Rocks release subtle geochemical warning signals before breaking. These signals, coming from naturally occurring nuclides (e.g., radon, helium, argon, and thoron), have often been reported before earthquakes, volcanic eruptions, landslides, and rock and ice avalanches. However, despite their high sensitivity to deformation, their detectability, as well as myriad promising observations over half a century, nuclide signals are still far from being applied to geohazard prediction or widely used for monitoring. Here, we first develop a decomposition and interpretation method for nuclide signals. By analyzing nuclide signal time series observed from a month-long laboratory rock failure experiment and year-long slope deformation in a field setting, we identify a universal paradigm unit of nuclide signal evolution. We find that this paradigm unit is characterized by two core characteristics: a transient pulse and equilibrium fluctuation which are intrinsically correlated to rupture area and crack aperture, respectively. Through analytical derivation and pore-scale simulations, we establish the constitutive equations that link these characteristic nuclide signals to key rupture structural parameters. Rooted in these constitutive relations, we further develop a diagnostic theory of rock rupture via nuclide signals. We apply the model to track rock failures at the laboratory and field scale. The proposed nuclide signal decomposition and rupturing model enable the unification of discrete signal units emitted by individual microrupturing events, with the integrated signal evolution observed during macroscopic failure. This integration may serve as a foundation for both the mesoscopic assessment of rock damage and the early warning of geohazards induced by rock ruptures.
Zhou et al. (2026) studied this question.