Landslide Hazards, Risks, and Disasters

Landslides are the most costly geo-hazard in the world, and they’re often the cause or the result of other hazards and disasters such as tsunamis, earthquakes, wildfires, and volcanic eruptions. Landslide Hazards, Risks, and Disasters makes a close and detailed examination of major mass movements and provides measures for more thorough and accurate monitoring, prediction, preparedness, and prevention. It takes a geoscientific approach to the topic while also discussing the impacts human-induced causes such as deforestation, blasting, and building construction—underscoring the multi-disciplinary nature of the topic.

• Contains contributions from expert geologists, seismologists, geophysicists, and environmental scientists selected by a world-renowned editorial board
• Presents the latest research on causality, economic impacts, fatality rates, and landslide and problem soil preparedness and mitigation
• Numerous tables, maps, diagrams, illustrations, photographs, and video captures of hazardous processes
• Discusses steps for prevention and treatment of problem soils, the most expensive geo-hazard in the world

• Language: English
• Publisher: Elsevier Science
• Release date: Oct 23, 2014
• ISBN: 9780123964755

Book preview

Landslide Hazards, Risks, and Disasters

Series Editor

John F. Shroder

Emeritus Professor of Geography and Geology, Department of Geography and Geology, University of Nebraska at Omaha, Omaha, NE 68182

Volume Editor

Tim Davies

Geological Sciences, University of Canterbury, Christchurch, New Zealand

Table of Contents

Cover image

Title page

Copyright

Contributors

Editorial Foreword

Preface

Chapter 1. Landslide Hazards, Risks, and Disasters: Introduction

1.1. Introduction

1.2. Understanding Landslide Hazards

1.3. Understanding Landslide Risks

1.4. Understanding Future Landslide Disasters

1.5. Conclusion

Chapter 2. Landslide Causes and Triggers

2.1. Introduction

2.2. Concept of Instability

2.3. Stability Factors

2.4. Summary and Conclusion

Chapter 3. Mass Movement in Bedrock

3.1. Introduction

3.2. Rock Materials

3.3. Mass Movement Characteristics

3.4. Mass Movement Types

3.5. Case Studies

3.6. Recognition and Response

3.7. Risk Management in Rock slopes

Chapter 4. Coseismic Landslides

4.1. Seismically Triggered Landslides

4.2. Mechanics of Earthquake-Induced Landslides

4.3. Stability Analysis and Hazard Assessment

4.4. Limitations of Current Understanding

Chapter 5. Volcanic Debris Avalanches

5.1. Introduction

5.2. Volcanic Debris Avalanches

5.3. Types of Volcanic Landslides

5.4. Deep-Seated Volcanic Landslide Deformation: Priming and Triggers

5.5. Deep-Seated Volcano Gravitational Deformation

5.6. Regional Tectonic Influences

5.7. Priming of Volcanic Landslides

5.8. Triggering Volcanic Landslides

5.9. The Structure of Volcanic Landslides

5.10. Volcanic Landslide Deposits

5.11. Debris Avalanche Textures and Structures

5.12. Secondary Hazards of Volcanic Landslides

5.13. Volcanic Landslide Transport Mechanisms

5.14. Hazards from Volcanic Landslides

5.15. Summary

Chapter 6. Peat Landslides

6.1. Introduction and Background

6.2. The Nature of Peat, Its Structure, and Material Properties

6.3. Morphology and Classification of Peat Landslides

6.4. Relationship Between Landslide Type and Peat Stratigraphy

6.5. Impacts of Peat Landslides

6.6. The Runout of Peat Landslides

6.7. Slope Stability Analysis of Peat Landslides and Geotechnical Properties

6.8. Historical Perspective on the Frequency of Peat Landslides

6.9. The Future Incidence of Peat Landslides

6.10. Conclusion

Chapter 7. Rock–Snow–Ice Avalanches

7.1. Introduction

7.2. Rapid Mass Movements on Glaciers

7.3. RSI Avalanche Propagation

7.4. Implications for Hazard Assessment

7.5. Conclusions

Chapter 8. Multiple Landslide-Damming Episodes

8.1. Introduction

8.2. Previous Work on Landslide Dams

8.3. Landslide-Dam Episodes: Lessons from Case Studies

8.4. Discussion

8.5. Conclusions

Chapter 9. Rock Avalanches onto Glaciers

9.1. Introduction

9.2. Processes

9.3. Consequences

9.4. Case Studies

9.5. Conclusions

Chapter 10. Paleolandslides

10.1. Introduction

10.2. Significance of Paleolandslides

10.3. Recognition and Mapping

10.4. Dating Paleolandslides

10.5. Temporal Bias

10.6. Role in Landscape Evolution

10.7. Risk Assessment

10.8. Conclusion

Chapter 11. Remote Sensing of Landslide Motion with Emphasis on Satellite Multitemporal Interferometry Applications: An Overview

11.1. Introduction

11.2. Brief Introduction to Differential SAR Interferometry and Multitemporal Interferometry

11.3. Examples of Different Scale MTI Applications to Landslide Motion Detection and Monitoring

11.4. Summary Discussion

Chapter 12. Small Landslides—Frequent, Costly, and Manageable

12.1. Introduction

12.2. Costs of Small-Medium Landslides

12.3. Frequency of Landslides

12.4. Management of Landslides

12.5. Size of Manageable Landslides

12.6. Conclusions

Chapter 13. Analysis Tools for Mass Movement Assessment

13.1. Introduction

13.2. The Computational Tools Available

13.3. Limit Equilibrium Methods

13.4. Limit Analysis

13.5. Continuum Numerical Methods

13.6. Distinct Element Method

13.7. Conclusions

Index

Copyright

Elsevier

Radarweg 29, PO Box 211, 1000 AE Amsterdam, Netherlands

The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, UK

225 Wyman Street, Waltham, MA 02451, USA

Copyright © 2015 Elsevier Inc. All rights reserved.

No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Details on how to seek permission, further information about the Publisher’s permissions policies and our arrangements with organizations such as the Copyright Clearance Center and the Copyright Licensing Agency, can be found at our website: www.elsevier.com/permissions.

This book and the individual contributions contained in it are protected under copyright by the Publisher (other than as may be noted herein).

Notices

Knowledge and best practice in this field are constantly changing. As new research and experience broaden our understanding, changes in research methods, professional practices, or medical treatment may become necessary.

Practitioners and researchers must always rely on their own experience and knowledge in evaluating and using any information, methods, compounds, or experiments described herein. In using such information or methods they should be mindful of their own safety and the safety of others, including parties for whom they have a professional responsibility.

To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any liability for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions, or ideas contained in the material herein.

Library of Congress Cataloging-in-Publication Data

Landslide hazards, risks, and disasters / volume editor, Tim Davies.

pages cm. -- (Hazards and disasters series)

ISBN 978-0-12-396452-6 (hardback)

1. Landslides--Risk assessment. 2. Landslides--Prevention. 3. Landslide hazard analysis. I. Davies, Timothy R. H., editor of compilation.

QE599.A2L363 2014

363.34'9--dc23

2014035690

British Library Cataloguing in Publication Data

A catalogue record for this book is available from the British Library

ISBN: 978-0-12-396452-6

For information on all Elsevier publications visit our web site at http://store.elsevier.com

Contributors

Fabio Bovenga,     National Research Council, CNR-ISSIA, Italy

Elisabeth T. Bowman,     Department of Civil and Structural Engineering, University of Sheffield, South Yorkshire, UK

Marc-André Brideau,     Simon Fraser University, Burnaby, BC, Canada

John J. Clague,     Centre for Natural Hazard Research, Simon Fraser University, Burnaby, BC, Canada

Giovanni B. Crosta,     Università degli Studi di Milano-Bicocca, Italy

Tim Davies,     Geological Sciences, University of Canterbury, Christchurch, New Zealand

Audray Delcamp,     Department of Geography, Faculty of Sciences, Vrije Universiteit Brussel, Brussel, Belgium

Philip Deline,     EDYTEM Lab, Université de Savoie, CNRS, Le Bourget-du-Lac, France

Kenneth Hewitt,     Geography and Environmental Studies, Wilfrid Laurier University, Waterloo, Ontario, Canada

Oliver Korup,     Institute of Earth and Environmental Science, University of Potsdam, Potsdam, Germany

Samuel T. McColl,     Physical Geography Group, Institute of Agriculture and Environment, Massey University, New Zealand

Bill Murphy,     Leeds University, UK

Natalya Reznichenko,     Department of Geography, Durham University, Durham, UK

Nicholas J. Roberts,     Simon Fraser University, Burnaby, BC, Canada

Dan Shugar,     Department of Geography, University of Victoria, British Columbia, Canada

Rosanna Sosio,     Università degli Studi di Milano-Bicocca, Italy

Stefano Utili,     School of Engineering, University of Warwick, Coventry, UK

Benjamin van Wyk de Vries,     Laboratoire Magmas et Volcans, Univeristé Blaise Pascal, Clermont-Ferrand, France

Gonghui Wang,     Disaster Prevention Research Institute, Kyoto University, Kyoto, Japan

Jeff Warburton,     Durham University, UK

Janusz Wasowski,     National Research Council, CNR-IRPI, Italy

Editorial Foreword

General hazards, risks, and disasters: Hazards are processes that produce danger to human life and infrastructure. Risks are the potential or possibilities that something bad will happen because of the hazards. Disasters are that quite unpleasant result of the hazard occurrence that caused destruction of lives and infrastructure. Hazards, risks, and disasters have been coming under increasing strong scientific scrutiny in recent decades as a result of a combination of numerous unfortunate factors, many of which are quite out of control as a result of human actions. At the top of the list of exacerbating factors to any hazard, of course, is the tragic exponential population growth that is clearly not possible to maintain indefinitely on a finite Earth. As our planet is covered ever more with humans, any natural or human-caused (unnatural?) hazardous process is increasingly likely to adversely impact life and construction systems. The volumes on hazards, risks, and disasters that we present here are thus an attempt to increase the understanding about how to best deal with these problems, even while we all recognize the inherent difficulties of even slowing down the rates of such processes as other compounding situations spiral on out of control, such as exploding population growth and rampant environmental degradation.

Some natural hazardous processes such as volcanoes and earthquakes that emanate from deep within the Earth's interior are in no way affected by human actions, but a number of others are closely related to factors affected or controlled by humanity, even if however unwitting. Chief among these, of course, are climate-controlling factors, and no small measure of these can be exacerbated by the now obvious ongoing climate change at hand (Hay, 2013). Pervasive range and forest fires caused by human-enhanced or induced droughts and fuel loadings, mega-flooding into sprawling urban complexes on floodplains and coastal cities, biological threats from locust plagues, and other ecological disasters gone awry; all of these and many others are but a small part of the potentials for catastrophic risk that loom at many different scales, from the local to planet girdling.

In fact, the denial of possible planet-wide catastrophic risk (Rees, 2013) as exaggerated jeremiads in media landscapes saturated with sensational science stories and end-of-the-world Hollywood productions is perhaps quite understandable, even if simplistically short-sighted. The end-of-days tropes promoted by the shaggy-minded prophets of doom have been with us for centuries, mainly because of Biblical verses written in the early Iron-Age during remarkably pacific times of only limited environmental change. Nowadays however, the Armageddon enthusiasts appear to want the worst to validate their death desires and prove their holy books. Unfortunately we are all entering times when just a few individuals could actually trigger societal breakdown by error or terror, if Mother Nature does not do it for us first. Thus we enter contemporaneous times of considerable peril that present needs for close attention.

These volumes we address here about hazards, risks, and disasters are not exhaustive dissertations about all the dangerous possibilities faced by the ever-burgeoning human populations, but they do address the more common natural perils that people face, even while we leave aside (for now) the thinking about higher-level existential threats from such things as bio- or cybertechnologies, artificial intelligence gone awry, ecological collapse, or runaway climate catastrophes.

In contemplating existential risk (Rossbacher, 2013) we have lately come to realize that the new existentialist philosophy is no longer the old sense of disorientation or confusion at the apparently meaninglessness or hopelessly absurd worlds of the past, but instead an increasing realization that serious changes by humans appear to be afoot that even threaten all life on the planet (Kolbert, 2014; Newitz, 2013). In the geological times of the Late Cretaceous, an asteroid collision with Earth wiped out the dinosaurs and much other life; at the present time by contrast, humanity itself appears to be the asteroid.

Misanthropic viewpoints aside, however, an increased understanding of all levels and types of the more common natural hazards would seem a useful endeavor to enhance knowledge accessibility, even while we attempt to figure out how to extract ourselves and other life from the perils produced by the strong climate change so obviously underway. Our intent in these volumes is to show the latest good thinking about the more common endogenetic and exogenetic processes and their roles as threats to everyday human existence. In this fashion, the chapter authors and volume editors have undertaken to show you overviews and more focused assessments of many of the chief obvious threats at hand that have been repeatedly shown on screen and print media in recent years. As this century develops, we may come to wish that these examples of hazards, risks, and disasters are not somehow eclipsed by truly existential threats of a more pervasive nature. The future always hangs in the balance of opposing forces; the ever-lurking, but mindless threats from an implacable nature, or heedless bureaucracies countered only sometimes in small ways by the clumsy and often feeble attempts by individual humans to improve our little lots in life. Only through improved education and understanding will any of us have a chance against such strong odds; perhaps these volumes will add some small measure of assistance in this regard.

Landslide hazards, risks, and disasters: The chapters in this volume on landslides provide the latest thinking on understanding these ever-so-common phenomena that seem to threaten societies almost everywhere in the world that slopes exist. In fact, as one of the most pervasive processes at work almost everywhere on the Earth's surface, slope failure is characteristically the one process that most people appear to ignore, all too commonly to their detriment.

Downslope movements of rock and soil debris and earth by falling, toppling, sliding, and flowing with great variations of velocities and water contents, whether liquid water, solid ice present such a plethora of movement mechanics, that mass-movement specialists have struggled to understand and protect against landslides for many decades. As one of those natural hazard specialists, it has indeed been a pleasure for me to see that continued new thinking on the phenomena has progressed so well; in fact, even though landslide processes are so common and so hard to avoid entirely, the new observations and understandings have led to a more intelligent siting of structures, as well as better engineering designs to cope with instability. As humans continue to expand their populations and move into regions with ever-increasing hazard potentials for slope stability, the observations and understandings presented in these chapters can help in detailing the means to avoid mass-movement problems.

Volume editor Tim Davies is from New Zealand, where the combinations of strong relief, active tectonic forces, high seismicity, strong climatic controlling factors, and many weak lithologies produce exceptionally active slope failure and mass movement at practically every turn. Dr Davies has given us a comprehensive volume that not only addresses the great variety of landslides in the world, but that also addresses a number of examples so that the reader will receive a good education on what has happened, what can happen, and how to best understand the many subtle variations in slope-failure phenomena so that avoidance of the dangerous or damaging processes can be effected best. The variety of chapters presented about a diverse and difficult hazard are sufficiently authoritative and new that even those quite accustomed to the rich literature of landslides will still find much that is new and interesting here.

John (Jack) Shroder

Editor-in-Chief

July 14, 2014

References

Hay W.W. Experimenting on a Small Planet: A Scholarly Entertainment. Berlin: Springer-Verlag; 2013 983 p.

Kolbert E. The Sixth Extinction: An Unnatural History. NY: Henry Holt & Company; 2014 319 p.

Newitz A. Scatter, Adapt, and Remember. NY: Doubleday; 2013 305 p.

Rees M. Denial of catastrophic risks. Science. 2013;339(6124):1123.

Rossbacher L.A. Contemplating existential risk. Earth, Geologic Column. October 2013;58(10):64.

Preface

This volume is one of a nine-volume set, each of which deals with the hazards, risks, and disasters associated with specific types of landscape process. The other volumes treat hydro-meteorological, volcanic, landslides, earthquakes, sea and ocean processes, snow and ice processes, wildfires, and biological and environmental processes; whereas the final volume presents a cross-disciplinary overview of natural hazards and disasters in society.

The present volume focuses on the hazards, risks, and disasters associated with landslides of various types (e.g., bedrock, volcanic, peat-slide, rock-ice, landslide-dam) and from a variety of perspectives (e.g., paleoslope failures, susceptibility, remote sensing). The purpose is to make available to readers, both students and professionals, a set of discussions by a range of experts on topics receiving current attention in the general area of landslide hazards, risks, and disasters. Authors from a range of backgrounds have contributed, with a number of contributions from younger researchers who have a refreshingly modern approach to landslide problems.

Acknowledging the recent publication of an excellent book on Landslides: Types, Mechanisms and Modeling edited by John Clague and Doug Stead, this volume has tried not to duplicate or overlap unnecessarily with the contents of that book. By contrast with the Clague and Stead book, the focus herein is to develop improved understanding of, and hence ability to foresee and thus make expectable, the effects of landslides on society. To that end some basic concepts are required, particularly where these may not be readily available elsewhere; but throughout, the aim is to further our understanding of landslides and the ways in which they affect humankind and its assets. Hence the Hazard–Risk–Disaster sequence, wherein the hazard is essentially a natural landslide process (perhaps quantified by its probability of occurrence); risk is the product of the probability and consequence of interaction between a landslide and a societal asset; and a disaster is the event that results from this interaction (usually inferred to be of major impact with respect to the part of society affected).

The introductory Chapter 1 is intended to provide a general context for the more specialized chapters that follow, by outlining some overarching concepts of hazards, risks, and disasters. It also introduces some developing concepts and ideas in a range of topics that may point the way to improvements in our attempts to reduce the damage, disruption, and deaths caused by landslides in the future.

In Chapter 2 McColl outlines the slope preconditioning, preparation, and triggering processes required to initiate a landslide that may have the potential to cause a disaster, depending on whether assets are likely to be affected by it. Brideau and Roberts describe the conditions and processes that lead to failure of rock slopes in Chapter 3, and discuss the strategies available to manage potential rock-slope instabilities; while in Chapter 4 Murphy focuses specifically on the processes by which earthquakes can destabilize slopes leading to coseismic landsliding, emphasizing the importance of vertical accelerations and topographic amplification, and sounding a cautionary note with respect to the reliability of presently available analysis techniques. In Chapter 5 Van Wyk De Vries and Delcamp outline the factors and processes that cause large-scale collapse of volcanic edifices, and the geometry of the resulting debris avalanche deposits; they also mention the range of secondary hazards resulting from such phenomena. In the following chapter Warburton describes the state of the art in understanding of peat landslides, emphasizing the remaining uncertainties that cause hazard assessment of these events to remain problematic at present.

In Chapter 7 Sosio describes the comparatively recently recognized class of events known as rock-snow-ice avalanches, which are represented by the Huascaran events of 1962 and 1970 in Peru, and the 2002 Kolka-Karmadon event in North Ossetia. This chapter complements that of Deline et al. (Chapter 9) describing the processes and effects of rock avalanches that travel over, and deposit on, glaciers, and both exemplify the reality that natural events rarely behave in such a way that they can be satisfactorily described within the confines of a single specialization.

Korup and Wang in Chapter 8 extend previous knowledge of landslide dam hazards and risks to the occurrence of multiple episodes of such events, as have been documented from recent severe earthquakes and storms. They emphasize the additional challenges posed to emergency response and recovery by multiple landslide-damming episodes. In Chapter 10 Clague discusses the nature and accuracy of information on hazards and risks that can be derived from study of the deposits of past landslides, discussing in particular the data bias resulting from unequal preservation of deposits of different ages and its application to risk analysis.

Wasowski and Bovenga provide in Chapter 11 a review of presently available methods for utilizing satellite-based remote sensing to detect and monitor, and outline the risk management and research opportunities offered by these rapidly advancing technologies.

In Chapter 12 Bowman addresses the comparatively neglected topic of assessment and management of the risks associated with the smaller but correspondingly more frequent landslides that make up most of the event population, focusing on the widespread impacts of these minor events on society, and on the techniques available and required for their economic mitigation. Finally Utili and Crosta (Chapter 13) outline and critically assess the methods currently available for assessing the stability of slopes, in the context of the technologies now available to facilitate these assessments, emphasizing the continuing necessity for adequate data on which to base even sophisticated analyses.

It is evident from many of these contributions that, even after a century or more of work, and with all the benefits of modern technologies, serious deficiencies remain in our ability to understand, and therefore to avoid or mitigate, future landslide disasters. Clearly landslide research needs to continue into the future, and we can confidently anticipate improvements in knowledge and even understanding of landslide processes. However, in order that these improvements are reflected in reduction in landslide disasters, this effort must be complemented with significantly increased involvement of landslide experts in the development of strategies that will allow society and communities to plan sustainable futures that are resilient to both anticipated and unexpected landslide events.

Tim Davies

Volume Editor

Chapter 1

Landslide Hazards, Risks, and Disasters

Introduction

Tim Davies     Geological Sciences, University of Canterbury, Christchurch, New Zealand

Abstract

Consideration of the nature and occurrence of landslide hazards leads to perspectives on the dominant role of landsliding in the geomorphology of active orogens, and consequently the major role that landsliding plays in determining hazard- and riskscapes both within and downstream of these areas. Discussion of landslide risks suggests that probabilistic analyses are only likely to be reliable in planning location-specific landslide risk management strategies for small, frequent events, and the potential for identifying sites of future landslides—both rainfall generated and coseismic—is examined. Finally the role of landslides in triggering consequential hazards, such as tsunami, river flooding, and debris flows, is emphasized.

Keywords

Landslides; Active landscapes; Hazards; Risks; Disasters

1.1. Introduction

Landslides are a ubiquitous phenomenon on a planet that, like Earth, is tectonically active. However society is generally inclined to view them as exceptional events that occur very infrequently, and usually elsewhere, and their inevitable impacts on society worldwide and over extra-human timescales have hitherto been considered rarely, if at all, in societal planning. In the last decade, however, global landslide occurrence and impact has been better documented (e.g. Petley, 2012), and the seriousness of this hazard underlined. Interestingly, climate change is shown to have much less influence on the number of landslide fatalities than population growth (Petley, 2010), and if this trend continues, landslide fatalities will continue to increase. There is thus good reason to examine the role of landslides as threats to society, and to seek for avenues whereby this threat can be reduced in the future.

1.2. Understanding Landslide Hazards

Landslide hazards are, in essence, landslides which have the potential to affect society detrimentally. One may debate whether or not all landslides constitute hazards to society, and whether in principle any landslide anywhere is a potential hazard if there is any possibility that humankind is now or will at some time in the future make itself vulnerable to the effects of that landslide thus generating a risk. There are very few if any places on Earth where this possibility is zero, thus to a fair approximation all terrestrial landslides can be considered to be hazards. So, on that basis, are landslides on the Moon, but we have to draw a line somewhere…

Landslides are a crucial component of Earth's geological cycle, in which tectonic plate motion causes parts of the crust to be continuously uplifted above a base level; they are then continuously eroded down again by gravity and gravity-driven water flow toward base level. Landslides represent the directly gravity-driven component of erosion, and they occur in sizes ranging from individual rocks falling to whole mountainsides collapsing. There is increasing evidence, by way of magnitude–frequency data, that larger landslides deliver more sediment to river systems over time than do smaller ones, so that large, infrequent events dominate the sediment supply spectrum (Korup and Clague, 2009); and, since the majority of river-transported sediment originates in slope failures, this emphasizes the significance of landsliding in geomorphology—including fluvial geomorphology—especially in and adjacent to active orogens. Even a mountain range such as the Southern Alps of New Zealand, which was heavily glaciated prior to 18  ka, today shows little evidence of any erosion process other than mass movement (Figure 1.1).

Thus, in steep, active terrain, the progress of geology requires that landslides will continue to occur on hillslopes in the future; and the increasing presence of people and their assets on, in the vicinity of and downstream of these hillslopes, means that landslide-generated disasters will inevitably occur—and to an increasing extent—in the future.

An interesting fact discussed at some length by Korup and Clague (2009) is the fairly consistent variation of probability of occurrence for landslides of different sizes. These, irrespective of type and trigger, appear to follow a common distribution for larger events (Figure 1.2), suggesting that there is some factor constraining the frequency of occurrence of landslides of various sizes. This obviously has relevance for assessing hazards and risks from landslides. Recently, research into complex systems has shown that distributions of this type are very common for such systems in many contexts (geomorphic, societal, financial, biological, ecological, etc.); and has in addition identified that the very largest events can depart significantly from this distribution. These mega-events, known as dragon-king events (Sornette, 2009), occur much more frequently than the distribution would suggest (indicated by the red line in Figure 1.2), and appear to reflect the fact that these events occupy a large proportion of the space available to them—thus such an event has as its environment the system boundaries, which is not the case for smaller events whose environment usually does not include these boundaries. In landslide terms this is equivalent to the probability distribution of landslides from a given hillslope being limited because the physical extent of the hillslope limits the volume of the largest landslides that can occur. The tendency for still larger events to occur is constrained by the system boundaries, so that these larger events are in fact manifested as smaller (but still very large) events, which thus acquire correspondingly higher (but still small) frequencies. This higher than expected frequency of the largest events is clearly a concern in anticipating future landslide disasters, as these events can give rise to the largest disasters; and, although they occur very rarely in a given place, they will inevitably occur and can occur at any time, so treating them as a low priority is not sensible. Hence the largest events occur more frequently than the probability function for smaller events would suggest.

FIGURE 1.1  The Southern Alps, New Zealand from the west. All of the landforms in the upper part of the picture are mass movement related, with a complete absence of glacial landforms (in spite of a small glacier being visible). The prominent high terraces in the center of the picture are glacial and/or tectonic in origin, while the sediment being reworked by the river at bottom is largely landslide derived.

FIGURE 1.2  Landslide area—probability density distribution. ( After   Malamud et al., (2004) .) The red curve indicates dragon-king events (see text).

To better deal with landslide hazards, it is first necessary to know where landslides are likely to occur; and, second, how big they will be. These two steps can lead to an event scenario for the hazard. Estimating a probability for an event of given size and location is much more difficult, and is in fact of lesser value from a disaster reduction perspective, because we are interested predominantly in reducing the impacts of the next disaster that will affect a locality, and probabilities give no reliable information about when that will occur or how big it will be, even in an ideal world with perfect magnitude–frequency information. Thus a landslide susceptibility map, together with an event scenario, provides local people and their governments with realistic information about what can happen there at any time; this can then be used, in conjunction with information on the location of societal assets, to develop a societal consequence scenario that forms the basis for designing hazard avoidance and/or damage reduction (assuming that preventing a major landslide is an unrealistic ambition) at all scales from personal to societywide and for thinking about disaster recovery frameworks.

The tricky bit in this train of logic lies in choosing the magnitude of the event scenario. A case can be made for selecting the worst-case scenario, on the basis that a community that has thought through how to cope with this scenario can also cope with anything smaller (although here it needs to be recognized that calculated worst-case scenarios—maximum credible events—have recently been dramatically exceeded in a number of earthquake disasters); however, this is likely to be criticized as scare mongering, on the grounds that the maximum possible event occurs incredibly rarely. Thus a somewhat less catastrophic scenario is likely to gain broader acceptance, although here it is easy to set off down the slippery slope of associating probabilities with scenarios. Recent geo-disasters have been spoken of by scientists as having return periods of tens of thousands of years, so it is quite clear that a useful disaster event scenario needs to be substantially worse than the commonly used 100-year event. People who experience major disasters learn that statistical improbability does not prevent rare events from happening at any time.

Developing a consequence scenario from an event scenario depends on having (or developing) information about where vulnerable assets are located, and where they will be located in the future. A distinct benefit of consequence scenarios is that if (as is often the case) the landslide location (event scenario) is uncertain, then the areal extent of the consequence scenario can be expanded accordingly to accommodate this uncertainty. Foreseeing landslide effects is clearly difficult; one needs to know the volume of the landslide and its location to predict the extent of its deposit and what assets and people it will affect. As the basis for an event scenario, the specific volume and location can be assumed on the basis of credible science, but the implications of this specific scenario then need to be extended across the full area that any landslide can possibly affect. Again it may be tempting to plan regionally for the most likely (highest probability) event; but there is a need to be aware that, although this event is indeed the most likely, its probability is nevertheless small, and the probability that something different will happen is very much larger. Thus planning specifically for the most likely event is not rational.

A potentially confusing factor in identifying and quantifying landslide hazards is the geomorphic similarity between moraine deposits and landslide deposits. In recent years many deposits previously classified as moraines (some of which have indeed been used as paleoclimatic indicators) have been reinterpreted as deposits of large landslides (e.g., Hewitt, 1999; McColl and Davies, 2011; Reznichenko et al., 2012). Obviously the hazard implications of such reinterpretation are significant; moraines indicate zero likelihood of repeat events until glaciers advance again, and even then the threat is minor due to the slow speed of glacial advance, but landslide deposits indicate the potential for further landslides at any time in the future, with potentially much more serious consequences.

Predicting the locations of future landslides is obviously difficult, but modern technologies allow at least the relative susceptibilities of different locations to landsliding to be estimated. For example, Kritikos et al. (in review) analyzed the spatial distribution of landslides caused by the Northridge and Wenchuan earthquakes with respect to ground shaking, topography, distance from active faults, slope gradient, and slope position, and were able on this basis to explain with >90% accuracy the relative spatial distribution of landslides resulting from the Chichi earthquake in Taiwan. This explanation used no data from Taiwan landslides, so the technique is applicable to regions with no previous landslide data; similar analyses for rainfall-generated landslides should be feasible. While providing only relative information, this technique nevertheless is useful for scenario-based planning using relative vulnerabilities of, for example, highways and utilities, so that preparation for future landslide events can take place.

It is relevant to note here that relatively little focus is to be found, in this book, on climate change as a factor in future landslide hazards, risks, or disasters. This is because, of the variables that affect the impact of landslides on society, the best quantified in the past, and thus the best predicted in the future, is the increase in global population and the degree of encroachment of society into landslide-prone territory. Recent data show unsurprisingly that landslides kill people mainly where people live in landslide-prone areas (Petley, 2012), and it follows that more people will be killed in future as a result of population increase in these areas. The effects of climate change on landsliding are in comparison poorly understood, poorly defined, and in some cases equivocal. A similar effect occurs with landslide damage, which is related to the growth in value of assets available to be damaged (Petley et al., 2007), and with the overall cost of landslides which is related to the increasing rate at which business is conducted round the planet.

1.3. Understanding Landslide Risks

Risk is defined as the product of (hazard) probability and (event) consequence, and so introduces as a variable the societal values potentially affected by a landslide (hazard). The consequence itself comprises products of exposure and vulnerability (fragility), so is conceptually more complex than the hazard. Nevertheless, some of its components are more readily quantifiable than many landslide characteristics—for example, the value of assets in a given area and the number of people in the area, at a given time—but these are of course also time-variable at a wide range of scales into the future. In addition, the cost of a landslide disaster includes a wide range of consequential costs, such as the effect on commerce of the cessation of some activities (including transport and power supply), perhaps for a long time; the effect on commerce nationally, if lifelines and infrastructure have been badly affected; and the disruption of community life and social activity, which can affect commerce even if other inputs are not affected. Landslide disasters also have intangible costs which are evidently difficult to quantify as risk, but which are nevertheless significant to society because they affect the way society functions—deaths, emotional damage, loss of quality of life, and environmental damage are examples.

Probabilistic analysis of landslide risk is the commonly accepted basis for making rational decisions about choosing what measures to take to reduce disaster risks. A fundamental difficulty with this procedure is that in any reasonable future planning time frame (i.e., a time frame that present inhabitants perceive as relevant to their community), the number of landslide events that will occur in a given place will be very small. Thus it is extremely unlikely that the events that actually occur in this time frame will follow the probability distribution used, and this would be the case even if this distribution were perfectly known. Thus the apparent precision deriving from probabilistic risk analysis is to a large extent spurious. As an example, consider a simple and perfect statistic: the average of all the numbers 0–100 is 50.000. If we take, say, 100 numbers at random between 0 and 100, what is their average? Well, it is almost never 50.000. We carried out this exercise, repeating the random number exercise 10 times and then choosing the highest value of the 10 averages of the 100 numbers. We then repeated this exercise for smaller and smaller samples of the numbers 0–100. The results are shown in Figure 1.3.

Here we plot the % error of the highest mean of the 10 repetitions against the size of the sample. With a sample size of 100 the error is about ±10%, with a sample size of 10 it is ±40%, and with a sample size of 2 it is ±95%. So any event that occurs only twice during the planning period is likely to depart from a perfectly known statistical distribution by ±95%. If our planning period is say 100  years, then we can predict 1-year events to ±10%, but 50-year events involve an error of up to ±95%. It is apparent that probabilistic risk analysis involves large and intrinsic imprecision for events that occur less than about 5 times during a planning period—but events of this scale happen so frequently that they can hardly be called disasters, because society will be expecting them and will have taken measures to mitigate them already.

FIGURE 1.3  Error in small samples of a distribution.

Other difficulties also lie in wait. Suppose we have somehow quantified a realistic landslide risk for a given site, what do we do with it? Two things, generally: if loss of life is a possibility, then there exist levels of acceptable risk that are used worldwide—for example, for an individual a risk of death due to a landslide is acceptable at a level of about 10−⁶ per year (Finlay and Fell, 1997). In this case landslide risk levels need to be managed so that they are below this figure. Alternatively, if the risks are to assets and values, a risk-based cost–benefit or utility–maximization exercise can be undertaken to find the optimal level of expenditure. Both of these, however, have problems.

1. The acceptable risk of loss of life varies dramatically with the number of lives involved. For example, a risk analysis of coseismic landslide-induced tsunami at Milford Sound, New Zealand, showed that the risk of death to an individual tourist is about 10−⁶ per year, so is more or less acceptable; however, such an event will on average kill about 400 people in one location all at once every 1000  years, and this is unacceptable by about six orders of magnitude (Figure 1.4). The calculated annual risk of 400 deaths at Milford is 400/1000  =  4  ×  10−¹; the acceptable risk level for this number of deaths is about 2  ×  10−⁷. So while the individual risk is acceptable, the societal risk is grossly unacceptable. What implications this has for risk management requirements—for example, who carries responsibility for societal¹ risk—is as yet unclear. The arguments regarding the imprecision of outcomes associated with probabilistically described risks are of little import in this case because of the gross differences between actual and acceptable risks; if these were smaller the implications would be much less clear.

FIGURE 1.4  Acceptable risk for dam failure. Because landslides at Milford Sound are known to occur at a given frequency at a given location (from dated seabed deposits), they are equivalent to known risks of dam failure. The acceptable risk of 400 deaths is 2   ×   10 − ⁷ , compared with the actual risk of 4   ×   10 − ¹ . After Munger et al. (2009).

2. Utility optimization involves comparing the net benefits of different levels of mitigation. However, the net benefit is the difference between benefits and costs associated with a given mitigation; these are both large and inevitably imprecise numbers, so subtracting them to derive the net benefit results in a very much smaller number—with a correspondingly very much larger imprecision. This can soon reach the stage where the end result is virtually meaningless—especially considering the imprecision intrinsic to probabilistic planning.

The real value of probabilistic risk analysis appears to lie in insurance, because an insurance (or reinsurance) company spreads local risk across a very large number of locations so that the number of hazard events addressed by the company is very large and widespread; and across all locations, event occurrences will match probabilities fairly well if the probabilities themselves are reliable. This is equivalent to using probabilities to plan mitigation for a given locality over an extremely long time period. However, insurance is a relatively ineffective mitigation for major disasters, because it addresses only with great difficulty the large volumes of claims that result; this is demonstrated by the fact that, several years after the 2011 Christchurch earthquake sequence in New Zealand, only a small proportion of insurance claims have been settled. In addition, insurance has no mitigatory value for many of the types of loss that occur, such as death and distress, and plays no role at all in reducing the physical impacts of disasters.

The only landslide risk management strategy with any realistic chance of being effective against major events is that of avoidance. Substantial landslides—especially earthquake-generated ones—often occur without warning, and are uncontrollable, so the prevention and evacuation procedures used for other hazards such as rainstorm-induced floods are not effective. If a potential landslide has been identified, monitoring of precursory motion may make evacuation a feasible option, but this may be inhibited by political and other considerations, as in the case of the Vaiont tragedy (e.g., Davies, 2013). For small and frequent events, by contrast, prevention is an option (Bowman, 2014)—and, if such events occur frequently in the same locality, a probabilistic risk analysis may be a realistic way of designing a mitigation strategy.

1.4. Understanding Future Landslide Disasters

Most of the major landslide disasters that affect society are unexpected. This statement is in fact not confined to landslides, but surprise is a particularly consistent feature of landslide disasters. It follows that landslide disaster impacts could probably be reduced substantially if they could be better expected—but how can that be achieved? Again, perhaps relying less on probabilistic information would reduce the natural tendency to underprioritize the inevitable occurrence of major but low-probability events at some time in the future. Identification of precursory behavior to a major landslide would be of great value, and in some places this is being done very effectively (e.g., the Åknes landslide, Norway: Kveldsvik et al., 2009. Here a major landslide seems possible, and would generate an extremely damaging tsunami in a highly populated fiord system). Obviously not all landslide-prone slopes can be identified and monitored, however; an alternative strategy available with modern remote sensing technologies is to search for geomorphic evidence of precursory deformation, such as slumping, developing cracks, and small-scale rockfalls (e.g., Wasowski and Bovenga, 2014). There is evidence that geomorphic detection of precursory motion is also feasible for earthquake-generated landsides, based on the recent realization that many earthquake-affected slopes may deform episodically in a sequence of earthquakes (perhaps hundreds of years apart over thousands of years) until eventually an earthquake (or, indeed, a rainstorm; Chigira et al., 2013) causes further deformation that completes a failure surface and causes catastrophic collapse. An example of this is shown in Figure 1.5. This is a slump at Roche Pass, South Island, New Zealand, that is known to have been mobilized by the 1929 M 7.1 Arthur's Pass earthquake, but has evidently not (yet) failed completely. In hundreds of years' time this feature will be much less conspicuous because the bare rock and debris will have vegetated (unless further slumping occurs), and the slump will only be detectable morphologically. Another example is shown in Figure 1.6, where the near part of the slope has failed in a 0.7-km³ rock avalanche giving the hummocky deposit at right center; however the remainder of the slope, although it has clearly slumped, has not yet failed completely. The 400-km long plate-boundary Alpine fault runs at the base of the slope, and generates M8 earthquakes several times per millennium (Berryman et al., 2012a,b), which probably caused the slumping and the rock avalanche. Identification of slopes with this characteristic morphology thus appears to be promising way to identify future coseismic landslides. Indeed Barth (2013) suggested that a steep slope overlooking Franz Josef Glacier township, Westland, New Zealand might pose a catastrophic collapse threat to the township in a future earthquake; again, the Alpine fault runs at the base of the slope (Figure 1.7). Given that the township hosts thousands of tourists each year, the potential hazard is clear; the volume of the potential landslide appears to be of the order of 10⁶  m³, so a runout of >1  km could occur, obliterating much of the township. As with the Milford Sound example above, the societal risk of this event appears to be unacceptable by several orders of magnitude, and the event itself is unmanageable, so relocation of the township seems to be the only feasible mitigation option. Clearly this is a very difficult sociopolitical situation—but the hazard potential revealed by the science is clear enough to cause the question to be asked.

FIGURE 1.5  Coseismic slump at Roche Pass, Southern Alps, New Zealand. Photo by Trevor Chinn.

FIGURE 1.6  Google Earth image of the 0.7-km ³ Cascade rock avalanche deposit (D), South Westland, New Zealand. The sackung (black dashed line) and the headscarp above it, immediately adjacent to the failed source area (S), indicate a partly deformed slope that could also fail catastrophically in a future earthquake. The Alpine fault is indicated by the white dashed line.

A peculiarity of some major mountain landslides that is now becoming apparent is that a significant number of them occur with no perceptible trigger. For example, since 1991 seven major (>10⁶  m³) rock avalanches have occurred in the Southern Alps of New Zealand, none of which was associated with either earthquake or rainstorm (e.g., McSaveney, 2002; Hancox and Thompson, 2013). Furthermore, given the relatively frequent occurrence of both earthquakes and severe rainstorms in the Southern Alps, the fact that these landslides did not occur during any previous potential triggering events suggests that whatever processes were reducing the slope factor of safety (e.g., freeze–thaw, permafrost degradation, stress corrosion), this reduction was occurring quite rapidly. Interestingly, this temporally clustered but spatially dispersed set of landslides could, if not historically recorded, have been interpreted by future scientists as evidence of a widespread trigger earthquake or rainstorm—a potential error that cautions against simplistic paleointerpretation of landslide deposits (Clague, 2014). In this case a seismic trigger would appear unlikely because all of the landslide source areas are shallow and broad, indicating surficial material failure, whereas major coseismic landslides often have deep-seated source-area scars (Turnbull and Davies, 2006).

FIGURE 1.7  Google Earth image of hillslope overlooking Franz Josef Glacier township, Westland, New Zealand. Sackung indicated by arrows; the Alpine fault is shown by the white dashed line.

Finally, it is important in considering future landslide disasters to realize that the dominant role of landslides in mountain geomorphology means that landslides in mountains are likely to trigger consequential geomorphic events of a variety of types, whose effects can propagate many tens or even hundreds of kilometers across a landscape, and can persist for many decades: Examples include

• A large landslide falling into a mountain valley is likely to affect, or even block, the river in the valley (Korup and Wang, 2014).

• A blockage forms a landslide dam, that can fail when overtopped (immediately or many decades later) causing a short but severe flood to pass along the valley (e.g., Hancox et al., 2005).

• Flooding upstream before the dam fails can also be troublesome, but less catastrophic.

• The large input of sediment to a river from a large landslide (or indeed a large number or small landslides), whether or not the river is blocked, causes the river behavior to alter—typically the river will aggrade to increase its slope and sediment transport capacity, increasing flood risk to the valley floor and to downstream floodplains (e.g., Davies and Korup, 2007; Robinson and Davies, 2013).

• A landslide in a small steep catchment can substantially increase debris flow risk in that catchment and on its fan.

• Large landslides falling into lakes or bays can cause catastrophic tsunami damage to assets close to water level (e.g., Lituya Bay, Alaska, 1954: Weiss et al., 2009).

• A large landslide falling onto a glacier can trigger a far-reaching rock-ice avalanche (e.g., Kolka-Karmodon: Huggel et al., 2005; Sosio, 2014; and Huascaran: Pflaker and Ericksen, 1978).

Thus, assessing the risks consequential on the occurrence of a landslide is no simple