reports of unusual animal behaviour and other possible precursors, alongside observations of smaller earthquakes. Many residents evacuated or took precautions before the damaging main shock.
The event is sometimes presented as proof that animals can predict earthquakes. The history is more complicated.
Researchers have debated which observations contributed to the warning and how much of the outcome depended on seismic activity, human decisions and other information available at the time. A successful warning in one case does not establish that the same signals will work before another earthquake.
The USGS notes that earthquakes are frequently preceded by no recognised warning signs. A method that appears promising in one event can fail to provide any useful warning in another.
For earthquake prediction to become dependable, a method must work repeatedly under clearly defined conditions—not merely produce a convincing explanation after a disaster has happened.
How Scientists Could Test the Animal Warning Theory
Researchers need more than witness reports to determine whether animal behaviour provides advance warning of disasters.
One approach is to monitor animals continuously using cameras, microphones, motion sensors and activity trackers. Scientists can compare behaviour with seismic records, weather data and other environmental measurements.
The crucial step is establishing a baseline: how does each animal normally behave when no disaster is approaching?
Researchers can then investigate whether measurable changes occur before particular events and whether those changes are more frequent than would be expected by chance.
Large datasets may help scientists identify patterns that people cannot easily notice. Machine-learning systems could analyse animal vocalisations or movement and compare them with environmental conditions.
However, a computer model finding a correlation would not be enough. Researchers would need to test it on new events, measure false alarms and determine whether it can provide useful warnings in real time.
The method would also have to work across different locations and animal populations. A behaviour pattern observed in one species or region might not apply elsewhere.
Could Animal Behaviour Ever Become an Early-Warning System?
It is possible that studying animal responses could help researchers discover environmental signals associated with certain hazards. Animals may provide clues about sounds, vibrations or chemical changes that deserve further investigation.
But there is a substantial gap between identifying a potentially useful signal and developing a reliable prediction system.
For earthquake forecasting, scientists would need to estimate where an event might occur, its likely magnitude and the relevant time window. A warning that animals appear restless without identifying when or where a quake may strike would have limited practical value.
False alarms could also create serious problems. If a system repeatedly warned people about earthquakes that never occurred, the public might eventually ignore warnings that mattered.
For now, official monitoring systems remain the appropriate source of disaster warnings. People should not evacuate or disregard safety advice solely because a pet or wild animal behaves unusually.
What the Science Really Tells Us
Animals can detect many environmental cues that humans miss. Their sensory abilities and survival instincts may help them respond to the earliest stages of some natural events.
The evidence is strongest when the explanation involves a detectable signal, such as the first seismic waves arriving before stronger shaking. Claims of reliable advance prediction days before an earthquake remain unproven.
That does not mean researchers should dismiss every unusual observation. Systematic monitoring could reveal useful patterns or identify environmental signals that deserve further study.
The real scientific opportunity is to replace folklore with measurable evidence. By recording animal behaviour continuously and comparing it with environmental data, researchers can determine whether particular responses offer meaningful warning—or whether they are simply normal animal behaviour interpreted after the fact.
Animals may sometimes notice danger before we do. But until reliable predictions can be demonstrated, their behaviour should be treated as a subject for scientific investigation, not a substitute for modern disaster-warning systems.
