Japan Hokkaido Nansei Oki Eer

Home , 1993 Kushiro earthquake, Okushiri Island

EERI SPECIAL EARTHQUAKE REPORT -AUGUST 1993

2 EERI SPECIAL EARTHQUAKE REPORT-AUGUST 1993

of Okushiri Island which lies west ~ARTHQUAKE MECHANISM of the Hokkaido mainland and south-southwest of the epicenter ..,-- --,.- , -, ,. -,. ", Shakotan-oki earthquake, which The earthquake began on July 12 (Figures 3a and 3b). The amount at 22: 17 local time at latitude 42 occurred immediately to the north of subsidence initially ranged from 47' North, 139° 12' E at a depth of the 1993 event, had a similar about 20 cm to 100 cm, but about of 27 km. The aftershocks define focal mechanism, but smaller mag- half of that subsidence has been a plane about 150 km long (north nitude and fault dimensions (Sata- recovered by uplift of the island. south) and 50 km wide (east-west) ke 1986). The occurrence of the Extensive marine seismic surveys dipping eastward at a shallow 1983 event led to the suggestion by the Japan Maritime Safety angle (about 10 ° ) , with most of of nascent subduction along the Agency following the earthquake the aftershock foci occurring at Japan Sea coast off northern Hon- have revealed extensive imbricate I shu (Kobayashi, 1983; Nakamura, depths between 10 km and 20 km , faulting of the sea floor extending based on preliminary information 1983) which the 1993 earthquake from east of Okushiri Island west- from Hokkaido University (Figure tends to confirm. The Sea of ward to the inferred surface ex- 1 ). A centroid depth of 1 5 km (a Japan side of this plate boundary pression of the subduction zone. 1"point source representative of the is the Eurasian plate" After the It is inferred that Okushiri Island is overall faulting) was estimated by 1983 earthquake, the northern located on a graben structure with- Harvard. The seismic moment of Japan side of the plate boundary in the accretionary wedge of the 5.6 x 1027 dyne cm corresponds tO was initially interpreted as the overriding plate and that this gra- an Mw of 7.8. Assuming a fault North American plate, but it may ben structure experienced coseis- length of about 150 km and a fault instead be the recently proposed mic subsidence. The sea floor width of 50 km based on after Okhotsk plate (Seno, 1987; De- structures discovered during this shock data, this moment and fault Mets et al., 1992)" survey resemble those that have geometry indicates an average slip been found in the sea floor off the of 2.5 m on the fault plane. The The earthquake caused subsidence Pacific coast of Japan. D ( ~MT focal mechanism estimated / ~ North AmeriCan-* J>y Harvard, representing the fault- / ~ or Okhotsk?) ,cI( I Plate f - ing process as a whole, has one / I nodal plane dipping eastward at 24 ° from the horizontal. This HOKKAIDO ( fault plane orientation is compat- ible with the aftershock distribu- 12 July 1993 M:7.8 tion and consistent with subduction of the floor of the Sea of v.JI Japan beneath northern Japan. However, the source process Eurasian Plate seems to have been complex, with the initial rupture occurring on a Plate Boundaries 78 steeper eastward dipping plane or even possibly on a shallow west- () ward dipping plane (Tanioka et al., Pacific Plate 1993; Kikuchi, 1993).

\ 2~ The subduction plate boundary along the Japan Sea coast of SHI~ northern Honshu and Hokkaido has J ~ been recognized only recently KYUSHU (Figure 2). The 1983 Nihon Kai , 48 Chubu earthquake occurred imme- 68/// '--- Philippine Sea Plate diately to the south of the 1993 ,event; the magnitude, fault dimen- 0 SOOkm )ions, and focal mechanism were d very similar to those of the 1993 Figure 2-Rupture zones of offshore earthquakes in Japan larger than earthquake (Satake, 1985; Kana- magnitude 7.4 since 1923. Source: modified from Japan Meteorological mori and Astiz, 1985). The 1940 . Agency.

~ EERI SPECIAL EARTHQU:AKE REPORT -AUGUST' 993

Map of Sea Floor Around Okushiri Island Inferred Cross Section of Okushiri Island and \ i I Coseismic Crustal Deformation Model \ y»" ~~EPicenter of Hokkaido ~~ ~ INanseiOki Okushiri Island Sea level

:~~~~~~~~~~~:::---O \;:.:- ,~a~t:~ apan ( Sea \, . I

~ ~ ~~ ~ 0'""'0'0000000'"0000

'Epicenter .--i :; , , -~ " .:Y', ~ -\;~;ijo~*i!io!o.'°'",""1 Okushiri Ridge IStudy AreC" ) \ Eurasian --I ESA h, Plate /6 North American Plate -' I ~~r~~t~~;t-"ke;~~i] Main Fault / / ./ Major Active Fault \(i Figure 3 -Sea floor faults inferred from marine surveys after the ,,:::::;::::::Large Scale Landslide Scar ,,~i;,\~;;': ":::::::::" earthquake. Source: Yomiuri Shlmbur, Japan MaritIme Safety Agency

RECONNAISSANCE TEAM

A combined EERI-UJNR team sur- to the field led by Yutaka lida the tsunami report, which was co- veyed damage and other effects (PWRI), leader, Kazuhiko authored by E. Bernard, F. Gonza- from the earthquake and tsunami. Kawashima (Head of the Earth- lez, and D. Sigrist of the U.S. side The team arrived in Japan on July quake Engineering Division, PWRI), of the UJNR team, and Hiroichi 18, and conducted field reconnais- Susumu lai (Head, Geotechnical Tsuruya, Port and Harbor Research sance and data gathering between Earthquake Engineering, P&HRI) S. Institute, Ministry of Transport, July 20 and 24. Unjoh (PWRI), T. Nakajima (PWRI), and Kenji Kato, University of and R. Tsunaki (PWRI). Tokyo, of the Japan side of the EERI team members included Les UJNR team, and the landslides Youd, Brigham Young University, The acronym UJNR refers to the report, which was authored by E. EERI team leader (geotechnical U.S.-Japan Panel on Natural Harp, USGS, of the UJNR team. D aspects); Jane Preuss, Urban Resources, a cooperative inter- Regional Research (tsunamis, land national program that celebrated use and response planning); Paul its 25th year in 1993. The UJNR Somerville, Woodward-Clyde Panel on Wind and Seismic Effects Consultants (seismology) ; and is one aspect of the larger UNJR Charles Scawthorn, EOE Inc. program. Members of the Japan- (structural engineering and fire side Wind and Seismic Panel response). Other members of the arranged outstanding logistical The research, publication, and UJNR team included Riley Chung support for the U.S. team, for distrlDution of this report was (NIST), leader; Richard Bukowski which all members were very funded by National Science (NIST); Eddie Bernard (NOAA); grateful. The UJNR cooperation Foundation Grant #BCS-9215158 Frank Gonzalez (NOAA); Dennis also made it possible for the U.S. as part of the EERI Learning from Sigrist (NOAA); Charles Barnes team members to work closely Earthquakes Project. (American Plywood Association); with Japanese investigators. David Tyree (American Forest and Complete copies of reconnaissancefi Paper Association); Edwin Harp The reports in this special edition reports submitted to EERI are V (USGS), and Peter Yanev, EOE Inc of the EERI newsletter were archived at the EERCINISEE lIbrary, (NCEER). Several Japanese authored by the EERI team Call EERC at 510-231-9401 for I engineers accompanied the team 'members, with the exception of further information.

~ EERI SPECIAL EARTHQUAKE REPORT -AUGUST 1993

damage, and few gravestones ( ~TRONG GROUND MOTION were overturned, suggesting that the ground motions in this region Strong ground motions were re- island lies just east of the lower were not unusually large. corded on a variety of instrument edge of the rupture surface, which types by numerous agencies in- had a depth of about 20 km. At Strong ground motion recordings cluding the Hokkaido Development the Aonae cemetery, at the south were made at several localities in Department, the Japan Meteoro- end of the island, 80% of the the eastern part of southwestern logical Agency I Japan Rail, the tombstones were overturned, and Hokkaido and farther to the north Ministry of Construction, and the in houses furniture was over- near Sapporo. The largest peak Port and Harbor Research Institute. turned, suggesting peak accelera- acceleration, slightly less than There were no strong motion re- tions in the range of 0.4g to 0.5g. 0.5g, was recorded at Kuromat- cordings from Okushiri Island or On the mainland of Hokkaido be- sunai which lies midway between the immediately adjacent south- tween Esashi and Motsutanosaki, Suttsu on the Japan Sea coast and west coast of Hokkaido. Okushiri there was no major structural Oshamanbe on Uchiura Bay. That locality is about 80 km from the nearest point on the fault. This 1 recording had an unusually large ratio of peak horizontal to vertical motions (a factor of about 5). 0 -Croule Suttsu, which is at a similar dis- -Youngl tance, had peak horizontal accel- -Cohee erations of about 0.2g and peak 00 displacements of about 8 cm. The ,0Q:DoO duration of strong ground motion 00 '0 was approximately one to two O~ minutes at most locations within ( ~ 0 .1 100 km of the rupture surface, as ~ 0 0 expected from an earthquake of :,j II 0 '\ ~ this size. The duration of high GI 0 - GI frequency motion was markedly CJ

~ longer at stations east of the epicenter than at stations to the ~ 0 § southeast, which may reflect the

.~ effect of rupture directivity from 0 00 = 0 north to south. 0 -= o~ GI 0. 01 ~ Very preliminary peak horizontal acceleration values without instru-

00 ment corrections are plotted as a function of closest distance to the 0 fault in Figure 4; these values will change after the data have been corrected and checked. Theyare compared with attenuation relations for subduction earthquakes 0.001 I for soil sites that are in current use 10 100 in the United States (Cohee et al., Closest Distance (km) 1991; Crouse, 1991; Youngs et al., 1988). The peak accelerations Figure 4 -Very preliminary uncorrected peak horizontal acceleration recorded during the Hokkaido- ~ ~' lotted ~gainst c~osest distance t,o the rupture surface, , compare.d with nansei-oki earthquake are generally :1 ttenuat/on relat/ons for subduct/on earthquakes used /n the UnIted larger in the distance range of 70 States. Sources: Hokkaido Development Agency, Japan Meteorological to 1 50 km and attenuate more Agency, Japan Rail, Ministry of Construction, Port and Harbor Research rapidly than the attenuation i Institute. ; relations. D

5 EERI SPECIAL EARTHQUAKE REPORT-AUGUST 1993

locations --on the beach, in TSUNAMI wrecked automobiles, and in -, homes and other buildings --whichU' A major tsunami was generated killed by salt water. Measure- are taken to be a rough indication and, within two to five minutes, ments of vertical height above sea of tsunami arrival times. extremely large waves engulfed level of the tsunami trace are the Okushiri coastline and the obtained by a series of measure- The tsunami was probably gener- central west coast of Hokkaido. ments down to the shoreline, using ated within the deformation area As of July 21, 185 fatalities were a surveyor's staff and level. The as defined by the aftershock pat- confirmed, with 120 attributed to effects of tides are normally tern shown in Figure 1 .Eyewit- the tsunami; this death toll is removed from such measurements, ness accounts collected by Y. Tsuji expected to rise, as missing per- but during the period of the (University of Tokyo) and F. Ima- sons are included among the fatali- survey, the predicted tidal range in mura (Tohoku University) indicate ties (Hokkaido Police Headquar- the study area was only 2-30 cm, that the tsunami hit the west coast ters) .Property losses have been referred to mean sea level. Since of Okushiri Island almost immedi- estimated at $60 million, due this range is a relatively small ately after the main shock. The principally to tsunami damage. fraction of the observed runup, the UJNR tsunami survey team found the electrical The Japanese clock at an elec- immediately dis- trical power sta- patched damage tion had stopped assessment and at 22:23, which survey teams. was 6 min after Most of these the start of the Japanese teams earthquake. were mobilized Along the west and began sur- coast, the tsu- veying tsunami nami runup runup by July measurements 13. On July 18, were the high- three U.S. scien- est from north tists joined the of Monai. The Japanese scien- village of Monai tists to comple- was totally des- ment the tsuna- troyed ( 10 per- mi survey teams. sons killed and A combined US- all 1 2 houses Japan team. . destroyed) . worked together F,gure 5 -Runup of 31 m was measured north of Monal. Photo: D. Sigrist Runup measure-

under the aus- ments around pices of the U.S.-Japan data presented in this report are the village were 20 m, and in a Cooperative Program in Natural not corrected for the tides. The small valley north of Monai, the Resources (UJNR). (See Recon- UJNR survey team made measure- runup was measured at 31 m. naissance Team, po 4, for names ments at 28 sites --24 on Okushiri Figure 5 shows the survey team in of team members.) Island and 4 on the west coast of action in the valley where the Hokkaido from July 20-23, 1993. maximum runup of 31 m was mea- The primary measurements in the sured. The staff in the photograph field surveys were the tsunami Another source of information is is 25 ft high, and the vegetation vertical and/or horizontal runup the approximate time of tsunami on both sides of the valley shows values, which are the maximum arrival, as inferred from the time at evidence of flooding. South of vertical height and/or horizontal which clocks have stopped due to Monai, tsunami runups between extent of flooding. In this report, saltwater flooding. The UJNR 15 and 20 m were measured all only vertical runup estimates are team conducted an active search along the coast. Vegetation was I presented. Traces left by the tsu- for such clocks in the tsunami stripped off the hillside, and large nami include water marks on build- debris and in the surrounding area boulders (up to 1 m in diameter) ings, debris lines along the coast, 'of each of the 28 sites visited. were deposited where the vegeta- or vegetation that is damaged or Eight clocks were found in various tion was flooded. These data are

6 EERI SPECIAL EARTHQUAKE REPORT -AUGUST 1993

consistent with the initial wave peninsula. The UJNR team ob- (according to Y. Tsuji) 13 persons ~rriving from the west of Okushiri served that runup values rose were killed and all houses were (if ,tery near the generation area . rapidly again to the 15-20 m level destroyed by waves that ran up a short distance northeast of over 1 O m. The eastern side of The tsunami was refracted by the Acnae; this is undoubtedly due to Okushiri Island was less affected shoaling bathymetry at both ends the absence of breakwaters or where the tsunami runup was of the Island. Hardest hit was the sand dunes along this part of measured between 2 and 5 m. town of Aonae (population 1,600), exposed coast. Subsidence was observed by Tsuji where the first tsunami wave at Acnae, Okushiri, and Monai, flooded the southern tip of the The extent of damage to Aonae is while uplift was observed at Inaho island and the entire first row of illustrated in Figure 6. The Point. These data again suggest houses in the harbor area within 4- breakwater in the top right corner that Okushiri Island was part of the 5 min after the main shock (see is 4.5 m. The houses in the deformation area, which is consis- Figure 6) .Tsuji reported (based on central part of the photograph tent with the aftershock data. eyewitness interviews) that the were flooded by tsunami waves tsunami arrived from the north- that ran up to a height of 5 m. The tsunami also hit the island of east, with flooding of 3- 7 m The area in the top left portion of Hokkaido, arriving at Ota Bay with- throughout the in 5 min of the town. About 7 main shock and min after the destroying five first wave, a " homes. The second, larger UJNR team mea- wave hit from sured runup in the east carrying this area at 9 m. boats into the Damage was al- main town. The so observed by second wave the UJNR team ~ ~ompletely at Setana due to flooded the first 6 m runup three rows of waves. The houses, and run- coastline from up was mea- Suttsu to south sured around 5- of Ota Bay was 10 m throughout hardest hit with the town. The runup values of UJNR survey 5-9 m. Outside team found bat- this area, the tery-operated I tsu-nami clocks in this Figure 6- Earthquake, tsunami: and fire damage in Aonae Photo: D. Sigrist intensity tapered area that had off rapidly, and stopped at 22:37 and 22:38. At the photograph had runup values runup values fall below 5 m. The 22:40, fires broke out; the combi- of 10 m. The tsunami destroyed a closest tide gauge to the nation of a strong northeast wind portion of the seawall to the right earthquake was Esashi (Japan and an ample supply of propane of the remaining building in the .Meteorological Agency), which and kerosene (used for heating) center of Figure 6. Notice that the recorded a 2-m wave approximate- Quickly spread the fire, which wave forces transported the barge ly 10 min after the main shock. burned throughout the night and with a crane from the harbor to Those data suggest that the east- destroyed 340 homes. Autopsies the center of the photograph . ern portion of the source was be- revealed that only 2 of the 114 Remnants of homes damaged by tween Okushiri Island and Hokkai- deaths in Aonae were caused by the quake, tsunami, and fire are do. The tsunami propagated to fire. This section of Aonae was seen throughout the harbor area Russia within 30 min, where 1-4 m the hardest-hit developed area in toward the land. The photographs tsunami runups were reported by ii -}'Wpite of the fact that a massive, were taken on July 20, 1993. Valentin Fedorey (Hydrometl. Af- 'V4.5 m breakwater and 10 m high ter 90 min, the tsunami struck the sand dunes were very effective in A similar refraction of the tsunami coast of South Korea, where B.H. reducing the runup to 5-10 m tools place on the northern point of Choi of Sung Kyun Kwan Univ. mea- along the southeastern tip of the Okushiri Island at Inaho where sured tsunami runup of 1-2 m. O

7 EERI SPECIAL EARTHQUAKE REPORT -AUGUST 1993

8 EERI SPECIAL EARTHQUAKE REPORT -AUGUST 1993

Warnings and Responses stopped. A second and larger inside their houses. Interestingly, wave arrived after the first waves. the water velocity was quite low, although the wave was approxi- Five minutes after the earthquake, With the initiation of strong shak- mately 3 meters high (it apparently which occurred at 10: 1 7 pm local ing, people reported that they passed over the houses) .The first time, the Japan Meteorological began to run immediately for wave arrived 1 to 2 minutes after Agency (JMA) issued a warning higher ground (not more than 100 the shaking stopped and did not that a major (over 3-meter high) m from most residences) .There recede. The next wave arrived tsunami had been generated. The was one reported incident, on the about 4 minutes later and rushed JMA transmitted the warning via 7 east coast in the vicinity of over the top of the first. different television and radio Torigashira, of a man who went stations to local jurisdictions on back to look between the first and In Aonae, people reported remem- Hokkaido and northern Honshu. In second waves and never returned . bering the tsunami of 1983, and addition, N TV issued a notice 1.5 therefore, when they felt the minutes after the earthquake that Interviews with residents in the strong earthquake, they knew that a possible tsunami had been northern and eastern parts of they needed to run very quickly to generated. Okushiri Island indicate that they the higher ground (as reported in felt the shaking, looked up, and interviews with adults, including The JMA has six local centers for saw the first wave arriving. In one elderly respondent) .The first issuing tsunami warnings: sap- Yamasedomari, the first wave wave arrived approximately 5 min- poro, Sendai, Tokyo, Osaka, Fuku- arrived approximately 3 minutes utes after the shaking; the second oka and Naha. Each local center is after the shaking began. However , arrived about 3 minutes later . responsible for issuing a warnings a larger second wave arrived for its region. Tsunami warnings approximately 1 minute later . When the shaking occurred, fire pre gra~ed according to three fighters in Aonae immediately Ucategorles: In Yamasedomari, the wave height drove to the waterfront area to was not high, and the people did warn residents to evacuate be- .Major tsunami -over 3 m high not evacuate, but watched from cause of a possible tsunami. As .Tsunami warning -1 m to 2 m .Tsunami alert -0.5 m to 1 m Table 1: Time-History of Tsunami Warnings Table 1 lists the times of tsunami warnings issued by Sapporo Dis- trict Meteorological Observatory Time Action (DMO) and by the Japan Meteoro- 22:17 Earthquake occurred. logical Agency (JMA). As indicated in the table, five minutes 22:18 N TV (local television) issued a notice that a possible after the earthquake, Sapporo 30 tsunami had been generated. DMO issued a warning that a major (over 3-m high) tsunami had 22:22 Sapporo DMO issued a major tsunami warning for the west been generated. The area to coast of Hokkaido, a tsunami warning for the east coast of which the major warning was Hokkaido, and a tsunami alert for the north coast of issued is shown in Figure 7. Hokkaido.

22:24 JMA issued a tsunami alert to mid-coastal sections of Okushiri Island western Honshu.

Okushiri Island is located near the 22:25 Sendai DMO issued a major tsunami warning for the source of the earthquake. Eyewit- northern segment of the west coast of Honshu, and a nesses reported that first small tsunami alert warning- to the next segment to the south { lwaves arrived on the east side of &\lthe island during the earthquake 23:41 JMA issued a tsunami alert for the west coast of Kyushu. shaking. The first waves arrived at 00:12 J~I1A upgraded the warnings for the mid-coastal areas of other parts of the island approxi- Honshu from an alert to a warning . mately 2 minutes after the shaking -- EERI SPECIAL EARTHQUAKE REPORT-AUGUST 1993

they approached the area, the first deaths were reported in Taisei pie seeking refuge on day two, on incoming wave arrived, and they Okushiri Island, was 1,070. By I retreated up the hill to safety. Housing day three, the number had risen to 1 ,205. Thus, approximately 90 Hokkaido Mainland According to preliminary informa- percent of displacees required tion obtained from the Laboratory shelter. These shelters were set The town of Taisei consists of 11 of Urban Safety Planning in Tokyo, up immediately after the event and districts, of which 3 coastal approximately 1 ,495 homes in at the time of our field investi- districts suffered damage. Approx- Hiyama Semi-Prefecture were se- gation, 9 days later, approximately imately 5 minutes after the earth- verely damaged or destroyed. The two-thirds of those displaced re- Quake, police cars with loudspeak- prefecture includes the following mained in the temporary shelters ers were dispatched to announce communities: Esashi, Kaminakuni, (data from Okushiri Town Hall). tsunami warnings from the streets. Assahu, Otobe, Kuaishi, Taisei, Tsunami warning sirens located on Okushiri, Setana, Kita Hiyama, and On Okushiri, approximately 300 poles adjacent to the fire station Imaganae. In these communities, temporary homes in 5 locations were not activated because electri- an estimated 2, 728 people in around the island were being con- cal power was interrupted at that 1,123 families were displaced. structed for the 400 displaced locality by the Quake. Residents families. Most of the houses were reported being aware that a police On Okushiri Island, 402 of 680 ready for occupancy by July 25. car was issuing announcements, housing units were destroyed. Of A national law permits occupants but could not understand the that total, an estimated 291 were to live in the houses rent free for message because their windows destroyed by the tsunami and an two years; then occupants are were closed. Those respondents additional 108 were initially required to move. reported feeling a strong earth- damaged by the tsunami and sub- QUake and preparing to evacuate sequently destroyed by fire. Of In Taisei, 124 housing units were immediately by car. One person these homes, 65 were located in reported destroyed displacjng a reported being stuck in traffic and an area that was inundated by the population of 291 people. An leaving his car to run to safety. 1983 tsunami. Assuming an esti- additional number of housing units 4 mate of 3.5 persons occupied each were evacuated because of land- Parts of southwestern Hokkaido destroyed home, this destruction slide damage. On day one, imme- and northern Honshu received a would have displaced 1 ,400 peo- diately after the tsunami, 425 warning of a major tsunami and ple. Temporary shelters were people sought shelter. This num- were evacuated. The following established in local meeting houses ber dropped to 170 people from information came from interviews and schools. The number of peo- 62 families the next day. 0 with residents of Hakodate. The earthquake knocked out the power, and a fire truck with a loudspeaker came by 5 minutes after the earthquake announcing a ~ warning to evacuate. No tsunami impacted this area, and people were angry at the inconvenience, especially those who suffered earthquake damage.

Deaths

Preliminary casualty estimates, as of July 24, 1993 were 196 fatalities, 46 missing, and 34 treated for injuries. On Okushiri alone 165 were confirmed dead and 39 missing. These totals include 14 I definite fatalities and 10 still missing in the Yo Yo Hotel which Figure 8- The fl.'e was fed by fuel from propane and kerosene tanks, which was buried by a landslide. Ten was used for heating and cooking in most homes. Photo: c. Sc8wthorn

10 EERI SPECIAL EARTHQUAK : REPORT -AUGUST 1993

present. Exterior coverings are carries two 4-m lengths of hard often non-combustible stucco or suction hose equipped with bam- cement board over wood, with boo strainer baskets. Relative to The only known fire ignitions corrugated metal roofing. Large US equipment, these fire engines during the earthquake occurred in amounts of exposed wood trim, are smaller in dimensions and Aonae on the southern tip of however, compromise the fire pro- capacity. This smaller size expe- Okushiri island. Aonae has a tection. Occupancies are generally dites passage through narrower population of approximately 500. commercial closer to the wharf Japanese streets, such as those in Most of the town is oriented north- area and residential behind (at the Aonae. A third fire engine was south and sited on or almost on base of the bluff) although many present in Aonae at the time of the the beach, only a few meters buildings are mixed occupancies. earthquake; this engine was in above sea level. The rest of the poor condition, however, and was town is located on a central bluff The town is protected against fire parked at the south end of town about 20 m high where a light- by a 38-member trained volunteer where it was destroyed by the house, the town offices, the fire fire department headed by a full- tsunami. station, the school, etc., are sited. time professional. The apparatus The lower part of Aonae is densely consists of two engines of typical Fire hydrants are located around built-up with narrow streets and Japanese size and configuration - the town but are not used because typical building spacings of about each pumper has a 2,000 liter the watermains are insufficiently 3 m. The buildings are generally booster tank and carries 10 sized and pressured to provide one and two story, typically with lengths of 20-m long 65-mm adequate water for fire control. Japanese wood post and beam diameter hose. The capacity of Small fires are fought from engine construction, although some steel the pumps is approximately 2,600 booster tanks, while the main fire and concrete structures were also liters per minute. Each engine also emergency water is stored in un-

,(,

Figure 9- Tsunami and fire-damaged areas ';n Aonae, Okush;r; Island. Source: c. ScBwthorn, EQE

~~"c-::;t ~ EERI SPECIAL EARTHQUAKE REPORT. AUGUST 1993

derground cisterns sited through The initial source of ignition is also documented (Figure 8). Reported- out the town. Individual cistern unknown (at this time}; however, Iy, every time the fire departmen~ capacity is 40,000 liters, which is villagers told of earthquake shaking seemed to be gaining headway, theW accessed through a concrete man- turning over all of their furniture, so fire would flare up again, probably hole cover . numerous ignition sources were due to successive involvement of available (e.g. cooking and heating these tanks. Additional materials Shortly after the earthquake. the appliances}. At the time of ignition, fueling the spread of the fire were fire department made a circuit of wind was from the east at about considerable scrap wood in and the town looking for fires. Seeing 1.5 mlsec with gusts up to about 5 among the buildings, and numer- none and concerned about a possi- mlsec. ous vehicles which added gasoline, ble tsunami, they returned to the tires, and flammable interiors to fire station. Within a few minutes Firefighting was from hand lines the conflagration . following the earthquake. the supplied from the pumpers on top of tsunami swept through the lower the bluff, drafting from the cisterns. Fire spread was southward at area wrecking many buildings and Fire progress was south-ward about 35 meters per hour, with scattering debris over a wide area. (cross-wind) and relatively slow; firefighting on the downwind edge. At approximately 10:40 pm the suppression efforts signifi-cantly Two hours into the fire, a second fire department received a citizen impeded fire progress, but the fire ignited behind the fire line. At alarm of a fire in the lower area. A firefighters were unable to stop the about 4 am (6 hours after the brigade of 10 men immediately fire. Fire progress was aided by earthquake), available water from responded and attempted to reach flammables normally stored in each the cisterns was exhausted. Citi- the fire by driving down the main home, particularly the fact that zen volunteers assisted in moving street but found the street blocked almost all houses had 490 liter the hose over debris from the bluff by debris. They then returned to elevated kerosene tanks for heating top to the port, where the two the top of the bluff and took a and propane tanks (20 kg) for pumpers drafted from the harbor . second route down the southern cooking. The kerosene tanks were At this point, the advancing fire part of the bluff . quite likely a principal factor in the front was about 90 m wide. The , fire spread. All such tanks were fire department used equipment to ti) The fire began in a structure above found empty after the fire, most move debris and two buildings, the area directly affected by the having vented safely through the creating a firebreak. Leading four tsunami, so it likely began as a top vent pipe. The venting was hand lines from the drafting result of the earthquake. The most likely caused by radiant heat pumpers, the fire was successfully causing the kerosene to boil. Eight stopped at about 9 am, saving 1~ precise site of initial ignition is unknown, although the approximate exploded propane tanks and two several dozen houses that were in location is located on Figure 9. ruptured kerosene tanks were the path of the advancing fire. D

RETAINING WALLS Tilting of retaining walls occurred Both walls tilted at several localities. In every port outward by as much we investigated, quay walls tilted as a few degrees. from about 10 cm to more than The tilting of the 1 m (see "Liquefaction"). In all of walls generated these instances liquefaction or fissures in the fill liquefaction-induced lateral spread that were parallel to was a major factor in generating the walls. At one increased lateral forces that locality, in the caused tilting of the walls. In two Kamiiso District near instances, however, gravity re- H a k o d a t e , t h e taining walls tilted, apparently as a cracks were about consequence of increased lateral 1 m from the wall; stresses generated in moist, but at the other location unsaturated backfill. In both of near Oshamanbe, Figure 10 -Combination retaining and sea wall ""-"' these instances the walls were the cracks were 2 m near Oshamanbe that tilted up a few degerees supporting 2- to 3-m high sections to, 3 m from the due to earthquake-generated increase of lateral of recently compacted highway fill. wall. D forces on the wall. Photo: T. i. Youd l' EERI SPECIAL EARTHQUAKE REPORT -AUGUST 1993

constructed of reinforced concrete. f"\;CSTRUCTURAL DAMAGE: BUILDINGS That buildin.g had no observable \~ damage. Directly to the west and , "-,-- "- Specific buildings of note in north of the fire station is a stone In general, shaking damage to buildings on Hokkaido and even on Okushiri-cho included the Ferry stairway leading to a hillside one- Okushiri Island was slight to non- building, a two story reinforced story wooden Shinto Temple. At existent. (It should be noted that concrete frame structure with the base of the stairs is.a granite due to the limited field investiga- relatively large, approximately 40 torii, or gateway, whose .stone tion, significant effects may not be cm square columns. Located on cross-member had broken. At the reported here.) Tsunami damage the main pier, the building's first top of the approximately 25 m on Okushiri Island varied from story was inundated by the tsuna- flight of stairs is an almost iden- complete destruction for severely mi, which swept away large glass tical torii of concrete whose cross- inundated wood buildings (typical- windows and all of the interior member is undamaged. The wood- Iy, all that remained was a con- furnishings. The building sustained en temple itself was undamaged. crete pad), to moderate damage settlement of one column on the Directly across the street to the for concrete or steel buildings well south side, resulting in shear south is the two-story elementary anchored to their foundations. For cracking of beams, but was other- school, built in 1970 of reinforced the latter buildings, tsunami inun- wise structurally undamaged. The concrete frame construction. This dation resulted in partial or total second floor was relatively undis- approximately 30-m by 100-m loss of contents and cladding on the first story, stripping the building to its bare structure, leaving the building above the inundation line almost undisturbed. One exception to the general lack of shaking damage was agricultural structures. In contrast to most of ())Japan, Hokkaido and Okushiri Island contain many farm buildings that resemble US barns, including cylindrical silos strikingly similar to, although smaller than, US farm silos. Japanese silos are typically built with unreinforced concrete hollow-cell masonry unit construction. These types of structures suffered severely in this earthquake. About half of the silos observed on Okushiri Island had collapsed. Figure 11 -Small unreinforced concrete block silo at a farm on Okush,.'i that sheared and collapsed during the earthquake. Note that the adjacent barn, OKUSHIRI-CHO is the main port of also constructed of block, is practically undamaged. Okushiri Island. The residential Photo: R.P. Orense,Kiso-Jiban Consultants, Tokyo construction on the island is typical of Japanese one- and two- turbed. The City Offices are in a building was being used to house iJ story wood frame buildings, with two story stucco-over-wood frame refugees, and had sustained very ~; metal roofing rather than heavier building, probably built in the minor cracking in columns and clay tile. Commercial buildings are 1950's and located approximately spandrels. Directly to the west of IU also one- and two-story wood 1 km from the shore in a narrow the school is the gymnasium, an frame structures. Shaking damage river valley. The building appears approximately 25 m by 40 m steel to Okushiri-cho residential and to be founded on firm soils. There framed structure with a barrel arch commercial construction was mini- was no observable damage to the lattice-truss roof. The gym was f );\mal, with no building collapses and building, other than some lost being used as housing for police ~only a few chimney collapses. stucco on the west side which officers brought in from other Chimneys were generally con- revealed badly decayed wood. Di- districts, and had no observable structed of lightly reinforced rectl,Y to the west is the two-story damage. Some spalling of stucco concrete. fire station, of similar vintage, but cover over concrete was observed '3 EERI SPECIAL EARTHQUAKE REPORT -AUGUST 1993

on a one-story structure that the tower to rotate en masse onto buckled near the base, while the connected the school and gym; the the one-story building. The tower other appeared undamaged. The "\:, spalling was at the junction of the came to rest at approximately a amount of cement in the tanks at ;) structure with the school. To the 45-degree angle. Our inspection the time of the earthquake was south of the school is the NTT indicated an apparent bond failure unknown. The hopper structure Telephone Exchange building (dis- of plain J anchorage bars (approxi- consisted of a first-story steel cussed in "Lifelines"). That struc- mately 24 mm in diameter) .Also moment frame structure (which ture was also undamaged. Lastly, on the bluff is the fire station, a permitted trucks to enter between directly opposite the ferry landing two-story reinforced concrete columns), with an overhead braced had been the two-story Yo Yo building, that had no apparent steel framework extending to a Hotel, approximately 25 m by 50 damage. Two schools are located height of about 20-m to house m in plan, which was buried under north of the town center, both of loading equipment and storage a massive landslide. similar construction to the Oku- facilities. The framework con- shiri-cho school described above. sisted of large cold-formed steel ACNAE was the most severely One of these buildings was two angular column sections, braced by impacted town on Okushiri island. stories high and was undamaged, hot-rolled steel angle diagonal Located at the south tip of Oku- while the second building was bracing. The columns buckled at shiri, this fishing community had a three stories high, and sustained the junction of the braced struc- population of approximately 1.500, severe shear cracking in spandrels, ture with the first-story moment frame, causing the braced structure to tilt about 5 degrees. A sloping 60-m long conveyor structure connected the top of the hopper to processing facilities on the ground. The conveyor has a cylindrical steel pipe-Iike wind shield, approximately 1 m in dia- , meter. The conveyor is vertically f;) supported at mid-span by a steel

A-frame truss. Due to the large I displacement of the canted hopper, the cylindrical steel pipe-Iike wind shield buckled at its junction with the A-frame. Repair costs for the cement plant complex were estimated at about 20% of replace- ment value.

OSHAMANBE is located on Hok- Figure 12 -Lighthouse at Aonae on Okushiri 'sland that tipped during the kaido, approximately 80 km from earthQuake. The tipping was caused by fracture of the tower structure at the epicentral area. As discussed the base along with pull-out of reinforcing steel from the foundation. Our elsewhere, extensive liquefaction inspection indicated an apparent bond failure of plain J anchorage bars was observed in this vicinity. Of (approximately 24 mm in diameter). particular interest was the occur- Photo: R.P. Orense,Kiso-Jiban Consultants, Tokyo rence of liquefaction at the Oshamanbe Nakonosawa Elemen- with most residential and commer- with the cracking extending into tary School, a modern complex cial buildings located directly bending and shear cracks in the consisting of a one-story class- around the port area, only a few columns, and continuing on as room building of concrete and glu- meters above sea level. However , shear cracking in the next lower lam construction, and a steel local government offices and sqme set of spandrels. The Okushiri framed gymnasium. These build- housing is located on a bluff above Cement Plant, in the same general ings were supported on concrete the port. A lighthouse on the vicinity, consists of two steel silos, columns and grade beams and, I )'1 bluff , a one-story reinforced con- a small office building, and a large despite extensive liquefaction and

crete building with a reinforced hopper for loading trucks with an lateral spreading, sustained only r

concrete tower approximately 8 m a~sociated conveyor structure. minor shear cracking in lower wall high, failed at the base, allowing One of the two steel silos had panels. O

~ EERI SPECIAL EARTHQUAKE REPORT -AUGUST 1993

BRIDGES

Few bridges exist on Okushiri and 1.8 m diameter reinforced concrete in Hokkaido bridge performance bridge piers (Figure 13). This was generally very good. No cracking caused spalling of con- bridges were closed due to struc- crete in the northern-most pier , tural damage, nor did any bridges leading to weight restrictions on suffer major distress. However , traffic crossing the bridge, but the the approaches to many bridges bridge was not closed. settled due to compaction of ap- proach fill, causing vertical separa- Near Oshamanbe, a bridge tilted tions in the pavement at the junc- sideways by about 45 cm during tion with the deck, that required the earthquake. This bridge was filling with asphalt before traffic constructed in about 1 960 and could cross the structure, at least then widened later by adding a at highway speeds. lane to the west side of the bridge, but without widening the founda- The most heavily damaged bridge tion. Thus, the bridge was subject observed was a six-span, two-lane, to eccentric loading. During the 156-m long, 1970 vintage steel earthquake, the bridge tilted west- plate girder highway bridge over ward, in the direction of the eccen- Figure 13 -Damaged piers beneath the Assabu river, at Assabu-cho, tricity. There was no apparent a bridge crossIng the Assabu River about 7 km northeast of Esashi. differential settlement of the piers. about 7 km northeast of Eashi. The Lateral spreading was observed in Liquefaction occurred beneath the X-pattern of cracks in the piers indi- the south river bank near the bridge, and probably around the cate that cyclic flexural stresses ()'t ridge, but did not appear to affect piers reducing lateral resistance to caused the damage i';cluding the D, ... spalled concrete and buckled 'y"(he superstructure. Significant horizontal loading induced by both cracking was observed just above the eccentricity and inertial forces reinforcement in the nearest pier. Photo: T.L. Youd the waterline in the single-column generated by the earthquake. D

bers of distribution poles in the WATER AND WASTEWATER -The LIFELINES tsunami inundation areas, particu- water supply to Aonae is furnished Due to our limited field investiga- larly on Okushiri, were swept by an underground pipe, approxi- tions of lifelines, some significant away. However, not all poles mately 20 cm diameter, lain along effects may not have been identi- were destroyed, and some over- the east shore of the island. (The fied and reported here . head lines were observed with source, somewhere to the north, vegetative debris hanging from the was not determined.) About 1.5 . HIGHWAYS- In general, highway lines, clearly deposited by the km north of Aonae the line, a performance was very good, with tsunami. From this effect, we welded steel pipe at that point, damage largely confined to wide- infer that poles were probably crosses a river channel beneath a ~i spread settlement at approaches to destroyed by large debris (e.g, highway bridge. The bridge, which bridge abutments and settlement houses) from the tsunami rather has about a 6 m clearance above of pavements in soft soil areas. than directly by the waves. Dam- the river was undamaged, where- Repairs were on-going ten days aged poles and lines were general- as, the pipeline, with about a 3.5 after the earthquake, with a large ly replaced within ten days of the m clearance above the channel, number of damaged pavement sec- earthquake. On Hokkaido, several was broken, presumably' by tsu- tions already repaired. poles tilted due to liquefaction of nami waves. No underground pipe supporting soils, causing local breaks were reported in Aonae. !\Q4LECTRIC -Electric power does outages until the poles could be not appear to have been signifi- righted. Reportedly, local outages About one mile north of Aonae is a cantly impacted in this event, with occIJrred due to transformers being wastewater treatment plant, set the exception that significant num- shaken loose from poles. back from the shore about 300 m

15 EERI SPECIAL EARTHQUAKE REPORT -AUGUST 1993

and consisting of two adjoining TELECOMMUNICATIONS -Two made to a buried welded steel fuel small concrete high-bay, one-story RCS's (Remote Control Switchesl line with a diameter of about 20 - buildings, two approximately 10 m exist on Okushiri --one in the town cm in the southern part of the port ~J' diameter 5 m high concrete diges- of Okushiri-cho, and other in of Okushiri-cho. The line was in ter tanks, an elevated methane gas Aonae. The Okushiri-cho RCS is the tsunami run-up zone as well as holder, and an outfall to the ocean . located in a reinforced concrete in an area of apparent extensive Moderate settlement of about 30 two-story building in the western ground settlement; either could cm occurred around the buildings, part of the community, approxi- have caused the damage. which also sustained some moder- mately 1 km from the shore up a ate damage due to pounding be- river valley, on apparently firm PORTS AND HARBORS -The ports tween the buildings, and also soils. Atop the building is an of Esashi, Aonae, and Okushiri-cho between the buildings and the approximately 40 m-high free- were impacted by the tsunami. digester tanks. The piping standing steel truss microwave Ferry service from Esashi port was (including significant amounts of tower. No damage or problems prevented for several days, due to ABS piping) and mechanical and were reported for the tower , automobiles and debris in the port. i electrical equipment all performed microwave, telecom, electrical, Aonae port had substantial debris well, with only minor to negligible HV AC, or other equipment in the in the port area, as well as general damage. Tsunami inundation in building, nor for the building itself . destruction of shoreside shops and the area of the buildings was a Telecom equipment in the building facilities. Breakwaters at Esashi few tenths of a meter which didn't was well-braced in both directions. and Aonae appeared undamaged. cause any significant damage. Backup diesel generators func- Okushiri-cho port had debris and However, a lower pump room be- tioned well following the quake. other problems. In addition, the neath the larger of the two build- northern concrete breakwater was ings was flooded to a depth of Telephone distribution on Okushiri damaged; several breakwater sec- about 1.6 m apparently due to was severely impacted by the tsu- tions were overturned and dis- tsunami flooding which entered the nami. Lines are carried on the placed up to tens of meters. The building through the outfall. same poles as the electrical distri- sections were approximately 3 m - Seawater entered the lower pump bution, and many were destroyed square, 6 m long concrete blocks. ~) room via the outfall, resulting in by the tsunami. Considerable time Most of the displacement was contamination of pumps, motors, was required to splice in new lines. apparently caused by tsunami electrical equipment, etc., such wave action, although geotech- that the plant was still shut down UNDERGROUND FUEL LINES -On nical effects may have played a ten days following the earthquake. July 23, we noted repairs being role. O

1983 Alota-Oki earthquake (Mw -7.8) and 249-260. its implications for systematics of subduction earthquakes, Earthquake Satake, K. (1986). Re-exemination of the 1940 Shakotan-oki earthquake and the fault Cohee, B.P., P.G. Somerville and N.A. Prediction Research 3, 305-317. parameters of the earthquakes along the Abrahamson (1991 ). Simulated ground motions for hypothesized Mw = 8 Kikuchi, M., (1992). The magnitude 7.8 eastern margin of the Japan Sea, Phys. subductio.n earthquakes in Washington and Hokkaido Nansei Oki earthquake of July 1 2, Earth. Planet. Interiors 43, 137.147. Oregon, Bull. Seism. Soc. Am. 81, 28-56; provisional analysis of broadband Seno, T ., C.R. DeMets, S. Stein, and D.F. printer's typos corrected p. 2529. teleseismic seismograms, Yokohama City University, Seismology Report No.23 Woods {1987). The movements of the Okhotsk plate {abstract in Japanese), Crouse, C.B. (1991). Ground motion (Appendix). attenuation equations for earthquakes on Seism. Soc. Japan Program and Abstracts 1987 (1), p. 87. the Cascadia subduction lOne, Earthquake Kobayashi, y (1983). On the initiation of Spectra 7.201-236. subduction plates. The earth monthly. 5,510-514 (in Japanese). Tanioka, Y., L. Ruff and K. Satake, (1993), The puzzling rupture process of the July 12, DeMets, C. A test of present-dey plate geometrics for northeast Asia and Japan, J. Nakamura, K. (1983). Possible nascent 1993 southwest Hokkaido earthquake, Geophysical Research, in press. trench along the eastern Japan Sea as the unpublished manuscript. convergent boundary between the Eurasian Youngs, R.R., S.M. Day and J.L. Stevens Hata, Mitsuo; Segawa, Hidoyoshi; and and North American plates, Bull. Earthquake (1988). Near fie~d ground motions on rock .jf Yajima, Junkieki, (1982), Geology of South- Research Institute, 58, 711-722. for large subduction earthquakes, ASCE ~ West Hokkaido; Okushiri Island; Geological Geotechnical Publication No.20, 445-462 Map (series and affiliation in Japanese), Satake, K. (1985). The mechanism of the (rock); and Washington Public Power Sup- Scale 1 ;50,000. 1983 Japan Sea earthquake as inferred from long-period surface waves and ply System (1988). Response to NBC Questions 230.1 and 230.2 (soil). Kanamori, H. and L. Astiz (1985). The tsunamis, Phys. Earth. Planet. Interiors 37 , EERI SPECIAL EARTHQUAKE REPORT -AUGUST 1993

Where liquefiable deposits had accumulated in the southern part of Hokkaido, surface effects of liquefaction were generally evident. These effects included sand boils, fissures, lateral spreads, ground oscillation, loss of bearing strength, ground settlement, and buoyant rise of manholes and tanks. The areas affected included several alluvial river valleys on the west coast of southern Hokkaido, alluvial plains on the east coast around Uchiura Bay, and infills at several ports. These effects caused varying amounts of damage to constructed works, but in Figure 15 -Anchored bulkheads tilted seaward at all of the ports we visited several notable instances, well- on the southern peninsula of Hokkaido. This bulkhead at the Port of Hako- built structures survived the date was one of the more severely damaged, with horizontal displacements occurrence of liquefaction with no at the top of the wall in excess of 1 m. A major fissure of similar width significant damage. paralleled the wall about 10 m inland (along the line of anchor points?). Widespread ground settlements and eruptions of sand boils indicate that ! At each of the ports we visited, compaction and liquefaction of fill materials were a principal cause of ~uay walls had tilted seaward from damage. Photo: T.L. Youd lf10 cm to 20 cm at ground surface with one notable instance of about disrupted pavements and storage Where highway grades crossed 1 m of displacement. Ground fis- areas behind these walls (Figure liquefiable areas, the road surface sures, settlements and sand boils 15). commonly differentially subsided generating wavy pavement with wave-lengths usually several meters in extent (Figure 16). This type of disturbance was noted in several areas on both sides of southwestern Hokkaido; at each area we specifically examined, sand boils, ground fissures and other evidences of liquefaction had developed near the roadway.

Several bridges crossed rivers and flood plains where liquefaction effects were evident. In a few instances, abutments had rotated inward as a consequence of in- rl creased lateral pressures due to liquefaction and lateral spread. In these instances abutments jammed the bridge girders causing minor ,Figure 16 ~ View of Highway 5 south of Oshamanbe where vertical waves, damage to the seating and/or to (pJWith~ at several troughs other as deeplocalities. as 0.6 m,Nearby disrupted fissures traffic. and Similarsand boilswaves at developedeach site the bridge girders. In several instances, however, the bridges examined indicate that liquefaction contfl"buted to the highway disruption. were undamaged indioating that Photo:R. P.'Orense,Kiso-Jiban Consultants, Tokyo the foundations and superstruc- EERI SPECIAL EARTHQUAKE REPORT -AUGUST' 993

~ foundation support. oil storage tank at an industrial I Oil storage tanks at facility popped mostly out of th- this and other ports ground (Figure 18). That tank was" were undamaged 7 .2 m long and 1.8 m diameter. even though ground The tank had been about 1/3 full of settlements and oil at the time of the earthquake. sand boils occurred in the near vicinity Flood control dikes along rivers on of the tanks. One the west side of southwestern Hok- tank we specifically kaido commonly cracked and settled examined at the Fer- severely in areas affected by lique- ry Port at Hakodate faction. This type of dike damage had oscillated hori- has been a common occurrence zontally relative to during several recent Japanese Figure 17 -The ground fissures and sand boils the ground with an earthquakes. The dikes were up to shown in the foreground continue beneath the bridge amplitude of about several meters high and paralleled where even larger fissures and sand boils developed. 10 cm to 20 cm . both sides of the rivers in lowland Free-field lateral spread displacemen~s in ~he n~ar The ground around valleys that cut through the gener- vicinity were as large as 1 m; the brIdge, Includ,ng the tank had settled ally mountainous terrain. Cultivated the piers and abutments, however, were undamaged. and sand boil depo- fields lay beyond the dikes in the The bridge crosses the Toshibetsu River about 5 km sits covered the protected areas. The dike settle- south of Setana. ground surface. ments commonly involved longitu- Photo: R.P. Orense,Kiso-Jiban Consultants, Tokyo The tank, which dinal splitting of the embankment at tures had sufficient strength to was apparently founded on piles, the crest and cracking and bulging resist ground displacements and was level and undamaged . of the ground surface along the other deleterious effects of flanks of the dikes. The mode of liquefaction (Figure 17). ' --.,-"--- Locally, several water and sewer failure appears to have been loss of lines were fractured or severed due bearing strength in the liquefied SOia Many houses and other small- to to differential ground displacements beneath the embankments with medium-sized buildings were located generated by liquefaction. How- penetration of the dike into the in areas of surface disturbances due ever, most liquefaction occurred in softened soil and lateral and upward to liquefaction. The foundation non-populated areas where under- thrusting of the foundation soils on walls or slabs of several buildings ground pipelines had not been con- either side of the dike. Sediment were pulled apart in extension or structed. The most spectacular compaction may also have added to vertically offset. In nearly all effects of liquefaction to appur- the subsidence of the dikes. D instances where this type of dam- tenant pipeline facilities was the rise of ,'Y- -- age occurred, the foundations were manholes relative to , unreinforced or poorly reinforced. Conversely, several buildings on the ground surface, well-constructed and well-reinforced either as a conse- foundations survived the ground quence of buoyant shaking and occurrence of liquefac- rise of the structure tion with no significant damage. or settlement of the Several of these structures were ground. These founded on piles with the ground effects were most around the structures subsiding and prominent in the city fracturing, but the structure of Oshamanbe. remaining in place and undamaged. In other instances, buildings on Several buried gaso- shallow, but well-reinforced and line or oil storage well-tied-together footings also tanks also rose Figure 18 -Ot1 storage tank that floated out of the remained undamaged. buoyantly by a few ground at an industrial fact1ity about 6 km north of centimeters at gaso- Esashi. Liquefaction of backft11 around the tan/4;:a A reinforced-concrete silo for line stations, crack- apparently allowed the tank and cover slab toW' storage of cement at the Port of ing overlying pave- buoyantly rise and tip. The tank was about 1/3 full Hakodate settled and tilted about m.ents and pipeline at the time of the earthquake. Photo: T.L. Youd 2.5° as a consequence of loss of connections. One

18 EERI SPECIAL EARTHQUAKE REPORT -AUGUST 1993

to scramble to safety after being briefly trapped within the Landslides were triggered through- remains of the hotel. out Okushiri Island as well as throughout much of the southwest- The hotel was lo- ern Hokkaido. The slides consisted cated just seaward mainly of rock falls and rock slides of a main road that on steep ( > 50. ) natural slopes and was aligned along engineered cut slopes. Volumes the base of the ranged from several cubic meters to slope. The pre- the 800,000 mJ rock slide that earthquake slope buried the Yo Yo Hotel at Okushiri inclination was Port (Figure 19). Numerous slumps about 60 0, parallel Figure 19 -Rock slide near the ferry terminal in occurred in engineered fill along to the slide surface Okushiri City that buried the Yo Yo Hotel. highways and railways. that was subse- quently exposed by the failure. The along the coast of the island where Rock falls and slides on height of the rock slide from toe to extremely steep slopes of volcanic Okushiri Island crown is approximately 100 m, its breccia, basalt, and tuffaceous sandstone shed debris that The largest landslide triggered was shaken apart along highly by the earthquake occurred weathered fractures. The about 70 m from the ferry rock debris blocked coastal terminal in Okushiri City. This highways in many places. At 800,000 mJ slide consisted of many of these localities, the Pliocene tuff, tuffaceous falling rock destroyed 11andstone, and conglomerate segments of concrete barriers ~ Hata et al., 1982). The slide that had been constructed for buried the yo Yo Hotel and rockfall protection. At several several vehicles. Two large localities, wire nets that had kerosene storage tanks at the been draped from the cliffs northern edge of the slide had been torn and penetrated were also damaged. The fatal- by large rock fragments. ities are estimated at 35-40 Figure 20 -Rock fall along seacoast north of These nets had been placed people, but these numbers are Setana leaves pinnacle virtually cleaved in half. to control the fall of rock uncertain because the hotel debris. records, as well as the hotel staff, width is about 800 m, and its were trapped in the debris. A few thickness ranges from 20 m to 40 Landslides on Southwestern guests on the upper floors were able m. The slide's morphology suggests that its motion was Hokkaido largely translational and parallel to its The occurrence of rock falls and 60° scarp. The rock slides triggered by the earth- slide mass moved quake on southern Hokkaido Penin- about 40 m with the sula was similar to that on Okushiri 11topmost part staying Island. These types of failures ! largely intact with appeared to be most numerous on the distal portion the western coast of the peninsula dilating and spread- where the steepest slopes exist. As ing as the material on Okushiri Island, most of the rock underwent crushing falls were generated from andesitic from the weight of volcanic breccias which stand as ! the overlying debris. coastal cliffs and erosional pinnacles Figure 21- Talus shed collapsed by rock fall along with slopes of 50 o to vertical. highway 229 north of Setana. Numerous rock falls Several of these pinnacles along the All photos this page by E. Harp, USGS and slides developed ocean shore north of Setana were

~ EERI SPECIAL EARTHQUAKE REPORT -AUGUST 1993

virtually cleaved in half by rock slides along widely space fracture r surfaces within the rock (Figure 20).

About 1 5 km north of Setana, on highway 229, one of the sections of a talus shed collapsed from the impact of a large rock fall (Figure 21 ). The shed was constructed as a cast-in-place concrete tube. The rock fall contained large pieces of andesitic breccia, some greater than 4 m diameter. The total volume was several hundred cubic meters and originated from a near vertical cliff about 60 m above the highway.

About 10 km northwest of Osha- Figure 22 -Slump north of Oshamanbe on highway 5 within engineered fill manbe, on highway 5, a principal in highway and ral1road embankments resulted in overturning of two trucks arterial highway, a slump de- and one car; only one truck remained when this photo was taken. veloped in the engineered highway Photo; R.P. Orense, Kiso-Jiban Consultants, Tokyo fill which overlies marsh deposits. An adjacent railroad fill was also thousand cubic meters of soil may underlying marsh deposits. The disturbed by slumping at this have been induced by liquefaction failure caused the overturning of ~ locality. This failure of several or shaking-induced softening of the two trucks and a car (Figure 22). LJ ~

EARTHQUAKE ENGINEERING FIRST CLASS RESEARCH INSTITUTE Postage Paid at 499 14th Street, Suite 320 Berkeley I CA Oakland, CA 94612-1902 Permit No.204 EARTHQUAKE ENGINEERING RETURN POSTAGE GUARANTEED RESEARCH INSTITUTE

PRESIDENT Lloyd S. Cluff

PAST PRESIDENT

J. Carl Stepp

VICE-PRESIDENT William T. Holmes

SECRETARY-TREASURER

Chris D. Poland

BOARD OF DIRECTORS Thomas L. Anderson Christopher Arnold A. Gerald Brady Lloyd S. Cluff C. B. Crouse Phillip L. Gould William T. Holmes Chris D. Poland ( J. Carl Stepp

EXECUTIVE DIRECTOR Susan K. Tubbesing