Document Actions

ijege-13_bs-crosta-et-alii.pdf

background image
Italian Journal of Engineering Geology and Environment - Book Series (6) www.ijege.uniroma1.it © 2013 Sapienza Università
Editrice
3
DOI: 10.4408/IJEGE.2013-06.B-01
MONITORING AND MODELLING OF
ROCK SLIDES AND ROCK AVALANCHES
G
iovanni
B. CRoSTa
(*)
, S
ilvia
iMPoSiMaTo
(**)
& D
enniS
RoDDeMan
(**)
(*)
Università degli Studi di Milano Bicocca - Department of Earth and Environmental Sciences - Milano, Italy
(**)
FeaT - The Netherlands (www.feat.nl)
specific sectors, and as isolated phenomena or within
larger slope instabilities. Depending on the rock mass
properties, the type of instability, local slope geom-
etry and the triggering conditions these phenomena
can evolve into rock avalanches posing extraordinary
risks where morphological conditions and valuable el-
ements occur. Therefore understanding rockslide be-
haviour and possible evolution is an important step to
be accomplished for both hazard and risk assessment,
as well as for civil protection and emergency actions.
As stated above, evolution of these phenomena
is controlled by slope morphology and its changes,
and by lithological, structural, geomechanical, hy-
drological and meteo-climatic factors. These slope
instabilities evolved over a timescale of thousands
of years, characterized by phases of different activ-
ity and geomorphological evolution, including load-
ing and unloading cycles, toe erosion, incision by
running waters, creep-like movement or long resting
phases interrupted by periodical reactivations, which
in some cases can evolve in catastrophic collapses.
During such long time intervals the rock mass prop-
erties can change because of water seepage, weather-
ing, disturbance and dislodging, as well as progressive
damaging. The progressive accumulation of displace-
ment can cause a gradual change in fracture intensity,
opening, and weathering with direct consequences on
rock mass strength and permeability, and controlling
the sensitivity to external perturbations (e.g. rainfall,
snowmelt, seismic shaking). In particular, snow melt-
ABSTRACT
Rockslides can appear in a large variety of types
and with different characters which make difficult
their understanding and the prediction of their behav-
iour. This contribution highlights some of the most
important steps in rockslide monitoring and modeling
showing the relationships between them. Monitoring is
a fundamental step in the understanding of such phe-
nomena. The analysis of the possible types of failure
and collapse are extremely important especially when
a rapid acceleration, a successive transformation in a
rock avalanche and a long runout are foreseen. Some
examples of ground surface and subsurface rockslide
displacements are shown and analysed with respect
to the possible definition of thresholds. In case of a
sudden collapse and long runouts erosion and impact
against water reservoirs should be considered and this
can be performed through different approaches. Some
results of laboratory tests, two dimensional and three
dimensional rockslide-rock avalanche numerical mod-
els are presented and tested against some case studies,
together with the interaction with water reservoirs.
K
ey
words
: rockslide, monitoring, slow to fast movement,
collapse, warning thresholds, rock avalanche, modeling, ero-
sion, impulse wave
INTRODUCTION
Rockslides occur along valley flanks in different
positions, affecting them for their entire length or at
background image
G.B. CROSTA, S. IMPOSIMATO & D. RODDEMAN
4
International Conference on Vajont - 1963-2013 - Thoughts and analyses after 50 years since the catastrophic landslide Padua, Italy - 8-10 October 2013
and possible evolution of landslide activity gives the
opportunity to define threshold values for pre-failure,
failure and post-failure phases to be adopted in land-
slide monitoring, early warning and risk management.
Therefore, a sound understanding of the future
evolution of a rockslide and its proneness to cata-
strophic collapse requires the availability of monitor-
ing data. Monitoring techniques are undergoing an
extremely rapid evolution and are providing a huge
amount of information to be treated and analysed. Tra-
ditional point-like monitoring activities are progres-
sively replaced by modern more spatially distributed
approaches. Point-like data can result insufficient or
misleading in the interpretation, being representative
of local behaviour, and are sometimes hindered by lo-
gistic and technical difficulties. Spatially-distributed
data (e.g. laser scanning surveys, radar interferometry)
can be suitable for general understanding and for early
warning scopes, providing a way to by-pass network
implementation and maintenance problems. The same
can be said for subsurface ground monitoring due to
the availability of continuous and real-time measure-
ments of displacement, pore pressure or piezometric
head along vertical profiles which can be coupled to
advanced geophysical surveys.
Finally, spatially-distributed and almost continu-
ous monitoring time series allow a more complete
calibration of numerical models as well as the de-
velopment of more advanced or suitable mechanical
models and constitutive laws. This finally results in a
more advanced predictive capability in terms of defi-
nition of the failure surface geometry, unstable vol-
ume, expected reaction to external perturbation and in
supporting collapse and then runout modelling (i.e. by
providing volume, geometry, water content data).
From all these points of view, the Vajont rock-
slide provides an incredible field laboratory because
of the amount of collected information both before
and after the landslide, till to our times, the number of
developed models, the performed laboratory tests and
the availability of monitoring data. At the same time,
the terrible consequences in terms of loss of human
lives, together with the complexity of the phenom-
enon, in terms of time evolution, the fast runout and
the involvement of the water reservoir make of this
rockslide an extremely complex phenomenon also for
what concerns the development of methods for a more
quantitative hazard and risk assessment.
ing and heavy rainfall are the most common triggers of
annually recorded displacements in alpine rockslides.
Shear zones at the base of rockslides can evolve from
intensely fractured and blocky to very fine grained.
This will cause a progressive localization, decrease
in strength and in permeability, with consequences on
sensitivity to groundwater level changes and with re-
corded displacement for comparable triggering events.
As a consequence, rockslide state of activity can
change in time following a seasonal trend, with accel-
eration and deceleration behaviour depending both on
internal and external controlling factors.
Furthermore, rockslides can affect very large vol-
umes of rock mass which through time can undergo
differential degree of damaging. Then, rockslides can
show a mix of different behaviours within their bounda-
ries, being composed by sectors formed by very differ-
ent materials (i.e. lithology, loose coarse or fine debris,
dismembered or sound rock mass, faulted or folded
rocks) and characterized by different geometry (e.g.
steep rock scarps, flat or even back tilted head, steep
slide toe). This complexity makes intrinsically difficult
the interpretation of the slope movements because of
both the different type of instability and the relative dis-
turbance between the different types of movement (e.g.
superficial movements masking deeper ones).
These conditions differentiate most of the natural
slope instabilities from those along artificial excava-
tions. At the same time, they strongly affect the pos-
sible monitoring strategy to be applied at each site,
in terms of type and frequency of monitoring, as
well as the modelling approach. The fact that most
of the natural slopes can reflect a long time evolution
makes then more difficult to extrapolate the initial
slope conditions as well as the precise evolutionary
step at which a specific landslide is examined. The
temporal pattern of displacement, the state of activ-
ity (i.e. active, reactivated, suspended, intermittent)
and the evolutionary trend of a landslide (slow vs.
catastrophic, continuous vs. episodic) depend on the
geomorphological history, the extent of the monitor-
ing time window, the local morphology or degrees of
freedom (B
RoaDBenT
& Z
avoDni
, 1982). Long periods
of inactivity or minor movements (difficult to be rec-
ognized in absence of a monitoring network) as well
as availability or incompleteness of monitoring data
series make the understanding prohibitive. Indeed, the
analysis of long monitoring time series, of the trend
background image
MONITORING AND MODELLING OF ROCK SLIDES AND ROCK AVALANCHES
Italian Journal of Engineering Geology and Environment - Book Series (6) www.ijege.uniroma1.it © 2013 Sapienza Università
Editrice
5
perturbations, changes in material properties and level
of perturbation. C
RoSTa
& a
GliaRDi
(2003) propose
a methodology to obtain physics-based alert velocity
thresholds for landslides with complex kinematics and
sensitivity to external triggers. The method, based on
the non-linear Fukuzono-Voight relationship (v
oiGhT
,
1988) between acceleration and displacement rate,
uses monitoring data to define thresholds. Non-linear
fitting allows to derive values for model parameters
which can be considered typical of the state and prop-
erties of the physical system evolving towards failure.
Synthetic velocity vs time curves can be generated
providing a basis to quantitatively establish alert ve-
locity threshold values at different time intervals be-
fore the expected failure.
To illustrate the problems and possibilities asso-
ciated to rockslide monitoring and modelling we start
from a series of case studies. These case studies are not
represented into detail but are brought as examples of
behavior and of possible consequences in modelling.
ROCKSLIDE MONITORING
As above mentioned, rockslides can evolve in rapid
and long runout phenomena, along some specific shear
zones and with changes in rock mass properties. This
requires a well designed monitoring network both for
the movement characterization and for the setting up
and management of an EWS. The general design of a
monitoring network is not the focus of this manuscript,
whereas the idea is just to show some special features
observed as some specific case study.
Localised subsurface displacement has been ob-
served at different rockslide sites but its occurrence
in time is not generally available at least on very short
time scales. Indeed, the temporal distribution is ex-
tremely important to evaluate the delay time between
perturbation and displacement, the sensitivity to the
perturbations, the way of accumulating displacement,
the possibility to generate alert signals in quasi real
time. The Mt de La Saxe rockslide is a complex insta-
bility for which a relatively complete dataset is avail-
able. The time series of displacement through a specific
accelerating event is here presented.
Some of the boreholes drilled within the rockslide
and crossing the deep failure surface have been equipped
with some multi-parametric columns (DMS by CSG)
with measuring elements 1 m in length and with a high
resolution and a continuous recording capability.
EARLY WARNING THRESHOLDS
Early Warning Systems (EWS) requires the defi-
nition of thresholds values for parameters and/or in-
dicators for which a substantial change in system be-
haviour can be observed and a series of action can be
undertaken.
Even when good quality monitoring data (i.e.
continuous, spatially distributed, real time) are avail-
able, prediction or forecasting a slope failure and col-
lapse in time is a difficult task. This is especially true
when the involved rockslides are sensitive to external
perturbations, and physics-based thresholds for some
suitable monitored variables are extremely sensitive
to changes or are under continuous evolution. At the
same time, for long-lived rockslides, it is sometimes
difficult to define the instant of initial failure or change
in behaviour, together with the most significant points/
areas to be monitored, the successive rate of displace-
ment and the correlations with specific triggering
events. Furthermore, as above mentioned these land-
slides are often characterised by complex geometry
and geology making difficult a robust geomechanical
characterization and a reliable quantitative modelling.
Several empirical/phenomenological approaches
based on the “slope creep” theory (S
aiTo
& U
eZawa
,
1961; F
UkUZono
, 1985; v
oiGhT
, 1988; C
RoSTa
et alii,
1999; C
RoSTa
& a
GliaRDi
, 2002; R
oSe
& h
UnGR
,
2007) have been suggested to forecast slope failures
by the analysis of time series of monitoring data. In-
verse velocity methods, based on inverse velocity vs
time plots, are based on empirical observations, which
can agree with damage theory and can be a useful tool
for deciding about early warning management and
civil protection actions. The same methods can be
used under some specific conditions to verify the ef-
fectiveness of stabilizing actions (see Rose and Hungr,
2007) under constant perturbations. At the same time
they do not offer a clear understanding of the rockslide
evolution for more complex scenarios.
These models about accelerating trends have been
reconsidered (P
eTley
et alii, 2002; a
MiTRano
, 2005;
a
MiTRano
& h
elMSTeTTeR
, 2006; F
ailleTTaZ
et alii,
2010) trying to understand the processes which lead
to a rupture and its propagation through a material or
a slope up to a possible final collapse. Unfortunately,
when forecasting a possible collapse a series of un-
knowns and uncertainties remain relative to the an-
tecedent landslide or slope behaviour, sensitivity to
background image
G.B. CROSTA, S. IMPOSIMATO & D. RODDEMAN
6
International Conference on Vajont - 1963-2013 - Thoughts and analyses after 50 years since the catastrophic landslide Padua, Italy - 8-10 October 2013
- with progressive acceleration and decelerations,
associated to each of the two previous modes.
This figure shows displacements at some specific in-
tervals, as recorded by the instrumentation, which could
suggest different mechanisms for their occurrence: a
stick slip behavior along the failure surface, with stress
accumulating progressively, or a progressive increase in
water pressure along the closed shear zone till the ef-
fective stress limit is reached for shearing to occur. The
successive pressure dissipation during the opening or
dilation of the shear zone can cause an increase in the
effective stress and the arrest of displacement.
These displacements are reduced with respect to
the larger ground surface displacements (see map of
displacements, Fig. 2 and the time histories at some
points in Fig. 3, upper inset), measured by a LisaLab
Ellegi GB-InSAR and total station (lower inset in Fig.
Figure 1 shows an example of displacement
records measured at different depth along two differ-
ent boreholes during a two weeks interval. These plots
show some interesting trends in the occurrence and
distribution of deep seated displacements. In particular,
they seem to occur:
- in a continuous-like mode;
- in a step-like mode;
Fig. 1 - Plots of hourly recordeddisplacements measured
at different depth (seelegend) during a reactiva-
tion event (May –June 2012) by two multi-para-
metric DMS columns at the La Saxe rockslide
Fig. 2 - GB-InSAR maps of ground surface displacement
(May-July 2012) for the La Saxe rockslide
Fig. 3 - Recorded ground surface displacements meas-
ured: upper figure) by the LisaLAB – ELLEGIGB-
InSAR in the same period (May-June 2012,aver-
age time interval 4 hrs) as in Figure 1, andlower
figure) by a total station (daily displacement)at
different points during the 2009-2013 interval at-
the La Saxe rockslide
background image
MONITORING AND MODELLING OF ROCK SLIDES AND ROCK AVALANCHES
Italian Journal of Engineering Geology and Environment - Book Series (6) www.ijege.uniroma1.it © 2013 Sapienza Università
Editrice
7
ROCKSLIDE - ROCK AVALANCHE MO-
DELING
Modeling the stability and evolution of rockslides
implies the availability of geometrical, kinematical
and mechanical data. Slope stability is controlled
by many factor that can be introduced in one dimen-
sional up to three dimensional analyses. Groundwater
recharge and flow are the main controlling factors for
seasonal acceleration at many rockslide sites. Nev-
ertheless, the progressive accumulation of displace-
ment and deformation causes a progressive change
in properties, namely: the decrease in strength along
the failure surface and the failing mass, the decrease
in permeability along the shear zone, the increase in
permeability within the rockslide mass. Therefore, a
constant properties approach does not usually allow
for a complete modeling of these phenomena and of
their evolution in time.
The complete description of the possible mode-
ling approaches is out of the scope of this manuscript,
and in the following the possibility to model displace-
ment trend in time and the evolution of rockslide into
rock avalanche on erodible materials, and within wa-
ter reservoirs will be considered.
EFFECTS OF ERODIBLE BASAL LAYER
Rapid collapse of a rock slide can cause a trans-
formation into a rock-avalanche characterized by high
velocity and mobility, variable material properties and
entrainment capability. Entrainment (C
RoSTa
, 1992;
M
C
D
oUGall
, 2006; h
UnGR
& e
vanS
, 2004; C
hen
et
alii, 2006; C
RoSTa
et alii, 2008a, b, C
RoSTa
et alii,
2013) causes an increase in volume, and a change in
properties, function of the material fed to the flow (e.g.
partially or fully saturated soil, rock, ice), which in
turn influence the emplacement mechanisms and the
runout. Changes in properties (e.g. thickness, physical
mechanical properties and behaviour, frictional, cohe-
sive, permeability), and behaviour of the erodible sub-
strate material (e.g. collapsible, liquefiable, dilatant)
control the quantity of material entrained, dragged,
sheared, ploughed, bulldozed or unaffected by the
landslide motion (e.g. D
UFReSne
& D
avieS
, 2009).
The influence of these factors on the evolution of a
rock slide – rock avalanche can be studied by laborato-
ry scale experimental tests (Figs 5 and 6) and numerical
modelling. Results of some simple tests with motion of
a dry granular material over an erodible horizontal bed
3), which accumulate also the effects of the presence of
soil cover, local topographic gradient and its changes.
At the same time, ground surface displacement
maps allow for rockslide zoning according to the differ-
ent state of activity (see black lines in Fig. 2).
The different resolution of the instruments and the
acquisition time adopted for the different type of instru-
mentation can explain some of the discrepancies and the
possibility to recognize the same behavior at the ground
surface and at depth. This stress both the relevance of
continuous monitoring and the resolution of the adopted
instrumentation in the understanding of mechanisms.
Figure 3 shows also some other important aspects,
namely: the seasonal accelerations/reactivation, the
change in seasonal behavior between different years,
the presence of other episodes out of the more regular
seasonal behavior, the possibility to have exceptional
accelerations (2013), the rapid initial acceleration
and progressive deceleration, the continuous slow
displacement in cold and dry periods.
This figure also resumes some of the difficulties at
defining displacement rate thresholds on the basis of
available data, or in absence of continuously recorded
displacements. The 2013 acceleration shows an impor-
tant increase in ground surface displacement rate, with
respect to the previous year, with values larger than 100
mm d-1. These values, together with their rate of in-
crease, are comparable with those relative to historical
rockslides which underwent a catastrophic collapse. In
general, the displacement rate just before failure varies
within a large interval (Fig. 4), from a few mm per day
to some thousands of mm per day, and this wide vari-
ability explains the difficulty in thresholding by means
of comparison with historical records.
Fig. 4 Histogram of relative frequency of displacement
rate values close to the catastrophic collapse for a
sample of 53 historical events
background image
G.B. CROSTA, S. IMPOSIMATO & D. RODDEMAN
8
International Conference on Vajont - 1963-2013 - Thoughts and analyses after 50 years since the catastrophic landslide Padua, Italy - 8-10 October 2013
have been performed.These tests demonstrate the role
of slope angle, deposit thickness, material properties
on the entrainment and the deformation of the erodible
substrate as well the strong modification in the motion
characteristics in time and space. Backward aggrada-
tion and progradation of the deposit are controlled by
the change in slope and the geometrical constrains.
The flow profile changes with time showing the pro-
gressive involvement of dynamic and static properties
till a complete arrest.
2D and 3D numerical simulations can allow to
study in depth the main physical and geometrical
controlling factors. A series of fully 3D numerical
analyses has been performed by using a Finite Ele-
ment code (R
oDDeMan
, 2011; C
RoSTa
et alii, 2003,
2006, 2009) considering a Mohr-Coulomb substrate
material from almost purely frictional to purely co-
hesive. C
RoSTa
et alii (2008a,b, 2009, 2013) used
these together with different thicknesses for the sub-
strate material to verify the effects.
Simulations (see example in Fig. 7) show that the
thickness of the substrate is generally: inversely related
to runout especially for weaker than avalanche materi-
als; the deposit extent grows for decreasing material
layers; very weak substrates cause larger spreading
especially for thin layers; a radial deformation zone
is developed; velocity and thickness of the entrained
material increase with saturation; wedge like failures
or folding can be observed in the substrate material.
WATER RESERVOIRS AND IMPULSE
WAVES
Rockslides can originate within water reservoirs
(artificial and natural, lakes, fjords) and rock avalanche
Fig. 5 - Physical model of the erosion process triggered by a
coarse granular avalanche on a fine sand substrate
Fig. 6 - upper) Longitudinal profiles (time interval: 0.05
s) showing the evolution of the deposit in time;
lower) plot of the change in profile height with
time and description of the main features describ-
ing the granular avalanche (fine uniform sand)
motion and deposition
Fig. 7 - Fully 3D simulation of the avalanche/substrate in-
teraction. upper) deformation accumulated by the
substrate at the end of the avalanche motion; lower)
as above but with the avalanche deposit in place
background image
MONITORING AND MODELLING OF ROCK SLIDES AND ROCK AVALANCHES
Italian Journal of Engineering Geology and Environment - Book Series (6) www.ijege.uniroma1.it © 2013 Sapienza Università
Editrice
9
A series of historical landslide tsunami events oc-
curred as a consequence of subaerial landslides: the
1959 Pondesei (Italy) debris slump generating a 20 m
high impulse wave; the 1963 Vajont rockslide (Italy,
about 2900 casualties; v
iPaRelli
& M
eRla
, 1968) char-
acterised by a low water depth to landslide thickness ra-
tio; the 1958 Lituya Bay event (a
laSka
, M
illeR
, 1960,
predated by other events on 1853-1854, 1936, 1958),
the 1934 Tafjord event (Norway, B
likRa
et alii, 2005),
the Loen events (Norway), and the volcano collapses in
deep ocean waters (Hawaii and Canary islands, k
eaT
-
inG
& M
C
G
UiRe
, 2000; w
aRD
& D
ay
, 2003).
Because of the intrinsic risk associated to these
phenomena experimental tests on 2D and 3D wave
generation and propagation have been completed
(k
aMPhUiS
& B
oweRinG
, 1970; h
UBeR
, 1980; M
ülleR
,
1995; h
UBeR
& h
aGeR
, 1997; F
RiTZ
, 2002; P
aniZZo
et
alii, 2005; h
elleR
, 2007; h
elleR
& k
inneaR
, 2010;
S
ælevik
et alii, 2009; B
oSa
& P
eTTi
, 2011), using both
rigid blocks and deformable granular” masses. These
tests resulted in the development of various analyti-
cal and numerical methods which in general consider
subaerial movement and material properties in a sim-
plified way (h
aRBiTZ
, 1992; J
ianG
& l
e
B
lonD
, 1993;
G
Rilli
et al., 2002; G
Rilli
& w
aTTS
, 2005; Q
UeCeDo
et
alii, 2004; l
yneTT
& l
iU
, 2005).
Based on the previous experiences a series of nu-
merical simulations has been completed with the aim
to simulate a continuous deformable landslide mass
spreading along a slope, both under subaerial and
submerged conditions, interacting with a water body.
2D experimental tests performed by F
RiTZ
(2002) and
S
ælevik
et alii (2009) with deformable masses or rigid
blocks, respectively, have been considered (C
RoSTa
et
alii, 2013). In our FE analyses water is considered an
inviscid almost incompressible fluid. Figure 8 shows
the following sequence of steps for a semi-rigid rock-
slide mass: impact, progressive submergence of the
mass accompanied by the water crater formation and
successive backward collapse and closure, backward
flow over the slide material, and finally splash rising
along the sloping ground. Figure 9 shows the evolu-
tion of a granular avalanche elongating and thinning
along the subaerial slope, and shortening after the im-
pact with the development of a pronounced steep and
convex snout front.
The avalanche front is torn off during subaqueous
flow developing a transient backward tilted plume.
can reach them along their path impacting the immobile
mass of water at high speed. This can originate large
landslide generated impulse waves or landslide tsu-
nami, generally subdivided in three stages: initiation,
propagation and runup. As for the case of substrate en-
trainment the impulse wave is controlled by the land-
slide initial position (i.e. high on a slope, partial or com-
ple tesubmergence of the toe) and speed, the type of
material, the slope subaerial and subaqueous geometry,
the relative size of the landslide mass with respect to the
depth of water. Landslide impulse waves and tsunamis
are characterized by larger height in the near field, rapid
decay, high turbulence, flow separation and subsequent
reattachment, strong mixing of air and water.
Fig. 8 - 2D modelling of S
ælevik
S
et alii. (2009) experiment.
Velocity field (m s-1, to the left) and velocity vectors
(to the right) are shown. The impact velocity of a qua-
si-rigid 1 m long rockslide is 3.38 m s
-1
(Landslide
Froude number, Fr = 1.4). Impact and backward col-
lapsing impact crater are very well simulated
background image
G.B. CROSTA, S. IMPOSIMATO & D. RODDEMAN
10
International Conference on Vajont - 1963-2013 - Thoughts and analyses after 50 years since the catastrophic landslide Padua, Italy - 8-10 October 2013
1965). The simulation (Fig. 10) well fits the observed
phenomenon in terms of displacement, total duration,
impact against the opposite valley flank, thrusting and
deformation of the paleo-landslide material. The water
wave overestimate the maximum wave runup and this
is a primary effect of the fully 2D simulation, which
preserves the water volume within the reservoir.
Because of this limitation a fully 3D rockslide-
water reservoir simulation has been considered. For
this model about 800.000 elements have been used
to discretize the rockslide, the water reservoir and the
dam. The rockslide was modelled as a Mohr-Coulomb
material (ρ = 24 kN/m
3
; ν = 0.23; E = 10
10
Pa; ϕ =
23°, c = 100 kPa) and along the basal plane (ϕ = 7.5 to
5.7°; see Skempton, 1966; h
enDRon
& P
aTTon
, 1985;
T
ika
& h
UTChinSon
, 1999; 8°-10° down to almost 0°
for shear rates above 0.01 m s
-1
, F
eRRi
et alii, 2011),
Crater collapse takes place when the material has
reached the bottom of the water tank.Then a back flow
to the shore and a large part of the flow generates a pri-
mary wave which propagates to the right along the tank.
2D and 3D slope stability analyses and runout
simulations, the last ones without considering the pres-
ence of the water reservoir, have been performed by
C
RoSTa
et alii (2007). Modeling has been performed
using available geomechanical properties, whereas
model calibration and validation have been accom-
plished by comparing the final rockslide geometry
(profiles and 3D deposit geometry).
The next step in the analysis consisted in the 2D
simulation of the Vajont rockslide and of the consequent
wave performed (C
RoSTa
et alii, 2013) starting from
the available topographic (maps and cross sections)
and historical records of the events (R
oSSi
& S
eMenZa
,
Fig. 9 - 2D simulation of a 0.118 thick granular avalanche with impact velocity = 5.4 m s-1 (Fr = 3.15). Material position and
velocity vectors are represented at different time steps from the impact to the primary wave propagation
background image
MONITORING AND MODELLING OF ROCK SLIDES AND ROCK AVALANCHES
Italian Journal of Engineering Geology and Environment - Book Series (6) www.ijege.uniroma1.it © 2013 Sapienza Università
Editrice
11
and lower properties for the old landslide material lo-
cated on the opposite valley flank (ϕ = 13° to 7.5°, c =
100 kPa to 10 kPa reduced according to a plastic strain
softening model).
The pre-failure and post failure topography have
been obtained by available topographic maps (R
oSSi
& S
eMenZa
, 1965), and recent Lidar surveys (Regione
Friuli Venezia Giulia) which together with borehole
data (B
Roili
, 1967), geological cross sections (R
oSSi
&
S
eMenZa
, 1967) and field checks allowed to trace the
failure surface geometry.
Figure 11 presents some of the results in terms of
velocity vectors at two different instants. The average
duration of the rockslide since its release to the arrest
is about 51 s, a duration quite well comparable with
previous estimations (C
iaBaTTi
, 1964) and records. The
maximum water wave runup on the opposite valley
side is very well simulated as well as the final deposit
geometry and the water front splitting in an upstream
and downstream direction. Back washing along the
rockslide surface is well catched by the model (see
bottom Figure 11; v
iPaRelli
& M
eRla
, 1968).
The comparison between the streamlines geometry
computed by the FE code for the rockslide and their real
counterparts is also used for calibration. These are repre-
sented by the relative displacement vectors of well rec-
ognizable points and rocky outcrops mapped by R
oSSi
&
Fig. 10 - 2D model results of the Vajont rockslide with simulation of the water wave at different time steps till water becoming
still. Landslide mass subdivided in different colors to evidence internal deformation. Light cyan colored mass on the
right hand flank of the valley represents the paleo landslide material described by R
oSSi
& S
emenza
(1965)
Fig. 11 - Fully 3D simulation of the Vajont rockslide and
water wave top) and middle) velocity vectors
after 21 s. In the middle the brown and cyan
colors indicate the slide material and water wave
respectively;bottom) water wave (in blue) propa-
gation, partitioning and back wash over the rock-
slide body (in green) after 30 s
background image
G.B. CROSTA, S. IMPOSIMATO & D. RODDEMAN
12
International Conference on Vajont - 1963-2013 - Thoughts and analyses after 50 years since the catastrophic landslide Padua, Italy - 8-10 October 2013
S
eMenZa
(1967) before and after the collapse and further
support the prediction capability of the model and of the
numerical approach.
Finally, the presented approach solve some of the
assumptions commonly adopted in landslide-reservoir
interactions, as the landslide rheology or mechanical
behavior and its geometry during motion, and genera-
tion of the impulse wave (w
aRD
& D
ay
, 2011; B
oSa
&
P
eTTi
, 2011).
CONCLUSIONS
The present contribution does not want to cover in
depth the existing knowledge and literature concern-
ing rockslides and rock-avalanche but suggests and
discusses the relationships between monitoring and
modelling as a continuous process leading to more sat-
isfying predictions and hazard assessment.
Rockslides in alpine environments are complex
phenomena which can require a careful monitoring, in-
terpretation and analyses. Because of they characteris-
tics a possible evolution in the form of a rock avalanche
must be considered involving the possible erosion and
impact against water reservoirs. As a consequence,
the prediction capabilities of a FE code are presented
against some experimental and real case studies. These
case studies involve different slope geometry, thick-
ness of erodible substrate, water depth to rockslide/
avalanche thickness. Results of fully 2D and 3D simu-
lations support the prediction capabilities and push fur-
ther limits for numerical modeling of such phenomena.
ACKNOWLEDGEMENTS
The authors wish to thanks the Geological Survey
of Valle d’Aosta Region for providing the data of the
La Saxe rockslide. Giorgio Volpi, Mattia De Caro and
Alberto Villa are thanked for support in laboratory ex-
periments of granular flows. Paolo Frattini and Fed-
erico Agliardi are thanked for their contribution in the
analysis of monitoring data.
REFERENCES
a
MiTRano
D., G
RaSSo
J.R. & S
enFaUTe
G. (2005) - Seismic precursory patterns before a cliff collapse and critical point
phenomena. Geophysical Research Letters, 32 (8):1-5
a
MiTRano
D. & h
elMSTeTTeR
a. (2006)- Brittle creep, damage, and time to failure in rocks. Journal of Geophysical Research.
111: B11201.
B
likRa
l., l
onGva
o., h
aRBiTZ
C. & l
øvholT
F. (2005) - Quantification of rock-avalanche and tsunami hazard in Storfjorden,
Western Norway. In Landslides and Avalanches, ICFL 2005 Norway, ed. F. K. . L. J. Senneset K. : 57-64.
B
oSa
S & P
eTTi
M. (2011) - Shallow water numerical model of the wave generated by the Vajont landslide. Environmental
Modelling & Software, 26: 406e418
B
RoaDBenT
C.D, Z
avoDni
Z.M. (1982) - Influence of rock structure on stability. Stability in Surface Mining, 3. Soc. of Mining
Engineers.
B
Roili
L. (1967) - New knowledge on the geomorphology of the Vaiont Slide slip surfaces. Rock Mechanics & Eng. Geol., Jour.
Int. Soc. Rock Mechanics., V (1): 38-88.
C
aSaGli
n., C
aTani
F., D
el
v
enTiSeTTe
C. & l
UZi
G. (2010) - Monitoring, prediction, and early warning using ground-based
radar interferometry. Landslides, 7: 291-301
C
hen
h., C
RoSTa
G.B. & C.F. l
ee
(2006) - Erosional effects on runout of fast landslides, debris flows and avalanches: a
numerical investigation. Geotechnique, 56 (5): 305-322.
C
iaBaTTi
M. (1964) - La dinamica della frana del Vaiont. Giornale di Geologia, XXXII (I), 139-154.
C
RoSTa
G.B. (1992) - An example of unusual complex landslide: from a rockfall to a dry granular flow. Geol. Romana, 30:
175-184.
C
RoSTa
G.B., a
GliaRDi
F. & F
RaTTini
, P. (1999) - Convenzione di studio sulla Frana del Ruinon (Valfurva, Sondrio). Technical
report, Regione Lombardia, Dip. di Scienze Geologiche e Geotecnologie, Univ. di Milano-Bicocca, 197 pp. (in Italian).
C
RoSTa
G.B. &, a
GliaRDi
F. (2002) - How to obtain alert velocity thresholds for large rockslides. Physics and Chemistry of the
Earth, 27: 1557-1565.
C
RoSTa
G.B. & a
GliaRDi
F. (2003) - Failure forecast for large rock slides by surface displacement measurements. Canadian
Geotechnical Journal, 40 (1): 176-191.
C
RoSTa
G.B., i
MPoSiMaTo
S. & R
oDDeMan
D.G. (2006) - Continuum numerical modelling of flow-like landslides. Landslides
background image
MONITORING AND MODELLING OF ROCK SLIDES AND ROCK AVALANCHES
Italian Journal of Engineering Geology and Environment - Book Series (6) www.ijege.uniroma1.it © 2013 Sapienza Università
Editrice
13
from massive rock slope failure. In: e
vanS
S.G., S
CaRaSCia
M
UGnoZZa
G., S
TRoM
a. & h
eRMannS
R.l. (
eDS
). NATO Science
Series, Earth and Environmental Sciences, 49: 211-232.
C
RoSTa
G.B., F
RaTTini
P., i
MPoSiMaTo
S. & R
oDDeMan
D. (2007) - 2D and 3D numerical modeling of long runout landslides – the
Vajont case study. In: C
RoSTa
G.B. & F
RaTTini
P. (
eDS
). Landslides: from mapping to loss and risk estimation. IUSS Press,
Pavia, 15-24.
C
RoSTa
G.B., i
MPoSiMaTo
S. & R
oDDeMan
D.G. (2008a) - Approach to numerical modelling of long runout landslides. Hong
Kong, GCO, Dec. 2007, Proc Inter. Forum on Landslide Disaster Management, 20 pp.
C
RoSTa
G.B., i
MPoSiMaTo
S., & R
oDDeMan
, D.G. (2008b) - Numerical modelling of entrainment/deposition in rock and debris-
avalanches. Engineering Geology, 109 (1-2): 135-145.
C
RoSTa
G. B., i
MPoSiMaTo
S. & R
oDDeMan
D.G. (2009) - Numerical modeling of 2-D granular step collapse on erodible and
nonerodible surface. J. Geophys. Res., 114: F03020.
C
RoSTa
G.B., i
MPoSiMaTo
S. & R
oDDeMan
D.G. (2013) - Interaction of landslide mass and water resulting in impulse waves. In:
M
aRGoTTini
C., C
anUTi
P. & S
aSSa
k. (e
DS
). Landslide science and practice. Vol. 5: Complex Environment. ISBN 978-3-
642-31426-1,28 DOI 10.1007/978-3-642-31427-8, Springer Heidelberg New York Dordrecht London : 49-56
C
RoSTa
G.B., i
MPoSiMaTo
S., R
oDDeMan
D.G. & F
RaTTini
, P. (2013) - On controls of flow-like landslide evolution by an erodible
layer. In landslide science and practice. Vol. 3 - Spatial Analysis and Modelling. In: M
aRGoTTini
C., C
anUTi
P. & S
aSSa
k. (e
DS
),
ISBN 978-3-642-31426-1,28 DOI 10.1007/978-3-642-31427-8, Springer Heidelberg New York Dordrecht London: 263-270
D
UFReSne
a. & D
avieS
T.R. (2009) - Longitudinal ridges in mass movement deposits. Geomorphology, 105: 171-181.
F
ailleTTaZ
J., S
oRneTTe
D. & F
Unk
M. (2010) - Gravity-driven instabilities: interplay between state- and velocity
dependent frictional sliding and stress corrosion damage cracking. Journal of Geophysical Research., 115 : B03409,
doi:10.1029/2009JB006512
F
eRRi
F., D
i
T
oRo
G., h
iRoSe
T., h
an
R. n
oDa
h., S
hiMaMoTo
T., Q
UaReSiMin
M., D
e
RoSSi
n. (2011) - Low- to high- velocity
frictional properties of the clay-rich gouges from the slipping zone of the 1963 Vajoont slide (Northern Italy). Journal of
Geophysical research, 116: B0920. doi 10.1029/2011jb008338.
F
RiTZ
h.M., (2002) - Initial phase of landslide generated impulse waves. Thesis Versuchsanstalt für Wasserbau, Hydrologie und
Glaziologie, ETH Zürich, Swiss ETH No. 14'871. Swiss Federal Inst. Techn., Zürich, ISSN 0374-0056.
F
UkUZono
T. (1985) - A new method for predicting the failure time of a slope. In: Proc. 4
th
Inter. Conference and Field Workshop
on Landslides, Tokyo, Tokyo University Press: 145-150.
G
Rill
S.T., v
oGelMann
S. & w
aTTS
P., (2002) - Development of a 3D numerical wave tank for modeling tsunami generation by
underwater landslides. Eng. Anal. Bound. Elem., 26 (4): 301-313.
G
Rilli
S.T. & w
aTTS
P. (2005). - Tsunami generation by submarine mass failure Part I: modeling, experimental validation, and
sensitivity analysis. J. Waterw. Port Coast. Ocean Eng. 131 (6): 283-297.
h
aRBiTZ
C.B., (1992) - Model simulations of tsunamis generated by the Storegga slides. Mar. Geol. 105, 1–21.
h
elleR
v. & k
inneaR
R.D. (2010) - Discussion of "Experimental investigation of impact generated tsunami; related to a
potential rock slide, Western Norway” by G. Sælevik, A. Jensen, G. Pedersen". Coastal Engineering, 57: 773-777. Corps of
Engineers, Washington, DC
h
enDRon
a.J. & P
aTTon
F.D. (1985) - The Vaiont slide, a geotechnical analysis based on new geological observations of the
failure surface. Tech Rep GL-85–5, 2 vols. Department of the Army, US Corps of Engineers, Washington, DC
h
UBeR
a. & h
aGeR
w. h. (1997) - Forecasting impulse waves in reservoirs. Comminsion Internationale des Grands Barrages,
19
th
Congres des Grand Barrages, Florence, 1997: 993-1005.
h
UBeR
a. (1980) - Schwallwellen in seen als floge von felssturzen (reservoir impulse waves caused by rockfall). Technical Report
Mitteilung 47, Lab. Hydraulics, Hydrology and Glaciology, ETH.
h
UnGR
o & e
vanS
S. G. (2004) - Entrainment of debris in rock avalanches; an analysis of a long run-out mechanism. Geological
Society of America Bulletin, 116(9-10): 1240-1252.
J
ianG
l. & l
eBlonD
P.h. (1993) - Numerical modeling of an underwater Bingham plastic mudslide and the wave which it
generates. J. Geophys. Res. 98: 304-317.
k
aMPhUiS
J.w.& B
oweRinG
R.J. (1970) - Impulse waves generated by landslide. Proc. of 12
th
Coastal Engineering Conf.: 575-588.
k
eaTinG
B.h. & M
CGUiRe
w. J. (2000)- Island edifice failures and associated tsunami hazards. Pure Appl. Geophys. 157: 899-955.
background image
G.B. CROSTA, S. IMPOSIMATO & D. RODDEMAN
14
International Conference on Vajont - 1963-2013 - Thoughts and analyses after 50 years since the catastrophic landslide Padua, Italy - 8-10 October 2013
l
yneTT
P. & l
iU
P. l.-F. (2005) - A numerical study of the run-up generated by three-dimensional landslides. J. Geophys. Res.,
110: C03006, 16 pp., doi: 10.1029/2004JC002443.
M
C
D
oUGall
S. (2006)- A new continuum dynamic model for the analysis of extremely rapid landslide motion across complex 3D
terrain. PhD thesis, University of British Columbia, 253 pp.
M
illeR
D. (1960) - Giant waves in Lituya Bay Alaska. USGS Prof. Paper, 354-C: 51-83.
M
ülleR
D. (1995)- Auflaufen und uberschwappen von impulswellen an talsperren (run-up and overtopping of impulse waves at
dams). Technical Report Mitteilung 137, Lab. of Hydraulics, Hydrology and Glaciology, ETH.
P
aniZZo
a., D
e
G
iRolaMo
P., D
i
R
iSio
M., M
aiSTRi
a. & P
eTaCCia
a. (2005) - Great landslide events in Italian artificial reservoirs.
Natural Hazards and Earth System Sciences, 5: 733-740.
P
eTley
D.n., B
UlMeR
M.h. & M
URPhy
w. (2002) - Patterns of movement in rotational and translational landslides. Geology,
30: 719-722.
R
oDDeMan
D.G. (2011) - TOCHNOG user’s manual. FEAT, 255 pp, www.feat.nl/manuals/user/user.html.
R
oSe
n.D.& h
UnGR
o. (2007) - Forecasting potential rock slope failure in open pit mines using the inverse-velocity method.
Technical Note. Int. Joir. Rock Mech. & Mining Sci., 44: 308-320.
Q
UeCeDo
M. P
aSToR
M. & h
eRReRoS
M.i., (2004) - Numerical modelling of impulse wave generated by fast landslides. Int. J.
Numer. Methods Eng., 59: 1633-1656.
R
oSSi
D. & S
eMenZa
e. (1965) - Carte geologiche del versante settentrionale del M. Toc e zone limitrofe, prima e dopo il
fenomeno di scivolamento del 9 ottobre 1963, Scala 1:5000. Ist. Geologia Universit`a di Ferrara, 2 Maps.
S
ælevik
G., J
enSen
a. & P
eDeRSen
G. (2009) - Experimental investigation of impact generated tsunami; related to a potential
rock slide, Western Norway. Coastal Eng., 56 (9): 897-906.
S
aiTo
M. & U
eZawa
h. (1961) - Failure of soil due to creep. In: Proceedings of the 5
th
International Conference on Soil Mechanics
and Foundation Engineering. 1: 315-318.
S
keMPTon
a.w. (1966) – Bedding-plane slip, residual strength and the Vaiont Landslide. Géotechnique, XVI (1): 82-84.
T
ika
T
h
.e & h
UTChinSon
J.n. (1999) - Ring shear tests on soil from the Vaiont landslide slip surface. Geotechnique, 49 (1): 59-74.
v
iPaRelli
M. & M
eRla
G. (1968) - L’onda di piena seguita alla frana del Vajont. Università degli Studi di Napoli (in Italian).
v
oiGhT
B. (1988) - A method for prediction of volcanic eruption. Nature, 332: 125-130.
w
aRD
S.n. & D
ay
, S. (2003) - Ritter Island Volcano-lateral collapse and the tsunami of 1888. Geophys. J. Int., 154 : 891-90.
w
aRD
S.n. & S. D
ay
(2011) - The 1963 landslide and flood at Vaiont reservoir, Italy: A tsunami ball simulation. Italian Journal
of Geoscience, 130 (1): 16-26.
Statistics