T13F-01
High-Vp/Vs Zone Accompanying Low-frequency Tremors, Short-term and Long-term Slow Slip Events Beneath the Southwestern Japan
With regard to the Philippine Sea (PHS) plate beneath southwestern Japan, deep low-frequency non-volcanic tremors (LFTs) are observed at depths of approximately 30 km along the strike of the PHS plate from Tokai to the western Shikoku regions (Obara, Science, 2002). Accompanying the LFTs, short-term slow-slip events (S-SSEs) are also observed beneath the Tokai region, eastern Kii peninsula, and Shikoku region. The source region of LFT is consistent with that of S-SSE. Some studies revealed that the LFTs beneath the southwestern Japan are located along a high-Vp/Vs zone by seismic tomography. Beneath the Tokai region, Ozawa et al. (Science, 2002) detected a long-term slow slip event (L-SSE) that started late 2000 and ended in 2005. Beneath the Bungo channel, at the west side of the Shikoku region, L-SSE was also observed during 1999 (Hirose et al., GRL, 1999). The National Research Institute for Earth Science and Disaster Prevention (NIED) has deployed the high- sensitivity seismograph network of Japan (Hi-net). Plenty arrival time data enable us to clarify the fine-scale 3D Vp and Vs structure beneath the southwestern Japan. In this study we estimate the relationship with velocity structures and L-SSE as well as LFT and S-SSE. The target region, 29–37°N and 129–139°E, covers the southwestern Japan. A total of 1,198,906 P-wave and 966,496 S-wave arrival times for 31,751 earthquakes recorded at 466 Hi-net stations are available for use in the tomographic method (Matsubara et al., Tectonophys., 2004; JGR, 2005) introduced spatial velocity correlation to the original code of Zhao et al. (JGR, 1992). We assume that artificial velocity discontinuities are absent since an incorrect configuration of the boundary would lead to an incorrect solution. Horizontal grid spacing is 0.1 degree and resolution is 0.2 degree. Slow earthquakes are distributed along the high-Vp/Vs zone beneath the southwestern Japan and they are considered to be caused by the high pore fluid pressure dehydrated from the oceanic crust of the PHS plate. The souce region of the S-SSEs are located in the high-Vp/Vs zone beneath the Tokai and Bungo channel, at the downdip landward end of the high-Vp/Vs beneath the southern Kii peninsula, at the updip seaward end of the high-Vp/Vs zone beneath the western Shikoku region. The high-Vp/Vs regions extend landward beneath the Tokai and western Shikoku region and Bungo channel. The mantle wedge has high-Vp/Vs zone with 10-30 % serpentinized peridotite exist if there is serpentinized. The S-SSE may occur at the plate boundary between the PHS plate and the serpentinized wedge mantle. High-Vp/Vs zone is extended from the S-SSE zone to the updip seaward beneath the Tokai region and Bungo channel along the plate boundary. L-SSEs were observed beneath both regions. The fluid also exists in the high- Vp/Vs (1.84) zone of the oceanic crust beneath the Tokai and the Bungo channel at depths approximately 25-30 km and L-SSEs occur owing to the high pore pressure. On the contrary beneath the Kii peninsula, L-SSE is not observed since the Vp/Vs is high as 1.80 and this value is lower than that beneath the Tokai region.
T13F-02
Scaling Relationships for Slow Slip Events and Tremor in Subduction Zones
Global compilations of faulting parameters for slow slip phenomena in subduction zones reveal that event duration is proportional to seismic moment rather than its cube-root, as for earthquakes (Schwartz and Rokosky, 2007; Ide et al., 2007). Ide et al. (2007) proposed two different models consistent with this scale dependent behavior that either assume direct proportionality between fault slip (d) and fault length (L), as observed for earthquakes or constant slip, independent of fault length. These two models predict different relationships between fault length and seismic moment, stress drop and rupture velocity that can be tested with available data. Combining data from 25 short-term slow slip events in Western Shikoku, the Kii Peninsula and Tokai, Japan with longer duration slow slip events globally reveals a nearly linear relationship between seismic moment (Mo) and fault length, inconsistent with the Mo α L3 and Mo α L2 predictions of the two previously proposed models. The observed relationship implies that fault slip is proportional to the inverse of fault length, stress drop is proportional to the inverse of fault length squared and rupture velocity is constant. Although at present the scatter in most of these slow slip event parameters is still too large to confidently differentiate between trends, the relatively small range in slow slip propagation velocities over a large range in fault lengths is consistent with this new model. The direct proportionality between seismic moment and duration (τ) implies that slow slip phenomena have constant moment rate and therefore constant far-field-displacements as a function of time. We test this prediction by calculating reduced displacements for tremor episodes accompanying short-term slow slip events in SW Japan. Consistent with Mo α τ scaling, we find that reduced displacements are relatively constant throughout each 3-10 day slow slip episode. Average values of reduced displacement vary between the different slow slip episodes and show a decrease in reduced displacement with increasing cumulative slip in the accompanying slow slip event. We are investigating whether this, as well as other systematic tremor amplitude behavior, reflects regional variations and what it reveals about the mechanics of tremor and slow slip events.
T13F-03
Tremor Constraints on Moment Release During the 2007 ETS from Surface and Borehole Seismometers
The 2007 ETS event, which began around Jan 15 beneath the southwestern Puget Basin and ended around Feb 5 beneath southern Vancouver Island, was well-recorded on local surface seismic arrays, EarthScope borehole- seismometers, strainmeters and long-baseline tiltmeters, and continuous GPS of the PANGA and PBO networks. Seismic tremor, however, offers the highest resolution for studying moment release through time, since tremor bursts lasting less than 10-seconds are often visible across stations. To test the hypothesis that tremor and transient deformation are two manifestations of the same faulting process, and to quantify the relative contribution of moment release during times of strain-transients versus other times, we systematically analyze the tremor bursts during the time period of 2005-2007.2, which includes the 2007 ETS event. We first consolidate daily seismic files from the Puget Basin of Washington State and SW British Columbia, where GPS density is highest. Seismic traces are included from the PNSN, the PBO borehole seismic network, and the EarthScope-funded CAFE experiment. We remove instrument gain, decimate the data to 10 sps, rectify it, compute its envelope using a Hilbert transform, and average the envelopes from regionally adjacent stations to provide a single metric indicative of tremor activity. This process is effective in quantifying small tremor bursts lacking GPS-inferred deformation and accurately identifies timing and duration of known events. We then compare tremor duration to equivalent moment slip inversions of corresponding GPS-derived deformation to obtain a model that relates hours of tremor to moment magnitude. To locate the tremor during the 2007 event, we use both picked waveform peaks and cross-correlated envelopes of band-pass filtered instruments. The location is determined by minimizing the L2-norm of the vector containing the differences between the measured and predicted stations offsets for a 3D grid of possible locations. Although the scatter is high, particularly in the depth, we find tremor during the 2007 event propagates in a northwesterly direction beneath the eastern Olympics Range over a three-week period. We find no instances of transient GPS deformation occurring in the absence of tremor, and a very linear relationship between tremor duration and GPS-estimated equivalent moment. However, the average amplitude of tremor, both during GPS-recognizable ETS and isolated bursts, varies little from the maximum velocity of ~1 micron/sec.
T13F-04
Scaling of the Tremor Source
The scaling of the tremor source quantifies the striking differences between tremor and regular earthquakes and offers clues on the underlying physics of this lately-discovered phenomenon. Recently, a compilation of various different types of slow slip events by Ide et al. (2007) suggested that durations of such events are linearly proportional to their seismic moments. Here I conduct a systematic study of source scaling of non-volcanic tremor using a consistent data stream for the September 2005 Cascadia ETS. Day-long horizontal-component seismograms from the small-aperture PA array above the Cascadia subduction zone are integrated, corrected for gain, and filtered from 1 to 5 Hz to highlight the tremor. Then envelopes of the horizontal records are stacked and lightly smoothed (low-passed below a varying frequency of 0.03 to 0.002 Hz). Tremor events are defined as intervals during which the envelope is greater than a threshold level. Durations of tremor events thus defined range from ~100 s to several hours. The envelopes of the tremor events are treated as time functions (moment- rates). Preliminary analysis of the shapes of the tremor time functions suggests a modest tendency for negative skewness – on average, these tremor events end slightly more abruptly than they start. An increasing peakedness (kurtosis) of the larger events is also seen. Moment of the tremor events is estimated assuming the signal consists of far-field direct S-waves emanating from the plate boundary below the array. I find that event duration is proportional to M00.85. This result does not depend on the absolute estimation of moment and appears robust - over a range of thresholds and smoothing periods, the exponent varies from 0.80 to 0.88, being commonly between 0.82 and 0.86. This allows for modest growth of the amplitude of tremor events with duration, which is clearly seen in the envelopes. This scaling contrasts strongly with that of regular earthquakes, which follow duration proportional to M00.33 over many orders of magnitude, and for which amplitudes grow strongly with increasing moment. That scaling occurs due to the rough proportionality of fault displacement, length, and width, and the imposition of a roughly constant value of rupture velocity by the dynamic stress propagation associated with earthquakes. One or more of these factors must be missing from tremor generation, and constraints on the spatial parameters will be useful in interpreting the scaling relation. The total moment released via tremor on a representative vigorous day of the 2005 ETS corresponds to that of a M5.3 earthquake. This suggests that the majority of slip in an ETS period occurs silently, without generating seismic signals in the bandwidth in which tremor is typically detected.
T13F-05
Source duration of deep very low frequency earthquakes in the western Shikoku region
Recent studies (e.g. Obara, 2002; Rogers and Dragert, 2003) report various types of "slow" events (tremors, low frequency earthquakes, very low frequency earthquakes and slow slips) at the subduction zones. Source duration of such events follows a scaling relationship different from the regular earthquakes (Ide et al., 2007). Ito et al. (2007) found very low frequency (VLF) earthquakes at the subduction zone of the Philippine Sea plate. Assuming that the source duration is sufficiently shorter than the frequency band of 0.02-0.05 Hz, they determined the hypocenters, the seismic moments, and the focal mechanisms by the centroid moment tensor inversion (CMTI) approach. However, the actual source duration of VLF earthquakes is still unclear. In this study, we estimated the source duration of VLF earthquakes, using the seismic data of the high-sensitivity seismograph network (Hi-net) which is operated by National Research Institute for Earth Science and Disaster Prevention. In the middle of March 2007, deep low frequency tremors became very active associated with a slow slip event in the western Shikoku region. Applying the CMTI approach, we determined the hypocenters and the focal mechanisms of VLF earthquakes during this period. Obtained focal mechanisms are thrust type according to the geometry of the subducting plate boundary. The hypocenters of VLF earthquakes migrated from the north to the south. Mw 3.8 VLF earthquake at 19:29 on March 14 (UT) is the largest event that we and Ito et al. (2007) have ever detected. At this event, body waves are clearly recognized in the seismograms of the high-sensitivity horizontal accelerometer (Hi-net TILT) which is installed at each Hi-net station. We estimate the source duration of this event through the comparison between the observed and the synthetic waveforms. The synthetic waveforms are directly calculated if a source time function is given, because the hypocenter, the seismic moment, and the focal mechanism of this event are already obtained by the CMTI analysis. We assume that the source time function is expressed by a linear B-spline basis function whose seismic moment is Mw 3.8. Employing the forward modeling approach, we found that 12 s of source duration explains the observed displacement waves well. We note that the source duration is longer than the S-P time, because the all stations are located within 80 km from the epicenter. Such long duration smoothes out the contribution from the P-wave, S-wave, and near field terms. We also compared the waveform of several smaller VLF earthquakes (Mw 3.4-3.5) in the western Shikoku region, and the source durations of around 10 s are suitable to fit the waveforms, though the estimations for such small events are less reliable than the case of Mw 3.8 events. Our analysis also supports the previous results of CMTI approach in terms of direct body waves, and shows that the source duration of M3 "slow" events is around 10 s. This duration is several tens times longer than that of the regular M3 earthquakes.
T13F-06
Slow earthquakes with duration of about 100 s suggested by the scaling law
Slow earthquakes in western Japan are considered as a group of interplate slip events that obey the scaling law proposed by Ide et al. (2007), in which the seismic moment is proportional to the event duration. However, the population of events in this group is not continuous. In the Nankai slow earthquake zone, we have found deep low-frequency earthquakes (LFE) below 1 s, very low-frequency earthquakes (VLF, Ito et al., 2006) between 20-50 s, and slow slip events (SSE) above a few days. Are there any slow events other than these? If a slow earthquake that satisfies the scaling relation with duration of about 100 s occurs within the Nankai slow earthquake zone, it is observable at low-noise stations with a vertical broadband sensor only if they are located near the maximum direction of the near-field signal. One station that satisfies these conditions is F-net KIS with STS-1 seismometers, maintained by National Research Institute for Earth Science and Disaster Prevention, Japan. This station records tremor activities a few times per year since 1996. During most of the activities, we can detect many large low-frequency signals. Longer events include VLFs and we can show that some previously reported VLFs are actually a part of a longer event. We installed a temporary observation station at 15 km from KIS station and recorded a sequence of low frequency tremor for July 17-20, 2007. Although the low frequency signals are visible at two stations, the amplitudes are quite different, which suggests that we can determine the location and orientation of the source using a small dense array of broadband seismometers. As expected, the moment magnitudes of 100 s events are around 4, which satisfy the scaling relation of slow earthquake. Existence of much larger and longer events is implied from the records of KIS, although large low- frequency noise less than 3 mHz impedes reliable judgments. The existence of such events suggests that any size of slow earthquakes may occur like regular earthquakes.
T13F-07
Modeling short-term silent slip events along the deeper parts of the Nankai subduction zone
Recent high-resolution observations of crustal movements have revealed short-term silent slip events (SSEs) with propagation velocities of around 10--15 km/day and with intervals of 3--14 months along the deeper parts of the Cascadia and Nankai subduction zones (Dragert et al., 2001; Obara et al., 2004). Shibazaki and Shimamoto (2007) have developed a model for these short-interval SSEs by considering the frictional behavior that was experimentally confirmed by Shimamoto (1987) using halite for the unstable-stable transition regime. They have modeled the SSEs considering a rate- and state-dependent friction law with a small cutoff velocity to an evolution effect. In their model, under conditions where the pore-fluid pressure is nearly equal to the lithostatic pressure and the critical weakening displacement is very small, short-interval SSEs with propagation velocities and slip velocities of 10 km/day and 5×10-7m/s, respectively, can be reproduced. They have found that the propagation velocity of short-interval SSEs is in proportion to the slip velocity. The activity of short-term SSEs in western Japan has been well investigated by Obara (1997). To understand the loading processes for great thrust earthquakes along the Nankai subduction zone, it will be important to model these short-term SSEs considering a realistic 3D geometry of the subduction interface. Based on the study by Shibazaki and Shimamoto (2007), we have developed a model of short-term SSEs on the 3D subduction interface beneath Shikoku, western Japan. The occurrence of SSEs is very complex: the generation zone of SSEs is divided into several segments in the horizontal direction. The mode of segmentation depends on the width of the generation zone. When the width of the generation zone is sufficiently small, significant segmentation occurs. On the other hand, when the width of the generation zone is large, larger events can occur that extend horizontally over the entire region. From the epicentral distribution of deep low-frequency tremors, we set the width of the generation zone of SSEs such that it is wider beneath the western part of Shikoku than beneath the eastern part. We can reproduce events with longer lengths in the horizontal direction beneath the western part of Shikoku with longer recurrence times. The numerical results are consistent with the observation by Obara (2007) that the events at longer segments have longer recurrence intervals. We have also attempted to model the very-low-frequency earthquakes observed by Ito et al. (2006) that are accompanied by short-interval SSEs. To model low-frequency earthquakes, we are required to consider a nonuniform fault zone structure, where local rupture with a high slip velocity occurs together with the propagation of short-interval SSEs. We consider a local patch where the critical displacement is very small. We then confirm that high-speed slips occur in the local patch with a small critical displacement. In some cases, slips occur at the same patch repeatedly within a short interval. We consider that, even in the transition zone, there are heterogeneities in the frictional properties and the critical displacement is scaled with the size of the events.¡¡We report the conditions for the friction parameters for which we can reproduce the low-frequency earthquakes that satisfy the scaling law proposed by Ide et al. (2007).
T13F-08
Dilatancy Stabilization of Frictional Sliding as a Mechanism for Slow Slip Events
Slow slip events and associated non-volcanic tremor have been discovered in a number of tectonic settings, yet the processes giving rise to these phenomena are as yet not understood. Transient slip in subduction zones appears to occur between the locked megathrust and the steadily creeping fault below, suggesting that slow slip occurs in regions near frictionally neutral stability. However, the transiently slipping zone must be large enough to allow non-steady slip but not so large that the rupture becomes dynamic. The size range for which transient, quasi-static slip occurs is small, particularly for the slip-law form of rate-state friction. We suggest that rate-state friction nucleates slip under drained conditions but that as slip accelerates deformation becomes effectively undrained, and dilatancy induced pore-pressure reductions quench the instability. We study this process assuming 2D elasticity, rate-state friction and the Segall-Rice [1995, JGR] constitutive law for dilatancy. Pore-pressure is treated mainly with simplified membrane diffusion: dp/dt = (p∞ - p)/tf + (1/β) d φ/dt, where p and p∞ are fault and remote pore-pressure, tf a characteristic diffusion time, β pore and fluid compressibility and φ fault zone porosity. For a step change in slip speed, v, the peak dilatant suction scales with (ε/β) \log (v θ/dc) g(v tf/dc), where ε is the dilatancy parameter, θ the state in front of the rupture, and g(v tf/dc) a function of the ratio of diffusion time to that for state evolution. Using drained results of Rubin -Ampuero [2005, JGR] we find that the ratio of dilatant strengthening to frictional weakening scales with E \equiv f0 ε/ β b (σ - p∞), where f0 is nominal friction and b the rate-state parameter. Indeed, numerical simulations with E ~ 1 exhibit slip that accelerates to limiting speeds well below inertial, followed by stable propagation. Simulations with E < ~ 0.1 accelerate to radiation damping limits. This suggests that stable slip is favored by low effective stress, consistent with some seismic observations. Transient sip nucleates at the down-dip end of the velocity weakening region, below which creep is imposed at v∞, and propagates updip toward the locked region. We observe stable transient slip for an effectively unbounded range of W/h* >~ 2, where W is the width of the velocity weakening region and h* is the drained critical nucleation dimension. With increasing W/h* the behavior transitions from periodic, to complex via period doubling. In the complex domain multiple slip events are spawned at the down-dip end, some overtaking early formed events. Increasing v∞ tf/dc inhibits drainage, thereby increasing the time between events while decreasing their amplitude. We have also developed finite difference methods that allow for one-dimensional fluid diffusion normal to the fault. In this case we have obtained analytical solutions only by excluding elastic fluid storage in the shearing zone. The peak suction scales similarly to the membrane diffusion case, although the decay in pore-pressure change is far slower, consistent with numerical simulations.