S23C-01
Quantifying Dynamic Earthquake Triggering in the Near and Far-Field
Remote earthquake triggering by dynamic strain from seismic waves is a regularly observed feature of earthquake interactions. However, the effectiveness of dynamic strains at triggering earthquakes has not been well quantified. In particular, it is unclear whether dynamic triggering can account for the majority of local aftershock production. Here we quantify the rate at which dynamic strains trigger earthquakes at long distances, using the ANSS and JMA earthquake catalogs, by comparing pre-and post-trigger interevent time statistics. The intensity of remote triggering scales continuously as a function of peak dynamic strain, estimated from the seismic wave amplitude. We then compare these rates to aftershock triggering rates very near a mainshock, and find that dynamic triggering can account for the majority of local aftershocks. Dynamic triggering also appears to be ubiquitous, rather than confined to geothermal and magmatic provinces. Some regions however, like California, are more prone to triggering than others, like Japan. This result points the way to a new methodology for earthquake forecasting based on the amplitude of the observed seismic waves.
S23C-02
Evidence of Earthquake Triggering by the Solid Earth Tides
Clear evidence for earthquake triggering by the earth tides has remained elusive for more than a century. Using the largest global earthquake catalog available (the NEIC catalog with 442,412 events), we observe a clear correlation (with approximately 99% confidence) between the phase of the Earth tide and the timing of seismic events: earthquakes occur slightly more often at the time of ground uplift by the Earth tide. We observe that this phase distribution anomaly is larger for smaller and shallower earthquakes. Although earthquakes in regions with dominantly normal and strike-slip faulting seem to exhibit more tidal trigerring than regions dominated by thrust faulting, there is no statistically significant evidence for a focal mechanism- dependence on earthquake triggering. Finally, we show that the tidal triggering observed in the present work is primarily caused by the solid Earth tide, rather than by loading from the ocean or atmospheric tides. Although an additional impact due to loading from ocean tides is possible and probable in oceanic regions, we cannot detect it here because the earthquake database is not sufficiently complete and homogeneous (more small magnitude earthquakes in oceanic areas are needed). The detection of tidally-triggered events suggests that earthquake initiation has a damped sensitivity to stress change, and that small tidal stresses are sufficient to create a small excess of earthquakes in phase with the tidal maximum, particularly small and shallow earthquakes.
S23C-03
Coulomb Stress Interactions Among Earthquakes in the Gorda Deformation Zone, the San Andreas, Mendocino Fracture Zone, and Cascadia Megathrust
During the last 30 years the most prolific source of large earthquakes in the western U.S. has been the Gorda deformation zone, a region of diffuse shear off the coast of northernmost California and southern Oregon. Fifteen M≥6 earthquakes have occurred there since 1980, including M≥7 shocks. The abundance of large earthquakes on different fault systems provides fertile ground for the study of earthquake interaction. We find five strong examples of triggering attributable to static stress transfer. A left- lateral Mw=7.3 earthquake in 1980 promoted slip on the right-lateral Mendocino Fracture Zone (MFZ) in an area where aftershocks were abundant, and inhibited slip on a section of the MFZ where aftershocks were absent. The 1980 earthquake appears to have inhibited slip on the Gorda zone faults except within an off- fault stress trigger lobe where three M≥6 shocks subsequently struck. The Mw=6.9 earthquake at Cape Mendocino in 1992 promoted failure on the rupture planes of Mw=6.5 and Mw=6.6 aftershocks. M≥7 earthquakes in 1994 and 2005 are found to have promoted the rupture of subsequent Mw=6.6 earthquakes. We also calculate that the 1906 San Andreas earthquake promoted slip on the MFZ and left-lateral slip on Gorda deformation zone faults, consistent with the M~5.8 shock offshore Cape Mendocino in 1909 and earthquakes in 1922, 1941 and 1954. However, the 1991 Mw=7.0 earthquake should have inhibited the rupture of the 2005 Mw=7.2 earthquake. There are also examples of earthquakes closely spaced in time that occurred more than two source dimensions apart, the approximate limit of static stress transfer for earthquakes with 30-bar stress-drops. A 16 Aug 1991 Mw=6.3 shock was followed 21 hr later by the 17 Aug event 200 km away; the 17 Aug 1991 Mw=6.1 shock was followed 3 hr later by a Mw=7.0 shock 200 km away; and a 25 Nov 1954 M=6.1 shock was followed 26 days later by a M=6.5 shock 120 km away. Together these counter-examples suggest that dynamic triggering is also important. We also infer, on the basis of continuous small earthquakes and three M>6 earthquakes since 1980, that the MFZ is capable of both large earthquakes and creep along the same sections, behavior typical of oceanic and some continental transforms such as the Hayward and Calaveras faults.
S23C-04
Self-Similar Slip Pulses During Earthquake Nucleation
For rate- and state-dependent faults that are close to velocity-neutral and loaded slowly, earthquake nucleation zones obeying either of the popular state evolution laws expand as they accelerate. Under the "slip" law this expansion takes the form of a unidirectional slip pulse. Motivated by the near self-similarity of numerical simulations, we obtain an approximate self-similar solution for slip pulses growing into regions initially sliding at steady state. In this solution the length scale behind the pulse front over which significant sliding occurs continually decreases, being inversely proportional to the logarithm of the maximum slip speed Vmax, while the total slip remains constant. This slip is approximately Dc/(1-a/b), where Dc is the characteristic slip scale for state evolution and a and b are the coefficients that determine the sensitivity of the frictional strength to changes in slip rate and state. The pulse has a "distance to instability" as well as a "time to instability", with the remaining propagation distance being approximately 2(1-a/b)-2[ln(Vmaxθbg/Dc)]-1, where θbg is the background state into which the pulse propagates. This solution provides a reasonable estimate of the total slip for pulses growing into regions not previously sliding at steady state.
S23C-05
Deep heterogeneous structure and earthquake generating properties in the Yamasaki fault zone, Southwest Japan
We have been estimating deep heterogeneous structure and earthquake generating properties in the Yamasaki fault zone, which is a left-lateral strike-slip active fault with a total length of ~80 km in southwest Japan. We deployed dense seismic observation network, composed of 32 stations with average spacing of ~10 km in and around the Yamasaki fault zone. We will estimate detailed fault-zone structure such as fault dip and shape, segmentation, and possible location of asperities and rupture initiation point, as well as generating properties of earthquakes in the fault zone, through analyses of accurate hypocenter distribution, focal mechanism, 3-D velocity tomography, coda wave inversion, and other waveform analyses. We also deployed a linear seismic array across the fault, composed of 20 stations with ~20 m spacing, in order to delineate the fault-zone structure in more detail using the seismic waves trapped inside it. We will also estimate detailed resistivity structure of the fault zone by AMT (audio-frequency magnetotelluric) and MT surveys. As preliminary analyses, we estimated distribution of accurate hypocenters, b values, velocity perturbation, and scattering coefficients around the fault zone using the routine network data. For the scattering analysis, for example, we analyzed 411 wave traces from 31 events, which occurred in 2003, recorded at 27 stations, and we estimated a distribution of scattering coefficients along the Yamasaki fault zone. Microseismicity is high and scattering coefficient is relatively larger in the upper crust along the entire fault zone. The distribution of strong scatterers suggests that the Ohara and Hijima faults, which are the segments in the northwestern part of the Yamasaki fault zone, have almost vertical extension from surface to the depth of ~20 km. These results will be improved by adding temporary dense network data. We used seismic network data operated by universities, NIED, and JMA. This study has been carried out as a part of the project "Study on evaluation of earthquake source faults based on surveys of inland active faults" by Japan Nuclear Energy Safety Organization (JNES).
S23C-06
Estimation of gas flow through Taiwan Chelungpu fault during Fluid Injection Test
Fluid Injection Test (FIT) was performed on from November 2006 to March 2007 to estimate permeability and
to understand hydrological and chemical properties between two boreholes penetrated through Chelungpu
fault in Taiwan. Murakami et al. (AGU, 2007) reported results of water quality, flow rate, gas and water
pressure monitoring at the observation hole during the FIT. As a result, they found the two different signals
suggesting arrivals of the injected water. The values of Oxidation Reduction Potential (ORP) and Dissolved
Oxygen (DO) as well as the gas pressure of oxygen, nitrogen and argon suddenly increased 3 days after the
start of the FIT. This suggested that the injected water, which well mixed with the air, arrived at the
observation hole 39 m distant from the injection hole. Then, the flow rate at the observation hole suddenly
increased 10 days after the start of the FIT. In this study, we estimated the water and gas flows through the
permeable zone during the FIT in order to estimate the permeability from the both data suggesting water
arrivals. Used model was based on the 2D water diffusion model and the 2D gas diffusion model in the water,
and we applied them to estimations of gas flow in an unsteady water flow during FIT. Here we used gas
diffusivity in water of 2 x 10-9 m2/s, and permeable zone of 1 m for the estimations. As a result, the
permeability of 1 x 10-16 m2 explained the increase of water flow rate 10 days after the start of the
FIT. In contrast, the estimated DO was fitted with that in the experience when the permeability was 1 x 10-
15 m2. These lines suggest that the permeability between two boreholes through Chelungpu fault was
between 1 x 10-16 m2 and 10-15 m2.
REFERENCE
Murakami, M., H. Tanaka, T.-C. Kuo, C.-W. Tsao, S. Giletycz, W.-M. Chen, C.-Y. Wang, C.-S. Chen, C.-S.
Chen, T.-Y. Yang and K.-F. Ma (2007), Hydrological and chemical monitoring during Fluid Injection Test in
Taiwan Chelungpu-fault Drilling Project, 2007 AGU, Fall Meet., Abstract S21B-0567.
S23C-07
San Jacinto Fault Zone Structures from Earthquake Relocation and Waveform Modeling.
We analyzed seismic waveform data of the 1999 San Jacinto Fault Zone (SJFZ) seismic recording experiment. This experiment consisted of three linear arrays, across three southern branches of the SJF respectively. Each array was about 350 m long and had 12 three-component short-period (L22) stations. We manually picked P and S arrival times of all earthquakes with S-P times less than 3 s. This allowed us to find about 500 recorded events that were close to the arrays. We used waveform cross-correlation to accurately measure their P and S differential arrival times and then used the double-difference relocation method to determine their locations. We found 40 events whose waveform record sections showed characteristic arrival time and waveform variations across the fault zone, similar to those that we identified in our previous Landers Fault Zone study. We applied the generalized ray theory (GRT) waveform modeling method that we developed before to the 40 earthquakes to determine fault zone structures of the San Jacinto fault zone. Preliminary results of fault zone width and velocity drops of each branch will be presented.
S23C-08
Crustal Anisotropy Measured Near the Calico Fault
We present preliminary results from a study of crustal anisotropy near the Calico Fault located in the Eastern California Shear zone. Shear wave splitting is used to approximate the in situ stress field and/or shear fabric near the fault and examine whether proximity to the main slip plane affects the observed anisotropy. 65 stations, a combination of L22 and 40T sensors, were deployed in a grid spanning 4 km across and 1.5 km along the Calico fault between June and November 2006 to explore seismic properties of a fault that has not experienced a major earthquake in hundreds to thousands of years. In this study we examine shallow crustal anisotropy determined using earthquakes that occur within the shear wave window, with an angle of incidence less than 45 degrees, during the experiment period. Splitting parameters are determined using an automated cross-correlation method that determines the fast direction and delay time for each station-event pair. Results show some spatial variation in shallow crustal anisotropy as the N30°W trending Calico fault is crossed from west to east. Fast directions rotate from parallel to the regional stress field, which is oriented approximately N30°E, towards north. Results also show significant scatter in fast directions and delay times within a region 1.5 km wide along the fault which supports a damage zone along the fault as found in previous studies.