Earthquake prediction

Earthquake prediction is a specialized branch of geophysics. It focuses on seismology, which is the study of earthquakes. The goal is to specify the exact time, location, and magnitude of future earthquakes. This is a very difficult task. Scientists try to determine the parameters for the next strong earthquake in a specific region. However, prediction is different from earthquake forecasting. Forecasting is a probabilistic assessment of general hazards. It looks at the frequency and size of damaging quakes over years or decades. There is also earthquake warning. These systems detect a quake as it starts. They provide real-time warnings of a few seconds to nearby areas.

Scientists use different methods to find signs of an earthquake. These signs are called precursors. An earthquake precursor is an unusual phenomenon that might signal an impending quake. Researchers look for two main types of methods. Precursor methods aim for short-term prediction. They look for specific events that happen just before a quake. Trend methods are used for long-term forecasting. These look for geophysical patterns over ten to 100 years. One famous idea was the dilatancy-diffusion hypothesis. In the 1970s, scientists thought highly of this theory. It suggested that stressed rock changes its volume, or dilatancy. This change might affect seismic velocity or electrical resistivity. Researchers even used these changes to attempt predictions in New York and California.

Many people have wondered if animals can sense earthquakes. There are thousands of reports of unusual animal behavior. Some reports date back to ancient times. For example, Claudius Aelianus wrote about the destruction of Helike in 373 BC. In modern science, researchers have studied over 130 species. They found that animals might sense P waves. These are pressure waves that travel twice as fast as damaging S waves. These waves arrive a few seconds before the main shaking. However, a 2018 review found insufficient evidence for long-term animal prediction. Some animals might be magnetoreceptive. This means they can detect electromagnetic waves. These waves might reach the surface before a quake. They could also cause air ionization or water changes that animals detect.

Evaluating whether a prediction works is a complex statistical process. A prediction is significant only if it succeeds better than random chance. Scientists use statistical hypothesis testing to check this. They compare the prediction to a "null hypothesis." The null hypothesis assumes the earthquake would have happened anyway. Earthquakes do not always occur randomly. They often show clustering in space and time. In southern California, about 6% of magnitude 3.0 earthquakes are followed by a larger one within 5 days and 10 km. In central Italy, 9.5% of magnitude 3.0 earthquakes are followed by a larger event within 48 hours and 30 km. While these statistics show patterns, they are not reliable for exact prediction. They often result in many false alarms.

History shows that the scientific community has changed its mind over time. In the 1970s, most scientists were optimistic. They believed a practical method for prediction would be found soon. By the 1990s, many failures led to doubt. Many scientists began to question if prediction was even possible. One famous claim involved the 1975 Haicheng earthquake. While some called it a success, later studies said there was no valid short-term prediction. Most scientists are now pessimistic. Some even believe that earthquake prediction is inherently impossible. The search for reliable precursors has had a checkered history with no convincing successes.

Predicting earthquakes creates a difficult dilemma for society.

If scientists warn the public, they must face the cost of false alarms. False alarms cause economic disruption and civil panic. In 1999, China introduced tough regulations to stop unofficial warnings. This followed three years of inaccurate warnings that caused panic. On the other hand, failing to warn people can be dangerous. In China, scientists were purged for ignoring predictions before the 1976 Tangshan earthquake. In Italy, seven people were convicted of manslaughter after the 2009 L'Aquila earthquake. They were not blamed for failing to predict the quake. They were blamed for giving people false assurance that there was no danger.

Some experts argue that we should focus our resources elsewhere. Stathis Stiros conducted a study in Greece in 1997. He looked at the cost-benefit ratio of prediction research. He found that earthquakes in Greece kill fewer than ten people per year on average. Most deaths happen in large buildings with structural issues. He argued that money would be better spent on highway safety. In Greece, highway deaths average more than 2,300 people per year. Stiros suggested that upgrading unsafe buildings is more cost-effective than trying to predict quakes. This highlights how society must decide the value of different safety measures.