English
Italiano
Forgot your password?
New user?
Document Actions
ijege-13_bs-hermanns-et-alii-1.pdf
001.png)
lanches or large rock falls, which directly affected
settlements, but also caused either a displacement
wave when impacting a water body with often fatal
consequences or damming of narrow valleys with a
high loss of property (B
catastrophic failure we follow here the definition
given by h
fragmentation of the rock mass and could impact with
high velocity an area larger than that of a rockfall with
typical shadow angles of ca. 28-34° (e.g, e
ize the source and invasion areas and its likelihood for
small scale rock slope failures (rockfall susceptibility
map, detailed hazard maps) (e.g h
gian Water and Energy Directorate (NVE) carries out
systematic geologic mapping of potentially unstable
rock slopes that might fail catastrophically (h
three of the 17 relevant counties of Norway has revealed
more than 300 sites of potential future rock slope fail-
ures. This number necessitates a systematic mapping
approach that focuses on the relevant geological data
that might undergo catastrophic failure in future and
can cause loss of life. The system is scenario-based
as the intensity and rate of displacement, as well as
the geological structures activated by the sliding rock
mass vary significantly on the slopes. In addition, for
each scenario the secondary effects, such as genera-
tion of displacement waves or landslide damming of
valleys with the potential of later outburst floods, are
evaluated. The hazard analysis is based on two types
of criteria: 1) Structural site investigations including
analysis of the development of a back-scarp, lateral
boundaries and basal sliding surface. This includes
a kinematic analysis for sliding and toppling based
on slope orientation, persistence of main structures
and morphologic expressions of the sliding surface.
2) Analysis of slope activity primarily based on slide
velocity, change of deformation rates, observation of
rockfall activity, and historic or prehistoric events.
The analysis of consequences focuses on the potential
fatalities to the rock slide scenarios and secondary ef-
fects. Based on the hazard and consequence analysis
each scenario is classified in a risk matrix into cat-
egory low, medium or high risk.
002.png)
ing spontaneous (seismically-triggered rockslides) for
that a minimum magnitude of M 6 was established by
k
with the premise of an acceleration phase prior to col-
lapse we could capture the vast amount of rock slope
failures in Norway (F
are not captured in this observational period, highlight-
ing that this classification system cannot be used as a
risk management tool alone, but has to be used espe-
cially in areas with higher levels of seismic activity in
connection with seismic hazard maps as presented by
s
features expressing slope deformations and changes
within the slope that can give a hint of the stability
state of the slope as well as on the area that will be
impacted by the direct impact of the rockslide or a re-
lated secondary effect. The risk classification focuses
on the potential loss of life only.
is dominated by crystalline rock and does not present
large rock slopes with weak sedimentary rocks such
as the Alps, Apennine or the Rocky Mountains. Other
observations would have to be included in mountain
terrains with thick weakly consolidated sedimentary
or volcanic rocks. The classification system might also
be applied in other areas in the world, but needs to be
adapted to local geologic, geographic and climatic con-
ditions. The classification system is flexible for such
adaptations by giving the possibility to exclude some of
the criteria used in Norway and to add new ones. We es-
pecially underline that today there is insufficient quan-
tity of information on geological occurrences to support
the prediction of large rock slope failures on geological
conditions alone and that instrumental monitoring is the
appropriate tool for monitoring changes in rock slope.
In the latter case, deformation varies between different
compartments of an unstable rock slope (also called
parts, blocks or similar). This difference in deforma-
requires prioritization of follow-up activities, such as
periodic or permanent monitoring, early-warning sys-
tems, and other mitigation measures. A first guideline on
the mapping approach and a hazard and risk classifica-
tion is given in a geological report (h
as well as hazard and risk classification will follow in
Norway in the upcoming years these guidelines until a
large number of sites are classified and related geologi-
cal data and data on potential consequences stored in a
related database (B
the likelihood of failure cannot be given quantitatively
in hundreds or thousands of years with today's scien-
tific knowledge, the risk analysis is built on a qualitative
hazard analysis and a quantitative consequence analysis.
The goal is to assemble enough data on historic and pre-
historic rock slope failures in Norway that will allow for
a calibration of the qualitative hazard analyses.
and international experience on large rock slope fail-
ures and a group of 18 Norwegian and 5 international
experts had participated in the discussion preceding
this classification system (see summary in h
systems that focus on long term slope stability of large
rock slopes have been taken as guide (h
stable rock slopes do not fail under aseismic conditions
without any pre-failure slope deformation (h
on aseismic failures because the timing of earthquakes
cannot be predicted up to now, making early-warning
of earthquake-triggered rockslides impossible. We have
to highlight here that in Norway seismicity rates over
the 20th century suggest that the region typically re-
veals one magnitude (M) 5 earthquake every 10 years
and one M 7 earthquake every 1100 years (B
regional differences with most of the seismic activity
concentrated in small areas located in the near-shore or
off shore area (s
003.png)
scenario has to be carried out independently.
analysis increases stepwise following the principles
outlined in Fig. 1. The term assessment is here used to
describe a semi-quantitative evaluation carried out by
the mapping geologist, while the term "analysis" is used
here for more thorough, quantitative investigations.
(Fig. 2). They can be arranged into two main groups:
1. the structural development of the unstable rock
slope; 2. displacement rates and other signs of activ-
ity. For each criterion (κ
conditions gives the total score, which is called hazard
geological records indicating that some unstable slopes
collapsed repeatedly while others slopes failed in a
single event (e.g. h
that at some rock slopes parts of the rock mass can get
to a critical state at different moments in time. These
compartments may have different failure probabilities,
different consequences and pose therefore also different
levels of risk. One can define a scenario by a compart-
ment of the unstable rock slope, which might fail in a
single event and individually from other compartments.
An additional hint to define failure scenarios is the
analyses of historic and prehistoric failures along slope
sections built by the same lithologies and controlled by
the same structures.
• Different deformation rates
• Varying structural conditions
• Internal scarps, cracks and depression which dissect
004.png)
therefore on the hazard score. In order to compute the
entire range of possible outcomes, the criteria are or-
ganized in a decision tree. Each criterion, κ
score, ρpath, and its probability, φpath, can be calcu-
lated using Equations (1) and (3), respectively:
probabilities for individual paths may be very low.
However, several paths may lead to the same path
hazard score, ρpath. Therefore, the total probability of
having a given hazard score corresponds to the sum of
an unstable rock slope to fail increases with ρ.
of the conditions (χ
score, ρ, is obtained by summing all the scores (νij)
multiplied by the conditions probabilities (Equation 2):
005.png)
ever this information from the geological past does
not necessarily indicate anything on the performance
of the slope in future and continuous fatigue of rock
in the past 10,000 years could have led to a critical
stability today. Similarly, rock slopes that failed cata-
strophically could define the very high hazard class, if
the slope conditions in the period of months/years pri-
or to the catastrophic failure are used. Unfortunately,
there is generally not enough information available on
past catastrophic rock slope failures, in order to assess
their hazard score with satisfying reliability.
an unstable rock slope, the proposed technique with
the decision tree analysis gives a range of probable
hazard classes.
whether more detailed analyses are necessary. For
example, often during early site investigations, no in-
formation is available on the displacement rate of the
slide. Hence, this high level of uncertainty should be
expressed in the analyses. If the result of the analyses
indicates that there is a probability that the sites might
be defined to be of moderate or high risk, then more
investigations become necessary focusing on defining
the velocity. If also under the worst case assessment the
site remains a low risk object, no further investigations
are required.
property or the environment". We focus in our conse-
quence/risk analyses on loss of life only. "Risk is often
estimated by the product of probability of a phenom-
enon of a given magnitude times the consequences"
(p. 86). The risk, R, can be calculated using the widely
used risk equation (modified from l
ondary effects reaching the element at risk); PE =
that is downloadable at the same website as the report
(see h
ρpath and φpath and creates the sum of all φpath lead-
ing to the same ρpath. Using the scores presented in
Figure 2, the path hazard score, ρpath, ranges from 0
to 12 with steps of 0.25. The final outcome is a prob-
ability for each of these 49 different hazard scores,
φscore. The probability distribution of φscore allows
obtaining the minimum and maximum hazard scores,
ρ, using the chosen probabilities, p
the probability distribution, while the mean value is
computed using Equation 2.
using equal intervals (Fig. 3). Equal intervals are
preferred over expert-knowledge-based class limits,
because the latter are more controversial and need to
be supported by calibrations of past rock slope fail-
ures. For example, one could define the very low haz-
ard class by slopes that move since more than 10,000
years and that did not yet fail catastrophically; hence
006.png)
nerability of the element at risk to the landslide event
(degree of loss from 0% to 100%); E = element at risk
(i.e. exposed population).
classification for unstable rock slopes in Norway, es-
pecially the probability of failure, P
unstable rock slopes within the timeframe of hundreds
to thousands years. For this hazard and risk classifica-
tion system the hazard score is used as a qualitative
measure of P
and medium to high risk objects that require more de-
tailed risk analyses. Therefore, a worst case scenario
is assumed for the preliminary risk assessment and P
affected area.
lanche or displacement wave (P
area (P
sequence analysis becomes necessary (Fig. 1). This
includes a more detailed quantification of the param-
eters in Equation 4. Detailed run-out modeling (and
displacement wave assessment if relevant) allows
the determination of P
school etc.) and can be determined roughly at a na-
tional level. Different vulnerabilities can be defined,
depending if a building is hit directly by a rock ava-
lanche and loss of life is nearly certain (V=1) or if it
is hit by a displacement wave that have an assumed
survival rate of 30% (V=0.7) (B
and its secondary effects. Areas frequently visited by
tourists and all infrastructure with persons in transit
are assessed in the same manner as buildings (e.g.,
B
In contrast to the direct impact of a rockslide on a build-
ing or the impact of a rockslide-triggered displacement
wave on a building, people affected by upstream and
downstream flooding related to landslide damming and
subsequent dam breaching can be evacuated from the
building. Hence this secondary effect is treated as a
flood hazard and in these cases the evaluation of haz-
ard and risk related to dam formation and dam failure
should be included as outlined in d
people which might lose their life in a potential event.
(Fig. 3). Isorisk lines are often used in a risk matrix
to distinguish between acceptable, tolerable and unac-
ceptable risks as proposed for example for landslides
and rock falls from natural slopes in Hong Kong (g
classification system, since the hazard score is only
a qualitative measure of the probability of failure and
the classification focuses on rock slope failures pre-
ceded by an acceleration phase only, thus excluding
earthquake-triggered rock slope failures. The risk can
therefore not be expressed in terms of number of casu-
alties per year, and this is not a risk management tool
in its own but a support for risk management
ing monitoring, further field investigations, and/or
possible mitigation measures. For that reason the risk
matrix is divided into three risk classes: low (green),
medium (yellow) and high (red). The limit between
the low and medium risk classes is set along the di-
agonal going from the high hazard class with very low
consequences (0.1 to 1 casualties) down to very low
hazard class with high consequences (100 to 1000 cas-
007.png)
order to decide on follow-up investigations and miti-
gation measures. It will also help municipalities and
other stakeholders as a basis for land use planning and
contingency planning.
classes will be presented in due time by NVE. This
will include implications related to land use planning,
monitoring and early-warning, contingency planning
and structural measures. All decisions on mitigation
measures will be based on cost-benefit reasoning that
will be explained in more detail in the NVE document.
remote sensing data (air photos, satellite data) every
10 to 20 years. For medium risk objects, periodic
monitoring is recommended and the techniques used
and the measurement intervals applied will depend
on geological conditions on the site, applicability of
the various methods under cost-benefit reasoning.
For high-risk objects mitigation measures are recom-
mended that will often be coupled monitoring and
early warning techniques. This has to be discussed
among risk owners and geoscientists.
rock slope failures occurred repeatedly in the past, of-
ten accompanied by secondary effects such as displace-
ment waves. Therefore, in contrast to other mountain
belts in the world, these rock slope failures resulted in
disasters with a high death toll far from the source area
of the rock slope failure. As such events will also oc-
cur in the future, systematic mapping of rock slopes has
been started in the first decade of the 21
This high number necessitated a quantitative classifica-
tion system based on hazard and risk related to the po-
tential failures that should help deciding on follow-up
activities. This system was elaborated in a large effort
combining national and international experts from vari-
ous disciplines in earth sciences. During the elabora-
tion of this system it became obvious that today there
fall into the low risk class. Those sites are either con-
sidered to have low consequences and further follow
up is not economically sustainable, or the site would
require dramatic changes in the geological conditions
prior to failure. Such changes could be captured with
a scanning of geological conditions by means of field
visit or remote sensing data interpretation every 10 to
20 years. Medium risk sites are expected to be less
common in Norway. However, potential consequenc-
es are higher or the probability of failure is higher so
that a low-level follow up is recommended to reduce
the risk level. The limit between the medium and high
risk classes is not precisely defined and is shown as a
yellow to red gradient. In this transition zone between
medium and high risk, in general further site-specific
geological criteria are needed to be studied in order
to have a good enough understanding for a final clas-
sification. These sites will generally require additional
expert judgement that will be used to classify the risk .
quence analysis. As both factors have uncertainties,
the minimum and maximum values for hazard and
consequences can also be plotted in addition to the
mean value (Fig 3).
tion and on the decision on follow-up activities. An
unstable rock slope might for example be classified
as low risk based on the most likely hazard class, but
there might be a certain probability that it ends up as
a medium risk. If this probability exceeds 5%, more
site investigations should be considered in order to
reduce the uncertainties on the assessment of condi-
tions for the different criteria. If this is not feasible, the
unstable rock slope might be classified with the higher
risk class. The 5% and 95% percentiles of the hazard
score are therefore also shown in the risk matrix (Fig.
3). Similarly, there is uncertainty related to the con-
sequences and more detailed consequence analyses
could be considered in order to reduce the uncertainty.
The decision on follow-up activities will be made af-
ter a thorough discussion of the uncertainties related
to both hazard and consequences.
008.png)
ed a high hazard for that slope (in h
sification system. Furthermore, the probability of the
Åknes rock slope in Norway was assessed earlier, and
independently of this system, and the results are com-
parable (B
cation system should be updated once more scientific
knowledge becomes available, and that more research
is necessary to better understand failure processes of
large rock slope failures through time. These efforts
will then hopefully allow replace the qualitative haz-
ard analysis with a quantitative hazard analysis.
proach by assuming that all people that might be hit by
a rock avalanche or a rockslide-triggered displacement
wave are likely to lose their lives. For potential high-
risk objects a more detailed analysis is carried out that
includes the probability of surviving the rockslide trig-
gered displacement wave. Both the qualitative hazard
analysis and the quantitative consequence analyses are
combined in a risk matrix for a risk analysis. Three dif-
ferent risk classes are defined. Low risk where no fur-
ther follow up is needed, medium risk where periodic
monitoring of the rock slope is suggested and high risk
that suggest for further follow up. Follow up for high
risk objects has to be discussed with the risk owners
and could be resettlement, periodic monitoring, con-
tinuous monitoring coupled with early warning or any
other mitigation measure that has to be decided after
cost-benefit reasoning. Often additional geological in-
formation to those summarized in this classification has
to be assembled for optimal monitoring practices and
a thorough slope stability analysis. This might include
subsurface information obtained from core logging,
geophysics and hydrological investigations.
ment tool in itself as it does not include seismically
triggered rock slope failures. It is thus a support tool
for risk management that will help to decide on follow
up (e.g. no follow up necessary, periodic monitoring
is recommended to reduce the risk, more studies are
required and/or risk mitigation measures should be
taken). This is also not a guideline for early warn-
ing practices as these are regulated in Norway by the
ing of large rock slope failures, and that more research
is needed and much can be learned from rock slope
failures that have been monitored in the years prior
to failure. Therefore, we qualitatively classified the
probability of failure in very high, high, moderate,
low and very low.
analysis of the development of a back-scarp, lateral
boundaries and basal sliding surface. This includes a
kinematic analysis that tests if rock sliding or toppling
is kinematically feasible with respect to the slope ori-
entation, the persistence of main structures and the
morphologic expression of the sliding surface. 2) The
analysis of the activity of the slope is primarily based
on the slide velocity, but also includes the change of
deformation rates (acceleration), observation of rock-
fall activity and historic or prehistoric events. For each
criterion several observations are possible to choose
from. Each observation is associated to a score and the
sum of all scores gives the total score for a scenario.
The weighting of these scores has changed from the
first proposal of the classification system (h
final version the historic and prehistoric events are
weighted much lower than in the first proposal. This
seemed necessary as the occurrence of a prehistoric
event alone should not raise a site by one hazard class
without any signs of present day activity. Further-
more, the displacement rates and morphological ex-
pressions/kinematic feasibility of failure are weighted
equally. This weighting should be revised once statis-
tically adequate information becomes available.
decision tree where probabilities to each observation
can be given. All possibilities in the decision tree are
computed and the individual probabilities giving the
same total score are summed. Basic statistics show the
minimum and maximum total scores of a scenario, as
well as the mean and modal value. The final output is
a cumulative frequency distribution divided into sev-
eral classes, which are interpreted as hazard classes.
been monitored for more than a decade (l
009.png)
thanked that initially have been in the discussion of
elaborating the classification system that later vol-
untarily dropped out as the discussion was very time
consuming. Also Carl Harbitz (Norwegian Geotechni-
cal Institute) and Hallvard Berg (NVE) contributed to
the classification system over a long period.
2010) and we thus do not include a discussion on trig-
gering mechanism here.
sible in 2010 that stared the discussion on the docu-
ment. Numerous scientist (Jordi Corominas, Olianne
955-964.
Sciences, Springer, Dordrecht, Netherland.
indudustrial partners). NORSAR and Norwegian Geotechnical Institute, Oslo, 187 pp.
010.png)
rocheux dans l'arc alpin. Programme Interreg IIC - "Falaises", Aosta, Italy: 132-154.
Mai 2010. Vienna, Austria. Geophysical Research Abstracts. 12: EGU2010-13657.
improved understanding. 459-465, Taylor & Francis Group, London.
The issue 16 volume 01 has been published