J Seismol (2009) 13:399–413 DOI 10.1007/s10950-008-9146-7

ORIGINAL ARTICLE

Geotechnical evaluation of slope and ground failures during the 8 October 2005 Muzaffarabad earthquake, Pakistan Ömer Aydan · Yoshimi Ohta · Masanori Hamada

Received: 27 April 2007 / Accepted: 17 November 2008 / Published online: 10 March 2009 © Springer Science + Business Media B.V. 2008

Abstract A large devastating earthquake with a magnitude of 7.6 struck in Kashmir on Oct. 8, 2005. The largest city influenced by the earthquake was Muzaffarabad. Balakot town was the nearest settlement to the epicenter, and it was the most heavily damaged. The earthquake caused extensive damage to housing and structures founded on loose deposits or weathered/sheared rock masses. Furthermore, extensive slope failures occurred along Neelum and Jhelum valleys, which obstructed both river flow and roadways. In this article, failures of natural and cut slopes as well as other ground failures induced by the earthquake and their geotechnical evaluation are presented, and their implications on civil infrastructures and site selection for reconstruction and rehabilitation are discussed. It is suggested that if housing and

Ö. Aydan (B) Department of Marine Civil Engineering, Tokai University, Orido 3-20-1, Shimizu-ku, Shizuoka, 424-8610, Japan e-mail: [email protected] Y. Ohta Nippon Geophysical Prospecting Company, Tokyo, Japan M. Hamada Department of Civil and Environmental Engineering, School of Science and Engineering, Waseda University, Shinjuku, Tokyo, Japan

constructions on soil slopes containing boulders as observed in Balakot and Muzaffarabad are allowed, there should be a safety zone between the slope crest and allowable construction boundary. Keywords Slope failure · Liquefaction · Geotechnical evaluation · Muzaffarabad · Rock slopes · Soil slopes

1 Introduction A large devastating earthquake of magnitude of 7.6 occurred near Muzaffarabad on Oct. 8, 2005. The hypocenter depth of the earthquake was estimated to be about 10 km. The earthquake killed more than 82,000 people in Kashmir (JSCE 2005). The earthquake resulted from the subduction of Indian plate beneath Eurasian plate, and the faulting mechanism solutions indicated that the earthquake was due to thrust faulting (Harvard 2005; USGS 2005). The surface expression of the causative fault extends between Bagh and Balakot through Muzaffarabad (GSP 2006). The city of Muzaffarabad and Balakot town were the nearest settlements to the epicenter, and they were the most heavily damaged (Fig. 1). Valleys are filled with weakly cemented conglomeratic deposits. Fast-flowing rivers cut through these deposits, resulting in very steep slopes. The

400

J Seismol (2009) 13:399–413

Fig. 1 Locations of the epicenter and fault of the earthquake, major settlements and liquefaction sites (base map arranged from Google Map)

earthquake caused extensive damage to housing and structures founded on steep soil slopes (Aydan and Hamada 2006). Furthermore, extensive rock slope failures occurred along Neelum and Jhelum Valleys, which obstructed both river flow and roadways (Aydan 2006a, b; Sato et al. 2007; Dunning et al. 2007). Ground liquefaction was also observed along Neelum and Jhelum Valleys, which resulted in some lateral spreading (Aydan 2006a, b; Sahoo et al. 2007). In this article, the characteristics of the failure of natural and cut slopes induced by the earthquake are presented and evaluated. Furthermore, their implications on civil infrastructures and site selection for reconstruction and rehabilitation are discussed. In addition, examples of ground liquefaction, which are quite limited in scale, are described.

2 Failures of slopes and embankments One of the most distinct characteristics of 2005 Muzaffarabad earthquake is the widespread slope failures all over the epicentral area. The Muzaffarabad earthquake of October 8, 2005

particularly caused extensive damage to housing and structures founded on sloping soil deposits and weathered/sheared rock masses. Extensive natural and cut slope failures occurred along Neelum, Jhelum, and Kunhar valleys, which obstructed both river flow and roadways. Furthermore, many slope failures associated with highly sheared and weathered dolomite occurred along the surface expression of the earthquake fault. The failures near Muzaffarabad were spectacular in both scale and distributions.

2.1 Classification and examples of slope failures Slope failures caused by the Muzaffarabad earthquake of October 8, 2005 may be classified into three categories: (1) soil slope failures, (2) weathered and/or sheared rock slope failures, and (3) rock slope failures as illustrated in Fig. 2 (Aydan 2006a, b; Aydan and Hamada 2006; Aydan and Kawamoto 1992; Aydan et al. 1989, 1991). Soil slopes failures occurred in either plane sliding mode or rotational sliding with surface fractures. Planar sliding modes are generally observed on soil slopes over bedrock (Fig. 3a). Deep-seated

J Seismol (2009) 13:399–413

401

Fig. 2 Illustration of slopes failure modes caused by Kashmir earthquake (re-arranged from Aydan 1989; Aydan et al. 1989, 1991; Aydan and Kawamoto 1992; Aydan and Hamada 2006)

Fig. 3 Soil slope failures. a Slope failure in Balakot. b Slope failure nearby Muzaffarabad. c Slope failure 3 km north of Balakot. d Slope failure 6 km north of Balakot

(a)

(c)

(b)

(d)

402

J Seismol (2009) 13:399–413

Fig. 4 Heavy damage near the crest of slopes in Muzaffarabad

circular type soil slope failures were observed where the soil thickness was quite large (Fig. 3c, d). Some peculiar soil slope failures were observed in both Balakot and Muzaffarabad in the form of lateral spreading of loose conglomeratic deposits (Fig. 2b). In addition, the damage was quite heavy near the crest of slopes due to permanent ground movement as well as ground amplification at such locations. These slope failures occurred

in conglomeratic soil deposits (Fig. 4). Since the original slope angles were quite steep (60–80◦ ), these slopes failed either by lateral spreading or shearing as illustrated in Fig. 2. The residual slope angle (repose angle) ranges between 40 and 45◦ , which may be considered to be roughly equivalent to its friction angles (Fig. 5). Some needle penetration tests were also carried out on soil matrix in both Balakot and Muzaffarabad (Fig. 2a, b).

Fig. 5 Relation between slope angle and frequency of conglomeratic soil deposits and actual examples

J Seismol (2009) 13:399–413 Fig. 6 Views of some rock slope failures. a Flexural toppling in Jhelum Valley (6 km southeast of Muzaffarabad). b Columnar toppling (10 km south of Muzaffarabad). c Planar sliding 3 km north of Balakot. d Surficial slope failure of fractured dolomite (3 km south of Balakot)

403

(a)

(c)

The cohesion of soil matrix of the conglomeratic deposit is inferred from the needle penetration index tests to be ranging between 60–100 kPa. Embankments of roadways failed along the rivers. Stone masonry or gabions are commonly used for supporting the embankments of roadways in steep terrain and along rivers. However, the embankments of roadways are not generally protected by retaining walls or gabions for a great length, which may make them prone to failures as a result of toe erosion due to fast river currents. They may also suffer from heavy rainfalls in the long term. No support or protection measures for most of slope cuts along roadways are undertaken. Furthermore, the slope cuts are generally very steep, and there are no catchments in case of rock falls and small-scale slope failures. Rock units in the epicentral area are mainly sandstone, shale, and dolomite (GSP 2005). Particularly red shale of Murree formation is prone to weathering, and the thickness of the weathered zone seems to be about 2–5 m. The whitish

(b)

(d)

Dolomite is intensively fractured, and this unit is thought to constitute the fault fracture zone. The surficial slope failures in dolomite were spectacular (Figs. 6d and 8) and continued for

Fig. 7 A view of Hattian landslide (picture by A.B. Kausar of GSP)

404

a

b

Fig. 8 Some slope failures (landslides) associated with the earthquake faulting. a Numerous slope failures south of Balakot in Kunhar Valley. b Numerous slope failures north of Muzaffarabad in Neelum Valley

several kilometers, as they are clearly noticed in satellite images. Except granite, all rock units have at least one thoroughgoing discontinuity set, namely, bedding plane or schistosity plane. Since rock units had been folded, they also include joint sets and fracture planes as a result of tectonic

Fig. 9 Views of localized slope failures in the vicinity of faulting model tests (from Aydan and Ohta 2006)

J Seismol (2009) 13:399–413

movements. Rock slope failures are mainly planar or wedge sliding failure or flexural or block toppling failure (Fig. 6). Planar sliding failures were observed mainly in schists, sandstone, and shale, while flexural toppling failure was observed in intercalated sandstone and shale depending upon the inclination of bedding planes with respect to slope geometries. The inclination of layers ranges between 30 and 65◦ , which implies that the sliding failure can be easily caused by a small intensity of disturbing forces resulting from such as earthquakes, heavy rainfall, or the both. Rock falls in the epicentral area generally resulted from the toppling of rock blocks due to strong shaking induced by the earthquake. Numerous rock falls were observed adjacent to roadways, and they were particularly common in sandstone slopes. Some flexural slope failures were also observed in intercalated sandstone and shale with undercutting. The satellite images indicated that there was a large-scale slope failure near Hattian in the vicinity of the SE tip of the causative fault. The slope failure was at Hattian (Dana Hill), and it is an asymmetric wedge sliding (Fig. 7). The estimated wedge angle is about 100◦ . The inferred trace of the fault is about 80– 90 km from aftershock activity, and its NW tip is located near Balakot (GSP 2006; Aydan et al. 2006). The inferred earthquake fault trace follows the line from Balakot town to Muzaffarabad City to Jhelum valley. Although the surface fracture as a result of the faulting was not easily distinguishable on land, extensive slope failures observed along the expected surface expression of the causative fault from aerial photographs and land surface observation along this line indicated

J Seismol (2009) 13:399–413

405

Table 1 Values of constants in Eq. 2 in relation to earthquake type Earthquake type

A

B (m/s)

C (km)

D

E

F(M) (Mw )

Intra-plate Inter-plate

2.8 2.8

1,000 1,000

40 66

0.5 0.5

1.5 1.5

1.16 1.05

numerous slope failures (Aydan 2006a, b; Aydan et al. 2006). There were numerous slope failures particularly on the hanging-wall side of the earthquake fault as compared with those on the footwall side of the earthquake fault. Furthermore, the areal extension of the slope failures is much larger on the hanging-wall side than that on the footwall side (Sato et al. 2007). They were generally associated with the whitish dolomite layer, which is a highly deformed and fractured rock unit (Fig. 8). Figure 9 also shows a laboratory test on ground deformation in loose non-cohesive ground on a rigid base during thrust faulting. As noted from the figure, in spite of upward motion of the bedrock, some surficial slope failure (similar

Fig. 10 Estimated maximum surface ground acceleration (rocky ground)

to small scale normal faulting or slope failures) occurs within the non-cohesive deposits due to gravity. Furthermore, the thickness of deposits increases above the tip of the moving hanging wall as a rigid body. Therefore the slope failures observed particularly on the hanging-wall side of the fault can be interpreted as the surface expressions of the earthquake faulting.

2.2 Geotechnical evaluation of slope failures The geotechnical evaluation of slope failures requires ground motion data for the stability analyses. The nearest strong motion record was taken at Abbottabad, which is about 58 km away from the hypocenter and located on the footwall of the causative fault. Since there was no strong motion data in the epicentral area, some estimation must be done. Aydan and Ohta (2006) suggested the following formula for estimating the maximum ground acceleration by considering the position (hypocentral distance and angle with respect to the strike of the earthquake fault) of

406

J Seismol (2009) 13:399–413

location with respect to the causative fault, ground shear wave velocity, and magnitude   amax = F (Vs ) × G (R, θ) × H (M) unit: gals (1)

Table 2 Inferred amax and vmax values for Balakot town Method

Amax (Gal)

Vmax (kine)

Block sliding model Block toppling model Slope failure model (Aydan 2006a, b)

> 0.7 > 0.9 > 1,280

280–340 330–390 > 289

where F (Vs ) = Ae−Vs / B ; 2 G (R, θ) = e−R(1−D sin θ+E sin θ )

 C

;

H (Ms ) = e M/ F − 1

(2)

Constants A, B, D, E, and F are empirical constants, and their values differ depending upon the characteristics of the earthquake and they are given in Table 1 (Aydan and Ohta 2006; Aydan 2007). Aydan (2006a, b) did such estimations for strong motions on the ground surface for the ground with a shear wave velocity of 250 m/s. In this study, the estimations have been done for the surface of ground with a shear wave velocity of 1,000 m/s as shown in Fig. 10. This estimation would be valid for rocky ground, which may also be regarded as base acceleration. The estimation indicates that the base acceleration could have been more than 0.6 Gal in Muzaffarabad and Balakot. This value would be amplified several times when ground becomes softer. An example of stability analyses for failed slopes of conglomeratic soil deposits observed in Balakot and Muzaffarabad is carried out using the planar slid-

ing method with the use of the strong motion records taken at Abbottabad strong motion station (Okawa 2005) and inferred cohesion and friction from in situ index tests and investigations. In the computation, the record was multiplied by a factor until the failure was induced and the postfailure configuration was simulated. The results are shown in Fig. 11 for a multiplication factor of 6.8. The inferred maximum ground acceleration, velocity, and displacement of ground were 1,280 Gal, 289 cm/s, and 500 cm, respectively. This estimation is in accordance with the estimations for soft ground by Aydan (2006a, b). Aydan (2006a, b) utilized some toppled or displaced structures to infer the strong motion characteristics in the epicentral area using the method proposed by Aydan (2002). Results are given in Table 2. On the bases of estimations for ground motions, the stability of soil slopes were backanalyzed. In the computations, the failure surface was obtained through a minimization procedure, and computational results are plotted as a function of slope height and lower slope angle as shown in Fig. 12. In the computation, the friction angle was taken as 40◦ , since the measured

cp=80kPa; p=40; =20kN/m3 cr=0kPa; p=40; =45; i=60

1000

15

Vertical Yc

500

10

0 5

-500 Displacement of Mass Center

-1000 -1500 0

10

20

30

40

0 50

TRAVELDISTANCE (m)

ACCELERATION (gal)

1500

REQUIRED SLOPE ANGLE (o)

NS-UD components of Abbotabad Record multiplied by 6.8

90

PLANAR SHEAR FAILURE (KASHMIR EARTHQUAKE) Gravity+Seismic (η=0.9)

75 c/γ

60

i

6 5 4

45 30

α

3 2

15 0

3

γ =20 okN/m φ =40

1

10

20 30 SLOPE HEIGHT (m)

40

50

TIME(sec)

Fig. 11 A back analyzed of soil slope failure

Fig. 12 Stability chart for sandy gravels with boulders (cohesion is variable)

J Seismol (2009) 13:399–413

repose angle in situ can be safely assumed to be equivalent to the friction angle of failed soil. The normalized cohesion by soil unit weight was varied between 1 and 6, and the relation between slope height and slope angle for safety factor of 1 for a lateral seismic coefficient of 0.9, which is likely to be the ground acceleration levels at Balakot as well as at Muzaffarabad, was computed. The normalized cohesion of soil with cobbles was inferred from needle penetration index tests to be ranging between 3 and 5. For these parameters, the slope failures would be observed when the slope height was greater than 7.7 m. This computational result implies that some restrictions on either slope angle or slope height in the redevelopment of settlements in sloping ground must be implemented. It is also recommended that geotechnical parameters of ground should be measured before the commencement of reconstruction. The wedge-like sliding failure at Hattian was quite large in scale. The sliding area was about 1.5km long and 1.0-km wide. Rock mass consisted of shale and sandstone and it constituted a syncline (GSP 2005, 2006). The estimated wedge angle was about 100◦ , and it was asymmetric. The friction angle of shale from tilting test was more than 35◦ with an average of 40◦ . Figure 13a shows a kinematic analysis of the Hattian slope through the projection of structural planes on equal angle stereo-net together with friction cones. The limiting equilibrium analysis indicated that the safety factor of the slope would be 1.55 under dry static conditions. However, the mountain wedge becomes unstable when acceleration is equivalent to the horizontal seismic coefficient of 0.3, and the safety factor becomes 0.9 under such a condition. Some parametric studies for the same wedge sliding at Hattian were carried out, and the results are shown in Fig. 13b. The result indicated that the seismic loading was the most critical parameter governing the wedge sliding. The author utilized the strong motion record of Abbotabad station by multiplying by a factor of 1.8, which corresponds the seismic coefficient factor of 0.3 by using the procedure proposed by Aydan and Ulusay (2002) and Aydan (2007). Computed velocity and displacement responses are shown in Fig. 14. The results imply that once the failure is initiated, the

407

(a) Kinematic analysis

(b) parametric analysis Fig. 13 Kinematic and parametric stability analysis for Hattian wedge sliding failure

motion of the failed body could not stop unless the failure surface changes its configuration. Keefer (1984) studied landslides in USA and other countries, and he proposed some empirical bounds for landslides, which are classified as disrupted or coherent. The empirical bounds of

408

J Seismol (2009) 13:399–413

a

Fig. 14 Displacement and velocity responses of the sliding body

Keefer are not specifically given as a formula. Besides the empirical bounds of Keefer, Aydan (2007) proposed the following empirical equation for the maximum distance of disrupted and coherent landslides as a function of earthquake magnitude and fault orientation:   R = A × 3 + 0.5 sin θ − 1.5 sin2 θ × e B·Mw

Disrupted landslides

b

(3)

Constants A and B of Eq. 3 for disrupted and coherent landslides are given in Table 3. Since ground accelerations differ according to the location with respect to fault geometry, the empirical bounds proposed herein can provide some bases for the scattering range of observations. Aydan (2007) compiled landslides (it would be better to name as slope failures) caused by earthquakes according to Keefer’s classifications and plotted in Fig. 15 together with the data for slope failures caused by the 2005 Muzarraffabad earthquake. It is of great interest that the data for Kashmir earthquake is within the limits proposed by Aydan (2007) and Keefer (1984). The hypocentral distance of the Hattian landslide is about 43 km, and it is within the empirical limits.

Table 3 Parameters of Eq. 3 for disrupted and coherent landslides Condition

A

B

Disrupted Coherent

0.10 0.08

0.9 0.9

Coherent landslides

Fig. 15 Comparison observations

of

empirical

relations

with

2.3 Simplified procedures for assessment of rock slope failures and counter measures The geometrical configuration of mountains in the epicentral area is strongly influenced by geological folding process. The fold axes of the bedrock formations generally range between N–S and NW– SE. The valleys are faulted, and they may also

J Seismol (2009) 13:399–413

409

90

=0o; =30o; k=0.2

=26.5o; =30o; k=0.2

90

60

60

30

30 Himalaya Kashmir

0 0

0 30 60 90 120 150 180 o INCLINATION OF BEDDING PLANE ( )

LOWER SLOPE ANGLE (o)

LOWER SLOPE ANGLE (o)

Fig. 16 Relation between bedding plane inclination and slope angle for natural slopes [limit lines from Aydan et al. (1989)] (ξ intermittency angle; k horizontal seismic coefficient; φ friction angle of bedding plane)

FLEXURAL TOPPLING FAILURE

REQUIRED SLOPE ANGLE (o)

correspond to either anticlines or synclines. As a result of these tectonic features, mountain slopes are entirely governed by a thoroughgoing discontinuity set, namely, bedding plane. The measurements on the planes at several locations (nearby Balakot and along the route between Murree and Muzaffarabad) indicated that their inclinations range between 34◦ and 65◦ . The in situ tilting tests revealed that the friction angle of bedding planes ranges between 30◦ and 40◦ , depending upon their surface morphology. The slope angle (lower part) of the natural mountain slopes in Kashmir together with those of Himalaya mountains in general are plotted in Fig. 16 as a function of bedding plane inclination, which may serve as guidelines for assessing the long-term slope angle. The natural slope of mountains in the epicentral area ranges between 30◦ and 75◦ . When the bedding planes dip toward valley side, the natural slope angle of the lower part of the mountain is almost equivalent to the inclination of bedding plane. This simply implies that the cohesion along the bedding planes is quite negligible. For the given height and tectonic features of rock mass, the natural analogy implies that the slopes would be resistant to pure shear failure or combined shear and sliding failure. However, when slopes are undercut, the planar sliding failure of bedded rock mass would be caused. Therefore, if the widening of present highways is to be carried out, the angle of slope cuts should be either parallel to the bedding plane, or reinforcement by rockbolts or rock anchors should be required.

65 Gravity Only Gravity+Water Gravity+Seismic (η=0.3) Gravity+Water+Seismic

60

σ t/γ =19 γ =25 kN/m3 φ =35o t=2m o α=135

55 50 45 0

100

200 300 SLOPE HEIGHT (m)

400

Fig. 17 Stability chart for rock slopes against flexural toppling

The mountains with bedding planes dipping into mountainside are more stable, and the slope angle is generally greater than 40◦ . It seems that the natural slope angle (i) of high mountains in relation to the bedding plane inclination (α) can be taken according to the following formula i = α − 90 + β

(4)

Where β is the rupture angle with respect to the normal to the bedding plane. Its value generally ranges between 10◦ and 15◦ in view of both model tests and case studies (Aydan et al. 1989; Aydan and Kawamoto 1992). The value of β for the natural slope angle of mountains in the epicentral area ranges between 10◦ and 12◦ . This implies that if the angle of slope cuts for roadways is in accordance with the above formula, there is no

410

J Seismol (2009) 13:399–413

Fig. 18 The concept of slope failure risk zoning along the transportation routes

road

need for any reinforcement measure in bedded rock mass. Figure 17 shows an example of computations of slope angle and slope height relations for bedding plane inclination of 135◦ by using the method of Aydan and Kawamoto (1992). Tensile strength of layers was inferred to be about 475 kPa in view of rock mass conditions in the epicentral area and in situ tests on shale in a hydraulic power plant project in Akaishi region of Shizuoka prefecture, Japan. While the lower angle of rock slope can be very steep (90◦ ) for a slope height of 6–10 m, the slope angle must be reduced in relation to the slope height when no reinforcement measure is implemented. It is recommended to prepare slope failure risk mapping along the transportation routes as illustrated in Fig. 18. In the first stage, the simplified procedure for assessment of rock slope failures risk can be used for identifying the high-risk zones. In high-risk zones, detailed investigations and stability analyses must be carried out depending upon the geological conditions of slopes and likely modes of instability under various combinations of loading conditions consisting of gravitational, rainfall, and earthquake shaking. In view of long-term stability of slopes, the requirement of slope stability without any artificial reinforcement is essential. When the slope angle and/or slope height are greater than those required for natural stability, the slope stabilization measures would be necessary. Furthermore, slopes of soil and soft-

sedimentary rocks may deteriorate or erode due to exposure to atmosphere, which may require the protection of slope surface. Probably slope angle reduction would be the most desirable measure, and the fundamental concept should be such that the overall stability of the slope must be attained by the self-resistance of ground (Fig. 19a). Therefore, the stabilization measures should be kept to a minimum. When there are some occasions, that is, it is impossible to reduce slope angle due to slope-cut height, other construction alternatives may be an effective way for dealing with the problem. Alternatively, if the construction next to slopes is allowed, there should be a safety zone between the slope crest and allowable construction boundary (Fig. 19b).

3 Ground liquefaction and lateral spreading Ground liquefaction was observed by the authors along Jhelum Valley at two locations, which were 12 and 16 km away from the epicenter of the earthquake. The ground liquefaction at a distance of 16 km from the epicenter occurred at a location where Jhelum river changes its flow direction from NW to NE. At this particular location, 100-m-long and 50-m-wide field laterally spread toward the river with a total displacement of 120– 160 cm. The lateral spreading induced separation cracks parallel to the river flow direction in the upper unsaturated layer, and sand boiling

J Seismol (2009) 13:399–413 Fig. 19 Some measures in reconstruction for earthquake affected area. a Slope angle reduction. b Implementation of safety zone

411

a

b

Fig. 20 Sand boiling and liquefaction induced lateral spreading (16 km from epicenter)

J Seismol (2009) 13:399–413

CUMULATIVE PERCENTAGE (%)

Clay

100

Silt

Sand

Gravel

100

Boundaries for most liquefiable soils

80

80

60

60

40

Boundaries for potentially liquefiable soils

40

20 0 0.001

20 Muzaffarabad 2005

0.01

0.1 1 GRAIN SIZE (mm)

0 10

CUMULATIVE PERCENTAGE (%)

412

Fig. 21 Comparison of grain size distribution of boiled soil samples with liquefiability bounds

occurred at the toe of laterally spread field (Fig. 20). Grain-size distribution, shear, and permeability tests were carried out on soil samples from sand boils. Figure 21 shows grain size distributions of soil samples together with liquefaction limits. As noted from the figure, grain-size distributions fall within easily liquefiable zone. There were some reports that ground liquefaction was also observed at several locations in the upperstream part of Jhelum River (Sahoo et al. 2007). Although the author had no chance to confirm the authenticity of these reports, the maximum distance for ground liquefaction from the hypocenter would be about 90 km (see also Fig. 1 for locations). Since there is no settlement in most of these areas, there is no structural damage resulting from ground liquefaction and associated lateral spreading of ground.

4 Conclusions and recommendations In this article, a geotechnical evaluation of failures of natural and cut slopes and ground induced by 2005 Muzaffarabad earthquake were presented. The maximum ground acceleration and velocity at Balakot were estimated to be, at least, 900 gals and 300 cm/s, respectively. The computational results indicated that the failure of soil slopes containing boulders was imminent under such strong ground motions. Furthermore, the loose surficial deposits and terrace deposits were later-

ally spread, which resulted in further damage in Balakot and in Muzaffarabad. Slope failures were observed along entire Neelum and Jhelum valleys. Particularly, slope failures associated with heavily fractured dolomite were spectacular in both scale and its areal distribution. However, these slope failures were aligned on locations, which may be interpreted as the surface expression of the causative fault. The implications of natural and cut slope failures on civil infrastructures and for the site selection of buildings and urban developments may be summarized as follows: 1. Both sides of the steep valleys should be connected to each other by bridges at certain intervals in order to facilitate by-pass routes in case of emergencies resulting from bridge collapses and/or slope failures. 2. Since the region is mountainous, it is recommended to implement other construction alternatives when there is a high risk of slope failures. 3. Embankment slopes are steep, and they are prone to fail either by ground shaking or heavy rainfall. It is recommended to either reduce slope angle of embankments or introduce support, reinforcement, or protection measures. Furthermore, measures should be introduced to eliminate the toe erosion. 4. The slope angle and slope height of slope cuts should be such that the slope is stable under its natural resistance. If such a condition is difficult to be fulfilled, some measures for support and reinforcement should be undertaken. Since the valleys are very steep, there is a high risk of surficial slope failures. 5. The design of slopes and the assessment of failure risk must be based on the principles of modern slope engineering. 6. Slopes of dolomite along the fault trace are susceptible to further failures. Therefore, it is recommended that no permission for housing or structure construction should be given in such locations. 7. Housing and constructions on soil slopes containing boulders as observed in Balakot and Muzaffarabad should not be allowed. Although these slopes can be stable for high

J Seismol (2009) 13:399–413

slope angles under static conditions, they are prone to failure during earthquakes. If the construction is allowed, there should be a safety zone between the slope crest and allowable construction boundary. Ground liquefaction was also observed during this earthquake. However, its effects were negligible, as there was no important engineering structure or settlement in such areas. Acknowledgements The author was a member of the reconnaissance and support teams dispatched by the Japan Society of Civil Engineers (JSCE) and Engineers without Borders-Japan (NPO) for 2005 Muzaffarabad earthquake in Azad Kashmir. Their financial supports for the travel expenses of the author are greatly acknowledged. The authors greatly acknowledge the information given by the members, in particular A.B. Kausar, of Geological Survey of Pakistan (GSP) during the visits and GSP Conference. The authors would also like to thank Prof. Harsh Gupta of National Geophysical Research Institute (NGRI) and Prof. Robert Yeats of Oregon State University for the invitation to contribute to the special issue on the 2005 Muzaffarabad earthquake.

References Aydan Ö (1989) The stabilization of rock engineering structures by rockbolts. Doctorate Thesis, Nagoya University, pp 204 Aydan Ö (2002) The inference of the earthquake fault and strong motions for Kutch earthquake of January 26, 2001. A Symposium on the Records and Lessons from the recent large domestic and overseas earthquakes. Japan Earthquake Engineering Society, Tokyo, pp 135–140 Aydan Ö (2006a) Geological and seismological aspects of Kashmir Earthquake of October 8, 2005 and a geotechnical evaluation of induced failures of natural and cut slopes. J Sch Mar Sci Technol 4(1):25–44 Aydan Ö (2006b) Strong motions, ground liquefaction and slope instabilities caused by Kashmir Earthquake of October 8, 2005 and their effects on settlements and buildings. Bull Eng Geol 23:15–34 (in Turkish) Aydan Ö (2007) The inference of seismic and strong motion characteristics of earthquakes from faults with a particular emphasis on Turkish earthquakes. The 6th national earthquake engineering conference of Turkey, Istanbul, pp 563–574 Aydan Ö, Hamada M (2006) Damage to civil engineering structures by Oct. 8, 2005 Kashmir Earthquake and recommendations for recovery and reconstruction. J Disaster Res 1(3):1–9 Aydan Ö, Kawamoto T (1992) The stability of slopes and underground openings against flexural toppling and

413 their stabilization. Rock Mech Rock Eng 25(3):143– 165. doi:10.1007/BF01019709 Aydan Ö, Ohta Y (2006) The characteristics of strong motions in the neighbourhood of earthquake faults and their evaluation. Symposium on the records and issues of recent great earthquakes in Japan and overseas, EEC-JSCE, Tokyo, June 7, 2006 Aydan, Ö, Ulusay, R (2002) Back analysis of a sesimically induced highway embankment during the 1999 Düzce earthquake. Environ Geol 42:621–631 Aydan Ö, Shimizu Y, Ichikawa Y (1989) The effective failure modes and stability of slopes in rock mass with two discontinuity sets. Rock Mech Rock Eng 22(3): 163–188. doi:10.1007/BF01470985 Aydan Ö, Ichikawa Y, Shimizu Y, Murata K (1991) An integrated system for the stability of rock slopes. The 5th int conf on computer methods and advances in geomechanics, Cairns, vol 1, pp 469–465 Aydan Ö, Hamada M, Konagai K (2006) An evaluation of strong ground motions and failures of natural and cut slopes induced by Kashmir earthquake of October 8, 2005. First European conference on earthquake engineering and seismology. Geneva, Switzerland, Paper Number:1352 Dunning SA, Mitchell WA, Rosser NJ, Petley DN (2007) The Hattian Bala rock avalanche and associated landslides triggered by the Kashmir Earthquake of 8 October 2005. Eng Geol 93:130–144. doi:10.1016/ j.enggeo.2007.07.003 Harvard (2005) Magnitude 7.6 PAKISTAN, Saturday, October 08, 2005 at 03:50:40 UTC. http://www. seismology.harvard.edu/ GSP-Geological Survey of Pakistan (2005) Tectonics, geology and seismicity of epicentral area. http://www. gsp.gov.pk/ GSP-Geological Survey of Pakistan (2006) Extended abstracts on International Conference on earthquake in Pakistan. http://www.gsp.gov.pk/ JSCE (2005) JSCE-AIJ Joint Investigation/Technical Support Team for Restoration and Reconstruction of the Affected Areas by the Pakistan Earthquake on Oct. 8, 2005. Japanese Society of Civil Engineers/ Architectural Institute of Japan, pp 76 Keefer DK (1984) Landslides caused by earthquakes. Geol Soc Amer Bull 95(2):406–421 Okawa I (2005) Strong earthquake motion recordings during the Pakistan, 2005/10/8, Earthquake. http:// www.bri.go.jp/ Sahoo RN, Reddy DV, Sukhija BS (2007) Evidence of liquefaction near Baramulla (Jammu and Kashmir, India) due to the 2005 Kashmir earthquake. Curr Sci 92(3):293–295 Sato HP, Hasegawa H, Fujiwara S, Tobita M, Koarai M, Une H, Iwahashi J (2007) Interpretation of landslide distribution triggered by the 2005 Northern Pakistan earthquake using SPOT 5 imagery. Landslides 4:113– 122. doi:10.1007/s10346-006-0069-5 USGS Global Seismographic Network (2005) Magnitude 7.6-Pakistan 2005. October 8 03:50:40 UTC. http://neic.usgs.gov/