Cover

Identification and Mitigation of Large

title: Landslide Risks in Europe : Advances in
Risk Assessment
Bonnard, Christophe; Forlati, F.; Scavia,
author:
C.
publisher: Taylor & Francis Routledge
isbn10 | asin: 9058095983
print isbn13: 9789058095985
ebook isbn13: 9780203024140
language: English
Landslides--Europe, Landslide hazard
subject
analysis.
publication date: 2004
lcc: QE599.E8I34 2004eb
ddc: 551.307094
Landslides--Europe, Landslide hazard
subject:
analysis.
Page i

IDENTIFICATION AND MITIGATION OF LARGE LANDSLIDE

RISKS IN EUROPE

IMIRILAND PROJECT
Page ii

This page intentionally left blank.

Page iii

Identification and Mitigation of Large

Landslide Risks in Europe
Advances in Risk Assessment

Ch.Bonnard, F.Forlati & C.Scavia, editors

European Commission

Fifth Framework Programme

IMIRILAND Project

A.A.BALKEMA PUBLISHERS LEIDEN/LONDON/NEW

YORK/PHILADELPHIA/SINGAPORE
Page iv

The IMIRILAND Project “Impact of large landslides in the mountain

environment: identification
and mitigation of risk” was financed by the European Commission
between 2001 and 2003 within
the Fifth Framework Programme “Environment Generic” (Contract
No EVG1-CT-2000-00035),
as well as by the Swiss Federal Office for Education and Science
(Contract OFES No 00.0361),
for which our thanks are expressed.

Management of the Project

Administrative Coordinator: Vincenzo COCCOLO—ARPA Piemonte,

Torino.
Scientific Coordinator: Prof. Claudio SCAVIA—Politecnico di Torino.
Editorial Board: G.Amatruda, Ch.Bonnard, M.Castelli, G.Moletta,
M.Morelli,
L.Paro, F.Piana and G.Susella.

ACKNOWLEDGEMENTS

Sincere acknowledgements have to be expressed to all the persons

who significantly contributed
to the IMIRILAND Project as authors, as well as to those whose
names do not appear as authors
in this book:

Technical assistants: L.Beccani, E.Fusetti, L.Mallen, H.Sarri,

F.Tamberlani (ARPA Piemonte);
A.Allodi, S.Marello (Politecnico di Torino).

Research assistants: J.Moreno Lapiedra, R.Rojas Fuentes (Ecole

Polytechnique Fédérale de
Lausanne); R.Russo, A.Stéphan (CETE Lyon).
Special thanks are also expressed to Prof. Oldrich Hungr and to Dr.
Pedro Basabe who contributed
to chapter 10, to K.Barone who mastered all the difficulties of the
final editing, as well as to
L.Martinenghi Stuckrath who revised the English form of the texts.
The administrative staff in charge
of the coordination at ARPA Piemonte and Politecnico di Torino also
deserves a vote of thanks.

We are equally grateful to all the communal authorities and to their

consultants, in particular
those of Ceppo Morelli, Rosone, Séchilienne and Sedrun, who
provided interesting information
for the development of this research.

Finally, the editors as well as the participants to the IMIRILAND

Project are deeply indebted to
Marta Castelli of Politecnico di Torino and Noemi Giordano of ARPA
Piemonte for their fundamental part in the success of the project.
Marta, by her continuous involvement in the scientific
coordination tasks as well as in the planning and management of the
different phases of the
research and the diffusion of its results, and Noemi, by her careful
and precise implication in the
administrative aspects of the project that have to be assumed in
order to ensure its completion.
Noemi also fulfilled the difficult task of translating some chapters.

Copyright © 2004 Taylor & Francis Group plc, London, UK

All rights reserved. No part of this publication or the information

contained herein may be
reproduced, stored in a retrieval system, or transmitted in any form
or by any means, electronic,
mechanical, by photocopying, recording or otherwise, without written
prior permission from the
publisher.

Although all care is taken to ensure the integrity and quality of this
publication and the
information herein, no responsibility is assumed by the publishers
nor the author for any damage
to property or persons as a result of operation or use of this
publication and/or the information
contained herein.

Published by: A.A.Balkema Publishers, a member of Taylor &

Francis Group plc www.balkema.nl and www.tandf.co.uk

This edition published in the Taylor & Francis e-Library, 2005.

To purchase your own copy of this or any of Taylor & Francis or

Routledge’s collection of thousands of eBooks please go to
www.eBookstore.tandf.co.uk.

ISBN 0-203-02414-1 Master e-book ISBN

ISBN 90 5809 598 3 (Print Edition)

Page v

The following institutions participated in the IMIRILAND Project:

AGENZIA REGIONALE PER LA PROTEZIONE

AMBIENTALE DEL PIEMONTE (ARPA
PIEMONTE)/ REGIONAL AGENCY FOR
ENVIRONMENTAL PROTECTION OF
PIEMONTE (Turin, Italy)

SETTORE STUDI E RICERCHE GEOLOGICHE,

SISTEMA INFORMATIVO PREVENZIONE RISCHI

Arpa is a leading centre of studies and applied

research in the field of natural risk. Its aim is the
development of methodologies and tools to assess,
manage and minimise the geological risk. It has a wide
experience in data collection, organisation and
elaboration with GIS techniques. It produces and
disseminates scientific and technical knowledge,
addressed to public and private end-users. Moreover it
coordinates national and international projects.

POLITECNICO DI TORINO/TECHNICAL
UNIVERSITY OF TURIN (Italy)

DIPARTIMENTO DI INGEGNERIA STRUTTURALE E

GEOTECNICA

The activities of the Department follow three main

directions: Structural Mechanics, Structural Design
and Geotechnics. Within each of these areas, focused
research is conducted, including fracture mechanics,
earthquake engineering and rock mechanics.
Theoretical, experimental and numerical aspects are
addressed in order to determine solutions of problems
related to structural and geotechnical phenomena
affecting, for example, the statics and dynamics of
bridges, large containers, buildings of strategic
importance and historical monuments. In particular,
the research in slope stability problems is dedicated to
the geomechanical modelling of rock masses and the
application of numerical methods to the definition of
hazard scenarios.

ECOLE POLYTECHNIQUE FÉDÉRALE DE

LAUSANNE (EPFL)/SWISS FEDERAL
INSTITUTE OF TECHNOLOGY (Lausanne,
Switzerland)

DEPARTEMENT ENVIRONNEMENT NATUREL,

ARCHITECTURAL ET CONSTRUIT (ENAC)—
LABORATOIRE DE MECANIQUE DES SOLS (LMS)

LMS has developed several metho dologies for the

hazard mapping and modelling of large zones affected
by landslides. Furthermore, it has developed a method
for the prediction of landslide behaviour based on
neural network analysis. LMS also works on coupled
thermo-hydro-mechanical modelling of saturated and
unsaturated soils, including constitutive and
thermodynamic aspects, as well as on geothermal
structures and heat exchange problems. Finally it deals
with soil reinforcement by cement injection and heat
effect.
Page vi

CONSIGLIO NAZIONALE DELLE RICERCHE

(CNR)/ ITALIAN NATIONAL RESEARCH
COUNCIL (Italy)

ISTITUTO DI GEOSCIENZE E GEORISORSE (IGG)

(SEZIONE TORINO)

CNR is involved in basic geological, structural,

petrographical and mineralogical researches. A relevant
activity of field survey for the 1:50000 scale National
Geological Map of Italy, represents the basis to develop
approaches for the analysis and mitigation of geologic
risk This topic has been worked out in the last years in co-
operation with some Italian Institutions.

TECHNISCHE UNIVERSITÄT WIEN

(TUW)/VIENNA UNIVERSITY OF TECHNOLOGY
(Austria)

INSTITUT FÜR INGENIEURGEOLOGIE

Institute for Engineering Geology is involved in research

activities concerning problems of slope stability and mass
movements, site exploration and engineering geological
mapping, numerical modelling, tunnelling (rock and soil
mechanics, hydrogeological and grouting aspects),
technical petrography (esp. for conservation of historical
monuments).

UNIVERSITAT POLITÈCNICA DE CATALUNYA

(UPC)/ TECHNICAL UNIVERSITY OF
CATALONIA (Barcelona, Spain)

DEPARTAMENT D’ENGINYERIA DEL TERRENY,

CARTOGRÀFICA I GEOFÍSICA
This UPC Department forms a leading centre of research
in the field of Soil and Rock Mechanics, in landslide
hazards and landslide modelling, in the numerical
analysis of coupled thermo-hydro-mechanical problems,
and in laboratory and fieldwork. The Department has
been considered as a Centre of Excellence for Research
within the Catalonian Research Plan framework.

LABORATOIRE CENTRAL DES PONTS ET

CHAUSSÉES/ CENTRAL LABORATORY OF
BRIDGES AND HIGHWAYS (Paris, France)

CENTRE D’ETUDE TECHNIQUES DE L’EQUIPEMENT

DE LYON (CETE)

LCPC is a State research organisation working for the

State and the local authorities in connection with
professionals involved in civil engineering, transport,
urban engineering and environment. One of the original
characteristics of LCPC resides in its role as technical
network co-ordinator for infrastructure, as far as natural
hazards are concerned. The LCPC is relying on the
regional CETE’s to carry out the main project studies and
some field research.
Page vii

List of contact persons and participants in the IMIRILAND Project:

Arpa Piemonte—(ARPA)

Settore Studi e Ricerche Geologiche—Sistema Informativo

Prevenzione Rischio

Contact Dr.Ferruccio Forlati

person:
Corso Unione Sovietica 216

10134 Torino—Italy

Tel: +39–011–3169335

Fax: +39–011–3169340

e-mail: [email protected]

URL: www.arpa.piemonte.it

Participants: Gianfranca Bellardone, Stefano Campus, Barbara Coraglia,

Ferruccio Forlati, Ermes Fusetti, Lidia Giacomelli, Noemi

Giordano,

Luca Mallen, Luca Paro, Manlio Ramasco, Gianfranco

Susella,

Herbert Sarri, Ferdinando Tamberlani, Carlo Troisi.

Politecnico di Torino (POLITO)

Dipartimento Ingegneria Strutturale e Geotecnica

Contact Prof. Claudio Scavia

person: C.so Duca degli Abruzzi 24

10129 Torino—Italy

Tel: +39–011–5644823

Fax: +39–011–5644899

e-mail: [email protected]

URL: http://www.polito.it/ricerca/dipartimenti/distr/

Participants: Andrea Allodi, Gaetano Amatruda, Marta Castelli, Stefania

Marello,

Marina Pirulli, Erminia Quacquarelli, Claudio Scavia.

Consiglio Nazionale delle Ricerche (CNR)

Istituto di Geoscienze e Georisorse (IGG)—Sezione di Torino

Contact person: Dr.Riccardo Polino

Via Accademia delle Scienze 5

10123 Torino—Italy

Tel: +39–011–530652

Fax: +39–011–530652

e-mail: [email protected]

URL: http://www.igg.cnr.it/

Participants: Michele Morelli, Fabrizio Piana, Riccardo Polino.

Page viii

Laboratoire Central des Ponts et Chaussées (LCPC)

Division de Mécanique des Sols et des Roches et de Géologie

de l’Ingénieur

Boulevard Lefebvre 58

75732 Paris—France

Tel: +33–01–40435246

Fax: +33–01–40436516

URL: www.lcpc.fr/

Centre d’Etudes Techniques de l’Equipement (CETE)

Contact Dr. Jean-Louis Durville

person:
CETE de Lyon

25 Av. Mitterand CSE 1

69674 Bron CEDEX—France

Tel: +33–04–72143215

Fax: +33–04–72143035

e-mail: [email protected]

URL: http://www.equipement.gouv.fr/

Participants: Jean-Louis Durville, Laurent Effandiantz, Pierre Pothérat,

Rosalba Russo, Ariane Stéphan.
Contribution:Philippe Marchesini (DDE-Isère).

Universitat Politècnica de Catalunya (UPC)

Departamento de Ingeniería del Terreno

Contact Prof. Pere Prat

person:
Jordi Girona 1–3, Building D2

08034 Barcelona—Spain

Tel: +34–93–4016511

Fax: +34–93–4017251

e-mail: [email protected]

URL: http://www.etcg.upc.es/

Participants: Jordi Corominas, Marcel Hürlimann, Alberto Ledesma, Pere

Prat.

Technische Universität Wien (TUW)

Institut für Ingenieurgeologie

Contact Prof. Rainer Poisel

person:
Karlsplatz 13

1040 Wien—Austria

Tel: +43–1–5880120–319

Fax: +43–1–5880120–399

e-mail: [email protected]
 URL: http://www.tuwien.ac.at/

Participants: Rainer Poisel, Alexander Preh, Werner Roth, Ewald

Tentschert.
Page ix

Ecole Polytechnique Fédérale de Lausanne (EPFL)

Faculté Environnement Naturel, Architectural et Construit

(ENAC)—Laboratoire de Mécanique des Sols (LMS)

Contact Christophe Bonnard

person:
EPFL—ENAC-LMS

1015 Lausanne—Switzerland

Tel: +41–21–6932312

Fax: +41–21–6934153

e-mail: [email protected]

URL: http://lmswww.epfl.ch/

Participants: Christophe Bonnard, Xavier Dewarrat, Jordi Moreno, Rafaël

Rojas.

Contribution: Francis Noverraz, Consulting geologist.

Page x

This page intentionally left blank.

Page xi

Contents

Preface xiii

1. Introduction 1
M.Castelli, & F.Forlati

2. The meaning of risk assessment related to large landslides 7

Ch.Bonnard

2.1 Specificities of most large landslides 7

2.2 Factors influencing the hazard assessment for large landslides 9

2.3 Expression of the consequences induced by large landslides 10

2.4 The meaning of risk for large landslides 11

3. A key approach: the IMIRILAND project method 13

G.Amatruda, Ch.Bonnard, M.Castelli, F.Forlati, L.Giacomelli,
M.Morelli, L.Paro, R.Piana, M.Pirulli, R.Polino, P.Prat, M.Ramasco,
C.Scavia, G.Bellardone, S.Campus, J.-L.Durville, R.Poisel, A.Preh,
W.Roth & E.H.Tentschert

3.1 Introduction 13

3.2 Definitions 15

3.3 QRA in the IMIRILAND project 16

3.4 Data collection 19

3.5 Hazard analysis 20

 3.6 Consequence analysis 34

3.7 Risk and total risk assessment 42

3.8 Scope of application of the proposed methodology 43

4. The Cassas landslide 45

F.Forlati, M.Morelli, L.Paro, F.Piana, R.Polino, G.Susella & C.Troisi

4.1 Introduction 45

4.2 Regional framework 47

4.3 Hazard analysis 53

4.4 Quantitative risk analysis 77

4.5 Indirect risk and regional impacts 85

4.6 Regulation and risk mitigation measures already available 87

5. The Rosone landslide 89

G.Amatruda, S.Campus, M.Castelli, L.Delle Piane, F.Forlati,
M.Morelli, L.Paro, F.Piana, M.Pirulli, R.Polino, M.Ramasco &
C.Scavia

5.1 Introduction 89

5.2 Geological background of the landslide 89

5.3 Historical background 92

5.4 Regional framework 92

Page xii

5.5 Hazard analysis 95

5.6 Quantitative risk analysis 118

6. The Oselitzenbach landslide 137

Gaetano Amatruda, Marta Castelli, Marcel Hurlimann, Alberto
Ledesma, Michele Morelli, Fabrizio Piana, Marina Pirulli, Rainer
Poisel, Riccardo Polino, Pere Prat, Alexander Preh, Werner Roth,
Claudio Scavia & Ewald Tentschert

6.1 Introduction 137

6.2 Regional framework 141

6.3 Hazard analysis 143

6.4 Quantitative risk analysis 171

6.5 Regulation and risk mitigation measures already available 180

7. The Ceppo Morelli rockslide 181

G.Amatruda, M.Castelli, F.Forlati, M.Hürlimann, A.Ledesma,
M.Morelli, L.Paro, F.Piana, M.Pirulli, R.Polino, P.Prat,
M.Ramasco, C.Scavia & C.Troisi

7.1 Introduction 181

7.2 Regional framework 183

7.3 Hazard analysis 185

7.4 Quantitative risk analysis 212

7.5 Regulation and risk mitigation measures already available 224

8. The Sedrun landslide 227

 Ch.Bonnard, X.Dewarrat & F.Noverraz

8.1 Introduction 227

8.2 Regional framework 228

8.3 Hazard analysis 230

8.4 Quantitative risk analysis 242

8.5 Regulation and risk mitigation measures 249

8.6 Conclusions 252

9. The Séchilienne landslide 253

Jean-Louis Durville, Laurent Effendiantz, Pierre Pothérat &
Philippe Marchesini

9.1 Introduction 253

9.2 Regional framework 255

9.3 Hazard analysis 256

9.4 Quantitative risk analysis 264

9.5 Regulation and risk mitigation measures already available 267

9.6 Conclusions 269

10. Specific situations in other contexts 271

Oldrich Hungr (10.1) & P.Basabe (10.2)

10.1 Rockslide hazard management: Canadian experience 271

10.2 Landslide risk reduction in Ecuador: from policy to practice 279

11. Suggestions, guidelines and perspectives of development 289

Ch.Bonnard, B.Coraglia, J.L.Durville & F.Forlati

11.1 General observations drawn from the studied cases 289

11.2 Consequences of the risk studies on land planning procedures 291

11.3 Consideration of risks in the development of exposed areas 293

Page xiii

11.4 Tendencies in risk management policy in several European 295

countries

11.5 Suggestions for future risk management studies 300

11.6 Perspectives and open questions 302

References 307

Author Index 317

Page xiv

This page intentionally left blank.

Page xv

Preface

Throughout history, all mountainous countries have been exposed to

natural hazards that often reached disastrous proportions. It is
undeniably a fact that the public awareness for natural disasters is
generally low and, therefore, progress in risk mitigation is mostly
based on the upsetting experience of the people concerned. This is
not only true for investments in risk mitigation measures but also for
the fields of research, legislation and standardisation. A significant
increase in damaging effects has been seen over the last two
decades. The reasons for the steady increase in damage due to
natural hazards are often ascribed to the supposed impact of climate
change. Additional and perhaps less spectacular reasons, such as
the increase in our living standards, in population density, and in
infrastructures and goods in economically privileged but hazardous
places, the development of settlements in disaster-prone regions,
the enormous increase in mobility on road and rail, the use of
hazardous technologies in industry and the increased susceptibility
to problems of our overall living space, have to be considered as well
(Fig. 1). In any event, to be able to develop efficient risk prevention
and mitigation strategies, it is important to know the true reasons for
this risk increase.

The disastrous events of the last few years have also clearly shown
that attainable security in natural hazard management remains
limited despite effective preventive measures. This is largely due to
the vulnerability and the sensitivity of our infrastructures (Fig. 2), as
well as our society and the environment which we live in. In addition,
it is more and more questionable to distinguish between natural risk
and that induced by human activity.

In the mountainous environment, landslides of various kinds, defined

in a general sense as “the movement of a mass of rock, debris or
earth down a slope” (Cruden, 1991) are frequently responsible for
considerable losses of both money and lives, and the severity of the
landslide problem worsens with increased urban development and
change in land use (Programme INTERREG 1, 1996). Given this
understanding, it is not surprising that landslides are rapidly
becoming the focus of major scientific research, engineering study
and practice, as is land use policy throughout the world, mainly in
rugged topographic areas.

Figure 1. Encampadana-Canillo landslide foot, Andorra–2002, a

region of intense touristic development.
Page xvi

When speaking of mountains, it should be realised that 14% of the

EU and 11% of the Accession Countries’ land area is formed of
mountainous relief (not including Switzerland!) (Source: EEA, 1999).
In Europe, major road and rail tunnels, high bridges and dams are
concentrated in the mountains, and are prone to widespread,
frequent and expensive damage. Furthermore, expansion of tourism
in mountainous villages has spread accommodation and
infrastructures into risk areas.

Risk assessment, prevention and mitigation require technical and

socio-economic answers to some basic questions. The question
“what risks can be considered as acceptable?” has to be answered
at a socio-cultural level and needs to be compared with the
questions “what could happen? how often? what are the
consequences?”, which require expert technical know-how. Science,
education and politics have to face this challenge and to support this
difficult decision-making process with adequate knowledge,
research, and the development of practical tools as well as through
insurance and legislation strategies. As a consequence, only a
multidisciplinary approach by specialists from different countries
using specific methods can increase confidence in risk assessment.
Therefore, the problem needs to be addressed not by single
countries, but at the European level.

This book is specifically dedicated to the impact of large landslides in

the mountainous environment in terms of identification and
assessment of risk. Several definitions exist for large landslides, for
instance the following one presented in the world landslide inventory
of the IUGS:

– volume of more than 1 million m3

– causes one victim or more

– induces a loss equal to the work of one man/year.

Figure 2. View of Cassas landslide, Italy–1983 where a major

motorway and railway are located at the toe.

This definition mainly focuses on sudden events of fairly large

magnitude but does not at all reflect the situation of complete
mountain slopes (in some cases affecting areas larger than dozens
of km2) in slow movement or threatened by huge rockfalls.
Therefore, the present text deals with another kind of large landslide
in which large communities or infrastructures are likely to be
Page xvii

exposed to the eventual extreme development of a phenomenon

characterised by a very slow basal movement induced by its
geological features.

A methodology carried out for the determination of hazard,

vulnerability and finally, risk, is presented, with some practical
applications to large landslide cases located in the Alps. The context,
derived from the first results of the project IMIRILAND, is supported
by the EU within the fifth framework programme.

The editors:
Christophe Bonnard, Ferruccio Forlati, Claudio Scavia.
Page xviii

This page intentionally left blank.

Page 1

1
Introduction

M.Castelli1, F.Forlati2
1Politecnico di Torino; 2Arpa Piemonte

Large landslides affect many mountain valleys in Europe. They are

characterised by a low probability of evolution into a catastrophic
event but can have very large direct and indirect impacts on man,
infrastructures and the environment (Table 1.1). This impact is
becoming more and more pronounced due to increasing tourist
development and the construction of new roads and railways in
mountainous areas. Methodologies for the assessment and
mitigation of risk are, therefore, a major issue. Since very large slope
movements are quite often directly or indirectly implicated in
disasters (not only large landslides but also secondary slides or
debris flows), their early identification, which is not always easy, is
essential to an adequate risk assessment of the zones involved.

Table 1.1. Some major landslides which have occurred over the past
centuries in the Alps.

Landslide Location Year Lives lost

Mont Granier French Alps 1248 About 3000

Salzburg Austria 1669 250

Goldau Swiss Alps 1806 457

Elm Swiss Alps 1881 115

Vaiont Italian Alps 1963 2100

Val Pola Italian Alps 1987 27

La Salle en Beaumont French Alps 1994 4

What do we know about landslide risks? A first theoretical definition

was given by Varnes in 1984: “Risk is the probability of an event of a
given magnitude multiplied by its consequences”.

Since then, many technical and scientific papers have been written
on this topic (i.e. Brabb, 1984; Brand, 1988; Bunce et al., 1997;
Carrara, 1992; Carrara et al., 1992; Cherubini et al., 1993; Cruden &
Fell, 1997; Cruden & Varnes, 1994; Einstein, 1988; Fell, 1994; Fell et
al., 2000; Finlay & Fell, 1997; Hungr, 1981; Hungr et al., 1993;
Hutchinson, 1992; Leroi, 1997; Mulder, 1991).

However, the evaluation of hazard is still made through different

approaches, both qualitative and quantitative, which lead to differing
results. Besides, only a few studies on vulnerability have been
carried out (i.e. Leone et al., 1996). Finally, no well-defined
procedure to include the results of risk analyses in land planning
from a legal point of view exists. As a matter of fact, many
experiences during the period of critical development of landslides
(Bonnard, 1994a; Vulliet & Bonnard, 1996) have shown a lack of
methodology and, above all, a nonsystematic approach of
interpreted risks. In practice, risk management is carried out by local
and regional authorities only during the actual critical event in a
necessarily improvised way. This “reactive” approach brings negative
consequences to the identification procedure and to preventive
actions. For example, very expensive monitoring systems have been
installed on several large landslides, but in spite of that:

– available data are often very dispersed (data banks, reports,

publications…) and their reliability is not always verified;
Page 2

– landslide studies are isolated, limiting their interpretation to local

factors;

– a well-established methodology linking the understanding of the

mechanisms of deformation and the interpretation of the data
obtained through the monitoring system to the practical questions
concerning the management of the risk does not exist (Fig. 1.1).

Figure 1.1. View of Ruines de Séchilienne landslide, France—1994,

where prevention works have been carried out but a global risk
management policy has not yet been adopted.

A literature review (IUGS/WGL/CRA, Working Group on Landslides,

committee on risk assessment, in Cruden & Fell, 1997) related to
these specific aspects has pointed out a lack of knowledge in the
following topics:

– experience obtained at a large scale to include the notion of actual

risk in land planning;
– comparative analyses of different hazard prediction techniques
carried out by multidisciplinary teams;

– application of risk assessment procedures to large landslide zones

by considering the potential damage in a quantitative manner and
the cost of protective measures in a complete economic balance
including direct and indirect costs;

– comparative assessments of the applicability of such techniques in

different frameworks;

– direct involvement of end users (regional and national technical

administrations);

– application of risk assessment and mitigation methods to real

cases, in which the administrative and legal aspects of the final
results can be checked.

Furthermore, risk can extend well beyond local damage (for

instance, risk of river damming which may induce major hydrological
hazards: floods, inundation of sewage plants, loss of drinking water
resources), so that it must be considered from a wider perspective.
In this sense, a multidisciplinary approach by specialists from
different countries using specific methods (i.e. tectonic analyses,
photogrammetry, numerical analyses, etc.) at some well-monitored
sites can improve the determination of the main landslide
mechanisms and thus the assessment of the hazard level, while the
contribution of administrators and end users in general can ensure
practical checks of the reliability and suitability of the developed
techniques.

Thus, the present text intends to examine the hazard analysis of

different types of landslides more closely by the application of
geological, geomechanical and statistical methods. Particular
Page 3

attention is given to the understanding of the structural setting of the

landslide sites and adjoining areas. Relations between “on site” and
“around site” structural features have been sought in order to define
the actual boundaries of the potential landslide areas.

These procedures are based on the application of basic techniques

of structural analysis (Hancock, 1985 and references therein), with a
special emphasis on the recognition of fundamental structural
associations according to Hobbs et al. (1976) (see also Forlati &
Piana, 2000). Furthermore, extensive use of satellite images has
been made to search for relations between the distribution of local
and regional structural elements (Morelli & Piana, 2003). Structural
models obtained for each landslide provided the basic inputs for the
geomechanical numerical models. Particular attention was given to
numerical models, where some of the most current finite element,
boundary element and finite difference methods were applied in
three-dimensional conditions on the same slope stability problems
(Castelli et al., 2001). A comparative analysis of different hazard
prediction techniques can then be carried out.

Finally, vulnerability and risk analyses have been developed for the
same landslides considering short and long-term perspectives, direct
and indirect consequences, as well as technical and social impacts
(Fig. 1.2). The combination of the results obtained in relation to
hazard analysis and through the risk approach will allow the
development of a new practical and quantified risk assessment
program which will be applied to several sites.

The book presents the results obtained by the IMIRILAND project,

supported by the EU from March 1st, 2001 to December 31st, 2003.

The project was born of the common interest of many European

countries in exchanging information and experiences in the field of
the use and development of risk assessment procedures. In
particular, the countries involved in the consortium (Italy, Austria,
France, Spain, Switzerland) are aware of the importance of these
problems and have been working on them for several years. The
consortium involves different and complementary expertise,
including some of the most scientifically advanced universities and
research infrastructures dealing with natural hazard problems as well
as two end users which must face practical land management
problems every day.

Figure 1.2. Dams on the Oselitzenbach landslide, Austria—2002,

needed to protect the valley from erosion phenomena which might
trigger this slide.
Page 4

The project allowed the selection of large landslide sites located in

various mountainous environments and legal contexts for the
development and illustration of risk assessment methodologies.
These areas ranged from the Pyrenees (Spain, France) to the West,
to the Austrian Alps to the East, passing through western (France,
Italy) and central (Switzerland) areas, in order to grant a high degree
of generality to the methodologies and to impart a European
dimension to the problem. In this way, the results can also be
diffused and applied outside the countries directly involved in the
project. The partnership, therefore, created a remarkable critical
mass for EU socio-economic development.

Eight landslide sites (Fig. 1.3) were selected (slides or rockfalls,

volume >106 m3) in the countries involved (Italy, France, Spain,
Austria, Switzerland), where geological, geomechanical and
monitoring data already exist. The selected cases (already occurred
or in a pre-failure phase, see Fig. 1.4) imply a complex risk situation
for which no reasonable stabilisation works can be executed. The
partners involved in the project already have some knowledge about
the sites, the data of which are already available but very dispersed
(data banks, reports, publications). In this way,

Figure 1.3. Landslide site locations.

Figure 1.4. Uphill-facing scarps in Sedrun landslide, Switzerland—
2002 whose movement has been significantly increasing over the
last 20 years.
Page 5

it has been fairly easy to collect a large amount of information

without expending excessive amounts of time and energy.

A general scheme for data description was created and compiled for
each site (Table 1.2). Each partner was then able to access data and
to manage them in a homogeneous manner. The objective was also
to check the reliability of the available data and to choose which
were to be considered as significant for the risk analysis.

The principal aims of the project are to:

– carry out a review of hazard analysis methodologies,

– compare the reliability of hazard assessment in different situations

according to various criteria,

– develop practical risk analysis methods considering direct and

indirect potential damage,

– apply the developed methods to real situations in different

countries,

– test the applicability of such approaches through close interaction

with administrators.

Table 1.2. Selected sites.

Site Country Approximate volumeMain elements at risk

of main landslide
[m3]/globally
affected area [km2]

ROSONE Italy 50×106 m3/5.5 km2 Rosone village and

main road
 SS 460

Hydroelectric power
plant

Valley downstream

CASSAS Italy 15×106 m3/0.9 km2 Highway A32

(Frejus)

International Torino-
Modane railway

Valley downstream

CEPPO MORELLI Italy 5×106 m3/0.5 km2 National Road N.

549, only link to the
Macugnaga tourist
resort

Prequartera and
Campioli villages

ENCAMPADANA Spain Several tens×106 Important ski resort

m3/1 km2
El Tarter, small
village

Main road Andorra—

France

Communities
downstream

OSELITZEN- Austria 0.5–2×106 m3/2.9 Villages of Tröpolach

BACH km2 and Watschig
downstream
 Main road “Naßfeld
—Bundesstraβe”,
important link to the
“Gailtal” Valley

SÉCHILIENNE France 25×106 m3/0.9 km2 National road RN 91

Valley downstream

CONTERS- Switzerland2×109 m3/25 km2 New A28 national

GOTSCHNAHANG road

Town of Lanquart

Valley downstream

SEDRUN Switzerland100×106 m3/1.5 Regional railway line

km2
National road

Sedrun tourist resort

Note: The volumes indicated above refer to the modelled active landslide
masses, whereas the areas indicated refer to the globally involved instability
phenomena.

To achieve such objectives, the following phases have been

foreseen:

1. Data collection. A data base has been created containing all

available data on the selected sites.

2. Development of risk assessment methodology. In order to

compare the results of different methodologies in terms of landslide
scenario evolution, the following approaches have been
Page 6

considered: (a) Field analysis, (b) Mechanical modelling. For each

approach, state-of-the-art, critical analyses of the methods and
applications to selected sites have been carried out. Vulnerability
and risk analyses were also made.

3. Application to management. This concerns to the application of

the developed methodologies to the management of endangered
landslide zones. The actions are defined in relation to the problem
urgency, the legal framework and the power of local and regional
authorities, as well as the relevant economical conditions.

4. Dissemination of risk management methodologies.

The global and multidisciplinary approach which characterises the

project represents the first important result for the entire consortium.
In fact, it allows the testing of a new method of collaboration at the
European level, gathering various researchers, technicians and
officials from different countries and making them study similar
problems (natural hazards in mountainous environments) in different
contexts (from a socio-political point of view). By analysing and
interpreting all of these experiences, a unified problem-solving
approach was developed. In particular, for universities and academic
organisations, the project represented a very useful opportunity for
developing new methodologies or applying existing ones to several
real situations, enabling a test of their applicability and reliability. As
far as the other members of the consortium were concerned, who
were also end users, these persons had the opportunity of
establishing a direct relationship with the academic world, in order to
work together to find practical solutions for the problems they must
face as a part of their administrative tasks.

Another important positive consequence of the project is the

improvement in the perception of scientific and technical problems
by officials and public administrators working in the field. A specific
work phase has been dedicated to this aspect; the comparison
among different approaches to similar phenomena within different
legal frameworks produced very interesting and useful results.

This special book is organised into eleven major sections, written by

several authors:

1. Introduction

2. The meaning of risk assessment related to large landslides

3. A key approach: the IMIRILAND project method

4.–9. Integrated quantitative risk assessment applied to the six

selected landslides

10. Specific situations in other contexts

11. Suggestions, guidelines and perspectives of development.

Page 7

2
The meaning of risk assessment related to large
landslides

Ch.Bonnard
Ecole Polytechnique Fédérale de Lausanne

2.1 SPECIFICITIES OF MOST LARGE LANDSLIDES

Most observed or inferred landslides (falls, slides, flows) generally

affect a very limited area; therefore, their management requires a
well-defined analysis and local action only in terms of prevention or
remediation. On the contrary, large landslides which cover more than
1 km2 or involve huge masses of rocks with extremely high potential
kinetic energy require a global consideration in the context of land
planning procedures, in which the aspects of risk assessment within
the prevention action play a major role.

Thus it is important and even essential to identify the specificities of

the majority of large landslides in terms of mechanisms, preparatory
and triggering causes and potential development. On the basis of the
case studies analysed within the IMIRILAND project (see Table 1.2),
four different typical situations related to large landslides may be
specified and analysed in order to understand the different
circumstances and conditions which rule the assessment of the risks
they involve.

2.1.1 Slow permanent movement of extensive areas on which

dwellings and infrastructures exist
This typical situation is often encountered on Alpine or Pre-alpine
slopes which have been affected by glacial retreat approximately ten
thousand years ago. The observed movements, provided they have
been identified and monitored over long periods, are generally fairly
regular, but some occasional accelerations may occur which last for
a short period of several months to a few years. Their specificity is
mainly their surface extension, from 1 to 40 km2, and their gentle
morphology, thus inducing the population to use this apparently
favourable land for settling villages on slopes with a shallow gradient
and developers to build infrastructures such as roads and cable cars
(see Fig. 2.1 showing the Lugnez landslide).

In many cases, the velocity pattern of the different sectors of this

type of landslide is fairly homogeneous, and differential velocities are
only recorded in specific zones, so that these zones are more
recognised by the population as local shear bands, which display
local damage to roads and buildings, rather than as the limits of
extensive moving areas.

The risks induced by such continuous phenomena are especially

related to serious climatic events of high intensity or long duration,
which may bring about a momentary acceleration and an increase of
differential movements, thus causing a disruption of the road network
and serious damage to houses located near the shear zones (see
Fig. 2.2 showing the acceleration of one point A located on the toe of
La Frasse Landslide (Canton of Vaud/Switzerland), whereas another
point B nearby does not show any acceleration) (Bonnard & al.,
1995).

2.1.2 Significant potential movement of large areas inducing major

cracks in slopes, change of geometry at the toe and consequent
damming of the valley
This critical case is often ignored or improperly understood as the
main movements of such a landslide are quite limited in terms of
average velocity and as the evidence of an activity of the mass
Page 8

Figure 2.1. Lugnez Landslide, located in Eastern Switzerland,

displays a gentle and regular slope over an area of more than 30
km2 on which several villages have developed since the Middle
Ages, despite a regular movement of several centimeters per year
(Noverraz & al., 1998).

is often limited to a localised zone near the scarp or at the toe. In the
case of a marked reactivation, the movement of the mass may
evolve into a rockfall phenomenon or a debris slide near the toe,
which, according to the size of the blocks, might be able to dam the
valley and cause flooding first upstream, and then downstream if a
sudden release of stored water occurs following a dam break event.
In this case, the impact of the phenomenon is much larger than its
own surface extension and may affect hundreds of km2 (see the
Sechilienne landslide commented on in Chapter 9). Such a case
occurred in 1993 in southern Ecuador at La Josefina as presented in
Chap. 10.2 below.

2.1.3 Possible rapid or very rapid mass movement (rockfall, rock

avalanche) causing a major threat to the communities and
infrastructures below the source area
This case is even more difficult to detect in many locations and
geological settings, as the preliminary movements are hardly
perceptible and the masses affected by typical preparatory
mechanisms, such as toppling, may be quite extensive. If the
unstable masses in one part of the slope experience a sudden
acceleration, sometimes related to a specific climatic event, the
induced rockfall may be massive and widespread, and can therefore
seriously threaten the villages or communities below, especially if the
slope is steep. As far as transportation infrastructures which cross
these unstable zones at the toe of the slope are concerned, they are
very difficult to protect due to the high energy of the falling blocks
and to the uncertainty of their trajectory, so that the only way to
handle this problem often implies the construction of a tunnel (see
the case of Ceppo Morelli commented on in Chapter 7).

The suddenness of such phenomena implies that any monitoring

program may be inefficient to trigger an alarm in due time and to
protect the users of the transportation routes located below the
source area.

2.1.4 Large debris flows induced by the presence of an active slide

on the side of a stream, causing damage in the valley downstream
A critical situation in terms of risk is also induced by the occurrence
of debris flows of large magnitude which are “fed” by erosional
processes touching active slides on one or both banks of the valley.
Page 9

Figure 2.2. Evolution over 25 years of the displacements of two

points located on the lower part of La Frasse Landslide (right scale
[cm]), as well as the monthly rainfall amounts (left scale [mm]). The
crisis of 1981–1982 seriously affected a main road crossing the
landslide (point A) whereas the movement at point B nearby was not
appreciably modified.

This “in-chain” mechanism, which may lead to a momentary

damming of the river at the toe of the slope, has serious
consequences on the population and infrastructures downstream, as
the extension and magnitude of the flow may far exceed previously
observed phenomena (see Oselitzenbach case commented on in
Chapter 6).

2.2 FACTORS INFLUENCING THE HAZARD

ASSESSMENT FOR LARGE LANDSLIDES

Considering the specificities of large landslides which involve huge

masses of soil or rock affected by gravity-induced movements and
taking into account that their behaviour is never repetitive, it is quite
difficult, if not impossible, to assess their hazard on the basis of
statistical monitoring data as is commonly done in flood prediction.
Therefore, hazard analysis must use particular criteria and consider
various factors in order to assess the hazard level for different
scenarios in a semiquantitative way, so as to specify relative orders
of magnitude of intensity and probability as explained in the
methodology developed in Chapter 3. This approach, implying a
combination of both parameters in order to quantify the hazard level,
will nevertheless allow a quantitative risk assessment which can be
used to manage the exposed areas.

The proposed hazard qualification through a matrix system can be

adapted in a flexible way to quantitative values when relevant
information exists concerning either intensity ranges, obtained for
instance by mechanical modelling, or probability ranges, based on
the analysis of statistical data.

Of course, the corresponding hazard value only has meaning as a

range which will qualify a certain surface extent on a map; such an
analysis cannot reasonably be done for a specific point or individual
object exposed to a certain danger. The significance of these hazard
zones on a map, which must not be too detailed or fragmented, will
then be considered in the vulnerability of the exposed objects, taken
in a general sense, in order to produce risk maps.
Page 10

Finally, the main factors influencing hazard assessment, in the case

of large landslides, are first the intensity and probability of different
events selected within reference scenarios, but also the
development or propagation of such events in a regional context,
and finally the combination of several events (extreme rainfall and
related flood plus landslide causing the damming of the river, for
instance). These catastrophic combinations, even if quite rare,
represent the determinant scenarios for most tragedies and the
cause of major disasters. Therefore, a comprehensive analysis of
hazards related to large landslides must be carried out in order to
assess the relevant risks properly.

2.3 EXPRESSION OF THE CONSEQUENCES INDUCED

BY LARGE LANDSLIDES

In order to assess the risks induced by large landslides, it is

necessary to determine in which way the hazard situations
investigated in a dangerous area may affect the population, the
buildings and infrastructures, as well as the environment. Indeed,
risks do exist only if a potential damage of a given magnitude may
be determined, whatever its characteristics are. Therefore, the notion
of direct and indirect impact has to be defined first.

2.3.1 Direct and indirect impact

Any natural phenomenon modifying the geometrical and
geomechanical conditions of a slope may have an impact on the
population, buildings and activities which can be classified as direct
or indirect consequences. Under direct consequences are grouped
all of the physical impacts leading either to destruction, damage
(cracks in buildings, failure of pipes) or even excessive deformation
(inclination of a house due to the movement of a landslide, without
any damage to the structure of the building, see Fig. 2.3).
These direct consequences include infrastructures hit by blocks and
buildings on a sliding slope affected by differential movements, but
also constructions in the valley downstream a slide which might be
destroyed by a flash flood induced by the failure of a dam caused by
a landslide blocking the valley floor. The impact on animals is also
included in direct damage.

Figure 2.3. Inclined chalet on the Chloewena landslide, following the

disaster of 1994; the building is considered as destroyed even
though it is not structurally damaged (Vulliet & Bonnard, 1996).
Page 11

The direct consequences include, of course, persons killed,

wounded or affected by the loss of their houses and properties due
to the landslide phenomenon. This even includes those persons that
need to be evacuated before the occurrence of the landslide,
because they might be affected by the mass itself or by the
consequences of the induced phenomena, such as the flash flood
mentioned above.

The direct consequences can also relate to the environment, in the

way that a landslide may destroy forests, animals or a habitat zone
for a certain type of protected animal; it may also affect the springs
of water available at the toe of a rockfall. The method of assessing
all of these direct consequences will be detailed in the following
paragraphs.

As far as indirect consequences are concerned, these deal with all

disruptions of or hindrances to the economic activities induced by the
considered landslide, mainly due to the interruption of traffic, the
reduction in tourist demand and other indirect consequences related
to the modification of the state of the landslide surface. These
indirect consequences are fairly difficult to assess because they may
imply a spatial extent much larger than the landslide zone itself.
Their nature may also change with time because when a hindrance
is suddenly introduced in an economic system other solutions are
found by all the partners to limit the negative consequences of this
perturbation. Thus, the magnitude of the potential impacts for a given
phenomenon must be expressed by different aspects, mainly social,
environmental and economic, as presented in Chapter 3 below.

2.4 THE MEANING OF RISK FOR LARGE LANDSLIDES

The main specificity of large landslides, such as those considered in

the IMIRILAND project, is that the probability of a critical global
movement of the whole mass is very low, whereas the
consequences may be catastrophic. It is, thus, indispensable to
assess several scenarios, and not only the worst case, with their
respective potential consequences, in order to quantify the
corresponding risk components in comparable terms, and to propose
preventive measures on a wide scale, so that such an extreme event
does not hit the site by surprise.

The comparison of the impacts of different scenarios computed on

parallel bases also allows the comparison among different
prevention strategies.

Such a risk assessment gives a correct and comprehensive

overview of the potential impact and thus provides a refined tool for
decision makers, which would otherwise either ignore the risks as a
politically incorrect perspective, or exaggerate the risks in order to
call for subsidies and often carry out inappropriate actions.

2.4.1 Possible management of risk

In the same way, the meaning of risk is difficult to express; the
management of the zones or regions exposed to landslide risk is
politically complex. First, nearly no reference situation in the past can
provide a guideline to adopt an adequate political and technical
action, as even similar situations which happened some centuries
ago (for instance in the case of rockfalls) are not always well
documented and occurred in a very different socio-economic
context. Then, it is certainly inappropriate to prohibit any
development on the largest potential impact zone, especially as the
people living in the mountainous areas in which large landslides
occur have already many hindrances to face and must be
encouraged to remain in these regions as much as possible. From
the other extreme perspective, it is also inappropriate to ignore the
potential hazards totally, even if they are quite rare, and to favour the
construction of any buildings and infrastructures in exposed areas,
without analysing alternative solutions or more careful layouts.
The ultimate goal of such research is thus to provide guidelines for
the management of the exposed zones in order to favour sufficient
attention to the risks without impeding any perspective of
development.
Page 12

2.4.2 Impact of risk assessment in the case of large landslides

It has often been observed that the consideration of landslide risks
by local authorities is either too optimistic (the risk is nearly denied)
or too pessimistic (major prohibitions are pronounced, major
protection works are carried out without a reasonable risk analysis).
It is, therefore, difficult to measure the real impact of a proper risk
assessment on local development, because the analysis and
consecutive political measures are not founded on a systematic
approach. This often leads to the execution of apparently efficient
protection works, which neither solve the stability problem nor limit
the hazard in case of rare events but rather aim at preserving the
good conscience of the authorities.

After a serious event with direct impact on exposed structures, even

if it implies only a small portion of the whole landslide phenomenon,
the authorities tend to favour extremely safe solutions and adopt
preventive measures which may be excessive with regard to the real
risk analysis results. This perspective is inappropriate for two
reasons; first, it leads to an unnecessary investment with respect to
the real necessity of protection of the exposed objects and, second,
it creates precedents in the risk management policy so that the need
of safety seems to grow unnecessarily. Thus, a rational approach is
needed, especially from a European perspective which must aim at
dealing the similar problems in a proportionate way.
Page 13

3
A key approach: the IMIRILAND project method

G.Amatruda1, Ch.Bonnard2, M.Castelli1, F.Forlati3, L.Giacomelli3,

M.Morelli4, L.Paro3, F.Piana4, M.Pirulli1, R.Polino4, P.Prat5,
M.Ramasco3, C.Scavia1,

G.Bellardone3, S.Campus3, J.-L.Durville6, R.Poisel7, A.Preh7,

W.Roth7, E.H.Tentschert7
1Politecnico di Torino; 2Ecole Polytechnique Fédérale de Lausanne;
3Arpa Piemonte; 4CNR-IGG sezione Torino; 5Universitat Politècnica
de Catalunya; 6Centre d’Études Techniques de l’Équipement de
Lyon; 7Technische Universitat Wien

3.1 INTRODUCTION

In order to supply quantitative data related to landslide risk to land

planners, it is necessary to consider the existing available tools first.

Quantitative Risk Assessment (QRA) is a method of quantifying the

degree of risk through a systematic examination of the factors
contributing to landslide hazard and affecting the severity of the
consequences, as well as the definition of probabilities for the
individual factors. The QRA technique was developed, and is now
used world-wide, to estimate risks from industrial plants, such as
petro-chemical, etc. QRA can provide a framework for assessing the
cost-effectiveness of risk mitigation options, formulating risk
management strategies and facilitating decision-making on resource
allocation to reduce risks posed by different types of landslide
hazards.

In simple terms, the following questions are addressed in a QRA of

landslides:
– What can cause damage?

– Where?

– How often?

– What could happen?

– How severe would the impact be?

The logical progression of landslide risk assessment has been

discussed by many authors (see, for instance Varnes et al., 1984;
Cruden & Fell, 1997). Most notable is the procedure developed by
Varnes et al. (1984); UN (1991) and Fell (1994). This consists of key
steps (Table 3.1), which can be covered quantitatively or
qualitatively.

Table 3.1. Steps in a hazard and risk assessment procedure for

landslides (Fell, 1994).

Action

1. Recognition of hazard (i.e. landslides of a certain type)

2. Estimation of magnitudes (volumes)

3. Estimation of the corresponding occurrence probabilities

4. Determination of elements at risk

5. Estimation of vulnerabilities

6. Calculation of specific risks

7. Calculation of the total risk

8. Assessment of risk acceptability

9. Mitigation of risk (if necessary)
Page 14

It should be noted that steps 8 and 9 deal with risk management

rather with than risk assessment, as explained in the definitions
given in § 3.2.

The evaluation of risk involves the notions of hazard, vulnerability

and cost of consequences. Theoretically, it is defined as follows
(Leroi, 1997):

where, R: risk; Aі: hazard i; Vji: vulnerability of object j exposed to

hazard i; Cj: cost or value of the object j.

Recent works (Cruden & Fell, 1997) propose the following

multidisciplinary procedure for the quantitative risk analysis of slopes
as a standardisation of the previous definitions:

– Hazard analysis: analysis of the probability and characteristics of

the potential landslides;

– Identification of the elements at risk, i.e. their number and

characteristics;

– Analysis of the vulnerability of the elements at risk;

– Calculation of the risk from the hazard, elements at risk and the
vulnerability of the elements at risk.

In the same reference, the Working Group on Landslides, Committee

on risk assessment, discussed the state-of-the-art on QRA for slopes
and landslides and detailed issues requiring research and
development. On the basis of these outcomes, the IMIRILAND team
addressed the theme aiming at the following proposed
improvements:
– “The process (i.e. QRA) would be carried out by suitably qualified
and experienced persons”

The researchers and technicians involved (representing different

countries) are aware of the importance of these problems, as they
have been working for several years on landslide risk analysis. They
believe that a good methodology could be derived from a
multidisciplinary approach taking into account the different know-
how, knowledge and results obtained by the different disciplines.
Another important aspect of the project lies in the implication of
endusers who have to face every day practical land management
problems;

– “QRA does not have to be based on large quantities of objective

data, it is equally valid to use subjective, as well as objective, data”

A first project step consists in defining a logical organisation of all

site-specific available data (historical and technical-scientific
material) to facilitate the activities of data collection, data analysis,
data processing updates and the creation of different outputs;
keeping in mind all information which could be involved in a QRA
process. A special effort was made to link subjective information (or
best interpreted data) with measured data and to create a new
working relationship between two fundamental disciplines: Geology
and Geotechnics.

–“The inability to recognise a significant hazard and thus

underestimate the risk [has to be considered]”

A very delicate step in the QRA process is related to the danger

characterisation and hazard evaluation. From a simplified point of
view, the calculation of risk is to be considered as a mathematical
computation, which is relatively easy when the hazard corresponding
to each danger, the elements at risk, their vulnerabilities and values
are known. The main objective taken into account in the IMIRILAND
project consists, therefore, in improving the hazard analysis, either
by specifying the role and contribution of the different disciplines
involved in such an investigation, or by defining criteria for the field
observation and modelling, aiming at the quantification of the
different scenarios.

Chapter 3 gathers the contributions of the IMIRILAND team on risk

analysis and explains the practical risk assessment framework
developed. Particular attention is given to pointing out simple but
rigorous tools devoted to risk analysis, and detailing the relevant
steps.
Page 15

3.2 DEFINITIONS

Several definitions of the terms used in QRA exist, sometimes with

differing meanings. On the basis of the results obtained by the
IMIRILAND project, the following definitions of the most important
terms, as used in this book, are given below, including the Italian,
French, German and Spanish terms. Some concepts proposed by
Cruden and Fell (1997) have also been developed:

Hazard (Pericolosità, Aléa, Gefährlichkeit, Peligrosidad)

For a given danger, the characterisation of hazard has to include the

probability of occurrence within a specific period of time, the
potentially involved area and the intensity of the damaging
phenomenon.

Danger (Fenomeno o Processo naturale, Phénomène ou Danger,

Gefahr, Peligro) is an existing or potential natural phenomenon, in
this case the landslide, which may induce negative consequences.
Danger characterisation (unlike danger identification) includes: a
relative and qualitative non-temporal probability of detachment
(predisposition), geometrical characterisation, geostructural and
geomorphological constrains, mechanism description. It, thus,
provides the basis for the construction of geomechanical models.

Intensity (Intensità, Intensité, Intensität, Intensidad) is a measure of

the destructive potential of a landslide, based on a set of physical
parameters, such as downslope velocity, thickness of the landslide
debris, volume, energy and impact forces. Intensity can be
expressed qualitatively or quantitatively Intensity varies with location
along and across the path of the landslide and, therefore, it should
ideally be described using a spatial distribution function or an
appropriate map.
Occurrence probability (Probabilità di accadimento, Probabilité
d’occurence, Eintretenswahrscheinlichkeit, Probabilidad) is the
probability that a certain danger will occur over a specified period of
time. It can be evaluated in both quantitative and qualitative terms
and it contributes to risk evaluation. Sometimes a relative probability
is used as a non-temporal concept. In this case, it should be clearly
specified as such by the term “predisposition”.

Hazard Scenario is the spatial representation of a specific hazard

situation and its development.

Vulnerability (Vulnerabilità, Vulnérabilité, Verwundbarkeit,

Vulnerabilidad)

It is the degree of loss for a given element at risk, or set of such

elements, resulting from the occurrence of a natural phenomenon of
a given magnitude. It is usually expressed in relative terms, using
words such as “no damage”, “some damage”, “major damage” and
“total loss”, or by a numerical scale between 0 (no damage) and 1
(total loss). It can be more significant if, instead of assuming only
one vulnerability coefficient, its different specific components (i.e.
physical, social, economic, environmental) are considered.

Elements at risk (Elementi a rischio, Eléments à risque, gefährdete

Elemente, Elementos expuestos a riesgo) include any land,
resources, environmental values, buildings, infrastructures,
economic activities and/or persons in the area that are exposed to
natural and technological hazards. The elements at risk can be
quantified by a monetary value or by a relative value scale.

Consequence is the resulting loss or injury, or the potential loss or

injury. It is the product of the value of the elements at risk and their
vulnerability, and can be quantified if the element at risk is expressed
as a value and if the vulnerability is expressed numerically. When a
consequence is expressed qualitatively, it is sometimes referred to
as a “consequence rating”.
Risk (Rischio, Risque, Risiko, Riesgo)

Landslide risk considers all the consequences of the landslide

hazard. Simply stated, risk is the product of the occurrence
probability and the consequences.

Physical Risk, Social Risk, Environmental Risk, Economic Risk are

the risks connected to each specific component of vulnerability. They
are separately computed for each scenario.

Total Physical risk is the sum of the respective risks for all the
scenarios, whenever relevant.

Total Economic risk is the sum of the respective risks for all the
scenarios, whenever relevant.

Total Social risk is the sum of the respective risks for all the
scenarios, whenever relevant.

Total Environmental risk is the sum of the respective risks for all the
scenarios, whenever relevant.
Page 16

Risk Analysis is the processing of all available information to

estimate the risk to individuals or populations, property or the
environment, from hazards. Risk analyses generally contain the
following steps: definition of danger, hazard identification and risk
estimation.

Risk Estimation is the process used to produce a measure of the

level of health, property, or environmental risks being analysed. Risk
estimation contains the following steps: frequency analysis,
consequence analysis and their integration.

Risk Assessment is the process of risk analysis and explicit or

implicit judgement, considering the various components of risk
(physical, social, economic, environmental), to relate the risk
perception with the results of risk analysis.

Risk Management is the iterative process of risk assessment and

risk control, including active and passive mitigation measures or
strategies, incorporating the risk acceptability whenever possible or
politically acceptable.

3.3 QRA IN THE IMIRILAND PROJECT

To deal with the problem of risk evaluation from a quantitative point

of view (QRA), we presuppose that, rigorously speaking and
according to the QRA definitions:

– each risk component should be numerically expressed;

– each risk component should be spatially represented.

In the light of this assumption, the methodology elaborated for the

IMIRILAND project is based on the development of the whole
process by means of integrated and consequential phases, referred
to as a simple mathematical operation with a matricial approach and
a spatial representation of risk through GIS techniques. The use of
GIS techniques can be considered as a relatively new approach in
slope risk assessment, with a logical structure in independent layers
of georeferred information. Different layers related to a single risk
parameter can be easily combined and cross each other in order to
derive a spatial quantitative risk representation (Fig. 3.1).

Each component of hazard and risk, represented by quantitative

values, as objectively as possible, can be considered as a numerical
attribute of a specific spatial representation.

The risk assessment process can be summarized in three steps:

1. Hazard analysis

2. Consequence analysis

3. Risk computation.

The most important question to be asked in each landslide hazard

and risk analysis relates to danger individuation and its spatial
representation (present or potential instability zones). A preliminary
step for the complete process consists, therefore, of a detailed
available data collection in order to identify the dangers. It is
important to mention that the majority of large landslides are
characterised by interacting specific mechanisms relating
preparatory and triggering causes. Each mechanism can evolve in a
specific way and, therefore, can represent a specific danger. In other
terms (Fig. 3.2), a large landslide could generate many dangers and,
as a consequence, the relation between a large landslide and the
induced dangers can be single or multiple. Indeed, one or more
hazards may correspond to each danger. This complex
correspondence derives from the fact that, in the hazard assessment
phase, several methods (e.g. geomechanical run-out modelling) can
be applied and, therefore, different results may be obtained in terms
of the affected area and intensity. Such results must be critically
examined in order to select the most representative hazard scenario.

The proposed methodology is based on the consideration of four

hazard components:

– Danger characterisation,

– The area affected in relation to a possible landslide evolution,

– Process intensity,

– Occurrence probability.
Page 17

Figure 3.1. Example of representation of different risk maps related

to the specific vulnerability factors considered obtained using GIS
techniques.
Figure 3.2. Principal steps in relation to the risk assessment process
used in the IMIRILAND Project.
Page 18

The danger characterisation points out the most prominent

geostructural and geomorphological constraints, allowing the
construction of a geomechanical model. The affected area describes
the limits of the possible evolution of a landslide of a given intensity.

The process intensity is a measure of the potential destructive

impact of the landslide.

The occurrence probability is the probability that a certain danger will

occur over a specified period of time. The hazard specific
components supply, moreover, their contributions to assess the
consequences of the landslide, in a specific way:

– The affected area permits the spatial identification of the elements

at risk.

– The process intensity contributes to the determination of the

vulnerability of the elements at risk.

– The occurrence probability, together with the consequences (the

product of the value of the elements at risk and the vulnerability),
defines the risk.

In order to perform the quantitative risk analysis, the process must

first be divided into consecutive steps, then a matrix is used in which
the row values increase vertically downwards and the column values
increase horizontally to the right (Fig. 3.3).
Figure 3.3. Matricial risk computation scheme.

The matrix is built by multiplying each row value by the

corresponding column value; thus a new risk-derived component is
obtained. In order to simplify the computation for some risk
parameters (i.e. the value of the exposed object), a relative index
can be used.

Theoretically, the value of risk is computed by multiplying the

probability that a certain phenomenon of a given intensity might
occur by the vulnerability and by the value of the assets in question
or of the exposed persons. This simple definition, introduced by
Varnes (1978), can be refined, first, by considering that a significant
probability can be established for each reference hazard scenario on
the basis of the recurrence of historic or similar past events or,
second, by considering the probability of triggering causes such as
rainfall.

Different vulnerability components (i.e. physical, social, economic,

environmental) are considered (see § 3.5.3).

The decision to resort to these different components arises from the

difficulty in allocating comparable values to all of the different
categories of goods and persons and, thus, in obtaining comparable
risk scenarios. The presentation of risk related to several scenarios
with a separate consideration of the different vulnerability factors
allows a matricial vision of the different risk components (Fig. 3.4).
Several sums may thus be computed. One possibility is the sum of
any risk for each scenario, connected to each specific component of
vulnerability. This sum expresses the global risk value of the
considered scenario, but requires the specification of a
homogeneous scale, namely by the expression of risks in monetary
terms. A second possibility is the sum of the respective risks for
Page 19

Figure 3.4. Matrix of individual risk values expressed as a function of

different specific vulnerabilities and scenarios (P: physical, S: social,
E: economic, En: environmental).

all scenarios considering specific components of vulnerability. This

sum (total risk) is justified, provided that the objects affected as a
consequence of one scenario will be repaired or rebuilt before the
next scenario occurs.

This type of sum is not appropriate when one scenario is conditioned

by the occurrence of another one (for example when a large rockfall
is triggered by a preliminary boulder fall).

As previously mentioned, an extensive use of GIS has been made to

perform such analyses and assessments. One of the advantages of
using a GIS technique is the possibility of producing an iterative
advanced risk mapping. The spatial overlapping of different risk
components/factors permits a quite simple and useful evaluation of
results. The possibility of organising the information into geo-
referenced vectorial information layers allows the association of
several attributes with each spatial entity.
In the following pages, a detailed description of the complete
IMIRILAND methodology with its different phases and tools is
presented.

3.4 DATA COLLECTION

Systematic collection, analysis and organisation of available data

allows a better hazard and risk assessment including their scientific,
technical, economical and political components and represents a first
logical and necessary step in the consequential phases devoted to
risk mitigation strategy.

The partners involved in the project generally have some knowledge

concerning the sites which permits an appropriate danger
identification, but the information is often very dispersed and
available from different government sources; it may include maps,
reports, aerial photographs, historical documents, etc. As spatial and
temporal comparisons of the data are of great interest for landslide
investigation and comprehension, it has been decided to define and
compile a general scheme for data collection and description for
each site.

A specific database structured in various sections to manage

alphanumerical data, eventually coupled with GIS software to
manage geographical data, has been developed. After the database
Page 20

compilation, landslide data organised in such a homogeneous

manner are available for consultation. Data organisation is aimed at
collecting all the information related to each landslide in order to
allow an easy and fast comparison among:

• the geological framework (geology, structural setting, morphology,

water conditions, hydrology, climate at regional and local scales) in
which landslides have formed, where high risk conditions have been
created,

• the type and rate of movement, landslide identification, history,

morphometry, morphology, damage, etc.,

• the different investigations carried out to study each site, monitoring

activities and instrumentation planning,

• the modelling approaches to understand the landslide kinematics

and its possible evolution, either to forecast triggering/failure (if and
when) or run-out (where) after failure,

• the stability analyses carried out to define the site “hazard”,

• the vulnerability coefficients for goods and activities affected,

• the assessment of risk scenarios,

• the landslide hazard and risk reduction approaches.

In each of these sections, there is the possibility to insert and

describe a great amount of data, according to different degrees of
detail, using either look-up tables, numerical fields or textual fields.

The database structure adopted allows the following actions:

• to check the reliability of the available data, to individuate which
items should be considered in the subsequent phases, as much
attention has been paid to data quality (such as source type, scale,
date) and data format (for example: a distinction has been made for
in cartographic data between traditional maps, available only as
images, and numerical ones, sometimes created with GIS software);

• to compare landslide studies and approaches easily, underlining

differences, lack of information or redundancy in characterising
different landslide typologies in different countries;

• to analyse available data in detail, to interpret and process

information easily according to different representations or different
aims;

• to extract all the information necessary to prepare exhaustive

reports containing a concise description of the main features/factors
identified in the studied sites, including information on land use and
elements at risk (either in qualitative or quantitative terms, i.e.
number and characteristics), a preliminary vulnerability analysis, a
preliminary (often qualitative) evaluation of the expected scenarios,
an indication of adopted regulations, and a critical analysis of what
has been done and what could be done further.

3.5 HAZARD ANALYSIS

In order to assume a rigorous risk analysis which will supply useful

elements to decision-makers, a complete and detailed hazard
analysis must be carried out. With this aim, the IMIRILAND project
develops a methodology based on a multidisciplinary approach for
the hazard analysis, at the same time showcasing the valuable
contributions of several professional experts (engineering geologists,
geologists, geotechnical engineers.) The project has also benefited
from experiences from different social, cultural and political contexts
(Spanish, French, Italian, Swiss and Austrian). The different types of
analysis used in the present methodology are detailed below; each
one supplies elements of different weight aiming at the determination
of the various hazard components (Table 3.2).
Page 21

Table 3.2. Contribution of the various disciplines to the hazard

component evaluation.

Analysis tools Aim Know-how Contribution to

quantitative risk
analysis

Geological— To understand —Definition of —Basic outline

structural model the basic landslide geometry of the situation,
hierarchical and and discontinuity
geometrical roles. —Input data for
relationships geomechanical
between the —Definition of
main regional typology and models
structural lithology of the (triggering and
features (around materials involved. run-out).
site) and
geostructural —Understanding of
landslide the kinematic
discontinuities mechanisms,
(on site).

GeomorphologicalQualitative —Landslide typology —Outlines a first

model evaluation of the classification. definition of
typology, hazard scenarios,
geometry and —Spatial-temporal
mechanisms of identification of the —Adds
evolution of landslide evolution. information to the
unstable areas. geological-
—Characterisation of structural model,
the slope in its initial
and present stages, —Input data for
according to different geomechanical
dangers (sliding models
 mechanisms, mass (triggering and
qualitative behaviour, run-out),
dimensional
parameters,
qualitative
nontemporal
probability: a sort of
predisposition).

Historical analysis To collect —Information about —Outlines the

(data collection) information movement typology, landslide
about damages, occurrence, effects phenomenon,
reactivation and damages.
frequency, —Establishes a
general —Definition of the spatial-temporal
instability relationships between frequency.
evolution. instability causes and
related phenomena —Refines the
(e.g. rainfall and results obtained
landslide). through other
surveying or
monitoring
methods.

Monitoring To collect —Spatial-temporal Refinement and

investigation quantitative data representation of the calibration of
about landslide activity. geomorphological
displacements, and
affected areas, —Identification of geomechanical
trend of the the sensitivity of models.
movements. measured parameters
with respect to the
variation of boundary
conditions.

—Comparison of
different
 measurement
typologies.

Geomechanical To carry out —Failure mechanism Validation of the

modelling: slope stability representation. previous
triggering analyses, starting qualitative
from the —Sensitivity hypotheses
geometrical and analyses. (geological and
physical- geomorphological
mechanical —Back-analyses of approaches) from
characteristics of the recorded a mechanical
the involved movements. point of view,
mass (through with the help of
different displacement
numerical data.
methods).

Geomechanical To evaluate the —Post-failure The results permit

modelling: run- run-out areas behaviour of the rock the definition of
out and the intensity mass. the mobilised
of the mobilized mass distribution
rock mass. —Sensitive analyses. and the
phenomenon
—Back-analyses. magnitude (in
terms of mass
velocity or
energy) for
several scenarios.
Page 22

Figure 3.5. Methodology followed in studying the geological aspects

of the IMIRILAND landslides.
Page 23

3.5.1 Geological analysis

In the framework of IMIRILAND project, the first fundamental step in
the definition of hazard (Fig. 3.2) is the danger characterization,
which leads to a connotation of the problem analyzed either from a
geometric point of view or from the point of view of the qualitative
understanding of the possible failure mechanisms. This phase
contributes, moreover, to qualifying the susceptibility to instability, i.e.
the potential instability phenomena. In other words, the danger
characterisation supplies a first schematisation of the studied
phenomenon, aimed at orienting and supporting the numerical
modelling. In particular, the danger characterisation of large
landslides requires the consideration and processing of the following
items:

– geometry of the potentially moving mass: volume, area, depth etc.;

– geomorphological constraints;

– geostructural constraints;

– comments on the assumed mechanism of deformation at or before

failure, with its boundary conditions;

– relative and qualitative non-temporal probability of failure

(predisposition).

In order to treat the aspects mentioned in a rigorous way, a specific

approach is required, focusing proper attention not only on the
landslide body itself, but also on the local and regional geological
context. The local analyses provide essential information, but they
are not really useful for establishing relations between the “on site”
and “around site” geological configurations. The regional analyses
can give a knowledge of these relations, which are necessary even
for a tentative assessment of landslide hazard conditions, based on
the lateral extrapolation (for instance over an entire mountain slope)
of the recognised geological features. Geological analyses of
landslide sites cannot, thus, concern only the “on site”
geomechanical properties of rocks and fractures, but relatively large
surrounding areas (“around site”) should also be investigated to
allow the understanding of the basic geometrical relations among the
main structural units and discontinuities. In particular, the around site
analysis becomes a necessity when the on site rock mass is so
damaged as to prevent the reconstruction of the original structural
framework of the landslide. This will be compared with those
geomorphologic elements that are thought to be strictly controlled by
the structural setting of the bedrock.

This is an integrated methodological approach, where meso and

macroscale data, regional models and remote sensed data are
linked. This approach is summarized in a flow chart (Fig. 3.5) and
will be described in the following paragraphs.

3.5.1.1 “On site” analysis

Geometric model

This results from the local characterization of the geometry and

physical properties of discontinuity systems (type, number of
systems, orientation, persistence, type of termination and
crosscutting relations), in order to define which are the most
recurrent shapes and dimensions of rock blocks, probable
kinematics and failure mechanisms. Exhaustive data collection is not
always possible on landslide sites due to logistic difficulties in
sampling, paucity of safe rock masses etc. In this case,
representative sites will have to be chosen within the landslide body
to describe the fracture network. If representative zones cannot be
found, a synthesis of single observations has to be carried out to
understand the fracture network.

Morphostructural model
In recent years, investigations on large landslides have more
frequently concerned the whole slope area or even the entire valley
on which they are located, in order to discover common (ubiquitous)
elements that could have driven the gravitational instability.

A morphostructural model of the landslide is then defined, where the

correspondence between morphological and structural elements is
mainly sought based on the assumption (indeed very often verified,
in the Authors’ experience) that the morphological features of the
landslide are the
Page 24

surface expressions of the structural subsurface elements (such as

schistosity, faults and fractures, etc.) of the bedrock.

However, the morphostructural approach also takes into account the

geomorphologic context due to physical and chemical erosion
agents. This approach consists of:

• Photo-interpretation of aerial photos at different scales and/or

analyses of multi-temporal and multi-spectral satellite images, in
order to achieve a general knowledge of those morphological
features that are peculiar for each slope or valley studied (for
instance those related to deep rock slope deformation or glacial
erosion). Successively, more detailed interpretations (scales
1:10’000; 1:20’000) are carried out to investigate the landslide site
properly at a scale where peculiar features such as scarps, debris,
trenches and undulations can be discerned.

• Field work in order to verify the relations between the features

observed on aerial photos and the structural setting of the site, as
well as to improve the early work hypotheses with new field data.
This allows the definition of a morphostructural model of the
landslide, i.e. the location of the main boundaries, the definition of
landslide sectors and the definition of distinct kinematic behaviour for
each sector (zonation of the landslide).

The integration of geometric and morphostructural models allows the

definition of the local features and kinematic behavior of the
landslides.

3.5.1.2 Around site analysis

These allow a better definition of the local geological model (Fig. 3.5)
that refers not only to the landslide site, but also to the surroundings
(slope or valley) and which is mainly based on the knowledge of:
– shape and extension of morphological, lithological and structural
rock units. These must be defined, since differences in their
lithological properties induce different mechanical behavior of the
rock units. This is the case, for example, of the Rosone landslide
where white mylonitic schists seem to control the triggering of the
gravitational instability (see Chapter 5). Tectonic units can also be
defined to recognize sectors subjected to instability, on the basis of
their different fracture intensity. The landslides that have been
studied in the IMIRILAND project are, in fact, placed in regional
geological settings characterized by distinct tectonic units which
have controlled the gravitational instability in a specific way.

– fault and fracture kinematics and degree of structural maturity.

Peculiar features such as thickness, type of fault rock, persistence
and subsidiary fracture networks correspond to each kinematic
group of faults. The knowledge of these features provides
constraints for defining rock block shapes which can be compared
with the morphostructural features of the landslide body. Structural
data are also organised into genetically coherent groups (structural
associations1). These represent key elements to model the
geometry of faults and fractures. This procedure leads to the
identification of one (or more) basic shapes of rock units, each one
characterized by distinct cross-cutting relations of their bounding
surfaces. This geometry can be reproduced graphically or by
numerical modeling. An example of the application of basic rock unit
has been applied to the bedrock of the Cassas landslide to infer the
type of instability mechanism (see Chapter 4).

The local geological model has to be compared with more recent

understanding of the regional geological framework (regional
geological model) in order to verify if the local interpretations can be
placed in a wider sphere.

Extrapolation of local data to a larger scale (and vice versa) is also

allowed by using remote sensed lineaments (interpretation of
geometry of lineament: orientation, density distribution, statistical
analysis). The geometrical and hierarchical relations of the lineament
systems (geometric

1 the structural association is a “geological tool” for the correlation of

the different discontinuity elements with different geometry,
kinematics and size into genetically coherent groups (Vialon et al.,
1976; Davis and Reynolds, 1996).
Page 25

model of lineaments) can thus be compared (qualitatively and

quantitatively) to the architecture of the regional faults (Morelli, 2000;
Morelli & Piana, 2003). Only when this comparison shows good
similarity, can the lineament systems be used as geological
structures. The lineaments present the advantage of being more
homogeneous than field data, of being suitable for use to greater
advantage in an automatic and computerized way for statistical and
numerical analyses, and of being characterized by a number of
elements that are statistically more representative and refer to a
context of wide extension requiring relatively short processing times.
This methodology has been applied on the Ceppo Morelli and
Cassas landslides, where geometrical and hierarchical relationships,
observed at the outcrop scale, are also confirmed by the statistical
distribution and geometric characteristics of the lineament systems
at both the regional and slope scales. This correspondence allows
the extrapolation of the geometry and hierarchy of the structural
elements to different scales (see Chapters 4 and 7).

The integration of local and regional geological models, as well as

the geometric model of lineaments, allows the definition of tectonic
and mechanical units and their relationships to the main
geomorphological unit (slope, valley, etc.) to which the landslide
belongs. Consequently, the landslide can be subdivided into sectors,
each one characterised by distinct morphology, dimensions, fracture
density and geometry. The danger intensity could thus depend on:
the position of the sectors with respect to the slope’s shape and
drainage network, the lithostratigraphic setting, fracture density and
intensity, structural maturity and kinematics along the sliding and/or
failure surfaces.

Comparison between “on site” and “around site” data and

interpretations
The geometric and kinematic features of the landslide are compared
with the main morphological and structural features (tectonic and
mechanical units) of the slope, watershed or valley to which the
landslide belongs, in order to verify if the inferred geologic context, to
be related with the peculiar instability conditions within the landslide
body, is recurrent (or not) over the whole slope.

When these conditions are satisfied, it can be assumed that the “on
site” and “around site” interpretations are consistent. In this case,
there is the possibility of extrapolating the “on site” interpretations to
broader areas or vice-versa.

In case they are non-consistent, it can be suggested that the

gravitational instability could be triggered by peculiar features of the
landslide body, for example by individual faults, mechanically
weakened layers, geomorphological features or, in the case of a very
mature landslide where the kinematic evolution is no longer
controlled by pre-existent structural features.

Contribution to the quantitative hazard analysis

The aim of the geological and structural approach is to understand

the basic hierarchical and geometrical relationships between the
main regional structural features (around site) and faults and
fractures on the landslide site (on site). This understanding leads to
a zonation of the landslide body where each zone could have distinct
mechanical features and kinematic behaviour. The contribution to
quantitative risk analysis consists in providing constraints for
geomechanical models as well as a qualitative interpretation for
linking different kinds of data, such as geomorphological,
geophysical, geotechnical and historical. The aim of the integrated
geological and morphostructural approach is to evaluate, in a
qualitative way, the typology, geometry and evolution mechanisms of
the unstable area. This approach leads to a classification of landslide
typology, a spatial-temporal identification of the landslide evolution
and a characterization of the slope in its initial and present stages,
according to different aspects (sliding mechanisms, mass qualitative
behavior, dimensional parameters, relative probability of
occurrence). The contribution to quantitative risk analysis consists of
a preliminary definition of hazard scenarios and of inputs for the
geological-structural and geomechanical models (triggering and run-
out).

3.5.2 Historical analysis

Two basic sources of documents are available for the reconstruction
of past events. First, the facts which can be found in the archives of
public or religious authorities, as well as the synthetic analysis
available in some textbooks (Heim, 1932; Abele, 1974; Eisbacher &
Clague 1984). Then, aerial
Page 26

and ground photographs and ancient maps allow a detailed

interpretation of the evolution of the slope over the period separating
the two documents related to same area (Fig. 3.6).

Figure 3.6. Document dated 1706 mentioning the occurrence of a

landslide in Bertodasco during autumn 1705 in the area of the
Rosone site (Italy).

Ancient cadastral maps, defining the limits of the plots of land which
indicate the location of old houses, also help establish past rates of
movement, as in the case of the La Frasse landslide in Switzerland,
where ancient maps date back to 1768 (Bonnard, 1984).

The data obtained from archives are often overemphasised and

reinterpreted, especially if they are ancient. Furthermore, the reading
itself of archives requires specific practice to understand the context
and exact circumstances of the past disasters. At the same time,
such investigations cannot be made only by historians, as the
interpretation of the recorded stories requires a detailed
understanding of the landslide characteristics. Finally, the exact
dates of the event are not always correct, so that it is difficult to
correlate the gathered information with other events related to the
disaster.

The comparison of aerial photographs allows a very precise

identification and quantification of displacements, provided some
well-identified fixed points exist around the landslide. However, the
exact date and circumstances of the observed disaster cannot be
determined. Moreover, such information of sufficient quality has only
existed for about 50 years. In very particular situations, related to
recent events, some television archives are very useful to
understand the mechanism and to calibrate the modelling of the
trajectories. Finally, the interviews of people who have lived near the
disaster area and who have a good memory of major past events are
useful to obtain details of the landslide sequence, but the testimony
has to be thoroughly cross-checked with the in situ observations to
be sure that the information is reliable. As an example, in the case of
the Ceppo
Page 27

Morelli landslide (Valle Anzasca, Piemonte, Italy), historical analysis

of archives and recent field observations supplied the following
results:

• Rockfall: 1940, Oct. 1971, Apr. 1977, Oct. 2000, Jun. 2002

• Movement of the entire mass (or of a large part of it): 312, 843,
1816, 2000

The reliability of all of this information is good, even if it is based on

very ancient documents, as the affected valley was heavily used in
the Middle Ages.

3.5.3 Monitoring investigation to assess long term behavior

Only past information collected by several monitoring techniques
may be significant to assess the long term behaviour of large
landslides. Through the new measurement of old topographical
landmarks which have been surveyed many times since the
beginning of the twentieth century (at least in Switzerland), for
instance, it is possible to reconstruct the total movement of some
landslides, as well as the evolution of the displacements of various
points with time. The most prominent case, related to the Lumnez
landslide (Canton of Graubünden, Switzerland), in which the spires
of the churches in five villages have been surveyed regularly since
1887, has permitted the confirmation of the regularity of the
movements for most of the surveyed points for over than a century
(Bonnard & Noverraz, 2001). In the case of Sedrun (Switzerland), as
detailed in Chapter 8, the regular check of the topographic points,
first for maintenance objectives and then for safety reasons,
permitted the observation of a marked and progressive acceleration
from 1973 onwards.

The precision of the displacements obtained depends on the quality

of the initial determinations, as well as on the structure of the survey
net, but even for small velocities (<1 cm/year) measured over a
period of more than 50 years, the results are quite significant.

The ancient cadastral maps still available in archives, indicating the

limits of plots of land as well as the location of buildings, can also be
compared to current maps, as well as to recent field surveys in
landslide areas. This technique, used for the La Frasse landslide,
allowed the determination of total horizontal displacements between
1768 and 1861 and from 1861 to 1981. Such a procedure requires
that several fixed points can be identified on the cadastral maps;
these have to be digitised and adjusted one by one or through global
compensation, as they generally cover only a limited area.

The precision checked by the comparison of the relative position of

fixed points indicates that the maximum error encountered in XVIII
century documents is 2–3 m, whereas it reaches 1 m on maps of the
end of the XIX century. As the movements over more than two
centuries reach 25 m the zone of the village, the accuracy is quite
acceptable.

A final source of information is related to the periodic check of the

position of infrastructures or transportation systems, such as roads,
railway lines or posts for cable cars. These data are often very
significant because they refer to a complete transversal (e.g. road) or
longitudinal (e.g. cable car) profile of the landslide.

The long term movement data, as well as piezometric readings

(which are rarely available over a long period), are fundamental to
establish the input parameters for the mechanical modelling and its
calibration.

3.5.4 Geomechanical modelling: numerical triggering models

Numerical triggering models, when used in a hazard assessment
framework, are often supposed to allow the determination of the
occurrence probability in time related to a catastrophic landslide, on
a mechanical basis.
The probability of failure in time of an existing slope is nil, unless a
variation in time of the existing situation can be envisaged, for
instance, the decrease in time of shear strength parameters or the
increase in pore pressure in the slope. As such variations cannot be
predicted in a quantitative way, only a non time-related probability is
usually computed (see, for instance Einstein, 1988 and Scavia et al.,
1990).
Page 28

Unfortunately, the evaluation of such a probability is impossible when

large landslides are studied, as it depends on the knowledge of a
number of both geometrical and mechanical parameters, including
their variability and that of the boundary conditions (for instance pore
pressure distribution in the slope, initial state of stresses), which are
very approximately known in reality. For these reasons, numerical
triggering models are used in the framework of the IMIRILAND
project as valuable tools to verify, from a mechanical point of view,
the hypotheses made through the geostructural and
geomorphological analyses, and as well as the monitoring results,
about the mechanism and the volume of the landslide. As a
consequence, if the numerical triggering results do not match the
geological hypotheses, a new drafting of the geo-structural and
geomorphological models is required. If the numerical results are in
good agreement with the geological hypotheses, on the basis of the
triggering analyses, more information about the shape and the
dimensions of the potentially unstable volume can be obtained and
used as further input for the run-out analyses (see examples in
Chapters 5, 6, 7 and 9).

In order to obtain significant results from these models, it is

necessary to incorporate the main geo-structural data precisely in
the construction of the model, to choose to work in 2D or 3D
conditions according to the scope of the analysis and to refer to a
discontinuous or equivalent continuum approach, as appropriate.

Numerical triggering models can indeed represent the rock slope as:

1. a discontinuous medium, where localized discontinuities (such as

joints or interfaces) are introduced as specific elements into the
geometrical model, allowing them to be taken into account for the
analysis;
2. an equivalent continuum model where the discontinuous medium
is substituted for the analysis by a continuous one, the material
properties of which are modified, so that its behavior is equivalent to
the discontinuous, real medium.

The choice of one of these methods depends on the size of the

structure in relation to spacing, orientation and strength of the
discontinuities and on the imposed stress level.

The main problem with the discontinuous approach is to determine

the location and geometry of the natural discontinuities; the main
problem with the continuum equivalent approach is the evaluation of
the mechanical characteristics of the rock mass, which cannot be
determined in the laboratory.

The most widely used numerical methods are the limit equilibrium
method (LEM), the finite element method (FEM), the boundary
element method (BEM) and the finite difference method (FDM).

Limit equilibrium methods do not allow the evaluation of stress and

strain in the slope, so they are not able to reproduce the crucial role
played by deformability in large landslides. Finite element methods
and finite difference methods are very suitable when the slope is
made of different materials, but the time of computation can be
cumbersome in 3D conditions. Boundary element methods are not
convenient when more than one material must be taken into
account, but are very useful in 3D conditions. In the IMIRILAND
project, three numerical codes were applied to the study of the
triggering in the selected sites: DRAC (FEM), MAP3D (BEM) and
FLAC3D (FDM).

The combined use of these 3 methods has permitted carrying out

analyses in both 2D and 3D conditions, on the basis of both
discontinuous and continuous equivalent approaches and adopting
elastic, viscous and plastic constitutive laws.
A brief description of the three numerical codes is presented in the
following sections.

3.5.4.1 Code DRAC

The computer package DRAC (Prat et al., 1993) is an analysis
system based on a finite element program designed for the analysis
of rock mechanics and other geotechnical problems.

It is a very flexible package, which can be used with a great number

of boundary conditions and different types of materials. The program
has an open structured architecture, allowing new modules serving
both future engineering needs and new computational techniques.
Page 29

DRAC can be used to solve two and three-dimensional problems.

The program can perform linear and non-linear analyses, for which
several techniques, such as line-search, are available, as well as
finite displacement analyses. Time-dependent analyses are also
possible. The element library has been built with the needs of the
problems specific to rock mechanics in mind, for instance, with the
development of a zero thickness joint element to model
discontinuities.

This interface model, a key constitutive model for joints, interfaces

and other discontinuities, is a simple but general model for
normal/shear cracking based on fracture mechanics recently
developed at UPC (Carol et al., 1997). It is defined in terms of the
normal and shear stresses on an average joint plane, and the
corresponding normal and shear relative displacements across the
discontinuity.

Besides the finite element program itself, there are a number of

auxiliary modules that complete the package: these are the pre- and
post-processing programs and the libraries (mathematical, finite
elements, material models, and solver). An external module allows
the analysis of water flow problems within the rock mass. The
package includes a post-processing graphics program specifically
designed to handle the output of the finite element program in two
and three dimensions, including a module to estimate the safety
factor for instabilities due to the presence of discontinuities in the
rock mass.

3.5.4.2 Code Map3D

The commercial code Map3D is based on BEM and in particular on
the Displacement Discontinuity Method (DDM), an indirect boundary
element methodology. Map3D, developed and supported by Mine
Modeling Pty Ltd. (Victoria, Australia) was first devoted to simulate
underground excavations and mining sequences, but has recently
been applied to slope stability analyses in 3D conditions.

The code allows the easy simulation of natural discontinuities in the

rock slope. Requiring the discretisation of the problem boundary, it
lends itself very well to analysis in 3D conditions, even when a visco-
elastic behavior of the discontinuities is adopted.

3.5.4.3 Code Flac3D

FLAC3D (Fast Lagrangian Analysis of Continua in 3 Dimensions) is
a powerful 3D continuum code for modeling soil, rock and structural
behavior, and can be used interactively or in a batch mode (ITASCA,
1997). It is a general analysis and design tool for geotechnical, civil
and mining engineers. FLAC3D can be applied to a broad range of
problems in engineering studies. The explicit finite difference
formulation of the codes makes it ideally suited for modeling
geomechanical problems that consist of several stages, such as
sequential excavation, backfilling and loading. The formulation can
accommodate large displacements and strains and non-linear
material behavior, even if yield or failure occurs over a large area or
if total collapse occurs. FLAC3D uses a mixed discretisation scheme
to provide accurate modeling of plastic collapse and flow. Materials
are represented by polyhedral elements within a three-dimensional
grid that is adjusted by the user to fit the shape of the object to be
modeled. Built-in primitive shapes allow the generation of a variety of
complex geometries. FLAC3D has a built-in macro programming
language called FISH that allows users to customize their analyses
to suit their needs. Constitutive models may be defined using FISH,
as can loading patterns, servo-control of test conditions, and grid
generation sequences.

3.5.5 Geomechanical modelling: numerical run out models

In the previous chapter, it was shown how a volume and a qualitative
probability of failure can be assessed for a certain danger through
geomorphological, geo-structural, historical and triggering numerical
analyses.
In order to provide a scenario for the danger under consideration, the
analysis of the run-out of the rock mass volume down to the valley
must be carried out. The results of such analyses are the area of
propagation of the landslide and its intensity (see examples in
Chapters 5, 6, 7 and 9).
Page 30

As a function of the triggering mechanism and of the entity of the

involved rock mass volume, rock mass movements can be
subdivided into three broad categories:

– Rockfalls, blocks of rock fall freely from cliffs or mountainsides and

accumulate at the base in talus slopes. Chemical and physical
weathering plays a large role in loosening the blocks. Freeze-thaw
cycles separate blocks along joints and the expansion of the water
when it freezes to ice is enough to trigger the fall.

– Rockslides, large masses of bedrock slide as a coherent material

along downward sloping joint planes or bedding surfaces.

– Rock avalanches, large masses of broken rock (millions of cubic

meters) flow as a viscous fluid at high velocities (tens to hundreds of
km/h).

In rockfall cases, the run-out can be simulated through the

computation of the trajectory of a single boulder, jumping and rolling
down to the valley bottom. Several non-interacting boulders can be
considered.

Many computer codes are used for the evaluation of boulder

trajectories, based on various simplifications of the phenomenon
(see, for instance, Scioldo, 1999).

As the IMIRILAND project mainly refers to the study of large mass

movements, the run-out methodologies described in the following
present a possible approach for the analysis of rockslide and rock
avalanche phenomena. Attention is particularly given to rock
avalanche behavior, where the rock mass can be simulated as a
massive, flow-like motion of interacting fragmented rock (Hungr,
2001).

3.5.5.1 Classification of the run out models

Run out prediction methods can be grouped into two broad classes
(Hungr, 2001):

– Empirical methods

– Analytical methods

Empirical methods provide a quick estimate of the total run-out

distance. They are based on simple correlations between the volume
of the mass involved in the movement and the geometrical
parameters of the slope. Empirical laws are then worked out through
back-analysis on documented cases. As a function of the aspect
taken into consideration, these methods can be further subdivided in
relation to:

– Path profile

– Deposition area

– Mass balance

Analytical methods should be able to reconstruct the motion of every

portion of the landslide mass in space and time, depending on path
topography, characteristics of the material, both within the sliding
mass and along the path, internal stresses within the mass and
entrainment or deposition of material. We may distinguish among:

– Lumped mass methods

– Continuum mechanics methods

– Granular methods

Lumped mass models are unable to account for internal deformation:

while they may provide a reasonable approximation for the
movement of the center of gravity of the landslide, they cannot
simulate the motion of the flow front, which is often the most
important aspect in a run-out analysis (Evans et al., 1994).

Continuum mechanics models use techniques developed for the

analysis of fluid flow in open channels. There are, however,
important differences between fluids and earth materials, even when
the latter are imbued with water and highly disturbed. It is necessary
to find an “equivalent” fluid, whose rheological properties are such
that the bulk behavior of the flowing body is able to simulate the
expected bulk behavior of the prototype landslide.

Continuum mechanics analyses adopt an Eulerian or a Lagrangian

approach to the study of landslides. The Eulerian approach
correlates the calculation to a reference frame fixed in space
Page 31

and is particularly suitable for slow movements. Given the highly

unsteady nature of landslide movements, it is preferable to use a
moving Lagrangian reference frame (Potter, 1972).

Granular models consider the moving mass as the result of the

interaction of single particles.

To model the possible run-out behavior of a landslide phenomenon,

two methodologies have been applied:

• The Particle Flow Code (PFC) from the ITASCA (1999a and 1999b)
Consulting Group in cooperation with FLAC (Itasca) as the Ball Wall
model and stand alone as the All Ball model;

• The computer code DAN (integrated with the rockfall program

ROTOMAP—Geo & Soft International, 1999 and 2003) based on a
continuum mechanics approach.

3.5.5.2 The Particle Flow approach

FLAC is a continuum mechanics finite difference code that, used
with the shear strength reduction technique, allows the estimation of
the failure mechanism, the distribution of displacements and the
critical parameters of the slope in three dimensions. The deduced
failure surface of the slope is the basis for the run-out analyses
carried out with PFC and DAN.

PFC is also a finite difference code which models the mass as an

assemblage of circular (2D) and spherical (3D) elements among
which exist bonds and friction; after the breaking of the bonds,
propagation is caused. To generate displacements and rotations, the
laws of motion take into account both the contacts of the particles
and the boundary conditions. In general, this is not a rockfall
program but it can be adjusted appropriately in order to generate the
run-out mechanisms in a more detailed way.
The computer program PFC2D and PFC3D have been used as an
All Ball and a Ball Wall model; in the PFC Ball Wall model the
detached rock mass is modeled by balls, while the bedrock is
simulated by linear (2D) and planar (3D) elements. Therefore, in the
Ball Wall model, an estimate of the failure mechanism is needed as
an input parameter. As previously explained, these estimations were
carried out with the aid of FLAC.

In contrast to the Ball Wall model, in the All Ball model, the relatively
stationary bedrock, the failure mechanism of the slope and the
detachment mechanism are modeled by balls. This allows the all in
one calculation of failure mechanisms, detachment and run-out. For
calculating the failure mechanism, however, quite a demanding
calibration of materials is necessary; this is particularly due to the
fact that the modeling of the bedrock by balls determines a rather
rough surface. When comparing the volume of the detached rock
with FLAC, both methods correspond closely.

One limitation of PFC is due to the fact that it is not able to model the
influence of water (e.g. pore pressure). On the other hand, as the
balls are supposed to model simple rock blocks, and as the number
of balls is limited by the computation capacity, a certain restriction
exists in the modeling of large landslides of dozens of millions of m3.
Finally, the All Ball model allows an all in one calculation but the
computational effort required obliges the modeling of the detached
mass with less and bigger balls than those used in the Ball Wall
model.

3.5.5.3 The DAN approach

The dynamic analysis model DAN (Hungr, 1995), was especially
developed to simulate the motion of flows, flow-like slides and
avalanches. In particular, it implements a one-dimensional
Lagrangian solution of the equations of motion, formulated in terms
of an open rheological kernel, in which the type of material (plastic,
frictional, Newtonian laminar, turbulent, Bingham, Coulomb viscous,
Voellmy) can vary either along the path or within the sliding mass.
The sliding mass is represented by means of a number of blocks
(mass blocks) in contact with one another that can deform freely and
retain fixed volumes of material in their downward path.

The main data used as inputs for the DAN code are the slope profile
and the top profile of the initial mass. In particular, to perform a
three-dimensional analysis, the width of the path must be determined
at each path profile point by assuming that the entire envelope
considered is enveloped by the moving mass.

Hungr’s approach to width definition stems from a visual analysis of

slope morphology and hence the lateral boundaries of the area
involved are defined on the basis of experience with real cases.
Page 32

In the proposed methodology, this evaluation has been replaced with

an analysis conducted with ROTOMAP (Geo & Soft International,
1999), a three-dimensional rockfall program which makes it possible
to define envelopes as a function of the trajectories of the rock
blocks released from the toe of the unstable area.

The final outputs are: run-out distance and run-up on the opposite
slope, landslide velocity and expiring time, limits of propagation and
depth of the displaced mass once it stops.

The main limitation of the Hungr model is due to the fact that it is
approximate, as it involves the reduction of a complex and
heterogeneous three-dimensional problem into an extremely simple
formulation. The simplicity of the model is, however, an advantage in
making possible an immediate and rapid numerical simulation of real
cases.

3.5.6 Occurrence probability assessment

Risk analysis requires the assessment of occurrence probability in
time; however time is of great uncertainty in landslide hazard/risk
analysis. Geomechanical models do not help in this case.

Occurrence probability, which can be defined as the chance or

probability that a landslide will occur, can be expressed in relative
(qualitative) or probabilistic (quantitative) terms. A rigorous,
quantitative, procedure could include the application of mechanical-
probabilistic methods, taking into account the uncertainty of all of the
geometrical and mechanical parameters (slope geometry, shear
strength, piezometric pressure, etc.). In the case of large landslides,
however, a similar procedure cannot be applied because the
uncertainty in the definition of the parameters involved is too large
and it is not possible to define any reliable variability for them; finally,
the number of events is not sufficient to obtain extreme values by
statistical extrapolation.
An alternative procedure takes into account the relationship between
the intensity of triggering events, such as rainfall or earthquakes, and
the landslide occurrence, through the use of empirical correlations or
more sophisticated methods (i.e. neural network methods in
Mayoraz et al., 1997). In any case, it is still difficult to relate a given
magnitude of a landslide phenomenon to a given intensity of the
triggering factor. In particular, in the case of rainfall-induced slides,
the real cause is often a combination of long antecedent rainfall and
intense spells of rain. Often, the lack of data makes this approach
inapplicable. Furthermore, from a rigorous point of view, it is not
always possible to determine a statistically-computed probability
value for most large landslides due to their non-repetitive behaviour.

The probability of sliding can be estimated using semi-quantitative

methods related to the assessment of the historic record of sliding in
conjunction with detailed geological and geomorphological
characterisation of the investigated area and considering the past
experience of similar events.

If several historical records exist on different instability process

activities (different dangers) and if, in the studied area, a non-
negligible hazard exists, the historical approach seems much more
suitable to obtain information related to the frequency of failure. In its
simplest form, this method consists in recording any type of sliding
which occurred in the past and in its detailed description (type,
volume and area involved, possible localization, etc.). Generally a
long representative period of recording is needed in order to
consider all phenomena types and a sufficient amount of data.
Historical approach results can be quantified using analyses of aerial
photographs.

A systematic multi-temporal interpretation of aerial photographs (it is

often possible to obtain photographs taken for the same area 10, 20
or 50 years earlier, allowing a stereoscopic viewing) indeed permits
an analysis of the factors leading to instability. It allows a subdivision
of the investigated area into “homogeneous” zones by using
interrelationships among geology and geomorphology. Morphometric
multi-temporal analyses of landslides, combined with geological field
investigation and surface observation, provides:

– surface and temporal characterisation of instability processes;

– quantitative movement displacement evaluation (between two

different dated stereo pairs);

– temporal activity identification and interpretation of historical slope

evolution;

– possible future evolution;

Page 33

– semi-quantitative volume estimation;

– possible landslide-prone areas;

– principal sliding mechanism interpretation;

– reasonable semi-quantitative estimation of the probability of sliding

and occurrence frequency for some repetitive instability processes.

This mixed method (historical and geomorphological approaches)

can be a useful way of estimating the annual average probability of
sliding, despite its limited accuracy. It also depends on the skill and
experience of the persons carrying out the studies.

For instance, as mentioned previously (see Chapter 7 concerning

the case of the Ceppo Morelli landslide/Valle Anzasca, Piedmont,
Italy), an historical analysis supplied the following results:

– Rockfall: 1940, Oct. 1971, Apr. 1977, Oct. 2000, Jun. 2002

– Movement of the entire mass (or of a large part of it): 312, 843,
1816, 2000.

At this point, it is possible to link together each danger, characterised

by geologic-structural and morpho-dynamic features and an
occurrence probability class. An occurrence range of time has been
obtained for each of the three scenarios (danger 1: rock fall, danger
2: movement of approx. 1×106 m3, danger 3: movement of the
entire mass, approx. 5×106 m3) considering the minimum and
maximum time intervals between two consecutive recorded events.
On the basis of their mean value, a frequency has then been
calculated in terms of events per year (Table 3.3). The computed
values are to be considered as relative values.
Table 3.3. Example of computation of frequency on the basis of
historical information as applicable to the Ceppo Morelli landslide.

Danger and Recorded Occurrence range Average Frequency

scenario No events (years) (years) (event/year)

1 Rockfall 1940 0–30 15 1/15=0,066

Oct. 1971 (2–31)

Apr. 1977

Oct. 2000

Jun. 2002

312

3 Rock 843 200–1000 600 1/600=0,0016

avalanche (B)
1816 (184–973)
5×106 m3
2000

2 Rock (assumed 115 1/115=0,009

avalanche (A) intermediate value)

1×106 m3 30–200

3.5.7 Relevance and choice of significant scenarios

The main characteristics of large landslides, such as those
considered in the IMIRILAND project, is that the probability of a
critical global movement of the whole mass is very low, whereas the
consequences may be catastrophic. The majority of large landslides
are characterized by a complex sort of interacting specific
mechanisms, as well as by preparatory and triggering causes. Any
mechanism can evolve in a specific way and, therefore, could
produce different potential dangers and, consequently, different
scenarios. It is, thus, indispensable to assess several scenarios, and
not only the worst case, with their potential consequences, so as to
quantify the respective and indirect risk components in comparable
terms, and to propose preventive measures on a wide scale so that
such an extreme event does not hit the site by surprise.

On the basis of the multidisciplinary application of the various

methods mentioned above, a synthetic model of the landslide
behaviour has to be determined for each danger identification. This
process requires a confrontation and critical assessment of results
that may not be in complete
Page 34

agreement. An assessment of the reliability of all of the results

produced must be carried out, so as to give more weight to the more
significant data.

The object of such a scenario selection is to qualify clearly the

different danger situations leading to different consequences, so that
each scenario represents an intrinsic phenomenon, for instance
rockfall scenarios vs. rock avalanche.

For each scenario, the probability of occurrence has to be set up on

the basis of historical information or past movements. This selection
process is certainly the most delicate operation within the risk
analysis. It can be related to the probability classes that are
proposed in several documents or recommendations set up in
different countries (Switzerland: OFAT, OFEE & OFEFP, 1997;
France: PPR, 1995). For example, in Switzerland, high probability
means a 1–30 year return period, medium probability 30–100, low
probability 100–300.

The interpretation of historical data gives a justification for the

selection of return periods, even though it is not always certain that
the reported events have the same magnitude. For very low
probabilities, for instance 1/1000 years, the probability of the
scenario has to be considered as more qualitative than quantitative.

Finally, there is a unique correspondence between a specific danger

situation and the related scenario. Similar mechanisms of different
magnitude will induce different scenarios.

3.5.8 Evaluation of the contribution of the various disciplines to

hazard analysis
On the basis of the experience gained by IMIRILAND project and the
knowledge of the partners of this project, the applicability of the
different disciplines proposed can be rated according to several
considered aspects: intensity definition, definition of the area
affected, relative probability of failure, occurrence probability
evaluation. Table 3.4 points out that all disciplines contribute to the

Table 3.4. Hazard analysis approaches: contribution to hazard

component evaluation.

Analysis type Contribution Contribution Contribution to Contribution to

to intensity to def occurrence relative failure
definition inition of probability probability
area affected evaluation (non-temporal
(temporal concept)
concept)

Geological— • • • •••
structural

Geomorphological•• •• • •••

Historical (data •• •• ••• •

collection)

Monitoring •• — • •
investigation

Geomechanical •• • — ••
modeling:
triggering

Geomechanical •••• •••• — —

modeling: run-out

intensity definition, but an appropriate geomechanical modelling of

the run-out does provide outstanding quantitative results. The
historical analysis definitely contributes in an essential way to
probability evaluations. Finally, run-out models are indispensable to
determine the spatial impact of the considered scenarios. It must be
recalled that no significant results can be obtained without
comprehensive geological and geomorphological studies.

3.6 CONSEQUENCE ANALYSIS

3.6.1 Introduction
Once the hazard scenarios have been identified, each one being
characterised by an area in which the distribution of the process
intensity is known, as well as the corresponding probability of
Page 35

occurrence, it is possible to proceed with the true and real

consequence analysis. The necessary steps to allow the definition of
consequences are the following:

• identification of elements at risk,

• evaluation of the value of elements at risk,

• vulnerability evaluation,

• consequences (element at risk value×vulnerability).

3.6.2 Exposed elements: identification and value

Superposing the involved area identified by the run out/morpho-
dynamic model for each scenario with territorial data (land use
according to town and country planning, population distribution
according to statistical demographic studies, strategic elements
defined by e.g. transport administration or electric power supplier,
etc.), elements at risk are recognised. The elements at risk are
classified in different categories (see below), and counted. The
actual state is either taken into account in the classification or
assessed through a depreciation coefficient expressing the
difference of value between the present and initial cost estimate of
each exposed object. The necessary inventory of the exposed
elements also considers potentially developed zones and buildings
under construction.

The elements at risk can be monetarily quantified or globally

assessed in order to obtain a consequence (expected impact)
calculation.

For the value computation, several approaches exist (Crosta et al.,

2001):

1. Computation of a specific value for the individual elements.

2. Use of a utility function.

3. Use of empirical formulas.

4. Qualitative assessment of the global value for a certain area.

3.6.2.1 Computation of a specific value for the individual elements

In such an approach, the value of the elements at risk is expressed
as the sum of the intrinsic value of each separate element. In most of
the published cases, a distinction is made between the material
goods/economic activities and human life. The evaluation of human
life is, in each case, a difficult problem; as a simple guide Table 3.5,
drawn from the project PER (DRM, 1990), is given

Table 3.5. Scale of relative costs of human victims exposed to risk

considered in France (PER project, DRM, 1990).

Human victims Relative cost

Dead persons 1

Wounded persons 2–3

Homeless persons 0.2–1

and puts forward the paradoxical aspect in which wounded persons

represent a higher value than dead persons because of the high
social costs related to medical care and rehabilitation of
handicapped persons.

In the assessment of the real value of buildings, the reference prices

needed, depending on the local situations, have been taken into
account (see, for example, the Sedrun site in Chapter 8).

3.6.2.2 Use of utility function

In this approach, the elements at risk are characterised by a utility
function u(x) in which the social or individual cost induced by the loss
of a specific element is expressed as a function and not as a single
value. Therefore, for each element, it is necessary to specify a
variation of the social or
Page 36

individual utility (e.g. linear, logarithmic, exponential variation, etc.).

The use of the utility functions imply a better flexibility and adaptation
to the complex conditions which determine the global cost of the
losses. In Figure 3.7, the utility functions used to express, first, the
cost of the interruption of a road depending on the time in hours and,
then, the cost related to the number of wounded persons, are
presented; in the first case, the variation is logarithmic because after
a certain number of hours of interruption of the road, alternative
routes are identified and/or reestablished; in the second case, the
variation is exponential because with the increasing number of
wounded persons the cost of rescue and medical care tends to
increase dramatically.

Figure 3.7. (a) Variation of cost of traffic hindrance as a function of

time; (b) Nonlinear variation of cost of rescue and medical attention
as a function of the number of wounded persons (Crosta et al, 2001).

3.6.2.3 Use of empirical formulas

The third approach foresees the use of empirical formulas for the
computation of a global quantitative value of the elements and risk.
As an example, the following formula is proposed (Del Prete et al.,
1992):

W=[Rm(Mm−Em)]*Nab+Ned*Ced+Cstr+Cmorf

Rm=average income of the inhabitants of the area;

Mm=average age of death of the inhabitants within the exposed
area;

Em=average age of the inhabitants within the exposed area;

Nab=number of inhabitants within the exposed area;

Ned=number of buildings present in the area;

Ced=average cost of existing buildings;

Cstr=cost of existing structures and infrastructures;

Cmorf=cost of morphological changes induced.

3.6.2.4 Qualitative assessment of the global value for a certain area

Such an approach appears to be somehow useful for particularly
vast areas in which it is complicated to analyse the value of
individual elements. In the methodology proposed here, this method
of assessment is selected by proposing a scale of values based on a
subdivision of the exposed areas according to land use and/or to the
categories determined in the local management plan.

The parametric value appraisal of the elements at risk, generally, is

carried out according to separate classifications (material assets and
persons), therefore, calculating separate risk both for the assets and
for the population. Moreover, in this methodology, environmental and
economic values (related to interruption of economic activity) are
individuated. In order to simplify the evaluation of the value elements
at risk, it is sometimes possible to determine a relative value scale
using several indices (in order to use indices correctly, a detailed
analysis is necessary).

In the following tables (Tables 3.6 and 3.7), relative values related to
some elements at risk categories are shown. For each element at
risk, a “relative value index” for assets, economic activity interruption
and environment has been indicated. The human lives relative
“value” has been applied on the basis of the persons involved (the
ranges can vary according to the area studied).
Page 37

Table 3.6. Assessment of relative values for exposed objects.

Element at risk Assets Relative value of Environmental

relative interruption of relative value
value economic activity

Densely built modern cities (with 4 4 1

high-rise buildings)

Historical city centers 4 2 4

Residential areas 4 1 1

Productive or industrial areas 4 4 1

Strategic services and facilities 4 4 2

Extra-municipal infrastructures 4 4 2
and plants

Valuable buildings or valuable 3 1 3

rural centers (historical,
architectural, artistic and/or
cultural value)

Tourist accommodation— 3 3 1
buildings

Valuable environmental areas— 3 2 2

buildings

Local infrastructures and plants 3 3 2

Transportation facilities (highway 3 4 1

and international railways with
operating services)

Strategic lifelines 3 4 1

Heavy traffic or strategic roads 3 4 1

Tourist accommodation—camp 2 2 1
grounds

Valuable environmental areas—no 2 1 4

buildings

Parks, sport and parking areas 2 1 3

Local services 2 2 1

Secondary roads 1 2 1

Secondary lifelines 1 2 1

Rural areas—farming and 1 2 2

domestic animals

Forests (private and public 1 1 3

properties)

Special risk objects

hospitals 3 4 1

schools 2 4 1

garbage dumps 2 3 4

Table 3.7. Assessment of relative values for exposed persons.

Human lives (number of persons involved) Relative value of human lives

0÷1 1

2÷9 2

10÷19 3

>20 4

When elements (assets and economic activities) can be monetarily

evaluated, the indices represent a relative cost. In these tables, the
relative values have been evaluated in an arbitrary way (using
indices 1 to 4), and, therefore, have only a relative meaning in each
of the considered categories. In one of the examples presented
below (see Chapter 8), the assessment of the value of all buildings
has been carried out in monetary terms, using average market
values of similar objects as a basis.

3.6.3 Vulnerability evaluation

The degree of loss to a given element at risk, or set of such
elements, resulting from the occurrence of a natural phenomenon of
a given magnitude, is defined by a vulnerability coefficient (V).
Vulnerability is usually expressed in relative terms as defined in the
quotation below, using words
Page 38

such as “no damage”, “some damage”, “major damage” and “total

loss”, or by a numerical scale between 0 (no damage, 0%) and 1
(total loss, 100%).

“Although the state of the art for identifying the elements at risk and their
characteristics is relatively well developed, the state of the art for the
assessment of vulnerability is in general relatively primitive.

The vulnerability is affected by the nature of affected facilities, whether it is

uphill, on, or downhill of the landslide, and the nature of the elements at
risk. The velocity of movement also affects the vulnerability, with higher
velocities usually leading to greater vulnerability. This can lead to different
degrees of damage on or in the travel path of a landslide. For structures
and persons onto which a landslide travels, the greater the depth of slide
material, generally the greater the damage and vulnerability.

For structures, the assessment of damage and, hence, vulnerability, depends

on modelling the landslide-structure interaction. This is relatively well
documented for rockfalls (where structures have been designed to withstand
impacts), and to a lesser extent for debris flows and extremely slow and
very slow movements. For higher velocity slides, spreads and topples, there
is very little guidance available. Often, it is necessary to use judgement. In
any case, the notion of direct and indirect impacts has to be defined first”
(IUGS/WGL/CRA, 1997).

As previously pointed out, the adopted methodology foresees the

assessment of the different vulnerability factors, which include
physical, social, environmental and economic components. In the
investigation work carried out, two approaches have been
considered for the assessment of the exposed objects, applying
either monetary values or coefficients expressing their relative
importance. The amount of the physical vulnerability coefficients
considered takes into account the intensity of the hazard and the
quality of the constructions.
3.6.3.1 Physical vulnerability
This term expresses the degree of loss or potential damage to a
given element or set of elements at risk when they are affected by
the behaviour of an unstable mass, which is qualified by a given
magnitude or intensity. Such an impact must be assessed first in
terms of structural failure, by the analysis of the effect of differential
settlements or movements on a structure crossing a scarp, or by the
impact of rock blocks hitting a building. It also has to be evaluated in
terms of operational failure, for instance, when the tilting of a house
or a road exceeds an acceptable value, even though no cracks are
observed. The physical damage also depends on the quality of the
materials used in the building or the infrastructures under study, as
well as on the maintenance of these structures, especially when
wooden buildings are concerned.

No theory has been elaborated up until now to model the damaging

effects due to all types of landslides, with the exception of the
sudden impact of a block on a wall; however, the number of variable
parameters for this case (velocity, block mass, impact angle, position
of the wall impact point, potential deformation of the whole structure,
detailed geometry of the wall, strength of the material) does not allow
the obtaining of a significant result in terms of global risk analysis for
a large landslide area. On the other hand, in most situations where
limited damage has occurred, no systematic monitoring procedure
has been carried out, which would have allowed a proper evaluation
of the effect of the landslide movements.

The main criteria to determine the value of the physical vulnerability

coefficients are, in decreasing order:

– the phenomenon intensity (velocity for slides, energy for rockfalls);

– the types and function of the structure, as far as its strength is

concerned;

– the state of maintenance and capacity of deformation.

Page 39

The classification of these criteria may vary when networks are

considered (electric lines, water pipes, sewage ducts) instead of
buildings and roads.

3.6.3.2 Social vulnerability

This term expresses the rate of impact due to a landslide on the
exposed population. In many rockfall cases, even of limited intensity,
this rate is 100% as it may cause death. Serious wounds leading to a
permanent important handicap are also assessed as nearly 100%,
because the long-term costs they induce for the society are high.
Temporary wounds (such as a broken leg) are considered, on the
contrary, as a minor vulnerability, as the persons can recover after a
short time.

The coefficient of social vulnerability also includes the psychological

consequences of the loss of a home, as it is often observed as a
more difficult situation for the victims to lose their effective roots than
to be wounded. Even the fact of being provisionally evacuated, for a
certain duration, constitutes an impact on the population that deals
with social vulnerability.

Thus, the main criteria to determine the value of the social

vulnerability coefficients are, in decreasing order:

– the phenomenon intensity (which is in relation to the warning time);

– the population sensitivity, depending on its age and capacity to

anticipate a landslide;

– the capacity of understanding the phenomenon and to move away

from the exposed zone.

3.6.3.3 Environmental vulnerability

In order to face all the direct impacts of a landslide, it is necessary to
add a third category after the physical and social domains, which
deals with the natural environment. Indeed, it is often the first “target”
of landslides, such as livestock, as the slopes where they occur are
generally not densely populated or crossed by major infrastructures.
On the other hand, the damage to these natural components cannot
be evaluated in monetary terms, especially as far as forests (which
now have nearly no merchant value), wild animals and rare plants
are concerned. For the forests, this impact may be important as
several zones are considered as protective forests against rockfall
and avalanches, so that their potential value has to be assessed
through the analysis of their global function (productive, protective,
recreational).

As far as impacts on water resources and flow conditions in rivers

are concerned, it is also important to analyse the potential loss of
springs due to a rockfall, even if they are not yet equipped for water
consumption, and the potential loss of fish if the river flow is
disturbed by a sliding mass obstructing its course.

Thus, the main criteria for determining the value of the environmental
vulnerability coefficients are:

– the intensity of the phenomenon, in relation to its dismpting effect

on nature,

– the function of the forest or of the endangered animal and plant

species,

– the sensitivity and rarity of these species.

3.6.3.4 Economic vulnerability

Beyond the potential destruction of assets on or near landslides,
such phenomena may cause indirect economic impacts when they
block a road or a railway line, destroy an electric line or a water pipe,
or dam a valley causing a lake that represents a danger
downstream, so that the economic activity in the valley below must
be stopped or reduced, but perhaps without final destructive
consequences.
It is important to realise that, in most real situations studied, the
indirect economic impacts are much larger than the direct impacts.
As an example, for the avalanches of the 1999–2000 winter in
Switzerland, it was assessed that the indirect impacts were 4 to 5
times higher than the direct ones. In the case of the La Josefina
landslide in Ecuador, the threat to the Paute hydroelectric plant
located downhill, which might have been seriously affected by the
overflow of the dam caused by the landslide in 1993, would have
represented the loss of 70% of all the electrical energy production in
the country; happily only limited damage was registered when the
flow reaching l0’000 m3/s filled the previously emptied reservoir, but
this flood brought the equivalent of 2 years of silting in
Page 40

the reservoir, which meant several hundreds of thousands of US

dollars for its pumping and evacuation in order to maintain the
regulation capacity of the lake (see Chapter 10.2, below).

The economic impact on a road blocked by a landslide depends, of

course, on the average expected traffic, but also on the existence of
an alternative route. For example, when the Ceppo Morelli rockfall
(Valle Anzasca, Italy) reached the access road to the Macugnaga
resort station, it impeded any traffic to and from this important station
for several weeks; an alternative road now exists on the other side of
the valley, but its position does not guarantee its operation in case of
large rockfall (see Chapter 7, below).

In other cases, an alternative road exists, but requires a very long

bypass of several dozens of kilometres, as it is the case for the La
Frasse landslide, in Switzerland, so that a large part of the tourist
traffic coming in the area for one day will give up and visit other
resorts, especially in winter and spring. If a threat of rockfall is known
and publicised, many people will not drive to the concerned area,
even if the risk is effectively limited. As far as transportation of goods
is concerned, some may be stockpiled for a certain time, such as
logs of wood, but other products, such as milk or cheese, need to be
transported continuously to the factories or warehouses, so that
there is a direct economic loss in case of disruption of traffic, or an
indirect economic loss if a longer transportation distance must be
foreseen by an alternative route.

The main criteria to be considered in the assessment of the

economic vulnerability are, thus:

– the types of services implied (industry, tourism, transport);

– the types of economic activities affected;

– the traffic and the cost of blocking a route (which is very high for
railways);

– the possibilities of alternative routes.

Finally, this aspect must include an analysis of the potential damage

due to the secondary consequences of a landslide, such as the
formation of a dam in a valley, inducing flooding upstream and a
potential catastrophic flow downstream.

3.6.3.5 Vulnerability computation

The vulnerability of elements at risk depends on the typology of the
element (T) (and, therefore, on its shock resistance features for
assets) and on the process intensity (I):

V=f (T, I)

From a practical point of view, two approaches can be followed in

order to evaluate vulnerability: (1) simply on the basis of effects on
the element at risk (see Table 3.8), and (2) on the basis of the
process intensity.

In the first case (1), 5 classes of percentage loss (0, 0.25, 0.5, 0.75,
1) have been applied to each vulnerability category. Through
detailed studies it could be possible, however, to attribute various
weights at various percentage classes based on “utility curves” [e.g.:
in the road interruption case—economic vulnerability—some studies
have evidenced a logarithmic type curve (Fig. 3.7a), as the
vulnerability diminishes with the passing of time for the opening of
the alternative roads (e.g.: variable indices–0, 0.5, 0.8, 0.95, 1–
according to elapsed time)].

In the second case (2), process intensity may be indicated as energy

thresholds (J) obtained from the run-out model. When no data
models are available, it is possible to “translate” the intensity
processes into qualitatively described effects (Fig. 3.8).
Several scales have been developed in order to qualify the intensity
of the different types of landslide phenomena. For instance, the
Swiss recommendations published in 1997 provide orders of
magnitude in order to differentiate low, medium and high intensity for
block fall, rockfall, slides and flows. They are expressed, either in
terms of impact energy, or in terms of average velocity, or in terms of
depth of the potentially erodible soil zone (OFAT, OFEE & OFEFP,
1997). For large rockfalls, it is evident that the intensity level in the
exposed zones is always high to very high, whereas for slides, many
large phenomena display low to medium intensity, according to the
proposed scale.
Page 41

Table 3.8. Assessment of vulnerability factors values for physical,

social, environmental, economic aspects.

Physical vulnerability Social vulnerability

Vulnerability description Loss IndexVulnerability description Index

range

Intact structures 0 0 Non-affected persons 0

Local damage 1÷25% 0.25 No physical damage, 0.25

evacuated persons

Serious damage (possible to 26÷50% 0.5 Physical damage (persons 0.5

repair) may continue their
activities)

Mostly destroyed structures 51÷75% 0.75 Seriously wounded 0.75

(difficult to repair) persons (up to 50%
disability)

Total destruction (out of 76÷100%1 Fatalities, 51–100% 1

use; e.g. >5% inclination) disability

Environmental vulnerability Economic vulnerability

Vulnerability description Loss IndexVulnerability description Index

range

Intact element 0 0 No interruption 0

Local loss 1÷25% 0.25 Short temporary 0.25

interruption (hours to day)

Serious damage (possible to 26÷50% 0.5 Average temporary 0.5

repair) interruption (days to
week)

Mostly destroyed elements 51÷75% 0.75 Long temporary 0.75

(difficult to repair) interruption (weeks to
months)

Total destruction 76÷100%1 Permanent interruption 1

Figure 3.8. (a) Quantification of impact in terms of unit mass energy;

(b) Qualification of impact based on the gradation of damage. In the
photograph: Randa rockfall in 1991 (Switzerland).
Page 42

A more detailed analysis implies, however, the consideration of the

acceleration phase, or the volume of potential unit blocks of a
rockfall, as well as their possible interaction. But in any case, the
limit values proposed for the intensity classes are indicative and
have to be considered as qualitative references rather than as
quantitative parameters.

3.6.4 Assessment of consequences

The assessment of consequences induced by a certain scenario is
based on a theoretical computation, simply the product of the
elements at risk value (VE) and the vulnerability (V):

C=VE×V

In this methodology, each element at risk value is multiplied by a

relative vulnerability category (e.g. physical vulnerability
index×assets value index=physical consequence index). The
operation is repeated for each hazard scenario, considering each
specific process intensity class as far as it influences the vulnerability
level (V) of the elements at risk. It is also repeated for each type of
vulnerability (physical, social, environmental, economic).

Therefore, the matricial approach on which the proposed

methodology is based foresees a simple multiplication, at this point
of the process, between value classes, the result of which
(consequence index) will enter in the next matrix for the computation
of the risk (Fig. 3.3 and 3.4). In this way, the assessment of
consequences results from an automated objective process. For this
reason, and in order that the consequence assessment be
significant, all the initial parts of the process (danger definition,
evaluation of the value of the elements at risk, vulnerability
assessment) which will inevitably suffer from a certain rate of
subjectivity, require very careful analyses and assessments.
3.7 RISK AND TOTAL RISK ASSESSMENT

As previously defined, risk is the product of landslide hazard and its

consequences:

R=H×C

In the methodology analysed, hazard (H) has been defined by its

three components: involved area, process intensity and occurrence
probability. The involved area (used for the identification of elements
at risk) and the process intensity (necessary for the vulnerability
identification) have already been used and included in the
consequences. So, risk is the result of the product of occurrence
probability (P) and its consequences (C):

R=P×C

In the same way as in the consequence computation, the matricial

approach used in the risk asessment foresees a simple multiplication
between two value classes the result of which is the risk index
defined for each vulnerability category, thus obtaining the physical,
social, environmental and economic risks. This operation must also
be repeated in this case for each hazard scenario, or at least for
each specific occurrence probability considered. In this way, four risk
maps are derived for each scenario.

The problem of the management of numerous data and of a large

quantity of resulting maps is solved thanks to the use of performing
GIS techniques which allow the superimposition of various levels of
spatial information and the development of mathematical operations
between each implied term in an automatic way.

In order to obtain a synthetic assessment of the risk (total risk) for

the studied area, it is sufficient to sum up the various risk indices
obtained in the different scenarios for each category of impact,
namely the physical, social, environmental and economic total risks.
This process of summing different risk index values corresponding to
several scenarios implies the hypothesis that each considered
scenario may occur independently without having the first
Page 43

occurred scenario modify a subsequent scenario. This is clearly the

case when the considered scenarios are characterized by very
different magnitudes. This assumption is also justified by the fact that
“frequent” and less intense scenarios can occur before a very rare
scenario. Moreover, it is also assumed that a certain lapse of time
will exist between one scenario and the next one. Finally, this
methodology supposes that the value of the exposed objects will not
significantly change over time.

3.8 SCOPE OF APPLICATION OF THE PROPOSED

METHODOLOGY

The methodology proposed by the IMIRILAND project has pointed

out the importance of a correct determination of the intensity and
probability for different scenarios in order to reach an appropriate
hazard and risk assessment in the zone of direct impact of large
landslides. However, these potential dangers have to be assessed in
a regional context with all their consequences, especially as they are
able to induce “in-chain” phenomena such as debris flows following a
slide reactivation, a secondary rockfall with a trajectory modified by a
first event, or above all the possible failure of a landslide dam
releasing a flash flood due to the water accumulated behind the dam
(see Chapter 10.2).

All of these complex scenarios must include the consideration of the

simultaneous occurrence of rare events such as extreme rainfall and
related flooding; but of course the probability of such combined
effects decreases, as expressed by the conditional probabilities.

The fundamental specificity of a landslide dam failure is that its

impact extends far beyond the landslide zone itself, either upstream
with the progressive inundation of extensive areas if the river course
dammed presents a reduced slope, or downstream with the sudden
and violent flooding following the dam failure due to overflow or
seepage. This final aspect requires a detailed study, not only of the
extension of the flooded area, but of all the infrastructures and
exposed objects threatened by such an event, especially as, in these
cases, the vulnerability coefficients of all kinds are very high.

When international roads or railways following the river course

downstream from the landslide site are damaged, when lifelines are
affected even locally, interrupting traffic, the indirect consequences of
even a very rare event may be huge on economic or tourism
activities and, thus, represent non-negligible risks.

The analysis of such “in-chain” phenomena therefore requires a

multiple assessment of the induced risks, as presented above, which
makes it even more difficult to compare different risk situations and
to evaluate their respective degrees of severity.
Page 44

This page intentionally left blank.

Page 45

4
The Cassas landslide

F.Forlati1, M.Morelli2, L.Paro1, F.Piana2, R.Polino2, G.Susella1,

C.Troisi1
1ARPA Piemonte, Torino, 2CNR—IGG Sezione Torino

4.1 INTRODUCTION

4.1.1 Landslide description

The Cassas landslide is located on the right side of the middle Susa
Valley, in the Salbertrand Municipality, in the “Gran Bosco” natural
park (county of Turin, Piedmont, Italy). The Cassas slope is deeply
touched by instability phenomena; the word “cassas” (probably from
French “casser”=to break), used over the ages to indicate this zone,
refers to debris accumulations (Brovero et al., 1996a).

The top of the landslide is located near a minor watershed at about

2’000 m a.s.l., while the landslide foot is placed at about 1’000 m in
the wide Salbertrand valley flat that was formed by the filling of a
landslide dammed lake in the locality Serre la Voute (Capello, 1941).

In this part of the Susa Valley, which is an ancient commercial and

military route, the A32 motorway that connects with the Turin to
Frejus Tunnel, the Turin-Modane international railroad and the
national road SS 24 “Montgenèvre Pass” are located. In particular, a
service station for the A32 motorway is located just at the foot of the
landslide (Fig. 4.1).
Figure 4.1. Location map of the Cassas landslide (red square), Susa
Valley.
Page 46

4.1.2 Historical background

The first records of the slope’s instability in this area are found in
documents from the 18th century, in which the disastrous effects of
the May 1728 flood are described (Bogge, 1975). In the 19th century,
the instability of the right slope of this part of the Susa Valley caused
several problems to the Turin-Modane railroad (Baretti, 1881, 1893;
Sacco, 1898). These authors wrote that the causes of the instability
were the intensively fractured rocks and the “dissolution of
calcareous formations”, outcropping at the slope’s foot.

At the beginning of the 20th century, additional information about the

instability phenomena in this area were extracted from documents of
the Corps of Engineers and forest authority offices: in the former
documents, the deformations of the railroad tunnels and the bypass
channel of the Chiomonte power plant (Segré, 1920) are fully
analysed; in the latter documents, a “huge landslide” of about 10
hectares is mentioned (Consorzio Forestale di Oulx, 1955; Corpo
Forestale dello Stato, 1956). Such evidences of instability are also
visible in the first available aerial photographs dated July 1954.

Between 12th and 14th June 1957 in the Susa Valley, a huge flood
caused great damage to the entire valley. On that occasion, the
western area of the Cassas landslide had a paroxystic phase:
millions of m3 of rock mass slid and caused a debris flow that
reached the valley floor. Re-activations of the landslide were
recorded until the middle Sixties (Peretti, 1967, 1969; Brovero et al.,
1996a) (Fig. 4.2).
Figure 4.2. General overview of the Cassas landslide in a photo from
1965, at the conclusion of the paroxystic phase of the period 1950–
1965 (Brovero et al., 1996a).
Page 47

In the Seventies, the Cassas landslide appeared much wider than

the instability phenomena of the 50’s and 60’s, as stated in a survey
about the “hydrologic instability” of the Susa Valley (Ramasco &
Susella, 1978).

In the Eighties, most of landslide accumulations on the right slope of

the middle Susa Valley, including the Cassas landslide, were
interpreted as associated to the evolution of the so-called “deep-
seated gravitational deformations” by Puma et al. (1984, 1989) and
Mortara & Sorzana (1987). These phenomena involved the Susa-
Chisone watershed for some kilometres and the slopes for hundreds
meters of depth.

In the early Nineties, in order to mitigate the risks on the motorway

and at the service station at the Cassas landslide foot, a monitoring
system was installed on the landslide to verify its movements. The
inclinometer data indicate a slow but constant movement up to the
present.

4.2 REGIONAL FRAMEWORK

4.2.1 Climatic and water conditions

The climatic features of the Susa Valley are variable, with
characteristics unique among the Piedmontese mountain valleys.
Such variability is due to the articulated valley floor morphology
(more than 70 km long and few metres to some kilometres wide) and
the slope’s altitude (more than 3’000 m a.s.l.). These morphological
features control to a large extent wind direction, solar radiation,
temperature distribution, humidity, etc. In Figure 4.3, the Piedmont
climatic map is shown, prepared using the Thornthwaite method
(Biancotti & Bovo, 1998a), based on the determination of
evapotranspiration (real and potential) and of its comparison with
precipitation amount. The map in Figure 4.3 shows valley areas
where special aridity conditions have been observed (xerothermic
areas of the Susa-Bussoleno valley floor and of the Bardonecchia
basin).

Figure 4.3. Map of Piedmont climatic regimes prepared using the

Thornthwaite method (Biancotti & Bovo, 1998a). Simplified legend of
climatic types: blue=perhumid; green=humid; yellow=from humid to
subhumid; red=from subhumid to subarid.
Page 48

On the basis of the rainfall data analysis, a bimodal trend is

observed, showing the two highest values in spring and autumn:
spring is the most rainy period of the year, while autumn records the
greatest rainy water heights over a few days (showers). In summer,
intense but short precipitations (storms) take place over small areas
(Peretti, 1969; Biancotti & Bovo, 1998a). Through the updating of the
regional weather monitoring net, it is possible to supply precise
indications about the climatic trend in the valley (Fig. 4.4). In Table
4.1 some climatic data that refer to a time span between 1990÷1999
are listed. These data come from weather stations located in the
upper parts of the Susa and Chisone Valleys (Fratianni & Motta,
2002). In Figure 4.5, a histogram of the annual precipitation data at
the Salbertrand-Graviere weather station, located at the Cassas
landslide foot, are reported.

Figure 4.4. Location of the weather stations in the upper part of the
Susa Valley used for climatic analysis.
4.2.2 Hydrological data
The main historical floods of the Dora Riparia have been recorded
since the 18th century; they are sometimes associated with
disastrous events (Brovero et al., 1996a; Fratianni & Motta, 2002)
listed as follows:

– XVIII century: 11/1705, 05/1728, 1748 (?);

– XIX century: 05–06/1827, 10/1839, 05/1856, 1863 (?), 06/1876,

05/1879, 07/1896;

– XX century: 09/1920, 05/1948, 06/1957, 08/1972, 05/1977,

04/1981, 09/1993, 06/2000, 09/2000, 10/2000.

Analysing the flood events seasonally, a greater frequency is

observed in the spring (April÷June), although in the autumn, rainfall
is generally more frequent. Therefore, the annual maximum flood of
Dora is the result of the spring snow melting and rainfalls as
confirmed by the analysis of the events of May 1728 and of June
1957 (Bogge, 1975; Brovero et al., 1996a). During
Page 49

Table 4.1. Climatic data in the upper Susa Valley.

No Weather station Municipaliy Alt. Tm Hr Rm Sm

[m] [°C] [%] [mm] [cm]

1 Pietrastretta Susa 520 12.2 59.2 710 —

2 Fniere Chiomonte 813 10.5 63.1 813 —

3 Graviere Salbertrand 1010 8.7 71.2 723 69

4 Gad Oulx 1065 7.9 68.4 602 —

5 Val Clarea Giaglione 1135 8.7 66.4 — —

6 Prerichard Bardonecchia 1353 7.3 66.2 786 98

7 Camini Frejus Bardonecchia 1800 4.9 63.2 916 199

8 Le Selle Salbertrand 1950 4.5 62.6 722 276

9 Sestriere Sestriere 2020 1.5 65.1 — 300

10 Clot della Soma Pragelato 2150 3.7 47.8 — 330

11 Colle Bercia Cesana T.se 2200 2.9 61.7 — 451

12 Lago Pilone Sauze d’Oulx 2320 2.3 — 499 456

13 Monte Fraiteve Sestriere 2701 −0.8 67.3 — —

14 Rifugio Giaglione 2750 −0.7 55.4 — —

Vaccarone

Tm: Average of the monthly temperature average (1990–1999)

Hr: Annual humidity average (1990–1999)

Rm: Annual rainfalls average (1990–1999)

Sm: Sum of the monthly average hights of snow (1990–1999).

Figure 4.5. Salbertrand-Graviere station annual rainfall (columns)

and annual average (756.9 mm) for the period 1914–2002. The lack
of data is due to instrument failure or to lack of measurement. Since
1992, a new pluviometric station has been functioning.

these events, the Cassas landslide immediately accelerated, as also

occurred during the October 2000 heavy meteorological event which
involved the whole Dora Riparia valley and a large part of the
Piedmont region.

4.2.3 Water condition

No surface waters have been observed within the landslide area but
there is an aquifer hosted in the debris cover of the landslide that lies
on the calcschistic bedrock. Groundwater height varies according to
rainfall regime. Waters related to rainfall and snow melting mainly
infiltrate in the upper part of the landslide, as high fractured bedrock
enables deep water circulation.

In the lower part of landslide, two springs run (estimated flow on July
1997: from 2×10−4 to 8×10−4 m3/s). There are also some springs
along the scarp that cut across the debris cover between the
altitudes 1’350 and 1’550 m a.s.l. (Brovero et al., 1996a; Paro,
1997).
Page 50

4.2.4 Regional morphology

The Susa Valley morphology is due to fluvial-glacial modelling and to
gravitational instability of the slopes (Fig. 4.6). On the right side of
the valley, the modelling process due to landslide deformations is
evident, whereas glacial morphologies and deposits are almost
absent (Carta Geologica d’Italia in scala 1:50’000). Also, the wide
Salbertrand valley flat was formed by a landslide dammed lake in
Serre la Voute (Capello, 1941). Furthermore, along the Susa-
Chisone watershed (Fig. 4.7), there are trenches, sliding steps,
depressions and spreading double-crested ridges related to deep-
seated gravitational phenomena (Puma et al., 1984; Mortara &
Sorzana, 1987; Puma et al., 1989).

The Cassas slope is not entirely affected by gravitational instability

as there is a sector on the left flank of the landslide, termed hereafter
as “1957 Cassas”, where polished rock steps due to fluvialglacial
modelling and glacial deposits are preserved (Paro, 1997; Carta
Geologica d’Italia in scala 1:50’000, 2002).
Figure 4.6. General overview towards the East of the Susa and
Chisone Valleys (2001 aerial photographs with overlapping Digital
Terrain Model). Gravitational phenomena deriving from IFFI Project
(Gravitational Phenomena Inventory in Italy. ARPA Piemonte, 2004)
are shown: yellow=deep-seated gravitational slope deformation;
red=gravitational phenomena of other typologies.

The sub-sectors of the Cassas slope will be described from East to

West in the following section (Fig. 4.8):

– on the eastern part, a landslide accumulation called here “Sapè

d’Exilles” (A) is distinguished; it extends from an altitude of about
2’350 m to the valley floor for about 2.5 km2 (height more than 1’000
m). This landslide accumulation and that of Eclause in front of it (on
the left side of the valley) formed the valley dam originating the Oulx-
Salbertrand plain.

– To the West of “Sapé d’Exilles”, other accumulations can be

recognized (B and D); these are smaller and lobe-shaped, and
gently overlap the alluvial deposits on the valley floor. The largest
one (D) is particularly articulated as it covers another accumulation
to the East (C), while to the
Page 51

Figure 4.7. Double-crested ridge (1 km long and 300 m wide) along

the Susa-Chisone watershed (Testa dell’ Assietta Mount) related to
the deep-seated gravitational phenomena (to the right).

West it is partially covered by the 1957 Cassas accumulation, which

partly re-mobilized this deposit. The main scarps of these
accumulations (C, D and “1957 Cassas”) are joined together and
aligned to the sub-rectilinear East-West direction.

The Cassas slope is also characterized by many scarps that bound

or split the accumulations described above. In particular, at the foot
of Sector A there is a scarp whose origin is related to the erosion of
the Dora Riparia river; the foot destabilization due to the erosion is
probably the cause of the deformations verified along the A.E.M.
bypass channel (the water of the Dora Riparia is transported to
hydroelectric plant of Chiomonte through this channel), located at the
foot of Sector A.

Also within Sectors B and D, two aligned scarps are recognized at

an altitude of about 1 ‘300 m. The alignment of such scarps is
probably due to the presence of a previous (pre-landslide)
morphology due to fluvial modelling, that had been successively
covered by B and D accumulations and then partially exhumed.

Another evident scarp, some tens of meters high, is located at

western part of the Cassas slope. The origin of this scarp is probably
due to past fluvial erosion. At the foot of this sub-vertical scarp, a
large talus and the motorway service station are located.

From a morpho-evolutionary point of view, the more interesting

escarpments are observable in “1957 Cassas”: one scarp is placed
in the upper part of the sector, while another is cuts across the sector
with a North-South direction (see § Geo-morphological analysis).
Page 52

Figure 4.8. View of the Susa Valley right slope, from Salbertrand to
Serre la Voute, with indications of the main morphological elements:
1. Cassas landslide main scarp, June 1957 event; 2. main scraps of
gravitational phenomena; 3. secondary scraps of gravitational origin;
4. scraps by fluvial-torrential erosion; 5. landslide accumulations; 6.
main rock outcrops; 7. actual fluvial-torrential deposits; 8. alluvial
fans; 9. fluvial torrential terraced deposits; 10. talus; 11. landslide
scar and transition area; 12. area characterised by morphological
features related to deep-seated gravitational deformations.

In the valley floor, some alluvial fans are observed, such as the large
fan formed by the Gran Plenei stream (right tributary of the Dora
Riparia river). Some of these, of smaller dimensions, are located at
the feet of landslide accumulations. Among those, the fan placed at
the foot of the “1957 Cassas”, was involved in the paroxystic event of
June 1957. Successively, this fan was modified by the construction
of the service station for the A32 Motorway (C in Fig. 4.15), which
also rests on the alluvial deposits of the Dora Riparia river.

4.2.5 Regional and structural setting

In the middle Susa valley, several tectonostratigraphic units (Dela
Pierre et al., 1997; Polino & al., 2002) of different paleogeographic
origin have been distinguished (Fig. 4.9):

A. oceanic units consisting of meta-ophiolites and calcschists

(Piedmont-Ligurian units);

B. continental margin units made up of Mesozoic dolomites and

calcschists;

C. polymetamorphic basement units consisting of gneisses,

quartzites, micaschists and metabasites (Ambin and Clarea
complex) and relative Mesozoic sedimentary covers (Briangonnais
units).

These tectonostratigraphic units are juxtaposed by west-dipping

thrust planes and displaced by high-angle strike-slip or normal faults
(see § local geology and structural analysis). These faults define the
geometric framework of the middle Susa valley and gave origin to
huge fault breccias and strongly fractured rock masses. This allowed
the triggering and evolution of the large landslides of the middle
Susa valley (Carraro & al., 1979; Giardino & Polino, 1997).
Page 53

Figure 4.9. Geological structural map of the Bardonecchia sheet n.

153 of the CGI at the scale of 1:50’000 and legend of the
tectonostratigraphic units (Carta Geologica d’Italia in scala 1:50’000,
2002). Blue circle: Cassas landslide.

4.3 HAZARD ANALYSIS

4.3.1 Geomorphological evolution

Since detailed historical documentation is lacking, other
morphological information has been deduced from topographic
maps.

The currently used topographic map (CTR 153110 Salbertrand—

1991) has been compared with the previous official map of the
Military Geographic Institute (IGM sheet N°54, II SE, Oulx–1934) in
order to underline the main morphological changes suffered by the
Cassas slope during the 1934–1991 timespan.

The 1934 official map of the Military Geographic Institute already

shows features that can be interpreted as instability clues on the
Cassas slope. In this area (Fig. 4.10), small scarps (A), a trench (B)
and a small landslide (C) are underlined.

The morphological evolution has been also reconstructed on the

basis of the interpretation of aerial photos made in different years
(1954, 1957, 1963, 1979, 1986, 2001). The revision of these
documents gave new important elements for the interpretation of the
Cassas landslide evolutionary processes.

4.3.1.1 Aerial photographs dated 1954

In these photos, some important features are shown (see Fig. 4.11):

– the upper part of the Cassas slope is already involved in incipient

sliding (L);

– a sliding surface is shown at the top of this area (Cassas main

scarp, F), with a vertical dislocation of some tens of meters;

– two sectors of active landslide (complex rock slide-rapid debris

flow, D);

– accumulation of large blocks on the eastern side of this slope,

probably derived from landslides of different age that reached the
valley floor; this accumulation is partly remodeled in its western
sector, where a scarp of about 80 m high is shown (S);
Page 54

Figure 4.10. 1934 map at scale of 1:25’000 (IGM sheet n.54, II SE,
Oulx). The shaded zones indicate the main instability features.

– below the S scarp, a wide wooden zone (from about 1’450 m to

1’200 m in altitude, A) is laterally bounded by two streams that
produced a fan at the bottom of the slope. It has been hypothesized
that the wooden zone is also a gravitational accumulation that
controlled the onset of a new hydrographical pattern and that still
shows evidence of instability;

– the accumulation area situated on the eastern side of Cassas

slope (C) also presents some scarps due to movements at the foot.

4.3.1.2 Photographs and surveys from 1962 to 2002

The comparison between the images of the Cassas area before and
after the event of June 1957, emphasises some important elements
about the kinematic evolution of the gravitational phenomena, whose
paroxystic phase ceased in the mid 60’s in the past century. In
Figure 4.14, the Cassas landslide appears at the end of the
1954÷1965 paroxystic phase: the debris accumulation is at its
greatest extension, since it covers 2/3 of the fan surface in the valley
floor.

Through the historical data and present-day morphology, it is

possible to describe the landslide of June 1957:

– during the event of June 1957, the upper part of the Cassas slope
(L in Fig. 4.11) underwent a total displacement of about 250 m,
mostly by sliding. A sub-vertical scarp, tens of meters high, was
formed on the right flank from which a part of the existing debris
flowed (S in Fig. 4.12).

– The movement involved the bedrock (slide surface), that is still

visible at present in the upper part of the slope. The strongly
fractured and released bedrock induces several metric open
fractures along the crown zone (Fig. 4.13).

– Most of the displaced rocks probably accumulated in the wide

depression just below Sector L, where they covered the S scarp (Fig.
4.11). The displaced rock mass is not excessively disarticulated, as
indicated by many preserved trees and by a large remnant of the soil
cover on the
Page 55

Figure 4.11. Geomorphological map based on aerial photographs

dated July 1954 (1–hydrographical pattern; 2–gravitational scarps;
3–Cassas main scarp; 4–secondary gravitational scarps; 5–erosional
and gravitational scarps; 6–landslide accumulations outside the
Cassas area; 7–accumulation of large blocks with no vegetation; 8–
sector of the Cassas landslide with evidences of movement; 9–
complex slide-rapid flow landslides; 10–bedrock main outcrops; 11–
presumed landslide accumulation; 12–alluvial fan; 13–talus; 14–
terraced fluvial-torrential deposits; 15–actual fluvial-torrential
deposits).
Page 56

accumulation body. Part of the displaced material reached the valley

floor by debris flow processes, and lies on the alluvial fan at the
slope foot with a maximum thickness of 1.5 m.

– Instability evidences, activated during the June 1957 event,

indicate a movement of the whole slope, also involving an area
located outside the Cassas landslide (Fig. 4.14).

Movements on the Cassas slope continued after the paroxystic

events of 1957÷1965. In Figure 4.14 it can be observed that inside
the accumulation body, some important scarps are delineating (N
and R in Figure 4.15). The largest one (R) cuts across the landslide
area, separating the upper

Figure 4.12. Overview of the main scarp on the right flank.

Heterogeneous debris accumulation is still visible today.
Figure 4.13. Metric open joints along the Cassas landslide crown
zone (photo by L.Paro).
Page 57

zone, characterised by prevalent sliding, from the lower one, in

which debris flows are prevalent. The scarp N is located in the same
position as the dip slope change shown in the 1954 aerial
photographs, indicating that this one has been covered for some
years by the sliding mass.

Figure 4.14. Cassas landslide evolution (1965 and 1982: Brovero et

al., 1996a; 2002: photograph by L. Paro).

The morphological evidence indicates that:

– the whole landslide accumulation is in movement (as confirmed by

monitoring data);

– the scarps separate zones with different kinematics;

– the scarps correspond to morphostructural elements buried below

the gravitational accumulation.

Finally, Figure 4.15 also shows an uphill-facing scarp (T), already

visible in photographs of 1954, placed at the western side of the
landslide, at an altitude of 1’600 m.
Figure 4.15. Synthetic sketch map of the main morphological
elements of the Cassas landslide (by aerial photos 1963÷2001). S:
main scarp; R and N: minor scarps; T: ancient up-hill facing scarp; U:
bedrock; W: recent forest colonisation; C: Lower position of debris in
the 1963 paroxysm.
Page 58

4.3.2 Local geology and structural analysis

The Cassas landslide has been intensively studied in recent years.
“On site” structural analyses have been carried out to characterise
the main discontinuity systems. The results have been compared to
the regional geological context (“around site” analyses) in order to
achieve a better knowledge of the physical and geometrical
characteristics of the main structural elements that concern the
landslide.

• “On site” geological data consist of available geological data mainly

collected on the landslide.

• “Around site” geological data consist of:

– regional geological and structural data and related mesostructural

and geomechanical data;

– mesostructural data collected in the area surrounding the Cassas

landslide;

– lineaments detected from aerial photo analysis;

– remotely sensed data.

4.3.2.1 “On Site” geological data

Data comes from geotechnical studies (Epifani, 1991; Brovero et al.,
1996a; S.I.T.A.F., 2000) and the geological survey for the
Bardonecchia sheet (Carta Geologica d’Italia in scala 1:50’000,
2002).

The Cassas landslide is placed in the Cerogne-Ciantiplagna

tectonostratigraphic unit bounded at the top and the base by tectonic
contacts (Fig. 4.9). This unit consists mainly of:

– prevalent calcschists;
– carbonatic massive calcschists with interlayered, cm-mm thick
micaceous schists;

– metabasites with local basaltic metabreccias;

– rare massive micaceous quartzites.

The available mesostructural data of the upper and lower part of the
landslide permitted the observation of six main fracture (s.l.) systems
(Fig. 4.16a):

– the main schistosity (S1) dips to the S-SE with an inclination of 40–
50°. In some places, the schistosity gently dips to the N; in this case
it is labelled S2;

– K1 system: sub-vertical (80°–85°) persistent fractures dipping

N350° and N170°;

– K2 system: fractures dipping N220° and N240° with an inclination

of 50°–70°; they are less persistent than K1 and K2;

Figure 4.16. a) Stereographic net of the discontinuity systems; b)

morphostructural lineaments observed in the upper part of the
Cassas landslide. S1 and S2: schistosity; K1, K2, K3, K4 and K: joint
systems (Brovero et al., 1996a).
Page 59

Figure 4.17. Extract of the geomorphologic and geological map of

the right slope of the Susa Valley between Salbertrand and Exilles
(Paro, 1997) updated in 2003. 1–trenches; 2–gravitational-structural
scarps; 3–scarp of fluvial-torrential erosion; 4–metric open-joints; 5–
strike and dip of main schistosity; 9–grey calcschists; 10–quartzitic
intercalations; 11–olistoliths of metabasites and serpentinites; 12–
brown-yellow calcschists; 13–phyllitic schists; 14–landslide
accumulations; 15–fan of mixed origin; 16–alluvial deposits; 17–
glacial deposits s.l.; 18–travertines. Hydrology: 19–water extinction
point; spring (20–<0.21/s; 21–0.2–0.81/s; 22–>0.81/s; carbonate
spring (23–<0.21/s; 24–0.2–0.81/s; 25–>0.81/s).

– K3 system: sub-vertical (80°–90°) fractures dipping N90° and

N270°;

– K4 system: fractures dipping N350° with an inclination of 40°–50°.

– K system: sub-vertical fractures dipping to the SE and NW.

These fracture systems correspond to major morphostructural

elements of the Cassas landslide, also confirmed by aerial photo
analyses (Fig. 4.16b and Fig. 4.17). They are:

– the E-W striking lineament system (K1) that corresponds to the

main scarp (western part) and open trench sub-parallel to the crown
of the landslide;

– the N150° striking lineament system (K2) that corresponds to

fractures and scarps within the landslide body and on the top of the
landslide. These systems are sub-parallel to the lateral boundaries of
the landslide.

– the N-S striking lineament system (K3) that corresponds to minor

rock scarps within the landslide body;

– the N45° striking lineament system (K) that corresponds to main

scarps (eastern part) and fractures within the landslide body and in
the upper part of the landslide.
Page 60

Rock mass classification

Available rock mass data have also been analysed to give

information on the geomechanical properties of the substratum of the
Cassas landslide.

These data refer to some measurement sites in the surroundings of

the landslide; these were chosen among those that were set up for
the survey for the Bardonecchia sheet (Fig. 4.18).

Figure 4.18. Portion of the geological map of Bardonecchia sheet

(see also Figure 4.9) with location of the some measurement sites
(yellow circle) collected during the geological survey of Bardonecchia
sheet. The sites 12, 13, 14 and 28 are sites used for rock mass
classification. Blue circle: Cassas landslide.

These data permitted the classification of the rocks of the slope

according to the SMR method (Slope Mass Rating1, see Table 4.2;
Romana, 1990) according to the I.S.R.M. (1978) rules.
Table 4.2. Description of the rock mass quality classes (Romana,
1990).

Class V IV III II I

SMR 0–20 21–40 41–60 61–80 81–

100

DescriptionVery bad Bad Normal Good Very

good

Stability Quite Unstable Partially unstable Stable Quite

unstable stable

Failure TranslationalTranslational slide Translational slide Some No

types slide and/or large wedge and/or small blocks one
wedge

1 This method permits the classification of the slope in classes of

quality by the following equation:

SMR—RMR+(F1·F2·F3)+F4

The RMR is the Rock Mass Rating (Bieniawski, 1974); F1, F2, F3
and F4 are corrective factors that define the type of slope failure
resulting from geometric relationships between discontinuity systems
and the slope (Romana, 1990).
Page 61

The collected data are summarised in Table 4.3 and in the relative
histogram (Fig. 4.18), where the RMR, SMR and RQD (rock quality
designation) are reported.

The comparison between the SMR value and the Romana table put
the rock mass of the Cassas landslide between classes of quality V
and III.

These data underline the great variability of the geomechanical rock

mass conditions as the measurement sites, although close to each
other, give very different SMR values.

Table 4.3. Geomechanical data collected at some sites in the area

surrounding the Cassas landslide (Lapenta & Trombetta, 1996).

Measurement Sites Lithotypes RQD RMR SMR

12 Carbonatic Calcschists 59 39 33.6

13 Calcschists 82 48 20.5

14 Carbonatic Calcschists 75 39 45

28 Calcschists 32 44 50
Figure 4.19. Histogram of the geomechanical data reported in Table
3.

4.3.2.2 “Around site” geological data

To improve the geometric and kinematic characterisation of the
geological model, a comparison between the on site structural data
and the around site geological and structural analyses has been
made with the aim of finding relations between the morphological
and tectonic elements of the landslide.

The amount of available regional data was high; those used here
have been critically revised, since they were originally produced for
very different purposes from those of the IMIRILAND project.

Regional structural and geological data

The Cassas landslide is located in the joining zone of two main

regional fault systems that control the morphology and geometry of
the drainage network and bound some regional tectonostratigraphic
units (Fig. 4.9).

These fault systems are:

– the N60° striking system (F1) that consists of some kilometre-scale

extended pervasive faults and fractures. These are conjugate normal
or left lateral faults dipping NW and SE, largely developed over the
whole Susa-Chisone watershed;
Page 62

– the N120° striking system (F2) that consists of some kilometre-

scale extended right lateral and normal faults clearly visible in the
Cassas landslide area. The geometric and kinematic features of
these faults have been studied at mesoscale in the surroundings of
the landslide.
Figure 4.20. a) Mesostructural sketch showing the peculiar cross-
cutting relationships between fault systems (western boundary of
Cassas landslide); b) Schmidt diagram (lower hemisphere) average
cyclograph trace of the main discontinuity systems; c) and b)
detailed pictures of the mesostructural sketch. s 1: schistosity; F1,
F1a, F2 and F3: fault systems; ZF1 shear zones (see text for further
explanations); vc: calcite veins. In the sketch a) the location of c) and
d) photographs is indicated.
Page 63

The F1 and F2 fault systems have been analysed at 6 sites on the

left side of the landslide where the rocks are relatively sound. These
analyses allowed the definition of the structural associations2,
described in the following paragraphs (see Fig. 4.20, 4.21 and 4.22).

The F1 mainly consists of transtensive conjugate faults dipping at

high angle towards the SE and NW. The F1 system is the most
represented at the outcrop scale; it defines metric to decametrescale
fault zones and disjoints the rock mass for several tens of metres
(Fig. 4.21).

In some places, these faults are associated with steep SE-dipping

metric shear zones (ZF1, in Fig. 4.20a), characterised by
anastomised shear planes. Decimetre-scale to metric
lenticularshaped blocks and tectonic breccias and gouges are
aligned on these surfaces. The thickness of these fault rocks, ranges
from a few centimetres to some decimetres. The geometry of the F1
system is indeed very complex since a minor fault system (Fla, in
Fig. 4.20a) is associated to the main one. The F1a consist of normal
and left lateral faults dipping from low to high angle to the N and S
(Fig. 4.20b), interpreted as synthetic Riedel shear planes (sensu
Ramsay & Huber, 1987) of the F1. The F1 system roughly
corresponds to the K, K1 and K4 fracture systems observed within
the landslide (see “on site” geological data, Fig. 4.16a).

The F2 mainly consists of discrete sinistral strike-slip faults dipping

about 40° to the NE and decametre-scale shear zones dipping to the
SW-SSW. The discrete faults have regular decimetrescale spacing
(Fig. 4.20), whilst the larger shear zones (F2) have a spacing of
several metres and are usually associated to fault rocks (Fig. 4.21
and 4.22). The F2 system roughly corresponds to the fracture
system (K2), observed within the landslide (Fig. 4.16a).
A third structural association, less evident at the regional scale but
widespread around the Cassas landslide, mainly consists of high
angle N-S to NNE-SSW striking normal faults (F3, Fig. 4.20). They
are mainly discrete faults with associate sub-vertical centimetre-
scale calcite veins (vc). The F3 system roughly corresponds to the
fracture system (K3), observed within the landslide.

Figure 4.21. Decametre-scale fault zones associated to subparallel

F1 fault systems (dashed red line) observed at the western boundary
of the landslide.

Figure 4.22. Decametre-scale shear zone associated to the F2 fault

systems observed at the western boundary of the landslide.
The schistosity (s1) dips at a low to medium angle to the SSE, both
within and outside the landslide area. The schistosity also dips to the
NE in some places near the landslide, probably due to

2The structural association is a “geological tool” for the correlation of

the different discontinuity elements with different geometry,
kinematics and size in genetically coherent groups (Vialon et al.,
1976; Davis & Reynolds, 1996).
Page 64

the presence of drag folds or tilting of rock blocks. Since the s1 has a
relatively constant orientation, it can be easily extrapolated along
strike and at depth for at least several hundreds of meters.

Aerial photo lineaments

The geometric relations and the geomorphological characters of the

discontinuity systems observed on the Cassas landslide have also
been analysed at the scale of the slope by available photo-lineament
analyses (Fig. 4.23) in the geotechnical study (S.I.T.A.F., 2000) and
at the regional scale by means of interpretation of Landsat Thematic
Mapper (Fig. 4.25). The integration of these different data led to a
clearer understanding of the distribution of morpho-structural
elements around the Cassas landslide.

Four main photo-lineament systems have been recognised (Fig.

4.23):

– a NE-SW system mainly consisting of regular vertical short scarps

that show kilometre-scale lateral continuity and are clearly apparent
on the left side and at the top of the landslide slope. These photo-
lineaments are sub-parallel to the eastern part of the Cassas main
scarp and trenches on the landslide body;
Figure 4.23. Geomorphologic map obtained from analysis of the
aerial photographs (S.I.T.A.F., 2000) where is shown: the main
landslide bodies (red colour); alluvial valley floor (light green colour);
fluvial incision in erosion (brown colour); slope due to glacial
modelling at low and height altitude (respectively yellow and light
blue colour); lineaments (red line) Cassas landslide (blue dashed
line).
Page 65

– a E-W system consisting of aligned fractures, depressions, double-

crested ridges and scarps with kilometre-scale lateral continuity.
Some of these alignments are well developed in the upper part of the
Cassas slope and they are also sub-parallel to the western part of
the main scarp of the landslide (see also Fig. 4.17);

– a NW-SE system consisting of longer morphostructural elements

that correspond to lateral boundaries of the Cassas landslide;

– a N-S system consisting of scarps (e.g. R in Fig. 4.15), trenches

and fractures well developed in the upper and middle parts of the
Cassas landslide.

Remotely sensed lineaments

On the map (Fig. 4.24), the lineament distribution points out a

paucity of lineaments on the SusaChisone watershed in comparison
to the adjoining areas. It is worth noting that this paucity could be
due to the uniform and wide distribution of fractures and faults that
characterise the whole Susa-Chisone watershed. This high fracture
intensity reduces the chance of detecting major spectral and
morphological lineaments at the scale of satellite images.

Although the lineaments are not evident on the watershed,

conversely, they show a good correspondence with regional
geological structures in the surroundings of the landslide as also
resulting from field observations.

Consequently, geometric relations between the regional fault

systems and the satellite lineament systems have been sought to
verify the statistic representativeness of the field structural data
(Morelli, 2000; Morelli & Piana, 2003). Morphostructural and spectral
criteria have been used for the detection of lineaments.
The identified lineaments were first analysed statistically and then
compared with photo-lineaments and structural field data. The
statistical analysis was carried out to characterise the relationships
between the azimuthal frequency distribution and lineament length
both for cumulative lengths and length classes. The statistical
description is illustrated in the rose diagrams of Figure 4.25.

The frequency and cumulative length diagrams show a similar

distribution (Fig. 4.25a, b). Both diagrams show a maximum of
frequency and cumulative length on the N60°–N70° direction and
another maximum on the N120°–N130° direction. Two other minor
clusters are on the N80° and N170° directions.

The lineaments have been distinguished in terms of their length to

determine the relations between frequency and length. Frequencies
were divided into three length classes (Fig. 4.25c, d, e, and f).

Figure 4.24. Landsat TM image processed (false colour composite

7–5–4 RGB,) with lineaments; the colours show the different
lineament systems: LnS1 (red), LnS2 (violet), LnS3 (green) and
LnS4 (blue). The blank dashed line is the landslide area.
Page 66

Figure 4.25. Landsat lineament diagrams showing the azimuthal

frequency and cumulative lengths (a and b) and azimuthal frequency
divided into classes of lengths (c–f). L: length of the lineaments; f:
azimuthal frequency of the lineaments.

The frequencies for the classes of lineament length are more widely
distributed and permit the identification of four lineament systems
(see Fig. 4.24):

– LnS1 (NE-SW) lineaments are uniformly distributed in the three

length classes. The LnS1 system consists of regularly spaced
lineaments, uniformly distributed, especially on the left side of the
Susa valley where they show geometric patterns similar to those of
the F1 fault system.

– LnS2 (NW-SE) lineaments are mainly distributed in the

intermediate and short classes. The LnS2 lineaments consists of
regularly spaced lineaments, well exposed on the left side of the
Susa valley where they cut the LnS1 lineaments or end against
them. The cross-cutting relations of the LnS1 and LnS2 systems
define lozenge-shaped domains and have geometric and hierarchic
relations similar to those of the F1 and F2 regional faults.
– LnS3 (N-S) lineaments are better represented in the shorter class.
The LnS3 system consists of minor lineaments particularly clearly
observable in the eastern and western parts of the Susa-Chisone
watershed. These show scarce evidence in the adjoining area of the
Cassas landslide although widespread at the mesoscale.

– LnS4 (E-W) lineaments are better represented in the longer class.

The LnS4 system consists of a cluster of lineaments mainly aligned
along the middle Susa Valley. This system does not show
cartographic evidence, but it is well exposed at the mesoscale in the
surroundings of the landslide.

4.3.3 Investigation and monitoring

The monitoring system of the Cassas landslide consists of different
instruments installed since 1991.

Five drilling surveys have been carried out until the present:

– 1991, 4 boreholes in the lower part of the accumulation;

– 1995, 11 boreholes in the middle-upper part of the landslide;

– 1998, 2 boreholes in the lower part of the accumulation;

– 1998, 2 boreholes in the upper part;

– 2002, 2 boreholes in the middle part of the landslide.

Page 67

In total, 21 drillings were carried out to a depth between 25 and 80

m, nine of them instrumented with inclinometers, eight with
piezometers and five with a geophones (piezometers and
geophones measure continuously in real time). Some boreholes
were supplied with fixed inclinometers. The last inclinometers
installed were supplied with TDR systems.

Figure 4.26. Longitudinal seismic profile of the middle-lower area of

the Cassas landslide (Brovero et al., 1996a).

At present, only one inclinometer is working (installed in 2002), as

the other instruments were cut by the landslide movements.

Starting from 1997, near the crown, 13 extensometers and tiltmeters

were installed: they measure the main fractures of the crown.

The rock wall at the left flank of the landslide is monitored by 12 joint
gauges and by 1 geophone.

Begun in 1994, a microseismic network is available to monitor the

rock noise, composed of 8 geophones (some of them located in
superficial positions), both triaxial and uniaxial ones.

Some refractive seismic surveys have been carried out (Epifani,

1991) showing four different reflective horizons (Fig. 4.26).

4.3.3.1 Monitoring results

Inclinometers placed in the middle-upper part of the landslide body
show a sliding surface at a depth of about 50–62 m. The
stratigraphic logs indicate that the moving mass consists of a chaotic
and heterogeneous material floating in a sandy-muddy matrix. Near
the sliding surface, breccias and fine-grained materials are found. In
the period 1995–2000, the average movement velocity was about 20
mm/y. Starting from the heavy meteorological event of October 2000,
the movements have accelerated up to 70–90 mm/y.

The inclinometers located at the landslide foot show another sliding

surface located at a depth of about 20 m, with an average movement
velocity of about 30 mm/y. These data indicate that the middle-upper
sliding surface of the mass is not related to the lower one. It seems
that the upper surface intersects the slope at an altitude of 1’2500–
1’300 m, where a spring alignment is observed.
Page 68

The surface instruments located on the Cassas landslide crown and

on the rock wall at the left flank have not recorded significant
movements (C.T.M., 2003).

Figure 4.27. Schematic section showing the most relevant

monitoring instruments and results.

4.3.4 Danger characterization

4.3.4.1 Introduction
The definition of the dangers for the Cassas landslide is difficult,
since it is complicated by the slow and continuous deformation at
depth. The different studies carried out over the last ten years have
supplied important elements for a correct understanding of the
kinematics and dynamics of the landslide, leaving, nevertheless,
several problems unsolved.

In this study, a detailed data analysis has been carried out; it

supplies some indications about the evolutionary features of the
phenomenon. Considering the difficulty of obtaining realistic results
due to the complexity of this study, no numerical modelling has been
carried out.
To obtain concrete conclusions about the dynamic and kinematic
evolution, the data are analysed in detail in the following chapter.

4.3.4.2 Morphostructural constraints

In this study, the Cassas landslide has been divided into zones and
particular attention has been given to the paroxystic phase of
movement in the period 1954–1965 to obtain a better danger
characterisation. Each of these zones is characterised by distinct
kinematics.

Referring to Figure 4.28, it is possible to define the following zones:

A—main scarp zone: intensely fractured and released calcschists,

with unstable blocks of variable volumes; in the crown zone many
deep fractures are apparent;

B—active landslide accumulation: made up of rock blocks of variable

size (max 15 m3) floating in a sandy-muddy matrix; it is bounded at
the base by a sliding surface at a depth of about 50–60 m,
approximately at the contact with the bedrock as inferred on the
basis of instrument data;

C—complex landslide accumulation: made up of rock blocks of

variable size (max 5 m3) in a sandy-muddy matrix; in this area the
kinematics are complex because of the combination of sliding and
slow-to-rapid flow movements;

D—alluvial fan area: area of debris flow expansion and accumulation

originating in the middle and upper parts of the slope;
Page 69

Figure 4.28. Subdivision of the Cassas landslide to identify danger.

Some inclinometers and the related movement data are located.

E—landslide accumulation in meta-stable equilibrium: block

accumulation and mega-blocks (20–30 m3) of older landslides, partly
mobilised during the event of 1957;

F—ancient landslide accumulation: old accumulation body reaching

the valley floor, partially reactivated by recent movements, probably
still active at present.

To the West of the Cassas landslide a stable area may be seen (G)
where superficial landslides are located; the monitoring data of the
sub-vertical rock wall above the motorway service station do not
indicate movement.

4.3.4.3 Structural constraints for kinematic behaviour and

displacement mechanisms
The local discontinuity systems that were taken into account for the
slope stability analysis (Fig. 4.29) belong to the main regional fault
systems described above (F1, F2, F3). At the Cassas site, they
correspond to:

– decametre-scale to metric NE fault zones (F1) and associated F1a

faults that displace the rock mass for several tens of metres. They
are sub-parallel to the scar zone of the landslide and to some
trenches observed within Zone B.

– NW strike-slip faults (F2) and shear zones (F2sz) parallel to the

lateral boundaries of the landslide that correspond to alignments of
many scarps more or less extended (i.e. the lateral boundary of
Zone C).

– N-S minor faults (F3) widespread all around the landslide; they are
sub-parallel to the trenches and scarps in the middle and upper parts
of the landslide. The most extensive scarp among those observed in
the landslide body separates Sector B from Sector C.
Page 70

Figure 4.29. Schmidt diagram (lower hemisphere) of the average

cyclograph trace taken into account for the slope stability analysis.
Legend: F1 fault zone with associated conjugate minor fault system
F1a; conjugate F2 fault system and F2sz shear zone; F3 minor fault
systems; s1 regional schistosity; the arrows shown the sliding
movements toward WNW and NW (see text for further explanations).

These tectonic features permit the drawing of a schematic geological

section of the bedrock (Fig. 4.30), where the structural framework is
depicted by graphic repetition, below the entire slope, of the main
structural associations observed in the field (a prototype structural
site is sketched in the inset, where the actual configuration and
cross-cutting relations of the features are reported). In this
interpretation, the bedrock is affected by first order structures mainly
dipping uphill, whilst second order structural features gently dip
toward the valley.

Moreover, a concave discontinuity has been traced at the base of the

shallower part of the bedrock, represented here as a strongly
disengaged rock mass (debris). This discontinuity could correspond
to the sliding surface inferred at the depth of about 50–60 m on the
basis of monitoring data; this surface constrained the estimation of
rock volumes for hazard scenarios (see below).

The structural setting of the bedrock described above would seem to

disagree with the interpretation of geophysical data reported in
Figure 4.26, where the presence of four different layers
characterised by progressively higher seismic velocity with depth is
pointed out. Indeed, the inferred surfaces that have been traced in
Figure 4.26 to separate layers characterised by different seismic
velocities, are not to be understood as potential sliding surfaces, but
simply as horizons (envelopes) that could separate rock layers
showing different mechanical properties (here thought of as strictly
dependent on fracture density).

Since the existence of a sliding surface has been postulated to justify

the gravitational instability of the landslide (see scenarios), a critical
analysis of the structural setting of the bedrock must be carried out
to verify which among the pre-existent structural features could act
as a potential sliding surface.

Furthermore, the physical and morphological characteristics of the

structural system (roughness, morphology, opening, infillings,
alteration etc.) are to be considered in order to evaluate qualitatively
how much the slope is prone to instability.

Searching for kinematic relations between bedrock structures and

landslide features is allowed only where the geometry of landslide
body is strictly controlled by the tectonic features of the bedrock. At
the Cassas site, this condition is verified, since the landslide has
been subdivided into morphostructural sectors bounded by steep
scarps parallel to regional fault systems and that correspond to local
(on site) tectonic features.
Page 71

Figure 4.30. Geometry and cross-cutting relations of the faults and

fractures in the bedrock refer to the actual configurations observed at
mesoscale in the most intensively deformed zones, here shown in
the right corner sketch.
Page 72

Consequently, the role of each structural feature with respect to

gravitational instability will be revised as follows:

– the schistosity is not prone to rock sliding, since it regularly dips to

the SE (i.e. uphill).

– most of faults and shear zones of the Cassas bedrock are

recemented by diffuse carbonate veins; furthermore, the fault
surfaces are often rough, wavy or irregular.

– the bedrock of the landslide is strongly fractured, mainly by steep

to medium-angle faults and shear zones whose dip direction is
opposite or oblique to the dip direction of the slope.

Nevertheless, the intersection of the main structural systems points

out that sliding movements toward the WNW and NW can be
suggested (Fig. 4.29). These sliding directions that result from the
intersection of F2sz shear zones with F1a discrete faults and by
F2sz with F2 conjugate faults, are, respectively, oblique or quite
parallel with respect to the dip direction of the slope (NNW). The
WNW sliding direction is roughly perpendicular to the F3 system
which corresponds to the main scarp which splits the landslide body
into two parts longitudinally (Zones B and C). This sliding direction
coincides with the displacement recorded by the monitoring system
(C.T.M., 2003). In this kinematic framework, rock blocks bounded at
the base by F2sz shear zones, laterally by F2 faults and at the rear
by the F3 system could slide toward the WNW; in this case, the left
lateral boundary of the landslide could hinder the sliding movement
and could induce partial toppling of blocks.

The NW-dipping intersection line is, instead, more favourable to rock

sliding, but could involve in fact only small-sized blocks, since the F2
faults are usually closely (dm) spaced.
In conclusion, fast sliding of huge rock masses is unlikely on the
Cassas slope since the theoretical sliding direction that can be
inferred from the intersection of the main fracture systems is oblique
to the dip direction of the slope, or where sub-parallel, only small-
sized blocks are involved.

Furthermore, the physical and morphological characteristics of

fractures are not generally favourable to rock sliding. Consequently,
complex debris slide and flows, instead of rock sliding, have been
imagined for the assessment of hazard scenarios, as described in
the following paragraphs.
Page 73

4.3.4.4 Scenario identifications

Concerning the dangers recognised in this study, three main
typologies of movement are described: rockfall, debris slide and flow.
The rockfall involves only the main scarp sector, the debris slide and
flow involve several zones of the Cassas accumulation.

Figure 4.31. Rockfall trajectories (Epifani, 1991).

Rockfall accumulations are mainly located near the foot of the

Cassas main scarp. The average volume of unstable blocks, as
observed in the deposit, ranges from some m3 to 10–15 m3. The
only rockfall accumulation of about 50’000 m3 took place between
1957 and 1963; it is located at the main scarp foot. Concerning the
movement of great volumes in the crown zone, the instrumentation
data related to the open joints do not indicate any displacement at
the moment. Rockfall trajectories (Fig. 4.31) have already been
described (Epifani, 1991).
More difficult to explain is an interpretation of a general Cassas
landslide evolution, in particular for Sectors B and C, inside the
Cassas landslide accumulation of 1957, where important instability
elements are shown.

The hypothesis expressed in the present work is that the movements

in the debris body that were originated during the event of June 1957
remain constant over time, with possible acceleration during heavy
rainfall (for example during the event of October 2000). The greatest
volume corresponds to area B (about 1×107 m3) which was also the
mobilised volume in June 1957.

The debris body C presents a complex evolution with debris slide in

the top sector, next to Sector B, and slow debris flow in the middle-
lower area. During the paroxystic phase of 957–1965, this area was
involved in rapid debris flows partly accumulated here. Actually, slow
flow movements seem mainly concentrated in the upper level, as
indicated by inclinometer I2 located at the foot of the accumulation
(movement of about 30 mm/y to a depth of about 13 m; C.T.M.,
2003). Volume estimation of the debris body is, therefore, about
1.5×106 m3.

The continued movements measured over the last ten years in the
upper part of the slope emphasised an acceleration starting from
June 2001 (approximately 20 to 80 mm/y), caused by the heavy
rainfalls of October 2000 and by the winter/spring snow melting.
Page 74

The evolutionary hypothesis for these sectors is summarised as

follows (Fig. 4.32):

– slow and continuous movements in area C towards the

accumulation body;

– the slides in the upper part of area C produce instability in area B;

– due to progressive deformation of the debris accumulation or to

triggering events (ex. rainfall) that produce acceleration, it is possible
that one or more sliding surfaces are in limit equilibrium until the
instantaneous failure and development of a paroxystic phase; parts
or the entire body B and likely part of the body C would slide towards
the valley floor;

– the water quantity would condition the typology and the velocity of
the debris movement in the middle-lower area and subsequently the
ways of accumulation in the valley floor;

– the sliding area B could also condition the stability of area E, as

already happened in June 1957; the total volume of area E is about
1.5×106 m3.
Figure 4.32. Longitudinal sketch cross-sections of the Cassas
landslide: present situation (the letters refer to Figure 4.28) and the
evolutionary hypotheses.

The movement could include a total volume of about 10÷12×106 m3

with accumulation both on the slope and on the valley floor. The
accumulation would be lobe-shaped, considering the material
typology and the shapes of the oldest accumulations on the same
slope (Fig. 4.8). It is difficult to foresee the real extension of the
accumulation area on the valley floor, as it will depend on the
phenomenon velocity and on the real quantity of the material that
reaches the floor. Besides, the “funnel-shaped” slope, tightening at
the slope base, will strongly condition the extension of the
accumulation.

In the risk analysis survey by S.I.T.A.F. (2000), a first attempt to

determine the extension of the accumulation body was carried out
using a numerical model (cellular automata modelling): it supplies
the limits of the accumulation areas for different volumes, from
1×105 to 12×106 m3. For volumes inferior to 0.15×106 m3, the
material accumulates entirely on the slope.
To complete the framework of the danger assessment, it is
necessary to evaluate the areas external to the Cassas landslide. In
the downhill sector (F in Fig. 4.28) some scarps and localised
landslides are shown apparently as a continuum with the
morphological elements of the Cassas landslide area. The
morphological and structural-geological elements available do not
permit the definition of a correct model of the whole slope; the lateral
continuity of the morph-structures remains to be demonstrated.
However, a generalised slope movement is improbable in the case of
Page 75

heavy rainfall, whereas, it is possible to observe local instability

phenomena in the middle-lower area of the old accumulation at the
eastern edge of the Cassas landslide. In the western area of the
Cassas landslide (G in Fig. 4.28), characterised by a general
stability, the only critical sector seems to be the rocky spur (T in Fig.
4.15), placed in the upper part of the slope, next to the main scarp of
the Cassas landslide. In this area, there are fractured bedrock,
trenches and small debris accumulations of gravitational origin. In
this case, the hypothesis is that during a general slope movement
this sector would also move and that part of the material would flow
into the Cassas main accumulation and part would directly reach the
valley floor. The volume of this unstable area is about 0.5×106 m3.

In summary, the dangers identified in this study are:

– blocks or rock masses falling from the main scarp of the Cassas
landslide (velocity of some m/s) with an accumulation zone at the
scarp foot or at the slope. Each block has a maximum volume of
some tens of m3; the rockfall has a maximum volume of 105 m3;

– collapse of parts of the slope with debris slide and flow (variable
velocity from some m/min to some tens of m/month) with
accumulation partly on the slope, mostly on the valley floor.
Maximum volume inferior to 10×106 m3 (indicative volume useful for
the delimitation of the area affected). It is excluded, at present, that
the phenomenon could evolve as a debris avalanche;

– collapse of the whole debris body, also involving some parts

external to the “1957 Cassas”. Debris slide and flow (variable
velocity from some m/min to some tens of m/month), rockfalls of the
rock spur to the West of the Cassas landslide with trajectories that
partly converge in the accumulation body of the Cassas landslide
and partly reach the valley floor after passing over the wall above the
service station; involvement of the E area (Fig. 4.28) with complex
movements of debris slide type. Volumes of about 10÷15×106 m3.

The affected areas have been defined according to considerations

based on the available documentation and the geomorphologic
analysis:

– for the rockfall phenomena: by rockfall trajectories (Fig. 4.31);

– for the complex debris slide and flow phenomena (<107 m3):
morphologically based hypothesis for volumes of about 1÷2×106 m3;

– for the complex debris slide and flow phenomena (>107 m3):
morphologically based hypothesis for volumes of about 1.5×107 m3.

4.3.5 Occurrence probability

In spite of the available information on the Cassas landslide, among
which those related to the monitoring system data and to some
weather data, the interpretations of the event probability of the
different dangers previously recognised remain an hypothesis.

On the basis of the considerations explained in Chapter 3, it is

possible to present the following interpretation.

4.3.5.1 Rockfall phenomenon

Detachment of rock blocks of variable sizes from 1 to 20 m3 are
assumed to have annual frequency. In fact, although no direct
observation or historical documents of such a phenomenon are
available, it is supposed that cryoclastic processes and bedrock
degradation cause continuous detachments of intensely fractured
and released rock elements from the main scarp of the Cassas
landslide; the absence of vegetation on the debris at the scarp foot
underlines the high frequency of the phenomenon. Falls of rock
blocks of larger size (of about 105 m3) have a lower frequency,
about 101 years (in fact a rockfall of about 50’000 m3 took place
during the 50s–60s of the past century).
The rockfall frequency was applied to the maximum extension of the
area affected by rockfall, without distinguishing between contiguous
areas (high probability of rockfall trajectories) and areas far from the
main scarp (low probability of rockfall trajectories).
Page 76

4.3.5.2 Complex debris slide and flow phenomena

It is difficult to establish an occurrence probability for phenomena
which would involve the whole slope. Considering the sparse
historical data available, the heavy meteorological events of May
1728 and of June 1957, an interval of about 230 years is obtained.
Such a recurrence time span is representative for the whole slope
around Cassas and not only for the specific Cassas landslide, as the
documents of 1728 refer to a generic “landslide opposite Eclause”
(Fig. 4.8), therefore, on the right slope, that “temporarily would have
dammed the Dora river” (Bogge, 1975).

An attempt to obtain the occurrence probability can be made by

linking the observed effects on the slope to the probable triggering
causes. It is not easy to discover a link between rainfall and the
movement velocity, as the inclinometer measurements are carried
out periodically at irregular intervals over a time span of less than 10
years. An attempt to correlate rainfall with movement was carried out
by the society that manages the monitoring system (C.T.M., 2003).
This study shows a relation between the “accelerations” recorded by
the inclinometer I3 (Fig. 4.27) and the average precipitation over the
last 90 days; in particular, a prominent acceleration of the movement
is recorded in correspondence with the maximum of the average
precipitation over the last 90 days in October 2000 (Fig. 4.33).
Considering this information and the rainfall data of the Salbertrand
weather station, the return time for such a maximum of the average
precipitation over the last 90 days was calculated using the Gumbel
method.
Figure 4.33. The Cassas landslide: 90 days average precipitation
and inclinometer data relationship (C.T.M., 2003, modified).

This elaboration gives a return time of 50 years which can be

considered the minimum threshold after which the Cassas landslide
will have important accelerations induced by heavy rainfalls.

The recurrence time span is then assumed as ranging from 50 to

about 250 years. This recurrence time span is calculated for an
event with movement of a debris volume less than that of the 1957
event (<1×107 m3). During the 1957 event, a volume of about 10
million m3 of bedrock, intensely fractured and released, slid along
pre-existing surfaces (structural surfaces). After a first phase of
acceleration, during which the landslide did not reach a velocity
higher than that of a rock-debris avalanche, a period of progressive
slowing of the movement followed which lasted about ten years.
Gravitational phenomena that would involve volumes superior or
equal to 1×107 m3 (until a max. of about 1.5×107 m3 for each event)
have, on the basis of previous considerations, a “return time”
greater than 230–250 years.

The return times cited above are indicative, considering the

complexity and the large uncertainty of the system. For example, a
hydrogeologic model of the Cassas landslide is not still available,
permitting a relation among rainfall-infiltration-movements. The
morphological features, the grain size distribution in the
accumulation and the high fracture intensity suggest that the
hydrogeologic
Page 77

system could be very complex. The available piezometric data

confirm this interpretation, as the variation of the groundwater depth
does not seem to be in direct relation with rainfall. Recent studies
(Gianquinto, 2000), aiming to establish a hydrogeologic model for
the Cassas slope, suggest the high permeability of both the landslide
accumulation and the intensively fractured rock mass. The role of the
latter in water circulation is unknown; it is not known if the bedrock
has either a draining role or a feeding one or, probably, both
functions, related to the overhanging debris accumulation.

For a quantitative risk assessment, the frequency of debris slides

and flow have been applied only to the expansion areas of the
gravitational phenomenon in the paroxystic phases (area affected on
the valley floor). The high frequency values (=max frequency value
recognised) have been applied to the whole accumulation body of
the “1957 Cassas” landslide which, according to the instrumental
data, is still continuously deforming.

Table 4.4. Occurrence probability.

Scenario Occurrence range Frequency

(average)

Rockfalls 1–50 (25)1/25=0.04

Debris slide & flow (<1×107 m3) 50–250 (150)1/150=0.007

Debris slide & flow (1÷1.5×107 2501/250=0.004

m3)

4.4 QUANTITATIVE RISK ANALYSIS

Three scenarios for the Cassas landslide evolution have been

assumed. Each scenario is characterised by a proper intensity, by a
triggering and run-out areas (derived from the geomorphological
model) and by an occurrence probability (as a result of the historical
analysis of recorded events).

According to the proposed quantitative risk analysis, each element of

the hazard scenario is used separately in the matrix calculation of
risk evaluation. This process is applied to each recognised scenario;
therefore, for the Cassas landslide three separate risk analyses have
been carried out.

At the end of the matrix calculation, a unique value is obtained (total

risk), divided into 4 categories: physical, economic, environmental
and social, useful for the total risk assessment. The three scenarios
are:

A. rockfall phenomenon: volumes 10°–105 m3; area affected:

obtained by rockfall trajectories;

B. complex debris slide and flow phenomena: volumes <107 m3;

area affected: morphologic hypothesis for volumes of about 1–2×106
m3;

C. complex debris slide and flow phenomena: volumes 1–

1.5×107 m3; area affected: morphologic hypothesis for volumes of
about 1.5×107 m3;

Risk analyses are based on land use defined by Salbertrand town

planning and by the areas affected as previously indicated.

4.4.1 Elements at risk

The elements at risk are recognised through the definition of the
area affected by each danger in the scenarios previously specified.
These elements at risk are shown in the following tables and in
Figure 4.34.
Page 78

Figure 4.34. Elements at risk next to the Cassas landslide.

Table 4.5. Rockfall scenario: element at risk identification.

Element at risk Name Notes

Forests (private Bosco Chapel Area affected: 0.4 km2 Persons

and public (Piccolo bosco) affected: 2 (occasional presence of
properties) monitoring operators)

Special elementsInstruments of the Inclinometers

Cassas monitoring
system Piezometers

Geophones

Extensometers

Laser distance measuring

Table 4.6. Debris slide & flow scenario (< 107 m3): element at risk
identification.
Element at risk Name Notes

Forests (private and Bosco Chapel Area affected: 0.23 km2 Persons
public properties) (Piccolo bosco) affected: 2 (occasional presence of
monitoring operators)

Rural areas — Area affected: 0.06 km2 Persons

affected: <5 (occasional farmer)

Transportation Torino-Frejus Area affected: 0.05 km2 Motorway

facilities(motorway motorway affected: 400 m Persons affected:
and international Motorway 50÷100 (service station operators,
railways with service station motorway and service station
corresponding customers)
services)

Valuable Dora Riparia Area affected: 0.1 km2

environmental areas river bed Inclinometers Piezometers
Special elements Instruments of Geophones Extensometers Laser
the Cassas distance measuring
monitoring
system
Page 79

Table 4.7. Debris slide & flow scenario (1÷1.5×10 m3): element at
risk identification.

Element at risk Name Notes

Forests (private and Bosco Chapel Area affected: 0.6 km2

public properties) (Piccolo bosco)
Persons affected: <5
(occasional presence of
monitoring operators, people
going to the Sapè of
Salbertrand and Exilles)

Rural areas Area affected: 0.08 km2

Persons affected: <10

(occasional farmer)

Transportation Torino-Frejus Area affected: 0.082 km2

facilities (motorway motorway Motorway
and international service station Torino-Motorway affected: 800 m
railways with Modane international
corresponding railway Persons affected: 50÷150
services) (service station operators,
motorway and service station
customers)

Railway affected: 900 m

Productive areas Area affected: 0.09 km2

Persons affected: 10

Residential areas Gravere (Salbertrand) Area affected: 0.005 km2

 Persons affected: <10

Valuable Dora Riparia river Area affected: 0.13 km2

environmental areas bed
Area affected: 0.03 km2
Local services and
plants Persons affected: <5

Special elements Instruments of the Inclinometers

Cassas monitoring
system Piezometers

Geophones

Extensometers

Laser distance measuring

4.4.2 Vulnerability of the elements at risk

To define the vulnerability of the elements at risk, an analysis about
involved assets typology and processes’ intensity has been carried
out.

In the rockfall scenario, the processes’ intensity is not very high, as

trees and debris would slacken the rock blocks rolling along the
slope. The monitoring instruments do not resist rock block impacts,
even those with low energy. In the technical reports related to the
monitoring, difficulties in obtaining measurements due to damages to
the instruments are recorded. Therefore, it has been decided to
assign the elements at risk in the area affected a vulnerability value
of 1 (100%).

In the debris slide and flow scenarios, a maximum intensity has been
given to the recognised processes, as the involved volumes are
huge (millions or tens of millions m3). Considering the elements at
risk involved in the gravitational phenomenon (buildings,
infrastructures, roads and railway), a vulnerability value of 1 (100%)
was considered: in the area affected (failure, depletion and
accumulation areas) the destruction is considered total. Also, at the
valley floor, when the velocity tends to zero, the accumulation can be
so big that all the assets can be completely destroyed, heavily
damaged or buried and, therefore, permanently useless. The velocity
and thickness distribution in the accumulation area depend on the
initial velocity and on viscosity (water quantity in the material in
movement).

In the quantitative risk analysis, it is important to consider the system

uncertainty and most of all the delimitation of the areas affected, as it
is the result of hypotheses and assessments. Therefore, the
elements at risk next to affected areas cannot be considered to have
zero vulnerability and, consequently, zero risk. In particular,
concerning the debris slide and flow scenario >107 m3, the area
affected reaches the railway embankment, but it is possible that the
whole railway would be impacted with heavy consequences for traffic
in both directions.
Page 80

4.4.3 Expected impact

Multiplying the values of the elements at risk and the vulnerability
percentage (obtained as described in the previous section), the
expected impact is obtained.

The affected areas recognised, even if hypothetical and

approximate, underline the problem related to the presence of a
motorway at the foot of the Cassas landslide. The A32 motorway
Turin-Frejus tunnel, the only tunnel between Italy and France during
the Mont Blanc tunnel closing due to an accident, is of international
economic importance.

Concerning social consequences, being a heavy traffic road, the

values are particularly elevated. If a landslide paroxystic phase
occurs, probably many people would be affected.

This probability rises in the sectors involved in the event of 1957,

when the service station “Gran Bosco” was built. Furthermore, the
accumulation area at the valley floor is next to the tunnels on the left
slope of the Susa Valley, being a problem for those who arrive from
Turin if the traffic would not be stopped.

The same consideration must be made concerning the Turin-

Modane railway, which runs in tunnels before arriving at the
Salbertrand plain next to the Cassas landslide.

In order to state a precise quantitative damage evaluation for

economic activities and assets, costs derived from economic
analyses are essential (e.g. asset costs, reconstruction costs,
turnover, profits loss, etc.). To assess environmental and social
expected impacts (where an economic value is difficult to obtain),
relative values indices are used.

4.4.4 Risk assessment

The last phase of the quantitative risk analysis concerns risk
evaluation. This evaluation is obtained by multiplying the expected
impact values, obtained previously, and the numeric values of the
occurrence probability related to each scenario.

In the case of the Cassas landslide, as previously reported (§ 3.5),

the values of the occurrence probability were obtained on the basis
of an analysis of historical data (Table 4.4).

Risk assessment results are summarised in the images from G.I.S.

and described in the related captions.

As shown in Figures 4.35, 4.36, 4.37 and 4.38, for the debris flow
and slide scenarios, the major risks are connected to the landslide
body on the slope: this situation is possible because the landslide on
the slope is active and, therefore, with high frequency (≥1).For the
correct considerations of the real risk, only that which exists in the
accumulation area on the valley floor must be evaluated.
Page 81
Figure 4.35. Map obtained by G.I.S. techniques concerning the
physical risk for the 3 scenarios analysed.

For the rockfall scenario, no distinction exists, as it concerns a

homogenous area. Risk values are not zero, as a very expensive
monitoring system exists which is very important for the motorway
and service station operation. A detailed analysis can be made
considering only the instruments correctly located.

In the two debris slide and flow scenarios, the major values of
physical risk are, first, those of the service station “Gran Bosco” and,
second, those of the motorway and the areas classified as
productive areas in the land planning.
Page 82
Figure 4.36. Map obtained by G.I.S. techniques concerning the
economic risk for the 3 scenarios analysed.

As it can be observed in the images, the areas with major risk values
are the A32 motorway (Turin-Frejus tunnel) and the service station.
Particularly important is the Dora Riparia river, as well as the intrinsic
value of the river, its waters are captured downstream of the Cassas
landslide by a power plant.
Page 83
Figure 4.37. Map obtained by G.I.S. techniques concerning the
environmental risk for the 3 scenarios analysed.

The main areas with major risk values are, obviously, the slope
areas (in the Natural Park “Gran Bosco di Salbertrand”) and the
valley floor areas where the Dora R. river runs.
Page 84
Figure 4.38. Map obtained by G.I.S. techniques concerning the
social risk for the 3 scenarios analysed.

Concerning the rockfall scenario, the risk values are not zero as
periodically 2 monitoring system operators are at the landslide.

For the debris slide and flow scenarios, the areas with elevated
social risk are the Turin-Frejus motorway tunnel and the related
service station. A detailed analysis shows how traffic variation due to
time and seasons leads to different considerations about risk
evaluation.
Page 85

4.4.5 Total risk

In the images (Fig. 4.39) the results of the quantitative total risk
analysis for each category (physical, economic, environmental and
social risk) are reported. As explained in the chapter concerning
quantitative risk analysis methodology, the total risk can be obtained
adding the risk value for each category of each scenario.

As already detailed for the risk analysis, the high risk index obtained
for the slope sectors involved in the active phenomenon must not be
taken into account. This derives from the fact that for the Cassas
landslide, a high occurrence probability value has been considered
and this leads to a high risk value, even if linked to a low value of the
affected assets. Therefore, for a correct risk analysis, it is necessary
to consider the affected areas at the valley floor whose occurrence
probability values are 10 times inferior to those on the slope and that
reach high risk values for the important consequences evaluated.

Concerning the assets evaluation, the analyses are not detailed, but
the relative values obtained by the methodologies have been
applied. Probably, to obtain a correct framework of the risk situations
related specifically to the Cassas area, careful social and economic
evaluation should be done on the basis of the considerations of the
next paragraph (§ “Indirect risk and regional impacts”).

In any case, it is important to underline that the motorway and the

service station affected by the landslide evolution present high risk
indices regarding the physical, economic and social total risk.
Particularly important in the analysis of the environmental total risk is
the Dora R. river bed.
Figure 4.39. Total risk maps.

4.5 INDIRECT RISK AND REGIONAL IMPACTS

It is particularly difficult to understand the indirect risks associated

with the processes related to the recognised scenarios. This
difficulty, in particular for debris slide and flow processes, is due to
Page 86

several uncertainties about the evolution of the phenomena, for

example:

– What volume will be exactly involved in the movement?

– How much material will deposit on the slope and how much will
reach the valley floor?

– Which is the maximum velocity of the material in movement?

This will depend on the water quantity and on the related viscosity
and velocity of the mass in movement.

In the worst case, in which the catastrophic scenario shows the

landslide blocking the valley, a dammed lake would be formed. It is
presumed that the landslide would occur during a critical
meteorological event, creating important flows along the Dora
Riparia river (some hundreds of m3/s).

The combination of these processes (valley floor dam with tens of

meters of landslide accumulation and river flood) would create a
basin of some hectares in few hours. Some villages in the valley
would be then overflowed with some damages to human assets and
the evacuation of several people, rising the social risk value.

Moreover, due to lake formation, downstream villages could

potentially be involved. Instantaneous lake depletion can, in fact, be
caused by a rapid overtopping dam erosion, or a landslide dam
collapse caused by piping and internal erosion. The processes that
lead to a sudden lake depletion are mostly:

– threshold erosion with instantaneous dam break when the lake

reaches the maximum level;

– piping erosion causing dam break.

In both cases, the propagation of a flash-flood wave in the valley
would create important problems which are difficult to evaluate. In
this specific case, no analysis of sudden collapse has been carried
out yet.

Furthermore, due to a valley dam, the power plant of Chiomonte,

which catches the water of the Dora Riparia river in Serre la Voute,
would be closed with important economic consequences for the
entire valley.

The presence of the motorway and of the railway in the landslide

area has already been described. They have great importance from
a military and economic point of view. The Frejus tunnel was the only
passage to France (open all year) during the closure of the Mont
Blanc tunnel following a bad accident. The economic damages
caused by the motorway and railway closing would be of about 106 €
with relapses on the economy of Italy and France (Fig. 4.40).

Figure 4.40. Synthesis of the meaningful elements at risk.

Page 87

The motorway and the railway play an important role in the economic
development of these mountain areas, as they are linked to tourist
resorts. The Susa Valley and the Chisone Valley are two important
ski resorts connected to other ski stations in France: international ski
competitions take place in this area every year and at this time new
infrastructures are developing for the winter Olympic games of 2006.

4.6 REGULATION AND RISK MITIGATION MEASURES

ALREADY AVAILABLE

The motorway and the service station “Gran Bosco” at the foot of the
Cassas landslide raise some problems related to risk mitigation,
problems never taken into account either in relation to the
Salbertrand village or in relation to other infrastructures in this area.
The S.I.T.A.F. (the society that manages the motorway) has been
studying the best protection both for the motorway and the service
station since the end of the ’80s of the last century.

At the beginning of the ’90s in the last century, S.I.T.A.F. built a

reinforced earth wall, 120 m long and 5–6 m high, to protect the
motorway service station from both debris flows and rockfalls.
Rockfall fences (Fig. 4.41) were also installed at the accumulation
foot and at the rock wall base over the service station.

Over the same period, a monitoring system was installed in order to

measure the phenomenon with the following aims:

– to improve the knowledge of the phenomenon in order to evaluate

the risk statistically and numerically;

– to create relations among instrumental data for landslide

forecasting.
The monitoring system is composed of: inclinometers, piezometers,
geophones, extensometers; recently, a distance measuring laser
was installed to evaluate the surface movements of the secondary
scarp crossing the “1957 Cassas”. The rock wall over the service
station is also monitored with automatic measurements in real time.

The monitoring system, managed by the motorway company, acts as

a warning system for both the service station and the motorway. In
future, it will be connected to a larger warning system, according to
the general civil protection plans of the area, but at the moment the
thresholds are not officially recognised.

Figure 4.41. Rockfall fence at the Cassas landslide foot.

Page 88

This page intentionally left blank.

Page 89

5
The Rosone landslide

G.Amatruda1, S.Campus2, M.Castelli1, L.Delle Piane3, F.Forlati2,

M.Morelli4, L.Paro2, F.Piana4, M.Pirulli1, R.Polino4, M.Ramasco2,
C.Scavia1
1Politecnico di Torino; 2Arpa Piemonte; 3Consulting geologist;
4CNR-IGG sezione Torino

5.1 INTRODUCTION

The entire southern side of the ridge bounded by the Orco and
Piantonetto Rivers, in the Province of Turin, has been undergoing a
slow process of deep-seated gravitational slope deformation
(DSGSD). This deformation process, a vast landslide movement, is
historically known as the “Rosone landslide” because it frequently
involved the small village of Rosone located at the toe of the slope.
At present, this phenomenon also involves the 99 MWh hydroelectric
power plant of the Electricity Agency (AEM) of the city of Turin (Fig.
5.1). Coming from the Ceresole Reale dam, the water reaches the
AEM facilities through a 17 km long pipe after spanning the entire
length of the gravitational deformation, where it falls towards the
power plant via a penstock with a drop of 813 m. At the toe of the
slope, the only National Road (No. 460) from Turin to Ceresole
Reale is located.

5.2 GEOLOGICAL BACKGROUND OF THE LANDSLIDE

The so-called Rosone landslide involves an area of about 5.5 km2

and reaches a depth of over several decametres (Ramasco et al.,
1989), and affects a 1300 m high slope, from an altitude of 2000 m
at the ridge crest down to 700 m at the valley bottom (Fig. 5.2 and
5.3).
The morphological and structural characteristics of the area suggest
subdividing it into three adjacent sectors, roughly corresponding to
the villages of Ronchi Perebella, and Bertodasco (Fig. 5.2 and 5.3).
These sectors reflect, respectively, final, early, and intermediate
stages of the evolution of a site subject to a deep-seated
gravitational process. On account of the presence of such widely
diversified conditions in a single area, this site turns out to be
particularly suited for stability studies and their comparative
evaluation (Forlati et al., 1993).

Examining the area from west to east, the Ronchi sector is

encountered first, characterised by a highly advanced stage of
evolution of the deformation that has caused the disruption of the
original rock formation. This sector may now be assumed to be
substantially stable.

The central sector, around Perebella, reflects an early stage in the

deformation process. It is clearly separated from the Ronchi sector
by a N-S striking scarp a few hundred meters long and 50–90 m
high. The bedrock in this sector is more damaged in the middle and
lower parts.

The eastern sector (Bertodasco), which is inhabited and hosts the

AEM electric power plant, shows a deformation stage intermediate
between those of the Ronchi and Perebella sectors. Previous studies
(Forlati et al., 1993) identify the Bertodasco sector as the one most
likely to undergo a catastrophic evolution. A morphostructural
analysis revealed the presence of three zones with different degrees
of mobility, from the top to the floor of the valley, referred to as A, B
and C (see below).
Page 90

Figure 5.1. a) Digital Terrain Model of the north-western Piedmont

with the location of the Rosone landslide area (red square), main
rivers and lakes (in blue), and villages and town (in red). b) General
overview of the Orco and Piantonetto watershed (air photographs
overlapped Digital Terrain Model) with instability phenomena
surveyed in the surroundings of the Rosone landslide area (IFFI
Project—Gravitational Phenomena Inventory in Italy, ARPA—
Piemonte, 2004). The coloured areas show: DSGSD (orange),
complex landslides (violet), rockfall (yellow), rock slide (light green),
fall and topple areas (red).
Page 91

Figure 5.2. Morphostructural map of the Rosone deep seated

gravitational slope deformation showing the subdivision into distinct
morphological sectors: Ronchi, Perebella and Rosone. Legend: a:
boundary of main rocks subjected to mass movements; b: trace of
the main slip surfaces or scarps; c: Bertodasco sector—Zone A; d:
Bertodasco sector—Zone B; e: Bertodasco sector—Zone C; f:
Ancient landslide deposits; g: (a) debris due to disruption of rock
mass (b) fragmented rock mass where the original structural features
are preserved; h: debris flows of the 1953 event (from 1954 aerial
photos); i: debris cones; 1: rockfalls; m: boulders; n: trenches and
opened fractures; o: traces of the main joint systems; p: zone of
extensive waste materials; q: boreholes location; r: M-M’: trace of
section of Figure 5.18 (Modified after Brovero & al., 1996b).
Page 92

Figure 5.3. Sketch of the Rosone landslide (Brovero et al., 1996b).

5.3 HISTORICAL BACKGROUND

The Rosone landslide underwent two major paroxysmal stages, one

in the early 18th century (1705–1706 from an inspection report “Atto
di visita” in Luino et al., 1993) and another in the fallwinter of 1953;
they caused severe damage to many buildings and the disruption of
cultivated fields.

From historical reports, in 1933–1934 the inhabitants were

evacuated for about 7 months. In the early 1940’s, a landslide
occurred at an altitude of 1300 m: huge boulders toppled downhill
and threatened Rosone and the eastern part of Bertodasco. In 1940,
the Bertodasco sector was affected by increasing slide movements
with the triggering of flows.

The landslide showed signs of re-activation in 1948 and 1951. In the

fall-winter of 1953, following abundant precipitations, a portion of the
slope collapsed. The movement of the ground in the Bertodasco
area damaged or even destroyed some houses. The fall of debris
and boulders from the slope overhanging Rosone caused the
evacuation of the 250 inhabitants and their cattle. Due to the
consequent damages, a new Rosone village (named Rosone nuovo)
was built in 1956. This village is close to the AEM power plant.

Between 1953 and 1957, these movements gradually slowed down,

as confirmed by topographic measurements carried out by AEM. In
1957, the two villages Rosone and Bertodasco were evacuated.

Between 1957 and the early 1960’s, the movements accelerated and
phenomena similar to those recorded in 1953 were observed, though
less severe. Further phenomena were recorded in the fall of 1963
and in the spring of 1964 and 1969, 1988, 1993.

In the 1993 and 2000 important displacements were also recorded

(several centimetres) immediately after heavy meteorological events.

At present, significant movements continue to take place in the

upper part of the Bertodasco sector, as borne out by the topographic
measurements performed on the anchoring blocks of the penstock.

A summary of the historical events is shown in Figure 5.4.

5.4 REGIONAL FRAMEWORK

5.4.1 Climatic and water conditions

The climate is characterised by a pre-alpine regime. The
thermometric regime is characterised by 4–6 frost months per year.
The Ceresole Reale station, located at an altitude of 1600 m,
Page 93

Figure 5.4. Historical data synthesis concerning slope instability

(Rosone landslide).

registered 184 frost days (1950–1986) with a probability of 80–100%

of frost days from December to March.

On the Graie Alps, the pluviometric rate is of a pre-alpine type and

the average annual number of rainy days varies from 90 to 110. On
the Orco basin, average annual precipitations are about 1224
mm/year and average annual rainy days are 96. The average daily
intensity is about 12.8 mm/day. Average annual rainfall has been
determined over a period of 42 years, from 1938 to 1980. The
above-mentioned results refer to a station placed at an altitude of
700 m, at the bottom of the slope in question.

Since 1990, a new pluviometric and thermometric station has been

activated at Bertodasco (1120 m) corresponding to the average
altitude of the slope.

The seasonal distribution of rainfall can be summarised as follows:

May: the highest monthly rainfall value (160 mm), the greatest
average number of rainy days (12 days) and the highest cumulative
rainfall values are reached in three or four days (up to nearly 195
mm).

September: heavy downpours of relatively short duration.

October: the monthly average is similar to that of May (155 mm–160

mm), but the average number of rainy days is modest.

On a regional scale, snow rates mainly vary according to altitude. On

the Graie Alps, snowfalls are frequent at an altitude between 2000 m
and 2300 m and they have their maximum in April and a secondary
maximum in February. At an altitude above 2300 m, snow rates have
a single maximum in April. At an altitude between 1200 m and 1700
m, snow rates are unimodal, showing a maximum in February
(Ceresole Reale at 1573 m). The maximum snow level recorded at
Rosone is about 2 m (1994–1995), the minimum level is 0.1 m
(1966) and the average is 0.35 m. In the Orco Valley, the maximum
snow level was recorded during the 1989–1990 winter, preceded by
the
Page 94

minimum recorded during the 1986–1987 winter. The absolute

monthly maximum in the last thirty years was measured in January
at the Rosone station, located at and altitude of 700 m.

Temperature increases (10°–15°) have been recorded in the last

decade for February, testified by rapid snow melting.

It is of interest to point out that two perpetual springs are found near
Perebella and Bertodasco and some temporary springs at the toe of
the slope and near the right flank of the Bertodasco sector. The flow
of water in the slope occurs through the main discontinuities, which
are generally open and persistent. Considering that the upper part of
the slope acts as a reservoir, the largest part of the groundwater is
cumulated when rainfalls coincide with the melting of the snow cover.

5.4.2 Regional geology and structural settings

The landslide site is located near the confluence of two glacial
valleys, namely the Orco and Piantonetto valleys. In fact, the
geomorphology features of the area are due to fluvial-glacial
morphogenesis although modified by gravity deformations. Slopes
are usually very steep and the valleys are narrow.

The Orco Valley is located in the central part of the Gran Paradiso
Massif (Fig. 5.5). The geological unit belongs to the Pennine domain
and it consists of a composite crystalline basement and a Permo-
Liassic cover, locally preserved in the peripheral areas. The Gran
Paradiso Massif consists of three different complexes (Compagnoni
et al., 1974):

– Augen Gneiss Complex: augen gneisses and fine-grained

gneisses with inter-bedded metabasites.

– Monometamorphic Complex of the “Money” area: albite

micaschists and graphite-bearing, gneisses inter-bedded with
quartzose meta-conglomerates.

– Erfaulet Ortogneiss Complex: leucocratic ortogneiss.

Figure 5.5. Geological map of a sector of the Western Alps (CNR,

Structural Model of Italy 1:500000 scale). In the red circle the
landslide site is placed within one of the main Alpine tectono-
metamorphic units (Gran Paradiso Unit GPU) in the axial part of the
chain, bordered on both sides by the Piemonte Zone (Penninic
domain, in yellow). East of GPU are the Austroalpine and South
Alpine domains (dark red and brown colour), the Canavese Line
(CL), a major fault of the Alps and Po Plain, while west of GPU,
Briançonnais (light blue) and Mont Blanc domains (red) are
represented.
Page 95

All of these units are locally covered by glacial deposits and they are
cut by two main joint sets, mutually orthogonal, corresponding to E-
W and N-S striking sub-vertical faults.

The metamorphic deformations are mainly recorded by a regional

schistosity (Sr in Fig. 5.11) coeval with the development of regional
folds that have been later refolded with structural styles (open folds).

The post-metamorphic deformations are characterised by major

strike-slip tectonics and minor thrusting (Perello et al., 2004). Later
extensional brittle faulting is also recorded.

5.5 HAZARD ANALYSIS

5.5.1 Geomorphological analysis

Following systematic geomorphological and structural analyses
carried out in recent years, the Ronchi, Perebella and Bertodasco
sectors are described here in detail:

– The Ronchi sector

In this area, the deep seated slope deformations have reached the
maximum evolution level. The gravitational processes, having lasted
here for a long time, have erased any evidence of glacial
morphology and strongly modified the profile of the slope. In ancient
times, the slope failed and rockslides occurred.

The slope shows, at present, large elongated depressions and

double-crested ridges in its upper part, while in the intermediate and
lower parts a prominent bulge formed and partially obstructed the
Orco valley.

This morphology is mainly due to a settlement in the upper zone

(immediately below the double-crested ridges), which occurred along
several hundred meters of quite continuous scarps of circular shape
and with sub-vertical slip surfaces; this settlement has also been
partitioned along the discontinuities of the highly fractured rock mass
and gave origin to many large undulations of the slope. It also
induces the formation of a prominent bulge in the lower part of the
slope, in response to dilatation.

Displaced material actually appears in the form of big rock blocks of

several cubic meters (up to 2000–3000 m3 meters for a single block)
distributed, together with heterogeneous debris, over the whole area
(Fig. 5.6). In the lower part of the Ronchi sector, some minor ancient
landslides are present

Figure 5.6. Aerial view of the high part of the Ronchi sector. The
boundary main scarp, on the right, as well as the intense disruption
of rock mass, is depicted.
Page 96

– The Perebella sector

This sector shows a preliminary stage of evolution (minor frequency

of released joints and minor disruption of rock mass). Ridge-top
trenches with graben-like depressions and doublecrested ridges,
also surveyed in the Rochi sector, characterise the upper part of the
slope, where along the main discontinuities dislocative behaviour
(shear and tension stress) is observed (Fig. 5.7). In the intermediate
and lower parts of the slope, the gravitational deformation gave
origin to many uphill facing scarps and in some places very unstable
piles of boulders (Fig. 5.8).

The control of the structural setting on the instability phenomena

appears quite clear in the middle part of the slope. The cross-cutting
relationships between the discontinuity system subparallel to the
schistosity, corresponding to the dip direction of the slope, and the
two main sets of sub-vertical discontinuities develops a toppling
mechanism of instability. The rock mass dilatancy due to severe
confining stresses determines a particular “arch” slope configuration.

Finally, since the original (tectonic-related) structural features are still

locally preserved in the Perebella sector, the cross-cutting relations
among the main joint systems have been extrapolated over the
whole substratum of the Rosone area (Fig. 5.9).
Figure 5.7. Double-crested ridges along the Chisone-Piantonetto
watershed of the Perebella sector.

Figure 5.8. Up-hill facing scarps in the upper part of the Perebella
sector.
Page 97

Figure 5.9. Aerial view of the medium part of the Perebella sector:
cross-cutting relations of the main steeply dipping joint systems are
well exposed.

– The Bertodasco sector

This sector shows an intermediate degree of evolution with respect

to that of the Ronchi and Perebella sectors. While the upper part of
the slope is characterised by trenches perpendicular to the direction
of the major movement, the middle part is affected by a rotational
slide movement superimposed on a general toppling and planar
slide. The disarticulation degree of the rock mass increases from the
top to some hundred meters over the slope foot.

The morphostructural evidences and the different kinematic

behaviour allowed the identification of three minor zones (A, B, C in
Fig. 5.2 and 5.3).

– Zone A includes the upper part of the Bertodasco sector, where the
movements of the slope (as a whole) and related morphological
evidences, are poorly defined. Moreover, although disrupted, this
zone still preserves its original structural features. This suggests the
presence of minor translational movements, as also confirmed by the
AEM monitoring system (Forlati et al., 1993).
The eastern lower part of Zone A shows more prominent
movements, as indicated by opened fractures, and is affected by
undulations and local bulges. The rocks here are split into blocks,
although still aligned along the main joint systems.

– Zone B is bounded by a prominent curved scarps and by rectilinear

scarps sub-parallel to the main joint systems. These scarps often act
(at least since the 1953 crisis) as slip surfaces, with movements of
3–4 meters that occurred along quasi-planar surfaces, sub-parallel to
the schistosity. These kinematics are strictly related to the shape and
orientation of the rock blocks resulting from the intersection of the K1
and K2 joint systems. The gravitational instability of Zone B also
induced a partial translation of the eastern lower part of Zone A, and
caused the displacement of the pipes of the AEM power plant. At the
eastern boundary of Zones A and B, toppling and rolling of rock
blocks have been observed. These rock blocks frequently moved
towards the ancient Rosone village and as a consequence the new
Rosone village (Rosone Nuovo) was established.

– Zone C consists of very heterogeneous mobilised mass, as it has

been affected by major gravitational movements since the 1930’s. It
consists of disengaged rock blocks of different size; a mixture of
chaotic coarse gravel and fine material. The movements mainly
affect the upper part of the zone along many concave and sub-
parallel scarps characterised by an offset of 1 to 10 meters. These
movements have also determined an abrupt change in slope
morphology.

The movements in the upper part of this zone have caused, during
the last fifty years, the break-down of part of the Bertodasco village,
which is now completely abandoned.

The lower part of Zone C is instead affected by rock-debris and

debris flows that have destroyed the National Road (SS. 460) that
runs along the Orco valley (Fig. 5.10).
Page 98

Figure 5.10. View of the Bertodasco sector where the origin and
deposit zones of debris flows are well exposed.

5.5.2 Local geology and structural analysis

5.5.2.1 Lithological features
The Rosone area is made up of intensively laminated orthogneiss,
within minor levels of white micaschists and rare chloritoschists. The
micaceous levels in the orthogneisses gave origin to anisotropy that
causes subdivision of the rock masses into differently thick (cm to
metre) massive slices, bounded by downstream dipping slip plains.
These slices are sub-parallel to the main schistosity, which here is
highly pervasive, as it represents the axial surface of isoclinal folds.
At the slope scale, the anisotropy of the rock masses gave origin to
plain-parallel surfaces that show high lateral continuity
(Geoengineering, 1984; Brovero et al., 1996b; Perello et al., 2004).
In some places, the white micaschists correspond to mylonitic levels.

5.5.2.2 Mesostructural analysis

On site available data

(a) Regional schistosity

The main schistosity (Sr) dips to N150 with an inclination of 35° and
is characterised by a regional metamorphic foliation corresponding to
the axial planes of metric to hectometre-scale tight to isoclinal folds
that refold an older pervasive foliation.

The schistosity is locally developed parallel to silvery micaschist

levels (see below for further details) that represent high strain shear
zones, due to metasomatic reactions along high pressure mylonitic
shear bands (Dal Piaz & Lombardo, 1986). Though discontinuous,
these silvery micaschists are to be carefully considered in the
stability analysis of the landslide, since they are characterised by
lower shear strength with respect to the average values of the Gran
Paradiso orthogneiss.

A joint system is often developed parallel to the schistosity (Studio

Geologico Italiano, 1984; Geoengineering, 1984).

(b) Fractures and fault systems

Two main joint systems have been recognised (Fig. 5.11). They are
the K1 and K2 steeply dipping (75°–85°) systems, which strike N10
and N90–N100 respectively (Fig. 5.9).

Two other minor joint systems are also present. They are the K3
system, that dips 70° to N240 and the K4 system, that is steeply
dipping and strikes N40 (Geoengineering, 1984).
Page 99

Figure 5.11. 3D sketch of the main tectonic discontinuities of the

Rosone area. The poles of joint systems and schistosity are shown
on a Schmidt diagram.

Around site data

As mentioned previously, the Rosone landslide affects a pre-

quaternary basement made of coarsegrained orthogneisses, deriving
from alpine metamorphic evolution of hercynian granites. These
granites were pervasively deformed along mylonitic shear bands,
where metasomatic processes were prevalent, and caused a total
rock volume reduction as great as 50%. The products of such
transformations are known as white schists (Schreyer, 1977) or
silvery micaschists (Compagnoni & Lombardo, 1974; Chopin, 1981;
Dal Piaz & Lombardo, 1986).

Geological surveys on the Rosone landslide and the adjoining areas

(SEA consulting, 2001) allowed the observation, within the
orthogneisses, of the presence of several decimeter to decameter
levels of silvery micaschists, locally folded with an axial plane
corresponding to the main schistosity, but generally transposed and
sub-parallel to it. The distinctive paragenesis is represented by
quartz, phengite, chloritoid, ±talc, ±Mg-chlorite; in the Rosone area,
the Mg-chlorite may predominate, originating chlorite-schists with
poor mechanical characteristics (very low shear and compressive
strength), with a general down-slope dip. Thick chlorite schists and
silvery micaschist horizons locally border the Rosone landslide,
cropping out in the Perebella area and in the adjoining upper
Piantonetto Valley, and can be held responsible for deep instability
processes within the pre-quaternary basement. The Rosone
landslide is not the only example of such relationships between
metamorphic structures and recent instability in the Orco valley: the
Noasca landslide, about 10 Km west of Rosone, is another huge
landslide whose sliding plane partially coincides with a metric level of
down-slope dipping silvery micaschists.

The unfavourable outcrop conditions make very difficult to follow

individual faults in the field, but correlations between faulting and
development of deep instability in the Rosone area are pointed out
by the parallelism between trenches and mesoscale faults and
fractures. Nevertheless, the mapping of the main fault sets of the
upper Orco valley (SEA consulting, 2001) highlighted the presence
of several regional fault zones affecting the pre-quaternary basement
and generating a related network of minor discontinuities (Fig. 5.12
and 5.13). The main fault zones of the area are the two E-W trending
fault systems of the Colle della Porta (left side of the Orco valley)
and Lago Fertà (right side) (Fig. 5.12). Between them, several
kilometric WNW-ESE striking faults are present. The Rosone
landslide is located in the step-over between two main WNW-ESE
faults, where the relatively strong frictional deformation may affect a
wide portion of the basement, generating several minor
discontinuities sub-parallel to the major ones.
Page 100

Figure 5.12. Structural sketch map of the upper Orco valley from
1:10.000 field mapping and 3D digital photo-interpretation.

Figure 5.13. Schmidt equal-area diagrams—Poles of faults. a: 126

data and joints b: 1127 data of the RosoneNoasca area. Data
correspond to faults and joints observed at the mesoscopic (metric to
decametric) scale. Great circles represent the average attitude of
joint and fault sets evidenced by the contour lines. Joints K3a
represent low-angle discontinuities sub-parallel to the regional
schistosity; they may not correspond to levels of silvery micaschists.

5.5.3 Investigation and monitoring

In the 1960–1980 twenty year period, attention had been almost
exclusively devoted to the sectors immediately adjacent to the AEM
systems (penstock and reservoirs): percussion and rotary drillings
with continuous coring were performed (maximum depth: 120 m) and
both surface and deep movements were investigated (topographic
and inclinometric measurements, respectively). The boreholes were
instrumented with inclinometers and piezometers.

Since 1985, attention has been focused on the whole slope, with
special emphasis on the Bertodasco sector: a systematic campaign
has been conducted on the entire rock mass involved in the sliding
activities.
Page 101

The methods employed included:

– surface investigations for the characterisation of the rock mass and

the main discontinuity systems;

– depth investigations consisting of: a) boreholes drilled by

continuous coring. Inside the boreholes, the piezometric pressures
and the deep-seated displacements by means of inclinometric
readings have been recorded. Moreover, at the top of some
boreholes the surface displacements have been measured by
means of a triangulation net; b) two triaxial accelerometers, located
on the main sliding body, useful to detect noises emitted by
movements of rock masses within the landslide;
Figure 5.14. Location of the boreholes, inclinometers and seismic
reflection profiling lines (A, B and C pink dashed lines) in the
Bertodasco sector.

– laboratory tests and geomechanical characterisation of the

samples obtained from the boreholes with the aim of evaluating the
shear strength of both the discontinuity surfaces and the altered
material obtained by drilling.

In order to survey the displacements around the Bertodasco sector,

during the year 2000, a new integrated monitoring system was
installed, with automatic data recording. It can be subdivided in:

– the geomechanical network, including inclinometers, piezometers

and extensometers. In particular, 7 wire extensometers were placed
along the main scarp which divides Zone B from Zone C and along
some fissures inside the latter one; three thermometers allow
compensation for thermal drift;
Page 102

– the topographical network, consisting of a topographic total station

and a GPS (Global Positioning System) net that uses 5 satellite
receivers and 19 bench marks controlled with manual GPS
measurements;

– the geophysical network, including a vertical deep geophone

located in borehole A3, two 3D deep geophones located in
boreholes B1 and B2 (see Fig. 5.14 for the location), 3D surface
geophones and a vertical surface micro-seismic detector.

The results of the laboratory tests carried out on samples and of in

situ tests on bedrock are summarised in Table 5.1.

Table 5.1. Average values of some mechanical parameters related to

the Rosone landslide.

Unit volume weight γ 26.2 kN/m3

Uniaxial compressive strength C0 74.9 MPa

Tensile strength T 7.1 MPa

Wave velocity p 3794 m/s

Base friction angle φb 35.1 ± 2.9 º

Joint compressive strength JCS 110 MPa

Barton joint roughness coefficient JRC 11 —

The surveys allowed the highlighting of the sliding surfaces’ depth in

the Bertodasco sector and in those immediately adjacent to the
penstock and reservoirs (Studio Geologico Italiano, 1984; Forlati et
al., 1993; Brovero et al., 1996b). Drilling campaigns were performed
in 1959, 1960, 1970, 1982, 1984, 1991 and 1999; the results of
those boreholes, later instrumented with inclinometers, are
summarised in Table 5.2. Figure 5.15 shows the core drilled from
borehole A2 in the depth range including the failure zone made of
destroyed rock.

On the basis of the displacement versus depth plot registered by the

inclinometers, it is possible to recognise the presence of a “knee”
that can be interpreted as a failure zone. In Figure 5.16, the
displacement versus depth plot from the A1 inclinometer is shown as
an example. In Table 5.3, the measurements related to the sliding
movements of the unstable rock mass, obtained by means of the
inclinometers, are summarised.

Furthermore, seismic reflection profiling and a downhole log,

performed along three lines and represented in Figure 5.17 (see Fig.
5.14 for the location), allowed the collection of data about the rock
mass physical properties and landslide body thickness. The results,
in terms of wave propagation velocity, underline two layers with
different behaviour: the superficial one is characterised by a greater
wave velocity, typical of a loose rock mass; the deeper layer reveals
a lower wave velocity value that can be linked to a structured rock
body. The surface between the two layers of rock can be linked with
the previously mentioned failure zone, as obtained from the
observation of the inclinometer measures and the drilling cores.

The recorded data of the monitoring system and the evaluations

derived from the investigations permit the confirmation of the
geomorphological hypotheses.

5.5.4 Danger characterization

5.5.4.1 Geomorphologic constraints
The danger characterisation was focused only on the Bertodasco
sector due to the peculiar hazard conditions and availability of
monitoring data recorded on AEM power plant and related pipes.

The Bertodasco sector

Two zones with major movements have been identified: Zone B
mainly characterised by planar sliding surfaces, and Zone C, where
rotational sliding is inferred (Fig. 5.2 and 5.3).
Page 103
Figure 5.15. Core drilled from borehole A2, including the failure
zone.

The lateral boundaries of these sectors correspond to deep and

opened fractures. The sliding surface (traced with a dashed line in
Figure 5.18 at the base of Zones B and C) is marked by a sharp
vertical change in the fracturing conditions of the rock mass. This
change is taken as a failure band below which the rock mass is
unaffected with respect to that involved in the landslide.
Page 104

Table 5.2. Features and measures of the boreholes carried out

immediately adjacent to the penstock and reservoirs. Here are
shown only those instrumented with inclinometers in 1991 and 1999.
The location of the mentioned boreholes is represented in Figure
5.14.

Sampling type DrillingAltitudeDepth Meaningful Values

date

A1 Rotary drilling with 1991 1143 102.2 0−48.7 m:

continuous coring. alternating soil/rock

Bored again in 1999 (solid 48.7–102.2 m: rock

drilling) until 51 m depth
Failure surface
intercepted at 38.98
m depth

A2 Rotary drilling with 1991 1336 120 0–120 m: rock

continuous coring
Failure surface
intercepted at 45.7
m depth

A3bisRotary drilling with 1991 14460 100.5 Failure surface

continuous coring intercepted at 71.3
m depth
Bored again in 1999 (solid
drilling) until 100 m depth

A4 Rotary drilling with 1991 1514 120 Failure surfaces

continuous coring intercepted at 30.0
m and 47.5 m depth

101 Rotary drilling with 1984 1540 31.3 0–3.55 m:

 continuous coring alternating soil/rock

3.55–31.3 m: rock
with sampling
>80%

Failure surfaces
intercepted at 3.1 m
depth

Water: 26.2 m

102 Rotary drilling with 1984 1506 39.2 0–22.9 m:

continuous coring alternating soil/rock

22.9–39.2 m: rock
with sampling
>80%

Sand and silt

cohesionless level:
16–22.9 m

Failure surfaces
intercepted at 22.6
m depth

103 Rotary drilling with 1984 1495 81 0–57.7 m:

continuous coring alternating soil/rock

57.7–81 m: rock
with sampling
>80%

Sand and silt

cohesionless level:
40–57.7 m
 Failure surfaces
intercepted at 52.4
m depth

104 Rotary drilling with 1984 1517 30.4 0–21.5 m:

continuous coring alternating soil/rock

21.5–30.4 m: rock
with sampling
>80%

Sand and silt

cohesionless level:
13–21.5 m

Failure surfaces
intercepted at 20.1
m depth

Water: 25.4 m

B1 Rotary drilling with 1999 1105 108.3 0−43.5 m:

continuous coring alternating soil/rock

Television probe 43.5–108.3 m: rock

with sampling
>80%

Sand and silt

cohesionless level:
37.3−43.5 m

B2 Rotary drilling with 1999 1290 99.7 0−40.3 m:

continuous coring alternating soil/rock

Television probe 40.3–99.7 m: rock

with sampling
>80%

Lugeon
permeability test:

depth [m]/UL
extrapolated to 10
bars

62.95–64.45/68–78

72.70–74.70/10–15
Page 105

Figure 5.16. Displacement versus depth plot from inclinometer A1.

Table 5.3. Measurements related to the sliding movements obtained

by means of the inclinometers.

Measurement period Surface displacements

Initial Last Failure surface [mm] Annual average Dip

depth [m] [mm/year] direction
[°]
A1 12/12/19913/6/1999 38.98 157.1 21.00 187.9

A2 12/12/19919/11/2000 45.72 131 14.69 170.1

A3 12/12/19915/10/1993 71.31Closed — —

A4 12/12/19916/11/2002 30.00 73.4 6.73 158.7

1016/12/1984 27/11/1993 3.05 57.7 6.42 147.9

1026/12/1984 6/11/2002 22.55 116.3 6.49 157.1

1036/12/1984 7/11/2000 52.42 117.1 7.35 165.5

1046/12/1984 7/11/2000 20.12 89.5 5.62 149.3

B1 17/11/199925/10/2000 40.23 131.5 — 144.3

B2 17/11/19996/11/2002 37.78 56.2 18.91 162.7

Page 106

Figure 5.17. Seismic reflection profiling performed along three lines

in the Bertodasco sector.

The geometry of this surface is similar to the one derived on the

basis of the geophysical data (see below).

In the lower part of Zone C, the main feature is the presence of

heterogeneous debris made up of boulders of different sizes, due to
rock-debris and debris flows.
5.5.4.2 Structural constraints
The main structural elements to be considered for modelling and
hazard analysis (sketched in Fig. 5.18 and represented in Fig. 5.11)
are:

– the schistosity and/or the joints that are sub-parallel to it;

– the mylonitic silvery micaschists interbedded in the orthogneisses

that lie sub-parallel to the slope in the Rosone area;

– the sub-vertical joint systems (K1, K2 and other minor systems);

Page 107

– the presence of an inferred deep failure zone at the base of the

Bertodasco sector (Zones B and C) that has been thought here to be
sub-parallel to the schistosity and related structures.

It is remarked here that the sliding surfaces of the Bertodasco sector

seem strictly related to the regional schistosity and associated joints,
dip parallel to the maximum inclination of the slope that in the
Bertodasco sector. In fact, the disjointing of the rocks results mainly
from the intersections of the schistosity and the perpendicular K1
and K2 joint systems. In this sense, the schistosity attitude (and
those of related features such as the silvery mylonitic schists) is
considered the triggering factor for the instability phenomena, which
could strongly control the kinematic evolution of the sliding
mechanism.

5.5.4.3 Geometrical characterization

The sliding surface, located at about 40–75 m below the ground
surface, has been thought to be sub-parallel to the schistosity and
the silvery micaschist levels. The geometry of this inferred surface is
very near the one obtained considering the results of the
inclinometers and the geophysical investigations performed in the
landslide area (see below also).
Figure 5.18. Schematic cross section of the Rosone landslide
(Bertodasco sector).

5.5.4.4 Scenarios
On the basis of geomorphological observation and field investigation,
the rock slope foot, over deepened by glacial erosion and weakened
by severe stress concentration, does not seem of particularly
efficient significance in contrasting a global movement of the slope.
As the deformation affecting the middle part of the slope will spread
througt the basal part, a quite defined failure surface might be
formed, leading eventually to a huge landslide and possibly severe
rockfall and/or rock avalanche phenomena.

In detail, three different scenarios of evolution will be taken into

account, with decreasing occurrence probability and increasing
impact on land planning:

Scenario 1: collapse of Zone C (Fig. 5.2). The rock mass of Zone C

is heavily fractured, therefore, continuous rockfalls can weaken the
rock mass located in this sector. An avalanche involving a rock mass
of about 2’200.000 m3 could then occur.

Scenario 2: collapse of Zones C and B (Fig. 5.2). The collapse of

Zone C may bring about the avalanche of Zone B at the same time.
The total volume involved would be about 9’300’000 m3.
Page 108

Scenario 3: collapse of the whole landslide area (Fig. 5.2). The

collapse of Zones C and B may induce the avalanche of the whole
rock body involving a volume of about 20’500’000 m3. In Zone A, the
quality of the rock body, which is much less fractured with respect to
the other zones, and the recorded displacement data, smaller than
those recorded in Zones B and C, suggest that this scenario must be
considered less probable than the previous ones.

The volumes of rock mass involved in the scenarios have been

roughly estimated. A precise estimation will be given on the basis of
the volume computation made through the numerical triggering
model.

5.5.5 Geomechanical modelling

5.5.5.1 Mechanical “triggering”model
The Rosone site was studied by means of a 3D equivalent
continuous model with discontinuity planes that cut off the unstable
mass from the whole rock slope.

The 3D analyses were carried out by means of the code Map3D

(described in Chapter 3) based on an indirect Boundary Element
technique, the Displacement Discontinuity Method. The numerical
simulations carried out to investigate the triggering of the rockslide
were focused on Zones B and C.

As previously seen, a complex description both from the

geomechanical and the kinematical point of view arose from the
geological investigations. At the present phase of the work, however,
a simplified geomechanical model is assumed and simple numerical
triggering simulations have been performed. Remarkable
characteristics of these models are the 3D analysis conditions and
the possibility of taking into account the measured and the simulated
displacements. In order to be able to focus attention on these
aspects, some simplifications had to be introduced from the
geometrical and the mechanical point of view:

– as the rock mass of Zones B and C is very fractured and

disjointed, in the 3D geometrical model it was represented as an
equivalent continuum and the discontinuity planes that belong to the
principal joint sets were introduced as boundaries of the unstable
volume;

– the global behaviour of the unstable rock volume was analysed:

this implies that it was possible only to reproduce the mean value of
the measured displacements due to sliding. The numerically
obtained value had to be referred to the whole investigated rock
mass. It must be underlined that the considered mean value was
evaluated with respect to time (i.e. during one year) and space (i.e.
over Zones B and C);

– the average displacements were supposed to be due to the overall

creep behaviour of the rock mass, which is assumed to be
concentrated only on the sliding surface. No influence on the
displacements of water pressure variations with respect to time was
considered, as the piezometric data were not sufficient.

On the basis of these considerations, the aim of the analyses was to

reproduce the displacements measured in situ by the monitoring
system and by means of the Bingham flow rule implemented in the
code Map3D. This can be carried out in two steps: in the first one,
the research of the mechanical parameters related to conditions of
unstable equilibrium of the rock mass is performed by means of
numerical simulations; on the basis of the results obtained, in the
second step, the analysis of the behaviour of the unstable rock mass
with respect to time is developed. In order to do this, a realistic value
of the creep coefficient C was checked by comparing the evaluated
displacements with those measured in situ.
In order to assess the 3D numerical simulation of Rosone, the first
effort was devoted to building the geometrical model of the landslide.
To attain this goal, the Digital Elevation Model (DEM) of the Orco
Valley and the available data derived from the geophysical analyses
and geotechnical site investigations were taken into account. The
recorded data of the monitoring system and of the continuous
corings reveal that the principal displacements occur 40–75 m below
the ground surface (Brovero et al., 1996b). This suggests the
existence of a deep-seated failure surface inside the slope. These
conclusions are strengthened by the results obtained from the
geophysical
Page 109

investigation. The interpretation of the position and shape of the

failure surface has been performed by means of an interpolation
methodology. Different interpolators were tested to estimate the
failure surface (IDW, Kriging, Spline, TIN), and finally Spline was
chosen because the in demand surface needs to be smooth and with
no local anomalies, characteristics that are necessary for the
foreseen numerical simulation.

The Spline interpolator is a general purpose interpolation method

that fits a minimum curvature surface through the input points.

The dataset utilised for contouring is composed of: 33 points from

seismic reflection profiling tests; 15 drillings; 5 inclinometers.
Moreover, 28 points were added where the outcrop of the failure
surface, according to the geomorphological studies, intersects the
topography. In all, the final dataset consists of 81 points for a
400’000 m2 surface. Each point gives the depth (meters) of the
failure surface from the top surface. Unfortunately, the distribution of
depth points is not homogeneous (Fig. 5.19) so, in those areas
where no information is available, the derived surface position
should not be very precise. The aspect of the derived failure surface
is depicted in Figure 5.20. The mechanical properties of the sliding
surface can be defined by a combination of the mechanical
characteristics of the silvery micaschists and of the discontinuities
sub-parallel to the schistosity.
Figure 5.19. Distribution of points utilised for contouring the inferred
failure surface.

The rock volume involved by the inferred failure surface coupling its
position with the topographic surface, obtained from DEM, has been
also calculated. The side boundaries of the moving mass have been
obtained from the geomorphological evidence (§ 5.5.2.2): hence,
vertical joints have been introduced to reproduce the boundaries of
Zones C and B (Figure 2). The estimated volume is about
11’000’000 m3, in good agreement with the volume evaluated on the
basis of the geomorphological studies (§ 5.5.4.4). In Figure 5.21, the
3D representation of the unstable volume obtained as described
above is shown.
Page 110

Figure 5.20. Sketch of the failure surface as derived by means of the

Spline interpolator.

The first aim of the 3D simulations was to obtain a realistic value of

the in situ friction angle. This value is related to a heterogeneous
material made of boulders, soil, water and voids considered as a
whole; therefore, it is impossible to evaluate the friction angle by
means of laboratory tests. A set of sensitivity analyses was
performed in order to calculate the value of the friction angle that
corresponds to the instability of the rock mass. These analyses were
conducted without creep simulation.

The following step was the assessment of the value of the creep
coefficient C. Activating the Map3D creep option, a simulation was
carried out with an estimated value of C and the results were
compared with the measured displacements. If these values were
not comparable, more simulations were carried out in order to find a
suitable C. The final set of mechanical parameters used for the
numerical analyses is shown in Table 5.4.

The numerically computed displacements can be compared with the

average annual displacement obtained by means of inclinometer
measurements. Taking into account inclinometers A1, A2 and B1
(Fig. 5.14), the only ones inside Zones B and C, the measured mean
value is about 18 mm/year with a dip direction of 167°; these values
are very close to that obtained numerically, which are equal to 20
mm/year with 156° of dip direction.

Under the hypotheses made in the framework of the simplified

geomechanical model, the obtained results allow the confirmation of
the consistency among measured displacements and kinematical
considerations, as carried out through geological studies, assuming
reasonable values of the mechanical parameters.

The obtained results are very far from giving the probability of failure
needed for the risk analysis, which is to be obtained through a
probabilistic procedure. For this purpose however the variation of
pore pressure in time and the decay of rock strength parameters in
time should be known. These data are seldom known and are not
known in the case of the Rosone landslide.
Page 111

Figure 5.21. The unstable volume considered in the 3D numerical

analyses.

Table 5.4. Mechanical parameters used for the Rosone landslide

analysis.

Equivalent continuous

Unit weight, γ 0.027 MN/m3

Young’s modulus, E 7000 MPa

Poisson’s ratio, v 0.25 —

Failure surface

Friction angle, φ 25 o

Normal modulus, Kn 5000 MPa

Shear modulus, Ks 1500 MPa

Creep coefficient, G 170 MPa×year

Page 112

5.5.5.2 Rosone run-out

The three scenarios described in § 5.5.4.4 have been taken into
consideration (Fig. 5.22), that is:

– Scenario C.

– Scenario C+B.

– cenario C+B+A.

Figure 5.22. Considered scenarios.

Each scenario was analysed by Enel.Hydro (2001) through the

application of a methodology elaborated by ISMES (Friz & Pinelli,
1993) that is based on the Perla & al. model (1980), the Li Tianchi
model (1983) and on well-documented historical cases contained in
a database drawn up by Dutto & Friz (1989).
In order to integrate the Enel.Hydro results, Scenario C was also
analysed with the Hungr model (1995).

The results described here were obtained considering a volume

estimates defined through the reconstruction of the possible shape
of the failure surface, the topography configuration and with the
hypothesis that, in correspondence of the boundary of the unstable
area, the depth of the unstable mass is zero. Furthermore, due to the
fact that the ISMES methodology considers the run-out volume (V),
the initial volume (V0) is multiplied by 1.3 (Table 5.5).

Table 5.5. Involved volume.

Scenario V0 [m3] V=V0×1,3 [m3]

C 1’864’000 2’423’000

C+B 8’987’000 11’683’000

C+B+A 17’429’000 22’658’000

Page 113

ISMES methodology

To carry out an analysis using the ISMES methodology, the input

data indicated in Figure 5.23 as F (front width), A and B are required.
The output data obtained are the maximum axial travel distance (Ra)
and the maximum lateral expansion (Sl).

Figure 5.23. Topographic parameters.

(a) Maximum axial travel distance estimate (Ra)

To define the maximum axial travel distance, Ra (Fig. 5.23):

– the unstable mass is considered as an adimensional block that

slides on an assigned topography and that is subject to gravity and
basal resisting forces (Perla & al., 1980).
Page 114

– the dynamic friction coefficient (µ) and the ratio of involved mass
(M) to aerodynamic strength coefficient (D) are considered constants
and their value is defined through the back-analyses of historical
cases.

The Ra minimum and maximum values were obtained by assuming

a normal probability distribution, with the hypothesis that at least
95% of the estimated values result inside the considered range (±2σ,
with σ standard error) (Table 5.6).

Table 5.6. Axial travel distance values.

Scenario Ramin [m] Raaverage [m] Ramax [m]

c 867 923 949

C+B 1297 1361 1438

C+B+A 1590 1861 1910

The Perla model was also applied to estimate the maximum lateral
travel distance (Rl), this aspect supposes that, once the bottom of
the valley is reached, the river direction is immediately followed by
the moving mass (Table 5.7).

Table 5.7. Lateral travel distance values.

Scenario Rlaverage [m] Rlmax [m]

c 1776 2245

C+B 2280 3475

C+B+A 2932 3526

(b) Maximum lateral expansion estimate (Sl)

To define the maximum lateral expansion, Sl (Fig. 5.23) a correlation
between Sl and the front width (F) is introduced, and the conclusions
of Li Tianchi (1983) about the relation between the run-out surface
and the involved volume is confirmed.

The Sl variability range was obtained assuming a normal probability

distribution, with the hypothesis that at least 95% of the estimated
values result inside the considered range (±2σ, with σ standard
error) (Table 5.8).

Table 5.8. Lateral expansion values.

Scenario Slmin [m] Slaverage [m] Slmax [m]

c 273 450 671

C+B 536 882 1315

C+B+A 660 1087 1619

Since it is not possible to foresee asymmetric behaviour of the mass

in the run-out area, Sl is considered symmetric with respect to the
axial direction of the movement.

Hungr method

In order to integrate the ISMES results, the rock avalanche run-out in

the case of Scenario C has been simulated by applying the DAN
code (Hungr, 1995).

The geometry of the considered unstable mass is approximately the

same as that assumed by Enel.Hydro but to allow a correct
representation in the Hungr model a geometrical simplification is
required (Fig. 5.24 and 5.25). In any case, it is important to underline
that the volume used by Enel.Hydro, and consequently by DAN, is
partially different from that estimated in the triggering model.
Page 115

Figure 5.24. Enel.Hydro results in the case of Scenario C.

To run an analysis, a frictional rheology has been chosen for the

whole path and, since the ISMES method supposes an entrainment
of material during the run-out phase, the code has been calibrated
by modifying not only the friction angle but also the volume change
rate.

The best results are obtained by considering a friction angle of 19.7°

and a mass entrainment of about 730 m3/m (Table 5.9).

The output data obtained by using these parameters are shown in

Table 5.10 and Table 5.11. In particular, a maximum front velocity of
about 47 m/s and a maximum depth in the run-out area of about 30
m result.
 Final remarks concerning the run-out investigations

The methodology applied by Enel.Hydro yields values to assign to

Ra and Sl. Since these are punctual parameters, Enel.Hydro
indicates the shape of the possible run-out area through graphical
interpretation of topography. In Figure 5.24 the continuous line
represents Ra and Sl maximum values, under the hypothesis that
the run-out area is symmetric with respect to the axial travel
distance, instead the dashed line considers the hypothesis of a
prevalent expansion of the mass in the direction of the valley.

The application of the Dan code allows an improvement of the

knowledge of the mass behaviour during the propagation and the
stop phases. Indeed, the obtained data are not only the run-out
distance and the run-up on the opposite slope but also the landslide
velocity, the expiring time and the depth of the displaced mass once
it stops.
Page 116

Figure 5.25. DAN results in the case of Scenario C.

Table 5.9. DAN input data.

Rheology Frictional

Gamma [kN/m3] 26.50

(unit weight)

Friction angle [°] 19.70

Pore pressure coefficient 0.00

(ratio of pore pressure to the total normal stress)

Volume change rate [m3/m] 730.90

(entrainment or deposition of material)

The analyses carried out underline that in the triggering case the
displaced mass would determine the river damming with disastrous
consequences.

5.5.6 Occurrence probability

The estimation of the probability of sliding is one of the critical
components of the assessment of landslide risk and hazard for
natural and constructed slopes. The probability of sliding can be
Page 117

Table 5.10. Run-out area depth and final front position.

Xprofile [m] H [m]

698.0 4.70

723.2 12.50

743.4 19.71

759.4 25.27

775.0 28.27

788.9 30.69

801.2 31.82

812.4 31.77

823.0 30.65

833.0 28.49

842.6 25.58

851.4 22.38

859.5 18.20

867.4 12.86

874.5 6.70

880.9 1.68
Table 5.11. DAN output data.

Run-out time [s] 61.40

Front final position [m] 880.90

Rear final position [m] 697.99

Maximum velocity [m/s] 47.84 at X [m]=640.63

Maximum front velocity [m/s] 46.57 at X [m]=644.28

Slide volume [m3] Initial=1856058.38

Final=2428449.50

Area in plan [m2] Initial=65443.79

Final=103757.18

estimated using formal probabilistic analysis approaches which are

inherently quantitative in nature, or using semi-quantitative methods
based on historical records, geomorphology, rainfall, slope geometry,
performance and other indications.

In the case of the Rosone landslide, quantitative methods cannot be

applied because the uncertainty in the definition of the parameters
involved is too large and it is not possible to define any reliable
variability for them.

Since several historical records exist on the landslide activity in the

area involving different volumes, the historical approach seems
much more suitable to the case of Rosone: in this way it is possible
to obtain some information related to periodic frequency.

From the historical data analysis (see Fig. 5.4), a frequency inferior
to 10 years for both rockfall/debris flow phenomena and the
Bertodasco sector instability is found. The latest signs were seen
about 40 years ago. This indicates a relationship among these
phenomena: the movements in Zone C cause the blockfall from the
scarp over the National road 460. This situation is particularly
apparent during heavy rainfalls.

The instability of Zone B is characterised by a recorded phenomena

frequency between 0 and 30 years: the last event was recorded 15
years ago. The whole slope (Zones A+B+C) showed important
movements during the 1953 event and probably at the beginning of
the XVIII century, with an interval of about 250 years. It must be
noted that these sectors are concerned by deformation movements
which are almost continuous in time. Under particular conditions,
they accelerate, causing morphological evidences on the slope and
sometimes provoke damages. The recorded historical data and the
temporal intervals of the occurrence probability refer to these
paroxystic
Page 118

phases; the last ones could cause sudden collapses of the slope
with the development of rock avalanche phenomena.

An occurrence time range was obtained for each scenario,

considering the minimum and maximum time intervals between two
consecutive recorded events. On the basis of their mean values, a
frequency was then calculated in terms of events per year (Table
5.12).

Table 5.12. Frequency calculated in terms of events per years.

Scenario Occurrence range Average Frequency

[years] [years] [event/year]

Rockfall & Debris flow 0÷50 25 1/25–0.04

Rock avalanche (Zone C)0÷50 25 1/25=0.04

Rock avalanche (Zones 50÷250 150 1/150=0.007

B+C)

Rock avalanche (Zones >250 1000 1/1000=0.001

A+B+C)

5.6 QUANTITATIVE RISK ANALYSIS

In the chapter dedicated to hazard analysis, various landslide

scenarios were identified. Concerning quantitative risk analysis, only
three rock avalanche scenarios were considered. These three
scenarios refer to the evolution of three slope sectors with
indications of instability or with movement data (recorded by
monitoring systems). The slope sectors are: the upper zone (A), the
intermediate zone (B) and the lower zone (C). The recognised
dangers are linked to the following scenarios:
– collapse of Zone C and rock avalanche accumulation at the valley
floor;

– collapse of Zones B and C and large rock avalanche accumulation

at the valley floor;

– collapse of Zones A, B and C and huge rock avalanche

accumulation at the valley floor.

The risk analyses of the scenarios are based on land planning of the
Locana Municipality and on the landslide run-out areas obtained
using the ISMES method (Friz & Pinelli, 1993).

5.6.1 Elements at risk

The elements at risk are recognised through the definition of the
area affected by each danger in the scenarios previously specified.
These elements at risk are shown in the following tables (Tables
5.13, 5.14 and 5.15).

Table 5.13. Rock avalanche scenario (Zone C).

Element at risk DenominationNotes

Residential areas [valuable buildings or Rosone Buildings affected:

valuable rural centres (historical, 2 buildings of the
architectural, artistic and/or cultural evacuated
value)] community (R.D.
445/1908).

Persons affected: 0

Forests (private and public properties) Area affected:

about 900,000 m2

Heavy traffic or strategic roads National road Length affected:

650 m
N° 450
 Annual traffic:
about 170,000
vehicles

Lifelines Power line Length affected:

“Valle 700 m

Locana” 15 Pylons affected: 1

kV
Page 119

Table 5.14. Rock avalanche scenario (Zones B+C).

Element at risk DenominationNotes

Residential areas [valuable buildings or Rosone Buildings

valuable rural centres (historical, affected: all the
architectural, artistic and/or cultural value)] buildings of the
evacuated
community

(R.D.445/1908).
32 buildings of
new Rosone
community

Persons affected:
88

Aghettini Buildings
affected: 8

Persons affected:
12

Forests (private and public properties) Area affected:

about 403,000
m2

Heavy traffic or strategic roads National road Length affected:

1506 m
N° 450
Annual traffic:
about 170,000
vehicles
Lifelines Power line Length affected:
“Valle 1000 m

Locana” 15 Pylons affected:

kV 3

Table 5.15. Rock avalanche scenario (Zones A+B+C).

Element at risk DenominationNotes

Residential areas [valuable buildings or Rosone Buildings affected:

valuable rural centres (historical, all the buildings of
architectural, artistic and/or cultural the evacuated
value)] community

(R.D.445/1908). 62
buildings of new
Rosone community.

Persons affected:
186

Aghettini Buildings affected:

8

Persons affected: 12

Casetti Buildings affected:

7

Persons affected: 18

Forests (private and public properties) Area affected: about

900,000 m2

Heavy traffic or strategic roads National road Length affected:

2200 m
N° 450
 Annual traffic:
about 170,000
vehicles

Lifelines Power line Length affected:

“Valle 2500 m

Locana” Pylons affected: 8

15kV

Secondary lifelines Power line forLength affected: 600

the funicular m
service
Pylons affected: 2

5.6.2 Vulnerability of the elements at risk

To define the vulnerability of the elements at risk, an analysis of the
typology of the assets affected and the processes’ intensity was
carried out.

In the rock avalanche scenarios, a distinction between various

degrees of energy based on landslide velocity is useless in defining
different degrees of vulnerability of the elements at risk. In the entire
area affected (failure, depletion and accumulation areas) the
destruction would be total. Also, at the valley floor, where the velocity
tends to zero, the accumulation (some tens of m) would cause the
complete destruction of the assets (heavily damaged or buried).
Therefore, the degree of vulnerability is equal to 1 (for all
vulnerability categories).

Considering areas next to the run-out zone, various hypotheses can

be made about the wind and dust effects. The models known to the
Authors, however, do not supply indications about the geometry of
the area affected and the process intensity of wind and dust effects.
Therefore, in this quantitative risk analysis, the effects associated
with a rock avalanche have not been taken into
Page 120

account; qualitative considerations about these processes will be

described in succeeding sections, deduced from case histories.

5.6.3 Expected impact

By crossing values for the elements at risk (assets, economic,
environmental and social value) with vulnerability percentage,
expected impact is obtained. It is possible to use a monetary index
for assets and economic evaluation, derived from detailed economic
analyses (e.g. costs of assets, reconstruction costs, turnover, profits
loss, etc.). To assess expected environmental and social impacts
where a monetary value is difficult to obtain, relative value indices
are used.

In the quantitative risk analysis of the Rosone landslide, a detailed

economic evaluation has not been considered at the moment.
Arbitrary relative value indices have been temporarily applied to
each category.

5.6.4 Risk assessment

The last phase of the quantitative risk analysis concerns the risk
evaluation. Such an evaluation is obtained by multiplying the
expected impact by the occurrence probability value related to each
scenario. In the case of the Rosone landslide, the occurrence
probability values have been calculated on the basis of the historical
data.

The results of the risk assessment are shown in the following figures
(Fig. 5.26, 5.27, 5.28 and 5.29) obtained by GIS layout and
described in the captions.
Page 121
Page 122

Figure 5.26. Maps by GIS of physical risk for the three rock
avalanche scenarios. The maps underline the areas with highest risk
value, i.e. the National road no. 450 and part of the new Rosone
community. Particularly relevant is the run-out area of the A+B+C
scenario: part of the accumulation involves the hydroelectric power
plant, part of the failure area involves the penstocks.
Page 123
Page 124
Page 125

Figure 5.27. Maps by GIS of economic risk for the three rock
avalanche scenarios. The maps underline the areas with highest risk
value, i.e. the National road no. 450 and the lifelines (electric power
line “Valle Locana” and secondary electric power line for the funicular
service). From an economic point of view the electric power plant of
Rosone and the related infrastructures are very important, as they
supply energy to a large part of Turin.
Page 126
Page 127
Page 128

Figure 5.28. Maps by GIS of environmental risk for the three rock
avalanche scenarios. The maps underline the areas with highest risk
value, i.e. the wooded areas and the valley floor. The Orco river is
completely buried by rock avalanche accumulation, therefore, the
risk value of the affected zones has no meaning.
Page 129
Page 130
Page 131

Figure 5.29. Maps by GIS of social risk for the three rock avalanche
scenarios. The maps underline the areas with highest risk value, i.e.
the urban areas of the new Rosone and Casetti communities. Also
the National road no. 450 involved in the rock avalanche
phenomenon has high risk values, being characterised by intense
traffic during the tourist season.

5.6.5 Indirect risk and regional impacts

In this paragraph, some considerations concerning the indirect
effects linked to rock avalanche are described (Fig. 5.30). The
evolutionary scenarios of the Rosone site, besides rockfall, involve
huge rock volumes (millions of m3) at high velocity (tens of m/s),
whose effects are not easily evaluated. The subsequent
considerations are based on descriptive and qualitative evaluations
derived from case histories (i.e. Hsü, 1975).
The large mass movement (tens of millions of m3) that could occur
in Rosone would produce a huge wind effect. The processes’
intensity cannot be easily calculated but it is probable that they
would be light laterally to the mass movement and very strong
frontally. The same would occur for the dust that follows the wind
effect.

At the moment, wind and dust effects cannot be obtained by a

numerical model with confidence. Nevertheless, it can be stated that
the damages produced by these indirect effects would be very
severe, as the area next to the accumulation is densely populated.

Another effect to be considered in risk assessment is the formation

of a landslide dam lake. This phenomenon can occur during a
meteorological event with important flows along the Orco river (some
hundreds of m3/s). Instant damming of the thalweg by tens of meters
of landslide accumulation and extraordinary flooding of the Orco
River would form a basin of some hectares in a few
Page 132

hours. Many buildings in the Fornolosa village would be inundated

and many people would have to be evacuated, increasing social risk.

Due to such a lake formation, downstream villages could potentially

be involved. Instantaneous lake depletion could be caused by a
rapid overtopping dam erosion, or a landslide dam collapse caused
by piping and internal erosion.

In both cases, flash-flooding along the valley would cause enormous

problems. At the Rosone site, a survey of the effects of flash-flood
events has not yet been carried out.

Figure 5.30. Maximum extension of landslide dam lake (A+B+C rock

avalanche scenario, Enel-Hydro, 2001).

5.6.6 Total risk

In the following maps (Fig. 5.31), the results of a quantitative
analysis of total risk for each category (physical, economic,
environmental and social risk) are shown. As clarified in Chapter 3,
total risk is calculated by adding the risk values obtained in each
category of each scenario.
The maps emphasise that the risk is strongly linked to the
occurrence probability. The occurrence probability of a Zone C rock
avalanche is 16 times superior to the occurrence probability of a
Zone A+B+C rock avalanche and 4 times superior to the occurrence
probability value of a Zone B+C scenario. Therefore, the major risk
values are recognised in the areas of Scenario C.

In the case of these three rock avalanche scenarios, the processes’

intensity is useless for the differentiation of the different risk degrees,
as it is very high in all scenarios and destructive in all of the areas
affected.
Page 133
Page 134
Page 135
Page 136

Figure 5.31. Maps by GIS of total risk. The maps underline the areas
with highest risk value for each category physical, economic,
environmental and social.
Page 137

6
The Oselitzenbach landslide

Gaetano Amatruda1, Marta Castelli1, Marcel Hurlimann2, Alberto

Ledesma2, Michele Morelli3, Fabrizio Piana3, Marina Pirulli1, Rainer
Poisel4, Riccardo Polino3, Pere Prat2, Alexander Preh4, Werner
Roth4, Claudio Scavia1, Ewald Tentschert4
1Politecnico di Torino; 2Universitat Politècnica de Catalunya; 3CNR-
IGG sezione Torino; 4Technische Universität Wien

6.1 INTRODUCTION

The area on the northern slopes of the so-called “Carnic Alps” in the
southern part of Austria’s Carinthia country is affected by numerous
mass movements.

The Carnic Alps, as a member of the (tectonically defined) “Southern

Alps” are divided from the “Central Alps” in the north of the area by a
continental lineament called the “Periadriatic Lineament” leading
from southern Switzerland to Slovenia.

Thus, the Naßfeld area, dewatered by a torrent called

Oselitzenbach, belongs to a different tectonic unit and contains rock
composition different from that of the Central Alps, i.e. mainly
sedimentary beds of paleozoic age containing clayey schists,
sandstones, marls and limestones. Due to the adjacent lineament
(distance 1–3 km) the degree of tectonic influence is rather high.

The Reppwand-Oselitzenbach sagging zone consists of interlayers

of competent and incompetent rocks which are highly deformed and
disturbed both by tectonics and by long-term sagging and sliding
processes.
The movements (measurements available since 1983) are very
inhomogeneous, reaching from some cm/y up to more than 1 m/y in
some areas.

The risks concern the destruction of the Naßfeld Road (National road
B 90, crossing the Alps to Italy and connecting to a large ski area),
as well as the possible spreading out of landslide material by the
Oselitzenbach torrent (which occurred in part after heavy rainfalls
and a consecutive flood in 1983).

Several measures have been taken to stabilise the landslide (drains,

deviation of the torrent, debris dams in the river, anchoring of the
road cuts and slopes), which led to some decrease in movement;
however, they are still continuing, but with reduced amounts.

6.1.1 Landslide description

The catchment area of the Oselitzenbach torrent is considerably
affected by two sagging slopes which were landslides in prehistoric
times (Reppwand landslide, Schlanitzenalm landslide). The main
sliding process of these slopes took place between the Rissian and
the Würmian Ice ages (i.e. between 70’000 y–150’000 y BP); this is
documented by Würmian age moraines partly overlaying the
landslide area with no signs of displacement.

The main scarp of the whole Reppwand landslide (Fig. 6.2 and 6.3)
shows a typical profile of the Southern Alpine sedimentary
sequences from Devonian sandy—marly calcareous sediments up to
lower Triassic carbonates.

The Reppwand sliding mass consists mainly of the marly elements

and covers the Silurian-Devonian Hochwipfel schist basement
(sandy-silty).
Page 138

Figure 6.1. Geological sketch of Austria with the location of the site.

Figure 6.2. Topographic map; original scale: 1:50’000.

Page 139

Figure 6.3. Panoramic view of the area (Kahler & Prey, 1963).

Legend: 1—landslide movements; 2—moraine; 3—dolomite; 4—

shelly limestone; 5—shelly limestone, conglomerate; 6—Werfener
layers; 7—Bellerophon dolomite; 8—Grödener layers; 9—Trogkofel
limestone; 10, 12—Pseudoschwagerinen limestone; 13—Auernig
layers; 14—Hochwipel layers; 15—layered limestone; 16—reef
limestone.

Especially the toe zone of the Reppwand landslide is still quite active
and is responsible for debris-generating slope failures and the
destruction of the Naßfeld Road. After sustained regional rainstorms
in Sept. 1983 with numerous embankment failures, an extensive
project of construction work and research was initiated along the
Oselitzenbach torrent and the sagging slope above.

Thus, the area of the eastern toe of the Reppwand landslide has
been the object of numerical investigations as the displacements of
this area are the largest and are, therefore, intensively monitored.

6.1.2 Historical background

The geological investigation of the area began in about 1850 by the
first determinations of paleozoic fossils, but the investigation of the
landslide area started only 100 years later (Kahler & Prey, 1963).
There are no reports available about the landslide before 1915.
During the Middle Ages up to the beginning of the 20th century, the
area of Naßfeld was only a summer pasture for the animals of the
valley villages. Only a small footpath led into the valley of
Oselitzenbach.

During World War I, a military road was constructed. Since then, the
movements along the road have been known, but have not been
investigated in detail.

In the seventies, the skiing area at the Naßfeld Pass was opened
and then enlarged and the road was upgraded to a primary road,
connecting Austria and Italy.

In September 1983, heavy rainfall and flooding in the catchment

area led to massive gravel deposits at the alluvial cone at
Oselitzenbach. As a consequence, erosion of the right abutment
took place and the Naßfeld Road became cut off (Fig. 6.4).
Additionally, the old torrent control dams along the upper and middle
courses of the tributary torrent “Rudingbach” were destroyed.

In 1985, planning was started for an extensive construction

programme carried out by the Austrian Service for Torrent, Erosion
and Avalanche Control as a reaction to the events in September
1983.

As a consequence of repeated flooding, an advancing undercutting

of the eastern toe of the landslide arose. A run-out of 60’000 m3 in
volume took place (August 1987) and as a result, large settlements
up to several metres occurred. Additionally, the formation of new
cracks up to 30 m above the Naßfeld Road occurred.

The implementation of the extensive construction programme started

in 1988. As a part of the construction measures, a 400 m long new
river channel was excavated in comparatively massive
Page 140

Figure 6.4. Destruction of Nassfeld Road Sept. 1983 by landslide

movements.

Hochwipfel formations (Fig. 6.5), as well as a placement of landfill at

the toe of the sagging mass using the excavated material (about
170’000 m3).

Additional measures have been: drainage of the Quellenbach

landslide (lower part of the ancient Reppwand landslide) between
elevations 830–920 m a.s.l. and the draining off of surface
watersabove 950 m a.s.l. as well to the area of the “Bodensee” Lake.
These measures were necessary
Figure 6.5. New rock channel of the Oselitzenbach torrent, fresh
landfill at the toe of the sagging zone (brown and yellow) filling in the
former ravine; in the background the Nassfeld Road is seen. Date of
photo: 1989.
Page 141

to prevent serious debris flow into the Oselitzenbach Torrent and in

the receiving stream “Gail” and to prevent debris flow reaching and,
possibly, destroying the villages “Tröpolach” and “Watsching”. For
economic reasons, the correction of all the debris sources from the
very top downwards was not considered possible or necessary
(Moser & Glawe, 1994).

6.2 REGIONAL FRAMEWORK

6.2.1 Climatic and water conditions

The Oselitzenbach torrent belongs to the Danube catchment system
(1st order: Danube River, 2nd order: Drau River, 3rd order: Gail
River, 4th order: Oselitzenbach Torrent).

The climatic conditions are influenced by an Alpine climate

(westward winds), influenced additionally by Adriatic lows.

The principal rainfall takes place in summer (>50% of annual

precipitation), but even heavy rainfalls can occur in October–
November, especially when westward autumn weather occurs and
meets Adriatic areas of low depression. The annual precipitation,
therefore, shows a high value of 2963 mm/y (the highest value
measured in Austria; the weather measurement station is situated at
1530 m a.s.l.)

The average annual temperature is: +7°C, the minima occur in

January with a value of −18°C, the maxima in summer can reach
+30°C.

The snow covering period depends on the altitude: at 1000 m a.s.l.

the area is covered from November until March; at 2000 m from
October to May. Snowfall may occur all the year, but mainly from
October to April.
6.2.2 Groundwater conditions
The groundwater conditions, as understood up until now, differ in the
various parts of the area.

The upper part shows a shallow infiltration, with groundwater coming

to daylight as springs in the middle part. In the lower part, water
losses are documented along deep rotational slide planes. The
groundwater velocities in the upper part are about 0, 2–5, 0 m/h
(evidenced by a tracer test, Hötzl & al., 1994).

There are three levels of springs: The highest level lies between an
elevation of 1150 m and 1170 m. At this level, there are two lakes,
the “Great Bodensee” and the “Small Bodensee”, also there are
springs west of the lake “Great Bodensee” and water outlets in the
water logging zone east of the lake “Small Bodensee”.

The area of the second level lies 20 m deeper (1130–1150 m), in the
north of the lake “Great Bodensee”.

At the third level, there are many selective water outlets below the
Naßfeld Road (at an elevation of 700–800 m).

The “Bodensee” lakes lie in a small depression, made up of

secondary zones of movement. The greater lake is fed by 2 small
tributaries and is dewatered by a small runoff river-the smaller lake
has no tributary at the surface, but has a small runoff (3–51/s).

6.2.3 Regional morphology

The Carnic Alps, forming the boundary between Austria (Carinthia
county) and Northern Italy (Province of Udine), are an East-West
striking mountain chain up to 2700 m a.s.l. The main valley (Gailtal,
600 m a.s.l.) in the north is parallel to these chains, whereas the
smaller tributaries cut transversally through the structures by steep
and narrow valleys, mainly being cut after the last ice age.

The morphology of the mountain chain is dominated by Triassic and

Devonian limestones in the highest peaks, and at the foothills and on
some less inclined slopes by (partly sandy) shales and slates. During
the Ice ages, the valleys were covered by ice up to 2000 m a.s.l.,
leaving only the
Page 142

highest tops uncovered. The carving of the valleys from U-shaped

glacier valleys to V-shaped erosion valleys took place after the last
period of the Ice Age (over the last 20’000 years). By these
processes, some slopes became partly eroded at their toe area.

6.2.4 Regional geology and structural setting

As a part of the Carnic Alps, there are evidences of a complicated
tectonometamorphic history since the first Variscian deformations
and by the Alpine orogeny. The Variscian record is partly preserved
because the Alpine overprint was generally weak.
Figure 6.6. Geology (after Schönlaub & al., 1987).

Legend: 1: young river deflection, torrent debris; 2: mudflow debris;

3: alluvial cone; 5: blocky material, rock fall material; 20: ground
moraine; 43, 44: Schlern dolomite; 45–47: shelly limestone; 49, 50:
Bellerophon dolomite; 51: Grödner sandstones; 53: Trogkofel
breccia: dolomite to conglomerate; 55–57: upper
Pseudoschwagerinen formation; 61: lower Pseudoschwagerinen
formation; 62–65: Auernig formations; 68–71: Hochwipfel formation;
81: Eder limestone (grey, banded limestone).
Page 143

The Variscian basement is represented by an epizonal tectonic unit

(laminated limestones, marbles and phyllites) and the anchizonal
Hochwipfel nappe (Fig. 6.6 and 6.7). The stratigraphic order begins
with light metamorphic rocks (lower paleozoic), overlain by intensely
folded and fractured schists (mainly sandstones and marlstones of
Devonian-Permian Age). The post-Variscian PermoCarboniferous to
Triassic Sequence rests on top of the Hochwipfel nappe above a
major unconformity (Schwarzwipfel fault, Fig. 6.7).

In the main valley of the Gailtal, near the site, the continental tectonic
divide of the “Periadriatic lineament” passes through (dividing the
Northern from the Southern Alps from the Piemont to Slovenia). This
leads to parallel faults of the periadriatic lineament (North West-
South East), to some Riedel shears (diagonal) and extension joints
(A-C- Joints, North-south) (see Fig. 6.7).

The investigated area mainly consists of two different geological

formations (see Fig. 6.6), the Hochwipfel schists (Silurian upper
Carbonian) and the Naßfeld schists (upper Carbonian to lower
Permian).

The Hochwipfel schists contain sandy siltstones; they have higher

strength than the more fractured Naßfeld schists. The heavily
fractured Naßfeld schists are a local term (a formation combining the
Aueringg schists, some intercalations and the Rattendorfer schists).
The rock menu mainly contains a melange of conglomerates,
limestones and sandy siltstones.
Figure 6.7. Tectonic overview of the area: the main fault zones are
sub-parallel to the “Gailtal fault” (part of the periadriatic lineament).

6.3 HAZARD ANALYSIS

6.3.1 Geo-morphological analysis

The Oselitzenbach has four small tributaries within the sagging
mass, forming the Ostelitzenbach torrent after being combined.

Through the sagging mass of the Reppwandgleitung (between the

last two Ice Ages, Riss and Würm) the valley floor became very
narrow and the torrent eroded the toe of the mass.
Page 144

The lowest part of the Oselitzenbach is a steep inclined valley

reaching the more deeply situated main Gail Valley, carved out after
the last Ice Age over the last 20’000 years.

Above the main scarp of the mass, there are outcrops of massive
limestones with steep slopes. Within the deformed mass, there are
many small trenches (Fig. 6.8). Within the sagging mass, there are
many secondary scarps, forming different elements of the deformed
mass. As shown in § 6.3.3, there are different movement rates in the
various sections.

Figure 6.8. (Kahler & Prey, 1963): geologic-geomorphologic map;

many trenches are indicated, resulting from different secondary
rotational slides;
Braun: Naßfeld layers; Yellow: moraine; Green: Hochwipfel layers;
Dark grey: Trogkofel limestone; Light grey with triangles: blocky
material; Pink: Grödner sandstones; Grey with points: detrital
formation from the Rattendorfer layers.

6.3.2 Local geology and structural analyses

6.3.2.1 “On site”available data
The project area covers an alpine valley with elevations ranging from
590 m to 2000 m a.s.l. within the catchment area of the
Oselitzenbach. The “rock menu” consists mainly of:

– Hochwipfel schists (see Fig. 6.9): dark greyish sandy-siltstones,

scarcely calcareous. Anchizonal light metamorph, well bedded, with
interlayers of clayey material; partly graphtolith schists. Mainly hard
and brittle behaviour. Occurrence: mainly in the north of the mass
movement, being the northern embankment of the river bed. Age:
Silurian to upper Carbonian

– Auernigg schists (Fig. 6.10,6.11 and 6.12): Conglomerates, partly

transgressive formed limestone beds; lower group, less limestone:
dark greyish fine graded sandstones, partly sandstone schists. lower
group, rich in limestone: sandy schists and sandstones, partly
limestone beds. middle group, less limestone: Sand & quartz
conglomerates, scarcely schists. upper group, rich in limestones:
sandy schists with mica and sandstones & conglomerates upper
group, less limestone: schists & sandstones, shales dominating Age:
upper Carbonian—lower Permian
Page 145

Figure 6.9. Hochwipfel schists.

Figure 6.10. Lower Auernigg schists, sandy.

– Pseudo-Schwagerinenkalk: Limestones, rich in the fossil
“pseudoschwagerina”: dark grey limestones, scarcely bedded. Age:
Upper Carbonian

– Rattendorfer Schists: limestones, dark, grey. Sometimes thin

interlayers of siltstone. Age: Lower Permian.

In the heavily disturbed, sagging areas, the Auernig and

Rattendorfer schists cannot be precisely distinguished in the field. In
this case, they are summarised by the term “Naßfeld schists”
Page 146

Figure 6.11. Upper Auernigg schists, right corner-centre: sandy

variety, left above: marly beds.

Figure 6.12. Auernigg (Nassfeld) schists, intensively fractured.

– Both stratigraphically and topographical above (outside the

landslide area), there are also Triassic sediments: The main member
is the Trogkofelkalk, consisting of limestones, partly dolomitic lesser
bedded, massive beds, light grey, wide jointing.
6.3.2.2 “Around site” structural analyses
Within the slide area itself, the structural features are rotated and/or
dislocated and, therefore, they do not represent the tectonic stress
field.

In the Reppwand (main scarp), the general bedding has a dip of 30–
50° with a dip direction of 200° to the south. The bedding planes are
usually flat, partly polished and may contain mylonitic layers. The
thickness of the bedding is in the range of cm-dm.

The main joint system dips 80° to the SW (dip direction 230°), is
medium rough and has an average trace length of 5–10 m. Some
clay seams may reach a thickness up to several cm.

A second joint system dips 75° to the NW (dip direction 310°), has
usually rough joint surfaces and can be covered by mm thick seams
of clay.
Page 147

The rocks underlying the slide consist of the “Hochwipfelschichten”

(grey schists and sandstones) and have good outcrops in the new
river channel. Foliation there dips to the N (350/50–60), the primary
joint system dips to the NW (300/45). A secondary system dips
steeply to the NE (035/75).

The main fault zones are (sub-) parallel to the Periadriatic lineament
(carving the main valley of the Gail River), striking WNW.
Figure 6.13. Joint diagrams in the wider area (Läufer et al., 1997);
the location of the landslide is indicated by a red ellipse.
Page 148

The fault-slip data presented in Figure. 6.13 indicate a dextral-

transpressional regime for the underlying units (Hochwipfel and Eder
nappes). Faults which reveal stress tensors with σ1 orientated sub-
horizontally (N-S) and σ3 switching between a sub-vertical (thrusting,
Fig. 6.13 A) and sub-horizontal, E to W orientated position (strike-
slip, Fig. 6.13 B) result from N to S directed contraction which is
responsible for the formation of several positive flower structures in
the vicinity of the Periadriatic lineament.

The geotechnical properties of the rocks in the front area of the

Reppwand landsl`ide can be defined as following:

– Naßfeld series: the rocks are heavily disturbed, therefore, they can
be defined as a cohesive soil

– Rock structure is disintegrated to a block-talus with a fine-grained

matrix

– Embankment (rock channel) with sandstones and schists, well-

jointed

– Well-jointed, loosed limestones with wide open fissures at the main

scarp

6.3.3 Investigation and monitoring

As a consequence of intense movements in August 1987, an
extensive construction and monitoring program was started (see §
6.1.2). Since October 1988, the toe zone of the landslide has been
intensively monitored by means of:

– 60 geodetic points (partly measured since 1985)

– 1 inclinometer
Figure 6.14. Location of the monitoring instruments.
Page 149

– 1 wire extensometer, 1=45 m, digital monitoring, measures every 2

min.

– 9 convergence scanlines (fixed endpoints), measurement with

steel tape (Fa. Soil instruments MK II), accuracy 1/100

Seven boreholes were drilled, one of them equipped with an

inclinometer (BL depth: 52 m). Figure 6.14 shows the investigated
area and the location of the monitoring instruments.

A comparison of the average values shows significant stabilisation

after the finalisation of the construction measures in most areas.
Periods A (1988–1991, before the construction measures) and B
(1991–2000, after the construction measures) have, therefore, been
compared showing the following:

– Areas of low velocity (<5 cm/y) significantly increased in Period B,

whereas zones of high velocity (>10 cm/y) decreased.

– Above the fill, an extensive stabilisation was recorded in Period B.

– The most active zone in both periods is the upper area of the
Quellenbach landslide, which, however, shows a stabilisation as well
in Period B from 10–15 cm/y down to 7–15 cm/y. Figure. 15 shows
the map of the movements, which were measured between 1991
and 2000.

– Relatively high movements can still be observed east of the point

RP below the Naßfeld Road.
Figure 6.15. Displacement rates of the investigated area (Moser,
2001).
Page 150

Within diverse homogeneous regions, there are different

homogeneous regions, having different displacement rates. Since
the construction measures were finished in 1991, the displacement
rates decreased.

Table 6.1. Displacement rates.

Homogeneous region 1988–1991 1991–2000

“Seebach” above road 0,9 cm/month 0,5 cm/month

“Quellenbach” 1,3 cm/month 0,65 cm/month

Roadmaster hut 0,4 cm/month 0,35 cm/month

The refraction seismic section through the investigated area (Fig.

6.16) shows a three layer structure for the Naßfeld schists (Weidner,
2000):

– The lowest layer does not show any displacements at the moment
although it is influenced by the Reppwand landslide.

– The middle layer (15–30 m thick) shows small displacements but is

not as fractured as the upper layer.

– The upper layer (10–15 m thick) is heavily fractured and behaves,

therefore, as soil with low cohesion. This fact is the main reason for
sustained displacements.
Figure 6.16. Seismic profile of the sagging slope (Weidner, 2000).

6.3.4 Danger identification

According to the results of the monitoring program and
morphological studies (activity of landforms, cracks, etc.) three
possible scenarios have been detected (Fig. 6.17 and 6.18):

1. Failure and detachment of the area displacing most at present

2. Bodensee landslide

3. Reactivation of the Reppwand landslide

Scenario 1: Failure and detachment of the area displacing most at

present

Topographic monitoring, as well as wire extensometer readings and

inclinometer readings, show that an area between 800 m and 1000
m a.s.l. is moving with displacement rates of 7 cm per year at
present (Fig. 6.15: displacement rates MOSER) endangering the
Naßfeld Road (see § 6.3.5).

Scenario 2: Bodensee landslide

Figure 6.17 shows that a sliding plane surfacing at the flattening of

the Bodensee (1107 m a.s.l.; see § 6.3.1) was assumed to exist by
various authors (e.g. Moser & al., 1988; Moser &
Page 151

Figure 6.17 Possible scenarios

Figure 6.18. Sliding planes of the possible scenarios at the toe zone
of the landslide.

Windischmann, 1989; Kahler & Prey, 1963). The monitoring

programme reveals that the area between Oselitzenbach and
Bodensee is moving with displacement rates of 5 cm per year at
present. Most probably, the Bodensee landslide will shear through
the Hochwipfel schists’ abutment at the toe of the slope for the
Reppwand landslide mass.
Scenario 3: Reactivation of the Reppwand landslide

Morphological studies have indicated Graben-like structures parallel

to the edge of the wall on the upper slope surface of the Reppwand
displaying no signs of activity at present. However, a reactivation of
the Reppwand landslide is seen as possible (Fig. 6.17; see § 6.1.1
and 6.3.2).
Page 152

6.3.4.1 Geometrical characterisation and geomorphologic constraints

Scenario 1: Geophysical investigations (see § 6.3.3) as well as
numerical modelling (see § 6.3.5) have shown that the sliding mass
has a depth of some 25 m. The monitoring programme, as well as
numerical modelling (comp. 3.5; Fig. 6.15: displacement rates
MOSER), gave an unstable area of some 52’850 m2 and a moving
mass of some 450’000 m3 (Fig. 6.19).

Figure 6.19. Scenario 1–Detachment and run-out of the most active

area at present.
Figure 6.20. Scenario 2–Bodensee landslide.
Page 153

Scenario 2: Due to the morphology (Vorderer Seebach), an unstable

area of some 536’000 m2 was assumed (Fig. 6.20).

Scenario 3: Geological investigations (Kahler & Prey, 1963) depicted

an unstable area of some 2’911’000 m2.

Figure 6.21. Scenario 3–Reactivation of the Reppwand landslide.

6.3.4.2 Structural constraints

Scenario 1: The whole sliding surface is running through the
fractured Reppwand sliding mass. Thus, there are no structural
constraints (Fig. 6.18).

Scenario 2: Almost the whole sliding surface is running through the

fractured Reppwand sliding mass. Joints and faults in the Hochwipfel
schists (see § 6.3.2) making an abutment at the toe of the slope and
retaining the Bodensee landslide mass will form a sliding plane (Fig.
6.18).
Scenario 3: In the lower part, the assumed sliding plane follows the
sliding plane of the ancient Reppwand landslide. In the upper part of
the sliding plane, Graben-like structures parallel to the edge of the
wall on the upper slope surface of the Reppwand coinciding with
joint sets indicate sliding planes in the Bellerophon dolomite and the
Trogkofel limestone (Fig. 6.17). Similar to the ancient Reppwand
landslide, the Auernig schists will be sheared through during the
development of the landslide (see Fig. 6.10).

6.3.4.3 Scenario identifications

Scenario 1: The run-out of the area displacing most at present was
numerically modelled (see § 6.3.5). The calculations have shown
that the Oselitzenbach ravine will be buried by 85’085 m3 of debris
flowing over the Oselitzenbach debris cone, burying 896’000 m2
(Fig. 6.19).

Scenario 2: The run out of the Bodensee slide will bury the
Oselitzenbach ravine by 3’375’000 m3 of debris flowing over the
Oselitzenbach debris cone, burying 200’000 m2 (Fig. 6.20).

Scenario 3: The run out of the reactivated Reppwand landslide will

bury the Oselitzenbach ravine by 16’500’000 m3 of debris flowing
over the Oselitzenbach debris cone, burying the villages of
Tröpolach and Watschig and 2’67 8’000 m2 of forest and rural areas
and damming up the river Gail by 5 m (Fig. 6.21), thus causing an
inundated area of some 4’852’000 m2.
Page 154

6.3.5 Geomechanical modelling

6.3.5.1 Mechanical “Triggering” model
The Oselitzenbach site has been studied by means of two 3D
continuum models: the first model was made using the code
FLAC3D (Itasca, 1997) based on the finite difference technique, the
second one has been carried out using the finite element code
DRAC (Prat et al., 1993).

Only Scenario 1 (Failure and detachment of the area displacing most

at present) has been numerically investigated because the
occurrence probability of this scenario is the highest. (see § 6.3.6).
Figure 6.22. Investigated area of the Reppwand landslide.

The investigated area mainly consists of two homogeneous regions

(see Fig. 6.24), the Naßfeld schists and the Hochwipfel schists. The
heavily fractured Naßfeld schists consist of a melange of
conglomerates, limestones and sandy siltstones. The Hochwipfel
schists consist of sandy siltstones; they have higher strength than
the more fractured Naßfeld schists.

Seismic surveys have revealed that the Naßfeld schists can be

divided vertically into three layers corresponding to the degree of
fracturation (Fig. 6.16). Since there are no main discontinuities at this
site, only a continuum model has been analysed. The models,
however, include four differ-ent material properties: the Naßfeld
schists with three material properties and the Hochwipfel schists with
one.
Page 155

Figure 6.23. View of the investigated area.

Figure 6.24. Digital surface model.

Blue: Naßfeld schists-low strength, green: Hochwipfel schists-high

strength.
Page 156

DRAC model

The Oselitzenbach test site has been analysed by a 3D continuum

model (Fig. 6.25). The topography of the model was defined by the
actual morphology of the slope utilising a digital elevation model (Fig.
6.24) with a cell size of 25 m. The final model consists of 15066
continuum elements.

Figure 6.25. Three-dimensional model of the Oselitzenbach slope

including different material types.

A two-stage approach was used during the modelling. In the first

stage, the upper limit of the glacier was set at 1700 m a.s.l.,
exceeding the surface of the entire study zone by several hundreds
of meters, whereas in the second stage the glacier was removed.
The boundary conditions were maintained constant during both
stages and were set fully restrained at the base and restrained in the
x and y directions at the lateral edges.

The values selected for the material properties are listed in Table
6.2. As mentioned above, the Naßfeld schists were divided into three
layers due to their different degree of fracturation.

Table 6.2. Material properties for the Oselitzenbach 3D model.

 y[kN/m3] E[GPa] v

Naßfeld schists

Upper layer 28 2,5 0,2

Middle layer 28 3,7 0,2

Lower layer 28 3,7 0,2

Hochwipfel schists 28 9 0,1

Glacier ice 9,2 9,6 0,33

The main objective of the analysis was the comparison of the

different computer codes through back-analysis of the displacements
measured during 1991 and 2000. Another aim was to check the
sensitivity of the stability of the slope with respect to the material
parameters. An elastic analysis will not present any failure
mechanism, but it will provide some insight into the landslide
behaviour, as areas with large displacements or tensile stresses may
be potential zones of failure (Zettler & al., 1999).

Figure 6.26 shows a top view of the computed displacements in the

northern (Y) direction (towards the Oselitzenbach). The results
indicate that the location of the maximum displacements coincides
rather well with the most unstable area detected by topographic
measurements.

Figure 6.27 shows the displacement contours in the northern (Y)

direction (towards the Oselitzenbach) on a section through the most
unstable area. The results can be compared qualitatively with the
results calculated by the FLAC3D model (Fig. 6.29 and 6.31) and
show that both numerical models reveal maximum displacements in
very similar zones of the landslide area.
Page 157

Finally, Figure 6.28 indicates the computed deviatoric stresses (J2)

along the same section as used in Figure 6.27, showing a measure
of the shear stress distribution in the zone. The results show
maximum shear in the channel bed of the Oselitzenbach Torrent and
expose high values in the zone of the most unstable area.

FLAC3D model

Two FLAC-models were investigated.

Figure 6.26. Displacement contours in the Y direction (towards the

valley)—top view.
Figure 6.27. Displacement contours in the Y direction—cut view
showing the sliding mass.
Page 158

Figure 6.28. Deviatoric stress (J2) contours, showing maximum

shear at the valley floor.

Model 1

The blocks of the Naßfeld schists and of the Hochwipfel schists are
homogeneous. The mesh of the model (Fig. 6.29) is based on the
digital surface model of the BEV (Fig. 6.24). The grid distance is 25
meters. A Mohr-Coulomb constitutive model was investigated using
the following material properties:

Table 6.3. Material properties applied for the Naßfeld schists.

E [GPa] v φ [°] c [kPa]

2,5 0,2 20 14

Table 6.4. Material properties applied for the Hochwipfel schists.

E [GPa] v φ [°] c [kPa]

9 0,1 40 1e3

The density of the material is 2500 kg/m3. Figure 6.29 shows the
distribution of the shear strain rate and the displacement vectors. In
general, the displacements decrease continuously with depth
resulting from creep (slope sagging). The distribution of the shear
strain rate indicates a zone of maximum shear strain rate in a certain
depth. Below this zone, displacements are zero, above they have a
value increasing to the surface. Thus, the zone of maximum shear
strain rate is the “sliding” zone.

Model 2

Based on the refraction seismic section, the area of the Naßfeld

schist was built as a three layer model (Fig. 6.30). The upper layer
was built with a continuous thickness of 15 m and the middle
Page 159

Figure 6.29. Shear strain rate and displacement vectors.

layer with a thickness of 30 m. The density of the material is 2500

kg/m3. A Mohr-Coulomb constitutive model was investigated using
the following material properties:

Table 6.5. Material properties applied for the Naßfeld schists.

E [GPa] v φ [°] c [kPa]

Upper layer 2,5 0,2 18 14

Middle layer 3,7 0,2 25 20

Lowest layer 3,7 0,2 40 20

Table 6.6. Material properties applied for the Hochwipfel schists.

E [GPa] v φ[°] c [kPa]

9 0,1 40 1e3

In general, Model 2 gave the same results as Model 1. The contour

of the displacement magnitude, Figure 6.31, shows continuously
decreasing displacements with depth.

As a result of the geological investigations and the monitoring

program, a continuum mechanical approach was used. Thus, the
analyses using FLAC gave an area with continuously decreasing
displacements with depth down to a certain depth (“sliding” zone).
The results are in good agreement with the results of the monitoring
program. Figure 6.32 shows the map of the movements, which were
measured between 1991 and 2000 and the area of movements from
the FLAC3D analysis. From this a characteristic profile was deduced
which was used for the two-dimensional run-out calculations.

The distribution of the shear strain rate indicates a zone of maximum

shear strain rate at a certain depth. Below this zone, displacements
and velocities are zero, above it they have a value increasing to the
surface. Thus, the zone of maximum shear strain rate is the “sliding”
zone.
Page 160

Figure 6.30. Homogeneous regions of the model.

Figure 6.31. Contour of displacement magnitude and displacement
vectors.
Page 161

This failure surface was used for the three-dimensional run-out

analysis. Because of the more exact gradient, the contour of the
velocity of the FLAC3D analysis was used to obtain the three-
dimensional failure surface. In several cross-sections of the velocity
contour, the position of the failure surface was defined and read out.
In Figure 6.32 the path of the failure surface in the cross-section A-A
is shown.
Figure 6.32. Map of movements (Moser, 2001) and cross-section of
the main movement directions.

Comparison of the methods

The main objective of the analyses with DRAC and FLAC3D was the
comparison of the different computer codes for the back-analysis of
the displacements measured during 1991 and 2000. DRAC and
FLAC are not directly comparable, as DRAC doesn’t provide any
plasticity models (e.g. Mohr-Coulomb). An elastic analysis will not
present any failure mechanism, but it will provide some insight into
the landslide behaviour, as areas with large displacements or tensile
stresses may be potential zones of failure.
Page 162

These results by DRAC (Fig. 6.26 and 6.27) can be compared

qualitatively with the results calculated by the FLAC3D model (Fig.
6.29 and 6.31) and show that both numerical models reveal
maximum displacements in very similar zones of the landslide area.
The area of movements determined by FLAC as well as that by
DRAC corresponds very well with the results of the monitoring
program (Fig. 6.32).

6.3.5.2 Mechanical “run-out” model

The computer programs PFC2D and PFC3D from ITASCA (1999a,
1999b) were used as an All Ball and a Ball Wall model by TUW and
the computer code DAN and the rock fall program ROTOMAP (Geo
& Soft International, 1999, 2003) was used by POLITO in order to
model the run-out behaviour of the Oselitzenbach landslide.

Generally PFC is not yet able to model the influence of water (e.g.
pore pressure) on the runout, whereas DAN needs an estimate of
the detached rock mass and of the run-out direction (Hungr, 1995).
An estimation of the detached rock mass is also needed for the PFC
Ball Wall model, which was provided by a FLAC3D investigation (see
§ 6.3.5.1). Thus, the methods described should be used in
combination, which makes a comprehensive assessment of the
runout as close to reality as possible.

PFC Ball Wall Model

In the PFC Ball Wall model, the bedrock is simulated by linear (2D)
and planar (3D) elements. In contrast to the All Ball model, where
the relatively stationary bedrock is modelled by balls as well in order
to model the failure mechanism of the slope and the detachment
mechanism also, in the Ball Wall model only the detached rock mass
is modelled by balls. Therefore, in the Ball Wall model, an estimate
of the failure mechanism of the slope and of the detachment
mechanism is needed as an input parameter. Consequently, in the
Ball Wall model, the detached mass can be modelled with the help of
more and smaller balls with the same computational effort. One goal
of the investigations of WP4, therefore, was to compare the two
different approaches.

The PFC Ball Wall model offers the possibility of making use of
know-how related to run-out relevant resistances (factors of
restitution, absorption, friction, etc.) applied in rock fall programs
and, consequently, makes a realistic calculation of the run-out
possible.

The combination with FLAC (see § 6.3.5.1) allows a realistic

estimate of the detached rock mass on the basis of already existing
experiences (see Fig. 6.32). An interaction of detachment and run-
out, however, is not possible.

Figure 6.33. Final state of the PFC2D calculation.

Page 163

Figure 6.34. Deposit in the valley after the rock mass fall.

Figure 6.35. 3D run-out calculation after 16.5 sec.

Page 164

Figure 6.36. 3D run-out calculation after 76 sec.

Figure 6.37. Final state of the Ball Wall model.
Page 165

Table 6.7. Measurement lines and run-out distances.

Measurement line Distance [m]

D1 346

D2 397

D3 573

D4 556

W2 377

PFC All Ball Model

The PFC All Ball model allows the all in one calculation of failure
mechanisms, detachment and run-out. For calculating the failure
mechanism, however, quite a demanding calibration of materials is
necessary. When comparing the volume of the detached rock with
FLAC both methods correspond closely, which verifies the All Ball
model. The surface of the model is rather rough due to the modelling
of the bedrock by balls, which has to be considered at the calibration
of the run-out parameters. This aspect needs further investigation.
Figure 6.38. Initial state of the 3D model.
Page 166

Figure 6.39. Final state of the All Ball model.

Figure 6.39 and Figure 6.40 show the final state of the All Ball
model: the run-out is described with the help of four horizontal
measurement lines (D1–D4), the path width by the lines W1 and W2
(see Table 6.8).

Table 6.8. Measurement lines and run-out distances.

Measurement line Distance [m]

D1 330

D2 389

D3 373

D4 378
W1 204

W2 364
Page 167

Figure 6.40. Run-out distance (D1–D4) & path width (W1, W2).

Comparison Ball Wall model—All Ball model

The area of the maximum displacement rates (light blue) in the All
Ball model corresponds closely to the direction and width of the run-
out in the Ball Wall model. The Ball Wall model, however, indicates a
far greater travel distance (Fig. 6.37). This is due to the collision of
moving particles with stationary ones (bedrock) in the All Ball model,
with the moving particles losing energy additionally. This aspect also
needs further investigation.
Page 168

Figure 6.41. Comparison of the displacements calculated by the ALL

BALL MODEL and the BALL WALL MODEL (plan view).

Rotomap and Dan Code

Since in the DAN method the run-out direction and the path width
are assigned a priori, in the investigations of WP4 they have been
determined with the aid of a 3D rockfall program called ROTOMAP.
The comparison of the results of the PFC2D Ball Wall model, which
also needs an estimate of the runout direction, and the results of the
DAN model shows that two different run-out directions have been
chosen and that these profile directions have an enormous influence
on the results.
The run-out directions determined by Rotomap correspond very well
to those of the 3D All Ball model in the West, whereas the run-out in
the 3D All Ball model in the East indicates a smaller width (Fig.
6.44). The same applies to the 3D Ball Wall model (Fig. 6.43).
Page 169

Figure 6.42. Detachment lines and block edges calculated with

ROTOMAP.
Figure 6.43. Comparison of the BALL WALL MODEL and
ROTOMAP (plan view).
Page 170

Figure 6.44. Comparison of the ALL BALL MODEL and ROTOMAP

(plan view).

In the case of the Oselitzenbach landslide, the comparison of the

results of the methods (Table 6.9) shows that:

– the FLAC3D simulation as the basis of the DAN and of the 3D Ball
Wall model give approximately the same detached rock volume as
the All Ball model,

– he PFC—Ball Wall model and the DAN Code (depending on the

pore pressure assumed) give the same travel distance of the run-
out, whereas the PFC—All Ball model gives a much smaller value
due to the rough surface of the bedrock caused by the bedrock built
up by balls,
Table 6.9. Comparison of the results of the methods.

Method PFC-Ball Wall PFC-All Rotomap+DAN Code

Ball

Detached rock 450’000 (input 377’000–450’000

volume [m3] from WP3, 495’000
FLAC) (input from WP3, FLAC)

(244’300–571’300 depending
on the shape factor assumed)

Travel distance 573 389 597

[m]
(for a pore pressure of 0.1)

470–833

(depending on the pore

pressure assumed)

Run-out width 377 364 340

[m]

Affected area 127’000 97’537 134.494 (Rotomap)

[m2]

Maximum 29 4 21
travel velocity
[m/s]
Page 171

– the run-out widths and the affected areas obtained by the models
correspond more or less and,

– the maximum travel velocities of the run-out obtained by the

models do not correspond. The slow travel velocity of the All Ball is
caused by the rough surface of the bedrock built up by balls. These
aspects should be considered during the calibration of the run-out
parameters and has to be investigated further.

Thus, the combination of all three methods yielded:

– a detached rock volume of some 450’000 m3,

– a travel distance of the run-out of 600 m (due to the blocky nature

of the Naßfeld schists a zero or low pore pressure is assumed to
develop in the run-out mass),

– a run-out width of 360 m,

– n affected area of 130’000 m2, and

– maximum travel velocity of the run-out of 20 m/s.

These data mean that the Nassfeld road will be destroyed or buried
with a maximum height of 7 m by the run-out in case of a slope
failure over a length of 400 m. The Oselitzenbach torrent will be
dammed up as well over a length of 460 m with a maximum height of
14 m, thus endangering the villages of Tröpolach and Watsching,
including the adjoining agricultural and forest areas by debris flows.

Occurrence probability

Scenario 1: Two events of debris flows have been recorded over an

observation period of some 100 years. Thus, an occurrence period
of 50 years has been assumed for Scenario 1.
Scenario 2: In historical times (about 1’000 years), no event as large
as the possible Bodensee landslide has ever been reported. Thus,
the worst case occurrence period for Scenario 2 is 1’000 years.

Scenario 3: The Würmian moraines show no dislocation at the edges

of the Reppwand landslide. Thus, the Reppwand landslide is older
than Würmian, meaning that it happened more than 70’000 years
ago and that the worst case occurrence period for Scenario 3 is
70’000 years.

Table 6.10. Definition of the occurrence probability through the

historical approach.

ScenarioObservation period Recorded Occurrence period Frequency

(years) events (years) (event/year)

1 100 19831987 50 1/50–0,02

2 Historic times None 1.000 1/1’000=1,0e−3

3 70’000 None 70’000 l/70’000=l,

43e−5

6.4 QUANTITATIVE RISK ANALYSIS

In Section 6.3.4, 3 scenarios of the Oselitzenbach landslide evolution

were described. Each scenario is characterised by failure and run-
out areas (derived from the geomorphological and geomemechanical
models) and by an occurrence probability (as a result of an historical
analysis of recorded events).

According to the proposed quantitative risk analysis, each element of

the hazard scenario is used separately in the matrix calculus of risk
evaluation. This process is applied on each recognised scenario;
therefore, at the Oseltzenbach site three separate risk analyses were
carried out.
6.4.1 Elements at risk
The elements at risk are recognised through the definition of the
area affected for each danger in the scenarios previously specified
(Fig. 6.19, 6.20 and 6.21). These elements at risk are shown in
Tables 6.11, 6.12 and 6.13.
Page 172

Table 6.11. Definition of the elements at risk: Scenario 1

(Detachment of the most active area at present).

Elements at risk Name Notes

Forests (private and public Area affected [km2]:

properties)
by landslide: 0,129

by mudflow: 0,896

Persons affected: 1

Heavy traffic or strategic roads National road Length affected [m]:

B90 400

Persons affected: 3

Tourist accommodation Persons affected: 1

Lifelines Length affected [m]:

600

Pylons affected: 4

Table 6.12. Definition of the elements at risk: Scenario 2 (Bodensee

landslide).

Elements at risk Name Notes

Residential areas [valuable buildings or valuable rural WatschigArea

centres: historical, architectural, artistic and/or cultural affected
value] [km2]:

by
mudflow:
 0,137

Persons
affected: 15

Forests (private and public properties) Area

affected
[km2]:

by
landslide:
0,536

by
mudflow:
1,652

Persons
affected: 1

Rural area Area

affected
[km2]:

by
mudflow:
0,204

Heavy traffic or strategic roads National Length

road B90affected
[m]: 730

Persons
affected: 3

Tourist accommodation Persons

affected: 5
Lifelines Length
affected
[m]: 1370

Pylons
affected: 9

Table 6.13. Definition of the elements at risk: Scenario 3

(Reactivation of the Reppwand landslide).

Elements at risk Name Notes

Residential areas [valuable buildings or valuable Watschig Area

rural centres: historical, architectural, artistic and/or affected
cultural value] [km2]:

by
mudflow:
0,500

Persons
affected:
129

Tröpolach Area
affected
[km2]:

by
mudflow:
0,120

by
inundation:
0,248
 Persons
affected:
180

RattendorfArea
affected
[km2]:

by
inundation:
0,271

Persons
affected:
50

Forests (private and public properties) Area

affected
[km2]:

by
landslide:
2,911

by
mudflow:
1,804

by
inundation:
0,2

Persons
affected: 1

(contd)
Page 173

Table 6.13. (contd)

Elements at risk Name Notes

Rural area Area affected [km2]:

by mudflow: 0,204

Heavy traffic or strategic roads National road B90 Length affected [m]:

by landslide: 3450

Persons affected: 3

Secondary railway Length affected [m]:

by mudflow: 2422

by inundation: 3440

Tourist accommodation Persons affected: 5

Lifelines Length affected [m]:

by mudflow: 1935

by inundation: 2628

Pylons affected:

by mudflow: 15

by inundation: 11

6.4.2 Vulnerability of the elements at risk

The vulnerability of the elements at risk was determined by an
estimation of the effects on the elements at risk. Table 6.14 shows
the vulnerability evaluation.

Table 6.14. Vulnerability evaluation as a function of estimated effects

on elements at risk.

Physical vulnerability Social vulnerability

Vulnerability Loss IndexVulnerability description Index

description range

Intact structures 0 0 Non affected persons 0

Local damages 1÷25% 0.25 Non physical damages; 0.25

evacuated persons

Seriously damages 26÷50% 0.5 Physical damages (person 0.5

(possible to repair) continue theiractivities)

Mostly destroyed 51÷75% 0.75 Seriously wounded persons 0.75

(difficult to repair) (50% disability)

Total destruction (out of 76÷100% 1 Died, 51–100% disability 1

use; e.g. >5% inclination)

Economical vulnerability Environmental vulnerability

Vulnerability description IndexVulnerability Loss Index

description range

Non interruption 0 Intact element 0 0

Short temporary interruption 0.25 Local loss 1÷25% 0.25

(hours to day)
Seriously 26÷50% 0.5
Average temporary interruption 0.5 damages (possible
(days to week) to repair)
Longtemporary interruption (weeks 0.75 Mostly destroyed 5 1÷75% 0.75
to months) (difficult to
repair)

Permanent interruption 1 Total destruction 76÷100% 1

The following tables (6.15–6.20) show the considered values of the

elements at risk and the evaluated vulnerability for each scenario.
Page 174

Table 6.15. Scenario 1: Considered values of the elements at risk.

Considered values [VE]

Elements at risk PhysicalEconomicEnvironm.Social

Forests/rural area (private and public 1 1 3 1

properties)

Heavy traffic or strategic roads 3 4 1 2

Tourist accommodation—buildings 3 3 1 1

Lifeline 1 2 1 0

Table 6.16. Scenario1: Vulnerability.

Vulnerability [V]

Physical Economic Environm. Social

Vulnerability 0,25 0,5 0,25 0,25

Table 6.17. Scenario 2: Considered values of the elements at risk.

Considered values [VE]

Elements at risk PhysicalEconomicEnvironm.Social

Residential areas 4 1 1 3

Forests/rural area (private and public 1 1 3 1

properties)

Heavy traffic or strategic roads 3 4 1 2

Tourist accommodation − buildings 3 3 1 1

Lifeline 1 2 1 0

Table 6.18. Scenario 2: Vulnerability.

Vulnerability [V]

Physical Economic Environm. Social

Vulnerability 0,5 0,5 0,5 0,5

Table 6.19. Scenario 3: Considered values of the elements at risk.

Considered values [VE]

Elements at risk PhysicalEconomicEnvironm.Social

Residential areas 4 1 1 4

Forests/rural area (private and public 1 1 3 1

properties)

Heavy traffic or strategic roads 3 4 1 2

Secondary railway 1 2 1 0

Tourist accommodation—buildings 3 3 1 2

Lifeline 1 2 1 0
Page 175

Table 6.20. Scenario 3: Vulnerability.

Vulnerability [V]

Physical Economic Environm. Social

Vulnerability 0,75 0,75 0,75 0,75

6.4.3 Expected impact

Multiplying the values of the elements at risk (VE, see Tables 6.15,
6.17 and 6.19), and vulnerability percentage V (obtained as
described in the previous section), the expected impact C is obtained
as C=VE×V

To state a precise quantitative damage evaluation for economic

activities and assets, costs derived from economic analyses are
essential (e.g. assets costs, reconstruction costs, turnover, profits
loss, etc.). To assess expected environmental and social impacts
(where an economic value is difficult to obtain), relative value indices
are used.

In the quantitative risk analysis of the Oselitzenbach landslide,

detailed economic evaluations were considered. Arbitrary relative
values indices were applied for each category. In Tables 6.21–6.23
expected impact values of each scenario are shown.

Table 6.21. Scenario 1: expected impact.

Expected impact [C=VE×V]

Elements at risk Physical Economic Environm. Social

consequ. consequ. consequ. consequ.

Forests/Rural area (private 0,25 0,5 0,75 0,25

and public properties)
Heavy traffic or strategic 0,75 2 0,25 0,5
roads

Tourist accommodation— 0,75 1,5 0,25 0,25

buildings

Lifeline 0,25 1 0,25 0

Table 6.22. Scenario 2: expected impact.

Expected impact [C=VE×V]

Elements at risk Physical Economic Environm. Social

consequ. consequ. consequ. consequ.

Residential areas 2 0,5 0,5 1,5

Forests/rural area (private and 0,5 0,5 1,5 0,5

public properties)

Heavy traffic or strategic 1,5 2 0,5 1

roads

Tourist accommodation— 1,5 1,5 0,5 0,5

buildings

Lifeline 0,5 1 0,5 0

Page 176

Table 6.23. Scenario 3: expected impact.

Expected impact [C=VE×V]

Elements at risk Physical Economic Environm. Social

consequ. consequ. consequ. consequ.

Residential areas 3 0,75 0,75 3

Forests/rural area (private and 0,75 0,75 2,25 0,75

public properties)

Heavy traff ic or strategic 2,25 3 0,75 1,5

roads

Secondary railway 0,75 1,5 0,75 0

Tourist accommodation— 2,25 2,25 0,75 1,5

buildings

Lifeline 0,75 1,5 0,75 0

6.4.4 Risk assessment

The last phase of the quantitative risk analysis concerns the risk
evaluation. This evaluation is obtained by multiplying the expected
impact values C (Tables 6.21–6.23) and the numeric values of the
occurrence probability P related to each scenario (Table 6.10):
R=C×P. Tables 24, 25 and 26 show the risk assessment results for
each scenario. The same results are also represented by a zonation
shown in Figures 6.45–6.48 for Scenario 1 and in Figures 49 and 50
for Scenario 2. Due to the extremely small risk values of Scenario 3,
the physical, economical, environmental and social risk are not
shown by a relative zoning (map of zones of equate risk) of the
values.
Table 6.24. Risk assessment for Scenario 1: R1= C1×1/50.

Risk assessment [R=C×P]

Elements at risk Physical Economic Environm. Social

risk risk risk risk

Forests/rural area (private and 0,005 0,01 0,015 0,005

public properties)

Heavy traffic or strategic roads 0,015 0,04 0,005 0,01

Tourist accommodation—buildings 0,015 0,03 0,005 0,005

Lifeline 0,005 0,02 0,005 0

Table 6.25. Risk assessment for Scenario 2: R2=C2×1/1.000.

Risk assessment [R=C×P]

Elements at risk Physical Economic Environm. Social

risk risk risk risk

Residential areas 0,0020 0,0005 0,0005 0,0015

Forests/rural area (private and 0,0005 0,0005 0,0015 0,0005

public properties)

Heavy traffic or strategic roads 0,0015 0,0020 0,0005 0,0010

Tourist accommodation—buildings 0,0015 0,0015 0,0005 0,0005

Lifeline 0,0005 0,0010 0,0005 0,0000

Page 177

Table 6.26. Risk assessment for Scenario 3: R3=C3×1/70.000.

Risk assessment [R=C×P]

Elements at risk Physical Economic Environm. Social

risk risk risk risk

Residential areas 4,3e–05 1,1e–05 1,1e–05 4,3e–05

Forests/rural area (private and 1,1e–05 1,1e–05 3,2e–05 1,1e–05

public properties)

Heavy traffic or strategic roads 3,2e–05 4,3e–05 1,1e–05 2,le-05

Secondary railway 1,1e–05 2,le-05 1,1e–05 0,0e+00

Tourist accommodation—buildings 3,2e–05 3,2e–05 1,1e–05 2,le-05

Lifeline 1,1e–05 2,le–05 1,1e–05 0,0e+00

Figure 6.45. Physical risk in Scenario 1.
Page 178

Figure 6.46. Economic risk in Scenario 1.

Figure 6.47. Environmental risk in Scenario 1.

Page 179

Figure 6.48. Social risk in Scenario 1.

Figure 6.49. Physical risk in Scenario 2.

Page 180

Figure 6.50. Social risk in Scenario 2.

6.5 REGULATION AND RISK MITIGATION MEASURES

ALREADY AVAILABLE

6.5.1 Viability
In order to prevent serious debris flow in the Oselitzenbach torrent
and in the receiving stream “Gail”, the following measures are
possible:

– Linear measures: “Steps” on stream bed and debris retention in

the flood outlet areas

– Forestry measures to improve the outlet conditions

– Measures such as draining and landfills to reduce the massive

slope movements.

6.5.2 Facilities
An extensive project of construction work and research was started
in 1988 to prevent debrisgenerating slope failures (especially at the
toe zone of the Reppwand slide) and the destruction of the “Naßfeld
Road”. As a part of the construction measures, a 400 m long channel
was excavated in massive Hochwipfel formations (comp. 1.2) and a
landfill was made at the toe of the excavated material as well (about
170’000 m3).

Additional measures were: Drainage of the Quellenbach landslide

between 830 and 920 m a.s.l. and the draining-off of water, also
above 950 m a.s.l. to the area of the “Bodensee” Lake. These
measures were necessary to prevent serious debris flow in the
Oselitzenbach Torrent and in the receiving stream “Gail” and to
prevent debris flow from reaching the villages “Tröpolach” and
“Watsching”. For economic reasons, the correction of all the debris
sources from the very top downwards was not deemed possible or
necessary.

Additionally, a monitoring program was started in October 1988. The

toe zone of the landslide has been intensively monitored as
presented above in § 6.3.3. At present a risk management system
according to the results of the IMIRILAND project is in preparation by
local authorities (e.g. WLV—National Authority for torrent and
avalanche control) and the project members.
Page 181

7
The Ceppo Morelli rockslide

G.Amatruda1, M.Castelli1, F.Forlati2, M.Hürlimann3, A.Ledesma3,

M.Morelli4, L.Paro2, F.Piana4, M.Pirulli1, R.Polino4, P.Prat3,
M.Ramasco2, C.Scavia1, C.Troisi2
1Politecnico di Torino; 2Arpa Piemonte; 3Universitat Politècnica de
Catalunya; 4CNR-IGG sezione Torino

7.1 INTRODUCTION

The Ceppo Morelli rockslide extends on the left side of the middle
Anzasca valley in the Pennine Alps, northern Piedmont (Italy), a few
kilometres from the border with Switzerland (Fig. 7.1). The national
road N° 549 runs on the valley bottom for about 30km, from the
Piedimulera village (Toce River plain) to the small village of Pecetto
di Macugnaga (at the foot of the Monte Rosa Massif). The road is the
only link to Macugnaga, a popular tourist resort close to Monte Rosa.
The Anza River drains the valley.

Attention has been focused on this well-known rockslide as a

consequence of relevant slope instability phenomena which occurred
during the October 2000 heavy meteorological event.
Figure 7.1. Digital Terrain Model of the north-western Piedmont with
location of the Ceppo Morelli landslide area (red star) main rivers
and lakes (in blue), and villages (in yellow). General overview of the
left slope of Anzasca valley (air photographs overlapped by Digital
Terrain Model) with Ceppo Morelli landslide (white arrow).

7.1.1 Landslide description

The direct effect of the catastrophic October 2000 heavy
meteorological event was the reactivation of a large ancient
rockslide, which affects the middle-lower part of Monte Rubi’s
southern slope (1850–1200 m a.s.l.). The slide involves a massive
gneissic rock mass. The overall surface is around 160’000 m2 and
the volume estimation is between 4 and 6×106 m3.
Page 182

Generally speaking, the whole unstable area can be divided in two

main portions, a lower one, where the rock mass, although fractured
and partially toppled and translated, is still ordered, and an upper
one, constituted by coarse blocky debris. The upper part (delimited
by tension cracks) moved due to the October 2000 event.
Displacements reaching more than several centimetres (up to 400–
500 cm) have been observed down a slope of about 30°–35°,
inducing a lower part reaction, especially some hundred meters over
the slope foot (1250 m). Numerous rockfalls reached the bottom of
the valley (800 m a.s.l.) with boulders up to 300 m3. Several
boulders reached the National Road No. 549. As the deformation
affecting the lower part of the unstable area increases, a composed
surface might be formed, leading to a huge rockslide and possibly
severe rockfall and/or rock avalanche phenomena.

Major mass falls may reach two small villages dating back to the XVI
century: Prequartera and Campioli located at the bottom of the
valley.

7.1.2 Historical background

The site has been involved in repeated rockslides and falls since
ancient times. A chronicle dating from the XV century reports failures
in the years 312 and 843 A.D.; in the latter episode 25 knights were
killed (allegedly carrying a sort of treasure). Remaining vivid in local
history after several centuries (though unfortunately lacking written
reports), the quoted events probably refer to mass failures rather
than simple rockfalls.

“The year of the fall, 1816,” is engraved on a 1’000 m3 boulder close

to the Anza River. Local residents report that rockfalls are extremely
common. In April 1977, severe rockfalls damaged the national road
and some boulders up to 1’000 m3 reached the valley-road (Fig.
7.2).
Figure 7.2. April 1977: rock fall involving the national road N 549
(boulder of about 50 m3, prismatic shape).
Page 183

Figure 7.3. Rock fall triggered during the October 2000 heavy
meteorological event. The boulder (right red arrow) damaged the
national road (red line) and stopped near a house.

During the October 2000 severe event, several boulders up to 300

m3 reached the river and the National Road (Fig. 7.3). Some local
rockfalls occurred in June 2002.

7.2 REGIONAL FRAMEWORK

7.2.1 Climate and water condition

Annual rainfall has been monitored over a period of 35 years, from
1951 to 1986. Its annual average value of about 1’594 mm, with 119
rainy days per year and average intensity of 13.4 mm/day, has been
derived from the basis of data recorded in 14 stations belonging to
the Idrografico e Mareografico National Service. Monthly rainfall
peaks have been monitored at the Anzino station, located at the
bottom of the valley, at an elevation of about 670 m and 6.5
kilometres away from the Ceppo Morelli rockslide. The months of
May and October are characterised by the highest monthly rainfall
value (198 mm and 190 mm) and by the largest number of rainy
days (12) (Biancotti & Bovo, 1998 a and b).

The high intensity event which occurred on 13–16 October 2000

produced a total precipitation of about 558 mm. The average daily
rainfall value recorded from 13 to 15 October was almost equal to
the highest monthly rainfall value recorded over a period of 35 years
(Anzino).

The annual mean temperature is 9°C and there is an average of 93

frost days per year. Snow information is related to the Alpe Cavalli
gauges (a station located at 1’510 m a.s.l.); data refer to 1966–1996
continuous measures: max. snow level 785 cm (1971–72), min. 137
cm (1985–86), average 438 cm. During the late spring days, the
snow cover melts away quickly. As a result, a considerable quantity
of water can seep down into the sliding body. The flow of water in the
slope can then occur through the main discontinuities, which are
generally open and persistent. Considering that the upper part of the
slope acts as a reservoir, the largest amount of groundwater is
cumulated when heavy meteorological events occur or when rainfalls
combine with snow melting.
Page 184

7.2.2 Regional morphology

The Anzasca valley is mainly characterised by fluvial and glacial
morphogenesis and by landslide activity (rockfalls, large rotational
and translational slides, deep seated gravitational deformations).
The Anzasca valley is steep and impervious, characterised by steep
rock walls seldom interrupted by scree deposits and dejection cones.
The bottom valley altitude ranges from 247 m (Piedimulera village) to
1’378 m (small village of Pecetto di Macugnaga) along a 30 km
length. Just beyond the village of Ceppo Morelli, the stream profile is
interrupted by the M orghen rocky step that bounds the Macugnaga
plain.

7.2.3 Regional geology and structural setting

The landslide site is found in the Monte Rosa basement nappe
(Upper Pennine Units) represented by polymetamorphic
augengneisses and paraschists. This basement is characterised by
several Alpine ductile deformation phases (Fig. 7.4).

At the regional scale, the basement is deformed by the Vanzone

antiform with a typical SW plunging axis and NW dipping axial plane.

The regional sin-metamorphic ductile deformation is characterised

by the development of regional schistosity related to large-scale
folding events that gave origin to local macrostructures such as the
Vanzone antiform.

The regional brittle tectonic deformation is characterised by

mesofaults and joint systems, partially parallel to the axial planes of
the nVanzone.

Recent alluvial-colluvium deposits are represented by

heterogeneous and incoherent masses of soil material and rock
fragment deposits.
Figure 7.4. Geological map of a sector of the Western Alps (CNR,
1990–Structural Model of Italy 1:500’000 scale). In the red circle is
the landslide site placed within one of the main Alpine tectono-
metamorphic unit (Monte Rosa Unit MRU), mainly made up of
augengneisses and micaschists folded by Vanzone antiform (VA)
and bordered on both sides by the Piemonte Zone (Penninic domain,
in yellow). East of MRU are the Austroalpine domain (dotted orange
and brown) while east of the Canavese Line (CL), a major fault of the
Alps, are the southern Alps with the granulites of the Ivrea-Verbano
Zone (Violet, light brown and green) and granitoids and micaschist of
the Serie dei Laghi (pink and dotted pink).
Page 185

7.3 HAZARD ANALYSIS

7.3.1 Geomorphological analysis

This high part of the Anzasca valley is characterised by a steep rock
wall and a narrow, almost flat alluvial valley floor. These
morphological features were caused by glacial erosion of highly
fractured bedrock. The toe of the slopes is generally covered by very
thick but discontinuous detrital fans and scree deposits, as in the
sector below the Ceppo Morelli landslide, found at the base of the
left slope, between the villages of Campioli and Prequartera.

This sector is studded with erosional scars separated by ridges of

relatively safe rocks, both of which features represent the peculiar
morphological aspects of the Ceppo Morelli landslide.

In order to understand the rockslide behaviour and to assess its

probable impact on land planning, an intensive field survey and a
systematic aerial photo analysis were conducted. The latest one was
carried out on different flights (1970, 1978, 2000, and 2001), as a
consequence of the 1977 and 2000 heavy meteorological events.
Topographic and geometric attributes of the slope were derived from
an available Digital Elevation Model (DEM). The entire studies and
investigations have evidenced a morpho-dynamic situation,
sufficiently clear in terms of recent and ancient slope evolution and in
agreement with the geological and structural observations carried
out during field surveys.

The landslide has been subdivided into three areas (see Fig. 7.5):

A. Detachment area

B. Rock mass translation area

C. Accumulation and landslide affected area

A. Detachment area (upper part)

The detachment area characterises the upper and the right

boundaries of the landslide. It presents two main discontinuity
systems with NE (K1) and NW (K2) directions (see Fig. 7.6).

The K1 system, together with minor N-S striking discontinuities,

corresponds to the right lateral boundary of the Sector A and has
hindered landslide movement towards the SW. A third, complex
system consists of two families of SSW-dipping joints that developed
on the schistosity planes. These joints are labelled here S1 and S2
since they are roughly coincident with the schistosity itself.

B. Rock mass translation area

This area, which corresponds to a dislocated rock mass, has been

subdivided into three main sectors characterised by different
morpho-structural features (Fig. 7.5):

• Sector B1 consists of an intensively fractured rock mass and it has

been affected by significant movements (up to 5 m) towards the
SSW (Fig. 7.7) that occur on a sliding surface roughly subparallel to
the schistosity. The displacements are well evidenced by continuous
open cracks in the upper part of this sector.

• Sector B2 consists of a less fractured rock mass; it is suggested

here that the B2 sector has been affected only by minor movement
toward the SW since it still preserves its original structural setting.

• Sector B3 consists of a dislocated rock mass and it is separated

from the B2 sector by a pronounced NE-striking open fracture. In
Sector B3, the inclination of the slope is greater than in Sector B2
and the rocks show bulging deformations and toppling. The lower
boundary of Sector B3 does not correspond to any failure surface,
but it represents the morphological toe of the landslide that, although
partially masked by plant cover, can be recognised by an alignment
of scars, source areas of debris flows and rockfalls.
C. Accumulation and landslide affected area

This fan-shaped area lies below the supposed landslide toe and
consists of debris featuring big rock blocks. On the basis of the
geomorphological features, it has been subdivided into two sectors:

• Sector C1 is characterised by fan-shaped debris deposits that

probably fell down from scars placed in the SE part of Sector B3 and
that should have reached the area of the Prequartera village. The
Page 186

Figure 7.5. Main features characterising the Ceppo Morelli landslide:

1) detachment area; 2) rock mass translation area; 3) landslide
boundary; 4) crown of an ancient rockfall; 5) accumulation and
landslide affected area; 6) continuous open crack delimiting the
landslide in the upper part; 7) main open cracks 8) main fissures
observed in the disaggregated rock mass; 9) debris flows and
rockfall trajectories; 10) debris flows and rockfall trajectories (1977
event) 11) main boulders fallen in October 2000; 12) sector where
relevant movements occurred in October 2000; 13) main schistosity
dip direction; 14) section S-S1 (see Fig. 7.18).
Page 187

Figure 7.6. “On-site” available data; Schmidt diagram (lower

hemisphere) counter of the poles distribution of joints. Legend: S1
and S2: schistosity; K1, K2 and K3: joint systems. The arrow
indicates the sliding direction (Regione Piemonte, 2000).

geomorphological features of debris suggest that they are relatively

old, although more rockfalls occurred as recently as during the 2000
meteorological crisis.

• Sector C2 is larger and more active than C1 and rockfalls and

debris flows currently occur. This sector shows an advanced central
lobe where larger blocks have reached and exceeded the valley
road.

7.3.2 Local geology and structural analysis

As a consequence of the 2000 severe meteorological event, a rapid
field reconnaissance was carried out (Regione Piemonte, 2000).
New structural analyses were then performed with the principal aim
of achieving improved knowledge of the physical and geometrical
characteristics of the main structural elements.

This analysis was based on two data sets:

a) available geological data mainly collected at the landslide site

during the October–November 2000 rapid field reconnaissance (“on
site data”),

b) new data consisting of:

– mesostructural “around site” data;

– data from remotely sensed and aerial photo lineaments.

Page 188

Figure 7.7. The topographic control points allowed the identification

of the direction and magnitude of the slip surface.

7.3.2.1 “On site” available data

The available data refer only to Sectors A and B1 of the landslide
and underline the presence of three discontinuity systems and a
composite metamorphic foliation, summarised in Figure 7.6. On the
basis of these observations, interpretations of the landslide
kinematics were attempted, since the orientations of the prominent
discontinuity systems roughly correspond to the morphological
boundaries of the rockslide.

7.3.2.2 “Around site” structural analysis

Structural analyses were performed around the unstable area, in
order to:

• obtain a better geological characterisation of the structural systems

described above;
• verify if the main morphological boundaries between the internal
areas and the sectors of the landslide could be related to main
tectonic structures;

• define the hierarchical and cross-cutting relations of the “around

site” structures to compare them with those observed in the landslide
area, and consequently derive some tentative “extrapolation rules”;

• give information on the mechanical properties of the rocks and the

structural elements to be used in the geomechanical model.

The results of these analyses will be described in the following

sections.

Mesostructural analyses

The analyses were performed at the boundary of the landslide area,

near the boundaries between the different sectors and, in few cases,
on isolated rock blocks that, although displaced, did not suffer major
tilting or rotations.

These analyses pointed out that the available data (Fig. 7.6),
although collected only in the upper part of the slope, are significant
for the whole landslide body and that they can consequently be
assumed as statistically representative.

Furthermore, a better geometrical and kinematic characterization of

all the structural elements was given in order to obtain conceptual
tools for the correlation of the different structural elements.
Page 189

Regional schistosity

The schistosity regularly dips to the SSW, both within and outside
the landslide area. Nevertheless, in the SW part of the landslide, it is
characterised by a steeper inclination (50–60°); in this sector it has
consequently been labelled as “S2”. On the other hand, the more
widely diffused low to medium angle schistosity has been indicated
as S1. The schistosity also dips to the NE in some places within the
landslide, probably due to the presence of open folds or the local
tilting of isolated rock blocks.

Since S1 has a relatively constant orientation, it can easily be

extrapolated along strike and at depth for at least several hundred
meters.

Minor quasi-planar fracture sets are locally developed on the

schistosity planes. They are mainly metric surfaces dipping to the
SSW at a low angle with respect to the schistosity inclination. These
surfaces, named here ZT (Fig. 7.8 and 7.9), are interpreted as
boundary planes of minor brittle shear zones which have been
locally reactivated by frictional processes having occurred on the
schistosity surfaces. Minor NNE-dipping conjugate shear zones are
also present.

Furthermore, subordinate, conjugate steep or medium angle joints

(K) are also developed. They consist of two conjugate systems
dipping to the SSW and the NNE, respectively. The crosscutting
relations of ZT and K allow their interpretation as synthetic and
antithetic Riedel shears within the ZT shear zones. This
interpretation is corroborated by the scale invariance of the ZT-K
crosscutting relations.

The schistosity is refolded by centimetre to decimetre-scale

crenulation folds that gave origin to sub-horizontal crenulation
cleavages, locally evolved into open joints (Ft, see below) due to
unloading processes.

Finally, hydrothermal alteration bands have been locally observed on

the schistosity planes. These are metre thick lenses or beds where
decimetre-scale carbonate nodules can be observed, locally affected
by karstic dissolution processes. These bands are characterised by
poor, or very poor, rock quality index.

Figure 7.8. Around-site data. Schmidt diagram (lower hemisphere)

average cyclograph trace of the main discontinuity systems. Legend:
S1 and S2: schistosity; ZT: shear zone; K, K1 and K2 joint systems;
F: regional faults and Fb: right-lateral boundary fault; Ft: release joint
system.
Page 190

Figure 7.9a-b. Crosscutting relationship between ZT and K at

different scales observed at the western boundary of the landslide.

Fractures and fault systems

One of the most prominent is the K1 (Fig. 7.8 and 7.10a) system that
is considered here as a comprehensive discontinuity system whose
strike ranges from NNE to NE, and which has been synthetically
represented in Figures 7.6 and 7.8 as a single average plane. K1 is
indeed characterised by different sub-systems:

• roughly NE-striking highly persistent fractures (hundreds of meters)

and largely spaced (from 10 to 100 m) joints;

• faults, well developed in the lower part of the slope, namely where
the slope shows a greater inclination (at about 1250 m, see Fig. 7.5
Sector B3). These last surfaces strike from NE to ENE and consist of
very steep planes, tens of meters long and dipping both SW and NE.
It is assumed here that these faults belong to a fault zone; this
interpretation is mainly based on the geometric array of F elements
(scattered and anastomosed discontinuities) at both the regional and
local scales.

The western boundary fault of the landslide (Fb) also strikes sub-
parallel to the K1 system. This fault is characterised by a cataclastic
belt tens of meters long; minor slip planes within this belt indicate S-
ward right-lateral movements.

Other metric NNE reverse faults (Fi) have been also locally observed
(Fig. 7.10a). The NNE striking steeply dipping Fb and Fi displace the
less inclined ZT shear zones.

A second system (K2) (Fig. 7.8 and 7.10a, b) consists of sub-vertical

NW striking fractures, that tend to the NE in the upper sector of the
landslide and that are more persistent and frequent on the left side
on the landslide. These fractures consist of discontinuities tens of
meters long that split the rock mass into blocks of different shape
and dimension, due to some irregularities in the fracture spacing. In
some cases, the K2 discontinuities correspond to reverse fault
surfaces.

A third discontinuity system is mainly made up of joints (Ft) (Fig. 7.8

and 7.10b and c) that strike sub-parallel to the K1 system but show
considerably lower inclination. The Ft mainly consists of release
joints (Engelder, 1985) originated by unloading processes, since
their attitude, generally less inclined than the slope, allows the sliding
of rock wedges defined by the intersections of the fracture systems
described above.
Page 191

Figure 7.10. Crosscutting relationship and geometry of the main

structural features. a) ZT, K1 and K2 observed in the western
surroundings of the landslide. This relationship shows one of the
peculiar rock block shape observed at the western of the landslide;
b) Fi reverse fault, K2 fracture and Ft release joints observed at the
western boundary of the landslide; c) major Ft release joints
observed in the western of the landslide; d) Sigmoidal fracture
systems associated to the F strike slip faults observed on the
opposite side of landslide.
Finally, a fourth discontinuity system consists of regional EW steep
strike-slip faults (F) (Fig. 7.8 and 7.10d), well exposed in the lower
part of the slope and particularly on the opposite side of landslide, in
the right slope of the Anzasca valley. Spacing and persistence of this
system are particularly evident at the plurimetric scale. Sigmoidal
fracture systems are usually associated with the F planes at different
scales, where they delimitate rock slices interpreted as strike-slip
duplexes (Woodcock & Fischer, 1986).

The hierarchical and geometrical relations of fault and fractures

described above need to be compared to the geomorphological data
and observations. Remotely-sensed satellite and aerial photograph
data are very useful for this purpose, since they allow geometrical
and morphological comparisons of the detected lineaments with
respect to the effective structures well exposed on the field. Such
comparisons allow the extrapolation of the fracture network that has
been locally
Page 192

observed in the field to relatively larger areas surrounding the

studied site, and vice-versa, in order to better constrain the fracturing
conditions in the numerical model.

Remotely sensed and aerial photo lineaments

The geometric and hierarchic relations between the discontinuity

systems observed on Ceppo Morelli landslide were also analysed at
the regional scale by means of interpretation and statistical analysis
of Landsat TM lineaments (Fig. 7.11) and at the scale of the slope by
photo lineament detection on the DVP stereo viewer (digital video
plotter, Geomatic Systems, Inc.) (Fig. 7.12). The photo lineament
analysis was carried out on the NE boundary of the landslide and
over its SW surroundings. Their combined use led to a clearer
identification of the geometry and distribution density of the structural
elements in the “around site” landslide area.

Figure 7.11. Landsat TM image processed (colour combination of

bands 7–5–4 RGB) with lineaments (in red colour). The green
dashed line is the area represented in Figure 7.12.
Morpho-structural and spectral criteria (the last only for satellite
Landsat images) were used for the lineament detection. The
association and correlation of lineaments was carried out using the
“Geometrical Lineament Identification” method (Morelli, 2000; Morelli
& Piana, 2003).

The identified lineaments, represented here in Figures 7.11 and

7.12, were first analysed statistically and then compared with
structural field data.

The statistical analysis was made to characterise the relationships

between the azimuthal frequency distributions and lineament lengths
for cumulative lengths and length classes. The statistical description
is illustrated in the rose diagrams in Figure 7.13.

There are evident similarities between the aerial photos and the
satellite lineament data. The former shows a distribution range from
N20 to N60 with a maximum at N40. These lineaments are mainly
concentrated in the NE. The greatest lineament length was found
along the N80–90 direction.

The satellite data display the same distribution for both frequency
and length diagrams with three average maxima at N290, N350–20
and N50. The lineaments can also be distinguished in terms of their
length to determine the relations between frequency and length.
Frequencies were divided into three length classes according to
statistical distribution.

The aerial photo data show shorter and intermediate lineaments

mainly at N20–60 and two relative maxima of shorter lineaments at
N330–10 and on the intermediate length at N290–310. The greatest
lengths (mostly>80 m) are along the N80–90 direction.
Page 193

Figure 7.12. Aerial photograph with the identified photo lineaments

(red lines) and the Ceppo Morelli landslide (blue dashed line).

The satellite data display a wider length distribution with short,

intermediate and long lineaments with an average direction of N50–
60 and mainly short and intermediate lengths along the N360–20,
N280–290 and 290–310 directions.

In summary, in both the aerial photo and the satellite data sets, four
main lineament systems can be recognised: the NE-SW (Ln1CM)
and WNW-ESE (Ln2CM) striking systems display higher azimuthal
frequency and longer lineaments; the NW-SE (Ln3CM) and N-S
(Ln4CM) striking systems show, instead, lower azimuthal frequency
and lengths.

The statistical and the geometrical pattern distribution of each

system have been compared with the structural field data.
• The Ln1 CM system consists of more widely distributed length and
direction lineaments. In plan view, this distribution is described by
linked lineaments that show an anastomosed geometric pattern, well
evident throughout the whole slope. This geometry points to the
presence of strongly fractured zones probably associated with steep
fault systems consisting of many hierarchic orders of fault elements.
This geometric pattern could be related among the F fault systems
and the K1 fracture system.

• The Ln2CM system, well represented in the longer length classes,

also consists of regularly and widely spaced lineaments, well
exposed in the surroundings of the Ceppo Morelli landslide. This
geometric pattern suggests that it could correspond to regional
tectonic structures. In the field, the Ln2CM system corresponds
locally to the F strike-slip fault system, not easily recognisable at the
landslide site but well exposed in the surrounding of the landslide.

• The Ln3CM system is uniformly distributed over the Ceppo Morelli

slope. It crosses the Ln1CM and Ln2CM lineaments or ends against
them. This system partially includes the K, K1 and ZT systems, when
the latter has greater inclinations.

• The Ln4CM system consists of fewer and shorter lineaments than

the others. Its lineaments often connect with those of the LnM2 and
the LnM3. This geometric relationship defines same
Page 194

Figure 7.13. Landsat and aerial photo lineaments diagrams showing

the azimuthal frequency and cumulative lengths (a, b and g, h) and
azimuthal frequency divided into classes of lengths (c-f, and i-n). L:
length of the lineaments; f: azimuthal frequency of the lineaments.
Page 195

hectometre to decametre-scale sectors whose geometry is similar to

that explained in the field by the crosscut relations among Fb, K2
and K1. This relationship is also in agreement with the shape of the
many tilted rock blocks observed at the toe of the landslide.

The geometric and hierarchic relationships observed at the outcrop

scale are also confirmed by the statistical distribution and geometric
characteristics of the lineament systems, analysed at both the
regional and the slope scales. This correspondence allows
extrapolation of the geometry and hierarchy of the structural
elements to different scales.

7.3.3 Investigation and monitoring

After the October 2000 intense rainfall event, the Ceppo Morelli
rockslide has been monitored by means of (Fig. 7.14):

• 25 topographic benchmarks, measured by means of an automated

total station located in a shelter on the facing slope;

• 9 wire extensometers, provided with 3 thermometers that allow

compensation for thermal drift;

• 1 rain gauge.

All of the instruments are connected to automated data recording

and transmission systems. All data are transmitted to a remote
monitoring station by means of a radio modem and GSM modules.

At the moment, only the topographic monitoring data are available.

These data are subdivided into three phases because of two
maintenance interventions, which have caused different placements
of the automated total station.

Figure 7.15 shows total displacement vectors from 5/8/2001 to

11/5/2003 (i.e. Phase 3); during this period the following
displacements were measured:

• the topographic benchmarks in the upper part of the sliding body

(Sector B1) show displacements on the order of 85–145 mm, with a
velocity of about 4 to 12 mm/month; displacement vectors are mainly
oriented southwards. Exceptions are represented by Point 21, which
shows

Figure 7.14. Position of the monitoring instruments located on the

Ceppo Morelli rockslide.
Page 196

the highest total displacement (262 mm), and Points 8, 14, 21 and
23, located on the scarps of the detaching niche, which seem to be
still;

• the topographic benchmarks in the lower part of the slide body

(Sector B3) show displacements up to 400 mm (Point 3), with a
velocity between 3 and 19 mm/month, displacement vectors are
scattered between SSE and SSW.
Figure 7.15. Total displacement vectors measured by means of the
topographic benchmarks from 5/8/2001 to 11/5/2003. Displacement
vector scale: 1:35.

The recorded displacements confirm what was found out by means

of the geomorphological analysis (see § 7.3.1): Sectors B1 and B3
show different behaviours, in particular with respect to velocity, which
is greater in the lower part. This is explainable by thinking of the
disengaging of the rock mass and of the greater inclination of the
slope in Sector B3.

Attention has been focused on Points 3 and 5, located in Sector B3,

and on Points 13 and 21, located in Sector B1 (Fig. 7.16) resulting in
more detailed observation of the variation of distance
Page 197

between the theodolite position and some topographic benchmarks:

it is possible to notice the influence of daily rain on the displacement
velocity. The highest observed rates of movement were reached
during the heavy peaks of rain recorded in May, June and November
2002, all greater then 115 mm of water. Concerning this topic, for
example, with regard to Point 3, the displacement rate measured in 6
days, from 4 June to 10 June 2003, is similar to that measured over
153 days, from 11 June to 10 November 2003, but the velocity is
much different: 11.6 mm/day versus 0.43 mm/day. Thus, the rain
seems to cause a velocity increase of more than 1 order of
magnitude. With reference to the peak rain in November, it should be
noted that measurements were interrupted for about 40 days. No
indications can therefore be obtained concerning the velocity,
although it is possible to recognise an increase in displacement rates
in correspondence with this period.

In the upper part of the landslide, both displacements and direction

of movements are quite homogeneous. The general direction of
movement is southward, i.e. down the slope. This seems to indicate
that the upper part of the landslide tends to move as a single body
following the main foliation but, since the south-eastward movements
are hampered by the cliff which limits the landslide on its right flank,
the movements then evolve following the line obtained from the
intersection of the foliation with the boundary fault, Fb.
Figure 7.16. Variation of distance between the theodolite position
and some topographic benchmarks versus daily rainfall from
5/8/2001 to 11/5/2003.

Here, it is remarked that the entity of the displacements measured

through the monitoring system is an order of magnitude smaller than
the displacement values estimated in situ as a consequence of the
intense rainfall event of October 2000.
Page 198

Figure 7.17. The new sub-vertical joint developed inside Sector B3

after the intense rainfall event of October 2000. Displacement
measured from 5/8/2001 to 11/5/2003. Displacement vector: 1:40.

The professionals responsible for the maintenance of the monitoring

have noticed a new subvertical joint inside Sector B3 during their
latest inspections (Fig. 7.17). Focusing attention on that sector, the
displacement paths show two homogeneous areas with different
behaviour patterns: points located downhill from the joint move with
a dip direction that is close to that of the Ft system, whereas the
others, near Sector B2, follow a southward direction, which results
from the combination of the foliation and/or the ZT system with the
boundary fault Fb. The different behaviours can be linked to different
kinematics: sliding on the foliation in the upper part and toppling for
the lower and more overhanging rock mass.

7.3.4 Danger characterisation

The knowledge of the landslide’s morphological and structural
features combined with the displacement analysis indicated useful
directions for the modelling of the phenomenon which will be
discussed in the following sections.

7.3.4.1 Geometrical characterisation

The geometrical and structural constraints allow the estimation of the
thickness of the landslide body, with a maximum (40 m) near the
crown and minimum (20 m) near the western landslide boundary (Fb
fault). The rock volume affected has been estimated at 5×106 m3.

7.3.4.2 Geomorphological constraints

The geomorphological analyses allow the subdivision of the
landslide body into several sectors, bounded by morphostructural
features (see also § 7.3.4.3). The geometry of the sectors is to be
considered as a constraint for the modelling of the landslide body.

The following description of the sectors corresponds to Figure 7.5.

Page 199

Sector A: defines the upper boundaries of the landslide; the eastern

boundary of the landslide corresponds to stiff scarps, roughly
coincident with the K2 discontinuity system.

Sector B1: shows evidences of SW-ward movements; these

movements are hindered to the West by the rock wall corresponding
to the Fb fault on the right side of the landslide. The N-S Fb can thus
be considered as the western boundary of the landslide, and it can
be prolonged at least down to the lower edge of Sector B3.

Sector B2: suffers minor translations and it is less damaged than the
other sectors.

Sector B3: is more damaged than the others, since it is cut by the
closely spaced K1 and F systems that subdivide it into small rock
blocks. The greater inclination of the slope induces relatively strong
gravity instabilities, as shown by toppling of blocks.

Sector C1: conveys most of the debris as indicated by its

pronounced lobe at the base of the slope.

Sector C2: the western boundary corresponds to a NS elongated

ridge that could constrain the falling trajectories of the boulders to
the right.

7.3.4.3 Structural constraints

The discontinuities to be taken into account in a slope stability
analysis of Sector B3 are (Fig. 7.8):

ZT (boundary planes of shear zones sub-parallel to the schistosity).

These are characterised by a low internal friction angle (inferred to
be about 16°, see for ref. Hobbs et al., 1976) and by low persistence.
The role of rock bridges aligned parallel to the ZT surfaces can thus
be taken into account in modelling.
Fb (boundary fault); it is characterised by a cataclastic belt tens of
meters in length (breccias and polished surfaces with a low internal
friction angle, inferred to be about 16°).

F (boundary fault); it is characterised by fault zones arranged by very

steep anastomosed faults. Spacing and persistence of this system
are particularly evident at the plurimetric scale.

Ft joints are closely spaced and intensively affect the rocks, mainly
within the landslide. These joints, probably originated by unloading
processes (release joints, Engelder, 1985) show inclinations ranging
from few to 30 degrees.

K1 joints and maybe faults which are more closely spaced than the
other systems in Sector B3. Their mutual array, at different scales,
suggests that they could be segments of a NE striking fault zone.

K2 fractures and minor faults which are more persistent and

pervasive on the left side of the Sector, namely in the upper part.

The combination of the surveyed discontinuity systems leads to the

definition of several kinematic scenarios of sliding and toppling:

• the intersection of surfaces ZT and Fb can be used for the

definition of the sliding direction of wedges isolated from the main
rock body by means of tensile joints belonging to the F system;

• the intersection of surfaces Ft and ZT can be used for the definition

of the sliding direction of wedges isolated from the main rock body
by means of tensile joints belonging to the K1 system;

• toppling of blocks bounded by surfaces K1, K2 and Ft.

7.3.4.4 Comments on the mechanism and on displacements

Different kinematic behaviours have been observed in each sector of
the landslide. In particular, the upper part of Sector B1 seems to
move toward SSW, along sliding directions subparallel with the line
resulting by intersection of ZT and Fb (Fig. 7.8). The lower part of
Sector B1 is affected by SSW directed sliding on the western side
and SSE directed sliding in the eastern one. This is interpreted as
due to the different orientation of the slope, that respectively dips to
SSW and SSE. In this interpretation, the gravitational instability
should be extended only at shallow depth within the landslide.

In the lower part of the landslide (Sector B3), both sliding and
toppling have been found. Sliding could occur toward SSW (Fig.
7.15), while toppling could be originated by the partial reactivation of
pre-existing vertical fractures (K1 and K2) and cracking of the rock
bridges along the Ft system.
Page 200

In Figure 7.18, a schematic cross-section of the Ceppo Morelli

landslide is shown, where the main structural elements and their
cross cutting relations are outlined. The highlighted sliding surface in
the upper part of the landslide (sliding zone in Figure 7.18) has been
inferred on the basis of geomorphological data. The occurrence of
an effective failure surface can be assumed in modelling the entire
landslide as a whole. In this case, this surface would form by partial
reactivation of pre-existing fractures or faults, and partially by
rupturing of the rock bridges eventually placed between the different
discontinuities. Orientation, shape and mechanical properties of rock
blocks can be extrapolated by combining the main discontinuity
systems described above.

It is here remarked that the kinematics analysis and modelling of the

landslide could require a basal sliding surface that should not be
thought as a continue sliding surface. It can be rather imagined as a
composed surface that might be formed in reliability terms.

7.3.4.5 Scenarios
On the basis of the previous considerations, three different hazard
scenarios are thus possible, with decreasing occurrence probability
and increasing impact on land planning:

• Scenario 1: toppling and sliding of rock boulders and debris falls

(features 9 and 10 in Figure 7.5). As pointed out by the historical
research, the site has been involved in repeated falls since
Figure 7.18. Schematic cross section of Ceppo Morelli landslide
(overall trace in Figure 7.5). The thicker lines in the sliding zone
indicate discontinuities which are generally open.
Page 201

ancient times. The sites from which rock blocks are released are
located all along the lower boundary of the unstable Sector B3.
Rockfalls range from small cobbles to large boulders hundreds of
cubic metres in size.

• Scenario 2: collapse of Sector B3 (Fig. 7.5). Continuous rockfalls

can weaken the rock mass located in this sector. An estimated
avalanche involving a rock mass of about 1’000’000 m3 could then
occur.

• Scenario 3: collapse of the whole Sector B (Fig. 7.5). The collapse

of Sector B3 could lead to a huge rock avalanche involving the entire
Sector B, which corresponds to a volume of about 5’000’000 m3.

7.3.5 Geomechanical modelling

7.3.5.1 Mechanical “Triggering” model
The Ceppo Morelli site has been studied by means of two
geomechanical models: the first is a 2D discontinuous model, the
second a 3D equivalent continuous model with discontinuity planes
that cut off the unstable mass from the whole rock slope.

The 2D analyses were carried out using the Finite Element Method
(code DRAC, Prat et al., 1993). The approach applied during the
numerical modelling consisted of two stages: a first stage during
which the calculation of the initial stress conditions of the slope was
carried out using topographical information (the actual morphology,
where possible) and incorporating the effect of the glacier weight; a
second stage in which the glacier was removed and the stress field
computed. The objectives of the simulations were twofold. First, the
general stress field of the slope was calculated in order to obtain
information on the failure mechanism. The interaction of the upper
part (B1) that consists of a rather well-structured rock mass and the
lower part (B3) that is characterised by a disjointed rock body, was
an additional aim. Second, the influence of different material
properties and mechanical models for the cracks (interfaces in the
discontinuous model) were analysed.

The 3D analyses were carried out by means of the code Map3D

(described in Chapter 3) based on an indirect Boundary Element
technique, the Displacement Discontinuity Method. Attention was
focused on the lower part of the unstable area (Sector B3), indicated
by the geological and geomorphological studies as the key volume,
in the sense that the failure of B3 could trigger that of the whole
landslide.

As previously seen, a complex description both from the

geomechanical and the kinematical point of view was arisen from the
geological investigations. At the present phase of the work, however,
simplified geomechanical model and numerical triggering simulation
are assumed. Remarkable characteristics of these models are the
3D analysis conditions and the possibility of taking into account the
measured and the simulated displacements. In order to be able to
focus the attention on these aspects, some simplifications had to be
introduced from the geometrical and the mechanical point of view:

• as the rock mass of B3 is very fractured and disjointed, in the 3D

geometrical model it was represented as an equivalent continuum
and the discontinuity planes that belong to the principal joint sets
were introduced as boundaries of the unstable volume;

• the global behaviour of the unstable rock volume was analysed:

this implies that it was possible only to reproduce the mean value of
the measured displacements due to sliding. The numerically
obtained value had to be referred to the whole investigated rock
mass. It must be underlined that the considered mean value was
evaluated with respect to time (i.e. during one month) and space (i.e.
over Sector B3);

• the average displacements were supposed to be due to the overall

creep behaviour of the rock mass, which is assumed to be
concentrated only on the sliding surface. No influence on the
displacements of water pressure variations with respect to time was
considered due to the lack of piezometric data.

On the basis of these considerations, the aim of the analyses was to

reproduce the displacements measured in situ by the monitoring
system: it is possible to evaluate the viscous plastic creep
Page 202

behaviour of the rock mass by means of the Bingham flow rule

implemented in the code Map3D. This can be carried out in two
steps: in the first one, the research of the mechanical parameters
related to conditions of unstable equilibrium of the rock mass is
performed by means of numerical simulations; on the basis of the
results obtained, in the second step, the analysis of the behaviour of
the unstable rock mass with respect to time is developed. In order to
do this, a realistic value of the viscous modulus G was checked by
comparing the evaluated displacements with those measured in situ.

The 2D model was developed along a N-S cross-section of the entire

slope. The topography of the model was defined using a DEM with a
cell size of 10 m. A two-stage approach was applied during the
modelling. In the first stage, the upper limit of the glacier was set at
1500 m a.s.l.; in the second stage, the glacier was removed. The
mesh consisted of 1780 continuum elements and two discontinuity
sets: the main schistosity, assumed as ZT, was introduced by
southward dipping joints with a spacing of 10 m; the fracture system
K2 was represented by sub-vertical joints with a spacing of 20 m.
These discontinuity sets were modelled with zero-thickness interface
elements. The constitutive law adopted was linear elastic for the
continuum material as well as for the interfaces. The values selected
for the material properties of the rock mass and the interfaces are
given in Tables 7.1 and 7.2, respectively.

Table 7.1. Material properties adopted for the 2D model continuum

elements.

Unit weight γ Young’s Modulus E Poisson’s Ratio

(kN/m3) (GPa) υ

Gneiss 28 50 0.3

Glacier 9.2 9.6 0.33

ice

Table 7.2. Material properties adopted for the 2D model interface

elements.

Normal stiffness Kn Tangential stiffness Ks Cohesion c Friction

(MPa/m) (MPa/m) (kPa) angle φ

1000 100 0 30°

The opening of the interfaces due to the removal of the glacier is

presented in Figure 7.19. During the simulations, it was noticed that
the role of the continuous elements was rather limited. That is,
deformation and displacement are controlled by the interfaces.
Therefore, the joint constitutive parameters and the joint geometry
became the key aspects of the analyses.

The 3D geometrical model of Sector B3 of the Ceppo Morelli

landslide was carried out on the basis of a simplified DEM (cell size
170 m) of the topography in order to reduce the computational time.
The boundary discontinuity planes were modelled as follows (Fig.
7.8): the upper tensile joint was simulated by a plane representing
the F system; the left side boundary was given by two planes
belonging to the K1 and Fb joint sets, respectively the discontinuity
between B2 and B3 and the boundary fault; the lower boundary was
represented by the outcrop of the sliding surface assumed as
belonging to the ZT system. In Figure 7.20, the 3D representation of
the unstable volume is shown.

The first aim of the 3D simulations was to obtain a realistic value of

the in situ friction angle. This value is related to a heterogeneous
material made of boulders, soil, water and voids considered as a
whole; therefore, it is impossible to evaluate the friction angle by
means of laboratory tests. A set of sensitivity analyses was
performed in order to calculate the value of friction angle that
corresponds to the instability of the rock mass. These analyses were
carried out with no creep simulation.
Page 203

Figure 7.19. Opening of the interfaces (red lines) at the final stage of
the Ceppo Morelli model. Width of the red lines is proportional to the
opening displacement of interfaces.
Figure 7.20. The unstable volume considered in the 3D numerical
analyses (Sector B3).
Page 204

The following step was the assessment of the value of the creep
coefficient C. Activating the Map3D creep option, a simulation was
made with an assumed value of C and the results were compared
with the measured displacements. If these values were not close to
each other, more simulations were carried out in order to find a
proper C. The final set of mechanical parameters used for the
numerical analyses is shown in Table 7.3.

Table 7.3. Mechanical parameters used for the 3D model of the

Ceppo Morelli landslide.

Equivalent continuum

Unit weight of volume γ 0.027MN/m3

Young’s modulus E 7000MPa

Poisson’s ratio υ 0.25–

Failure surface

Friction angle φ 25°

Normal modulus Kn 5000MPa

Shear modulus Ks 1500MPa

Creep coefficient C 100MPa×Month

The computed displacement due to one month’s creep can be

compared in magnitude and direction with the mean value of the
displacements obtained by means of topographic measurements.
Taking into account points 1, 1bis, 2, 5, 10, 10bis and 12 (Fig. 7.15),
the only ones in Sector B3 referred to a sliding kinematics, the
measured mean value is about 4 mm/month with a dip direction of
182°; these values are very close to that obtained numerically, which
are equal to 3.8 mm/month with 180° of dip direction.

The obtained results are very far from giving the probability of failure
needed for the risk analysis, which is to be obtained through a
probabilistic procedure. For this purpose however the variation of
pore pressure in time and the decay of rock strength parameters in
time should be known. These data are seldom known and are not
known in the case of the Ceppo Morelli landslide.

7.3.5.2 Mechanical “Run Out” model

The morphostructural and kinematical characteristics of the Ceppo
Morelli instability phenomena have allowed the recognition of three
possible scenarios (see § 7.3.4.5). The run-out phase for the three
scenarios has been simulated as follows:

• Scenario 1 (rock fall). The trajectories of the falling rock blocks

have been simulated through a 3D lumped mass method
(ROTOMAP code, Geo & Soft Int., see § 3.5.5.3). Values of the
restitution coefficients in the normal and tangential direction equal to
0.32 and 0.80 respectively, and a friction coefficient of 0.65 have
been assumed. The choice has been done on the basis of data
obtained from the literature and through back analyses of the
trajectories of already fallen rock blocks.

• Scenarios 2 (rock avalanche of about 1×106 m3) and 3 (rock

avalanche of about 5×106 m3). The rock avalanche run-out has
been simulated through the coupled use of ROTOMAP and DAN
(Hungr, 1995) codes. A frictional reology, with friction angle equal to
30° and pore pressure coefficient equal to 0 (i.e. no water) has been
assumed in the analyses carried out through DAN code.

Scenario 1

The aim of the analysis is to reproduce the rock falls occurred in the
past and to forecast new rock falls affecting a larger area.
To get realistic results the assumed range of velocity for the released
blocks changes if they originate from Line 1 (Fig. 7.21) or from the
Line 2 (Fig. 7.22) of the considered unstable area as indicated in
Table 7.4.
Page 205

Table 7.4. Range velocity for the released blocks.

Line Vmin [m/s] Vmax [m/s]

1 1.0 1.8

2 0.6 1.0

These different ranges of velocity are justified by the possibility that

some of the blocks crossing Line 1 can originate in the upper part of
the unstable area and than gain a higher velocity.

In addition to rock fall trajectories (Fig. 7.21 and 7.22), the analyses
allow the distribution of the kinetic energy of the blocks (Fig. 7.23
and 7.24): the slope can then be subdivided in homogeneous
sectors, each associated to a specific value of the medium energy.
The distribution obtained in terms of energy represents an important
tool to be taken into account for risk and vulnerability evaluations.

Due to the fact that ROTOMAP is a “lumped mass” code, the energy
values are formulated as E=v2/2 [J/kg]. Hence, in order to link the
obtained results to a possible real situation, an indicative mass
should be hypothesised for the boulders. The maximum mass of the
boulders observed at the bottom of the valley (around 2.7×106 kg,
corresponding to a volume of 1000 m3) seems a reasonable value.
Figure 7.21. Scenario 1: rock fall trajectories from Line 1 (in red:
blocks in flight).

Figure 7.22. Scenario 1: rock fall trajectories from Line 2 (in red:
blocks in flight).
Page 206

Figure 7.23. Scenario 1: kinetic energy distribution (Line 1).

Page 207

Figure 7.24. Scenario 1: kinetic energy distribution (Line 2).

Page 208

Scenario 2

On the basis of geomorphological and geostructural considerations

(see § 7.3.4.5), Scenario 2 is considered the most prone to evolve as
a catastrophic rock avalanche.

The direction of propagation chosen for the run-out analysis is

identical to the dip direction of the joint system Ft (Fig. 7.25).

Together with the run-out area (Fig. 7.26), the value of the avalanche
front velocity is defined along the section taken into account (Fig.
7.27); the front velocity distribution is then supposed to be the same
for all the points located at the same height.

As the energy distribution for the rock fall case, the homogeneous
areas obtained in terms of velocity distribution are an important
aspect, that should contribute to the risk and vulnerability evaluation.

Figure 7.25. Section in the direction of propagation in case of

Scenario 2.
Page 209

Figure 7.26. Run out area in case of Scenario 2.

Page 210

Figure 7.27. Velocity distribution (m/s) in case of Scenario 2.

Scenario 3

This is the most catastrophic hypothesis, considering the possibility

that the whole volume is involved in a rock avalanche phenomenon.
Although extremely unlikely, this scenario has been considered since
it would generate particularly catastrophic consequences for the
hamlets, the river and the road that are located in the bottom of the
valley. The run-out area has been obtained as described above for
Scenario 2 (Fig. 7.28). In the same way as described above, the
front velocities can be obtained.

Conclusions

The results obtained applying this methodology represent a useful

contribution to vulnerability and risk evaluations since the rock fall
analyses give the possibility of calculating the energy
Page 211

Figure 7.28. Run out area in case of Scenario 3.

distribution per unit mass, and the avalanche analyses give both the
definition of the run-out areas and the front velocity distribution.

In particular, the simulations allow the determination of different

hazard situations for the hamlets located at the bottom of the valley.
The Prequartera village can be involved in a rock fall phenomenon
(scenario 1) and in the most catastrophic rock avalanche (scenario
3), whereas the expansion of the mobilized mass in both the rock
avalanche cases (scenarios 2 and 3) could be the cause of a danger
condition for the Campioli area.

Finally, an evolution in terms of rock avalanche (scenarios 2 and 3)

would determine the road interruption and the river damming with
disastrous consequences.

7.3.6 Occurrence probability

As previously discussed (§ 3.1.5.6), the probability of occurrence can
be defined as the chance or probability that a landslide will occur. It
can be expressed in relative (qualitative) or probabilistic
(quantitative) terms.
Page 212

In the case of large landslides, a rigorous, quantitative, procedure

should include the application of formal mechanical-probabilistic
methods, taking into account the uncertainty in all the geometrical
and mechanical parameters (slope geometry, shear strength,
piezometric pressure). In the case of the Ceppo Morelli landslide,
however, a similar procedure cannot be applied because the
uncertainty in the definition of the parameters involved is too large
and it is not possible to define any reliable variability for them.

An alternative procedure takes into account the relationship between

the intensity of triggering events such as rainfalls or earthquakes and
the landslide occurrence through the use of empirical correlations or
neural network methods. Again, the lack of data makes this
approach inapplicable.

Since several historical records exist of the rockfall activity in the

area involving different volumes, the historical approach seems
much more suitable to the case of the Ceppo Morelli landslide: in this
way it is possible to obtain some information related to periodic
frequency.

An occurrence range of time (see also Corominas et al., 2002) was

obtained for each of the three scenarios of evolution described in §
7.3.4.5, considering the minimum and maximum time interval
between two consecutive recorded events. On the basis of their
mean value, a frequency was then calculated in terms of events per
year (Table 7.5).

Table 7.5. Definition of the occurrence probability using the historical

approach.

Scenario Recorded Occurrence range Average Frequency

events (years) (years) (event/year)
1 1940 0–30 151/15=0,066

Oct. 1971

Apr. 1977

Oct. 2000

Jun. 2002

2 (Intermediate) 30–200 1151/115=0,009

3 312 200–1000 6001/600=0,0016

843

1816

2000

It should be noted that the events recorded are related to two

different types of danger: the fall of single blocks and catastrophic
events involving a larger mass. The first has been associated to a
rockfall “Scenario 1” (low energy phenomena and high frequency)
and, conservatively, the second has been considered as
representative of phenomena involving the whole mass (rock
avalanche “Scenario 3”: high energy phenomena and low
frequency). The intermediate occurrence range has finally been
associated to the rock avalanche “Scenario 2” (medium energy
phenomena and frequency).

7.4 QUANTITATIVE RISK ANALYSIS

Each of the described scenarios is characterised by its own intensity,

failure and run-out areas (derived from the geomorphological and
geomechanical models) and by an occurrence probability (as a result
of an historical analysis of recorded events).
According to the proposed quantitative risk analysis (see Chapter 3),
each element of the hazard scenario is used separately in the matrix
calculus of risk evaluation. This process is applied on each
recognised scenario; therefore, for the Ceppo Morelli site, 3 separate
risk analyses were carried out.

At the end of the matrix calculation, a unique value is obtained (total

risk), divided into 4 categories: physical, economic, environmental
and social, useful for the total risk assessment phase.
Page 213

7.4.1 Elements at risk

The elements at risk (Fig. 7.29) are recognised through the definition
of the area affected for each danger in the scenarios previously
specified. These elements at risk are described in Tables 7.6, 7.7
and 7.8.

Figure 7.29. Elements at risk.

Table 7.6. Definition of the elements at risk: Scenario 1 (Rockfall).

Elements at risk Name Notes

Residential areas [valuable buildings or Campioli Area

valuable rural centres: historical, architectural, affected:
artistic and/or cultural value] 3400 m2

Number of
buildings
affected: 5
 Persons
affected: 12

Area
affected:
6300 m2

Prequartera Number of
buildings
affected: 18

Persons
affected: 42

Forests (private and public properties) Area

affected:
228000 m2

Heavy traffic or strategic roads National Length

road No. 549 affected: 800
m

Annual
traffic:
100000
vehicles

Secondary roads Mondelli Length

municipal affected: 230
road m

Annual
traffic: 1000
vehicles
Lifelines Power line Length
Valle affected:
Anzasca 15 1025 m
kV
Pylons
affected: 4

7.4.2 Vulnerability of the elements at risk

To define the vulnerability of the elements at risk, an analysis
concerning typology and process intensity of the concerned assets
was carried out. Geomechanical models give important data about
the energy of impact of rock blocks, while structural and architectural
analyses of the elements at risk give data about strength at impact.
Page 214

Table 7.7. Definition of the elements at risk: Scenario 2 (Rock

avalanche 1×106 m3).

Elements at risk Name Notes

Residential areas [valuable buildings or valuable Campioli Area

rural centres: historical, architectural, artistic affected:
and/or cultural value] 2300 m2

Number of
buildings
affected: 3

Persons
affected: 3

Forests (private and public properties) Area

affected:
248500 m2

Heavy traffic or strategic roads National Length

road No. affected: 543
549 m

Annual
traffic:
100000
vehicles

Secondary roads Mondelli Length

municipal affected: 400
road m

Annual
traffic: 1000
 vehicles

Lifelines Power line Length

Valle affected: 527
Anzasca m

15 kV Pylons
affected: 1

Table 7.8. Definition of the elements at risk: Scenario 3 (Rock

avalanche 5×106m3).

Elements at risk Name Notes

Residential areas [valuable buildings or valuable Campioli Area

rural centres: historical, architectural, artistic affected:
and/or cultural value] 2500 m2

Number of
buildings
affected: 3

Persons
affected: 3

Area
affected:
6300 m2

PrequarteraNumber of
buildings
affected: 18

Persons
affected: 42

Forests (private and public properties) Area

 affected:
406280 m2

Heavy traffic or strategic roads National Length

road No. affected: 810
549 m

Annual
traffic:
100000
vehicles

Secondary roads Mondelli Length

municipal affected: 400
road m

Annual
traffic: 1000
vehicles

Lifelines Power line Length

Valle affected: 722
Anzasca m

15 kV Pylons
affected: 3

Starting from energy data, different considerations concerning

damages and losses of the elements at risk were obtained for the 3
scenarios for the Ceppo Morelli site, as impacts are strongly linked to
rock volumes and distribution.

Concerning the rockfall scenario, the run-out model yields an energy

zonation (J/kg) per unit mass, according to rockfall velocity.
Considering a mass of 2.7×106 kg (maximum mass of the boulders
observed at the bottom of the valley, see § 7.3.5.2), impact energy
can be evaluated and a structural analysis supplies different values
for physical, economic, environmental and social vulnerability of the
elements at risk, through the indexes (0.5, 0.75, 1) described in
Table 3.8 (§ 3.6.3.5).

For instance, buildings, mostly old, do not have a particular strength

at impact, as they are made of stones and wood, without concrete.
Blocks impacting on roads usually damage the wearing course as
happened during the 1978 and 2000 events, interrupting road use
(see Fig. 7.30).

Regarding human lives, vulnerability depends on where people are

(inside or outside the buildings): in this case, the worst situation was
considered, i.e. all persons without adequate protection.

Concerning rock avalanche scenarios (both 1 and 5 million m3) and

the different energy degrees on the basis of the landslide velocity
(measured in m/s in the run-out geomechanical model), it is
Page 215

Figure 7.30. Few damages next to a house in Campioli village

(Ceppo Morelli Municipality). This block of about 4 m3 rolled towards
left, slightly damaging the road without reaching the house.

useless to define the different elements at risk by vulnerability

degrees. In the rock avalanche area (failure, depletion and
accumulation areas, see Fig. 7.26 and 7.28) the destruction would
be total. Also along the road at the bottom of the valley, where the
velocity tends to zero (see, for example, Fig. 7.27), the accumulation
(some tens of meters) is so great that all the assets would be
completely destroyed, heavily damaged or buried and, therefore,
permanently useless. Following these considerations, a vulnerability
degree equal to 1 is used for all the vulnerability categories.

7.4.3 Expected impact

Multiplying the values of the elements at risk (VE, see § 3.1.6.2), and
of the vulnerability percentage (V), the expected impact (C) is
obtained by C=VE×V

In the quantitative risk analysis for the Ceppo Morelli rock slide,
detailed economic evaluations were not considered for elements at
risk and arbitrary relative value indices have been temporarily
applied for each category (Table 7.9).

Concerning rockfall danger (Scenario 1), 3 different expected impact

values were calculated, as 3 vulnerability degrees (0.5, 0.75, 1) had
been previously differentiated on the basis of the energetic impact
thresholds of the block falls.

All the considerations were made through the automatic application

of a G.I.S. technique, whose graphics are not included in this
section.

Table 7.9. Considered values of the elements at risk.

Element Physical Economic Environm.Social

cons. cons. cons. cons.

Residential areas (valuable buildings 4 1 1 4

or valuable rural centres)

Heavy traffic or strategic roads 3 4 1 3

Secondary roads 1 2 1 2

Secondary lifelines 1 2 1 0

Forests/Rural area (private and public 1 1 3 1

properties)
Page 216

7.4.4 Risk assessment

The last phase of the quantitative risk analysis concerns the risk
evaluation. This evaluation is obtained multiplying the expected
impact values (C) and the numeric values of the occurrence
probability (P) related to each scenario (Table 7.5): R=C×R

Figures 7.31–7.33 show the physical risk assessment results as a

G.I.S. layout for the 3 scenarios (rockfall: P=1/15; rock avalanche,
1×106 m3: P=1/115; rock avalanche, 5×106 m3: P=1/600). In the
same way, Figures 7.34–7.36 show the economic risk assessment
results, Figures 7.37–7.39 show the environmental risk assessment
results and Figures 7.40–7.42 show the social risk assessment
results for the same 3 scenarios.

Figure 7.31. Map obtained by G.I.S. techniques concerning the

physical risk for Scenario 1 (rockfall). The areas with elevated risk
values represent the National Road No. 549 and the uphill part of the
Prequartera village, where a church is located.
Figure 7.32. Map obtained by G.I.S. techniques concerning the
physical risk for Scenario 2 (1 million m3 rock avalanche). The areas
with elevated risk values are the National Road No. 549 and part of
the Campioli village involved in the rock avalanche process.
Page 217

Figure 7.33. Map obtained by G.I.S. techniques concerning the

physical risk for Scenario 3 (5 million m3 rock avalanche). The areas
with elevated values are the National road No. 549, part of the
Campioli village and the entire Prequartera village.

Figure 7.34. Map obtained by G.I.S. techniques concerning the

economic risk for Scenario 1 (rockfall). The main areas with elevated
risk values represent communications and services. The National
Road of Valle Anzasca is highlighted, being the only important road
of the valley from an economic point of view.

Figure 7.35. Map obtained by G.I.S. techniques concerning the

economic risk for Scenario 2 (1 million m3 rock avalanche). The
main areas with elevated risk values are communications, services
and the main Valle Anzasca River (the Anza stream). The
interruption of the river due to the rock avalanche accumulation
produces important economic consequences, as the river feeds the
Prequartera power plant reservoir.
Page 218

Figure 7.36. Map obtained by G.I.S. techniques concerning the

economic risk for Scenario 3 (5 million m3 rock avalanche). The
main areas with elevated risk values are communications (0.8 km. of
the National Road No. 549), services (more than 700 m. of power
lines) and the main river of the valley (the Anza stream). As
previously stated, the Anza stream is economically essential to this
valley, feeding a power plant reservoir also involved in the rock
avalanche phenomenon.
Figure 7.37. Map obtained by G.I.S. techniques concerning the
environmental risk for Scenario 1 (rockfall). The main areas with
elevated risk values represent the wooded areas near the
Prequartera village and part of the village itself (it is a medieval
village of great environmental and cultural value).

Figure 7.38. Map obtained by G.I.S. techniques concerning the

environmental risk for Scenario 2 (1 million m3 rock avalanche). The
main areas with elevated risk values are the wooded areas and parts
of the Anza stream involved in the gravitative phenomenon
evolution.
Page 219

Figure 7.39. Map obtained by G.I.S. techniques concerning the

environmental risk for Scenario 3 (5 million m3 rock avalanche). The
main areas with elevated risk values are the wooded areas and a
part of the Anza stream involved in the rock slide evolution.

Figure 7.40. Map obtained by G.I.S. techniques concerning the

social risk for Scenario 1 (rockfall). The areas with elevated values
correspond to the villages of Campioli and Prequartera. In the last
one, it is possible to distinguish the risk value according to the
impact energy of different blocks.

Figure 7.41. Map obtained by G.I.S. techniques concerning the

social risk for Scenario 2 (1 million m3 rock avalanche). The areas
with elevated risk values are situated in the Campioli village. The
National Road No. 549 of Valle Anzasca is also involved in the rock
avalanche phenomenon and, therefore, presents elevated risk
assessment values, due to the process type (characterised by a
rapid evolution), to the large part of the involved road (0.55 km) and
to the traffic of this road at certain periods of the year.
Page 220

Figure 7.42. Map obtained by G.I.S. techniques concerning the

social risk for Scenario 3 (5 million m3 rock avalanche). The areas
with elevated values are the villages of Campioli and Prequartera;
the last one would be completely involved and destroyed. About 1
km of the National Road No. 549 is also affected by the rock
avalanche phenomenon, being a heavy traffic road in the tourist
season.

7.4.4.1 Total risk

To obtain a risk assessment synthesis for the Ceppo Morelli site,
total risk was calculated as described in § 3.1.7. The result is shown
in Figure 7.43, showing the risk value sum for each
Figure 7.43. Maps (obtained by GIS technique) about total risk
assessment, divided into 4 risk categories.
Page 221

scenario. Again, total risk is divided into 4 risk categories: physical,

economic, environmental and social. In these maps, the comparison
of the risk among the different areas is simplified.

The areas affected by high intensity and high occurrence probability

processes present the most elevated risk values. At the Ceppo
Morelli site, the area affected by rockfall shows the highest risk
values, as the occurrence probability of this scenario is more than 40
times higher than the occurrence probability of Scenario 2 and about
7 times higher than the occurrence probability of Scenario 3.

7.4.5 Indirect risk and regional impacts

In order to complete the risk analysis, some considerations
concerning the indirect effects associated with the rock slide
evolution (not considered by the geomechanical models) are
nonetheless necessary. The landslide scenarios of Ceppo Morelli
show, as well as rockfalls, some phenomena involving huge rock
masses (millions of m3) at high velocity (40 m/s corresponding to
about 150km/h), whose effects are not easily estimated.

In particular, the great mass movement (5 millions m3) that could

occur at Ceppo Morelli according to the most catastrophic scenario
would produce a huge wind effect. The process intensity cannot be
easily estimated but a qualitative evaluation can be made on the
basis of rare case analyses described in the literature (e.g. Hsü,
1978). It is probable that the phenomenon would be light laterally
with respect to the mass movement and very strong frontally. The
same would occur for the dust that follows the wind effect.

At the moment, wind effect and dust cannot be input in a numerical

model without mistakes. Nevertheless, it can be stated that the
damages produced by these indirect effects are of small extent, as a
wooded slope without buildings stands in front of the mass
movement. Some damages could occur to Campioli and Prequartera
buildings not directly involved in the mass movement.

Another effect to be considered in risk assessment is the formation

of a landslide dam lake (Fig. 7.44). This phenomenon can occur
during a meteorological event with important flows along the Anza
stream (flow of about 900 m3/s, see Bossalini & Cattin, 2002). These
processes together (instantaneous damming of the valley-road by
tens of meters of landslide accumulation and extraordinary flood of
the Anza stream) would form a basin of about 2 million m3 in few
hours. Many buildings in the Campioli village would be inundated
and many people would have to be evacuated, increasing the social
risk (see, for example, the case of Randa (CH), shown in Figure
7.45).

Moreover, due to this lake formation, downstream villages could

potentially be involved. Instantaneous lake depletion can, in fact, be
caused by a rapid overtopping dam erosion, or a landslide dam
collapse caused by piping and internal erosion.

In both cases, flash-flood along the valley would cause enormous

problems. In Ceppo Morelli, a study has been carried out to detail a
civil protection plan in case of a flash-flood event (Bossalini & Cattin,
2002). Trying to delineate the areas affected by flash-flood, a
simplified approach has been used in this study, similar to the rock
slide risk assessment previously detailed. The elements at risk (land
use, human lives, buildings and civil engineering structures) have
been identified along the entire valley bottom (Fig. 7.46 and 7.47). In
this risk analysis, physical,
Figure 7.44. Mechanism and consequences of lake damming due to
a huge landslide. (source http://www.kingston.ac.uk)
Page 222

Figure 7.45. Flood effect in alpine area due to landslide dam (Randa,
CH, 9th May 1991). Effects also on buildings 1 km far from landslide
site. (source: www.crealp.ch).
Figure 7.46. Areas that might be affected by a flash flood due to
rockslide dam breaking (blue colored). The zones in orange squares
are detailed in Figure 7.47.
Page 223

economic, environmental and social vulnerability was not evaluated

separately due to a very large potentially affected area.

To complete the description of indirect effects associated with the

Ceppo Morelli rock slide, it is important to recall that a so-called
“Vajont effect”: waves produced by the rockfall in the reservoir can
get over the dam located near Prequartera causing flooding along
the Anza stream. Nevertheless, as the water volume in this reservoir
is only some hundred of m3, strong consequences at the bottom
dam are not foreseen.

In Figure 7.48, a map of the main direct and indirect risks

considering the worst catastrophic rock avalanche scenario is
shown. This is obtained through the qualitative evaluation described
in this section.
Figure 7.47. Risk maps of zones in orange squares in Fig. 7.46.
Green=moderate risk, yellow=medium risk, orange=high risk,
red=very high risk (Bossalini & Cattin, 2002).
Page 224

Figure 7.48. Map of direct and indirect effects associated to rock

avalanche phenomenon. Borderlines are drawn by qualitative
evaluation.

7.5 REGULATION AND RISK MITIGATION MEASURES

ALREADY AVAILABLE

7.5.1 Viability
After the many rockfalls of October 2000, the risk conditions of the
National Road were considered too high to be accepted. Since the
road is the only possible connection for Macugnaga, a well-known
ski resort, a provisional by-pass road was constructed along the right
banks of the Anza River, in order to avoid the most dangerous zone.
The road is about 1 km long and construction took about 30 days;
the total cost was on the order of 1.25 M€. The river is crossed twice
by means of Bayley-type bridges provided by army engineers.
Before construction, the Regional Geologists made a running
evaluation of the risk condition of the provisional road. Results
showed that the road is protected from rockfalls but may be affected
by major mass falls (e.g. a rock avalanche). A rock avalanche would
obviously destroy the provisional road.

Opening of the road was thus made conditional on the results of the
monitoring system. If some defined threshold values (both for rainfall
and displacements) are exceeded, the road is blocked by local
authorities. Up until now this has happened only once, during the
heavy rains of November 2002.

In order to provide a final solution to road practicability problems, a

tunnel is currently being studied. The tunnel, about 800 m long, will
totally avoid the rockslide and will allow traffic flow to Macugnaga
even in the case of large failure scenarios. The total cost will be
about 38 M€.

7.5.2 Endangered buildings

After the October 2000 flood, the mayor evacuated two small
villages, Campioli and Prequartera, since a 70 m3 boulder stopped a
few meters from a house of in Campioli and a 10 m3 boulder
Page 225

almost reached Prequartera. An analysis of possible remedial

measures was carried out by the regional geologists and by the
Politecnico of Turin, as follows.

Concerning Prequartera, which has about twenty houses, the

analysis indicated that, due to its position and to the general features
of the landslide, the small village may be affected by rockfalls but is
out of the main area of influence of mass falls. The small village has
been protected by means of a reinforced earth rockfall protection
wall.

The wall is about 120 m and 5 m high; average base width is about 8
m. The wall was designed in order to withstand impact energies up
to 8 MJ; the total cost was about 0.25 M€.

As for Campioli, the situation is much more critical, for it is just on the
edge of the zone of possible invasion in the case of failure of the
lower part of the landslide body. The evacuated part of the small
village consists of seven houses. One of them, built in the seventies,
may also be affected in the case of common rockfalls; in fact, a 70
m3 boulder stopped a few meters from the building in October 2000.
The house will be demolished and relocated elsewhere.

The remaining six buildings pose more problems, in that they form a
historic nucleus built around 1650, a marvellous and perfectly
preserved sample of typical Walser architecture, protected by the
Sovraintendendenza ai Beni Architettonici (National agency for the
protection of the Architectural heritage).

The conditions of risk related to this nucleus are too high to be

accepted for residential use but acceptable for discontinuous use, so
the more likely solution will be:

– purchase, by the local town authority, of the houses;

– construction of a reinforced earth rockfall protection wall, to offer
partial protection.

7.5.3 Alarm system

The monitoring system described in § 7.3.3 is connected to an alarm
procedure.

The data are examined on a regular basis by a group of

professionals: the level of attention depends on the warning codes
emitted daily by the Sala situazione rischi naturali della Regione
Piemonte (Regional natural risks monitoring office). If any threshold
values (relative to both rain and displacement rates) are exceeded,
an alarm system is triggered in order to close the main road to traffic
and to activate a proper civil protection plan. The plan also includes
the possible effects of river damming.
Page 226

This page intentionally left blank.

Page 227

8
The Sedrun landslide

Ch.Bonnard & X.Dewarrat

Ecole Polytechnique Fédérale de Lausanne

F.Noverraz
Consulting geologist, Lausanne, Switzerland

8.1 INTRODUCTION

8.1.1 Landslide description

The slope movement phenomenon named the Sedrun Landslide is
located in eastern Switzerland, namely in the western part of the
Canton of Graubünden, in the higher stretch of the Vorderrhein
valley. It takes place in the rock slope called Cuolm da Vi overlooking
the villages of Sedrun and Camischolas which form part of the
community of Tujetsch. The unstable area extends between the Val
Strem valley to the west, which deeply intersects the mountain range
limiting the northern slope of the Rhine valley, and the temporary
Drun Tobel creek to the east (Fig. 8.1).
Figure 8.1. Layout of the Sedrun Landslide and its main
characteristics (map published with the authorization of the Swiss
Office of Topography, October 2001).
Page 228

The instability in this mountainous and austere area, covered by very

scarce vegetation and completely without buildings, is mainly the
result of a toppling effect concerning the whole Cuolm da Vi zone.
Due to sub-vertical schistosity and to several fault families, this
phenomenon has produced a thick bending in the rock layers, quite
visible from Val Strem, in particular in the upper part of this rock
mass. This process could trigger rockslides, rockfall or local slides in
some parts of the rock mass. Such sliding appears to be more likely
in the top of the area, in particular at the highest point of the zone
called Cuolm Parlets Dadens, but also on the right side of Drun
Tobel creek, in the zone called Plauncas Tgamos.

The unstable area, covering 1.5 km2, can be quite well determined
on the basis of its morphology, the presence of major cracks more or
less opened by the movements in the upper part and due to
numerous monitoring data (see § 8.3.3). As no boreholes have been
carried out, its volume can only be roughly assessed at some 100
million m3. The important movements recorded cause the formation
of typical uphill facing scarps in the upper part of Cuolm da Vi (see
Fig. 1.3).

Some major infrastructures are located in the region of Sedrun, such

as a construction site for the new Gotthard Tunnel (shaft for
intermediate access), the Oberalp cantonal road, connecting the
Cantons of Graubünden and Uri, and a regional railway line
connecting Zermatt to St. Moritz (Glacier Express). Moreover, the
village of Sedrun has been developed extensively over the last few
decades, especially as a winter sport resort.

This landslide site has been studied particularly between 1993 and
1997 as a part of a large research project funded by the Swiss
National Science Foundation, within the PNR 31 national project on
climate change and natural disasters (VERSINCLIM project—
Noverraz et al., 1998; Bonnard et al, 1996).
8.1.2 Historical background
The temporal origin of the movements observed in the Cuolm da Vi
unstable zone is not easy to determine. Due to the late installation of
the Swiss triangulation network in this area in 1942, and to the lack
of a renewed determination of the position of the landmarks until
1973, past information concerning the evolution of movements is
quite limited. No damage or special catastrophic event has taken
place in Cuolm da Vi itself, as no building or infrastructure exists in
that zone. The population was aware that some movements were
occurring in this zone but did not pay much attention and was not
overly concerned. The occasional snow avalanches originating from
Val Strem sparked much greater concern.

However, a very remarkable evolution has recently been observed

following several topographic surveys. The rock mass of Cuolm da
Vi has shown some important movements in the last years,
movements that were first visually perceived in 1957 (see Fig. 8.4; at
that time the crack was only slightly open). The first geodetic control
measurements which date from 1983 proved the existence of
significant horizontal displacements, in particular 1.83 m for point No.
514 over a period of 41 years (see Fig. 8.5 below). A large number
of movement measurements later confirmed this fast evolution over
the last few years (see § 8.3.3). These movements seem to be much
more important on top of the Cuolm da Vi zone, at an altitude of
about 2’458 m a.s.l. (point No. 26), than in the lower part of the
unstable slope. During the last 20 years, the elevation of this point
has decreased from 2’45 8m to 2’453 m a.s.l. However, no
catastrophic landslide event has occurred yet, so this very active
phenomenon only implies potential risk which is mostly ignored by
the local population and even more so by the numerous tourists.

8.2 REGIONAL FRAMEWORK

8.2.1 Climatic and water conditions

At the top of the Cuolm da Vi area (altitude approximately 2’500 m
a.s.l.), monthly average temperatures range from −9°C in January to
+6°C in August, with important variations throughout
Page 229

the year, inducing a significant change in periods of snowmelt,

especially as the main unstable slope faces south. The annual
sunshine duration can be estimated to be 1’700 h, which is fairly
high. The average snow cover lasts from six to seven months and
may be quite thick (600 cm for a return period of 100 years at 2’500
m a.s.l.). Therefore, this zone is located at the lower level of the
permafrost and might be affected by the progressive rise of this limit
due to climate change. As far as the annual evapotranspiration is
concerned, it ranges between 200 and 300 mm [data from the Swiss
Hydrological Atlas-SHGN, 1992].

The average annual rainfall measured in the area of the Sedrun

Landslide reaches a value of approximately 2’000 mm, dropping to
1’227 mm at the base of the valley, where the Sedrun rain gauge
station is located. The high spatial variation is due to a marked
topographic effect, which is typically encountered in Alpine valleys.

The extreme rainfall values for a return period of 100 years are 70
mm for a rain duration of 1 hour and 175 mm for a rain duration of 24
hours. The maximum annual precipitation amounts recorded at the
Sedrun station were 1’844 mm in 1935, 1’809 mm in 1999 and 1’836
mm in 2002, this last value being mostly influenced by the extremely
high precipitation in November (544 mm), but which fell mainly as
snow in the area of Cuolm da Vi. Since the beginning of the
measurements, no significant long-term tendency of variation has
been observed, except a long dry period between 1959 and 1964
and a wet period between 1999 and 2002 (133% of the average long
term rainfall over a period of four years). Therefore, it can be inferred
that rainfall is not directly the cause of the increasing movements
over the last 20 years.

Most of the extreme rainfall events in the area are produced by the
same type of meteorological process: warm and wet air masses
coming from the Mediterranean encountering the Alps are forced to
reach high altitudes, leading to sudden condensation and discharge
of their water content.

In the vicinity of the Cuolm da Vi area, there are two main streams
determining the limiting groundwater conditions. The first one is the
Val Strem stream, which runs along the western side of the unstable
zone of Cuolm da Vi, but at a much lower level, producing a deep
drainage effect for the landslide mass. To the east, the Drun Tobel
creek, which is less important in terms of flow, but presents a very
steep slope of 60% over 1’000 m, limits the unstable zone (Fig. 8.2).
The Drun Tobel creek displays a temporary flow, being especially
important in the snow melting season and

Figure 8.2. Drun Tobel creek seen from the village of Sedrun. The
unstable area extends up to the top of the mountain at the back.
Page 230

collecting an important volume of surface water with a high bedload

related to the mylonitic zone, which explains the presence of this
creek (see § 8.2.2). It may induce important debris flows; its lower
stretch requires regular maintenance.

On the right side of the Drun Tobel, as well as along the east slope of
Val Strem, a series of temporary water outlets can be inferred by the
presence of several gullies located between elevations 1’900 m and
2’000 m. In the Cuolm da Vi area itself, absolutely no drainage
pattern is visible, which means that the net precipitation (rainfall and
snow less evapotranspiration) infiltrates completely in the moving
mass, in particular during snowmelt. However, no groundwater level
data are available.

8.2.2 Regional morphology

The Sedrun Landslide is located in the basin of the Rhine river,
namely on the slope forming the left bank of its tributary called
Vorderrhein, near the Gotthard massif, just south of the Oberalpstock
peak (3’327 m). The region consists of a deep glacier valley parallel
to the main original geological structures of the Swiss Alps—Aar
massif (N 70° E). The general slope presents a drop of nearly 2’000
m from the highest summit to the north (Oberalpstock) and of some
1’000 m from the top of the studied Cuolm da Vi landslide area
(Cuolm Parlets Dadens, the peak of which is at an elevation of
approx 2’450 m a.s.l.). Perpendicular to the Rhine Valley, a deep
glacier valley (Val Strem) with steep slopes of some 40° intersects
the main regional geological structures, displaying their conformation
(see Fig. 8.3). The landslide area has been affected by the Würm
glaciation which developed until some 12’000 years ago, as can also
be seen on the gentle slope to the west of Val Strem.

The toe of the slope is incised by a deep erosion creek (Drun Tobel)
resulting from a recent torrential erosion. It displays kakiritic and
mylonitic rocks and favours the destabilization of the slope (Fig. 8.2).
The villages of Sedmn and Camischolas are located on a vast
alluvial fan, formed both by the Val Strem stream and the Drun Tobel
creek. In most of the area, the forest cover is absent.

8.2.3 Regional and structural setting

As far as the stratigraphic units present in the zone are concerned,
the metamorphic southern border of the Aar massif is in contact with
the Urseren-Garvera-Furka zone and with the Tavetsch massif,
which separate the Aar massif in the NW from the Gotthard massif to
the SE.

The lithotypes encountered from north to south can be summarized

as follows: granitic rocks of the Aar massif, gneiss and crystalline
schists of the southern metamorphic border of the Aar massif,
Verrucano and Permo-Trias units of the Urseren-Garvera-Furka
zone, granitic and metamorphic rocks of the Tavetsch massif,
granitic and gneissic rocks of the Gotthard massif.

The geological age of the region can be assumed as Pre-Hercynian

and Permo-Trias.

In the area of Drun Tobel, chloritoschists and schistic gneisses are

observed, as well as mylonites and kakirites resulting from an
intense tectonic crushing without further crystallisation, so that this
rock displays very weak mechanical characteristics,

By means of field mapping, it has been possible to detect in the

region that the main schistosity is sub-vertical, and displaying an
ENE-WSW direction, even nearly E-W at the scale of the massif, that
is more on less parallel to the Vorderrhein Valley and perpendicular
to the lateral small valley of Val Strem.

8.3 HAZARD ANALYSIS

8.3.1 Geomorphological analysis

The longitudinal profile of the Sedrun Landslide displays a fairly
regular slope of around 30% in its central part, with a zone at an
altitude of 2’200 m a.s.l. which is nearly horizontal over a length of
100 m. The global longitudinal profile between the summit of Cuolm
da Vi (initial altitude
Page 231

2’458 m a.s.l.) and the bottom of the Drun Tobel outlet, with its floor
at 1’550 m a.s.l., leads to an average slope of 60% over a distance
of 1’500 m. The exposure of the unstable zone is oriented towards
the south (Fig. 8.3).

In the transversal direction, i.e. ESE-WNW, the profile of the

unstable area displays lateral slopes of 78% in the direction of Drun
Tobel to the east, and 68% in the direction of Val Strem to the west.
Therefore, the most critical potential failure mechanisms are related
to these lateral slopes rather than to the main NS slope, in whose
direction the movements are observed.

The general instability corresponds to a quite particular type of

mechanism, consisting of a deep toppling effect in the sub-vertical
rock layers of gneiss. This toppling effect has its origin in a deep
bending of rock layers under the effect of gravity. There is, thus, no
failure surface in this kind of phenomenon, but it is possible to
delineate an approximate limit of the zone affected by bending, as
shown in Fig. 8.3.

The present instability process results from a more or less marked

evolution of the toppling of rock layers, which will lead to rockslides
or rockfall. The progressive inclination of the layers in the
downstream direction induces the well-known “uphill facing scarp”
type of morphology, with sometimes open grabens (see Fig. 1.3).
This process has led some geologists to believe in an active alpine
tectonic uplift, as this zone is located between the two most seismic
areas in Switzerland, but this opinion can be contradicted on the
basis of the displacement monitoring data.

In the described instability processes, some features have appeared

or are likely to develop in the rock mass of Cuolm da Vi, being a
direct consequence of this instability mechanism. These features are
briefly described below and may be classified into four different
groups: faults, rockslides, uphill facing scarps and rockfalls.

Figure 8.3. Geological profile of the Sedrun/Cuolm da Vi Landslide

with vectors of surface displacements and approximate limit of the
unstable zone (red dotted line). The red vectors indicate the
displacements measured during the geodetic campaigns: A in 1942;
B in 1983; C in 1988; D in 1989; E in 1990; F in 1993. Black thin
horizontal vectors indicate the displacements measured by
photogrammetry between 1973 and 1990 (see § 8.3.3).

8.3.1.1 Faults in Cuolm da Vi

Apart from the three different systems of faults which cut into the
rock mass of Cuolm da Vi (see § 8.3.2), there are other much larger
faults which occur at particular places in the rock mass.
Page 232

At an altitude of 2’404 m a.s.l., a very recent large fault appears, first

seen in 1995, with a NS direction, transversal to the ancient fault
already discovered in 1957 at the same location (NW-SE
orientation). This old fault limits the unstable zone in movement to
the east. It is due to the coexistence of two systems of fractures in
the rock mass and has a length of 120 m and a width of 15–20 m
(Fig. 8.4).

On the terrace, with a fairly horizontal inclination that extends at an

altitude of 2’200 m a.s.l., there are some cracks filled with morainic
material with some compression bulges. The orientation of these
faults follows one of the families of fractures in the rock mass (S1)
(see Fig. 8.5 and § 8.3.2). These presently active faults disappear
towards the west in Val Strem, and to the east in Drun Tobel.

In 2002, a new major vertical crack, some 200 cm wide, oriented SW

then turning W, has been observed uphill of Cuolm Parlets Dadens,
in the small Parlets depression; its extension towards the W crosses
the ridge along the Strem Valley leading to Piz Pardatschas. This
crack was hardly visible 10 years before. However, it is not the upper
scarp of the Sedrun Landslide as other recent cracks appear uphill of
this major feature.
Figure 8.4. Large open fault in the upper part of the Sedrun
Landslide due to the joint effect of the intersection of faults and to
gravity induced phenomena. Behind this crack, very reduced
movements are observed (11 cm in 20 years for point No. 14).

8.3.1.2 Rockslides
To the west, along Val Strem, there are some instances of rock
sliding, most of all at the top of the slope. Unfortunately there are no
monitoring points in this area, but secondary scarps clearly appear
and the rock mass is deeply cut by numerous faults and cracks.
Limited phenomena have already occurred as the toe of the slope in
Val Strem is covered with scree (Fig. 8.3).

To the east side of Cuolm da Vi, along the northern slopes of Drun
Tobel, this phenomenon is even larger, in terms of spatial extension
and activity, with several rockslides and rockfalls of limited depth
being reactivated by seepage and rainfall, as this zone shows the
presence of formations of mylonites and kakirites. A horizontal limit
below which an activity is not perceptible is clearly visible in Drun
Tobel (Fig. 8.2).
Page 233

8.3.1.3 Uphill facing main scarps

Due to the mechanism of toppling in the upper rock layers, some
uphill facing scarps appear near the top of the slope (see Fig. 1.3).
They are the consequence of the bending of rock layers with initially
sub-vertical schistosity on top of Cuolm da Vi. This toppling produces
movements that are concentrated on the main tectonic joints, where
the rock is less strong, due to the schistosity, faults and systems of
faults. This effect produces a dislocation of the rock mass, leading to
the apparition of some scarps or uphill facing scarps, which favour
the potential development of mass movements. Most of these faults
are widely opened inducing small grabens between 2’200 and 2’350
m. At a higher level, these phenomena disappear due to the more
complete toppling and sliding process.

8.3.1.4 Rockfalls
Although no specific evidence of large rockfall is visible in the
Sedrun Landslide area, except for some screes of larger blocks at
the toe of the left bank of Val Strem and for minor events visible in
Figure 8.5, in this zone, it is clear that main rockfall events are in a
preparatory phase to the west of Cuolm Parlets Dadens at an
elevation of approximately 2’400 m a.s.l., in the site called Gion
Giachen. The very high drop down to the bottom of Val Strem
(approx. 600 m), the very steep slope (some 80%) and the intensely
cracked rock mass have to be interpreted as major features for a
potential rockfall. However, it is difficult to forecast if these rockfalls
could occur as a single main event or as a succession of smaller
events.

8.3.2 Local geology and structural analysis

At the site of the Sedrun Landslide, the Aar massif forming the slope
is mainly composed of granite and granodioritic gneiss with biotite or
with two micas to the north, and the southern part is composed of
chloritoschists and schistic porphyritic gneiss, mylonites and
kakirites.
The degree of weathering of the rock of Cuolm da Vi (on the basis of
examined rock exposures and by simple strength tests) could be
classified as slight to moderate, and is related to the toppling effect.
The rock presents a density of around 27 kN/m3. The strength of the
rock (estimated with the aid of a ball-pen or hammer) is extremely
variable.

The rock in the northern part of the site displays a rough schistosity
and a high strength (50–100 MPa). In the southern part, the rock
displays a more developed schistosity and high tectonisation. This
tectonisation leads to the formation of deep series of mylonites and
kakirites (superficial tectonical crushing inducing the formation of a
pulverulent rock without recrystallisation).

It is important to point out the presence of a schistosity in the rock

layers, with a direction ENE-WSW. The dip of the schistosity is 75°–
80° SSE at the end of Val Strem, then tends to be vertical below
Cuolm da Vi, at a low altitude, and even dips NNW in the higher
parts of the Cuolm da Vi slope that are visible from Val Strem, under
the effect of a deep toppling mechanism (Fig. 8.3). This toppling
progressively increases towards Drun Tobel, with a parallel increase
of the kakiritisation of the rocks. This toppling effect even leads to
almost horizontal rock layers near Cuolm Parlets Dadens.

Moreover, the main following structural systems have been detected

as several families of main faults intersect the massif:

– S1: sub-vertical faults N 120°–130°

– S2: sub-vertical faults N 150°–165°

– S3: sub-horizontal faults dipping 15–25° E, sometimes SE

The S1 and S2 faults are present in the same zones and are
apparently not related to a local evolution of the same system.
Finally, some NS isolated faults can be observed.
These various fault families defining the local structural setting,
detected by field survey in 1994 and on the basis of aerial
photographs (Fig. 8.5), can be briefly described as follows:

Family SO: Main schistosity

Displays a direction between N 40° and N 90°, essentially N 70°, and

a dip between 75° and 90° SSE (regional schistosity).
Page 234

Figure 8.5. Aerial photograph of the Sedrun Cuolm da Vi Landslide

(1990) showing the main fault systems, as well as the displacements
obtained by photogrammetry (black vectors—1973/1990) and by
geodetic measurement on 4th order topographic points Nos. 514,
516 and 615 (blue vectors—1942/1993).
Page 235

Family S1

It consists of sub-vertical faults with a direction N 120–130° and a dip

of around 90°. These faults are not related to the schistosity of the
massif, despite their quite similar orientation.

Their persistence varies from 100 m to several hundreds of meters;

the opening of the faults varies from 0 to 1 m; the roughness can be
classified as very rough near the surface. The infilling and water
conditions in the massive and in the faults are unknown.

Family S2

It consists of sub-vertical faults with a direction N 150–165° and a dip

of around 90°.

The persistence varies between 10 m and probably some dozens

meters and the opening is some centimetres wide.

Family S3

It consists of sub-horizontal faults dipping 15–25° E, sometimes SE.

The persistence reaches several hundreds of meters, whereas the

opening is not significant. They are visible in the eastern slope of Val
Strem. Due to the toppling phenomena, these faults, which were
originally dipping slightly opposite to the slope, display an inclination
towards the slope, thus becoming able to induce a sliding
phenomenon.

Family S4

It consists of isolated NS faults. One of them gave rise to a very

large opening. They present a direction N 180° and a dip of around
90°. The persistence extends for some hundreds of meters.
The intersection of family S1 and S2 faults with S4 faults brought
about the opening of a very large “fault” or hole due to gravity
induced phenomena, below the crest at an altitude of 2’400 m (see
Fig. 8.4).

8.3.3 Investigation and monitoring

The only monitoring systems that have been used in Cuolm da Vi
consist of measuring the horizontal and vertical surface movements
by means of different methods. Indeed no borehole has ever been
carried out and no information on groundwater level is available,
except for seepage horizons in Drun Tobel. Summing up the different
geodetic measurement campaigns undertaken in Cuolm da Vi, the
following actions and results can be mentioned.

8.3.3.1 Measures carried out by the ETHZ (Swiss Federal Institute of

Technology, Zurich)
• Measurements organised to reconfirm the coordinates of the 4th
order triangulation points around Cuolm da Vi.

• Local topographic measurements in 1988 and 1989 to determine

the opening of some faults, by MM. Schwendimann, Pluss, Mark,
Köchle and Lautenschlager.

The first geodetic measurements took place in 1983, when the ETHZ
began to reconfirm the coordinates of the 4th order triangulation net
in the area. The purpose of this campaign was to analyse the
movements induced by probable slide processes and not those
related to active faults already described in the zone for a research
project for the National Science Foundation, by Eckardt, Funk and
Labhart (1980 and 1983). This was the reason why this study
focused on the zone of Cuolm da Vi, where a series of some 20
complementary points were installed. During these measurements,
the movements of the triangulation points of the 4th order, namely
Nos. 514, 516 and 615 were verified. Additional measurements were
carried out again in 1988, then partially in 1989 and 1990 (only
points 6 and 26 for this last year, by the survey office of the canton of
Graubünden—see Fig. 8.6) [Bovay-Huguenin surveyors, 1994].

In 1997, new geodetic and GPS measurements were carried out.

Point No. 26, which experiences the fastest movements, had to be
replaced by point No. 70, located some meters away and which
could still be measured in 2003 (its elevation decreased by 4.45 m
over 20 years!).

Geodetic measurements were also extended to the zone of Alp

Caschlé, to the west of Val Strem, which displays a similar glacier
valley morphology. But no significant displacements were observed
there.
Page 236

In 1989, the survey office of the canton of Graubünden checked the

triangulation network and signalled some points for a
photogrammetric flight.

8.3.3.2 Measures carried out by the VERSINCLIM project of the

EFPL (Swiss Federal Institute of Technology, Lausanne)
• Reconfirmation, by means of GPS measurements, of the position of
the 4th order triangulation points and of the points installed by the
ETHZ. This campaign was carried out by the office Bovay-Huguenin,
official surveyors (Epalinges/Lausanne) in 1993, with the contribution
of local surveyors (office Grünenfelder and Partners, in Domat-Ems).

• Determination of movements by photogrammetric measurements

for the period 1973–1990 (Institute of photogrammetry, EPFL).

• Synthesis of movement measurements between 1943 and 1993 for

the VERSINCLIM research project funded by the Swiss National
Science Foundation (Noverrez et al., 1998).

• Field survey of major geological features and interpretation of the

mechanism and causes of movements related to the geological
context.

In 1993, within the VERSINCLIM project (Noverraz et al., 1998), a

campaign including GPS (with Leica S200) and classic geodetic
measurements was carried out by the Bureau BovayHuguenin,
allowing a reconfirmation of 7 out of 12 points of the triangulation net
(except some displaced or disappeared points and others with
difficult access), and also of the majority of the points that had been
laid out in 1983 by the ETHZ, namely 22 significant points. For all the
surveyed points, a double independent determination has been
carried out including either:

– two GPS determinations with 30 minute sessions

– one GPS session and one terrestrial geodetic measurement

– two terrestrial geodetic measurements.

At the same time, a photogrammetric determination of movements

based on the comparison of photos taken in 1973 and 1990 was
made by the Institute of Photogrammetry of the Swiss Federal
Institute of Technology, Lausanne (EPFL) checking some 60 points
on the slope of Cuolm da Vi.

Thus, considering the combination of the various measurements

carried out, the periods of time for which a computation of the
displacements was made are the following:

1942–1983; 1983–1988; 1988–1989; 1989–1990; 1990–1993, as

well as 1973–1990, and, by deduction, for neighbouring points:
1942–1973 and 1973–1983.

The total significant displacements that were obtained by means of

GPS measurements for the period 1942–1993 were 3.2 m, 0.9 m
and 0.45 m, respectively, for points Nos. 514, 516 and 615 of the 4th
order triangulation net, the global precision reaching less than 10 cm
(including the uncertainty of the original 1942 determination). They
vary between 0.94 m and 5.18 m for 10 out of the 22 ETHZ
significant points between 1983 and 1993. The most important
displacement (5.18 m in 10 years) is the one measured at the
summit of Cuolm da Vi.

The displacements obtained by means of photogrammetry

(Intergraph Image Station) between 1973 and 1990 give a more
complete image of the spatial distribution of movements, showing a
fairly homogeneous tendency within the unstable mass. The
standard deviation on the measurements is ±20 cm, which is 10 to
15 times less than the observed movements.

Despite the particular care in carrying out all these measurements

and in analysing them, it is quite possible that the short term
observations (1988–1989, 1989–1990, 1990–1993), especially for
the local points, may give too extreme velocity values, such as for
point No. 26, which require further observations. This is due in
particular to the fact that there may be a combination of global and
local movements due to the intense cracking of the rock mass. The
measurement conditions are also quite difficult and the number of
really fixed points is reduced (often slowly moving pointed have been
taken as reference points).

8.3.3.3 Measurements carried out by the consultants of the

community of Sedrun
During the last five years, the geological office Bonanomi AG in
Sedrun and the survey engineering office Donatsch in Landquart
have been appointed by the Community of Tujetsch, including
Page 237

the village of Sedrun, to carry out a monitoring programme and

follow the evolution of the landslide of Cuolm da Vi (Bonanomi &
Donatsch, 1998).

The monitoring system first included the continuation of the survey

measurements on the points located in the landslide area by the
ETHZ and by the survey office of the canton of Graubünden (in 1989
and 1994), as well as on a few new points to the NW of the
landslide, and then the operation of continuous distance
measurement devices (since 2000) between a fixed point located in
the valley near the village and two points near the top of Cuolm da
Vi. This automatic system also includes a camera to visualize the
evolution of the slope.

These data allowed the confirmation of the very important

displacements recorded at point No. 26, which increased again, after
a relatively quieter period from 1993 to 1998, to nearly 80 cm/year in
2000 and 75 cm/year in 2001. The other points on the slope below
the peak are less active, as already observed, but several of them
still reach annual velocities between 15 and 30 cm/year, which vary
very little and, thus, do not show any clear acceleration trend.

Figure 8.6 below presents a synthetic view of the long-term tendency

of variation of the velocities for several significant points, which
clearly marks a considerable change in the development of the
phenomenon since the 1970’s, putting forward a potentially high
hazard level.
Figure 8.6. Annual displacements of several points on the Sedrun
Cuolm da Vi Landslide. The location of some of these points is
visible in Figure 8.5.

The continuous measurements related to two specific points show a

fairly constant tendency between June 2000 and the end of 2002
with a slight acceleration at the end of the spring. The average
velocity recorded for these two points is included between 25 and 30
cm/year. Therefore the short term evolution seems less worrying
than the long term one.

8.3.4 Danger identification

Because no boreholes are available at the Sedrun Cuolm da Vi
Landslide, the volume of the unstable rock mass can only be
assessed approximately to some 100 or 110·106 m3, as a
consequence of the small uncertainties related to the spatial limits of
the landslide zone and above all to the presence or absence of a
basal failure surface. But the existence of very important
displacements, of major cracks extending over hundreds of meters
and of severely dislocated rock zones visible at the surface, near the
top of the unstable area, clearly proves the presence of a potential
danger. In order to analyse the possible occurrence of hazardous
phenomena, it is first necessary to understand the possible failure
mechanisms that may occur, which leads to the description of
several
Page 238

landslide scenarios that are likely to cause direct or indirect

important damage to the village of Sedrun. Several realistic
scenarios have been elaborated, based on the geology, morphology
and survey data, and three of them have finally been selected as the
most representative potential mechanisms, to be considered in the
risk assessment procedure.

8.3.4.1 First potential scenario with a high probability of occurrence

This scenario implies the development of a limited slide, i.e. not a
shallow phenomenon, but a massive slide (approx. 20’000 m3) on
the right side of Drun Tobel, in the zone called Plauncas Tgamos
where the granite gneiss formation to the north and the schistic
gneiss formation to the south are separated by a zone of mylonites
and kakirites prone to instability due to their low shear strength (Figs
8.2 and 8.7).
Figure 8.7. Scenario No. 1 for the risk assessment studies.

This slide of limited volume, triggered on a very steep slope of the

Drun Tobel, where similar events seem to have already occurred
following intense rainfall, might suddenly dam this creek and induce
the immediate formation of a small lake. As soon as the overflow
begins, the dam will be eroded and cause a fast debris flow with an
important mass of granular material, a part of which will be washed
away from the creek bed itself.

Due to the presence of a sharp bend at the outlet of Drun Tobel

Creek, at the approximate elevation of 1’500 m a.s.l., the high
velocity debris flow will not stay in the thalweg and overflow on the
alluvial fan of the Strem River, below the site called Valtgeva,
upstream of the village of
Page 239

Figure 8.8. Bend of Drun Toble Creek where the debris flow
considered in Scenario 1 will flow directly towards the village of
Sedrun (the first visible houses are only barns).

Sedrun (Fig. 8.8). Therefore, the debris flow formed of coarse

material would affect a part of the village downstream of the railway
station. As the Val Strem alluvial fan is not very steep, the debris flow
would reduce its speed and, therefore, only hit the basement of the
first exposed houses, as well as damage the railway line.

This event has never occurred there in the past, according to

historical information, but the presence of large blocks in the lower
part of the Drun Tobel creek leads to the assumption that that it is a
very reliable scenario, as the Drun Tobel streambed shows narrow
stretches at some places. The very large fan deposits of the Drun
Torrent indirectly prove that similar phenomena occurred in the past.
Similar debris flows have seriously affected the Vorderrhein Valley
downstream in November 2002, in particular in the village of
Schlans, following extremely intense rainfall that had never been
observed before, over a period of three days. But in the Cuolm da Vi
area, the precipitation of this event only fell as snow. However this
first scenario has not been modelled in a debris flow code, as the
necessary assumptions related to the expected volume of implied
materials and to the overflow process might induce a large range of
flow velocities and very different potential consequences. The
selected scenario has thus to be considered as a typical danger
situation, but not implying a unique path, with a relatively high
probability of occurrence (10 to 100 year return period).

8.3.4.2 Second potential scenario with a medium probability of

occurrence
This scenario considers the occurrence of a large slide (more than
100’000 m3) on the right side of Drun Tobel, namely in approximately
the same zone as in Scenario 1, that is in a weak rock zone, but
extending higher and wider in the lateral slope of the creek and
implying a larger volume due to a regressive failure mechanism. This
scenario takes the movement of point No. 27 into account, which
reaches more than 5 m in 18 years (1983–2001) and is located in
the upper part of the expected slide zone, as being a significant
preparatory sign; but as no acceleration is perceived, the occurrence
is not expected in the short term.

This potential event caused by intense or long lasting rainfall may

induce a major debris flow which will not be contained by the natural
thalweg, so that massive overflow will occur on the
Page 240

Figure 8.9. Scenario 2 for the risk assessment studies.

alluvial fan of Val Strem, reaching the course of this river to the west
of Sedrun; it will also extend along the Drun Tobel valley where it
crosses the village of Sedrun down to the Rhine river. The narrow
zone of impact in this last stretch may be widened by erosion
phenomena which would destabilise the creek banks (Fig. 8.9).

The affected zone has been determined on the basis of the local
topographical conditions of the slope in the lower part of Val Strem,
referring to other similar situations which have occurred in
Switzerland, in particular in August 1987 in Poschiavo and in
September 1993 in Brig. For such a case, and considering the large
uncertainties affecting the volume as well as the debris flow
mechanism, mathematical modelling would not provide more reliable
results to determine the expected zone of impact. Indeed these two
first scenarios will be used in a comparison of quantitative risks with
respect to phenomena of high and medium probabilities of
occurrence (100 to 300 year return period for Scenario 2).

8.3.4.3 Third potential scenario with a low probability of occurrence

This quite different scenario considers a potential rockfall from the
top of the Cuolm da Vi area, in particular from its peak called Cuolm
Parlets Dadens at an initial elevation of 2’458 m a.s.l. It will occur in
the direction of the Val Strem valley, as the slope is very steep to the
south-west (about 80%), with an average height of fall of some 600
m. Due to the intense cracking and
Page 241

de-structured aspect of the gneiss rock mass in this area of the

Sedrun Landslide, as well as the recently appeared movements to
the west of the peak following the deep toppling mechanism which
weakens the rock strength properties, and despite of the fact that the
present movements are not clearly oriented towards the SW, a total
volume of approximately 5·106 m3 may be involved, either as a
single event or as a succession of smaller rockfall events (Fig. 8.10).

Figure 8.10. Scenario 3 for the risk assessment studies.

Considering the induced energy of compaction related to the drop,

such a phenomenon might certainly cause the formation of a large
dam in Val Strem, which is very narrow at the expected point of
impact of the rockfall (less than 3 00m for a dam height of 100 m). It
would thus form a large lake in the Val Strem valley, which would be
filled rapidly, especially if this occurs in the spring during the
snowmelt when the flow of the river is important (no discharge data
exist, but according to the size of the drainage area and the amount
of precipitation, it is expected to carry more then 1 m3/s in normal
spring conditions—average annual discharge=0.54 m3/s—and up to
15–20 m3/s during storm events).

Considering the original gneissic rock material, the dam would be

formed of fairly coarse material, but the experience of Randa rockfall
in 1991 with a similar rock formation shows that an intense
compaction allows the dam to be fairly impervious except at the
surface of the deposit. Even if the
Page 242

seepage develops when the lake level reaches the lowest point of
the deposited rockfall mass, it would probably not prevent an
overflow which would cause a fast erosion and thus create a
massive flood downstream. Considering the velocity of such flow
induced by the steep inclination of the stream bed in Val Strem
(approx. 10%), which would exceed 10 m/s, the flood would dash
down the river bed and pass over its banks, devastating a part of
Sedrun (Community of Camischolas) down to the Rhine river. This
catastrophic scenario, assessed to have a return period of some 300
to 1’000 years, may have serious consequences on the Rhine valley
downstream, in particular for a hydro-electric plant as well as one of
the building sites of the Gotthard base tunnel and for its respective
construction facilities. The present evolution of the affected zone
near Cuolm Parlets Dadens may even lead to the assumption that
the probability of occurrence might increase in the future.

8.3.5 Geomechanical modelling

In order to try to understand the development of deformations in the
Cuolm da Vi area, geomechanical modelling has been attempted at
the Polytechnical University of Catalunya (UPC), aiming at the
determination of the deformation field after the retreat of the glacier
which occupied the Rhine Valley. However, this 2D-computation,
taking the various crack and fault systems into account, has yielded
minor displacements at the top of Cuolm da Vi and higher
displacements at the toe of the slope, whereas the present
monitoring data (extending over the last 60 years as well as over the
last few years), indicate that the major displacements occur at the
top of the slope. Such a situation can be explained as several
mechanisms are affecting the slope (topple, eventually slide, as well
as possible melting of permafrost) which are not directly related to
the past history of the glacial retreat. Indeed, in such cases the
classical “triggering” numerical modelling is not appropriate to supply
a displacement pattern that can be directly used to express the
hazard. The reference to the experience of potential landslide
scenarios, based on historical events in other contexts, is more
significant to serve as a basis, for the risk assessment, than the
results of elastoplastic mechanical modelling that is not sophisticated
enough to consider a possible melting of the permafrost and creep
effects.

8.3.6 Occurrence probability

As indicated in the three most reliable scenarios selected for the risk
analysis described above, the occurrence probability considered
here is not based on known past events at that site, but on
experience of similar events that are quite frequent when implying
limited volumes (Scenario 1), less frequent when the slide displays a
larger mass (Scenario 2) and rare in the case of a rockfall of several
million m3 (Scenario 3).

For the risk computation, the following return periods have thus been
assessed in a conservative way, especially for the last scenario.

– Scenario 1:10 year return period.

– Scenario 2:100 year return period.

– Scenario 3:1’000 year return period.

It is clear that these investigation hypotheses should be confirmed at

a later stage, for instance by more sophisticated numerical modelling
or on the basis of new field observations, but above all when
appropriate geological and geotechnical data will be obtained by
boreholes, including underground temperature data.

8.4 QUANTITATIVE RISK ANALYSIS

8.4.1 Elements at risk

In the case of Sedrun, a large village with 1’100 inhabitants in 1950
and 1860 inhabitants in 2000, including the sector of Camischolas
(figure for the whole community of Tujetsch), the potentially
Page 243

affected zones when considering the three selected scenarios are

fairly limited, so that a detailed census of the elements directly at risk
has been carried out. On the basis of the local management plan of
the Community of Tujetsch (Fig. 8.11) and following investigations on
site, a comprehensive list of the different exposed “objects” has been
established, classified into 23 categories, as shown in Table 8.1.
They first include the population, then the buildings and
infrastructures, and finally the implied economic activities
(transportation lines, for example). In winter, the population
significantly increases, as the total number of holiday beds reaches
5’400.

Figure 8.11. Local management plan showing the exposed objects. It

can be noted that natural hazard red zones are shown on this map,
but they are presently based on snow avalanche and debris flow
hazard only. The purple zones along the Rhine River correspond to a
hydro-electric plant, technical equipments and the construction
facilities for the new Gotthard base Tunnel.
As far as the exposed persons are concerned, a fairly low average
rate of occupancy is considered, due to the fact that most of the
potentially affected buildings are holiday properties. Of course it
cannot be excluded that many persons might be present outside of
the buildings in the potentially affected are; but this assumption is
extreme if the weather conditions causing at least Scenarios 1 and 2
are recorded. An appropriate alarm system may reduce such
numbers significantly.

8.4.2 Vulnerability of the elements at risk

According to the type of structures built in the exposed areas and to
the mechanism and intensity of the landslide events considered in
the different scenarios, the vulnerability coefficients specifying the
expected degree of damage have been assessed, based on the
reference values given in the literature (Leone et al., 1996), as well
as on the consideration of the type of building structure observed in
the field. The selected values appear in Tables 8.2 to 8.4. It is clear
that the
Page 244

Table 8.1. Categories of exposed “objects” considered in the risk

analysis.

People Population at risk

Buildings House 1st class

House 2nd class

House 3rd class

Restaurant

Railway annex building

Garage

Industrial building and equipments

Barn, farm

Hydro-electric plant

Networks Road (value/m)

Forest road (value/m)

Bridge (value/m)

Railway line (value/m)

Electrical line (value/m)

Natural areas Fields (value/m2)

Forest (value/m2)
 River (value/m)

Vehicles, movables

Socio-economic activities Closed roads (value/day)

Closed railway (value/day)

Housing (value/day)

Unemployment (value/day)

Tourism (value/day)

considered scenarios described in § 8.3.6 which will not imply thick

masses of debris in the village area justify fairly low vulnerability
coefficients for most of the buildings. At this stage, no difference has
been introduced between the buildings hit in a first line by the debris
flows and those located downstream which will probably be less
affected, due to the protection offered by these upstream buildings,
especially in Scenarios 1 and 2. For Scenario 3, the considered
vulnerability coefficients are higher due to the important impact of the
flow.

8.4.3 Expected impact

The impact caused by each considered scenario results from the
multiplication of the value of each “object” by the respective
vulnerability coefficients; the sum of all expected impacts gives the
global impact corresponding to each scenario (see Figs. 8.12, 8.13
and 8.15, as well as Tables 8.2 to 8.4). The monetary values
included in the tables are in Swiss Francs and, therefore, must be
multiplied by 0.64 to obtain amounts in €.

It is evident that Scenario 1 implies a reduced impact in terms of

physical damage, mainly to buildings (1.82 mio CHF or 1.65 mio €)
which are insured. The indirect impact on the economic and
transportation activities will also be limited, as it will be possible to
carry out the necessary repair works at very short notice (traffic
interrupted during one day for the railway line). For the Drun Tobel
creek, the damage to the right bank where the debris flow will leave
the stream bed will be significant. But the suddenness of the event
considered in Scenario 1 may imply human victims if the exposed
persons are outside of their houses; however, this is fairly
improbable considering the bad weather conditions expected. As far
as the indirect impact on tourism is concerned, it is very difficult to
assess, but the possibility of repairing the damage during the
summer should not affect the next winter season.
Page 245

Figure 8.12. Impact of Scenario 1.

Table 8.2. Expected costs related to Scenario 1, excluding costs of

directly affected persons. The number of persons at risk takes an
average rate of occupancy into account.

NumberValue VulnerabilityCost

People 25 ?

Buildings House 1st class 41’000’0000.2 800’000

House 2nd class 2 750’0000.25 375’000

House 3rd class 2 500’0000.2 200’000

Restaurant 12’000’0000.15 300’000

Annex building 1 100’0000.15 15’000

Garage 2 100’0000.15 30’000

 Barn 2 100’0000.5 100’000

Networks Road (value/m) 150 5000.6 45’000

Forest road 100 2000.8 16’000

(value/m)

Railway line 200 5000.4 40’000

(value/m)

Natural areas Fields (value/m2) 75’000 11 75’000

River (value/m) 200 2’0000.5 200’000

Vehicles,
movables

Socio- Closed roads

economic (value/day)
activities

Closed railway 10’0001 10’000

(value/day)

Housing
(value/day)

Unemployment
(value/day)

Tourism 750’000(1’500 p/day)

(value/day)

Total costs 2’206’000

(CHF)

Total cost (€) 1’412’000

Page 246

Figure 8.13. Impact of Scenario 2.

Table 8.3. Expected costs related to Scenario 2, excluding costs of

directly affected persons. The number of persons at risk take an
average rate of occupancy into account.

NumberValue VulnerabilityCost

People 150 ?

Buildings House 1st class 21’000’0000.15 300’000

House 1st class 11’000’0000.2 200’000

House 1st class 11’000’0000.25 250’000

House 2nd class 1 750’0000.15 112’500

House 2nd class 12 750’0000.2 1’800’000

House 3rd class 2 500’0000.2 200’000

 Garage 1 100’0000.15 15’000

Barn 11 100’0000.5 550’000

Networks Road (value/m) 350 5000.6 105’000

Forest road 600 2000.8 96’000

(value/m)

Bridge (value/m) 4/80 16’0000.2 256’000

Railway line 250 5000.4 50’000

(value/m)

Natural areas Fields (value/m2) 120’000 11 120’000

River (value/m) 2’100 2’0000.5 2’100’000

Vehicles,
movables

Socio- Closed roads 2 2’0001 4’000

economic (value/day)
activities

Closed railway 10 10’0001 100’000

(value/day)

Housing
(value/day)

Unemployment
(value/day)

Tourism 750’000 (500 p/day)

(value/day)

Total costs 6’258’500

(CHF)

Total cost (€) 4’005’000

Page 247

Figure 8.14. Bridge for the cantonal road crossing the Val Strem
creek that would be damaged by the debris flow considered in
Scenario 2.

Figure 8.15. Impact of Scenario 3.

In Scenario 2, the damage will affect several bridges, as the one
shown in Figure 8.14, but it can be assumed that they will be only
partially destroyed, due to their reinforced concrete structure and to
the deep foundations of their piles. However, the damage to the river
courses of the Strem River and the Drun Tobel Creek will be
important in terms of environmental impact (2.1 mio CHF or
Page 248

Table 8.4. Expected costs related to Scenario 3, excluding costs of

directly affected persons. The total cost is given with and without the
losses corresponding to a decrease in tourism activity.

NumberValue VulnerabilityCost

People 120 ?

Buildings House 1st class 2 1’000’0000.3 600’000

House 1st class 1 1’000’0000.4 400’000

House 1st class 1 1’000’0000.5 500’000

House 1st class 1 1’000’0000.6 600’000

House 2nd class 1 750’0000.3 225’000

House 2nd class 9 750’0000.4 2’700’000

House 3rd class 2 500’0000.4 400’000

Barn 5 100’0000.8 400’000

Barn 2 100’0000.5 100’000

Annex building 1 500’0000.4 100’000

Small annex 3 10’0000.3 9’000

building

Small industrial 1 1’000’0000.25 250’000

cable lift

Electric 116’000’0000.3 4’800’000

plant
Networks Road (value/m) 650 5000.6 195’000

Forest road 100 2000.8 16’000

(value/m)

Bridge 3/60 16’0000.2 192’000

(value/m)

Railway line 150 5000.6 45’000

(value/m)

Electrical line 200 1500.5 15’000

(value/m)

Natural Fields 10’000 11 10’000

areas (value/m2)

River (value/m) 3’000 2’0000.5 3’000’000

Vehicles, 3 100’0001 300’000

movables

Socio- Closed roads 10 4’0001 40’000

economic (value/day)
activities

Closed railway 20 10’0001 2000’000

(value/day)

Housing 2’000 150 300’000

(value/day)

Unemployment 4’000 120 480’000

(value/day)

Tourism 750’000 (1’500 p/day)(22’500’000)

(value/day)
 Total costs (CHF) without loss related to tourism 15’877’000

Total cost (€) 10’161’000

Total cost (CHF) With expected loss related to 38’377’000

tourism

Total cost (€) 24’561’000

1.344 mio €). The damage to buildings will represent the higher part
of the expected costs (approx. 3.43 mio CHF or 2.2 mio €). As far as
socio-economic activities are concerned, a limited amount has been
considered for closed roads, because the main access road to
Sedrun along the Rhine valley to the east should not be affected and
as it will be possible to re-establish a by-pass road through
Camischolas for the traffic coming from the Oberalp Pass. No
amount has been computed for the loss of tourist activity as the
major facilities will not be concerned; but the reputation of the resort
might be affected.

In Scenario 3, the direct damage to buildings is much higher than in

Scenario 2 because of the higher expected vulnerability coefficients
(5.93 mio CHF or 3.8 mio €), but one of the major indirect impacts
will be the partial destruction of the hydro-electric plant near the
Rhine River, as well as the damage to the construction facilities for
the Gotthard base tunnel which includes losses of machinery (in the
items of movables) and partial unemployment. Moreover, a delay in
the achievement of this very important infrastructure for the whole
European economy may have huge economic consequences that
cannot be assessed in this research. The interruption of electricity
production may also imply losses of millions of CHF. Finally if the
expected cost without considering the loss
Page 249

related to tourism reaches nearly 16 mio CHF, the possible additional

impact on tourism, assuming 30 days of total loss of tourist income,
will be 22.5 mio CHF or 14.4 mio €. This value has to be taken into
account as a rockfall may occur even in winter time. This last figure
clearly puts forward the considerable importance of the indirect
impacts on the local economy, which requires careful attention in
management and protection plans.

8.4.4 Risk assessment

Table 8.5 below sums up the risks corresponding to the three
selected scenarios in terms of physical, environmental and socio-
economic damage. The number of potentially directly affected
persons is also computed, but no monetary value is given. It clearly
appears that Scenario 1 implies a relatively higher risk of assets and
life because of its high assessed probability. In terms of social risk,
Scenario 2 may appear more important as the impacted houses are
numerous (19, dwelling houses) and will be more severely affected;
but when the probability is considered, the risk to life is smaller than
in Scenario 1. Nevertheless, the aversion factor for which a large
number of victims in one event is less acceptable that the same
number for several disasters will have to be taken into account as far
as the risk perception is concerned. Finally in terms of indirect risks,
not only at a local scale, but at a European scale, Scenario 3 will
definitely imply very severe consequences in particular on regional
tourism, electricity production and on the facilities for the future
transportation system through the Alps.

Table 8.5. Computation of the risks involved by the three considered

scenarios.

Scenario considered 1 2 3

Costs related to the corresponding scenario 1.4 4.0 24.6

(approximate value) million € million € million €
Number of persons directly at risk 25 150 120

Assessed probability 0.1 0.01 0.001

Corresponding monetary risk 140’000 40’000 € 25’000 €

€

Corresponding risk to life 2.5 1.5 0.12

As the three scenarios are independent of one another and will not
necessarily occur within a short period during which damage will not
yet have been repaired, the total risk implied corresponds to the sum
of the three quoted figures, namely a monetary risk of some 205’000
€ (yearly amount to spare in order to cover future damage) and a risk
to life of 4.1 victims per year. Preventive measures are thus definitely
required.

8.5 REGULATION AND RISK MITIGATION MEASURES

8.5.1 Existing measures

After a long period (1983–1993) during which the investigations
carried out at the Sedrun Landslide in order to understand and
determine the displacements were of interest only to scientists, and
which led to the expression of various opinions concerning the
possible mechanisms, the authorities of the community of Tujetsch
became concerned about the marked evolution of displacements in
the Cuolm da Vi area. Therefore, they appointed a consulting
geologist (Bonanomi AG at Buguei-Sedrun) and a surveyor
(Donatsch Ingenieur- und Vermessungsbüro SA, Landquart), in
order to check the area and follow the evolution of the movements
more closely. In accordance with this aim, several survey campaigns
were organised annually (from 1998 to 2003) and from July 2000 on,
a continuous distance measurement of two points located near the
summit as well as one reference fixed point were installed and
operated, with the possibility for access on-line to this
Page 250

information and also of visualising the slope by a picture taken at

daily intervals from the site of the distance meter.

Moreover, it is foreseen by the community to carry out rockfall

modelling in order to produce a hazard and risk map, according to
information received in August 2002. But the community and its
consultants are convinced that boreholes would neither be cost-
effective nor contribute to the main aim, that is the safety of the
population. This aspect is, however, still in discussion at the federal
level.

8.5.2 Proposed investigations to plan mitigation measures

On the basis of all the existing information—as far as it is accessible
—and considering the possible consequences of any hazard
scenario as analysed in this chapter, four main recommendations in
terms of need can be expressed at this stage to improve the safety
conditions in the area of Sedrun:

1. Need for further investigation.

2. Need for preliminary protection measures.

3. Need for specific regulations.

4. Need for extended monitoring and warning system.

8.5.2.1 Need for further investigation

The present analysis, as shared by the IMIRILAND partners, does
not conceal the fact that the real mechanism affecting the slope at
Cuolm da Vi is not yet fully understood. This is due in particular to
the lack of subsurface information concerning the geological
conditions, the groundwater levels, the displacement pattern with
depth and the temperature distribution. Surface displacement data
are also lacking in some parts of the landslide, namely in the areas
from which the expected scenarios presented in § 8.3.6 are likely to
be triggered.

Before these investigations are carried out, any geomechanical

modelling, such as the one which was attempted for the Sedrun
Landslide within IMIRILAND project, will not reach a sufficient level
of reliability and therefore will not help in producing significant hazard
and risk maps. For the time being, an analysis based on possible
feasible scenarios is still the best approach.

8.5.2.2 Need for preliminary protection measures

Although the potential landslide mechanisms and respective rock
masses considered in the three selected scenarios are not yet
completely defined, it is possible to plan preliminary protection
measures along the expected trajectories, which will definitely
contribute to slowing down the process or to limiting its extension.
Therefore, there is a need to design and carry out such appropriate
measures, according to the principles described in § 8.5.3. Indeed,
even if the initial design is not the best, it will improve the safety
conditions and should allow later modifications, when more reliable
landslide scenarios are confirmed by geomechanical modelling.

8.5.2.3 Need for specific regulations

Presently the local management plan includes the consideration of a
snow avalanche hazard, as it occurred in 1749, 1808 and 1817,
uphill of the eastern part of Sedrun and of a debris flow hazard along
the Val Strem and Drun Tobel river beds (see Fig. 8.11). The
corresponding danger zones in the plan should also be extended to
consider the hazards related to the presented scenarios, even if their
delimitation cannot be expressed in a definitive way. Indeed, several
examples exist in which, after a serious landslide event, provisional
zoning has been adopted by the local authorities and building zones
have been temporarily suspended until the real hazard could be
properly determined.
These regulations do not concern the local management plan only,
but also the preparedness plans, including the organisation of
evacuation routes, the training of the population in presence of an
alarm signal, the planning of emergency protection measures, etc.

8.5.2.4 Need for extended monitoring and warning system

According to the presently considered landslide scenarios, it is likely
that the critical movements will not affect the whole zone of Cuolm
da Vi, but only a small part of it. In these specific areas,
Page 251

nearly no monitoring data are recorded due to the particularly difficult

access; furthermore, no continuous measurements are available and
the evolution of the respective slopes prior to failure will be difficult to
detect visually. It is, therefore, necessary to install more monitoring
systems; their cost-effectiveness has been proven on many
occasions, when it is linked to the potential number of victims which
may be evacuated in time on the basis of reliable monitoring data.
On the other hand, these new devices with automatic transmission
must include a two-level warning system allowing a preliminary
scanning of the data by the authorities or their consultants before the
public alarm is triggered.

8.5.3 Proposed mitigation measures

With respect to the three analysed scenarios, the following
preliminary mitigation measures can be proposed for further design
and implementation, according to the priorities set up for the most
urgent protection plans.

8.5.3.1 Measures corresponding to Scenario 1

• Construction of check dams on the Drun Tobel creek so as to
cause the deposition of sediments and to reduce the erosion
mechanism upstream of the possible overflow zone (see Fig. 8.8);
these works will also reduce the peak flow after the failure of the
landslide dam, by ensuring the proper conditions for kinetic energy
dissipation.

• Construction of a protection dam on the right bank of the possible

overflow zone, in order to reduce or suppress the extension of the
debris flow on the alluvial fan of Val Strem.

• An alternative proposal, or in conjunction with the former one,

implies a riverbed improvement in the possible overflow zone so as
to increase the flow velocity and avoid any sediment deposition. In
such a case, all the downstream part of the Drun Tobel creek has to
be protected so as to avoid scouring where the creek crosses the
village.

8.5.3.2 Measures corresponding to Scenario 2

• Excavation of a 300 m long rectilinear channel between the Drun
Tobel creek, at the beginning of its bend below the site of Valtgeva,
and the Val Strem River, upstream of the railway bridge (see Fig.
8.9). This channel should be some 15 m deep; it will require the
displacement of three ski-lifts on the alluvial fan of Val Strem. The
normal flow of the Drun Tobel creek should be maintained in its
original riverbed.

• River protection works along the lower part of Val Strem, in the
area of Camischolas-Zarcuns, down to the hydro-electric plant near
the Rhine.

• An alternative proposal may include the building of a high catch

dam on the Val Strem River, upstream of Camischolas, to collect the
debris flows and sediments coming from the channel connecting the
Drun Tobel creek to the Val Strem River. This dam could be built with
the excavation materials from the channel. It could also be useful for
Scenario 3. Its height is difficult to assess, but will certainly exceed
20 m, which will cause a certain impact in the scenery.

8.5.3.3 Measures corresponding to Scenario 3

It is difficult to justify any major protection measures for such a
process with a very low probability of occurrence. But the catch dam
considered for Scenario 2 might prove to be efficient for this scenario
also, even if it would not retain all the debris flow induced by the
failure of the landslide dam. It would certainly reduce the peak
discharge of this flood wave, block a large part of the transported
sediments, and give some additional time to the population for its
evacuation.

However, this dam will require a protected spillway to avoid its

destruction in case of overflow and of obstruction of the torrent
culvert by snow avalanches.

It is clear that all these mitigation measures have to be studied not

only in the perspective of reducing the hazard, but also in a
development prospective allowing a sustainable increase of tourist
activities which are the major industry for the community of Tujetsch.
Page 252

8.6 CONCLUSIONS

The Sedrun Landslide case study represents a typical situation of

the difficult paradox that the populations of the alpine range have to
face if they want to continue developing in their homeland. As the
hazards implied do not seem obviously threatening, the population
and the local authorities tend to ignore or understate the phenomena
in order to avoid panic or fear and thus reduce the attractiveness of
their resort. Meanwhile, they limit their investigations and do not
favour a scientific debate of the risks involved. They rather favour
new constructions which increase the value of the exposed objects.
On the other hand, as time passes, the slope movements continue
and their exceptional magnitude, especially at the top of the slope,
necessarily implies an increased probability of occurrence of a
significant event.

Therefore, the only way to face the landslide risks is to recognize

their existence, to improve their knowledge by appropriate
investigation and monitoring programs, to take provisional prevention
measures in order to reduce the short term potential impacts, and
once the potential mechanisms are better understood and modelled,
to carry out protection works so as to limit the consequences of any
possible landslide event.

The scientific investigations carried out for the Sedrun Landslide

within the IMIRILAND project do not pretend to give final
recommendations or advice as far as the mitigation actions are
concerned. This case study was considered to show the applicability
of landslide risk methodologies and demonstrate the possibility of
expressing the relative magnitude of such risks according to the
various scenarios analysed. It is clear that several parameters may
change with time with respect to the evolution of the unstable slope,
to the development of the tourist area and to the application of
preliminary prevention measures. But it is vital to initiate a long-term
process of risk analysis and management, so that the population and
the tourists can be convinced that the situation is really “under
control”.
Page 253

9
The Séchilienne landslide

Jean-Louis Durville, Laurent Effendiantz, Pierre Pothérat

Centre d’ÉtudesTechniques de l’Équipement de Lyon

Philippe Marchesini
Direction Départementale de l’Équipement de l’Isère

9.1 INTRODUCTION

The Séchilienne landslide has developed on the right side of the

Romanche valley, 20 km to the south-east of Grenoble in the Isère
Departement of the French Alps (Fig. 9.1). The national road RN 91
runs along the valley bottom and then to the Lautaret Pass (the link
from Grenoble to Briançon and Italy), one of the higher alpine
passes which is not closed in winter. Except for a narrow winding
road, the RN 91 is the only access to numerous ski resorts as
L’Alpe-d’Huez and Les Deux-Alpes.

Figure 9.1. Location map of Séchilienne.

9.1.1 Landslide description

The landslide is occurring on a slope extending from an elevation of
330 m a.s.l. at the bottom of the valley to 1150 m a.s.l. at Mont Sec
(Fig. 9.2). The landslide itself extends from 600 m a.s.l. up to 1130 m
a.s.l., over an area of approx. 70 ha. The limits are as follows:

– to the east: a major tectonic shear zone (N 20°), which is a clear-

cut border,

– to the north (upwards): a scarp which has probably been formed by

a large post-glacial sagging,

– to the south (downwards): the slope below 600 m a.s.l. is not

moving,

–to the west-south-west: the movement is assumed to decrease

continuously.

The landslide can be divided into three principal areas (Fig. 9.3): the
frontal mass, the most active (several decimetres per year; source of
many rockfalls) and disrupted part, the volume of which can be
estimated at approximately 3 hm3; an intermediate zone with
medium activity; the upper and north-western part, corresponding to
an elliptic (probably) post-glacial sagging, with low velocities.
Page 254

Figure 9.2. General map of the site of Les Ruines de Séchilienne

(IGN map).
Figure 9.3. Aerial view of the frontal mass. The Séchilienne plain can
be seen in the background.
Page 255

9.1.2 Historical background

Many rockfalls have been reported during the last centuries resulting
from the right side of the Romanche valley at the site currently called
“Les Ruines”. Some mining activity (metallic sulphurs) was carried
out in the XIXth century and ended after the 1st World War. The
oldest aerial photographs (1937 and 1948) show a slope morphology
with elliptic sagging and NE-SW trenches and active screes. But it
may be seen that a footpath remained passable through the frontal
mass until after World War II.

An important reactivation was observed in the 1980’s: rockfalls hit

the national road RN 91 during the winter of 1985 and the existence
of a large slope deformation was recognized (Antoine et al., 1987).
Monitoring of the slope was set up and some protective measures
were installed: a fence with an electrical wire alarm and traffic lights
along the original portions of the RN 91, a dam with a storage
volume capacity of about 2 hm3 of debris and a diversion river bed
for the Romanche River.

In 1986, a diversion road was opened to traffic; it lies in the middle of

the valley on the other side of the river (the initial road ran along the
foot of the slope) and it is, therefore, safe from rockfalls but remains
exposed to rockslides of more than 1 or 2 hm3 (Fig. 9.4). In 2003, a
diversion gallery was driven in the opposite slope for the Romanche
River.
Figure 9.4. Sketch of the main features of the site of Les Ruines: see
particularly the original trace of the RN 91 (ancienne route nationale)
and the dam (Merlon).

9.2 REGIONAL FRAMEWORK

9.2.1 Climatic and water conditions

The annual rainfall in Grenoble is nearly 1 m per year. Average
rainfall (and snowfall) on Mont Sec can be estimated at about 1200
mm per year. High intensities occur mainly during the autumn
rainfall, but water infiltration may be important during snow melting
periods. No springs are present on the landslide slope.

9.2.2 Regional morphology

The higher peaks in the vicinity of Les Ruines are the Pic de
1’Oeuilly (1500 m a.s.l.), north of the valley, and the Grand Serre
(2140 m a.s.l.) to the south.
Page 256

The Romanche valley has an average east-west orientation. It is a

typical glacial valley originating near the Lautaret Pass. In front of
Les Ruines, the valley is rather narrow (220 m) but enlarges
upstream (Séchilienne village) and downstream (L’lle-Falcon, small
village) (Fig. 9.5).

Figure 9.5. General view of the Séchilienne landslide (from uphill). In

the foreground, the small village of Grand Serre in Séchilienne (see
also Fig. 9.2).

9.2.3 Regional geology and structural setting

The landslide takes place in the external part of the crystalline
Belledonne Massif in the French Alps. It consists mainly of
micaschists resulting from the metamorphism of old sedimentary
deposits (proterozoic or early paleozoic age). The main metamorphic
episode is the hercynian one; the alpine metamorphism is less
pronounced (INTERREG 1996; Pothérat & Alfonsi, 2001).
Several tectonic episodes folded and faulted the massif during the
hercynian and alpine tectonic phases. The quaternary uplift of this
part of the alpine chain is not yet complete.

The geological unit (“série satinée”), which includes the Séchilienne

landslide, is bordered on the western side by the Vizille fault and on
the eastern side by the “median” syncline; both structures have a N
20° azimuth, which is the elongation direction of the Belledonne
Massif (Fig. 9.6).

9.3 HAZARD ANALYSIS

9.3.1 Geomorphological analysis

The Romanche River flows from east to west in the alluvial plain.
Geophysical investigations performed in the valley in front of Les
Ruines showed that the bedrock lies nearly 100 m below the alluvial
plain.

The south-facing slope of Les Ruines has an angle of 45° in its lower
part and about 20° in its upper part and near the crest. The main
feature of the slope is the sagging of Mont Sec, bounded
Page 257

Figure 9.6. Geological sketch of the Séchilienne area.

by a 40–50 m high scarp with an elliptic shape. Several trenches N

70° cross the upper part of the landslide. The frontal zone (south-
eastern part of the landslide) is presently wholly disrupted in its
upper part, with deep fissures in the ground which are broadening
and collapsing. Rockfalls and small debris flows run downhill from
the frontal part (corridor of Les Ruines) and frequently reach the
abandoned stretch of the RN 91.

Ancient quaternary glaciations (Riss and early Würm) covered, the

Mont-Sec mountain and the present Romanche valley. During the
Würm II period (−90 000 to −40 000 BP) the glacier covered the
Mont-Sec mountain and filled the Romanche valley up to 1200 m
a.s.l. During the Würm III period (−35 000 to −25 000 BP) the glacier
was not higher than 600 m a.s.l, which is the level of the bottom of
the moving zone.

9.3.2 Local geology and structural analysis

The metamorphic rocks (greenschist facies) that make up the slope
are rather heterogeneous: micaschists (ancient pelitic rocks and
sandstones) with variable quartz content producing variable
resistance to weathering and erosion. To the north-east of Mont Sec,
the metamorphic rocks are locally covered by carboniferous
conglomerates or Mesozoic sediments.

Two main structural alpine directions are present at the site:

– strike-slip faults striking N 20–30° to the right, parallel to the

“median” syncline,

– strike-slip faults striking N 120–140° to the left, conjugated to the

preceding ones.

These faults divide the moving zone into four parts (Fig. 9.7). The
geodetic measurements carried out over the past 20 years are fully
coherent with these structural features.

The elliptic sagging of Mont Sec may be related to a cone-sheet

structure due to a deep magmatic intrusion, possibly of Permian age
(Fig. 9.8). This may be correlated to the radial and concentric filonian
structures with sulphur mineralisation.
Page 258

Figure 9.7. Structural sketch of the Séchilienne landslide. Zone A is

stable. Zones B1, B2, C and D are formed by two conjugated faults.
Thick lines: main faults. Thin lines: secondary faults.
Figure 9.8. Geological section through the “sagging” of Mont Sec.
Page 259

9.3.3 Investigation and monitoring

The geological survey included aerial photo interpretation, field
observations, geological mapping, geophysical exploration and the
boring of a 240 m long horizontal gallery (710 m a.s.l.). The cost of
the gallery was about 380 k€.

Basic monitoring of the slope was set up by the Centre d’Etudes

techniques de 1’Équipement de Lyon (CETE) in 1984 (Evrard et al.,
1990; Duranthon et al., 2003) and it was then progressively
increased: wire extensometers, geodetic measurements of points on
the slope and in the gallery, tiltmeters in the gallery, tacheometers
and a new technique based on microwave radar (measures of
distance from the opposite side of the valley: see Fig. 9.9), rain and
snow gauges. Acoustic emission has been tested but did not give
reliable results.

Some of these instruments are connected to an automated data

recording and transmission system (33 extensometers, 54 distance
measurements). Others are periodically surveyed, e.g. the geodetic
measurements in the gallery.

The monitoring made it possible to define a zonation of the moving

area (see § 9.3.4) and to make some inferences about the
mechanism of deformation (see § 9.3.5.1): exclusion of simple
Figure 9.9. Microwave radar located on the opposite slope.
Page 260

sliding on a plane or circular surface for instance, proposal of a

complex deformation including some kind of toppling.

The history of the velocities measured by extensometer A 13 is

shown in Figure 9.10. One can see, apart from the dispersion of
values due to the actual accuracy of the extensometer, the large
seasonal variations and a trend to increasing mean annual
velocities.

Geophysical prospecting (electrical and seismic) has recently been

performed by Grenoble University. The results are not yet completely
available but it seems that the electrical and seismic properties of the
rock are quite changed by the deformation process, in comparison to
those of the intact formations.

Figure 9.10. Rate of deformation of the A 13 extensometer

(mm/day), located in the frontal zone. Day 2000=1990/12/05.

9.3.4 Danger identification

The hazard analysis is based on the existence of moving zones of
various dimensions and velocities:
– the most active frontal zone (about 3 hm3 with velocity of 0.15 to 1
m/y) periodically releases rockfalls (10–100 m3) through Les Ruines;
larger rockfalls (104 to more than 106 m3) are expected to occur
over the next few years,

– an active volume of 20–25 hm3 (0.05–0.15 m/y),

– a large slowly moving zone (the ancient sagging): 0.02 to 0.04 m/y.

An important question is to know the volume of the frontal mass

which could be destabilised in the short term. Many estimates have
been made, ranging from 2 to 5 hm3. In order to obtain the best
estimate, it is assumed that the frontal zone is limited by a diamond-
shaped polygon on the surface of the slope and by a basal surface
dipping 30° towards the valley. This results in a volume of 3 hm3.

This volume is used as a basis in the short-term reference scenario

for the local authorities.

9.3.5 Geomechanical modelling

9.3.5.1 Mechanical triggering model
Monitoring of the slope and of the survey gallery showed some
toppling movement of large rock masses. Numerical modelling with
the distinct element method (DEM) yielded some interesting
Page 261

results despite the simplified 2D geometry and structure (only two

families of discontinuities). It was shown that glacial melting induces
slope deformation consisting of toppling in the middle of the slope,
bulging of the lower part of the slope and sagging of the upper part,
creating a major scarp at the top of the slope (Fig. 9.11).

Figure 9.11. Vertical displacement as given by a DEM numerical

analysis (UDEC code).

There is no need of a basal failure surface to obtain such

deformations, which reflect the postglacial evolution of the slope
fairly well. But the reactivation of the deformation observed since
about 1950 is not explained by this type of model, which stabilises
after an equilibrium deformed state has been reached; new
processes (weathering and fracture growth) must be added to the
model to obtain the recent evolution. The present deformation
combines some toppling of the frontal zone (increasing from west to
east), which induces the deformation of the sagged upper part. The
geodetic measurements in the survey gallery confirm the toppling
movement of large rock masses separated by weathered clayey
joints (Fig. 9.12). One can notice that the end of the gallery is not in
the stable rock mass (it is moving slowly).

An empirical hydrogeological model has been proposed which

makes it possible to find a relationship between the velocities of the
frontal zone of day n and the water input (net rainfall and snow
melting) of days n−1, n−2, etc. including 40 days of data. Recently,
however, the reaction of the moving mass to precipitation seems to
be quicker, probably due to the deformation process and increase in
permeability.

9.3.5.2 Mechanical run-out models

Various numerical analyses have been performed with different
hypotheses for the unstable volume. Due to the narrowness of the
valley at the foot of Les Ruines, the river may be dammed if a 3 hm3
failure volume is assumed (it may be noticed that the estimated
volume of the frontal mass is about
Page 262

Figure 9.12. Vertical displacement (mm) along the survey gallery in

relation to the end of the gallery (point 237 m on the right).

3 hm3, i.e. around the threshold between damming and no damming

of the Romanche valley!). Figure 9.13 shows the result of one run-
out model, supposing that 3.2 hm3 (4 hm3 in the debris cone due to
dilatancy) falls down in one step, a condition which strongly
influences the shape of debris because of the increasing energy
dissipation with volume (cf. the classical Heim or Scheidegger
diagram of “Fahrböschung” versus volume); however, the model
takes the progressive
Figure 9.13. Extension of the debris in the case of 3.2 hm3 failure
volume (from R&R consultants): the valley is assumed to be
completely dammed. Topography: IGN map.
Page 263

change of topography during the spreading process into account.

For larger volumes, the steep opposite slope will induce the
diversion of the debris upstream in the Romanche valley (towards
the village of Séchilienne) and downstream (towards Ile-Falcon).

Figure 9.14 shows another result for the spreading of the debris
using a digital elevation model and a mechanical process with
N=1000 pieces of rock sliding one after the other, given a law of
energy dissipation.

Figure 9.14. Debris cone resulting from a 1 hm3 volume i.e., 1.15
hm3 of debris simulation (courtesy of J.-F.Serratrice). It can be seen
that the debris cone does not reach the earth protection dam in this
case.

9.3.6 Occurrence probability

9.3.6.1 Scenarios of rupture of the slope
Different scenarios have been considered over the last years,
including catastrophic failure of the whole landslide (nearly 100
hm3). Two main groups of scenarios (in relation to the volume of
rock involved) may be put forward at present, according to whether
one considers the short term or the middle—long term (Panet et al.,
2000). These groups have been defined in consideration of present
scientific knowledge. They are based on physical considerations,
such as measured velocities and opening of fractures.

– Group 1 (short term i.e. <10 years): toppling and falls originating
from the frontal zone; many rockfalls; failure of the whole frontal
zone (about 3 hm3), probably in several steps. There are two main
possibilities: continuation of the current behaviour (rockfalls of a few
cubic meters to hundreds or thousands of cubic meters) or a
significant rockslide involving the whole fastmoving zone: the volume
could be several million m3 (2 to 3 in the more recent evaluation).

– Group 2 (middle term: 10 to 50 years): the entire high or moderate

velocity zone collapses. The volume could be 20 to 25 million cubic
meters. Of course, intermediate scenarios could also occur.

9.3.6.2 Consequences according to the volume of the rockslide

In the same way as for the definition of the two groups of slope
failure mechanisms, scenarios of the failure consequences may be
defined according to the influence of the fall of debris in the bottom
of the valley:

– Scenario 1: rockfalls of limited volume (e.g., less than 1 hm3) do

not dam the valley. The river will use the emergency riverbed. The
debris cone does not reach or hardly reaches the RN 91.

– Scenario 2: a significant rockfall (3 hm3) creates a small dam

through the valley and a lake upstream. The dam is not very high
(about 10 m) and the storage of water in the lake is about 200 000
m3. A moderate volume of materials covers the RN 91.
Page 264

– Scenario 3: a large rockslide occurs (about 5 hm3) and creates a

large dam (approximately 15 to 20 m). The water storage is about 3
million cubic meters. The filling rate of the reservoir depends heavily
on the discharge of the river, and may last between one or two days
and a few hours; the upstream village of Séchilienne is partly
flooded. The overtopping of the dam can cause its own destruction
and then generate a dramatic flood which reaches the town of Vizille
in about ten minutes.

– Scenario 4: a catastrophic rockslide occurs (10–25 hm3) and

creates a very high dam (40 to 50 m). The reservoir (water storage
between l0 hm3 and 20 hm3 of water) is filled within a few days (less
than a day if it occurs during a rising period of the river). The village
of Séchilienne is flooded. The overtopping of the dam and its
destruction generate a dramatic flood, which reaches Vizille in about
ten minutes and the industrial suburbs of Grenoble in 30–40
minutes.

Scenarios 1 and 2 are linked to the Group 1 slope behaviour.

Scenarios 3 and 4 are related to Group 2. The occurrence of
Scenarios 1 and 2 is considered to be probable in the next 10 years.
Taking account of the current evolution of the slope, the occurrence
of Scenario 3 is not considered to be realistic before about 10 years.
It does not seem possible to define probabilities of failure for
Scenario 4 because it is very difficult to get a precise idea of the
morphology and of the hazard degree after a first rockslide involving
around 3 hm3.

The possibility of windblast due to the rock avalanche is

controversial; in any case it would only be significant in the case of
Scenario 4.

9.3.6.3 Scenarios related to the planning of the prevention measures

The authorities in charge of the emergency planning have to make a
distinction between the following: closure or not of the RN 91,
flooding or not of the plain upstream, flooding or not of Vizille, etc.

Many studies have been devoted to this subject. In particular,

evaluations of the risk of flood downstream have been carried out
with three values of water storage (3 million, 9 million and 20 million
m3).

From the knowledge of the slope and the results of these studies,
various scenarios were defined. These scenarios make it possible to
plan the actions of the authorities, either preventive ones (i.e., land
use control) or curative ones (emergency plans). For each scenario,
the authorities take into account a reasonably elevated evaluation of
the risk, to guarantee a high level of safety, while not spending
financial resources in a useless or premature way.

9.4 QUANTITATIVE RISK ANALYSIS

9.4.1 Elements at risk

The main elements at risk are, first, those affected by the run-out of
the debris (hypothesis of 4 hm3 or more): the RN 91 (nearly 10’000
vehicles/day, more than 20’000 during the winter holidays), the small
village of Ile-Falcon (initially 90 houses), a paper factory (initially 51
employees) and a small electric power plant. Considering the
importance of the risk to the people, in 1997, the French State
decided to purchase all the installations directly threatened by the
rockfalls. This operation is about to be completed at present (2003).

Other zones are exposed to secondary (indirect) phenomena, in the

case of Scenarios 3 or 4:

– In the case of high damming and rising water behind the natural
dam formed by the debris: flooding of the major part of the village of
Séchilienne (670 inhabitants),
– In the case of overtopping and rapid erosion of the dam: flooding
downstream in the Romanche valley affecting the town of Vizille and
its surroundings (10’000 inhabitants) and chemical industries near
Grenoble, etc.

Economic consequences could be very important if the road traffic is

blocked because no other route of reasonable capacity exists during
the winter. About 10’000 people live in the upper valley. The winter
tourist activity of the ski resorts in the upper valley (Les Deux-Alpes,
l’Alpe-d’Huez, etc.: 80’000 beds) provides an essential income to the
regional economy.
Page 265

There are also large economic consequences according to

Scenarios 3 and 4 for the industrial facilities located downstream
(chemical industries, electric power plant, etc.).

The environmental consequences could also be very heavy: 4 km

downstream of Les Ruines there is a drinking water supply zone
serving 200’000 people. This pumping area would be flooded in the
case of Scenarios 3 and 4. There are also chemical industries
downstream: these Seveso type industries could be hit by the flood
in Scenario 4 and this could have a huge environmental impact.

According to the given scenarios, the following consequences may

be considered:

Scenario 1: SMALL ROCKFALL (1000 to 1000000 m3)

This scenario may actually be considered as a non-event: it would

(probably) be a first step—though already significant—which would
not result in real consequences for property, as described in Table
9.1.

Table 9.1. Consequences of Scenario 1.

Romanche Cone scree reaches the river, likely locally diverting the flow
River towards the earth dam, inducing a beginning of diversion by
the new bed. Pollution (suspended matter) of the Romanche
River.

RN 91 Fall of blocks, temporary closure (several days); no

significant damage.

ConsequencesDust. Closure of short duration → economic impact on the

downstream ski resort of Oisans (80’000 tourist beds).
and upstream
Scenario 2: ROCKFALL (3 million m3)—SMALL DAM

Scenario resulting in a cone scree totally sealing the bed of the

Romanche River and damming the valley at a moderate height. A
lake of roughly 200’000 m3 of water would form, as described in
Table 9.2 (Fig.9.15).

Table 9.2. Consequences of Scenario 2.

Romanche Temporary cut, diversion and divagation in Ile-Falcon,

River pollution (suspended matter).

RN 91 Road covered and destroyed (along about 80 m) by the

rockfall. Closure of long duration (several months):
significant economic impact on the ski resort of Oisans.

Inhabited Dust, etc. Possible evacuation during the crisis for the closest
zones areas. No significant consequence.
upstream

Nearby Approximately 1 km of road destroyed and a bridge

downstream damaged. Divagation of the Romanche river flowing out of
(1–2 km) the failure point in the dam.

Water supply Possible effects on drinking water for 200’000 inhabitants.

of Jouchy

Side dams in Possible localised degradations, without overflow.

Romanche

Inhabited Dust, etc. Temporary increase in the flow of Romanche

zones River, without significant consequence.
downstream
(2–5 km)
Other Possible damage to a small hydroelectric power plant.
consequences Damage to local power lines.

Scenario 3: ROCKSLIDE (5 million m3)—MEDIUM SIZED DAM

Scenario with significant hydraulic risk of flood downstream (city of

Vizille). Formation of an already high dam, with the creation of a
significant lake (approximately 3 million m3), as described in Table
9.3.

Note: The diversion tunnel now makes it possible to be protected

against a failure of the dam if the discharge of the Romanche River
remains low.
Page 266

Figure 9.15. Lake formed in a 3 hm3 rockfall hypothesis. See also

the debris cone across the valley (see § 9.3.5.2), overflowing the
catch dam, and the location of the eastern portal of the diversion
gallery.

Table 9.3. Consequences of Scenario 3 (in case of dam failure).

Romanche Temporary cut, diversion and divagation in Ile-Falcon,

River important pollution (suspended solids).

RN 91 Road covered by the debris: about 150–200 m long and 15 m

high. Closure of long duration (several months): significant
economic impact on the ski resorts of the Oisans area.

Inhabited Dust, etc. Evacuation during the crisis (several weeks) for the
zones closest sectors. Possibility of flood for some of the houses
upstream located at the lowest level.

Immediate Approximately 1 km of road destroyed and a bridge

downstream destroyed. Divagation of the Romanche River, damping out
(1–2 km) the peak flow following the dam failure.

Water supply Pollution of drinking water for 200’000 inhabitants. Partial

area of destruction of the pumping wells and pipes.
Jouchy

Side dams in Overflowed, destruction (4–5 km).

Romanche
River

Inhabited Dust, etc. Evacuation during the crisis (several weeks) for the
zones closest areas: 5’000 to 15’000 inhabitants affected. Damage to
downstream numerous buildings (several hundred). Damage to roads,
(2–5 km) bridges (destruction) within a radius of 20 km (local or
national level).

Other Possible damage to several hydroelectric power plants.

consequencesDamage to local and regional power lines and other electric
installations (30’000 inhabitants affected). Flood of chemical
factories, risk of chemical pollution, cessation of activity for a
long duration.
Page 267

Scenario 4: ROCK AVALANCHE (20 million m3)—HIGH DAM

The consequences of the failure of the dam are assumed to be very

significant, not only within the close downstream valley (areas of
Vizille and Jarrie), but also in the urban area of Grenoble. This
means that more than 100’000 people would undergo damage due
to a significant flood. Buildings, chemical factories, electric
installations would be damaged or destroyed. Roads and motorways
of national importance and railways would also be destroyed or
damaged. The importance of the zone in question makes it difficult to
define an exhaustive list of the potential consequences of this
scenario.

This scenario highlights very heavy consequences. However, it is

currently not considered to be possible for a long time (several
decades), because of the need of undergoing several stages of
deformation in the slope before it could occur.

9.4.2 Vulnerability of the elements at risk and expected

consequences
Table 9.4 roughly shows the vulnerability of the main exposed
elements for the various scenarios, as far as direct and immediate
effects are concerned. The scale that is used is a relative scale and
ranges from 0 to 100 (conventionally the value of 100 has been
attributed to the catastrophic consequences of Scenario 4 in the
Grenoble area). Finally, the table presents the situation as it was five
years ago: today some preventive actions have already been
implemented (see § 9.5).

Table 9.4. Vulnerability of elements at risk.

Element ScenarioScenarioScenarioScenario
1 2 3 4
National road RN 91 and other roads, 0,05 0,2 1 5
railway

Upstream economic consequences 0,1 0,8 10 20

Upstream local consequences (flood) 0,05 0,2 0,5 5

Small village of Ile-Falcon (94 houses, 1 0 0,5 2 2

school, 1 restaurant)

Area of Vizille and surroundings 0 0,1 1 5

Local factories (a paper factory and a 0 0,05 1 5

small hydroelectric power plant)

Water supply 0 0,2 1 5

Grenoble and surroundings 0 0 1 100

9.4.3 Risk assessment

No precise evaluation of the total cost of the scenarios has been
made. One can roughly evaluate the cost of Scenario 3 at several
hundreds of millions of euros.

One should also add that the social and economic impact already
exists: the development of the Séchilienne village has been stopped
and it is also probable that the chemical industries of Claix and Jarrie
downstream in the Romanche valley are no longer investing at their
sites. Moreover, large expenses have been devoted over the past
twenty years to investigation and monitoring of the landslide.

9.5 REGULATION AND RISK MITIGATION MEASURES

ALREADY AVAILABLE

Taking account of:

– the volume of rock material involved,

– the jointed nature of the rock mass,

– the hydrogeological complexity,

Page 268

no stabilisation technique seems to be either technically or

economically feasible. We can mention that even a solution of
massive blasting has been assessed.

The new alignment of the RN 90 (which was initially located on the

right side of the Romanche River) was opened in 1986; it has been
diverted to the left side of the river and two new bridges had to be
built. A diversion channel for the Romanche River has been
prepared and is protected by a catch dam. It is supposed that these
protective measures would be efficient for a failure situation involving
not more than 1 or 2 hm3 of rock.

The houses in Ile-Falcon have been removed and people (200)

relocated in safe areas (Fig. 9.16). A factory has been closed. The
government has purchased all these buildings from the private
owners (cost: roughly 20 M€).

Figure 9.16. Ile-Falcon viewed from the frontal zone of Les Ruines.

A 2 km long exploration tunnel (Effendiantz et al., 2000) has recently

been completed (10 M€); it was driven on the left side of the
Romanche valley (see Figure 9.4). It could be used in case of
catastrophic failure by the Romanche River but the discharge
capacity (60 m3/s) is only about a third of the maximum annual
discharge. The need for a diversion suited to the floods of the
Romanche River appears at the present time only for the scenarios
likely to occur in the long run. One of the aims of monitoring is to
determine the time when the implementing of this project would have
to be decided.

The first emergency plan was worked out in 1993. A second version
was prepared in 1999, corresponding to Scenario 3. Because of the
recent analyses of the slope stability, and of the need for having an
action plan adapted to the short term risks, a new version has
recently been prepared. It corresponds to Scenarios 1 or 2.

Real time monitoring is a major component of this plan. Data from

extensometers, distancemeters and rain and snow gauges are
transmitted every 2 hours to the CETE in Lyon. In case of alert,
safety measures would be progressively taken: closure of RN 91,
call for the on site presence of experts, evacuation of people from
the most dangerous zones, information for people living
downstream, etc. The cost of monitoring is evaluated at 400 k€ per
year (equipment value: 300–400 k€).
Page 269

In the case of the occurrence of Scenario 1 or 2, the RN 91 would

have to be reopened, probably after some clearing of debris and
protective works. The main question would be to provide safe
conditions for the workers and then to decide if the residual danger
of rockfalls is over.

9.6 CONCLUSIONS

The Séchilienne and the La Clapière (Alpes-Maritimes) landslides

are the two major active landslides in France. At both sites, the
landslide risk has been efficiently managed for more than s25 years,
in spite of all imaginable types of difficulties. This has represented a
scientific challenge as much as a political and financial headache…

Our knowledge of these large slope deformations is rather poor,

therefore, there is no certain prediction of the future evolution of the
system: the best approach is the construction of scenarios, the
assessment and evaluation of these scenarios and their continuous
updating in order that the local and regional authorities may take
appropriate preventive measures and organise the preparedness in
the manner most suited to the situation.

The history of Les Ruines de Séchilienne has not yet come to its
end. The scientific experts have sometimes been mistaken, but the
observations and data collected from the site allow continual
understanding of the phenomenon and our ability to define the most
probable scenarios, and therefore to prepare the mitigation decisions
which are the responsibility of the political authorities.
Page 270

This page intentionally left blank.

Page 271

10
Specific situations in other contexts

10.1 ROCKSLIDE HAZARD MANAGEMENT: CANADIAN

EXPERIENCE

Oldrich Hungr
University of British Columbia, Vancouver, Canada

ABSTRACT: Over the last 30 years a number of hazard and risk

studies involving large rockslides have been completed in Western
Canada. Several examples are summarised. Probably the most
detailed hazard studies have been completed by BC Hydro, the
publicly-owned electricity producer in the Province of British
Columbia. Some hazard studies have also been commissioned by
government regulatory agencies. No standard methodology can as
yet be recommended for such work. Each case must be investigated
on its own merits. Detailed geological work and monitoring is an
important part of hazard studies. Recent developments in run-out
assessment techniques will be a useful additional tool.

10.1.1 In troduction
Some 1.2 million km2 of Canadian territory, an area equal to more
than half of Western Europe, contains mountainous terrain with
sustained local relief in excess of 1000 m. Glacial erosion during the
Holocene and Early Pleistocene left deep valleys and oversteepened
mountain slopes. Large rockslides are relatively common in this
area. Surprisingly, damage due to large rockslides is not a common
occurrence. Evans (2003)—see references at the end of the book—
reports that 238 lives have been lost to rockslides and rockfall in all
of Canadian history, more than half of which occurred in only two
events (the 1903 Frank Slide in Alberta and the 1915 slide at Jane
camp, near Vancouver). The probable reason for this rather limited
mortality is the remoteness and low population density of the
mountain lands. The Pandemonium Creek rock avalanche (Evans et
al., 1989) is an example that illustrates the point. The rock avalanche
probably occurred in 1958 in a mountainous area approximately 300
km north of Vancouver. Several million of m3 of rock failed from a
cirque headwall, passed over a glacier and traveled 11 km along a
mountain valley, devastating some 8 km2 of area (Fig. 10.1.1). Not
only were there no fatalities, but the landslide was not even noticed
by anyone! The extensive damage to the natural landscape was first
identified on air photos only in the 1970’s.

Nevertheless, the potential for damage is now increasing with

growing population and increasing development of the area. Future
damage reports could be considerably more serious than those in
the past. Government and private agencies responsible for
development regulation are aware of this and, more and more
frequently, call on engineers and geologists to assess the hazards
and recommend suitable remedial measures. This article
summarises some examples of this process.

10.1.2 Examples
10.1.2.1 Large rockslides on hydroelectric dam reservoirs
B.C.Hydro, a crown corporation of the British Columbia Provincial
Government, owns 60 hydro-electric dams. The reservoirs behind
these dams form thousands of km of bank slopes in mountainous
Page 272

Figure 10.1.1. Aerial view of the Pandemonium Creek rock

avalanche in British Columbia. The landslide path is 11 km long.

terrain. Ever since the Vaiont disaster of 1963, there has been
awareness of the danger posed to dams by large landslides
displacing reservoir water. In the case of the dams on the Columbia
River, for example, a landslide-generated displacement wave could
potentially destroy a large earth or rockfill dam, generating a
cascade of dam failures along a river corridor 1000 km long,
reaching into the territory of the United States. Consequently,
B.C.Hydro operates an intensive program of investigation and
monitoring of reservoir slopes. Over the last 30 years, more than
2000 km of slopes have been subjected to systematic landslide
hazard evaluations by B.C.Hydro and its consultants (Imrie and
Moore, 1997).

Some very large specific rockslides have been studied in depth and
subjected to remedial construction. The first of these, the Downie
Slide, situated on the Columbia River 70 km upstream of Revelstoke
Dam, is one of the largest recognized landslides in Canada. It
involves a mass of 1.4×109 m3 of metamorphic rock, situated above
a fault dipping at about 20° towards the reservoir and more than 200
m beneath the slope surface (Imrie et al., 1992). The rockslide
underwent a displacement of about 100 m in prehistoric time, but
was in a dormant state until disturbed by water impoundment. During
filling of the reservoir, a part of the slide mass began to move at
rates of up to 100 mm/year. Although such slow movement posed no
direct danger to the dam, there was concern about a Vaiont-like
sudden displacement. B.C.Hydro carried out a detailed investigation
with drill holes, testing and detailed suface mapping and then
designed a dewatering system with a tunnel and fans of horizontal
and inclined drainage holes. The total cost of the drainage system
was 30 million $CDN. The drainage decreased the slope movement
rates to about 10 mm/year. Design changes were implemented at
the dam, in order to accommodate a displacement wave, predicted
by physical modeling. Full time remote monitoring of displacement
rates and piezometric levels is continuing.

Two somewhat smaller rockslides were identified by systematic

mapping a short distance above the Mica Dam on the same river.
Both were subjected to detailed investigations during the 1990’s. In
one case, the drilling information showed no adversely-oriented
structure in the slope and the slide is believed to be benign. In the
second case, called the Dutchman’s Ridge landslide, a shear zone
dipping 29° downslope was found (Moore and Imrie, 1992). The
slope was drained using a drainage system similar to that of the
Downie Slide, at a similar cost.

The small Wahleach power plant near Chilliwack, British Columbia,

was in operation for 40 years, when cracks and buckling of an
underground penstock were discovered during a routine
Page 273

inspection. Subsequent detailed investigations showed that massive

toppling of the granitic slope, separated by closely-spaced anaclinal
joints, was occurring (Moore et al., 1991). An overall failure of the
slope would endanger the powerhouse and enter the floodplain of
the Fraser River, destroying a major highway, railway and pipeline
corridor. The slope movement rates were reduced by improving
drainage, by means of relocation of the power tunnel and penstock.

In each of the above cases, the slow slope movements observed

were not of significant direct consequence. The main concern was
about the possibility of a brittle catastrophic failure, endangering the
area at the foot of the slope. The responsible authorities and their
review boards concluded that there was insufficient proof of the
absence of potential for such catastrophic detachment.
Consequently, remedial measures were implemented at very
considerable costs.

10.1.2.2 Cheekye River valley, Mt. Garibaldi

Mt.Garibaldi is a dormant volcano located approximately 100 km
north of Vancouver. Following the last period of volcanism at the end
of the Pleistocene glaciation, the stratovolcanic edifice was deeply
dissected by erosion. The most prominent of the erosion features is
the valley of the Cheekye River, whose headwall cuts nearly to the
summit of the 2600 m high mountain (Fig. 10.1.2). At the outlet of the
valley, only some 50 m a.s.l., there is a large recent fan built of
sediment transported by the Cheekye River. Both ridges of the
Cheekye Valley exhibit extensive sets of tension features, consisting
of series of normal scarps and tension cracks (Fig. 10.1.3). Both the
ridges and the high headwall are composed partly of uncemented
volcanic breccias of Pleistocene age and partly of foliated and
deeply altered metamorphic basement rock. Small and medium-
scale rock instabilities are very common on the upper slopes of the
valley and several small debris flows have occurred on the Cheekye
River during the last 100 years.
The Cheekye Fan, with a surface area of about 10 km2, is a very
desirable area for urban development, being close to the rapidly
growing community of Squamish. During the 1970’s, several major
developments, including a hospital, were being planned here.
However, excavations revealed that the fan contains deposits of
large debris flows, originating in the Cheekye Valley. A study of

Figure 10.1.2. The headwall of Cheekye valley, showing a cross-

section of Quaternary volcanics in the central part of the Mt.
Garibaldi stratovolcano.
Page 274

Figure 10.1.3. A zone of cracks and scarps on the south ridge of the
Cheekye valley. The valley slope is visible in the upper left corner of
the nearly-vertical photograph.

landslide fan hazards on the fan was commissioned by the British

Columbia Ministry of Environment in the early 1990’s (Hungr and
Rawlings, 1995). This was probably the most detailed quantitative
study of landslide risk carried out in Canada. It was based on the
assumption that the large prehistoric debris flows were generated by
landslides originating from the slopes of the Cheekye Valley.

The study comprised two parts. The first was a detailed surface and
geophysical investigation of the two ridges above the Cheekye
Valley. Stability analyses based on the data collected were used to
predict the likely maximum volume of detachment corresponding to
the extent of the cracked zones. The second part was a detailed
stratigraphic study of the Cheekye Fan deposits to establish the
magnitude and frequency of past debris flows.
The stability investigation in the headwaters showed that the
cracking was caused by an incipient translational failure of the
volcanic breccia deposits over a zone of heavily altered basement
rock, just below the gently sloping contact between the two units. It
was concluded that a sufficiently large volume could detach during a
future failure retrogression, with the possibility of sending several
million m3 of rock into the valley. Damming of the Cheekye stream
and subsequent erosion of alluvial and colluvial deposits from the
valley floor and sides could generate a large debris flow in the river.
However, it was not possible to predict the probability of occurrence
of such an event, especially because the tension features on the
ridge were found to be currently inactive.

The stratigraphic study on the fan, using some 60 excavated test

pits, revealed the presence of several large debris flow deposits,
interbedded with fluvial gravels. Two of these could be dated using
radiocarbon dating of organic remains buried in the debris. The
largest deposit, called the Surface Diamicton, lies on the surface of
the fan. It consists of an unsorted mixture of silty sand, gravel and
boulders, similar in texture and mineralogy to the pyroclastic
breccias of the ridges. Its total volume is about 7 million m3 and its
date of placement ranges between 1100 and 1300 years BP. A
similar, although slightly smaller debris flow occurred about 5000
years ago. A compilation of the available data, combined with the
estimate of total deposition volume during the last 5000 years,
allowed the construction of a cumulative frequency-magnitude (CFM)
curve (Sobkowicz et al., 1995).

Thus, the two parts of the study were found to be mutually

compatible and the CFM curve was accepted as a basis for a
quantitative hazard and risk assessment, assuming that the past
behaviour
Page 275

was likely to extend into the future. Using the curve, a prediction of
the probability of occurrence of three magnitude classes of debris
flows was made. With the guidance of the identified deposits, a run-
out analysis of debris flows between the fan apex and the hazard
area was conducted. Based on this, a hazard intensity map was
constructed as shown in Figure 10.1.4. Each of the hazard zones on
the map is associated with estimated hazard intensity parameters
(flow velocity, thickness of deposits) for each of the three classes of
debris flow magnitude. This map in itself can be used by the
planning authorities to determine whether a certain level of
development is suitable or not.

Figure 10.1.4. Hazard zones as identified on the Cheekye Fan below

Mt. Garibaldi (Sobkowicz, Hungr and Moyan, 1995).

Further analyses were carried out, assuming vulnerability of

structures and inhabitants and calculating specific and total risks
(Sobkowicz et al., 1995).
10.1.2.3 Hope Slide
The Hope Slide, which occurred in 1964 in the Coast Ranges,
200km east of Vancouver, contained 60 million m3 of metamorphic
rock (Fig. 10.1.5). Cracks and linears existed on the crest of the
slope prior to failure, but these were inactive and covered by forest
(S.G. Evans, Geological Survey of Canada, pers. comm.). As a
result, no warning of the catastrophic detachment was available and
four travelers on a regional highway lost their lives in the slide.

The right flank of the slide contains a large mass of rock which
moved several meters during the slide, opening large tension cracks,
but did not accelerate and remained on the slope (Mathews and
McTaggart, 1978). The British Columbia Ministry of Highways carried
out a detailed investigation of this remnant mass. Run-out studies
were conducted using a fragmental rockfall model. The highway was
relocated to the opposite side of the valley based on the results.

10.1.2.4 Frank Slide

The Frank Slide of 1903 destroyed a part of the village of Frank in
the Southern Canadian Rocky Mountains, causing the deaths of
more than 70 people (McConnell and Brock, 1904) (Fig. 10.1.6).
Page 276

Figure 10.1.5. The source area of the 1964 Hope Slide. The
disturbed rock mass is visible in the upper left corner of the image.
Figure 10.1.6. An overall picture of the 1903 Frank Slide in Southern
Alberta. A disturbed rock mass exists near the peak of the Turtle
Mountain, in the upper left part of the picture (Photo courtesy Dr.
D.M.Cruden, University of Alberta).
Page 277

As in the case of Hope Slide, a portion of the limestone ridge east of

the landslide source remained in a disturbed condition, with large
open cracks. A study of the stability of this ridge was carried out
recently and run-out estimates were made for a 5 million m3 rock
avalanche, using empirical techniques and a numerical Cellular
Automata model (BGC, 2000). The latter model is a dry frictional
model and is considered to represent a lower limit of possible run-out
distance. Development constraints have been placed on the hazard
area at the foot of the slope.

10.1.2.5 Mt. Breakenridge

Harrison Lake is an inland fjord located about 100 km north-east of
Vancouver and occupying a deep glacial valley. The lake is about
40km long. In 1992, an extensive area of slope disturbance was
discovered on Mt. Breakenridge, above the northern part of the lake.
(Fig. 10.1.7). Concern arose about the possibility of a catastrophic
failure of the disturbed rock mass, causing a displacement wave and
endangering a community at the south end of the lake, some 30km
away. A detailed surficial geological study was carried out. A parallel
hydraulic study determined that danger to the community would arise
only if the entire mass (200 million m3) were to fail simultaneously. It
was determined that the slide is a large flexural topple and that there
is no large structural feature enabling an overall catastrophic failure
of the unstable mass (Nichol and Hungr, 2002). Rapid failures are
considered possible, but involving only parts of the slope. Such
failures could create waves in the lake, but not of sufficient
magnitude to endanger the distant community.
Figure 10.1.7. The Mt.Breakenridge slope above Harrison Lake,
British Columbia, disturbed by massive toppling.

10.1.3 Conclusions
Over the last 30 years, a body of experience with the management
of large rockslide hazards has been gradually developing in Western
Canada. It is, however, a difficult field and no general
Page 278

methodology can be recommended. Each case must be considered

as a unique problem. In particular, the determination of probability of
occurrence of a catastrophic failure remains exceedingly difficult. In
a few cases, it was possible to discount the potential for a large-
scale catastrophic detachment by means of identifying a non-brittle
failure mechanism. Elsewhere, however, it was necessary to make
the (sometimes conservative) assumption that a brittle failure could
occur. Monitoring is an important part of risk reduction for large
rockslides, although monitoring programs are not often adequately
supported by public institutions. Corporations, such as B.C. Hydro,
are an exception (Moore et al., 1991).

The estimation of possible failure consequences relies heavily on the

prediction of run-out velocity and distance. None of the cases
reviewed here could benefit from a proven technique of run-out
prediction. Such techniques have only relatively recently become
available and are currently being calibrated. (e.g. Hungr, 1995;
Hungr and Evans, 1996; Hungr, 1997). Future hazard studies will
benefit from these research advances.
Page 279

10.2 LANDSLIDE RISK REDUCTION IN ECUADOR: FROM

POLICY TO PRACTICE

P.Basabe
United Nations International Strategy for Disaster Reduction,
Geneva, Switzerland

ABSTRACT: Within the framework of a multihazard project

(PRECUPA) in the region of Cuenca City, in the south of Ecuador,
geological and hydrometeorological hazards were studied, mapped
and monitored between March 1994 and December 1998, including
landslide, flood and earthquake assessments, real time network
installation and the implementation of mitigation measures. One of
the main components of the project included the detailed
identification and mapping of landslides in inhabited or productive
zones and the implementation of landslide risk reduction measures
(non structural ones); such as the development of public awareness
and community work, strengthening of local capacities, and
encouragement of public commitment including the development and
application of urban law and land-use planning.

10.2.1 Introduction
The location of Ecuador, a small country in South America located
along a convergence zone of tectonic plates, has played a principal
role in producing not only its beautiful geography, but also a high
susceptibility to natural disasters such as earthquakes, tsunami and
volcanic eruptions. Moreover, the influence of ocean currents
exposes this country to intense hydrometeorological phenomena
which cause landslides, debris flows and floods. These natural
phenomena turn into hazards as population and activities increase
without appropriate planning actions, causing a growing occupation
of land in inappropriate zones, so that the vulnerability and risk
factors continue to increase.
The economic impact of “natural disasters” in Ecuador is very
important, as shown in Figure 10.2.1 for the period of 1980 to 2001.
The gross domestic product (GDP) increase in percentage per year
normally varies between 2.0 and 4.3; nevertheless, most of the
GDP’s decrease coincides with the occurrence of the main disasters
having natural geneses in the same period. In 1982–83, the year of
the El Niño phenomenon, the GDP decreased and reached a
variation of 1.2% and −2.8%, respectively. In 1997–98 the impact of
the “El Niño” phenomenon which caused floods, landslides and
related damage was responsible for direct losses of 2’882 million
dollars (CAF-CEPAL, 2000), i.e. 15% of the GNP, which is the
highest value recorded in a South American country (Basabe, 1998).
The GDP variation decreased to less than 2.0% in 1998 and even
−7.0% in 1999.

In 1987 the decrease of the GDP (−6.0%) coincided with a mid-

magnitude earthquake in the north-east of the country which severed
the national oil pipeline into various parts. In 1988 the reexportation
of oil, including the 1987’s quota, produced an abnormal growth of
10.5%. Finally, in 1993 the decrease in the GDP (from 4.3 to 2.0% in
two years) coincided with the La Josefina large landslide which
affected the third most productive region of the country, in the south
of Ecuador.

This situation, and particularly the impact of La Josefina landslide,

motivated the implementation of a pilot project in order to study and
monitor natural hazards in a specific watershed as a geographic unit.

In the south of Ecuador, the main watersheds display loose rock

formations which produce large landslides due to intense and long-
lasting rainfall as well as to inappropriate human actions; thus,
productive agricultural zones, inhabited areas and infrastructures are
frequently affected.A serious event, called “La Josefina”, occurred in
March 1993, after a period of major rainfall, when a huge landslide
blocked the Paute and Jadan valleys (Fig. 10.2.2), in the vicinity of
Cuenca, the
Page 280

Figure 10.2.1. Comparison of “natural disaster” occurrence (floods

related to El Niño phenomena in 1983 and 1998, earthquake in 1987
and La Josefina Landslide in 1993) and gross domestic product %
change per year between 1980 and 2001.
Figure 10.2.2. Major landslide “La Josefina” which took a heavy toll
of some 80 persons and then caused intense destruction by flooding
the valley upstream and downstream.
Page 281

third major city of the country in the Southern Andes. Some 20 to 30

million m3 of rock which fell during the night produced the formation
of a large lake causing extensive inundations upstream. After an
emergency excavation of an outlet channel, some 33 days after the
landslide, the huge natural dam was partially washed out by a
regressive erosion mechanism, giving way to 170 million m3 of water
in a few hours, which caused flooding and destruction down to 140
km downstream, in the Amazonian forest (Almeida et al., 1996).

As the United Nations had declared the 1990’s the “International

Decade for Natural Disaster Reduction”, Switzerland answered the
national and international plea by sponsoring the cited pilot project
called “PRECUPA” which represents the abbreviation of “Prevention-
Ecuador-Cuenca-Paute”. This project gathered national, regional,
local and university partners as Ecuadorian counterparts of Swiss
Federal Institutes of Technology of Lausanne and Zürich, as well as
Federal Institutes in a joint cooperation, through the support of the
Swiss Humanitarian Aid and Disaster Relief Unit (now called Swiss
Humanitarian Aid Unit—SHA), which is an entity of the Swiss
Agency for Development and Cooperation (SDC).

10.2.2 PRECUPA Project aims and components

The objectives of the project included the assessment of the different
natural hazards which could affect the high and central watershed of
the Paute River, over an area of 3’700 km2 (Fig. 10.2.3) in which
some 700’000 inhabitants are living.
Figure 10.2.3. Study zone of the PRECUPA project and location of
areas where landslide and hazard maps at 1:25’000 scale were
produced. The six components of the project are indicated below, as
described in the text.

Moreover, this region is important because the lower part of the

Paute watershed contains the major hydroelectric scheme of
Ecuador which produced some 70% of the national electricity needs
at the time of the La Josefina landslide (1993).

The PRECUPA project was developed in six components or fields of

action (Fig. 10.2.3, lower part), following a systematic hazard
assessment methodology (Basabe et al., 1998):

– Establishing complementary mapping and creating databases.

– Landslide identification and mapping, including rapid and slow

onset as rockfall and debris flows.

– Landslide monitoring through geodetic networks for movement

definition.
Page 282

– Study and installation of real time hydrometeorological network,

monitoring, data analysis and flood hazard identification and
mapping.

– Study and installation of real time seismic network, monitoring,

data analysis and seismic hazard identification and mapping.

– Monitoring of contamination of the residual lake which remained

upstream of the slide after the natural dam failure.

These first five components required the planning, installation,

operation and maintenance of several specific monitoring networks
in order to obtain the data necessary for the identification and
assessment of the corresponding hazards.

But the major aspect of the project included the development and
application of non-structural disaster risk reduction measures. A
large inter-institutional cooperation was developed with national,
regional and local partners to create a multidisciplinary approach.
Above all, primary importance was given to the professional training,
in order to ensure the continuation of monitoring, analysis and
implementation of disaster reduction measures at the end of the
project. The measures also included Civil Defence strengthening, as
well as the training of local and regional professionals operating in
the application of hazard assessment and mapping within land use
planning actions, as well as the promotion of regulations and laws as
part of sustainable development and related programmes.

As the geodynamic phenomena represented the major hazard, the

detection of landslides was extensive, including geological and
geotechnical mapping, landslide and hazard maps as well as pilot
studies on vulnerability aspects. This led to the production of seven
complete maps at 1:25’000 scale which cover the most populated
areas (Fig. 10.2.3).
Only the part of the project related to landslides, as well as the
component of hazard management in land planning, will be
developed in the following chapters.

10.2.3 Assessment and mapping of landslide phenomena and

hazards
The stability of a slope is directly conditioned by the geological
nature of the materials encountered, by their geomechanical
behaviour as well as by the impact of external factors such as rainfall
that causes the saturation of the ground, earthquakes and anthropic
factors (Bonnard, 1994b). Thus, a slope will become unstable when
the intrinsic preparatory causes combine with triggering causes,
inducing a landslide phenomenon; even a small triggering cause
may be sufficient to induce a marked acceleration of the instability
phenomenon (Bonnard & al., 1995).

In the PRECUPA project, the most important preparatory causes

were determined by the geological, morphological and geotechnical
features of the slopes (presence of finely fractured marine siltstones,
deep cuts in slopes due to rivers-canyons-, intense weathering and
unfavourable dip of geological structures, leading to low
geomechanical strength parameters).

The main triggering causes are the intense and long-lasting rainfall,
as well as the anthropic actions such as deforestation, inappropriate
cuts and fills for roads, quarry exploitation and undue use of soil.

Thus, 8 specific types of instability phenomena were identified, which

can be grouped into three main categories (UNESCO, 1993):

– Slides, characterised by relatively slow movements, to which creep

and superficial erosion can be associated.

– Rockfall, characterised by discontinuous fast movements

developing mainly along planar discontinuities (dip, cracks), implying
fresh or weathered rock.
– Flows, formed of mud or debris, which can develop at a high
velocity along natural drainage channels.

10.2.3.1 Methodology applied

In order to follow a well-established methodology for the detection of
landslide-prone areas, as well as to detect their level of hazard, the
Swiss recommendations published by the Federal Offices
Page 283

of Water Economy (OFEE), of Land Planning (OFAT) and of

Environment, Forest and Landscape (OFEFP) were used (Lateltin,
1997; Loat & Petraschek, 1997). These recommendations were
adapted to the local geological context for the PRECUPA project.
Thus, some 28 maps at a 1:25’000 scale were produced, covering
900km2; in addition 20 specific landslide zones were studied and
monitored, leading to maps at a 1:5’000 scale.

The methodology implied four phases:

1. Detection of landslide-prone areas and characterisation of rock

outcrop and soil cover.

2. Improvement of available maps (especially 1:5’000 scale) and

analysis of air and satellite photographs; comparison of photographs
of different epochs in highly endangered zones; planning and
installation of geodetic monitoring networks in most active zones (64
landmarks), including crack opening measurements (Fig. 10.2.4).
Figure 10.2.4. Location of geodetic landmarks [J] and crack opening
devices [P] at the La Josefina Landslide, showing the residual lakes
after the catastrophic outflow of May 1993.

3. Preparation of phenomena maps, including location of scarps,

limits and accumulation zones, and determination of level of activity
(active, dormant, relict) and of probable depth (Turner & Schuster,
1996). The main mechanisms are specified by a letter and the major
phenomena studied in detail are identified by a code number.

4. Assessment of danger level depending on the preparatory and

triggering factors, the observed level of activity, the intensity of the
phenomenon, allowing the determination of the level of hazard in
relation to the assessed probability and intensity (Fig. 10.2.5).

The reference values of probability and intensity obtained are

introduced in standard diagrams which include both variables
(Petraschek, 1995), and which were adapted for their use in the
PRECUPA project.

10.2.3.2 Final product

The hazard map legend includes three colours as shown in Figure
10.2.5, as well as three letters for each zone. The first letter identifies
the type of phenomenon; the second, as an index, indicates the
probability of occurrence and the third, also as an index, states the
intensity of the phenomenon.
Page 284

Figure 10.2.5. Assessment of level of hazard for landslides

according to (Petraschek, 1995), adapted for the project, considering
relative values of probability and intensity. “Baja” means “low”.
“Media” means “medium”. “Alta” means “high”.

Although this codification is qualitative, the values of probability and

intensity are backed by characteristic values determined within the
project which are based on quantitative conditioning or triggering
parameters.

The index letters have the following meaning:

Qualitative Quantitative

A High 0.66<x<1.0

M Medium 0.33<x<0.66

B Low 0.00<x<0.33

10.2.3.3 Vulnerability studies for geodynamic hazards

Looking at more significant parameters to introduce in natural
disaster prevention studies, the PRECUPA project developed a
methodology applying a vulnerability factor (Mora & Wahrson, 1994).
It was used in particular in a case study in which the unstable zones
were moving up to 84 m/year, affecting a rural zone of some 13 km2
with a high hazard level, especially as it was in a process of
urbanisation; this zone called Paccha is located 10km east of the
town of Cuenca.

For the computation of the vulnerability factor, a first assessment of

the various hazard levels was carried out in the Paccha area (Fig.
10.2.7), dividing the landslide into six zones of potentially high
vulnerability in which a population census was carried out, as well as
a census of houses, economic activities and social conditions, with
verifications in the field. Such an assessment allowed a better
knowledge of the exposed zones which was expressed through four
specific vulnerability factors, namely human, socio-economic,
physical and preparation factors, the values of which vary from 0 (not
vulnerable) to 1 (totally vulnerable). In the hazard map, these four
factors, established by a standard characterisation, are shown as a
dial (Fig. 10.2.6).

The human factor considers the level of knowledge of the population

when facing an eventual hazard, considering that the landslide may
be slow (as in the Paccha case) or violent (as in the La Josefina
case), thus inducing a different level of impact. The socio-economic
factor analyses the potential damage in the field, in terms of daily
work carried out by the population, as well as by the impact on
housing, health, education, employment, economic activities and
standard of living. The physical factor considers potential damage to
buildings and infrastructures, namely roads, electric lines, sewage
and drinkable water systems, as well as schools, churches, health
centres and houses. The preparation factor considers the level of
knowledge of the population with respect to the Civil Defence
programs and the preparatory activities that may save their lives,
such as evacuation routes.
Page 285

Figure 10.2.6. Graphical representation of the factors used in the

vulnerability study.

Finally, it may sometimes be necessary to adopt a global

vulnerability value with a weighted average, considering specific
coefficients established according to the experience gained in the
studied area. This global value was assessed as low when included
between 0 and 0.4, as medium if it varied between 0.5 and 0.7 and
as high when it reached a value between 0.8 and 1.0.

10.2.4 Landslide management and disaster risk reduction measures

10.2.4.1 Awareness raising and community work
The project took advantage of the traumatic situation and shock
created by the La Josefina’s impacts to raise awareness and
contribute to the development of a culture of prevention. Every single
result, main activity and training courses were accompanied with
awareness campaigns through the media in the region. Bulletins of
information, interviews in the radio and local television were
common.

Every year the project delivered a detailed report to inform the

community and authorities on the activities, results and progress
made. The presentation of the report was organised by the
municipality of Cuenca city by using the major meeting hall of the
town. Authorities, local and regional governments and parliaments,
national institutions, universities, scientific and social communities,
as well as representatives of the neighbourhoods and of landslides
prone areas were invited and informed in detail as appropriate. On
the third year the project and its director were awarded with the
“Honorific insignia VIRREY HURTADO DE MENDOZA for the high
civil labor and technical assistance in the prevention of natural
disaster project in Paute basin (PRECUPA), City of Cuenca
Municipality and Parliament, Ecuador”, 1997.

The first two years of the project were focussed in mapping and
hazard assessment of landslides by covering 384 km2 in 1994 and
512 km2 in 1995 (Fig. 10.2.3). The communities located in main
hazard prone areas were informed through a joint effort with Civil
Defence and the Ministry of Education, regional branch. Specific
booklets and comics (40’000 copies each) were created to inform the
population.

The task of getting involved and not creating panic in the population
is not easy. The project took advantage of a specific programme
between Civil Defence (CD) and the Ministry of Education in which
bachelor students need to develop a thesis or a social work on Civil
Defence during one year. Every year some 800 hundred students
were informed on landslide assessment and trained on CD and risk
disaster reduction management. Every student had to apply her/his
knowledge for the community. The work consisted to take care of 3
or 4 families living in the landslide prone areas to inform them about
hazards and risk globally, as well as on landslide hazards, with
emphasis on disaster risk reduction and CD measures. At the end of
the social work the student delivered a report to the Ministry of
Education and CD office who took care of the follow-up.
Page 286

In the case of Paccha, the biggest landslide prone area (13 km2)
discovered (see § 10.2.3.3), the vulnerability assessment was based
on the compilation of information through the census of population,
housing, infrastructures, economic activities and social conditions
(Fig. 10.2.7). The census was developed with volunteers of Civil
Defence (CD) and the Ministry of Education, in conjunction with the
social work, to inform the population on hazards, vulnerability,
landslides and simple disaster risk reduction measures.

At the end of the project, the region of Paccha, which had been
designated within the expansion urban zone of the city of Cuenca
before the project, was redesignated as agricultural and recreational
forest area. The same designation was done for flood prone areas
along the streams; in this case the ordinance indicates longitudinal
parks.

10.2.4.2 Knowledge development

One of the main goals of the project was to contribute to the
development of local and national capacities, both by using
national/local capacities in the execution of hazard mapping,
assessment and monitoring and by training new professionals and
future decision-makers, many of them were directly active in the
project by secondments. The project also encouraged and used
interns in all its components.

During the execution of the project, more than 80 technicians and

professionals working in local or national institutions and interns
participated and were trained in: techniques for the identification,
mapping and monitoring of mass movements, use of satellite
images, hazard and vulnerability analysis, community work, land-use
planing and disaster risk reduction measures and management.
Thus the gained know-how could lead to a practical application of
the results at different levels (Basabe et al., 1998).
10.2.4.3 Development of inter-institutional cooperation and
application of results in disaster risk reduction measures
From the beginning of the project, the development and
implementation of disaster risk reduction measures were duly
considered. Capacity building was therefore one of the main tasks;
the project fostered both the participation of local and national
institutions and professionals to encourage the inter-institutional co-
operation and information sharing. This task in addition to knowledge
development (see § 10.2.4.2) were the key factors to look after the
implementation of the results in disaster risk reduction measures and
the sustainability of the monitoring networks and related studies.

• National level

The studies carried out, as well as the maps, the GIS and the
monitoring systems were handed over to the national institutions
participating in the project, namely the National Direction of Civil
Defence, the National Institute of Electrification and the National
Meteorological and Hydrological Institute (INAMHI). These
institutions have continued to use the results obtained and, in the
case of INAMHI, to be part of the maintenance, data treatment and
management of the hydrometeorological monitoring networks and
forecast.

Furthermore, the information gathered was delivered to higher

education centres and to the National Planning Council, now called
the National Planning Secretary of the Presidency of the Republic
(ODEPLAN). An agreement was signed with this institution so that
the information and maps obtained could be used and the risk factor
included in the development agenda and planning. Presently,
ODEPLAN leads a ministerial and institutional working group in order
to carry out a national plan for disaster prevention and risk reduction.

• Regional level
The studies carried out by PRECUPA project allowed the
development of other projects in the region of Cuenca, in particular
the one, financed by the European Union, which was set forth by the
institution created for the rehabilitation of the zone affected by the La
Josefina Landslide and which used the results obtained by the
PRECUPA Project. Thus, the destroyed road was rebuilt,
Page 287

Figure 10.2.7. Landslide-prone area hazard and vulnerability map in

the Paccha zone.

considering the landslide phenomena and hazard maps and

recommendations produced, so as to connect the cities of Cuenca
and Azogues, upstream of the La Josefina site, with the towns of
Gualaceo and Paute downstream of this point. In the same project,
levees and dams were built to stabilise the riverbed. The obtained
results were also useful for the Development and Redeployment
Centre for Southern Ecuador (CREA) which used the installed
monitoring systems and data.

• Provincial level
The provincial council of Azuay, namely the provincial government,
received constant support by the PRECUPA project in order to
assess the situation of landslide-prone areas in rural surroundings,
even outside the zone of study. Such advice was directly useful to
the population for opening new roads, building protection works or
providing support to local communities. The results and maps of the
project were also handed to the Provincial Council of Cañar, north of
Azuay Province.

• University level

Several activities of the project were carried out with the participation
of the regional universities, which also contributed to the training of
students and assistants. For instance, the maintenance and
operation of the seismic monitoring network, data base treatment,
hazard and vulnerability assessment and further investigations are
managed by the Cuenca University.

Up to now, three specialisation courses, in which the head of

PRECUPA project intervened, were organised on disaster
assessment, risk management, effect on environment and resisting
structures, so that a deeper awareness was fostered concerning the
sense of risk reduction.
Page 288

• Local level

The main local counterparts of the PRECUPA Project were the

Municipality of the Cuenca Canton and the Telecommunication,
Water and Sewage Company of this canton. First, they participated
in the studies, then they profited from the given courses, obtained
useful results, satellite images, phenomena, hazard and vulnerability
maps and studies, as well as some monitoring networks and
equipment of the project in several fields. Both institutions mentioned
assisted the Direction of Environment Management and with the help
of the project, the Interinstitutional Commission of Environmental
Management was created which also deals with risk aspects. This
commission implies public and private institutions working in the
canton.

Finally, the Cuenca Municipality used the results obtained by the

PRECUPA Project in order to define ordinances for the use of the
land at the level of the plot, in rural as well as in urban areas. These
ordinances were considered in the process to obtain the decision of
classification of Cuenca City as a Humanitarian Heritage by
UNESCO in 1999, as well as cultural capital of America in 2001.

10.2.5 Conclusions
The participation of national and local institutions in the project, the
knowledge development, the training of operational professionals
and future decision makers, as well as the community and civil
defence work towards a large dissemination of the results obtained,
have guaranteed the continuity of activities and the implementation
of specific risk reduction measures (Basabe et al., 1996). Thus much
of the data and results could be incorporated in municipal ordinances
and landuse planning. The conscience of the importance of natural
hazard and risk management has clearly increased, allowing a better
use of the land. This is expressed by a widespread popular slogan:
“The knowledge and respect of Nature is fundamental to protect
ourselves”.
Page 289

11
Suggestions, guidelines and perspectives of
development

Ch.Bonnard1, B.Coraglia2, J.L.Durville3, F.Forlati2

1Ecole Polytechnique de Lausanne; 2Arpa Piemonte; 3Centre
d’Études Techniques de l’Équipement de Lyon

11.1 GENERAL OBSERVATIONS DRAWN FROM THE

STUDIED CASES

11.1.1 Assessment of obtained results in the studied cases with

respect to hazard
The presentation of the six most important case studies considered
in the IMIRILAND Project shows that all sites imply a clearly
hazardous situation which may induce significant damageable
consequences in some of the studied scenarios. However, the
assessment of the hazard intensity carried out in Chapters 4–9
cannot be considered as equally reliable in all cases. Indeed, several
factors may impair the quality of the analyses developed:

– At several sites, a lack of basic data which could be technically

obtained by field observation, long-term comprehensive monitoring
and, if possible, boreholes including geotechnical laboratory tests
and groundwater measurements, but which are not available for
several reasons, implies the necessity of formulating hypotheses in
order to determine the characteristics of the landslide mechanisms
and then to model landslide behaviour. Some of these hypotheses
are soundly based on careful geological, morpho-structural and
geomorphological investigations, but the degree of uncertainty of the
resulting models is difficult to quantify.
– The major sources of uncertainty in the models considered lie in
general in the pattern and evolution of groundwater conditions,
especially for specific conditions which induce critical behaviour of
the landslide; some of the models considered do not express
groundwater pressures explicitly. Another main source of uncertainty
lies in the dissipation of kinetic energy for rockfalls, either at the
impact point of single blocks, or within the whole moving mass;
conservative assumptions will lead to extensive hazard zones at the
toe of the slope, whereas an optimistic selection of restitution
coefficients or equivalent angle of friction may result in ignoring the
residual level of hazard existing beyond the zone of impact.

– Finally, the selection of the models themselves aimed at describing

the evolution of landslide masses may induce uncertain results
which do not depend on the randomness of the material
characteristics, as was shown by the Wall Ball and All Ball models
used for the modelling of the Oselitzenbach landslide impact zone.
Even though the extent of the zone was fairly similar in both
computations, the velocity of the landslide process was quite
different, which may induce radically different consequences in terms
of infrastructure damage.

However, the difficulties in modelling natural dynamic processes

such as large landslides will not be solved satisfactorily in the short
term, especially as basic data will always be lacking due to financial
reasons. It is thus more important to express the formulated
hypotheses clearly and to build several scenarios of the landslide
mechanism in order to qualify the meaning and reliability of the
hazard zoning resulting from the modelling of such mechanisms.
Considering the complexity of the case studies selected in the
IMIRILAND project and the paucity of data related to the potential
failure mechanisms, it can be stated that the determined hazard
levels for the different scenarios are sufficiently significant to allow
the development of a comprehensive risk analysis.
Page 290

11.1.2 Relative gradation of the scenarios in the risk assessment

As it appears in all the case studies analysed, the potential
development of large landslides into catastrophic phenomena cannot
be expressed as a single, well-defined mechanism, as is the case,
for example, when a minimum factor of safety nearly equal to one is
reached for one specific failure circle in a homogeneous soil slope.
Indeed, when the complexity of the geological and morphostructural
conditions is considered, as well as the possible triggering factors
and failure sequences, it is evident that the landslide may evolve
according to various mechanisms. The quantitative probability of
each mechanism which may imply various intensities according to
the volume mobilised, is difficult to establish, except when a series of
past phenomena has been recorded.

It is, therefore, indispensable in a risk analysis to model several

significant scenarios following a well-defined mechanism for which a
correct hazard assessment can be performed, and then to attribute a
relative probability to each of the scenarios. This procedure
corresponds to real conditions as they may be observed by
morphological studies in mountainous areas: phenomena of limited
intensity (block falls, debris flows) are frequently observed at a given
site (order of magnitude of return period: 10 years), medium-sized
phenomena (rockfall, local slides) occur when ordinary triggering
conditions are very severe (order of magnitude of return period: 100
years), and huge phenomena (rock avalanches, generalized
catastrophic slides) only occur under extraordinary preparatory and
triggering conditions or a combination, such as during glacial retreat
or following a high magnitude earthquake (order of magnitude of
return period: 200 to 1’000 years).

This type of analysis including the consideration of several scenarios

is also essential to the risk analysis as it dictates not only the hazard
level, but also the number and the vulnerability coefficients of the
exposed objects. Even if an extraordinary catastrophic event has an
extremely low probability of occurrence, which could seemingly
justify that it not be taken into account (in the same way as the fall of
a meteorite is not considered as a relevant criterion in most risk
analyses), the importance of its expected direct and indirect
consequences is such that the resulting risk may well be significant,
with respect to the risk related to frequent phenomena.

It is not the responsibility of engineering geologists and scientists to

fix the criteria for acceptable risks, but that of government officials
and politicians. Nevertheless, all the possible risk scenarios have to
be investigated so that progressive prevention and preparatory
measures are taken, even if they are not all communicated to the
population to avoid the development of an inappro priate feeling of
panic.

11.1.3 Consideration of the real value of exposed objects

The notion of the monetary value of objects and beings (buildings,
infrastructures, persons, animals, economic activities, the
environment) exposed to a certain degree of hazard is very difficult
to formulate in absolute terms.

As far as buildings are concerned, a large number of procedures

aimed at assessing their intrinsic value exist based on the cost of
construction and that of the plot of land, but they are often valid for
short term evaluations only, as it is difficult to anticipate what will be
the evolution of market prices, the attractiveness of the endangered
zones and the consecutive possible annual rents as permanent or
holiday housing in the long term; the value of the buildings also
depends on their maintenance level. Therefore, a possible analysis
grid consists in assessing the value of the building at the present
cost of reconstruction, in case of total destruction, as is used by fire
insurances; in this case the influence of the market is virtually
neglected, except through the level of construction costs depending
on the level of the local economy.
But this approach is open to criticism for zones exposed to the
consequences of large landslides: indeed, if a chalet on a large slide
has been seriously affected by surface movements, it is most
probable that a new construction or even repair work will not be
allowed in the same zone until it is properly stabilised. A chalet may
even be tilted without any structural damage and completely lose its
value (see Fig. 2.3). If the considered house proves to be located in
a zone seriously exposed to rockfall, it will lose its intrinsic value
even if it is not damaged, because most people
Page 291

are not ready to accept living with such risk (unless they have good
reasons to do so, or do not believe in the threat).

In the case of the Sedrun landslide, for which the hazard level is not
considered as severe by the inhabitants and, consequently, the risks
do not affect the market prices in this tourist resort, the value
considered for the buildings has been assessed on the basis of
average construction prices in the area, which is quite appropriate as
a large majority of the houses are recent and fairly standard (3
storied chalets or 5 storied buildings with 8 apartments). But, of
course, the risk is assessed without considering the variation of
prices over time. As far as the cost related to traffic interruption, it
can be computed using the tables established for that purpose by
the railway companies and road administrations.

In other situations where the value of exposed objects is much more

difficult to assess (old buildings, complex infrastructures, industrial
buildings), it is inappropriate to try to define a real value which might
induce conflicts in the community. Therefore, relative values, as
detailed in § 3.6, are more relevant for the comparison of the risks
related to different scenarios. However, it is important to avoid
practising an automatic analysis on the basis of incomplete data
bases and to assess the relative importance of the exposed objects
carefully not only based on their possible monetary value but also
considering their patrimony value. This is particularly the case for
forests, whose economic value is now very low, but which have a
high environmental value because of their contribution to the
landscape and to the safety of slopes against snow avalanches.

With the separate consideration of social vulnerability and the use of

indicative coefficients, the value of life does not need to be
expressed in monetary terms, but the social impacts can be
analysed properly considering, in addition, the burden of an
evacuation from an endangered zone as a social impact, even if no
lives are directly threatened.

Finally, reference values for environmental vulnerability do not yet

exist apart from forest maintenance costs, but the modern tendency
to consider all ecological components as a patrimony of human ity
encourages the identification and qualification of the value of the
exposed environmental resources.

11.2 CONSEQUENCES OF THE RISK STUDIES ON LAND

PLANNING PROCEDURES

11.2.1 Possible actions to incorporate risk studies in land planning

Even though some landslide zones may have been developed over
several decades without any recognised evidence of potential
dangers, the ignorance of such phenomena cannot be the rule and
cannot serve only to cover the negligence of local authorities. So the
major challenge, once a risk study related to a large landslide has
been carried out, deals with the management of land in threatened
areas.

The possible attitudes and actions when facing such a situation are
very different when a master plan or a local plan is considered. At
the national or regional level, landslide risk situations, once identified
and recognised, are considered as local constraints which will be
duly taken into account in the planning of new infrastructures
(motorways, railway lines), as well as in specific development
programs, for instance for flood protection. But such constraints do
not substantially modify the development of projects, as they can be
solved technically and, socially speaking, do not represent a major
hindrance.

At the local level, on the contrary, large landslides may constitute a

major preoccupation and induce risks which must be taken into
account in land planning, for safety reasons as well as to comply
with the legal prescriptions. The population of developed countries
now demands better protection against risks, but at the same time
tends to ignore the basic behaviours which could prevent the
damageable consequences of specific natural hazards. Therefore,
the local authorities have to develop an appropriate strategy to face
this potential threat. Three distinct responses exist:

– To accept passively the occurrence of an unpredictable and very

rare event which the population generally does not believe in,
without assessing the potential consequences in an anticipated way.
Such a policy must be considered as unacceptable in a perspective
of sustainable development.
Page 292

– To resist and oppose physical actions to the development of a

dangerous potential event, so as to protect the primarily exposed
elements. Such an attitude can lead to the construction of protection
works, but in the case of large landslides it is often unrealistic and
too costly except, for instance, when an important diversion road
may be constructed as a tunnel below the landslide area; this is
presently being carried out in the case of the Conters-Gotschnahang
landslide for the bypass road avoiding the village of Klosters
(Graubünden, Switzerland) and it is planned in the case of the
Ceppo Morelli rockfall (see Chapter 7).

– To take into account the existence of the hazard and of the

related risks and to adapt the land use and buildings to its possible
occurrence. Such an attitude is, in most cases, the only reasonable
one provided an adequate mitigation strategy is applied which
considers the different protection objectives and analyses the
residual risks in the case of an extremely rare scenario. Implicitly,
this last option means that the population can live with certain risks
in an exposed zone, even though, from a political point of view, the
authorities state that they have taken all the required measures to
avoid any risk. Explicitly, this means that the risk level has to be
considered quantitatively in land planning policy, which is one of the
goals of the IMIRILAND project.

This appropriate mitigation strategy thus requires different types of

actions:

– Technical solutions, such as the construction of protection works

(tunnels, protective dams) or the installation of warning systems and
the implementation of an emergency plan.

– Political solutions, such as the adoption of an appropriate local

development plan by the community, specifying different hazard and
land use zones.
– Legal solutions, such as the adoption of laws and ordinances
prohibiting construction in high hazard zones and the evacuation and
expropriation of seriously endangered existing buildings.

11.2.2 Practical policy development

On the basis of the case studies examined within IMIRILAND
project, as well as considering several other experiences related to
large landslides, the local authorities in charge of the safe
development of an exposed area have to analyse the following
practical aspects:

– The construction of protection works, such as a tunnel for a

diversion road or a protective dam to limit the extension of a zone
exposed to rockfall, requires the definition of clear design criteria
related to the landslide scenario selected, according to the needed
degree of protection of the exposed objects. The determination of
the protection objectives, depending on the value and political
significance of the threatened elements, is a task for local authorities
who must understand that protection works cannot be fully efficient
in the case of extreme scenarios and that residual risks thus have to
be considered as inevitable. Protection works may induce an
exaggerated feeling of safety within the population if it is not duly
informed of the limitations of these works. They may even be the
indirect cause of an unforeseen disaster if these protection works are
damaged by an unexpected scenario which then concentrates the
destructive action of the landslide mass, particularly debris flows, on
a limited portion of land.

– The installation of a monitoring and warning system requires not

only the definition of an alarm strategy, which cannot be fully
automatic due to the negative consequence of false alarms, but the
preparation of appropriate evacuation plans and the elaboration of
alternative routes in order to avoid congestion in exposed areas and
to save lives, even if physical damage may occur. Such evacuation
plans must be worked out in collaboration with the population and
the scientists studying the landslide hazard, and tested during
simulation exercises.

– The use of the endangered zones must be adapted according to

the type of activities carried out or to the function of the existing or
planned buildings. Indeed it is neither appropriate nor feasible to
consider the exposed zones as banned areas in which no activity
(e.g. farming, sports) and no constructions (e.g. secondary road,
lifelines) should be carried out. But the users of these exposed
zones, in particular when occasional sports or leisure activities are
concerned, such as for camp grounds, must be aware of the
possible dangers and of the conditions under which they might
occur.
Page 293

– In some cases for which an evacuation at short notice might prove

to be inapplicable, such as for hospitals or retirement homes for
elderly people, a preventive evacuation and a modification of
building use must be carried out without waiting for a dangerous
situation. These buildings may be still used for other functions, which
do not induce the same risks.

– Easy access to appropriate information concerning landslide risks

and the related consequences for persons and buildings must be
organised by the local authorities, without dramatising the situation,
but without hiding the reality of the danger and its potential
development. All exposed persons, whether they reside in exposed
areas or are tourists, must be duly informed and instructed on how
they have to react in the case of a landslide event.

– Finally, in all of studied situations, it is not appropriate to apply one

single solution but it is advisable to combine several measures within
a comprehensive management strategy which has to be developed
jointly by the local authorities and the scientists or experts in charge
of the investigations on the landslide area. The efficiency and
reliability of each solution should be properly assessed.

11.2.3 Difficulties of application (population, authorities)

It is necessary, however, to be aware that when facing landslide risk
situations, particularly of very low probability of occurrence, many
difficulties arise in the planning and implementation of safety
measures. First of all, the population of the exposed area, which has
lived there for centuries, might not believe in the seriousness of the
hazard, in particular when no evident warning signs are perceptible
for non-specialists or when the proneness of a serious event
increases without a clear change of phenomenology (e.g. when
cracks are opening exponentially in a cliff without releasing any
blocks). Then, if the inhabitants are aware of a certain risk, they tend
to understate it so as to preserve future possibilities of development
on the land they own in exposed areas. Finally, in the clearly
exposed areas, for which an evacuation is required, the owners will
resist either because of a strong affective bond to their homeland, or
because the compensation proposed by the state or the insurance
companies does not reach their expectations. These difficulties
should lead the authorities to develop communication actions
involving the scientists which have to present what they know, but
also their uncertainties.

As far as the authorities are concerned, the situation varies from one
country to another, depending on the respective competencies of the
local and regional authorities for the execution of safety measures
(see § 11.4). It is clear that the municipal authorities, who have to be
re-elected every four to six years, are not induced to take drastic
safety measures limiting the use of the land, as this will displease the
threatened owners and induce a feeling of injustice between the
affected and non-affected citizens. Otherwise, they will take partial
measures and then declare that the landslide risks are “under
control”, whereas the level of safety is only slightly or locally
increased.

The fact that the adaptation of local plans in order to include new
hazard zones or modify their requirements must be approved by
municipal councils, makes it difficult to proceed to swift changes
when the nature of the phenomenon evolves. Of course, the mayor
generally has full power for the enacting of safety or evacuation
measures when a critical situation occurs, but such actions are of
limited duration. Therefore, it is not easy to reach an approved
consensus or the rapid application of restrictive measures in order to
reduce the landslide risks. But when the local authorities, and in
particular the mayor, have not taken the required safety measures to
avoid a disaster, they are legally considered to be responsible for the
damage, in particular if human victims are reported.

11.3 CONSIDERATION OF RISKS IN THE DEVELOPMENT

OF EXPOSED AREAS
11.3.1 Possibilities of reducing risks by protection works or
measures
Risks induced by large landslides such as rockfalls or slides may
theoretically be reduced by a modification of one of the five factors
which intervene in the risk computation, namely:

– intensity of hazard assessed at the fall source zone or at the

surface of the slide;

– intensity of hazard assessed in the foreseen fall impact zone;

Page 294

– probability of occurrence of the hazard;

– vulnerability of the exposed elements (physical, social, economic,

environmental);

– value of the exposed elements.

However, some of the corresponding actions, through landslide

protection or stabilisation works, prevention, reinforcement or
modification works of the exposed elements, as well as
reassignment of the exposed structures and land, which induce a
change in their value, are not always technically and financially
feasible, mainly due to the considerable size of the landslides
studied. Some induced aspects of the corresponding actions may
also appear as unfeasible due to environmental protection
constraints.

Practically, the economic feasibility of protection works depends

essentially on the importance of the physical and economic risks
involved, whereas the risk to life can be essentially limited by a
comprehensive monitoring and warning system allowing the safe
evacuation of the exposed inhabitants and the interruption of the
threatened traffic. One significant case of reduction of hazard
intensity was the stabilisation of the slope above the Tablachaca
Dam on the Mantaro River in Peru, which represented a high hazard
for the hydro-electric scheme as well as implying a potential loss of
energy production. The slope was, therefore, stabilised, among other
works, by a huge fill placed at the toe of the slope, in the reservoir,
with a total volume of half a million m3, by a series of drainage
galleries at three different levels, extending over a total length of
1’400 m, from which 3’000 m of radial drainage boreholes were
drilled. In addition, more than 400 prestressed anchors were
installed, some of which extended over a length of more then 100 m
(Carrillo-Gil & Carrillo-Delgado, 1988).
In the case of the large slides considered in the IMIRILAND project,
the reduction of hazard intensity related to the landslide mass itself
may only be achieved by extensive underground drainage, as
recently carried out at the Campo Vallemaggia landslide in
Switzerland, but only if certain favourable conditions prevail, related
to the permeability and groundwater pressure pattern (Bonzanigo &
al., 2001). In fact the knowledge of water distribution and flow in
large rock slopes is low and drainage may sometimes be an
uncertain action. In the cases of rockfalls considered, no practical
action can reasonably be developed in the source zone to reduce
the hazard.

A possible action to reduce the intensity of the hazard in the potential

impact zone is either the construction of a transversal earth dam to
limit the extension of the rockfall, as carried out at the toe of the
Séchilienne landslide (see Chapter 9), or the construction of a
longitudinal earth dam to divert the falling mass from an area at risk,
as built in several sites in Switzerland. For large rockslides, the
common solution of transversal wire nets to retain blocks is not
applicable due to the high energy of the falling rock mass.

If there is no possible direct action to reduce the probability of

occurrence of the main landslide hazard cause, i.e. rainfall, the
construction of deep drainage systems, even though they may not
stop the continuous movement of a large slide, may significantly
reduce the probability of experiencing exceptionally high
groundwater pressure along the slip surface. So, this solution
represents a significant contribution toward the reduction of the
probability of occurrence of disaster, as shown for the Downie Slide
by Patton (1984) (see § 10.1).

A major possibility of reducing risks exists by removing the exposed

objects from the impact zone, thus bringing their vulnerability to nil,
such as in the case of the construction of a tunnel for a road or
railway below the landslide. In extreme scenarios, however, such a
measure may not be sufficient and partial damage to the tunnel
portals may occur. In the same way, the evacuation of the population
in due time significantly reduces their vulnerability, but does not
decrease it to nil as enduring long-term evacuation in temporary
shelters may cause important psychological impacts. Economic
vulnerability may also be reduced by organising alternate routes for
traffic that might be blocked by the landslide mass or by avoiding the
storage of goods in exposed zones, so that carefully planned
management measures, based on the information provided by
warning systems, may significantly reduce the economic risks.

Finally, the risk may be reduced, in the case of temporary threats, by

reducing the value of the exposed elements, for instance the goods
produced and stored on the landslide mass. A well-known
extraordinary case was reported at the Cornillo landslide in Italy,
when the acceleration of the
Page 295

moving mass induced severe damage to a ham factory; major efforts

were done to evacuate all the stored meat before the building and
access road were too seriously damaged as the value of the meat
was much higher than that of the processing plant itself (Larini & al.,
2001).

11.3.2 Legal aspects related to protection measures

In order to enforce a danger zone permanently in a local
management plan, the procedures vary from one country to another,
but the approval of the legislative municipal authority is often
required, which means that the population has to be convinced of the
pertinence or relevance of the proposed limits. Therefore, detailed
information must be given by the scientists or the experts to the
municipal authorities and the population in order to justify the
proposed limits.

This is even more complicated to explain when different limits with

different criteria are set up, as is recommended in Switzerland (see
Fig. 8.11, showing the Sedrun present local management plan where
two limits (red and blue lines) for avalanche, flood and debris flow
danger zones appear). In this case, the results of the risk analysis
will be useful to justify the proposed limits, economically and socially.

According to Swiss legislation, the local management plans must be

revised every ten years so that, according to the evolution of the
situation, it is possible to adapt the danger zones, as well as the
rules which have to apply where prescriptions are only given for the
exposed land use.

In the case of France, a specific law allows the expropriation of

endangered areas by the State, which permits a satisfactory
monetary settlement for the evacuated owners (see § 9.5 and 11.4).
However, several open legal questions still arise in relation to the
obligation of the French State to carry out preventive actions if any
natural danger may possibly affect persons, whatever its probability
of occurrence, according to a decision of the State Council. In
particular, the application of the principle of precaution, which implies
the application of preventive actions even if the threat cannot be
demonstrated, may lead to an excess in terms of limitation of land
use.

11.3.3 A cceptable risks

The use of several possible scenarios in hazard analysis stems
implicitly from the fact that the maximum intensity of a given natural
phenomenon, considered as a value that would never be exceeded,
is difficult to assess; thus it is often judged as too extreme to
consider as a reference value in land planning. Consequently, less
critical scenarios are generally accepted to fix the limits of the
protection zones in the local management plans.

This practice indeed means that, when facing an extreme scenario

with a very low probability of occurrence, such as the failure of
protection dams, damage will extend to zones which were supposed
to be protected, implying residual risks. Such risks are, therefore,
considered as acceptable and must imply limited consequences,
such as material or environmental damage only, when the
evacuation of the exposed population has been carried out.

It is clear that, despite the difficulty in terms of legal reference, the

application of any methodology based on risk analysis necessarily
means that some risks will be considered as acceptable by the
authorities. Such risks can only be accepted by the population when
they are duly informed of this possibility and have clearly expressed
their will to assume the respective consequences. This situation
often occurs in mountain villages, but tourists coming from safe
areas and not aware of the potential consequences of natural
phenomena may not be ready to tolerate a risk that is indeed
unavoidable.
11.4 TENDENCIES IN RISK MANAGEMENT POLICY IN
SEVERAL EUROPEAN COUNTRIES

11.4.1 Introduction
In the face of a natural hazard, risk management can be divided in
several stages:

a) hazard assessment and vulnerability analysis,

b) risk evaluation and assessment,

Page 296

c) risk prevention (protective works, land use regulation, monitoring,

etc.),

d) crisis and post-crisis management,

e) feedback from experience.

The public authorities have to choose which kind of policy for risk
prevention and mitigation they will apply. One can define three
degrees of increasing involvement by the authorities:

• To produce and disseminate information about hazards and risks

(items a) b) and e) above): in principle the citizens or the
communities are therefore supposed to be aware of the dangers and
they are supposed to take them into account in their behaviour
(“liberal” policy);

• To make regulations (item c)—it is prohibited to…, it is compulsory

to…—in order to oblige the citizens or the community to apply
mitigation measures (“dictatorial” policy);

• To bring money for mitigation measures in order to incite and to

help the citizens or the community to take preventive (item c) or
rehabilitation (item d) actions (“Etat-providence” or Welfare State).

Generally, there exists a mixing of these attitudes in a given country,

but each of them can be more or less stressed.

The comparison of risk management in different countries brings our

attention to some particular points:

• Hazard mapping has developed in Europe since the 1970’s. It is

divided in most countries into regional mapping (typically 1/25 000
scale) and local mapping. The regional level is presented in Table
11.1.
Table 11.1. General hazard maps produced by different European
countries.

Italy Austria Switzerland France

Decision Autorità di Austrian Canton Préfet of the

maker bacino/ geological Département
Regione survey

Maps Hazard map Hazard maps Danger (indicative) Hazard map

(1:25000) (1:25000 to map (1:10000 to (1:10000 to
1:50000) 1:50000) 1:25000)

• Landslide inventories or databases are in progress in several

countries. They are a useful tool to keep the information available
and to prepare hazard maps.

• The vulnerability analysis is, in actual practice, most simplified,

generally reduced to an identification of “zones of conflicts”, where a
high hazard level and some exposed properties coexist. Experience
shows that it is only where very large landslides are threatening the
lives and property of numerous people that an actual vulnerability
and risk analysis is performed.

• The protective works are financed at different levels. Most

frequently the cost is far too high for private persons and even for
small communities. For instance:

– in Switzerland: protective works funded by the federal, cantonal,

municipal levels

– in Italy: planning in Piano di Assetto di Bacino, funding by the

Region

– in France: recent possibility of funding by “fonds Barnier” (but the

available budget in 2003 is rather small…).
• Land use regulations exist in all countries: it is probably the most
effective means of regulating development but it is a long-term
strategy which has been used in Austria and France since the 1980’s
and in Italy and Switzerland since the 1990’s. The different kinds of
hazard mapping should be used to update the municipal land use
plans:

– Italy: the “Piano Regolatore Generale” (from the “Piano di Assetto

Idrogeologico”),

– Switzerland: the “Plan d’affectation communal” (from the “Carte

des dangers” generally established at the cantonal level),
Page 297

– France: the « Plan Local d’Urbanisme » (from the « Plan de

Prevention des Risques »),

– Austria: The « Flächenwidmungsplan » (from the «

Gefahrenzonenplan »)

• In all countries, the different zones are typically:

– zones where building is forbidden,

– zones where building is permitted with conditions (preventive

passive or active techniques),

– zones without any particular problem.

In Switzerland, a fourth zone has been defined: zones of low

probability of occurrence of a major landslide, where the
preparedness for the event has to be developed (emergency plans).

• Insurance against natural hazard: in some countries the public

authorities intervene strongly:

– Switzerland: in 19 of the 26 cantons there is a cantonal insurance

for buildings,

– ance: insurance against natural disasters is compulsory (law of

1982).

In other countries, only private insurance is available.

From a general point of view, some final remarks can be drawn from
the comparison of different risk management policies in Europe:

• Legislation relating to hazard prevention has developed after the

occurrence of some large disasters. In European countries, the
legislative output corresponds to an increasing involvement of the
governmental authorities in the face of the growing social demand
for security.

• We may contrast France, which is a typical, centrally-run country

(the government agencies decide where the hazardous zones are
and what must be done), and countries like Italy and Switzerland,
where the government defines the general framework, but the
regional and municipal levels are the main players. The amount of
negotiation among the different players (gov-ernment or region,
municipality, citizens) is certainly different from one country to
another.

• The local level authority (i.e. the mayor of the community) is

generally primarily responsible for the crisis management.

• The legal framework is probably fitted to the most frequent events

but it may not be suitable for very large landslides.

• The back-analysis of landslide disasters is not as well organised as

for earthquakes, dam failures, etc.

In the next paragraphs, a brief description of risk prevention policy is

given for Italy, France and Switzerland.

11.4.2 Italy
Before the 1990’s, the policy in Italy was characterised by two
principal behaviours:

– take no action at all,

– to provide relief and rehabilitation assistance after disasters

occurred.

The Government provided payment for damage suffered by private

human activities or buildings also in known landslide prone areas.
At the beginning of 1970’s, land use management was transferred to
the regions. For example, in Piedmont the law 45/1989 regulates
land use modification and transformation in areas subject to
environmental protection. It does not define common lines on land
use planning, as each case is considered separately and
prescriptions are different.

As a result of technical and sociological advances, the concept of

prevention and risk by appropriate land use is becoming increasingly
important. The law 183/1989 introduces land use planning on a
basin scale: the Government sets the standards and aims but it does
not fix a methodology. The same law designates the Basin Authority,
whose main goal is to draw up the Basin Plan, a tool for planning
actions and rules for conservation and territorial protection.
Concerning the Po basin, the last plan adopted is called PAI: it tries
to verify the geological instability of the whole territory and to
upgrade the land use planning. Moreover, this project classifies the
territory into risk classes.
Page 298

The catastrophic event of May 1998, which caused heavy damage

and death in the municipalities of Sarno (Fig. 11.1) and Quindici
(Campania), urged the government to provide answers for
development regulation (to reduce or eliminate landslide losses).
According to a decree named “the Sarno Decree”, the Government
detailed legislative measures at the national level, including the
procedure to define landslide risk areas.

Figure 11.1. Large debris flow which affected the town of Sarno.

Another important aspect of law 267/1998 promulgated after the

Sarno Decree regards the development of “Extraordinary Plans” to
manage the situation of higher risk where safety problems or
functional damage are possible. In these areas, protection measures
must be applied, executing projects for risk mitigation. In some
regions, the actions have been extended to the whole area and in
others they have been applied only to some significant cases; in the
Piedmont Region, the Ceppo Morelli landslide has thus been
classified as a very high risk area.
The national legislation defines general principles, functions,
activities and the authorities involved. As a consequence, the
regions apply restrictions on land use through different regional laws.
In Piedmont, the local management plan includes danger/hazard
zoning in order to identify landslide prone areas and to recognise
danger/hazard zones on the basis of morphological features. For
example, the Rosone and Cassas areas are classified as higher
hazard zones.

In a state of emergency, a Piedmont regional law allows the stopping

or the suspension of development in landslide prone areas.
Consequently, new land planning must be implemented.

At present in Italy, a national inventory landslide map does not exist.

Innovations in land use regulation could be introduced by a new
project named IFFI (ARPA-Piemonte, 2004), an inventory of the
landslides on the whole national territory. It represents a very
important tool for the planners who finally have the detailed and most
complete knowledge of the landslide occurrence on the whole
territory. The recognition of different landslide types, the definition of
their relationship with human interactions, the geomorphological and
the geological structural setting, have allowed detailed territorial
zoning. As a consequence, the Geological Services intend to use the
inventory data to improve land use management and to evaluate the
hazard and risk areas at the regional scale, as well as to support
Government decisions.
Page 299

11.4.3 France
Since the 1980’s, several laws related to natural hazard prevention
have been passed. Citizens are more and more concerned with their
own safety and the idea of “fate” is no longer accepted. The
prevention of natural or technological hazards has become a
politically important topic.

In 1982, when H.Tazieff was the (first) minister in charge of hazard

prevention, a law created and organised:

– the insurance against “natural disasters” (the word disaster has

been broadly interpreted); property insured includes buildings and
moveable property (including vehicles) insured against fire or any
other type of damage (theft, etc.); the rates of additional premium are
set by the Government (12%).

– the “Plans d’Exposition aux Risques” which regulate and plan the
land use at the municipal level; they were changed into “Plans de
Prevention des Risques” (PPR) in 1995; more than 10’000
communities are involved (5000 PPR have been produced up to
now). The PPR are the main tool for the Government to encourage
better land use in relation to natural hazard. In the most dangerous
zones, building is forbidden; in zones of medium hazard, preventive
measures have to be taken.

In 1995, a new law simplified the PPR procedure and created a

possibility of expropriation by the Government when private buildings
are exposed to a short-term danger.

In 2003, a law devoted to natural and technological hazard

prevention was established. Under particular conditions, the
Government could help citizens or communities by partially funding a
local prevention policy. Moreover, the information for the citizens
about natural hazards is strengthened: for instance the mayor has to
publish information for the inhabitants every two years. In addition, a
national database related to landslides is in progress; it deals with
data about the landslides themselves, consequences and human
response. Data are freely available on the Internet for every citizen.

Finally, in France there exist powerful legal tools in order to

encourage a prevention policy. Nevertheless, land use regulations
that were developed in the 1980’s and 1990’s have a long-term
efficiency and are frequently not well accepted. Presently the public
authorities are trying to set up a more positive policy based upon
technical and financial help to favour the development of preventive
measures by citizens or communities. The case of a very large
landslide such as Séchilienne shows, however, that it is very difficult
to organise the funding for monitoring and preventive works
(investigation gallery, for instance) because of the lack of clarity
related to the relevant responsibilities of the different partners
(Ministry of Public Works and Ministry of the Environment,
Département, Commune).

11.4.4 Switzerland
The 1979 Federal Law on Land Planning (LAT) requires that the
areas seriously endangered by natural hazards be designated
specifically in cantonal master plans. Their management, as well as
the type of qualification and representation, then depends on the
cantons. But a federal recommendation has been produced in 1997
specifying how to take landslides into account in land planning
(OFAT, OFEE, OFEFP, 1997).

Several cantons have developed mapping projects in order to

identify not only landslide-prone areas, but to establish danger maps
(indeed, hazard maps) which will be considered in local
management plans; the canton of Freiburg is one of the most
advanced in this task. These maps distinguish red zones, in which a
prohibition to build will be declared, blue zones, in which
prescriptions will be enacted and only some buildings or
constructions implying limited risks will be accepted, and yellow
zones, in which the owners have to be informed of the existing
limited danger.

If such zoning is already applied extensively for snow avalanche and

flood hazards in local management plans, it is not yet common as far
as landslide risks are concerned. In the two Swiss communities
considered in the IMIRILAND project, the landslide risk was not
explicitly considered in the local management plan. But in the
planning of the diversion road to avoid the centre
Page 300

of Klosters, which was first designed at the toe of Conters-

Gotschnahang landslide, the potential difficulties for such a
construction were duly evaluated and finally a route passing partly
on the other side of the valley and partly in a tunnel below the
Gotschnahang landslide was adopted and is presently being carried
out (Bonnard & Noverraz, 2001).

In other communities affected by slow movements due to large

landslides, such as Grindelwald (Canton of Bern) or Lugnez (Canton
of Graubünden, see Fig. 2.1), the zones reserved for construction
are never located in red zones and only rarely in blue zones, so that
no major conflict arises between the promoters and the authorities.
But in the Canton of Freiburg, near the Black Lake (Schwarzsee), a
development project was rejected after a long struggle as being too
exposed to landslide hazard. Some years later, the slide suddenly
accelerated…

11.4.5 Conclusions on risk management policies

A comparison between the main features of the risk management
tools applicable to land planning allows the following conclusions
that are summarised in Table 11.2 through their respective
advantages and disadvantages:

Table 11.2. Comparison between positive and negative aspects in

risk management.

Italy Austria Switzerland France

Positive Regions Classification by the The hazard is The hazard

aspects have “Brown zones of classified in a mapping program
technical reservation”: Areas homogeneous is under way and
know how, of danger, such as way (red should be
so the risk rockfall, landslides… zone, etc.) completed in a
analysis can (statal bylaw— few years. People
 give a regulations 1976) and local
contribution Similar to the well authorities are
to risk known zones for more and more
management avalanche protection convinced that
risk prevention is
not only the
government’s
duty but also their
own’s

Negative A common Only qualitative No The PPR’s are

aspects way to system, no vulnerability strongly
define risk vulnerability costs and costs are heterogeneous.
levels does Not yet all zones taken into Nobody is really
not yet exist mapped account controlling their
application for
land use

11.5 SUGGESTIONS FOR FUTURE RISK MANAGEMENT

STUDIES

11.5.1 Hazard studies

The definition and evaluation of a landslide hazard includes several
parts:

– The precise nature and intensity of the expected phenomena: the

landslide itself and the secondary or indirect effects (air-blast
induced by the rapidly moving mass, damming of the valley by
debris, etc.),

– An evaluation of the probability of occurrence for a given time

period, generally a qualitative estimation of this probability.

With respect to cases like Rosone (Italy), Encampadana (Andorra) or

Séchilienne (France), it appears that the deformation process of
large rock slopes is not well understood yet. An effort should be
made in order to elaborate more appropriate mechanical models and
to estimate global
Page 301

(macroscopic) properties of the rock mass that could account for the
observed deformation and morphology. This clearly relies upon a
detailed structural field survey and should involve geologists in
collaboration with mechanical and numerical modelling specialists,
within multidisciplinary teams.

As for the time occurrence, a quantitative probability approach is not

suited to the large (and rare) landslides of the alpine region because
of the lack of statistical data. The prediction of a catastrophic failure
at sites like Séchilienne or Rosone is based on extensive monitoring
of the slope, which should give warning about failure a few months
(or weeks, or days) before it occurs. It is also essential to obtain
quantitative data related to the movement (direction, velocities,
dependence on rainfall, etc.) for the understanding of the
deformation process. New terrestrial, aerial or satellite imagery
techniques will probably become an essential part of the monitoring
systems of large landslides.

As for run-out modelling, several numerical tools are available

deduced from solid, fluid or particulate mechanics, but not all give
relevant results. The priority is to collect detailed information related
to past or future events: comparison of pre- and post-topography,
estimation of velocities, etc. These case histories will enable the
calibration and validation of the mechanical models.

11.5.2 Consequence studies

One of the most significant consequences of large landslides in
mountainous areas is the damming of the valley. It is, therefore,
necessary to address the question of the stability and durability of
the natural dam formed by the landslide debris. Hydrological
simulations can estimate the flooding hazards, both upstream and
downstream, in case of dam failure, but the results are directly
dependant on the behaviour and failure process (permeability of the
dam, overtopping consequences, internal erosion) of the dam. A
careful analysis of past events should be carried out in order to be
able to make reliable predictions about the dam stability.

The quantitative data related to the exposed objects may exist but
are difficult to find and to collect. In the near future, data bases and
geographical information systems will probably facilitate the
obtaining of information related to the spatial distribution, the
functionality and the cost of the main objects. As an example, the
Swisstopo database includes a GIS layer containing all of the
buildings in Switzerland. The evaluation of indirect costs, because of
the closure of a road for instance, has yet to be formalised;
economists should be involved in such analyses in conjunction with
engineers.

In the case of slow movements, such as in Conters and

Gotschnahang (Switzerland), it is difficult to evaluate the long-term
effect of the moving slope and the potential consequences on
infrastructures (tunnels, for instance). It should be possible to live
and to develop buildings and infrastructures on such sites, but a
better evaluation of the maintenance and protection costs due to the
movement should be carried out.

11.5.3 Total risk and prevention

In order to develop risk prevention, some feedback is needed from
the experience of past events, in the scientific, technical, social and
economic fields. Lessons should be learned from risk management
before, during and after the crisis. The appropriateness of the legal
framework could be assessed and many aspects should be
analysed; for instance, the changes in public opinion concerning the
risk level, which is an essential “parameter” of risk management,
should be analysed: these changes may be induced by some local
(e.g. a small but visible rockfall) or external events (e.g. the landslide
in a neighbouring country: see the influence of the 1987 Valtellina
landslide in France), by a public statement coming from a well-
known expert, by the realisation of some preventive works, etc. Cost
estimates of past events, including all direct and indirect items,
should also be made in order to assess the importance of landslide
risks better.

In the face of a large landslide, involving several million m3 of

moving rock, most prevention engineering techniques have a limited
efficiency. However, deep drainage may be considered. In
Page 302

order to optimise the drainage system, research should be pursued

concerning the hydrogeology of rock slopes in mountainous areas
(hydrogeology of discontinuous material).

As far as passive techniques are concerned, earth catch dams have

been extensively used for rockfalls with energies less than 100 MJ or
so. Data are lacking concerning the efficiency of large catch dams,
particularly their resistance to an impacting mass of debris; a better
understanding of the run-out process should give an estimate of the
induced pressure on obstacles and allow an increase in the field of
applicability of this technique.

11.6 PERSPECTIVES AND OPEN QUESTIONS

11.6.1 Relevance of overall risk analysis in development policy

The concept of vulnerability was indirectly referred to first by
engineers, in considering construction value and building design
criteria related to the level of resistance and deformation to physical
forces exerted by the landslide moving mass. Over the last years
there has been a significant and important development in the
understanding about what makes people, as well as social,
economic and environmental assets, susceptible to hazard of any
nature (landslide, flood, avalanche). This has expressed a new
range of socio-economic and environmental concerns which could
be also expressed by the notion of vulnerability. So, the vulnerability
meaning was extended to a reflection on the state of the individual
and the collective physical, social, economic and environmental
conditions. Governed by human activity, vulnerability cannot be
isolated from ongoing development efforts and, therefore, it plays a
critical role in the social, economic and environmental context of
sustainable development.

There is a fundamental need in landslide risk assessment to

recognise the relationships between population growth, the physical
demands of human settlements, short and longer term economic
trade-off, social and cultural questions, as well as the most
appropriate use of available land. The determination and wide
acceptance of the most suited use of land, whether it is privately or
publicly held, is demanding enough. It becomes even more daunting
if there are various points of view about the role that land use can or
should play, in terms of reducing collective exposure to risk. Too
often, the desire for short-term gains is prone to override anticipated
benefits that stretch further into the future.

Anyway, land use planning that is carefully designed and rigorously

implemented is the most useful approach to managing population
growth and the physical demands of human settlements minimising
associated risk. For this reason, land use management and the
related aspects of regional or territorial planning have to be
considered as a natural extension of conducting hazard assessment.
A failure on the part of government to implement effective land use
and planning practices could bring about unexpected effects.
Consequently, inadequate, ill-informed or nonexistent land use
planning can contribute to increasing the vulnerability of the
communities exposed to landslide hazard.

When studying land use planning measures, it is necessary to take

into account the dangers and their effects, because of the high
population density of the potentially involved area. The land planning
must be adapted to the territorial and environmental reality. In order
to increase land planning efficiency in minimising risk, the following
important principles, generally valid for different kind of dangers
(flood, flash-floods, avalanches, earthquakes, etc.), have to be taken
into account:

• Land use management operates at different geographic scales

which require different ranges of suitable management tools and
operational mechanisms.
• Land use management involves legal, technical, and social
dimensions. The legal and regulatory dimensions include laws,
decrees, ordinances adopted by national and local administrations.
The technical and instrumental dimension includes planning tools
and instruments that regulate uses of land and balance the private
and public interests. The social and institutional dimension includes
people participation in land use management practices.
Page 303

• The practice of land use management proceeds through three

consequential and interrelated stages: strategic planning,
administration and fiscal control and monitoring.

• Successful strategic land use management requires a clear legal,

procedural and regulatory framework that defines the competencies
of the various stakeholders and the “rule of the game”, including the
role of each actor in the different phases of the planning.

11.6.2 Gaps in landslide risk management studies

The IMIRILAND project develops a generalised method for risk
assessment providing a basis in relation to large landslides, but also
accounts for an insufficiently emphasised source of uncertainties and
gaps to the global risk assessment process. Many of these gaps
depend on the multidisciplinary and multisectorial nature of the
approach and should be ascribed to the following themes.

a) Crucial quantification of the socio-economic and environmental

impacts and related interconnected aspects

To take into account the large landslide risk means to change the
proper assessment scale, i.e. to extend it from a local perspective
due to the directly exposed elements at risk to a more extensive
regional one, often not easy to identify, but usually of relevant
impact. These risks are related to the indirect effects, e.g. chain
effects, implying further impacts on social and economic assets
etc…To assess the global risk thus means to take into account more
vulnerability aspects, which often interact with one another, often
implying different zones of influence both in terms of dimensions and
spatial identity. These interconnected aspects also include the
quality and availability of regional socio-economic, environmental
and cultural data, as well as the capacity for understanding and
managing this topic correctly.
While hazard analysis and physical aspects of vulnerability
evaluation have been substantially facilitated and improved due to
the use of GIS techniques, the inclusion of social, economic and
environmental variables into GIS conceptual models remains a major
methodological challenge. The need to assign a quantitative value to
the variables analysed in the spatial model used by GIS is not
always feasible for some social/economic dimensions of
vulnerability. For instance, how can the social impact of a destroyed
international railroad or the impact on cultural aspects be quantified?

Moreover, the diverse scales (individual, community, regional,

national, international) at which different aspects of socio-economic
and environmental vulnerability operate, make a representation
through these techniques very difficult. The quality and detail of the
information required by the analysis facilitated by GIS is in many
cases non-existent. In general, the quality and availability of regional
socio-economic and cultural aspects and respective data limit the
accuracy of risk analysis and assessment. On the other hand, the
possibility to measure and to quantify the socioeconomic and
environmental impacts correctly has been proved to be a very
difficult task for risk analysis and assessment. Risk assessments
need to reflect the dynamic and complex consequences involved, in
order to incorporate them into a large landslide reduction strategy
properly. Multihazards and comprehensive vulnerability/capacity
assessments that take into account the changing patterns in
landslide risk are departure points for raising risk awareness at all
scales.

b) Real perspectives of risk management with respect to short and

longer term

Prevention focuses mainly on repetitive events of low to medium

impact. However, past experiences reveal that the safety of the
community is threatened primarily by rare or extremely rare events,
and in these cases prevention strategies and measures are quite
insufficient. People, therefore, become unconsciously accustomed to
the devastating effects of these extremely rare events. Scientific
studies and research still need to be conducted to determine the
conditions under which such events occur and to evaluate their
effects.

Concerning risk management tools, one of the most effective

methods of reducing the landslide losses is to delineate the
danger/hazard zones in a management plan. In order to ensure short
and long-term control, management plans have to be revised and
updated (for example: every 5–10 years) so that, according to the
evolution of the situation, it is possible to adapt the danger/hazard
Page 304

zones, as well as the rules which have to apply, where only

prescriptions are given for exposed land use.

c) Lack of knowledge and of available tools to manage the temporal

component of the studied landslides

The spatial scale of the examined instability problem and its related
effects, the role exerted by climatic and meteorological triggering
factors and the time-dependent rock mass behaviour influence the
temporal scale of the analysis. What is the likely future behaviour of
the slope (acceleration leading to a catastrophic failure, deceleration
then stabilisation, change of speed with seasonal fluctuations, etc.)?
Even though this question is raised systematically, at the present
time it is rarely solved. Risk analysis requires the assessment of
occurrence probability in time; however, the time element is of great
uncertainty in landslide hazard/risk analysis. Occurrence probability
which can be defined as the chance or probability that a landslide
hazard will occur can be expressed in relative (qualitative) or
probabilistic (quantitative) terms. A rigorous, quantitative, procedure
could include the application of formal mechanical-probabilistic
methods, taking into account the uncertainty in all the geometrical
and mechanical parameters, including the variability of the boundary
conditions (for instance pore pressure distribution in the slope, initial
state of stresses etc.).

In the case of large landslides, however, a similar procedure cannot

be applied because the uncertainty in the definition of the
parameters involved is too large and it is not possible to define any
reliable variability for them; finally, the number of events is not
sufficient to obtain extreme values by extrapolation.

Moreover, the use of empirical correlations is often still difficult

because of the lack and/or reliability of monitoring data. Otherwise, a
periodic frequency of landslide occurrence can be estimated using
semi-quantitative methods related to the assessment of historical
data in conjunction with a detailed geological and geomorphological
characterisation. In any event, occurrence probability assessment
introduces severe uncertainties and, therefore, is almost unable to
answer the following question raised by managers at this time: When
will failure occur? Only when detailed displacement data are
available for several days before failure is it possible to express a
reasonable answer on the date of failure, as it was put forward
during the second stage of the Randa rockfall (Noverraz & Bonnard,
1992).

d) Limits in the process of modelling

The term “modelling” may be used in many contexts with different

meanings. On the one hand, for a field geologist, the model of a
mountain area includes the geological origin of the massifs, the
tectonic and morphologic process, the main structural units, etc. On
the other hand, a model for a finite element expert involves a
characterisation of the geometry of a body and its material properties
as well as initial and boundary conditions. Both experts are required
in order to produce a final comprehensive result, and the quality of
that final result will depend of their intercommunication and
exchange of experiences.

Moreover, when defining the modelling work, it is important to point

out that the quality of the final result depends on the quality of
geological and numerical analyses. First, a geological model should
be built from field observation that requires some expertise in field of
geology, since this process requires understanding of the landslide
site at a very global level and implies a simplification of reality.
Second, the results of the geological model have to be “converted”
into a numerical model framework. This is not without uncertainties,
because of the limitations and intrinsic drawbacks of numerical
models simplifying the geological models (e.g. the too high number
and the different levels of importance of geological parameters).
Despite all limitations, to obtain a good characterisation of landslide
behaviour, it is necessary to compare the results of numerical
simulation with field information and to decide whether further
analyses are required. This is done as feedback for the geological
model, which may be redefined with this new information. Feedback
for the geological model is an added value in large landslide
analyses.

Nevertheless, it is important to underline that this type of analyses

are full of difficulties despite the capabilities of the numerical
techniques and the quality of geological models. Indeed, our
Page 305

knowledge is still far from being able to reproduce instability

phenomena. For instance, it is impossible to predict, at the moment,
the temporal occurrence of displacements leading to a failure
condition, and thus the development and operation of alarm systems
based on these analyses is still a matter of further research.

11.6.3 Future potential developments

Through the discussion of the results obtained (applying the risk
assessment methodology pointed out) into different landslide risk
situations in mountainous contexts, the IMIRILAND working group
highlighted a need of improvement for the following principal
aspects:

a) A global, integrated and multidisciplinary approach involving

different disciplines

Sound methodology can only be derived from a multidisciplinary

approach taking into account the results obtained by different
disciplines.

In the light of this assumption, the risk management is based on the

development of integrated phases. Moreover, each of these phases
consists in a further multidisciplinary approach which must provide a
synthesis of data and knowledge for the next step. These phases are
described in the following points:

– analysis of instability phenomena which consists in the

construction of a geological and numerical model of the landslide in
order to understand the triggering and evolution of the instability
phenomena. This phase is carried out by experts in geological,
geotechnical and meteorological disciplines which define the hazard
scenarios.
– risk analysis of the instability phenomena which consists of the
evaluation of the direct and indirect consequences, in a wide sense,
determined from the hazard scenarios and risk levels. This process
should be carried out by experts in economic, ecological, political
and social disciplines which define a quantitative risk analysis.

– Land planning devoted to mitigation of risk. This phase is carried

out by local, regional and national authorities that provide legal,
technical and administrative instruments to make decisions, in order
to choose the most appropriate strategies.

The risk management quality depends not only on the quality of each
phase but also on the communication among the experts of the
consecutive phases. Practical difficulty exists since each phase is
performed by experts of different disciplines and the integration of
the works may suffer from communication problems.

From the experience gained in the IMIRILAND project and also from
the previous experience of the groups involved, it is possible to
improve the integration of knowledge in order to guarantee
straightforward and efficient communication among the disciplines.

b) The role of the risk perception in the community

Hazard assessment uses specific procedures which are mostly

restricted to a scientific community. On the other hand, vulnerability
and consequence analysis makes use of more conventional
methodologies and techniques, through which the community at risk
may also play an active role. The distinction between risk
assessment and risk perception has an important implication for
disaster risk reduction. In some cases, risk perception may be
formally included in the assessment process by incorporating
people’s own ideas and the perception of risk they are exposed to. In
fact, the daily “living together” between the population and the
environment has developed a high sensitivity in the community to
search the safest areas and to perceive any incipient signs of slope
instability. This sensitivity produces an increase of the acceptable
risk level. The case of the Langhe area (southern Piedmont), for
instance, underlines the importance of this sensitivity. In fact, despite
the high number and the size of landslides (over 900 cases) as a
consequence of heavy meteorological events of November 1994, no
victims were recorded.

c) Maintenance and systematic updating of landslide impact data

sets

Historical analysis of disaster data provides the information to

deduce levels of risk based on past experience. Post-analysis of a
catastrophic event can be essential for the data connected to the
Page 306

phenomenon, its impact and the way in which it is managed. In

addition, historical databases are essential to identify the dynamic
aspects involved in vulnerability. In this context, the refinement,
maintenance and systematic updating of landslide data sets are vital
for risk assessment as a whole. Data is the primary input for
identifying trends in hazard and vulnerability, as well as driving the
risk assessment and landslide impact analysis. There is a need to
work towards the standardisation of methodologies and processes
related to the collection, analysis, storage, maintenance of data.

11.6.4 Contribution of IMIRILAND project to the management of

large landslides
The large number of still open questions related to the adequate
management of large landslides might suggest that no final
conclusions may be drawn from the research developed within
IMIRILAND project. However the practice of multidisciplinary
investigations and the interchange of methodologies have led to a
much sounder assessment of involved risks which will be profitable
to all land planners involved with such landslide problems. Moreover
the complete analysis of several significant case studies with respect
to hazard assessment and risk management shows that the
scientists and the implied authorities can develop well-designed
strategies together that will prevent increasing damage. It is thus
possible to express confidence in the future improvement of large
landslide management actions.
Page 307

REFERENCES

Abele, G., 1974. Bergstürze in den Alpen. Wissenschaftliche

Alpenvereinshefte, München. H 25, 2308.

Almeida, E., Basabe, R, Jaramillo, H., Ramón, P. & Serrano, C,

1996. Desestabilización de laderas, por la crecida debido a la
ruptura de “La Josefina”. Libro: “Sin plazo para la esperanza”,
reporte sobre el desastre. Escuela Politécnica Nacional (EPN).
Quito-Ecuador, p. 213–223.

Antoine, R, Camporota, R, Giraud, A. & Rochet, L., 1987. La

menace d’écroulement aux Ruines de Séchilienne (Isère). Bull
liaison Labo. R et Ch., n° 150–151, pp. 55–64.

ARPA-Piemonte, 2004. IlProgetto IFFI in Piemonte: Inventario dei

fenomeni franosi in Italia. Archives Settore Studi e Ricerche
Geologiche—Sistema Informativo Prevenzione Rischi, ARPA-
Piemonte [Agenzia Regionale per la Protezione dell’ Ambiente],
www.arpa.piemonte.it.

Baretti, M., 1881. Relazione sulle condizioni geologiche del versante

destro della valle della Dora Riparia tra Chiomonte e Salbertrand.
Tip. e Lit. Camilla e Bertolero, Torino.

Baretti, M., 1893. Geologia della Provincia di Torino—con atlante

fuori testo di 27 carte e 8 pro-fili geologici Torino.

Basabe, R, 1998. Amenazas por inundaciones y plan de

contingencia ante el fenómeno “El Niño”. Comité de Emergencia
Nacional del Paraguay y PNUD, Asunción-Paraguay, UNDHA, p.
69+ anexos.
Basabe, R & al., 1996. Prevención de desastres naturales en la
cuenca del río Paute. Libro: “Sin plazo para la esperanza”, reporte
sobre el desastre de “La Joseflna”-Ecuador. EPN.QuitoEcuador,
p.271–287.

Basabe, P., Neumann, A., Almeida, E., Herrera, B., García, E. &
Ontaneda, R, 1998. Informe Final del proyecto PRECUPA de
cooperación entre el CSS, DNDC, INECEL, INAMHI, Munic. de
Cuenca/ETAPA, U.Cuenca y Consejo de Programación por “La
Josefina”, para la prevención de desastres naturales en la cuenca
del río Paute. Temas: Topografía/geodesia, Geología/geot-ecnia,
Hidrometeorología, Sismología, Limnología, Capacitación/difusión y
apoyo a Defensa Civil, Cuerpo Suizo de Socorro, 369 p. más
anexos, Cuenca-Ecuador.

BGC Engineering Inc., Calgary, Alberta, Canada, 2000.

Geotechnical hazard assessment, south flank of Frank Slide,
Hillcrest, Alberta. Unpublished report to Alberta Environment.

Biancotti, A. & Bovo, S. (eds), 1998a. Distribuzione regionale di

piogge e temperature. Studi climatologici in Piemonte, Vol. 1, 79 pp.
Regione Piemonte, Università degli Studi di Torino.

Biancotti, A. & Bovo, S. (eds), 1998b. Le precipitazioni nevose sulle

Alpi piemontesi. Studi climatologici in Piemonte, Vol. 2, 80 pp.
Regione Piemonte, Università degli Studi di Torino.

Bieniawski, Z.T., 1974. Geomechanics Classification of Rock Masses

and its Application in Tunnelling. Proc. 3rd Int. Congr. Rock Mech.,
ISRM, Denver U.S.A. (II A): 27–32.

Bogge, A., 1975. L’alluvione del 1728 in Val di Susa. Centro Studi
Piemontesi 4(2): 379–396.

Bonnard, Ch., 1984. Determination of slow landslide activity by

multidisciplinary measurement techniques. In Kovari, K. (ed.), Field
Measurements in Geomechanics. Proc. Int. Symp, Zurich (CH), 5–8
September 1983 (1): 619–638. Rotterdam: Balkema.

Bonnard,Ch., 1994a. Recognition, Identification, and Control of

Slope Instability Movements General Report, Proc. IVth Geoengrg.
Int. Congress, Torino, Vol. 3, pp. 685–692.

Bonnard, Ch., 1994b. Los deslizamientos de tierra: Fenómeno

natural o fenómeno inducido por el hombre? Memorias 1ersimposio
panamericano de deslizamientos de tierra. Soc. ecuatoriana de
Mecánica de Suelos y Rocas, Guayaquil, Ecuador, Vol. II, Vol. II, pp.
1–15.
Page 308

Bonnard, Ch. & Noverraz, F., 2001. Influence of climate change on

large landslides: assessment of long-term movements and trends.
Proc. Int. Conf. on Landslides: Causes, Impacts and
Countermeasures, Davos (CH), 121–138. Ed. VGE.

Bonnard, Ch., Noverraz, F. & Dupraz, H., 1996. Long-term

movements of substabilized versants and climatic changes in the
Swiss Alps. Proc. VIIth Int. Symp. on Landslides, Trondheim,
Norway, Vol 3, pp. 1525–1530, Rotterdam: Balkema.

Bonnard, Ch., Noverraz, F., Lateltin, O. & Raetzo, H., 1995. Large
Landslides and Possibilities of Sudden Reactivation. Proc. 44th
Geomechanics Colloquy, Salzburg. Felsbau No 6/95, pp. 401−407.

Bonzanigo, L., Eberhardt, E. & Löw, S. 2001. Hydromechanical

Factors Controlling the Creeping Campo Vallemaggia Landslide.
Proc. Int, Conf on Landslides: Causes, Impacts and
Countermeasures, Davos, Switzerland, Ed. VGE, pp. 13–22.

Bossalini, G. & Cattin, M., 2002. Studio dell’onda di piena

conseguente ad una ipotetica frana in località Prequartera. Internal
report, Regione Piemonte, Comunità montana Valle Anzasca.

Brabb, E.E., 1984. Innovative approaches to landslide hazard

mapping. Proc. IVth Int. Symp. on Landslides, Toronto, Canada.
Can. Geot. Soc., Vol. 1, pp. 307–324.

Brand, E.W., 1988. Special lecture: Landslide risk assessment in

Hong Kong. Proc. Vth Int. Symp. on Landslides, Lausanne,
Switzerland. Vol. 2, pp. 1059–1074. Rotterdam: Balkema.

Broilli, L., 1974. Ein Felssturz im GroBversuch. Rock Mechanics,

Suppl 3, Wien, pp. 69–78.
Brovero, M., Campus, S., Forlati, F., Ramasco, M. & Susella, G.,
1996a. La Frana del “Cassas”, Salbertrand, Val di Susa. In Regione
Piemonte & Université J. Fourier (Eds.), Rischi Generati da Grandi
Movimenti Franosi, Programma Interreg I Italia-Francia: 71–103.

Brovero, M., Campus, S., Forlati, F., Ramasco, M., Susella, G.

Scavia, C., 1996b. La Frana di “Rosone”, Valle Orco, In Regione
Piemonte & Université J.Fourier (Eds.), Rischi Generati da Grandi
Movimenti Franosi, Programma Interreg I Italia-Francia: 143–177.

Bunce, C.M., Cruden, D.M. & Morgenstern, N.R., 1997. Assessment

of the hazard from rock fall on a highway. Can. Geotech. J., (34), p.
344–356.

Bureau Bovay-Huguenin, géomètres officiels, 1994. PNR 31

Versinclim-Etude du glissements de terrain de Sedrun (GR)
(unpublished report, commenting ETHZ reports).

Büros Bonanomi, A.G. & Donatsch, 1998. Gemeinde Tujetsch-

Deformationsmessung Cuolm da Vi (unpublished report).

Büros Bonanomi, A.G. & Donatsch, 2003. Measurement data at

Cuolm da Vi (unpublished CD).

CAF-CEPAL, 2000. Las lecciones de El Niño, Volumen IV: Ecuador.

Memorias del Fenómeno El Niño 1997–1998 Restos y propuestas
para la región andina. Corporación Andina de Fomento (CAF),
Caracas, Venezuela.

Capello, C.F., 1941. Il lago quaternario della conca di Salabertano

(Valle di Susa). Boll. Com. Glac. It. (21): 155–160.

Carol, I., Prat, P.C. & López, C.M., 1997. A normal/shear cracking
model. Application to discrete crack analysis. Journal of Engineering
Mechanics (123–8): 765–773.
Carrara, A., 1992. Landslide hazard assessment. Proc, Ist Symp. Int.
Sensores Remotos y Sistema de Inform. Geogr. para el Estudio de
Riesgos Natur., March 10–12, 1992, Bogotá, 329–355.

Carrara, A., Cardinali, M. & Guzzetti, F., 1992. Uncertainty in

assessing landslide hazard and risk. ITC Jour., v. 1992:2, 172–183.

Carraro, F., Dramis, F. & Pieruccini, U., 1979. Large-scale landslides

connected with neotectonic activity in the Alpine and Apennine
ranges.In Proc. 15th Plenary Meeting of IGU-UNESCO Commission
“Geomorphological Survey & Mapping”, Modena, 7–15 September
1979: 213–230.

Carrillo-Gil, A. & Carrillo-Degado, E., 1988. Landslide risk in the

Peruvian Andes. Proc. Vth Int. Symp. on Landslides, Lausanne,
Switzerland, Vol. 2, pp. 1137–1142. Rotterdam: Balkema.

Carta Geologica d’Italia in scala 1:50’000, 2002. Foglio 132–152–

153 Bardonecchia. Servizio Geologico Nazionale APAT, Roma.

Castelli, M., Forlati, F. & Scavia, C., 2001. Landslides: the questions
of the decision-maker and the answer of the modelisation, Proc. 1st
Int. Conf. Albert Caquot, Paris, 3–5 October 2001.
Page 309

Cherubini, C., Giasi, C.I. & Guadagno, F.M., 1993. Probabilistic

approaches of slope stability in a typical geomorphological setting of
Southern Italy. Risk and reliability in ground engineering, pp. 144–
150, Thomas Telford, London.

Chopin, C., 1981. Talc-Phengite: a Widespread Assemblage in High-

Grade Pelitic Blueschists of the Western Alps. J. of Petrology, 22(4):
628–650.

Compagnoni, R. & Lombardo, B., 1974. The Alpine age of the Gran
Paradiso eclogites. Rend. Soc. It. Min. Petr., 30 (1): 223–237.

Compagnoni, R., Elter, G. & Lombardo, B., 1974. Eterogeneità

stratigrafica del complesso degli “gneiss minuti” nel massiccio
cristallino del Gran Paradiso. Mem. Soc. Geol It., 13 (1): 227–239.

Consiglio Nazionale Delle Ricerche (CNR), 1990. Structural model of

Italy, scale 1:500’000, sheet 1. Progetto Finalizzato Geodinamica,
Firenze, Italy.

Consorzio Forestale di Oulx, 1955. Capitolato di vendita di

soprasuolo boschivo. Not published, Archives Consorzio Forestale di
Oulx, 13 febbraio 1955.

Corominas, J., Copons, R., Vilaplana, J.M., Altimir, J. & Amigò, J.,
2002. Integrated landslide susceptibility analysis and hazard
assessment in the principality of Andorra. Natural Hazard, Kluwer
Academics Publisher: 1–15.

Corpo Forestale dello Stato, 1956. Lettera inviata al Consorzio

Forestale di Salbertrand del 26/1/1956. Not published, Archives
Consorzio Forestale di Oulx.

Crosta, G., Frattini, P. & Sterlacchini, S., 2001. Valutazione e

gestione del rischio da frana. Principi e metodi. Volume 1. Milano:
Regione Lombardia and Università di Milano Bicocca.

Cruden, D., 1991.A simple definition of a landslide. Bull. Int. Ass. of

Engrg. Geol., No. 43, pp. 27–29.

Cruden, D.M. & Fell, R. (Eds.), 1997. Proc. Int. Workshop on

Landslide Risk Assessment, Honolulu (Hawaii, USA), 19÷21
February 1997. Rotterdam: Balkema.

Cruden, D.M. & Varnes, D.J., 1996. Landslide types and processes.
In: “Landslides: Investigation and Mitigation”. Transportation
Research Board. National Academy of Sciences, pp. 36–75.

C.T.M. [Compagnia Torinese Monitoraggi], 2003. Impianto di

monitoraggio territoriale, geotecnico e sismico della Val Susa—
Frana del Cassas. Impianto di monitoraggio inclinometrico—Misura
strumentazione, Rapporto Interpretativo N° 14, giugno 2003 (not
published). SITAF (Soc. It. Traforo Autostradale Frejus) s.p.a.,
Archives Settore Progettazione Interventi Geologico-Tecnici e
Sismico, ARPA-Piemonte, Torino.

Cundall, P.A., 1987. Distinct Element Models of Rock and Soil

Structure, Analytical and Computational Methods in Engineering
Rock Mechanics, Ch. 4, pp. 129–163, E.T. Brown, Ed. London: Allen
& Unwin.

Dal Piaz, G. & Lombardo, B., 1986. Ealy Alpine ecloite

metamorphism in the Penninic Monte Rosa-Gran Paradiso
basement nappes of the northwestern Alps. Geol Soc. Am.,
164:249–265.

Davis, G.H. & Reynolds, S.J., 1996. Structural geology of rocks and
regions. 776 p. New York: Wiley & Sons.

Dela Pierre, F, Lozar, F. & Polino, R., 1997. L’utilizzo della tettono-
stratigrafia per la rappresentazione cartografica delle successioni
meta-sedimentarie nelle aree di catena. Mem. Sci. Geol. (49): 195–
206.

Del Prete, M., Giaccardi, E. & Trisoro-Liuzzi, G., 1992. Rischio da

frane intermittenti a cinematica lenta nelle aree montuose e collinari
urbanizzate della Basilicata. Pubbl. n. 841 GNDCI (Gruppo
Nazionale per la Difesa dalle Catastrofi Idrogeologiche). Potenza:
Centro Nazionale delle Ricerche.

DRM (Delegation aux risques majeurs), 1990. Les études

préliminaires a la cartographie régle-mentaire des risques naturels
majeurs. Secrétariat d’Etat auprès du premier Ministre charge de
l’Environnement et de la Prévention des Risques technologiques et
naturels majeurs. La Documentation Française.

Duranthon, J.-P, Effendiantz, L., Memier, M. & Previtali, I., 2003.

Apport des méthodes topographiques et topométriques au suivi du
versant rocheux instable des Ruines de Séchilienne. Revue XYZ, n°
94.

Dutto, F. & Friz, E., 1989. Ricerche sull’area di invasione di valanghe

di roccia. Rapporto interno CNR-IRPI/ISMES, Febbraio, 16p.
Page 310

Eckardt, P., Funk, H.P. & Labhart, T., 1983. Postglaziale

Krustenbewegungen an der Rhein-RhoneLinie. Vermessung,
Photogrammetrie und Kulturtechnik 2/83, pp. 43–56.

Effendiantz, L., Marchesini, P. & Pera, J., 2000. La galerie de

reconnaissance de Séchilienne. Tunnels et ouvrages souterrains, n°
161, pp. 303–306.

Einstein, H.H., 1988. Special lecture: Landslide risk assessment

procedure. Proc. Vth Int. Symp. on Landslides, Lausanne,
Switzerland, Vol. 2, pp. 1075–1090. Rotterdam: Balkema.

Eisbacher, G.H. & Clague, J.J., 1984. Destructive mass movements

in high mountains: hazard and management. Ottawa: Geological
Survey of Canada, Paper 84, 16, p. 230.

Enel.Hydro, 2001. Attività di progettazione, fornitura ed installazione

di un sistema di monitoraggio integrato del movimento franoso di
Rosone (TO); Scenari di rischio. Prog. ISMES 2338. Archives of
Settore Studi e Ricerche Geologiche ARPA-Piemonte, Italy.

Engelder, T., 1985. Loading paths to joint propagation during a

tectonic cycle: an example from the Appalachian Plateau, U.S.A. J.
Struct. Geol. 7:459−476.

Epifani, F.D., 1991. Studio geologico e geomorfologico del versante

a monte delle aree di servizio di Salbertrand—Studio geologico
Epifani—SITAF (not published), Archives Settore Progettazione
Interventi Geologico-Tecnici e Sismico, ARPA-Piemonte, Torino.

Evans, S.G., 2003. Characterizing Landslide risks in Canada. Proc.

Geohazards 2003, Edmonton Conference, Canadian Geotechnical
Society, pp. 20.
Evans, S.G., Hungr, O. & Enegren, E.G., 1994. The Avalanche Lake
rock avalanche, Mackenzie Mountains, Northwest Territories,
Canada: description, dating and dynamics. Canadian Geotechnical
Journal (31): 749–768.

Evans, S.G., Clague, J.J., Woodsworth, G.J. & Hungr, O., 1989. The
Pandemonium Creek Rock Avalanche, British Columbia. Canadian
Geotechnical Journal, 26:427–446.

Evrard, H., Gouin, T., Benoit, A. & Duranthon, R, 1990. Séchilienne.

Risques majeurs d’éboule-ment en masse. Point sur la surveillance
du site. Bull liaison Labo. P et Ch., n° 165, pp. 7–16.

Fell, R., 1994. Landslide risk assessment and acceptable risk. Can.
Geot. J., (31): 261–272.

Fell, R., Hungr, O., Leroueil, S. & Riemer, W., 2000. Keynote
Lecture015—Geotechnical engineering of the stability of natural
slopes, and cuts and fills in soil. Proc. GeoEng 2000 Conf,
Melbourne, Vol. l, pp. 21–120.

Finlay, P.J. & Fell, R., 1997. Landslides: risk perception and
acceptance. Can. Geot. J., (34): 169–188.

FLAC3D, 1997. Itasca: Fast Lagrangian Analysis of Continua in 3

Dimensions (FLAC3D), Version 2.0, Manual, Itasca, Minnesota,
Minneapolis.

Forlati, F. & Piana, F., 1998. Vincoli geologico-strutturali e

caratterizzazione degli ammassi rocciosi. In: G. Barla Ed., MIR 98,
Conf. Naz. Mecc. Ing. Rocce, VI ciclo, Torino, 1–11.

Forlati, F, Ramasco, M., Susella, G., Barla, G., Marino, M. & Mortara,
G., 1993. La deformazione gravitativi di Rosone. Un approccio
conoscitivo per la definizione di una metodologia di studio. Studi
Trentini di Scienze Naturali, Acta Geologica 68:71–108.
Fratianni, S. & Motta, L., 2002. Andamento climatico in Alta Val Susa
negli anni 1990–1999. Regione Piemonte, Università degli Studi di
Torino. Studi climatologici in Piemonte (4).

Friz, E. & Pinelli, P.F., 1993. Riceerche sull’area di invasione di

valanghe di roccia. Atti del convegno “Fenomeni Franosi”, Riva del
Garda, 23–25 maggio 1990. Studi Trentini di Scienze Naturali, Acta
Geologica, 68, 1, pp.71–108.

Geo & Soft International, 1999. ROTOMAP for Windows. Turin, Italy.

Geo & Soft International, 2003. ISOMAP & ROTOMAP for Windows
(3D Surface Modelling & Rockfall Analysis). User’s Guide.

Geoengineering, 1984. Studio geologico-tecnico del settore di

versante retrostante alla frazione di Rosone vecchio. Internal report
Comune Locana.

Gianquinto, F, 2000. Studio idrogeologico dei settori di versante

interessati dalle frane del Cassas e di Serre la Voute. ECOPLAN-
SITAF, May 2000 (not published), Archives Settore Progettazione
Interventi Geologico-Tecnici e Sismico, ARPA-Piemonte, Torino.

Giardino, M. & Polino, R., 1997. Le deformazioni di versante dell’alta

valle di Susa: risposta pellicolare dell’evoluzione tettonica recente.Il
Quaternario, Italian Journal of Quaternary Sciences (10–2): 293–
298.
Page 311

Hancock, R L., 1985. Brittle microtectonics: principles and practice.

J. Struct. Geol., 7:437−457.

Heim, A., 1932. Bergsturz und Menschenleben. Zurich: Fretz &

Wasmuth Verlag, p. 227.

Hobbs, B.E., Means, W.D. & Williams, P.F., 1976. An Outline of

Structural Geology. New York: Wiley & Sons.

Hötzl, H., Moser, M., Reichert, B. & Rentschler, K., 1994.

Hydrologische Markierungsversuche in Massenbewegungen.
Reppwandgleitung, Kärnten und Stubnerkogel, Salzburg. Steir.
Beiträge z. Hydrogeologie, H. 45, S. 69–92.

Hsü, K.J., 1975. Catastrophic debris streams (Sturzstroms)

generated by rockfalls. Bull Geol. Soc. of Am. 86 (1): 129–140.

Hsü, K.J., 1978. Albert Heim: observations on landslides and

relevance to modern interpretation. In Voight (Ed.), Rockslides and
avalanches (1): 71–93.

Hungr, O., 1981. Dynamics of rock avalanches and other types of

mass movements. Ph. D. thesis, University of Alberta, Canada.

Hungr, O., 1995. A model for the runout analysis of rapid flow slides,
debris flows, and avalanches. Canadian Geotechnical Journal,
32:610–623.

Hungr, O., 1997. Some methods of landslide hazard intensity

mapping. Proc. Int. Workshop on Landslide Risk Assessment, D.M.
Cruden & R. Fell (Eds.), pp. 215–226. Rotterdam: Balkema.

Hungr, O., 2001. A review of the classification of landslides of the

flow type. Environmental and Engineering Geoscience (7): 221–238.
Hungr, O. & Evans, S.G., 1996. Rock avalanche runout prediction
using a dynamic model. Proc. 7th Int. Symp. on Landslides,
Trondheim, Norway, V 1, p. 233–238. Rotterdam: Balkema.

Hungr, O. & Rawlings, G., 1995. Terrain hazards assessment for

planning purposes: Cheekye Fan, B.C. Proc. 48th. Canadian
Geotechnical Conf. Vancouver, B.C. 1:509–517.

Hungr, O., Sobkowicz, J. & Morgan, G., 1993. How to economise on

natural hazards. Geotechnical news, v. 7/1, pp. 54–57.

Hutchinson, J.N., 1992. Landslide hazard assessment. Proc VIth Int.

Symp. on Landslides, Christchurch, New Zealand. Vol. 3, pp. 1805–
1841. Rotterdam: Balkema.

Imrie, A.S. & Moore, D.P., 1997. BC Hydro’s approach to evaluating

reservoir slope stability from a risk perspective. Proc. Int. Workshop
on Landslide Risk Assessment, D.M. Cruden & R. Fell Eds., pp.
197–205. Rotterdam: Balkema.

Imrie, A.S., Moore, D.P. & Enegren, E.G., 1992. Performance and
maintenance of the drainage system at Downie Slide. Proc. 6th. Int.
Symp. on Landslides, Christchurch, New Zealand, Vol. 1:751–757.
Rotterdam: Balkema.

I.S.R.M. [International Society for Rock Mechanics], 1978.

Suggested Methods for the Quantitative Description of Discontinuity
in the Rock Masses. Int. J. Rock Mech. Min. & Geomech. (Abstract,
15): 319–368.

ITASCA, 1997. Fast Lagrangian Analysis of Continua in 3

Dimensions (FLAC3D), Version 2.0, Manual. Minneapolis: Itasca.

ITASCA, 1999a. PFC2D, Particle Flow Code in Two Dimensions,

Version 2.0, Manual. Minneapolis: Itasca.
ITASCA, 1999b. PFC3D, Particle Flow Code in Three Dimensions,
Version 2.0, Manual. Minneapolis: Itasca.

IUGS/WGL/CRA (International Union of Geological

Societies/Working Group on Landslide/ Committee on Risk
Assessment), 1997. Quantitative risk assessment for slopes and
landslides—The state of the art. In Cruden, D.M. & Fell, R. (Eds.),
Proc. Int. Workshop on Landslide Risk Assessment, Honolulu
(Hawaii, USA), 19÷21 February 1997:3–12. Rotterdam: Balkema.

Khaler, F. & Prey, S., 1963. Erläuterungen zur geologischen Karte

des Naßfeld-GartnerkofelGebietes in den Karnischen Alpen. Wien
(GBA) (Erläuterungen geol. Karten).

Lapenta, M.C. & Trombetta, A., 1996. Caratterizzazione

geomeccanica dei versanti rocciosi dell’alta Valle di Susa (Piemonte)
in relazione ai problemi di stabilità. GEAM (Associazione Georisorse
& Ambiente) (91): 31–39.
Page 312

Larini, G., Malaguti, C, Pellegrini, M. & Tellini, C., 2001. La Lama di

Corniglio (Appennino parmense), riattivata negli anni 1994–1999.
Quaderni di Geologia Applicata, 8.2 (2001), pp. 59–114. Bologna:
Pitagora Editrice.

Läufer, A.L., Frisch, W., Steinitz, G. & Loeschke, J., 1997. Exhumed
fault-bounded Alpine blocks along the Periadriatic lineament: the
Eder unit (Carnic Alps, Austria). In: Geol. Rundschau, Bd. 86, S.
612–616.

Leone, F., Asté, J.-P. & Leroi, E., 1996. Vulnerability assessment of
elements exposed to massmovement: working toward a better risk
reception. Proc. VIIth Int. Symp. on Landslides, Trondheim, Norway,
Vol. 1, p. 263–270. Rotterdam: Balkema.

Leroi, E., 1997. Landslide risk mapping: problem, limitations and

developments. In Cruden, D.M. & Fell, R. (Eds.), Proc. Int. Workshop
on Landslide Risk Assessment, Honolulu (Hawaii, USA), 19÷21
February 1997:239–250. Rotterdam: Balkema.

Li Tianchi, 1983. A mathematical model for predicting the extent of a

major rockfall. Zeitschrift für Geomorfologie, 27:473−482.

Luino, F., Ramasco, M. & Susella, G., 1993. Atlante dei centri abitati
instabili piemontesi. Gruppo per la difesa delle catastrofi
idrogeologiche. Prog. Spec. Studio centri abitati instabili, pubb. n°
964, CNR-IRPI, Regione Piemonte, Settore Prevenzione Rischio
Geol. Meteorol. e Sismico. p.245.

Mathews, W.H. & McTaggart, K.C., 1978. The Hope rock slides,
British Columbia, Canada. In Rockslides and Avalanches, Voight, B.
Ed., 1:197–258. Amsterdam: Elsevier.

Mayoraz, F., Cornu, Th., Djukic, D. & Vulliet, L., 1997. Neural
networks: A tool for prediction of slope movements. Proc. 14th Int.
Conf. on Soil Mechanics and Foundation Engineering, Hamburg. Vol.
1, pp. 703–706.

McConnell, R.G. & Brock, R.W., 1904. The great landslide at Frank,
Alberta. Canadian Parliament Sessional Paper, No.25, Department
of the Interior, Government of Canada.

Moore, D.P. & Imrie, A.S., 1992. Stabilization of Dutchman’s Ridge.

Proc. VIth. Int. Symp. on Landslides, Christchurch, New Zealand,
Vol. 3:1783–1788. Rotterdam: Balkema.

Moore, D.P., Imrie, A.S. & Baker, D.G., 1991. Rockslide risk
reduction using monitoring. Proc. Canadian Dam Safety Conference.
Vancouver: BiTech Publishers.

Mora, S. & Wahrson, W., 1994. Macrozonation methodology for

landslide hazard determination. Memorias lersimposio panamericano
de deslizamientos de tierra, Sociedad Ecuatoriana de Mecánica de
Suelos y Rocas, Guayaquil, Vol. I, pp. 406−431.

Morelli, M., 2000. Analisi sulla possibilità di integrazione dei dati

telerilevati in studi geologicostrutturali: applicazione nel dominio del
Monferrato e delle Langhe. Ph.D. Thesis, University of Torino, p.
168.

Morelli, M. & Piana, F. 2003. “Geometric filtering” of remote-sensed

lineaments and search for geological rules of their distribution.
Application to the Monferrato succession marlyarenaceous
succession (NW-Italy). Abstract in Proc. of Tectonic Studies Group
AGM, Liverpool (UK), 8–10 January 2003.

Mortara, G. & Sorzana, P.F, 1987. Fenomeni di deformazione

gravitativa profonda nell’arco alpino occidentale italiano.
Considerazioni lito-strutturali e morfologiche. Boll Soc. Geol. It.
(106): 303–314.
Moser, M., 2001. Talzuschub Reppwand-Gleitung, Geologisch-
geotechnischer Bericht für 2000 unter besonderer Berücksichtigung
der Bewegunsmessungen (Periode 14–15, Juni 99 bis Oktober
2000). Bericht WLV (Austrian Service for Torrent, Erosion and
Avalanche Control), unpublished.

Moser, M. & Glawe, U., 1994. Das Naßfeld in Kärnten-geotechnisch

betrachtet. In: Abh. Geol B.-A., Bd. 50, pp. 319–340.

Moser, M. & Windischmann, Th., 1989. Die

Reppwandgleitung/Kärnten. Geologische und geotechnische
Betrachtungen. In: Oberrhein. geol. Abh., 35. pp. 157–176.

Moser, M, Angerer, J. & Seitz, S., 1988. Geotechnische

Untersuchungsergebnisse im Rahmen des Verbauungsprojektes
Oselitzenbach/Kärnten. Interprävent (Ed): Interprävent 3, Graz. pp.
77–102.
Page 313

Mulder, H.F.H.M., 1991. Assessment of landslide hazard. Geograph.

Sci. Fac., Elinkwijk Utrecht, p. 150.

Nichol, S. & Hungr, O., 2002. Brittle and ductile toppling of large rock
slopes. Canadian Geotechnical Journal, 39:1–16.

Noverraz, F. & Bonnard, Ch., 1991. L'écroulement rocheux de

Randa, près de Zermatt. Proc. VIth Int. Symp. on Landslides,
Christchurch, New Zealand. Vol. 1, pp. 165–170. Rotterdam:
Balkema.

Noverraz, F., Bonnard, Ch., Dupraz, H. & Huguenin, L., 1998.

Grands glissements de versants et climat. Rapport final PNR 31—
Projet VERSINCLIM, 314 p. Zurich: v/d/f Verlag.

OFAT (Office Fédéral de l’Aménagement du Territoire), OFEE (Office

Fédéral de l’ Economie des Eaux) & OFEFP (Office Fédéral de
l’Environnement, des Forêts et du Paysage), 1997. Prise en compte
des dangers dus aux mouvements de terrain dans le cadre des
activités de l’aménage-ment du territoire, Recommandations
Fédérales. Série dangers naturels (O. Lateltin, Ed.), OCFIM (Office
Central Federal des Imprimés et du Matériel), Bern, Switzerland, 42
p.

OFEE (Office Federal de l’Economie des Eaux), OFAT (Office

Federal de l’Aménagement du Territoire) & OFEFP (Office Federal
de l’ Environnement, des Forêts et du Paysage), 1997. Prise en
compte des dangers dus aux crues dans le cadre des activités de l
’aménagement du territoire, Recommandations Fédérales. Série
dangers naturels (R. Loat et A. Petraschek, Eds.), OCFIM (Office
Central Fédéral des Imprimés et du Matériel), Bern, Switzerland, p.
32.
Panet, M., Bonnard, Ch., Lunardi, P. & Presbitero, M., 2000.
Expertise relative aux risques d’éboulement du versant des Ruines
de Séchilienne. Rapport pour le ministère de l’aménage-ment du
territoire et de l’environnement. See the website:
www.environnement.gouv.fr.

Paro, L., 1997. Ricostruzione dell’evoluzione geologica quaternaria

del versante destro della Valle di Susa nel tratto compreso tra
Salbertrand ed Exilles. Thesis in Geological Science, University of
Torino; 4 maps published by Regione Piemonte.

Patton, F.D., 1984. Climate, Groundwater Pressures and Stability

Analyses of Landslides. Proc. IVth Int. Symp. on Landslides,
Toronto, Canada. Can. Geot. Soc., Vol. 3, pp. 43–59.

Perello, P, Delle Piane, L., Piana, F., Stella, F. & Damiano, A., 2004.
Brittle post-metamorphic tectonics in the Gran Paradiso massif
(north-eastern Italian Alps). Geodinamica Acta, in press.

Peretti, L., 1967. Collegamento autostradale del traforo del Frejus

con Torino. La Rivista della Stmda (36, 310).

Peretti, L., 1969. Premesse geoapplicative per la realizzazione

coordinata dell’autostrada Torino Oulx e della sistemazione
idrogeologica della Valle di Susa. Cronache da Palazzo
CisternaPeriodico della Provincia di Torino (1).

Perla, R., Cheng, T. & McClung, D., 1980. A two-parameter model of

snow-avalanche motion. Journal of Glaciology, Vol. 26, n.94, 197–
207.

Petrascheck, A., 1995. Mapas de peligros: Encuesta respecto de los

niveles de peligrosidad. Offices fédéraux suisses de l’économie des
eaux (OFEE), de l’amenagement du territoire (OFAT) et de
l’environnement, des forêts et du paysage (OFEFP), Berna. Inédito.
16 p. y anexos.
Polino, R., Dela Pierre, F., Borghi, A., Carraro, F, Fioraso, G. &
Giardino, M., 2002. Note Illustrative della Carta Geologica d’Italia
alla scala 1:50’000–foglio 132–152–153 Bardonecchia. Servizio
Geologico Nazionale APAT, Roma.

Pothérat, P & Alfonsi, P, 2001. Les mouvements de versant de

Séchilienne (Isère). Revue française de géotechnique n° 95–96, p.
117–131.

Potter, D., 1972. Computational physics. London: John Wiley &

Sons.

PPR (Plans de Prevention des Risques Naturels Prévisibles), 1995.

Loi n° 95–101 du 2 février 1995. Ministère de l’Ecologie et du
Développement Durable, Direction de la Prévention, des Pollutions
et des Risques (DPPR), France.

Prat, P.C., Gens, A., Carol, L, Ledesma, A. & Gili, J.A., 1993. DRAC:
A computer software for the analysis of rock mechanics problems. H.
Liu (Ed.), Application of computer methods in rock mechanics. Xian
(China): Shaanxi Science and Technology Press. 1361–1368.

Programme INTERREG I Italie-France, 1996. Risques générés par

les grands mouvements de versants. Regione Piemonte, Université
Joseph Fourier. p. 207.
Page 314

Puma, F., Ramasco, M., Stoppa, T. & Susella, G., 1984. Carta dei
movimenti di massa nelle alte valli del Chisone e di Susa.—in
Compagnoni et al., Geotraversa della Zona Piemontese nella Valle
di Susa, escursione pre-congresso (10÷11/9/1984) 72° Congresso
Soc. Geol. It., Serv. Geol. Reg. Piemonte.

Puma, F., Ramasco, M., Stoppa, T. & Susella, G., 1989. Movimenti di
massa nelle alte valli di Susa e Chisone. Boll Soc. Geol It. (108):
391–399.

Ramasco, M. & Susella, G., 1978. Studi geologici per il collegamento

stradale tra il traforo del Frejus e Torino (tratto Bardonecchia-Susa).
Regione Piemonte, Dipartimento Organizzazione e Gestione del
territorio.

Ramasco, M., Stoppa, T. & Susella, G., 1989. La deformazione

gravitativa profonda di Rosone in Valle Orco. Boll Soc. Geol It. 108:
401–408.

Ramsay, J.G. & Huber, M.I., 1987. The techniques of modern

structural geology. Volume 2: Folds and Fractures. London:
Academic Press.

Regione Piemonte, 2000. La frana di Prequartera, analisi del

movimento franoso e valutazione delle potenziali fasi evolutive e
schema del sistema di controllo del fenomeno franoso. A cura del
Gruppo Interdisciplinare di Studio Regione Piemonte e Politecnico di
Torino e del settore Regionale Progettazione Interventi Geologico
Tecnici e Sismici, Internal report (unpublished).

Rochet, L., 1987. Développement des modèles numériques dans

l’analyse de la propagation des éboulements rocheux. Proc. of the
VIth Int. Congress on Rock Mechanics, Montreal, Vol. 1, pp. 479–
484.
Romana, M., 1990. Practice of SMR Classification for Slope
Appraisal. Proc. 5th Int. Symp. on Landslides, Lausanne,
Switzerland, Vol. 2:1227–1231. Rotterdam: Balkema.

Sacco, F., 1898. La geologia e le linee ferroviarie in Piemonte.

Torino.

Scavia, C., Barla, G. & Bernaudo, V, 1990. Probabilistic stability

analysis of block toppling failure in rock slopes. Int. Journal of
Mechanics Mining Science & Geomechanical Abstracts (27–6): 465–
478. Oxford: Pergamon Press.

Schönlaub, H.P. & al., 1987. Geologische Karte der Republik

Österreich 1:50’000, Blatt 198 (WeiBbriach). Geol.BA.Wien.

Schreyer, W., 1977. Whiteschists: their composition and pressure-

temperature regimes based on experimental, field and petrographic
evidence. Tectonophysics, 43:127–144.

Scioldo, G., 1991. ROTOMAP: analisi statistica del rotolamento dei

massi. Ass. Min. Subalpina: Atti Convegno “La meccanica delle
rocce a piccola profondità”, Torino, 8184.

SEA consulting, 2001. Ductile structural setting of the Gran Paradiso

unit and relationships with large scale landslides in the Orco valley
(Italian western Alps). In Giulio Elter (Ed.), Dalla Tetide alle Alpi;
Proc. symp. Cogne, 21–22, Giugno 2001, 37–38.

Segré, C., 1920. Considerazioni geognostiche sul tronco Bussoleno-

Salbertrand (ferrovia TorinoFrejus) con riguardo speciale ai tratti
franosi: provvedimenti. Giornale del Genio Civile (LVIII): 19–40.

Service Hydrologique et Géologique National, 1992. Swiss

Hydrological Atlas.

S.I.T.A.F. [Società Italiana Traforo Autostradale del Frejus], 2000.

Frane di Serre la Voute e del Cassas, modellazione matematica e
valutazione del rischio geologico. Authors: C.T.M. – Polithema studio
associato—Oboni associés, not published, Archives Settore
Progettazione Interventi Geologico-Tecnici e Sismico, ARPA-
Piemonte, Torino.

Sobkowicz, J., Hungr, O. & Morgan, G.C., 1995. Probabilistic

mapping of a debris flow hazard area. Proc. 48th. Canadian
Geotechnical Conference, Vancouver, B.C. 1:519–529.

Spang, R.M. & Rautenstrauch, R.W., 1988. Empirical and

mathematical approaches to rockfall protection and their practical
applications. Proc. Vth Int. Symp. on Landslides, Lausanne,
Switzerland, Vol. 2, pp. 1237–1243. Rotterdam: Balkema.

Studio Geologico Italiano (SGI), 1984. A.E.M.Torino Impianto

S.Lorenzo (Locana). Relazione geologica, Allegato 1, Internal report
Regione Piemonte.

Turner, A.K. & Schuster, R.L. (Eds.), 1996. Landslides—Investigation

and Mitigation. Special Report 247. Transportation Research Board,
Academy of Sciences, Washington D.C., p. 673.
Page 315

UN (United Nations), 1991. Mitigating Natural Disasters.

Phenomena, Effects and Options. A Manual for Policy Makers and
Planners. UNDRO/MND/1990 Manual. New York: UN (1991-VIII).

UNESCO, 1993. Multilingual landslide glossary. The International

Geotechnical Societes. UNESCO Working party for world landslide
inventory. BiTech Publishers. ISBN 0–920 505–10–4, Canada.

Varnes, D.J. & The International Association of Engineering Geology

Commission on Landslides and other Mass Movements, 1984.
Landslide Hazard Zonation: A Review of Principles and Practice.
Natural hazards (3), p. 63 Paris, France: UNESCO.

Varnes, D.J., 1978. Slope movement types and processes. In

Schuster, R.L. and Krizek, R.J. (Eds.), Landslides. Analysis and
Control. Transportation Research Board, Special Report 176:12–33.
Washington, D.C.: National Academy of Sciences.

Vialon, R, Ruhland, M. & Grolier, J., 1976. Elements de tectonique

analytique. p. 118 Paris: Masson.

Vulliet, L., & Bonnard, Ch., 1996. The Chlöwena Landslide:

Prediction with a Viscous Model. Proc. 7th Int. Symp. on Landslides,
Trondheim, Norway, Vol. 1, pp. 397–02. Rotterdam: Balkema.

Weidner, S., 2000. Kinematik und Mechanismus alpiner

Hangdeformationen unter bes. Berücksichtigung der
hydrogeologischen Verhältnisse. PhD Thesis. Uni Erlangen,
unpublished.

Woodcok, N.H. & Fischer, M., 1986. Strike-slip duplexes. J. Struct.

Geol 8:725–735.

Zettler, A.H., Poisel, R., Roth, W. & Preh, A., 1999. Slope stability
analysis based on the shear reduction technique in 3D. Proc. of the
International Flac Symposium on Numerical Modeling in
Geomechanics, Minnesota, Minneapolis, pp. 11–16.
Page 316

This page intentionally left blank.