Dynamic rupture analysis: Inversion for the source process of the 1990 Izu‐Oshima, Japan, earthquake ( M = 6.5)

A waveform inversion has been applied to strong motion data using a dynamic shear crack model. We studied the 1990 Izu‐Oshima earthquake ( M JMA = 6.5), which has vertical strike‐slip faulting with unilateral rupture propagation. The inversion has two steps, a waveform inversion and a crack inversion, that are applied iteratively. A waveform inversion is used to determine the distribution of rupture starting times and slip dislocations using the slip functions calculated by the initial crack model, or by previous crack inversion. A crack inversion is used to calculate dynamic crack propagation that explains the results of the above inversion. In this step, we use the estimated rupture times as a locking fracture criterion; the maximum shear stress attained before a fault segment breaks gives a lower bound estimate of the peak shear strength at each fault segment. Then the dynamic stress drop distribution is estimated from the slip distribution obtained from waveform inversion assuming a dynamic crack model. From the results, we determine the rise time distribution and the distribution of a dimensionless stress ratio S defined as (strength excess)/(stress drop). Our analysis gives the following picture of the rupture process of the 1990 Izu‐Oshima earthquake: (1) An asperity‐type faulting having large slip and high stress drop was detected in the region around the initiation point of rupture. (2) South of the asperity zone, barrier‐type faulting characterized by incoherent propagating rupture, small slip, long rise time, and high strength excess was detected. This zone corresponds to the intersection of the fault with the 1978 earthquake ( M JMA = 7.0).