Baba, T., and P. R. Cummins (2005), Contiguous rupture areas of two Nankai Trough earthquakes revealed by high-resolution tsunami waveform inversion, Geophys. Res. Lett., 32, L08305, doi:10.1029/2004gl022320. Park, J.-O., T. Tsuru, N. Takahashi, T. Hori, S. Kodaira, A. Nakanishi, S. Miura, and Y. Kaneda (2002), A deep strong reflector in the Nankai accretionary wedge from multichannel seismic data: Implications for underplating and interseismic shear stress release, J. Geophys. Res., 107(B4), doi:10.1029/2001jb000262. Takahashi, N., S. Kodaira, J.-O. Park, and J. Diebold (2003), Heterogeneous structure of western Nankai seismogenic zone deduced by multichannel reflection data and wide-angle seismic data, Tectonophysics, 364, 167 190. Original paper: Obana et al. (2006), Gophys. Res. Lett., 33, L23310, doi:10.1029/2006gl028179. Seismicity in the incoming/subducting Philippine Sea Plate off the Kii Peninsula, central Nankai Trough Along the Nankai Trough, southwestern Japan, the Philippine Sea Plate (PHS) is subducting beneath the Eurasian Plate, and large interplate earthquakes have occurred repeatedly. The latest large thrust earthquake at the central Nankai Trough off the Kii Peninsula was the 1944 Tonankai earthquake (Mw = 8.2). Although a previous ocean bottom seismograph (OBS) experiment from November 2001 to February 2002 showed seismicity around the rupture area of the 1944 Tonankai earthquake, there were large uncertainties regarding hypocenter locations, especially with respect to depth. We conducted another OBS experiment to obtain accurate locations of earthquakes and their focal mechanisms. The new results show earthquake clusters at the trough axis within the oceanic crust (Fig. 1-1-3). The clusters were located near faults cutting the oceanic crust, and thus the earthquakes may have occurred along preexisting faults in the incoming PHS. Seismicity in the subducting PHS is characterized by seismic quiescence in the oceanic crust of the coseismic rupture area of the 1944 Tonankai earthquake. The change in seismicity in the oceanic crust can be explained by changes in the stress of the incoming, subducting PHS. The stress in the oceanic crust of the PHS changes from trough-normal tension at the trough axis, caused by the bending of the PHS, to downdip compression beneath the Kii Peninsula. The seismicity in the subduction zone likely relates to the balance of stress in the incoming and the subducting plate as well as to coupling between the subducting and overriding plates. Fig. 1-1-3. Hypocenter distribution determined by the OBS experiment. Red circles indicate earthquakes. The size of the symbol corresponds to the magnitude of the earthquake. The P-wave velocity model for X= 0 km is projected as a cross section along the Y axis. The iso-velocity contour interval is 0.5 km/s. Original paper: Obana et al. (2005), J. Geophys. Res., 110, B11311, doi:10.1029/2004jb003487 Random inhomogeneity in southwestern Japan revealed from the peak delay time analysis of high-frequency S-wave envelope The characteristics of the medium inhomogeneities of which characteristic scale is less than a few km can be inferred from the envelopes of high-frequency seismic waves (>1Hz). This study examines the path-dependence of envelope broadening, and investigates the spatial distribution of random inhomogeneities in southwestern Japan. We mainly focus on the peak delay time ( t p ) defined as the time lag from S-onset to the maximum amplitude arrival of its envelope. This quantity is the best measure of the accumulated scattering effect due to random inhomogeneities. We measure the peak delay time in 2-4Hz, 4-8Hz, 8-16Hz and 16-32Hz bands by using velocity seismograms recorded by Hi-net (NIED) and ocean bottom seismograph (OBS) stations. As an average feature of peak delay times, most of the peak delay times show smaller values than those in northeastern Japan. We also find that peak delay times show larger values in higher frequencies for the ray-path propagating beneath the Quaternary volcanoes in northern Kyushu (Fig. 1-1-4). Applying the simple method proposed in Takahashi et al. (2007), we show that the region beneath the Quaternary volcanoes along the rift from Yufu-dake to Aso volcano demonstrates large velocity fluctuations at 0-20km depths (Fig. 1-1-4). The deeper parts beneath this rift-zone show weak inhomogeneities, as does its surrounding area. In the case of northeastern Japan, the medium inhomogeneities beneath Quaternary volcanoes tend to be strong as depth increases. This difference between the rift-zone and northeastern Japan is due to the differences in the 23
genesis of the Quaternary volcanoes. interplate earthquakes. Seismic swarm activities have been observed inside the subducting oceanic crust along the bending axis, and we infer that some of these activities are caused by the bending of the subducting plate. In addition, P-wave velocities of the upper plate vary across the bending axis, implying that the position of the bending axis is closely related to physical properties of the upper plate and to seismic activities. Fig. 1-1-4. Spatial distribution of minimum values of! logt p in 4-8Hz at depths of 0-20km. The! logt p represents the relative strength of the accumulated scattering effect (Takahashi et al. 2007). Red and yellow regions indicate where small peak delay times are not observed. These regions are expected to be strong inhomogeneities. Open triangles are Quaternary volcanoes. Reference Takahashi T., H. Sato, T. Nishimura and K. Obara, Strong inhomogeneity beneath Quaternary volcanoes revealed from the peak delay analysis of S-wave seismograms of microearthquakes in northeastern Japan, Geophys. J. Int., 168, 90-99, 2007 Original paper: Takahashi T., SSJ Fall Meeting, C51, 2006. Sharp plate bending and the seismicity in the northern Japan Trench subduction zone. Many destructive earthquakes have occurred in the northern Japan Trench region owing to the subduction of the Pacific Plate. To reveal the relation between the distribution of earthquakes and the geometry of the plate boundary along the subduction direction, we conducted wide-angle seismic experiments in the regions off Aomori (40N), off Iwate (39N) and off Miyagi (38N). We analyzed the acquired data by applying a new method, in which observed reflection phases are directly projected onto the depth-distance domain from the time-distance domain. As a result, we found plate-bending points in all the three regions (Fig. 1-1-5). Distances between the three bending points and the trench axis are almost uniform and sharp plate bending is a common feature in the northern Japan Trench subduction zone. The bending axis, which is defined by connecting bending points, approximately coincides with the updip limit of the rupture zones of large Fig. 1-1-5. Plate bending points and seismic activities in the northern Japan Trench subduction zone. (a) Three straight lines crossing the trench axis are wide-angle seismic experimental lines. Dark ellipses indicate bending points and a dashed line connecting bending points is a bending axis. Gray contour lines show the slip distribution for eight M>7 earthquakes that occurred after 1930; the contour interval is 0.5m and the areas within the value of half the maximum slip are filled with light gray (Yamanaka and Kikuchi, 2004). Small open circles and crosses show hypocenters of microearthquakes (Nishizawa et al., 1992; Hino et al., 1996). (b,c,d) Depth sections. Large arrows indicate the position of the bending point, and the gray dashed lines are the plate boundary interface. Most of the microearthquakes occur inside the subducting oceanic crust around the bending point. References Hino, R., Kanazawa, T., Hasegawa, A., 1996. Interplate seismic activity near the northern Japan Trench deduced from ocean bottom and land-based seismic observations. Physics of the Earth and Planetary Interiors 93, 37--52. Nishizawa, A., Kanazawa, T., Iwasaki, T., Shimamura, H., 1992. Spatial distribution of earthquakes associated with the Pacific Plate subduction off northeastern Japan revealed by ocean bottom and land observation. Physics of the Earth and Planetary Interiors 75, 165--172. Yamanaka, Y., Kikuchi, M., 2004. Asperity map along the subduction zone in northeastern Japan inferred from regional seismic data. J. Geophys. Res. 109, B07307, doi:10.1029/2003jb002683. Original paper: Fujie et al., Phys. Earth Planet. Intr., 157, 72-85, 2006 Last stage of the Japan Sea back-arc opening deduced from the seismic velocity structure The Japan Sea is one of the very well studied back-arc basins in the northwestern Pacific. The seismic crustal model, however, has been inadequate in elucidating the detailed opening model of the Japan Sea. In 2002, to clarify the late 24
stage of the formation style of the Japan Sea opening, a seismic experiment using 35 ocean bottom seismographs (OBSs), an airgun array, and a multi-channel hydrophone streamer was undertaken in the areas from the southwestern Yamato Basin, the Oki Ridge, and the southwestern Oki Trough to the coastal area of the southwestern Japan Island Arc. The crusts beneath the southwestern Yamato Basin and the Oki Ridge are estimated to be approximately 13 km and 19.5 km, respectively (Fig. 1-1-6 (a)). The upper and lower crusts of the southwestern Yamato Basin are approximately 3.2 km and 8 km thick, respectively. Those of the Oki Ridge are approximately 8.2 km and 10.5 km thick, respectively. The upper crust of the Oki Ridge thickens more steeply than that of the southwestern Yamato Basin; however, the lower crust thickens more gently. The crustal structure of the southwestern Yamato Basin shows the extended continental crust accompanied by the Japan Sea opening because this crustal thickness is less than that of a typical continental crust and greater than that of a typical oceanic crust. The upper crust being thinner than the lower crust is a remarkable structural characteristic, which is caused by listric or complicated normal faults that have developed in the upper crust of southwestern Yamato Basin. This deformed upper crust is a common structural characteristic in the southern Japan Sea, which includes Yamato Basin. Southern Yamato Basin, including southwestern Yamato Basin, has the thinnest upper and lower crusts in the Japan Sea. For that reason, it is suggested that southern Yamato Basin had the strongest deformation by a back-arc opening and that the period of the opening in southern Yamato Basin had been longest in the southern Japan Sea. The formation process of southern Yamato Basin is inferred to have two stages: rifting and extending the continental crust separating the northeastern and southwestern Japan Island Arcs from the Asian continent, and furthering the extension affected by the rotation of the southwestern Japan Island Arc (Fig 1-1-6 (b, c and d)). respectively. Numerals are P-wave velocities (km/s), and the lines are iso-velocity contours whose interval is 0.1 km/s in (a). Before the opening of the Japan Sea, later modified by Yamakita and Otoh [2000] (b). First stage of the formation process of the Basin area (c), and second stage of the formation process of the Basin area (d). NEJ, SWJ and YR in (b, c and d) show the northeastern and southwestern Japan Island Arc, and Yamato Rise, respectively. Reference Yamakita, S., and S. Otoh, Cretaceous rearrangement processes of pre-cretaceous geologic units of the Japanese Islands by MTL-Kurosegawa left-lateral strike-slip fault system (in Japanese with English abstract), Mem. Geol. Soc. Jpn., 56, 23-38, 2000. Original paper: Sato, T., et al., Geochem. Geophys. Geosyst., 7, Q06004, doi:10.1029/2005gc001135, 2006 1-2 Subduction factory study Crustal structure and evolution of the Mariana intra-oceanic island arc We proposed a new high-resolution velocity model of the Mariana arc-backarc system obtained from active-source seismic profiling demonstrating velocity variations within the arc middle and lower crusts of intermediate to felsic and mafic compositions. The characteristics of the oceanic island-arc crust are a middle crust with a velocity of about 6 km/s, and a laterally heterogeneous lower crust with velocities of about 7 km/s and unusually low mantle velocities (Fig. 1-2-1). Petrologic modelling suggests that the volume of the lower crust composed of restites and olivine cumulates after the extraction of the middle-crust should be significantly larger than is observed, suggesting that a part of the lower crust, especially the cumulates, is seismically a part of the mantle. Fig. 1-1-6. P-wave velocity structure from the southwestern Yamato Basin, the Oki Ridge toward the southwestern Oki Trough (a) and a simplified sketch of the formation process of the Yamato Basin area (b, c and d). The vertical and horizontal axes of (a) are the depth from the sea surface and the distance from the north end of the survey line, Fig. 1-2-1. Active seismic experiment carried out around the Mariana arc-backarc system. (a) Bathymetry of the Mariana arc-backarc system around our wide-angle seismic profile. The thick black line and open circles indicate locations of the airgun shooting line and OBSs, respectively. Numerals denote the site numbers of OBSs. Inset shows our profile superimposed on bathymetry of the entire IBM arc, and locates the northern Izu arc (NIA), Kyushu Palau Ridge (KPR), Parece Vela Basin (PVB), West Mariana Ridge (WMR), Mariana Trough (MT), Mariana arc (MA) and Mariana Trench. Open arrows indicate boundaries between the arc and backarc areas used to calculate crustal volumes. (b) Final velocity model of Mariana arc-backarc system and its resolution. (A) Final model. Solid circles indicate OBS locations. Numerals denote P-wave velocity (Vp) (km/s). Open arrows indicate boundaries between the arc and backarc areas used to calculate crustal volumes. UC, OL2 and OL3 are upper crust and oceanic layers 25
2 and 3, respectively. Dark shading shows region of seismic ray penetration; bsl-below sea level. (c) Resolution of the final model. The color scale with contour spacing of 0.1 indicates resolutions of velocity nodes; and those of the interface nodes are shown by the size of the open circles. Red circles indicate OBS locations. average velocities (~7.1 km/s) were obtained beneath rhyolitic volcanoes (e.g., Nii-jima, Kurose, South-Hachijo caldera, Myoji Knoll, and South Sumisu caldera). We concluded from these observations that continental crust grows predominantly beneath the basaltic volcanoes of the Izu arc, and that rhyolitic volcanism may be indicative of a more juvenile stage of crustal evolution, or re-melting of pre-existing continental crust, or both. Table 1 Comparison of seismologically measured crustal volumes. One petrologic scenario for crustal growth is that tonalitic middle crust is produced by the anatexis of the basaltic lower crust that is produced by differentiation of a primary, mantle-derived basaltic magma. This anatexis model allows us to calculate the expected volumes of restites and cumulates (together forming the petrologic lower crust) from the seismologically measured volume of the upper and middle crusts. A crucial observation made in all the three arcs is that the seismologically measured volume of the lower crust is much smaller than the petrologically inferred combined volume of the restites and cumulates. Original paper: Takahashi et al., Geology, in press. Fig. 1-2-2 (a) Final seismic velocity model. Shaded area indicates the poorly resolved area identified by the checkerboard test. (b) Average wt.% SiO2 of volcanic rocks sampled and dredged from Quaternary volcanoes. Data are described in Tables 1 and 2. The horizontal axis indicates distance from the northern end of the profile. (c) Average crustal velocity and thickness relationships. Black line, average crustal seismic velocity below the middle crust; red line, thickness of the middle crust; dotted black line, average crustal velocity excluding the lower part of the lower crust; black horizontal line within pink band, average crustal velocity (and standard deviation) of typical continental crust. (d) Map showing location of wide-angle seismic profile. Red circles show OBSs used in this study. Orange circles indicate OBSs that malfunctioned and provided unusable data. Os, Oh-shima; Nij, Nii-jima; Myk, Miyake-jima; Mkr, Mikura-jima; Krs, Kurose Hole; Hcj, Hachijo-jima; Shc, South Hachijo caldera; Ags, Aoga-shima; Myn, Myojin Knoll; Myj, Myojin-sho; Sms, South Sumisu; Ssc, South Sumisu caldera; Trs, Torishima; SB, Shikoku Basin; KS, Kinan seamount chain; PB, Parece Vela Basin. Original paper: Kodaira et al., Jour. Geophys. Res., in press Seismological evidence for variable growth of crust along the Izu intra-oceanic arc The processes that create continental crust in an intra-oceanic arc setting are a matter of debate. To address this issue we conducted an active source wide-angle seismic study to examine along-arc structural variations of the Izu intra-oceanic arc. The data used were acquired from a 550-km-long profile along the volcanic front from Sagami Bay to Tori-shima (Fig. 1-2-2(d)). The obtained structural model showed the existence of felsic to the intermediate composition middle crust with a P wave velocity (Vp) of 6.0 6.5 km/s in its upper part and 6.5 6.8 km/s in its lower part (Fig. 1-2-2). The thickness of the middle crust varied markedly from 3 to 13 km. The underlying lower crust also consisted of two layers (Vp of 6.8 7.2 km/s in the upper part and Vp of 7.2 7.6 km/s in the lower part). The upper of these layers was interpreted to consist of plutonic gabbro, and the lower layer to be mafic to ultramafic cumulates. Average crustal velocities calculated from our model showed remarkable lateral variations, which correlated well with arc volcanism. Low average crustal seismic velocities (~6.7 km/s) due to thick middle crust were obtained beneath basaltic volcanoes (e.g., O-shima, Miyake-jima, Hachijo-jima, Aoga-shima), while higher 2. Lithosphere Dynamics Research 2.1 Mechanisms of separation of the rupture area and variations in time intervals and size of great earthquakes along the Nankai Trough, southwest Japan According to the more than 1000-year earthquake history along the Nankai Trough, two great interplate earthquakes occurred east and west off Kii Peninsula within a few years of each other, and such earthquakes have occurred repeatedly at 26
intervals of every 100-200 years. Recent seismic structure surveys along the Nankai Trough reveal that there are some heterogeneous structures several tens of kilometers in scale at the boundary of the separated rupture areas. We consider that such large-scale heterogeneous structures may control the rupture area separation. To investigate the mechanisms of rupture area separation and also time- and size-variations of great earthquakes along the Nankai Trough, we conducted large-scale numerical simulation of earthquake generation cycles. Heterogeneous distribution in frictional properties is assumed based on the heterogeneous structures obtained by seismic surveys. The results show that rupture area separation can be reproduced within a reasonable range with frictional parameters. Significant variations in recurrence time and earthquake size occur only when large fracture energy areas exist in the deeper portion of the seismogenic zone. The existence of such areas causes heterogeneous stress distribution after one earthquake cycle. The variation pattern obtained here is similar to that of the last three earthquake cycles in the historical data. Great earthquakes, whose magnitude (M) is equal to or more than 8.0, have repeatedly occurred along the southern coast of southwest Japan, where the Philippine Sea Plate is subducting beneath it from the Nankai Trough (e.g. Ando, 1975). The history of earthquake generation cycles can be traced back more than 1000 years (Ishibashi, 2004). A typical earthquake occurrence pattern in this area is that two great earthquakes occur one after another within a few years, while recurrence intervals are 100-200 years (Fig. 2-1-1). The recurrence interval and size have changed somewhat systematically in recent cycles. The recurrence intervals and earthquake sizes have become smaller over time while the time intervals between Tokai (C+D+E) or Tonankai (C+D) and Nankai (A+B) earthquakes become longer. For both the 1944 Tonankai and 1946 Nankai earthquakes, rupture started from off Kii Peninsula and propagated unilaterally in the east and west, respectively (Kanamori, 1972). Fig. 2-1-1 Space-time distribution of great earthquakes along the Nankai-Suruga Trough (Ishibashi, 2004). Roman and italic numerals indicate earthquake occurrence years and time intervals between two successive series, respectively. Thick solid, thick broken, and thin broken lines show certain, probable, and possible rupture zones, respectively. Thin dotted lines indicate unknown. Hori et al. (2004) constructed a model of earthquake generation cycles on a large fault plane of 691km x 307.2km, which represents the subducting Philippine Sea Plate beneath southwest Japan from Tokai to Shikoku. Their model also included depth variations in frictional properties. Although heterogeneity in frictional properties related to large-scale (~100km) variations of plate geometry is included as depth variations of frictional parameters, ruptures propagate through the whole area with every earthquake cycle. Here, we introduce smaller scale (~50km) heterogeneity in frictional properties based on the results of structure surveys and examine whether rupture area separations can be reproduced or not. Moreover, mechanisms for variations in recurrence intervals and earthquake size are also investigated. The modeling procedure is described in Hori et al. (2004) or Hori (2006). The frictional parameters a,b and L, the effective normal stress " is assumed to be depth-dependent. The region where a " b is negative corresponds to the seismogenic zone, where stick and slip occur. The maximum value of b " a = 0.5 #10 "3 ( ) is determined so as to fit the average recurrence time interval (117 years) of historical earthquakes along the Nankai Trough. The characteristic slip distance, L, which is a memory distance over which the contact population changes, is constant in shallower portions of the seismogenic zone, but larger in the deeper parts, based on laboratory experiments. The effective normal stress is given by " = (# $ # w )gz, where ", " w, g and z are rock density, water density, gravity acceleration and depth. The former three values are 2.8 "10 3 kg m 3, 1.0 "10 3 kg m 3 and 9.8m s 2, respectively. Recent dense structural surveys along the Nankai Trough reveal that there are significant heterogeneities in seismic structures in the rupture segmentation boundaries. Thus, other than the depth dependence of frictional properties above, we introduce here heterogeneous frictional properties based on structural heterogeneity as shown in Fig. 2-1-2. 27
the Nankai Trough, southwest Japan, J. Earth Simulator, 5, 8-19, 2006. Hori, T., N. Kato, K. Hirahara, T. Baba, and Y. Kaneda, A numerical simulation of earthquake cycles along the Nankai Trough, southwest Japan: Lateral variations in frictional properties due to the slab geometry controls the nucleation position, Earth Planet. Sci. Lett., vol.228, pp.215-226, 2004. Ishibashi, K. Status of historical seismology in Japan, Ann. Geophys., vol.47, pp.339-368, 2004 Kanamori, H. Tectonic implications of the 1944 Tonankai and the 1946 Nankaido earthquakes, Phys. Earth Planet. Inter., vol.5, pp.129-139, 1972. Original paper: Hori et al., J. Earth Simulator, 5, 8-19, 2006. Hori et al., Earth Planet. Sci. Lett., 228, 215-226, 2004. Fig. 2-1-2 (a) The distribution of the value of at each cell. (b) The distribution of the value of L. ( b " a)# Fig. 2-1-3 Distribution of coseismic slip (slip velocity higher than 1 cm/s). Each numeral and time unit accompanied by an arrow shows the time interval between two earthquakes. Significant variations in the recurrence times of each event and the time intervals between the Tonankai and Nankai events are seen. All the coseismic slip patterns are shown in Fig.2-1-3. The time interval between events and event sizes seem to change systematically. After the Tonankai and Nankai events, the time intervals for which are significantly short, the time interval becomes longer for the following two earthquake cycles. The recurrence interval and size of the Nankai event also systematically changes. The recurrence interval becomes shorter over the two cycles and the seismic moment becomes smaller. This calculated variation pattern in the time interval and size of earthquakes (Fig.2-1-3) is similar to the historical earthquakes in 1707 (Fig. 2-1-1). References Ando, M. Source mechanisms and tectonic significance of historical earthquakes along the Nankai Trough, Japan, Tectonophysics, vol.27, pp.119-140, 1975. Hori, T., Mechanisms of separation of rupture area and variations in time intervals and size of great earthquakes along 2.2 Shear resistance reduction due to vibration in simulated fault gouge In order to investigate the mechanisms of dynamic triggering for earthquakes or creep events on natural faults with gouge layers, the frictional behavior of granular materials and their sensitivity to vibrational disturbances were examined by means of a direct shear test with crushed quartz sand and precise measurement equipment. The disturbances were created by light tapping on the lower part of a shear box. No displacement was observed from the tapping without shear load. From a series of experiments we found an acceleration in horizontal displacement just after a small vibration under shear. The acceleration indicates reduced shear resistance in the gouge layer, by about 3 % of the resistance just before the small vibration. This reduction in shear resistance seemed to recover quickly and did not affect the long-term behavior of the gouge layer. The response of the vertical displacement to the small vibration depends on the amount of accumulated dilatation of the gouge layer. The mechanism of shear resistance reduction and the variable response of the vertical displacement due to the vibration can be explained by the intrinsic feature of pillar-like structure of a force chain network in the gouge layer. Our results indicate that there might be dynamic triggering of fault motion due to shear resistance reduction of the gouge layer in a natural fault zone. It is well known that the presence of a gouge layer significantly effects the mechanical behavior of faults. Indeed, a vast amount of work has been devoted to characterize the frictional behavior of a gouge layer under various conditions [e.g., Marone, 1998]; however, the frictional behavior of a gouge layer under vibration has not been extensively studied. Some researchers have reported that an increase in seismicity or a dynamic rupture of a fault was triggered by seismic waves (e.g., Gombers et al., 2001). Thus it is very important to know how an elastic wave or a dynamic stress change affects the frictional behavior of a gouge layer and whether a dynamic stress change reduces or increases the frictional strength of the fault. A gouge layer of dry crushed quartz sand was directly sheared between two blocks made of brass, labeled upper (UB) and lower (LB) (Fig 2-2-1). A linear motor (LM) moved horizontally with a given constant velocity and pushed the 28
upper block via a stainless-steel plate (SP). The simulated gouge layer had an area of 136 70.5 mm 2 and a thickness of 2 mm. The particle size of the crushed quartz sand was between 0.15 and 0.25 mm. After the gouge layer was fully compacted, the shear load was applied to the upper block. While the shear load was applied, one side of the lower block was struck with a spherical point impactor made of brass, with a kinetic energy of 3 10-2 J. The impact energy was so small that no displacement of the upper block both for horizontal and vertical directions was observed in the absence of shear loading conditions. Fig.2-2-1 Schematic illustration of the apparatus. LM, linear motor; UB, upper block; LB, lower block; SP, stainless-steel plate; HDT, horizontal displacement transducer; GS, gap sensor; PZT, piezo-electric transducer; LC, load cell; GL, gouge layer. The cross indicates the impact point. This figure was modified from Yoshioka and Sakaguchi [2006]. We first conducted direct shear tests of the gouge layer without impact during the shear loading. A typical example of the tests is shown in Fig.2-2-2a. Fig. 2-2-2b shows a typical example of the tests in which a series of impacts were given during shear loading. The entire experiment is divided into four stages based on amplitude variations of horizontal and vertical steps caused by the impacts. Except for the existence of the spikes, which are the responses to the impacts, the overall shape of the shear load and displacements in Fig.2-2-2b seems to be identical to the one without impact, shown in Fig.2-2-2a. This correspondence indicates that although each impact instantaneously reduces the shear resistance, the gouge layer soon recovers the original resistance before the impact. In Fig. 2-2-3, we define a step due to an impact as a difference in the position of the upper block averaged for 0.5 seconds before and after the impact for both horizontal and vertical directions. Based on the regional subdivision in this diagram, we divided the stages in Fig. 2-2-2b into II, III and IV. Stage I is the period with no significant step. Stage II is defined as the period where both the vertical and horizontal steps monotonically increase. In stage III, the horizontal steps increase slightly with some fluctuations, as the vertical steps decrease gradually. Finally, in stage IV, the horizontal steps oscillate about a certain mean value around 0.035mm, and the vertical steps converge on the negative value, -0.002mm. Fig.2-2-2 (a) A typical example of tests for shearing without impact. Shear load acting on the upper block, and the resulting horizontal and vertical displacements of the upper block are shown. Vertical displacement is magnified ten times. Three sudden decreases in load and steps in the displacements show small spontaneous instabilities. (b) A typical example of tests for shearing with a series of impacts. The same variables as (a) are shown. The broken lines and Roman numerals indicate the stages mentioned in the text and in Fig 2-2-3. Sudden decreases in load and steps in the displacements show the instabilities caused by the impacts. 29
Fig.2-2-3 Detailed relationship between the horizontal steps and the vertical steps for the case in Fig.2.2.2b. The roman numerals with colored background correspond to the stages in Fig. 2-2-2b. Since the impact does not work to give extra force to push the upper block, it is natural to assume that the impact reduced the gouge layer shear resistance to some extent to mobilize the horizontal step of the upper block. We estimated this shear resistance reduction based on the assumption that the acceleration of the upper block is equal to the true shear resistance reduction divided by the mass of the upper block. The true shear resistance reduction due to an impact is about 3 % of the total shear resistance from the horizontal displacement data. This result is common for both stages III and IV (Figs. 2-2-4 a and 2-2-4b). However, in terms of the vertical displacement, as plotted by the red line in Figs. 2-2-4a and 2-2-4b, the response for an impact is significantly different between stages III and IV. For stage III, the upper block rises (positive dilatancy) 0.004 mm for 0.02 seconds after the impact. On the other hand, the upper block sinks (negative dialatncy) for stage IV. Nonetheless, it should be noted that both stages III and IV show negative dilatancy for 0.003 seconds immediately after the impact. Fig.2-2-4 (a) Variation of the horizontal and vertical displacements just before and after an impact given in stage III with high sampling rate (5kHz). Vertical displacement is magnified five times. Variation of PZT before and after the impact is shown below. (b) The same variables are shown for stage IV. 30 All the findings mentioned above cannot be explained by a simple macroscopic friction model, and suggest to us the existence of some specific microstructures in the sheared gouge layer. According to previous studies, granular materials under shear form an inhomogeneous contact network, which carries most of the external load by way of strong force chains. Among many features of force chain microstructure in sheared granular materials, the most important fact is that a force chain network consists of many pieces of one dimensional pillar-like structure in the direction of compressive forces between particles. In the gouge layer of our experiment, for example, this pillar-like structure develops mainly in the direction of the superposition of gravitational force and shearing force. As a
result, it facilitates to resist compaction and to shear simultaneously. Why is this relatively trivial fact important? Because a one-dimensional pillar-like structure is stiff only in the direction of axial compression. To force in the other direction, it is sensitive to buckling or failure, especially when it is long or has less support. This intrinsic feature of the pillar-like structure explains all the interesting findings in our experiments (see Hori et al., 2006). References Gomberg, J., P. A. Rosenberg, P. Bodin, and R. A. Harris (2001) Earthquake triggering by seismic waves following the Landers and Hector Mine earthquakes, Nature, 411, 462-466. Hori, T., H. Sakaguchi, N. Yoshioka, and Y. Kaneda (2006) Shear resistance reduction due to vibration in simulated fault gouge, in Radiated Energy and the Physics of Earthquake Faulting, ed. by R. E. Abercrombie, A. McGarr, G. Di Toro, and H. Kanamori, Geophyscal Monograph Ser., 170, AGU, Washington D.C., 135-142, 2006. Marone, C. (1998), Laboratory-derived friction laws and their application to seismic faulting, Annu. Rev. Earth Planet. Sci., 26, 643-696. Yoshioka, N. and H. Sakaguchi (2005), An Experimental Trial to Detect Nucleation Processes by Transmission Waves, Proc. AOGS, 2 nd Meeting, in press. 3. The pseudotachylyte is derived from the frictional melting of an illite-rich ultracataclasite layer. The variation in the volume fraction of unmelted grains in the pseudotachylyte matrix primarily represents the difference in the initial volume fraction of illite in the ultracataclasite layer prior to frictional melting. The minimum melting temperature is 1100 C, which is ~850 920 C greater than the maximum temperatures recorded in the host rocks. 4. The viscosity and shear resistance of the melt layer are very low (Fig. 2-3-2); therefore, the dynamic weakness of the fault, acceleration of seismic slip, and propagation of instability can possibly occur during an earthquake. This would contribute, at least locally, to the efficiency with which stored strain energy is released and hence to the earthquake magnitude in subduction zones. 5. Such frictional melting of the illite-rich slip zone may be applicable to subduction thrusts and faults in other accretionary complexes. The melting of the illite-rich slip zone is likely to form a hydrous melt layer, possibly leading to a high H 2 O content in these pseudotachylytes. Original paper: Hori et al., Geophysical Monograph, 170, AGU, Washington D.C., 135-142, 2006. Yoshioka and Sakaguchi, Proc. AOGS, 2 nd Meeting, in press 2.3 Pseudotachylytes in an ancient accretionary complex and implications for melt lubrication during subduction zone earthquakes Pseudotachylytes (i.e., solidified frictional melts produced during seismic slip) have recently been discovered in ancient accretionary complexes. They are considered to have developed in underplated rocks at seismogenic depths and thus may help our understanding of the dynamics of earthquake faulting in subduction zones. However, descriptions of pseudotachylytes in accretionary complexes are as yet severely limited; there is little information on their microstructures, compositions, and melting temperatures. At present, it is uncertain how the frictional melting of subducted material affects seismic slip in a subduction zone. Therefore, we analyzed pseudotachylytes from the Shimanto accretionary complex of eastern (Mugi area) and western (Okitsu area) Shikoku, southwest Japan. The conclusions obtained by our study are shown as follows: 1. At seismogenic depths in the subduction zone, a cataclastic thrust zone develops at the top of the mélange, and the coseismic slip is concentrated into a narrow zone less than a few millimeters thick. 2. The pseudotachylyte displays a fragment-laden, glass-supported texture resulting from rapid cooling of the frictional melt (Fig. 2-3-1), which is consistent with a very short cooling time of the melt layer calculated using thermal modeling. The rapid cooling of the melt layer is due to its narrow thickness, resulting in the fast healing of the coseismic slip zone by the solidified melt layer. 31 Fig. 2-3-1. Evidence of frictional melting recorded in the pseudotachylytes in the Shimanto accretionary complex. (a) Backscattered electron image of the pseudotachylyte, characterized by vesicles (black arrow) and embayed grains in the homogeneous matrix. Kf: K-feldspar. (b) Transmission electron micrograph of pseudotachylyte (bright-field image). Note the presence of euhedral microcrystals in a glassy matrix. The lower-right corner shows the diffraction pattern of the microcrystals. Fig. 2-3-2. Estimated viscosity and shear resistance of the melt layer. (a) Viscosity versus temperature for seven samples from the Mugi and Okitsu areas. (b) Shear stress versus effective normal stress for the melt layer, compared with Byerlee s frictional strength. Reference Ujiie, K., Yamaguchi, H., Sakaguchi, A., Toh, S., 2007. Pseudotachylytes in an ancient accretionary complex and
implications for melt lubrication during subduction zone earthquakes. Journal of Structural Geology 29, in press. Original paper: Ujiie K. et al., Journal of Structural Geology 29, in press. 2.4 Estimation of apparent elastic strain of the rock after stress release Elastic deformation is defined as deformation that rebounds to its original shape completely when stress is removed. Estimating the elastic strain after stress release is an unanswerable question; however, if elastic strain meter is found, it will be applied across broad fields, such as seismology, tectonic research and concrete engineering. An earthquake is an example of elastic rebound along a fault in the Earth's crust. An estimation of accumulated strain energy along the fault helps us to understand an earthquake s energy. A quantitative discussion between stress and rock deformation will enhance structural geology. An estimation of the inner stress condition of concrete will help to build solid construction. We propose a method by which to estimate elastic strain of the rock after stress release. A rock is composed of various mineral grains of various strengths, and some weak grains deform plastically during apparent elastic deformation. A calcite is a weak mineral that undergoes intracrystalline planar dislocation to make twin responses to an applied stress (Jamison and Spang, 1976; Rowe and Rutter, 1990; Ferrill, 1998), and this twin density has been used as a paleo-stress indicator for pure-calcite rock in plastic deformation. This method, however, has never been applied to elastic rock composed of various mineral grains. It was thought that the inner-stress field of mixed granular materials must be more complicated than the pure-calcite rock (Burkhard, 1993). A computer simulation model shows that inner-stress fields are complicated, even pure-calcite rock (plastic) and mixed rock composed of rigid grains (elastic). The mean stress value area, including heterogeneous stress distribution, is in response to the loading stress of the entire rock. Actual twin behavior within sandstone was documented by tri-axial compression experiments. The mean value of calcite twin density area, including heterogeneous twin distribution, can be obtained by multi-point analysis. The mean value of twin density was proportional to loading stress, except when there was much higher stress before failure. Twin density can become the stress indicator during elastic deformation, and obtained stress and elastic modules of the rock can estimate apparent elastic strain (Fig.2-4-1). A B Fig. 2-4-1 The rock s inner stress field was considered by computer simulation experiments using discrete element modeling (A, B). A virtual rock sample is composed of 100,000 particles and is compressed under confining pressure (A). Obtained inner stress field is very complicated and heterogeneous; however, the average value of the stress area, including heterogeneity, can be represented by the whole loading stress of the sample (B). Actual calcite twin density is investigated by tri-axial experiments. Average twin density increases with loading stress (C). The positions of the black dots are analyzed to create the twin density distribution map (D). References Burkhard, M. Calcite twins, their geometry, appearance and significance as stress-strain markers and indicators of tectonic regime: a review, Journal of Structural Geology, 15, p 351-368, 1993. Ferrill, D. Critical re-evolution of differential stress estimates from calcite twins in coarse-grained limestone, Tectonophysics, 285, p 77-86, 1998. Jamison, R. W. and Spang, H. John. Use of calcite lamellae to infer differential stress, Geological Society of America Bulletin, 87, p 868-872, 1976. Rowe, K. J. and Rutter, E. H. Palaeostress estimation using calcite twinning: experimental calibration and application to nature, Journal of Structural Geology, 12, p 1-17, 1990. Original paper: Arito Sakaguchi, Hide Sakaguchi, Masao Nakatani and Shingo Yoshida, Estimation of apparent elastic strain of the rock after stress release (in preparation) 2.5 The afterslip distribution of the 2003 Tokachi-oki earthquake estimated from ocean-bottom pressure gauges data A major shortcoming concerning studies of fault slip associated with subduction zone earthquakes is the lack of offshore, near field data. The lack of offshore geodetic data is an especially serious problem when studying slow slip, such as afterslip following a subduction zone earthquake. To overcome this problem, the Japan Agency for Marine-Earth Science and Technology (JAMSTEC) has developed cabled seafloor observatory systems equipped with high-precision pressure gauges that transmit data in real time to land-based observatories via optic fiber (Hirata et al., 2002; Monma et al., 1997). Measurement of the water pressure on the seafloor is an C D 32
innovative geodetic observation because the pressure increase is directly proportional to vertical displacement of the seafloor. During September 2003, a M w 8.0 earthquake (hereafter, the 2003 Tokachi-oki earthquake) occurred very close to this observatory system. This event was an interplate earthquake that ruptured the plate boundary between the Pacific Plate and the North American Plate. Two ocean-bottom pressure gauges of the observatory system successfully recorded data before, during, and after the earthquake (Fig. 2-5-1). This is the first offshore, near-field data recorded for a M8-class interplate earthquake at a subduction zone, and provides the first opportunity to use such observations to investigate the slow slip behavior on an offshore boundary following a major earthquake. Here, we estimated the offshore afterslip distribution by using this data combined with onshore GPS data. Fig. 2-5-1. Daily averages of water-pressure-residual (circles) and temperature (crosses) fluctuations for ocean-bottom pressure gauges. One-sigma standard deviation bars are also shown. The day of the 2003 Tokachi-oki earthquake is set as the origin (time=0) on the time axis. Tide components have already been removed. A water depth change of 100 cm is approximately equivalent to a pressure change of 10 4 Pa. The afterslip distribution shown in Fig. 2-5-2a was estimated by using only the onshore GPS data. Significant afterslip zones of more than 0.45 m appear to the east and west of Cape Erimo. Small (< 0.3 m) or non-slip subfaults are imaged between the two significant afterslip zones. The maximum slip was estimated to be 0.8 m on the eastern significant slip zone. Small slips less than 0.3 m extend extensively trenchward. In the simultaneous inversion estimated by using onshore GPS and offshore PG data (Fig. 2-5-2b), a slip distribution similar to Fig.2-5-2a was retrieved other than around PG stations. In the offshore part near the PG stations, the afterslip does not extend trenchward when compared with Fig. 2-5-2a. This difference implies that land-based geodetic data alone are not enough to resolve offshore slip. The simultaneous inversion of the onshore GPS and offshore PG data suggests that the afterslip occurs in a U-shaped pattern with a width of about 200 km. A large afterslip area, with slip amounts of almost 0.9 m, is imaged on the eastern side of the afterslip region, and slip amounts of about 0.4 0.7 m were estimated on the southern and western sides. Fig. 2-5-2 One-year afterslip distributions estimated from (a) onshore GPS data only and (b) both onshore GPS and offshore PG data. The color scale bar shows the slip amount on the subfaults. Green circles show the observation points. A U-shaped afterslip distribution appears after the GPS & PG inversion analysis. Contours in (b) show the 2003 coseismic slip distribution, with a contour interval of 0.6 m, estimated from seismic inversion [Yagi, 2004]. Coseismic slips of large subduction zone earthquakes are means for releasing strain energy accumulated from interplate contact due to relative plate motion. This coseismic slip is associated with stick-slip frictional properties from shallow interplate contact. Slow slip on the plate interface, such as afterslip or episodic slip, also releases accumulated strain energy. A numerical simulation assuming a rate- and state-dependent friction law showed that afterslip occurs in a stable sliding region [Yoshida and Kato, 2003]. As shown in Fig. 2-5-2b, the afterslip of the Tokachi-oki earthquake is distributed in a U-shaped pattern encircling its coseismic slip area. This shows that stick-slip frictional properties that yield coseismic ruptures are non-uniformly distributed along the plate interface in the southern Kuril subduction zone. Such a patchy distribution has been suggested by previous works [Lay and Kanamori, 1981; Song and Simons, 2003], but to our knowledge has never been verified through direct observation of the afterslip, as has been achieved in this study. References Hirata, K., M. Aoyagi, H. Mikada, K. Kawaguchi, Y. Kaiho, R. Iwase, S. Morita, I. Fujisawa, H. Sugioka, K. Mitsuzawa, K. Suehiro, H. Kinoshita, N. Fujiwara, Real-time geophysical measurements on the deep seafloor using submarine cable in the southern Kuril subduction zone, IEEE J. Ocean Eng. 27 (2002) 170-181. Lay, T. and H. Kanamori, An asperity model of great earthquake sequences, in: D.W. Simpson, P.G. Richards (Eds), Earthquake prediction; an international review, Maurice Ewing Series 4, 1981, pp. 579-592. Monma, H., N. Fujiwara, R. Iwase, K. Kawaguchi, S. Suzuki, H. Kinoshita, Monitoring system for submarine earthquakes and deep sea environment, Proc. MTS/IEEE OCEANS 97 2 (1997) 1453-1459. Song, T.A. and M. Simons, Large trench-parallel gravity variations predict seismogenic behavior in subduction zones, Science 301 (2003) 630-633. Yagi, Y., Source rupture process of the 2003 Tokachi-oki earthquake determined by joint inversion of teleseismic body 33
wave and strong ground motion data, Earth Planets Space 56 (2004) 311-316. Yoshida, S. and N. Kato, Episodic aseismic slip in a two-degree-of-freedom block-spring model, Geophys. Res. Lett. 30 (2003) doi:10.1029/2003gl017439. 34