Unravelling the prediction of strong earthquakes

Due to their chaotic nature and self-similarity, predicting strong earthquakes in the hours before they strike is almost impossible. The findings, published in the journal Moscow University Geology Bulletin, highlight that the failure to develop accurate short-term forecasts for strong earthquakes is due to the fundamental characteristics of these events rather than research challenges, such as a lack of observational data or adequate equipment.

For decades, researchers have gone to great lengths in their attempts to develop reliable and accurate methods for predicting strong earthquakes just hours before they happen. This work has involved searching for distinct and measurable signals, such as unusual seismic, geophysical, and deformational changes, among others. While hundreds of these potential precursors have been identified, they have largely failed to pinpoint the exact location, time, and magnitude of impending earthquakes.

“The study of precursors gave hope for the creation of methods for forecasting earthquakes,” says study co-author, Vladimir Zakharov, at the Moscow State University. But he adds that there have been more erroneous predictions than successful ones, with one analysis of 548 strong earthquakes occurring between 2007 and 2010 finding just 20 “acceptable” predictions.

For short-term forecasts to be accurate, they must be able to identify clear earthquake precursors that can be separated from signals generated by background events. But because earthquakes take place in fractal systems where each component is similar to the whole, identifying precursors and disentangling them from background states is inherently difficult as they are all products of the same process.

Unlike forecasts that predict earthquakes over months and years, short-term predictions must include rigid and precise parameters to provide meaningful information. While straightforward in theory, setting and applying these parameters to a real-world system is notoriously difficult. This is because seismic processes are fundamentally nonlinear systems where even the smallest inaccuracy in parameter settings can lead to massive errors in the forecast. “For real systems, it is impossible to set these parameters precisely,” says Zakharov.

Taking these challenges into account in their literature review, Zakharov and his colleagues conclude that the lack of success in developing useful short-term forecasts for strong earthquakes is due to their self-similarity, nonlinearity, and chaotic nature, rather than gaps in observational data, equipment limitations, or funding problems.

“We are convinced that the problem has already come to its solution: a reliable and accurate forecast of short-term strong earthquakes is impossible,” says Zakharov. But it’s important to continue studying the seismic process and focus on improving forecasts that predict strong earthquakes over the medium- and long-term, he says.