Comparison of Long-Period Ground Motions in the Kanto Basin during the 2004 and the 2011 Fukushima Hamado ri Earthquakes Yuka Esashi Supervisors: Kazuki Koketsu and Yujia Guo Department of Earth and Planetary Science, University of Tokyo, Japan (Dated: August 27, 2015) Amplification effects by the Kanto basin of the 2004 and the 2011 Fukushima Hamado ri earthquakes were compared. It was found that the predominant period was longer and amplification larger for waves of the earthquake, which entered the basin from the north-western direction, compared to those of the Fukushima Hamado ri earthquake, which entered from the north-eastern direction. These differences may have been due to the steeper north-western basin edge that is more suited to produce basin-induced surface waves. Also, correspondence was seen between long-period amplification and the depth of the bottom two layers of the basin (at 900 m/s and 1 S-wave velocities), both at depths of 1 km or deeper. I. INTRODUCTION Long-period ground motions are typically defined to have periods longer than 2 or 3 seconds. They have been studied as a topic in seismology because they cause damage in tall buildings and liquid storage tanks by resonance [1]. Long-period waves are often generated by shallow earthquakes, as the main component of long-period waves tend to be surface waves. They are also generated by large earthquakes that have rupture durations of sufficient lengths. Long-period waves are attenuated less compared to short-period waves, and in addition tend to be affected more by deep ground structures since their displacements are not as reduced by depth as short period waves. 4 km, and it has been said that the predominant period is around 7 seconds or above [3]. The basin has an irregular shape with two strips stretching to the north, and another one to the north-west. Additionally, the basin slope is gentler at the north-eastern edge compared to the north-western edge. Previous research has suggested the possibility that the degree of wave amplification caused by the basin differs depending on the direction of the incoming waves [4, 5]. X Niigata Chuetsu X Fukushima Hamadōri FIG. 1. Oil tanks damaged by long-period ground motions and the resulting fire in the Tokachi-oki Earthquake of 2003. [2] Because sedimentary basins have low density, when long-period waves are generated near a basin or when they enter a basin, they are often amplified. This study investigated this amplification effect in the Kanto basin. The Kanto basin lies underneath Tokyo and the greater Kanto area in Japan (FIG. 2). The deepest parts exceed FIG. 2. Distribution map of network of K-net and KiK-net stations. The color gradient shows the bedrock depth under the Kanto basin. The epicenter locations of the two earthquakes are shown by two crosses. Based on map. Therefore, this study compared two earthquakes that occurred in different directions relative to the Kanto basin, and investigated the differences in wave amplification with respect to the basin structure. The main source of data was the K-net and KiK-net seismometer stations that are located in high density throughout Japan.
2 II. THE TWO EARTHQUAKES The two earthquakes chosen for this study were the Niigata Chuetsu earthquake that occurred on 2004/10/23 17:56 with M w 6.6 at a depth of 13 km, and the earthquake that occurred on 2011/04/11 17:16 with M w 6.6 at a depth of 6 km (FIG. 2). They occurred north-west and north-east of the Kanto basin respectively. They were chosen because of their similarity in magnitude, depth and distance from the Kanto basin [6]. The Niigata earthquake is said to have generated strong long-period ground motion in the Kanto basin [7], and this has been predicted for earthquakes that occur north-west of the Kanto basin in general [4]. On the other hand, the Fukushima earthquake is said to have generated less long-period ground motion [5], and this has again been predicted for earthquakes that occur northeast of the Kanto basin [4]. III. WAVEFORM COMPARISON First, waveform data were analyzed to capture the characteristics of long-period waves that were generated during each of the earthquakes. Acceleration data from multiple sets of K-net and KiK-net stations that lie in straight lines from the epicenters through the Kanto basin were taken for each of the earthquakes. The most representative sets of accelerograms were chosen for each of the two earthquakes and are shown in FIG. 3. Acceleration data are typically used to look at shortperiod components of the wave, but in this case longperiod ground motion could be seen for the Niigata earthquake, especially at SIT003 and CHB016. This shows that long-period waves were prominent in the Niigata earthquake. However, such long-period waves were not seen in the Fukushima accelerograms. Next, the acceleration data were integrated to obtain velocity data, and a filter applied to include only wave components with periods of 3-10 seconds (FIG. 4). Here, the long-period waves can be seen with more clarity. Only the radial component result is shown here, however a similar result could be obtained from the transverse component. In the Niigata graph of FIG. 4 (a), long-period surface waves first seem to have been generated around IBR009 and SIT003, where the basin depth sharply increases. As the waves travel through the basin, they increase in period and also in amplitude relative to the S-waves. The group velocity of the waves seem to decrease at around CHB003, when the bottom two layers of the basin (900m/s and 1500m/s S-wave velocity layers) increase in depth once again. On the other hand, in the Fukushima graph of FIG. 4 (a), the periods of the waves were much shorter compared to Niigata. Surface waves were generated around CHB003, again where the bottom two layers of the basin increase in depth. As they traveled through the basin, the period remained short, and the amplitude did not exceed that of the S-waves. The peak velocities at each station in FIG. 4 (a) are plotted in (b), but with finer filters of 3-5, 5-7, and 7-10 seconds applied, and with peak velocities plotted separately. In the Niigata graph, waves of all three period ranges were greatly amplified in the first peak in the basin depth, around GNM010, IBR009, and SIT003. Waves in the 7-10 seconds range seem to be amplified around CHBH10 as well, again displaying correspondence with the bottom two layers of the basin. This correspondence was seen in the Fukushima graph where there was amplification around IBR006, and again at CHB028 and TKY025. However these amplifications were not as strong as the ones seen in the Niigata case. Also, the predominant period seems to have been around 3-5 seconds for the Fukushima case, but 5-7 seconds for the Niigata case. One possible cause of the differences noted above between the two earthquakes is that the basin edge is much steeper in the Niigata basin profile than it is in the Fukushima profile. It has previously been noted that sharper basin edges are more likely to generate stronger basin induced waves [8]. IV. VELOCITY RESPONSE SPECTRA A velocity response spectrum is a spectrum of maximum velocity response of a series of oscillators, each with slightly different natural frequencies, subjected to seismic waves. The distribution of mean response in the different period ranges in the spectra was mapped out in FIG. 5. The largest response occurred at the 5-7 seconds period range for the Niigata earthquake, but at 3-5 seconds for the Fukushima earthquake. This confirmed that the predominant period is longer for the Niigata earthquake than for the Fukushima earthquake. This was also confirmed by the individual response spectra at three stations at some of the deepest parts of the basin (FIG. 6). There were clear peaks around period of 7 seconds in the Niigata spectra; on the other hand, the Fukushima spectra did not have any peaks. The response was higher for the Fukushima earthquake at shorter periods of 1-5 seconds, but this was probably due to the fact that the three stations that were chosen were closer to the Fukushima epicenter than the Niigata epicenter, meaning that the shorter period waves had less distance to be attenuated over. Correspondence with the bottom two layers of the basin was again apparent in the maps of FIG. 5 as well; while the response radially decreased from the epicenter in the Fukushima case, the distribution of the high response sites mapped out the bottom two layers of the basin.
3 (a) Niigata Chuestu Epicenter! Through the center of the basin Niigata!Gunma!Saitama!Chiba Radial Acceleration Transverse Acceleration Vs layer 1 (b) Epicenter! Through the center of the basin Ibaraki!Chiba!Tokyo!Kanagawa Radial Acceleration Transverse Acceleration Vs layer 1 FIG. 3. Accelerograms of stations that lie in a straight line from each of the (a) and (b) Fukushima Hamado ri epicenters through the basin. Shown together are the maps of the stations, peak acceleration at each station (individual waveform graphs are normalized), and the underlying basin profile.
4 (a) Filter: 3 10 seconds, Radial Velocity 1 1 (b) Filter: 3-5, 5-7, 7-10 seconds, Radial Component 1 1 3-5 sec 5-7 sec 7-10 sec 3-5 sec 5-7 sec 7-10 sec FIG. 4. (a) Integrated velocity data from the same stations shown in FIG. 3. Filter of 3-10 seconds period have been applied. (b) Peak velocity for each of the stations in (a), separately shown for three period ranges of 3-5,5-7, and 7-10 seconds.
5 3-5 sec 5-7 sec 7-10 sec h=5% Vs Layer Depth 1 FIG. 5. Mean velocity response distribution for the two earthquakes, separately shown for three period ranges of 3-5, 5-7, and 7-10 seconds. On the right are the maps showing the depth of each layer in the Kanto basin. TKY007 KNG006 h=5% CHBH10 FIG. 6. Velocity response spectra at three stations at the deepest parts of the basin (shown on the top right map) for the two earthquakes.
6 V. CONCLUSION AND FUTURE WORK This study showed that long-period ground motion was amplified by the Kanto basin both in the 2004 Niigata Chuetsu and the 2011 earthquakes, but the predominant period was longer and amplification larger for waves from the earthquake, which entered the Kanto basin from the north-western direction. These differences may have been due to the steeper north-western basin edge that is more suited to generating basin-induced surface waves. Correspondence was seen between long-period propagation and the bottom two layers of the basin, at and 1 S-wave velocities, at depths of 1 km or deeper. This may suggest that long-period wave are affected by deeper ground structures. Future work should investigate a greater number of earthquakes around the Kanto basin, although the number of earthquakes around the Kanto basin with sufficiently large magnitude and good seismic records is limited. Similar analysis should be done on other basins too to investigate whether the same conclusions can be generalized across multiple basins. Furthermore, a previous study has suggested that wave conversion within the Kanto basin may be one reason why wave amplification was so large in the Niigata earthquake [7], arguing that seismic waves were refracted in such a way that they interfered with one other in the center of the basin. Another study has also suggested the effect of basin shape in directing waves to certain parts of the basin through refraction [9]. It would be interesting to repeat a similar analysis on the Fukushima earthquake, since the concave north eastern basin edge may refract waves in such a way that the waves do not converge within the basin. [1] Koketsu, K., and Miyake, H. A seismological overview of long-period ground motion. Journal of Seismology Vol. 12, Issue 2, pp. 133 143, April 2008. [2] Koketsu, K., Hatayama, K., Furumura, T., Ikegami, Y., and Akiyama, S. Damaging Long-period Ground Motions from the 2003 Mw8.3 Tokachi-oki, Japan Earthquake. Seismological Research Letters Vol. 76, No.1, pp. 67 73, January/February 2005. [3] Yoshimoto, K., and Takemura, S. A study on the predominant period of long-period ground motions in the Kanto Basin, Japan. Earth, Planet and Space 66:100. 2014 [4] Denolle, M. A., H. Miyake, S. Nakagawa, N. Hirata, and G. C. Beroza Long-period seismic amplification in the Kanto Basin from the ambient seismic field. AGU Geophysical Research Letters Vol. 41, Issue 7, pp. 2319 2325, 2014. [5] Tsuno, S., Yamanaka, H., Midorikawa, S., Yamamoto, S., Miura, H., Sakai, S., Hirata, N., Kasahara, K., Kumura, H., and Aketagawa, T. Characteristics of Long- Period Ground Motions in the Tokyo Metropolitan Area and its Vicinity, by Recording Data of the 2011 Off the Pacific Coast of Tohoku Earthquake (Mw 9.0) and the Aftershocks. JAEE Journal Vol. 12, No. 5, pp. 102 116, 2012. [6] NIED F-net. http://www.fnet.bosai.go.jp/top.php [7] Furumura, T., and Hayakawa, T. Anomalous Propagation of Long-Period Ground Motions Recorded in Tokyo during the 23 October 2004 Mw 6.6 Niigata-ken Chuetsu, Japan, Earthquake. Bulletin of the Seismological Society of America Vol. 97, No. 3, pp. 863 880, June 2007. [8] Graves, R. W., Pitarka, A., and Somerville, P. G. Ground- Motion Amplification in the Santa Monica Area: Effects of Shallow Basin-Edge Structure. Bulletin of the Seismological Society of America Vol. 88, No. 5, pp. 1224-1242, October 1998. [9] Koketsu, K., and Kikuchi, M. Propagation of Seismic Ground Motion in the Kanto Basin, Japan. SCIENCE Vol. 288, pp. 12371239, May 2000.