Earthquake Prediction

The term Earthquake Prediction usually refers to the ability to predict when, where and how big a future earthquake will be. The bottom line is that currently there is no scientifically reliable way to do this. There has always been great interest in trying to predict when and where these events will take place — nobody likes an unpleasant surprise. Moreover, knowing when to take additional precautions would have obvious life- and property-saving implications. 

In a probabilistic sense, a great deal is now known about where earthquakes are likely to occur some time in the future. But there is currently no reliable way to predict the location, timing, and size of any particular earthquake. This is because we do not fully understand the processes that lead a fault to slip suddenly in an earthquake. Because of the chaotic nature of earthquakes there are some seismologists who think there never will be a reliable way to predict them, though scientific investigations are on going.

Each year, on average, about 17 magnitude 7.0 or larger earthquakes happen worldwide. Although we are not able to predict individual earthquakes, the world's largest historical earthquakes do have a spatial pattern that allows for probabilistic "forecasts" of the likely locations and magnitudes of future large earthquakes.

Once again, there is currently no reliable way to predict the day or month when an event will occur in any specific location. However, earthquake prediction is not the same as earthquake early warning, forecasts, and probabilities.

Earthquake early warning (EEW) is a notification that is issued after an earthquake starts to rupture along a fault. One example of EEW is the ShakeAlert Earthquake Early Warning System, which detects significant earthquakes so quickly that alerts can reach many people before shaking arrives. ShakeAlert is not earthquake prediction, but rather a rapid detection system that indicates an earthquake has begun and shaking is imminent. 

Scientists are able to use probabilities to describe the chance of an earthquake of a certain magnitude occurring within a region over a certain span of years. There are several approaches to calculating the probability of earthquake occurrence, the most common of which is based on the average rate of past events. By assuming that a certain event occurs at a constant rate in an arbitrary period of time, scientists can estimate the likelihood of such an event in the next given number of years. 

Probabilistic seismic hazard analysis (PSHA) takes earthquake probabilities a step further. By considering the rates of all possible earthquakes that can occur in an area as well as the level of shaking that each of those earthquakes would create, scientists can determine the likelihood that ground shaking at or above some level will occur within some window of time. PSHA is important for engineers who design buildings to withstand earthquakes, and it forms the basis of seismic building codes and other long-term earthquake planning efforts.

Earthquake forecasts are similar to probabilities but are used for shorter time windows. Forecasting is typically applied to the aftershocks that follow a large earthquake. After most large events, there is a sequence of aftershocks that become less frequent over time. Many aftershock sequences follow systematic patterns, so the probability of an aftershock in a time window following an earthquake can be estimated. 

According to the U.S. Geological Survey (USGS), the sequence of aftershocks usually follows a few general rules:

Long-term earthquake probabilities and medium-term earthquake (or aftershock) forecasts are analogous to how scientists understand long-term climate or medium-term patterns like El Niño. However, unlike day-to-day weather predictions, seismologists can not yet make accurate statements about the time, location, and size of specific earthquakes.

And yet, earthquake prediction is a popular pastime for psychics and pseudo-scientists, who commonly make extravagant claims of past success. Such claims of earthquake prediction are not based on empirical evidence or a coherent understanding of the physical processes that produce earthquakes. For example, earthquakes have nothing to do with clouds, bodily aches and pains, or slugs. Likewise, neither tidal forces nor unusual animal behavior have been useful for predicting earthquakes. 

Predictions claimed as "successes" may rely on a restatement of well-understood long-term geologic earthquake hazards, or may be so broad and vague that they are trivially fulfilled by background seismic activity. Some examples of trivial earthquake predictions include:

When a more specific unscientific prediction is made, scientists cannot state unequivocally that the predicted earthquake will not occur, because an event could possibly occur by chance on the predicted date, though there is no reason to think that the predicted date is more likely than any other day. Scientific earthquake predictions should state where, when, how big, and how probable the predicted event is, and why the prediction is made. The National Earthquake Prediction Evaluation Council reviews such predictions, but no useful method of predicting earthquakes has yet been found.

In 1985, the USGS attempted a formal earthquake prediction experiment. Between 1857 and 1966, the Parkfield segment of the San Andreas Fault produced a series of six similar earthquakes (each around magnitude 6.0) at fairly regular time intervals. Using a set of assumptions about fault mechanics and the rate of stress accumulation, the USGS predicted that a Parkfield earthquake of about magnitude 6.0 would occur before 1993. USGS initiated a set of monitoring experiments to test whether evidence of a process leading to a short term prediction would emerge. Scientists monitored Parkfield for a wide variety of possible precursory effects, but the forecasted earthquake did not materialize until 2004, long after the forecast window expired. "Capturing" the magnitude 6.0 Parkfield earthquake in a dense network of instrumentation was a significant accomplishment, providing data to determine whether precursory effects exist (none were found) and yielding new insights on the mechanics of fault rupture.

Other official attempts at short-term earthquake forecasts have had mixed success. The section of the San Andreas Fault System that broke in the 1989 magnitude 7.1 Loma Prieta earthquake had been identified by the USGS as one of the more likely segments of the San Andreas to rupture. Magnitude 5+ earthquakes that occurred 2 and 15 months before the damaging earthquake were treated as possible foreshocks, and the USGS issued 5-day Public Advisories through the California Office of Emergency Services. But even in areas where foreshocks are fairly common, there is no way of distinguishing a foreshock from an independent earthquake. A “foreshock" is a label that can only be applied after the fact. In the Pacific Northwest, there is no evidence of foreshock activity for most historic earthquakes.

One well-known successful earthquake prediction was for the Haicheng, China earthquake of 1975, when an evacuation warning was issued the day before a magnitude 7.3 earthquake. This evacuation was based on reports from scientists and lay observers about unusual observations. In the preceding months, changes in land elevation and ground water levels, accounts of peculiar animal behavior, and other possible precursors to an earthquake had been reported. A regional increase in seismicity (which later was recognized as foreshocks) then triggered a low-level alert. Subsequently, an increase in foreshock activity triggered the evacuation warning. 

Unfortunately, most earthquakes do not have such obvious precursors. In spite of their success in 1975, there was no warning of the 1976 Tangshan earthquake, magnitude 7.6, which caused an estimated 250,000 fatalities and 164,000 injuries. A team of scientists from the U.S. visited laboratories in China in 1976 to investigate the Haicheng prediction. Their report concluded that the 1975 Haicheng prediction was based mainly on the pronounced foreshock sequence; other aspects of the described methodology were more difficult to assess.

It may never be possible to reliably predict the exact time when a damaging earthquake will occur, because when enough strain has built up, a fault may become inherently unstable, and any small earthquake that starts on the fault may or may not continue rupturing and turn into a large earthquake. While it may eventually be possible to accurately diagnose the strain state of faults, the precise timing of large events may continue to elude us. In the Pacific Northwest, earthquake hazards are well known and future earthquake damage can be greatly reduced by identifying and improving our most vulnerable and dangerous structures.

For more information on how to prepare for an earthquake, visit here: Preparedness.