J Seismol (2008) 12:133–143 DOI 10.1007/s10950-007-9080-0

ORIGINAL ARTICLE

A seismological overview of long-period ground motion

Kazuki Koketsu & Hiroe Miyake

Received: 12 May 2007 /Accepted: 29 November 2007 / Published online: 11 January 2008 # Springer Science + Business Media B.V. 2007

Abstract Long-period ground motion has become an 1 Introduction increasingly important consideration because of the recent rapid increase in the number of large-scale The long-period component of seismic ground motion structures, such as high-rise buildings and oil storage generated by causes damage in near- tanks. Large subduction-zone earthquakes and moder- regions through source effects such as the directivity ate to large crustal earthquakes can generate far-source effect of rupture propagation and the near-field term long-period ground motions in distant sedimentary of body wave radiation. In addition, the long-period basins with the help of path effects. Near-fault long- ground motions attenuate slowly with distance be- period ground motions are generated, for the most part, cause of certain path effects, and site effects amplify by the source effects of forward rupture directivity. Far- these motions in distant basins so that they can cause source long-period ground motions consist primarily destruction over a much greater range. of surface waves with longer durations than near-fault Previously, most structures in -prone long-period ground motions. They were first recog- regions were low-profile structures, and so relatively nized in the seismograms of the 1968 Tokachi-oki and short-period (1 s or shorter) ground motions, with 1966 Parkfield earthquakes, and their identification has which these structures might be resonant, were been applied to the 1964 earthquake and earlier important. However, considering the increasing num- earthquakes. Even if there is no seismogram, we can ber of large structures, such as high-rise buildings, oil identify far-source long-period ground motions storage tanks, suspension bridges, off-shore oil through the investigation of tank damage by liquid drilling platforms, and recent base-isolated structures, sloshing. long-period (1 to 10 s or longer) ground motions have been increasingly important (e.g., Kanamori 1979; Keywords Far-source long-period ground motion . Fukuwa 2008). Near-fault long-period ground motion . Source effect . The worst example of destruction caused by long- Path effect . Site effect . Liquid sloshing period ground motion occurred in Mexico City,

400 km from the 1985 Michoacan earthquake (MW= 8.0; e.g., Beck and Hall 1986). Another example is the K. Koketsu (*) : H. Miyake 2003 Tokachi-oki earthquake (MW=8.3) that occurred Earthquake Research Institute, University of Tokyo, in Hokkaido, (e.g., Koketsu et al. 2005). In the 1-1-1 Yayoi, Bunkyo-ku, Tokyo 113-0032, Japan present study, we will review this long-period ground e-mail: [email protected] motion from a seismological point of view. 134 J Seismol (2008) 12:133–143

2 Nature of long-period ground motion from propagating fault rupture. Hanks (1975) recov- ered 234 components of long-period ground motion The examples listed in the previous section indicate in the source region of the 1971 San Fernando the existence of far-source long-period ground mo- earthquake (MW=6.6), and the neighboring Los tion. Far-source long-period ground motion was Angeles basin in California. He coined the term identified, for the first time in Japan, in seismograms “long-period ” in this paper of the 1968 Tokachi-oki earthquake (MW=8.2) ob- (Zama 1993). served with large amplitudes and a predominant The upper panels of Fig. 1 compare the typical period of 2.5 s at Hachinohe, northeastern Japan. velocity seismograms of far-source (left) and near- They were also observed by strong motion seismo- fault (right) long-period ground motions. The most graphs installed in the first super high-rise building in obvious difference is the duration of ground motion. Japan. The Kasumigaseki building was located in The far-source long-period ground motions continue Tokyo, 650 km from the earthquake source (Shima for 1 min or longer, whereas the near-fault long- 1970). Trifunac and Brune (1970) observed long- period ground motions last only for 10 to 20 s. period ground motion in distant seismograms of the Accordingly, the far-source long-period ground

1940 Imperial Valley earthquake (MS=7.1) in Cal- motions have velocity response spectra that are ifornia. Both the Japanese and Californian authors comparable to those of the near-fault long-period attributed these far-source long-period ground ground motions despite their smaller amplitudes. motions to regional surface waves. We will describe these two types of long-period On the other hand, Aki (1968) discovered near- ground motion using actual seismograms. Seismic fault long-period ground motion in strong motion ground motion can be considered to be composed of accelerograms observed very close to the 1966 Park- source, path, and site effects, as mentioned earlier. As field earthquake (MW=6.1) in California. This short- regional surface waves are significant beyond tens of duration ground motion was interpreted as resulting kilometers, we will explain the far-source long-period

Fig. 1 Examples of far-source long-period ground motions at and 1 km (upper right). Their velocity response spectra are also epicentral distances of 400, 250, and 50 km (upper left) and shown in the lower half of the panels near-fault long-period ground motions at fault distances of 1, 5, J Seismol (2008) 12:133–143 135 ground motion based primarily on the effect of paths the exceptional tank damage. The 1983 Coalinga earth- and sites and their velocity structures. The near-fault quake shows motion of a similar duration (Manos long-period ground motion will be explained based and Clough 1985), but this earthquake was of moderate primarily on the source effect, some site effects, and magnitude, so its duration was due primarily to local velocity structures because near-fault long- site effects. period ground motion is significant at distances less than 50 km (Somerville et al. 1997). However, even when there is no seismographic 3 Far-source long-period ground motion evidence, the generation of far-source long-period ground motion can be identified based on damage to The potential impact of far-source long-period ground large tanks. This damage is caused primarily by motion was first recognized worldwide in 1985. The sloshing of the liquid inside the tanks. Because the Michoacan earthquake may have caused as many as excitation of liquid sloshing appears to require long- 20,000 fatalities (Beck and Hall 1986), although the duration seismic ground motion, it can be linked to official estimate of human fatalities was 8,000. In far-source long-period ground motion. Ohta and Mexico City, at a distance of 400 km from the Zama (2005) documented 14 cases of tank damage earthquake, approximately 300 buildings collapsed, because of liquid sloshing (Table 1). Among these and 800 buildings were later demolished because they cases, there are four obvious exceptions, namely, the were beyond repair (Celebi et al. 1987). The 1923 Kanto earthquake, the 1979 Imperial Valley waveguide produced by the subduction of the Cocos earthquake, 1983 Coalinga earthquake, and the 1999 plate lengthened the duration of ground motion

Kocaeli earthquake (MW=7.9,6.5,6.2,and7.6, (Furumura and Kennett 1998). Its long-period com- respectively). In these cases, the damaged oil tanks ponent with a period of 2 to 4 s was then amplified were located in the source regions of these earth- and further lengthened by the sediments of an ancient quakes. The near-fault ground motion observed near lake that once existed under Mexico City (Beck and the tank damaged by the 1999 Kocaeli earthquake, Hall 1986). which is the middle trace of the right-hand panel in These path and site effects resulted in far-source Fig. 1, shows considerable later phases, which long-period ground motions with large amplitudes resulted in motion of longer duration compared to and long duration, as shown in Fig. 2. As partially the other traces. This may be because of the long destroyed buildings having 6 to 15 stories resonate causative fault and site effect, and may have caused with the predominant periods of 2 to 4 s, many of

Table 1 List of tank damage by liquid sloshing (Ohta and Zama 2005)

Earthquake Year MW Damage Far-source? Reference

Kanto 1923 7.9a 6,000 t oil tank No Hirano (1982) Long Beach 1933 6.2 Water tank Yes? Steinbrugge (1970) Kern County 1952 7.5 Oil tanks Yes? Steinbrugge and Moran (1954) Alaska 1964 9.2 Many oil tanks, fires Yes Rinne (1967) Niigata 1964 7.6b Many oil tanks, fires Yes FDMA (1965) Central Chile 1965 7.1c Oil tanks Yes Shibata (1974) San Fernando 1971 6.6 Oil tank Yes Shibata (1971) Miyagi-oki 1978 7.4d Oil tanks Short period FDMA (1979) Imperial Valley 1979 6.5 Oil tank No Horoun (1983) Coalinga 1983 6.2 Many oil tanks No Manos and Clough (1985) Japan Sea 1983 7.7e Many oil tanks, fire Yes Yoshiwara et al. (1984) Kocaeli 1999 7.6 Many oil tanks, fires No JSCE (2000) Chi-Chi 1999 7.7 Oil tanks Yes Yoshida et al. (2000) Tokachi-oki 2003 8.3 Many oil tanks, fires Yes Ohta and Zama (2005)

The moment magnitudes were retrieved from the USGS earthquake database except for a Wald and Somerville (1995), b Ruff and c d e Kanamori (1983), MS, Seno et al. (1980), and Dziewonski et al. (1983). 136 J Seismol (2008) 12:133–143 Fig. 2 Schematic diagram of accelerograms observed along the propagation path from the epicentral station at Caleta de Campos to Mex- ico City (Celebi et al. 1987; reprinted with permission from Earthquake Engineer- ing Research Institute)

these buildings eventually collapsed or were severely which damaged many oil storage tanks at Akita and damaged (Beck and Hall 1986). The velocity struc- Niigata, northeastern Japan (Fig. 4). In Akita, a tures along the propagation path and in the Mexico damaged storage tank caught fire, and Niigata is City valley controlled these damaging features of the located 300 km away from the earthquake. far-source long-period ground motion (Furumura and This far-source ground motion reminded seismol- Kennett 1998; Shapiro et al. 2002; Kawase and Aki ogists of long-period ground motions caused by the

1989). 1964 Niigata earthquake (MS=7.4), which had been Before the 1985 Michoacan earthquake, however, considered to result from local liquefaction. At as mentioned earlier, Japanese seismologists observed Kawagishi-cho in Niigata, several apartment build- far-source long-period ground motion during the 1968 ings were toppled or tilted by the liquefaction, and the Tokachi-oki earthquake (Fig. 3). This motion did not strong motion records observed there show long- cause any significant damage, but the 1983 Japan Sea period features (Fig. 5). The effect of liquefaction earthquake (MW=7.7) generated damaging long- appears only during the initial 13 s of short-period period ground motions, causing large fluid sloshing, ground motion (Fig. 6) but was also blamed for the later long-period ground motion. The seismograms shown in Fig. 5 are currently considered to be the first observation of far-source long-period ground motion in Japan (Kudo et al. 2000). Kawagishi-cho is not far from the source region of the 1964 Niigata earth- quake, but we specify the long-period ground motions as in this category, as they mostly consist of surface waves. Liquid sloshing by them and the associated liquefaction caused severe damage to large oil storage tanks, five of which caught fire and burned for 2 weeks. As verified by the 1964 Niigata and 1983 Japan Sea earthquakes, the Niigata basin efficiently excites far-source long-period ground motion. A further example was provided by the 1993 Hokkaido Nansei- Fig. 3 Strong motion records of the 1968 Tokachi-oki oki earthquake (M =7.7). The far-source long-period earthquake at Hachinohe 180 km from the epicenter. Dashed W lines represent horizontal long-period components (Sakajiri et ground motion reached this basin at a distance of al. 1974) 500 km from the earthquake epicenter (Fig. 7), causing J Seismol (2008) 12:133–143 137 Fig. 4 Acceleration (upper) and velocity (lower) seis- mograms of far-source long- period ground motion at a distance of 300 km from the 1983 Japan Sea earthquake (Kudo and Sakaue 1984)

liquid sloshing of oil storage tanks. A maximum in a basin, either directly or by conversion at the sloshing height of 1.7 m was recorded and oil margins, which lead to large amplitudes and long scattering was observed on the floating roofs of four durations of shaking (Fig. 8). The direct and con- of the tanks. verted surface waves also imply the importance of The 2003 Tokachi-oki earthquake provided the velocity structures both along the propagation path southern coast of Hokkaido, Japan, with one of the and within the basin, respectively. Fluid sloshing in most significant examples (Koketsu et al. 2005; damaged oil storage tanks occurred with a period Hatayama 2008) of far-source long-period ground similar to that of the dominant ground motion (7 to 8 s), motion. This largest earthquake in the world in 2003 producing displacements of a few meters and a long occurred in the southeast of Hokkaido along the Kuril fluid oscillation that contributed to the destruction of trench, although the 1968 Tokachi-oki earthquake floating roofs and fire in two tanks. mentioned earlier occurred near the junction of the The latest offshore example is the 2004 off Kii

Japan and Kuril trenches. The city of Tomakomai in Peninsula earthquake (MW=7.4), which occurred in a Hokkaido suffered serious damage to large oil tanks shallow part of the subducting Philippine Sea plate, from long-period ground motions generated by the southwestern Japan. This also demonstrates that a earthquake more than 250 km away offshore. A very large earthquake associated with an effective propa- important contribution to this far-source long-period gation path can generate damaging long-period ground motion comes from surface waves stimulated ground motions in a distant sedimentary basin

Fig. 5 Acceleration (top), velocity (middle), and dis- placement (bottom) seismo- grams of the 1964 Niigata earthquake observed at the basement of a damaged apartment house in the Niigata basin 50 km from the epicenter (Kudo et al. 2000) 138 J Seismol (2008) 12:133–143

Fig. 6 Close-up of the early 13 s of N–S(top), E–W(middle), and vertical (bottom) components of the acceleration observed at the basement of the damaged apartment house. Dashed lines represent the components at the roof of the apartment house Fig. 8 Sections of NS velocity seismograms of the 2003 (Kudo et al. 2000) Tokachi-oki earthquake observed at the stations along the southern coast of Hokkaido (Koketsu et al. 2005)

(Fig. 9). Yamada and Iwata (2005) demonstrated that An inland crustal earthquake called the 1999 a low-velocity accretionary prism in this subduction Hector Mine earthquake (MW=7.1) generated far- zone worked as a waveguide for seismic energy source long-period ground motions in the San transfer. Shapiro et al. (2000) pointed out this path Bernardino basin 100 km from the earthquake effect for the Mexican subduction zone. (Fig. 10). Graves and Wald (2004) attributed the In addition, site conditions of a sedimentary basin main characteristics of these motions, such as ampli- also affect far-source long-period ground motions fication and surface waves, to strong basin response (Miyake and Koketsu 2005). This can be confirmed effects. They then developed the 3D velocity structure by the distributions of pseudo-velocity response model in this region by comparing the observed and spectra observed during the earthquake. The velocity simulated long-period ground motions. The 1999 Chi- structure in the Kanto basin under Tokyo developed Chi earthquake (MW=7.6) provided further examples ground motions at periods of 7 to 10 s. As the Osaka and Nobi basins under Osaka and Nagoya have thinner sediments and a smaller extent, the peak responses appear at shorter periods (Kawabe and Kamae 2008; Iwaki and Iwata 2008). Mamula et al. (1984) investigated these types of site conditions in sedi- mentary basins all over Japan, using ground motions from the 1961 Kita Mino earthquake (M=7.0).

Fig. 9 Section of EW acceleration seismograms observed at Fig. 7 Ground accelerations observed in the Niigata basin stations along a propagation path from the 2004 off Kii 500 km from the 1993 Hokkaido Nansei-oki earthquake (Zama Peninsula earthquake to the Kanto basin (Miyake and Koketsu and Inoue 1994) 2005) J Seismol (2008) 12:133–143 139

in the distant Taipei and Ilan basins (Furumura et al. 2002).

The 2004 Chuetsu earthquake (MW=6.6) in the central part of Japan also generated far-source long- period ground motions in a large basin 200 km from the earthquake (Furumura 2005; Furumura and Hayakawa 2007). A main wave train from the earthquake was separated into S and surface waves at the northern margin of the Kanto basin, and the surface wave was developed into far-source long- period ground motion in propagation toward Tokyo at the center of the basin (Fig. 11). The ground motion damaged an elevator in a new super high-rise building in downtown Tokyo. This secondary surface wave occurring in California was thoroughly investigated by Joyner (2000).

4 Near-fault long-period ground motion

The other type of long-period ground motion is found in the near-field of an earthquake source fault. This near-fault long-period ground motion for an inland Fig. 10 Ground velocities for stations located along a profile from the Hector Mine earthquake at a 154 azimuth. The San crustal earthquake has been widely discussed with the Andreas (SAF) and San Jacinto faults indicated by the arrows advance of source theory and interpretation. Aki bound the basin (Graves and Wald 2004; © Seismological (1968) was one of the earliest examples of the study Society of America) of near-fault ground motion with the finite-source

Fig. 11 Record section of ground velocity motions from the 2004 Chuetsu earthquake toward Tokyo at the center of the Kanto basin (Furumura 2005) 140 J Seismol (2008) 12:133–143

pulse has a dominant period of 0.6 s or longer, and the velocity response in the fault-normal components is larger than in the fault-parallel component. Recent studies on attenuation relationships tried to incorpo- rate this near-fault rupture directivity effect into the long-period component. Near-fault long-period ground motions for large earthquakes were also observed during the 1999 Kocaeli earthquake and the 1999 Chi-Chi earthquake. The dominant period of these velocity pulses was longer than 3 s. Based on the compilation of long-period pulses, Somerville (2003) found that the dominant period of the rupture directivity pulse can be expressed as a function of earthquake magnitude. The rupture directivity pulse with a period of around 1 s dominates near-fault long-period ground motions for earthquakes of moderate magnitude. The rupture directivity pulses of the 1994 Northridge and Fig. 12 Near-fault ground motions at a distance of 80 m from 1995 Kobe earthquakes (MW=6.6 and 6.9, respec- the 1966 Parkfield earthquake (Aki 1968; © American tively) have been examined most thoroughly. The Geophysical Union) rupture directivity effects from the source and basin edge effects caused severe damage in the San theory. He recovered the long-period ground motions Fernando and Kobe regions. With the help of detailed shown in Fig. 12 from strong motion accelerograms models of the source processes and velocity struc- observed at a distance of only 80 m from the source tures, deterministic ground motion simulations repro- fault of the 1966 Parkfield earthquake. He then duced the rupture directivity pulses well (e.g., Graves compared these long-period ground motions with et al. 1998; Furumura and Koketsu 1998; Pitarka et al. theoretical ground motions computed for unilaterally 1998; Iwata et al. 1999). propagating fault rupture. This favorable comparison The impact of near-fault long-period ground led to the establishment of earthquake source theory. motion on structures has been described by Heaton Hanks (1975) proved that strong motion accelero- et al. (1995). Just after the 1994 Northridge earth- grams could provide sufficiently accurate long-period quake, they focused on near-fault long-period ground ground motions for the analysis of earthquake source motions with dominant periods of 1 s or longer. They parameters and velocity structures deterministically. examined the responses of high-rise and base-isolated He showed 234 components of near-fault long-period buildings to large displacements and ground velocities ground motion within the source region of the 1971 because of a hypothetical MW=7.0 blind thrust San Fernando earthquake and the neighboring Los earthquake. As a result of the very long recurrence Angeles basin. These components exhibit some time of large crustal earthquakes, deterministic complexities reflecting the details of the earthquake ground motion simulation for earthquake scenarios source and ground motion propagation. has been a useful tool for estimating the ground The rupture directivity pulse is one of the most response of a future large earthquake. Olsen et al. significant features of near-fault long-period ground (2006, 2008) performed large-scale ground motion motions. The ground motion characteristic of the simulations (TeraShake) for scenario earthquakes of rupture directivity pulse is well summarized by M=7.7 along the southern San Andreas fault. They Somerville (2003; Fig. 13). During the 1992 Landers simulated the rupture directivity effect and its modi- earthquake (Mw=7.2), near-fault long-period ground fication by interactions with a chain of sedimentary motion because of rupture directivity effects was basins. Love waves with a dominant period of 4.5 s clearly observed at Lucerne station. Somerville et al. are channeled into the Los Angeles basin and excite (1997) pointed out that the forward rupture directivity strong far-source long-period ground motions. J Seismol (2008) 12:133–143 141

Fig. 13 Recent observation of rupture directivity pulses. The traces show fault- normal velocity ground motions at fault distances of 4 (Loma Prieta), 1 (Landers), 1 (Kobe), 5 (Kocaeli, Turkey), 7 (Northridge), and 6 km (Chi-Chi, Taiwan) (Somerville 2003; reprinted with permission from Elsevier)

Regarding subduction-zone earthquakes, near-fault 5 Conclusions and discussion long-period ground motions from a megathrust earthquake were evaluated for the 1923 Kanto We divided long-period ground motions into far- earthquake. Both long-period features of the earth- source and near-fault classes. Most far-source long- quake source just beneath the Tokyo metropolitan period ground motions were generated by large area and ground motion response because of a deep earthquakes and effective propagation paths, such as sedimentary basin resulted in a significant level of accretionary prisms. Therefore, far-source long-period long-period ground motion, as reported by Sato et al. ground motions are generally associated with offshore (1999). As there were no super high-rise buildings or earthquakes in subduction zones. However, inland huge oil tanks at the time of the 1923 Kanto crustal earthquakes of moderate size, such as the 1999 earthquake, the damage caused by near-fault long- Hector Mine earthquake and the 2004 Chuetsu earth- period ground motion was small (Table 1; Hirano quakes, have also generated far-source long-period 1982). However, the above-mentioned report cautions ground motions in distant basins because of signifi- that damaging long-period ground motion caused by a cant basin response effects. In the vicinity of future megathrust earthquake near an urban area is a earthquake source faults, near-fault long-period considerable threat to countries in and around ground motions are generated mainly by rupture subduction zones. directivity effects. Accordingly, they consist primarily 142 J Seismol (2008) 12:133–143 of rupture directivity pulses, which can be damaging, International Workshop on Strong ground Motion Prediction – especially when combined with site effects and/or and Earthquake Tectonics in Urban Areas, pp 143 148 Furumura T, Hayakawa T (2007) Anomalous propagation of basin edge effects. long-period ground motions recorded in Tokyo during the Far-source long-period ground motions consist 23 October 2004 MW 6.6 Niitgata-ken Chuetsu, Japan, primarily of surface waves excited by these path and earthquake. Bull Seismol Soc Am 97:863–880 site effects, and have longer durations than near-fault Furumura T, Kennett BLN (1998) On the nature of regional seismic phases—III. The influence of crustal heterogeneity long-period ground motions. Far-source and near- on the wavefield for subduction earthquakes: the 1985 fault long-period ground motions were first identified Michoacan and 1995 Copala, Guerrero, Mexico earth- in the records of the 1968 Tokachi-oki and 1966 quakes. Geophys J Int 135:1060–1084 Parkfield earthquakes, respectively. The Mexico City Furumura T, Koketsu K (1998) Specific distribution of ground motion during the 1995 Kobe earthquake and its genera- records of the 1985 Michoacan earthquake lead to far- tion mechanism. Geophys Res Lett 25:785–788 source long-period ground motions being known Furumura T, Koketsu K, Wen K-L (2002) Parallel PSM/FDM around the world, whereas knowledge of near-fault hybrid simulation of ground motions from the 1999 Chi-Chi, – long-period ground motions was dispersed through Taiwan, earthquake. Pure Appl Geophys 159:2133 2146 Graves RW, Wald DJ (2004) Observed and simulated ground the records of the 1992 Landers earthquake. Even in motions in the San Bernardino basin region for the Hector the absence of a seismographic record, we can Mine, California, earthquake. Bull Seismol Soc Am confirm the generation of damaging long-period 94:131–146 ground motions based on tank damage because of Graves RW, Pitarka A, Somerville PG (1998) Ground-motion amplification in the Santa Monica area: Effects of shallow liquid sloshing. basin-edge structure. Bull Seismol Soc Am 88:1224–1242 Hanks TC (1975) Strong ground motion of the San Fernando, Acknowledgment This study was supported in part by the California, earthquake: ground displacements. Bull Seismol SCEC-ERI Cooperation Program, Grants-in-Aid for Scientific Soc Am 65:193–225 Research (C) no. 18631006 and (A) no. 19201034 from the Hatayama K (2008) Lessons from the 2003 Tokachi-oki, Japan, Japan Society for the Promotion of Science, and Special earthquake for prediction of long-period strong ground Coordination Funds for Promoting Science and Technology motions and sloshing damage to oil storage tanks. J from the Japan Science and Technology Agency. Takeshi Seismol (this issue) Kimura kindly helped improve several figures. We also thank Heaton TH, Hall DF, Wald DJ, Halling MW (1995) Response Kim Olsen, Steven Day, Naomi Sakajiri, Kazuyoshi Kudo, of high-rise and base-isolated buildings to a hypothetical Shinsaku Zama, and Takashi Furumura for their reviews and M 7.0 blind thrust earthquake. Science 267:206–211 figures. Hirano M (1982) The great Kanto earthquake disaster and Yokosuka naval port. Suiro 41 (in Japanese) Horoun MA (1983) Behavior of unanchored oil storage tanks: Imperial Valley earthquake. J Tech Topics Civil Eng 109:23–40 References Iwaki A, Iwata T (2008) Validation of 3D basin structure models for long-period ground motion simulation in the Aki K (1968) Seismic displacements near a fault. J Geophys Osaka basin, western Japan. J Seismol (this issue) Res 73:5359–5376 Iwata T, Sekiguchi H, Pitarka A, Irikura K (1999) Ground motion Beck JL, Hall JF (1986) Factors contributing to the catastrophe simulations in the Kobe area during the 1995 Hyogoken- in Mexico City during the earthquake of September 19, Nanbu earthquake. In: Irikura K, Kudo K, Okada H, 1985. Geophys Res Lett 13:593–596 Sasatani S (eds) The effects of surface geology on seismic Celebi M, Prince J, Dietel C, Onate M, Chavez G (1987) The motion, vol. 3. Balkema, Rotterdam, pp 1369–1376 culprit in Mexico City—amplification of motions. Earthq JSCE (Japan Society of Civil Engineers) (2000) Damage to Spectra 3:315–328 industrial facilities. Kocaeli (Turkey) Earthquake, JSCE, Dziewonski AM, Franz JE, Woodhouse JH (1983) Centroid- Tokyo 7:1–20 moment tensor solutions for April–June, 1983. Phys Earth Joyner WB (2000) Strong motion from surface waves in deep Planet Inter 33:243–249 sedimentary basins. Bull Seismol Soc Am 90:S95–S112 FDMA (Fire and Disaster Management Agency) (1965) Study Kanamori H (1979) A semi-empirical approach to prediction of on the 1964 Niigata Earthquake Fire. FDMA, Tokyo, p long-period ground motions from great earthquakes. Bull 224 (in Japanese) Seismol Soc Am 69:1645–1670 FDMA (Fire and Disaster Management Agency) (1979) Report Kawabe K, Kamae K (2008) Prediction of long-period ground on damage of oil storage tanks in the Sendai refinery. motions from huge subduction earthquakes in Osaka, FDMA, Tokyo (in Japanese) Japan. J Seismol (this issue) Fukuwa N (2008) Key parameters governing the dynamic Kawase H, Aki K (1989) A study on the response of a soft response of long-period structures. J Seismol (this issue) basin for incident S, P, and Rayleigh waves with special Furumura T (2005) Computer simulation of long-period ground reference to the long duration observed in Mexico City. motions associated by large subduction zone earthquakes. 2nd Bull Seismol Soc Am 79:1361–1382 J Seismol (2008) 12:133–143 143

Koketsu K, Hatayama K, Furumura T, Ikegami Y, Akiyama S Seno T, Shimazaki K, Somerville P, Sudo K, Eguchi T (1980) (2005) Damaging long-period ground motions from the Rupture process of the Miyagi-oki, Japan, earthquake of 2003 Mw 8.3 Tokachi-oki, Japan, earthquake. Seismol Res June 12, 1978. Phys Earth Planet Int 23:39–61 Lett 76:67–73 Shapiro NM, Olsen KB, Singh SK (2000) Wave-guide effects Kudo K, Sakaue M (1984) Oil-sloshing in the huge tanks at in subduction zones: evidence from three-dimensional Niigata due to the Nihonkai-chubu earthquake of 1983. modeling. Geophys Res Lett 27:433–436 Bull Earthq Res Inst Univ Tokyo 59:361–381 Shapiro NM, Olsen KB, Singh SK (2002) On the duration of Kudo K, Uetake T, Kanno T (2000) Re-evaluation of nonlinear seismic motion incident onto the Valley of Mexico for site response during the 1964 Niigata earthquake using the subduction zone earthquakes. Geophy J Int 151:501–510 strong motion records at Kawagishi-cho, Niigata City. Shibata H (1971) General aspects of San Fernando earthquakes, Proc 12th World Conf Earthq Eng no.0969 Feb. 9, 1971. Seisan Kenkyu 23:314–316 (in Japanese) Mamula L, Kudo K, Shima E (1984) Distribution of ground-motion Shibata H (1974) Survey report on earthquake damage of amplification factors as a function of period (3–15 sec), in industrial facilities in the world—spherical vessels and Japan. Bull Earthq Res Inst Univ Tokyo 59:467–500 cylindrical vessels. Seisan Kenkyu 26:259–264 (in Japanese) Miyake H, Koketsu K (2005) Long-period ground motions from a Shima E (1970) Seismic surface waves detected by the strong large offshore earthquake: the case of the 2004 off the Kii motion acceleration seismograph. Proc 3rd Jpn Earthq Eng peninsula earthquake, Japan. Earth Planets Space 57:203–207 Symp, pp 277–284 Manos GC, Clough RW (1985) Tank damage during the May 1983 Somerville PG (2003) Magnitude scaling of the near fault rupture Coalinga earthquake. Earthq Eng Struct Dyn 13:449–466 directivity pulse. Phys Earth Planet Inter 137:201–212 Ohta T, Zama S (2005) Large earthquakes and large-scale Somerville PG, Smith NF, Graves RW, Abrahamson NA (1997) structures. Kyoritsu Shuppan, Tokyo, p 287 (in Japanese) A modification of empirical strong ground motion atten- Olsen KB, Day SM, Minster JB, Cui Y, Chourasia A, Faerman uation relations to include the amplitude and duration M, Moore R, Maechling P, Jordan T (2006) Strong effects of rupture directivity. Seismol Res Lett 68:199–222 shaking in Los Angeles expected from southern San Steinbrugge KV (1970) Earthquake damage and structural perfor- Andreas earthquake. Geophys Res Lett 33:L07305 mance in the United States. In: Wiegel RL (ed) Earthquake Olsen KB, Day SM, Minster JB, Cui Y, Chourasia A, Okaya D, engineering. Prentice-Hall, Englewood Cliffs, pp 167–226 Maechling P, Jordan T (2008) TeraShake2: simulation of Steinbrugge KV, Moran DF (1954) An engineering study of the Mw7.7 earthquakes on the southern San Andreas fault southern California earthquake of July 21, 1952 and its with spontaneous rupture description. Bull Seismol Soc aftershocks. Bull Seismol Soc Am 44:201–462 Am (in press) Trifunac MD, Brune JN (1970) Complexity of energy release Pitarka A, Irikura K, Iwata T, Sekiguchi H (1998) Three- during the Imperial Valley, California, earthquake of 1940. dimensional simulation of the near-fault ground motion for Bull Seismol Soc Am 60:137–160 the 1995 Hyogo-Ken Nanbu (Kobe), Japan, earthquake. Wald DJ, Somerville PG (1995) Variable-slip rupture model of Bull Seismol Soc Am 88:428–440 the great 1923 Kanto, Japan, earthquake: Geodetic and body- Rinne JE (1967) Oil storage tanks. In: Wood FJ (ed) The Prince waveform analysis. Bull Seismol Soc Am 85:159–177 William Sound, Alaska, Earthquake of 1964 and Aftershocks. Yamada N, Iwata T (2005) Long-period ground motion vol. 2A. US Government Printing Office, Washington, 2A: simulation in the Kinki area during the MJ 7.1 foreshock 245–252 of the 2004 off the Kii peninsula earthquakes. Earth Ruff L, Kanamori H (1983) The rupture process and asperity Planets Space 57:197–202 distribution of three great earthquakes from long-period Yoshida S, Zama S et al (2000) Report on damage to oil storage diffracted P-waves. Phys Earth Planet Inter 31:202–230 tanks by the Chi-Chi, Taiwan, earthquake. J High Pressure Sakajiri N, Naruse S, Takeuchi F, Yoshikawa K, Goto N, Ohta Y Inst Jpn 38(6):23–34 (1974) Observation of 1- to 5-sec microtremors and their Yoshiwara H, Zama S, Kamei A (1984) Sloshing and its damage application to , Part I. Preliminary in oil storage tanks due to the 1983 Japan Sea earthquake. observation in Hachinohe. Zisin 27:338–351 Tech Rep Nat Res Inst Fire Disas 14:31–46 (in Japanese) Sato T, Graves RW, Somerville PG (1999) Three-dimensional Zama S (1993) Long-period strong ground motion. Zisin finite-difference simulations of long-period strong motions 46:329–342 (in Japanese) in the Tokyo metropolitan area during the 1990 Odawara Zama S, Inoue R (1994) Sloshing of liquid in oil storage tanks earthquake (MJ 5.1) and the great 1923 Kanto earthquake due to the 1993 Hokkaido Nansei-oki earthquake. Rep (MS 8.2) in Japan. Bull Seismol Soc Am 89:579–607 Fire Res Inst Jpn 77:1–10 (in Japanese)