Global investigation of foreshock acceleration prior to large earthquakes

Predicting large earthquakes remains one of the most important yet unresolved challenges in geoscience1,2,3. Accelerating aseismic slip (preslip) and associated accelerating foreshock activity have long been discussed as possible precursory phenomena preceding large earthquakes4–11. Previous studies have reported that many large earthquakes worldwide were preceded by accelerating foreshock sequences, suggesting a potential for earthquake forecasting12,13. However, how frequently such acceleration occurs and whether it is unique to the periods preceding large earthquakes remain unclear. Here we statistically analyse seismic activity preceding large earthquakes worldwide, showing that significant foreshock acceleration is very rare and not unique to large earthquakes. We extend the epidemic-type aftershock-sequence model with an acceleration term based on laboratory and simulation results14,15 to quantitatively evaluate the number and characteristic timescale of foreshocks, revealing that such acceleration occurs before only 3–4% of large earthquakes. Comparisons with randomly selected seismic activity and synthetic earthquake catalogs demonstrate that the observed acceleration cannot be explained by aftershock cascades, implying that it results from aseismic processes, yet it is not regarded as unique to large earthquakes. These findings reveal the limitations of the preslip-based forecasting framework supported by theory, observation, and experiments, highlighting the need for a new paradigm in earthquake forecasting.