Keiko Kuge

Data‐dependent non‐double‐couple components of shallow earthquake source mechanisms: Effects of waveform inversion instability

Seismic moment tensors are routinely determined for global earthquakes using waveform inversions, and the resulting double-couple and non-double-couple characteristics are extensively used in tectonic analyses. For shallow events, several moment tensor components are intrinsically poorly resolved, especially for surface wave inversions, which can lead to artificial non-double-couple components or can obscure actual non-double-couple components in the source representations. The relatively greater instability of surface wave inversions can cause differences between inversions based on only body waves and joint inversions of body and surface waves. The Harvard CMT catalog exhibits systematic variations in non-double-couple components with seismic moment, which can be attributed to use of only body waves in inversions for small events and combined body and surface waves for large events. Systematic behavior of non-double-couple components with regional strain environment is significantly less dependent on seismic moment when solutions influenced by less stable surface waves are omitted.

Significance of non-double couple components of deep and intermediate-depth earthquakes: implications from moment tensor inversions of long-period seismic waves

Abstract Analysis of long-period seismic waves suggests that non-double couple components in source moment tensors of intermediate-depth and deep earthquakes are significant and appear to respond to the state of predominant strain release within slabs. The strain regime is partially or fully induced by the sources themselves or by slab structures near the sources. We observe consistency in the non-double couple components from three different seismic wave inversions. Among 21 earthquakes studied, 17 events have the same sign in the three non-double couple components, and the signs show a correlation with the state of strain release within the slabs. The effects of unmodeled propagation structure and instability of the inversion procedure are unlikely to be responsible for the consistent non-double couple components because the suites of seismic waves traverse different paths and the various inversion schemes have the different resolvabilities of the moment tensors.

Systematic non-double-couple components of earthquake mechanisms: The role of fault zone irregularity

Several recent studies have revealed statistical correlations between earthquake mechanism type and associated non-double-couple components in catalogs of seismic moment tensors. This systematic behavior may result either from biases in the solutions due to Earth structure effects in different tectonic regimes, or from source radiation effects. For certain large non-double-couple events, detailed analyses have shown that multiple subevents with different fault orientations produce the non-double-couple radiation. We generalize this idea and simulate non-double-couple components (NDCC) resulting from subfaults with variable geometry, showing that the statistical behavior of the NDCC in moment tensor catalogs can be generally accounted for by such source complexity. Assuming that an earthquake fault consists of many subfaults with random fluctuations about some mean geometry, the total moment tensor for failure of the system under a regional stress state can be represented by the sum of moment tensors of the subfaults. The sign of the NDCC in the composite moment tensor directly reflects the stress state applied to the fault system, and the NDCC amplitude is expected to be systematic over a wide range of cumulative seismic moment (Mo ) given the basic fractal nature of fault systems. These simulation results are consistent with the global behavior of the NDCC reported in the Harvard centroid moment tensor catalog for 1977–1991. Parameters such as the applied stress state, parameterization and randomness of the subfault geometry, and seismic moment distribution among the subfaults affect the predicted NDCC amplitude. Change in the parameters as a function of subfault seismic moment controls the shape of the NDCC - Mo relationship. Regional variations of the NDCC - Mo relationship may thus reflect regional variations in the fault zone parameters, raising the possibility that the regional NDCC behavior may be used to infer stress state and subfault distribution in various source regions.

Source Modeling Using Strong-Motion Waveforms: Toward Automated Determination of Earthquake Fault Planes and Moment-Release Distributions

The geometry, size, and slip distribution of earthquake faults control the generation of strong ground motions around earthquake source regions. Fast determination, which is inevitably made by automated methods, of such source char- acteristics is important for estimating characteristics of strong ground motions. This study suggests that determination of the source parameters can be automated using the waveform data from modern strong-motion instruments. The method consists of three separate linear inversions, which provide data on moment tensor solutions, fault planes and their length, and distributions of moment release on the finite faults, respectively. Tests for five inland earthquakes (M 6) beneath Japan show that the fault planes and distributions of moment release, which were obtained in the present method without human inspections, are in good agreement with the aftershock dis- tributions and the results from other studies. The successful application of this method is produced by the broadband responses of modern strong-motion instru- ments that are deployed at a regular spacing. The method works best for shallow strike-slip earthquakes. Success with shallow dip-slip earthquakes is supported by a numerical test and the result for a reverse-fault earthquake. The second inversion, which models waveforms using aligned point sources to determine fault planes and their length, can contribute to speeding up the analyses of source processes. The present results suggest that using realtime data from strong-motion instruments will enable fast determination of source processes and consequently lead to the rapid estimation of ground motions.

Determination of the isotropic component of the 1994 Bolivia Deep Earthquake

We determine the isotropic component of the moment tensor of the 1994 Bolivia deep earthquake using various seismic waves, including body waves, surface waves, and normal mode data, in the period band between 20 s and 1000 s. We carefully investigate whether the isotropic component can be obtained independently from the other components by checking the correlation matrices. We show that it is possible to obtain a precise estimate of the isotropic component by using normal mode data in the period band between 550 and 1000 s. We find that the Bolivia earthquake did not have a significant isotropic component in this period band.

Rupture characteristics of the 2005 Tarapaca, northern Chile, intermediate-depth earthquake: Evidence for heterogeneous fluid distribution across the subducting oceanic plate?

[1] We examined the rupture of the 2005 Tarapaca, northern Chile, earthquake at about 110 km depth with respect to both kinematic and dynamic characteristics by using regional and teleseismic waveforms. The earthquake has a downdip tensional focal mechanism. The subhorizontal rupture is characterized by two patches of large slip and high stress drop which are aligned nearly in the east-west direction, being perpendicular to the direction of the Chile Trench. Rupture initiated in the eastern patch and then propagated to the western patch. Between the two patches, there exists a region of nonpositive stress drop and high strength excess, which can cause subshear rupture to propagate from the eastern to the western patches but radiates little seismic waves. Seismic radiation energy from this earthquake tends to be low, which is consistent with the nonpositive stress drop and high strength excess between the two patches. While the physical mechanism of intermediate-depth earthquakes is still controversial, current leading hypotheses are associated with dehydration within subducting plates. The rupture characteristics of the Tarapaca earthquake can be related to heterogeneous fluid distribution due to the dehydration. The spatial separation and dominant stress of the two large-slip patches agree with the characteristics of the previously reported double seismic zone beneath Chile. The two patches may be the manifestation of the double seismic zone where dehydration reactions can release fluid. Using a numerical simulation of 3-D dynamic rupture, we have shown that weakening due to fluid can account for the rupture characteristics of the Tarapaca earthquake.

Non‐double‐couple moment tensor of the March 25, 1998, Antarctic Earthquake: Composite rupture of strike‐slip and normal faults

The 1998 Antarctic intraplate earthquake (Mw 8.1) has a significant non-double-couple component in the moment tensor solution as determined from long-period surface waves. In the teleseismic P and SH wave-forms, there are two distinctive wave packets that can be explained by multiple strike-slip subevents. We propose a composite rupture model comprising two different episodes: two predominant clusters of en echelon strike-slip segments (∼16 × 1020 Nm) and normal faulting with a long duration of approximately 100 s (∼4 × 1020 Nm). This model agrees with observations of both teleseismic body waves and long-period surface waves, including the large non-double-couple component. The normal faulting with as long a duration as the main strike-slip rupture might be secondarily induced by the en echelon segments of strike-slip faulting.

Spontaneous Dynamic Rupture Propagation beyond Fault Discontinuities: Effect of Thermal Pressurization

Abstract A range of several fault segments often sequentially ruptures during an earthquake. We investigated the effects of thermal pressurization (TP) on dynamic rupture propagation beyond fault discontinuities by simulating spontaneous rupture propagation on two vertical strike‐slip fault segments. We revealed that a rupture can jump wider stepovers owing to TP, and that TP on a primary (nucleating) fault enables a rupture to jump at deep portions. In previous numerical studies on dry fault systems, it was found that a rupture sometimes fails to propagate to an unconnected fault, which is observed in the case of real earthquakes, and a rupture that successfully propagates is usually triggered near the surface of the Earth, unlike rupture evolution images obtained by seismic waveform modeling. TP can explain the inconsistencies between the previous numerical simulations and the observations, without depending on the heterogeneity of the initial stress and/or friction. Under depth‐dependent stress, we showed that TP enables a rupture to jump much wider stepovers at deep portions. If TP is in effect on faults, hydraulic diffusivity along with fault geometry can strongly control the characteristics of rupture propagation at fault discontinuities.

Lateral segmentation within the subducting lithosphere: three-dimensional structure beneath the North Island, New Zealand

Abstract Three-dimensional (3-D) velocity structure beneath the North Island, New Zealand is determined by the inversion of travel-time data from local earthquakes and the result is compared with the 3-D attenuation structure previously obtained. In the upper mantle, a high-velocity and high-Q zone represents the subducting Pacific plate. The slab is dipping steeply to the northeast of the island, possibly continuing to the Tonga-Kermadec trench. In the southwestern part, it dips gently and seems to continue to the Alpine Fault in the South Island. Two discontinuities in the velocity structures are found beneath the North Island. Between these boundaries is a high velocity slab bending both the ends deeper. The discontinuous structure, along with the attenuation structure and the other seismological evidence, indicates the segmentation within the subducting lithosphere. The slab is likely to be torn at the segmentation boundary.

Suppression of slip and rupture velocity increased by thermal pressurization: Effect of dilatancy

We investigated the effect of dilatancy on dynamic rupture propagation on a fault where thermal pressurization (TP) is in effect, taking into account permeability varying with porosity; the study is based on three-dimensional (3-D) numerical simulations of spontaneous ruptures obeying a slip-weakening friction law and Coulomb failure criterion. The effects of dilatancy on dynamic ruptures interacting with TP have been often investigated in one- or two-dimensional numerical simulations. The sole 3-D numerical simulation gave attention only to the behavior at a single point on a fault. Moreover, with the sole exception based on a single-degree-freedom spring-slider model, the previous simulations including dilatancy and TP have not considered changes in hydraulic diffusivity. However, the hydraulic diffusivity, which strongly affects TP, can vary as a power of porosity. In this study, we apply a power law relationship between permeability and porosity. We consider both reversible and irreversible changes in porosity, assuming that the irreversible change is proportional to the slip rate and dilatancy coefficient ɛ. Our numerical simulations suggest that the effects of dilatancy can suppress slip and rupture velocity increased by TP. The results reveal that the amount of slip on the fault decreases with increasing ɛ or exponent of the power law, and the rupture velocity is predominantly suppressed by ɛ. This was observed regardless of whether the applied stresses were high or low. The deficit of the final slip in relation to ɛ can be smaller as the fault size is larger.