G22A-01 INVITED
Discovery of slow earthquakes by dense high-sensitivity broadband borehole seismic observation network in Japan, NIED Hi-net
National Research Institute for Earth Science and Disaster Prevention (NIED) has constructed the nation-wide high-sensitivity seismograph network (Hi-net) with 700 borehole stations, which uniformly covers the Japanese Islands with a spacing of 20–30 km. The borehole sensor is composed of three-component high-sensitivity velocity seismometer, three-component strong motion accelerometer and horizontal component high-sensitivity accelerometer which is available to measure the ground tilt and long-period seismic waves. The sensor vessel is placed at the bottom of the borehole deeper than 100m. There are four advantages of Hi-net; high sensitivity, high signal-to-noise ratio, broadband property of sensors and high density of stations. As a result, detection capability for micro earthquakes has been dramatically improved and some new geophysical phenomena have been discovered. Most remarkable discovery by Hi-net is wide variety of slow earthquakes; non-volcanic deep low- frequency tremors (LFT) [Obara, 2002], short-term slow slip events (SSE) [Obara et al., 2004], and very low- frequency (VLF) earthquakes [Ito et al., 2007]. These slow earthquakes occur simultaneously with a certain recurrence interval at the transition zone on the deeper plate interface along the strike of the subducting Philippine Sea plate, southwest Japan. The tremor is characterized by randomly wave trains lasting for hours to weeks with a predominant frequency of around 2 Hz. The tremor activity is clustered spatially and temporally within the narrow belt-like zone. Peak of tremor activity recurs with a time interval of six months accompanying to the short-term SSE lasting for a several days. VLF earthquake has a predominant period of 20s and occurs coincident with peak of tremors. During the active stage, the source of LFT migrates with a propagation velocity of around 10km/day along the strike of the plate geometry at the downdip side of the locked seismogenic zone. The space-time property of SSE and VLF earthquakes is also coincident with the migration of tremor. On the other hand, another kind of VLF earthquake with a predominant period of 10s has been detected at the updip side of the seismogenic zone [Obara and Ito, 2005]. This shallow VLF earthquake mainly occurs inside the accretionary prism on the landward side of the Nankai trough. We have succeeded to detect some slow earthquakes at updip and downdip sides of the seismogenic zone on the subducting plate interface. These slow earthquakes reflect the stress relaxation process around the plate boundary. Therefore, it is very important to watch these slow earthquakes with high accuracy by using the Hi-net in order to monitor the status of stress accumulation at the seismogenic zone on the plate interface.
G22A-02 INVITED
The Mechanics of Deep Tremor and Slow Slip
Deep tremor is a weak, extended-duration seismic signal observed episodically on some major faults, often in conjunction with slow slip events. We use borehole Hi-Net data to study tremor beneath Shikoku, Japan and through the use of a matched-filter algorithm, find that it can be explained as a swarm of small, low-frequency earthquakes (LFEs). Precise relocation places these LFEs on the subduction interface, immediately down-dip of the locked portion of the plate boundary. In order to determine the mechanism, we analyze stacked LFE waveforms and compare them with the waveforms of nearby earthquakes of known mechanism within the subducting Philippine Sea Plate. Both P-wave first-motion focal mechanism and the S-wave moment tensor analysis indicates that LFEs are shear slip on a low-angle thrust fault dipping to the northwest. We conclude that LFEs, and hence deep non-volcanic tremor, is generated directly by shear slip on the plate boundary, and represents a seismic signature of the accompanying slow slip events (SSEs). SSEs that accompany tremor have durations of up to weeks. They have been observed in Japan using borehole accelerometers and in Cascadia using GPS. Recently discovered very low frequency earthquakes (VLFs) with characteristic durations of 10s of seconds have been demonstrated to occur as shear slip in the same location as LFEs and SSEs, but with a longer time constant. Taken together, these events define a spectrum of slow shear-slip events spanning 8 orders of magnitude in moment and duration. Borehole observations of seismic/aseismic transients should allow our results from Japan to be generalized to other tectonic settings.
G22A-03
Estimate of interplate coupling along the Nankai Trough, Southwest Japan using GEONET data
Over the last 1000 years, larger magnitude M~8 earthquakes have occurred every 100-200 years at the Nankai trough, southwest Japan, where the Philippine Sea plate is subducting at a rate of ~60mm/yr beneath the southwestern Japan. In addition to large earthquakes, several transient slip events have been observed in this region, possibly due to spatially varying frictional properties on the subduction interface. Using the dense GEONET continuous GPS network, we estimate the spatial extent of interplate coupling along the Nankai trough. The GPS time series are from the JPL-developed GPS Network Processor, which uses GIPSY-OASIS and cluster computing to quickly and efficiently analyze large quantities of data. Using QOCA to remove non-tectonic signals such as common mode error and seasonal signals, we have determined accurate horizontal and vertical interseismic velocities for the entire continuous GPS network. In this study, we use the results of this analysis to look at steady-state subduction plate coupling in southwestern Japan. We construct plate interface geometry from the west of Shikoku to the Tokai region, based on the composite plate boundary model by Wang et al. [2004]. For the inversion procedure, we use both horizontal and vertical GPS site velocities, referenced to the Amurian plate. By employing elastostatic Green's function of triangular dislocation elements [Jeyakumaran et al., 1992] and spatial smoothing based on the Fujuwara operator [Desbrun et al., 1999], we estimate the back slip along the Nankai trough. We find strong coupling at the depth of ~10-30 km on the plate interface beneath Shikoku, Kii Peninsula and Tokai area. There is good spatial correspondence between highly coupled regions with the rupture zones of past large earthquakes (e.g., 1946 Nankai and 1944 Tonankai earthquakes). It appears that there is complementary pattern between short-and long-term slow slip patches [Hirose and Obara, 2005] and strong coupling zone, which may suggest variations in frictional properties on the plate interface.
G22A-04
The New National Seismic Network of Chile
The new National Seismic Network of Chile, sponsored by the Government and executed by the Departments of Geophysics and Civil Engineering of the University of Chile, will consist of 65 broadband stations, sampled at 50 sps, colocated with more than 200 strong ground motion sensors sampled at 200 sps as well as 140 GPS instruments sampled at 1 sps. These instruments will be deployed along the country (4200-km long) with near real time access. Satellite communications will be used for the BB and GPS devices directly linked to a central site in Santiago with a mirror system in a different site. The strong ground motion instruments will be accessed via phone lines. This network will be supported by the efforts of regional universities along the country that will run their own local seismic short-period arrays. It is expected that the three integrated networks will be installed during the next three years and be fully operational by the end of the period 2008-2010. The additional effort of installing regional networks is planned for a longer period.
G22A-05 INVITED
Areal and Shear Strain Coupling of PBO Borehole Strainmeters From Teleseismic Surface Waves
In order to compare borehole strainmeter data with tectonic models, we must know the coupling parameters relating elastic deformation of the strainmeter to strain in the surrounding rock. At least two coupling parameters are required: the ratios of instrument areal and shear strain to formation areal and shear strain, respectively. These coupling parameters depend on the relative elastic moduli of the formation, grout, and strainmeter, and typical elastic moduli yield nominal coupling parameters of 1.5 and 3. More accurate coupling parameters must be determined by analyzing each strainmeter's response to a known deformation source after the instrument has been grouted into the borehole. Borehole strainmeters installed by the National Science Foundation-funded Earthscope Plate Boundary Observatory (PBO) consist of four gauges, sampled at 20 Hz, that measure extension along distinct azimuths. Teleseismic Love and Rayleigh waves that produce fractional gauge elongations > 10-7 , such as those from the M8.3 Kuril Islands earthquake of November 15, 2006, can be used to constrain the coupling parameters. A planar Love or Rayleigh wave is expected to have a simple strain field that produces the same waveform on all four gauges of a strainmeter. The two-parameter coupling model is consistent with the variation of surface wave amplitudes as functions of azimuth for the borehole strainmeter data analyzed to date, although most of the PBO strainmeters require that differences in the relative gains of the four gauges be estimated as well. Fits to the data can be improved for some strainmeters by allowing for two distinct shear strain coupling parameters, and/or for a small (<10 degrees) error in the orientation of the strainmeter as measured during installation. However, data from more earthquakes will need to be analyzed before these refinements can be called significant. The Rayleigh wave data provide tight constraints on the ratio of shear to areal coupling. For borehole strainmeters in northern Washington state, these ratios are in the range 5 to 10, in contrast to the ratio of 2 for the nominal coupling parameters. This behavior could be explained by grout with a relatively low elastic modulus. Surface waves are well recorded by some strainmeters that do not record earth tides, demonstrating that for some borehole strainmeter installations, coupling may decrease at long periods.
G22A-06
Groundwater-Level Anomalies Associated with a Hypothetical Preslip Prior to the Anticipated Tokai Earthquake: Evaluation of Detectability Using the Groundwater Observation Network of AIST
A large earthquake of M 8 has been anticipated along the Suruga trough, offshore from central Japan. This anticipated earthquake is referred to as the Tokai earthquake, and the Japanese Government has maintained an earthquake prediction program for this earthquake since 1978. The National Institute of Advanced Industrial Science and Technology (AIST) has been monitoring groundwater levels in the Tokai region since 1978. As of 2007, AIST continuously monitors 15 wells at nine observation sites in the Tokai region. We infer the groundwater-level anomalies associated with a hypothetical preslip prior to the anticipated M8 Tokai earthquake, and evaluate the detectability of the anomalies using data from seven groundwater wells. We evaluate the detectability of the anomalies under the following assumptions: (1) an Mw 5.5 - 6.5 aseismic preslip event occurs at the plate boundary in and around the hypothetical focal zone of the Tokai earthquake; (2) the total amount of the strain step at each observation associated with the preslip can be calculated by tensile and shear faulting based on the dislocation model; (3) a normalized strain history associated with the preslip is defined from the results of numerical simulations based on rate- and state-dependent friction laws; and (4) the groundwater-level anomaly prior to the earthquake is proportional to the estimated history of the strain change associated with the preslip. We investigate the detection time of the anomaly at seven wells given an Mw 5.5, 6.0, or 6.5 aseismic preslip at one of the 272 grid points in and around the area of the hypothetical focal zone of the Tokai earthquake. As a result, over the time interval between 1 and 48 hours prior to the hypothetical Tokai earthquake, we are able to detect at each of the seven wells a hypothetical Mw 6.5 preslip at 10-86 of the 272 grid points, an Mw 6 preslip at 0-19 grid points, and an Mw 5.5 preslip at 0-5 grid points.
G22A-07
Studies of spatial sensitivity of the hydrological response to earthquakes
Numerous observed earthquake induced hydrological changes inferred some sensitive wells could detect the tiny signals of the crustal deformation. But the reasons and mechanisms included for the sensitive wells are not well-known. Some previously study (Matsumoto et al., 2003; Roeloffs et al., 2003) show individual mechanism could be the reasons made the wells with high sensitivity to earthquake. Also the earthquake induced hydrological changes shows the high variances in spatial distribution (Lai et al., 2005). To make a criteria for choosing sensitive sites or wells are still quite a big challenge. In this study, we using the observations of the groundwater and earthquake monitoring network in Taiwan act as the dense observation dataset to the earthquake induced hydrological responses. The comparison of the all events (detectable and non detectable) in groundwater level and the inferred volumetric strain step in response to the earthquake events were discuss. These observations had been geostatistically analysis for estimate the sensitivity of the observation well to different spatial regions. From the spatial analysis of the detect ability, they shows the highly anisotropy and heterogeneity in four wells. They could partly explain the different responses of earthquake induced groundwater changes. Tectonic and structural geology setting could be the main reason control the spatial difference of earthquake induced groundwater changes (Fault-Barrier Effect). The strain model usually could explain the type of the coseismic change, but the amplitudes usually not fit to the homogeneous assumption. The Structural anisotropy and mechanical heterogeneity should be considered to improve the volumetric strain estimation. The results issued the limitation and ability of the detection of the observation well spatially. Base on the spatial distribution of the detection, we should choose fitted sites or wells for the purpose. http://www.dprc.ncku.edu.tw
G22A-08 INVITED
Mapping Geodetic Strain Rates into Long and Intermediate Term Earthquake Potentials
We modify our method of long and intermediate term earthquake forecast using geodetically derived crustal strain rates. The method is based on the assumptions that: (a) seismicity rate is steady and proportional to the horizontal maximum shear strain rate; and b) earthquake magnitude distribution is spatially invariant except for an amplitude constant which is proportional to the maximum horizontal shear strain rate. This method is relatively easy to implement if crustal deformation is steady, and not so if significant postseismic deformation is involved. The difficulty is two folds: First a populated continuous GPS network, not always available in most parts of the world, is needed to monitor postseismic deformation both in space and time. Second causal relationship between the transient strains and earthquake occurrence is not quite clear. We tackle the problem by (a) implementing a postseismic deformation model such that transient strain rates could be derived from limited observational data epochs of a geodetic network, and (b) treating the postseismic strain the same as the long term tectonic strain for earthquake forecast, since the former is imposed onto the brittle upper crust the same way as the latter to cause earthquakes. Data analysis in southern California reveals that postseismic deformation is best described by a logarithmic function, with a decay time constant of 10 days regardless of earthquake magnitudes and locations. Combining the postseismic and secular strain rates one could assess the long and intermediate earthquake potentials, which evolve with time. Application of the method to earthquake forecast in southern California and its retrospective test result will be presented at the meeting. We will also assess the possible gain for earthquake forecast using fault-model aided strain rates rather than the strain rates derived directly from the geodetic data.