Taka'aki Taira

Remote triggering of fault-strength changes on the San Andreas fault at Parkfield

Fault strength is a fundamental property of seismogenic zones, and its temporal changes can increase or decrease the likelihood of failure and the ultimate triggering of seismic events. Although changes in fault strength have been suggested to explain various phenomena, such as the remote triggering of seismicity, there has been no means of actually monitoring this important property in situ. Here we argue that ∼20 years of observation (1987–2008) of the Parkfield area at the San Andreas fault have revealed a means of monitoring fault strength. We have identified two occasions where long-term changes in fault strength have been most probably induced remotely by large seismic events, namely the 2004 magnitude (M) 9.1 Sumatra–Andaman earthquake and the earlier 1992 M = 7.3 Landers earthquake. In both cases, the change possessed two manifestations: temporal variations in the properties of seismic scatterers—probably reflecting the stress-induced migration of fluids—and systematic temporal variations in the characteristics of repeating-earthquake sequences that are most consistent with changes in fault strength. In the case of the 1992 Landers earthquake, a period of reduced strength probably triggered the 1993 Parkfield aseismic transient as well as the accompanying cluster of four M > 4 earthquakes at Parkfield. The fault-strength changes produced by the distant 2004 Sumatra–Andaman earthquake are especially important, as they suggest that the very largest earthquakes may have a global influence on the strength of the Earth’s fault systems. As such a perturbation would bring many fault zones closer to failure, it should lead to temporal clustering of global seismicity. This hypothesis seems to be supported by the unusually high number of M ≥ 8 earthquakes occurring in the few years following the 2004 Sumatra–Andaman earthquake.

Fluid-faulting interactions: Fracture-mesh and fault-valve behavior in the February 2014 Mammoth Mountain, California, earthquake swarm

Faulting and fluid transport in the subsurface are highly coupled processes, which may manifest seismically as earthquake swarms. A swarm in February 2014 beneath densely monitored Mammoth Mountain, California, provides an opportunity to witness these interactions in high resolution. Toward this goal, we employ massive waveform-correlation-based event detection and relative relocation, which quadruples the swarm catalog to more than 6000 earthquakes and produces high-precision locations even for very small events. The swarms main seismic zone forms a distributed fracture mesh, with individual faults activated in short earthquake bursts. The largest event of the sequence, M 3.1, apparently acted as a fault valve and was followed by a distinct wave of earthquakes propagating ~1 km westward from the updip edge of rupture, 1–2 h later. Late in the swarm, multiple small, shallower subsidiary faults activated with pronounced hypocenter migration, suggesting that a broader fluid pressure pulse propagated through the subsurface.

Detecting seismogenic stress evolution and constraining fault zone rheology in the San Andreas Fault following the 2004 Parkfield earthquake

[1] We investigate temporal changes in seismic scatterer properties at seismogenic depth attributed to the 2004 M 6 Parkfield earthquake, making use of the San Andreas Fault Observatory at Depth repeating-earthquake target sequences, as well as nearby similar-earthquake aftershock clusters. We use a two-step process: (1) observing temporal variations in the decorrelation index, D(t), reflecting changes in the scattered wavefield of repeating-earthquake sequences and (2) estimating the spatial distribution of timedependent scatterers by using a larger-aperture source array. We focus on three scatterers exhibiting clear time dependence, using pairs of earthquakes that span or follow the 2004 Parkfield earthquake. They are found to be located on the fault at the northernmost extent of coseismic rupture, beneath Middle Mountain, with a depth range of 11 to 17 km. The shallowest and most prominent scatterer is located near a region of increased Coulomb stress, as well as significant postseismic slip following the 2004 Parkfield earthquake, and a large M = 5 aftershock. The other two deeper ones are also in regions of increased Coulomb stress. We show that D(t) 1/2 is expected to be proportional to the level of stress in the fault zone, and then we constrain the form of fault zone rheology by comparing the time dependence of D(t) 1/2 with geodetic or seismic measures of strain rate, assuming a power law rheology between stress and strain rate characterized by exponent n. Such a comparison yields n ranging from 1.6 through 3.3, a value that is more consistent with ductile behavior, rather than frictional sliding, at the base of the seismogenic zone.

Interseismic coupling and refined earthquake potential on the Hayward‐Calaveras fault zone

Interseismic strain accumulation and fault creep is usually estimated from GPS and alignment arrays data, which provide precise but spatially sparse measurements. Here we use interferometric synthetic aperture radar to resolve the interseismic deformation associated with the Hayward and Calaveras Faults (HF and CF) in the East San Francisco Bay Area. The large 1992–2011 SAR data set permits evaluation of short- and long-wavelength deformation larger than 2 mm/yr without alignment of the velocity field to a GPS-based model. Our time series approach in which the interferogram selection is based on the spatial coherence enables deformation mapping in vegetated areas and leads to refined estimates of along-fault surface creep rates. Creep rates vary from 0 ± 2 mm/yr on the northern CF to 14 ± 2 mm/yr on the central CF south of the HF surface junction. We estimate the long-term slip rates by inverting the long-wavelength deformation and the distribution of shallow slip due to creep by inverting the remaining velocity field. This distribution of slip reveals the locations of locked and slowly creeping patches with potential for a M6.8 ± 0.3 on the HF near San Leandro, a M6.6 ± 0.2 on the northern CF near Dublin, a M6.5 ± 0.1 on the HF south of Fremont, and a M6.2 ± 0.2 on the central CF near Morgan Hill. With cascading multisegment ruptures the HF rupturing from Berkeley to the CF junction could produce a M6.9 ± 0.1, the northern CF a M6.6 ± 0.1, the central CF a M6.9 ± 0.2 from the junction to Gilroy, and a joint rupture of the HF and central CF could produce a M7.1 ± 0.1.

Potential for larger earthquakes in the East San Francisco Bay Area due to the direct connection between the Hayward and Calaveras Faults

The Hayward and Calaveras Faults, two strike-slip faults of the San Andreas System located in the East San Francisco Bay Area, are commonly considered independent structures for seismic hazard assessment. We use Interferometric Synthetic Aperture RADAR to show that surface creep on the Hayward Fault continues 15 km farther south than previously known, revealing new potential for rupture and damage south of Fremont. The extended trace of the Hayward Fault, also illuminated by shallow repeating micro-earthquakes, documents a surface connection with the Calaveras Fault. At depths greater than 3–5 km, repeating micro-earthquakes located 10 km north of the surface connection highlight the 3-D wedge geometry of the junction. Our new model of the Hayward and Calaveras Faults argues that they should be treated as a single system with potential for earthquake ruptures generating events with magnitudes greater than 7, posing a higher seismic hazard to the East San Francisco Bay Area than previously considered.

Ambient noise‐based monitoring of seismic velocity changes associated with the 2014 Mw 6.0 South Napa earthquake

We perform an ambient noise-based monitoring to explore temporal variations of crustal seismic velocities before, during, and after the 24 August 2014 Mw 6.0 South Napa earthquake. A velocity drop of about 0.08% is observed immediately after the South Napa earthquake. Spatial variability of the velocity reduction is most correlated with the pattern of the peak ground velocity of the South Napa mainshock, which suggests that fracture damage in rocks induced by the dynamic strain is likely responsible for the coseismic velocity change. About 50% of the velocity reduction is recovered at the first 50 days following the South Napa mainshock. This postseismic velocity recovery may suggest a healing process of damaged rocks.

Variability of fault slip behavior along the San Andreas Fault in the San Juan Bautista Region

An improved understanding of the time history of fault slip at depth is an essential step toward understanding the underlying mechanics of the faulting process. Using a waveform cross-correlation approach, we document spatially and temporally varying fault slip along the northernmost creeping section of the San Andreas Fault near San Juan Bautista (SJB), California, by systematically examining spatiotemporal behaviors of characteristically repeating earthquakes (CREs). The spatial distribution of pre-1998 SJB earthquake (1984–1998) fault slip rate inferred from the CREs reveals a ~15 km long low creep or partially locked section located near the 1998 Mw 5.1 SJB earthquake rupture. A finite-fault slip inversion reveals that the rupture of the 1998 SJB earthquake is characterized by the failure of a compact ~4 km2 asperity with a maximum slip of about 90 cm and corresponding peak stress drop of up to 50 MPa, whereas the mean stress drop is about 15 MPa. Following the 1998 earthquake, the CRE activity was significantly increased in a 5–10 km deep zone extending 2–7 km northwest of the main shock, which indicates triggering of substantial aseismic slip. The postseismic slip inferred from the CRE activity primarily propagated to the northwest and released a maximum slip of 9 cm. In this 5–10 km depth range, the estimated postseismic moment release is 8.6 × 1016 N m, which is equivalent to Mw 5.22. The aseismic slip distribution following the 1998 earthquake is not consistent with coseismic stress-driven afterslip but represents a triggered, long-lasting slow earthquake.

On the Systematic Long-Period Noise Reduction on Ocean Floor Broadband Seismic Sensors Collocated with Differential Pressure Gauges

Vertical-component broadband seismic data collected on the ocean floor at regional distances from the shore and water depths on the order of 1-2 km are polluted by noise due to ocean infragravity (IG) waves in a period range (20-200 s) that is critical for regional structure studies and the determination of moment tensors for regional earthquakes. Using broadband seismic and differential pressure gauge data collected at the Monterey Ocean Bottom Broadband (MOBB) observatory, which has been operational since 2002, we identify the optimal length of time window to obtain the transfer function between the vertical seismic and pressure records. The maximum reduction in IG-wave-induced noise is obtained by using a one-day stack of transfer function inferred from data segment collected in the last 24 hours prior to data in which the noise removal method is applied. We show that a one-year stack of transfer function effectively reduces the noise level in the period band 50-100 s by 20 dB and by up to 15 dB at shorter periods. This resultant 20 dB noise reduction is comparable to that obtained from the one-day stack of transfer function. Our result suggests that the removal of noise induced by IG waves on MOBB vertical-component data can be done routinely without continuously calibrating the transfer function, also in near-real time, and is therefore also useful for routine regional moment tensor de- termination. This noise removal method has been implemented in the moment tensor determination system of the Northern California Seismic System.

Imaging of crustal heterogeneous structures using a slowness‐weighted back‐projection with effects of scattering modes: 2. Application to the Nagamachi‐Rifu fault, Japan, area

[1] This is the second paper in a two-part series on a newly developed imaging approach for small-scale heterogeneities (<1 km) in the crust with effects of scattering modes. In the present paper, we estimated a detail three-dimensional spatial distribution of small-scale heterogeneities around the Nagamachi-Rifu fault, northeastern Japan, in a frequency range of 2-16 Hz, using the imaging approach presented in the first paper. We used seismograms recorded by dense three-component seismic arrays that were deployed in this region for 15 explosion sources. As one of our important results, there are concentrations for only P-S but not P-P scatterers near the surface trace of the fault. P-S scatterers in a frequency range of 8-16 Hz are localized near the surface trace of the fault, implying the possibility of strong heterogeneities with a size of 0.08 km there. In the Shirasawa caldera region, the characteristics of seismic scatterers seem to convert from large P-S to large P-P relative scattering coefficients with its transition depth range of 5-8 km. This feature implies that the materials composed of seismic scatterers may show a systematic variation with depth. Finally, the strength of scattering coefficients is rather weak in the coseismic area of the 15 September 1998 earthquake with a magnitude of 5.2, the largest recent event in this area. This result suggests that this coseismic area is rather homogeneous and can hold local stress larger than in the surrounding portions of the fault system.

Characteristics of Small-Scale Heterogeneities in the Hidaka, Japan, Region Estimated by Coda Envelope Level

We estimated the spatial distribution of small-scale heterogeneities as anomalous amplification of coda level in the Hidaka, Japan, region. We analyzed 2768 seismograms in a frequency range of 1-32 Hz for 24 earthquakes recorded at 62 stations. First, we estimated the site effect of each station, using the coda nor- malization method with regional earthquake data of large epicentral distances. All the data processing was done after the correction of this site effect. Next, we deter- mined coda amplitude factor (CAF), that is, the amplitude ratio of coda waves on each source-station pair relative to the averaged coda amplitude over stations, for earthquakes inside the region. We found systematic spatial variations of CAF, im- plying the existence of localized heterogeneities. At frequencies lower than 2 Hz, the CAF is relatively large in the west of the Hidaka Mountains, implying the possibility of strong heterogeneities with a scale of 0.6-1.3 km there. In a high-frequency range (� 16 Hz), CAF values are large on paths crossing the Hidaka Mountains. From the corresponding lapse time of coda (65 sec), there may be a zone of concentrated heterogeneities beneath the Hidaka Mountains at the depth of 100-120 km.