Seismically Induced Failure of Rock Slopes: Insights From Field Observations and Analytical Modeling
Seismically Induced Failure of Rock Slopes: Insights From Field Observations and Analytical Modeling
Seismically Induced Failure of Rock Slopes: Insights From Field Observations and Analytical Modeling
C. Massey
GNS Science, New Zealand
M.L. Olsen
Oregon State University, USA
M. McLaughlin
Montana Tech, USA
D. Keefer
University of Maine, USA
ABSTRACT: Seismically-induced rock slope failures have caused the deaths of tens-of-
thousands of people and economic losses in the billions over the last century. This paper pre-
sents insights on rock slope behavior in earthquakes both from field reconnaissance missions
and analytical modeling. In the first part of the paper, a detailed case study is presented for
the 2010–2011 Canterbury earthquakes, which triggered in thousands of landslides in the
populated Port Hills region of Christchurch resulting in both human and capital losses. Data
is presented on rock slope performance and the impact of co-seismic rockfall on buildings. In
the second part of this article discusses a straightforward, yet powerful, two-dimensional
single block framework to assess the co-seismic mode of failure (sliding, toppling, slumping)
and yield acceleration equations for rock blocks. Noteworthy is that complexly shaped blocks
can be evaluated knowing just their centers of mass and contact points with supporting
fractures.
1 INTRODUCTION
In May 2008, the Mw7.9 Wenchuan Earthquake shook a mountainous region of China
located approximately 80 km northwest of the city of Chengdu. The event triggered thousands
of rock slope failures that were collectively responsible for an estimated third of the 88,000
total fatalities (Fan et al. 2009) and for the emplacement of over 200 unstable landslide debris
dams that forced massive post-earthquake evacuations. Events such as this are not isolated.
While these types of landslides disproportionately impact the most densely populated moun-
tain regions of the world, other locations are by no means immune to rock slope hazards. The
1959 M7.5 Hebgen Lake, Montana earthquake triggered a rock slope failure that killed 28
people and rockslides from the 1989 Loma Prieta earthquake dammed several valleys in the
mesoseismal region. More recently, the 2010-2011 Christchurch Earthquake sequence trig-
gered thousands of landslides on the south island of New Zealand (Massey et al. 2014).
Worldwide, seismically induced rock slope failures have been responsible for approximately
30% of the most significant landslide catastrophes of the past century (USGS 2007). They are
491
among the most common and dangerous of all seismic hazards. Rock slope failures1 differ
fundamentally in two critical respects from landslides in unconsolidated, soil materials: (i)
rock slope stability is controlled principally by discontinuities in the rock mass rather than the
properties of the larger continuum, and (ii) owing to their potentially large volumes, high vel-
ocities, long travel distances, and impact forces, the consequences of rock slope failures can be
severe. These consequences routinely include burying of roadways and canals, impacts to
structures, and formation of landslide debris dams. While earthquake-induced soil landslides
have been well studied, there has been significantly less fundamental research on the more
common (Keefer 1984a) and often more significant problem of rock slope stability under seis-
mic conditions (Carro et al. 2003). The sparse treatment of this hazard in the literature is
partly due to the large displacements that rock masses often undergo at failure, which make it
challenging to estimate pre-earthquake properties and to ‘reconstruct’ and interpret triggering
processes and mechanisms in the field.
2 BACKGROUND
Figure 1. Rock slope failures from the Tohoku, Japan (a) and Pisco, Peru (b) earthquakes.
1. A wide range of terminologies exists to describe ‘rock slope failures’ including, ‘rock slides,’ ‘rock slumps,’
‘rock block slides,’ and ‘rock falls,’ ‘rock topples,’ and ‘rock spreads’. For brevity and consistency, the gen-
eric phase rock slope failures will be used throughout the paper. Rock slope failures include a range of land-
slide technical subcategorizations including rock slides, slumps and block slides, as defined by Keefer
(1984a). Consistent with Keefer (2002), rock refers to bedrock that is firm and intact prior to failure.
Although this paper focuses on initiation and immediate progression, it should be recognized that rock-slope
failures could mobilize into a long-run out rock avalanches, another landslide descriptive category.
492
post-event rescue efforts, but they also offer essential support to a region’s population and
thus play a critical role in both immediate and long-term recovery following an earthquake.
Figure 1b shows a rock slope failure triggered by the 2007 Pisco, Peru earthquake that closed
a major roadway for several days, entirely isolating several devastated nearby communities.
This same rock slope failure also buried a conveyance channel, cutting off the mountain water
supply to residents of nearby coastal cities. The earthquake reconnaissance literature contains
many other similar accounts. Rock slope failures also affect the natural environment by such
mechanisms as emplacement of landslide debris dams (Korup 2002). Because these landslide
dams are entirely uncontrolled (i.e., ‘non-engineered’), they are significant post-earthquake
hazards due to risks of their eventual breaching or overtopping. More pragmatically, they fur-
ther compound problems in regions already suffering direct losses from the earthquake itself.
493
Nevertheless, there currently exist few practical alternatives, and thus the pseudostatic
method has been used in many studies. For example, Chen et al. (2003) perform a pseudostatic
back-analysis of the Tao-Ling rockslide triggered by the Chi-Chi earthquake, while Mavrouli
et al. (2009) use a similar approach to assess the seismic stability of rock slopes in the Princi-
pality of Andorra. Sepulveda et al. (2005) use both a pseudostatic and Newmark-type sliding
block deformation analysis (Newmark 1965) to investigate the role of topographic amplifica-
tion in the triggering of rock slope failures in the Northridge earthquake. The discrete element
method, a tool that was largely limited to research applications, is beginning to find more
common use in practice for analyzing rock slope stability.
3.1 Introduction
The 2010–2011 Canterbury (New Zealand) earthquakes initiated thousands of landslides in
the populated Port Hills region of Christchurch.
These included rockfalls and debris avalanches, and to a lesser degree, soil slumps
(Figure 2). At least 200 homes were damaged by rockfalls, leading to the temporary
Figure 2. Location of rock slope failures in the Port Hills region of Christchurch (New Zealand).
Source areas are shown in orange and resulting runout paths are shown in yellow. The names shown near
the coast (e.g. “Quarry Road”) pertain to major rock slope failure sites. A northwest-facing cross section
through the “Richmond Hill” site is shown in Figure 3. With the exception of the LPCC station, who site
conditions closely mimic a rock outcrop, all seismic stations shown on the map are located on soil.
494
evacuation of hundreds of residents. The first significant event of the sequence, which
occurred on 4 September 2010, is known as the Mw 7.1 Darfield earthquake. This event was
followed about six months later by the devastating 22 February 2011, Mw 6.2 Christchurch
earthquake, which centered in the Port Hills. This second major event triggered widespread
landsliding in the region, including thousands of rock slope failures (Dellow et al. 2011). Peak
horizontal and vertical ground accelerations (PGA values) recorded in the Port Hills during
the February 2011 earthquake were in the range of 0.3–0.9 g but reached values as high as 1.4
g (horizontal) and 2.2 g (vertical) at one station (Heathcote Valley School station). These
large PGA values and the high vertical: horizontal acceleration ratios (> 1) are due, in part, to
the low epicentral distance in the Port Hills (epicentral < 10 km).
The intense, high-frequency nature of ground motion in the Port Hills is thought to have
been a key reason for the large number of rockfalls and landslides triggered in the region
(Cubrinovski et al. 2011). Several weaker earthquakes occurred on 16 April, 13 June, and 23
December 2011 triggering further landslides (Massey et al. 2014). The observations and sur-
veys made of rock slope performance during the Canterbury earthquake sequence are perhaps
the highest quality case studies available in the literature.
495
where the displacement of the mass is likely to have occurred along multiple discontinuities -
both preexisting and those formed by earthquake-induced movement - at depth within the
cliff. At the cliff crests, the crack frequency, persistence, and the displacement magnitudes
tend to increase toward the cliff face. In some locations, the cliff face had fallen away, indicat-
ing displacements greater than those measured. An eyewitness to a massive coastal rock cliff
during the 13 June 2011 earthquake observed that failure of the cliff started at the outside
edge, where displacement was more vertical and retrogressed into the slope away from the
cliff edge where movement became more horizontal. In the more massive debris avalanches,
failure of the basalt appears to have occurred due to a failure of the under-lying breccias
during the strong earthquake shaking. Failures in the rock mass forming the cliffs were also
noted to have occurred along discrete preexisting and sometimes clay-infilled discontinuities
and along material boundaries or through weaker epiclastic layers separating the lava-flow
sequences. In some of the cliffs, the observed failures comprised combinations of all of the
different failure modes discussed. Figure 3 depicts rock slope failure and damage to the Rich-
mond Hills cliff (see Figure 2 for location) during earthquake sequence. Such damage was typ-
ical for large rock slopes throughout the Port Hills region.
496
Figure 3. Northwest-facing cross section of Richmond Hills cliff (see Figure 2 for site location) showing
rock slope failures during earthquakes in the Canterbury sequence. The diagram in the upper middle
shows the measured accumulated displacement across the upper part of the cliff.
497
Figure 4. Examples of rockfall impacts on dwellings in the Port Hills of Christchurch. A and B: two
rock impacts on building affecting a total of three rooms causing significant damage, C: rock impacting a
corner of building causing exterior damages, D: damages caused by a through-going rock.
distribution of the simulated rock blocks at the point of impact was then used to compute
a likely distribution of energy on each building. The resulting back-analyzed rock impact
energies ranged from 0.15 to 1412 kJ, with a median impact energy of 49 kJ. The volume
of rocks ranged from 0.0015 to 14.4 m3, with a median value of 0.26 m3. Rockfall runout
distances into dwellings ranged from ~0.01 m (minor impact where the rock failed to
penetrate building) to complete dwelling penetrations (in the back and out the front) with
final resting positions more than 75 m beyond the initial impact with the dwelling. The
median runout distance was 2 m into a house.
The rock impact data for both wall and roof penetrations were combined to create
Figure 5, which shows runout distance into homes as a function of kinetic energy. These data
indicate a power law relationship between kinetic energy and the runout distance into and
through the dwelling. Overall, the data show that for an order of magnitude increase in
impact energy, a 5.5-fold increase in runout distance.
Independent regressions that consider wall and roof impacts separately do not differ in a
statistically significant manner, and therefore were combined into the single, best-fit regression
presented in Figure 5. It is worth noting that roof impacts are measured from their point of
entry and not the rear of the dwelling and therefore in some cases, affect portions of the dwell-
ing further down- slope than a comparable wall impact.
These data can be used to support quantitative risk assessment analysis to help quantify
the probability of a person being hit by a rock block (assume they are present in a home).
Examining Figure 5, the data suggest that rock blocks at impact kinetic energies of less than
2 kJ are unlikely to penetrate more than 0.1 m into a dwelling, which is similar to the typical
thickness of the timber-frame wooden buildings in the database. The data also suggests a
kinetic energy threshold of ~10 kJ, below which rocks tend to become lodged in the wall
system of the house (with penetration up to 50 cm) rather than fully enter a dwelling. It
worth noting that data points lying to the right of the best fit line (i.e., for rocks having
high-impact energies for relatively low penetrations) were most likely shifted by vegetation
effects along the runout path, which were not captured in our numerical simulations (Perret
et al. 2004).
498
Figure 5. Runout distance into dwellings as a function of kinetic energy. The three upper data points
on the runout distance diagram correspond to rock impacts that traveled through and beyond dwellings.
Correlation equations for the disaggregated data are presented above in text. Error bars in the Figure
shows +/- 1 standard deviation of impact kinetic energy.
4.1 Overview
Reconnaissance observations have played an essential role in inspiring and informing analytical
models. Observations from the Port Hills indicated that failure of rock blocks could initiate by
sliding, toppling, and slumping (sliding consists of translational motion; toppling consists of for-
ward, downslope rotational motion; and slumping consist of both translational and back-rota-
tional motion). These observations are supported by a database of earthquake-induced
landslide failures compiled by Keefer (1984a), which was further reinforced by others such as
Rodrıguez et al. (1999). To assess such failures, methods based on simplified, discrete, block
models such as a rectangular block on a plane or wedge blocks are commonly used to approxi-
mate the behavior of rock slopes and determine the factor of safety (Hoek and Bray, 1981;
Yagoda-Biran and Hatzor, 2013). These models assume rectangular blocks on a plane or sliding
prismatic three-dimensional (3D) wedge blocks. However, a significant limitation common to
rectangular and wedge-shaped blocks is that the assumed geometry and boundary conditions
restrict the potential failure modes to sliding or toppling. Thus, there is a need to expand the
geometry assumptions such that all possible failure modes are kinematically admissible.
To address this issue, a simple, yet powerful, two-dimensional (2D) single block analytic
framework was introduced that does not restrict the geometry to orthogonal fracture sets.
This formulation allows for the slumping failure mode to naturally occur and can identify a
new single block failure mode, confined toppling. Through the use of simple failure mode
charts, this framework can be applied easily by practitioners including engineers performing
499
pre- and post-failure field reconnaissance. In addition to the identification of new failure
modes, this formulation allows for the calculation of the corresponding pseudostatic acceler-
ations that lead to block motion.
The range of discrete rock block geometries found in rock slopes is the result of the intersec-
tions formed by fractures throughout the rock mass. In some cases, these fracture networks
create orthogonal junctions that form rectangular blocks. However, when the fractures inter-
sect each other at arbitrary angles, many more block geometries become possible. As a matter
of terminology, the fracture set closest to a horizontal orientation is termed the base fracture,
while the other fracture set is termed the back fracture. For example, Figure 6 shows a rock
slope in the Andes of southern Peru having two predominate fracture sets. This
Figure illustrates a slope where the base fracture set is dipping out of the slope (i.e., ‘daylight-
ing fracture’) while the back fracture set is oriented either side of vertical (not that the blocks
are not rectangular). The discrete blocks formed by these intersecting fracture sets at the slope
surface are highlighted. In addition to distinct block failures, these separate blocks can inter-
act with each other and create complex behavior (Aydan et al., 1989). As shown in the Port
Hills, New Zealand, the cascading failure of multi-block slopes during earthquakes can be
catastrophic. Often these slopes are evaluated with either simplified failure charts based on
rectangular-shaped blocks (e.g., Yagoda-Biran and Hatzor, 2013) or slope specific numerical
analyses (e.g., Pal et al. 2011).
Figure 6. Blocky slope with indications of rock blocks where sliding and slumping modes are suspected.
500
Figure 7. Discrete rock block geometry parameterized by forward and backward block angles, α1 and α3.
primary weight support to the block. Point 4 is not in contact with a fracture but is noted to
define the block geometry.
The edges of the block are not necessarily assumed to be perfectly planar. It is assumed that
the block makes contact with the fractures in only three contact locations. Three position vec-
tors, r1, r2, and r3, are identified as the vectors from the center of mass, C, to the three corners
in contact with the two fracture planes. Since the block is a parallelogram and assuming a
uniform density distribution throughout the block, r2 = −r3 and r1 = −r4. With these relation-
ships, the geometry of this 2D discrete rock block is entirely described by r1 and r3. Because
r1 and r3 completely describe the geometry of the block, then using equations, the entire
shape of a block is characterized by the fracture spacings and relative fracture angle. A feature
of writing the equations in this manner is that vector components can be used to describe a
unit block, thus allowing equations governing the mechanics of the system to be derived as
presented by Gibson (2016).
501
Figure 8. Mode of failure for rock blocks influenced by both base and back fractures. Alpha angles are
as defined in Figure 7. The diagonal line representing α1 = −α3 describes blocks that are rectangular in
shape. This line creates a ‘mirror geometry’ that describes blocks with shapes leaning forward (left of the
dashed line) and blocks leaning backward (right of the dashed line). The grey shapes depict block geom-
etry for combinations of α1 and α3. Chart was based on vector mechanics derivations detailed by Gibson
(2016).
Figure 9. Chart solution showing yield accelerations (contoured values) for all modes of failure for the
condition of φ = 40◦ and base fracture angle = 0◦ as a function of forward and back block angles (see
Figure 7 for definition of alpha angles). Chart solutions of other combinations of φ and base fracture
angle are presented in Gibson (2016). “Mirror geometry” is as defined in Figure 8.
• As α1 approaches and exceeds φ, support from the back fracture plane is required for static
stability. The increase in the normal force on the back fracture plane reduces the normal
force on the base fracture plane, and since shear resistance is based on this normal force,
the same trend occurs. The orientation of the shear resistance on the base fracture plane is
the most productive at resisting horizontal acceleration than that of the back fracture
502
plane; therefore, the transfer of shear resistance from the base to back fracture plane
reduces the overall horizontal resistance of the system.
• The lower yield acceleration of the slumping block keeps in step with that of a log-spiral
failure surface. The interesting connection between these previous results and those of the
slumping block is that they are all systems undergoing back rotations.
• For all failure modes, increasing the base angle reduces the magnitude (peak ground accel-
eration) of the earthquake that induces motion.
• For the toppling, sliding, and slumping failure modes, as the base angle increases and
exceeds the friction angle, the yield equations become negative which means that the spe-
cific geometry is not statically stable.
Seismically-induced rock slope failures are among the most common and dangerous of all seis-
mic hazards. For example, the 2010–2011 Canterbury earthquakes triggered thousands of
mass movements in the Port Hills, with a majority being in rock. These rock slope failures
resulted in both capital and human losses.
The most abundant mass movements with the highest risk to people and buildings were
rockfalls originating from steep cliffs. The majority of the larger volume failures occurred in
volcanic materials and most happened as a result of fracturing through the intact breccia
forming persistent defects. In some of the larger cliffs, the cliff-top deformation patterns and
displacement vectors, measured across cracks and from instruments, support relatively deep-
seated failure mechanisms, where the displacement of the mass is likely to have occurred along
multiple discontinuities - both preexisting and those formed by earthquake-induced movement
- at depth within the cliff. At the cliff crests, the crack frequency, persistence, and the displace-
ment magnitudes tend to increase toward the cliff face.
Over 200 homes (typically simple timber-frame structures) in the Port Hills were impacted
by co-seismic rockfall. Rockfall impacts on the dwellings were observed to follow a power law
relationship between kinetic energy and the runout distance into and through the dwelling.
The variability in observed consequences (rock runout and area impacted) due to rock impacts
on dwellings is primarily accounted for by back-analyzed impact kinetic energy. These data
provide a fundamental input for rockfall risk analyses and will better constrain future esti-
mates of loss from rockfall impacts (under both seismic and non-seismic conditions).
Reconnaissance observations have inspired and informed analytical models. A two-dimen-
sional single block framework was developed for the seismic evaluation of discrete rock
blocks. The framework is unique in that its geometry assumptions are less restrictive such that
a wide range of commonly observed failure modes, including sliding, toppling, and slumping,
are be kinematically admissible. Some of the critical inferences that can be drawn from the
formulation to determine pseudostatic failure modes include:
• the angle of the base fracture is not needed to determine the pseudostatic failure mode for
blocks on a frictional fracture plane.
• the scale of a block does not contribute to the determination of failure mode or yield accel-
eration on when the fracture’s shear resistance is described by a friction-only relationship.
However, this is not the case for blocks having fracture shear resistance that is best
described by friction and cohesion parameters, such as rocks types where fracture gouge
forms or chemical bonding occurs.
• the orientation of fracture sets relative to an open slope face, and the friction angle of the
fractures are the main determinants of the seismic mode of failure.
• Yield accelerations for rock failure modes show a non-linear relationship with their geom-
etry whereas the yield acceleration for sliding is constant. Between the sliding, toppling and
slumping failure modes, the sliding yield acceleration is always the greatest. Key inferences
drawn from the yield charts include:
503
• the pseudostatic modes can change because of slight changes in geometry or friction angle
whereas the yield accelerations are equal at the sliding/slumping, sliding/toppling, and slid-
ing/confined toppling transition points.
• the likelihood of a mode occurring is based on the possibility of a specific joint orientation,
spacing, and open slope face orientation. As different rock types might have different typ-
ical geometries of joint set formation, different types of rocks may have different dominant
modes of failure under seismic loading. For example, in rocks that typically show columnar
jointing, such as basalts the expected modes of failure will be sliding or toppling depending
on the relative spacing of those fractures. In rocks such as slates that have relatively small
fracture spacing and secondary joints at conjugate angles, slumping and sliding failure
would be expected.
ACKNOWLEDGEMENTS
Financial support for this research was provided by the U.S. National Science Foundation
under grants nos. 1156413 and 1439773.
REFERENCES
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
Carro, M., De Amicis, M., Luzi, L., Mazorati, S., 2003. The application of predictive modeling tech-
niques to landslides induced by earthquakes: the case study of the 26 September 1997 Umbria-Marche
earthquake (Italy), Engineering Geology, 69: 139–159
Chen., T., Lin., M., Hung., J. 2003. Pseudostatic analysis of Tsao-ling rockslide caused by Chi-Chi earth-
quake, Engineering Geology, 71: 31–47
Cubrinovski, M., Bradley, B., Wotherspoon, L., Green, R., Bray, J., Wood, C., Pender, M., Allen, J.,
Bradshaw, A., Rix, G., Taylor, M. (2011). Geotechnical aspects of the 22 February 2011 Christchurch
earthquake. Bull New Zealand Society for Earthquake Engineering, 44(4): 205–226
Dellow, G., Yetton, M., Massey, C., Archibald, G., Barrell, D.J.A., Bell, D., Bruce, Z. et al. (2011) Land-
slides caused by the 22 February 2011 Christchurch earthquake and management of landslide risk in
the immediate aftermath. Bull New Zealand Society for Earthquake Engineering, 44: 227–238
Eberhardt, E., 2008. Twenty-Ninth Canadian Geotechnical Colloquium: The role of advanced numerical
methods and geotechnical field measurements in understanding complex deep-seated rock slope failure
mechanisms. Canadian Geotechnical Journal, 45(4): 484–510.
Fan, X., T. Gorum, C.J.van, Westen, Q., Xu, Tang, C., Huang, R. 2009. Distribution of large landslides
and landslide dams triggered by the Wenchuan earthquake, Sichuan, China’, Geophysical Research
Abstracts, Vol. 11, EGU2009-2863.
Gibson, M., 2016. Seismic Rock Slope Failure Modes and Time-dependent Displacements Using Single
Block Methods. Ph.D. Thesis, University of Washington.
Gibson, M., Wartman, J., McLaughlin, M., Keefer, D., 2018. Pseudo-static failure modes and yield
accelerations in rock slopes. Int. Journal of Rock Mechanics and Mining Sciences, 102: 1–14
Grant, A., Wartman, J., Olsen, M., O’Banion, M., Motley, M., 2017. The impact of rockfalls on dwell-
ings during the 2011 Christchurch, New Zealand, earthquakes. Landslides. 15(1): 31–42
Harp, E. & Noble, M., 1993. An engineering rock classification to evaluate seismic rock-fall susceptibility
and its application to the Wasatch Front, Bulletin of the Association of Engineering Geologists 30(3):
293–319
Hewitt, K., Clague, J., Orwin, J., 2008. Legacies of catastrophic rock slope failures in mountain land-
scapes, Earth-Science Reviews 87: 1–38
Hoek, E. & Bray, J.W., 1981. Rock Slope Engineering, Revised 3rd edition. The Institution of Mining
and Metallurgy, London.
Hoek, E., 2006, Practical Rock Engineering, e-book (www.rocscience.com), last accessed January 2019.
Hungr O., Leroueil S, Picarelli L., 2013. The Varnes classification of landslide types, an update. Land-
slides, 11:167–194
Keefer, D., 1984a. Landslides caused by earthquakes, Geol. Soc. of America Bull. 95(4): 406–421
504
Keefer, D., 1984b. Rock Avalanches Caused by Earthquakes: Source Characteristics, Science, 223: 1288–
1290
Keefer, D., 1993. The susceptibility of rock slopes to earthquake-induced failure, Assoc. of Eng. Geo.
Bull., 30(3): 353–361
Keefer, D., 2002. Investigating landslides caused by earthquakes – a historical review, Surveys in Geo-
physics 23: 473–510
Keefer, D. & Larsen, M., 2007. Assessing Landslide Hazards, Science, 316(5828), 1136–1138
Korup, O., 2002. Recent research on landslide dams–a literature review with special attention to New
Zealand, Progress in Physical Geography 26(2): 206–235
Newmark, N.M., 1965. Effects of earthquakes on dams and embankments. Geotechnique 15(2): 139–160.
Massey C., McSaveney M., Yetton M., Heron D., Lukovic B., Bruce Z., 2012. Canterbury Earthquakes
2010–2011 Port Hills Slope Stability: pilot study for assessing life-safety risk from cliff collapse. In:
GNS Science Report 2012/57
Massey, C., Della Pasqua, F., Holden, C., Kaiser, A., Richards, L., Wartman, J., McSaveney, M., Archi-
bald, G., Yetton, M., Janku, L., 2017. Rock slope response to strong earthquake shaking, Landslides,
14(1),249–268
Massey, C.I., McSaveney, M.J., Taig, T., Richards, L., Litchfield, N.J., Rhoades, D.A., McVerry, G.H.,
Lukovic, B., Heron, D.W., Ries, W. and Van Dissen, R.J. 2014. Determining rockfall risk in Christ-
church using rockfalls triggered by the 2010–2011 Canterbury earthquake sequence. Earthquake Spec-
tra 30(1): 155–181.
Mavrouli, O., Corominas, J. and Wartman, J., 2009, The seismic effect on rockfall hazard: The Solá de
Santa Coloma, Natural Hazards and Earth System Science. 9(6) DOI: 10.5194/nhess-9-1763-2009
Pal, S., Kaynia, A., Bhasin, R., Paul, D., 2011. Earthquake stability analysis of rock slopes: a case study.
Rock Mech Rock Eng. 2011. 45(2): 205–215
Perret, S., Dolf, F., Kienholz, H., 2004. Rockfalls into forests: analysis and simulation of rockfall trajec-
tories—considerations with respect to mountainous forests in Switzerland. Landslides 1:123–130
Plafker, G. and Ericksen, G. E. 1978. in ‘Rockslides and Avalanches 1, Natural Phenomena’. B. Voight,
Ed., Elsevier, Amsterdam, 277–314
Rocscience, Inc., 2016. RocFall-computer program for risk analysis of falling rocks on steep slopes. Ver-
sion 5.0. Toronto.
Rodríguez, C., Bommer, J., and Chandlerb, R.0., 1999. Earthquake-induced landslides: 1980–1997, Soil
Dynamics and Earthquake Engineering, 18(5):325–346
Schuster, R., 1996. Landslides Investigation and Mitigation, A. K. Turner, R. L. Schuster, Eds., National
Research Council, Washington, DC.
Sepulveda, S., Murphy, W., Jibson, R. and Petley, D., 2005. Seismically induced rock slope failures
resulting from topographic amplification of strong ground motions: The case of Pacoima Canyon,
California, Engineering Geology 80: 336–348
Wasowski, J. and Del Gaudio, V., 2000. Evaluating seismically induced mass movement hazard in Cara-
manico Terme (Italy), Engineering Geology 58: 291–311
USGS, 2007, Catastrophic Landslides of the 20th Century. Worldwide, http://landslides.usgs.gov/learn
ing/majorls.php
Yagoda-Biran, G., Hatzor, Y., 2013. A new failure mode chart for toppling and sliding with consider-
ation of earthquake inertia force. Int. Journal of Rock Mechanics and Mining Sciences. 64:122–131.
505