BROADBAND SOURCE MODEL AND STRONG MOTIONS

Transcription

1 BROADBAND SOURCE MODEL AND STRONG MOTIONS OF THE 1855 ANSEI-EDO EARTHQUAKE ESTIMATED BY THE EMPIRICAL GREEN S FUNCTION METHOD Toshimi Satoh 1 1 Chief Researcher, Institute of Technology, Shimizu Corporation, Tokyo, Japan toshimi.satoh@shimz.co.jp ABSTRACT We estimate the broadband source model and strong motions in the periods of 0.05 to 10 s by the empirical Green s function method. The seismic intensity of Mw5.9 earthquake on July 23, 2005 in the northwest of Chiba prefecture occurring within the Philippine Sea plate exhibits the anomaly pattern similar to that of the Ansei-edo earthquake. Therefore by using this earthquake as an element event we estimate locations and sizes of two strong motion generation areas (SMGAs) with the same stress drop assuming Mw7.0, 7.1, and 7.2 to fit the seismic intensity distribution. In the model we assume a high-angle reverse fault in northwest of Chiba prefecture as an intraplate earthquake of the Philippine Sea plate based on F-net CMT solution of Mw5.9 earthquake. The estimated centroid depth of SMGAs is 60 km. The estimated seismic intensity distribution agrees with the observed one from the source region to 200 km far beyond Kanto district. The A of [Nm/s 2 ] of the 1855 Ansei-edo earthquake is much larger than A derived from the M 0 -A relation for interplate earthquakes of the Pacific plate (Satoh, 2010) and is slightly smaller than that for intraplate earthquakes of the Pacific plate (Satoh, 2013). The absolute value of A is comparable to A estimated for 1923 Mw7.9 Kanto earthquake (Satoh, 2016). The peak ground velocities at the engineering bedrock at Chiba, Otemachi and Shinjuku in Tokyo metropolitan area are comparable to those during the 1923 Mw7.9 Kanto earthquake, but the velocity pulses with periods of 1 to 2 s, which depend on the size of SMGAs, are more predominant than those during the 1923 Kanto earthquake. Keywords: Strong motion, The 1855 Ansei-edo earthquake, Empirical Green s function method INTRODUCTION The 1855 Ansei-edo earthquake caused severe damage in the Tokyo metropolitan area. About 10,000 people in center of Tokyo passed away due to this earthquake (CAO, 2004). The magnitude M is estimated to be 6.9 to 7.4 (e.g., Hikita and Kudo, 2001; Usami et al., 2013). The M7-class earthquake occurrence probability within 30 years in the Tokyo metropolitan area is estimated to be 70 % (HERP, 2014). The maximum seismic intensity scale of 6 upper during the 1855 Ansei-edo earthquake in center of Tokyo (Nakamaura et al., 2003) was the biggest among historical M-7 class earthquakes whose seismic intensity scale were known. However there are several views on the hypocenter, mechanism and M of this earthquake. For example Hikita and Kudo (2001) estimated M and the hypocenter by assuming a strike slip fault to fit the seismic intensity distribution in the south Kanto district by the empirical Green s function method. They concluded that M is 7.4 and that it occurred in the northwest of Chiba prefecture with a depth of 68 km within the subducting Pacific slab. However, they did not simulate the seismic intensity distribution in the regional area and the maximum seismic intensity in center of Tokyo was estimated to be 6 lower. Nakamura et al. (2007) estimated the hypocenter of the point source to fit the seismic intensity distribution in the Kanto district by the stochastic Green s function method considering three dimensional Q structure and concluded that the hypocenter was northwest of Chiba prefecture with a depth of 70 km. On the other hand, Furumura and Takeuchi (2007) pointed out that the Ansei-edo earthquake should be a shallow crustal earthquake to reproduce the wide region of the seismic intensity of 4 based on the numerical method. However, it has not been proposed broadband source models which can reproduce in both the Tokyo metropolitan area and the regional area. The Ansei-edo earthquake exhibited anomaly seismic intensity distribution in the Kanto district due to the

2 complex structure underneath there. It has been pointed out (e.g., Nakanishi and Horie, 1980: Nakamura et al., 2007; Furumura and Takeuchi, 2007) that the seismic intensity of the intermediatesize earthquakes occurring in the northwest or central Chiba prefecture with depths of 60 to 80 km exhibited anomaly seismic intensity distribution. Therefore we firstly examine the seismic intensity distribution of the intermediate-size earthquakes occurring underneath the Tokyo metropolitan area. Then we estimate broadband source models composed of strong motion generation areas and background and the strong motions to fit the seismic intensity distribution in both the Tokyo metropolitan area and the regional area by the empirical Green s function method using intermediate earthquakes whose seismic intensity distribution are similar to the Ansei-edo earthquake. DATA Fig.1 shows the seismic intensity distribution and the hypocenter of the 1855 Ansei-edo earthquake Figure 1. Seismic intensity scale (SI) and the epicenter (a star) of the Ansei-edo earthquake by Usami et al. (2013). Question marks mean the low reliability (Usami and Daiwa Exploration & Consulting Co. Ltd., 1994) of SI. Figure 2. Epicenters and CMT solutions of four events shown in Table 1 and the epicenters of the Ansei-edo earthquake estimated by Usami et al.(2013) and Hikita and Kudo (2001). Red lines and blue dotted lines denote top depths of the Pacific (PAC) plate (Nakajima et al, 2009) and the Philippine Sea (PHS) plate (Hirose et al, 2008), respectively. Table 1. Event list Unified hypocenter catalog byjma CMT by JMA CMT by F-net This study Event ID Date and Time M J Depth[km] Depth[km] Depth[km] M 0 [Nm] M W Stress drop[mpa] A[Nm/s 2 ] Aug Non E E Jul E E May E E Sep Non E E+18

3 5th IASPEI / IAEE International Symposium: Effects of Surface Geology on Seismic Motion estimated using historical materials by Usami et al. (2013). E denotes the seismic intensity 4 and e denotes the seismic intensity 3 (Usami et al., 2013). Question mark means the low reliability (Usami and Daiwa Exploration & Consulting Co. Ltd., 1994). Fig.2 shows the location and F-net CMT solution of four earthquakes listed in Table 1 and the epicenters of the Ansei-Edo earthquake estimated by Usami et al. (2013) and Hikita and Kudo (2001). In this figure contours of the upper depth of the Philippine Sea plate (PHS) (Hirose et al., 2008) and the Pacific plate (PAC) (Nakajima et al., 2009) are also shown. Four earthquakes near the epicenters estimated by previous studies (e.g., Hikita and Kudo, 2001; Nakamura et al., 2007; Usami et al., 2013) with Mw5-6 are selected for the examination of the seismic intensity distribution. Based on the examination we determine the appropriate events as element events for the empirical Green s function method (Dan and Sato, 1998) in Table 1 is the event used by Hikita and Kudo (2001) as the element event. INTERMEDIATE-SIZE EVENTS AROUND THE ANSEI-EDO EARTHQUAKE Estimation of source parameters In previous our paper (Satoh, 2015a) we separated source, path, and site effects from observed strong motion records of intraplate events with Mw and depths from 30 to 60 km occurring within the Philippine sea plate. The estimated Q was similar to that estimated by other researchers including little deeper events as shown in Satoh (2014). Therefore we calculate acceleration source spectra of four events in Table 1 using the Q and empirical amplification factors estimated by Satoh (2015a) from strong motion records. Strong motion records at K-NET, KiK-net and JMA-95 type stations with hypocentral distance < 100 km which is the same data selection conditions as Satoh (2015a) are used. Then we estimate the short period spectral level A which is the flat level of acceleration source spectrum (Dan et al., 2001) and the Brune s stress drops to fit the acceleration source spectra to -2 model. Estimated values are listed in Table 1. Fig. 3 compared the seismic intensity calculated from strong motion records and predicted by the (a) Observed seismic intensity (b) Predicted seismic intensity Figure 3. Seismic intensity derived from strong motion records and predicted by revised stochastic Green s function method (Satoh, 2015a) for four events listed in Table 1.

4 revised stochastic Green s function method for PHS intraplate earthquakes (Satoh, 2015a). In the revised stochastic Green s function method, surface waves and scattering waves as well as S waves are empirically considered and the empirical site factors for both Fourier amplitude and phase spectra are used. The observed seismic intensity distribution of is similar to that of the 1855 Ansei-edo earthquake shown in Fig.1. The predicted seismic intensity distribution for is larger than the observed at western Tokyo and Kanagawa prefecture. The pattern of predicted seismic intensity of 4 for is almost the same as that of predicted seismic intensity of 3 for This result means that the observed anomaly seismic intensity of is not reproduced by only empirical site factors and would be affected by complex structure under the seismic bedrock. Since the seismic intensity distribution for is totally different from that of the 1855 Ansei-edo earthquake, we estimate broadband source models and strong motions to fit the seismic intensity distribution of the 1855 Ansei-edo earthquake by the empirical Green s function method using three evens except for as element events. THE 1855 ANSEI-EDO EARTHQUAKE Method of estimation of broadband source models and strong motions Since the corner frequency of the 1855 Ansei-edo earthquake would be longer than 5 s, the short period spectral level A is more influential to the seismic intensity than Mw or seismic moment M 0. Therefore we firstly estimate the broadband source model assuming Mw7.1 using three events as element events. As a result the seismic intensity estimated using as an element event reproduced best the seismic intensity scale by Usami et al. (2013), Nakamura et al. (2003) and Nakamura et al. (2007). Therefore we also estimate the broadband source models assuming Mw7.0 and Mw7.2 using as the element event. Fault area S is derived from the M 0 -S relation for intraplate earthquakes by Iwata and Asano (2011). Strike and dip angles are set to be the same as those of one fault plane of F-net CMT solution of the element event. The location of the fault is set that the element event is on the fault plane. We regard that is the intraplate earthquakes occurring within the Philippine sea plate based on the focal depth, centroid depth in Table 1 and the depth contours of the Philippine sea plate (Hirose et al., 2008) as shown in Fig. 2. Since the seismic intensity distribution of the Ansei-edo earthquake is large in the north and south directions, we select the high-angle reverse fault dipping to west. We regard that is the intraplate earthquake occurring within the Philippine sea plate because the depth of 68 km estimated by the DD method is shallower than unified hypocenter catalog by JMA shown in Table 1 and the location estimated by the DD method is further west from unified hypocenter catalog by JMA as pointed out JMA (2006). The centroid depth of CMT solution estimated by JMA shown in Table 1 is also shallow, 50 km. In addition Koketsu and Miyake (2005) pointed out the possibility that is the high-angle reverse fault dipping to east based on the aftershock distribution. The is the typical thrust fault occurring interface between the Philippine sea plate and the Pacific plate. The Brune s stress drop of is the smallest among four events shown in Table 1. Then SMGAs and the location of the rupture starting point are estimated by the empirical Green s function method (Dan and Sato, 1998) to fit the seismic intensity scale of the Ansei-edo earthquake by Usami et al. (2013), Nakamura et al. (2003), and Nakamura et al. (2007). The rupture starting point is assumed to be located at the hypocenter of the element event and the center of the every third element fault. For each the rupture starting point, the location, length and width of two SMGAs with the same stress drop are estimated. The other parameters are derived from the equations by Madariaga (1979) and Boatwright (1988) together with the relation that the slip of SMGAs is twice of the total slip based on the strong motion prediction recipe (e.g. Irikura and Miyake, 2011). Background parameters are also derived based on the strong motion prediction recipe (e.g. Irikura and Miyake, 2011). We assume that the rupture velocity is 2.88 km/s which is 0.72 times of S-wave velocity (Geller, 1979) and that the rupture propagates concentrically. The geometric spreading factor and frequency f dependent Q (=107f 0.51 ) (Satoh, 2015a) are used in the empirical Green s function method. Strong motion records of K-NET, KiK-net, JMA-95 type observed at the surface of the element events are used. The seismic intensity at each station is derived from the estimated strong motions. Since the strong motions are estimated without considering nonlinear site effects, the seismic intensity considering the

5 nonlinear effects are derived by an empirical relation between the seismic intensity affected by linear and nonlinear site responses derived by Satoh (2015b). Broadband source model and estimated seismic intensity Fig.4 shows the broadband source models estimated assuming Mw7.1 together with the epicenters and the CMT solutions of the element events. Table 2 shows the source parameters of the element events and the Ansei-edo earthquake. The total A is the almost the same to each other for three cases. This means that A is the influential parameter to seismic intensity. Fig.5 shows the seismic intensity distribution for three cases. The distribution pattern is similar to that of the element event shown in Fig.3(a). The seismic intensity distribution estimated using agrees with the seismic intensity of the Ansei-edo earthquake best. Especially the seismic intensity estimated using reproduces the feature that the seismic intensity abruptly decreases in west of Tokyo. The seismic intensity distribution estimated using in the regional area within about 200 km far from the fault is shown in Fig.6 with the same scale bar to Fig.1. It is confirmed that the estimated seismic intensity reproduces that of the Ansei-edo earthquake in both Tokyo metropolitan and regional area. Table 2. Source parameters used in the case of Mw7.1 Figure 4. Locations of outer fault, two SMGAs and rupture starting point (a star) for the best broadband source model estimated using each element event shown by F-net CMT solution with JMA epicenter. Element event Fault Rupture starting point SMGA-1 SMGA-2 Total SMGAs Background Event ID Length[km] Width[km] M 0 [Nm] 5.62E E E+19 M W Length[km] Width[km] Strike[ ] Dip[ ] Rake[ ] Total A[Nm/s 2 ] 6.64E E E+19 Top depth[km] Bottom depth[km] Longitude[ ] Latitude[ ] depth[km] Length[km] Width[km] Stress drop[mpa] Length[km] Width[km] Stress drop[mpa] M 0 [Nm] 9.49E E E+18 A[Nm/s 2 ] 5.55E E E+19 M 0 [Nm] 4.67E E E+19 A[Nm/s 2 ] 3.64E E E+19 Stress drop[mpa]

6 5th IASPEI / IAEE International Symposium: Effects of Surface Geology on Seismic Motion Figure 5. Seismic intensity of the Ansei-edo earthquake estimated by using (left), (middle), and (right) assuming Mw7.1. A large rectangle, small rectangles, and a blue star in each figure denote a fault, SMGAs and a rupture starting point, respectively. A bold line of the fault means the upper line Figure 6. Seismic intensity scale and the rupture starting point (a star) of the Ansei-edo earthquake estimated by using assuming Mw7.1. Figure 7. Seismic intensity of the Ansei-edo earthquake estimated by using assuming with Mw7.0 (upper) and Mw7.2 (lower). Fig.5 shows the seismic intensity estimated using assuming Mw7.0 and Mw7.2. The estimated seismic intensity distribution assuming Mw7.0, Mw7.1 and Mw7.2 are similar to each other except that the seismic intensity of 6 upper at stations just beneath the fault in the case of Mw7.2. In Table 3 source parameters estimated using assuming Mw7.0 Mw7.1 and Mw7.2 are shown.

7 Table 3. Source parameters estimated using assuming different Mw for the Ansei-edo earthquake Fault SMGA-1 SMGA-2 M W Total A [Nm/s 2 ] 6.46E E E+19 Length [km] Width [km] Stress drop [MPa] Length [km] Width [km] Stress drop [MPa] (a) M 0 -A (b) Depth-A/A dan Figure 8. Relation between M 0 and A and relation between depth and A/A dan. A dan is the short period spectral level derived from M 0 -A relation by Dan et al.(2001). The total A is the almost the same to each other for three cases. This result also means that Mw is not an influential parameter to seismic intensity. The Cabinet Office (2013) sets a pure strike-slip fault with the strike of north-south direction under the Tokyo station in order to reproduce the seismic intensity in center of Tokyo during the Ansei-edo earthquake. Since the Cabinet Office used stochastic Green s function method, the anomaly seismic intensity distribution was not be reproduced. Hikita and Kudo (2001) estimated the epicenter (Fig.2) of a pure strike-slip fault with the strike of north-south direction by the grid of 0.2 degree intervals for the longitude and 0.1 degree intervals for the latitude. Considering with the intervals, the locations of SMGAs estimated in this study are consistent with the location of the fault by Hikita and Kudo (2001). The seismic intensity distribution estimated using which is used by Hikita and Kudo (2001) is similar to that estimated by Hikita and Kudo (2001). Since Hikita and Kudo (2001) estimate the Mw of the Ansei-edo earthquake assuming the same Brune s stress drop of 26 MPa for both the element event and the Ansei-edo earthquake, Mw must be estimated to be 7.4. Fig.8(a) shows the M 0 -A relation derived from this study and previous studies. A for the Anseiedo earthquake estimated using is almost the same to A for the 1923 Mw7.9 Kanto earthquake (Satoh,2016), which is the interplate earthquake of the Philippine Sea plate.. The M 0 -A relation of the Ansei-edo earthquake is much larger than that for the PAC interplate earthquakes (Satoh, 2010). The large A of the Ansei-edo earthquake is the typical feature of intraplate earthquakes (e.g., Satoh, 2010, 2013). Fig.8(b) shows the depth-a/a dan relation for earthquakes along the Sagami Tough. Here A dan is short period spectral level derived from M 0 -A relation by Dan et al. (2001). A/A dan for intraplate earthquakes are larger than interplate earthquakes. This feature is the same to earthquakes along the Japan Trench. There is no depth dependency within the same types of earthquakes. This is different from PAC intraplate earthquakes whose A/A dan increases as the depth is deeper. The A/A dan for the Ansei-edo earthquake assuming Mw7.0 to 7.2 are larger than the average of those for small and

8 5th IASPEI / IAEE International Symposium: Effects of Surface Geology on Seismic Motion intermediate-size intraplate earthquakes. This feature is the same to A/Adan for M>7 intraslab earthquakes along the Japan trench and Kurile trench (e.g., Satoh, 2013). Predicted strong motions Since the difference of strong motions estimated by assuming Mw7.0 and 7.1 using is small, we show the strong motions estimated by assuming Mw7.1 at four stations shown in Fig.5. The horizontal and vertical ground motions at the engineering bedrock with S-wave velocity > 400 m/s are derived using one-dimensional site responses for S-waves and P-waves, respectively. The subsurface structure is the same as that used in Satoh (2016) for the estimation of strong motions at the bedrock during the 1923 Kanto earthquake. In Satoh (2016) the subsurface structure is inverted using H/V spectral ratios based on diffused field theory (Kawase et al., 2011) by fixing the shallow structure by the PS logging results and the deep structure by three-dimensional structure model estimated by HERP (2009). Then the horizontal motions at the surface are estimated using the revised equivalent linear analysis code DYNEQ by Yoshida. Dynamic properties are assumed based on Yasuda and Yamaguchi (1985) who derived the empirical relations parameterized by nature of soils and the effective confining pressure. The vertical motions at the surface are regarded as the same to ones estimated by empirical [s] [s] (a) Acceleration filtered from 0.05 to 10 s (b) Velocity filtered from 0.05 to 10 s Figure 9. Acceleration and velocity time histories estimated at the surface (GL0m) and the engineering bedrock based on the best broadband source model with Mw7.1 using as the element event. Number upper each time history is peak ground acceleration [cm/s2] and peak ground velocity [cm/s]. Figure 10. Pseudo velocity response spectra (h=5%) estimated at the surface (GL0m) and the engineering bedrock based on the best broadband source model with Mw7.1 using as the element event.

9 5th IASPEI / IAEE International Symposium: Effects of Surface Geology on Seismic Motion Green s function method without considering nonlinearity. Fig.9 shows the acceleration and velocity time histories estimated at the surface and the bedrock for the Ansei-edo earthquake. The estimated seismic intensity is 6 upper at KNG002 (Yokohama) and 6 lower at CHB009 (Chiba), E4E (Otemachi) and TKY007 (Shinjuku). The PGAs and PGVs of horizontal components at the bedrock at four stations are about 250 to 350 cm/s2 and 30 to 50 cm/s, respectively. Fig.10 shows pseudo velocity response spectra with damping factor of 5 % at the surface and the bedrock. Building standard law notification spectrum for level-2 at the engineering bedrock in Japan is also shown in Fig.10. The spectra of horizontal components at the bedrock at CHB009, E4E and TKY007 have peaks around 1 to 2 s. The spectra at the bedrock at four stations are comparable to the level-2 spectrum in the periods shorter than 2 s and are smaller than that in the periods longer than 2 s. Fig.11 shows the acceleration and velocity time histories estimated at the surface and the bedrock for the 1855 Mw7.1 Ansei-edo earthquake and the 1923 Mw7.9 Kanto earthquake (Satoh 2016). Since there are no proper events as element events around SMGAs of the 1923 Kanto earthquake, strong motions for the 1923 Kanto earthquake are estimated by the revised stochastic Green s function method (Satoh, 2016). The duration of strong motions during the 1923 Kanto earthquake is much longer than the 1855 Ansei-edo earthquake. PGAs and PGVs of horizontal components for the Anseiedo earthquake are comparable to those of the 1923 Kanto earthquake at three stations except for KNG002 located near SMGAs of the 1923 Kanto earthquake. However, the velocity pulses with [s] [s] (a) Acceleration filtered from 0.05 to 5 s (b) Velocity filtered from 0.05 to 5 s Figure 11. Acceleration and velocity time histories estimated at the engineering bedrock for the 1855 Mw7.1 Ansei-edo earthquake and the 1923 Mw7.9 Kanto earthquake (Satoh, 2016). Number upper each time history means the same as Fig.9. Figure 12. Pseudo velocity response spectra (h=5%) estimated at the engineering bedrock for the 1855 Mw7.1 Ansei-edo earthquake and the 1923 Mw7.9 Kanto earthquake (Satoh, 2016).

10 periods of 1 to 2 s, which depend on the size of SMGAs, are much dominated especially in east-west (EW) components during the Ansei-edo earthquake. Fig.11 shows pseudo velocity response spectra with damping factor of 5 % at the surface and the bedrock. The spectra of EW components for the Ansei-edo earthquake at E4E and TKY007 are larger than those for the 1923 Kanto earthquake in the periods of 0.5 to 2 s. The velocity pulses with a period of 1 to 2 sare most influential for the structural damage to ordinary buildings. (Kawase and Nagato, 2000). In addition natural periods of half of super high-rise buildings in the Tokyo metropolitan area are 1.5 to 2.5 s (Koyama, 2014). Therefore strong motions of the Ansei-edo earthquake in center of Tokyo would be important as input motions for super high-rise buildings. CONCLUSIONS The 1855 Ansei-edo earthquake was estimated to occur beneath Tokyo metropolitan area and the seismic intensity was 6 upper there (e.g., Nakamura et al., 2003). There are several different studies on the hypocenter, magnitude, or mechanism, but are still unknown factors. No broadband source models which can simulate the seismic intensity in both the Tokyo metropolitan and regional area were proposed. In this study we estimate the broadband source model and the strong ground motions in the periods of 0.05 to 10 s by simulating the seismic intensity using K-NET, KiK-net and JMA95-type strong motion records by the empirical Green s function method. The seismic intensity of Mw5.9 earthquake on July 23, 2005 in the northwest of Chiba prefecture occurring within the Philippine sea plate exhibits the anomaly pattern similar to that of the Anse-edo earthquake. Therefore using this earthquake as an element event we estimate locations and sizes of two strong motion generation areas (SMGAs) with the same stress drop assuming Mw7.0, 7.1, and 7.2. The estimated seismic intensity agrees with the observed seismic intensity from the source region to 200 km far beyond Kanto district. In the model we assume a high-angle reverse fault in northwest of Chiba prefecture as an intraplate earthquake of the Philippine Sea plate based on the F-net CMT solution and the depth distribution of the aftershocks of Mw5.9 earthquake (Koketsu and Miyake, 2005). The estimated centroid depth of SMGAs is 60 km. The short period spectral level A, which is the flat level of acceleration source spectrum is well constrained by the seismic intensity, but Mw is not so much. The A of [Nm/s 2 ] of the 1855 Ansei-edo earthquake is much larger than A derived from the M 0 -A relation for interplate earthquakes of the Pacific plate (Satoh, 2010). The absolute value of A is comparable to A estimated for 1923 Mw7.9 Kanto earthquake (Satoh, 2016), which is the interplate earthquake of the Philippine Sea plate. The large A of the Anse-edo earthquake is the typical feature of intraslab earthquakes. Strong motions at the engineering bedrock are calculated by the one-dimensional site responses from the strong motions at the surface estimated using the broadband source model with Mw7.1. The peak ground velocities at the engineering bedrock at Chiba, Otemachi and Shinjuku in Tokyo metropolitan area are comparable to those during the 1923 Mw7.9 Kanto earthquake, but the velocity pulses with periods of 1 to 2 s, which depend on the size of SMGAs, are more predominant than those during the 1923 Kanto earthquake. ACKNOWLEDGMENTS This research was supported by the Ministry of Education, Science, Sports and Culture, Grant-in-Aid for Scientific Research (A), (P.I., Prof. Kawase). Strong motion records observed at K-NET and KiK-net by NIED (National Research Institute for Earth Science and Disaster Prevention) and JMA-95 type by Japan Meteorological Agency (JMA) are used. The CMT solutions by NIED and JMA and the unified hypocenter catalog by JMA are also used. We also used the depth data of plate by Hirose et al (2008) and Nakajima et al. (2009) in /fhirose/ja/platedata.html. I would like to thank these organizations to provide data and information. Some figures are plotted with GMT (Wessel and Smith, 1998). REFERENCES

11 Boatwright, J. (1988). The seismic radiation from composite models of faulting, Bull. Seism. Soc. Am., 78, Brune, J.N. (1970). Tectonic stress and the spectra of seismic shear waves from earthquakes, J. Geophys. Res., 75 (26), Cabinet Office (2004). The 1855 Ansei-edo earthquake, /kyoukunnokeishou/rep/1855-ansei-edojishin/ (in Japanese) Cabinet Office (2013). dansoumodel_01.pdf (in Japanese) Dan, K. and Sato, T. (1998). Strong-motion prediction by semi-empirical method based on variable-slip rupture model of earthquake fault, J. Struct. Constr. Eng., AIJ, 509, (in Japanese with English abstract) Dan, K., M. Watanabe, T. Sato and T. Ishii (2001). Short-period source spectra inferred from variable-slip rupture models and modeling of earthquake faults for strong motion prediction by semi-empirical method, J. Struct. Constr. Eng., AIJ., 545, (in Japanese with English abstract). Furumura, T. and Takeuchi, H. (2007). Large earthquakes occurring beneath Tokyo metropolitan area and strong ground motions: Ansei Edo earthquake and Meiji Tokyo earthquake, Journal of Geography, 116 (3/4), (in Japanese with English abstract). Geller, R.J. (1976). Scaling relations for earthquake source parameters and magnitudes, Bull. Seism. Soc. Am., 66, Headquarters for Earthquake Research Promotion (2009). Map of Long-period ground motions, (in Japanese) Headquarters for Earthquake Research Promotion (2014). Long-term evaluation of seismic activity along the Sagami trough (The second edition), htttp:// (in Japanese) Hikita, T. and K. Kudo (2001). Estimation of source parameters and strong ground motions during the 1855 Ansei-edo earthquake by the empirical Green's function method, J. Struct. Constr. Eng., AIJ., 546, Hirose, F., J. Nakajima and A. Hasegawa (2008). Three-dimensional velocity structure and configuration of the Philippine Sea slab beneath Kanto district, central Japan, estimated by double-difference tomography, Zisin, 60 (3), (in Japanese with English abstract). Irikura, K. and H. Miyake (2011). Recipe for predicting strong ground motion from crustal earthquake scenarios, Pure Appl. Geophys., 168, , doi: /s Iwata, T. and Asano, K. (2011). Characterization of the heterogeneous source model of intraslab earthquakes toward strong ground motion prediction, Pure Appl. Geophys., 168, Japan Meteorological Agency (2006). An outline of the earthquake of M6.0 occurred in the northwestern part of Chiba Prefecture on July 23, 2005, Report of the Coordinating Committee for Earthquake Prediction, 75, , (in Japanese) Kawase, H., F.J. Sánchez-Sesma and S. Matsushima (2011). The optimal use of horizontal-to-vertical (H/V) spectral ratios of earthquake motions for velocity structure inversions based on diffuse field theory for plane waves, Bull. Seism. Soc. Am., 101, Kawase, H. and K. Nagato (2000). Structural damage impact of strong motions evaluated by the nonlinear analysis of a set of building models, in Proc. Sixth International Conference on Seismic Zonation, Palm Springs, CA, CD-ROM, pdf, November Koketsu, K. and H. Miyake (2005), (in Japanese) Koyama, S. (2014). /lecture/h26/pdf/t6.pdf (in Japanese) Madariaga, R. (1979). On the relation between seismic moment and stress drop in the presence of stress and strength heterogeneity, J. Geophys. Res., 84, Nakamura, M., I. Kayano and R. Matsu ura (2003). Damage of the Ansei-Edo earthquake (1855/11/11) in capital region, Historical Earthquakes, 19, (in Japanese with English abstract). Nakamura, R., T. Uetake, K. Satake, S. Toda, T. Usami, K. Shimazaki and K. Watanabe (2007). Characteristics of 3D attenuation structure, abnormal seismic intensity distribution of the Kanto district, and the depth

12 estimate of the 1855 Ansei-Edo earthquake, Historical Earthquakes, 22, (in Japanese with English abstract). Nakanishi, I. and A. Horie (1980). Anomalous distributions of seismic intensities due to the descending Philippine Sea plate beneath the southern Kanto district, Japan, J. Phys. Earth, 28, Nakajima, J., F. Hirose and A. Hasegawa (2009). Seismotectonics beneath the Tokyo metropolitan area, Japan: Effect of slab-slab contact and overlap on seismicity, J. Geophys. Res., 114, B08309, DOI: /2008JB Satoh, T. (2010). Scaling law of short-period source spectra for crustal earthquakes in Japan considering style of faulting of dip-slip and strike-slip, J. Struct. Constr. Eng., AIJ., 651, (in Japanese with English abstract). Satoh, T. (2013). Short-period spectral level fmax and attenuation of outerrise, intraslab and interplate earthquakes in the Tohoku district, J. Struct. Constr. Eng., AIJ., 689, (in Japanese with English abstract). Satoh, T. (2014). Generation method of stochastic greens function considering into surface waves and scattering waves using records of moderate-sized interplate earthquakes along the Sagami trough, J. Struct. Constr. Eng., AIJ., 705, (in Japanese with English abstract). Satoh, T. (2015a). Stochastic Green's functions considering surface waves and scattering waves using strong motion records of moderate-sized intraplate earthquakes along the Sagami trough, Journal of Japan Association for Earthquake Engineering, 15 (1), (in Japanese with English abstract). Satoh, T. (2015b). Relation between seismic intensity by linear and nonlinear site responses using transfer functions derived from KiK-net records, J. Struct. Constr. Eng., AIJ., 713, (in Japanese with English abstract). Satoh, T. (2015c). Strong motion simulation of the 1987 Chiba-ken-toho-oki earthquake (M J 6.7) using stochastic green's function generation method empirically considering surface waves and scattering waves, Journal of Japan Association for Earthquake Engineering, 15 (1), (in Japanese with English abstract). Satoh, T. (2016). Estimation of strong motion generation areas and strong motions during the 1923 Kanto earthquake using revised stochastic Green's function method, J. Struct. Constr. Eng., AIJ, 719, (in Japanese with English abstract). Usami, T. and Daiwa Exploration & Consulting Co. Ltd., (1994). Map of seismic intensity distribution of historical earthquakes in Japan, The Japan Electric Association, pp.647 (in Japanese) Usami, T., H. Ishii, T. Imamura, M. Takemura and R. Matsu ura (2013). Materials for comprehensive list of destructive earthquakes in Japan , University of Tokyo Press, Tokyo (in Japanese) Yasuda, S. and I. Yamaguchi (1985). Dynamic soil properties of undisturbed samples, in Proc. of the 20th Annual Convention of Japanese Society of Soil Mechanics and Foundation Engineering, (in Japanese). Yoshida, N., DYNEQ computer program for DYNamic response analysis of level ground by EQuivalent linear method, (in Japanese). Wessel, P. and W.H.F. Smith (1998). New, improved version of Generic Mapping Tools released, EOS.