Characterization of slip zone associated with the 1999 Taiwan Chi-Chi earthquake: X-ray CT image analyses and microstructural observations of the Taiwan Chelungpu fault

Available online at www.sciencedirect.com

Tectonophysics 449 (2008) 63 – 84 www.elsevier.com/locate/tecto

Characterization of slip zone associated with the 1999 Taiwan Chi-Chi earthquake: X-ray CT image analyses and microstructural observations of the Taiwan Chelungpu fault Tetsuro Hirono a,⁎, Masumi Sakaguchi b , Kenshiro Otsuki c , Hiroki Sone d , Koichiro Fujimoto e , Toshiaki Mishima f , Weiren Lin g , Wataru Tanikawa g , Masaharu Tanimizu g , Wonn Soh g , En-Chao Yeh h , Sheng-Rong Song h a

g

Department of Earth and Space Science, Graduate School of Science, Osaka University, Toyonaka, Japan b Marine Works Japan Ltd., Nankoku, Japan c Department of Geoenvironmental Science, Graduate School of Science, Tohoku University, Sendai, Japan d Department of Geophysics, School of Earth Sciences, Stanford University, Stanford, California, USA e Faculty of Education, Tokyo Gakugei University, Koganei, Japan f Research Center for Inland Seas, Kobe University, Kobe, Japan Kochi Institute for Core Sample Research, Japan Agency for Marine-Earth Science and Technology, Nankoku, Japan h Department of Geosciences, National Taiwan University, Taiwan Received 19 September 2006; received in revised form 15 November 2007; accepted 5 December 2007 Available online 14 December 2007

Abstract To characterize the fault-related rocks within the Chelungpu fault, we performed X-ray computed tomography (CT) image analyses and microstructural observations of Hole B core samples from the Taiwan Chelungpu-fault Drilling Project. We identified the slip zone associated with the 1999 Chi-Chi earthquake, within the black gouge zone in the shallowest major fault zone, by comparison with previous reports. The slip zone was characterized by low CT number, cataclastic (or ultracataclastic) texture, and high possibility to have experienced a mechanically fluidized state. Taking these characteristics and previous reports of frictional heating in the slip zone into consideration, we suggested that thermal pressurization was the most likely dynamic weakening mechanism during the earthquake. © 2007 Elsevier B.V. All rights reserved. Keywords: Chelungpu fault; Chi-Chi earthquake; X-ray CT; Microstructure; Fault zone

1. Introduction The Taiwan Chi-Chi earthquake (Mw 7.6; epicenter, 23.853 °N, 120.816 °E; focal depth, 8 km) occurred on 21 September 1999 (Ma et al., 1999) (Fig. 1). The earthquake initiated on the southern Chelungpu fault and ruptured both upward and laterally northward (Chen et al., 2001; Kikuchi et al., 2000; Ma et al., 2000). This earthquake was well recorded by a dense seismic observatory ⁎ Corresponding author. Department of Earth and Space Science, Graduate School of Science, Osaka University, Toyonaka 560-0043, Japan. Tel.: +81 6 6850 5796; fax: +81 6 6850 5480. E-mail address: [email protected] (T. Hirono). 0040-1951/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.tecto.2007.12.002

network (Taiwan Strong Motion Network) operated by the Central Weather Bureau (Shin et al., 2000). The stations near the northern end of the Chelungpu fault recorded the largest ground velocities and displacements, up to 3 m/s and 8 m, respectively (Shin and Teng, 2001). In contrast, ground accelerations were higher along the southern part of the fault, even though the ground velocities and fault displacements there were less than in the north. Ma et al. (2003) attributed the low level of high-frequency radiation to an elasto-hydrodynamic lubrication-type slip mechanism, in which increased lubrication pressure widened the gap between the fault surfaces, thereby reducing the areas of asperities in contact and decreasing the level of high-frequency radiations. Andrews (2005), however, proposed that the high slip velocity and large

64

T. Hirono et al. / Tectonophysics 449 (2008) 63–84

Fig. 1. Location of the Chelungpu fault with geological map, an E–W cross section through drilling site location, and fault zone architectures of three major fault zones (FZB1136, FZB1194, and FZB1243) in TCDP Hole B. CF, Chelungpu Fault; FDZ, fracture-damaged zone; BZ, breccia zone; GGZ, gray gouge zone; BGZ, black gouge zone; BM disk, disk-shaped black material.

displacement resulted from thermal pressurization, by means of which fluid pressure generated by shear-related heating reduced the fault strength during seismic slip.

The 2002 Taiwan Chelungpu-fault Drilling Project (TCDP) was undertaken to investigate the faulting mechanism of the 1999 Chi-Chi earthquake. One of the main aims of TCDP was

T. Hirono et al. / Tectonophysics 449 (2008) 63–84

65

to assess the above-mentioned hypotheses through observations and analyses of rock samples. TCDP penetrated the Chelungpu fault (Fig. 1) and recovered core samples from two holes, Hole A (total depth, 2003.00 m) and Hole B (total depth, 1352.60 m), which were 40 m apart at the surface. In Hole B, drilled core samples were recovered only from depths between 948.42 and 1352.60 m, and a series of three major fault zones, at depths of 1134–1137 m, 1194–1197 m, and 1242–1244 m, was identified within the Chelungpu fault (Hirono et al., 2006b). These zones were referred to as FZB1136 (fault zone around 1136 m depth in Hole B), FZB1194, and FZB1243, respectively (Hirono et al., 2007). The core samples from Hole B were sent to the Kochi Core Center, for nondestructive continuous physical property measurements of density, porosity, magnetic susceptibility, and natural gamma ray, and the results of these measurements were reported in Hirono et al. (2006b, 2007). To characterize the fault-related rocks within the Chelungpu fault, we performed X-ray computed tomography (CT) image analyses and microstructural observations of faultrelated samples from TCDP Hole B. In this paper, we report the characteristics of all three major fault zones in Hole B (FZB1136, FZB1194, and FZB1243), and consider the most recent slip zone associated with the 1999 Chi-Chi earthquake. We also discuss the faulting mechanism during the earthquake. 2. Methods 2.1. X-ray CT imaging X-ray CT (XCT) is a radiological imaging technique. The attenuation of two-dimensional X-ray fan-beams penetrating a sample is measured by an array of detectors. X-ray projection data from various directions are obtained by the 360° rotation of the X-ray source (Fig. 2a). A twodimensional image representing the linear distribution of X-ray attenuation is reconstructed using Fourier transformation of the projection data. A three-dimensional data set of the sample is obtained by stacking consecutive twodimensional images (Fig. 2b). The degree of X-ray attenuation depends on the density and the atomic number of the materials composing the samples. Material with higher density and higher atomic number generally cause higher attenuation of the X-rays. An X-ray CT scanner (Pratico, Hitachi Medical Co., Tokyo, Japan) was used for this study (Fig. 2a). The X-ray was produced by electrons striking a Mo-W alloy target in an X-ray tube. The electron current was 100 mA, the accelerating voltage was 120 kV, and the scan times were 4 s. The slice images were captured at 1-mm intervals (Fig. 2b), and each image was 1 mm thick. The resolution of the image was 0.31 × 0.31 cm, and thus the resulting voxel size was 0.31 × 0.31 × 1.0 mm. The output XCT images were digitized as DICOM-formatted 16-bit image files. Stacked images (Fig. 2c) were produced with VGStudioMAX software (Volume Graphics GmbH, Heidelberg, Germany).

Fig. 2. (a) Medical X-ray computed tomography scanner. (b) Capturing slice images of a core sample. (c) A reconstructed three-dimensional XCT image of a core sample.

The amount of attenuation in the XCT image is expressed as the CT number (Nct), which is defined as: Nct ¼ ðA  Aw Þ=Aw  1000 where µ is the linear X-ray absorption coefficient of the sample and µw is the linear absorption coefficient of pure water used as a standard reference. The CT number of water is defined as 0, and the CT number of air as − 1000. The CT number is a function of the density and chemical composition of the material in a voxel. 2.2. Optical microscope observation We made sequential thin sections from each major fault zones within the Chelungpu fault, and conducted the microstructural observation under an optical microscope. 3. FZB1136 3.1. Association with the 1999 Chi-Chi earthquake Kano et al. (2006) measured borehole temperatures in Hole A and observed a slight heat signal around FZA1111 (fault zone around 1111 m depth in Hole A; corresponding to FZB1136 in Hole B; correlation of the fault zones in

66

T. Hirono et al. / Tectonophysics 449 (2008) 63–84

Fig. 3. XCT images in and around the gouge zone in FZB1136. (a) Photo image. (b) Color values of CT numbers in XCT images of parts (d) and (f). (c) Color values of CT numbers in XCT images in parts (e) and (g). (d) and (e) XCT images of a cross section in a plane parallel to the direction of dip. (f) and (g) XCT images of a cross section in a plane perpendicular to the direction of dip. High CT number in the outer part of the core sample is a beam hardening artifact. LZ, zone with relatively low CT number; OF, open fracture; BS, bright spot. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Holes A and B is discussed later). They suggested that the heat signal had been produced by frictional heating during the 1999 Chi-Chi earthquake. However, they did not report the results of temperature measurements within the deeper two fault zones (FZA1153 and FZA 1222 in Hole A). On the other hand, Ma et al. (2006) found a 12-cm-thick primary slip zone in the shallowest fault zone in Hole C (corresponding to FZA1111 and FZB1136), and within the

zone identified a 2-cm-thick major slip zone associated with the Chi-Chi earthquake. Wu et al. (2007) found low seismic velocity, low electrical resistivity, and a major stress orientation anomaly around FZA1111 in geophysical logs of Hole A, supporting the inference that the shallowest fault zone was most likely related to the 1999 Chi-Chi earthquake. Thus, FZB1136 (FZA1111) is though to be most likely related to the 1999 Chi-Chi earthquake.

T. Hirono et al. / Tectonophysics 449 (2008) 63–84

67

3.2. Mesoscopic characteristics

3.4. Microstructural observations

In FZB1136, the following subzones were encountered, from top to bottom (Fig. 1): upper fracture-damaged zone (1134.40– 1134.93 m), upper breccia zone (1134.93–1136.22 m), upper gray fault gouge (1136.22–1136.26 m), black fault gouge (1136.26– 1136.40 m), lower gray fault gouge (1136.40–1136.46 m), lower breccia zone (1136.46–1136.70 m), and lower fracture-damaged zone (1136.70–1137.90 m). Both fracture-damaged zones display layer-parallel and -subparallel open fractures, which gradually increase in frequency toward the breccia and fault gouge zones, but which are sparse outside of the fault zones in the hanging wall and footwall, which are characterized by an alternation of sandstone and mudstone with bioturbation. In both breccia zones, the clasts are composed of very fine-grained sandstone or siltstone with minor shear bands, and the foliations within the matrixes become more intense closer to the gouge zones. Both black and gray gouge zones display foliation or random fabric textures, usually with a gradual transition from the black gouge zone to the gray gouge zone.

We made sequential thin sections from the black gouge zone, including the upper and lower boundaries with the gray gouge zones, in FZB1136 (Fig. 6). We observed all thin sections under an optical microscope, and found three zones corresponding to LZ1, LZ2, and LZ3 on the XCT image. Their representative characteristics are described below. LZ1 corresponds to the sedimentary lamination, observed mesoscopically, because of their same depth and position (Figs. 6 and 7a). Microscopically, the lamination is composed of an abundant fine-grained matrix supporting rounded to subangular lithic fragments, composed mainly of quartz, plagioclase, and potassium feldspar (Fig. 7b, c). The grain size ranges from clay to coarse silt (thus, 1/256–1/16 mm), which is slightly smaller than that of the surrounding zones. A bright layer with coarse silt-sized grains is observed within the lamination (Fig. 7b, c). Dark portions outside of the layer are composed mainly of clay-sized grains with fewer fine silt-sized grains (1/256–1/64 mm) (Fig. 7b, c). Fig. 7d shows the part at the same depth and position with LZ3 in the lower gray gouge zone. This part is composed of an abundant fine-grained matrix supporting subangular lithic fragments, composed mainly of quartz, plagioclase, and potassium feldspar. The grain size ranges from clay to coarse silt. A preferred orientation in shape of finer platy minerals is recognized, but sedimentary laminations are partially preserved (top-right in Fig. 7d). We thus conclude that LZ3 corresponds to a shear zone with a weak degree of deformation. Fig. 8a shows intense shear zone with 20 mm thick around the boundary between the black gouge and the lower gray gouge zones, at the same depth and position with LZ2. This zone corresponds to a 2-cm-thick major slip zone in Hole C by Ma et al. (2006), because of their spatial relationship. Microscopically, the shear zone (here named MSZ) is composed of an abundant fine-grained matrix supporting rounded to subangular lithic fragments, composed mainly of quartz, plagioclase, potassium feldspar, and clay minerals (Fig. 8b–d). The grain size ranges from clay to coarse silt, which is smaller than those of the surrounding zones. Fragmentation of mineral grains such as quartz, plagioclase, and potassium feldspar is locally observed. The upper boundary of MSZ with other portion within the black gouge zone is sharp, and the grain size reduces toward the contact. A preferred orientation in shape of platy fine grains is well developed within the shear zone adjacent to the boundary (Fig. 8c). MSZ also has a sharp contact with the lower gray gouge zone, but no distinct change in grain size or degree of shearing toward the boundary is observed. Anastomosing network structures with thin clay seams are observed primarily in the middle part of MSZ (Fig. 8d). The seams are oriented roughly parallel to the dip direction of MSZ. In the middle part, the matrix is composed of well-rounded clayey blocks. On the basis of these features, MSZ is recognized as cataclasite or ultracataclasite. We therefore conclude that LZ2 corresponds to cataclastic or ultracataclastic shear zone.

3.3. XCT images XCT images of FZB1136 are shown in Figs. 3–5, and the position of each image is indicated on Fig. 1. Each figure shows two cross sections, one in a plane parallel to the direction of dip (e.g., Fig. 3d) and another in a plane perpendicular to the direction of dip (e.g., Fig. 3f), which were reconstructed from each three-dimensional image with VGStudioMAX software. Each cross section in each image is shown with the CT number values expressed in two ways (e.g., Fig. 3d, e). The black gouge zone is characterized by pervasively lower CT numbers ranging approximately from 1700 to 2000, compared with other parts of the images (Fig. 3). In particular, three zones with significant lower CT numbers (LZ1, LZ2, and LZ3) are recognized. The dips of these zones are approximately 35°. Open fractures with relatively low CT numbers (OF) are identified in the black gouge zone and the lower breccia zone, because they have extremely lower CT numbers nearly zero, thus corresponding to the void. Clast and matrix structures are well recognized in both the upper and lower breccia zones (Fig. 3e, g). Bright spots with relatively high CT numbers (BS) are observed in the upper breccia zone (Fig. 3d, f). XCT images of the upper breccia zone are shown in Fig. 4. Some zones with relatively low CT numbers are observed in the lower core, while brecciated structures are well recognized in the upper core. Bright spots with relatively high CT numbers are also seen in the lower core (Fig. 4e, g). XCT images of the lower breccia zone and the lower fracture-damaged zone are shown in Fig. 5. The boundary between these zones is gradual, and brecciated structures within the breccia zone are not obvious. Some zones with relatively low CT numbers and an open fracture are recognized in both zones. A few bright spots with slightly high CT numbers are observed in the fracture-damaged zone (Fig. 5d, f).

68

T. Hirono et al. / Tectonophysics 449 (2008) 63–84

Fig. 4. XCT images in the upper breccia zone in FZB1136. (a) Photo image. (b) Color values of CT numbers in XCT images of parts (d) and (f). (c) Color values of CT numbers in XCT images of parts (e) and (g). (d) and (e) XCT images of a cross section in a plane parallel to the direction of dip. (f) and (g) XCT images of a cross section in a plane perpendicular to the direction of dip. High CT number in the outer part of the core sample is a beam hardening artifact. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

T. Hirono et al. / Tectonophysics 449 (2008) 63–84

69

Fig. 5. XCT images in the lower breccia zone and lower fracture-damaged zone in FZB1136. (a) Photo image. (b) Color values of CT numbers in XCT images of parts (d) and (f). (c) Color values of CT numbers in XCT images of parts (e) and (g). (d) and (e) XCT images of a cross section in a plane parallel to the direction of dip. (f) and (g) XCT images of a cross section in a plane perpendicular to the direction of dip. High CT number in the outer part of the core sample is a beam hardening artifact. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 9a shows the other two shear zones within the central part of the black gouge zone, which are not appeared on the XCT image. Both zones are composed of an abundant finegrained matrix supporting rounded to subangular lithic fragments, composed mainly of quartz, plagioclase, and potassium feldspar. The grain size ranges from clay to coarse silt, which is smaller than that of the surrounding zones.

Fragmentation of mineral grains is locally observed. On the basis of these features, both shear zones are recognized as cataclasite or ultracataclasite. On the other hand, lower shear zone, OSZ in Fig. 9a, is newer than the upper zone, SZ in Fig. 9a, as determined by their cross-cutting relationship. Elongated or diamond-shaped black clasts, which we have named BM clasts, are observed within OSZ (Fig. 9c). The BM

70

T. Hirono et al. / Tectonophysics 449 (2008) 63–84

Fig. 6. Photo images from in and around the black gouge zone of FZB1136. The left image was captured on the split surface of the core sample, and the small images on the right are of thin sections. MSZ, major shear zone; Weak SZ, a shear zone with a weak degree of deformation.

clasts are composed of clay-sized grains with a few fine-siltsized mineral grains. The clasts are well rounded, and partly in contact with shear foliation. 4. FZB1194 4.1. Mesoscopic characteristics In FZB1194, subdivisions observed from top to bottom are (Fig. 1) upper fracture-damaged zone (1194.00–1194.73 m),

black fault gouge (1194.73–1194.87 m), gray fault gouge (1194.87–1195.13 m), breccia zone (1195.13–1195.50 m), and lower fracture-damaged zone (1195.50–1197.25 m). The upper sandstone unit includes minor shear bands. The black fault gouge contains a disk-shaped black material (1194.73– 1194.75 m), referred to as the BM disk (Hirono et al., 2006b), which has been identified as pseudotachylyte with a low degree of melting (Hirono et al., 2006a). It is a 2 cm thick and relatively stiff compared with the gouge. The upper boundary between the BM disk and the adjacent upper cohesive sandstone unit is

T. Hirono et al. / Tectonophysics 449 (2008) 63–84

71

The XCT image of the BM disk shows a change in thickness, an intrusion into the upper fracture-damaged zone, indicated by the red arrow in Fig. 10e, and the development of cracks overprinting both the BM disk and the upper zone, indicated by blue arrows in Fig. 10e. The presence of overprinted cracks implies that the disk formed before the last earthquake, as has already been suggested by Hirono et al. (2006a). The dip of the BM disk is approximately 45°. An open fracture is observed around the boundary between the BM disk and the black gouge

Fig. 7. Photo (a) and optical microscope images (b, c) of lamination (LZ1) around the boundary between the gray gouge zone and the black gouge zone within FZB1136. (d) An optical microscope image of the shear zone with a weak degree of deformation (LZ3) in the gray gouge zone within FZB1136.

extremely sharp, whereas fragments of the disk are contained in the underlying black gouge. Other parts of the two gouge zones display foliation or random fabric texture. The breccia zone in FZB1194 is thinner than that in FZB1136, less shear banding or foliation seems to have developed within the zone itself, and the clasts are mainly fine-grained sandstone. The fracture-damaged zones display similar characteristics to those of FZB1136. 4.2. XCT images XCT images of FZB1194 are shown in Fig. 10, and the positions of the images are indicated on Fig. 1.

Fig. 8. Optical microscope images (a–c) of MSZ (LZ2) in the black gouge zone within FZB1136.

72

T. Hirono et al. / Tectonophysics 449 (2008) 63–84

Fig. 9. Photo (a) and optical microscope images (b, c) of the other two shear zones in the middle part of the black gouge zone within FZB1136.

zone, but its origin is seemed to be artificial as it was created during transport of the core sample, as indicated by the presence of the overprinted crack. CT values of the BM disk range approximately from 1900 to 2100, which are similar to those in the upper fracture-damaged zone. The black gouge zone is characterized by the presence of clasts with moderate CT numbers, similar to the upper zones, and of bright spots with relatively high CT numbers. Brecciation is also recognized within each clast within the zone. The gray gouge zone shows relatively low CT numbers, which range approximately from 1600 to 2000. Clasts are observed, and brecciation within each clast is also recognized. No intense distinct shear plane is observed within either gouge zone. 4.3. Microstructural observations We made sequential thin sections from the black gouge zone and BM disk, including the boundaries with the upper fracturedamaged zone and the gray gouge zones, in FZB1194 (Fig. 11). Their representative characteristics, observed under an optical microscope, are described below. Fig. 12a shows the sharp boundary between the upper fracture-damaged zone and the BM disk. The upper zone is composed of very fine sand-sized grains, which are composed mainly of quartz, plagioclase, and potassium feldspar. No shear fabric is observed within the upper zone. Some grains along the boundary with the BM disk have been fractured (Fig. 12b), which indicate that the boundary originated from brittle fracturing and subsequent contact.

The BM disk is composed of an abundant fine-grained matrix supporting well-rounded lithic fragments, which are composed mainly of quartz, plagioclase, and potassium feldspar. Foliations, indicated by the preferred orientation in shape of fine platy minerals, are observed parallel to the boundary with the upper zone (Fig. 12b). The grain size of the lithic fragments in the BM disk is extremely smaller than that in the upper zone, and ranges from fine silt to very fine sand. The microscopic features of the BM disk under optical microscope are consistent with its being identified as ultracataclasite or cataclasite. However, Hirono et al. (2006a) discovered hourglass structures around the edges of individual grains under scanning electron microscope, which might be evidence of quenching immediately after melting, and concluded that the BM disk was pseudotachylyte with a low degree of melting. In contrast, the boundary between the BM disk and the black gouge zone is not sharp (Fig. 12c), even though an artificial open fracture is observed on the XCT image (Fig. 10). Fragments of the BM disk, which are recognized around the top left of Fig. 12c, and flow structures are observed in the gouge. These features suggest that fragmentation might have originated from high pressure in the gouge. Based on the observed cross-cutting relationship, the BM disk is older than the black gouge zone, which supports the interpretation that the BM disk predated the latest seismic event (Hirono et al., 2006a). The black gouge zone is composed of an abundant matrix of clay-size grains supporting well-rounded lithic fragments, which are composed mainly of quartz, plagioclase, and

T. Hirono et al. / Tectonophysics 449 (2008) 63–84

73

Fig. 10. XCT images in and around the gouge zone in FZB1194. (a) Photo image. (b) Color values of CT numbers in XCT images of parts (d) and (f). (c) Color values of CT numbers in XCT images of parts (e) and (g). (d) and (e) XCT images of a cross section in a plane parallel to the direction of dip. (f) and (g) XCT images of a cross section in a plane perpendicular to the direction of dip. Red arrow indicates the intrusion of the BM disk into the upper fracture-damaged zone. Blue arrows indicate the development of cracks overprinting both the BM disk and the upper zone. High CT number in the outer part of the core sample is a beam hardening artifact. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

potassium feldspar (Fig. 12d). Intense foliations, indicated by the preferred orientation in shape of fine platy minerals, are observed parallel to the boundary with the BM disk. Elongated or diamond-shaped black clasts are observed in the black gouge zone (Fig. 12e), and they resemble the BM clasts in FZB1136.

The black gouge zone is thus characterized by cataclastic texture with BM clasts. On the other hand, the gray gouge zone does not dominantly exhibit a shear fabric (Fig. 12f). Sorted grains and a weak preferred orientation in shape of very fine platy minerals are

74

T. Hirono et al. / Tectonophysics 449 (2008) 63–84

Fig. 11. Photo images from in and around the black gouge zone of FZB1194. The left image was captured on the split surface of the core sample, and the small images on the right are of thin sections.

observed, suggesting that the origin of the fabric may be not shearing but sedimentary. 5. FZB1243 5.1. Mesoscopic characteristics In FZB1243, subdivisions observed from top to bottom are (Fig. 1) upper fracture-damaged zone (1242.70–1243.33 m),

upper gray fault gouge (1243.33–1243.38 m), black fault gouge (1243.38–1243.50 m), lower gray fault gouge (1243.50– 1243.60 m), and lower fracture-damaged zone (1143.60– 1244.30 m). The black fault gouge includes a disk-shaped black material (1243.38–1243.41 m), which is 3 cm thick and relatively stiff compared with the gouge, similar to that encountered in FZB1194. It was also identified as pseudotachylyte with a low degree of melting (Hirono et al., 2006a). Other parts of both gouge zones display foliation or random

T. Hirono et al. / Tectonophysics 449 (2008) 63–84

75

Fig. 12. Optical microscope and photo images from in and around the black gouge zone of FZB1194. (a, b) Boundary between the upper fracture-damaged zone and the BM disk. (c) Boundary between the BM disk and the black gouge zone. (d, e) BM clasts within the black gouge zone. (f) Gray gouge zone.

fabric texture. The fracture-damaged zones display similar characteristics to those in the other two fault zones. The overall scale of FZB1243 is smaller in comparison with the shallower two fault zones, as no breccia zone is present between the intact rocks and the fault gouge zones and the fault rocks are confined within a region approximately 25 cm thick. 5.2. XCT images XCT images of FZB1243 are shown in Fig. 13, and the positions of the images are indicated on Fig. 1.

The XCT image of the BM disk shows the development of fractures within the disk. The upper boundary with the gray gouge zone is extremely sharp on the photo (Fig. 13a), while it appears as an open fracture on the XCT images. The lower boundary with the black gouge zone is unclear. The occurrence of cracks overprinting the upper and lower zones, indicated by blue arrows in Fig. 13g, implies that the disk formed before the last earthquake, as has been suggested by Hirono et al. (2006a). It also suggests that the open fracture at the boundary between the upper gray gouge zone and the BM disk is artificial, perhaps originating during transport of the core sample. The

76

T. Hirono et al. / Tectonophysics 449 (2008) 63–84

Fig. 13. XCT images in and around the gouge zone in FZB243. (a) Photo image. (b) Color values of CT numbers in XCT images of parts (d) and (f). (c) Color values of CT numbers in XCT images of parts (e) and (g). (d) and (e) XCT images of a cross section in a plane parallel to the direction of dip. (f) and (g) XCT images of a cross section in a plane perpendicular to the direction of dip. High CT number in the outer part of the core sample is a beam hardening artifact. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

dip of the BM disk is approximately 30°. CT numbers of the upper gray gouge zone are slightly lower than those of the other zones.

Within the black gouge zone and gray gouge zone, clasts with brecciation and an open fracture are observed. The lower fracture-damaged zone does not show any dominant

T. Hirono et al. / Tectonophysics 449 (2008) 63–84

deformation structures. CT numbers of the lower fracturedamaged zone are significantly high, ranging approximately from 2000 to 2200. A few bright spots are observed in the lower fracture-damaged zone. 5.3. Microstructural observations Fig. 14 shows the core image of the main damaged zone in FZB1243 and thin section images. Their representative

77

characteristics under the optical microscope are described below. The upper gray gouge zone is composed of fine silt to fine sand grains, which are composed mainly of quartz, plagioclase, and potassium feldspar (Fig. 15a). Some minor shear bands with cataclastic fabrics are locally observed in the zone, but an intense shear fabric is not recognized. Fig. 15b shows the boundary between the upper gray gouge zone and the BM disk. The boundary is sharp, and fractured

Fig. 14. Photo images from in and around the black gouge zone of FZB1243. The left image was captured on the split surface of the core sample, and the small images of the right are of thin sections.

78

T. Hirono et al. / Tectonophysics 449 (2008) 63–84

Fig. 15. Optical microscope and photo images from in and around the black gouge zone of FZB1243. (a) Upper gray gouge zone. (b) Boundary between the upper gray gouge zone and the BM disk. (c) Boundary between the BM disk and the black gouge zone. (d, e) BM clasts within the black gouge zone. (f) Boundary between the black gouge zone and lower gray gouge zone.

mineral grains are observed along it. The BM disk is composed of an abundant fine-grained matrix supporting well-rounded lithic fragments, which are composed mainly of quartz, plagioclase, and potassium feldspar. Foliations, indicated by the preferred orientation in shape of fine platy minerals, are observed. The grain size of the lithic fragments ranges from fine silt to coarse silt. As with the BM disk in FZB1194, the microscopic features of the BM disk of FZB1243 under the optical microscope are consistent with identification as ultracataclasite or cataclasite. Hirono et al. (2006a) also discovered hourglass structures around the edges of individual grains under SEM in the BM disk, and concluded that it is pseudotachylyte with a low degree of melting.

In contrast, the boundary between the BM disk and the black gouge zone is not sharp (Fig. 15c). A layer of the black gouge is sandwiched between two blocks of the BM disks and the fragments of the disk are observed within the gouge layer. The open crack between the upper block of the BM disk and the black gouge zone may correspond to the open fracture on the XCT image (Fig. 13), because of their spatial relationship. Around the lower boundary, between the lower block of the BM disk and the black gouge zone, fragments of the disk are observed. Based on the cross-cutting relationships, the BM disk is older than the black gouge zone, which supports the interpretation that the BM disk predated the latest seismic event (Hirono et al., 2006a).

T. Hirono et al. / Tectonophysics 449 (2008) 63–84

The black gouge zone is composed of abundant matrix of clay-sized grains supporting well-rounded lithic fragments, which are composed mainly of quartz, plagioclase, and potassium feldspar (Fig. 15d). Foliations and shear fabric are not dominant. Elongated or diamond-shaped black clasts, BM clasts, are observed in the black gouge zone, as in the two shallower fault zones (Fig. 15d). The BM clasts are composed of clay-sized grains with a few fine-silt-sized mineral grains, and these clasts are also well rounded (Fig. 15e). The black gouge zone therefore corresponds to a shear zone by weak degree of deformation with random fabrics and BM clasts. Fig. 15f shows the sharp boundary between the black gouge zone and the lower gray gouge zone. The gray gouge zone is composed of fine silt to fine sand grains, which are composed mainly of quartz, plagioclase, and potassium feldspar, and do not dominantly exhibit a shear fabric. 6. Discussion 6.1. Correlation of fault zones in Holes A and B Before considering the characteristics of slip zone of the 1999 Chi-Chi earthquake, we first discuss the correlation of fault zones between in Holes A and B. The characteristics of the three major fault zones in Hole A, FZA1111, FZA1153, and FZA1222, were reported in Yeh et al. (2007) and Sone et al. (2007). Hirono et al. (2007) considered that these fault zones in Hole A could correspond to FZB1136, FZB1194, and FZB1243, respectively. We therefore compared the architecture in the shallower, middle, and deeper fault zones in Holes A and B (Fig 16a, b, and c, respectively). All fault zones in Hole A are thicker than those in Hole B. The spatial relationship of the fault zones between the two holes is illustrated in Fig. 16d. Only the middle fault zones (FZA1153 and FZB1194) have a steep angle of dip, approximately 45°, and, if extrapolated, the middle zone would apparently merge into both the shallower and deeper fault zones. If this is the case, the fault system may have a duplex structure (Fig. 16e). 6.2. Characteristics of slip zone associated with the 1999 ChiChi earthquake As described above, FZB1136 (FZA1111) is thought to be most likely related to the 1999 Chi-Chi earthquake, and we found that MSZ (LZ2) could correspond to the 2-cm-thick major slip zone associated with the 1999 earthquake by Ma et al. (2006), which was also appeared as LZ2 on the XCT image (Fig. 3). Our microstructural examination revealed that MSZ (LZ2) was cataclasite or ultracataclasite. For further characterization of MSZ, we here adopt an advanced analyses: probability to find the fragmented counterparts by fluidization. Monzawa and Otsuki (2003) proposed criteria for identifying the mechanical fluidization of granular fault materials. During fault slip, particles become fragmented, and the fragmented counterparts move apart from each other. The probability, P(r), that fragmented counterparts with radius,

79

r, can be identified within a given area under the microscope at a given magnification is defined as PðrÞ ¼ Nf ðrÞ=Nt ðrÞ where Nf (r) is the number of fragmented counterparts and Nt(r) is the total number of total particles. Otsuki et al. (2003) suggested that P(r) is inversely proportional to the interfragmentation slip (fault slip during one fragmentation event to the next for each other) and to the mean relative velocity of a pair of fragments. This may indicate that fault gouge with larger P(r) is less likely to have experienced mechanical fluidization. The resulted P(r) is presented in Fig. 17. The P(r) values for MSZ are nearly zero over the whole range of particle sizes. The maximum is 0.02 at a particle size of 7.1 µm. As Otsuki et al. (2003) suggested that such a low P(r) value (e.g., 0.02 in the gouge of the Nojima fault) could indicate that mechanical fluidization had occurred, if follows that these low P(r) values in MSZ indicate that they are very likely to have experienced a mechanically fluidized state. On the basis of these results, MSZ, most likely slip zone related to the 1999 Chi-Chi earthquake, is therefore characterized by a lower CT number, cataclastic (or ultracataclastic) texture, and high possibility to have experienced a mechanically fluidized state. 6.3. Estimation of faulting mechanism during the 1999 Chi-Chi earthquake We here discuss the seismic slip behavior of the slip zone during the 1999 earthquake. As described above, a relatively high slip velocity, 3 m/s, and large displacement, 8 m, and a low level of high-frequency radiation were recorded in the northern part of the Chelungpu fault, where the surface rupture occurred (Ma et al., 2000; Shin and Teng, 2001). We first discuss the probability that frictional melting occurred during the 1999 earthquake, because Kano et al. (2006) reported a heat signal in the shallowest fault zone (FZA1111). Frictional melting is one dynamic weakening mechanism during fault slip. In simulated faults in gabbro, Hirose and Shimamoto (2005) and DiToro et al. (2006) demonstrated experimentally slip weakening, which resulted from a reduction in the shear strain ratio in molten layer, during frictional melting. However, we did not find direct evidence, such as pseudotachylyte, of frictional melting in MSZ. Hirono et al. (2006a) found disk-shaped black material, identified as pseudotachylite with a low grade of melting, only in FZB1194 and FZB1243, not in FZB1136. Even though Sone et al. (2007) reported a black material, similar to the materials in FZB1194 and FZB1243, in the shallowest fault zone in Hole A (FZA1111), microscopic examination of the material was not performed, nor was it determined whether it was pseudotachylyte. Therefore, we cannot cite results from Hole A in our discussion of the likelihood of frictional melting during the 1999 earthquake. Taking only the results from Hole B into consideration, frictional melting seems not to have occurred during the 1999 Chi-Chi earthquake. If the black material in FZA1111 is correctly identified as pseudotachylyte, it will

80

T. Hirono et al. / Tectonophysics 449 (2008) 63–84

Fig. 16. Correlations of the three major fault zones between in Holes A and B. (a) FZA1111 and FZB1136. (b) FZA1153 and FZB1194. (c) FZA1222 and FZB1243. (d) Spatial relationship of the fault zones in the two holes. (e) Diagram based on the hypothesis that the Chelungpu fault system is a duplex structure.

T. Hirono et al. / Tectonophysics 449 (2008) 63–84

81

fault-related core samples recovered from another pilot shallow hole, indicating that clay gouge had been subjected to positive pressure. However, in our meso- and microscopic examinations, evidence for clay injection was not observed in MSZ so we cannot address the validity of this hypothesis. Therefore, the results of this study suggest that thermal pressurization was the most likely dynamic weakening mechanism during the 1999 Chi-Chi earthquake. 6.4. Characteristics of other shear zones in FZB1136 and black gouge zones in FZB1194 and FZB1243

Fig. 17. Relationship between probability, P(r), and particle size. Fragmented counterparts were counted under the optical microscope within an area of 1 mm2 by optical microscope with 4 × 10 magnification.

suggest that frictional melting occurred heterogeneously along the shallowest fault zone, perhaps as a large-scale melt patch. We next discuss the possible mechanisms of mechanical fluidization and thermal pressurization. Because MSZ is characterized as cataclastic (or ultracataclastic) texture and high possibility to have experienced a mechanically fluidized state, comminution and particle dispersion have occurred during the 1999 earthquake. Both mechanisms occur under mechanical fluidization, because comminution occurs when particles collide with each other and particle dispersion occurs simultaneously (Davies, 1982). However, both mechanisms also occur under thermal pressurization mechanism, because comminution occurs during frictional sliding and particle dispersion also occurs by subsequent pressurization of the pore fluid. Therefore, both mechanical fluidization and thermal pressurization may be valid mechanisms. However, frictional heating is thought to have occurred in FZB1136 because of a heat signal reported there by Kano et al. (2006). Mishima et al. (2006) also suggested that the black gouge zone in FZB1136 experienced at high temperature (N 400 °C) on the basis of magnetic mineral analyses. These evidences for frictional heating may support that thermal pressurization is more likely rather than mechanical fluidization, because the frictional heating is required for high pressuring of the interstitial fluid during the process of thermal pressurization but is not necessary for the process of mechanical fluidization. Finally we discuss elasto-hydrodynamic lubrication as a possible mechanism. Brodsky and Kanamori (2001) and Ma et al. (2003) proposed that the fault zone contained viscous material (e.g., gouge), in which pressure increased in accordance with the Reynolds lubrication equation; when the displacement exceeded a threshold value, lubrication pressure became high enough to widen the gap between the fault surfaces, thereby reducing the area of contact of asperities, and applied this concept to the Chelungpu fault. Otsuki et al. (2005) also proposed a lubrication mechanism on the basis of their discovery that many clay injections were developed within the

Although the other shear zones in FZB1136 (Fig. 9) and the black gouge zones in FZB1194 and FZB1243 might not be related to the 1999 earthquake, we summarize here their characteristics. We found the other two shear zones in the middle part of the black gouge zone in FZB1136 (SZ and OSZ in Fig. 9a), corresponding to old cataclasites or ultracataclasites on the basis of their microscopic examinations. One shear zone (OSZ) included a number of BM clasts (elongated or diamondshaped black clasts) inside. On the other hand, the black gouge zone in FZB1194 was characterized by cataclastic textures with the BM clasts, while the black gouge zone in FZB1243 was characterized by random fabrics with the BM clasts. Because these zones might not be related to the 1999 earthquake, we do not discuss their frictional mechanisms but only examine the origin of the BM clasts. Fig. 18 shows the backscattered electron images that resulted from semi-quantitative chemical mapping of Mg, Fe, Ca, Na, K, Ti, Al, and Mn in the BM clasts from FZB1136 and FZB1194. The microscopic images of the two analyzed clasts are shown in Figs. 9c and 12e. The BM clast from FZB1136 has remarkably higher concentrations of Mg, Fe, and K than the surrounding matrix (Fig. 18b). The Ti concentration is also slightly higher in the clast, whereas the Al concentration is slightly lower. No differences in Ca, Na, and Mn concentrations are recognized between the clast and its matrix. The BM clast within FZB1194 has higher concentrations of Mg, Fe, and K than those in the matrix (Fig. 18d). The Ti concentration is also slightly higher, the Ca concentration is much lower, and the Al concentration is slightly lower. No differences in Na and Mn concentrations are recognized. High concentrations of Mg, Fe, and Ti are common in the BM clasts in FZB1136 and FZB1194. Because high concentrations of these elements are well reported to be typical of the chemical composition of pseudotachylyte (e.g., Otsuki et al., 2003; Moecher and Sharp, 2004), we interpret the BM clasts to have originated from ancient pseudotachylyte. The pseudotachylyte formed during a past earthquake, and was incorporated into the gouge zone accompanied and then became fragmented by subsequent events. 6.5. Correlation between zone with high or low CT number and its microstructure The CT number is a function of the density and chemical composition of the material: Higher density and higher atomic

82

T. Hirono et al. / Tectonophysics 449 (2008) 63–84

Fig. 18. Back-scatter images and chemical mappings of the BM clasts in FZB1136 (a, b) and FZB1194 (c, d). An electron probe microanalyzer (EMPA; JEOL JXA8200) was used. The accelerating voltage, specimen current, and beam diameter for the mapping were kept at 15 kV, 5.0 × 10− 7 A, and 1 µm, respectively. The analysis step diameter was 1 µm, and each step time was 50 ms long. Eight components, TiO2, Al2O3, FeO, MgO, MnO, CaO, Na2O, and K2O, were detected under these conditions.

number follow higher CT number (e.g., Hirono et al., 2003a). As MSZ (LZ2) and LZ3 shows significantly lower CT number, shear zones in the fault-related samples from Hole B accompany relatively low CT numbers. The CT number in a shear zone generally depends on the internal deformation structure (e.g., Hirono et al., 2003b). A shear zone composed of compacted fabric has a relatively higher CT number because of its higher density, while a shear plane composed of a fabric displaying dilation, such as granular flow accompanying dilatancy and microscale open cracks, has a relatively lower CT number

because of its lower density. On the other hand, Hirono et al. (2006b, 2007) reported low densities, calculated from attenuation of gamma ray (GRA), in the black gouge zones within all three major fault zones, and attributed them to extensive microscale fracturing and cracks and/or to the loss of atoms with high atomic number. The causes of low CT numbers and low GRA densities in the shear zones may be the same. Ishikawa (personal communication) performed the chemical analyses of the major fault zones in Hole B by usage of X-ray fluorescence spectroscopy, and presented that contents of the

T. Hirono et al. / Tectonophysics 449 (2008) 63–84

83

Fig. 19. (a) An optical microscopic image of bright spot in FZB1136. (b) Result of the chemical analysis by scanning electron microscopy with energy dispersion X-ray spectrometry (JEOL JSM-6500F/JEOL EX-23000BU). The bright spot was identified as pyrite. (c) Disappearing of pyrite in the black gouge zone in FZB1136. An Xray diffraction spectroscopy (Spectris PANalytical X'Pert PRO MPD) was used. All scans were run at 45 kV/40 mA with Cukα radiation at the following machine settings: divergence slit, 1°; receiving slit, 0.1 mm; step width (Δ2q), 0.01°. Ill, illite; Chl, chlorite; Qtz, quartz; Pl, plagioclase; Cal, calcite; Dol, dolomite. The samples for fault gouge and host rock were collected from the black gouge zone in FZB1136 (1136.55 m depth) and the lower fracture-damaged zone in FZB1136 (1137.88 m depth), respectively.

major elements did not show large differences among the gouge zone, breccia zone, and fracture-damaged zone within all three major fault zones. Therefore, the loss of atoms with high atomic number is not likely for the causes of low CT numbers and low GRA densities in the shear zones. On the basis of our microscopic examinations, we found anastomosing network structure in the black gouge zone. This fabric might be accompanied by dilative shear deformation, resulting in low densities, low CT numbers, and low GRA densities in the shear zones. As for the relationship between CT number and density, we unfortunately cannot analyze the CT number quantitatively, because of usage of X-ray with a wide energy range. The XCT images in this study include beam hardening artifacts: X-rays with lower energy (longer wavelength) become relatively more attenuated, and penetrate less distance into the sample, but the detector counts only the number of photons without discriminating between different energy levels (wavelengths). As a result, the outer part of the sample has an apparently higher CT number. Therefore the CT number in this study should be treated semi-quantitatively. On the other hand, bright spots with high CT numbers were observed within the upper and lower breccia zones and the lower fracture-damaged zone in FZB1136 (Figs. 3–5), within the black gouge zone in FZB1194 (Fig. 10), and within the lower fracturedamaged zone in FZB1243 (Fig. 13). We performed a microscopic examination using optical microscopy and chemical

analysis using scanning electron microscopy with energy dispersion X-ray spectrometry (SEM-EDS) to identify the material. It appeared nontransparent under optical microscopy and was identified as pyrite (FeS2) by SEM-EDS (Fig. 19). As described above, a material with not only higher density but also higher atomic number generally causes higher attenuation of the X-ray, resulting in a higher CT number. The higher CT number of the bright spot can be explained by the presence of iron, because iron has a higher atomic number than typical components of sedimentary rocks, such as silicon and aluminum. In FZB1136, pyrite was observed within both the breccia and the fracture-damaged zones, but was not within the gouge zones. The X-ray diffraction spectroscopy data on the samples from the host rock and the black gouge zone supported the lack of pyrite within the gouge zone (Fig. 19c). Because thermal decomposition of pyrite occurs at 445 °C at 1 bar in an open system (Golden et al., 2004) and the gouge zone was reported to have experienced at high temperature (N 400 °C) from magnetic mineral analyses (Mishima et al., 2006), the lack of pyrite within the black gouge zone in FZB1136 may be caused from the decomposition induced by frictional heating during the earthquake. 7. Conclusions We performed X-ray CT image analyses and microstructural observations of all three major fault zones from TCDP Hole B.

84

T. Hirono et al. / Tectonophysics 449 (2008) 63–84

We found three zones with significant lower CT numbers in the black gouge zone within FZB1136, and concluded that the middle one (LZ2 on the XCT image, MSZ on the petrographic image) could correspond to the 2-cm-thick major slip zone associated with the 1999 Chi-Chi earthquake by Ma et al. (2006). We revealed that the slip zone was characterized by lower CT number, cataclastic (or ultracataclastic) texture, and high possibility to have experienced a mechanically fluidized state, and suggested that thermal pressurization was the most likely dynamic weakening mechanism in the slip zone, that is, during the 1999 Chi-Chi earthquake. For further investigation of the possibility of thermal pressurization, friction experiment of the sample from the slip zone and theoretical modeling will be needed. Acknowledgments We thank Kuo-Fong Ma, Jih-Hao Hung, and Chien-Ying Wang for their support in our analyses of Hole B core samples. We greatly thank Anne-Marie Boullier and one anonymous reviewer for their constructive comments, and we also thank Editor Kevin P Furlong and Mike Sandiford for editing this paper. This research was supported by the Japan Ministry of Education, Science, Sports and Culture, Grant-in-Aid for Young Scientists (B) 18740323, 2007. References Andrews, D.J., 2005. Thermal pressurization explains enhanced long-period motion in the Chi-Chi earthquake. Eos Trans. AGU 86 (52) Fall Meet. Suppl., Abstract S34A–04. Brodsky, E.E., Kanamori, H., 2001. Elastohydrodynamic lubrication of faults. J. Geophys. Res. 106, 16357–16374. Chen, K.C., Huang, B.S., Wang, J.H., Huang, W.G., Chang, T.M., Hwang, R.D., Chiu, H.C., Tsai, C.P., 2001. An observation of rupture pulses of the 20 September 1999 Chi-Chi, Taiwan, earthquake from near-field seismograms. Bull. Seism. Soc. Am. 91, 1247–1254. Davies, T.R.H., 1982. Spreading of rock avalanche debris by mechanical fluidization. Rock Mech. 15, 9–24. DiToro, G., Hirose, T., Nielsen, S., Pennacchioni, G., Shimamoto, T., 2006. Natural and experimental evidence of melt lubrication of faults during earthquakes. Science 311, 647–649. Golden, D.C., Ming, W., Lauer Jr., H.V., Morris, R.V., 2004. Thermal decomposition of siderite–pyrite assemblages: implications for sulfide mineralogy in Martian meteorite ALH84001 carbonate globules. Lunar Planet. Sci. XXXV 1396.pdf. Hirono, T., Ikehara, M., Otsuki, K., Mishima, T., Sakaguchi, M., Soh, W., Omori, M., Lin, W., Yeh, E., Tanikawa, W., Wang, C., 2006a. Evidence of frictional melting within disk-shaped black materials discovered from the Taiwan Chelungpu fault system. Geophys. Res. Lett. 33, L19311. doi:10.1029/2006GL027329. Hirono, T., Lin, W., Yeh, E., Soh, W., Hashimoto, Y., Sone, H., Matsubayashi, O., Aoike, K., Ito, H., Kinoshita, M., Murayama, M., Song, S., Ma, K., Hung, J., Wang, C., Tsai, Y., 2006b. High magnetic susceptibility of fault gouge within Taiwan Chelungpu fault: nondestructive continuous measurements of physical and chemical properties in fault rocks recovered from Hole B, TCDP. Geophys. Res. Lett. 33, L15303. doi:10.1029/2006GL026133. Hirono, T., Takahashi, M., Nakashima, S., 2003a. Direct imaging of fluid flow in geomaterials by X-ray CT. Geological Society of London Special Publication 215, pp. 107–115. Hirono, T., Takahashi, M., Nakashima, S., 2003b. In-situ visualization of fluid flow image within deformed rock by X-ray CT. Eng. Geol. 70, 37–46.

Hirono, T., Yeh, E., Lin, W., Sone, H., Soh, W., Hashimoto, Y., Matsubayashi, O., Aoike, K., Ito, H., Kinoshita, M., Murayama, M., Song, S., Ma, K., Hung, J., Wang, C., Tsai, Y., Kondo, T., Nishimura, M., Moriya, S., Tanaka, T., Fujiki, T., Maeda, L., Muraki, H., Kuramoto, T., Sugiyama, K., Sugawara, T., 2007. Nondestructive continuous physical property measurements of core samples recovered from Hole B, Taiwan Chelungpu-fault Drilling Project. J. Geophys. Res. 112, B07404. doi:10.1029/2006JB004738. Hirose, T., Shimamoto, T., 2005. Growth of molten zone as a mechanism of slip weakening of simulated faults in gabbro during frictional melting. J. Geophys. Res. 110, B05202. doi:10.1029/2004JB003207. Kano, Y., Mori, J., Fujio, R., Ito, H., Yanagidani, T., Nakao, S., Ma, K., 2006. Heat signature on the Chelungpu fault associated with the 1999 Chi-Chi, Taiwan earthquake. Geophys. Res. Lett. 33, L14306. doi:10.1029/ 2006GL026733. Kikuchi, M., Yagi, Y., Yamanaka, Y., 2000. Source process of the Chi-Chi, Taiwan, earthquake of September 21, 1999 inferred from teleseismic body waves. Bull. Earthq. Res. Inst. 75, 1–13. Ma, K.F., Lee, C.T., Tsai, Y.B., Shin, T.C., Mori, J., 1999. The Chi-Chi Taiwan earthquake: large surface displacements on inland thrust fault. EOS 80, 605–611. Ma, K.F., Song, T.R., Lee, S.J., Wu, S.I., 2000. Spatial slip distribution of the September 20, 1999, Chi-Chi Taiwan earthquake: inverted from teleseismic data. Geophys. Res. Lett. 27, 3417–3420. Ma, K.F., Brodsky, E.E., Mori, J., Song, T.R., Kanamori, H., 2003. Evidence for fault lubrication during the 1999 Chi-Chi, Taiwan, earthquake (Mw 7.6). Geophys. Res. Lett. 30 (5), 1244. doi:10.1029/2002GL015380. Ma, K.F., Tanaka, H., Song, S., Wang, C., Hung, J., Song, Y., Yeh, E., Soh, W., Sone, H., Kuo, L., Wu, H., 2006. Slip zone and energetics of a large earthquake from the Taiwan Chelungpu-fault Drilling Project. Nature 444, 473–476. Mishima, T., Hirono, T., Soh, W., Ikehara, M., Lin, W., Tanikawa, W., Yeh, E., Song, S., Wang, C., 2006. Thermal history estimation of the Taiwan Chelungpu Fault using rock-magnetic methods. Geophys. Res. Lett. 33, L23311. doi:10.1029/2006GL028088. Moecher, D.P., Sharp, Z.D., 2004. Stable isotope and chemical systematics of pseudotachylyte and wall rock, Homestake shear zone, Colorado, USA: Meteoric fluid or rock-buffered conditions during coseismic fusion? J. Geophys. Res. 109, B12206. doi:10.1029/2004JB003045. Monzawa, N., Otsuki, K., 2003. Comminution and fluidization of granular fault materials: implications for fault slip behavior. Tectonophysics 367, 127–143. Otsuki, K., Monzawa, N., Nagase, T., 2003. Fluidization and melting of fault gouge during seismic slip: identification in the Nojima fault zone and implications for focal earthquake mechanisms. J. Geophys. Res. 108 (B4), 2192. doi:10.1029/2001JB001711. Otsuki, K., Uduki, T., Monzawa, N., Tanaka, H., 2005. Clayey injection veins and pseudotachylyte from two boreholes penetrating the Chelungpu Fault, Taiwan: their implications for the contrastive seismic slip behaviors during the 1999 Chi-Chi earthquake. Isl. Arc 14, 22–36. Shin, T., Kuo, K., Lee, W., Teng, T., Tsai, Y., 2000. A preliminary report on the 1999 Chi-Chi (Taiwan) earthquake. Seismol. Res. Lett. 71, 24–30. Shin, T., Teng, T., 2001. An overview of the 1999 Chi-Chi, Taiwan, earthquake. Bull. Seism. Soc. Am. 91, 895–2001. Sone, H., Yeh, E., Nakaya, T., Hung, J., Ma, K., Wang, C., Song, S., Shimamoto, T., 2007. Mesoscopic structural observations of cores from the Chelungpu fault system, Taiwan Chelungpu-fault Drilling Project Hole-A, Taiwan. Terr. Atmos. Ocean. Sci. 18, 359–377. Wu, H., Ma, K., Zoback, M., Boness, N., Ito, H., Hung, J., Hickman, S., 2007. Stress orientations of Taiwan Chelungpu-Fault Drilling Project (TCDP) hole-A as observed from geophysical logs. Geophys. Res. Lett. 34, L01303. doi:10.1029/2006GL028050. Yeh, E., Sone, H., Nakaya, T., Ian, K., Song, S., Hung, J., Lin, W., Hirono, T., Wang, C., Ma, K., Soh, W., Kinoshita, M., 2007. Preliminary results of core description from the Hole-A of the Taiwan Chelungpu-Fault Drilling Project. Terr. Atmos. Ocean. Sci. 18, 327–357.