A NEW SEISMICITY MAP IN THE KANTO. Tokyo, Japan (Received October 25, 1978)

Transcription

1 A NEW SEISMICITY MAP IN THE KANTO DISTRICT, JAPAN Tokyo, Japan (Received October 25, 1978) In order to improve the accuracy of the hypocenter locations in the Kanto district, Japan, station corrections are estimated from the data of artificial explosions. About 500 earthquakes which occurred during the period from January to August 1975 have been relocated by using the station corrections. The area covered in this study is bounded by latitudes error in the spatial coordinates of the relocated epicenter is estimated to be about 1.5km. Depth of the local explosions is estimated to be within 2km of the actual location. The good agreement between the calculated and the actual depths suggests that the accuracy of the depth is significantly improved from that determined when the station corrections were not included. An aseismic zone in the middle of the Kanto district which was suggested by the old seismicity map is not substantiated by the present study. The Tokyo area has repeatedly suffered from earthquakes which occurred in the Kanto district (Fig. 1). Forty eight earthquakes with intensity 5 and larger (Japanese scale) occurred in the Tokyo area since 818 (KAYANO, 1975). In general the degree of damage caused by an earthquake depends on its magnitude. However, other factors such as the depth of the event and the population in the epicentral area drastically affect the degree of damage. Unfortunately, Tokyo has large population, heavy traffic, densely built tall structures etc. all of which go to enhance the effect of earthquakes. In spite of the relatively moderate magnitude 6.9, the Edo earthquake which occurred at a very shallow depth beneath the Tokyo Metropolitan area in 1855 caused considerable damage, and about seven thousand people were killed (KATSUMATA, 1974).

2 Fig. 1. Distribution of epicenters of earthquakes which caused damage in Tokyo during the period from 799 to 1974 (after USAMI, 1975). The distribution of earthquakes in the Kanto district has already been studied by many investigators (ISHIBASHI and TSUMURA, 1971; TSUMURA, 1973; KATSUMATA, 1974; TSUMURA, 1974a,) Figure 2 shows epicenters of microearthquakes which were located by using the data from highsensitivity seismic stations of the Earthquake Research Institute (ERI), the University of Tokyo. All earthquakes with magnitude larger than 2.7 were located for the Kanto district, and larger than 2.0 for the Tokyo area

3 A New Seismicity Map in the Kanto District, Japan Fig. 2. Distribution of epicenters of small earthquakes determined by the ERI network data during the period from January 21 to August 31, 1975.

4 Kanto district (Tsumura, personal communication, 1975; ISHIDA and ASANO, 1976; ZAMA and SHIMA, 1976). In the present study, we shall develop a method of hypocenter determination for the Kanto district and investigate the distribution of earthquakes in greater detail. The effect of surface heterogeneity near the station will be reduced by using the travel-time residuals from the explosions observed at the individual stations. About 500 earthquakes which occurred in the Kanto district during the period from January to August 1975 are relocated. 2. Data Figure 4 shows the locations of the explosions used in the present study. Numbers 1, 2 (IIZUKA et al., 1975) and 4 (ITo et al., 1976) were Fig. 4. Distribution of explosions and temporary observations used in the present study. Large and small symbols show the shot point and the temporary observations at each explosion, respectively. Dotted belts show the seismic profiles made by R.G.E.S. ( ). The numbers correspond to those in Table 1.

5 Fig. 5. Distribution of permanent stations used in the present study (Two stations, Fujigawa (FUJ) and Shizuoka (SHZ), which are used in the present study are not plotted in this figure because they are located outside of this figure). shots for the purpose of monitoring velocity changes in the Kanto district and numbers 3, 5 and 6 (SHIMA et al., 1976a, b) were shots to study the velocity structure in the Tokyo area. The coordinates of the explosions are listed in Table 1. The P waves from the explosions were recorded by the temporary stations shown in Fig. 4 and the permanent stations belonging to the Earthquake Research Institute (ERI) and the National Research Center For Disaster Prevention (NRCDP). Figure 4 includes

6 the seismic profiles which were made by the Research Group For Explosion Seismology (R.G.E.S., ) to study crustal structures. In addition to these explosions, the Cannikin nuclear explosion on Amchitka Island on November 6, 1971, was used in the present study. Seismic waves from this explosion were recorded at the permanent stations belonging to the Japan Meteorological Agency (JMA) and the Earthquake Research Institute (ERI), shown in Fig. 5. The earthquakes which occurred in the area bounded by latitudes 3. Crustal Structure A number of seismic explosion studies have been made along many Kanto made by R.G.E.S. On the basis of refraction studies using these profiles, MIKUMO (1966) proposed a structure for the Kanto district called Model E-3A3 (Fig. 6(a)). He also used gravity and surface wave dispersion data in the study. However the seismic profiles employed in Mikumo's study do not cover the middle part of the Kanto district. Furthermore the profiles have not been reversed and the derived structure is not unique. Nevertheless, use of Model E-3A3 seems to improve the accuracy of the hypocenter locations in the Kanto district (ISHIBASHI and TSUMURA, 1971; TSUMURA, 1973). It is considered that Model E-3A3 is an accurate representation of the general crustal structure for the entire Kanto district. No other extensive explosion profile going through the Kanto district is available. For this reason, Model E-3A3 is used as the standard crustal the travel-time and the travel-time anomaly at the Yumenoshima site for a vertical path through the top 4.5km, respectively.

7 441 structure of the Kanto district in the present study. The effects of structural heterogeneity beneath the stations will be discussed in the following section in relation to this model. The velocity structure deeper than 32 km is assumed to be the same as that of Jeffreys-Bullen. 4. Station Corrections Significant errors in the hypocentral locations may result since the actual structure is more complex than the model used for the determination of hypocentral parameters. In the present study, this effect is reduced by applying station corrections to the data. The station corrections were determined using data obtained from various explosions. The high sensitivity seismic stations in the Kanto district recorded P waves from the six explosions shown in Fig. 4. By using the observed travel-times and Model E-3A3, the travel-time residuals at each station can be estimated. The travel-time residuals thus obtained include the effect of the whole path from the explosion to the station. Although the details of the structure along the whole path are not known, the effects of the surface geology at the explosion and station sites can be estimated using the explosion data as follows. Figure 6(b) compares Model E-3A3 and the velocity structure derived from the Yumenoshima explosion (SHIMA et al., 1976(a)). The travel-time along a vertical path through the top 4.5km of the crust is 0.82sec and 1.51sec for Model E-3A3 and the crustal model derived from the Yumeno- Table 2. Travel-time residuals of the six explosions listed in Table 1.

8 442 M. ISHIDA and S. ASANO shima explosion, respectively. If the velocity structure below 4.5km at the Yumenoshima site is assumed to be identical to that of Model E-3A3, the travel-time anomaly at the site of the Yumenoshima explosion is 0.69 sec for a vertical path (Fig. 6(b)). Since the emergence angles from the explosion to the stations are about 60 degrees, the travel time anomaly caused by the local geology at the shot point is estimated to be about 1.38sec. In the same way, the travel-time anomaly at each shot point was estimated. These source terms were subtracted from the travel-times to obtain the residuals at the stations. These travel-time residuals for the Fig. 7. The profile of upper crustal structure derived from each explosion. The numbers correspond to those in Table 1. Abbreviations show the same stations as those shown in Fig. 5.

9 A New Seismicity Map in the Kanto District, Japan 443 six explosions at the stations are listed in Table 2. We adopt the mean values of the residuals for the six explosions as the station corrections at the individual seismic stations. They are listed in column 3 of Table 3. This travel-time anomaly represents not only the effect at the station site but also the path effect from the shot point to the station. Figure 7 shows the velocity structures derived from the explosions along the indi- Table 3. Station corrections.

10 444 M. ISHIDA and S. ASANO vidual profiles. The two profiles at the top were derived from the Oshima explosion in 1968 along the paths to Dodaira (NNW) and Narada (NW) (ASANO et al., 1976). The third profile from the top was derived from et al., 1976). The fourth profile from the top was derived from the Yumenoshima explosions in 1975 (Feb. and Dec.) and the Yoshikawa explosion in 1975 (SHIMA et al., 1976a, b). The bottom profile was derived from the Ogishima explosion (ISHIDA, 1975). The effects of local geology beneath the stations were estimated using the individual profiles in a manner similar to that used for the estimation of the source term. The traveltime anomalies for a vertical path are listed in column 5 of Table 3. In order to compare these travel-time anomalies with those obtained above from the difference between the observed and calculated travel-times (listed in column 3 of Table 3), the latter is reduced to the value for vertical incidence with the assumption that the incidence angles at the individual stations are about 60 degrees. These corrected values are listed in the last column of Table 3. The good agreement between the value in column Table 4. The difference between the real and relocated short points.

11

12

13

14 Fig. 10. The crosses and squares show the earthquakes located by the high sensitivity seismic stations, which could and could not be detected at Iwatsuki (IWT). Thick line shows the maximum hypocentral distance where the onset of P wave can be identified at IWT. Thin lines show the maximum hypocentral distances estimated in proportion to the magnifications of the instruments of the individual stations.

15 A New Seismicity Map in the Kanto District, Japan artifact of systematic mislocation due to the structural heterogeneity. 6. Detection Capability The detection capability and the distribution of stations around the epicenters also have important effects upon the hypocenter determination. The detection capability is defined by the magnitude of an earthquake, the 10 shows the relationship of these elements at the Iwatsuki (IWT) station where the instrument is installed at a depth of 3,500m. The crosses and squares show the earthquakes located by the high-sensitivity seismic stations, Fig. 11. Closed circles and crosses show the high sensitivity seismic stations and assumed epicenters for the estimation of the accuracy of hypocenter determination, respectively.

16 M. ISHIDA and S. ASANO which could and could not be detected at IWT, respectively. The solid line wave could be identified. The detection capabilities of the other highsensitivity seismic stations were estimated in proportion to the magnification of the instruments and are shown in Fig. 10 by the thin lines. In order to estimate the accuracy of each hypocenter determination the timing accuracy of the onset of the P wave was assumed as follows, Fig. 12. Accuracy of hypocenter determination where the depth is constrained at 20km and the magnitude is assumed to be 2.5. The stations used for computation are shown in Fig. 11. The digits show the standard deviations in hundreds of meters. Triangles represent events which can be detected at four stations but are not located (after F. Yamamizu, personal communication).

17 A New Seismicity Map in the Kanto District, Japan hypocentral distance shown. The distribution of the timing error is assumed to be Gaussian. By using the Monte Carlo method, fifty cases were randomly tried for each earthquake denoted by crosses shown in Fig. 11. In the example shown in Fig. 12 (F. Yamamizu, personal communication, 1976), the focal depth is constrained at 20km and the magnitude of the earthquake is assumed to be 2.5. The locations of the stations used in the computation are shown in Fig. 11. In order to simplify the computation, a single-layer crustal structure with a P wave velocity of 7.0km/sec is assumed. The digits in Fig. 12 represent the standard deviations in hundreds of meters of the locations. The triangles in Fig. 12 show the earthquakes which could be identified at four stations but could not be located. The result shown in Fig. 12 may suggest the accuracy of the hypocenter determination used in the present study although the real situation is a little different from this example; the crustal structure is more complex and the timing accuracy may be worse than the assumed value due to the thick sedimentary layer which covers the Tokyo area. 7. Discussion and Conclusion The accuracy of hypocenter determination strongly depends on the method of computation and the travel-time table. If a three-dimensional seismic velocity structure is very well known, methods using recently developed two or three-dimensional seismic ray tracing (ENGDAHL and LEE, 1976; Hamada, personal communication, 1976) may be used to improve the accuracy. Unfortunately the structure in the Kanto district is not well enough known to use these methods. Furthermore, these methods are prohibitively time consuming for routine analyses. The master event method (e.g., JOHNSON and HADLEY, 1976) is another useful method. This method provides good relative location accuracy for events which are clustered in the same area. However, the earthquakes examined in the present study are distributed all over the Kanto district and the master event method is not adequate. The S-arrival times, if identified correctly, are sometimes very useful to constrain the focal depth or the origin time. However, when stations are located on a thick sedimentary layer, the onset of S waves often becomes very ambiguous. Since many stations in the Kanto district are located on a thick alluvium, S waves were not used in the present study. About 500 earthquakes which occurred in the Kanto district were relocated using the station corrections obtained in this study (listed column 6 of Table 3). The accuracy of the relocation was examined by locating

18 M. ISHIDA and S. ASANO the local explosions. The absolute error of the relocation was estimated to be about 1.5km. In particular, the absolute error in the depth was within 2km. This suggests that the accuracy of the depth is significantly improved from that when the station corrections are not included. An aseismic area in the middle of the Kanto district which was suggested by the old seismicity maps becomes very ambiguous in the relocated map. We are very grateful to Dr. H. Kanamori who critically reviewed the manuscript. We also thank Mr. F. Yamamizu who kindly allowed us to use his program for numerical computation of the accuracy of hypocenter determination. Dr. K. Tsumura provided us with the data of hypocenters used in this study and gave us many useful comments. We express sincere thanks to him. We have benefitted greatly from many comments about crustal structures by Dr. T. Mikumo. Dr. Christine Powell made helpful comments on the manuscript. Assistance by Miss M. Tatsukawa is also gratefully acknowledged. The computations were partially made at the computing center of the National Research Center For Disaster Prevention. This work was partially supported by the U.S. Geological Survey. Contribution Number 3,160, Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena, California REFERENCES AOKI, H. and T. TADA, P-wave travel-time anomaly in Japan, Observation of the Cannikin nuclear explosion, J. Phys. Earth, 21, , ASANO, S., Y. ICHINOSE, I. HASEGAWA, S. IIZUKA, and H. SUZUKI, The crustal structure in the southern Kanto district derived from explosion seismic observations, Abstract, Spring Meeting Seismol. Soc. Japan, p. 143, 1976 (in Japanese). ENGDAHL, E.R. and W.H.K. LEE, Relocation of local earthquakes by seismic ray tracing, J. Geophys. Res., 81, , IIZUKA, S., I. HASEGAWA, K. ITO, K. ICHIKAWA, H. SUZUKI, S. KIKUCHI, S. ASANO, H. MATSUMOTO, and M. TAKAHASHI (Research Group for Seismic Wave Velocity), Precise measurements of changes in seismic wave velocities by means of explosionseismic method, Rept. Geol. Surv. Japan, 254, 1-74, 1975 (in Japanese). ISHIBASHI, K. and K. TSUMURA, Temporary observation of micro-earthquakes in the southern part of the Kanto district, Bull. Earthq. Res. Inst., 49, , ISHIDA, M., Travel-time anomaly in and around Kawasaki based on the Ogishima explosion, Abstract, Fall Meeting Seismol. Soc. Japan, p. 31, 1975 (in Japanese). ISHIDA, M. and S. ASANO, The hypocentral determination in the Kanto district, Abstract, Fall Meeting Seismol. Soc. Japan, p. 15, 1976 (in Japanese). ITO, K., K. ICHIKAWA, I. HASEGAWA, T. KAKIMI, K. KASAHARA, S. IIZUKA, and T. TADA, Abstract, Spring Meeting Seismol. Soc. Japan, p. 52, 1976 (in Japanese). JOHNSON, C.E. and D.M. HADLEY, Tectonic implications of the Brawley earthquake swarm, Imperial Valley, California, Bull. Seismol. Soc. Am., 66, , KATSUMATA, M., Seismic activity in the Kanto district, ed. T. Kakimi and Y. Suzuki, in Earthquakes and Crustal Movements in the Kanto District, pp , Rateisu Press, Tokyo, 1974 (in Japanese). KAYANO, I., Intervals between earthquakes with intensity V or more in the Tokyo area, ed. Tokyoto-Bosaikaigi, in Researches on the Earthquakes beneath Tokyo, Vol. 3, pp , 1975 (in Japanese).

19 A New Seismicity Map in the Kanto District, Japan KIKUCHI, T., Investigation based on the geology, ed. Tokyoto-Bosaikaigi, in Researches on the Earthquakes beneath Tokyo, Vol. 1, pp , 1974 (in Japanese). MIKUMO, T., A study on crustal structure in Japan by the use of seismic and gravity data, Bull. Earthq. Res. Inst., 44, , NAGAMUNE, T., P-waves to seismological stations in Japan from the underground explosion of November 6, 1971, at Amchitka Island, J. Phys. Earth, 21, , SHIMA, E., M. YANAGISAWA, K. KUDO, K. SEO, and K. YAMAZAKI, On the base rock of Tokyo; observations of seismic waves generated from the 3rd Yumenoshima and Yoshikawa explosions, Bull. Eathq. Res. Inst., 51, 45-61, 1976 a (in Japanese). SHIMA., E., M. YANAGISAWA, K. KUDO, T. YOSHII, Y. ICHINOSE, K. SEO, K. YAMAZAKI, N. OHBO, Y. YAMAMOTO, Y. OGUCHI, and M. NAGANO, On the base rock of Tokyo; observations of seismic waves generated from the 1st and 2nd Yumenoshima explosions, Bull. Earthq. Res. Inst., 51, 1-11, 1976b (in Japanese). TAKAHASHI, H. and K. HAMADA, Deep borehole observation of earth's crust activities around Tokyo; Introduction of Iwatsuki observatory, Pure Appl. Geophys., 113, , TSUMURA, K., Microearthquake activity in the Kanto district, Publications for the 50th anniversary of the Great Kanto earthquake, 1923, pp , TSUMURA, K., Seismic activity in the Kanto district, ed. T. Kakimi and Y. Suzuki, in Earthquakes and Crustal Movements in the Kanto District, pp , Rateisu Press, Tokyo, 1974a (in Japanese). TSUMURA, K., Distribution of microearthquake activity in the Kanto district, ed. Tokyoto- Bosaikaigi, in Researches on the Earthquakes beneath Tokyo, Vol. 3, pp , 1974b (in Japanese). USAMI, T., Descriptive Catalogue of Disaster Earthquakes in Japan, Univ. Tokyo Press, Tokyo, 1975 (in Japanese). ZAMA, S. and E. SHIMA, The accuracy of hypocentral determination, Abstract, Fall Meeting Seismol. Soc. Japan, p. 133, 1976 (in Japanese).