Deep Seismic Profiling in the Tokyo Metropolitan Area for Strong Ground Motion Prediction Hiroshi Sato 1, Naoshi Hirata 1, Takaya Iwasaki 1, Kazuki Koketsu 1, Tanio Ito 2, Keiji Kasahara 3, Kiyoshi Ito 4, Takeshi Ikawa 5, Taku Kawanaka 5, Masazumi Onishi 5, Susumu Abe 5, Hideo Saito 5, David Okaya 6, Tomonori Kawamura 1, Steven Harder 7 and Kate Miller 7 1 Earthquake Research Institute, University of Tokyo, Tokyo 113-0032, Japan 2 Department of Earth Science, Chiba University, Chiba 263-8522, Japan 3 National Res. Inst. for Earth Science and Disaster Prevention, 3-1, Tsukuba 305, Japan 4 Disaster Prevention Research Institute, Kyoto University, Uji 611-0011, Japan 5 JGI Inc., Tokyo 112-0012, Japan 6 Dept. Earth Sciences Univ. Southern California, Los Angeles, CA 90089-0740, USA 7 Dept. Geological Sciences, Univ. Texas El Paso, El Paso, TX 79968-0555, USA Summary To reveal the location and geometry of source faults and velocity structures of the crust for reliable estimation of strong ground motion, deep seismic profiling was carried out in the Tokyo metropolitan area along four seismic lines; Boso 2002, Sagami 2003, Tokyo bay 2003 and Kanto Mts. 2003. The seismic sources were four vibroseis trucks, explosives and air guns. From the 60% of the total seismic line, CMP (common mid-point) seismic reflection data by dense shooting was acquired. Low-fold seismic reflection and refraction data by high-energy sources, such as an explosive shot and more than one hundred of sweeps by vibroseis trucks, was obtained except for Tokyo bay 2003 seismic line. Seismic data were processed by standard CMP seismic reflection methods. Four seismic sections clearly portray the upper surface of the Philippine Sea plate, corresponds to the source faults of the Kanto earthquake of 1923 (M7.9) and Genroku earthquake of 1703 (M7.9-8.2), down to 11 seconds (TWT: Two-way travel time) beneath the metropolitan areas. The deep geometry of the inland active faults, such as the Kouzu-Matsuda fault, was also demonstrated in the seismic sections. Together with data of the velocity structures, geometry of the Neogene sedimentary basins, these data will contribute for more reliable estimation of strong ground motions. 1. Introduction The metropolitan areas in Japan, such as Tokyo (33 million in population) and Osaka-Kyoto (20 million in population), have high risk of seismic hazards. Large earthquakes in these areas have a potential to produce devastative seismic hazards, associated with worldwide economic confusion and recession. The Headquarters for Earthquake Research Promotion Japan determined to start the new program targeting the reduction of seismic hazards in the metropolitan areas. As a part of this program, the project to reveal the regional characterization of metropolitan area, including the deep seismic profiling, began from 2002 as a five years project. A long-term goal is to produce a map of reliable estimations of strong ground motion. This requires accurate determination of: source, propagation path, ground motion response. This projects focuses on identification and geometry of: source faults, subducting plates and mega-thrust faults, crustal structure: seismogenic zone, sedimentary basins, 3D velocity properties. Reconstruction of source fault and velocity models allow for more realistic 3D EQ wave simulations. All of these information will be synthesized and provided to communities involved in probabilistic hazards analysis, risk assessment and societal response. The deep seismic lines, including planning seismic lines, are shown in Fig. 1. Since 2002, deep seismic profiling was carried out along four seismic lines: Boso 2002, Sagami 2003, Tokyo Bay 2003 and Kanto Mts. 2004. Here we present the basic results of these seismic experiments. 2. Boso 2002 Seismic Line Along the Boso Peninsula, located 50-km east of Tokyo, the successions of Neogene accretionary complex and fore-arc basin fill are well exposed. Deep seismic profiling was carried out along a 165-km-long seismic line (Fig.1). In the southern 79
Figure 1. Location of the seismic lines in the "Metropolitan Project", including the planning seismic line. half of the seismic line, CMP reflection data was acquired using four vibroseis trucks and air-guns (1500 cu. In., 2000 psi). To get the continuous image from offshore to on-shore, seismic signals from two sources were recorded by Ocean Bottom Cables and land digital telemetry system. To obtain deeper image, a low-fold seismic reflection and refraction experiment was carried out along the whole seismic lines. In the northern half of the seismic line, 800 of Texan recorders from UTEP and IRIS were deployed with 100-m-interval. Twelve shots of dynamite (50-300 kg) were recorded by total 2500 channels. Figure 2. Post-stack, migrated, depth converted seismic reflection profile along the southern half of the Boso 2002 seismic line. The post-stack migrated, depth converted seismic section is shown in Fig. 2. In the northern half of the section, 4-km-thick Neogene to Quaternary basin fill is represented by continuous and coherent reflections. The relatively low frequency and large amplitude reflections at 4 km (Fig. 2) are interpreted from the 80
Figure 3. Post-stack, migrated, low-fold time section along the Boso 2002 seismic line. Red arrow: upper surface of the Philippine Sea plate, green arrow: upper surface of the pre-neogene rocks. upper surface of pre-neogene rocks. The central part forms an uplifted zone called Hayama-Mineoka uplifted zone. From this zone to the fore-arc side, the reflections show a chaotic pattern, corresponding to the Neogene accretionary complex. Fig. 3 is a stacked, migrated, low-fold seismic section of the Boso 2002 seismic line. From four seconds (TWT: two-way travel time) at the southern end to 10 seconds (TWT) at the 40 km from the southern end of the line, northward-dipping clear reflections are identified. Further north, it becomes horizontal and extends to the northern end of the seismic line at 11 seconds (TWT). Judging from its continuous nature at depth, the upper most part of the reflectors is interpreted as the upper surface of the Philippine Sea plate. Figure 4. Post-stack, migrated, seismic time section along the Sagami 2003 seismic line. Arrows represent the upper surface of the Philippine Sea plate. 3. Sagami 2003 seismic line A 77-km-long Sagami 2003 seismic line is located just above the source fault of the 1923 Kanto earthquake and crossing the Kouzu-Matsuda active fault, which shows one of the largest slip-rate in inland active faults of Japan. Along the southern flank of Hakone volcano in the western part of the seismic line, CMP-seismic reflection data was acquired using four vibroseis trucks and digital telemetry recording system. Along the Sagami Bay, due to severe traffic noise air-gun signals were used as seismic source and signals were recorded by land geophones deployed 81
Figure 5. Post-stack, migrated, depth converted seismic section across the Kozu-Matsuda fault. This seismic section is the part of Sagami 2003 seismic section (Fig. 4). along the shoreline. The post-stack, migrated, seismic time section is shown in Fig. 4. East-dipping reflectors are clearly identified beneath Odawara at 2 sec (TWT) to beneath Kamakura at 4.5 sec (TWT). These east-dipping clear reflections are interpreted as reflections from the upper part of the Philippine Sea plate. According to the source fault model of the 1923 Kanto earthquake proposed Matsu'ura et al (1980), the location and geometry of the reflectors almost coincidence with their fault model II. The Kozu-Matsuda fault is identified by the discontinuity of the pattern of reflection and can be traced down to 6 km in depth. As shown in Fig. 5, the deeper extension of this fault merges above-mentioned east-dipping reflectors. Thus, the Kouzu-Matsuda fault is interpreted as a spray fault from the boundary between Philippine Sea plate and its overlying plate. Figure 6. Post-stack, migrated time section along the Tokyo Bay 2003 seismic line. A: the upper surface of the Philippine Sea plate, B: Possible thrust beneath Miura Peninsula, C: the upper surface of the pre-neogene rocks. 82
4. Tokyo Bay 2003 seismic line The Tokyo Bay 2003 seismic line extends from the south-western end of Miura Peninsula to the northern end of Tokyo Bay over 90 km. Along the Tokyo bay, CMP reflection data was acquired using the Ocean Bottom Cables and air-guns. Due to heavy marine traffic, the length of cable was restricted for 3 km in length. To obtain large shot-receiver distance, the shooting of air-guns were carried out for 18-km-distance with 50-m-spacing for 3-km-long receiver line. The resultant number of folds reached to 200-300. This horizontal stacking enabled us to get the deeper image of the crust. Fig. 6 shows the post-stack, migrated, time section along the Tokyo Bay 2003 seismic line. The general feature of the section is very similar to the Boso 2002 seismic section. In the southern part, the Hayama-Mineoka uplift zone forms a structural high and its northern part shows thick accumulation of Neogene and Quaternary sediments of the Kanto sedimentary basin beneath the Tokyo Bay. The upper surface of the pre-neogene basement is traced at 2.2 seconds (TWT) in the northern end of the seismic section to the north of the uplift zone at 3.7 seconds (TWT; arrows C in Fig. 6). As same as Boso 2002, this reflector does not extend beneath the Hayama-Mineoka uplift zone. Beneath the Miura peninsula pattern of the reflection is chaotic except for a gently north-dipping reflectors. This part is interpreted as Neogene accretionary complex. The uplift movement of the Hayama-Mineoka uplift zone was started at middle Pliocene; the uplifting was probably produced by thrusting, represented as a low-angle north dipping reflectors (arrow B in Fig. 6). Most prominent features of this section is a gently northward dipping reflectors (arrows A in Fig. 6), which is almost continuously traced in the whole section from four seconds (TWT) in the southern end of the section to 11 seconds (TWT) at the northern end of the section. This reflector is interpreted as the upper surface of the Philippine Sea plate. Figure 7. Post-stack, migrated seismic time section along the Kanto Mts. 2003 seismic line. The seismic sections obtained by dense shooting of vibroseis trucks in the Ashigara and Kanto plain were pasted on the low-fold seismic section by mainly explosive sources. 83
5. Kanto Mountains 2003 seismic line The Kanto Mts. 2003 seismic line extends from the Ashigara plan in the south to the southern end of the Ashio Mountains crossing the eastern part of the Kanto Mountains and North-western part of the Kanto plain over 140 km. CMP-reflection data by dense shooting was acquired across the northwestern part of Kanto plain and northern part of the Ashigara plain. Wide-angle reflection and refraction data was also obtained by nine explosive shots (200-300 kg) and many sweeps of vibroseis trucks at seven locations. The post-stack, migrated time section is shown in Fig. 7. The upper surface of the Philippine Sea plate is marked by northward-dipping reflections from two seconds (TWT) at the southern end to eight seconds (TWT) beneath the Kanto Mountains. From regional geology, the Tanzawa Mountains is considered as a volcanic arc block of Izu-Bonin arc, collided into Honshu arc in latest Miocene (e.g. Niitsuma, 1989). Beneath the Kanto Mountains, southward-dipping reflectors at the lower crust and northward-dipping reflectors at the middle to upper crust are identified. Based on the P-wave velocity model by refraction analysis, wedge-shaped zone, corresponding to the deeper extension of the Tanzawa block, shows lower velocity than that of the Honshu arc crust at the same depth. These structural features have a potential to show the arc-arc collision processes associated with a wedge-shaped thrusting. 7. Conclusions Through the deep seismic profiling along the four seismic lines, we could successfully obtain the image of the upper surface of the Philippine Sea plate beneath the Tokyo metropolitan areas. To obtain the high-resolution image of the subduction mega-thrust by controlled source beneath these areas was an important first step for precise determination of the location and physical properties of the source fault. The acquired data packages also provide the basic information to construct the velocity model. Through further data processing, we will continue our research for constructing the 3D velocity and source fault models as a next step. Acknowledgments We thank to JGI seismic crews for data acquisition. We are also grateful for the usage of IRIS, PASSCAL instruments. This research was supported by the Special Project for Earthquake Disaster Mitigation in Urban Areas from the Ministry of Education, Culture, Sports, Science and Technology of Japan. References Matsu'ura, M., T. Iwasaki, Y. Suzuki and R. Sato (1980). Statical and dynamical study on faulting mechanism of the 1923 Kanto earthquake, J. Phys. Earth. 28, 119-143. Niitsuma, N. (1989). Collision Tectonics in the South Fossa Magna, Central Japan, Modern Geology 14, 3-18. 84