Abstract

Understanding the causes of strong ground motion is crucial for mitigating earthquake damage and devastation. This necessitates unraveling the physics of the seismic source, the primary driver of ground shaking variations. However, limited data from real earthquakes compels researchers to develop computational tools for deeper exploration. Crucially, simulations lack empirical constraints due to constant parameters or arbitrary domain limitations, leading to potentially inaccurate representations of real-world earthquakes and erroneous interpretations. This work aims to identify the rupture process characteristics that closely resemble natural earthquakes using empirical tests. The chosen test, the Somerville (or asperity) criteria, stipulates that 20-30% of the ruptured area releases over 50% of the seismic moment. Dynamic simulations revealed two rupture types: self-arrested, which halt before reaching the domain boundaries, and runaway, which traverse the entire domain. Only self-arrested ruptures met the Somerville criteria. Moreover, their parameters exhibited stronger correlation during evolution, suggesting runaway ruptures might not be physically realistic due to seemingly independent parameter evolution. Limited by the difficulty of generating self-arrested ruptures in dynamic simulations, simpler kinematic simulations were also explored. Both approaches revealed that the spatial distribution of residual energy (available energy minus dissipation) acts as the key constraint governing rupture parameters during arrest.