It is currently impossible to say when and where an earthquake might strike, but having a rough idea of where it might strike could save countless lives; even a few hours’ warning can be extremely valuable.New study evaluates years of seismic data from around the world to pick out statistical patterns in forecasts and aftershocks.
Some large earthquakes are preceded by a series of smaller tremors commonly called foreshocks; however, they have so far been identified as such only after a larger event. For years, scientists have hoped to record these precursors when they occur, potentially helping them predict earthquakes.
Some researchers argue that there are small but noticeable differences in the characteristics of tremors that could telegraph warnings of a larger earthquake, such as the pattern of energy release and the distribution of earthquake magnitudes in an earthquake sequence.
“Just about any earthquake at a certain level will have several earthquakes associated with it,” said Thorne Lay, a seismologist at the University of California, Santa Cruz and co-author of the study. “The magic would be if there was some unique process or distinct character of events that ended before the greatest event.”
Global earthquake analysis searches for an elusive prediction signal
However, other studies argue that these predictors are not real, or if they are, they are not useful enough to matter. An earthquake is defined as a foreshock only by its association with a larger earthquake, which is called a mainshock. Predictions are all that precedes the main shock.
For predictions to be useful, seismologists would need a reliable physical or statistical signal to distinguish the precursor from the mainshock. To do this, they need better data about when predictions occur and what they look like. To that end, Lay and his colleagues applied three different algorithms that automatically detect clusters of earthquakes in the US Geological Survey’s global earthquake catalog to pick out predictions from other earthquakes. The researchers followed up on their earlier work with eight regional earthquake catalogs.
Observing tremors as they could help seismologists predict large earthquakes
•The researchers found that anywhere from 15% to 43% of the earthquakes they analyzed had recognizable predictions.
•Lay said, although the three algorithms produced different numbers, their results tended to match each other.
•The researchers also found indications that some faults may be more prone to shaking than others, useful information for potential earthquake predictions.
•Earthquakes caused by tectonic plates pushing together had more likely predictions than earthquakes caused by plates sliding past each other, and earthquakes in the western Pacific had more aftershocks than those in the eastern Pacific.
•However, these differences are on the order of a few percentage points, which may be too small for earthquake prediction purposes.
•”This is really a question of statistical significance versus practical significance,” Trugman said.
•Indeed, Lay notes that he is less optimistic now that forecasts sometimes prove useful for earthquake prediction than when he first began researching them.
Building for a breakthrough?
This kind of research is nevertheless useful, says Martijn van den Ende, a seismologist at the Université Côte d’Azur. More observations could help scientists discern patterns in forecasts, if they exist, and disentangling regional differences is also important for understanding forecasts in different environments.
“I think there’s a lot of potential, but because we rarely see these very bright, beautiful sequences, it’s very difficult to look at the data in a very focused way,” he said.
Trugman also noted that Lay and his co-authors limited their data set in a way that made it difficult to find patterns. They did not include earthquake swarms or sequences that included two main shocks very close to each other. Those events could provide additional insights into the predictions, he said.
It’s also possible that many of the predictions are simply too small for scientists to reliably pick up. A 2019 paper co-authored by Trugman argues that including earthquakes up to magnitude 2.0 or 3.0 greatly expands the prediction presence, although others disagree.
It is possible that with more and better data, predictions of the earthquake signals that seismologists have long sought could be made. At this point, however, scientists can do no better than guess whether a given earthquake represents a true prediction or is merely an isolated incident.