Scholarly article on topic 'Reconnaissance report on damage in and around river levees caused by the 2011 off the Pacific coast of Tohoku earthquake'

Reconnaissance report on damage in and around river levees caused by the 2011 off the Pacific coast of Tohoku earthquake Academic research paper on "Earth and related environmental sciences"

Share paper
Academic journal
Soils and Foundations
{"Earthquake damage" / Liquefaction / Levee / Damage / Mitigation / "2011 Off the Pacific coast of Tohoku earthquake" / "(IGC: E08)"}

Abstract of research paper on Earth and related environmental sciences, author of scientific article — Yasushi Sasaki, Ikuo Towhata, Kenya Miyamoto, Masami Shirato, Akiyoshi Narita, et al.

Abstract The gigantic earthquake on March 11, 2011, caused significant damage in and around river levees over a vast area. Because the amount of damage was on such a huge scale, emergency restoration was difficult and many lessons were learnt. The cause of the damage in most cases was liquefaction either in the foundation or in the levee body, and the latter was recognized as a new technical problem. Many levees in the coastal area experienced the combined effects of the tsunami with the co-seismic subsidence of the earth's crust. Among the many examples of damage, one positive issue was that damage mitigation measures such as drainage and soil improvement were found to be effective. The present text addresses the findings and lessons learnt from the authors' emergency activities after the quake.

Academic research paper on topic "Reconnaissance report on damage in and around river levees caused by the 2011 off the Pacific coast of Tohoku earthquake"



The Japanese Geotechnical Society

Soils and Foundations journal homepage:

Reconnaissance report on damage in and around river levees caused by the 2011 off the Pacific coast of Tohoku earthquake

Yasushi Sasakia, Ikuo Towhatab n, Kenya Miyamotoc, Masami Shiratoc, Akiyoshi Naritac,

Tetsuya Sasakid, Shunsuke Sakoa

aJapan Institute of Construction Engineering, Japan University of Tokyo, Japan cMinistry of Land, Infrastructure, Transport and Tourism, Japan dPublic Works Research Institute, Japan

Received 29 January 2012; received in revised form 23 June 2012; accepted 1 September 2012 Available online 11 December 2012


The gigantic earthquake on March 11, 2011, caused significant damage in and around river levees over a vast area. Because the amount of damage was on such a huge scale, emergency restoration was difficult and many lessons were learnt. The cause of the damage in most cases was liquefaction either in the foundation or in the levee body, and the latter was recognized as a new technical problem. Many levees in the coastal area experienced the combined effects of the tsunami with the co-seismic subsidence of the earth's crust. Among the many examples of damage, one positive issue was that damage mitigation measures such as drainage and soil improvement were found to be effective. The present text addresses the findings and lessons learnt from the authors' emergency activities after the quake.

© 2012 The Japanese Geotechnical Society. Production and hosting by Elsevier B.V. All rights reserved.

Keywords: Earthquake damage; Liquefaction; Levee; Damage; Mitigation; 2011 Off the Pacific coast of Tohoku earthquake; (IGC: E08)

1. Introduction

The damage that was produced by the 2011 off the Pacific coast of Tohoku earthquake on March 11, 2011, is classified into two groups, namely that caused by the tsunami and that caused by ground shaking. Most of the damage belonging to the latter was caused by soil liquefaction in artificial islands and along river channels. These liquefaction-prone situations are both characterized by

something inexperienced in the past. While those concerning artificial islands are described elsewhere (e.g., Towhata et al., 2011), the issues relevant to river levees are going to be addressed in this paper.

In previous times, the earthquake resistant design of river levees was not mandatory unless there was a high risk of flooding in the area behind levees (Ministry of Construction, 1985). In other words, earthquake resistant measures were considered unnecessary if the probability of simultaneous occurrence of strong earthquake and flooding is low. Hence, an effort was made to repair the seismic damage done to the levees within two weeks so that those levees would regain their original height before heavy rain and high water events which may have come at a later time.

Since the 1990s, this situation has changed significantly. Levees of the Tokachi and Kushiro rivers, after being

nCorresponding author. E-mail address: (I. Towhata). Peer review under responsibility of The Japanese Geotechnical Society.

0038-0806 © 2012 The Japanese Geotechnical Society. Production and hosting by Elsevier B.V. All rights reserved.

damaged by the 1993 Kushiro-oki earthquake, were not able to be restored in such a short time as intended. Moreover, at the time of the 1995 Kobe earthquake, the significant subsidence of the Yodo-river levee in Osaka demonstrated that high water level can take place every day when the river is affected by sea tides (Matsuo, 1996). Accordingly, it has been agreed that levees should have sufficient resistance against earthquake action in case that the protected area behind the levees is likely to be inundated when the levee is seismically damaged. Hence, in the 21st Century, the seismic performance of levees has been inspected (Ministry of Land, Infrastructure, Transport and Tourism (2007); Japanese Geotechnical Society (2009)).

It should be noted that the concept of seismic performance has been introduced into the inspections of levees by stating that the risk of overtopping, breaching, and flooding has to be sufficiently low even if a levee is shaken by a strong earthquake. This implies that minor seismic distortion is still allowed even if the seismic factor of safety is less than unity. Because the seismic performance of geotechnical structures are based on residual deformation, efforts have been made to develop practical methods to predict the seismically-induced deformation of levees as well as to reduce deformation at a reasonable cost. This is considered reasonable because of the tremendous lengths of the entire levees and since financial limitations do not allow the kind of expensive seismic retrofitting that would keep the seismic factor of safety greater than unity.

Assessing the seismically induced deformation of levees requires significant knowledge about the stress-strain behavior of the constituent soils, the cross sections, and the seismic loads. However, such detailed knowledge is not available in most cases. Scarce information is recorded about the damage done by past earthquakes because the

first priority is quick restoration, as stated above. Therefore, today, it is important to assemble data from levee damage and interpret it for future use. Despite this intention, the

Fig. 2. Liquefaction in the Tone river channel (Sawara city at Right Bank 39.0 km). (a) Sand boils; (b) Distortion in levee.

Fig. 1. Distribution of seismically affected levees. (For interpretation of the references to color in this figure, the reader is referred to the web version of this article.)

page limitations imposed on this paper restricts the authors to describing only some of the damage that happened. For more damage information, refer to National Institute for Land and Infrastructure Management (2011) for examples.

The authors have been engaged in emergency activities on damaged river levees as a part of a governmental program. The data cited in this paper is therefore the product of those activities that apparently deserve to be published at this point in time.

River channel Liquefaction in 2011

Downtown Sawara without Significant damage in 2011 (Topography in late 19 th Century)

Fig. 3. Old river channels of the Sawara site (Zinsoku-Sokuzu 1/20000 made available to public by National Institute for Agro-Environmental Sciences).

2. Observed damage in river levees

This chapter reports the damage observed immediately after the earthquake. Fig. 1 demonstrates the location of affected rivers and levees in the Northeastern (Tohoku) and Kanto (near Tokyo) regions, respectively. Therein, the damaged levees are marked in red and distributed from the downstream area to upstream. Note, however, that this does not mean the entire levees in the red parts were affected; the extent of the damage extent varied considerably, depending upon the local geology, the configuration of levees, the intensity of shaking, and other factors.

Like the damage in other types of structures, those in river levees are characterized by its vast spatial distribution and simultaneous occurrence. The total number of levee damage events, whether significant or minor, was 2115. Moreover, different types of damage impacted on other types of damage, as exemplified by the shortage of the fuel supply and construction equipments hindering rescue and restoration activities elsewhere. Aftershocks were another reason for delay. Accordingly, damaged river levees were not restored within two weeks after the quake, as originally intended; only the most severely affected parts of levees were taken care of (recovery of the original height by temporary earth filling) prior to the rainy season.

As is often stated, the seismic magnitude of MJMA = 9.0 (at 14:46 on March 11th Japanese time) induced a very long duration of shaking. It seems that liquefaction-vulnerable structures continued to be shaken after the onset of subsoil liquefaction for as long as two minutes or longer, though this varied from place to place. Together with the aftershock (MJMA = 7.3 at 15:15 on the same day) that occurred 30 minutes after the main shock, the extreme

Fig. 4. Cross section and subsoil condition of the right-bank levee at the Sawara site on the Tone river (In the Japanese tradition of river engineering, cross sections are drawn as seen from the upstream side).

long duration of the shaking most probably made the deformation and subsidence of levees more serious.

Another issue is the limited time between the earthquake occurrence (March 11th) and the rainy season that usually starts in early June. During this period, the damaged levees were recovered to their original height by urgently filling with soil. However, the mechanical strength and permeability of the levees affected were not properly controlled. Hence, it was decided to achieve the required safety during high water levels by more careful patrol and by watching the levees. It was fortunate that no significant damage occurred during the rainy season in 2011.

2.1. Levee along the Tone river

The discussion starts with the cases in the Kanto region; see Fig. 1. The Tone river is one of the biggest rivers in Japan. It forms many swamps and lakes along its route. Works in modern times have made the river straighter by constructing levees on former river channels, which are now vulnerable to subsoil liquefaction.

Fig. 2 indicates significant liquefaction and the induced distortion which occurred in Sawara city (R 39.0 km). This levee is situated in a compound of flood plain and abandoned river channels (Fig. 3). The significant subsoil liquefaction led to a 1m subsidence of the 3.8-meter-high levee and cracks in the slope on the river side. This area on both sides of the levee is a former river channel and is of typical liquefaction-prone condition. The subsurface data in Fig. 4 indicates that the alluvial sandy layer (As1 and Bs) that is vulnerable to liquefaction is 1 to 4 m in thickness. For symbols of types of soil layers, refer to Table 1.

2.2. Levee along the Naka river

The Naka river is located in the north part of Kanto region. There are liquefaction-prone soil conditions in the downstream area. A typical site of liquefaction is at Shimo-Ohno-Jisaki at R 3.5 km (Fig. 5). The original height of the levee was 1.9 m and the seismic subsidence was 0.6 m. Longitudinal cracks 1 m in depth occurred at the crest because of lateral spreading. This area is located in a flood plain and an abandoned river channel (Fig. 6). Much evidence of liquefaction was observed on both the river and the land sides of the levee (Fig. 7).

Table 1

Names of soil layers.

Bs Sandy embankment and fill

Bc Fine-soil embankment and fill

Bsc Sandy fine soil in embankment

Bsnc N: newer part of Bsc (levee)

As Alluvial sand

Ac Alluvial clay

Ag Alluvial gravel

Acs Alluvial sand with fines

Fig. 5. Distortion of Naka river Levee at Shimo-Ohno-Jisaki (R 3.5 km).

Fig. 6. Surface geology of the Shimo-Ohno-Jisaki site (written on landform classification map for flood control made by Geospatial Information Authority of Japan).

2.3. Levee of the Hinuma lake

Hinuma is a small lake connected to the Naka river near its mouth. The Shimo-Ishizaki site on the left bank of Hinuma (L 7.5 km) is underlain by a thick clayey layer. In spite of this, the crest of the levee subsided 1 m from its original height of 2.6 m (Fig. 8). The subsoil is composed of thick clay and there is no sandy layer down to more than 20 m below the surface (Fig. 9). The significant distortion, in spite of the lack of liquefiable sandy subsoil, has led to the interpretation that the soil developed liquefaction inside the levee.

Sasaki et al. (1994) together with Kaneko et al. (1996) and Finn et al. (1997) initiated discussion on the mechanism of substantial damage in levees resting on unliquefiable subsoil. Fig. 10 was schematically illustrated to demonstrate the idea of liquefaction inside the levee, taking examples of the levees

Fig. 7. Sand boils around the Shimo-Ohno-Jisaki site along the Naka river (Fig. 5). (a) On the river side; (b) On the land side.

Fig. 8. Significant distortion of the Hinuma levee at Shimo-Ishizaki.

of Tokachi and Kushiro rivers that distorted significantly during the 1993 Kushiro-Oki earthquake. Initially, the surface of peaty deposit was level and a river levee was constructed upon it. As long as this initial geometry lasts, the levee will be situated above the ground water level, and its water content is not very high. Hence, the risk of liquefaction is not high.

In reality, however, the weight of the levee induces consolidation and settlement. The base of the levee comes into the subsoil beneath the ground water level which is originally located at a shallow depth (about 50 cm below the ground surface in the peaty layer), and the ground water flows into this part of the levee. The ground water level in the levee rises further because of the infiltration of rain water. The soil in this part of the levee may get looser during this subsidence and lateral extension. Moreover, it has been considered that, during seismic shaking, the inertial force of the embankment transmitted to the underlying peaty layer might be reduced by liquefaction in the base soil of the levee, leading to less seismic disturbance in the peaty foundation soil.

2.4. Levee along the Naruse river

The discussion proceeds to the Tohoku region. The Naruse river is located in a flood plain to the north of Sendai city. At its Shimo-Nakanome site at L30.0 km, the crest subsided 5.5 m out of its original height of 7.5 m associated with multiple deep cracks (Fig. 11a). Lateral expansion induced longitudinal development of cracks in which an eyewitness found sand boils (Fig. 11b) that suggested liquefaction inside the levee. Moreover, the lateral spreading was overwhelmingly oriented towards the land side, while the river side had no distortion at all. The subsoil condition is illustrated in Fig. 12, which shows that the soil in the foundation is profoundly clayey and unlikely to liquefy. The ground water surface is situated inside the levee body and higher than the original ground surface. Moreover, the levee body has sunk into the clayey subsoil. Thus, this site is possibly another example of aforementioned liquefaction inside the levee.

2.5. Levee along the Abukuma river

The Abukuma river has its mouth to the south of Sendai city. This area is located in a compound of a flood plain and natural levees. At the Edano site (R 30.6 km), the levee distortion as illustrated in Fig. 13 occurred. This figure shows that the levee deformation towards the land side was more profound than towards the river, although tension cracks in the longitudinal direction occurred on both sides.

Fig. 10. Schematic illustration of liquefaction in the body of the levee.

Moreover, a subsidence of 2.9 m occurred out of the original height of 5.7 m (Fig. 14). The depth of the cracks reached 1 m as well. An eyewitness reported sand boils in the crack at the crest as well as at the foot of the slope on the land side, suggesting liquefaction inside the levee. The cross section as shown in Fig. 15 suggests that the subsoil is made of more than 15 m of clayey soil. Hence, the induced damage is likely to have been induced by liquefaction inside the levee.

2.6. Levees along the Eai river

The Eai river flows in the Ohsaki plain that is to the north of Sendai city, and merges with the Kitakami river near its mouth. At the Fukunuma site (R 26.6 km) on an abandoned river channel, a significant subsidence of 2.3 m occurred out of the 3.8-m height together with very deep cracks of 1.9 m (Fig. 16). This site is situated in a compound of abandoned river bed and flood plain, and therefore is surrounded by liquefaction-prone soil conditions. The subsurface conditions here are illustrated in Fig. 17, indicating a thick clayey layer with interbedding sandy soil under the levee. Noteworthy is that the Bsc soil on the river side is composed of volcanic ash and is prone to large distortion under earthquake loading (Kazama et al., 2006; Verdugo, 2008; Unno et al., 2008). Because the river water level rose by retention behind a small dam near this site, the ground water level in the levee was

higher than usual and liquefaction in the levee might have been possible.

The site of Kita-Wabuchi (Fig. 18) is located near the confluence of the Eai river and the Kitakami river. The levee here did not incur any seismic damage, but the area behind the levee was an abandoned river channel and liquefaction occurred for the 4th time in the recent decades (twice in 1978 during main- and after-shocks, in 2008, and in 2011). Thus, liquefiable soil repeatedly liquefied, and the minor densifica-tion during the dissipation of excess pore water pressure is not significant enough to prevent further liquefaction.

3. Earthquake motion in the concerned region

One of the important features of the earthquake of M — 9 was the long duration time of shaking, which produced a significant number of cyclic loading on soils and resulted in extensive liquefaction and related ground deformation. Although the details of ground motion at the concerned river levees varied from place to place and the exact motion in respective levees is unknown, the present paper shows general feature of motion in the Sendai region in Tohoku, and the Kanto region particularly along the Tone river.

Fig. 19 indicates the time history of EW acceleration to the K-Net Sawara site near the damage in Fig. 2. The maximum acceleration of this EW motion was 301 cm/s2, which was greater than the maximum acceleration of the NS component (277 cm/s2). The duration time of the record at

Fig. 11. Significant distortion of the Naruse river levee at Shimo-Nakanome. (a) Damaged shape of the levee (viewed in the upstream direction); (b) Sand boils found in the crack (NILIM, 2011).

Sawara is expressed by two indices that are D5 95 and the duration of the amplitude exceeding 50 cm/s2. Both took into account the combination of NS and EW components and the former was defined by D5 95 — Elapsed time between 5% and 95% development of /{(NS Acceleration)2 + (EW Acceleration)2}dt (Fig. 20) which is a kind of Arias Intensity (Arias, 1970).

The latter supposes that the acceleration of VNS2 + EW2 being greater than 50 cm/s2 is strong enough to maintain high excess pore water pressure and produces more deformation of levees (Okada et al., 1999).

Moreover, the time history of the EW motion (571 cm/s2 at maximum) at the K-Net Furukawa station to the north of Sendai, Tohoku region, is illustrated in Fig. 21. By comparing the records in Figs. 19 and 21, it is found that the motion in the Kanto region to the south consists basically of one earthquake event, while the motion in the Tohoku to the north is of two major phases. This difference was caused by the proximity to the causative rupture zone, which had more than one seismic event.

The distribution of the seismic indices such as the maximum horizontal accelerations, D5 _ 95 and the duration of acceleration greater than 50 cm/s2 is illustrated in Fig. 22. The latter is often employed in the authors' practice as giving the idea of duration of strong acceleration (Okada et al., 1999). Both the acceleration and the duration times were greater in the Sendai, Tohoku, region than in the Kanto region, and made the earthquake more destructive there.

4. Tsunami overtopping

The tsunami was the most important cause of damage during the earthquake in 2011. Many sea walls were destroyed by the overtopping of sea water. Similarly, there are river levees that were destroyed by the tsunami. Fig. 23

Fig. 12. Cross section of subsoil at the Shimo-Nakanome site on the Naruse river.

Fig. 13. Distortion of Abukuma river levee at Edano site (seen in upstream direction) (The yellow dotted line shows the location of excavation study in Fig. 29). (For interpretation of the references to color in this figure, the reader is referred to the web version of this article.)

Fig. 14. Subsidence of the Abukuma river levee at Edano site.

indicates the Kitakami river levee at Kamaya (R 3.8 km). Note that the entire body of the levee was lost due to the onslaught of the tsunami attack and scouring. Fig. 24 shows the significant scouring and erosion of a levee of the Kesen river in Rikuzen-Takata, which was devastated by tsunami. It is thought at present that the scouring in Fig. 24 occurred when the tsunami water was receding towards the sea. A vortex was most likely significant on the upstream side of the bridge abutment, eroding the levee soil therein.

The case of the Natori river levee at Yuriage (near the river mouth south of Sendai) appears different from the above-mentioned cases. Fig. 25 illustrates the evidence of subsoil liquefaction that occurred immediately after the earthquake. It seems that subsoil liquefaction made the levee unstable. However, when the tsunami overtopped the levee later, the levee was able to maintain its shape (Fig. 26). Although further study is necessary, this finding is interpreted as follows:

(1) The entire Yuriage area was inundated by tsunami water that came in directly from the coastline.

(2) When tsunami water overtopped the levee from the river channel, the ground surface behind the levee had been already covered and was protected by water and hence scouring did not occur.

(3) Moreover, the direction of tsunami propagation along the Natori river channel was parallel to the levee, thus reducing the impact of the tsunami.

(4) In contrast, the Kitakami River mouth is of rias shape, with its open entrance collecting and concentrating the tsunami energy in the channel. The Kamaya site is located upon the bend of the channel and was impacted by the tsunami energy more directly.

One important lesson for future is the consideration of the regional and tectonic subsidence of the earth crust, which is otherwise called co-seismic subsidence. In the

Fig. 15. Subsoil condition at Edano site on the Abukuma river.

past, this type of subsidence happened during the 1946 Nankai earthquake, the 1960 Chile earthquake, the 1964 Alaska earthquake, and the 2004 Indian Ocean earthquake together with other gigantic events. The present earthquake was associated with the subsidence of a maximum of 1.2 m in the coastal area. This implies that the height of levees near the mouth was suddenly lowered by 1.2 m, followed by the tsunami. Where this compound hazard mechanism is expected, the levees have to be made higher to compensate appropriately.

5. General discussion on damage of levees

It is important to study the effects of local geology on the significance of levee damage. Fig. 27 was drawn by

Fig. 16. Levee distortion at Fukunuma levee of the Eai river.

using data from relatively significant damage sites. This damage summary indicates that the majority of damage occurred in the flood plain where there are many abandoned channels. Note that the current geographical database on geomorphology classification is not so accurate as to precisely define the border between each kind of geology to the order of a few meters. This is probably the reason why Fig. 27 shows a great amount of damage in the natural levee where the liquefaction risk is not considered

Fig. 18. Repeated liquefaction and sand boils at the Kita-Wabuchi site on the Eai river.

Fig. 17. Subsoil conditions at the Fukunuma site on the Eai river.


® s 8 1

Main shock on 20110311 EW acceleration K-Net Sawara CHB004

0 50 100 150 200 250 300

Time (second)

Fig. 19. Time history of EW acceleration at K-Net Sawara station.


«2 .13 50

1 1 1 1 1 1 1 1 I Ii II in i i i I i i i I

- f 9 5% ■

Accumulation of ns2+ew2 with time

- 5% -

1 1 1 1 i i i i i i i i i i i i i i i i i i i i

^ 0 50 100 150 200 250 300

Time (second)

Fig. 20. 5% and 95% development of Arias Intensity in K-Net Sawara record.

TT—|-T-|—I I I I

Furukawa2011 .qpc

in shock on 20110311 EW acceleration K-Net Furukawa MYG006


0 50 100 150 200 250 300 Time (second)

Fig. 21. Time history of EW acceleration at K-Net Furukawa station.

high, as is empirically known (Wakamatsu, 1992). The natural levee is not the most prone to liquefaction probably because of the ageing of sand (Mulilis et al., 1977; Tatsuoka et al., 1988). Further note that Fig. 27 is a damage summary and should not be misinterpreted as a statistic database.

It has been empirically known that the earthquake-induced subsidence of river levees (S) does not exceed 75% of the original height (H); see the Fig. 28. The new data obtained from the earthquake in 2011 remains within 75% as well at the worst case, and still supports this empiricism.

Fig. 10 illustrates the possible mechanism of liquefaction in the body of the levee. This mechanism was proposed after the 1993 Kushiro-Oki earthquake in which levees of Kushiro and Tokachi rivers resting on peaty soft soil were distorted significantly (Sasaki et al., 1994, Kaneko et al., 1996, and Oshiki and Sasaki, 2001), and a numerical analyses by Finn et al. (1997) was in agreement with the idea. To further validate the liquefaction inside the levee,

Fig. 22. Distribution of earthquake indices in Tohoku and Kanto regions. (a) Tohoku region (b) Kanto region.

excavation was conducted in 2011 at several places, although the range of excavation was limited by the depth of ground water. Fig. 29 shows one of them where the bottom of the levee had sunk into the clayey subsoil. Upon the earthquake, significant depression in the central part of the levee induced a squeeze-out of bottom sandy materials towards the landside, while the boundary between the levee and the subsoil (Ac1), which was determined after the earthquake by geolicers, bore hole drillings (Fig. 15), and soundings, did not deform significantly.

A more interesting study was made at the Shimo-Nakanome site (Fig. 11). Fig. 30 indicates that the ground

Fig. 23. Kitakami river Levee at Kamaya under restoration after total destruction by the tsunami attack (view in the upstream direction).

Fig. 24. Significant erosion of levee upon retreat of the tsunami water (looking upstream of Kesen river, Rikuzen-Takata, from a bridge abutment).

Fig. 25. Liquefaction behind the Yuriage Levee prior to the attack by the tsunami (looking towards the downstream area on the right bank, Photograph was taken by a local resident, Mr. SeishinKosai, and supplied to the authors by Tohoku Regional Development Bureau).

water level (GWL) was situated in the levee body on April 30th, and still remained at the bottom of the levee on July 13th (Fig. 31), although it got lowered possibly because the

Fig. 26. Appearance of the Yuriage Levee after the attack by the tsunami (looking towards the downstream area on the right bank.

Rivers :


I Abukuma I Naruse □ Kitakami Tone Kuji Naka

Surface geology


Former swamp Reclaimed land

Flood plain

Former river channel and scoured pond Slightly high place in abandoned channel

Natural levee

0 5 10 15 20 25 30 Number of significant damages

Fig. 27. Relationship between number of levee damage and geomorphology.

failed levee soil had been removed. The accumulated precipitation between March 11th and April 30th was merely 76 mm, while the precipitation in one month prior to July 13th was only 131 mm. Thus, it may be stated that the subsided bottom of the river levee had plenty of pore water when the earthquake occurred. Note, however, that the observed subsidence is a post-earthquake configuration and that further study is needed on the pre-earthquake situation of levees resting on clayey subsoil, including the density of the soil. Finally, Fig. 32 demonstrates the cross section of the levee at the Shimo-Nakanome site. Although the depth of excavation was limited by the level of ground water, the disrupted soil in the lower part of the levee supports the likelihood of liquefaction inside the levee; see displacement of the fill materials (designated by B) as well as the null deformation in the Ac1 layer at the bottom. In

contrast, the river side with more clayey materials exhibited much less deformation.

To shed light on the idea of liquefaction inside the body of levees, the factor of safety against liquefaction, FL, was

Significant cases in 2011 acldecl to data from Japanese earthquakes since 1891

• Tolioku Regional Development Bureau as of April 12tli

• Kanto Regional Development Bureau as of May 8tli

• * ♦ ♦ ■ • *

/ ЩО * 1 • о v.? • ■ * ш Л а & А

/ 4/ C' or* * о e<l • ОС- 0 1 • • ' Те оо гЧ Ч « Л» ТЭ ■ I : ■ С ° • >

/ / * Ф л i, Й Оо Ой/ а >*• м ж г° ^ о а * Л : л С ф □ » . • • • • • •

01 23456789 10 Height of levee, H (m)

Fig. 28. Relationship between levee height and seismic subsidence at crest (new data in 2011 was added to the original diagram by MLIT, 2007).

assessed near the bottom of levees where the soil type was sandy and hence liquefaction was considered likely. The FL value was calculated by using the Highway Bridge Design Code of Japan Road Association (2002 version). The intensity of maximum acceleration at the ground surface, k, was determined by referring to nearby K-Net data. Note, therefore, that the calculation is of limited accuracy because the intensity of ground motion at the exact site of levees is not available. Moreover, to take account of the significantly

Fig. 31. Observed ground water on July 13th at Shimo-Nakanome site. (For interpretation of the references to color in this figure, the reader is referred to the web version of this article.)

Ongnal Shape

Ranje of depression

Fig. 29. Excavated cross section of damaged levee at the Edano site on the Abukuma river (the same site as in Figs. 13-15) (Geoslicer was employed for the study below the ground water table).

OWL oil April 30 th

Fig. 30. Excavated cross section of damaged levee at the Shimo-Nakanome site on the Naruse river (see Fig. 12 as well as Table 1 for symbols).

Fig. 32. Excavated cross section of the levee at the Shimo-Nakanome site on the Naruse river.

Table 2

Assessment of possibility of liquefaction in the sandy bottom part of levees.

Rivers Borehole data in this paper Surface acceleration k (G unit) Depth (m) SPT-N Fl

Hinuma lake Fig. 9 0.546 4.3 6 0.588

Naruse river Fig. 12 0.458 2.3 3 0.162

3.3 2 0.120

Abukuma river Fig. 15 0.411 3.3 3 0.264

4.3 4 0.263

4.8 2 0.207

5.3 5 0.309

5.8 6 0.321

Eai river Fig. 17 0.571 3.3 3 0.182

4.3 3 0.341

5.3 4 0.341

long duration of time and the large number of cycles of shear stress application induced by the earthquake of M— 9, the liquefaction resistance of sand was reduced by 20% from the conventional design assessment by using Cw—0.8. The values of FL in Table 2 are less than unity in those damaged levees and suggest that liquefaction inside the levee is likely to have occurred during the earthquake.

There is an opinion that the induced subsidence of levees was possibly the consequence of a collapse mechanism in which partially saturated soil develops volume contraction upon increasing water content towards saturation. This opinion is not useful at all in the present study because, first, the bottom of the levee, lying below the ground water level, had been saturated with water, and, second, there was no such phenomenon as rainfall to increase the water content in the damaged levees at the same time as the earthquake. Moreover, note that those damaged levees in Table 2 were located in the upstream areas too far from the tsunami-prone coastal area.

From a practical viewpoint, liquefaction within the levee

body poses the following problems:

(1) It is difficult to identify the possible sites of this problem because levees are designed and constructed above the ground surface and the situations that occur afterwards are not visible.

(2) There are no detailed construction records for many levees. This is particularly the case when the levee was first constructed in the 19th Century or earlier, followed by further filling and widening in modern times.

(3) It is therefore difficult to assess the potential of liquefaction inside the levee body over the entire length of levee resting on soft subsoil.

(4) A detailed mechanism of the loosening of once-compacted levee soil is yet to be known. In some regions, levees on soft clayey subsoil have been constructed without intense compaction, different from soil-mechanic tradition, in order to avoid consolidation and subsidence.

(5) If liquefaction risk is high, pore water in the levee body has to be removed. If the liquefiable part is located above the ground surface, drainage is feasible. However,

if it is below the ground surface, drainage of pore water at an appropriate cost is not an easy task.

6. Successful mitigations

Seismic retrofitting of levees has been underway during the recent decade at places where seismic inspection identified potential problems. Generally, mitigations have been designed by using the pseudo-static seismic coefficient equal to 0.18 that corresponds to a design earthquake hitting the site once during the service period.

6.1. Gravel drains at the Omigawa bridge site on the Tone river

Fig. 33a illustrates the successful soil improvement on the land side of a levee by using gravel drains that successfully dissipated excess pore water pressure and prevented liquefaction in 2011. Its location is on the upstream side of Omigawa bridge of the Tone river (R 27.8 km), which is situated within a flood plain and abandoned river channel. For the subsurface cross section, see Fig. 34. This situation is in clear contrast with the heavy damage at a nearby site where no mitigation had been installed (Fig. 33b).

7. Drainage from levee at Mitanda site on the Naka river

Lowering of ground water level in the body of a levee and keeping the soil unsaturated are important in order to prevent liquefaction inside the levee (Fig. 10). The installation of a gravel toe drain at the foot of the slope (Fig. 35) has been widely practiced in order to lower the phreatic surface in the levee during high water level in the river. Fortunately, in the case of Fig. 36, this measure was successful in lowering the water level at the time of the

Fig. 33. Successful performance of gravel drains as mitigation of liquefaction in river levees (Omigawa on the Tone river). (a) Levee section with gravel drains installed to the right of the crest; (b) Damaged section without soil improvement.

Fig. 35. Conceptual sketch of gravel toe drain.

Fig. 34. Cross section of the Tone river levee at Omigawa (R 27.75 km) with soil improvement by gravel drains.

Affected section Drained levee without damage

Fig. 36. Successful installation of toe drain in levee slope at the Mitanda site on the Naka river (no photograph is available at sites without damage).

Fig. 37. Longitudinal cracks in a levee section without gravel drainage (for location, see the levee section marked by red in Fig. 36.

Fig. 38. Earthquake damage during the 2003 North Miyagi earthquake at the Kimazuka site on the Naruse river.

earthquake as well and consequently avoiding the problem of liquefaction. For comparison, see the post-earthquake situation of a nearby site (Fig. 37).

River side

Fig. 39. Effects of soil improvement under Kimazuka levee in 2007.

7.1. Shallow soil stabilization by cement mixing at the Kimazuka site on the Naruse river

The Kimazuka levee on the Naruse river (R 12.9 km) was damaged by the 2003 North Miyagi earthquake with MJMA = 6.2 (Fig. 38 and Takahashi and Sugita, 2009). After that event, the surface 6 m, at maximum, of the foundation soil was improved by mixing soil with cement. Because of this improved liquefaction resistance in the foundation, the 2007 earthquake did not cause damage in the levee, although a neighboring area was affected by liquefaction as well (Fig. 39). This success was repeated in 2011.

7.2. Sand compaction piles at the Naka-Shimo site on the Naruse river

This section of the levee (near the river mouth) was damaged by the 1978 Off-Miyagi earthquake and it was partially improved by using sand compaction piles (Fig. 40); the improvement ratio (defined by cross sectional area of piles/ entire improved area) being 10%. Because no damage occurred in 2011, there is not much information about its behavior except that the levee was safely used for evacuation

Fig. 40. A cross section of the Naka-shimo levee with partial improvement by sand compaction pile (L and D stand for locations of pore pressure transducers in loose and dense part of the levee, respectively) (drawn after documents owned by the Research Center for Disaster Risk Management belonging to the National Institute for Land and Infrastructure Management).

—1—1—1—1— L. —i—i—i—i—J—i—i—i—i—J—i—i—i—i— Naka-shimo levee of Naruse River _

T EW accel. at crest 2003 North Miyagi earthquake -

5Й ЧЯ ^ И

и Ь ^ &

я SÉ

0.6 0.4 0.2 0

Time (second)

Naka-shimo levee of Naruse River i—i—i—i—i—r

-Capacity limit of sensor—


-2003 North Miyagi earthquake-j_i_i_i_i_I-

Time (second)

Fig. 41. Time histories of crest acceleration and excess pore water pressure ratio in Naka-shimo levee during an earthquake in 2003. (a) Acceleration; (b) Excess pore pressure ratio.

the damage to individual levees, the study further addressed the successful performance of soil improvement and other mitigation measures that were practiced prior to the quake. The chief conclusions drawn from the study are as follows.

(1) Most damaged levees are situated in flood plains and abandoned river channels, which are susceptible to subsoil liquefaction.

(2) Thus, liquefaction in the foundation subsoil is one of the major causative mechanisms of levee damage.

(3) There were many damaged levees without subsoil liquefaction. It is inferred that the body of those levees liquefied.

(4) The tsunami effect was varied, being substantial or minor, depending probably upon the geometrical relationship between the levee and tsunami propagation.

(5) Because of the significant number of damage events and their spatial distributions, it was difficult to restore them within the short time between the earthquake itself and the rainy season.

(6) Mitigation measures such as compaction, solidification, and drainage successfully prevented the occurrence of levee distortion during the earthquake.

from the tsunami attack. However, the experience during the 2003 North Miyagi earthquake (MJMA = 6.2) provides some information. Fig. 41 compares the records of excess pore water pressure in the sandy part during the 2003 event. Evidently, the excess pore water pressure in the densified part ("D" in Fig. 40) is lower than in the uncompacted part ("L"). It is reasonable to infer that the same situation occurred in 2011 as well, successfully preventing the onset of liquefaction and reducing the deformation.

8. Conclusions

A damage investigation has been carried out on river levees after the earthquake on March 11, 2011. While focusing on


The present paper summarizes the damage investigation activities that were conducted by the authors. Most of the activities were performed as a governmental (MLIT) activity. However, there are certainly products of individual activities, some of which were organized by JGS and JSCE, and they are included in this paper. Mr. Yuichi Taguchi of Fudo-Tetra Corporation and Mr. Shogo Aoyama, who is a graduate student of University of Tokyo contributed by participating in the field investigations, drawing some of the figures and supplying maps. The authors express their sincere gratitude to those institutes and individuals who were involved in the

related damage investigations and supported the authors' activities.


Arias, A., 1970. A measure of earthquake intensity. In: Hansen, R.J. (Ed.), Seismic Design for Nuclear Power Plants. MIT Press, pp. 438-483.

Finn, W.D.L., Sasaki, Y., Wu, G., 1997. Simulation of response of the Kushiro River Dike to the 1993 Kushiro-oki and 1994 Hokkaido Toho-oki earthquake. In: Proceedings of the 14th ICSMFE, Hamburg, vol. 1, pp. 99-102.

Japanese Geotechnical Society, 2009. Technical proposal towards mitigation of geotechnical hazards caused by earthquakes, heavy rains, and floods, p. 145 (in Japanese).

Kaneko, M., Sasaki, Y., Nishioka, J., Nagase, M., Mamiya, K.,1996. River dike failure in Japan by earthquake in 1993. In: Proceedings of the 3rd International Conference on Recent Advances in Geotechnical Earthquake Engineering and Soil Dynamics, pp. 495-498.

Kazama, M., Takamura, H., Unno, T., Sento, N., Uzuoka, R., 2006. Liquefaction mechanism of unsaturated volcanic sandy soils. Journal of Geotechnical and Geoenvironmental Engineering, JSCE 62 (2), 546-561.

Matsuo, O., 1996. Damage to river dikes, Soils and Foundations, Special Issue 1 on Geotechnical Aspects of the January 17 1995. Hyogoken-Nambu Earthquake, pp. 235-240.

Ministry of Construction, 1985. Technical standards for river and erosion-control structures (in Japanese).

Ministry of Land, Infrastructure, Transport and Tourism, 2007. Guideline for inspection of seismic resistance of river structures, II, river levees (in Japanese).

Mulilis, J.P., Mori, K., Seed, H.B., Chan, C.K., 1977. Resistance to liquefaction due to sustained pressure. Journal of Geotechnical Engineering ASCE 103 (GT7), 793-797.

National Institute for Land and Infrastructure Management, 2011. Technical Report, Prompt report on damage of civil structures caused by the 2011 Great East Japan earthquake (NILIM and PWRI). In: Chapter 6, Damage of River Structures, p. 264.

Okada, S., Orense, R., Kasahara, Y., Towhata, I., 1999. Prediction of liquefaction-induced deformation of river embankments. In: Proceedings of the 2nd International Conference on Earthquake Geotechnical Engineering, Lisbon, vol. 2, pp. 543-548.

Oshiki, H., Sasaki, Y., 2001. Restoration works of seismically damaged river dikes using remedial treatment of liquefiable layer. Proceedings of JSCE 686/VI (52), 15-29.

Sasaki, Y., Oshiki, H., Nishikawa, J., 1994. Embankment failure caused by the Kushiro-Oki earthquake of January 15, 1993. Session publication during the 13th ICSMGE, Delhi, pp. 61-68.

Takahashi, A., Sugita, H., 2009. Behaviour of SCP-improved levee during 2003 Miyagiken-Hokubu earthquake, earthquake geotechnical case histories by performance-based design-from case history to practice. In: T. Kokusho, ed. Proceedings of the IS Tokyo 2009, Tsukuba, Taylor and Francis, pp. 177-184.

Tatsuoka, F., Kimura, H., Pradhan, T.B.S., 1988. Liquefaction strength of sands subjected to sustained pressure. Soils and Foundations 28 (1), 119-131.

Towhata, I., Goto, H., Kazama, M., Kiyota, T., Nakamura, S., Wakamatsu, K., Wakai, A., Yoshida, N., 2011. On gigantic Tohoku Pacific earthquake in Japan. ISSMGE Bulletin 5 (2), 46-66.

Unno, T., Kazama, M., Uzuoka, R., Sento, N., 2008. Liquefaction of unsaturated sand considering the pore air pressure and volume compressibility of the soil particle skeleton. Soils and Foundations 48 (1), 87-99.

Verdugo, R., 2008. Singularities of geotechnical properties of complex soils in seismic regions. Journal of Geotechnical and Geoenviron-mental Engineering, ASCE 134 (7), 982-991.

Wakamatsu, K., 1992. Evaluation of liquefaction susceptibility based on detailed geomorphological classification. In: Proceedings of the Annual Meeting on Architectural Institute of Japan, pp. 1443-1444 (in Japanese).