ABSTRACT
Many large earthquakes have been preceded by an increase in the number of intermediate-sized events. In the seismological literature this phenomenon has been termed “accelerating moment release” (AMR), and it is currently being studied as a promising approach to earthquake forecasting. The AMR signal is observed over such a large area surrounding the impending mainshock that the intermediate events are generally viewed, not as triggers for the large event, but rather as indicators that the state of the fault network is such that a large event is possible. The size of the active area that optimizes the AMR signal has been shown to scale with the magnitude of the mainshock.We show here that the time over which the AMR signal develops also scales, but in this case with both the magnitude of the main shock and with the slip rate on the dominant fault in the network. One proposed explanation for the AMR phenomenon, known as the “inter-
mittent criticality model”, is based on an analogy to critical point phenomena in statistical physics. In this model, a large earthquake is possible only when the fault network is in a critical state where patches of the crust that are at the failure level are connected at all length scales. An earthquake nucleated in this critical state can grow large by jumping geometrical barriers that would have arrested the rupture in a less connected non-critical stress field. This large event, and its aftershocks, lowers the stress over the network moving it away from the critical state. Tectonic loading and small events moves the system back toward the critical state and the possibility of the next network-wide large event. The fault networks on which these spatial and temporal seismicity patterns
develop have been shown to have a spatial fractal structure. Any discrete hierarchy in this structure can be shown to produce log-periodic fluctuation in the AMRsignal. Mechanically heterogeneous systems in the laboratory show similar patterns of acoustic emissions prior to failure.
