3-D finite-difference simulation of seismic fault zone waves—Application to the fault zone structure of the Mozumi-Sukenobu fault, central Japan—

Abstract

Fault zone waves have the potential to be a powerful tool to reveal the fine structure of a fault zone down to the seismogenic depth. Seismic fault zone waves include head waves, trapped waves and direct body waves propagating in the fault zone. 3-D numerical simulation is necessary to interpret the waveforms in the presence of low-velocity zones with relatively complex fault structure. We computed finite difference (FD) synthetic seismograms to fit the seismograms of explosions, which contain frequencies up to 25 Hz, recorded by a linear seismometer array across the Mozumi-Sukenobu fault, central Japan. We find fault zone head waves, direct P waves propagating within the low-velocity zone and wave trains following the direct P waves associated with the fault for both observed and synthetic waveforms. Thus, modelling of fault zone waves is expected to determine details of complex fault zone structure.

Deep structure of the Mozumi-Sukenobu fault, central Japan, estimated from the subsurface array observation of fault zone trapped waves

“…Li et al (2000) carried out forward modelling of the trapped waves using a 3‐D finite difference code developed by Graves (1996) to infer the depth variation of fault zone structures. In this study, we used a parallelized 3‐D FDM code developed by Mamada et al (2002) to reduce the calculation time. We modelled the seismogram for the event 000408.163500, the nearest earthquake to the array.…”

“…The cumulative displacement of the Mozumi‐Sukenobu fault estimated to be ranging from 0.2 to 0.5 km with a right‐lateral slip (The Research Group for Active Faults in Japan, 1991). Through analysing trapped waves generated by explosives, the crush zones A and B, and the zone sandwitched between the two crush zones can be recognized as a low velocity zone (Mamada et al 2002). Mamada et al (2002) estimated the width of the fault to be 200 m, and determined the P ‐wave velocity of the shallow part of the fault as 3.6 km s −1 , approximately 75 per cent of the host rock velocity.…”

“…Through analysing trapped waves generated by explosives, the crush zones A and B, and the zone sandwitched between the two crush zones can be recognized as a low velocity zone (Mamada et al 2002). Mamada et al (2002) estimated the width of the fault to be 200 m, and determined the P-wave velocity of the shallow part of the fault as 3.6 km s −1 , approximately 75 per cent of the host rock velocity. The Atotsugawa fault system is located along the highly deformed Niigata-Kobe tectonic zone (Hashimoto & Jackson 1993;Sagiya et al 2000).…”

Finite Difference Numerical Simulation of Trapped Waves in the Kunlun Fault Zone

“…In the recent decade, studying the fine structure of fault zone through the trapped waves has attracted more attention of seismologists. They have applied this method to exploring many fault zones, such as the rupture zone of the Landers [1∼3] , the Hector Mine [4,5] , the San Jacinto near Anza [6,7] , the San Andreas near Parkfield [8,9] , USA, and the Karadere-Duzce branch of the North Anatolian Fault, western Turkey [10] , Nojima Fault [11] and the Mozumi-Sukenobu branch of the Atotsugawa fault, central Japan [12,13] . Seismologists performed numerical simulations of fault-zone trapped waves based on observation and analysis.…”

A numerical approach for modelling fault-zone trapped waves

“…Full waveform solvers allow investigations into the properties of FZTWs as influenced by different velocity models or source types and locations (Li & Vidale 1996;Igel et al 1997Igel et al , 2002Jahnke et al 2006). The computationally intensive 3-D full waveform solvers have also been used for the estimation of fault zone parameters through forward modelling (Li et al 2000(Li et al , 2003(Li et al , 2014Mamada et al 2002Mamada et al , 2004Mizuno et al 2004;Li & Malin 2008). The high computational time (hours) means that full waveform solvers are generally impractical for estimating fault zone parameters through inversion which requires many evaluations of the forward model.…”