NATURAL HAZARDS by : EDWARD BRYANT

http://www.cambridge.org/9780521537438

This page intentionally left blank

Natural hazards afflict all corners of the Earth; often unexpected, seemingly unavoidable and frequentlycatastrophic in their impact.

This revised edition is a comprehensive, inter-disciplinary treatment of the full range of natural hazards.Accessible, readable and well supported by over 150 maps, diagrams and photographs, it is a standard text forstudents and an invaluable guide for professionals in the field.

Clearly and concisely, the author describes and explains how hazards occur, examines prediction methods,considers recent and historical hazard events and explores the social impact of such disasters. This revised editionmakes good use of the wealth of recent research into climate change and its effects.

Edward Bryant is Associate Dean of Science at Wollongong University in Australia. Among his other publicationsis Tsunami: The Underrated Hazard (Cambridge University Press, 2001). He has particular interest in climaticchange and coastal evolution.

Praise for the First Edition:‘Professor Bryant’s heroic compilation is an excellent guide.’Scientific American

NATURAL HAZARDS

EDWARD BRYANT

S E C O N D E D I T I O NNATURALHAZARDS

Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo

Cambridge University PressThe Edinburgh Building, Cambridge , UK

First published in print format

- ----

- ----

© Edward Bryant 2005

2005

Information on this title: www.cambridge.org/9780521537438

This book is in copyright. Subject to statutory exception and to the provision ofrelevant collective licensing agreements, no reproduction of any part may take placewithout the written permission of Cambridge University Press.

- ---

- ---

Cambridge University Press has no responsibility for the persistence or accuracy ofs for external or third-party internet websites referred to in this book, and does notguarantee that any content on such websites is, or will remain, accurate or appropriate.

Published in the United States of America by Cambridge University Press, New York

www.cambridge.org

paperback

eBook (EBL)eBook (EBL)

paperback

http://www.cambridge.org

http://www.cambridge.org/9780521537438

To Dianne, Kate and Mark

List of illustrations ixPreface xiiiAcknowledgements xv

1 Introduction to Natural Hazards 1

Rationale 1Historical background 2Chapter outlines 10References and further reading 13

Climatic Hazards

2 Mechanisms of Climate Variability 17

Introduction 17Models of atmospheric circulation and change 17Astronomical cycles 37Concluding comments 41References and further reading 42

3 Large-scale Storms as a Hazard 44

Introduction 44Tropical cyclones 44Extra-tropical cyclones 58Snowstorms, blizzards and freezing rain 67Storm surges 73Probability of occurrence 76

Dust storms 78Concluding comments 83References and further reading 83

4 Localized Storms 85

Introduction 85Thunderstorms, lightning and hail 85Tornadoes 93Concluding comments 102References and further reading 102

5 Drought as a Hazard 103

Introduction 103Pre-colonial response to drought 103Post-colonial response 105Drought conditions exacerbated by modern societies 106Modern response to droughts 106International response 113Private responses: Bob Geldof 115Concluding comments 118References and further reading 119

6 Flooding as a Hazard 120

Introduction 120Flash floods 120High-magnitude, regional floods 131

Contents

Concluding comments 137References and further reading 137

7 Fires in Nature 139

Introduction 139Conditions favoring intense bushfires 140Causes of fires 143Bushfire disasters: world perspective 144Concluding comments 153References and further reading 154

8 Oceanic Hazards 155

Introduction 155Waves as a hazard 156Sea-ice as a hazard 162Sea level rise as a hazard 166Beach erosion hazard 170Concluding comments 174References and further reading 175

Geological Hazards

9 Causes and Prediction of Earthquakes and Volcanoes 179

Introduction 179Scales for measuring earthquake intensity 179Distribution of earthquakes and volcanoes 182Causes of earthquakes and volcanoes 184Prediction of earthquakes and volcanoes 189Concluding comments 200References and further reading 200

10 Earthquakes and Tsunami as Hazards 202

Types of shock waves 202Seismic risk maps 204Earthquake disasters 204Liquefaction or thixotropy 211Tsunami 214

Concluding comments 224References and further reading 225

11 Volcanoes as a Hazard 227

Introduction 227Types of volcanic eruptions 229Volcanic hazards 230Volcanic disasters 238Concluding comments 247References and further reading 247

12 Land Instability as a Hazard 249

Introduction 249Soil mechanics 249Shear strength of soils 251Classification of land instability 254Subsidence 267Concluding comments 268References and further reading 269

Social Impact

13 Personal and Group Response to Hazards 273

Introduction 273Before the event 273Dealing with the event and its aftermath 277Additional impacts 283References and further reading 286

14 Epilogue 288

Changing hazard regimes 288Modern consequences of natural hazards 290References and further reading 291

Select glossary of terms 292Index 304

viii Contents

1.1 Incidence of natural hazards by region

1975–2001 7

1.2 Reporting incidence of natural hazards

1900–2001 8

1.3 Deaths from natural hazards 1900–2001 9

2.1 Wind movements relative to isobars 18

2.2 Palmén–Newton schematic model of

general air circulation 19

2.3 Mid-latitude westerly and easterly trade

wind belts 20

2.4 Path of jet stream in the northern hemisphere 21

2.5 Mobile polar highs and resulting trade wind

circulation 23

2.6 Dynamic structure of mobile polar highs 24

2.7 El Niño–Southern Oscillation event 25

2.8 Indices of atmospheric pressure oscillations 26

2.9a Fans deposited during La Niña event 30

2.9b Landslides and washouts resulting from

La Niña 30

2.10 Southern Oscillation index, drought and

rainfall indices and river discharges 31

2.11 Time series of Southern Oscillation index 33

2.12 Hurricane days in North Atlantic–Carribean

Sea region 34

2.13 Phases of North Atlantic Oscillation 36

2.14 Winter storms in north Atlantic 36

2.15a Zurich sunspot numbers, 1500–2002 38

2.15b Zurich sunspot record between 1860

and 2002 38

2.16 Precession of the Moon’s orbit 39

2.17 The 18.6-year lunar orbit relative to the Earth 40

3.1 Hazards related to occurrence of tropical

cyclones 45

3.2 Location map 46

3.3 Origin and number of tropical cyclones 46

3.4 Potential temperature structure in

Hurricane Inez 47

3.5 Wind and pressure cross sections through

Hurricane Anita 48

3.6 Damage caused by Cyclone Tracy 52

3.7 Satellite image of Hurricane Andrew’s path 53

3.8 Sheet of plywood driven through a

Royal Palm 53

3.9a Effects of wind damage caused by

Cyclone Tracy 55

3.9b Building peeled back to beams by high winds 55

3.10 Path of 1970 cyclone in Bay of Bengal 58

3.11 Development of an extra-tropical depression 59

3.12 Incidence of severe North Sea and English

Channel storms 60

3.13 Historical, erosive North Sea storms 61

3.14 Pressure pattern over northern Europe, 1588 62

3.15 Schematic development of an east-coast low 63

3.16 Satellite image of Halloween storm, 1991 64

3.17 Ash Wednesday storm damage, 1962 65

3.18 Storms of 25 May–15 June 1974, Sydney 66

3.19a Undermining and collapse of beachfront

houses 66

Illustrations

3.19b Collapsed picnic shelter 66

3.20 Precipitation patterns, mid-latitude cyclonic

depression 68

3.21 Pressure patterns for 1984 snowstorm 69

3.22 Effect of a heavy snowstorm, Newfoundland 69

3.23 Equivalent temperature or wind chill chart 72

3.24 Effect of freezing precipitation and wet snow 73

3.25 Track of Hurricane Carol 75

3.26 Effect of wave shoaling on long wave height 75

3.27 Recurrence intervals of North Sea storm

surges 77

3.28 Probability exceedence diagram for Fig 3.27

data 78

3.29 Dust storm bearing down on Melbourne 79

3.30 Trajectories of dust storms 80

3.31 Extent and frequency of dust storms in

United States 82

4.1 Annual frequency of thunder days in

United States 86

4.2 Average annual thunder days in Australia 87

4.3 Location map 88

4.4 Separation of charged water droplets 89

4.5 Annual frequency of hailstorms in

United States 91

4.6a Large hailstones 92

4.6b Layered hailstone 92

4.7 Tornado generation within a supercell

thunderstorm 94

4.8 Development of a mountainado 95

4.9 Waterspout striking the Trombes 95

4.10 First tornado captured by NSSL 96

4.11 Model for multiple-vortex formation in a

tornado 96

4.12 Average tornadoes per year and preferred

paths 98

4.13 Tornadoes and tornado deaths in United

States 98

4.14 Effect of topography on passage of a tornado 101

5.1 Map of semi-arid regions and the Sahel

in Africa 104

6.1 Maximum rainfall amounts versus time 121

6.2 Location map 121

6.3 Stream power per unit boundary area 123

6.4 Channel depth verus current velocity 124

6.5 Abnormal patterns of airflow, leading to

flash flooding 125

6.6 Maximum probably rainfalls per unit time 126

6.7 Forty-eight-hour isohyets for the Dapto flood 128

6.8 Synoptic pattern for Sydney flash floods 129

6.9 Isohyets for three flash flood events in Sydney 130

6.10 Flood locations 1985–2003 131

6.11 Mississippi River drainage basin and flood

extents 132

6.12 Occurrence of major floods on Mississippi

River 132

6.13 Major rivers of eastern Australia 133

6.14 Hwang Ho River channel: 2300 BC to

present 136

7.1 William Strutt’s painting Black Thursday 139


7.3 Synoptic weather pattern conducive to

bushfire 141

7.4a Crown fire in a Eucalyptus forest 142

7.4b Crown fire in a boreal forest 142

7.5 Whirlwind associated with a clearing

after burning 143

7.6 Destruction caused by southern Californian

fires 148

7.7 Pattern of seasonal bushfires in Australia 149

7.8 Extent of bushfires, location of spot fires 150

7.9 Devastation of Fairhaven following Ash

Wednesday 152


8.2 Schematic representation of wave

characteristics 157

8.3 Wave behavior in shallow water 158

8.4 Worldwide distribution of altimeter wave

heights 160

8.5 Deep-water swell waves, Wollongong 162

8.6 Worldwide distribution of sea-ice 163

8.7 Composite April sea-ice extent 163

8.8 Pack-ice Cornwall Island, Canadian Arctic 166

8.9 Annual sea level changes 167

8.10 Changes in global sea level 168

8.11 Global sea level change 1992–2002 169

8.12 The Bruun Rule for shoreline retreat with

sea level rise 171

8.13 Difference in beach volume, Stanwell Park 173

8.14 Linear regression lines, Stanwell Park Beach 174

9.1 Plate boundaries 181


9.3 Return intervals of earthquakes 183

9.4 Origin of volcanism: mid-ocean plate

spreading 184

9.5 Origin of volcanism over a subduction zone 185

9.6 Movement of Hawaiian Islands drifting over

hot spot 186

9.7 Hot spots and island strings on Pacific Plate 187

x Illustrations

TABLES

1.1 Frequency of natural hazards during the twentieth century 7

1.2 Cost of natural hazards, 1900–2001 71.3 Number of people killed, injured or displaced

due to natural hazards 91.4 Ranking of hazard characteristics

and impact 112.1 Comparison of drought in Indonesia with

El Niño–Southern Oscillation events 32

xi

2.2 Timing of flood and drought with 18.6 year lunar tide 41

3.1 Lowest central pressures recorded in tropical cyclones 49

3.2 Scales for measuring cyclone intensity 504.1 Fujita Tornado Intensity Scale and

frequency of tornadoes in the USA 978.1 Wave energy and power for typical

ocean waves 159

Illustrations

9.8 Faults associated with earthquake activity 188

9.9 Probability of eruptions of individual

volcanoes per century 190

9.10 Recent volcanism (1600–2000) 191

9.11 Magnitude on the Richter scale of major

earthquakes 193

9.12 Major seismic gaps in the Pacific ‘ring

of fire’ 194

10.1 Seismic waves and their travel behavior 203

10.2 Global seismic risk map 204


10.4 Earthquake and tsunami characteristics 207

10.5 Head scarp of landslide, Anchorage, Alaska 207

10.6 Historically active fault systems of southern

California 208

10.7 Collapsed sections of Cypress viaduct 209

10.8 Distribution of faulting for Tokyo

earthquake, 1923 210

10.9 Cumulative effect of compressional P-waves

on pore water pressure 212

10.10 Liquefaction following 1964 Niigata

earthquake 213

10.11a Spatial distribution of tsunami since 47 BC 215

10.11b Source areas for Pacific-wide events 215

10.12 Plots of tsunami wave trains in Pacific region 217

10.13 Tsunami wave refraction pattern 217

10.14 Sequential photos of 1957 tsunami, Hawaii 218

10.15 Ships stranded by tsunami, 1868 218

10.16 Run-up heights of tsunami, Hawaiian

Islands 219

10.17 Pacific Tsunami Warning System 223


11.2 Tephra ejecta from Mt St Helens 233

11.3 Beginning of pyroclastic flow 234

11.4 Location map of Mt St Helens 236

11.5 Lahar following eruption of Mt Pinatubo 237

11.6 Region affected by Santorini eruption 239

11.7 Location map of Mt Vesuvius 241

11.8 Sundra Strait coastline affected by tsunami 242

11.9 Volcanoes and earthquakes in Carribean

region 244

11.10 Location map of Mt Pelée 245

11.11 St Pierre after eruption of Mt Pelée 246

11.12 North side of Mt St Helens after 1980

eruption 247

12.1 Effect of gravity on an object 250

12.2 Strength of bonding on particles of

different sizes 251

12.3 Schematic representation of Mohr–Coulomb

equation 252


12.5 Schematic representation of Atterberg limits 254

12.6 Classification of land instability 255

12.7 Extent of expansive soils in the United

States 256

12.8 Effect of alternate wetting and drying of

expansive soils 257

12.9 Downhill movement caused by soil creep 258

12.10 Widespread debris flows in Kiwi Valley area 260

12.11 Erosion and debris flow deposition after

urban flood 261

12.12 Process of rotational sliding 261

12.13 Complexity of landslide initiation over time 262

12.14 Mega-landslide, northern slopes of

Nakanai Range 263

12.15 Extent of landslides in the coterminous

United States 264

12.16 Process and impact forces generated by

snow avalanche 266

13.1 Drinkers in Lorne watching Ash Wednesday

bushfire 274

14.1 Imbricated boulders stacked against cliff face 288

8.2 Factors controlling beach erosion 1729.1 Annual frequency of earthquakes 1809.2 Some of the largest earthquakes 1809.3 The Mercalli scale of earthquake intensity 18110.1 Major historical earthquakes by death toll 20510.2 Percentage of tsunami in the world’s

oceans and seas 21510.3 Earthquake magnitude, tsunami magnitude

and tsunami run-up heights in Japan 219

10.4 Soloviev’s scale of tsunami magnitude 22010.5 Causes of tsunami in the Pacific 22010.6 Large death tolls from tsunami in the Pacific

Ocean region 22012.1 US Highway Research Board Landslide

Committee classification of mass movements 255

13.1 Frequency of psychological problems amongst young survivors of Cyclone Tracy 285

xii Illustrations

I have reread the preface to the first edition manytimes: extreme events, dire warnings about Green-house warming, El Niño–Southern Oscillation predic-tion . . . Little has changed in the fifteen years sinceI wrote about them. I am still perplexed becauseextreme events continue to happen and globalwarming is no closer to occurring. As Sydney inFebruary 2004 experienced a heat wave of a magnitudeunprecedented since at least 1939, I was chasing myfavourite research topic – cosmogenically inducedmega-tsunami – on Stewart Island, New Zealand sometwo thousand kilometres away where an unprece-dented cold snap was occurring. One event witnessedby four million people got all the publicity; the otherplayed out in a remote cabin in front of half of dozentrekkers got none. Yet both climatic extremes wereproduced by the same pattern of atmospheric circula-tion controlled by the same sequence of mobile polarhighs. Sydney lay on the equatorial ‘greenhouse’ side ofthe highs and Stewart Island lay on the poleward ‘IceAge’ side. This book covers two of the phenomenaI experienced in my February of extremes – mobilepolar highs and tsunami. As with the first edition, thebook does not cover the third phenomena, Green-house warming. This book is about everyday climaticand geological hazards that can be explained, pre-dicted, and alleviated. Global warming is mentionedand is covered by the concept of changing hazardregimes. However, heat waves – and cold snaps – are

about everyday hazards that we have lived with, willcontinue to experience, and hopefully can survive.These concepts are what this second edition is about.

In order to convey this point of view clearly in thebook, adherence to academic referencing has been keptto a minimum. Usually each section begins by listing themajor papers or books on a topic that have influencedmy thinking and writing. Full reference to these publi-cations can be found at the end of each chapter. I apol-ogize to anyone who feels that their crucial work hasbeen ignored; but the breadth of coverage in thistextbook precluded a complete review of the literatureon many topics including some of my own.

Manuscript preparation is quite different now fromwhat it was in the late 1980s when the first edition wasbeing published. For one thing, the software programsfor scanning, image enhancement, graphics, and wordprocessing are far more comprehensive and efficientat doing tasks. The diagrams in the first edition werehand-drawn, a technique that is rarely used today.Readers will find that many of those diagrams remainin this version. However, many have been revampedusing graphics software. New computer-prepareddiagrams have also been added. Word processingpackages now allow spelling and grammar to bechecked uniformly without the assistance of a copy-editor. Minor changes have been made to text retainedfrom the first edition using this capability. TheInternet was in it infancy when the first edition was

Preface

prepared. It is now very easy to capture all the argu-ments or theories related to a hazard topic via thismedium. Where the Internet was used to prepare thecurrent edition, the web sites have been referenced inthe text and their full addresses appended to individ-ual chapters. The reader should be aware that some of

these addresses may not be available to them becausethey have changed, or because of the lack of anarchival tradition on the Internet.

Ted Bryant9 August 2004

xiv Preface

Several people generously provided data from theirown research as follows: Jane Cook, formerly of theFaculty of Education, University of Wollongong for herKiama landslip data in figure 2.11; Professor Eric Bird,formerly of the Department of Geography, Universityof Melbourne for updating table 8.2; Dr. GeoffBoughton, TimberED Services Pty Ltd., Glengarry,Western Australia, for information on the lowerlimits of seismically induced liquefaction; Col Johnson,New South Wales South Coast State EmergencyServices for pointing out the importance of CriticalIncident Stress Syndrome amongst rescue workers;Rob Fleming, Australian Counter Disaster College,Mt Macedon, Victoria for access to photographic fileson hazards in Australia; and Bert Roberts, Schoolof Earth and Environmental Sciences, University ofWollongong for pointing out the relative literature onuniformitarianism.

The following people or publishers permitteddiagrams or tables to be reproduced: University ofWisconsin Press, for figure 2.4; Methuen, London forfigures 3.12 and 3.14; Harcourt Brace JovanovichGroup, Academic Press, Sydney for figures 9.9 and11.7; Butterworths, Sydney for figures 12.2 and 12.13;and Dr. John Whittow, former at the Department ofGeography, University of Reading for figures 12.12 and12.16. Acknowledgment is made to the following orga-nizations who did not require permission on copy-righted material: NOAA for figures 3.7, 3.8, 4.6a & b,

4.9, 4.10, and 11.5; United States Jet PropulsionLaboratory for figure 8.4; the United States GeologicalSurvey for figures 10.5, 10.7, 10.10, 10.14, 10.15, 11.2,11.3, and 12.11; the United States Marine FisheriesService, Seattle for Table 2.1; and the American Geo-physical Union for Table 2.2. All of the above diagramsand tables are acknowledged fully in the text.

The following people or organizations providedphotographic material: John Fairfax and Sons Limited,for figures 3.6, 7.9 and 13.1; Dr. Geoff Boughton,TimberED Services Pty Ltd., Glengarry, WesternAustralia, for figures 3.9a & b; the Canadian Atmos-pheric Environment Service for figures 3.22 and 3.24;the Australian Bureau of Meteorology for figure 3.29;John Telford, Fairymeadow, New South Wales forfigure 8.5; Dr. Ann Young, for figure 8.13; Bob Webb,Mal Bilaniwskyj and Charles Owen, New South WalesDepartment of Main Roads, Wollongong and BellambiOffices for figure 2.9a and b; the State Library ofVictoria who kindly permitted the reproduction ofWilliam Strutt’s painting entitled Black Thursdayappearing as figure 7.1; N.P. Cheney, Director, andA.G. Edward, National Bushfire Research Unit,CSIRO, Canberra for figures 7.4a and 7.5 respectively;Jim Bryant, for figure 7.4b; Director Naval VisualNews Service, US Navy Office of Information forfigure 7.6; J.L. Ruhle and Associates, Fullerton, Cali-fornia for figure 11.12; A.G. Black, Senior Soil Conser-vator, Hawke’s Bay Catchment Board (Hawke’s Bay

Acknowledgements

Regional Council), Napier, New Zealand for figure12.10; and Dr. Peter Lowenstein for figure 12.14,courtesy C.O. McKee, Principal Government Volca-nologist, Rabaul Volcanological Observatory, Depart-ment of Minerals and Energy, Papua New GuineaGeological Survey.

I am indebted to the staff and associates of Cam-bridge University Press, Melbourne: Peter Debus whoencouraged a second edition of the hazards text, SusanKeogh who oversaw the publishing process, MoniqueKelly who processed the figures, and Malory Westonwho copyedited the manuscript. The majority of this

text has been presented as lecture material. I wouldlike to thank many students in my first and secondyear Hazards and Climatology classes who challengedmy ideas or rose enthusiastically to support them.Unfortunately, current students have no access to thesesubjects. I hope that this textbook makes up for thedeficiency. Finally, I would like to thank my familywho kindly tolerated my ‘next textbook’ and deflectedmany of the requests upon my time by friends whodid not realize the magnitude of the task of writing asecond edition.

xvi Acknowledgements

RATIONALE

The field of environmental studies is usually intro-duced to students as one of two themes. The firstexamines human effects upon the Earth’s environ-ment, and is concerned ultimately with the question ofwhether or not people can irreversibly alter that envi-ronment. Such studies include the effect of humanimpact on climate, of land-use practices on the land-scape in prehistoric and recent times, and of nuclearwar upon the Earth’s environment. The second themetotally disregards this question of human impact on theenvironment. It assumes that people are specks of dustmoving through time subject to the whims of nature.In this sense, calamities are ‘acts of God’, events thatmake the headlines on the evening news, events youmight wish on your worst enemy but would never wantto witness yourself.

University and college courses dealing with this lattertheme usually treat people as living within a hostileenvironment over which they have little control. Suchcourses go by the name of ‘“Natural” Hazards’. The dif-ference between the two themes is aptly summarizedby Sidle et al. (2003). Both themes describe hazards.The first theme can be categorized as chronic while thesecond is episodic or periodic. Chronic hazards wouldinclude desertification, soil degradation, and melting ofpermafrost. The causes could be due to humans orglobal warming. Periodic hazards are large magnitude

events that appear over a short time period. Theyinclude phenomena such as earthquakes, tsunami,volcanic eruptions and flash floods. This book dealsmainly with episodic hazards, events that wreak havocupon humans. However, the message is not one of totaldoom and gloom. Humans choose hazardous areasbecause they often offer benefits. For example, ashfrom volcanoes produces rich soils that can grow threecrops per year in the tropics, and floodplains provideeasily cultivated agricultural land close to a watersupply. Humans in these environments are forced topredict and avoid natural calamities such as landslides,cyclones, earthquakes, and drought. Or as Middleton(1995) states, ‘a hazard should be seen as an occasion-ally disadvantageous aspect of a phenomenon, whichis often beneficial to human activity over a differenttimescale.’ For those hazards that cannot be avoidedby humans, this book also examines how peoplecan minimize their effects and rectify their negativeconsequences.

The book is organized around climatic and geologi-cal hazards. Intense storms and winds, oceanographicfactors such as waves, ice and sea level changes,together with extreme precipitation phenomena, areall investigated under the heading of climatic hazards.Earthquakes, volcanoes, and land instability areexamined under the topic of geological hazards. Thisbook is not concerned overly with biological hazardssuch as plague, disease, and insect infestations; but one

Introduction toNatural Hazards

C H A P T E R 1

should realize that death and property loss from theselatter perils can be just as severe as, if not worse than,those generated by geological and climatic hazards. Afinal section examines the social impact of hazards.

HISTORICAL BACKGROUND

The world of myths and legends(Holmes, 1965; Day, 1984; Myles, 1985; Milne, 1986;Bryant, 2001)

Myths are traditional stories focusing on the deeds ofgods or heroes, often to explain some natural phenom-enon, while legends refer to some historical eventhanded down – usually by word of mouth in traditionalsocieties. Both incorporate natural hazard events. Pre-historic peoples viewed many natural disasters withshock and as a threat to existence. Hazards had to beexplained and were often set within a world of animisticgods. Such myths remain a feature of belief systems formany peoples across the world – from Aborigines inAustralia and Melanesians in the South Pacific, to theindigenous inhabitants of the Americas and the ArcticCircle. With the development of written language,many of these oral myths and legends were incor-porated into the writings of the great religions of theworld. Nowhere is this more evident than with storiesabout the great flood. Along with stories of Creation,the Deluge is an almost ubiquitous theme. The biblicalaccount has remarkable similarities to the BabylonianEpic of Gilgamesh. In this epic, Utnapishtim, whoactually existed as the tenth king of Babylon, replacesNoah, who was the tenth descendant of Adam. Thebiblical account parallels the epic, which tells of thebuilding of an ark, the loading of it with different animalspecies, and the landing of the ark upon a remotemountaintop. Babylonian civilization was founded inwhat was called Mesopotamia, situated in the basin ofthe Euphrates and Tigris Rivers, which historically havebeen subject to cataclysmic flooding. Clay deposits inexcess of 2 m have been found at Ur, suggesting atorrential rain event in the headwaters of the valley thatcaused rivers to carry enormous suspension loads athigh concentrations.

Such an event was not unique. The epic and biblicalaccounts are undoubtedly contemporaneous and, infact, a similar major flood appears in the writtenaccounts of all civilized societies throughout MiddleEastern countries. For example, broken tablets found in

the library of Ashurbanipal I (668–626 BC) of Assyria, atNippur in the lower Euphrates Valley, also recount theBabylonian legend. The Deluge also appears worldwidein verbal and written myths. The Spanish conquista-dores were startled by flood legends in Mexico that borea remarkable similarity to the biblical account. Becausethese staunchly Catholic invaders saw the legends asblasphemous, they ordered the destruction of all writtennative references to the Deluge. The similarities weremost remarkable with the Zapotecs of Mexico. The heroof their legend was Tezpi, who built a raft and loaded itwith his family and animals. When the god Tezxatlipocaordered the floodwaters to subside, Tezpi released firstlya vulture and then other fowl, which failed to return.When a hummingbird brought back a leafy twig,Tezpi abandoned his raft on Mt Colhuacan. A Hawaiianlegend even has a rainbow appearing, which signalsa request for forgiveness by their god Kane for hisdestructiveness. Almost all legends ascribe the flood tothe wrath of a god. The Australian Aborigines have astory about a frog named Tiddalik drinking all the waterin the world. To get the water back, all the animals tryunsuccessfully to get the frog to laugh. Finally, a cavort-ing eel manages the task. However, the frog disgorgesmore water than expected and the whole Earth isflooded. In some myths, floods are attributed to animalsurinating after being picked up. Many myths areconcerned with human taboo violations, women’smenstruation, or human carelessness. Almost all storieshave a forewarning of the disaster, the survival of achosen few, and a sudden onset of rain. However, theBible has water issuing from the ground, as do Chinese,Egyptian, and Malaysian accounts. The latter aspectmay refer to earthquakes breaking reservoirs orimpounded lakes. Many Pacific island and Chileanlegends recount a swelling of the ocean, in obvious ref-erence to tsunami. Despite the similarities of legendsglobally amongst unconnected cultures, there is no geo-logical evidence for contemporaneous flooding of theentire world, or over continents. It would appear thatmost of the flood myths recount localized deluges, andare an attempt by human beings to rationalize the floodhazard.

Almost as ubiquitous as the legends about flood areones describing volcanism or the disappearance ofcontinents. North American natives refer to sunken landto the east; the Aztecs of Mexico believed they weredescended from a people called Az, from the lost landof Aztlán. The Mayas of Central America referred to a

2 Natural Hazards

lost island of Mu in the Pacific, which broke apart andsank with the loss of millions of inhabitants. Chineseand Hindu legends also recount lost continents.Perhaps the most famous lost continent legend is thestory of Atlantis, written by Plato in his Critias. Herecounts an Egyptian story that has similarities withCarthaginian and Phoenician legends. Indeed, if theAztecs did migrate from North Africa, the Azteclegend of a lost island to the east may simply be theMediterranean legend transposed and reorientedto Central America. The disappearance of Atlantisprobably had its origins in the catastrophic eruption ofSantorini around 1470 BC. Greek flood myths refer tothis or similar events that generated tsunami in theAegean. It is quite possible that this event enteredother tales in Greek mythology involving fire from thesky, floating islands and darkening of the sky by Zeus.The Krakatau eruption of 1883 also spawned manylegends on surrounding islands. Blong (1982) describesmodern legends in Papua New Guinea that can berelated to actual volcanic eruptions in the seventeenthcentury. The Pacific region, because of its volcanicactivity, abounds in eruption mythology. For example,Pele, the fire goddess of Hawaii, is chased by her sisterfrom island to island eastward across the Pacific. Eachtime she takes refuge in a volcano, she is found byher sister, killed, and then resurrected. Finally, Peleimplants herself triumphantly in the easternmostisland of Hawaii. Remarkably, this legend parallels thetemporal sequence of volcanism in the HawaiianIslands.

Earthquakes also get recounted in many myths.They are attributed by animistic societies in India toanimals trying to get out of the Earth. Sometimes thewhole Earth is viewed as an animal. Ancient Mexicansviewed the Earth as a gigantic frog that twitched itsskin once in a while. Timorese thought that a giant sup-ported the Earth on his shoulder and shifted it to theother side whenever it got too heavy. Greeks did notascribe earthquakes to Atlas, but to Poseidon disturb-ing the waters of the sea and setting off earth tremors.The oscillating water levels around the Greek islandsduring earthquakes were evidence of this god’s actions.In the Bible, earthquakes were used by God to punishhumans for their sins. Sodom and Gomorrah weredestroyed because they refused to give up sins of theflesh. The fire and brimstone from heaven, mentionedwith these cities’ destruction, were most likely associ-ated with the lightning that is often generated above

earthquakes by the upward movement of dust. Alter-natively, it may have come from meteorite debris. TheBible also viewed earthquakes as divine visitation. Anearthquake preceded the arrival of the angel sent toroll back Christ’s tombstone (Matthew 28: 2).

Myths, legends, and the Bible pose a dilemmabecause they do not always stand up to scientificscrutiny as established in the last three centuries. Todayan incident is not credible unless it can be measured,verified by numerous witnesses, viewed as repeatable,and certainly published in peer-reviewed literature.Media accounts, which may be cobbled together andsubject to sensationalism because of the underlying aimto sell a product, are regarded with suspicion. There isdistrust of events that go unpublished, that occur inisolated regions, or that are witnessed by societies withonly an oral tradition occurring in isolated regions.Legends, however, can become credible. Hence, theKwenaitchechat legend of a tsunami on the west coastof the United States suddenly took on scientific accept-ability when it received front-page coverage in Nature.It was shown – using sophisticated computer modelling– that the source of a tsunami that struck the east coastof Japan on 26 January 1700 could have originated onlyfrom the Cascadian subduction zone off the west coastof the United States. The fact that the legend mentionswater overrunning hilltops has still not been explained.Sometime in the future, we will find it just as hard tobelieve that published statements were accepted asarticles of faith when they were scrutinized by only tworeferees who often came from an entrenched innercircle of associates. Certainly, the rise of informationon the Internet is challenging twentieth centuryperceptions of scientific scrutiny.

Catastrophism vs. uniformitarianism(Huggett, 1997)

Until the middle of the eighteenth century, westerncivilization regarded hazards as ‘acts of God’ in thestrict biblical sense, as punishment for people’s sins.On 1 November 1755, an earthquake, with a possiblesurface magnitude of 9.0 on the Richter scale,destroyed Lisbon, then a major center of Europeancivilization. Shortly after the earthquake, a tsunamiswept into the city and, over the next few days, fireconsumed what was left. The event sent shock wavesthrough the salons of Europe at the beginning of TheEnlightenment. The earthquake struck on All Saints’Day, when many Christian believers were praying in

3Introduction to Natural Hazards

church. John Wesley viewed the earthquake as God’spunishment for the licentious behavior of believersin Lisbon, and retribution for the severity of thePortuguese Inquisition. Immanuel Kant and Jean-Jacques Rousseau viewed the disaster as a naturalevent, and emphasized the need to avoid building inhazardous places.

The Lisbon earthquake also initiated scientific studyof geological events. In 1760, John Mitchell, GeologyProfessor at Cambridge University, documented thespatial effects of the earthquake on lake levels through-out Europe. He found that no seiching was reportedcloser to the city than 700 km, nor further away than2500 km. The seiching affected the open coastline ofthe North Sea and shorelines in Norwegian fjords,Scottish lochs, Swiss alpine lakes, and rivers and canalsin western Germany and the Netherlands. He deducedthat there must have been a progressive, wave-liketilting movement of the Earth outward from the centerof the earthquake, different to the type of waveproduced by a volcanic explosion.

Mitchell’s 1760 work on the Lisbon earthquake effec-tively represents the separation of two completely dif-ferent philosophies for viewing the physical behavior ofthe natural world. Beforehand, the Catastrophists dom-inated geological methodology. These were the peoplewho believed that the shape of the Earth’s surface, thestratigraphic breaks appearing in rock columns, andthe large events associated with observable processeswere cataclysmic. Catastrophists believed events had tobe cataclysmic to allow the geological record to fit withthe date for the Earth’s creation of 4004 BC, as deter-mined by Biblical genealogy. Charles Lyell, one of thefathers of geology, sought to replace this ‘catastrophetheory’ with gradualism – the idea that geological andgeomorphic features were the result of cumulative slowchange by natural processes operating at relativelyconstant rates. This idea implied that processes shapingthe Earth’s surface followed laws of nature as defined byphysicists and mathematicians as far back as Bacon.William Whewell, in a review of Lyell’s work, coined theterm uniformitarianism, and subsequently a protracteddebate broke out about whether or not the slowprocesses we observe at present apply to past unobserv-able events. The phrase ‘The present is the key to thepast’ also arose at this time and added to the debate.

In fact, the idea of uniformitarianism involves twoconcepts. The first implies that geological processesfollow natural laws applicable to science. There are no

‘acts of God’. This type of uniformitarianism wasestablished to counter the arguments raised by theCatastrophists. The second concept implies constancyof rates of change or material condition through time.This concept is nothing more than inductive reasoning.The type and rate of processes operating today charac-terize those that have operated over geological time.For example, waves break upon a beach today in thesame manner that they would have one hundredmillion years ago, and prehistoric tsunami behave thesame as modern ones described in our written records.If one wants to understand the sedimentary deposits ofan ancient tidal estuary, one has to do no more than goto a modern estuary and study the processes at work.Included in this concept is the belief that physical land-scapes such as modern floodplains and coastlinesevolve slowly.

Few geomorphologists or geologists who study earthsurface processes and the evolution of modern land-scapes would initially object to the above concept ofconstancy. However, it does not withstand scrutiny.For example, there is no modern analogy to thenappe mountain building processes that formed theEuropean Alps or to the mass extinctions and suddendiscontinuities that have dominated the geologicalrecord. Additionally, no one who has witnessed afault line being upthrusted during an earthquake, orMt St Helens wrenching itself apart in a cataclysmiceruption, would agree that all landscapes developslowly. As Thomas Huxley so aptly worded it, gradual-ists had saddled themselves with the tenet of Naturanon facit saltum – Nature does not make suddenjumps. J. Harlen Bretz from the University of Chicagochallenged this tenet in the 1920s. Bretz attributed theformation of the scablands of eastern Washington tocatastrophic floods. For the next forty years, he borethe ridicule and invectiveness of the geological estab-lishment for proposing this radical idea. It was not untilthe 1960s that Bretz was proved correct when VicBaker of the University of Arizona interpreted spaceprobe images of enormous channels on Mars asfeatures similar to the Washington scablands. At theage of 83, Bretz finally received the recognition of hispeers for his seminal work.

Convulsive events are important climatic and geolog-ical processes, but are they likely to occur today? This isa major question cropping up throughout this book.The climatic and geological processes responsible formost hazards can be described succinctly. However, do

4 Natural Hazards

these processes necessarily account for events occur-ring outside the historical record (in some places thisis a very short record indeed)? In some cases, theevidence for catastrophic events cannot be explained.For example, there is ongoing debate whether or notmega-tsunami or super storms can move boulders tothe top of cliffs 30 or more meters high. The mega-tsunami theory has been attacked – mainly because nohistorical tsunami generating similar deposits has beenwitnessed. However, the same argument can be appliedto the alternative hypothesis of mega-storms. Thedilemma of changing hazard regimes will be discussedat the end of this book.

The relationship between humansand natural hazards (Susman et al., 1983;Watts, 1983)

The above concepts emphasize the natural aspect ofnatural hazards. Disasters can also be viewed from asociological or humanistic viewpoint. An extensivechapter at the end of this book deals with the humanresponse to natural hazards at the personal or grouplevel. While there are great similarities in naturalhazard response amongst various cultures, societiesand political systems, it will be shown that fundamen-tal differences also occur. This is particularly evident inthe ways that countries of contrasting political ideology,or of differing levels of economic development, copewith drought. Both Australia and Ethiopia wereafflicted by droughts of similar severity at the begin-ning of the 1980s. In Ethiopia, drought resulted instarvation followed by massive international appeals forrelief. However, in Australia no one starved anddrought relief was managed internally.

In dealing historically with the scientific and mathe-matical description of natural hazards, it is apt todiscuss a sociological viewpoint formulated in thetwentieth century. This sociological viewpoint statesthat the severity of a natural hazard depends upon whoyou are, and to what society you belong at the time ofthe disaster. It is exemplified and expressed mostforcibly by Marxist theory. Droughts, earthquakes andother disasters do not kill or strike people in the sameway. The poor and oppressed suffer the most, experi-ence worst the long-term effects, with higher casualtiesmore likely as a result. In Marxist philosophy, it ismeaningless to separate nature from society: peoplerely upon nature for fulfillment of their basic needs.Throughout history, humans have met their needs by

utilizing their natural environment through labor input.Humans do not enter into a set contract with nature.In order to produce, they interact with each other, andthe intrinsic qualities of these interactions determinehow individuals or groups relate to nature. Labor thusbecomes the active and effective link between societyand nature. If workers or peasants are able to controltheir own labor, they are better adapted to contend withthe vagaries of the natural environment. The separationof workers from the means of production impliesthat others, namely those people (capitalists or bosses)who control the means of production, govern theirrelationship with nature. This state of affairs is self-perpetuating. Because capitalists control labor andproduction, they are better equipped to survive andrecover from a natural disaster than are the workers.

The main point about the Marxist view of naturalhazards is that hazard response is contingent upon theposition people occupy in the production process. Thissocial differentiation of reactions to natural hazards is awidespread phenomenon in the Third World, where itis typical for people at the margins of society to live inthe most dangerous and unhealthy locations. A majorslum in San Juan, Puerto Rico, is frequently inundatedat high tide; the poor of Rio de Janeiro live on theprecipitous slopes of Sugarloaf Mountain subject tolandslides; the slum dwellers of Guatemala live on thesteeper slopes affected by earthquakes; the Bangladeshifarmers of the Chars (low-lying islands at the mouth ofthe Ganges–Brahmaputra Delta in the Bay of Bengal)live on coastal land subject to storm surge flooding.

The contrast in hazard response between the ThirdWorld and westernized society can be illustrated bycomparing the effects of Cyclone Tracy – which struckDarwin, Australia, in 1974 – to the effects of hurricaneFifi, which struck Honduras in September of the sameyear. Both areas lie in major tropical storm zones, andboth possess adequate warning facilities. The tropicalstorms were of the same intensity, and destroyedtotally about 80 per cent of the buildings in the mainimpact zones. Over 8000 people died in Honduras,while only 64 died in Darwin. A Marxist would reasonthat the destitute conditions of under-development inHonduras accounted for the difference in the loss oflife between these two events. Society is differentiatedinto groups with different levels of vulnerability tohazards.

One might argue, on the other hand, that manypeople in Third World countries, presently affected by


major disasters, simply are unfortunate to be living inareas experiencing large-scale climatic change ortectonic activity. After all, European populations werebadly decimated by the climatic cooling that occurredafter the middle of the fourteenth century – andeveryone knows that the Los Angeles area is welloverdue for a major earthquake. One might also arguethat Third World countries bring much of the disasterupon themselves. In many countries afflicted by largedeath tolls due to drought or storms, the politicalsystems are inefficient, corrupt, or in a state of civil war.If governments in Third World countries took nationalmeasures to change their social structure, they wouldnot be so vulnerable to natural disasters. For example,Japan, which occupies a very hazardous earthquakezone, has performed this feat through industrializationin the twentieth century. One might further argue thatAustralians or Americans cannot be held responsiblefor the misery of the Third World. Apart from foreignaid, there is little we can do to alleviate their plight.

The Marxist view disagrees with all these views.Marxism stipulates that western, developed nations arepartially responsible for, and perpetuate the effect ofdisasters upon, people in the Third World – because weexploit their resources and cash-cropping modes of pro-duction. The fact that we give foreign aid is irrelevant. Ifour foreign aid is being siphoned off by corrupt officials,or used to build highly technical projects that do notdirectly alleviate the impoverished, then it may only bekeeping underprivileged people in a state of povertyor, it is suggested, further marginalizing the alreadyoppressed and thus exacerbating their risk to naturalhazards. Even disaster relief maintains the status quoand generally prohibits a change in the status of peasantsregarding their mode of production or impoverishment.

The basis of Marxist theory on hazards can be sum-marized as follows:• the forms of exploitation in Third World countries

increase the frequency of natural disasters as socio-economic conditions and the physical environmentdeteriorate;

• the poorest classes suffer the most;• disaster relief maintains the status quo, and works

against the poor even if it is intentionally directed tothem; and

• measures to prevent or minimize the effects ofdisasters, which rely upon high technology, reinforcethe conditions of underdevelopment, exploitation andpoverty.

Marxist philosophy does not form the basis of thisbook. It has been presented simply to make the readeraware that there are alternative viewpoints about theeffects of natural hazards. The Marxist view is basedupon the structure of societies or cultures, and howthose societies or cultures are able to respond tochanges in the natural environment. The frameworkof this book, on the other hand, is centered on thedescription and explanation of natural hazards that canbe viewed mainly as uncontrollable events happeningcontinually over time. The frequency and magnitude ofthese events may vary with time, and particular typesof events may be restricted in their worldwide occur-rence. However, natural hazards cannot be singled outin time and space and considered only when they affecta vulnerable society. An understanding of how, whereand when hazards take place can be achieved only bystudying all occurrences of that disaster. The arrival ofa tropical cyclone at Darwin, with the loss of 64 lives,is just as important as the arrival of a similar-sizedcyclone in Honduras, with the loss of 8000 lives.Studies of both events contribute to our knowledge ofthe effects of such disasters upon society and permit usto evaluate how societies respond subsequently.

Hazard statistics (Changnon & Hewings, 2001;CRED, 2002; Balling & Cerveny, 2003)

Figure 1.1 plots the number of hazards by region overthe period 1975–2001. The map excludes epidemics orfamine. Despite its large population, Africa is relativelydevoid of natural hazards. This is due to the tectonicstability of the continent and the fact that tropicalstorms do not develop here, except in the extremesouth-east. However the African population is affecteddisproportionately by drought. The frequency ofevents over the rest of the world reflects the domi-nance of climate as a control on the occurrence ofnatural hazards. This aspect is illustrated further forthe twentieth century in Table 1.1. The most frequenthazards are tornadoes, the majority of which occur inthe United States. Over the latter half of the twentiethcentury, their frequency – more than 250 events peryear – exceeds any other natural hazard. Climaticallyinduced floods and tropical cyclones follow this phe-nomenon in frequency. Tsunami, ranked fourth, is themost frequent geological hazard, followed by damagingearthquakes with nine significant events per year. Alltotalled, climatic hazards account for 86.2 per cent ofthe significant hazard events of the twentieth century.

6 Natural Hazards

The economic relevance of the biggest hundred ofthese hazard events is summarized over the sameperiod in Table 1.2. Values are reported in US dollarsand do not include inflation. Earthquakes, while

ranking only fifth in occurrence, have been the costli-est hazard ($US249 billion), followed by floods($US207 billion), tropical storms ($US80 billion), andwindstorms ($US44 billion). In total, the hundred mostexpensive natural disasters of the twentieth centurycaused $US631 billion damage. The single largestevent was the Kobe earthquake of 20 January 1995,which cost $US131.5 billion. While this event isfamiliar, the second most expensive disaster of thetwentieth century – floods in the European part ofthe former Soviet Union on 27 April 1991 costing$US60 billion – is virtually unknown.

Figure 1.2 presents the number of hazards reportedeach year over the twentieth century. Apparently,natural hazards have increased in frequency over this


1000

500250

No. of Events

Fig. 1.1 Incidence of natural hazards by region 1975–2001 (based on CRED, 2002).

Type of Hazard No of EventsTornadoes (US)1 9476Flood 2389Tropical Cyclone 1337Tsunami 986Earthquake 899Wind (other) 793Drought 782Landslide 448Wild fire 269Extreme temperature 259Temperate winter storm 240Volcano 168Tornadoes (non-US) 84Famine 77Storm surge 18

1 for F2–F5 tornadoes 1950–1995

Table 1.1 Frequency of natural hazards during the twentiethcentury. Tsunami data comes from the IntergovernmentalOceanographic Commission (2003). Tornado statistics forthe USA are for the 1990–1995 period only and derivedfrom the High Plains Regional Climate Center (2003). Allother data based on WHO (2002).

Type CostEarthquake $248,624,900,000Flood $206,639,800,000Tropical storm $80,077,700,000Wind Storm $43,890,000,000Wild Fire $20,212,800,000Drought $16,800,000,000Cold wave $9,555,000,000Heat wave $5,450,000,000Total $631,250,200,000

Table 1.2 Cost of natural hazards, summarized by type of hazardfor the 100 biggest events, 1900–2001 (based onWHO, 2002).

time span from around ten events per year at the begin-ning of the century to over 450 events at the end. Whileincreased vigilance and perception of natural hazardsaccounts for part of this increase, other data suggest thatnatural hazards are on the increase. The fact that thesefigures include both geological and climatic hazardsrules out such a simple explanation as global warming asa major cause of this increased incidence. On the otherhand, these figures should be treated with some cautionbecause they do not always hold under scrutiny. Forexample, in the United States, the magnitude and fre-quency of thunderstorms, hailstorms, intense tornadoes,hurricanes and winter storms have not increased, butmay have decreased. Over the same period, there is noincrease in the cost of climatic hazards after the figuresare adjusted for inflation. Instead, damaging eventsappear to recur at twenty-year intervals around1950–1954, 1970–1974, and 1990–1994. This timingcoincides with the peak of the 18.6 MN lunar cycle,which will be described in Chapter 2.

Figure 1.3 presents the number of deaths due tonatural hazards over the twentieth century. The timeseries does not include biological hazards such as epi-demics or insect predation; however, it does includedrought-induced famines. The number of deaths isplotted on a logarithmic scale because there have beenisolated events where the death toll has exceeded thelong-term trend by several million. On average 275 000people have died each year because of natural hazards.

If famine is removed from the data set, this numberdeclines to 140 200 deaths per year. This reductionsuggests that famine kills 134 800 people per year,roughly 50 per cent of the total number. Although thegreatest number of deaths occurred in the 1930s – afigure that can be attributed to political upheaval andcivil war – there is no significant variation over thetwentieth century. This fact suggests that improvedwarning and prevention of natural hazards hasbalanced population increases resulting in a constantdeath toll. Many world organizations would considerseveral hundred thousand deaths per year due tonatural hazards as unacceptable.

Table 1.3 presents the accumulated number ofdeaths, injuries and homeless for each type of hazardfor the twentieth century. Also presented is the largestevent in terms of death for each category. The greatesthazard during the twentieth century was flooding;however much of this was due to civil unrest. Halfof the 6.9 million death toll occurred in China inthe 1930s where neglect and deliberate sabotageaugmented deaths. Earthquakes and tropical cyclonesaccount for the other significant death tolls of thetwentieth century. Interestingly, during the first threeyears of the twenty-first century 4242 deaths werecaused by cold waves. This is 60 per cent of the totalfor the whole of the twentieth century despite the per-ception of global warming. In contrast, the death tollfrom a heat wave in France in 2003 resulted in 15 000

8 Natural Hazards

100

200

300

400

500

0

Num

ber

of d

isas

ters

rep

orte

d

1900 1920 1940 1960 1980 2000Year

Exponential trend line

Fig. 1.2 Reporting incidence of natural hazards over time 1900–2001 (based on CRED, 2002).

deaths, more than could be attributed to this causeduring the whole of the twentieth century. Obviously,the statistics presented in Table 1.3 underestimateactual deaths, mainly because data were simply notcollected for some hazards until recent times.

Table 1.3 also presents the number of injured anddisplaced people due to natural catastrophes in thetwentieth century. Often hazard statistics concentrateupon death, not taking into account the walkingwounded and the homeless who put a greater, more

lasting burden, upon society. Eighteen times morepeople were made homeless by floods in the twentiethcentury than were killed. This ratio rises to 30 and344 times respectively for tropical cyclones and extra-tropical storms. For example, the January 1998 icestorm that paralyzed Montreal killed twenty-five people.However up to 100 000 people were made homelessfor up to a month afterwards because of the failure ofelectricity supplies as temperatures dipped to –40°C.


10

100

1000

10000

100000

1000000

10000000N

umbe

r of

ann

ual d

eath

s

1900 1920 1940 1960 1980 2000Year

Significant deaths due to drought or famine

Fig. 1.3 Deaths from natural hazards over time 1900–2001 (based on CRED, 2002).

Type of Hazard Deaths Injuries Homeless Largest death toll event and date Death tollFloods 6 851 740 1 033 572 123 009 662 China, July 1931 3 700 000Earthquakes 1 816 119 1 147 676 8 953 296 Tangshan, China, July 1976 242 000Tropical cyclones 1 147 877 906 311 34 272 470 Bangladesh, Nov 1970 300 000Volcano 96 770 11 154 197 790 Martinique, May 1902 30 000Landslides,avalanches,mud flows 60 501 8 071 3 759 329 Soviet Union, 1949 12 000

Extra-tropical storms 36 681 117 925 12 606 891 Northern Europe, Feb 1953 4 000Heat wave 14 732 1 364 0 India, May 1998 2 541Tsunami 10 754 789 – Sanriku Japan, Mar 1933 3 000Cold wave 6 807 1 307 17 340 India, Dec 1982 400Tornado 7 917 27 887 575 511 Bangladesh, Apr 1989 800Fires 2 503 1 658 140 776 USA, Oct 1918 1 000Total 10 052 401 3 257 714 183 533 065

Table 1.3 Number of people killed, injured or displaced due to natural hazards during the twentieth century (based on WHO, 2002).

CHAPTER OUTLINES

While this book has been written assuming little priorknowledge in Earth science at the university level,there are occasions where basic terminology (jargon)has had to be used in explanations. More explicitdefinitions of some of these terms can be found in theglossary at the end of the book. You may have noticedalready that some words in this book have beenitalicized. These terms are defined or explained furtherin the glossary. Note that book titles, names of shipsand botanical names are also italicized, but they do notappear in the glossary.

Individual chapters are arranged around a specificconcept or type of hazard. The mechanisms controllingand predicting this particular hazard’s occurrence areoutlined and some of the more disastrous occurrencesworldwide are summarized. Where appropriate, someof humankind’s responses to, or attempts at mitigating,the hazard are outlined.

The hazards covered in this book are summarized inTable 1.4, with a chapter reference for each hazardshown in the last column. The hazards, assessedsubjectively, are listed in the order that reflects theemphasis given to each in the text. Table 1.4 also gradesthe hazards, on a scale of 1 to 5, as to degree of severity,time span of the event, spatial extent, total death toll,economic consequences, social disruption, long-termimpact, lack of prior warning or suddenness of onset,and number of associated hazards. The overall socio-economic–physiological impact of the hazard is rankedusing these criteria. The most important global hazardis drought, followed by tropical cyclones, regionalfloods, and earthquakes. Three of these four tophazards rank highest for cost (Table 1.2) and humanimpact (Table 1.3). These latter two tables do notevaluate drought because it develops so slowly andinsidiously that it is often ignored when event statisticsare collected. Of the top ten hazards, six are climaticallyinduced. Table 1.4 has been constructed for worldhazards. The ranking differs for individual countriesand latitudes. For example, the ranking of hazards inthe United States in terms of economic loss is tropicalstorms, floods, severe thunderstorms, tornadoes, extra-tropical storms, hail, and wind.

The hazards in the book are organized under twomain parts: climatic hazards and geological hazards.Climatic hazards are introduced in Chapter 2, whichoutlines the mechanisms responsible for climatic vari-

ability. This chapter covers the basic processes of airmovement across the surface of the globe, the conceptof mobile polar highs, air–ocean temperature inter-actions resulting in the Southern, North Atlantic andNorth Pacific oscillations, and the effect of astronomicalcycles (such as sunspots and the 18.6-year lunar tide) onthe timing of climate hazard events.

A description of large-scale storms is then coveredin Chapter 3. This chapter discusses the formation oflarge-scale tropical vortices known as tropical cyclones.Tropical cyclones are the second most importanthazard, generating the greatest range of associatedhazard phenomena. Familiar cyclone disaster eventsare described together with their development, magni-tude and frequency, geological significance, andimpact. The response to cyclone warnings in Australiais then compared to that in the United States andBangladesh. Extra-tropical cyclones are subsequentlydescribed with reference to major storms in thenorthern hemisphere. These storms encompass rain,snow, and freezing rain as major hazards. Storm surgesproduced by the above phenomena are then described.Here, the concepts of probability of occurrence andexceedence are introduced. The chapter concludeswith a description of dust storms as a significant factorin long-term land degradation.

Chapter 4 summarizes smaller hazards generatedby wind. Attention is given to the description of thun-derstorms and their associated hazards. These includetornadoes, described in detail together with some ofthe more significant events. The chapter concludeswith a description of the measures used to averttornado disasters in the United States.

These chapters set the scene for two chapters dealingwith longer lasting climatic disasters, namely droughtand floods. Chapter 5 deals with impact of humanactivity in exacerbating drought and people’s subsequentresponses to this calamitous hazard. Emphasis is placedon pre- and post-colonial influences in the Sahel regionof Africa, followed by a discussion of modern impacts forcountries representative of a range of technologicaldevelopment. The second half of the chapter deals withhuman response by a variety of societies, from thosewho expect drought as a natural part of life, to those whoare surprised by its occurrence and make little effort tominimize its impact. These responses cover Third WorldAfrican countries as well as developed westernizedcountries such as the United States, England, andAustralia. The chapter concludes by describing the way

10 Natural Hazards

the international community alleviates drought. It isonly apt that this section emphasizes the inspirationaland enthusiastic role played by Bob Geldof during theEthiopian drought of the mid-1980s. The reaction to his

initiatives by individuals, regardless of their economic orpolitical allegiance, gives hope that the effects of eventhe worst natural calamities affecting humanity can beminimized.


1 – largest or most significant 5 – smallest or least significantNumber of

Degree of Length of Total areal Total loss Total Social Long term associated Chapter Rank Event severity event extent of life economic loss effect impact Suddenness hazards location

1 Drought 1 1 1 1 1 1 1 4 3 5, 6

2 Tropical 1 2 2 2 2 2 1 5 1 2cyclone

3 Regional 2 2 2 1 1 1 2 4 3 5, 8flood

4 Earthquake 1 5 1 2 1 1 2 3 3 10, 11

5 Volcano 1 4 4 2 2 2 1 3 1 10, 12

6 Extra- 1 3 2 2 2 2 2 5 3 2tropicalstorm

7 Tsunami 2 4 1 2 2 2 3 4 5 11

8 Bushfire 3 3 3 3 3 3 3 2 5 9

9 Expansive 5 1 1 5 4 5 3 1 5 13soils

10 Sea-level 5 1 1 5 3 5 1 5 4 4rise

11 Icebergs 4 1 1 4 4 5 5 2 5 4

12 Dust storm 3 3 2 5 4 5 4 1 5 4

13 Landslides 4 2 2 4 4 4 5 2 5 13

14 Beach 5 2 2 5 4 4 4 2 5 4erosion

15 Debris 2 5 5 3 4 3 5 1 5 13avalanches

16 Creep & 5 1 2 5 4 5 4 2 5 13solifluction

17 Tornado 2 5 3 4 4 4 5 2 5 3

18 Snowstorm 4 3 3 5 4 4 5 2 4 7

19 Ice at shore 5 4 1 5 4 5 4 1 5 4

20 Flash flood 3 5 4 4 4 4 5 1 5 8

21 Thunderstorm 4 5 2 4 4 5 5 2 4 7

22 Lightning 4 5 2 4 4 5 5 1 5 7strike

23 Blizzard 4 3 4 4 4 5 5 1 5 7

24 Ocean waves 4 4 2 4 4 5 5 3 5 4

25 Hail storm 4 5 4 5 3 5 5 1 5 7

26 Freezing rain 4 4 5 5 4 4 5 1 5 7

27 Localized 5 4 3 5 5 5 5 1 5 3strong wind

28 Subsidence 4 3 5 5 4 4 5 3 5 13

29 Mud & debris 4 4 5 4 4 5 5 4 5 13flows

30 Air supported 4 5 5 4 5 5 5 2 5 13flows

31 Rockfalls 5 5 5 5 5 5 5 1 5 13

Table 1.4 Ranking of hazard characteristics and impact.

Chapter 6 examines flash flood and regional floodingevents that are usually severe enough to invoke statesof emergency. In addition, regional flooding accountsfor the largest death toll registered for any type ofhazard. This chapter begins by introducing the climaticprocesses responsible for flash flooding. This isfollowed by examples of flash flooding events in theUnited States and Australia. Regional flooding is dis-cussed in detail for the Mississippi River in the UnitedStates, the Hwang Ho River in China, and Australiawhere floods are a general feature of the country.

Drought-induced bushfires or forest fires are thentreated as a separate entity in Chapter 7. The conditionsfavoring intense bushfires and the causes of such disas-ters are described. Major natural fire disasters, includingin the United States and Australia, are then presented.Two of the main issues discussed in this chapter are thecontinuing debate over prescribed burning as a strategyto mitigate the threat of fires (especially in urban areas),and the regularity with which fire history repeats itself.Both the causes of fires and the responses to themappear to have changed little between the ends of thenineteenth and twentieth centuries.

These climatic chapters are followed by Chapter 8’sdiscussion of aquatic hazards – encompassing a varietyof marine and lacustrine phenomena. Most of thesehazards affect few people, but they have long-termconsequences, especially if global warming becomessignificant over the next 50 years. Because of theirspatial extent and long period of operation, most ofthe phenomena in this chapter can be ranked asmiddle-order hazards. Waves are described first withthe theory for their generation. The worldwide occur-rence of large waves is discussed, based upon ship andsatellite observations. This is followed by a briefdescription of sea-ice phenomena in the ocean and atshore. One of the greatest concerns today is theimpending rise in sea level, supposedly occurringbecause of melting of icecaps or thermal expansion ofoceans within the context of global warming. World-wide sea level data are presented to show that sealevels may not be rising globally as generally believed,but may be influenced in behavior by regional climaticchange. Finally, the chapter discusses the variousenvironmental mechanisms of sandy beach erosion(or accretion) caused mainly by changes in sea level,rainfall, or storminess. This evaluation utilizes anextensive data set of shoreline position and environ-mental variables collected for Stanwell Park Beach,

Australia. Much of the discussion in this chapter drawsupon the author’s own research expertise.

Geological hazards – forming the second part of thisbook – are covered under the following chapterheadings: causes and prediction of earthquakes andvolcanoes; earthquakes and tsunami; volcanoes; andland instability. Chapter 9 presents the worldwide distri-bution of earthquakes and volcanoes and a presentationof the scales for measuring earthquake magnitude. Thisis followed by an examination of causes of earthquakesand volcanoes under the headings of plate boundaries,hot spots, regional faulting, and the presence of reser-voirs or dams. Next, the presence of clustering in theoccurrence of earthquakes and volcanoes is examined,followed by a discussion of the long-term prediction ofvolcanic and seismic activity. The chapter concludeswith a presentation of the geophysical, geochemical, andgeomagnetic techniques for forecasting volcanicand seismic activity over the short term.

Two chapters that are concerned separately withearthquakes and volcanoes then follow this introductorychapter. Both hazards, because of their suddenness andhigh-energy release, have the potential to afflict humanbeings physically, economically, and socially. Both rankas the most severe geological hazards. Chapter 10 firstdescribes types of seismic waves and the global seismicrisk. Earthquake disasters and the seismic risk forAlaska, California and Japan are then described. Theassociated phenomenon of liquefaction or thixotropy,and its importance in earthquake damage, is sub-sequently presented. One of the major phenomenagenerated by earthquakes is tsunami. This phenomenonis described in detail, together with a presentationof major, worldwide tsunami disasters. The prediction oftsunami in the Pacific region is then discussed. Chapter11, on volcanoes, emphasizes types of volcanoes andassociated hazardous phenomena such as lava flows,tephra clouds, pyroclastic flows and base surges, gasesand acid rains, lahars (mud flows), and glacier bursts.The major disasters of Santorini (~1470 BC), Vesuvius(79 AD), Krakatau (1883), Mt Pelée (1902), and Mt StHelens (1980) are then described in detail.

The section on geological hazards concludes, inChapter 12, with a comprehensive treatment of landinstability. This chapter opens with a description ofbasic soil mechanics including the concepts of stressand strain, friction and cohesion, shear strength ofsoils, pore-water pressure, and rigid, elastic, and plasticsolids. Land instability is then classified and described

12 Natural Hazards

under the headings of expansive soils, creepand solifluction, mud and debris flows, landslides andslumps, rockfalls, debris avalanches, and air-supportedflows. Many of these hazards rank in severity andimpact as middle-order events. For example, expansivesoils, while not causing loss of life, rank as one of thecostliest long-term hazards because of their ubiqui-tousness. For each of these categories, the type of landinstability is presented in detail, together with adescription of major disasters. The chapter concludeswith a cursory presentation on the natural causes ofland subsidence.

Because the emphasis of the previous chapters hasbeen on the physical mechanisms of hazards, the bookcloses in Chapter 13 with a discussion of the socialimpact of many of these phenomena. Here is empha-sized the way hazards are viewed by different societiesand cultures. This section is followed by detaileddescriptions of responses before, during, and after theevent. As much as possible, the characterization ofresponses under each of these headings is clarifiedusing anecdotes. It is shown that individuals or familieshave different ways of coping with, or reacting to,disaster. The chapter concludes by emphasizing thepsychological ramifications of a disaster on the victims,the rescuers, and society.

REFERENCES AND FURTHERREADING

Balling, R.C. and Cerveny, R.S. 2003. Compilation and discussion oftrends in severe storms in the United States: Popular perceptionv. climate reality. Natural Hazards 29: 103–112.

Blong, R.J. 1982. The Time of Darkness, Local Legends andVolcanic Reality in Papua New Guinea. Australian NationalUniversity Press, Canberra.

Bryant, E.A. 2001. Tsunami: The Underrated Hazard. CambridgeUniversity Press, Cambridge.

Changnon, S.A. and Hewings, G.J.D. 2001. Losses from weatherextremes in the United States. Natural Hazards Review 2:113–123.

CRED 2002. EM-DAT: The OFDA/CRED International DisastersData Base. Center for Research on the Epidemiology of Disas-ters, Université Catholique de Louvain, Brussels, <http://www.em-dat.net/>

Day, M.S. 1984. The Many Meanings of Myth. Lanham, NY.High Plains Regional Climate Center 2003. Annual average number

of strong–violent (F2–F5) tornadoes 1950–1995. University ofNebraska, Lincoln, http://www.hprcc.unl.edu/nebraska/USTORNADOMAPS.html

Holmes, A. 1965. Principles of Physical Geology. Nelson, London.Huggett, R. 1997. Catastrophism: Asteroids, Comets and Other

Dynamic Events in Earth History. Verso, London.Intergovernmental Oceanographic Commission 2003. On-line

Pacific…, Atlantic…, and Mediterranean Tsunami Catalog.Tsunami Laboratory, Institute of Computational Mathematicsand Mathematical Geophysics, Siberian Division, RussianAcademy of Sciences, Novosibirsk, Russia, <http://omzg.sscc.ru/tsulab/>

Middleton, N. 1995. The Global Casino. Edward Arnold, London.Milne, A. 1986. Floodshock: the Drowning of Planet Earth. Sutton,

Gloucester.Myles, D. 1985. The Great Waves. McGraw-Hill, NY.Sidle, R.C., Taylor, D., Lu, X.X., Adger, W.N., Lowe, D.J., de Lange,

W.P., Newnham, R.M. and Dodson, J.R. 2003. Interaction ofnatural hazards and society in Austral-Asia: evidence in past andrecent records. Quaternary International 118–119: 181–203.

Susman, P., O’Keefe, P. and Wisner, B. 1983. Global disasters, aradical interpretation. In Hewitt, K. (ed.) Interpretations ofCalamity. Allen and Unwin, Sydney, pp. 263–283.

Watts, M. 1983. On the poverty of theory: natural hazards researchin context. In Hewitt, K. (ed.) Interpretations of Calamity. Allenand Unwin, Sydney, pp. 231–261.


CLIMATICHAZARDS

P A R T 1

INTRODUCTION

Climatic hazards originate with the processes thatmove air across the Earth’s surface due to differentialheating and cooling. Surprisingly, examination of theseprocesses has focused upon heating at the tropics anddownplayed the role of cold air masses moving out ofpolar regions due to deficits in the radiation balance inthese latter regions. Fluctuations – in the intensity ofpulses of cold air moving out of polar regions orof heating at the equator – and the location of theinteraction between these cold and warm air masses,are crucial factors in determining the magnitude, fre-quency, and location of mid-latitude storm systems.While most of these factors are dictated by internalfactors in the Earth–atmosphere system, modulationby 11-year geomagnetic cycles linked to solar activityand by the 18.6 year MN lunar tide also occurs. Thischapter examines these processes and mechanisms.The responses in terms of centers of storm activity willbe examined in the following chapter.

MODELS OF ATMOSPHERICCIRCULATION AND CHANGE

(Bryson & Murray, 1977; Lamb, 1982)

How air moves

Barometric pressure represents the weight of air abovea location on the Earth’s surface. When the weight of airover an area is greater than over adjacent areas, itis termed ‘high pressure’. When the weight of air islower, it is termed ‘low pressure’. Points of equalpressure across the Earth’s surface can be contouredusing isobars, and on weather maps contouring isgenerally performed at intervals of 4 hectopascals (hPa)or millibars (mb). Mean pressure for the Earth is1013.6 hPa. Wind is generated by air moving fromhigh to low pressure simply because a pressure gradientexists owing to the difference in air density (Figure 2.1):the stronger the pressure gradient, the stronger thewind. This is represented graphically on weather mapsby relatively closely spaced isobars. In reality, wind doesnot flow down pressure gradients but blows almostparallel to isobars at the surface of the Earth because ofCoriolis force (Figure 2.1). This force exists becauseof the rotation of the Earth and is thus illusionary. Forexample, a person standing perfectly still at the polewould appear to an observer viewing the Earth fromthe Moon to turn around in a complete circle every

Mechanisms ofClimate Variability

C H A P T E R 2

18 Climatic Hazards

24 hours as the Earth rotates. However, a personstanding perfectly still at the equator would alwaysappear to be facing the same direction and not turningaround. Because of Coriolis force, wind tends to bedeflected to the left in the southern hemisphere and tothe right in the northern hemisphere. Coriolis force canbe expressed mathematically by the following equation:

Coriolis force = 2 � sin�� 2.1where � = rate of spin of the Earth

sin � = latitude� = wind speed

Clearly, the stronger the pressure gradient, thestronger the wind and the more wind will tend to bedeflected. This deflection forms a vortex. Also, thestronger the wind, the smaller and more intense isthe resulting vortex. Very strong vortices are known ashurricanes, typhoons, cyclones, and tornadoes. Coriolisforce varies across the surface of the globe because oflatitude and wind velocity. Coriolis force is zero at theequator (sin 0° = 0), and maximum at the pole(sin 90° = 1). The equator is a barrier to inter-hemispheric movement of storms because vortices

rotate in opposite directions in each hemisphere. Forexample, tropical cyclones cannot cross the equatorbecause they rotate clockwise in the southern hemi-sphere and anti-clockwise in the northern hemisphere.

Palmén–Newton model of globalcirculation

Upward movement of air is a prerequisite for vortexdevelopment, with about 10 per cent of all airmovement taking place vertically. As air rises, it spiralsbecause of Coriolis force, and draws in adjacent air atthe surface. Rising air is unstable if it is warmer thanadjacent air, and instability is favored if latent heat canbe released through condensation. The faster air rises,the greater the velocity of surface winds spiraling intothe center of the vortex. Of course, in the oppositemanner, descending air spirals outward at the Earth’ssurface. These concepts can be combined to account forair movement in the troposphere. The Palmén–Newtongeneral air circulation model is one of the morethorough models in this regard (Figure 2.2). Intenseheating by the sun, at the equator, causes air to rise andspread out poleward in the upper troposphere. As thisair moves towards the poles, it cools through long wave

1012 hPa

1008

1004

1000

1012 hPa

1008

1004

1000

Weak

Strong

Potential wind

Actual wind

Wind strengthWind direction

Low

Low

High

High

wind deflected to the rightNorthern hemisphere

wind deflected to the leftSouthern hemisphere

Fig. 2.1 Wind movements relative to isobars, with and without Coriolis force (i.e. potential and actual).

emission and begins to descend back to the Earth’ssurface around 20–30° north and south of the solarequator, forming high pressure at the ground. Uponreaching the Earth’s surface, this air either movespoleward or returns to the equator to form a closed cir-culation cell, termed a Hadley cell. Because of Coriolisforce, equatorial moving air forms two belts of easterlytrade winds astride the equator. Air tends to convergetowards the solar equator, and the uplift zone here istermed the intertropical convergence. Over the westernsides of oceans, the tropical easterlies pile up warmwater, resulting in convection of air that forms lowpressure and intense instability. In these regions, intensevortices known as tropical cyclones develop preferen-tially. These will be described in Chapter 3.

At the poles, air cools and spreads along the Earth’ssurface towards the equator. Where cold polar air meetsrelatively warmer, subtropical air at mid-latitudes, a coldpolar front develops with strong uplift and instability.Tornadoes and strong westerlies can be generated nearthe polar front over land, while intense extra-tropicalstorms develop near the polar front, especially overwater bodies. Tornadoes will be described in Chapter 4,while extra-tropical storms are described in the next

chapter. A belt of strong wind and storms – dominatedby westerlies poleward of the polar front – forms around40° latitude in each hemisphere. These winds, known asthe roaring forties, are especially prominent in thesouthern hemisphere where winds blow unobstructedby land or, more importantly, by significant mountainranges. Maps of global winds support these aspects ofthe Palmén–Newton model (Figure 2.3).

The Palmén–Newton model has two limitations.First, the location of pressure cells within the modelis based upon averages over time. Second, becausethe model averages conditions, it tends to be static,whereas the Earth’s atmosphere is very dynamic.Additionally, the concept of Hadley circulation isoverly simplistic. In fact, rising air in the tropics is notuniform, but confined almost exclusively to narrowupdrafts within thunderstorms. At higher latitudes,large-scale circulation is distorted by relatively smalleddies. Other factors also control winds. For example,over the Greenland or Antarctic icecaps, enhancedradiative cooling forms large pools of cold air, whichcan accelerate downslope under gravity because of thelow frictional coefficient of ice. Alternative models thatovercome some of these limitations will be presented.

19Mechanisms of Climate Variability

air flow

heat transport

Polar front

Polar jet

Subtropicaljet

Intertropicalconvergence

Tropopause

HADLEYCELL

Easterlytrades

Roaringforties

Polareasterlies

90˚ 60˚ 30˚ 0˚

POLAR HIGH

SUB TROPICALHIGH

EXTRATROPICALLOW MONSOONAL

LOW

Fig. 2.2 Palmén–Newton schematic model of general air circulation between the pole and the equator.

20 Climatic Hazards

Ice Ice

Ice Ice

January

2 4 6 8 10 12 14 16

Meters s-1

A

Ice Ice

Ice Ice

July

2 4 6 8 10 12 14 16

Meters s-1

B

Fig. 2.3 Location of mid-latitude westerly and easterly trade wind belts based on TOPEX/POSEIDON satellite observations A) January 1995,B) July 1995 (Jet Propulsion Laboratory, 1995a, b).

Changes in jet stream paths

Strong winds exist in the upper atmosphere adjacentto the tropopause boundary (Figure 2.2). The moresignificant of these is the polar jet stream on the equa-torial side of the polar front. This jet stream consists ofa zone of strong winds no more than 1 km deep and100 km wide, flowing downwind over a distance of1000 km or more. Wind speeds can reach in excessof 250 km hr–1. The polar jet stream is most prominentand continuous in the northern hemisphere. Here, itforms over the Tibetan Plateau, where its position islinked with the seasonal onset and demise of theIndian monsoon. The jet loops northward over Japanand is deflected north by the Rocky Mountains. It thenswings south across the Great Plains of the UnitedStates, north-eastward (parallel to the AppalachianMountains), and exits North America off Newfound-land, dissipating over Iceland. Both the TibetanPlateau and the Rocky Mountains produce a resonanceeffect in wave patterns in the northern hemisphere,locking high pressure cells over Siberia and NorthAmerica with an intervening low pressure cell over thenorth Pacific Ocean. Barometric pressure, measuredalong the 60°N parallel of latitude through these cells,reveals a quasi-stable planetary or Rossby wave.

In the northern hemisphere, the jet stream tends toform three to four Rossby waves extending through5–10° of latitude around the globe (Figure 2.4). Rossbywaves move a few hundred kilometres an hour fasterthan the Earth rotates and thus appear to propagatefrom west to east. Any disturbance in a Rossby wavethus promulgates downwind, such that a change inweather in North America appears over Europeseveral days afterwards. This aspect gives coherenceto extreme events in weather across the northernhemisphere. Because there is a time lag in changes toRossby waves downwind from North America, fore-casters can predict extreme events over Europe days inadvance. In some winters, the waves undergo amplifi-cation and loop further north and south than normal.Many researchers believe that changes in this loopingare responsible, not only for short-term drought(or rainfall), but also for semi-permanent climaticchange in a region extending from China to Europe.Sometimes the looping is so severe that winds in the jetstream simply take the path of least resistance, andproceed zonally (parallel to latitude), cutting off theloop. In this case, high and low pressure cells can be

left stranded in location for days or weeks, unable tobe shifted east by the prevailing westerly airflow. High-pressure cells are particularly vulnerable to thisprocess and form blocking highs. Blocking highsdeflect frontal lows to higher or lower latitudes thannormally expected, producing extremes in weather.Blocking is more characteristic of winter, especially inthe northern hemisphere. It also tends to occur overabnormally warm seas, on the western sides of oceans.

Expansion of the westerlies because of enhancedlooping of Rossby waves can be linked to recent failure


A

B

Fig. 2.4 Path of jet stream in the northern hemisphere showing likely Rossby wave patterns A) for summer, B) for extreme winter (Bryson & Murray, ©1977 with the permission of The University of Wisconsin Press).

of monsoon rains and resulting drought in the Sahel inthe early 1970s. In the Sahara region, rain normallyfalls as the intertropical convergence moves seasonallynorthward. If westerlies expand, then such movementof the intertropical convergence is impeded anddrought conditions prevail. Drought in Britain can alsobe linked to displacement northward of the jet streamand polar-front lows – shifting into the Arctic insteadof crossing Scotland and Norway. This displacementleaves Britain under the influence of an extension ofthe Azores high-pressure system that is stable andrainfall-deficient. Blocking of this high leads to droughtfor several months. Blocking highs off the Californiancoast were also responsible for drought in the mid-United States in 1997, while highs over easternAustralia have led to some of the worst droughts in thatcontinent’s recorded history. The social response tothese types of droughts will be examined in Chapter 5.

Mobile polar highs (Leroux, 1993, 1998;Bryant, 1997)

The monsoonal circulation described above does notfit well within the Palmén–Newton general circulationmodel. In fact, the positioning of Hadley cells, and thesemi-permanency of features such as the Icelandic orAleutian Lows, are statistical artefacts. There is not anIcelandic Low, locked into position over Iceland. Norare there consistent trades or polar easterlies. Aircirculation across the surface of the Earth is dynamic,as is the formation and demise of pressure cells. Inreality, the Icelandic Low may exist as an intense cell oflow pressure for several days, and then move eastwardtowards Europe within the westerly air stream.Climate change in the Palmén–Newton model impliesthe movement, or change in magnitude, of thesecenters of activity. For example, weakening of the Ice-landic Low conveys the view that winter circulation isless severe, while expansion of Hadley cells towardshigher latitudes suggests that droughts shoulddominate mid-latitudes.

There are other problems with terminology in thePalmén–Newton model; these have an historical basis.The depression, extra-tropical cyclone, or polar lowwas initially explained as a thermal phenomenon, andthen linked to frontal uplift along the polar front, toupper air disturbances, and recently to planetary wavesor Rossby waves. The jet stream is supposedly tiedto the polar front, and related to cyclogenesis at mid-latitudes. However, the connection between the jet

stream and the polar front is approximate, and a clearrelationship has not been established between the jetstream in the upper troposphere and the formation oflow pressure at the Earth’s surface. Indeed, no onetheory adequately explains the initial formation of lowsat mid-latitudes. The Palmén–Newton general circula-tion model is a good teaching model, but it is not anideal model for explaining the causes of climatechange.

Conceptually, the Palmén–Newton model in itssimplest form has two areas of forcing: upward airmovement at the equator from heating of the Earth’ssurface, and sinking of air at the poles because ofintense cooling caused by long wave emission. At theequator, air moves from higher latitudes to replacethe uplifted air, while at the poles subsiding air movestowards the equator as a cold dense mass huggingthe Earth’s surface. The lateral air movement at theequator is slow and weak, while that at the poles isstrong and rapid. Polar surging can reach within 10° ofthe equator, such that some of the lifting of air in thisregion can be explained by the magnitude and locationof polar outbursts. Hence, the dominant influence onglobal air circulation lies with outbursts of cold airwithin polar high-pressure cells. Each of these out-bursts forms an event termed a mobile polar high.There is no separate belt of high pressure or Hadleycell in the subtropics. These cells, as they appear onsynoptic maps, are simply statistical averages over timeof the preferred pathways of polar air movementtowards the equator.

Mobile polar highs developing in polar regions areinitially maintained in position by surface cooling, airsubsidence and advection of warm air at higher alti-tudes. When enough cold air accumulates, it suddenlymoves away from the poles, forming a 1500 m thicklens of cold air. In both hemispheres, polar high out-breaks tend to move from west to east, thus conservingvorticity. Additionally, in the southern hemisphere,pathways and the rate of movement are aided by theformation of katabatic winds off the Antarctic icecap(Figure 2.5). In the northern hemisphere, outburstpathways are controlled by topography with mobilepolar highs tending to occur over the Hudson Baylowland, Scandinavia and the Bering Sea (Figure 2.5).The distribution of oceans and continents, with theirattendant mountains, explains why the mean trajectoriesfollowed by these highs are always the same. Forinstance, over Australia, a polar outbreak always

22 Climatic Hazards

approaches the continent from the southwest, and thenloops across the continent towards the equator, anddrifts in the Hadley belt out into the Tasman Sea.Highs rarely move directly northward across the conti-nent, or sweep up from the Tasman Sea. In NorthAmerica, mobile polar highs regularly surge southwardacross the Great Plains for the same reason.

Because mobile polar highs consist of dense air, theydeflect less dense, warm air upward and to the side.The deflection is greatest in the direction they aremoving. Hence, the polar high develops an extensivebulbous, high-pressure vortex, surrounded downdriftby a cyclonic branch or low-pressure cell (Figure 2.6).Typically, the high-pressure cell is bounded by anarching, polar cold front with a low pressure cellattached to the leading edge. However, in the northernhemisphere, individual outbreaks tend to overlap sothat the low-pressure vortex becomes containedbetween two highs. This forms the classic V-shaped,wedged frontal system associated with extratropicallows or depressions (Figure 2.6). Thus lows, and upperwesterly jet streams, are a product of the displacementand divergence of a mobile polar high. The intensity ofthe low-pressure cell becomes dependent upon thestrength of the polar high, and upon its ability todisplace the surrounding air. Strong polar highs

produce deep lows; weak highs generate weak lows. Ina conceptual sense, a deeper Icelandic or Aleutian Lowmust be associated with a stronger mobile polar high.If a mobile polar high is particularly cold, it can causethe air above to cool and settle. This creates lowpressure above the center of the high-pressure cell,which can develop into a trough and then a cell asupper air flows inwards. If a mobile polar high has lostits momentum and stalled forming a ‘blocking high’,the low-pressure cell can intensify, propagate to theground, and generate a surface storm. This processoccurs most often on the eastern sides of continentsadjacent to mountain ranges such as the Appalachiansin the United States and the Great Dividing Range inAustralia. These storms will be discussed further in thenext chapter.

Mobile polar highs tend to lose their momentumand stack up or agglutinate at particular locations overthe oceans. A blocking high is simply a stagnant mobilepolar high. Air pressure averaged over time thusproduces the illusion of two stable, tropical high-pressure belts – known as Hadley cells – on each sideof the equator. Mobile polar highs can propagate intothe tropics, especially in winter. Here, their arrivaltends to intensify the easterly trade winds. Moreimportant, yet little realized, is the fact that strong


Main outbursts

MPH pathways

MPH boundaries mid-latitudes

Trade Winds

MPH boundaries in tropics

Relief affecting MPH pathways

Fig. 2.5 Pathways for mobile polar highs (MPH) and resulting trade wind circulation in the tropics (based on Leroux, 1993).

polar highs produce stronger monsoons, although in amore restricted tropical belt.

The Southern Oscillation

Introduction

The Earth’s general atmospheric circulation in thetropics and subtropics, as shown above, can be simplydescribed. However, the actual scene is slightly morecomplex. The intensity of mobile polar highs withina hemisphere varies annually with the apparentmovement of the sun north and south of the equator.There is also a subtle, but important, shift in the inten-sity of heating near the equator. In the northernhemisphere summer, heating shifts from equatorialregions to the Indian mainland with the onset of theIndian monsoon. Air is sucked into the Indian subcon-tinent from adjacent oceans and landmasses, to returnvia upper air movement, to either southern Africa orthe central Pacific. In the northern hemisphere winter,this intense heating area shifts to the Indonesian–northern Australian ‘maritime’ continent (Figure 2.7A),

with air then moving in the upper troposphere, to theeast Pacific. Convection is so intense that updraftspenetrate the tropopause into the stratosphere, mainlythrough supercell thunderstorms. These convectivecells are labeled stratospheric fountains.

Throughout the year, mobile polar highs tend tostack up over the equatorial ocean west of SouthAmerica. The center of average high pressure hereshifts seasonally less than 5° in latitude. The highs arelocked into position by positive feedback. Cold water inthe eastern Pacific Ocean creates high pressure thatinduces easterly airflow; this process causes upwellingof cold water along the coast, which cools the air. Onthe western Pacific side, easterlies pile up warm water,thus enhancing convective instability, causing air torise, and perpetuating low pressure. Air is thus contin-ually flowing across the Pacific, as an easterly tradewind from high pressure to low pressure. This circula-tion is zonal in contrast to the meridional airflowinherent within the Palmén–Newton general circula-tion model. The persistent easterlies blow warmsurface water across the Pacific Ocean, piling it up in

24 Climatic Hazards

x

x

N

N

x

x

x

Northern hemisphere Southern hemisphere

Low

High

Low

High

Center of high or low pressure

Low pressure corridor

Surface wind flow

Path of high or low pressure

Warm front

Cold front

Fig. 2.6 Schematic view of the dynamic structure of a mobile polar high and its associated mid-latitude low (based on Leroux, 1993).

the west Pacific warm pool in the Philippine Sea. Thesupra-elevation of sea level in the west Pacific Oceanis approximately 1 cm for each 1°C difference intemperature between the west Pacific and theSouth American coast. Normally this supra-elevationamounts to 13–20 cm.

This atmospheric system is very stable, existingbeyond the annual climatic cycle. The easterly tradewind flow is termed Walker circulation, after theIndian meteorologist, Gilbert Walker, who describedthe phenomenon in detail in the 1920s–1930s. Forsome inexplicable reason, this quasi-stationary heatingprocess weakens in intensity, or breaks down com-pletely, every 3–5 years – a frequency very suggestiveof a chaotic system. High pressure can become estab-lished over the Indonesian–Australian area, while lowpressure develops over warm water off the SouthAmerican coast. In the tropics, the easterly winds abateand are replaced by westerly winds. The rainfall beltshifts to the central Pacific and drought replaces

normal or heavy rainfall in Australia. Such conditionspersisted in the Great World Drought of 1982–1983,which affected most of Australia, Indonesia, India, andSouth Africa.

Because it tends to fluctuate, this phenomenon iscalled the Southern Oscillation (SO). Simply subtract-ing the barometric pressure for Darwin from that ofTahiti produces an index of the intensity and fluctua-tions of the SO. This value is then normalized to amean of 0.0 hPa and a standard deviation of 10. Thisnormalized index, from 1851 to the present, is pre-sented in Figure 2.8.

Warm water usually appears (albeit in reducedamounts) along the Peruvian coast around Christmaseach year. This annual warming is termed the El Niño,which is Spanish for ‘Christ Child’. However, whenWalker circulation collapses, this annual warmingbecomes exaggerated, with sea surface temperaturesincreasing 4–6°C above normal and remaining thatway for several months. This localized, above average,


A

B

C

D

Normal January

September pre-ENSO Post ENSO

December ENSO event

LowLow

Low

Low

Low

Low

Low High

High

HighHigh

Wet or underflood

Dry or indrought

Warm surfacewater

Very warmSurface water

Relative pressure

Jet stream

Surface winds

Tropical cyclone

Fig. 2.7 Idealized representation of evolution and effects of an El Niño–Southern Oscillation event.

warming is termed an El Niño event. In the early1960s, Jacob Bjerknes recognized that warmer waters,during some El Niño events, were linked to theSouthern Oscillation and the movement of warm watereastward from the west Pacific Ocean. These more

widely occurring events are termed El Niño–SouthernOscillation, or for brevity, ENSO events. Greateramounts of warm water – along what is normally anarid coastline – lead to abnormally heavy rainfall,causing floods in the arid Andes rainfall region and

26 Climatic Hazards

SO Index

-50.0

-40.0

-30.0

-20.0

-10.00.0

10.0

20.0

30.0

40.0

1850 1870 1890 1910 1930 1950 1970 1990 2010

NAO Index

-30.0

0.0

30.0

1850 1870 1890 1910 1930 1950 1970 1990 2010

NPO Index

-30

0

30

1850 1870 1890 1910 1930 1950 1970 1990 2010

A

B

C

Fig. 2.8 Indices of atmospheric pressure oscillations over time A) the Southern Oscillation, B) the North Atlantic Oscillation (from Hurrell, 2002a), and C) the North Pacific Oscillation (from Hurrell, 2002b). The indices have been normalized to a mean of 0.0 hPa and a standard deviation of 10. See relevant sections in the text for further explanation.

drought in southern Peru, Bolivia and north-easternBrazil. For example, in 1982–1983, in the coastaldesert region of Ecuador and Peru, flash floods rippedout roads, bridges and oil pipelines. Towns were buriedin mud, and shallow lakes appeared in the deserts.Irrigation-based agriculture was devastated, first by theflooding, and then by swarms of insects that prolifer-ated under the wet conditions. The warm water alsoforces fish feeding on plankton to migrate to greaterdepths, and causes vast fish kills. Guano-producingbirds feeding on anchovies die in their thousands,causing the collapse of the anchovy and fertilizerindustries in Peru.

These climatic changes worldwide can be responsiblefor the increase of such isolated phenomena assnakebites in Montana, funnel-web spider bites inSydney, and schizophrenia in Brazil (the latter caused bya lack of UV light due to increased cloudiness). Moresignificantly, these changes can generate extremeeconomic and social repercussions, including damageto national economies, the fall of governments, andthe deaths of thousands through starvation, storms andfloods.

El Niño–Southern Oscillation (ENSO) events(Philander, 1990; Glantz et al., 1991; Allan et al., 1996;Glantz, 1996; Bryant, 1997; Couper-Johnston, 2000)

Zonal, ocean–atmosphere feedback systems in thesouthern hemisphere are powered by three tropicalheat-sources located over Africa, South America andthe Indonesian–Australian maritime continent. Thelatter site, by far the strongest of the three systems,initiates Walker circulation. While it is relatively stable,Walker circulation is the least sedentary, because it isnot anchored directly to a landmass. It is highly migra-tory during transition periods of the Indian monsoon,in either March–April, or August–September. Six pre-cursors to the failure of Walker circulation stand out.First, sea-ice in the Antarctic and snow cover in centralAsia tend to be more extensive beforehand. Second,the strength of the Indian monsoon relies heavily uponformation of an upper air, thermally driven anticyclone(between 100 and 300 hPa) over the Tibetan Plateauthat leads to the formation of an easterly upper jet streamover southern India. Zonal easterly winds at 250 hPa overthe Indian monsoon area also have been found toweaken up to two months before the onset of an ENSOevent. Third, the Southern Oscillation appears to be acombination of the 2.2 year Quasi-Biennial Oscillation

and a longer period cycle centered on a periodicity offive years. Fourth, changes in the behavior of southernhemisphere, mobile polar highs are linked to ENSOevents. Strong Walker circulation is related to strongwesterlies between 35° and 55°S. These westerliesappear to lock, over the Indonesian–Australian‘maritime’ landmass, with heat lows developing in theaustral spring. If the normally strong latitudinal flow inthe mobile polar highs is replaced by longitudinallyskewed circulation in the Australian region, then lowpressure can be forced into the Pacific during the tran-sition from northern to southern hemisphere summer.Increased southerly and south-westerly winds east ofAustralia preceded the 1982–1983 ENSO event, andflowed towards a South Pacific Convergence Zone thatwas shifted further eastwards. While highs were asintense as ever, they had a stronger meridional compo-nent, were displaced further south than normal, andwere stalled over eastern Australia. Fifth, ENSO eventsare more likely to occur one year after the south polarvortex is skewed towards the Australian–New Zealandsector. A significant correlation has been found amongstthis eccentricity, the Southern Oscillation, and thecontemporaneous occurrence of rainfall over parts ofsouthern hemisphere continents. Sixth, excess salinitynorth and south of the equator is correlated withwarmer waters, presaging an ENSO event about twelvemonths later. The sinking of this cold salty water aroundthe equator draws in warmer, less saline, surface watertriggering the ENSO event. Finally, the causes may beinterlinked. For example, more snow in Asia the winterbefore Walker circulation collapses may weaken thesummer jet stream, leading to failure of the Indianmonsoon. Alternatively, more sea-ice in the Antarcticmay distort the shape and paths of mobile polar highsover the Australian continent, or displace the southpolar vortex.

Whatever the cause, the movement eastward of lowpressure, beyond the Australian continent, leads towesterly airflow at the western edge of the PacificOcean. Because there is no easterly wind holding supra-elevated water against the western boundary of thePacific, warm water begins to move eastward alongthe equator (Figure 2.7b). Normally, the thermoclineseparating warm surface water from cooler water belowis thicker (200 m) in the west Pacific than in the east(100 m). As warm water shifts eastward, the thermoclinerises in the west Pacific. One of the first indications of anENSO event is the appearance of colder water north of


Darwin, Australia. In the east Pacific, a layer of warmwater extending 50–100 m below the ocean surfaceswamps cold water that is normally at the surface andmaintaining the high-pressure cell off the SouthAmerican coast (Figure 2.7C). As a result the thermo-cline deepens. As low pressure replaces high pressureover this warm water in the east Pacific, the easterliesfail altogether because there is no pressure differenceacross the Pacific to maintain them. Westerly windsmay even begin to flow at the equator. So strong isthe change that the Earth’s rotation speeds up by0.5–0.7 milliseconds as the braking effect of the easterlytrades diminishes.

Cyclone development follows the location of thepool of warm water as it moves eastward towards Tahitiand Tonga. The 1982–1983 event caused widespreaddestruction in these areas, with Tonga receiving fourtropical cyclones, each as severe as any single event inthe twentieth century. The Tuamotu Archipelago, eastof Tahiti, was devastated by five cyclones betweenJanuary and April 1983. Not since 1906 had a cycloneoccurred this far east in the tropics. In unusual cir-cumstances, paired tropical cyclones may develop oneach side of the equator in November. Because windsin tropical cyclones in each hemisphere rotate inopposite directions, coupled cyclones can combine(like an egg beater) to generate, along the equator,strong westerly wind that may accelerate watermovement eastward.

At the peak of an ENSO event, a Kelvin wave istrapped along the east Pacific coast. Warm water andelevated sea levels spread north and south, reaching asfar north as Canada. In the 1982–1983 event, tempera-tures were 10°C above normal and sea levels rose by60 centimetres along the South America coastline, andby 25–30 cm along the United States west coast. Theintense release of heat by moist air over the centralPacific Ocean causes the westerly jet stream in theupper atmosphere, and the position of the wintertimeAleutian Low, to shift towards the equator (Figure 2.7D).Westerly surface winds and the jet stream, rather thanbeing deflected north around the Rocky Mountains,cross the continent at mid-latitudes. This can lead todramatic changes in climate. For example, in the1982–1983 ENSO event, coastal storms wreaked havocon the luxurious homes built along the Malibu coastlineof California while associated rainfall caused widespreadflooding and landslides. The storms were made all themore destructive because sea levels were elevated by

25–30 cm along the entire United States west coast,from California to Washington State. Heavy snowfallsfell in the southern Rocky Mountains, record-breakingmild temperatures occurred along the American eastcoast and heavy rainfall fell in the southern UnitedStates.

Finally, with the spread of warm water in thenorthern Pacific, easterly air circulation begins to re-establish itself in the tropics. The depression of thethermocline in the east Pacific during an ENSO eventcauses a large wave, also called a Rossby wave, to prop-agate westward along the thermocline boundary. Thiswave is reflected off the western boundary of thePacific and, as it returns across the Pacific, it slowlyraises the thermocline to its pre-ENSO position.However, the return to Walker circulation can beanything but normal. The pool of warm water nowmoving back to the west Pacific drags with it excep-tional atmospheric instability, and a sudden return torainfall that breaks droughts in the western Pacific. In1983, drought broke first in New Zealand in February,one month later in eastern Australia with rainfalls of200–400 mm in one week, and then – over the next twoto three months – in India and southern Africa.

The Southern Oscillation causes extreme, short-term climate change over 60 per cent of the globe,mainly across the southern hemisphere and certainlyin North America. However, some events influenceEurope. For example, the El Niño of 1997–1998caused widespread flooding in central Europe con-comitantly with flooding in central Africa, South-EastAsia, and Peru. The climatic effects of an ENSO eventare widespread, and its influence on both extremeclimate hazards and short-term climate change issignificant. Of South American countries, southernPeru, western Bolivia, Venezuela, and north-easternBrazil are affected most by ENSO events. The1997–1998 ENSO event destroyed $US1 billion inroads and bridges in Peru alone. The widest impactoccurs over southern Africa, India, Indonesia, andAustralia. For instance, in Indonesia, over 93 per centof monsoon droughts are associated with ENSOevents, and 78 per cent of ENSO events coincide withfailure of the monsoon. In Australia, 68 per cent ofstrong or moderate ENSO events produce majordroughts in the east of the continent. ENSO eventshave recently been shown to contribute to drought inthe African Sahel region, especially in Ethiopia andSudan.

28 Climatic Hazards

Once an ENSO event is triggered, the cycle ofclimatic change operates over a minimum of two years.In historical records, the longest ENSO event lastedfour years: from early 1911 to mid-1915. However,recent events have dispelled the belief that ENSOevents are biennial phenomena. Following the1982–1983 ENSO event, the warm water that travelednorthward along the western coast of North Americaslowly crossed the north Pacific, deflecting theKuroshio Current a decade later. As a result, northernPacific sea surface temperatures increased abnormally,affecting general circulation across the North Americancontinent. It is plausible that the drought on the GreatPlains of the United States in the summer of 1988, andeven the flooding of the Mississippi River Basin in thesummer of 1993 (attributable by most to the 1990–1995ENSO event) were prolonged North American climaticresponses to the 1982–1983 El Niño. Warm water leftover from the 1982–1983 ENSO event persisted in thenorth Pacific until the year 2000. The oceanographiceffects of a major El Niño thus can have large decadalpersistence outside the tropics.

The 1990–1995 ENSO event was even moreanomalous. This event appeared to wane twice, butcontinued with warm central Pacific waters for fiveconsecutive years, finally terminating in July 1995 – anunprecedented time span. Probability analysis indi-cates that an ENSO event of five years’ duration shouldonly recur once every 1500–3000 years. The changes inclimate globally over these five years have been asdramatic as any observed in historical records. Two ofthe worst cyclones ever recorded in the United States,Hurricanes Andrew in Florida and Iniki in Hawaii,both in August 1992, occurred during this event.Hurricane Andrew was unusual because Atlantic hurri-canes should be suppressed during ENSO events. TheMississippi River system recorded its greatest floodever in 1993 (and then again in 1995), surpassing theflood of 1973. Record floods devastated westernEurope in 1994–1995. Eastern Australia and Indonesiasuffered prolonged droughts that became the longeston record. Cold temperatures afflicted eastern NorthAmerica in the winter of 1993–1994, together withrecord snowfall, while the western half of the continentregistered its highest winter temperatures ever. Recordhigh temperatures and drought also occurred in Japan,Pakistan, and Europe in the summer of 1994.

Not all of the climatic extremes of 1990–1995 wereconsistent with that formulated for a composite ENSO

event. For example, the Indian monsoon operatednormally in 1994, while eastern Australia recorded itsworst drought. Even locally within Australia, whilemost of the eastern half of the continent was in droughtin 1992–1993, a 1000 km2 region south of Sydneyreceived its wettest summer on record. It is question-able whether or not all of these climatic responses canbe attributed to the 1990–1995 ENSO, but they haveoccurred without doubt in regions where ENSOlinkages or teleconnections to other climate phenom-ena operate. Additionally, the unprecedented natureof the 1982–1983 and 1990–1995 ENSO events maybe linked to significant volcanic eruptions. While theMexican El Chichon eruption of 1981 did not triggerthe 1982–1983 El Niño, it may have exacerbated itsintensity. Similarly, the 1990–1995 event correspondedwell with the eruption and subsequent global coolinggenerated by significant volcanic eruptions in 1991 ofthe Philippines’ Mt Pinatubo and Chile’s Mt Hudsonand, in 1992, of Mt Spurr in Alaska.

The events of the late twentieth century appearexceptional. They are not. Historical and proxy recordsare showing that mega-ENSO events occur every400–500 years. For example, around 1100 AD, rivers inthe Moche Valley of Peru reached flood levels of 18 m,destroying temples and irrigation canals built by theChimu civilization. In the same location, the worstfloods of the twentieth century (in 1926) reacheddepths of only 8 m. The latter event resulted in massivefires in the Rio Negro catchment of the AmazonianBasin. However, similar if not more extensive fires haveoccurred in 500, 1000, 1200, and 1500 AD. The 500and 1100 AD events appear in palaeo-records fromVeracruz and Mexico. In Veracruz, the latter eventsproduced flooding and laid down sediments over ametre thick. Compared to this, the 1995 event, whichhas been viewed as extreme, deposited only 10–15 cmof sediment. Other mega-ENSO events occurredaround 400, 1000, 1600, 2400, 4800, and 5600 BC inVeracruz. As with the prolonged 1990–1995 ENSOevent, it is probable that more than one ENSO eventwas involved at these times. The recent prolonged1990–1995 ENSO cycle may not be unusual, but ongeological timescales represents the upper end of ashifting climate hazard regime.

La Niña events Exceptionally ‘turned on’ Walker circulation is nowbeing recognized as a phenomenon in its own right.


The cool water that develops off the South Americancoastline can drift northward, and flood a 1–2° bandaround the equator in the central Pacific with waterthat may be as cold as 20°C. This phenomenon istermed La Niña (meaning, in Spanish, ‘the girl’) andpeaks between ENSO events. The 1988–1990 La Niñaevent appears to have been one of the strongest in40 years and, because of enhanced easterly trades,brought record flooding in 1988 to the Sudan,Bangladesh and Thailand in the northern hemispheresummer. In the Sudan, the annual rainfall fell in fifteenhours, destroying 70 per cent of that country’s houses.Regional flooding also occurred in Nigeria, Mozam-bique, South Africa, Indonesia, China, CentralAmerica, and Brazil. An unusual characteristic of thisevent was the late onset of flooding and heavy rain.Both the Bangladesh and the Thailand floodingoccurred at the end of the monsoon season. Bangladeshwas also struck by a very intense tropical cyclone inDecember 1988, outside the normal cyclone season.

In eastern Australia this same La Niña eventproduced the greatest rainfall in 25 years, with contin-uation of extreme rainfall events. Over 250 mm ofrainfall on 30 April 1988 caused significant damagealong a stretch of the coastal road between Wollongongand Sydney. Numerous landslides and washouts under-mined the roadway and adjacent main railway line,leading to a total road repair cost, along a 2 kilometerstretch of coastline, of $A1.5 million (Figure 2.9).Slippage problems were still in evidence there15 months later. The Australian summers of1989–1990 not only witnessed record deluges but alsoflooding of eastern rivers on an unprecedented scale.Lake Eyre filled in both years, whereas it had filledonly twice in the previous century. Towns such asGympie, Charleville (Queensland), and Nyngan (NewSouth Wales) were virtually erased from the map.

Global long-term links to drought and floods(Quinn et al., 1978; Pant & Parthasarathy, 1981; Bhalme et al., 1983; Pittock, 1984; Adamson et al., 1987)

Where the Southern Oscillation is effective in control-ling climate, it is responsible for about 30 per cent ofthe variance in rainfall records and some of the largestdeath tolls in the resulting famines. Over 10 millionpeople died in the state of Bengal during the1769–1770 ENSO event, while the event of 1877–1878– one of the strongest ENSO events recorded – wasresponsible for the deaths of 6 and 13 million people in

India and China, respectively. In China people ate theedible parts of houses, while in the Sudan there wereso many corpses that vultures and hyenas became fussyabout the ones they would eat.

Severe droughts can occur simultaneously aroundthe globe. In the twentieth century, the 1997–1998ENSO event brought severe drought to South Americafrom Colombia to north-eastern Brazil, the Caribbean,Central America, Hawaii, northern China andKorea, Vietnam, Indonesia, Papua New Guinea, andBangladesh (East Pakistan). Southern Peru, western

30 Climatic Hazards

Fig. 2.9A The La Niña event of 1987–1988 was one of the strongest at the end of the twentieth century. Extreme rainfall fell throughout eastern Australia. A) Fans deposited on the coastal road between

Wollongong and Sydney following more than 250 mm of rainfall on 30 April 1988.

Fig. 2.9B B) The same rainfall event in A) caused numerous land-slides and washouts that undermined the roadway andadjacent main railway. The total cost in road repairsalong this 2 km stretch of coastline was $A1.5 million.Slippage problems as a result have continued for15 years (photographs by Bob Webb, courtesy of MalBilaniwskyj and Charles Owen, New South Wales Depart-ment of Main Roads, Wollongong and Bellambi Offices).

Bolivia, Venezuela, and north-eastern Brazil areusually affected by drought during an ENSO event;however, greatest devastation occurs in the westernpart of the cell in southern Africa, India, Indonesia,and Australia. There is a strong coherence in thetiming of both flood and drought on the Nile in Africa,the Krishna in central India and the Darling River inAustralia. As well, rainfall over parts of Australia,Argentina and Chile, New Zealand and southern Africacoincides (Figure 2.10). Most notable are the years1894, 1910, 1917, 1961 and 1970. There is an evenbetter correspondence between droughts in India and‘turned off’ Walker circulation, or ENSO events. Thesevere droughts in India of 1899, 1905, 1918, 1951,1966, and 1971 all occurred before, or at times when,El Niño events were at their peak around Christmas orwhen the Southern Oscillation index was negative. Allof these years, except for 1951, were also times whendischarges on the Nile, Krishna, and Darling Riversystems were either very low, or nonexistent.

A similar relationship can be traced back even furtherin Indonesia. This is surprising because most climaticclassifications label Indonesia as a tropical country with

consistent yearly rainfall. Table 2.1 summarizes therelationship between monsoon droughts and ENSOevents during the period 1844–1983. In Indonesia, over93 per cent of monsoon droughts are associated with ElNiño events and 78 per cent of ENSO events coincidewith failure of the monsoon. The relationships are sostrong that in Java the advent of an ENSO event can beconfidently used as a prognostic indicator of subsequentdrought. In recent years, both India and Indonesia haveestablished distribution infrastructures that preventmajor droughts from causing starvation on as large ascale as they once did; however, the 1982–1983 eventcaused the Indonesian economy to stall as the countryshifted foreign reserves towards the purchase of riceand wheat.

Links to other hazards(Gray, 1984; Marko et al., 1988; Couper-Johnston, 2000)

The effect of ENSO events and the Southern Oscilla-tion goes far beyond just drought and flood. They areresponsible for the timing of many other naturalhazards that can influence the heart and fabric ofsociety. Three culturally diverse examples illustrate this


River discharge

SO index

Area indexIndian floodsand droughts

Indian floods

Indian droughts

Discharge on Nile, Darlingand Krishna Rivers coincidence

HighLow

1890 1900 1910 1920 1930 1940 1950 1960 1970 1980

60

50

40

30

20

10

420

-2-4

Fig. 2.10 Southern Oscillation index, drought and rainfall indices for India plus timing of coinciding low or high discharges on the Nile, Darling, andKrishna Rivers, 1890–1980 (based on Bhalme et al., 1983; Adamson et al., 1987).

fact. First is the effect of ENSO events on civilizationin Peru. The story is one of successive developmentand collapse of city-states. By the end of the eleventhcentury, most of Peru’s desert coast had come underthe influence of the Chimu civilization. They devel-oped a vast network of irrigation canals stretchinghundreds of kilometres between river valleys. Around1100 AD, extraordinary flooding exceeding 18 mdepths wiped out all of their engineering achievementsin the Moche Valley. Rival city-states were alsoaffected; however the Chimu undertook military actionto expand their area of influence so that they couldrelocate. While the rival states sank into revolt andanarchy, the Chimu conquests succeeded to rival theIncas in power. This example was but one phase in asequence of rapid cultural advancement, abandon-ment, and expansion of opportunistic city-states trig-gered by ENSO events along the west coast of SouthAmerica. Over four centuries, this process has coveredan area of 30 000 km2 of coastal plain.

The second example is the historical development ofEthiopia. The ENSO events of 1888 and 1891 killeda third of this country’s population at a time whenthe Italians hoped to establish a colonial empire in the

Horn of Africa. Other European countries – mainly theBritish – had taken advantage of ENSO-relateddroughts to expand their empires in Africa. However,Emperor Menilek took advantage of his weakenedenemies to consolidate his position as the country’sabsolute ruler. When the La Niña of 1892 brought rainand harvests to feed his army, Menilek was able todefeat the Italians at the Battle of Adowa and becomethe first African ruler to repel colonialism in Africa.

Third is the effect of the 1982–1983 ENSO event onAustralia. This brought the worst drought Australiahas experienced. Approximately $A2000 million waserased from the rural economic sector as sheep andcattle stocks were decimated by the extreme aridity.Wheat harvests in New South Wales and Victoria were,respectively, 29 per cent and 16 per cent of the averagefor the previous five years. The resulting lack of capitalexpenditure in the rural sector deepened what wasalready a severe recession, and drove farm machinerycompanies such as International Harvester andMassey Ferguson into near bankruptcy. By January1983, dust storms in Victoria and New South Waleshad blown away thousands of tons of topsoil. Finally,in the last stroke, the Ash Wednesday bushfires of16 February 1983, now viewed as one of the greatestnatural conflagrations observed by humans, destroyedanother $A1000 million worth of property in SouthAustralia and Victoria. At the peak of the event,Malcolm Fraser’s Liberal–National Party coalitiongovernment called a national election. The timingcould not have been worse, and the government of theday was swept convincingly from power.

Other natural hazards are associated with theSouthern Oscillation. This is exemplified in Figure 2.11for Australia between 1851 and 1980. (The index here ispart of the one used in Figure 2.8.) Shaded along thisindex are the times when eastern Australia was either indrought or experiencing abnormal rainfall. Sixty-eightper cent of the strong or moderate ENSO eventsbetween 1851 and 1974 produced major droughts ineastern Australia. Sixty per cent of all recoveries after anENSO occurrence led to abnormal rainfall and floods.Figure 2.11 also includes the number of tropicalcyclones in the Australian region between 1910 and1980; the discharge into the Murray River of Victoria’scentral north Campaspe River between 1887 and 1964;beach change at Stanwell Park, south of Sydney,between 1930 and 1980, described in Chapter 8; rainfallthis century at Helensburgh, south of Sydney; and the

32 Climatic Hazards

1844–96 1902–83Drought El Niño Drought El Niño

event event1844 1844 1902 19021845 1845–46 1905 19051850 1850 1913–14 19141853 none 1918–19 1918–191855 1855 1923 19231857 1857 1925–26 1925–261864 1864 1929 1929–301873 1873 1932 19321875 1875 1935 none1877 1877–78 1940 1939–401881 1880 1941 19411883 none 1944 1943–441884–85 1884–85 1945–46 19461888 1887–89 1953 19531891 1891 no data 1954–551896 1896 1976 1976

1982–83 1982

Table 2.1 Comparison of drought in Indonesia with El Niño–SouthernOscillation events, 1844–1983 (Quinn et al., 1978).

incidence of landslip in the Kiama area, south of Sydney,between 1864 and 1978. Drawn on the diagram are linespassing through all data at the times when the SouthernOscillation was strongly ‘turned on’. At this time, tropicalcyclones were more frequent, the Campaspe River

discharge high, Sydney rainfall heavy, beach erosion atStanwell Park more severe, and Kiama landslip morefrequent. The diagram indicates that ‘turned on’ Walkercirculation results in this predictable geomorphicresponse. The 1933, 1951 and 1961 periods of strong


SOUTHERN OSCILLATION INDEX Australian rainfallAbnormal rain

Drought

Major bush fires in Southeast Australia4.0

2.0

0.0

-2.0

-4.0

1840 1860 1880 1900 1920 1940 1960 1980

1840 1860 1880 1900 1920 1940 1960 1980

Strong ENSOModerate ENSO

Number of tropical cyclones

20

10

Discharge - Campaspe River, Victoria 750

500

250meg

alite

rsx

1000

Beach change - Stanwell Park, NSW-20

-100

10

20met

ers

Erosion -

Accretion +

Rainfall at Helensburgh, NSW3.0

2.0

1.0

0.0

met

ers

Landslip at Kiama, NSW

High

Moderate

Low

Fig. 2.11 Time series of the Southern Oscillation index, major bushfires (from Luke & McArthur, 1978), Australian cyclone frequency (from Lourensz,1981), Campaspe River discharge (from Reichl, 1976), beach change at Stanwell Park (from Bryant, 1985), rainfall near Sydney, andlandslip incidence at Kiama (courtesy Jane Cook, School of Education, University of Wollongong).

Walker circulation show up in all five time series at bothregional and localized levels. The reverse conditionsapply during ENSO events, except for the frequency oftropical cyclones in the Australian region; historicallythis is not well correlated to ENSO events. This is owingto the fact that many cyclones are still counted withinthe Australian region even though they have shiftedeastward during these events. More importantly, theSouthern Oscillation appears to switch months inadvance of the arrival of the meteorological conditionsdriving these responses. For instance, the heavy rainfallin eastern Australia, beginning in the autumn of 1987,was predicted 12 months in advance. Unfortunately,while the timing of rain can be forecast, its preciselocation or the amounts that may fall remain unpre-dictable because of the spatial variability of precipitationalong the coast. Even during the worst of droughts,some areas of eastern Australia received normal rainfall.

Links to other hazards are also appearing globallyas researchers rapidly investigate teleconnectionsbetween hazards and the Southern Oscillation. Forexample, droughts are often followed by extensive firesin Florida, eastern Russia, and Indonesia. There isalso heavier precipitation, mainly falling as snow innorth-eastern United States and south-eastern Canada.The 1997–1998 ENSO event produced five days offreezing rain in this region that downed 120 000 kmof power and telephone lines, paralyzing the city ofMontreal. The duration and incidence of tropicalcyclones in the equatorial Atlantic is profoundlydepressed during ENSO years (Figure 2.12). Here,upper tropospheric winds between 0° and 15°N mustbe easterly, while those between 20° and 30°N must be

westerly, before easterly wave depressions and distur-bances develop into tropical cyclones (hurricanes).During ENSO events, upper westerlies tend todominate over the Caribbean and the western tropicalAtlantic, giving conditions that suppress cycloneformation. Years leading up to an ENSO event havetended to produce the lowest number of tropicalcyclone days in the Atlantic over the past century.Additionally, since 1955, the number of icebergspassing south of 48°N has been highly correlated tothe occurrence of ENSO events. ENSO events occurconcomitantly with a strengthened Icelandic Low thatresults in stronger winds along the North Americaneast coast. This exacerbates the production and south-ward movement of icebergs. Both the 1972–1973 and1982–1983 ENSO events produced over 1500 icebergssouth of 48°N, a number that dramatically contrastswith non-ENSO years when fewer than one hundredicebergs per year were recorded this far south. Earlydetection of the onset of ENSO events can thusprovide advance warning not only of drought andrainfall over a significantly large part of the globe, butalso of the occurrence of numerous other associatedhazards.

Other Oscillation phenomena (Trenberth & Hurrell, 1994; Hurrell, 1995; Villwock, 1998;Stephenson, 1999; Boberg & Lundstedt, 2002; Hurrell et al.,2002)

North Atlantic Oscillation

There are two other regions: the north Atlantic andPacific Oceans where pressure oscillates on the same

34 Climatic Hazards

1900 1920 1940 1960 1980

- ENSO year

Num

ber

of D

ays 60

40

20

2000

Fig. 2.12 Number of hurricane days in the North Atlantic–Caribbean Sea region, 1900–2003 (based on Gray, 1984; Hurricane Research Division, 2003).

scale as the Southern Oscillation. Variations in theintensity of atmospheric circulation at these locationsare known as North Atlantic Oscillation (NAO) andthe North Pacific Oscillation (NPO). Whereas theSouthern Oscillation characterizes zonal pressure fluc-tuations in the tropics across the Pacific, the NAO andNPO are meridional air circulation phenomena reflect-ing inter-annual variability in the strength of Rossbywaves in the northern hemisphere. The North AtlanticOscillation also reflects major temperature differencesin winter. An index of the NAO since 1864 hasbeen constructed using the normalized atmosphericpressure between the Icelandic Low, measured atStykkisholmur/Reykjavik, Iceland, and the AzoresHigh, measured at Lisbon (Figure 2.8). Positive valuesof the index indicate stronger-than-average westerlies.Air temperature fluctuations are best characterized bythe winter temperature between Jakobshavn, WestGreenland, and Oslo, Norway. Significant environ-mental changes throughout the twentieth century havebeen associated with the NAO. These include varia-tions in wind strength and direction, ocean circulation,sea surface temperatures, precipitation, sea-ice extent,and changes in marine and freshwater ecosystems. TheNorth Atlantic Oscillation influences an area fromSiberia to the eastern seaboard of the United States.Given the magnitude and extent of the NAO, it issurprising that little attention was given to it untilthe 1990s.

The NAO undergoes large variation from year toyear and, more importantly, at decadal time scales.However, large changes can also occur during a winterseason. Hence it is difficult to characterize northernhemisphere winters as completely severe or benign.This probably reflects the fact that climate processesare random over time. Large amplitude anomalies inwintertime stratospheric winds precede anomalousbehavior of the NAO by one to two weeks. Changesto the strength and pathway of mobile polar highs andthe amplitude of Rossby waves in the polar jet streamalso correspond to winter anomalies. Similar to theSouthern Oscillation, short-term alterations are alsopreceded by changes in sea surface temperature aboutnine months in advance. Together the NAO and SOaccount for 50 per cent (34 per cent and 16 per cent,respectively) of the variation in winter temperatures inthe northern hemisphere outside the tropics. However,the Southern and North Atlantic Oscillations act inde-pendently of each other. The eastern United States and

adjacent Atlantic region are the only areas where bothphenomena always act together.

There are two distinct decadal phases associatedwith the NAO. When the index is positive, atmosphericpressure is low around Iceland and high in the Azores.Sea surface temperatures off the United States easternseaboard and western Europe are warmer thannormal, while those in the western north Atlantic arecooler than average (Figure 2.13a). These aspectsenhance meridional air circulation in the atmosphere,strengthening westerly winds at temperate latitudesand forcing warm, moist air from the Atlantic Oceanover northern Europe. Hence land temperatures arewarmer adjacent to these pools of warmer oceanwaters, leading to milder winters. Temperatures arecool over North Africa, the Middle East, easternCanada, and Greenland. The flux of heat across thenorth Atlantic produces more intense and frequentstorms, a fact that has been more obvious since 1970(Figure 2.14). As a result, wave heights increase in thenorth-east Atlantic and decrease south of 40°Nlatitude. This is paralleled by changes in precipitation– wetter over Scandinavia and drier in the Mediter-ranean. Over northern Africa, more dust moves fromthe Sahara into the Atlantic Ocean. Sea-ice alsoextends further south over the west Atlantic, whileretreating east of Greenland. These effects are eitherwind- or temperature-driven. In the negative phase,the pressure gradient between Iceland and the Azoresis weakened, the Mediterranean region is wetter andwarmer, and northern Europe is cooler and drier(Figure 2.13b). While storms off the east coast of theUnited States may be more frequent, they are virtuallyabsent off western Europe. These periods of anom-alous circulation patterns have persisted over longperiods. For example, the NAO index was positivefrom the turn of the twentieth century until about the1930s. During the 1960s, the NAO index was negativewith severe winters across northern Europe. Since theearly 1970s, circulation has been locked into a positivephase with warmer temperatures than average overEurope and cooler ones over eastern Canada andGreenland.

North Pacific Oscillation

The North Pacific Oscillation (NPO) measures thestrength and position of the Aleutian low-pressuresystem, mainly in winter. Because there are fewrecording stations in the north Pacific, an index


36 Climatic Hazards

characterizing this oscillation has been built usingaverage surface air pressure over the region30°N–65°N, 160°E–140°W (Figure 2.8). Over thenorth Pacific Ocean, atmospheric pressure nearthe Aleutian Islands varies out-of-phase with thatto the south, seesawing around the mean position ofthe Pacific subtropical jet stream. Over North America,variations in atmospheric pressure over westernCanada and the north-western United States are nega-tively correlated with those over the south-easternUnited States; but positively correlated with thesubtropical Pacific. Changes in these centers occurconcomitantly with changes in the amplitude of Rossbywaves and the intensity of mobile polar highs. TheNPO is linked to changes in tropical Pacific sea surfacetemperatures associated with the ENSO phenomenon.

Westerlies Trades

NAO Positive phase

NAO Negative phase

L

H

L

H

H

Cold air or water

Warm air or water

L HCenter of low pressure

Center ofhigh pressure

Area of increased rain

A

B

Fig. 2.13 Atmospheric circulation patterns and climatic impacts associated with A) positive and B) negative phases of the North Atlantic Oscillation (basedon Stephenson, 1999).

Inde

x

20

0

-20

1960 1970 1980 1990

15

10

0

Num

ber

of s

torm

s

NAO

Storms

5

Fig. 2.14 The number of winter storms in the north Atlantic between 1956and 1994 with pressure below 950 hPa (Villwock, 1998).

In this respect, the NPO is probably the temperatenorth Pacific extension of the Southern Oscillation,with sea surface temperature changes in the northPacific lagging those in the tropics by three months.However, the NPO does have its own independentfeatures. Of all atmospheric indices, the North PacificOscillation shows the greatest well-defined trend overtime (Figure 2.8). The index has been steadily declin-ing, although there is significant year-to-year variation.Concomitantly with this trend, the Aleutian Low hasintensified and shifted eastward in winter. As a result,storm tracks have shifted southward. This has resultedin warmer and moister air being transported north-ward along the west coast of North America intoAlaska. Sea surface temperatures in the north Pacifichave cooled, but sea-ice has decreased in the BeringSea.

ASTRONOMICAL CYCLES

Solar cycles(Shove, 1987; Hoyt & Schatten, 1997; Boberg & Lundstedt,2002)

Sunspots are surface regions of intense disturbance ofthe sun’s magnetic field: 100–1000 times the sun’saverage. The spots themselves appear dark, becausethe magnetic field is so strong that convection is inhib-ited and the spot cools by radiation emission. However,the total amount of radiation given out by the sunincreases towards peaks in sunspot activity. Thesun has a 22-year magnetic cycle, consisting of two 11-year sunspot cycles. In Figure 2.15, these are plottedback to 1500 AD. The 22-year periodicity known as theHale Cycle is evident as an enhanced peak in sunspotnumbers in this figure. There are also other periodici-ties, at 80–90 years and 180 years, in the number ofsunspots. The interaction of these cycles has producedperiods where there was little sunspot activity. Twoof these, the Maunder and Dalton Minimums, in thelate seventeenth and early nineteenth centuries,respectively, are evident in Figure 2.15. Both arecorrelated to periods of cooler climate, at least in thenorthern hemisphere. The Maunder Minimum is alsoknown as the Little Ice Age. Periods of high activitycorrelate with warmer climate. This occurred in thethirteenth century during the Medieval Maximum andover the last 150 years concurrent with modern globalwarming.

Other phenomena characterize our sun. Solar flares– representing ejection of predominantly ionizedhydrogen in the sun’s atmosphere at speeds in excessof 1500 km s-1 – develop in sunspot regions. Flaresenhance the solar wind that ordinarily consists ofelectrically neutral, ionized hydrogen. Accompanyingany solar flare is a pulse of electro-magnetic radiationthat takes eight minutes to reach the Earth. Thisradiation, in the form of soft X-rays (0.2–1.0 nmwavelengths), interacts with the Earth’s magneticfield, increasing ionization in the lowest layer of theionosphere at altitudes of 65 km. The enhanced solarwind arrives one to two days after this magnetic pulseand distorts the magnetosphere, resulting in large,irregular, rapid worldwide disturbances in the Earth’sgeomagnetic field. Enhanced magnetic and ioniccurrents during these periods heat and expand theupper atmosphere. The solar wind also modulatescosmic rays affecting the Earth. When the solarwind is strong, cosmic rays are weak – resultingin decreased cloud formation and warmer surfaceair temperatures. Cloud cover decreases around3–4 per cent between troughs and peaks in geo-magnetic activity. Solar flare activity and the strengthof the solar wind are weakly correlated to the numberof sunspots (Figure 2.15b). Thus, geomagnetic activityis a better indicator of solar influence on climate thanthe number of sunspots. While solar activity on atimescale of weeks can affect the temperaturestructure of the stratosphere, there is no provenmechanism linking it to climatic change over longerperiods near the surface of the Earth. This fact hastended to weaken the scientific credibility of researchinto the climatic effects of solar cycles.

Despite this, the literature is replete with examplesshowing a correlation between climate phenomena andsolar activity in the form of sunspots. Much of theemphasis has been upon a solar sunspot–globaltemperature association. High sunspot activity leadsto warmer temperatures and more precipitation,although this relationship is consistent neither overtime nor the surface of the globe. This discussion isbeyond the scope of this text and the curious readershould see Hoyt and Schatten (1997) for further infor-mation. More substantial is the fact that thunderstorm,lightning and tropical cyclone activity worldwideincreases during periods of sunspot activity. Inaddition, 15 per cent of the variation in the position ofstorm tracks in the north Atlantic and Baltic Sea can be


accounted for by solar variations. Storm tracks shiftsouthward by 3–4° latitude in both regions whensunspot activity is highest. Forest fires are morenumerous in North America at peaks in the sunspotcycle. The decadal variance in the North Atlantic

Oscillation is also positively correlated to enhancedgeomagnetic activity. Finally, there are solar periodici-ties in the frequency of earthquakes and volcaniceruptions. The reason for the latter is beyond the scopeof this text.

38 Climatic Hazards

AA Geomagnetic Index

0

10

20

30

40

Sunspot Number

0

40

80

120

160

200

1860 1880 1900 1920 1940 1960 1980 2000

B

A Zurich, Annual Average, Sunspot Numbers 1500-2002

020406080

100120140160180200

1500 1600 1700 1800 1900 2000

Maunderminimum

Daltonminimum

Fig. 2.15 A) Time series of the annual average, Zurich sunspot number between the years 1500 and 2002. Values between 1500 and 1749 are fromShove (1987), between 1750 and 2002 from the National Geophysical Data Centre (2003a). B) Detail of the Zurich sunspot record between1860 and 2002 with the annual AA geomagnetic index superimposed. Latter from National Geophysical Data Centre (2003b).

The 18.6-year MN lunar cycle(Tyson et al., 1975; Currie, 1981, 1984; Wang & Zhao, 1981)

The 18.6-year lunar tidal cycle (MN) represents afluctuation in the orbit of the moon. The moon’s axis oforbit forms a 5° angle with the sun’s equator as shownin Figure 2.16; however, with each orbit the moon doesnot return to the same location relative to the sun.Instead, it moves a bit further in its orbit. The processis analogous to a plate spinning on a table. The platemay always be spinning at the same angle relative tothe table, but the high point of the plate does not occurat the same point. Instead, it moves around in thedirection of the spin. The moon’s orbit does the samething, such that 9.3 years later the high point of theorbit is at the opposite end of the solar equator, and 9.3years after that it returns to its original position. Thus,there is an 18.6-year perturbation in the orbit of themoon. This perturbation appears trivial until youconsider the moon’s orbit in relation to the Earth(Figure 2.17). Relative to the Earth, the solar equatormoves seasonally, reaching a maximum of 23.46°N ofthe equator on 22 June and a minimum of 23.46°S ofthe Earth’s equator on 22 December. If the moon–sunorbital configuration in Figure 2.16a is superimposed

onto the Earth, then the moon is displaced in its orbitclosest to the Earth’s poles, 28.5° (23.5° + 5°) northand south of the equator. The moon’s gravitationalattraction on the Earth is slightly greater when themoon is displaced poleward. In 9.3 years’ time, themoon will move to the opposite side of the sun’sequator. Relative to the Earth, the moon will becloser to the equator, or in its minimum position(Figure 2.17b). Although it is only 3.7 per cent thedaily effect, the gravitational effect of the MN tide hastwo to three years to act on any standing, planetarywave phenomena.

Standing atmospheric waves, such as Rossby wavesembedded in the jet stream (Figure 2.4), are amplifiedby the MN lunar tide. An 18–20 year periodicityappears in sea level, air pressure, and temperaturerecords at locations beneath the jet stream from Japanto Scandinavia. Detailed examination of the occur-rence of drought on the United States Great Plainssince 1800 shows a strong link between drought andthe 18.6-year MN lunar cycle. A 20-year cycle is alsoevident in rainfall over the Yangtze River Basin ofChina, coinciding with the American pattern and theIndian flood record. In the southern hemisphere, a 20-year periodicity appears in summer rainfall in


Moon’s positionJune 22

Moon’s orbit relative to sun

5˚ angle

A

B

Sun’s equator

Sun’s equator

Moon’s positionnext orbit June 23

Precession in moon’s orbit

Sun’s equator

Successive positions relative to Sun's equator

5˚ angle

Moon’s position 9.3 years later

Fig. 2.16 Schematic diagrams showing the precession of the Moon’s orbit relative to the Sun’s equator for A) lunar maxima and B) lunar minima.

southern Africa and eastern Australia. A sampling ofthese associations worldwide is presented for thewestern United States and Canada, northern China,India, the Nile region of Africa, and the mid-latitudesof South America (Table 2.2). In all cases, except forIndia, the mean discrepancy between the occurrenceof the hazard and peaks in the lunar tide is less thanone year. Not only is the 18.6-year cycle dominant, butthe data also show temporal and spatial bistablephasing. In this process, drought may coincide withmaxima in the 18.6-year lunar cycle for one period, but

then suddenly switch to a minimum. This ‘flip-flop’happens every 100–300 years, with the most recentoccurring at the turn of the twentieth century in SouthAmerica, China, Africa, and India. On the GreatPlains, no bistable ‘flip-flop’ has occurred since 1657;however, the last lunar cycle maximum in 1991 wit-nessed flooding instead of drought in this region. Atpresent, disparate regions of the globe’s major wheatgrowing areas undergo synchronous periods of droughtthat have major implications for the price of grain andthe relief of famine-stricken regions.

40 Climatic Hazards

A

B

North Pole

North Pole

South Pole

South Pole

23.5˚angle

Moon’s angle toEarth’s equatoris 28.5˚

5˚ angle

5˚ angle

9.3 years later

Moon

Sun’s equator seasonal path

Earth’s equator

Sun’s equator

Earth’s equator23.5˚angle

Moon’sangle toEarth’sequatoris 18.5˚

Fig. 2.17 Schematic presentations of the 18.6-year lunar orbit relative to the Earth for A) lunar maxima and B) lunar minima.

CONCLUDING COMMENTS

The dominant factor controlling air movement acrossthe surface of the Earth is the movement of cold polarair in the form of mobile polar highs from the polestowards the equator. This drives global air circulationand satisfies the requirement in the atmosphere tobalance the inequality in heating and cooling betweenthe equator and poles, respectively. The concept ofmobile polar highs implies that the climate of highlatitudes is the key to world climate and global climatechange. Superimposed upon this meridional exchangeare zonal processes controlled by ocean–atmosphereinteractions. Chief amongst these is the SouthernOscillation. While droughts and floods have plaguedpeople throughout recorded history, and are probablyresponsible for the greatest loss of life of any natural

hazards, the realization is that they represent insepa-rable hazards that follow each other in time as nightdoes day. Not only are the two events linked, but theiroccurrence is also remarkably coincident across theglobe. They are intricately connected to the SouthernOscillation and other spatial cyclic phenomena such asthe North Atlantic and North Pacific Oscillations.Other hazards – such as tropical and extra-tropicalcyclones, wave erosion, and land instability – arealso correlated to the Southern Oscillation, and thussubject to prediction. At mid-latitudes in the northernhemisphere, storminess is more important. Here, theNorth Atlantic Oscillation is the controlling factor.Similar influences may occur in western NorthAmerica due to changes in sea surface temperature inthe north Pacific.


Lunar Canadian United States N. China Patagonian Nile in IndiaMaxima Prairies Great Plains Andes Africa1583 1581D 1582D1601 1601D 1600D 1606F1620 1622D 1620D 1621F1638 1640D 1640D 1637F1657 1652F 1659D 1656F1676 1674F 1678D 1675F1694 1694F 1693F1713 1712F 1710F 1713D1731 1729F 1727F 1733D1750 1752F 1752D 1750D1768 1768F 1768D 1766D1787 1786F 1784D 1789D1806 1805F 1805D 1806F 1802D 1806D1824 1823F 1824D 1823F 1822D 1822D1843 1843F 1844D 1846D 1841D 1837D1861 1859F 1861D 1862D 1864D1880 1881F 1879D 1881D 1880D 1885D1899 1900F 1901D 1900D 1895D 1902D 1902D1918 1916F 1919D 1918D 1918F 1917F 1913F1936 1932F 1935D 1935F 1937F 1936F 1939F1955 1953F 1955D 1954F 1956F 1953F 1958F1973 1975F 1975D 1974F 1973F 1975F 1976F1991 1995F 1993F 1990F 1992F 1996F 1990FSources: Currie (1984) and various internet searches. Since 1980, plotted rainfall anomalies for each region using

http://climexp.knmi.nl/fieldplot.cgi?someone@somewhere+gpcp

Table 2.2 Timing of flood (F) and drought (D) in North America, northern China, Patagonia, the Nile valley, India and the coincidence with the18.6 year lunar tide

There is now indisputable proof that droughts andsubsequent heavy rainfall periods are cyclic, not onlyin semi-arid parts of the globe, but also in thetemperate latitudes. This cyclicity correlates wellwith the 18.6-year lunar tide in such diverse regionsas North America, Argentina, northern China, theNile Valley, and India. There is also evidence thatthe 11-year sunspot cycle is present in rainfall timeseries in some countries. Long-term astronomicalperiodicities, especially in the northern hemisphere,permit scientists to pinpoint the most likely year thatdrought or abnormal rainfall will occur, while theonset of an ENSO event with its global consequencespermits us to predict climatic sequences up to ninemonths in advance. These two prognostic indicatorsshould give countries time to modify both long- andshort-term economic strategies to negate the effectsof drought or heavy rainfall.

Unfortunately, very few governments in the twen-tieth century have lasted longer than 11 years – thelength of the shortest astronomical cycle. Realistically,positive political responses – even in the United States– to dire drought warnings are probably impossible.Societies must rely upon the efficiency and power oftheir permanent civil services to broadcast thewarnings, to prepare the programs reducing theeffects of the hazards, and to pressure governments ofthe day to make adequate preparations.

Finally, it should be realized that the astronomicalcycles and oscillations such as the Southern or NorthAtlantic Oscillations account for only 15–30 per centof the variance in rainfall in countries where theireffects have been defined. This means that 70 per centor more of the variance in rainfall records must be dueto other climatic factors. For example, during thedrought that affected eastern Australia in 1986–1987,Sydney was inundated with over 400 mm of rainfall,breaking the previous 48-hour rainfall record, andgenerating the largest storm waves to affect the coastin eight years. This randomness exemplifies theimportance of understanding regional and localclimatic processes giving rise to climatic hazards.These regional and local aspects will be described inthe following chapters.


Adamson, D., Williams, M.A.J. and Baxter, J.T. 1987. Complex lateQuaternary alluvial history in the Nile, Murray–Darling, andGanges Basins: three river systems presently linked to theSouthern Oscillation. In Gardiner, V. (ed.) InternationalGeomorphology Pt II, Wiley, NY, pp. 875–887.

Allan, R., Lindesay, J. and Parker, D. 1996. El Niño SouthernOscillation and Climatic Variability. CSIRO Publishing,Melbourne.

Bhalme, H.N., Mooley, D.A. and Jadhav, S.K. 1983. Fluctuations inthe drought/flood area over India and relationships with theSouthern Oscillation. Monthly Weather Review 111: 86–94.

Boberg, F. and Lundstedt, H. 2002. Solar wind variations related tofluctuations of the North Atlantic oscillation. GeophysicalResearch Letter 29(15) 10.1029/2002GL014903.

Bryant, E.A. 1985. Rainfall and beach erosion relationships,Stanwell Park, Australia, 1895–1980: worldwide implications forcoastal erosion. Zeitschrift für Geomorphologie Supplementband57: 51–66.

Bryant, E.A. 1997. Climate Process and Change. CambridgeUniversity Press, Cambridge.

Bryson, R. and Murray, T. 1977. Climates of Hunger. AustralianNational University Press, Canberra.

Couper-Johnston, R. 2000. El Nino: The Weather Phenomenon ThatChanged the World. Hodder and Stoughton, London.

Currie, R.G. 1981. Evidence of 18.6 year MN signal in temperatureand drought conditions in N. America since 1800 A.D. JournalGeophysical Research 86: 11055–11064.

Currie, R.G. 1984. Periodic (18.6-year) and cyclic (11-year) induceddrought and flood in western North America. Journal GeophysicalResearch 89(D5): 7215–7230.

Glantz, M.H. 1996. Currents of Change: El Niño’s Impact onClimate and Society. Cambridge University Press, Cambridge.

Glantz, M.H., Katz, R.W. and Nicholls, N. 1991. TeleconnectionsLinking Worldwide Climate Anomalies: Scientific Basis andSocietal Impact. Cambridge University Press, Cambridge.

Gray, W.M. 1984. Atlantic seasonal hurricane frequency Part I: ElNiño and 30 mb quasibiennial oscillation influences. MonthlyWeather Review 112: 1649–1667.

Hoyt, D.V. and Schatten, K.H. 1997. The Role of the Sun in ClimateChange. Oxford University Press, Oxford.

Hurrell, J.W. 1995. Decadal trends in the North Atlantic Oscillation:Regional temperatures and precipitation. Science 269: 676–679.

Hurrell, J.W. 2002a. North Atlantic Oscillation (NAO) indicesinformation: Winter (Dec–Mar). <http://www.cgd.ucar.edu/~jhurrell/nao.stat.winter.html>

Hurrell, J.W. 2002b. North Pacific (NP) index information.<http://www.cgd.ucar.edu/~jhurrell/np.html>

Hurrell, J.W., Kushnir, Y., Ottersen, G. and Visbeck, M. (eds) 2002.The North Atlantic Oscillation: Climatic Significance and Environ-mental Impact. American Geophysical Union, Washington.

Hurricane Research Division 2003. What are the most and leasttropical cyclones occurring in the Atlantic basin and strikingthe USA?. United States Department of Commerce. <http://www.aoml.noaa.gov/hrd/tcfaq/tcfaqE.html>

42 Climatic Hazards

Jet Propulsion Laboratory 1995a. TOPEX/POSEDON: Wind Speed,January. <http://education.gsfc.nasa.gov/experimental/all98invProject.Site/Pages/trl/inv4-7WIND_SPEED_JAN.html>

Jet Propulsion Laboratory 1995b. TOPEX/POSEDON: Wind Speed,July. <http://education.gsfc.nasa.gov/experimental/all98invProject.Site/Pages/trl/inv4-7WIND_SPEED_JUL.html>

Lamb, H.H. 1982. Climate, History and the Modern World.Methuen, London.

Lamb, H.H. 1986. The causes of drought with particular referenceto the Sahel. Progress in Physical Geography 10(1): 111–119.

Leroux, M. 1993. The Mobile Polar High: a new concept explainingpresent mechanisms of meridional air-mass and energyexchanges and global propagation of palaeoclimatic changes.Global and Planetary Change 7: 69–93.

Leroux, M. 1998. Dynamic Analysis of Weather and Climate:Atmospheric Circulation, Perturbations, Climatic Evolution.Wiley-Praxis, Chichester.

Lourensz, R.S. 1981. Tropical Cyclones in the Australian RegionJuly 1909 to June 1980. Australian Bureau of Meteorology,Australian Government Publishing Service, Canberra.

Luke, R.H. and McArthur, A.G. 1978. Bushfires in Australia.Australian Government Publishing Service, Canberra.

Marko, J.R., Fissel, D.B. and Miller, J.D. 1988. Iceberg movementprediction off the Canadian east coast. In El-Sabh, M.I. andMurty, T.S. (eds) Natural and Man-made Hazards. Reidel,Dordrecht, pp. 435–462.

National Geophysical Data Center 2003a. Sunspot numbers.<ftp://ftp.ngdc.noaa.gov/STP/SOLAR_DATA/SUNSPOT_NUMBERS>

National Geophysical Data Center 2003b. AA Index. <ftp://ftp.ngdc.noaa.gov/STP/SOLAR_DATA/RELATED_INDICES/AA_INDEX/>

Pant, G.B. and Parthasarathy, B. 1981. Some aspects of an associationbetween the Southern Oscillation and Indian summer monsoon.Archiv für Meteorologie, Geophysik und Bioklimatologie. B29:245–252.

Quinn, W.H., Zopf, D.O., Short, K.S. and Kuo Yang, R.T.W. 1978.Historical trends and statistics of the southern oscillation, El Niño, and Indonesian droughts. U.S. Fisheries Bulletin 76:663–678.

Philander, S.G. 1990. El Niño, La Niña and the Southern Oscillation.Academic, San Diego.

Pittock, A.B. 1984. On the reality, stability and usefulness of southernhemisphere teleconnections. Australian Meteorological Magazine32(2): 75–82.

Reichl, P. 1976. The Riverine Plains of northern Victoria. In Holmes,J.H. (ed.) Man and the Environment: Regional Perspectives.Longman, Hawthorn, pp. 69–95.

Shove, D.J. 1987. Sunspot cycles. In Oliver, J.E. and Fairbridge,R.W. (eds) Encyclopedia of Climatology. Van NostrandReinhold, New York, pp. 807–815.

Stephenson, D.B. 1999. The North Atlantic Oscillation thematicweb site. <http://www.met.rdg.ac.uk/cag/NAO/>

Trenberth, K.E. and Hurrell, J.W. 1994. Decadal atmospheric–oceanvariations in the Pacific. Climate Dynamics 9: 303–319.

Tyson, P.D., Dyer, T.G.S., and Mametse, M.N. 1975. Secularchanges in South African rainfall 1880–1972. Quarterly Journalof the Royal Meteorological Society 101: 817–833.

Villwock, A. 1998. CLIVAR initial implementation plan: Chapter 5,The North Atlantic Oscillation. <http://www.clivar.org/publications/other_pubs/iplan/iip/pd1.htm>

Wang, S-W. and Zhao, A-C. 1981. Droughts and floods in China1440–1979. In Wigley, T.M.L., Ingram, M.J. and Farmer, G. (eds)Climate and History, Studies in Past Climates and Their Impacton Man. Cambridge University Press, Cambridge, pp. 271–288.


INTRODUCTION

The previous chapter concentrated upon large-scalepressure patterns and their changes across the surfaceof the Earth. While these patterns regionally controlprecipitation, large-scale vortices amplify the amountsand add a wind component to produce devastatingstorms. In addition, moderate-to-strong winds on aregional scale have the capacity to suspend largequantities of dust. This material can be transportedthousands of kilometers, a process that represents asignificant mechanism for the transport of sedimentacross the surface of the globe. Dust storms are exac-erbated by drought in low rainfall, sparsely vegetatedregions of the world. For that reason, they pose aslowly developing but significant hazard in sparselysettled semi-arid regions. Unfortunately, many ofthese areas have been marginalized by the degradingagricultural practices of humans over thousands ofyears. However, in some regions where settlementhas occurred only in the last 200 years, we arepresently witnessing this marginalization process. Duststorms are one of the most prominent and long-lastingsignatures of human impact on the landscape.

In this chapter, the process, magnitude, and fre-quency of tropical cyclones are described first,followed by a description of some of the more familiarcyclone disasters. This section concludes with acomparison of the human response to cyclones in

Australia, the United States, and Bangladesh –formerly East Pakistan. Extra-tropical cyclones arethen described with particular reference to majorstorms in the northern hemisphere and exceptionaldevelopment of lows or bombs off Japan, the UnitedStates, and the east coast of Australia. Storm surgesdriven over bodies of water by these weather phenom-ena are then described. It is appropriate at this stage tointroduce more formally the concepts of probability ofoccurrence and exceedence using storm surges as anexample. These concepts are central to evaluating justhow big and frequent is a particular hazard event.Finally, the chapter concludes with a description ofdust storms and their impact.

TROPICAL CYCLONES

Introduction (Nalivkin, 1983)

Tropical cyclones are defined as intense cyclonicstorms that originate over warm tropical seas. In NorthAmerica, the term ‘hurricane’ is used because cyclonerefers to an intense, counterclockwise-rotating, extra-tropical storm. In Japan and South-East Asia, tropicalcyclones are called ‘typhoons’. Figure 3.1 summarizesthe hazards relating to tropical cyclones. These can begrouped under three headings: storm surge, wind, andrain effects. Storm surge is a phenomenon wherebywater is physically piled up along a coastline by lowpressure and strong winds. This leads to loss of life

Large-scaleStorms as a Hazard

C H A P T E R 3

through drowning, inundation of low-lying coastalareas, erosion of coastline, loss of soil fertility due tointrusion by ocean saltwater, and damage to buildingsand transport networks. In the largest cyclones, windscan exceed 300 km hr-1. High wind velocities candirectly cause substantial property damage and loss oflife, and constitute the main agent for crop destruction.Surprisingly, strong winds – simply because they are sostrong – can also exacerbate the spread of fires in urbanand forested areas, even under heavy rainfall.

On average, a tropical cyclone can dump 100 mmper day of rain within 200 km of the eye, and 30–40 mm per day at distances of 200–400 km. Theserates can vary tremendously depending upon localtopography, cyclone motion, and the availability ofmoisture. In 1952, a tropical cyclone at Reunion Islanddropped 3240 mm of rain in three days. Rainfall isresponsible for loss of life, property damage, and cropdestruction from flooding – especially on densely pop-ulated floodplains. Contamination of water suppliescan lead to serious disease outbreaks weeks after thecyclone. Heavy rain in hilly or mountainous areas isalso responsible for landslides or mudflows as flood-waters in stream and river channels mix with excesssediment brought down slopes. The destruction ofcrops and saline intrusion can also result in famine thatcan kill more people than the actual cyclone event.This was especially true on the Indian subcontinentduring the latter part of the nineteenth century.

Earthquakes are not an obvious consequence ofcyclones; however, there is substantial evidence for theiroccurrence during cyclones. Pressure can vary dramati-cally in a matter of hours with the passage of a cyclone,bringing about a consequentially large decrease in the

weight of air above the Earth’s surface. The deloadingcan be as much as 2–3 million tonnes km-2 over a matterof hours. In addition, storm surges about 6–7 m in heightcan occur in shallow seas with a resulting increase inpressure on the Earth’s surface of 7 million tonnes km-2.In total, the passage of a cyclone along a coast can inducea change in load on the Earth’s crust of 10 million tonneskm-2. In areas where the Earth’s crust is already understrain, this pressure change may be sufficient to triggeran earthquake. The classic example of a cyclone-inducedearthquake occurred with the Tokyo earthquake of 1923(see Figure 3.2 for the location of major placenamesmentioned in this chapter). A typhoon swept throughthe Tokyo area on 1 September, and an earthquakefollowed that evening. The earthquake caused therupture of gas lines, setting off fires that were fanned bycyclone-force winds through the city on 2 September. Inall, 143 000 people lost their lives, mainly through incin-eration. The events of this tragedy will be described inmore detail in Chapter 10. There is also evidence thattropical cyclones have triggered earthquakes in otherplaces along the western margin of the Pacific Plate andalong plate boundaries in the Caribbean Sea. In CentralAmerica, the coincidence of earthquakes and cycloneshas a higher probability of occurrence than the jointprobability of each event separately.

The Tokyo earthquake is also unusual for anotherreason. The typhoon winds blew the fires out ofcontrol. This dichotomous occurrence of torrentialrain and uncontrolled fire is quite common in Japan.Typhoons moving north-east in the Sea of Japanproduce very strong winds in the lee of mountains. InSeptember 1954 and again in October 1955, cyclone-generated winds spread fires that destroyed over 3300buildings in Hokkaido and 1100 buildings in Niigata,respectively. Nor is the phenomenon of cyclone–fireunique to Japan. The great hurricane of 1938 that dev-astated New England, in the United States, startedfires in New London, Connecticut, that raged for sixhours, and would have destroyed the city of 30 000people if it were not for the fact that the fires wereturned back on themselves as the winds reversed direc-tion with the passage of the cyclone.

Mechanics of cyclone generation(Anthes, 1982; Nalivkin, 1983; Gray, 1984; Gray et al.,1994; Chang et al., 2003; Simpson, 2003)

Tropical cyclones derive their energy from the evapo-ration of water over oceans, particularly the western

45Large-scale Storms as a Hazard

Storm surge Wind Rain

Flooding of coastallowland

Saline intrusion

Earthquakes

Coastal erosion

Damage to structures

Death fromdrowning

Disruption totransportation

Death, injuries

Urban fire

Loss of crops& livestock

Severe flooding

Contaminatedwater supplies

Spread of disease

Landslides

Death from drowning & disease

Fig. 3.1 Hazards related to occurrence of tropical cyclones.

parts, where surface temperatures are warmest. As thesun’s apparent motion reaches the summer equinox ineach hemisphere, surface ocean waters begin to warmover a period of two to three months to temperaturesin excess of 26°C. Tropical cyclones, thus, developfrom December to May in the southern hemisphereand from June to October in the northern hemisphere.At the same time, easterly trades on the equatorial sideof agglutinated mobile polar highs intensify and blowthese warm surface waters to the westward sides ofoceans. Hence, the warmest ocean waters accumulate

to a great depth in the Coral Sea off the east coast ofAustralia, in the Caribbean, and in the west PacificOcean, south-east of China. A deep layer of warmingreduces the upward mixing of cold sub-surface waterthat can truncate the cyclone before it fully develops.Warm water also develops in shallow seas becausethere is a smaller volume of water to heat up. Fig-ure 3.3 presents the origin of tropical cyclones. Themajority originate either at the western sides of oceans,or over shallow seas, where temperatures are enhancedby one of the above processes.

46 Climatic Hazards

FloridaMiami

Buffalo

TorontoMontreal

Galveston

Hawaii

Alaska

New YorkGreat Lakes

Winnipeg

North Amazon Basin

Hobart

Stanwell ParkSydney

Mackay

Darwin Bathurst Bay

Caribbean Sea

North Sea

Newfoundland

Sahara Desert

LakeChad

Aral Sea

Mediterranean Sea

Arabian Sea

Rajputana Desert

Gobi Desert

Bay ofBengal

Tokyo

Sea of Japan

Shanghai

ShantouGulf of Mexico

Fig. 3.2 Location map.

CaribbeanEast Pacific

Arabian Sea

Bay of Bengal

Madagascar

21

216 2

62

76

Western Australia

Southwest- Pacific

Eastern Asia

Southwest- Pacific

Fig. 3.3 Origin and number of tropical cyclones per year for the 25-year period 1952–1977 (after Gray, 1975).

At least eight conditions are required to develop atropical cyclone. First, efficient convective instability isrequired causing near-vertical uplift of air throughoutthe troposphere. Cyclonic depressions begin todevelop when ocean temperatures exceed 26–27°Cbecause the lower atmosphere becomes uniformlyunstable at these values. With further temperatureincreases, the magnitude of a tropical cyclone intensi-fies. Tropical cyclones tend not to form above 30°Cbecause smaller-scale turbulence begins to dominatewith the breakdown of near-vertical air movement.Once developed, cyclones rapidly decay below tem-peratures of 24°C. Cyclones rarely develop polewardof 20° latitude because ocean temperatures seldomreach these latter crucial values. These relationshipshave implications for global warming. While the area ofocean favorable for cyclone formation may increase,there is an upper temperature limit constraininggrowth. Tropical cyclones can also dramatically coolsea surface temperature by 3–4°C because of theenormous amount of heat involved in evaporatingocean water. Hurricane Gilbert cooled the ocean from31°C to 26°C off the coast of Mexico in 1988.

Second, there must be convergence of surface airinto areas of warm water. The formation of numerous

thunderstorms over warm waters provides this mecha-nism. Convergence causes upward movement of air andcreates instability in the upper atmosphere. However,in the case of tropical cyclones, this instability isenhanced by the release of latent heat of evaporation.Because of warm ocean temperatures and intense solarheating, evaporation at the sea surface is optimum.Each kilogram of water evaporated at 26°C requires2425 kilojoules of heat energy. As air rises, it cools, andif water vapor condenses, then this energy is released aslatent heat of evaporation. Vast amounts of heat energyare thus transferred to the upper atmosphere causingconvective instability that forces air upwards. Figure 3.4illustrates the type of temperature structure that candevelop in a tropical cyclone because of this process.The temperatures shown schematically represent themaximum possible temperature, in degrees Celsius, ifall evaporated moisture were condensed. The release oflatent heat of condensation in the zone of uplift canraise air temperatures up to 20°C above those at equiv-alent levels. Convective instability will persist as long asthis heating remains over a warm ocean.

Third, a trigger mechanism is required to beginrotation of surface air converging into the uplift zone.This rotation most likely occurs in an easterly wave


85

80

75

70

65

100

9590

85

80

75

70

6565

85

90

80 60 40 20 0 20 6040 80Distance from center of eye (km)

<70˚C >90˚C

200

400

600

800

1000

Pre

ssur

e (h

Pa)

Fig. 3.4 Potential temperature structure due to latent heat of evaporation in Hurricane Inez, 28 September 1966 (adapted from Hawkins & Imbembo, 1976).

developed within the trade winds. On surface charts,this wave is evident as an undulation in isobars parallelto degrees of latitude.

Fourth, while easterly waves begin rotation, Coriolisforce has to be sufficient to establish a vortex. Cyclonesrarely form within 5° of the equator. Nor cancyclones cross the equator, because cyclones rotate inopposite directions in each hemisphere. Since obser-vations began (in 1886 in the Atlantic and 1945 in thePacific), only two tropical cyclones – Typhoon Sarah in1956 and Vamei in 2001 – have traveled within 3° of theequator. Typhoon Vamei was unprecedented in thatits center came within 1.5° of the equator east ofSingapore and its zone of convection actually spilledacross the equator into the southern hemisphere. Thistropical cyclone’s development was forced by a strongmobile polar high that swept out of China into thetropics.

Fifth, cyclones cannot be sustained – unless the landis already flooded with water – at temperatures above24°C because they are dependent upon transferof heat, from the surface to the upper atmosphere,through the process of evaporation and release oflatent heat of condensation.

Sixth, tropical cyclones must form an ‘eye’ structure(Figure 3.5). If the convective zone increases to around10–100 km distance, then subsidence of air takes placein the center of the zone of convection, as well as tothe sides. Subsidence at the center of uplift abruptly

terminates convection, forming a ‘wall’. Upwardlyspiraling convection intensifies towards this wall, wheresurface winds become strongest. Once the convectivewall forms with subsidence in the core, the cyclonedevelops its characteristic ‘eye’ structure. Subsidencecauses stability in the eye, cloud evaporates, and calmwinds result. The ‘eye’ is also the area of lowest pressure.

Seventh, cyclones cannot develop if there aresubstantial winds in the upper part of the troposphere.If horizontal wind speed aloft increases by more than10 m s-1 over surface values, then the top of the con-vective column is displaced laterally, and the structureof the ‘eye’ cannot be maintained. Cyclones tend todevelop equatorward of, rather than under, the directinfluence of strong westerly winds.

Finally, the central pressure of a tropical cyclonemust be below 990 hPa. If pressures are higher thanthis, then uplift in the vortex is not sufficient tomaintain the ‘eye’ structure for any length of time.Some of the lowest pressures recorded for tropicalcyclones are presented in Table 3.1. The record lowpressure measured in a tropical cyclone in the Pacificwas 870 hPa in Typhoon Tip, north-east of the Philip-pines, in October 1979. In North America, HurricaneGilbert in September 1988 produced the lowestpressure of 880 hPa just before making landfall overJamaica. Note that cyclones with the lowest pressureare not always the most destructive in terms of the lossof human lives. This is mainly owing to the fact that notall cyclones occur over populated areas.

Magnitude and frequency(Dvorak, 1975; Simpson & Riehl, 1981; Gray, 1984;Donnelly et al., 2001; Ananthaswamy, 2003; AustralianBureau of Meteorology, 2003)

Tropical cyclones occur frequently. Figure 3.3 also sum-marizes the average number of cyclones per year world-wide in each of seven main cyclone zones. While thenumbers in any one zone appear low, when summedover several decades, cyclone occurrence becomes veryprobable for any locality in these zones. For example,consider Hurricane Hazel, which I witnessed as a childin 1954 when I was living at the western end of LakeOntario, Canada – a non-cyclone area 600 km from theAtlantic Ocean. Hurricane Hazel was a rogue event andis still considered the worst storm witnessed in southernOntario. This cyclone not only traveled inland, but it alsocrossed the Appalachian Mountains, which should havediminished its winds and rainfall. In fact, Hurricane

48 Climatic Hazards

150 100 50 0 50 100 150

1020

1000

980

960

940

80

60

40

20

0

Win

d sp

eed

(m s

-1)

Distance from center of cyclone (km)

WindPressureThe 'wall'

Pre

ssur

e (h

Pa)

Fig. 3.5 Wind and pressure cross sections through Hurricane Anita,2 September 1977 (based on Sheets, 1980).

Hazel’s path over the Great Lakes was not that rare anevent. Over a 60-year period, 14 cyclones have crossedthe Appalachians into the Great Lakes area. Not all wereof the magnitude of Hurricane Hazel, but all broughtheavy rain and winds. In Australia, at least one cycloneevery 20 years has managed to penetrate to AliceSprings in the center of the continent, and it is possiblefor west-coast cyclones to cross the whole continentbefore completely dissipating their winds and rain. Inthe New York region, major hurricanes, though rare,have struck twice in the last thousand years. Thehistorical 1821 hurricane produced winds reaching180–210 km hr-1. Worse was a storm sometime between1278 and 1438. Were it to recur, it would flood majorcities along the eastern seaboard. Even southernCalifornia is not immune. It lies close to Mexico, thespawning center of some of the most intense tropicalcyclones in the world. Fortunately, the majority trackwest into uninhabited parts of the Pacific Ocean.However, a tropical storm from this region moved intosouthern California in 1939, killing 45 people. Since1902, the remnants of forty hurricanes have droppedrain on the state.

The actual number of cyclones within the zones inFigure 3.3 is highly variable. In Australia between 1909and 1980, there have been on average ten cyclones peryear. The number has been as low as one per year andas high as 19. In the Caribbean region, the presence oftropical cyclones has fluctuated just as much between1910 and 2003 (Figure 2.12). In 1950, the aggregatednumber of days of hurricanes totaled 57, while nohurricanes were recorded at all in 1914. These patternsfor both areas are not random, but linked to the occurrence of El Niño–Southern Oscillation (ENSO)events. Recent research indicates that hurricanes inthe United States also follow decadal cycles. Periodswith numerous and intense hurricanes were 1865–1900,1930–1960, and 1992 onwards. Quiescent times were1850–1865, 1900–1930, and 1960–1992.

The zones depicted in Figure 3.3 are also restrictive.Tropical cyclones have appeared in the 1970s and1980s outside these limits, and outside the usuallydefined cyclone seasons for each hemisphere. Thecyclone season around Australia usually lasts from1 December to 1 May. Cyclone Namu devastated theSolomon Islands as late as 21 May 1986 and the 1987


Event Location Date Pressure (hPa)1 Typhoon Tip NE of Philippines October 1979 8702 Typhoon June Guam November 1975 8763 Typhoon Nora NE of Philippines October 1973 8774 Typhoon Ida NE of Philippines September 1958 8775 Typhoon Rita NE of Philippines October 1978 8786 unnamed Philippines August 1927 8877 Hurricane Gilbert Caribbean September 1988 9028 Typhoon Nancy NW Pacific September 1961 8889 Labour Day storm Florida September 1935 892

10 Typhoon Marge NE of Philippines August 1951 89511 Hurricane Allen Caribbean August 1980 89912 Hurricane Linda Baja Peninsula September 1997 90013 Hurricane Camille Gulf of Mexico August 1969 90514 Hurricane Mitch Caribbean October 1998 90515 Typhoon Babe NE of Philippines September 1977 90616 unnamed Philippines September 1905 90917 Cyclone Vance Western Australia March 1999 91018 Typhoon Viola NE of Philippines November 1978 91119 Cossack Cyclone Australia June 1881 914?20 Hurricane Janet Mexico September 1955 91421 Cyclone Mahina Australia March 1899 914?

Table 3.1 Lowest central pressures recorded in tropical cyclones.

Australian season ended with Cyclone Blanche on26 May. The occurrence of these storms has effectivelymeant that the cyclone season for Australia extends tothe end of May. Australian cyclones are also increasingin frequency. The worst season to date occurred in1989, which witnessed more than 14 cyclones in theCoral Sea. Certainly, the number of cyclones affectingcentral Pacific islands such as Fiji and Tonga hasincreased significantly in the 1980s – coincidently withmore frequent ENSO events.

Several scales measure the intensity of tropicalcyclones. These are presented in Table 3.2. The mostcomprehensive is the Dvorak Current Intensity, whichcategorizes the magnitude of both tropical storms andcyclones. The scale is related to central pressure anddiffers between the Atlantic and the north-west Pacificoceans. Scaled storms are more intense in the latter.The schema utilizes satellite photography and is calcu-lated using either cloud patterns discernible on visibleimages or the difference between the temperature ofthe warm eye and the surrounding cold cloud tops.Unfortunately, the Dvorak scale does not increment atregular intervals of pressure or wind speed. The mostfamiliar scale is the Saffir–Simpson scale developed inthe 1970s. The scale has five categories and applies onlyto hurricanes. The scale has the advantage of beingrelated to property damage, so it can be used in a pre-dictive capacity. A category 1 storm produces minimal

damage, while a category 5 storm equates with the mostdestructive hurricanes witnessed in the United States.These will deroof most buildings, flatten or removemost vegetation, and destroy windows and doors. Notethat Hurricane Andrew – which struck the FloridaPeninsula in 1992 and is the second costliest hurricanein United States history – has recently been reclassifiedas a category 5 event. The Saffir–Simpson scale can alsobe related to storm surge potential depending upon thewidth of the continental shelf and how the storm tracks.For example, a category 5 hurricane can produce stormsurge in excess of 5.5 m. This aspect is very convenientfor laypeople in flood-prone areas. A variation of thisscale is used in the southern hemisphere, mainlyAustralia. The Australian scale includes higher windspeeds than the Saffir–Simpson scale and is related tomaximum wind gust. Unfortunately, wind gauges in thetropics of Australia are sparsely distributed so that it isdifficult to verify the scale. The accuracy and viability ofwind gauges is also questionable at wind speeds towardsthe upper end of the scale.

World cyclone disasters(Cornell, 1976; Anthes, 1982; Whipple, 1982; Nalivkin,1983; Holthouse, 1986)

Tropical cyclones are responsible for some of thelargest natural disasters on record. In the UnitedStates, the deadliest hurricane struck Galveston Island,

50 Climatic Hazards

Mean sea Mean sea Dvorak Australia-level pressure level pressure Current Mean wind Southern

(hPa) (hPa) Intensity speed Saffir-Simpson Wind speed Storm surge Hemisphere Strongest gust Atlantic NW Pacific Scale (km hr-1) Scale (km hr-1) height (m) Scale (km hr-1)

1 46.31.5 46.3

1009 1000 2 55.61005 997 2.5 64.81000 991 3 83.3 1 125

994 984 3.5 101.9987 976 4 111.1 1 119–153 1.2–1.5 2 125–170979 966 4.5 142.6970 954 5 166.7 2 154–177 1.8–2.4 3 170–225960 941 5.5 188.9 3 178–209 2.7–3.6948 927 6 213.0935 914 6.5 235.2 4 210–249 3.9–5.5 4 225–280921 898 7 259.3 5 >249 >5.5906 879 7.5 287.0 5 >280890 858 8 314.8

Table 3.2 Scales for measuring cyclone intensity (Dvorak, 1975; Simpson and Riehl, 1981; Australian Bureau of Meteorology, 2003).

Texas, on 8 September 1900, killing 6000–8000 people.A rapidly rising storm surge trapped inhabitants, whothen either drowned as waves swept through the city orwere killed by the 170 km hr-1 winds. Apathy and igno-rance exacerbated the death toll. There have been atleast four cyclones in recorded history that, as singleevents, killed over 300 000 people – all but oneoccurred over the Indian subcontinent. The mostdestructive cyclone in the twentieth century was theone that struck Bangladesh – formerly East Pakistan –in November 1970, killing over 500 000 people. It ismemorable for two reasons. Firstly, most people werekilled by the storm surge rather than strong winds orfloodwaters. Secondly, although viewed by inhabitantsas a rare event, the disaster was repeated in May 1985and on 29 April 1991, resulting in another 100 000 and140 000 deaths, respectively. The press also graphicallydocumented the 1970 disaster. The ensuing ineffectualrescue and reconstruction effort by the Pakistangovernment led to political upheaval and indepen-dence for Bangladesh. Ironically, while Bangladeshobtained its independence because of a cyclone,ultimately it became the poorest country in the world,mainly because it never recovered economically fromthe disaster.

Historically, cyclones account for the greatestnumber of deaths resulting from any short-termnatural hazard. Flooding over swathes of 400–500 kmalong the cyclone track causes most of this death toll.The most horrific cyclone-induced flooding occurredin China along the Chang River between 1851 and1866. During this period, 40–50 million peopledrowned on the Chang River floodplain because oftropical cyclone-related rainfall. Another 1.5 milliondied along the Chang River in 1887 and, in 1931,3–4 million Chinese drowned on the Hwang Ho Riverfloodplain because of cyclone flooding.

Tropical cyclones have even changed the course ofhistory in the Far East. In 1281 AD, Kublai Khaninvaded Japan, and after seven weeks of fierce fightingwas on the verge of breaking through Samurai defenses– until a typhoon swept through the battlefield. Thestorm destroyed most of the 1000 invading ships andtrapped 100 000 attacking soldiers on the beaches,where they either drowned in the storm surge or wereslaughtered by the Japanese. The Japanese sub-sequently believed their homeland was invincible,protected by the Kamikaze (or Divine Wind) thathad saved them from this invasion.

Tropical cyclones can even affect non-tropical areas.As mentioned above, Hurricane Hazel swept throughsouthern Ontario, Canada, in October 1954. Thisstorm occurred during a period of remarkable super-storms that afflicted Japan and North Americabetween 1953 and 1957. Hazel drifted northwardalong the eastern coast of the United States, where itkilled 95 people, raised a storm surge of 5 m, andproduced the strongest winds ever recorded at Wash-ington and New York. It then turned inland over theAppalachians and headed across the Great Lakes toToronto, 600 km from the nearest ocean. Here 78people lost their lives in the worst storm in 200 years.

In Australia, on 4–5 March 1899, Cyclone Mahina(or the Bathurst Bay Cyclone) crossed the northernGreat Barrier Reef and killed 407 people. The stormsurge reached 4 m, which is high for Australia. Smallcyclones, with eyes less than 20 km in diameter, arecalled ‘howling terrors’ or ‘kooinar’ by the Aboriginesand appear the most destructive. For instance, theQueensland town of Mackay, with 1000 inhabitants, wasdestroyed in 1918 by one of these small cyclones.Cyclone Ada in 1970 was so small that it slipped unno-ticed through the meteorological observation network,and then surprised and wrecked holiday resorts alongthe Queensland coast. The Yongala Cyclone of 1911,when it reached landfall south of Cape Bowling Green,cleanly mowed down mature trees 2 m above theground in a 30-kilometre swathe. In recent times,Cyclone Tracy has been the most notable cyclone.Tracy, with an eye diameter of 12 km, obliteratedDarwin on Christmas Day 1974 (Figure 3.6), killingover 60 people and reaching wind speeds over217 km hr-1. In world terms, its wind speeds were smalland the loss of life meager. There was also no surgedamage because the tidal range in Darwin is about 7 mand the cyclone occurred at low tide, with a storm surgeof 4 m. Tracy is one of the best-documented Australiancyclones, with its graphic evidence of wind damageforming a benchmark for engineering research intothe effects of strong winds. It also resulted in one ofthe largest non-military air evacuations in history, as themajority of the population of 25 000 was airlifted southto gain shelter and avoid the outbreak of disease.

Hurricane Andrew, 24 August 1992(Gore, 1993; Santana, 2002)

In recent times, two events are noteworthy: HurricaneAndrew in the United States in 1992 and Hurricane


Mitch in Central America in 1998. Hurricane Andrewstruck after a 27-year lull in intense cyclones in theUnited States. The storm began on 13 August 1992 asa region of thunderstorms off western Africa andrapidly developed into an easterly wave. It became aweak tropical cyclone on 17 August; but by 23 Augustit had developed central winds of 240 km hr-1 and acentral pressure of 922 hPa. Andrew made landfallin Dade County, 32 km south of Miami, just aftermidnight on Monday 24 August (Figure 3.7). Whileup to one million people had evacuated to shelters inthe Miami area, most of the 370 000 residents of DadeCounty were still in their homes under the assumptionthat they were not in the path of the hurricane.Warnings were issued to them on television onlyminutes before winds began gusting to 280 km hr-1 –simultaneously with the arrival of a storm surge of4.9 m. This made it a category 5 hurricane. Over25 500 homes were destroyed and another 101 000

severely damaged. Many of the homes were of sub-standard construction for a hurricane-prone area.For example, plywood had been stapled to supportsinstead of being secured using screws or bolts. Muchof the damage was caused by vortices that developedin the main wind stream, then tightened, and woundup to speeds of 120 km hr-1 in 10–20 seconds. Wherethe wind in these mini-cyclones flowed with the mainair stream, wind velocities exceeded 200 km hr-1 – withbuildings torn apart, seemingly sporadically, anddebris flung at tornadic speeds (Figure 3.8). Theofficial death toll was only 55, although there isevidence that it may have been higher because theregion contained a high number of illegal immigrants.Over $US30 billion worth of property was destroyed.The hurricane then swept into the Gulf of Mexico and,two days later, came ashore in Louisiana, killing15 people and causing another $US2 billion ofdestruction. This was the biggest insurance disaster inthe country. What followed became one of the biggestfederal government ‘non-responses’ in history.Looting was rampant and temperatures and humiditysoared; but for a week state and federal authoritiesignored the extent of the disaster. Over 250 000people were homeless, there was no food or safedrinking water, and sanitation was failing due to foodrotting in disconnected freezers and a lack of service-able toilets. Usually, when response to a disasterrequires resources beyond the capability of a stategovernment, assistance is provided by the FederalEmergency Management Agency (FEMA). FEMAwas established in 1979 as an independent federalagency reporting directly to the President. It has aclear mission to reduce loss of life and property andprotect critical infrastructure from all types of haz-ards through a comprehensive, risk-based, emergencymanagement program of mitigation, preparedness,response, and recovery. However, it had no capabilityto provide emergency accommodation, food, andmedical assistance on the scale needed in southernFlorida. Even private relief agencies such as the RedCross did not know how to distribute donations theywere receiving. Finally, Dade County’s EmergencyDirector, Kate Hale, made a passionate public plea:‘Where the hell is the cavalry on this one?’ Only thenwere military personnel increased to 16 000 to assistwith clean-up and relief. So poor was FEMA’sresponse that Congress considered eliminating theagency.

52 Climatic Hazards

Fig. 3.6 Damage done to a typical suburb in Darwin by Cyclone Tracy,December 1974 (photograph courtesy John Fairfax and SonsLimited, Sydney, FFX Ref: Stevens 750106/27 #4).

Hurricane Mitch, October/November 1998

(Lott et al., 1999)

Hurricane Mitch was the second most destructivehurricane ever recorded in the Atlantic Ocean. Itbegan as a tropical depression on 21 October 1998 andwreaked havoc across the Caribbean for two weeksbefore dissipating in the Atlantic on 4 November.During this time, it reached wind speeds of 290 km hr-1,

making it a rare Saffir–Simpson category 5 event.Mitch made landfall on 28 October over Honduras,slowed its westward progress and dumped, during thesubsequent week, up to 1.9 m of torrential rain onthe mountains of Central America. This led tocatastrophic flooding and mudslides in Honduras,Nicaragua, Guatemala, Belize, and El Salvador.Hundreds of thousands of homes were destroyed andup to 18 000 people died, making the storm the worstin the Americas since 1780. Over three million peoplewere made homeless. In Nicaragua, damage was worsethan that resulting from the 1972 Managua earth-quake. So calamitous was the disaster that appeals forinternational aid were made by all countries affectedby the hurricane. In Honduras, at least 30 per centof the population was homeless, 60 per cent had noaccess to safe water, 66 per cent of the crops weredestroyed, and 100 per cent of the road networkrendered impassable. The World Food Programmeestimated that 12 per cent of the population requiredlong-term food supplies, while UNICEF estimatedthat recovery would take 40 years. The disasterepitomizes the effects of severe natural hazards upondeveloping countries characterized by poverty, civilwar, and economies dominated by a single commodityprimary industry.


Miami

22 Aug23 Aug24 Aug

25 Aug

Fig. 3.7 NOAA composite satellite image of Hurricane Andrew’s path, 22–25 August 1992 (NOAA, 2002).

Fig. 3.8 Sheet of plywood driven through the trunk of a Royal Palmby extreme winds during Hurricane Andrew in South Florida(NOAA, 2003).

Impact and response(Burton et al., 1978; Simpson & Riehl, 1981; The DisasterCenter, 1998)

The human impact of cyclones can be assessed interms of deaths, financial damage, reliance uponinternational aid, and the prospects of further impov-erishment. The response to cyclones depends mainlyon these four factors. Deaths due to cyclones in theUnited States decreased dramatically in the twentiethcentury to fewer than a 100 people per year, mainlybecause of a policy of total evacuation together withearlier warning systems. Costs, on the other hand, haveincreased considerably – as illustrated by HurricaneAndrew – because of the escalating development incyclone-prone areas after the Second World War. Ofcourse, the damage bill from a tropical cyclone alsodepends upon the type of building construction.Cyclone Tracy – the most destructive cyclone experi-enced in Australia – destroyed only 10 per cent of thevalue of infrastructure in Darwin. While most houseswere damaged or destroyed, larger commercial andpublic buildings with a high capital investment werenot badly damaged. Other non-residential buildingswere mildly affected because they had to comply withstrict building codes formulated to minimize the effectof cyclone winds.

The impact of cyclones goes far beyond just deathsand building damage. In developing countries,destruction of infrastructure and primary agriculturecan lead to a decrease in exports and gross nationalproduct, while increasing the likelihood of forfeiture ofinternational loan repayments. Figure 3.1 indicatesthat cyclones can cause increased soil salinitythrough storm-surge inundation, an effect that canhave long-term economic consequences for agricul-tural production – with such consequences virtuallyimmeasurable in monetary terms. Contamination ofwater supplies and destruction of crops can also lead todisease and starvation. In fact, some of the large deathtolls caused by cyclones on the Indian subcontinent inthe nineteenth century were due to starvation after-wards. In certain cases, the destruction of crops canaffect international commodity markets. For example,Hurricane Flora’s passage over Cuba in 1963 wiped outthe sugar cane crop and sent world sugar pricessoaring. Of all hazard events, the effects of tropicalcyclones differ substantially from each other depend-ing upon the development of a country’s economy,

transport network and communication infrastructure.This can be illustrated most effectively by comparingthe response and impact of tropical cyclones forthree countries: Australia, the United States, andBangladesh.

Australia

(Australian Bureau of Meteorology, 1977; Western & Milne,1979)

One of the effects of tropical cyclones not covered inFigure 3.1 is long-term human suffering caused bydislocation. Studies of hazards where people aredispossessed indicate that the incidence of psycho-somatic illness (illness, such as by anxiety, induced bychanges in mental well-being) increases afterwards.More importantly, people who have lost a relativerecover more completely than people who have lostthe family home. A death can be mourned during afuneral occurring within several days, but a familyhome has to be rebuilt over months or years. Australiahas experienced this phenomenon on a large scale. Innorthern coastal Australia, it is accepted that cycloneswill inevitably occur. Evacuation is normally notencouraged except for the most hazardous areas nextto the ocean. Strict building codes have been estab-lished to minimize damage and loss of life. Such codesare policed and adhered to. A network of remote-controlled, satellite and staffed stations has beenestablished to detect and track cyclones. Theseobservations are continually reported over radio andtelevision if a cyclone is imminent. Some criticism hasbeen expressed regarding the Australian Bureau ofMeteorology’s inability to predict the landfall of allcyclones, and there is some fear that the staffedstations are being run down as a cost saving measure.However, it should be realized that tropical cyclonesin Australia, as elsewhere, often travel unpredictably.Residents have been instructed on how to prepare andweather a cyclone. For example, most people uponhearing about the forecast arrival of a cyclone willremove loose objects in the house and yard that canbecome missiles in strong winds, will tape windows,and will store at least three days’ supply of water andfood.

Cyclone Tracy, which struck Darwin on 25 December1974, demonstrated that apathy can negate all thesepreventative procedures. The cyclone also led to large-scale psychological damage that cannot be measured ineconomic terms. Despite three days of warnings, and

54 Climatic Hazards

because of the approaching Christmas holiday, only8 per cent of the population was aware of the cyclone.About 30 per cent of residents took no precautions toprepare for the storm’s arrival; fewer than 20 per centbelieved that people took the cyclone warnings seri-ously; and 80 per cent of residents later complained thatthey had not been given enough warning. Fortunately,Australia had established a disaster coordination centerthat, depending on the severity of the disaster, couldrespond in varying degrees to any event. The directorof this center, once requested to take charge, couldadopt dictatorial powers and even override federalgovernment or judicial decisions. This organizationresponded immediately to Cyclone Tracy. Within24 hours, communications had been re-established withthe city even though it was the height of the Christmasvacation season. Major-General Stretton, the head of theNational Disasters Organisation, reached Darwin within24 hours on a military aircraft with emergency suppliesloaded at Sydney and Mt Isa. Stretton took total controlof Darwin and made the decision to evacuate non-essential residents because of the serious threat tohealth. The cyclone destroyed 37 per cent of homes andseverely damaged most of the remainder (Figure 3.9).There was no shelter, water, power, or sanitation. Withina few days, most of the food in deep-freezers began todecay, so that a stench of rotten meat hung over the city.Stray cats and dogs were shot and special army teamswere organized to clean up rotten food. In addition, thedecision was made to facilitate evacuation to permitorderly and manageable rebuilding. Commercial planeswere commandeered for the evacuation and detailstaken of all vehicles leaving Darwin by road. Volunteeremergency relief organizations ensured that all evacueeshad food, adequate transport, accommodation, andmoney. The response in the first week was efficientlyorganized and carried out.

In the medium term, the evacuation caused moreproblems than it solved. A major problem was the factthat, while residents were efficiently and safely evacu-ated, no register of their whereabouts was established.Over 25 000 people were flown out, mainly to southerncities of their choice 3000–4000 km away. Communi-cation with, and among, evacuees became impossible.Hence, there was no way to contact people personallyafter the cyclone to assist them with adjustment tonew circumstances. Residents who stayed in Darwinsuffered the least, while those who did not return inthe medium term suffered the most. Non-returned

evacuees reported greater negative changes in lifestyle.These people also tended to be disadvantaged by thestorm. They were more likely not to have insuredhouses for cyclone damage and more likely to havesuffered total property loss. They ended up with lessincome after the event and suffered the greatest stress.While the stress may have existed beforehand orbeen exacerbated by the disaster, the differencebetween this group and the others was striking. Bothpsychosomatic illness and family problems increasedsignificantly for non-returnees. The evacuation rein-forced and increased the levels of stress and anxietyresidents were experiencing because of their con-frontation with Cyclone Tracy. Some of the stress was aconsequence of the dictatorial powers assumed after


Fig. 3.9 Effects of wind damage caused by Cyclone Tracy strikingDarwin, Australia, 25 December 1974.A) High winds in excess of 217 km hr–1 tended to detach

outside walls from the floor, leaving only the innerframework surrounding the bathroom.

Fig. 3.9 B) In extreme cases, the high winds progressively peeled offthe complete building, leaving only the iron floor beamsand support pillars. (Photographs courtesy of Assoc. Prof.Geoff Boughton, TimberED Services Pty Ltd., Glengarry,Western Australia.)

Tracy by a few. Unintentionally, a victim–helper rela-tionship became established to the point that littleencouragement was given to victims to play an activerole in their own recovery. Disaster victims in Australiaare resourceful. To treat them otherwise impedes thesocial recovery from disaster.

The long-term response to Cyclone Tracy was alsonegative. The decision to rebuild and the method ofallowing rebuilding did not take into considerationthe long-term effects on human mental and physicalhealth. Large sums of money were poured intorebuilding lavish government offices and facilities. Therebuilding effort was taken over by outsiders, andthe growth of Darwin beyond 1974 population levelswas encouraged. The rebuilding became a ‘sacredcow’, with the visible commitment to the rebuilding bythe federal government leading to unlimited growth.Locals who still lived in Darwin objected to redevelop-ment plans, stating that they had no say in the range ofplans put forward. Residents evacuated to the southhad no say in the redevelopment at all. Only menwere allowed to return to the city to rebuild. The men,isolated from their families for up to one year, groupedtogether socially. When their families finally returnedhome, many of these men could not break backinto family groupings, most of which had completelychanged (for example, children had physically andmentally grown in the absence of a male role figure).Even though within the first year, returnees sufferedless physically and emotionally than non-returnees,returnees in the long term suffered the same conse-quences as non-returnees. By 1985, over 50 per cent ofthe families who experienced Tracy had broken up.The personal problems resulting from Cyclone Tracyare now being imparted to the second generation, thechildren, to be passed on in future years to anothergeneration unless rectified. Today, Darwin is a trans-formed city. It is estimated that 80 per cent of thepeople now living there, never lived through CycloneTracy.

United States

(American Meteorological Society, 1986; Carter, 1987)

The United States government believes that earlywarning and monitoring is the best method toreduce loss of life due to tropical cyclones. Weathersatellites were originally deployed to predict and trackcyclones, thus giving enough warning to permit the

total evacuation of the population, either horizontallyor vertically. Horizontal evacuation involves movinglarge numbers of people inland from flood-pronecoastal areas. In August 1985, millions of people wereevacuated, some more than once, as Hurricane Dannywandered unpredictably around the Florida coast andthen headed into the Gulf States. This type of evacu-ation can be especially difficult where traffic tendsto jam on escape routes. The alternative is verticalevacuation into high-rise, reinforced buildings. Thisapproach is taken in Miami Beach, where limitedcauseways to the mainland restrict the number ofpeople who can flee, where the population is elderlyand hence physically immovable, and where safestructures can easily accommodate the majority of thepopulation. There is a concerted effort to build coastalprotection works and even try to modify the intensityof hurricanes through cloud seeding. A high priority isalso put on research activities that model, predict, orquantify hurricanes and related hazards. Some eastcoast states, such as North Carolina and Maine, havedevelopment plans that exclude development fromlow-lying coastal zones subject to storm surge and waveerosion due to hurricanes. These setback lines areequivalent to 30–100 years of average erosion.

The above measures have effectively reduced theloss of life due to tropical cyclones in the United States.However, there are serious flaws with the evacuationprocedures. Most evacuations actually take up to30 hours to carry out, whereas the general populacebelieves that it would require less than one day to fleeinland to safety. The evacuation system also reliesheavily upon warnings of hurricane movement put outby the National Hurricane Center, which attempts togive 12 hours’ warning of hurricane movements. Infact, hurricanes move unpredictably and can shiftdramatically within six hours. At one point in time,Hurricane Hazel moved at speeds of 50 km hr-1. In theUnited States, the time required for evacuationcurrently falls well short of the monitoring and forecastlead times for hurricanes. There is also the problemthat 80 per cent of the 50 million people living today inhurricane-affected areas have never experienced anevacuation. These same people, together with develop-ers, have become complacent about the hurricanethreat because the frequency and magnitude ofhurricanes has decreased in recent decades. The1970s, in fact, experienced the lowest frequency ofhurricanes ever recorded. At present along the whole

56 Climatic Hazards

eastern seaboard and much of the Gulf Coast of theUnited States, record construction has taken place withdwellings being built either at the back of the beach, orjust landward of ephemeral dunes. Studies have shownthat there is a general lack of awareness of cyclonedanger and evacuation procedures. This is despitewide publicity, as the hurricane season approaches,about the dangers of cyclones. Publicity includes tele-vision specials, distribution of free literature, andwarnings on supermarket bags and parking tickets.These warnings are meeting with resistance because,in the absence of any storm, municipalities view thepublicity as bad for the tourist trade and damaging toproperty investment.

The realities of evacuation, and the poor public per-ception of the ferocity of hurricanes in the UnitedStates, implies that – as far as the cyclone hazard isconcerned – many people are sitting on a time bomb.When a large tropical cyclone strikes the United Stateseast coast, evacuation procedures may fail and theresulting loss of life may reach unprecedented levels.

Hurricane Andrew also exposed flaws in the federalgovernment’s capability to respond to a large disaster.When the response to a disaster requires resourcesbeyond the capability of a state government, theFederal Emergency Management Agency (FEMA)can assist. FEMA failed its mandate during HurricaneAndrew. While FEMA has been overhauled since1992, the prospect still looms that two large hurricanesmay occur so close together that they will overwhelmthe capacity of any of the United States’ resources torespond to the disaster in the short term.

Bangladesh (East Pakistan)

(Burton et al., 1978; Blong & Johnson, 1986; Milne, 1986)

Third World countries suffer exorbitantly from tropicalcyclones because of depressed economic conditionsand the lack of an effective domestic response capabil-ity. In 2000, cyclones Eline and Hudah devastatedMozambique, which has one of the lowest living stan-dards in the world. In 2003, Cyclone Zoe swept theouter islands of Tikopia, Anuta, and Fatutaka in theSolomon Islands. As the government was bankrupt andwracked by lawlessness, the inhabitants were forced tofend for themselves. But nowhere has the plight ofimpoverished countries been exemplified more than inBangladesh. The Bay of Bengal has a history ofdestructive storm surges. The 300 000 death toll from

Calcutta’s October 1737 cyclone was caused by a surge,13 m high, racing up the Hooghly River. In 1876, astorm surge near the mouth of the Meghna River killed100 000 people, while another 100 000 died in the sub-sequent famine. Small cyclone-related surges drowned3000 people on the Ganges River in October 1960 andanother 15 000 people in May 1965. The Chars, thedeltaic islands in the mouth of the Ganges, were firstviewed as potentially arable in the nineteenth century(Figure 3.10). Since 1959, 5500 km of dykes andembankments have been built to protect low-lying salt-water marshland from flooding, with the World Bankfinancing the scheme. The reclaimed land, however,was not protected against storm surge. Settlement,mainly by illiterate farmers, was permitted and encour-aged to relieve overcrowding in the country. By theend of the twentieth century, over four million peoplelived in areas at high risk from storm surge. Many itin-erant workers searching for temporary employmentflock into the Chars at the time of harvesting and fishprocessing, swelling the population by 30 per cent.

The 13 November 1970 cyclone was detected bysatellite three days in advance, but a warning was notissued by the Pakistani Meteorological Bureau untilthe evening that the cyclone struck. Even then, themessage was given to the sole radio station after it hadshut down for the day at 11:00 pm. If the warninghad been broadcast, the majority of the population,being asleep, would not have heard it. Much of theChars’ reclaimed land had inadequate transport linksacross channels to higher ground, such that effectiveevacuation was impossible. Recent migrants had noknowledge of the storm surge hazard and were notprepared, or organized, for evacuation. In the middleof the night a storm surge, 15 m high, struck thesouthern Chars, obliterating 25 islands, swamping afurther 2000 and drowning approximately 500 000people. Over 400 000 hectares of rice paddies wereinundated with salt water and 1 000 000 head of live-stock killed. Over 50 million people were affected bythe storm surge, flooding, and winds.

The pressure for land in Bangladesh is so great thatthe deltaic land was resettled rapidly after the 1970event. Nothing had really changed by 1985 and, in Mayof that year, a similar cyclone caused the same sort ofdestruction. Again, while there was adequate satellitewarning of the approaching cyclone, evacuation wasimpossible. Many illiterate, itinerant workers knewnothing of the hazard, the method of escape, or the


event. In the 1985 event, 100 000 people lost theirlives.

The response in Bangladesh to tropical cyclonesappears to be one of ignorance and acceptance ofthe disaster as the ‘will of God’. Cyclones cannot beavoided and one must simply wait out the storm andsee what happens. With a bit of luck a family that hasexperienced one cyclone storm surge in its lifetime willnot experience another. In 1970, only 5 per cent of thepeople who survived had evacuated to specific shelterswhile 38 per cent had climbed trees to survive. Manyview evacuation as unwise because homes could belooted. Even if evacuation had been possible, religioustaboos in 1970 would have prevented Moslem womenfrom leaving their houses because the cyclone struckduring a month when women were forbidden by estab-lished religious convention from going outside. Since1970, 1841 cyclone shelters have been built by CaritasBangladesh and other non-government organizations.These shelters are built of reinforced concrete andraised on piers at least 4.6 m above historical stormsurge levels. They can double as schools, family welfarecenters, mosques, markets, and food stores duringnormal times. Each shelter is designed to give a person0.5 m2 of floor space, which is sufficient to allow themto sit comfortably on the floor during storms. When

filled to capacity with people standing shoulder-to-shoulder, this personal space decreases to 0.14 m2.Toilet facilities and potable water supplies poseproblems under these latter circumstances. A sheltercan hold 1000–1500 people and some livestock. Up to50 per cent of the population can now be accommo-dated in cyclone shelters. Their effectiveness wasproven on 28 May 1997 when tropical cyclone Helenhit the Chittagong region. A million people took refugein 700 cyclone shelters for up to 20 hours. Winds of upto 230 km hr-1 damaged 600 000 houses in south-eastBangladesh. Instead of a death toll in the tens ofthousands, only 106 people were killed.

EXTRA-TROPICAL CYCLONES

Extra-tropical cyclones or depressions are low-pressure cells that develop along the polar front. Theyalso can develop over warm bodies of water outside thetropics, usually off the east coast of Australia, Japan,and the United States. This section will describe theformation of low-pressure storms due to upwardforcing of warm air along the polar front, and theformation of east-coast lows that develop or intensifyover bodies of warm water along limited sections of theworld’s coastline.

58 Climatic Hazards

0 100 200km

Brahmaputra R.

India

Bay of Bengal

Ganges R.

Bangladesh

Kolkata

Dacca

Cyclone track

Dyked or reclaimed landSubject to storm surge flooding

Fig. 3.10 Path of 13 November 1970 cyclone in the Bay of Bengal and location of embankments protecting low-lying islands in Bangladesh.

Polar-front lows

Formation (Linacre & Hobbs, 1977; Whipple, 1982)

Low-pressure cells developing along the polar fronthave cold cores and an effective diameter that is muchlarger than that of a tropical cyclone (2000 km versus400 km, respectively). Their intensity depends uponthe difference in temperature between colliding coldand warm air masses. They always travel in an easterlydirection, entrapped within the westerlies. Theirlocation depends upon the location of Rossby waves inthe polar front around 40°N and S of the equator. Inthe southern hemisphere, this occurs south of land-masses, but in the northern hemisphere, it coincideswith heavily populated areas.

The theory for the formation of polar-front lows wasformulated by Bjerknes and his colleagues at Bergen,Norway, in the 1920s. Outbreaks of cold polar air flowrapidly towards the equator and clash with muchwarmer air masses at mid-latitudes. Coriolis forcedictates that this polar air has an easterly componentwhile the movement of warmer air on the equatorialside of the cold front has a westerly component (Figure3.11). At the same time, the warm air will be forced

upwards by the advancing cold air. Winds on each sideof the cold front blow in opposite directions and spiralupwards in a counterclockwise direction because ofCoriolis force. This rotation establishes a wave orindentation along the front. Warm air begins toadvance eastward and poleward into the cold air masswhile cold air begins to encircle the warm air from thewest. A V-shaped frontal system develops with warmand cold fronts at the leading and trailing edges,respectively. The lifting of warm, moist air generateslow pressure and convective instability as latent heat ofevaporation is released in the upper atmosphere. Thisuplift enhances the rotational circulation, increasingwind flow and producing precipitation over a widearea. The cold front at the back of the system pushesup warm air rapidly leading to thunderstorm form-ation. If the temperature difference at this locationbetween the two air masses is great enough, tornadoescan be generated. The cold front, being backed bydenser air, eventually advances faster than the warmfront and begins to lift off the ground the warm airwithin the ‘V’, occluding the fronts. In the final stage ofa depression, the polar front is re-established at theground with decreasing rotational air movementstranded aloft. Without continued convection, cloud


Warm

Cold

Air movementFrontsWarm

Cold

Stationary

Occluded

Northern hemisphere Southern hemisphere1 2

3 4

5 6

1 2

3 4

5 6

NLow Low

Low Low

LowLow

Low

Low

Fig. 3.11 Stages in the development of an extra-tropical depression as viewed in each hemisphere 1) and 2) initial development of wave form, 3) and 4) mature front development and full extent of low, 5) and 6) occlusion.

formation and precipitation slowly decrease and thestorm cell disappears. Under some circumstances,the advancing cold front may entrap another lobe ofwarm air, generating a secondary low to the south-eastof the original one that is now dissipating. It is evenpossible for the rotating air aloft to circle back to theoriginal site of cyclogenesis, and trigger the formationof a new low-pressure wave along the polar front.These extra-tropical lows or depressions, because oftheir method of formation, have the potential toproduce the same wind strengths and precipitationamounts as tropical cyclones. However, because theyare larger and more frequent, they pose a greaterhazard. Such extra-tropical storms also can generateocean waves and storm surges similar in magnitude tothose produced by tropical cyclones.

Historical events

(Wiegel, 1964; Lamb, 1982; Whipple, 1982; Milne,1986; Bresch et al., 2000)

Mid-latitude depressions have also been as devastatingas tropical cyclones. Their effects are dramatic in twoareas, namely along the eastern seaboard of the UnitedStates and in the North Sea. Historical accounts forthese areas document the strongest storms recorded.Figure 3.12 plots the number of recorded cyclonicdepressions in northern Europe that have producedsevere flooding over the past 2000 years. The MiddleAges was an unfortunate period of cataclysmictempests that caused great loss of life and immenseerosion. Four storms along the Dutch and Germancoasts in the thirteenth century killed at least 100 000people each. The worst of these is estimated to havekilled 300 000 people. North Sea storms on 11 Nov-

ember 1099, 18 November 1421 and in 1446also killed 100 000 people each in England and theNetherlands. By far the worst storm was the All SaintsDay Flood of 1–6 November 1570. An estimated 400 000 people were killed throughout westernEurope. These death tolls rank with those produced inrecent times in Bangladesh (East Pakistan) by stormsurges from tropical cyclones. An English Channelstorm on 26–27 November 1703, with an estimatedcentral pressure of 950 hPa and wind speeds of170 km hr-1, sank virtually all ships in the Channel withthe loss of 8000–10 000 lives (Figure 3.13). Otherstorms with similar death tolls occurred in 1634, 1671,1682, 1686, 1694 and 1717 at the height of the LittleIce Age. Much of the modern-day coastline of northernEurope owes its origin to this period of storms. Erosionin the North Sea was exacerbated by 30 notable,destructive storm surges between 1000 and 1700 AD.North Sea storms reduced the island of Heligoland,50 km into the German Bight, from a length of 60 km around the year 800 to 25 km by 1300 and to 1.5 km by the twentieth century. The Lucia Flood stormof 14 December 1287 in northern Europe was ofimmense proportions. Before that time, a continuousbarrier island bordered the Netherlands coast. Thestorm breached this coastal barrier and formedthe Gulf of Zuider Zee (Figure 3.13). It is only recentlythat the inlets at these breaches have been artificiallydammed. Smaller embayments further east at Dollartand Jade Bay were also initiated and the Eider Estuarywas widened to its present funnel shape. The samestorm in northern France eroded several kilometres ofmarsh, making one headland a distant offshore island.In the Great Drowning Disaster storm of 16 January

60 Climatic Hazards

0 500 1000 1500 2000Channel coastNorth Sea coast

15

10

5

0

Num

ber o

f sto

rms

Year

Fig. 3.12 Incidence of severe North Sea and English Channel storms per century for the last 2000 years (Lamb, ©1982; with permission fromMethuen, London).

1362, parts of the North Sea coast of Schleswig–Holstein eroded 15 km landward, turning a low-lyingestuarine coastline into one bordered by irregularbarrier islands separated from the mainland by lagoons5–10 km wide. Over 60 parishes, accounting for halfthe agricultural income in this area, were destroyed.Storms in 1634 severely eroded both the Danish andGerman coasts and, in 1825, a North Sea storm sepa-rated part of the northern tip of Denmark from themainland, leaving behind several islands. These stormsalso directed the course of history. In 1588, the SpanishArmada of 130 ships, after being forced into the NorthSea by marauding attacks from the outnumberedBritish fleet, was all but rendered ineffective by a five-day storm off the east coast of Scotland (Figure 3.14).What remained of the Armada was finished off bymore of these storms as the fleet tried to escape westaround Ireland. The summer of 1588 was character-ized by an exceptional number of cyclonic depressions,corresponding to a polar jet stream diverted muchfurther south and wind speeds much stronger thancould be expected at present.

None of the extra-tropical storms mentioned abovehave been matched in the twentieth century for loss of

life. However, there have been big storms. Mostnoteworthy was the North Sea storm of 1 February1953. This storm developed as a low depression northof Scotland and then proceeded to sweep directlysouth-east across the North Sea (Figure 3.13). Thelowest pressure recorded was 966 hPa. While this isnot as low as pressures produced by tropical cyclones(see Table 3.1), it represents an extreme low pressurefor a cold-core depression. Wind speeds at Aberdeen,200 km from the center of the storm, exceeded200 km hr-1. These velocities are similar to thoseexperienced in tropical cyclones. Tides in the NorthSea were superimposed upon a seiching wave (alsoknown as a Kelvin wave) that moved counterclockwisealong the coast because of Coriolis force. The stormmoved in the same direction as this Kelvin wave. Indoing so, it set up a storm surge, exceeding 4 m inheight, which flooded The Wash in England andbreached the dams fronting the Zuider Zee. (Theeffect of this surge will be discussed in more detail atthe end of this chapter.) In Britain, 307 people died,32 000 inhabitants had to be evacuated, and 24 000homes were damaged. Over 83 000 hectares ofvaluable agricultural land were submerged under salt


986 18:00Jan 30

979 hPa971

966

96612:00Jan 31 North Sea

Heligoland

Zuider ZeeThe Wash 979

The Netherlands

970

975

2

3

4

England

1

Aberdeen

6:00Feb 1

1) English Channel –17032) Zuider Zee –1287

3) N. Frisian Is. –13624) Jutland, Denmark –1825

Fig. 3.13 Location of some historical, erosive North Sea storms including the path of the 1 February 1953 storm.

water. In the Netherlands, dykes were breached inover 100 places, flooding 1 600 000 hectares or one-sixth of the country. In this country, 2000 lives werelost and 250 000 head of livestock perished. The stormwas estimated to have had a recurrence intervalof 1:500 years, but does not rank as severe as some ofthose that occurred in the Middle Ages. It did becomea benchmark for modern engineering design in theNetherlands and Britain and gave rise to renewedinterest in coastal engineering in these countries.

The North Sea region still has the potential to spawnsome of the most intense storms ever witnessed. Thelast years of the twentieth century have seen an increasein these intense mid-latitude storms, with some of thelessons of past events ignored. For example, on thenight of 15 October 1987, very warm, humid air fromthe west of Africa clashed with cold Arctic air off thecoast of France, triggering the formation of a veryintense low-pressure cell off the Bay of Biscay. Thestorm, which swept up the English Channel and overthe south of England, eventually developed a centralpressure of 958 hPa, the lowest ever recorded inEngland. Wind gusts of 170–215 km hr-1 affected a

wide area between the coast of France and England.Only 18 people were killed, but the storm virtuallydestroyed all large trees growing in the south ofEngland. The French Meteorology Department hadissued a very strong wind warning two days in advance.The same information was available to the BritishMeteorological Office, but went unheeded until thestorm actually hit the English coastline. The storm wasreported as the worst to hit England, breaking pressureand wind records, and was followed by similar storms inJanuary–February 1990.

Two recent windstorms, on 26–27 December 1999,rank as intense as any medieval storm. Both stormswere given names – Lothar, which crossed northernFrance, southern Germany and Switzerland on26 December, and Martin, which crossed central andsouthern France, northern Spain and northern Italy aday later. Each developed in front of a rapidly movingmobile polar high and developed wind speeds of180 km hr-1 and 160 km hr-1, respectively. The stormscaused 80 deaths and damaged 60 per cent ofthe roofs in Paris. Lothar and Martin blew over 120and 80 electricity pylons, respectively. More than

62 Climatic Hazards

L

L

H

24 June 1588

H

29 June 1588

L

L

H

L

H

L

L

L

H

27 June 1588

2 July 1588

Fig. 3.14 Pressure pattern over northern Europe during the Spanish Armada-attempted invasion of Britain, 24 June–2 July 1588 (Lamb, ©1982; withpermission from Methuen, London).

three million insurance claims were filed, bank-rupting several insurance companies. The storms cost$US12 and $US6 billion damage, respectively, ofwhich less than 50 per cent was insured. Destructiveas these storms were, they had a theoretical returnperiod of 5–10 years and were matched by similarmagnitude events on at least eight other occasions,including the 16 October 1987 windstorm thatdevastated southern England. The 1:100 year stormevent in Europe is currently estimated to cost$US100 billion. Intense mid-latitude storms stillplague Europe today as much as they did at any pointover the past millennium.

East-coast lows or ‘bombs’

Formation

(Sanders & Gyakum, 1980; Holland et al., 1988)

East-coast lows are storm depressions that develop,without attendant frontal systems, over warm bodiesof water – usually off the east coasts of continentsbetween 25 and 40° latitude. The development of suchlows exhibits the forcing upon atmospheric instabilityby both topography and sea surface temperatures. The

lows gain latent heat from warm ocean currents, whichtend to form on the western sides of oceans. For someregions, where easterly moving low-pressure systemspass over a mountainous coastline and then out to seaover this warm poleward flowing current, intensecyclonic depressions can develop with structuralfeatures and intensities similar to tropical cyclones.This development is shown schematically in Fig-ure 3.15. These regions include the east coasts ofJapan, characterized by the Japanese Alps and theKuroshio Current; the United States, characterized bythe Appalachians and the Gulf Stream; and south-eastern Australia, characterized by the Great DividingRange and the East Australian Current. Lows may alsointensify off the east coast of South Africa, dominatedby the Drakensburg Range and the Agulhas Current.The lows are not associated with any frontal structure,but tend to develop under, or downstream of, a cold-core low or depression that has formed in the upperatmosphere. The depressions are associated on thesurface with a strong mobile polar high that may stallor block over the adjacent ocean. The initial sign that alow is developing is a poleward dip in surface isobarson the eastward side of these highs.


weak inland low

d

e

a

c

b

d

f

c

14

east-coast low

warm offshorewater

coldinshorewater

isotherm ˚C

a) Strong onshore temperature gradient

b) Upper atmospheric low pressure forms

c) Strong surface air flow

d) Strong orographic uplift

e) Explosive development of low

f ) Intense surface low forms

2624

2220

18

16 14

Fig. 3.15 Schematic development of an east-coast low.

The lows intensify and reach the ground near thecoastline due to convection that is enhanced by onshorewinds forced over steep topography at the coast.Topographic control also deflects low-level easterliestowards the equator, causing convergence and convec-tion that can explode the formation of the cyclone.Some lows become ‘bombs’, developing rapidly withina few hours. This explosive development occurs whenpressure gradients reach 4 hPa per 100 km. Convectionof moisture-laden air is enhanced by cold inshore waterthat maximizes condensation and the release of latentheat. Along the east Australian coastline, the magnitudeof the sea-surface temperature anomaly does notappear to be important in triggering a bomb. Rather,the temperature gradient perpendicular to the coast(the zonal gradient) is the crucial factor. This gradientmust be greater than 4°C per 0.5° of longitude, within50 km of the coastline, for a low to intensify.

Fig. 3.16 Infrared satellite image of the Halloween storm of October1991 off the north-east coast of the United States. Thestorm developed as an east-coast low and, shortly afterthis photograph was taken, was reclassified as a tropicalstorm (source: McCowan, 2001).

East coast lows develop preferentially at night, attimes when the maritime boundary layer is mostunstable. East coast lows tend to form in late autumn orwinter when steep sea-surface temperature gradientsare most likely. However, they are not necessarilyrestricted to these times, and can occur in any month ofthe year. In Australia, a 4.5-year cycle is apparent inrecords of these storms over the last forty years, with atendency for such lows to develop in transition yearsbetween ENSO and La Niña events. In extreme cases,the lows can develop the structure of a tropical cyclone.

They may become warm-cored, obtain central pressuresbelow 990 hPa, develop an ‘eye’ that appears on satelliteimages (Figure 3.16), and generate wind speeds inexcess of 200 km hr-1. These ‘eye’ structures have beendetected off the east coast of both Australia and theUnited States. Once formed, the lows travel polewardalong a coastline, locking into the location of warmestwater. Such systems can persist off the coastline for upto one week, directing continual heavy rain onto thecoast, producing a high storm surge, and generatingwaves up to 10 m in height. Heaviest rainfalls tend tooccur towards the tropics, decreasing with increasinglatitude. This rainfall pattern shifts towards the equatorfrom summer to winter, reflecting the intensificationand seasonal movement of high-pressure systems.

East coast lows have produced exceptional storms.The Ash Wednesday storm of 7 March 1962 in theUnited States and the 25 May 1974 storm in Australiaare classic examples of this type of event. The mostrecent, significant, east-coast lows to occur in the UnitedStates were the Halloween storm of 28–31 October1991 (The Perfect Storm), and the 21–22 December1994 storm. The latter ‘turned on’ explosively, reachinga central pressure of 970 hPa, developed a warm corewith attendant ‘eye’ structure, and generated winds of160 km hr-1 and waves 11.9 m high. These events arefurther described below.

United States Ash Wednesday storm of 7 March 1962(Dolan & Hayden, 1983; Nalivkin, 1983)

North America has experienced many severe, extra-tropical storms. For instance, in 1868–1869, cyclonicstorms on the Great Lakes sank or damaged more than3000 ships and killed over 500 people. However, theAsh Wednesday storm of 7 March 1962 is consideredthe worst storm recorded along the east coast of theUnited States. At no time did the storm reach land as itparalleled the coast some 100 km offshore. Its passagealong the coast took four days, with waves exceeding4 m in height for most of this time. In the decadesleading up to the storm, housing development hadtaken place on barrier islands along the coast. Thedestruction of the storm was immense. Whole townswere destroyed as storm waves, superimposed on astorm surge of 1–2 m, overwashed barrier islands(Figure 3.17). The loss of life was minimal since most ofthe damage occurred to summer homes in the winterseason. Coastal retreat was in the order of 10–100 m.

64 Climatic Hazards

However, this value varied considerably over shortdistances. The Ash Wednesday storm represented aturning point for coastal geomorphology and engineer-ing in the United States. Afterwards, the federalgovernment made research into coastal erosion andbeach processes a priority. Since then, the concept ofa coastal setback line has been developed andimplemented in some states. This setback line is basedon the probability, derived from historical evidence, ofcoastal flooding or erosion. Building development is pro-hibited seaward of the line, with the legality of such aprohibition being tested and upheld in the courts, specif-ically in North Carolina, where established insurancecompanies now will not insure buildings against wavedamage if they are built seaward of the setback line.

The Halloween storm of October 1991(The Perfect Storm)

(McCowan, 2001)

The Halloween storm of October 1991 was comparableto the Ash Wednesday storm. It began with the passageof an intense cold front preceding a mobile polar highthat swept across warm seas off the north-east coast ofthe United States. In North America, these east coastlows are called ‘nor’-easters’. The low intensified offNova Scotia in the late evening of 28 October 1991. Sostrong was this extra-tropical low that Hurricane Grace,which had formed on 27 October and was movingnorthward along the east coast of the United States,swung dramatically towards the storm. Grace wasalready producing waves 3–5 m high. Grace mergedwith the low, which continued to deepen to a centralpressure of 972 hPa by midday on 30 October, backedby a high-pressure cell with a central pressure of1043 hPa. Within the next 24 hours, winds reached

speeds of 120 km hr-1 and waves reached heights of12 m. Storm conditions extended along the coast fromNorth Carolina to Nova Scotia on a storm surgeexceeding 2 m. This surge had been exceeded onlyduring the Great Atlantic Hurricane of 1944 and thenor’-easter of March 1962. For hundreds of kilometersof coastline, sea walls, boardwalks, piers, and homeswere reduced to rubble. Inland, high winds downedcommunication infrastructure, tore out trees, and blewroofs off buildings. Finally, in one last twist, the centerof the storm passed over the warm Gulf Stream andacquired the characteristics of a hurricane.

Australian east-coast storms of May–June 1974(Bryant & Kidd, 1975; Bryant, 1983)

East-coast lows off the Australian coast are as intenseas tropical cyclones. However, the three May–Junestorms of 1974, while mainly confined to New SouthWales, remain the most significant and widespreadseries of such lows to strike the east coast of Australiain 100 years. Each storm event consisted of a classiceast-coast low that developed along the western edgeof the Tasman Sea over warm pools of water. In eachcase, blocking highs over south-eastern Australiaexacerbated the duration of the storms by directingsouth-east winds onto the coast for several days.Figure 3.18 presents the wave heights generated bythe three storms and the elevation of the associatedstorm surge as shown by the difference between actualand predicted tide heights. Maximum wave heightsreached 7 m on 25 May and again on 12 June. For mostof the three weeks, wave heights were above 4 m. Forthe whole period, tide levels exceeded predicted ones,with the greatest deviation being 0.7–0.8 m during the25 May storm. This is the highest recorded surge alongthe New South Wales coast. In embayments, this surgewas accompanied by seiching and wave set-up suchthat water levels at the peak of each storm did notfall towards low tide but kept rising. In Sydney, manyhomes built close to the shoreline were threatened;however, only a few homes were destroyed (Figure3.19). The sandy coastline suffered the greatestdamage. In the three-month period from April to June,Stanwell Park beach high tide line retreated over100 m. At Pearl Beach in Broken Bay, north of Sydney,a dune 7 m in height was overtopped by storm waves.Measured shoreline retreat amounted to 40 m on somebeaches (Figure 2.11). During the third storm,Cudmirrah Beach south of Jervis Bay underwent dune


Fig. 3.17 Storm damage produced by the Ash Wednesday storm of7 March 1962 along the mid-Atlantic United States coast.

scarping along a face 2 m in height at the rate of1 m min-1. Despite these large rates of retreat, coastalerosion was highly variable, with some beachesescaping the storm completely unscathed. Similarresults were found for the Ash Wednesday storm alongthe east coast of the United States.

Since 1974, there have been similar east-coast lowsjust as intense as the May 1974 storm. The autumnstorms of 1978 and 1985 both produced 10-meter wavesat Newcastle. Historical documents also suggest thatstorms rivaling the 1974 storms for intensity have beencommon over the past century. The fact that the fre-quency and magnitude of storms appears to be constantover time is not unique to New South Wales. A similareffect has been observed along the east coast of NorthAmerica. In the long term, there is no evidence thatstorms are solely responsible for coastal erosion in NewSouth Wales. Additional causes of beach retreat will bedescribed in more detail in Chapter 8.

The Sydney Hobart Yacht Race storm of 27–29 December 1998(Australian Bureau of Meteorology, 2001)

Boxing Day in Sydney heralds the beginning ofthe Sydney Hobart Yacht Race that pits amateur andprofessional sailors against each other in a leisurelysummer sail down the south-east coast of Australia.The deadly storm that beset the fleet in 1998 not only

66 Climatic Hazards

Inshore

Wav

e H

eigh

t (m

)

B Tide ConditionsActual

May June

May June

Tid

e he

ight

(m

)

7

6

5

4

3

2

1

0

2

1

0

Predicted

Actual

Offshore

Predicted

25 30 1 10 155

25 30 1 10 155

A Wave Conditions

Fig. 3.18 Storms of 25 May–15 June 1974 at Sydney, Australia. A) wave heights B) actual and predicted tide heights (adapted from Bryant & Kidd,© 1975).

Fig. 3.19 A) Undermining and collapse of beachfront houses at Bilgolabeach, Sydney following the 25 May 1974 storm.

Fig. 3.19 B) The same storm caused up to 40 m of erosion on otherbeaches. The collapsed structure shown here was apicnic table shelter situated about 5 m shoreward of theprevious dune scarp at Whale Beach, Sydney.

turned many off sailing forever, but was also astonish-ing in its intensity. A rescue helicopter provided one ofthe highest measurements of a wave ever recorded. Asit hovered 30 m above a stricken yacht buffeted by highwinds, a single wave, for a brief second, rose suddenlytowards the helicopter and touched its skids. However,the worst failing surrounding the race was the fact thatno warning was given at the final, pre-departurebriefing – despite ominous signs that an unusualmeteorological situation was developing.

The storm began with the passage of a cold frontacross the south of the continent on Christmas Day, theday before the race began. At the same time, an east-coast low had developed off the Queensland coast witha central pressure of 992 hPa. This latter low movedsouth-west and linked up with a low-pressure cellpreceding the cold front. In the early hours of27 December, both cells merged in Bass Strait betweenthe mainland and Tasmania, to explosively produce acomplex storm cell with a central pressure of 982 hPa.This cell then drifted east into the Tasman Sea and thepath of most of the yachts sailing down the coast. Tem-porarily, the center of the maelstrom developed an eyestructure characteristic of tropical cyclones. As it did so,the fleet was hammered by strong westerly windsgusting to 167 km hr-1. At this time, average maximumwave heights reached 14 m, with isolated rogue wavestheoretically reaching the 30 m height cited above.Many of the waves were over-steepened, probably bythe East Australian current moving in the oppositedirection. Not since the 1979 Fastnet Race off Britain –in which a freak storm killed 15 competitors – had thebest intentions of organizers of a yacht race been sodestroyed. Six people drowned in the Sydney Hobartstorm, five boats sank, and fifty-five sailors were rescuedunder some of the most treacherous conditions everexperienced by helicopter rescue crews. The subse-quent coroner’s inquiry found that the Cruising YachtClub of Australia, which organized the race, was blindto reports of impending disaster and had ‘abdicated itsresponsibility to manage the race’.

SNOWSTORMS, BLIZZARDSAND FREEZING RAIN

Snowstorms(Rooney, 1973; Whittow, 1980; Eagleman, 1983; Lott, 1993)

While many mid-latitude cyclonic depressions can giverise to exceptionally heavy rain and widespread

flooding, conditions will always be worse if the precip-itation falls in the form of snow. The fact that snowvolume exceeds rainfall by a factor of 7–10 implies thateven small amounts of precipitated water falling assnow can totally paralyze large sections of a continent.Even if the volume of snow in individual storms issmall, it can accumulate over many falls to present aserious flood hazard lasting one to two weeks duringthe spring melting season. If snowstorms occur toofrequently, then the effects of the previous storm maynot be cleared away and urban transport systems canbe slowly crippled. Whereas snow generally incapaci-tates transportation, freezing rain can cause severe andwidespread damage to power transmission lines.

Most large snowstorms originate as mid-latitudedepressions following the meandering path of the jetstream across the continents. In winter, the jet streamtends to be located across North America and Europe,so that the area affected by snowstorms is likely topersist over periods of several weeks. There is also therisk that abnormal or unseasonable jet stream pathscan cause snowstorms to occur in areas unprepared forthem. In North America, the preferred path for wintersnowstorms follows jet stream looping down overthe United States Midwest towards Texas andthen northward parallel to the Appalachians, GreatLakes–St Lawrence River Valley, and eastward overNewfoundland. This jet stream path is topographicallycontrolled by the Appalachian Mountains. Movementof the jet to the eastern side of the Appalachians canbring exceptionally heavy snowfalls to the east coastof the United States. In Europe, the jet stream tendsto loop down over England and northern Europe.Movement southward is partially hampered by theAlps. Extended looping southward can bring very coldconditions to northern Europe, and unseasonablesnowfalls to the Mediterranean region of Europe.

In winter, mid-latitude cyclonic depressions canaffect very large areas of the continental United States.Figure 3.20 shows areas of intense snowfall and thedevelopment of a low-pressure cell. In this diagram,rain is falling in the warm part of the air mass. As thewarm air is pushed up over the cold air, there is a smallzone of freezing rain (to be discussed in detail later). Asthe air is forced higher along the warm front, temper-atures drop below freezing point and precipitation fallsas snow. The area in front, and to the poleward side ofthe advancing warm front, is the area of heaviestsnowfall, which can cover a distance of 1000 km. Cold


air is rapidly pushed into the low behind the warm airmass. In this region, there is rapid uplift of moist, warmair. However, condensing moisture is rapidly turned tosnow. Here, snow may fall during thunderstorms, butthe amounts are generally less than that accumulatingin the path of the warm front. In exceptionally intensecases, another cyclonic frontal system can developto the equatorial side of the first, forming twin low-pressure cells.

Figure 3.21 is typical of the big snowstorms to haveaffected eastern North America in the late twentiethcentury. It illustrates a cyclonic situation that broughtsnow to the eastern half of the United States at the endof February 1984, paralyzing transport. The easternlow was associated with very warm air from the Gulf ofMexico that brought widespread rain to the southernpart of the United States. The western low was domi-nated by uplift of moist air along the polar front. Heavysnow extended across 1500 km, from New York toChicago, and from Canada to the southern states. Insome cases, as the cyclonic depression moves down theSt Lawrence Valley and occludes, spiraling winds mayturn the snow-bearing cloud back over parts of thecontinent already affected by the main storm, andprolong the fall of snow.

Notable events

The United States is most susceptible to large snow-storms because of the ability of mobile polar highs to

interact with moisture-bearing air originating from theGulf of Mexico. One of the worst snowstorms was ‘TheGreat Snow of 1717’, which resulted from four late-winter snowstorms that dumped 1.5 m of snow. NewEngland’s Blizzard of 1888 dropped between 1.0 and1.5 m of snow over several days. It is most notable forthe high winds that turned it into a blizzard. A centurylater the 12–15 March 1993 ‘Storm of the Century’replicated the havoc. This latter storm covered twenty-six states – from Texas to Maine – and the easternprovinces of Canada. The storm equated to a category3 hurricane on the Saffir–Simpson scale. It began inthe western Gulf of Mexico, tracked across Louisiana,and then rapidly intensified up the eastern seaboardbefore moving north-east over eastern Canada. Itgenerated maximum winds of 232 km hr-1 at MtWashington, waves 19.8 m high off Nova Scotia, a3.7 m storm surge in Florida and a minimum pressureof 961 hPa over New York State. The snowfalls weremost notable, reaching 1.4 m in Tennessee. Theseoccurred with record-breaking cold temperatures inconjunction with power failures that left three millionpeople without heating. The storm killed 270 people,paralyzed communications and air traffic, and impairedpeople’s ability to get to work. Fifteen tornadoesformed over Florida, accounting for 44 of those killed.The storm cost $US6 billion, making it the most expen-sive extra-tropical storm in United States history.Three years later, on 6–8 January 1996, a similar storm

68 Climatic Hazards

Snow Rain Freezingrain

Snow

< 0˚ C< 0˚ C> 0˚ C

A B

LowDepression

A

B

Rain

Freezing rain

Snow

1008

1004

1000

996

Fig. 3.20 Schematic representation of precipitation patterns for a mid-latitude cyclonic depression in winter in the northern hemisphere.

sweeping along the same track produced heavier snow-falls in the eastern United States.

Impact of snowstorms

Heavy, northern hemisphere snowstorms have sub-stantial effects. As soon as snow begins to fall, trans-portation networks in the United States are disrupted.Local roads can be completely closed off within hoursof a storm striking and, depending upon the severity ofthe storm, closed for several days (Figure 3.22).Accident rates increase 200 per cent above averagewithin a few hours of a storm. Geographically, motorvehicle insurance rates reflect the likely incidence ofstorms. During moderate falls, vehicles caught onroads can become stranded and impede snow removal.In extreme cases, stranded vehicles can become deathtraps. Motors left idling for warmth can suffocate occu-pants with carbon monoxide, while accompanying coldtemperatures can freeze people to death. In the severeIllinois winter of 1977–1978, 24 people died in motorcars from these causes.

Airports are also affected within hours, and can bekept closed for several days because of the timerequired to clear long, wide runways. Closure of majorairports such as Chicago can disrupt airline schedulesnationwide. Blocking of roads leads to closure ofschools, industry and retail trade. The effect on retail

trade, however, is often short-term because shoppersusually postpone purchases until after the storm. Infact, the seasonal onset of snow is seen in manynorthern states as a blessing, because it triggersconsumer spending for Christmas. Purchase of winterclothes, snow tires, and heaters can also increase.


H

H

L L

L

H

SnowSnow

Rain

Snow

Flurries Showers

Showers

Flurries

Snow

HL

PrecipitationHigh PressureLow PressureCold frontWarm Front

Fig. 3.21 Pressure patterns for February 1984 snowstorm that paralyzed the eastern United States.

Fig. 3.22 The effect of a heavy snowstorm on roadways in St John’s,Newfoundland (photograph from Canadian AtmosphericEnvironment Service).

Motel business can boom during major storms.Surprisingly, garage stations record heavier business,too, because people unable to reach work can stillmanage to travel to the local service station tocomplete delayed vehicle maintenance. For mostindustries, a major storm means lost production.Contingencies must be made by certain industries –such as steelworks – to begin shutting down blastfurnaces before a storm affects the incoming shift ofworkers. Unless production is made up, industry canlose profits and contracts, workers can lose pay andgovernments, income-tax revenue. Worker and studentabsenteeism can overload domestic power demand, asthese people are at home during the day utilizingheating and appliances not normally used at that time.Severe, rapidly developing, or unpredicted snowstormsmay even prevent the workers required for snowremoval from reaching equipment depots.

Annually, many cities in North America budgetmillions of dollars to remove snow from city streets.Most cities have programs of sanding and salting roadsand expressways to keep the snow from beingcompacted by traffic and turning into ice. For the citiessurrounding Lake Ontario, the volume of salt used onroads is so large that increased salinity now threatensfreshwater life. Snow-removal programs necessitate abureaucracy that can plan effective removal of snow atany time of day or night during the winter months.Inefficiency in this aspect has severe economicrepercussions on industry and retail trading.

All of these disruptions depend upon the perceptionof the storm. Cities experiencing large annual snowfallsare generally more prepared to deal with them as ahazard. Snowfalls under 10 cm – considered minor inthe north-east of the United States – will paralyzesouthern states. However, an awareness of snowstormscan be counterproductive. In the western United States– where on a personal level snowfall is considered aneveryday element of winter life – large snowfalls cancripple many major cities because they do not preparefor these higher magnitude events. In the eastern states,large storms are expected, but continuous small falls arenot. Here, snow clearing operations and responses aregeared to dealing with larger storms.

The variation in perception and response to thesnow hazard is best illustrated around the Great Lakes.Because of their heat capacity, the Great Lakes remainunfrozen long into the winter. Cool air blowing overthese bodies of relatively warm water can cause them

to reach their saturation point and lead to rapid accu-mulation of snow downwind. As the prevailing windsare westerly in the southern lakes, and northerlyaround Lake Michigan and Lake Superior, snowbeltsdevelop on the eastern or southern sides of the Lakes.In these snowbelts, average annual snowfalls canexceed 3 m – double the regional average. Often thissnow falls during a few events, each lasting severaldays. Cities such as Buffalo on the eastern side of LakeErie are prepared for the clearing of this type ofsnowfall, on top of which may be superimposed majormid-latitude storms. Cities such as Hamilton at thewestern end of Lake Ontario have to cope only withthe irregular occurrence of mid-latitude storms. Eachcity has built up a different response to snow as ahazard, and each city allocates different sums of moneyfor snow-removal operations. Buffalo removes its snowfrom roads and dumps it into Lake Erie; if it did not,snow would soon overwhelm roadsides. Hamilton usessanding and salting operations to melt smaller amountsof snow, and only resorts to snow removal for the largerstorms. These two cities, within 60 km of each other,thus have very different responses to snow as a hazard.

However, these average winter conditions are notguaranteed. Too many mid-latitude cells – even oflow intensity – tracking across the Great Lakes willdiminish the duration of westerly winds and increasethe duration of easterly winds. Hamilton then becomespart of a localized snowbelt at the western end of LakeOntario, while Buffalo receives little snowfall. Buffalomay end the winter with millions of unspent dollarsin its snow-removal budget, while Hamilton has torequest provincial government assistance, or increasetax rates to cope with the changed conditions of onewinter’s snowfall.

The winter of 1976–1977 was a dramatic exceptionfor Buffalo. Abnormally cold temperatures set recordsthroughout the north-east. Buffalo’s well-planned snowremoval budget was decimated by a succession ofblizzards. On 28 January, one of the worst blizzards tohit an American city swept off Lake Erie. The stormraged for five days and occurred so suddenly that17 000 people were trapped at work. Nine deathsoccurred on expressways cutting through the city,when cars became stalled and people were unable towalk to nearby houses before freezing to death. Over3.5 m of snow fell. Winds formed drifts 6–9 m high.For the first time in United States history, the federalgovernment proclaimed a disaster area because of a

70 Climatic Hazards

blizzard. A military operation had to be mounted, with2000 troops digging out the city. Not only was Buffalocut off from the rest of the world, it was also completelyimmobilized within. Over one million people werestranded in streets, offices, their own homes, or onisolated farms. It took two weeks to clear the streets. Tominimize the local spring flooding risk, snow had to beremoved to other parts of the country using trains. Thefinal clearance bill totalled over $US200 million, morethan five times the budgeted amount for snow removal.In contrast, the city of Hamilton was relativelyunaffected by this blizzard, and ended up loaning snowremoval equipment to Buffalo as part of an inter-national disaster response.

The Buffalo blizzards of 1976–1977 illustrate theadditional hazard caused by abnormal amounts ofsnow: flooding in the subsequent spring meltingperiod. While snow accumulation can be easilymeasured, the rate of spring melting is to some degreeunpredictable. If rapid spring warming is accompaniedby rainfall, then most of the winter’s accumulation ofsnow will melt within a few days, flooding major rivers.The historic flooding of the Mississippi River in thespring of 1973 occurred as the result of such con-ditions, after a heavy snowfall season within itsdrainage basin. Snowmelt flooding is a major problemin the drainage basins of rivers flowing from moun-tains, and in the interior of continents with significantwinter snow accumulation. This aspect is particularlysevere in the Po Basin of northern Italy, along thelower Rhine, and downstream from the Sierra NevadaMountains of California. On a continental scale, theflooding of the Mississippi River and its tributariesis well known. However, for river systems drainingnorth to the Arctic Ocean, snowmelt flooding can be acatastrophic hazard. Because the headwaters of theserivers lie in southern latitudes, melt runoff and icethawing within the river channel occur first upstreamand then downstream at higher latitudes. The ice-dammed rivers, swollen by meltwater, easily floodadjacent floodplains and low-lying topography. InCanada, spring flooding of the Red River passingthrough Winnipeg, and the Mackenzie leading tothe Arctic Ocean, is an annual hazard. Nowhere is theproblem more severe than on the Ob River systemin Russia. All northward flowing rivers in Russiaexperience an annual spring flooding hazard; but onthe Ob River, the problem is exacerbated by theswampy nature of the low-lying countryside.

Blizzards(Orville, 1993; Environment Canada, 2002; Henson, 2002)

There is one special case of snowfall which does notdepend upon the amount of snow falling but upon thestrength of the wind. The term ‘blizzard’ is given to anyevent with winds exceeding 60 km hr-1, visibility below0.4 km for more than three hours because of blowingsnow, and temperatures below –6°C. The wordderives from the description of a rifle burst and wasfirst used to describe a fierce snowstorm on 14 March1870 in Minnesota. The term ‘purga’ is used in Siberia.A severe blizzard occurs if winds exceed 75 km hr-1

with temperatures below –12°C. Strong mid-latitudedepressions with central pressures below 960 hPa (lessthan some tropical cyclones), and accompanied bysnow, will produce blizzard conditions. On the prairiesof North America strong, dry north-westerly windswithout snow are a common feature of winter out-breaks of cold Arctic air. These winds can persist forseveral days. Snow is picked up from the ground andblown at high velocity, in a similar manner to dust indust storms. Surprisingly, the death toll annually fromblizzards in North America is the same as that causedby tornadoes. The 28 January 1977 blizzard at Buffalo,described above, resulted in over 100 deaths across theeastern United States, and produced winds in excess of134 km hr-1. In March 1888, the late New Englandwinter snowstorm mentioned above turned into ablizzard as winds gusted 128 km hr-1. Temperatures inNew York City dropped to –14.5ºC – the coldest onrecord for March.

The main hazard of blizzards is the strong wind,which can drop the wind-chill factor, the indexmeasuring equivalent still-air temperature due to thecombined effects of wind and temperature. Wind chillquantifies terms such as ‘brisk’, ‘bracing’ and ‘bone-chilling’. Since 1973, wind chill has been measuredusing the Siple–Passel Index based upon how long ittakes to freeze a water-filled plastic cylinder positioned10 m above the ground. It is assumed that, at temper-atures of less than –35°C, lightly clothed skin or bareflesh will freeze in 60 seconds. If body temperaturedrops by more than 5°C, hypothermia and death willresult. However, water in a plastic cylinder bears norelationship to reality. In addition, body heat forms alayer of warm air adjacent to exposed flesh. This layeris dispersed faster by increasing wind speeds. Finally,there is biofeedback such that flesh exposed to wind


feels less cold as time progresses. A new index has beenconstructed that is referenced to a human walking at5–6 km hr-1 with their face exposed. This index wasestablished using volunteers walking on a treadmill in awind tunnel. Figure 3.23 presents this version of thewind chill index. For example, a face exposed to a50 km hr-1 wind at a temperature of –25°C experi-ences the equivalent, still-condition temperature of –42°C. For this scenario, the old index gave a greaterwind chill of –51°C. The chart also indicates thetemperatures and wind speeds at which human fleshwill freeze after 2 minutes, 10 minutes, and prolongedexposure to the elements. For example, if the air tem-perature near the ground is –25°C and the windspeed is 20 km hr-1, then frostbite will occur within10 minutes. Other factors are being incorporated intothis schema. One’s flesh cools more quickly at nightthan in sunshine, an effect that is, surprisingly, exacer-bated by precipitation but not by increased humidity.In addition, nothing substitutes for acclimatization.Hence, the original inhabitants of Tierra del Fuego atthe tip of Patagonia were discovered walking aroundnaked – by Europeans who found the cold unbearable.

Freezing rain(Environment Canada, 1998; Jones & Mulherin, 1998;Fell, 2002–2003)

In some cases in winter or early spring, mid-latitudedepressions will develop rain instead of snow along theuplifted warm front. To reach the ground, this rain must

fall through the colder air mass (Figure 3.20). If thiscolder air is below freezing and the rain does not freezebefore it reaches the ground, then it will immediatelyturn to ice upon contacting any object with a tempera-ture below 0°C. If the warm front becomes stationary,then continual freezing rain will occur over a period ofseveral days, and ice can accumulate to a thicknessof tens of millimetres around objects. The addedweight on telephone and power lines may snap the lineswhere they connect to a pole. Collapsing tree branchesexacerbate the problem and disrupt road transport(Figure 3.24). The area most susceptible to cripplingice storms occurs in a band from central Texas, northover the Appalachian Mountains and along the north-eastern United States–south-eastern Canada border.This is the region where low pressure can develop atthe leading edge of mobile polar highs and stall againstthe Appalachians or a preceding high-pressure cell.Severe ice storms occurred in Kentucky and Tennesseein January 1951, Georgia and South Carolina inDecember 1962, Maine in December 1964 and January1979, Texas and western New York in March 1976,Iowa in March 1990, Mississippi in February 1994, andthe US Midwest in December 2000.

The worst ice storm occurred on 4–10 January 1998along the Canadian–US border east of the GreatLakes. In the United States, federal disaster areas weredeclared in 37 counties, 11 people died and 18 millionacres of forest worth $US1–2 billion was damaged byice. Analysis indicates that the storm has a return

72 Climatic Hazards

Temperature (˚C)

5 0 -5 -10 -15 -20 -25 -30 -35 -40 -45 -50

10 3 -3 -9 -15 -21 -27 -33 -39 -45 -51 -57 -63

20 1 -5 -12 -18 -24 -31 -37 -43 -49 -56 -62 -68

30 0 -7 -13 -20 -26 -33 -39 -46 -52 -59 -65 -72

40 -1 -7 -14 -21 -27 -34 -41 -48 -54 -61 -68 -74

50 -1 -8 -15 -22 -29 -35 -42 -49 -56 -63 -70 -76

60 -2 -9 -16 -23 -30 -37 -43 -50 -57 -64 -71 -78

70 -2 -9 -16 -23 -30 -37 -44 -51 -59 -66 -73 -80

80 -3 -10 -17 -24 -31 -33 -45 -52 -60 -67 -74 -81

Frostbite possible after prolonged exposure

Frostbite possible after 10 minutes

Frostbite possible after 2 minutes

Windspeedkm hr-1

Fig. 3.23 Equivalent temperature or wind chill for given air temperatures and wind speeds. Wind chill temperatures at which frostbite is possible after10 minute and 2 minute exposures are shaded (based on Environment Canada, 2002).

period of between 35 and 85 years, and that thedamage was comparable to a storm that struckthe same region in December 1929. In Canada, thestorm paralyzed the city of Montreal and its populationof four million for several weeks. Here, the accumula-tion of ice around objects was unprecedented andamounted to 100 mm at some locations – double anyprevious storm. Around Montreal, more than 1000power transmission towers and 30 000 wooden utilitypoles crumbled under the weight. Over 120 000 km ofpower lines and telephone cables were downed. Closeto 1.4 million people in Quebec and 230 000 in Ontariowere without electricity for over a week. Temperaturesdropped to –40°C afterwards and 100 000 people hadto be evacuated to shelters for a month because of alack of home heating. Twenty-five people died – mainlyfrom hypothermia, 20 per cent of Canada’s work forcemissed work, 25 per cent of its dairy industry wassuspended and five million sugar maples lost thecapacity to produce syrup for 30 to 40 years. Over16 000 Armed Forces personnel had to be brought into assist overwhelmed utility workers. The total coastto the economy was in excess of $CAD2 billion.

Two recent innovations may go a long way towardsminimizing the downing of power lines by ice. The firstutilizes a property of ice whereby the top layer of watermolecules exposed to air is quasi–liquid and conductselectricity. Ice can be melted from power lines simplyby passing a high-frequency electrical current at50 watts m-1 along the cables. The second innovationcompletely does away with power lines. The French

are developing a system whereby electricity can betransferred between towers using microwaves, thuseliminating cables.

STORM SURGES(Wiegel, 1964; Coastal Engineering Research Center, 1977;Anthes, 1982)

Introduction

Storm surge, as alluded to above, plays a significantrole in tropical and extra-tropical cyclone damage.Storm surge was the main cause of death inBangladesh (East Pakistan) in the 1970 and 1985cyclones; the main cause of destruction in theFebruary 1953 North Sea storm; and the main reasonthat waves were so effective in eroding beaches inthe May–June 1974 storms in New South Wales. In theUnited States storm surge is considered the mainthreat necessitating evacuation of residents fromcoastal areas preceding tropical cyclone landfall. Here,evacuation was instituted in response to the Galvestonhurricane disaster of September 1900, during which6000 people died because of storm-surge inundation.The phenomenon is also a recurring hazard in theembayments along much of the Japanese and south-east Chinese coastlines. For example, approximately 50 000 people lost their lives around Shantou(Swatow), China, on 3 August 1922 and, as recently asSeptember 1959, 5500 people lost their lives in IseBay, Japan, because of storm surges. In this concludingsection, the causes of storm surge and the concept ofprobability of occurrence will be discussed.

Causes

Storm surge is generated by a number of factorsincluding:• wind set-up,• decreases in the atmospheric weight on a column of

water,• the direction and speed of movement of the

pressure system,• the shallowness of the continental shelf, bay or lake,

and• the shape of the coastline. This discussion will ignore any addition effects due toriver runoff, direct rainfall, wave set-up inside the surfzone, or Coriolis force.

The main reason for storm surge would appear to bethe piling up of water by wind. The exact amount of


Fig. 3.24 The effect of freezing precipitation and wet snow on wiresand tree branches in Ontario, Canada (photograph fromCanadian Atmospheric Environment Service).

water piled up depends upon the speed of the wind, itsduration, and its location relative to the center of acyclone. Wind set-up is difficult to calculate, butWiegel (1964) gives an equation that shows the magni-tude of wind set-up in a channel as follows:

h2 ~ [(2.5 ��-1 g-1) (x + C1)] – d (3.1)where h = height of wind set-up in metres

d = water depth of channel� = 0.0025Uo

2

� = density of salt water g = gravitational constant

Uo2 = wind speed

(x + C1) = the fetch length of the surge

Equation 3.1 does not help one to easily conceptualizewind-induced surge heights, because the actual surgedepends upon where you are relative to the movementof the cyclone and its center. If the wind is movingaway from a coast, it is possible to get set-down. Ingeneral, for a cyclone with 200 km hr-l winds, onecould expect a wind set-up of at least 2 m somewherearound the cyclone. Sea level elevation depends moreupon the weight of air positioned above it. As airpressure drops, sea level rises proportionally. This isknown as the inverted barometer effect. It must beremembered that some tropical cyclones have apressure reduction of up to 13 per cent. Equation 3.2expresses simply the relationship between sea leveland atmospheric pressure:

hmax = 0.0433(1023 – Po) (3.2)where hmax = the height of the storm surge

due to atmospheric effects Po = the pressure at the center of the

hurricane in hPa

From Equation 3.2, it can be seen that the lowestcentral pressure of 870 hPa ever recorded for a tropicalcyclone could have produced a theoretical surgeheight, due to atmospheric deloading, of 6.6 m. Thetheoretical surge height for the North Sea storm of1953 is 2.47 m, while for the 25 May 1974 storm inNew South Wales it is 1.2 m. The last value is higherthan the recorded tide level difference and reflectscyclone movement.

If a storm moves in the direction of its wind speed,then it will tend to drive a wall of water ahead of it.This wall behaves as a wave and travels with a speed

similar to that of the storm. The height of the wave isrelated to the size of the disturbance. Along the UnitedStates east coast, this long wave can have a height ofseveral metres, becoming highest where land juts outinto the path of the tropical cyclone. Figure 3.25 showsthe track of Hurricane Carol in 1954 along the UnitedStates east coast, together with records of sea level ele-vation at various tide gauges. Note that as the cycloneapproaches land, the surge height increases, but itincreases most where land at Long Island and CapeCod intercepts the storm path. At these locations thelong wave, which has been moving with the storm,piles up against the shoreline. The 21 September 1938east-coast hurricane, moving along a similar path, senta wall of water 6 m in height plowing into the LongIsland coastline, where it made landfall. Along theCape Hatteras barrier island chain that sticks out intothe Atlantic Ocean, and along the Gulf of Mexicocoastline of Florida, planning dictates that houses mustbe constructed above the 6.5 m storm-surge floodlimit. In New South Wales, Australia, most stormsmove offshore; so, the long wave is in fact travelingaway from the coast, and there is minimal storm surgefelt at the shoreline. This is the reason storm-surgeelevation during the May 1974 storm reached only0.7–0.8 m instead of the theoretically possible value of1.2 m due to atmospheric deloading.

Figure 3.25 also reflects two other factors affectingsurge. As Hurricane Carol moved along the coast, itbegan to cross a shallower shelf. The storm-surge waveunderwent shoaling (shallowing) as it moved shore-ward. This raised its height. Figure 3.26 illustrates theeffect of shoaling on a long wave. The movement ofthe storm-surge wave throughout the water column isdictated by the speed of the storm. As this wave movesthrough shallower water, its speed decreases and,because the energy flux is conserved through adecreasing water depth, wave height must increase.Any shallow body of water can generate large surgesfor this reason. The Great Lakes in North America arevery susceptible to surges of the order of 1–2 m.In December 1985, a major storm struck Lake Erie,producing a surge of 2.5 m at Long Point, Ontario,coincident with record high lake levels. Lakeshorecottages were floated like corks 0.5–1.0 km inland. Inthe shallow Gulf of Finland towards St Petersburg,surges 2–4 m in height can be generated. Since 1703,there have been at least 50 occasions when surgesgreater than 2 m have flooded this city.

74 Climatic Hazards


29 2826

27

Time of day

Long Is.

31

Nova Scotia

Bay of Fundy

L

31

30

30

25

240730

1930

Hurricane Carol, 1954

Aug 27 28 29 30 31 Sept 1 2

K

I

H

G

FEDC

B

A

LK

IH G

F

E

DC

B

A

M

N

O

P

Q

RS

MNO

PQ

RS

2

1

0

-1

(m)

0 1000 km

Fig. 3.25 Track of Hurricane Carol, 1954, and tide records along the United States east coast (after Harris, 1956).

wave height increases as water depth decreases

Hs

Ho

Continental shelf

Continental slopeSeabed

Shoreline

mean sealevel

Open ocean ordeep water

H

Fig. 3.26 Effect of wave shoaling on the height of long waves crossing the continental shelf.

Finally, the shape of the coastline accounts for thelargest storm surge. The Bay of Bengal, where stormsurges have drowned so many people, is funnel-shaped(Figure 3.10). Any surge moving up the funnel, as the1970 cyclone storm surge did, will be laterally com-pressed. In addition, the size of the basin can lead toresonance if its shape matches the period of any waveentering it. A similar enhancement effect happened in1770 to the tidal bore on the Qiantang River, south ofShanghai, China. A storm surge lifted the tidal bore onthe funnel-shaped estuary to heights in excess of 4 m.The protecting dykes on each side of the river wereoverwashed and over 10 000 people drowned withinminutes. The highest tides in the world are recorded inthe Bay of Fundy in Canada because the basin shapematches, within six minutes, the diurnal tidal period of12.42 hours. On 4 October 1869, a cyclone, called theSaxby Gale, moved up the United States coast in asimilar fashion to Hurricane Carol in 1954, but slightlyeastward. The storm traveled up the Bay of Fundy,moving a mass of water at about the resonance fre-quency (13.3 hours) of the Bay of Fundy–Gulf ofMaine system. The resulting storm surge of 16 m wassuperimposed on a tide height of 14 m. In all, waterlevels were raised 30 m above low tide level, almostoverwashing the 10 km isthmus joining Nova Scotia tothe mainland.

PROBABILITY OF OCCURRENCE(Leopold et al., 1964; Wiegel, 1964)

The probability of occurrence of a surge height ishighly dependent upon the physical characteristics of acoastal site. To define this probability, knowledge of thesize of past events and how often they have occurredover time (magnitude–frequency) is also required. Thisinformation is usually obtained from tide gauges. Themaximum storm-surge values have been calculatedfor many tide stations around the United States Gulfand east Atlantic coast. This type of information canbe used for planning; however, it is fraught with thedanger that one may not have records of the mostextreme events. Take the example of the 1953 stormsurge in the Netherlands. Engineers correctly designeddykes to withstand the 100–200-year storm-surgeevent; but the 1953 event exceeded those limits. Theprobability of occurrence and magnitude of this event,but not its timing, could have been foreseen from asimple analysis of past historical events.

Recurrence intervals

The probability of rare events of high magnitude can beascertained in one of two ways: by determining therecurrence interval of that event or by constructinga probability of exceedence diagram. In both methods,it is assumed that the magnitude of an event can bemeasured over discrete time intervals – for example,daily in the case of storm waves, or yearly in the case ofstorm surges. All the events in a time series at suchintervals are then ranked in magnitude from largest tosmallest. The recurrence interval for a particularranked event is calculated using the following equation:

Recurrence interval = (N + 1) M-1 (3.3)where N = the number of ranks

M = the rank of the individual event (highest = 1)

The resulting values are then plotted on special loga-rithmic graph paper. This type of plot is termed aGumbel distribution. An example of such a plot isshown in Figure 3.27 for 70 years of maximum fort-nightly tide heights in the Netherlands before the 1953storm surge event. Note that the recurrence interval isplotted along the x-axis, which is logarithmic, while themagnitude of the event is plotted along the y-axis.Often the points plotting any natural hazard time serieswill closely fit a straight line. The extrapolation of thisline beyond the upper boundary of the data permitsthe recurrence interval of unmeasured extreme eventsto be determined. For example, in the century beforethe 1953 storm surge in the Netherlands, the greatestrecorded surge had a height of 3.3 m. This eventoccurred in 1894 and had a recurrence interval of onein 70 (1:70) years. The ranked storm-surge data for theNetherlands fit a straight line which, when extendedbeyond the 70-year time span of data, permits one topredict that a 4 m surge event will occur once in800 years. The 1953 surge event fits the straight linedrawn through the existing data and had a recurrenceinterval of once in 500 (1:500) years. If the dykes hadbeen built to this elevation, they would be expected tobe overtopped only once in any 500-year period. Notethat the exact timing of this event is not predicted, butjust its elevation. The engineers who designed thedykes in the Netherlands before the 1953 North Seastorm cannot be blamed for the extensive floodingthat occurred because they had built the dykes to with-stand only the 1:200 year event. At present, dykes in

76 Climatic Hazards

the Netherlands have been elevated to withstand the1:10 000 year event with a predicted surge height of 5 m.

Probability of exceedence diagrams

Recurrence intervals can be awkward to interpret,especially if the period of measurement is limited. Inthis case, a probability of exceedence diagram isfavored because it expresses the rarity of an event interms of the percentage of time that such an event willbe exceeded. Probability of exceedence diagrams areconstructed by inverting Equation 3.4 as follows:

Exceedence probability = M(N + 1)-1 100% (3.4)

Data in this form cannot be plotted on logarithmicpaper, but instead are plotted on normal probabilitypaper. This paper has the same advantage as logarith-mic paper, in that the probability of occurrence of rareevents beyond the sampling period can be readilydetermined, because the data tend to plot along oneor two straight lines. This is shown in Figure 3.28, aprobability plot for the Netherlands storm-surge dataplotted in Figure 3.27. For these data, two lines fit the

data well. Coincidentally, where the two lines intersectat a high-water height of 2.4 m is the point separatingstorm surges from astronomical tides. The advantageof a probability plot is that the frequency of an eventcan be expressed as a percentage. For example, a stormsurge of 3 m – which in Figure 3.27 plots with a recur-rence interval of 1:30 years – has a probability of beingexceeded only 3.8 per cent of the time, or once every26 years (1:26) (see Figure 3.28). Engineers find thistype of diagram convenient and tend to define an eventas rare if it is exceeded 1 per cent of the time. Proba-bility plots have an additional advantage in that theprobability of low-magnitude events can also be deter-mined. For instance, annual rainfalls, when ranked andplotted as recurrence intervals, will not provide anyinformation about the occurrence of drought. If thesame data are plotted on probability paper, boththe probability of extreme rainfall as well as deficientrainfall can be determined from the same graph.

Both types of plots have limitations. Firstly, anextreme event such as a 1:10 000 year storm surge,while appearing rare, can occur at any time. Secondly,once a rare event does happen, there is nothingto preclude that event recurring the next day. Thisconstitutes a major problem in people’s perception ofhazards. Once high-magnitude, low-frequency eventshave been experienced, people tend to regard them asbeyond their life experience again. The farmers of theChars, having lived through the 1970 storm-surgeevent that killed 500 000 people, came back and re-established their flooded farms with the notion thatthey were now safe because such an event was so infre-quent that it would not recur in their lifetimes. Only 15years later, a storm surge of similar magnitudehappened again. The 1973 flooding of the MississippiRiver was the greatest on record in 200 years. Withintwo years that flood had been exceeded. High-magnitude events in nature tend to cluster over timeregardless of their frequency. Such clustering is relatedto the persistence over time of the forcing mechanismcausing the hazard. In the case of storm surges in theBay of Bengal, the climatic patterns responsible forthe cyclones may be semi-permanent over the span ofa couple of decades. A third drawback about proba-bility diagrams is that the baseline for measurementsmay change over time. For example, deforestation of amajor drainage basin will cause rare flood events tobecome more common. In the case of the Nether-lands, storm surges in the Middle Ages were certainly


1894

1953

5

4

3

2

1

0

Mean high water

Hei

ght o

f hig

h w

ater

(m

)

Recurrence interval (yr)

Stor

m s

urge

s

.001 .01 .1 1 10 100 1000 10000

Fig. 3.27 Recurrence intervals of storm surges in the North Sea alongthe Netherlands coast for the 70 years before the 1953February surge event (after Wemelsfelder, 1961).

more frequent and to higher elevations than at present.The 1:10 000 storm-surge elevation planned for todaycould also be exceeded by an event smaller than 5 m ifmean sea level were to rise. Finally, neither diagrammay be the most appropriate method in engineeringdesign to determine the frequency of rare events. Inthe case where high-magnitude events tend to clusterover time, engineers will resort to frequency distribu-tions that are more complex to permit the recurrenceof clustered events to be evaluated more realistically.

DUST STORMS(Lockeretz, 1978; Goudie, 1983; Middleton et al., 1986;Pearce, 1994; Tegen et al., 1996; Griffin et al., 2002)

Introduction

Dust storms are windstorms accompanied by sus-pended clay and silt material, usually but not alwayswithout precipitation. Presently, between 130 and

800 million tonnes of dust, with extremes as high as5000 million tonnes, are entrained by winds each year.The average annual amount of clay-sized particlesmoved is about 500 million tonnes. Dust storms areresponsible for most of the terrigenous material foundin ocean basins, contributing over 75 million tonnes ofmaterial per year to the Atlantic Ocean alone. Atdistances 5000 km out into the Atlantic Ocean, falloutfrom the Sahara is still deposited at a rate of3000 tonnes km-2 yr-l. Dust from the Sahara is fre-quently deposited 7000 km away on Caribbean islandsand in the North Amazon Basin. In the latter location,it accumulates at the rate of 13 million tonnes everyyear. Saharan dust often falls over western Europe withsingle storms being used as marker horizons in glaciersin the Alps. Dust from China exceeds these statistics.Dust from the Gobi Desert has been recorded inHawaii and Alaska, a distance in excess of 10 000 km.A dust plume, which left China around 25 February

78 Climatic Hazards

Single line of fit

Sto

rm s

urge

s

2 lines of best fit

Height of high water (m)1.0 2.0 3.0 4.0 5.0

99.999.8

95.0

90.0

99.5

99.0

98.0

80.0

70.0

60.0

50.0

40.0

30.0

20.0

10.0

5.0

2.0

1.0

0.5

0.20.1

% A

ccum

ulat

ed p

roba

bilit

y

Fig. 3.28 Probability of exceedence diagram for the same data set as depicted in Figure 3.27 (data from Wemelsfelder, 1961).

1990, was tracked across North America using satelliteimagery. It arrived in the French Alps ten days later –having traveled a distance of 20 000 km.

Formation

Dust storms commonly form as the result of thepassage of cold fronts linked to mobile polar highsacross arid or drought-affected plains. The passage ofthese fronts can give rise to dust storms lasting forseveral days. In the northern Sahara region, duststorms are mainly produced by complex depressionsassociated with the westerlies. The depressions origi-nate in winter in the eastern Mediterranean Sea or theAtlantic Ocean. In the southern Sahara, low-pressurefronts are responsible for harmattan winds. Thesedepressions track south during the northern hemi-sphere winter and easterly during the summer. Aparticularly common feature of dust storms in NorthAfrica is the haboob, generated by cold downburstsassociated with large convective cells, and linked to theadvance of the intertropical convergence. These typesof winds also account for many of the dust storms inthe Gobi and Thar deserts in central Asia, in theSudan, and in Arizona. The seasonal movement ofthe monsoon into the Sudan region of the Sahara isresponsible for dust storms in eastern Africa, while theIndian monsoon controls the timing of dust storms onthe Arabian Peninsula. Dust transport out of northernAfrica to the Mediterranean Sea and the Atlantic Oceanis correlated to stages in the North Atlantic Oscillation(NAO) described in the previous chapter. When theNAO index is high in winter, dust transport out ofAfrica is greater. East of the Mediterranean, depres-sions moving across Turkey and northern Iraq generatemost dust storms. In Australia, dust storms are gener-ated by the passage of intense, cold fronts across thecontinent, following the desiccating effects of hot, drywinds produced by the subtropical jet stream in the leeof high-pressure cells. The Victorian dust storm of8 February 1983 (Figure 3.29), and the devastatingbushfires of Ash Wednesday exactly one week later,both originated via this mechanism. In mountainousregions such as western North America, katabaticwinds can generate dust storms in places such asCalifornia and along Colorado’s Front Ranges.Similarly, cold, dense winds flowing off the mountainsin the Himalayas and the Hindu Kush generate duststorms southward over the Arabian Sea. Such localizedwinds also produce dust storms in the valleys leading

from the Argentine foothills. There is a correlationworldwide between dust storms and rainfall. Whererainfall exceeds 1000 mm yr-l, dust storms occur on lessthan one day per year, mainly because vegetationprevents silt entrainment by winds into the atmos-phere, and convective instability tends to give rise tothunderstorms rather than dust storms. Dust stormfrequency is greatest where rainfall lies between100 and 200 mm. Except under unusual conditions,dust storms in both hemispheres tend to have theirgreatest frequency of occurrence in late springand early summer. This pattern corresponds to thechangeover from minimum rainfall in early spring tointense evaporation in summer. In Australia, however,the pattern also represents the most likely period ofdrought associated with the Southern Oscillation.

Where cold fronts form dust storms, cold air forceswarm air up rapidly, giving a convex, lobed appearanceto the front of the dust storm. As the air is forced up, itsets up a vortex that tends to depress the top of thecold front and further enhance the lobe-like appear-ance of the storm. The classic image is one of a sharpwall of sediment moving across the landscape withturbulent overturning of dense clouds of dust (seeFigure 3.29). Wind speeds can obtain consistentvelocities exceeding 60 km hr-1 or 30 m s-1. Materialthe size of coarse sand can be moved via saltation inthe first tens of centimeters above the ground surface.Fine sand can be moved within 2 m of the ground,while silt-sized particles can be carried to heights inexcess of 1.5 km. Clay-sized particles can be suspendedthroughout the depth of the troposphere and carriedthousands of kilometers. Most of the dust is suspended


Fig. 3.29 Dust storm from the mallee country of western Victoria aboutto bear down on the city of Melbourne for only the secondtime in a century, 8 February 1983 (photograph courtesythe Australian Bureau of Meteorology).

in the lower 1 km of the atmosphere. Concentrationscan be such that noonday visibility is reduced to zero.Gravel-sized material cannot be transported by winds,so it is left behind forming a stony or reg desert. Sand-sized particles in saltation are also left behind by theremoval of silt and clay. These sand deposits becomenon-cohesive and mobile. Once the fine material hasbeen removed from topsoil its cohesiveness andfertility cannot easily be re-established. The removal offines often means the removal of organics as well. Thisis why the initial occurrence of dust storms in an areasignifies the removal of soil fertility and, over extendedperiods, the desertification of marginal semi-arid areas.

The role of dust

Dust in storms diffuses into the upper tropospherewhere it absorbs heat during the day, but blocksincoming solar radiation. At the top of the atmosphere,the dust blocks incoming solar radiation by –0.25 W m-2

and absorbs outgoing long-wave radiation by+0.34 W m-2. The global mean cooling of dust, at theground, is –0.96 W m-2, rising above –8.00 W m-2 oversuch arid locations as the Arabian Peninsula andadjacent sea (Figure 3.30). Because of dust, the groundsurface is slightly cooler than normal during daytime,resulting in less convection. At night, high-altitude dust

radiates long wave radiation in the upper atmosphere,cooling air and causing it to sink. At the ground,however, the dust traps in long-wave radiation thatwould normally escape from the Earth’s surface,causing the air above the surface to remain warmerthan normal, thus preventing dew formation. Thesinking air aloft leads to conditions of stability, whilethe lack of dew keeps the ground surface dry andfriable. These conditions favor aridity.

Human activity is also a major contributor of dust tothe atmosphere. This effect was espoused in the 1970sduring a debate about global cooling. At this time, itwas postulated that humans were increasing the dustcontent of the atmosphere at an accelerating ratethrough industrial and agricultural activity. In the1970s, archaeological evidence was used to show thatsome of the earliest civilizations and societies reliantupon agriculture in semi-arid regions collapsedbecause their poor land management practices led toincreased atmospheric dust and aridity. Insidiously,arable topsoil was blown away, and the slow processof marginalization or desertification of semi-aridland took place. The long-term social consequencesinvolved the dislocation of communities, famine, andultimately the destruction of civilizations dependentupon such areas for their existence. This was the fate of

80 Climatic Hazards

Pampas

GreatPlains

0.1

0.1

-2

1

1

1

0.1

0.1

10 -2

Kalahari

Sahara

-2Simpson

10

1

10 1

-4-8

-4 Arabian

-2

-4

Gobi

Kasakhstan

0.1

Main dust trajectoriesAnnual amount of dust deposited in the world's oceans

0.1-1 tonnes km2 yr-1 >10 tonnes km2 yr-11-10 tonnes km2 yr-1

Surface radiation forcing due to atmospheric soil dust (W m-2)-2

Fig. 3.30 Trajectories of dust storms and annual amounts of atmospheric dust transported to the world’s oceans (based on Middleton et al., 1986; Goudie,1983; and Pearce, 1994). Also mapped is the radiative cooling effect at the Earth’s surface due to soil dust (from Tegen et al., 1996).

early agricultural civilizations in Mesopotamia, depen-dent upon the Euphrates and Tigris Rivers for theirwater. Deserts now occur in humid regions and aremaintained by this human-induced dust effect. TheRajputana Desert on the Indian subcontinent exempli-fies this best. Historically, the Rajputana was one of thecradles of civilization, with a well-developed agrariansociety based on irrigation around the Indus River.Successive cultures occupied the desert, but each timethey collapsed. Monsoons affect the Rajputana area;however, rainfall generally amounts to less than400 mm yr-1. At the time of the development ofcivilization by the Harappan culture, 4500 BP, rainfallexceeded 600 mm yr-1. The dust put into the atmos-phere by agricultural activities inhibited convectionand rainfall, leading to collapse of this, and successive,civilizations. Extrapolated to the modern world, thesecase studies imply that anthropogenic dust fromindustrialization and intensifying agriculture is creatingan enhanced global cooling effect.

Figure 3.30 also maps the tracks of the main duststorms worldwide and the annual amount of dustcurrently being deposited in the world’s oceans. Abroad band of dust, amounting to over 10 tonnes km-2,occurs up to 1000 km off the east coast of Asia. Asimilar swathe extends south-east of the Indiansubcontinent and through South-East Asia. High dusttransport rates also occur off the west coast of northernAfrica, associated with thunderstorms generated infront of mobile polar highs as these sweep along1000 km-wide fronts. The above areas are notnecessarily downwind of deserts or cultivated semi-arid regions. They represent the global signature ofdust output due to both industrial and agriculturalactivities. Carried with this dust are substantiallyincreased amounts of iron, both from industrial activityand soil weathering. In addition to atmospheric coolingcaused directly by dust, windblown material may beindirectly cooling the atmosphere by ‘fertilizing’ thegrowth of phytoplankton and, with it, increasingdimethylsulphides and cloudiness in the marine atmos-phere. Dust fertilization of algae may be responsiblefor red tides off Florida – with the incidence increasingconcomitantly with droughts in northern Africa.

Dust storms also carry heavy metals, fungi,bacteria, and viruses. These particles, along with toxicpollutants from the burning of rubbish, and pesticidesand herbicides used in semi-arid regions, are absorbedon dust particles and transported across the oceans. The

problem is being exacerbated by the exposure of sedi-ments on drying lake beds such as the Aral Sea in centralAsia and Lake Chad in northern Africa. Most importantis the transport of pathogens. Following major Africandust storms, sugar cane rust, coffee rust and banana leafspot originating in Africa have been found in the air overthe Caribbean. The fungus Aspergillus sydowii, foundin African dust, is also linked to sea fan disease that killscorals in the Mediterranean and Caribbean seas.Airborne dust may also be responsible for asthma, itsincreased incidence in some countries possibly follow-ing the increased volumes of dust originating from semi-arid countries due to human activities.

Frequency of dust storms

The frequency of dust storms cannot be overempha-sized. Between 1951 and 1955, central Asia recorded3882 dust storms. Turkmenistan in central Asiarecorded 9270 dust storms over a 25-year period. Inthe United States Midwest, as many as ten dust stormsor more per month occurred between 1933 and 1938.Dust storms occur, geographically, in the southern aridzones of Siberia, European Russia, Ukraine, westernEurope, western United States, north-western China,and northern Africa. Frequencies in the Sahara rangefrom 2–4 per year in Nigeria to 10–20 per year in theSudan. In Iraq, the incidence of dust storms rises toover 30 per year. Many parts of central Asia averageabout 20 storms per year, with the frequency rising toover 60 per year in Turkmenistan on the eastern side ofthe Caspian Sea. North-western India, adjacent to theThar Desert, is affected by as many as 17 events peryear, with the frequency increasing northward intoAfghanistan to over 70 storms per year. The highestfrequency of dust storms occurs in the Gobi Desert ofChina, which averages 100–174 dust storms annuallyunder the influence of the Mongolian–Siberian high-pressure cell. Dust has been blown as far as Beijingand Shanghai, where as many as four storms per yearhave been recorded. In Australia, more than five duststorms per year are recorded around Shark Bay, inWestern Australia, and Alice Springs, north of theSimpson Desert.

It is debatable whether the incidence of dust stormsis increasing because of growing populations. Chinesedata show no such increase; but intense cultivationof the United States Great Plains in the 1920s, theopening up of the ‘Virgin Lands’ in the former SovietUnion in the 1950s, and the Sahel drought of the


1960s–1970s, all led to an increase in the frequency ofdust storms in these locations. While the latter tworegions have produced the most dramatic recentincrease in dust storm activity because of human activ-ities, the best-documented effect has been on theUnited States Great Plains during the dust bowl yearsof the 1930s Great Depression. Here, it can be shownthat dust storms have a cyclic occurrence. In the 1870s,dust storms during drought drove out the first group offarmers. While no major dust storm period wasrecorded until the 1930s, drought has tended to occursynchronously with maxima in the 18.6-year MN lunarcycle. Dust storm degradation of arable land alsooccurred in the mid-1950s and mid-1970s. While thedust bowl years of the 1930s have become part ofAmerican folklore, the drought years of the mid 1950sand 1970s actually produced more soil damage. This isdespite the fact that reclamation and rehabilitationprograms had been initiated by the federal govern-ment in response to the 1930s dust bowl. Theseprograms included the return of marginal land tograssland, and the use of strip farming, crop rotation,and mulching practices. These techniques build up soilmoisture and nutrients while protecting the surfacesoil from wind deflation. In most cases, the dust stormsfollowed a period of rapid crop expansion duringfavorable times of rainfall and commodity prices.

Major storm events

There have been many notable dust storms. Forinstance, in April 1928 a dust storm affected thewhole of the Ukrainian steppe, an area in excess of1 million km2. Up to 15 million tonnes of black cher-nozem soil were removed and deposited over an area of6 million km2 in Romania and Poland. In the affectedarea, soil was eroded to a depth of 12–25 cm in someplaces. Particles 0.02–0.5 mm in size were entrained bythe wind. In March 1901, 1 million tonnes of red dustfrom the Sahara were spread over an area fromwestern Europe to the Ural Mountains in Russia. Inthe 1930s, the dust bowl of the United States Midwestproduced some of the most dramatic dust storms of thetwentieth century. Their effect was enhanced bythe fact that the area had not previously experiencedsuch extreme events. Most of the storms occurred inlate winter–early spring, just as the snow cover hadmelted, and the frozen ground inhibiting soil deflationhad thawed. However, at this time of year, the polarwesterlies were still strong. The most severe stormsoccurred between 1933 and 1938 on the southernPlains and between 1933 and 1936 on the northernPlains. The number and extent of storms during theworst period of March 1936 is shown in Figure 3.31.These storms covered an area from the Gulf of Mexico

82 Climatic Hazards

4

6810

1214

1618

2022

2

≥ 10 Dust storms ≥ 20 Dust storms

2

0 1000 km

Fig. 3.31 Extent and frequency of dust storms in March 1936 in the United States (after Lockeretz, 1978).

north into Canada. During this month, the panhandleof Texas and Oklahoma experienced dust storms on28 out of 31 days. At one point, as the United StatesCongress debated the issue in Washington, dustoriginating 2000 km to the west drifted over thecapital. By 1937, 43 per cent or 2.8 million hectares ofthe arable land at the center of the dust bowl had beenseriously depleted. In some counties over 50 per centof the farm population went bankrupt and onto relief.

CONCLUDING COMMENTS

The phenomena described in this chapter account formost of the deaths, economic disruption, and societalchange due to climatic hazards. While it is relativelyeasy today to model the global pressure changesdescribed in the previous chapter, the storm systems inthis chapter pose a problem because they containelements of chaotic systems. East coast bombs are oneof the best examples of this, while the Great Ice Stormof 1998 in eastern North America illustrates the unpre-dictability of many storms. It is thus relatively simple topredict temperature increases in a ‘Greenhouse’-warmed world fifty years from now using sophisticatedcomputer modeling; but it is impossible to predict sub-sequent changes in the magnitude and frequency oftropical cyclones, mid-latitude depressions or icestorms stalled at the boundary between air masses. TheGreat Ice Storm of 1998 also illustrates how vulnerablemajor cities are to climatic phenomenon. The world’slargest cities of economic importance operate on elec-tricity. Cut off power supplies and these cities cease tooperate at any economically viable level for days, weeksor months. However, such disruptions are amenable toengineering solutions. Storms similar to those thatstruck northern Europe in the Middle Ages are not.Were a storm the size and intensity of the All SaintsDay storm of November 1570 to recur today, theeffects on modern European economies would becatastrophic. At present, the types of large stormsdescribed in this chapter are not so deadly as they havebeen historically. One of the main reasons for this is thefact that emergency services in many countries aregeared to respond and militate against extreme deathtolls. This has been a trend in developed countries suchas the United States as well as impoverished ones suchas Bangladesh. High death tolls are now more likely tooccur during localized events such as thunderstorms.These localized storms will be discussed in the follow-ing chapter.


American Meteorological Society 1986. Is the United States headedfor hurricane disaster?. Bulletin American MeteorologicalSociety 67(5): 537–538.

Ananthaswamy, V. 2003. Historic storms live again. New Scientist27 September, pp. 14–15.

Anthes, R. A. 1982. Tropical cyclones: their evolution, structureand effects. American Meteorological Society MeteorologicalMonograph 19 No. 41.

Australian Bureau of Meteorology 1977. Report on Cyclone Tracy,December 1974. AGPS, Canberra.

Australian Bureau of Meteorology 2001. Preliminary report onmeteorological aspects of the 1998 Sydney to Hobart yacht race.<http://www.bom.gov.au/inside/services_policy/marine/sydney_hobart/contents.shtml>

Australian Bureau of Meteorology 2003. Tropical cyclone severitycategories. <http://www.bom.gov.au/info/cyclone/#severity>

Blong, R.J. and Johnson, R.W. 1986. Geological hazards in thesouthwest Pacific and southeast Asian region; identification,assessment, and impact. Bureau Mineral Resources JournalAustralian Geology and Geophysics 10: 1–15.

Bresch, D.N., Bisping, M. and Lemcke, G. 2000. Storm overEurope: An Underestimated Risk. Swiss Re, Zurich.

Bryant, E.A. 1983. Coastal erosion and beach accretion Stanwell Parkbeach, N.S.W., 1890–1980. Australian Geographer 15: 382–390.

Bryant, E.A. and Kidd, R.W. 1975. Beach erosion, May–June, 1974,Central and South Coast, NSW. Search 6(11–12): 511–513.

Burton, I., Kates, R.W. and White, G.F. 1978. The Environment asHazard. Oxford University Press, NY.

Carter, R.W.G. 1987. Man’s response to sea-level change. In Devoy,R.J.N. (ed.) Sea Surface Studies: A Global Review. Croom Helm,London, pp. 464–498.

Chang, C.P., Liu, C.H. and Kuo, H.C. 2003. Typhoon Vamei: Anequatorial tropical cyclone formation. Geophysical ResearchLetters 30(3) 10.1029/2002GL016365.

Coastal Engineering Research Center 1977. Shore ProtectionManual (3 vols). United States Army, Washington.

Cornell, J. 1976. The Great International Disaster Book. Scribner’s,NY.

Dolan, R. and Hayden, B. 1983. Patterns and prediction of shorelinechange. In Komar, P.D. (ed.) CRC Handbook of Coastal Processesand Erosion. CRC Press, Boca Raton, Florida, pp. 123–150.

Donnelly, J.P., Roll, S., Wengren, M., Butler, J., Lederer, R. andWebb, T. III 2001. Sedimentary evidence of intense hurricanestrikes from New Jersey. Geology 29: 615–618.

Dvorak, V.F. 1975. Tropical cyclone intensity analysis and forecastingfrom satellite imagery. Monthly Weather Review 103: 420–430.

Eagleman, J.R. 1983. Severe and Unusual Weather. Van NostrandReinhold, NY.

Environment Canada 1984. Atlantic seasonal hurricane frequencyPart I: El Niño and 30 mb quasi-biennial oscillation influences.Monthly Weather Review 112: 1649–1667.

Environment Canada 1998. The worst ice storm in Canadian history<http://www.msc-smc.ec.gc.ca/media/icestorm98/icestorm98_the_worst_e.cfm>


Environment Canada 2002. Wind chill calculation chart. <http://www.msc-smc.ec.gc.ca/education/windchill/windchill_calculator_e.cfm>

Fell, N. 2002–2003. Piste lightning. New Scientist 21 December2002: 42; 25 January 2003: 27.

Gore, R. 1993. Andrew aftermath. National Geographic 183(4): 2–37.Goudie, A.S. 1983. Dust storms in space and time. Progress in

Physical Geography 7: 503–530.Gray, W.M. 1975. Tropical Cyclone Genesis. Atmospheric Science

Paper No. 4, Department of Atmospheric Science, ColoradoState University, Fort Collins, Colorado.

Gray, W.M. 1984. Atlantic seasonal hurricane frequency Part 1: ElNiño and 30 mb quasi-biennial oscillation influences. MonthlyWeather Review 112: 1649–1667.

Gray, W.M., Landsea, C.W., Mielke, P.W. and Berry, K.J. 1994.Predicting Atlantic Basin seasonal tropical cyclone activity by1 June. Weather and Forecasting 9: 103–115.

Griffin, D.W., Kellogg, C.A. Garrison, V.H., and Shinn, E.A. 2002.The global transport of dust. American Scientist 90: 227–235.

Harris, D.L. 1956. Some problems involved in the study of stormsurges. United States Weather Bureau, National HurricaneResearch Project Report, No. 54. Washington.

Hawkins, H.F. and Imbembo, S.M. 1976. The structure of a small,intense hurricane, Inez 1966. Monthly Weather Review 104:418–425.

Henson, R. 2002. Cold rush scientists search for an index that fitsthe chill. Weatherwise 55 January/February pp. 14–20.

Holland, G.J., Lynch, A.H. and Leslie, L.M. 1988. Australian east-coast cyclones. Part I: Overview and case study. Monthly WeatherReview 115: 3024–3036.

Holthouse, H. 1986. Cyclone: A Century of Cyclonic Destruction.Angus and Robertson, Sydney.

Jones, K.F. and Mulherin, N.D. 1998. An evaluation of the severityof the January 1998 ice storm in northern New England. USArmy Cold Regions Research and Engineering Laboratory<http://www.crrel.usace.army.mil/techpub/CRREL_Reports/reports/IceStorm98.pdf>


Leopold, L.B., Wolman, M.G. and Miller, J.P. 1964. FluvialProcesses in Geomorphology. Freeman, San Francisco.

Linacre, E. and Hobbs, J. 1977. The Australian Climatic Environment.Wiley, Brisbane.

Lockeretz, W. 1981. The lessons of the dust bowl. In Skinner, B.J.(ed.) 1981 Use and Misuse of Earth’s Surface. Kaufmann,Los Altos, California, pp. 140–149.

Lott, N. 1993. The big one! A review of the March 12–14, 1993‘Storm of the Century’. National Climatic Data Center ResearchCustomer Service Group Technical Report 93-01 <ftp://ftp.ncdc.noaa.gov/pub/data/techrpts/tr9301/tr9301.pdf>

Lott, N., McCown, S., Graumann, A. and Ross, T. 1999. Mitch: Thedeadliest Atlantic hurricane since 1780. United States National

Climatic Data Center <http://www.ncdc.noaa.gov/oa/reports/mitch/mitch.html#DAMAGE>

McCowan, S. 2001. The Perfect Storm. US National ClimaticData Center, <http://www.ncdc.noaa.gov/oa/satellite/satelliteseye/cyclones/pfctstorm91/pfctstorm.html>

Middleton, N.J., Goudie, A.S. and Wells, G.L. 1986. The frequencyand source areas of dust storms. In Nickling, W.G. (ed.) AeolianGeomorphology. Allen and Unwin, London, pp. 237–260.

Milne, A. 1986. Floodshock: The Drowning of Planet Earth. Sutton,Gloucester.

Nalivkin, D.V. 1983. Hurricanes, Storms and Tornadoes. Balkema,Rotterdam.

NOAA 2002. NOAA composite satellite image of HurricaneAndrew’s path August 22–25, 1992. <http://www.noaa.gov/images/andrew-comp0822-2592.jpg>

NOAA 2003. Image ID: wea00544, historic NWS collection.http://www.photolib.noaa.gov/historic/nws/wea00544.htm

Orville, R.E. 1993 Cloud-to-ground lightning in the Blizzard of ’93.Geophysical Research Letters 20: 1367–1370.

Pearce, F. 1994. Not warming, but cooling. New Scientist 9 July: 37–41.Rooney, J. F. 1973. The urban snow hazard in the United States: an

appraisal of disruption. In McBoyle, G. (ed.) Climate in Review.Houghton Mifflin, Boston, pp. 294–307.

Sanders, F. and Gyakum, J.R. 1980. Synoptic-dynamic climatologyof the ‘Bomb’. Monthly Weather Review 108: 1589–1606.

Santana, S. 2002. Remembering Andrew. Weatherwise 55 July/August: 14–20.

Sheets, R.C. 1980. Some aspects of tropical cyclone modification.Australian Meteorological Magazine 27: 259–286.

Simpson, R. (ed.) 2003. Hurricane! Coping with Disaster. AmericanGeophysical Union, Washington.

Simpson, R.H. and Riehl, H. 1981. The Hurricane and Its Impact.Louisiana State University Press, Baton Rouge.

Tegen, I., Lacis, A.A., and Fung, I. 1996. The influence on climateforcing of mineral aerosols from disturbed soils. Nature 380:419–422.

The Disaster Center 1998. Hurricane Mitch reports. <http://www.disastercenter.com/hurricmr.htm>

Wemelsfelder, P.J. 1961. On the use of frequency curves of stormfloods. Proceedings of the Seventh Conference on CoastalEngineering, The Engineering Foundation Council on WaveResearch, ASCE, NY, pp. 617–632.

Western, J.S. and Milne, G. 1979. Some social effects of a naturalhazard: Darwin residents and cyclone ‘Tracy’. In Heathcote,R.L. and Thom, B.G. (eds) Natural Hazards in Australia.Australian Academy of Science, Canberra, pp. 488–502.

Whipple, A.B.C. 1982. Storm. Time-Life Books, Amsterdam.Whittow, J. 1980. Disasters: The Anatomy of Environmental

Hazards. Penguin, Harmondsworth.Wiegel, R. L. 1964. Oceanographical Engineering. Prentice Hall,

Englewood Cliffs, N.J.

84 Climatic Hazards

INTRODUCTION

The previous chapter was mainly concerned with theeffects of strong winds generated by secondaryfeatures of general air circulation. These winds wereassociated with the development of low-pressure cellsspanning areas of 10 000–100 000 km-2. While tropicalcyclones and extra-tropical depressions produce someof the strongest winds and highest amounts of precipi-tation over the widest areas, they are by no means theonly source of high winds or heavy precipitation. Thesevalues can be matched by thunderstorms, which coverno more than 500 km2 in area and rarely travel morethan 100–200 km before dissipating. They can producehigh-magnitude, short-period rainfalls leading to flashflooding. Thunderstorms are also associated with awide range of climatic phenomena such as lightning,hail and tornadoes that bring death and destruction.Tornadoes generate the highest wind speeds and canproduce localized wind damage just as severe as thatproduced by a tropical cyclone. This chapter willexamine first the development and structure ofthunderstorms resulting in lightning and hail. This isfollowed by a description of tornadoes and the majordisasters associated with them. The chapter concludeswith a discussion of warning and response to thetornado threat, an aspect that has been responsiblefor decreasing death tolls throughout the twentiethcentury.

THUNDERSTORMS,LIGHTNING AND HAIL

Thunderstorms(Whipple, 1982; Eagleman, 1983)

Thunderstorms are a common feature of the Earth’senvironment. There are about 1800–2000 stormsper hour or 44 000 per day. In tropical regions, theyoccur daily in the wet season. However, in theseregions thunderstorms may not represent a hazardbecause they do not intensify. As thunderstorms repre-sent localized areas of instability, their intensity isdependent upon factors that increase this instability.On a world scale, instability is usually defined by therate at which the base of the atmosphere is heated byincoming solar radiation, especially where evaporationat the ground and condensation in the atmosphereboth occur. In this case, the saturated adiabatic lapserate prevails. Under these conditions, large quantitiesof heat energy (2400 joules gm-1 of liquid water thatcondenses) are released into the atmosphere. Thisprocess causes convective instability, which terminatesonly when the source of moisture is removed.

On a localized scale, the degree of instability isalso dependent upon topography and atmosphericconditions such as convergence. If air is forced over ahill, then this may be the impetus required to initiateconvective instability. Convergence of air masses by

Localized StormsC H A P T E R 4

topography, or by the spatial arrangement of pressurepatterns, can also initiate uplift. The most likely occur-rence of instability takes place along cold fronts, mainlythe polar front where it intrudes into moist tropical air.Such intrusions are common over the United StatesMidwest and along the south-eastern part of Australia.

Figure 4.1 illustrates all of the above processes overthe United States. The highest incidence of thunder-storms – greater than 60 per year – occurs in Florida,which is dominated by tropical air masses, especially insummer. A smaller area of high thunderstorm fre-quency over the Great Plains corresponds to an areaaffected by topographic uplift. From here across tothe east coast, thunderstorm activity occurs becauseof the interaction of polar and tropical air. In Australia,the distribution of thunderstorms follows a similarpattern (Figure 4.2). The greatest intensity of storms –greater than 30 per year – occurs in the tropics andalong the eastern Divide, where topographic uplift isfavored. In terms of forcing thunderstorm develop-ment, the passage of cold fronts plays a minor role inAustralia compared to the United States. While theincidence of storms in Australia is less than inthe United States, thunderstorms over tropical

Australia are some of the most intense in the world.They have been referred to as stratospheric fountainsbecause their associated convection is strong enough topierce the tropopause and inject air from the tropo-sphere into the normally isolated stratosphere. Thisprocess is highly effective during periods when Walkercirculation is ‘turned on’.

Thunderstorms globally generate electricity betweena negative ground surface, and a more positivelycharged electrosphere at about 50 km elevation. Tropicalthunderstorms are responsible for most of this upwardcurrent of electricity, each averaging about 1000amperes (A). This current creates a positive charge inthe ionosphere reaching 250 kV with respective to theground. There is, thus, a voltage gradient with altitude.A steepening of this gradient enhances thunderstormactivity. This can be accomplished by increasing theamount of atmospheric aerosols such as smoke frombushfires, pollution, or thermonuclear devices. Thegradient can also be increased by solar activity throughthe interference of the solar wind with the Earth’smagnetic field. During sunspot activity, the electro-sphere becomes more positively charged, increasingthe frequency of thunderstorms. Thus thunderstorm

86 Climatic Hazards

1020 30

30 30

30

20

30

40

5060

708070605040

30

40

50

3020

10

10

40

50

70–79days

>80days

50–59days

60–69days

<40days

40–49days

0 1000 km

20

Fig. 4.1 Annual frequency of thunder days in the United States (from Eagleman, 1983).

frequency strongly parallels the 11-year geomagneticcycle.

Because thunderstorms in the United States are soclosely linked to tornado occurrence, the threat ofthunderstorms is forecast daily by evaluating thedegree of atmospheric instability. This is calculated at93 stations using air temperature measured at the500 hPa level, and comparing this to a theoretical valueassuming that air has been forced up at the forecasttemperature and humidity for that day. A difference intemperature of only 4°C has been associated withtornado development. The largest tornado occurrencein the United States (when over 148 tornadoes wererecorded in 11 states on 3–4 April 1974) was associatedwith a predicted temperature difference of 6°C in the

Mississippi Valley. As well, the movement of thunder-storms is predictable. Most move in the direction of,and to the right of, the mean wind. The effect is duemainly to the strength of cyclonic rotation in thethunderstorm. In the United States Midwest, thunder-storm cells also tend to originate in the same placebecause of topographic effects. Thunderstorms, liketornadoes, tend to follow lower lying topography.Cities, which develop an urban heat island, also attractthunderstorms. The urban heat island initiatesupdrafts, which then draw in any thunderstormsdeveloping in the area. This phenomenon has beenobserved, for example, in both Kansas City, UnitedStates, and London, England (see Figure 4.3 for thelocation of major placenames mentioned in this

87Localized Storms

0 1000 km

<40days

40–49days

50–59days

60–69days

70–79days

>80days

20

10

10

10

10 10

20

30

40

40

30

20

20

4030

20

304050

30

40

50607080

807060

50

50

30

20

403020

10

Fig. 4.2 Average annual thunder days in Australia (based on Oliver, 1986).

chapter). Thunderstorms can also be enhanced by thewarmer temperatures and increased surface roughnessfound in urban areas. For example, the amount ofrainfall from thunderstorms in Sydney, Australia,increases with increase in the percentage of landcovered by concrete and buildings. The highest rain-falls occur over Sydney’s central business district whereturbulence increases over a regular grid of skyscrapersand streets.

Thunderstorms accompanied by their attendantphenomena have produced some colossal damage. InSydney, on 21 January 1991, a thunderstorm sweptuphill through the northern suburbs of the cityaccompanied by wind gusts of 230 km hr-1, hail upto 7 cm in diameter and rainfall intensities up to205 mm hr-1. In twenty minutes, 50 000 mature treeswere destroyed, uprooting gas and water mains andbringing down 140 km of power lines concentrated inan area of only 100 km2. Over 7000 homes weredamaged by wind or fallen trees. The storm cost$A560 million and the clean-up involved 4000 utilitypersonnel and emergency workers.

Lightning(Geophysics Study Committee, 1987; Black & Hallett, 1998;Anagnostou et al., 2002)

Thunder is generated when lightning briefly heatsthe air to temperatures of 30 000 K or five times thesurface temperature of the sun. As a result, the width

of the conducting channel expands from a fewmillimetres to a few centimetres in a few millionths ofa second. This expansion produces a shock wave andnoise. A build-up of charged ions in the thundercloudcauses lightning (Figure 4.4). The process is known ascharge separation or polarization and is aided byupdrafts and the presence of ice or graupel. Graupelconsists of aggregated ice or hail that is coated in super-cooled water. If this freezes, expansion causes the outerlayer of the graupel to shatter into small pieces that arecarried in updrafts at speeds of 50 m s-1. Shatteringoccurs at temperatures between –3 and –10°C. Theseshattered particles tend to carry a positive chargeaway from the larger graupel particle leaving behind anegative charge. Thus, positive charges migrate to thetop of the cloud leaving negative charges at the bottom.Positive charges can also build up in a shadow zone atthe ground under the thundercloud. The latter build-up, with its concomitant depletion of negative charge,has been postulated as the reason for an increasein suicides, murders and traffic accidents beforethe occurrence of thunderstorms. This distribution incharge produces an electrical gradient, both within thecloud, and between the base of the cloud andthe ground. If the gradient reaches a critical threshold,it is reduced by sparking between the area of negativecharge and the two areas of positive charge build-up.The amount of energy involved in a lightning dischargevaries greatly, but is quite small compared to the

88 Climatic Hazards

Venice

Topeka

IrvingKansas City

S. Dakota

Edmonton

MassacuhsettsNew York State

FloridaJudsonia

Rock SpringsOklahoma City

Boulder

Xenia

LondonRussian steppes

Bangladesh

Queensland

OrangeSydney

Murrumbidgee

Munich

Chartres


volume of air included in the thunderstorm. Forinstance, the energy density of the charge is 100 milliontimes smaller than that produced in the same volume ofthe air by convection or latent heat due to condensa-tion. However, the total energy of the bolt, because it isconcentrated in such a small channel, can reach 9 � 108 J. Under extreme wind shear, the area ofpositive charge at the top of the atmosphere may movedownwind faster than the central core of negativecharge. If negative charge builds up at the Earth’ssurface, a discharge can take place from the top of theatmosphere to the ground with ten times the energy ofa normal lightning flash. Often the imbalance in chargeadds an attractive force to moisture in the cloud. Theopposing charges tend to hold moisture in the cloud. Alightning spark may weaken this hold, and lightningflashes are often accompanied by localized increases inprecipitation.

Initially, the sparking originates in the area ofnegative charge at the base of the cloud, and surges injerky steps towards the top of the cloud or to theground at the rate of 500 km s-1. Once a leader channel(no more than a few millimeters wide but up to 100 kmlong) has been established, a current races along thischannel at a speed of 100 000 km s-1. Currents in the

channel can reach 300 000 A and be driven by a voltagepotential of hundreds of millions of volts between theground and the cloud. On average, a channel, onceestablished, lasts long enough to carry three distinctpulses of lightning between the positive and negativeionized areas. There are on average 100–300 flashes persecond globally, representing a continuous power flowof about 4000 million kilowatts in the global electricitycircuit. On average, in the United States, the greatestnumber of cloud-to-ground lightning flashes per year –more than 10 km-2 – occurs in Florida; however, thegreatest daily incidence of discharges to date happenedin June 1984, when 50 836 flashes were recorded –mainly in and around New York State. The ‘Storm ofthe Century’ of 12–15 March 1993 mentioned in theprevious chapter generated 59 000 flashes across theAtlantic seaboard, with a peak rate exceeding 5100flashes per hour. Finally, lightning frequency relates toincreases in temperature and pollutants. Urban areaswith large heat islands, more dust, and increasedpollution have more cloud and lightning. This impliesthat regions subject to recent global warming will besubject to more lightning during thunderstorms.

Lightning also generates high-frequency radionoise, gamma rays and X-ray bursts. The radio noise

89Localized Storms

Wind shearing

Direction ofstorm

movement

Positive chargezone

Negative charge zone

Positively charged shadow

Updrafts

Fig. 4.4 Schematic representation of the separation of charged water droplets by updrafts in a thundercloud, giving rise to lightning.

can be monitored by satellites, and located to ahigh degree of accuracy. A network of automaticdirection-finding, wide-band VHF, magnetic receivershas been established, across North America and othercontinents, to locate and measure the direction,polarity, intensity, and stroke patterns of lightning.In the United States, 100 stations in the NationalLightning Detection Network measure the exact timeand direction of electromagnetic energy burstsproduced by lightning. Lightning can also be locatedby measuring the difference in the time of arrival ofthe electromagnetic wave between two sensors. Thelatter method uses radio noise produced by lightning inthe 7–16 kHz very low frequency (VLF) spectrum. InEurope, a network of six very low frequency receivershas been established in conjunction with globalpositioning systems and signal processing algorithmsto locate lightning flashes globally. This network, inreal time, can detect lightning within severe stormsresulting in flash flooding. This can be verified usingthe Lightning Imaging Sensor (LIS) on the TropicalRainfall Measuring Mission (TRMM) satellite.

Lightning is a localized, repetitive hazard. Becausethe lightning discharge interacts with the best con-ductors on the ground, tall objects such as radio andtelevision towers are continually struck. When light-ning strikes a building or house, fire is likely as thelightning grounds itself. However, this effect, whilenoticeable, is not severe. In the United States, about$US25 million in damage a year is attributable tolightning. Insurance companies consider this a lowrisk. However, because most of the damage can beavoided by the use of properly grounded lightning rodson buildings, insurance companies in North Americaeither will not insure, or will charge higher rates for,unprotected buildings. Lightning, besides ignitingforest fires, is also a severe hazard to trees and othervegetation. In North America, tree mortality in someareas can reach 0.7–1.0 per cent per year because oflightning strikes, and up to 70 per cent of tree damageis directly attributable to lightning. Scorching is evenmore widespread. A single lightning bolt can affect anarea 0.1–10 hectares in size, inducing physiologicaltrauma in plants and triggering die-offs in crops andstands of trees. Lightning can also be beneficial tovegetation, in that oxygen and nitrogen are combinedand then dissolved in rain to form a nitrogen fertilizer.

The greatest threat from lightning is the fact that thehigh voltage kills people. Lightning enters a person

through the body’s orifices, flowing along blood vesselsand the cerebral spinal fluid pathways. The currentbuilds up in the brain and then discharges along theskin, leading to a dramatic drop in charge insidethe body. This causes failure of the cardiac andrespiratory systems linked to the brainstem. Victims oflightning die from suffocation because breathing stops.About 40 per cent more people die in the UnitedStates and Australia from lightning than from any othermeteorological hazard. This incidence is decreasingover time despite a growth in population. Decreasedfatalities reflect increased urbanization where protec-tion is provided by tall buildings, an increase in safetyeducation, and the improved availability of medicalservices. Despite the misconception that standingunder a tree during a lightning storm is playing withdeath, statistics indicate that the number of peoplewho die from lightning-induced charges inside housesor barns is equal to the number of those who dieoutside standing under trees. This statistic may merelyreflect the fact that most people know about the risksof standing under trees during a storm, but not aboutother hazardous activities. In Australia, Telecom had tomount a major campaign in 1983 warning subscribersabout the risk of using telephones during thunder-storms. Risky behavior also includes standing nearelectrical wiring or metal piping (including toilets)inside homes, or standing at the edges of forests, nearmountains or on beaches during thunderstorms. In thelatter cases, an observer may be the first tall objectavailable for lightning discharge for some kilometers.

Hail(Eagleman, 1983; Yeo et al., 1999)

The formation of hail depends upon the strength ofupdrafts, which in turn depends upon the amountof surface heating. The likelihood of hail increases withmore intense heating at the Earth’s surface and coolertemperatures aloft. The surface heating produces theupdrafts, while the cooling ensures hail formation.Almost all hailstorms form along a cold frontshepherded aloft by the jet stream. The jet streamprovides the mechanism for creating updrafts. Windshear, which represents a large change in wind speedover a short altitude, also facilitates hail formation.The degree of uplift in a thunderstorm can be assessedby the height to which the storm grows. In colderclimates, however, this height is less than in moretemperate or subtropical latitudes.

90 Climatic Hazards

Locations receiving hail do not coincide with areasof maximum thunderstorm or tornado development.Figure 4.5 illustrates the annual frequency of hail-storms in the United States. Areas with the highestincidence of hailstorms on this map do not correspondwell to those areas of maximum thunderstorm activityshown on Figure 4.1. The middle of the Great Plains,underlying the zone of seasonal migration of the jetstream, is the most frequently affected region. Mosthailstorms in the United States occur in late spring andearly summer as the jet stream moves north. A similarseasonal tendency occurs in Australia.

The size of hail is a direct function of the severityand size of the thunderstorm. General physics indi-cates that a hailstone 2–3 cm in diameter requiresupdraft velocities in excess of 96 km hr-1 to keep it insuspension. An 8 cm hailstone would require windspeeds in excess of 200 km hr-1, and the largest stonesmeasured (>13 cm) require wind velocities greaterthan 375 km hr-l. If graupel is present, hailstones cangrow quickly to large sizes as pieces freeze together(Figure 4.6a). Multiple-layer hailstones indicate thatthe stone has been repetitively lifted up into the thun-derstorm (Figure 4.6b). Ice accumulates less when the

stone is being uplifted because it often passes throughair cooler than the freezing point. The ice in thisprocess can be recognized from its milky appearance.When the stone descends again, water freezes rapidlyto the surface forming a clear layer.

Because hail can do tremendous damage to crops,there has been substantial research on hail suppressionin wheat- and corn-growing areas. Supercooled water(water cooled below the freezing point but still liquid)is very conducive to hail formation, and attempts havebeen made to precipitate this water out of the atmo-sphere before large hailstones have a chance to form.This process is accomplished by seeding clouds con-taining supercooled water with silver iodide nuclei,under the assumption that the more nuclei present ina cloud, the greater will be the competition for waterand the smaller the resulting hailstones. In the UnitedStates seeding is carried out using airplanes, while inRussia and Italy the silver iodide is injected into cloudsusing rockets. In the former Soviet Union, seedingreduced hail damage in protected areas by a factor of3–5 at a cost of 2–3 per cent of the value of the cropssaved. In South Africa, crop loss reductions of about20 per cent have been achieved by cloud seeding. In

91Localized Storms

11

2

2

< 2days

2–3days

4–5days

6–7days

≥ 8days

1

1

2

4

4

4

2

1

2

1

1 22

1

22

1

0 1000 km

Fig. 4.5 Annual frequency of hailstorms in the United States (from Eagleman, 1983).

the United States, hail suppression programs were ter-minated at the end of the 1970s because of politicalcontroversy. The main reason was marginal evidencethat cloud seeding experiments actually increased theamount of hail, especially if thunderstorms hadreached a mature or supercell stage. However, datacollected over a three-year period indicated that croplosses had been reduced by 48 per cent in westernTexas and by 20 per cent in South Dakota. In Australia,major hail suppression programs have not progressedbeyond the experimental stage. There is some question

of cost-effectiveness, given the lower crop yields overlarge areas. However, in the Griffith and Murrumbidgeeirrigation areas, the cost of hail suppression wouldcertainly be less than the losses experienced. To date,the purpose of cloud seeding in Australia is usually notfor hail suppression, but for rain enhancement to relievedrought in New South Wales, and to prevent bushfiresin Victoria.

Hail can be deadly. On two occasions since 1960,hail has killed over 300 people in Bangladesh, whichexperiences the most frequent and intense hailstormsof any location in the world. Nothing, however, haseclipsed the disaster that struck the army of EdwardIII in 1359 near Chartres, France. In a matter ofminutes hail the size of goose eggs killed 1000 men and6000 horses, effectively ending his campaign to takeFrance in the early years of the Hundred Years’ War.Hail is also destructive to crops, animals, buildings, andvehicles. About 2 per cent of the United States’ cropproduction is damaged by hail each year. On the GreatPlains of the United States, losses in some years haveamounted to 20 per cent of the value of the crop. InAustralia in 1985, 20 per cent of the apple crop waswiped out by a single hailstorm around Orange, NewSouth Wales. Following the drought of 1982–1983,some wheat farmers in eastern Australia suffered themisfortune of seeing bumper wheat crops destroyed byhail. Property damage is less extensive than crop losses;however, hailstorms are the major cause of windowbreakage in buildings, and the major natural cause ofautomobile damage. The largest single insurancepayout – made by the major international insuranceunderwriter Munich Reinsurance Corporation –occurred following the 12 July 1984 hailstorm thatstruck the city of Munich, West Germany. Thiscompany paid out $US500 million, mainly on damageto automobiles.

Sydney, Australia, has the misfortune of being rec-ognized as the hail capital of the world. Insurancebrokers in Australia call hailstorms ‘glaziers’ picnics’because of the damage they do to windows. Here,there is an additional risk to property because homesare traditionally roofed with clay tiles that can bebroken by hail. Worse, low-cost public housing usesfibrocement sheeting as exterior wall cladding.Between 1980 and 1990, three or four localized stormsin the Sydney metropolitan area led to claims eachamounting to $A40–60 million. This was accepted asthe norm until two devastating events in the 1990s.

92 Climatic Hazards

Fig. 4.6 A) Large hailstones that have built up quickly from theaggregation of graupel particles (photo credit: Eric Beach,28 April 2002, Pomfret, MD http://www.erh.noaa.gov/er/lwx/publicpix/index.htm).

Fig. 4.6 B) A layered hailstone created by hail passing through themelting–freezing boundary at least five times (photo credit: http://www.crh.noaa.gov/mkx/slid-eshow/tstm/images/slide22.jpg).

The first occurred on 18 March 1990 and dropped hailup to 9 cm in diameter, breaking windows, roofing tilesand fibro cladding. The damage bill reached $A500million with houses completely smashed beyondrepair. This event paled in significance comparedto the damage done during the Great Hailstorm of14 April 1999.

The storm reached the outskirts of Sydney around7:20 pm, just after the Meteorological Bureau’s storm-watch roster changed and the evening televisionstations issued weather forecasts for plenty of sunshinelasting the next week. The forecasts were ironcladbecause the Bureau had already dismissed volunteers’phoned reports of the storm’s intensity up to 200 kmsouth of the city hours beforehand. Over a distanceof 1 km, between pristine bushland at the edge ofthe urban heat island and the first row of houses insouthern Sydney, the storm unleashed its fury. The lackof warning led to 30 aircraft at Sydney Internationalairport being damaged to the tune of $A100 million.Hailstones reaching 9 cm in diameter punched throughcar roofs and houses at velocities of 200 km hr-1.Hailstones in Sydney had reached this size only twicebefore: in January 1947 and March 1990. So intensewas the hail that flocks of seagulls, cormorants, andflying foxes were killed in the open where they stood.By the end of the evening, 2500 km2 of urban Sydneyhad been declared a natural disaster area with manyhomes uninhabitable as rain poured in through brokenroofs. In the hail zone, 34 and 62 per cent of homeshad broken windows and roofing tiles, respectively,while 53 per cent of cars were damaged. The storm sta-tistics were massive: 500 000 tonnes of hail fell, 20 415homes and 60 000 cars were damaged, 25 000 callswere made for assistance, 170 000 tarpaulins werefitted to roofs, 55 000 m2 of slate and 11 000 000terracotta tiles were replaced, and 32 000 home and3000 commercial insurance claims were processed. Itwas the costliest natural disaster in Australian history,resulting in a total insurance bill ($A1.5 billion rising to$A3 billion when non-insured damage was included)that surpassed the damage caused by the Newcastleearthquake of 1989 and by Cyclone Tracy in Darwinin 1974. Despite this damage, the storm does noteven rank in the top 40 most costly natural disasters inthe world since 1970. It ranks equal fourth with lossesdue to tornadoes in the United States over the sameperiod. The storm actually stimulated the regionaleconomy, providing work for home renovators and car

salespeople. In the weeks following the storm, thesupply of roofing tiles, tarpaulins, and householdwindows was exhausted, and the reputation of theState Emergency Services was in tatters. Untrained toclimb to the top of 3–4 storey blocks of flats to fixtarpaulins, and understaffed by the extent of thedisaster, the emergency services had to defer opera-tions to the Rural Fire Brigade which at least hadextension ladders and professionals trained to skipacross roofs. A week later, the army relieved everyoneas the enormity of the task unfolded. In the end, themajor occupational hazard for emergency servicespersonnel – besides falling off buildings – was beingstung by the thousands of wasps nesting on the sunnysides of damaged buildings.

TORNADOES

Introduction(Miller, 1971; Eagleman, 1983; Nalivkin, 1983)

A tornado is a rapidly rotating vortex of air protrudingfunnel-like towards the ground from a cumulonimbuscloud. Most of the time, these vortices remainsuspended in the atmosphere, and it is only whenthey connect to the ground or ocean surface that theybecome destructive. Tornadoes are related to largervortex formation in clouds. Thus, they often form inconvective cells such as thunderstorms, or in the right-forward quadrant of a hurricane at large distances(> 200 km) from the area of maximum winds. In thelatter case, tornadoes herald the approach of the hurri-cane. Often, the weakest hurricanes produce the mosttornadoes. Tornadoes are a secondary phenomenon, inwhich the primary process is the development of avortex cloud. Given the large number of vortices thatform in the atmosphere, tornadoes are generally rare;however, because vortices can be generated by amyriad of processes, tornadoes have no one mecha-nism of formation. On average, 850 tornadoes arereported annually, of which 600 originate in the UnitedStates. This frequency is increasing in the UnitedStates; the increase, however, is due mainly toincreased monitoring. The width of a single tornado’sdestructive path is generally less than 1 km. Theyrarely last more than half-an-hour and have a pathlength extending from a few hundred meters to tensof kilometres. In contrast, the destructive path of atropical cyclone can reach a width of 200 km, spread

93Localized Storms

over thousands of kilometers, and last for up to twoweeks. However, tornadoes move faster than cyclonesand generate higher wind velocities. Tornadoes moveat speeds of 50–200 km hr-1 and generate internalwinds at speeds exceeding 400–500 km hr-1. Torna-does, similar to tropical cyclones, are almost alwaysaccompanied by heavy precipitation.

Form and formation(Nalivkin, 1983; Grazulis, 1993)

Supercell tornado formation

By far the best-studied and most common process oftornado formation is that associated with supercellthunderstorms. The parent cloud may be 15–20 km indiameter with strong rotational uplift that pierces thestratosphere as a dome cloud capping the storm(Figure 4.7). This mesocyclone links to the jet streamproducing an anvil shape that is skewed in the direc-tion of movement. As precipitation falls from the stormit cools air, which sinks to the ground in front ofthe storm producing a gust front that precedes thethunderstorm. Destructive winds in this zone areknown as a downburst and when they hit the groundcan produce dangerous wind shear. Drier air behindthe thunderstorm is drawn into the mesocyclone inthe middle atmosphere. This evaporates cloud, giving asharply defined outline to the storm. More importantly,evaporation causes the air to cool, become unstableand cascade to the ground as a downdraft. Tornadoesdevelop between this rear downdraft and the main

updraft. As the downdraft expands, it collapses thesupercell and at this point the tornado descends tothe surface.

Dust devils, mountainadoes, fire tornadoes andwaterspouts

There are also a number of small vortex phenomenathat do not originate under the above conditions. Hor-izontal vortices tend to form over flat surfaces becauseair at elevation moves faster and overturns without theinfluence of frictional drag at the Earth’s surface. Ifthe horizontal vortex strikes an obstacle, it can be splitin two and turned upright to form a dry tornado-likevortex (Figure 4.8). Because of its small size, Coriolisforce does not affect these vortices, so they can rotateeither clockwise or counterclockwise. On very flatground, where solar heating is strong, thermal convec-tion can become intense, especially in low latitudes. Ifthe rate of cooling with altitude of air over this surfaceexceeds the usual rate of –1°C per 100 m (the normaladiabatic lapse rate), then the air can become veryunstable, forming intense updrafts that can begin torotate. These features are called ‘dust devils’ or, inAustralia, ‘willy-willys’. While most dust devils arenot hazardous, some can intensify downwind to takeon tornado-like proportions. Tornado-like vortices canalso be generated by strong winds blowing overtopography. Strong wind shearing away from the top ofan obstacle can suck air up rapidly from the ground.Such rising air will undergo rotation, forming smallvortices. Around Boulder, Colorado, strong winds are

94 Climatic Hazards

am

Dry air inflow

o

Zone of tornado formation0

2

4

6

8

10

12

14

16

Ele

vatio

n (k

m)

Eastt

Fig. 4.7 Generation of a tornado within a supercell thunderstorm (after Grazulis, 1993).

a common feature in winter, blowing across theRocky Mountains. Vortices have often been producedwith sufficient strength to destroy buildings. Suchvortices behave like small tornadoes and are called‘mountainadoes’.

Strong updrafts can also form because of intenseheating under the influence of fires. Most commonly,this involves the intense burning of a forest; however,the earthquake-induced fires that destroyed Tokyo on2 September 1923, and the fires induced by thedropping of the atomic bomb on Hiroshima on6 August 1945, also produced tornadoes. Experiments,where the atmosphere has been heated from belowusing gas burners, commonly have generated twinvortices on either side of the rapidly rising columns ofair. These fire whirlwinds can vary from a few metresin height and diameter to fire tornadoes hundreds ofmetres high. The most intense and hazardous firetornadoes are generated by forest fires, and usuallyform on the lee side of hills or ridges during extremebushfires or forest fires such as that of the AshWednesday disaster in southern Australia in February1983. Rising wind velocities can exceed 250 km hr-1,with winds inflowing into the maelstrom reachingvelocities of 100 km hr-1. Such firestorms pose aserious, life-threatening hazard for firefighters on theground. Fire whirlwinds have been known to liftflaming debris high into the sky and have snapped offfully grown trees like matchsticks. In many cases, theyare linked to the convective column above the fire,

and are fuelled by massive gaseous explosions withinthis rising thermal.

Finally, vortices can easily develop over warm seas,bays, or lakes under conditions of convective instabilitybecause of the lower friction coefficient of water.The waterspouts that develop can be impressive(Figure 4.9); however, wind velocities are considerablylower than their land counterparts – around 25–25 m s-1 – and damage is rare. They often form as multiplevortices (Figure 4.9). Florida has 50–500 sightingseach year and they are probably just as frequent inother tropical environments, especially on the easternsides of continents where warm water accumulatesunder easterly trade winds. Waterspouts that moveinland and kill people are probably tornadoes that havesimply developed first over water. These events arerare. One of the most notable waterspouts occurredduring Australia’s 2001 Sydney Hobart Yacht Race.The crew of the 24 metre maxi-yacht Nicorette video-taped the approach of an enormous waterspout.Despite efforts to avoid it, the yacht sailed through thewaterspout, was buffeted by 130 km hr-1 winds andhail, but came out the other side with only its mainsailshredded.

Fig. 4.8 Development of a mountainado from the splitting of ahorizontal vortex rolling along the ground (after Idso, 1976).

Fig. 4.9 An artist’s impression of the sailing vessel Trombes beingstruck by multiple waterspouts. Sketch first appeared in Les Meteores, Margolle et Zurcher, 3rd edition, 1869, page 126 (NOAA, 2003).

Structure of a tornado

Simple tornadoes consist of a funnel, which is virtuallyinvisible until it begins to pick up debris. Many torna-does are associated with multiple funnels. In June 1955a single tornado cloud in the United States produced13 funnels that reached the ground and left still moresuspended in midair. Generally, the larger the parentcloud, the greater the number of tornadoes that can be

95Localized Storms

produced. The funnel has horizontal pressure andwind profiles similar to a tropical cyclone (Figure 3.5),except that the spatial dimensions are reduced to a fewhundred meters. In addition, tornadoes developgreater depressed central pressure and more extremewind velocities at the funnel boundary. The sides ofthe funnel enclose the eye of the tornado, which ischaracterized by lightning and smaller, short-livedmini-tornadoes. Air movement inside the funnel isdownward. Vertical air motion occurs close to the wallwith upward speeds of 100–200 m s-1. The most visiblepart of a tornado, the funnel, is actually enclosedwithin a slower spinning vortex. The air moving downthrough the center of the funnel can kick up a curtainof turbulent air and debris at the ground – in a processsimilar to that of a hydrofoil moving over water(Figure 4.10). This feature – termed a ‘cascade’ – canalso form above the ground before the main funnelactually develops. If the cascade spreads laterally farenough, its debris can be entrained by the outer vortex,and if wind velocities in this outer vortex are strongenough, the main funnel may slowly be enshrouded bydebris from the ground upwards to form an ‘envelope’.Envelopes are more characteristic of waterspouts thantornadoes that develop over land.

Often tornado structure is more complex than asingle funnel. Air may descend through the center of

Fig. 4.10 The first tornado captured by the NSSL Doppler radar andNSSL chase personnel. Note the defined funnel shape andthe cascade formed at the ground by the incorporation ofdebris. Union City, Oklahoma, 24 May 1973. Image ID:nssl0062, National Severe Storms Laboratory (NSSL) Collection, NOAA Photo Library, NOAA.

the funnel at great speed. At the surface, this airspreads out and causes the vortex to break down. Airspinning into the vortex undergoes a rapid change indirection or shearing where it meets this downdraft,leading to the formation of mini-vortices around themain zone of rotating air. Up to half a dozen mini-vortices may form, rotating in a counterclockwisedirection around the edge of the tornado itself. Thesesmaller vortices are unstable, often appearing as wispyspirals. However, their presence can be devastating.Mini-vortices can rotate at speeds of 20–40 m s-1. Onthe inner side of the mini-vortex, rotation opposes thatof the main funnel leading to wind reversal or shearover very small distances (T-V in Figure 4.11). On theouter edge of the mini-vortex, air moves in the samedirection as that of the main funnel and enhances windspeeds (T+V in Figure 4.11). The former conditiontears structures apart, while the latter destroysstructures that might resist lower wind speeds.

T + V

T + V

T - V

T - V

T, rotational speed of parent vortexV, rotational speed of secondary vortex

Parent vortex

Secondary vortex Secondary

vortex

Fig. 4.11 Model for multiple-vortex formation in a tornado (based onFujita, 1971 and Grazulis, 1993).

The highest wind velocities generated withintornadoes have not been measured; but they can bedetermined theoretically. For example, eggs orwindows cleanly pierced by a piece of debris withoutshattering or a wooden pole pierced by a piece of strawrequires wind speeds in excess of 1300 km hr-1.However, this wind speed exceeds the speed of sound,a phenomenon that has not been reported during thepassage of a tornado. The maximum wind speedmeasured using photogrammetry is 457 km hr-1. Thesewind speeds can singe the skin of defeatheredchickens, move locomotives off railway tracks and shiftbridges. Because of the lack of measurement, a visual

96 Climatic Hazards

scale of tornado intensity has been developed; calledthe Fujita Tornado Intensity Scale, it is based on windspeeds (Table 4.1). The most intense tornado ranks asan F5 event on this scale. Such tornadoes tend tocompletely destroy houses, leaving nothing but theconcrete slab, and badly damage steel-reinforcedconcrete structures. Wind speeds during such mael-stroms range from 419–510 km hr-1. Note that windsduring Cyclone Tracy (mentioned in the previouschapter) not only twisted steel girders but alsotransported them through the air. These windsmust have exceeded F5 on the Fujita scale. Only0.4 per cent of tornadoes in the United States reachcategory F5; however with 1200 tornadoes observedeach year, this statistic equates to five extreme eventsannually. Over 50 per cent of tornadoes produce onlyminor damage to homes.

Mini-vortices are responsible for the haphazard andindiscriminate damage caused by tornadoes. Peopledriving a horse and cart have witnessed a tornadotaking their horse and leaving the cart untouched. Onetornado that swept through a barn took a cow, and leftthe girl milking it and her milk bucket untouched.Numerous photographs exist that show houses andlarge buildings cleanly cut in half with the remainingcontents undisturbed.

Occurrence(Miller, 1971; Nalivkin, 1983; Grazulis, 1993)

Tornadoes do not form near the equator becauseCoriolis force is lacking. Nor do they form in polar or

subpolar regions where convective instability is sup-pressed by stable, cold air masses. About 80 per cent ofall tornadoes occur in the United States, particularly onthe Great Plains. Figure 4.12 illustrates the density oftornadoes and the preferred path of travel in thisregion. The area with five or more tornadoes per year,which is referred to as ‘tornado alley’, takes in Kansas,Oklahoma, Texas, Missouri, and Arkansas. Tornadoesoccur commonly in this region, because here coldArctic air meets warm air from the Gulf of Mexico,with temperature differences of 20–30°C. The flatnature of the Plains enhances the rapid movement ofcold air, while the high humidity of Gulf air createsoptimum conditions of atmospheric instability. Torna-does in the United States tend to occur in late spring,when this temperature difference between air massesis at its maximum. At this time of year, the polar jetstream still can loop far south over the Great Plains.

Between 1950 and 1999 there were 40 522 tornadoesrecorded in the United States, an average of 810 eventsper year. Of these, approximately 43 per cent wereF2–F5 events. Since 1950, the frequency of tornadoeshas been increasing from 600 tornadoes per year to acurrent average of 1200 per year (Figure 4.13). WhileTexas has the greatest number of tornadoes, the greatestdensity of tornadoes occurs in Florida. Kansas reportsthe most tornadoes in North America by population:3.4 per 10 000 people. Surprisingly, the greatest numberof deaths per unit area occurs in Massachusetts. Thegreatest number of tornadoes reported in a singleweek occurred 4–10 May 2003 (384 reports), while the

97Localized Storms

F-scale Wind Speed Type of Damage FrequencyF0 64–116 km/h MINIMAL DAMAGE: Some damage to chimneys, TV antennas, 82.0%

roof shingles, trees, and windows.F1 117–180 km/h MODERATE DAMAGE: Automobiles overturned, carports destroyed, 11.0%

trees uprooted.F2 181–253 km/h MAJOR DAMAGE: Roofs blown off homes, sheds and outbuildings 4.0%

demolished, mobile homes overturned.F3 254–332 km/h SEVERE DAMAGE: Exterior walls and roofs blown off homes. 1.8%

Metal supported buildings severely damaged. Trees uprooted.F4 333–418 km/h DEVASTATING DAMAGE: Well-built homes levelled. Large steel 0.9%

and concrete missiles thrown far distances.F5 419–512 km/h INCREDIBLE DAMAGE: Homes leveled with all debris removed. 0.4%

Large buildings have considerable damage with exterior walls and roofs gone. Trees debarked.

Table 4.1 Fujita Tornado Intensity Scale and frequency of tornadoes on the Fujita Scale in the USA (from Fujita, 1981; Grazulis, 1993).

greatest number in one day occurred on 29 March 1974(274 reports). However, these trends must be set withinthe context of variability from year to year. The 2003season mentioned above occurred following a year that

witnessed the fewest tornadoes – adjusted for increasedreport trends – in 50 years.

Canada now experiences the second largest numberof tornadoes: 100–200 events annually as far north as

98 Climatic Hazards

0 1000 km

1

2

3

4

5

6

>7

Tornadoes per year

Southern Ontario

Rocksprings, Texas

Oklahoma City

Irving, Kansas

Topeka, Kansas

Illinois

3

254

Alabama

Judsonia, Arkansas

Xenia, Ohio

1

Edmonton, Alberta

Fig. 4.12 Average number of tornadoes per year and preferred path of movement in the continental United States (adapted from Miller, 1971).

1920 1930 1940 1950 1960 1970 1980 1990 2000

Deaths

Number Of Tornadoes

Year

Num

ber

Of T

orna

does

200

400

600

800

1000

1200

1400

1600

00

200

400

600

800

Num

ber

Of T

orna

do D

eath

s

Fig. 4.13 Number of tornadoes and deaths due to tornadoes in the United States between 1916 and 2002 (U.S. Severe Weather Meteorology andClimatology, 2002a, b).

Edmonton, Alberta, at 53.5°N. This is far north ofthe zone where sharp temperature differences favortornado development. Here, in 1987, a F4 tornadokilled 27 people. Other areas of the world are sus-ceptible to tornadoes. Surprisingly, a region such asQueensland, Australia, reports 4.0 tornadoes per 10 000 people – more than Kansas. The RussianSteppes are similar in relief to the Great Plains. Here,however, there is no strong looping of the jet stream orany nearby warm body of water supplying highhumidity. Great Britain and the Netherlands average50 and 35 tornado-like vortices a year, respectively. InEurope overall, about 330 vortices are reported eachyear, of which half occur as waterspouts and fewerthan ten cause noteworthy damage. One of thegreatest tragedies on this continent occurred in Veniceon 11 September 1970 when a tornado sank awaterbus near St Mark’s Square and swept laterthrough a campground, killing 35 people. Surprisingly,tornadoes are rare in Russia, Africa and the Indiansubcontinent. In the case of the last, tornadoes arerarely reported outside the northern part of the Bay ofBengal where they are associated with tropicalcyclones. However, the deadliest tornado occurred inBangladesh on 26 April 1989, killing 1300 people andinjuring 12 000 others.

Tornado destruction(Whipple, 1982; Nalivkin, 1983)

Tornadoes are one of the most intense and destructivewinds found on the Earth’s surface. The fact that theymost frequently form along cold fronts means thatthey can occur in groups over wide areas. Theirdestructive effects are due to the lifting force at thefunnel wall, and to the sudden change in pressureacross this boundary. The lifting force of tornadoes isconsiderable, both in terms of the weight of objectsthat can be moved, and the volume of material thatcan be lifted. Large objects weighing 200–300 tonneshave been shifted tens of metres, while houses andtrain carriages weighing 10–20 tonnes have beencarried hundreds of meters. Tornadoes are also able tosuck up millions of tonnes of water, which can becarried in the parent cloud. Tornadoes have beenknown to drain the water from rivers, thus temporarilyexposing the bed.

The most destructive mechanism in a tornado isthe dip in barometric pressure across the wall of thefunnel. It is not how low the pressure gets in the eye

(about 800 hPa), but how rapidly this change inpressure occurs. The fact that the tornado is socompact and has a sharp wall means that a pressurereduction of tens of hPa s-1 could occur, although thisdip has been difficult to substantiate by measurement.Air pressure in buildings affected by a tornado ends upbeing much higher than the outside air pressure, andcannot escape from the structure quickly enoughwithout causing an explosion. This exploding effect isthe main reason for the total devastation of buildings,whereas adjacent structures or remaining remnantsappear untouched. The pressure reduction can be sorapid that live objects appear burnt because of dehy-dration. One of the major causes of death from atornado is not due to people being struck by flyingdebris or tossed into objects, but to their receivingsevere burns. It is even possible that the defeatheredchickens described above could be cooked because ofthis phenomenon.

Between 1916 and 2000, tornadoes killed 12 766people in the United States. This death toll hasdecreased by 200 people per decade because ofincreased warning and awareness (Figure 4.13). InApril 1965, 257 people lost their lives as 47 tornadoesdeveloped along a cold front in the central UnitedStates. The ‘Three States’ tornado of March 1925created a path of destruction over 350 km long. Over700 people were killed by the tornadoes spawned fromthe movement of this single parent cloud. One tornadoin March 1952 destroyed 945 buildings in Judsonia,Arkansas. Another tornado – in Rock Springs, Texas, inApril 1927 – killed 72 people, injured 240 others,and wiped out most of the town of 1200 people in90 seconds. In all, 26 per cent of the populationwere either killed or injured. Cities are not immune,although high-rise buildings tend to survive better thansingle-unit dwellings. For example, in June 1966, atornado went through Topeka, Kansas, destroying mosthomes and shops in its path but leaving multistoriedflats and office buildings undamaged. The worsttornado disaster in the United States occurred duringa 16-hour period on 3–4 April 1974. A total of 148 tor-nadoes was recorded from Alabama to Canada, killing315 people and causing damage amounting to $US500million. Six of the tornadoes were amongst the largestever recorded. Complete towns were obliterated, theworst being Xenia, Ohio, where 3000 homes weredestroyed or damaged in a town with only 27 000residents.

99Localized Storms

Warning (Golden & Adams, 2000)

Tornado prediction in the United States has dramati-cally improved since 1970 with the explosion in privatemeteorological forecasting and improved federallyfunded technology. In the 1950s, the US Air Forceinstigated tornado forecasting. A radar network wasestablished in which tornadoes could be detectedbased upon their ‘hock echo’ signature on a radar map.However, in most cases, these identified only tornadoesthat had already touched down. In the 1970s, lightningdetectors were established to detect tornado-bearingstorms. In the 1990s, infrasound and seismic grounddetector systems augmented a range of remote sensingsystems. The latter included precursor GeostationaryOperational Environmental Satellites (GOES) 8 and 10forecasting in tandem with 161 WSR88-Doppler radarinstallations, all of which were backed up by improvedpersonnel training, increased numbers of local stormspotters and enhanced public awareness. The satellitemonitoring permits severe convective systems to bedetected days in advance. No longer is a simple vortexapproach used to model tornadoes. Instead, satellitedetection and computer modelling provides, hours inadvance, evidence of mesocyclones favorable totornadoes. The Doppler radar network monitors theinstantaneous development of severe thunderstorms,large hail, heavy rain, and high winds associated withtornadoes. These signals are processed locally usingsophisticated algorithms at over 1200 sites usingthe Automated Surface Observing System. In 1978, theprobability of detecting a tornado was 22 per cent witha lead time of three minutes. In 1998 this probabilityhad risen to 65 per cent with a lead time of 11 minutes.The events that occurred on 3 May 1999 aroundOklahoma City provide an example of how effectiverecent procedures have become. That night 116warnings, with lead times of 20–60 minutes, wereissued for 57 tornadoes. No warning is useful unless itcan be communicated effectively to the public. Radiomessages can be transmitted by a network of 504transmitters to 80–85 per cent of the US population.About 25 per cent of households can automaticallyreceive these signals. Many TV stations also receivethese automatic warnings and display them on tele-vision screens as either a text crawler or a flashing icon.In the United States the fear of tornadoes occurringwithout any warning has virtually disappeared.

Response(Sims & Baumann, 1972; Oliver, 1981)

For most of the world, tornadoes can be put into the‘act of God’ category. For instance, in Australia, it isjust bad luck if a tornado strikes you or your property.In the United States, especially in ‘tornado alley’,tornadoes are a frequent hazard. While it is difficultto avoid their occurrence, there are many ways tolessen their impact. For instance, in the United StatesMidwest, schoolchildren are taught at an early age howto take evasive action if they see, or are warned about,the approach of a tornado. I can remember, as aschoolchild living in southern Ontario, being senthome from school one day when tornadoes wereforecast. Southern Ontario, then, was not consideredhighly vulnerable; but the school authorities wereaware that a tornado hitting a school full of childrenwould generate a major disaster. We were sent home tominimize the risk. We were also drilled in proceduresto take if a tornado did strike during school hours – thesame procedures we would have taken if a nuclearattack were imminent! We knew that tornadoestraveled mainly north-east; so, if we could see atornado approaching and were caught in the open, wewere instructed to run to the north-west. In the UnitedStates Midwest, most farms or homes have a basementor cellar separate from the house that can be used asshelter. All public buildings have shelters and workersare instructed in procedures for safe evacuation. Safetypoints in buildings, and in the open, are well sign-posted. In the worst threatened areas such as ‘tornadoalley’, tornado alerts are broadcast on the radio as partof weather forecasts. The actual presence of a tornadocan be detected using Doppler radar or tornadowatches maintained by the Weather Service. Theweather watch consists of a network of volunteerham radio operators known as the Radio EmergencyAssociated Citizen Team (REACT), who will move tovantage points with their radio equipment on high-riskdays. Immediately a tornado is detected, local weathercenters, civil defense and law enforcement agenciesare alerted and warnings then broadcast to the publicusing radio, television, and alarm sirens.

For many rural communities in the Midwest, atornado is the worst disaster that can strike the area.State governments will declare a state of emergencyand, if outbreaks of tornadoes occur over several states,federal assistance may be sought through the Federal

100 Climatic Hazards

Emergency Management Agency (FEMA). In ruralareas, community response to a tornado disaster isautomatic and there are government agencies, inaddition to police and fire services, which have beenspecially created for disaster relief. Generally, however,tornadoes are accepted as a risk that must be livedwith. The above observations are not universal; theymay be modified by the cultural attitudes of peopleliving in various threatened areas. Sims and Baumann(1972) analyzed the attitudes of people to the tornadorisk between Illinois and Alabama. The death rate fromtornadoes apparently was higher in the latter state.They concluded, based upon interviews, that there wasa substantial attitude difference between the twostates. People in Illinois believed that they were incontrol of their own future, took measures to minimizethe tornado threat, and participated in communityefforts to help those who became victims. In Alabama,however, people did not heed tornado warnings, tookfew precautions, were more likely to believe that Godcontrolled their lives, and were less sympathetictowards victims. Oliver (1981) has analyzed this studyand pointed out some of its fallacies. For instance, thedeath toll from tornadoes is higher in Alabama becausethe tornadoes there are more massive than in Illinois.However, the Sims and Baumann (1972) study doeshighlight the fact that not all people in the UnitedStates have the same perception of the tornadohazard or response to dealing with its aftermath. Thesedifferences will be elaborated further in Chapter 13.

Your attitude to tornadoes may make little differ-ence to your survival; rather, where you live in theMidwest dictates your chances of survival. Inadver-tently, Europeans settling the Midwest may havechosen the most susceptible town sites. For example,Oklahoma City on the Canadian River has been struckby more than 25 tornadoes since 1895, even though theState of Oklahoma ranks seventh in the United Statesfor deaths per unit area due to tornadoes – 7.3 deathsper 10 000 km2 compared to 20.5 per 10 000 km2 forMississippi, which ranks first. Indian tribes in theMidwest rarely camped in low-lying areas such asthe bottoms of ravines, or along river courses, becausethey were aware that tornadoes tend to move towardsthe lowest elevation once they touch ground. Thismakes sense since the tornado vortex is free-movingand will tend to follow the path of least resistance.Figure 4.14 dramatically illustrates this tendency. Thefirst tornado to pass through Irving, Kansas, on 30 May1879, ended up being trapped in the valley of the BigBlue River, and eventually in a small gully off to theside. Tornadoes do not always adhere to this pattern, asillustrated in the same figure by the path of the secondtornado that came through the town minutes later. Forhistorical reasons such as access to transport and water,many towns in the United States Midwest are builtnear rivers or in low-lying areas. The fact thattornadoes tend to travel along these paths may be oneof the reasons why they have been, and will continue tobe, a major hazard in the United States Midwest.

101Localized Storms

N

Irving, Kansas

2nd tornado

1st tornado

Big Blue River

Fig. 4.14 The effect of topography on the passage of a tornado at Irving, Kansas, 30 May 1879 (after Finley, 1881).

CONCLUDING COMMENTS

The thought that a series of tornadoes – such as thosethat struck the United States in April 1974 or ahailstorm as devastating as the Sydney hailstorm of1999 – could strike again demolishes complacencythat our present climate is stable or that humans canprevent extreme climatic events. Thunderstorms willcontinue to drop rain at the upper level of physicalpossibility, produce hail that can kill crowds in theopen, and spawn tornadoes that are certain to setrecords for damage and loss of life. The uncertaintyis not if, but when, such events will recur. Whilescientists are tempted to invoke global warming asthe cause of an increase in the severity of thunder-storms, thunderstorm magnitude simply representsthe natural occurrence of random natural events.King Edward III’s defeat due to a hailstorm atChartres, France, in 1359 was not a signature ofglobal warning. History is replete with similar events.Today, were such a disaster to strike the final matchof the baseball World Series or the soccer WorldCup, the blame would certainly be directed towardsglobal warming as the scapegoat.

While short-lived events such as hailstorms andtornado swarms are easily dealt with in modernsocieties, the effects of flash flooding, catastrophicregional rains, and droughts always will test ourclimatic modelling, our societal altruism and,ultimately, the capacity of the international com-munity to respond in any meaningful way. Thefollowing chapters examine the nature of droughtsand regional floods, and our attempts to mitigate, andrespond to, such large-scale disasters.


Anagnostou, E.N., Chronis, T., and Lalas, D.P. 2002. New receivernetwork advances in long-range lightning monitoring. EOS,Transactions American Geophysical Union 83: 589, 594–595.

Black, R.A. and Hallett, J. 1998. The mystery of cloud electrification.American Scientist 18: 526–534.

Eagleman, J.R. 1983. Severe and Unusual Weather. Van NostrandReinhold, NY.

Finley, J.P. 1881. Report on the tornadoes of May 29 and 30, 1879,in Kansas, Nebraska. Professional Paper of the Signal ServiceNo. 4, Washington.

Fujita, T.T. 1971. Proposed mechanism of suction spots accompaniedby tornadoes. Preprints, Seventh Conference on Severe LocalStorms, American Meteorological Society, Kansas City, pp. 208–213.

Fujita, T.T. 1981. Tornadoes and downbursts in the context ofgeneralized planetary scales. Journal of Atmospheric Research38: 1511–1534.

Geophysics Study Committee 1987. The Earth’s ElectricalEnvironment. United States National Research Council,National Academy Press.

Golden, J.H. and Adams, C.R. 2000. The tornado problem: forecast,warning, and response. Natural Hazards Review 1(2): 107–118.

Grazulis, T.P. 1993. Significant Tornadoes 1680–1991. EnvironmentalFilms, St. Johnsbury.

Idso, S. 1976. Whirlwinds, density current and topographicaldisturbances: A meteorological mélange of intriguing interactions.Weatherwise 28: 61–65.

Miller, A. 1971. Meteorology (second edn). Merrill, Columbus.Nalivkin, D.V. 1983. Hurricanes, Storms and Tornadoes. Balkema,

Rotterdam.NOAA 2003. Image ID: wea00300, historic NWS collection.

<http://www.photolib.noaa.gov/historic/nws/wea00300.htm>Oliver, J. 1981. Climatology: Selected Applications. Arnold, London.Oliver, J. 1986. Natural hazards. In Jeans, D.N. (ed.) Australia:

A Geography. Sydney University Press, Sydney, pp. 283–314.Sims, J.H. and Baumann, D.D. 1972. The tornado threat: coping

styles of the North and South. Science 176: 1386–1392.U.S. Severe Weather Meteorology and Climatology 2002a. United

States tornadoes, 1916–2000. <http://www.hprcc.unl.edu/nebraska/ustornadoes1916-2000.html>

U.S. Severe Weather Meteorology and Climatology 2002b. UnitedStates tornado deaths 1916–2000. <http://www.hprcc.unl.edu/nebraska/ustornadoes1916-2000.html>

Whipple, A.B.C. 1982. Storm. Time-Life Books, Amsterdam.Yeo, A., Leigh, R., and Kuhnel, I. 1999. The April 1999 Sydney

Hailstorm. Natural Hazards Quarterly 5(2), <http://www.es.mq.edu.au/nhrc/web/nhq/nhq5-2tables.htm>


INTRODUCTION

Drought and famine have plagued urban–agriculturalsocieties since civilizations first developed. While manydefinitions exist, drought can be defined simply as anextended period of rainfall deficit during which agri-cultural biomass is severely curtailed. In some parts ofthe world, such as the north-east United States andsouthern England, a drought may have more of aneffect on urban water supplies than on agriculture. Thedefinition of drought, including the period of rainfalldeficit prior to the event, varies worldwide. In southernCanada, for instance, a drought is any period where norain has fallen in 30 days. Lack of rain for this length oftime can severely reduce crop yields in an area wherecrops are sown, grown, and harvested in a period ofthree to four months. In Australia, such a definition ismeaningless, as most of the country receives no rainfallfor at least one 30-day period per year. Indeed, intropical areas subject to monsoons, drought conditionsoccur each dry season: most of tropical Australia, evenin coastal regions, endures a rainless dry season lastingseveral months. In Australia, drought is usually definedas a calendar year in which rainfall registers in thelower 10 per cent of all the records. Unfortunately, inthe southern hemisphere, a calendar year splits thesummer growing season in two. A more effective crite-rion for drought declaration should consider abnor-mally low rainfall in the summer growing season.

Drought onset is aperiodic, slow, and insidious. As aresult, communities in drought-prone areas must beconstantly prepared to insulate themselves from theprobability of drought and finally, when overwhelmed,to ensure survival. However, a community’s response todrought varies depending upon its social and economicstructure. This chapter examines first the pre-colonialresponse of societies to drought. This is followed by anexamination of drought response in modern societies.While westernized countries have generally faredbetter than underdeveloped countries, it will be shownthat national perception and policy ultimately deter-mine success in drought mitigation. The chapterconcludes by comparing the success of drought relieforganized by world organizations, mainly the UnitedNations, to that of private responses such as the BandAid appeal headed by Bob Geldof in response to theEthiopian drought of 1983–1984.

PRE-COLONIAL RESPONSE TODROUGHT

(Garcia, 1972; Morren, 1983; Watts, 1983; Scott, 1984)

In nomadic settings, there is a heavy reliance uponsocial interaction as drought develops. The !Kung ofthe Kalahari Desert of south-west Africa congregatedaround permanent water holes only during the dryseason (see Figure 5.1 for the location of major Africanplacenames). If conditions became wetter and more

Drought as a Hazard

C H A P T E R 5

favorable for food gathering and hunting, !Kung clanssplit into progressively smaller groups with as few asseven people. During severe droughts, however, the!Kung gathered in groups of over 200 at two perma-nent water holes, where they attempted to sit out thedry conditions until the drought broke.

In societies that are more agrarian, there is a criticalreliance upon agricultural diversification to mitigate allbut the severest drought. Pre-colonial Hausa society innorthern Nigeria (on the border of the Sahara Desert)exemplifies this well. Before 1903, Hausan farmersrarely mono-cropped, but planted two or three differ-ent crops in the same field. Each crop had differentrequirements and tolerances to drought. This practiceminimized risk and guaranteed some food productionunder almost any conditions. Water conservationpractices were also used as the need arose. If replantingwas required after an early drought, then the spacingof plants was widened or fast-maturing cereals plantedin case the drought persisted. Hausan farmers alsoplanted crops suited to specific microenvironments.Floodplains were used for rice and sorghum, inter-fluves for tobacco or sorghum, and marginal land fordry season irrigation and cultivation of vegetables.Using these practices, the Hausan farmer couldrespond immediately to any change in the rainfallregime. Failing these measures, individuals could relyupon traditional kinship ties to borrow food should

famine ensue. Requests for food from neighborscustomarily could not be refused, and relatives workingin cities were obligated to send money back to thecommunity or, in desperate circumstances, wererequired to take in relatives from rural areas asrefugees.

A resilient social structure had built up over severalcenturies before British colonial rule occurred in 1903.Hausan farmers lived in a feudal Islamic society. Emirsruling over local areas were at the top of the social‘pecking order’. Drought was perceived as an aperiodicevent during which kinship and descent groupinggenerally ensured that the risk from drought was dif-fused throughout society. There was little reliance upona central authority, with the social structure givingcollective security against drought. Taxes were col-lected in kind rather than cash, and varied dependingupon the condition of the people. The emir mightrecycle this wealth, in the form of loans of grain, in alocal area during drought. This type of social structurecharacterized many pre-colonial societies throughoutAfrica and Asia. Unfortunately, colonialism made theseagrarian societies more susceptible to droughts.

After colonialism occurred in Nigeria, taxes werecollected in cash from the Hausan people. These taxesremained fixed over time and were collected by acentralized government rather than a local emir. Theemphasis upon cash led to cash cropping, with returnsdependent upon world commodity markets. Produc-tion shifted from subsistence grain farming, whichdirectly supplied everyday food needs, to cashcropping, which provided only cash to buy food. Ifcommodity prices fell on world markets, people had totake out loans of money to buy grain during lean times.As a result, farmers became locked into a cycle ofseasonal debt that accrued over time. This indebtednessand economic restructuring caused the breakdown ofgroup and communal sharing practices. The process ofsocial disruption was abetted by central governmentintervention in the form of food aid to avert famine.Hausan peasant producers became increasingly vulner-able to even small variations in rainfall. A light harvestcould signal a crisis of famine proportions, particularly ifit occurred concomitantly with declining export prices.The Hausan peasantry now lives under the threat ofconstant famine.

During drought, a predictable sequence of eventsunfolds. After a poor harvest, farmers try to generateincome through laboring or craft activity. As the famine


Semi-arid

Saheliancountries

Senegal Mauritania

Mali Niger Chad Sudan

Port SudanEritrea

Djibouti

SomaliaKenya

EthiopiaNigeria

Botswana

Burkina Faso

Gambia

KalahariDesert

30˚

10˚

20˚

0˚

10˚

20˚

30˚

Fig. 5.1 Map of semi-arid regions and the Sahel in Africa.

intensifies, they seek relief from kin and subsequentlybegin to dispose of assets. Further steps includeborrowing of money or grain and, if all else fails, eitherselling out or emigrating from the drought-afflictedregion. With outward migration, governments noticethe drought and begin organizing food relief. After thedrought, the farm household has few resources left tore-establish pre-existing lifestyles. Rich households inthe city have the resources (the farmer’s sold assets) towithstand poor harvests. The natural hazard of droughtnow affects exclusively the Hausan farmer. In thisprocess, society has become differentiated such thatthe wealthy beneficially control their response todrought, while the Hausan farmer becomes largelypowerless and subjected to the worst effects of adrought disaster. This sequence of events now charac-terizes drought response throughout Africa and otherThird World countries, even though colonialism hasdisappeared. Dependency upon cash cropping, exportcommodity prices or the vagaries of war have repeat-edly brought large segments of various societies tocrisis point during the major droughts that haveoccurred towards the end of the twentieth century.

POST-COLONIAL RESPONSE

(Garcia, 1972; French, 1983)

In contrast, westernized societies rely heavily upontechnology to survive a drought. This is exemplified inAustralia where individual farmers – who withstand theworst droughts – have developed one of the mostefficient extensive agricultural systems in the world.Self-reliance is emphasized over kin or community ties,and it is efficiency in land management that ultimatelydetermines a farmer’s ability to survive a drought. As adrought takes hold, these management practices aresubtly altered to maximize return and minimizepermanent damage to the land and to the farming unit.Where land is cultivated, land management practicesare directed towards moisture conservation and cropprotection. Practices promoting the penetration ofrain into the subsoil include mulching, tillage, andconstruction of contour banks and furrows thatminimize runoff. In certain cases, catchments will becleared and compacted to increase runoff into farmdams that can be used for irrigation during drought.Because most crops mature at times of highestmoisture stress, the time of sowing is crucial to the finalcrop yield. Early sowing in winter ensures grain flow-

ering and maximum growth before high water-stressconditions in summer. In addition, application ofphosphate fertilizers can double crop yield for eachmillimetre of rainfall received. Where land is used forgrazing, some pasture is set aside in spring so that haycan be cut to use as feed over a dry summer period. Inmixed farming areas, grain must also be held over fromthe previous year’s harvest to supplement sheep andcattle feed. When a drought persists for several years,stock are either dramatically culled or breeding stockare shipped considerable distances and grazed at a cost(agistment) in drought-free areas.

Technological advances also play a key role infarming drought-prone, semi-arid areas. These tech-niques include aerial spraying of herbicides to controlweed growth, and aerial sowing of seed to preventsoil compaction by heavy farm machinery. Aerialmonitoring is used to track livestock and give earlywarning of areas that are threatened by overgrazing.Computer management models are also utilized tomaximize crop yields, given existing or forecastclimate, moisture, and crop conditions. Finally, effi-cient financial management becomes crucial as muchin drought survival as in recovery. Most farmers aim forthree profitable years per decade and must be able tosustain the economic viability of the farm for upto seven consecutive drought years. Such a mode of operation requires either cash reserves or a line ofcredit with a bank to permit breeding stock to becarried for this period of time. Until the rural crisis ofthe 1980s, fewer than 15 per cent of farmers inAustralia entered financial difficulty following a majordrought.

The above management practices set the westernizedfarmers apart from those in nomadic and under-developed countries. Consequently, the Australianfarmer can survive droughts as severe as those that havewracked the Sahel region of Africa in recent years.However, the Australian farm does not have to supportthe large number of people that units in Africa do. As aresult, drought-induced famine in Africa can detrimen-tally disturb the entire social and economic fabric of acountry, whereas in Australia or the United States, muchof the population may be unaware that a major droughtis occurring. In the 1970s and 1980s, the intensity ofdrought worldwide not only crippled many Africancountries but, for the first time since the depressionyears of the 1930s, it also perturbed many westernizedsocieties.

105Drought as a Hazard

DROUGHT CONDITIONSEXACERBATED BY MODERNSOCIETIES

(Bryson & Murray, 1976; Glantz, 1977; Lockwood, 1986)

Despite these technological solutions, human activitycan greatly exacerbate droughts through over-croppingof marginalized land, massive vegetation clearing, andpoor soil management. These activities have affectedall semi-arid regions of the world. In the RajputanaDesert of India, which was a cradle of civilization,successive cultures have flourished, but each timecollapsed. The region is humid and affected seasonallyby the monsoon. However, the atmosphere isextremely stable because overgrazing leads to dustbeing continually added to it. Dust in the uppertroposphere absorbs heat during the day, lessening thetemperature gradient of the atmosphere and pro-moting atmospheric stability. At night, dust absorbslong wave radiation, keeping surface temperatureselevated above the dew point. An artificial desert hasbeen produced that is self-maintaining as long as dustgets into the atmosphere.

Even without massive dust production, clearing ofvegetation can itself establish a negative, bio-geophysicalfeedback mechanism, locking a region into aridity.In the Sahel of Africa, decreasing precipitation since1960 has reduced plant growth, leading to reducedevapotranspiration, decreased moisture content in theatmosphere, and a further decrease in rainfall. Overtime, soil moisture slowly diminishes, adding to thereduction in evaporation and cloud cover. As the soilsurface dries out and vegetation dies off, the surfacealbedo – the degree to which short wave solar radiation isreflected from the surface of plants – is reduced, leadingto an increase in ground heating and a rise in near-ground air temperatures. This process also reduces pre-cipitation. The destruction of vegetation exposes theground to wind, permitting more dust in the atmosphere.

Drought in the Sahel occurred concomitantly withan increase in population and deleterious economicconditions. In the 1970s, world fuel prices soared andpeople switched from kerosene to wood for cookingand heating. Such a change led to rapid harvesting ofshrubs and trees. With diminishing crop yields, landwas taken out of fallow, further reducing soil moisture.Western techniques of plowing – introduced to increasefarming efficiency – destroyed soil structure, leading tothe formation of surface crusts that increased runoff

and prevented rain infiltration. All of these practiceshave enhanced the negative feedback mechanisms,favoring continued drought and desertification.

Similar physiographic conditions exist at present inAustralia, but without the social outcomes. Here,humans have interfered more broadly with the eco-system. The reduction of the dingo population allowsthe kangaroo population to increase, competing withdomestic stock for food and water during drierconditions. Construction of large numbers of dams andponds for stock watering allows native animals tosurvive into a drought; and introduction of feralanimals – such as rabbits, pigs, goats, cats, cattle, waterbuffalo and donkeys – means passive native specieshave been replaced with more aggressive species.Many of these introduced species are ungulates(hoofed animals) and cause soil compaction. Wholesaleringbarking of trees has occurred, to such an extentthat in many areas there are not enough mature treesleft to naturally re-seed a region. Dieback is endemic.Without deep-rooted trees and evapotranspiration,watertables have risen to the surface, flushing salt fromthe ground to be precipitated in insoluble form atthe surface. This salt scalds vegetation, leaving bareground susceptible to wind erosion. Soil structure isaltered and surface crusting takes place with the resultthat wetting and water penetration of the groundduring rain becomes more difficult. In Victoria, salinityaffects about 5200 km2 of land and is increasingby 2–5 per cent per year, while in Western Australia,3000 km2 of arable land has been abandoned. Largesections of the Murray River drainage basin arethreatened by this increasing problem, which isexacerbated by irrigation. As a result, arable land insome agricultural river basins within Australia is beingruined at a rate ten times greater than Babyloniancivilization’s degradation of the Euphrates–Tigrissystem, and five times faster than it took to change theIndus River floodplain into the Rajputana Desert.

MODERN RESPONSE TODROUGHTS

Societies that expect drought: theUnited States(Rosenberg & Wilhite, 1983; Warrick, 1983)

The above examples of national response to droughtwould imply that underdeveloped countries have great


difficulty in coping with drought conditions. Somecountries – such as Ethiopia in the 1980s – even go sofar as to ignore the plight of victims and deny to theoutside world that they are undergoing a majorcalamity. The western world should not sit smug justbecause our societies are developed to a higher techni-cal level. Often, by delving into history, it becomesevident that western countries have responded todrought in exactly the same manner that some ThirdWorld countries do today.

The national response to drought is well-documentedin the United States. In Chapter 2, it was pointedout that there was a well-defined cyclicity to droughtoccurrence on the Great Plains, synchronous withthe 18.6-year lunar cycle. Major droughts haveoccurred in the 1890s, 1910s, 1930s, 1950s, and 1970s.Most people have a vivid picture of the 1930s droughtwith mass outward migration of destitute farmers froma barren, windswept landscape. However, the 1930sdrought was not the worst to occur. Before the 1890s,the Great Plains were opened up to cultivation on alarge scale. Settlement was encouraged via massiveadvertising campaigns sponsored by governmentsand private railway companies in the eastern citiesand abroad. During the 1890s drought, widespreadstarvation and malnutrition occurred in the central andsouthern High Plains. Similar conditions accompaniedthe drought in the 1910s in the Dakotas and easternMontana. There was no disaster relief in the form offood or money from the federal government. Duringthe 1890s drought, many state governments refused toacknowledge either the drought, or the settlers’ plight,because government was trying to foster an image ofprosperity to attract more migrants. In both the 1890sand 1910s, mass outward migration took place withsettlers simply abandoning their farms. In somecounties, the migration was total. Those that stayedbehind had to shoulder the burden individually or withmodest support from their own or neighboring states.The similarity between responses to these twodroughts and to the 1980s Ethiopian one is striking.

The 1930s drought on the Great Plains is notablesimply because the government began to respond todrought conditions. Outward migration was not somassive as some people think, but merely followed apattern that had been established in the precedingyears. Many over-capitalized farmers, as their financescollapsed, simply liquidated or signed over their farmsto creditors. At a national level, the 1930s drought had

little impact. It received massive federal aid but thiswas minor in comparison to the money being injectedinto the depressed American economy. There were nofood shortages (in fact, there was a surplus of foodproduction), no rising costs and few cases of starvationor malnutrition. For the first time, the governmentassisted farmers so that by 1936, 89 per cent of farmersin some counties were receiving federal funds. The aidwas not in the form of food, but cash. The governmentundertook a number of programs to aid farmers’staying on the land. Irrigation projects were financed,government-sponsored crop insurance initiated, and aSoil Conservation Service established. Programs weredesigned to rescue farmers, re-establish agriculturalland, and mitigate the effect of future droughts.

The 1950s drought was very different. Theprograms begun in the 1930s were accelerated. Therewas a six-fold increase in irrigated land between the1930s and the 1980s. The amount of land under cropinsurance increased, while marginal land was tuckedaway under the Soil Bank program. In the 1950s, fewfarms were abandoned: they were sold off to neighborsto form larger economies of scale. People selling theirproperties retired into adjacent towns. Governmentaid poured into the region, but at a lesser scale thanin the previous drought. Again, there was littleeffect upon the national economy. The governmentresponded by initiating more technical programs.Irrigation and river basin projects were started, andweather prediction, control and modification programsresearched and carried out. The Great PlainsConservation Program established mechanisms forrecharging groundwater, increasing runoff, minimizingevaporation, desalinizing soil, and minimizing leakagefrom irrigation canals. Since the 1950s, these programshave increased in complexity, to the point that droughtnow has little impact on the regional or nationaleconomies.

It would appear that, in the United States, the effectof drought on the Great Plains has been dramaticallyreduced to a level unachievable at present in mostAfrican countries. The evolution of drought-reductionstrategies indicates an increasing commitment togreater social organization and technological sophisti-cation. Hence, the local and regional impacts ofrecurrent droughts have diminished over time;however, the potential for catastrophe from rarerevents may have increased. The latter effect is due tothe reliance placed on technology to mitigate the


drought. If the social organization and technologycannot cope with a drought disaster, then the conse-quences to the nation may be very large indeed.Whereas droughts at the turn of the century had to beborne by individuals, and had little effect outside theregion, society today in the United States is structuredso that the failure of existing technology to minimize adrought will severely disrupt the social and economicfabric of that nation. A ripple effect will also permeateits way into the international community, becausemany Third World countries are reliant upon the GreatPlains breadbasket to make up shortfalls should theirown food production fail. This international effect willonly be exacerbated if droughts occur worldwide atthe same time. Until the 1990s, this appeared unlikelybecause the timing of drought on the Great Plainswas out of sequence with that in other continents.However, since the early 1990s, droughts haveoccurred globally at the same time. This is soberingconsidering the international community’s inability torespond to more than one major drought at a time.

Societies that don’t expect drought:the United Kingdom(Whittow, 1980; Morren, 1983; Young, 1983)

The United Kingdom is not usually associated withdrought; however, the south and south-east sections ofEngland have annual rainfalls of less than 500 mm. In1976 England was subjected to 16 months of drought,the worst recorded in its history. For the previous fiveyears, conditions had become drier as the rain-bearing,low-pressure cells, which usually tracked acrossScotland and Norway, shifted northward into theArctic. The low-pressure cells were replaced withunusually persistent high-pressure cells that oftenblocked over the British Isles, and directed the north-ward shift of the lows. By the summer of 1976, poorrainfalls affected many farmers and temperaturesuncharacteristically soared over 30°C. Domestic watersupplies were hard hit as wells dried up. By the end ofJuly, water tankers were carrying water to townsthroughout the United Kingdom, and water restrictionswere brought into force. In some cases, municipalwater supplies were simply turned off to conservewater. In mid-August, at the height of the drought, theThames River ceased to flow above tide limits. BySeptember, there were calls for national programsto mitigate the worsening situation. When the govern-ment finally shifted into action, the drought had

broken. By November of 1976, most of the country hadreturned to normal rainfall conditions.

The drought was severe, not because of a lack ofrain, but because of increased competition for availablewater supplies that in some parts of the country hadtraditionally been low. Since the Industrial Revolution,public priority ensured water supplies to cities andindustry. Commercial water projects were commonand, where they were not available, councils wereempowered to supply water. The Water Act of 1945encouraged amalgamation of local authorities andnationalization of private companies. A severe droughtin 1959 resulted in the Water Resources Act of 1963;this led to the establishment of river authorities toadminister national water resources, and a WaterResources Board to act in the national interest. By1973 there were still 180 private companies supplyingwater, and through the Water Act of 1973, ten regionalwater authorities were set up to take control of allmatters relating to water resources. In addition, acentralized planning unit and research center were setup. It would appear that the United Kingdom shouldhave been well-prepared for the drought; but, despitethis legislation, per capita consumption of water hadincreased as standards of living increased. Water mainshad been extended into rural, as well as suburbanareas, so that agriculture was reliant upon mains supplyfor irrigation and supplementary watering.

As the drought heightened in June of 1976, regionalwater authorities complained that their legal responseswere either too weak or too extreme. Either they couldban non-essential uses of water such as lawn wateringby ordinary citizens, or they could overreact and cut offdomestic water supplies and supply water from streethydrants. They had no control whatsoever over indus-trial or agricultural usage. Except in south Wales,industry at no time had water usage drasticallycurtailed. On the other hand, water mains in manyplaces were shut down, with special wardens appointedto open water mains when a fire broke out. In August,the national government gave the water authoritiesmore powers through the Drought Act. The authoritieswere free to respond to local conditions as they sawfit. Publicity campaigns managed to get domesticconsumers, who were collectively the largest waterusers, to decrease consumption by 25 per cent.Because there were virtually no water meters in theUnited Kingdom, it was not possible to reduceconsumption by imposing a limit on the volume of


water used at the individual household level andcharging heavily for usage beyond that limit. Smallertowns had their water shut off, but at no time was itsuggested that London should have its water supplyreduced. When canals and rivers dried up, there wasconcern that the clay linings would desiccate andcrack, leading to expensive repairs to prevent futureleakage or even flooding. Ground subsidence on claysoils became a major problem, with damage estimatedat close to £50 million. This latter aspect was one of thecostliest consequences of the drought.

The agriculture sector was badly affected and poorlymanaged. From the onset, grain and livestock farmerswere disadvantaged by the drier weather, receivinglittle assistance. Just when they needed water most – atthe time of highest moisture stress – irrigation farmersfound their water supplies cut by half. Britain’s agricul-ture had become so dependent upon mains watersupply that it virtually lost all ability to respond to thedrought. This was not only a problem with the 1976drought – it had been a growing concern throughoutthe 1960s. Fortunately, because of Britain’s entryinto the European Common Market, the agriculturalsector was insulated economically against the fullconsequences of the drought. The early part of thesummer saw production surpluses, but at no time didagricultural production for the domestic marketdecrease to unbearable levels. In fact, many farmersresponded to the drought in its early stages andincreased acreages to make up for decreased yields.

The government’s response to the drought wasminimal. It began to act only at the beginning ofAugust 1976: looking in detail at the plight of farmersand giving water authorities additional powers throughthe Drought Act. By October, there was considerabledebate in parliament about subsidies and food pricing,but in fact the drought was over. The drought of 1976cannot be dismissed as inconsequential; it did disruptpeople’s lives at all levels of society. The consequenceswere inevitable given the changes in water usage inthe United Kingdom, especially the dependence ofagriculture upon mains supplies and the increases indomestic water consumption. The national govern-ment was poorly prepared for the drought and still hasnot taken steps to rationalize water management inthe United Kingdom to prepare for inevitable futuredroughts. This case illustrates two poorly realizedfacts about drought. Firstly, drought is ubiquitous: allcountries can be affected by exceptional aridity.

Drought is as likely in the United Kingdom (orIndonesia or Canada) as it is in the Sahel, given theright conditions. Secondly, the national response todrought in westernized countries is often similarto that of many Third World countries. The maindifference is that in the Third World, governments areoften inactive, usually because they are not economi-cally able to develop mechanisms of drought mitigationand alleviation, while governments in westernizedcountries such as the United Kingdom (or otherwestern European countries, which also experiencedthe 1976 drought) are not prepared to accept the factthat droughts can happen.

Those that lose(Glantz, 1977; Bryson & Murray, 1977; Whittow, 1980;Gribbin, 1983; Hidore, 1983)

Africa’s Sahel has succumbed to drought since coolingsea surface temperatures in the adjacent Atlanticbecame pronounced in the early 1960s. The seasonalswitch of the intertropical convergence over Africafailed as a result. The consequences first affected,between 1968 and 1974, the region betweenMauritania and Ethiopia. Human-induced feedbackmechanisms, as described above, exacerbated droughtconditions. The coping responses of people in theSahel soon became inadequate to ensure the survivalof populations. One of the reasons that Saheliancountries have suffered so dramatically is the relativeinaction of national governments in responding todrought. In many of these countries, local groups havetraditionally played a crucial role in minimizing theeffects of drought. Farmers have always been able toadjust their mode of farming to increase the probabil-ity of some crop production, to diversify, to resortto drought-tolerant crops, or to respond instantly tochanging soil and weather conditions. Nomads mayresort to cropping under favorable conditions, followregional migratory patterns to take advantage ofspatially variable rainfall or, under extreme conditions,move south to wetter climates. All groups have strongkinship ties, which permit a destitute family to fall backon friends or relatives in times of difficulty. There isvery little national direction or control of this type ofactivity, especially where national governments areweak and ineffective. In many of these countries,colonialism has weakened kin group ties and forcedreliance upon cash cropping. Westernization, espe-cially since the Second World War, has brought


improved medical care that has drastically reducedinfant mortality and extended lifespan. Unfortunately,birth rates have not decreased and all countries areburdened with some of the highest growth rates in theworld: in the 1980s, 3 per cent and 2.9 per cent forSudan and Niger, respectively. Technical advances inhuman medicine also applied to veterinary science,with the result that herd sizes grew to keep pace withpopulation growth. In the Sudan between 1924 and1973, livestock numbers increased fivefold, abetted byinternational aid development and the reduction ofcattle diseases. The government responded to theseincreases during past droughts by providing moneyfor construction of wells and dams. Otherwise, thenational neglect was chronic. Subsistence farming isthe primary economic activity of 80–90 per cent ofpeople in the Sahel. Almost 90 per cent of the popula-tion is illiterate, and four of the countries making upthe Sahel are among the twelve poorest countries inthe world, and getting poorer.

During the first series of droughts, in the 1960s, thecollapse of agriculture was so total that outward migra-tion began immediately. By 1970, 3 million people inthe west Sahel had been displaced and needed emer-gency food. The international response was minimal.When the drought hit Ethiopia, aid requests to thecentral government were ignored. As the droughtcontinued into 1971, large dust storms spread south-ward. The United States was responsible for80 per cent of aid and Canada, a further 9 per cent,even though many of the afflicted countries had oncebeen French colonies. The fifth year of the droughtsaw not only community groups but also nationalgovernments overwhelmed by the effects. Outwardmigration of nomads led to conflict with pastoralistsover wells, dams and other watering areas. Intertribalconflicts increased. Migration to the cities swelledurban populations, leading to high unemploymentand large refugee camps, often resented by locals.Migrations occurred beyond political – but notnecessarily traditional – borders. Up to a third ofChad’s population emigrated from the country.Chronic malnutrition and the accompanying diseasesand complaints – such as measles, cholera, smallpox,meningitis, dysentery, whooping cough, malaria, schis-tosomiasis, trypanosomiasis, poliomyelitis, hepatitis,pneumonia, tuberculosis, and intestinal worms –overwhelmed medical services. Chronic malnutritionusually results in death, but not necessarily by

starvation. Lack of nutrition makes the body moreprone to illness and disease, and less able to recover.

As far as aid efforts were concerned, ‘the silence wasdeafening’. By 1972, every national government andinternational aid group knew of the magnitude of thedisaster, but no plan for drought relief was instituted.The countries were either politically unimportant, oreveryone believed the drought would end with the nextrainy season. A survey of the natural disasters coveredin the New York Times between 1968 and 1975 showsno references to the Sahelian drought. After 1972,national governments gave up and relied completelyupon international aid to feed over 50 million people.Six west African countries were bankrupted. In thesixth year of the drought, the first medical survey wascarried out. In Mali, 70 per cent of children, mostlynomads, were found to be dying from malnutrition.In some countries, food distribution had failedcompletely and people were eating any vegetationavailable. Until 1973, only the United States wascontributing food aid. The Food and AgriculturalOrganization (FAO) of the United Nations did notbegin relief efforts until May 1973, the sixth year ofthe drought, when it established the Sahelian TrustFund. The difficulties of transporting food aid wereinsurmountable. Sahelian transport infrastructure waseither non-existent or antiquated. Seaports could nothandle the unloading of such large volumes of food.The railroad inland from Dakar, Senegal, was over-committed by a factor of 5–10. Where the railwaysended, road transport was often lacking or brokendown. Eventually, camel caravans were pressed intoservice to transport grain to the most destitute areas. Insome cases, expensive airlifts were utilized. In 1973,only 50 per cent of the required grain got through; in1974, the seventh year of the drought, this proportionrose to 75 per cent. In some cases, the aid was entirelyineffective. For instance, grain rotted in ports, foundits way to the black market or was of low quality.

The environmental impact of the drought wasdramatic. Grasslands, overgrazed by nomads movingout of the worst hit areas, subsequently fell victim towind deflation. The process of desertification acceler-ated along all margins of the Sahara. Between 1964 and1974, the desert encroached upon grazing land 150 kmto the south. Yet the lessons of this drought wereignored. By the early 1980s, and despite the death tollof the early 1970s, the population of the Sahel grew by30–40 per cent. At this time, five of the world’s nine


poorest countries made up the Sahel. Land degrada-tion continued, with more wells being drilled tosupport the rebuilding and expansion of livestockherds. Governments could not control the populationexpansion, and relied heavily upon food imports tofeed people – even under favorable conditions. The1983 drought was a repeat performance. The samescript was dusted off; the same national and interna-tional responses were replayed. This time, however,rather than beginning in the west and moving east, thedrought began in Ethiopia and moved westward intothe Sudan. If it had not been for a chance filming by aBritish television crew, the drought would again havegone unnoticed. This chance discovery evoked twointernational responses never previously seen: the firstattempt to implant drought aid in an area before adrought, and the efforts of Bob Geldof. Both of theseresponses will be discussed at length towards the endof this chapter.

Besides the obvious point about weak, impoverishednational governments being incapable or unwilling todeal with drought, there are two points to be madeabout the continuing drought in the African Sahel.Firstly, Sahelian droughts have large temporal vari-ability. It has already been emphasized, in Chapter 2,that most droughts worldwide appear to coincide withthe 18.6-year lunar cycle. What has not beenemphasized is the fact that this lunar cycle may besuperimposed upon some longer term trend. TheSahelian droughts persisted for three decades.Secondly, some developed countries may not haveexperienced their worst drought. For instance, the1982–1983 Australian drought was the worst in200 years. The next worst drought in 200 yearsoccurred in 1993–1995 and again in 2002–2003. Thereis evidence from tree-ring studies that some droughts inAustralia in the last two to three centuries have lasted aslong as the present Sahelian conditions. Western coun-tries should not be so ready to chastise the governmentsof Sahelian countries about their inability to deal withdrought. Australia, and many other countries, has yet tobe tested under similar conditions over such a longperiod – conditions that should be viewed as likelyunder our present climatic regime.

Those that win(Charnock, 1986)

Given the persistence of the African drought, and thefact that so many people face a losing battle, the

question arises whether any African country can reallycope with drought. There is one notable exception tothe bleak news, namely what has been achieved inBotswana. Botswana, in 1986, was in its fifth consecu-tive year of drought – a record that easily matches thatof any country in the Sahel. Yet no one appeared tobe dying from starvation, although two-thirds of itsinhabitants were dependent upon drought relief.Botswana occupies the southern hemisphere equiva-lent of the Sahel region of Africa, at the edge of theKalahari Desert. Long-term droughts have continuallyafflicted the country, with average rainfall varyingerratically from 300 to 700 mm yr-1. The country hasbecome newly independent in the last two decades;but, unlike other decolonized African countries, it hasnot been wracked by intertribal conflict and has anadministration that is open to self-criticism. After asingle year of drought in 1980, the governmentcommissioned an independent study on drought-reliefefforts, finding measures severely wanting. Followingthat report, the government overhauled the storageand distribution network, to the point that 90 per centof the population requiring food aid during the 1986drought year received uninterrupted supplies. Thegovernment also instituted a program of early warningthat tracked the intensity of the drought in Botswana.Every month rainfall figures were collected andanalyzed to produce a contour map that pinpointedthose areas favorable for cropping and those areas instress. In 1983, a computer model was initiated thatpredicted soil moisture and maximum crop yield. Themodel takes rainfall data and makes projections of cropyield until harvesting, based upon the premise of ‘nostress’ and ‘continuing stress’. This model is updatedwith daily rainfall readings fed into a centralizedcollecting agency via two-way radio from all parts ofthe country. The model permits farmers to ascertainthe risk to crops if they are planted now, the risk tocrops if they are cultivated, and the risk to crops ifharvesting is postponed. Because soil moisture iscalculated in the model, it is possible to determinethe reserves of soil moisture from the previous cropyear.

Botswana monitors the state, not only of its agricul-ture, but also of its people. Each month, two-thirds ofits children under the age of five are weighed at healthclinics, to detect undernourished children with a bodyweight less than 80 per cent the expected weight fortheir age. Areas where drought effects are manifesting


as undernourishment can be pinpointed and reportedto a central drought committee. This committee, withaccess to government channels, relays informationabout supplies back to local committees. The govern-ment firmly believes that the loss of rural incomeduring drought, and its attendant unemployment,exacerbates the shortage of food. In the 1986 drought,the government temporarily employed over 70 000people in rural areas, replacing one-half of lost incomes.These programs were not in the form of food-for-workbut operated strictly on a cash-for-work basis. Projects,selected by local committees, were usually drought-mitigating. In addition, the drought program offeredfree seed, grants for land cultivation, subsidies for oxenhire, and mechanisms for selling drought-weakenedanimals at above market prices. Government-trainedagricultural officers made efforts to enthuse the ruralcommunity, to provide the expertise to maintainfarmers on the land during the drought, and to makecertain that maximizing the farmers’ potential did notresult in degradation of available arable land. Theemphasis of this program is on keeping the ruralpopulation on the land, and keeping them self-reliantat existing living standards. In the 1980s, two factorsjeopardized the program: growing national apathyand a 3.9 per cent annual growth rate in population.Unfortunately, all these efforts took place before theoutbreak of AIDS in Africa. At present, the greatest riskto drought mitigation in Botswana is the breakdown insophisticated government-sponsored programs due tothe death of talented civil servants as the result of thistwentieth century scourge.

Laissez-faire: the Australian policy(Waring, 1976; Tai, 1983)

Australia is a country plagued by drought, whereefficient grazing and cultivation systems have beendeveloped to cope with a semi-arid environment.Droughts rarely occur at the same time across thecontinent. Only the 1895–1903 drought affected thewhole country. Most Australian farmers are financiallyviable as long as they experience a minimum of threeconsecutive harvests in a decade. This expectationmakes it difficult to separate dry years from droughtyears. The 1982–1983 drought, which was the worstin Australian history up to that point in time, waspreceded by several dry years in the eastern states.Even during the severest times in 1982, the federalgovernment was slow to instigate drought relief

schemes. This inability to perceive the farmer’s plightwas one of the reasons Malcolm Fraser’s Liberal–Country Party coalition lost power in the March 1983federal election. Response to drought in Australia,thus, falls into two distinct categories. The first line ofdefence occurs with the individual farmer’s response,as discussed above. Second, when the farmer fails toprofitably manage a property, then the state andfederal governments attempt to mitigate a drought’sworst effects and ensure that quick economic recoveryis possible when the drought breaks. Unfortunately,this response is too piecemeal and at times politicallymotivated.

One of the first responses to drought is theperception that areas are being affected by exceptionaldeficiencies in rain. The Bureau of Meteorologymaintains an extensive network of rain gauges, andclassifies an area as drought-affected when annualrainfall is below the tenth percentile. Many statesuse this information and data on dam levels, crop,pasture and range condition, and field reportsthrough government agencies and politicians, todeclare districts drought-affected. This status isusually maintained until sufficient rainfall has fallento recharge groundwater or regenerate vegetation tosustain agriculture at an economic level. State reliefto farmers can take many forms. Initially, attempts willbe made to keep stock alive. Subsidies may be given tosink deeper bores, import fodder to an area, carry outagistment of livestock in unaffected areas, or importwater using road or rail tankers.

The 1982–1983 drought exemplifies the range ofmeasures that can be adopted to lessen the hardship ofdrought. Over 100 000 breeding sheep were shippedfrom the eastern states to Western Australia, a distanceof 3000 km. The Australian National Railways arrangedspecial freight concessions and resting facilities enroute, while the federal government provided a75 per cent freight subsidy. In addition, assistance wasgiven to farmers paying interest rates over 12 per cent,and a 50 per cent subsidy was provided for importedfodder for sheep and cattle. Despite such measures,droughts can be so severe that it is only humane todestroy emaciated stock. In the 1982–1983 drought,local councils forced to open and operate slaughteringpits were reimbursed their costs, and graziers weregiven a sheep and cattle slaughter bounty. In SouthAustralia, each head of sheep and cattle generated asubsidy of $A10 and $A11, respectively. In addition,


the federal government matched this sheep subsidy upto a maximum of $A3 per head. However, the main aimof the federal government’s drought aid package, whichtotalled approximately $A400 million, was to preventthe decimation of breeding stock so vital to rebuildingAustralia’s sheep and cattle export markets.

Attempts at conserving and supplying water werealso energetically pursued. The Victorian governmentcancelled indefinitely all leave for drilling-rig operatorsworking in the mining industry, and redirected them tobore-drilling operations. The number of water tankersproved insufficient to meet demand, and the stateresorted to filling plastic bags with water and shippingthem where needed. In New South Wales, waterallocations from dams and irrigation projects were cutin half and then totally withdrawn as the capacity ofmany dams dwindled to less than 25 per cent. BrokenHill’s water supply was threatened as the Darling Riverdried up. Water in Adelaide, which is drawn from theMurray River, reached unsafe drinking limits as salinityincreased through evaporation. To save trees withindowntown Adelaide, trenches were dug around thetrees and roots were hand-watered. Many other countrytowns and cities suffered a severe curtailment of waterusage except for immediate domestic requirements.

While the 1982–1983 drought caused farmerseconomic hardship, the financial response of state andfederal governments was relatively minor. The droughtcost farmers an estimated $A2500 million and resultedin a $A7500 million shortfall in national income.Government relief (both state and federal) probably didnot exceed $A600 million, with most of this aid directedtowards fodder subsidies. In Australia, in most cases,the individual farmer must bear the financial brunt ofdroughts and 1982–1983 drought was no exception.Many farmers were forced to increase their debt load,with personal interest rates as high as 20 per cent, to thepoint where their farming operations became econom-ically unviable even when favorable times returned.The continued high interest rates following this droughtand the over-capitalization of many Australian farmershave ensured a rural decline throughout the latter partof the twentieth century.

INTERNATIONAL RESPONSE

International relief organizations

By far the largest and most pervasive internationalrelief organizations are the ones associated with the

United Nations or the Red Cross. The United NationsDisaster Relief Office (UNDRO) was established bythe General Assembly in 1972, and financed partlythrough the United Nations budget and the VoluntaryTrust Fund for Disaster Relief Assistance. The latterchannel allows countries, sensitive about contributingto the overall running cost of the United Nations, toparticipate in disaster relief funding. UNDRO gathersand disseminates information about impendingdisasters (in the form of situation reports) to govern-ments and potential donors. It conducts assessmentmissions and, working through 100 developing coun-tries, uses trained, experienced personnel who canopen channels of communication with governmentofficials in disaster-prone areas, irrespective of thepolitical beliefs of that government or area. It has aresponsibility for mobilizing relief contributions andensuring the rapid transport of relief supplies. Duringtimes of disaster, it can call upon other agencies forpersonnel to implement programs. It also has the dutyof assessing major relief programs, and advisinggovernments on the mitigation of future naturaldisasters.

There are a number of other United Nations officesthat are involved less directly in drought relief. TheUnited Nations Children’s Fund (UNICEF), Foodand Agriculture Organization (FAO), World HealthOrganization (WHO), and World Food Programme(WFP), while mainly involved in long-term reliefin underdeveloped countries, can provide emergencyservices and advice. In addition, the World Meteoro-logical Organization (WMO) monitors climate, and theUnited Nations Educational, Scientific and CulturalOrganization (UNESCO) is involved in the construc-tion of dams for drought mitigation. The InternationalBank for Reconstruction and Development (WorldBank) finances these mitigation projects, as well aslending money to Third World countries forreconstruction after major disasters.

Temporary infrastructure to handle specific disas-ters can also be established from time to time. Inresponse to the Ethiopian–Sudanese drought of theearly 1980s, the United Nations Office for EmergencyOperations in Africa (UNOEOA) was establishedto coordinate relief activities. This organizationchanneled $US4.5 billion of aid from 35 countries,47 non-government organizations and half-a-dozenother United Nations organizations, into northernAfrica during this drought. Credited with saving


35 million lives, this office is one of the success storiesof the United Nations. It was headed by MauriceStrong, a Canadian with a long-time involvement in theUnited Nations and a reputation for being able to ‘pullstrings’ – especially with uncooperative governmentswith obstructive policies. He persuaded Sudan’s GaarNimeiny to open more refugee camps to relieve thedesperate and appalling conditions in existing camps,and coerced Ethiopia’s Colonel Haile Mengistu toreturn trucks (commandeered by the army for its wareffort against Eritrean rebels) to the task of clearingthe backlog of food stranded on Ethiopian docks.

The United Nations is not always the altruisticorganization implied by these descriptions. Many of itsbranches, most notably the Food and AgriculturalOrganization (FAO), are politicized and riddled withinefficiency, if not outright cronyism. FAO, which has a$US300 million annual budget to develop long-termagricultural relief programs, in the 1980s spent40 per cent of its budget on paternalistic and corruptadministration. At the height of the Ethiopian faminein 1984, FAO administrators delayed for 20 days insigning over $US10 million in emergency food aid,because the Ethiopian ambassador to FAO hadslighted the FAO. The most efficient FAO field workerin Ethiopia was recalled during the drought because heinadvertently referred to himself as a UN, instead of aFAO, worker in statements to the press. Much ofthe FAO administration operates as a Byzantine merry-go-round, where the mode of operation is moreimportant than the alleviation of drought in membercountries.

Second in effort to the United Nations is theInternational Red Cross (Red Crescent in Islamiccountries), which constitutes the principal non-government network for mobilizing and distributinginternational assistance in times of disaster. Foundedin 1861, it consists of three organizational elements:the International Committee of the Red Cross (ICRC),an independent body composed of Swiss citizens con-cerned mainly with the victims of armed conflict;the 162 national societies, which share the principlesand values of the Red Cross and conduct programs andactivities suited to the needs of their own country;and the League of Red Cross Societies (LORCS),which is the federating and coordinating body for thenational societies and for international disaster relief.The International Red Cross, because of its strict non-aligned policy, plays a unique and crucial role in

ensuring disaster relief can reach refugees in areasstricken by civil war, or controlled by insurgencygroups with little international recognition.

The aims of the Red Cross are duplicated by anumber of religious and non-affiliated organizations.One of the largest of these is Caritas (InternationalConfederation of Catholic Charities), which wasestablished for relief work during the Second WorldWar, and which afterwards continued relief effortsduring natural disasters. The Protestant equivalent isthe World Church service. Other denominationalgroups maintaining their own charities include theSalvation Army, Lutheran Church, Adventist Develop-ment and Relief Agency (ADRA) and the Quakers(who in the United States sent aid to Cuba after itwas devastated by hurricanes). Religious charities alsoexist in Islamic, Hindu and Buddhist countries. Finally,there are a number of non-profit organizations thatcan raise money for immediate disaster relief, long-term reconstruction and disaster mitigation. Theseorganizations include AustCare, CARE, World Vision,Oxfam, and Save the Children. Some have internalizedadministrations that pass on as much as 95 per centof all donations, while others utilize professionalfundraising agencies that may soak up as much as75 per cent of donations in administrative costs. Theseorganizations periodically collect monies using door-knock appeals, telethons, or solicitations throughregular television, radio, newspaper, and magazineadvertisements.

International aid ‘flops’

Whenever the international community is mobilized torespond to large droughts and impending famine formillions of people, the question arises ‘Does the aid getthrough?’. There are three difficulties with supplyingaid to Third World countries. Firstly, the suddenincrease in the volume of goods may not be efficientlyhandled because existing transport infrastructure inthose countries is inefficient or non-existent. If relief issent by boat, it may sit for weeks in inadequate harborsawaiting unloading. Once unloaded, the aid may sit formonths on docks awaiting transshipment. Such wasthe case with grain supplies sent to Ethiopia duringthe drought at the beginning of the 1980s. If a droughthas occurred in an isolated area, there simply may beno transport network to get relief to the disaster site.This problem may be exacerbated by the destruction oftransport networks because of civil war.


Secondly, Third World countries may have anelement of corruption or inefficiency to their adminis-tration. The corruption, if it exists, may lead to reliefaid being siphoned off for export, ransom or sale on theblack market. Finally, inefficient administrations mayslow the relief aid with paper work or be unaware thatrelief aid is even necessary.

By far one of the biggest ‘flops’ to occur for thesereasons was the international aid effort for theSudanese drought of 1984–1985. This drought wasthe first in Sudan to be predicted and prepared forbefore its occurrence. In the previous two years,Ethiopia had been wracked by drought that wentvirtually unnoticed for a year, and was then exposed onBritish television. The Ethiopian drought generatedconsiderable attention about why it had occurred, andhow relief was getting through to the people. Itbecame a foregone conclusion that the drought wasspreading westward and would eventually enter Sudan,a country that had magnanimously permitted over1 million Ethiopian refugees to cross its borders forrelief aid. The management of the Ethiopian faminewas a disaster in itself. As it had in 1975, the Ethiopiangovernment at first failed to admit that it had a famineproblem, mainly because the drought was centeredin rebellious northern provinces. It continued to sellgarden produce to other countries instead of divertingfood and produce to the relief camps that had beenset up or had spontaneously appeared. Authoritiesdeliberately strafed columns of refugees fleeing thedrought-affected areas and allowed relief aid to pile upin its eastern ports. At the same time, an enormoussum of money was spent to dress up the capital for ameeting of African nations.

The United States government recognized theseproblems and decided that a similar situation wouldnot hinder relief aid for the impending drought inSudan. It pledged over $US400 million worth of foodaid and set about overcoming the distribution problem,which consisted in getting the food from Port Sudan onthe Red Sea to the western part of Sudan, which wasisolated at the edge of the Sahara Desert. The UnitedStates hired consultants to evaluate the distributionproblem, and it was decided to use the railway betweenPort Sudan and Kosti, or El-Obeid, to ship grain westbefore the onset of the rainy season. The grain wouldthen be distributed westward using trucks. The UnitedStates went so far as to warn the residents of thewestern part of Sudan that a drought was coming, and

that food aid would be brought to their doorstep.The inhabitants of that part of the country were notencouraged to migrate eastward. Thus, there was goingto be no repeat of the large and deadly migrations thatoccurred in Ethiopia.

However, no one assessed the Sudanese railwaysystem’s ability to handle transportation of such a largevolume of grain. In fact, scrapped trains in westerncountries were in better shape than those operating onthe Sudanese railway system. A decision was made torepair the railway, but inefficiencies delayed this formonths. Finally, it was decided to move the food formost of the distance by truck. No one had evaluated thetrucking industry in the Sudan. The operation absorbedmost of the trucks in the country. The local truck drivers,realizing they now had a monopoly, went on strike forhigher wages. By the time the railway and truckingdifficulties were overcome, the drought was well-established, and few if any people had migrated to theeast. They were now starving. When the transportsituation was sorted out, the rains set in, and many trucksbecame bogged on mud tracks or while attempting toford impassable rivers, or hopelessly broken down.Meanwhile, grain supplies were rotting in the openstaging areas in the east, because no one had consideredthe possibility that the grain would still be undistributedwhen the rainy season broke the drought.

The decision was finally made to airlift food suppliesinto remote areas using Hercules aircraft, a decisionthat increased transport costs fourfold. But this was notthe end of the story. The rains made it impossible forthe planes to land, and there was no provision toparachute supplies to the ground. Like sheep in anAustralian flood awaiting airdropped bales of hay, thestarving inhabitants of west Sudan waited for bags ofgrain to be airdropped. The Hercules would fly as lowand as slowly as possible, as bags of grain were shovedout without any parachutes to slow them down. Thebags tumbled to the ground, bounced and broke,scattering their contents. The Sudanese then raced onfoot to the scene to salvage whatever grain they couldbefore it was completely spoilt in the mud.

PRIVATE RESPONSES: BOBGELDOF

(Geldof, 1986)

Without Bob Geldof there would have been no BandAid or Live Aid: his name is now synonymous with


spontaneous, altruistic, international disaster reliefefforts – although he was not the first celebrity to takeon massive famine relief. In 1891, Leo Tolstoy and hisfamily organized soup kitchens, feeding 13 000 peopledaily in the Volga region of Russia following destruc-tion of crops by inclement weather.

Bob Geldof was a musician from Ireland whoworked with a band called the Boomtown Rats thathad one international hit single in 1979, ‘I don’t likeMondays’. They were the number one band inEngland that year. Because he was a musician, Geldofhad contacts with most of the rock stars at the time andwas also aware of the sporadic efforts by musicians tohelp the underprivileged in the Third World. Forexample, following the 1970 storm surge in EastPakistan (now Bangladesh) and that country’s indepen-dence, musicians organized a concert, the ‘Concert forBangladesh’, to raise money to help overcome thedisaster. It was staged free by several famous rockgroups, and raised several million dollars throughticket sales and royalties on records.

Bob Geldof, like many other people in 1984, wasappalled by the plight of the Ethiopians – and the factthat the famine remained virtually unknown to theoutside world until a British television crew, led byreporter Michael Buerk, chanced upon it. Buerk’s filmshowed the enormity of the disaster, and painted adismal, hell-like picture. Children were being selec-tively fed. Only those who had a hope of surviving werepermitted to join food queues. People were dyingthroughout the film. In the midst of this anarchy, theEthiopian people stoically went on trying to live inthe camps, which were over-crowded and lackingmedical aid. Their honesty formed a striking contrastto the greed of most people viewing the show. Starving,Ethiopian refugees would not touch food in openstores unless it was given to them. No one appeared tobe doing anything to help the refugees. The Ethiopiangovernment had denied the existence of the drought,and the international monitoring agencies had failed tobring it to world attention. UNDRO in particularappeared hopelessly inefficient.

The emotional impact of the BBC film stunned allviewers, including Bob Geldof; however, a rockmusician was hardly the most obvious candidate toorganize and lead the largest disaster relief effort inhistory. The day after the screening of the documen-tary, Geldof began to organize, coerce, and lie to puttogether a group of musicians to produce a record for

Christmas 1984, the profits of which were to bedonated to Ethiopian relief. The group was made up ofthe most popular rock musicians in Britain at the time:Sting, Duran Duran, U2, Style Council, Kool and theGang, Boy George and Culture Club, Spandau Ballet,Bananarama, Boomtown Rats, Wham, Ultravox,Heaven 17, Status Quo, David Bowie, Holly, and PaulMcCartney. The band was called Band Aid, an obviousplay on words, but chosen in reality because the namewould make people aware of the futility of this one actof charity in overcoming the famine. In the end, BobGeldof managed to convince all people associated withthe record to donate their services free. The Britishgovernment, after protracted lobbying, waived thevalue added tax. The proceeds of the record were to beput into a special fund to be distributed for relief aid asa committee saw fit. The record contained the song‘Do They Know It’s Christmas?’ on side one, and somemessages from rock stars on side two. It cost £1 30pand reached number one the day it was released, sixweeks before Christmas. Within weeks it was selling320 000 copies per day and utilizing the pressingfacilities of every record factory in Britain, Irelandand Europe to meet demand. People bought 50 copies,kept one, and gave the rest back for resale; they wereused as Christmas cards, sold in restaurants, andreplaced meat displays in butchery windows. Musi-cians in the United States (USA for Africa) and Canada(Northern Lights) subsequently got together toproduce a similar type of album, which had the samesuccess in North America. Over twenty-five BandAid groups (‘Austria für Afrika’, ‘Chanteurs SansFrontiers’, and three German groups amongst others)were formed worldwide, raising $US15 million forfamine relief.

Following the success of the Christmas album, in1985 Bob Geldof organized a live concert. It was tobecome the first global concert and would run for17 hours on television, using satellite hook-ups. Itwould start in Wembley Stadium in London, and thenend in Philadelphia with five hours of overlap. Famousbands would alternate on a rotating stage. The concertwould also draw in television coverage of smallerconcerts held in Australia and Yugoslavia. Countrieswould pay for the satellite link-up, and at the sametime be responsible for organizing telethons to raisemoney. Paraphernalia would also be sold, with allprofits going to the Band Aid fund. As much aspossible, overheads were to be cost-free.


Critics said that it could not be done, that theorganization would be too horrendous, the musicianstoo egotistical, that no one would watch it or givedonations. A group of middle-aged consultants hired toevaluate the income potential of such a concert, said itwould raise less than half a million US dollars. Theconcert was organized over three months and it wasnot known until the day of the show, 13 July 1985,exactly what groups would be participating. Manycountries only telecast the concert at the last minute.The chaos, especially in Philadelphia, spelt disaster;but through the efforts of talented organizationalpeople in television and concert promotion, the showwas viewed by 85 per cent of the world’s televisionaudience.

The show blatantly, but sincerely, played upon thetelevision audience’s emotions. At one point, whentelephone donations were lagging, a special video clipof a dying Ethiopian child was played to the words ofthe song ‘Drive’ by the group The Cars. As the childtried repeatedly to stand up, only to fall down eachtime, the song asked who was going to pick him upwhen he fell down. The clip brought a response thatjammed the telephone lines worldwide. Donationswere received from both communist and non-communist countries. One private donation fromKuwait totaled $US1.7 million. Ireland gave the mostmoney per capita of any country, in total $US10million. People there even cashed in their weddingrings and houses. In twenty-four hours Live Aid, as theconcert became known, raised $US83 million.

Live Aid was to ad hoc disaster relief whatWoodstock was to rock concerts. It became just one ofmany unique fundraising efforts. Geldof convincedpeople in many industries and recreational pursuits torun their own types of money-raising programs,spawning an alphabet soup of organizations includingActor Aid, Air Aid, Art Aid, Asian Live Aid, Bear Aid,Bush Aid, Cardiff Valley Aid, Fashion Aid, FoodAid, and so on. In France, which had not successfullylinked in with the fundraising activity for Live Aid,students raised money in School Aid, whose successwas ensured by the refusal of teachers to participate. InMay 1986, Sport Aid saw joggers worldwide paying forthe privilege of participating in a 10 kilometre run.Over 20 million joggers participated worldwide.Sport Aid raised nearly $US25 million, with donationsbeing split between the Band Aid organization andUNICEF.

Bob Geldof personally oversaw the distribution ofrelief aid in Ethiopia and Sudan, and was responsiblefor ensuring, in a last desperate measure, thatHercules planes were used to salvage the Sudandrought relief operation. A committee, Live AidFoundation, was established to distribute monies forrelief efforts. It was run by volunteers, including pro-fessional business people, lawyers, and accountants.Bureaucratic administration was avoided as much aspossible. A board of trustees (Band Aid Trust) wasestablished to oversee the relief distribution. Thisboard acted on the advice of a team of eminent acad-emics who had regular working experience in Africa,from various institutions including the universities ofSussex and Reading; the School of Hygiene andTropical Medicine, and the School of Orientaland African Studies in London; and GeorgetownUniversity in Washington. Of the money raised,20 per cent went for immediate relief, 25 per cent forshipping and transport, and the remainder for long-term development projects vetted by these experts.In addition, the Band Aid Trust approached countriesto undertake joint aid. For instance, the UnitedStates government was ideologically opposed to theEthiopian government, but not to the Sudanese gov-ernment. Band Aid teamed up with United States aidagencies to ensure that relief efforts were directedequally to these two countries. Band Aid handledthe Ethiopian relief, while the United States tackledSudan. Other countries were approached for transportand funding. The Australian government was askedto refurbish ten Hercules transport planes for fooddrops; but in the end donated only one plane withoutsupport funding. Other countries, faced with similarrequests, donated all the support needed to keep theirplanes in the air. President Mitterrand of France, inone afternoon of talks with Geldof, pledged $US7million in emergency relief aid for areas and projectsrecommended by Band Aid field officials.

The Band Aid organization also decided thatmoney should support drought reconstruction andmitigation. Proposals for financing were requested,and then checked for need by a four-person field staffpaid for by private sponsorship. Over 700 proposalswere evaluated, with many being rejected because oftheir lack of relevance or grassroots participation. Bythe end of 1986, $US12 million had been allocatedfor immediate relief, $US20 million for long-termprojects, and $US19 million for freight. Project


funding went mainly to Burkina Faso, Ethiopia andthe province of Eritrea, and Sudan. The largest sumwas for $US2.25 million for a UNICEF program toimmunize children, and the smallest was for $US3000to maintain water pumps in refugee camps in easternSudan.

All these efforts had their critics. Many argued thatGeldof’s fundraising efforts were siphoning money thatwould have gone to legitimate aid organizations.Geldof always contended that the efforts were raisingthe level of consciousness amongst ordinary people.In fact, following the Live Aid concert, donations toOxfam and Save the Children increased 200 per centand 300 per cent, respectively. Geldof foresaw criti-cism that he might personally profit from his fundrais-ing activities and refused to use one cent of the moneyfor administration or personal gain. All work had to bevoluntary, using equipment, materials, labor and faresthat had been donated free and with no stringsattached.

Of all the international relief programs, BobGeldof’s has been the most successful at pricking theconsciences of individuals worldwide – regardless oftheir political ideology. Initially he tapped the musicindustry for support, but has shown that the methodsfor raising aid, with involvement of people from allwalks of life, are almost endless. Geldof’s program ofraising money and distributing funds without anysignificant administration costs is one of the most suc-cessful and efficient ever, putting to shame the effortsof individual countries and the United Nations. Theeuphoria of Band Aid and Live Aid has not worn off;however, Geldof saw them as temporary measures. Atthe end of 1986, the Band Aid umbrella organizationwas wound up and converted to a standing committee.Donations are now being channeled, through this com-mittee, to permanently established aid organizations.Since Ethiopia, there has been no other calamitousdrought, no need for an international response, notesting of the First World’s capacity for altruism in theface of a Third World disaster. Since the Ethiopiandrought, communism has fallen, sectarianism hasformally fragmented nations, greed has overwhelmedcapitalistic economies, and terrorism has permeatedthe voids opened by poverty, disenfranchisement andfailed nationalism. The inevitable specter of calamitousdrought will rise again to challenge the sincerity ofthe ideals that Bob Geldof – for one brief moment –established in response.

CONCLUDING COMMENTS

The effects of drought go far beyond the immediatecrisis years when communities and nations are tryingto survive physically, socially and economically. Of allhazards, drought has the greatest impact beyond theperiod of its occurrence. This does not necessarilyaffect population growth. For instance, while manySahelian countries suffered appalling death tolls inthe 1968–1974 drought, high birth rates of 3 per centensured that populations had grown to pre-droughtlevels within 10–13 years. However, the populationwas skewed, with the number of children under theage of 18 making up 50 per cent of the population insome countries. Such a young population taxesmedical and school facilities and overwhelmseconomic development. Unless exploited, suchnumbers of young people do not contribute directlyto the labor force or to a nation’s economic growth.In addition, survivors of droughts can weigh downfamily units and community groups with an excessivenumber of invalids, especially amongst the young.Many of the diseases that afflict malnourishedchildren can lead to permanent intellectual andphysical impairment. Malnourished children whosurvive a drought may never overcome their malnu-trition, especially if they are orphaned, have manysiblings, exist within a family unit that cannot recovereconomically from the drought, or are displacedfrom home areas where kinship ties can traditionallyprovide support.

Additionally, droughts pose difficult problems inrecovery. If seed stock is consumed, it must beimported quickly to ensure that a country can re-establish the ability to feed itself. While many peoplemigrate out of drought areas to towns where foodcan be more easily obtained, the transport infra-structure is not adequate to ship seed into isolatedrural areas when these people return home. A similarsituation applies to livestock. During major droughts,even in developed countries such as Australia,breeding stock may either perish or be used for food.Livestock herds require several years to be rebuilt, aprocess that can put a heavy financial strain upona farm’s resources because this period of rebuildinggenerates little income. In the longer term, droughtsare often accompanied by human-induced or naturalland degradation. Rangeland may be overgrazedduring a drought to keep livestock herds alive, and in


some cases foliage stripped from shrubs and treesto feed stock. Landscapes can be denuded of anyprotective vegetation, resulting in wholesale loss oftopsoil during dust storms. In some places in NewSouth Wales, Australia, up to 12 centimetres oftopsoil blew away during the 1982–1983 drought.This soil can never be replaced; while growth ofvegetation, required to protect the remaining topsoil,can be a slow process taking years to decades. Thisproblem is especially acute in semi-arid areas.Even in the United States, where a major federalgovernment program was initiated on the GreatPlains following the mid-1930s drought, land rehabil-itation was continuing when the next major droughtstruck in the mid-1950s. For developing countriesstrapped for foreign aid and burdened by debt,such long-term recovery and mitigation programsare receiving such a low priority that complete landdegradation, caused by successive droughts, maybecome a permanent feature of the landscape.


Allison, G.B. and Peck, A.J. 1987. Man-induced hydrologic changein the Australian environment. Bureau of Mineral ResourcesGeology and Geophysics Report 282: 29–34.

Bryson, R. and Murray, T. 1977. Climates of Hunger. AustralianNational University Press, Canberra.

Charnock, A. 1986. An African survivor. New Scientist 1515: 41–43.French, R.J. 1983. Managing environmental aspects of agricultural

droughts: Australian experience. In Yevjevich, V., da Cunha, L.,and Vlachos, E. (eds) Coping with Droughts. Water ResourcesPublications, Littleton, pp. 170–187.

Garcia, R. 1972. Drought and Man, Vol. 1 Pergamon, Oxford.Geldof, B. 1986. Is That It? Penguin, Harmondsworth.Glantz, M. H. 1977. Desertification: Environmental Degradation in

and Around Arid Lands. Westview, Boulder, Colorado.Gribbin, J. 1983. Future Weather: The Causes and Effects of

Climatic Change. Penguin, Harmondsworth.Hidore, J.J. 1983. Coping with drought. In Yevjevich, V., da Cunha,

L., and Vlachos, E. (eds) Coping with Droughts. WaterResources Publications, Littleton, pp. 290–309.

Lockwood, J.G. 1986. The causes of drought with particularreference to the Sahel. Progress in Physical Geography 10(1):111–119.

Morren, G.E.B. 1983. The bushmen and the British: problemsof the identification of drought and responses to drought. InHewitt, K. (ed.) Interpretations of Calamity. Allen and Unwin,Sydney, pp. 44–65.

Rosenberg, N. and Wilhite, D.A. 1983. Drought in the US GreatPlains. In Yevjevich, V., da Cunha, L., and Vlachos, E. (eds)Coping with Droughts. Water Resources Publications, Littleton,pp. 355–368.

Scott, E.P. (ed.). 1984. Life Before the Drought. Allen and Unwin,London.

Tai, K.C. 1983. Coping with the devastating drought in Australia. InYevjevich, V., da Cunha, L. and Vlachos, E. (eds) Coping withDroughts. Water Resources Publications, Littleton, pp. 346–354.

Waring, E.J. 1976. Local and regional effects of drought in Australia.In Chapman, T.G. (ed.) Drought. AGPS, Canberra, pp. 243–246.

Warrick, R.A. 1983. Drought in the U.S. Great Plains: shifting socialconsequences?’. In Hewitt, K. (ed.) Interpretations of Calamity.Allen and Unwin, Sydney, pp. 67–82.

Watts, M. 1983. On the poverty of theory: natural hazards researchin context. In Hewitt, K. (ed.) Interpretations of Calamity. Allenand Unwin, Sydney, pp. 231–262.

Whittow, J. 1980. Disasters: The Anatomy of EnvironmentalHazards. Pelican, Harmondsworth.

Young, J.A. 1983. Drought control practices in England and Wales in1975–1976. In Yevjevich, V., da Cunha, L. and Vlachos, E. (eds)Coping with Droughts. Water Resources Publications, Littleton,pp. 310–325.


INTRODUCTION

The earlier chapter on large-scale storms did notconsider in detail the effects of flooding associated withtropical cyclones, extra-tropical storms or east-coastlows. Nor was it appropriate to discuss large-scaleflooding in that chapter because some of the worstregional flooding in the northern hemisphere hasoccurred in spring in association with snowmelt. Inaddition, rainfall associated with thunderstorms is amajor cause of flash flooding. This chapter examinesflash flooding events and regional floods. Flash floodingrefers to intense falls of rain in a relatively short periodof time. Usually the spatial effect is localized. Modifica-tion of the overall landscape during a single event isminor in most vegetated landscapes; however, in arid,semi-arid, cultivated (where a high portion of the landis fallow) or urban areas, such events can be a majorcause of erosion and damage. Steep drainage basinsare also particularly prone to modification by flashfloods because of their potential to generate the highestmaximum probable rainfalls. Large-scale regionalflooding represents the response of a major continentaldrainage basin to high-magnitude, low-frequencyevents. This regional flooding has been responsiblehistorically for some of the largest death tolls attribut-able to any hazard. When river systems alter course as aresponse to flooding, then disruption to transport andagriculture can occur over a large area.

FLASH FLOODS

Magnitude and frequency of heavyrainfall(Griffiths, 1976)

Short periods of heavy rainfall occur as a result of veryunstable air with a high humidity. Such conditionsusually occur near warm oceans, near steep, highmountains in the path of moist winds, or in areas sus-ceptible to thunderstorms. A selection of the heaviestrainfalls for given periods of time is compiled inFigure 6.1. Generally, extreme rainfall within a fewminutes or hours has occurred during thunderstorms.Events with extreme rainfall over several hours, but lessthan a day, represent a transition from thunderstorm-derived rain to conditions of extreme atmosphericinstability. The latter may involve several thunderstormevents. In the United States, the Balcones Escarpmentregion of Texas (see Figure 6.2 for the location ofmajor placenames mentioned in this chapter) has expe-rienced some of the heaviest short-duration rainfallsin this category. Extreme rainfall over several daysis usually associated with tropical cyclones, whileextreme rainfall lasting several weeks to months usuallyoccurs in areas subject to seasonal monsoonal rainfall,or conditions where orographic uplift persists. In thelatter category, Cherrapunji in India, at the base ofthe Himalayas, dominates the data (9.3 m rainfall in

Flooding as a Hazard

C H A P T E R 6

July 1861 and 22.99 m for all of 1861). In this region,monsoon winds sweep very unstable and moist air fromthe Bay of Bengal up over some of the highest moun-tains in the world. In Australia, the highest rainfall inone year (11.3 m) occurred near Tully in Queensland,a location where orographic uplift of moist air from theCoral Sea is a common phenomenon in summer.

The following equation (6.1) represents the line ofbest fit to the maximum rainfall data in Figure 6.1:

R = 0.42 D0.475 (6.1)where R = rainfall in meters

D = duration in hours

While this equation fits the data well for periods aboveone hour, the line overestimates the amount of rainthat can fall in shorter periods. The latter may merelyrepresent the undersampling – by the existing networkof pluviometers – of spatially small events dropping

121Flooding as a Hazard

DurationMinutes Hours Days Months

Rai

n (c

m)

5 10 20 40 60 3 6 9 12 24 5 10 20 30 2 4 6 12

10

20

50

100

200

500

1000

2000

5000

Fussen, BavariaPlum Point, Jamaica

Curtea de Arges, RomaniaHolt, Mo

Rockport, W. VaD'Hanis, Texas

Dapto, AustraliaSmethport, Pa

Rapid City

Réunion Island

Cherrapunji, India

Fig. 6.1 Maximum rainfall amounts versus time (adapted from Griffiths, 1976).

Los AngelesLake Bonneville

Ousel Cr.St. Lawrence R.

Johnstown

Balcones Escarpment

Pecos R.

Teton R.Great Lakes

Cherrapunji

Tully

SydneyWollongong

Vaiont Dam

Three Gorges

Lake Agassiz

Lake Missoula

Rapid City

Bay of Bengal Katherine Gorge


these large rainfall amounts over this shorter timespan. The line of best fit probably represents a physicallimit to the rate of condensation and droplet formationpossible under the Earth’s present temperature andpressure regime. The amounts of rainfall possible inless than one day are staggering. Within an hour,0.42 m of rain is possible, and within a day, 1.91 m ofrain can fall. The latter value represents more than theaverage yearly rainfall of any Australian capital city.

The large amounts of rainfall at Cherrapunji, India(or, for that matter, at Tully, Queensland), while excep-tional, probably do not disturb the environment nearlyas much as these intense but short falls. Over long timeperiods, locations normally receiving heavy rainfallhave evolved dense vegetation that can absorb theimpact of falling rain, and have developed drainagepatterns that can handle the expected runoff. Mostextreme short-term rainfalls occur in places where veg-etation is sparser, and where drainage systems may notbe best adjusted to contain large volumes of runoff.Here, dislodgement of topsoil by raindrop impactcan also be high. Sheet flow occurs within very shortdistances of drainage divides and overland channelflow will develop within several meters downslope. Asa result, sediment erosion and transport is high andrapid. For these reasons, flash flooding in arid andsemi-arid regions can become especially severe.

In urban areas, where much of the ground is madeimpervious by roads or buildings and where drainagechannels are fixed in location, flash flooding becomesmore likely with much lower amounts of rainfall thanindicated in the above graph. Since 1970, flash floodingin semi-arid or urban catchments appears to beincreasing worldwide, including in the United Statesand Australia.

Flood power(Baker & Costa, 1987)

The amount of work that a flood can perform, and thedestruction caused by a flash flood, are not necessarilydue to the high amounts of rainfall described above.Nor are high flood discharges a prerequisite forerosion. Rather, it is the amount of shear stress, and thestream power – the amount of energy developed perunit time along the boundary of a channel – that ismore significant. It is because of stream power thatfloods in small drainage basins, as small as 10–50 km2,can be more destructive than major floods on theMississippi or Amazon rivers, with discharges

10–1000 times larger. This is particularly so wherechannels are narrow, deep and steep. Stream powerper unit area of a channel is defined by the followingequation:

� = � v (6.2)where � = power per unit boundary area

v = velocity� = boundary shear stress

= �R Swhere � = specific weight of the fluid

(9800 N m-3)S = the energy slope of the flow R = the hydraulic radius of the

channel= A (2d + w)-1

where A = the cross-sectional area of the wetted channel

d = the mean water depthw = the width of the channel

The parameters in Equation 6.2 are not that difficult tomeasure, because many rivers and streams havegauging stations that measure velocity and water depthduring floods. Once the height of a flood has beendetermined, it is relatively simple to calculate thehydraulic radius. The slope of a channel is normallyused in place of the energy slope; however, this mayincrease errors in discharge by 100 per cent duringcatastrophic flash floods. Problems may also arise indetermining the specific weight of the fluid. The valuequoted above is standard for clear water. Duringfloods, however, waters can contain high concen-trations of suspended material that can double thespecific weight of floodwater.

Figure 6.3 plots stream power against drainage basinarea for small flash floods, mainly in the United States,and for the largest floods measured in recent times.Some of these floods, such as the Teton River floodin Idaho on 5 June 1976, can be associated with thecollapse of dams following heavy rains. The collapse ofa dam can greatly increase the magnitude of streampower, because of its effect on hydraulic radius and theenergy slope of the flow. In fact, dam collapses have ledto the largest flash flood death tolls. For example, theJohnstown, Pennsylvania, flood of 31 May 1889 killedover 2200 people. The failure of dams during heavyrains can be attributed to neglect, inadequate designfor high-magnitude events, geological location (for


example, built across an undetected fault line), or mis-chance. The greatest stream power calculated for anyflood in the United States was for the 8 June 1964 OuselCreek flood in Montana. During this flood, streampowers reached 18 600 W m-2. Most of the powerfulfloods in the United States have occurred on quite smalldrainage basins because they have steeper channels. Incomparison to the Ousel Creek flood, the large flood onthe Mississippi River that occurred in 1973, had astream power of 12 W m-2.

For larger drainage basins, two conditions must bemet to attain high stream powers. Firstly, the channelmust be constrained and incised in bedrock. Thisensures that high velocities are reached because watercannot spill out across a wide floodplain. Secondly, theupstream drainage basin must be subject either tohigh-magnitude rainfall events, or to high discharges.These two features severely limit the number of highstream power events occurring on large rivers. Duringthe Teton Dam failure in Idaho in 1976, a streampower was reached in excess of 10 000 W m-2 becausethe river flowed through an incised channel. Similarmagnitudes have occurred through the narrow Kather-ine River gorge in northern Australia. The Pecos Riverin Texas is also incised into a limestone escarpmentsubject to flash flooding, mainly during tropicalcyclones such as Hurricane Alice, which penetratedinland in 1954. Finally, the Chang Jiang (Yangtze)River produces stream power values as high as flashfloods on streams with drainage basins less than1000 km2. At the Three Gorges section of the Chang

Jiang River, snowmelt and rainfall from the TibetanPlateau concentrates in a narrow gorge. This river,in the 1870 flood, flowed at a depth of 85 m withvelocities of 11.8 m s-1.

Figure 6.3 does not plot the largest known floods.Stream power values for cataclysmic prehistoric floodsassociated with Lake Missoula at the margin of theLaurentian icesheet on the Great Plains, and with LakeBonneville, were an order of magnitude larger. Theseresulted from the sudden bursting of ice-dammed lakesas meltwater built up in valleys temporarily blockedby glacier ice during the Wisconsin glaciation. TheMissoula floods reached velocities of 30 m s-1 anddepths of 175 m, producing stream powers in excess of100 000 W m-2, the largest yet calculated on Earth.Similar types of floods may have emptied from theLaurentian icesheet into the Mississippi River andfrom Lake Agassiz eastward into the Great Lakes–St Lawrence drainage system. Equivalent large floodsalso existed on Mars at some time in its geologicalhistory; however, if the high stream powers recordedon Earth were reached, the flows may have been asdeep as 500 m – as gravity is lower on Mars.

The ability of these catastrophic floods to erode andtransport sediment is enormous. Figure 6.4 plots meanwater depth against the mean velocity for these floodevents. Also plotted on this diagram is the point whereflow becomes supercritical and begins to jet. Super-critical flow is highly erosive, and cannot be sustainedin alluvial channels because the bed will be eroded sorapidly. It is also rare in bedrock channels because,


Amazon R.

Yangtze R.

Mississippi R.

Pecos R., Texas

Katherine GorgeTeton Dam

105

104

103

102

101

100

100 101 102 103 104 105 106 107

Ousel Cr., Montana

Str

eam

Pow

er (

W m

-2)

Drainage area (km2)

Fig. 6.3 Plot of stream power per unit boundary area versus drainage basin area. The curve delineates the upper limit in flood power (adapted fromBaker & Costa, 1987).

once floodwaters deepen a small amount, the flow willthen revert to being subcritical. Finally, Figure 6.4plots the region where cavitation occurs. Cavitation isa process whereby water velocities over a rigid surfaceare so high that the vapor pressure of water isexceeded, and bubbles begin to form at the contact. Areturn to lower velocities will cause these bubbles tocollapse with a force in excess of 30 000 times normalatmospheric pressure. Next to impact cratering, cavi-tating flow is the most erosive process known on Earthand, when sustained, is capable of eroding bedrock.Fortunately, such flows are also rare. They can be wit-nessed on dam spillways during large floods as themass of white water that develops against the concretesurface towards the bottom of the spillway. Note thatthe phenomenon differs from the white water causedby air entrainment due to turbulence under normalflow conditions at the base of waterfalls or other dropstructures. Cavitation is of major concern on spillwaysbecause it will shorten the life span of any dam, andmay be a major process in the back-cutting of water-falls into bedrock. Only a few isolated flash floods inchannels have generated cavitation. However, some ofthe Missoula floods reached cavitation levels over longdistances.

Large flood events have enormous potential tomodify the landscape. The stream powers of cata-clysmic floods, as well as many other common flashfloods in the United States, are substantial enough totransport sand-sized material in wash load, suspendedgravel 10–30 cm in size, and boulders several metres indiameter as bedload. This material can also be movedin copious quantities. Even stream powers of

1000 W m-2 are sufficient to move boulders 1.5 m indiameter. Many flash floods can reach this capacity. Ifcavitation levels are reached, bedrock channels can bequickly excavated. The Missoula floods, in the westernUnited States, were able to pluck basaltic columnsof bedrock, each weighing several tonnes, fromstreambeds, and carve canyons through bedrock in amatter of days. More importantly, as will be discussedbelow, present-day flood events are capable ofcompletely eroding, in a matter of days, floodplainsthat may have taken centuries to form.

Synoptic patterns favoring flashflooding(Hirschboeck, 1987)

The severity of catastrophic flooding is ultimatelydetermined by basin physiography, local thunderstormmovement, past soil moisture conditions, and degreeof vegetation clearing. However, most floods originatein anomalous large-scale atmospheric circulations thatcan be grouped into four categories as follows: (i)aseasonal occurrence or anomalous location of a defin-able weather pattern; (ii) a rare concurrence of severalcommonly experienced meteorological processes; (iii)a rare upper atmospheric pattern; or (iv) prolongedpersistence in space and time of a general meteoro-logical pattern. The degree of abnormality of weatherpatterns is very much a function of three types ofpressure and wind patterns that are often part of thegeneral atmospheric circulation. These patterns areshown in Figure 6.5. Meridional air flow representsnorth–south air movement and is usually associatedwith Rossby wave formation linked to the meandering


0.1 1 10 100 1000

Mean depth (m)

Mississippi R.

Amazon R.

Katherine Gorge

Yangze R.

Missoula floods

Pecos R., Texas

CavitationNo cavitation

Supercritical

SubcriticalM

ean

velo

city

(m

s-1

)

2

5

10

20

50

100

1

Fig. 6.4 Channel depth versus current velocity for various floods plotted together with boundaries for cavitation and supercritical flow (adapted fromBaker & Costa, 1987).

path of the polar jet stream, especially over continentsin the northern hemisphere. As stated in Chapter 2,the jet stream path can be perturbed by ElNiño–Southern Oscillation events, and can be associ-ated with persistent weather patterns giving rise todroughts. Exaggerated meridional flow also appears tobe one of the most frequent atmospheric patterns gen-erating flash flooding in the United States. Frequently,it is associated with blocking of a low- or high-pressurecell. The former occurs over warmer water while thelatter is associated with the stalling or agglutinationof mobile polar highs. Blocking in North Americacommonly occurs over the north Atlantic and Pacific

Oceans. In the north Atlantic, blocking increases extra-tropical storm activity and enhances persistence ofpatterns.

Within this general picture of anomalous circulationare smaller features that are responsible for localizedflash floods. These features can be classified into fourcategories: synoptic, frontal, mesohigh, and westernevents. Synoptic events occur when an intense mid-latitude cyclone and a semi-stationary front are linkedto an intense, low-pressure trough overhead. Rainfall iswidespread, persistent and, in local cases, heavy asthunderstorms develop repeatedly in the same generalarea. Such storms occur in the seasonal transitionsbetween winter and summer pressure patterns overNorth America. Thus, they are common from springto early summer, and in autumn. Frontal events aregenerated by a stationary or very slow-moving frontwith zonal air circulation. Upper air stability may exist,and rainfall is usually heavy on the cool side of anywarm front, as warm air is lifted aloft. Such eventsoccur most frequently in July–August in the UnitedStates. Mesohigh floods are caused by instability andconvection following the outbreak of cold air that mayor may not be associated with a front. For instance, astationary front may develop supercell thunderstormsthat force out, in a bubble fashion, a high-pressure cell.These storms occur in the late afternoon or eveningand are a localized summer feature. Western-typeevents refer to a range of regional circulation patternsover the Rocky Mountains. These events tend to peakin late summer and are associated with eitherextremely meridional or zonal circulation.

Maximum probable rainfall(Australian Bureau of Meteorology, 1984)

There are limitations to the concepts of recurrenceinterval and probability of exceedence presented inChapter 3. In many cases, calculations of recurrenceintervals for flash flooding events within the historicalrecord are erroneous because they give no idea of thepossible rainfalls within a given time interval. The useof probability of exceedence graphs as a technique tohypothesize rainfall amounts is not viable for fourreasons. First, the occurrence of a large and rare eventdoes not exclude one of the same magnitude or largerhappening soon after. Second, large and rare events areclustered in time. Third, rare events may or may notappear in rainfall records because the rain gaugenetwork may sample rainfalls at too gross a scale to


High

HL

Low

Trough

Ridge

HeavyrainLow

High

Meridional

Zonal

Blocking

High

Heavy rain

Low

Heavyrain

Fig. 6.5 Abnormal patterns of airflow, within general air circulation,leading to flash flooding (adapted from Hirschboeck, 1987).

detect flash flood events in small catchments. Finally,and more importantly, if the rainfall regime becomeswetter, then rare events of say 1:100 years maysuddenly become common events of 1:20 years or less.For example, in eastern Australia, rainfall between1948 and 1993 increased by 30 per cent over theprevious thirty years. The period witnessed exceptionallocalized and regional rainfalls that could not possiblybe accounted for using standard and accepted proba-bility of exceedence diagrams. Some of these eventswill be described later in this chapter.

An alternative approach uses the concept ofmaximum probable rainfall that can fall in a catchmentwithin a set period, given optimum conditions. Theprocedure was developed in the United States inthe 1940s. Extreme rainfalls are generated by large,virtually stationary thunderstorms; or by mesoscalesynoptic storm systems containing convective cells.The amount of precipitation that can fall over a settime in a given area is a function of the availablemoisture in a column of air, and the efficiency of anystorm system in causing condensation and aggregationof water particles. Because intense cells have a limitedsize, the amount of rainfall falling over a catchmentbecomes a function of the catchment size. Small catch-ments approximating the area of a thunderstorm willreceive more rainfall than large catchments in whicha thunderstorm occupies only part of the area. Therate of condensation, while dependent upon climaticrelationships, is also dependent upon topography.Rough terrain, in which elevation changes by 50 mor more within 400 m – especially in close proximityto the ocean – can trigger thunderstorms or trapconvective cells for several hours. Many locations inthe Wollongong area of Australia and along theBalcones Escarpment in Texas favor maximumprobable rainfall because of rough topography.Figure 6.6 presents the maximum probable rainfallsthat theoretically can fall within fixed periods forsmooth or rough topography. For example, a smallurban catchment of about 10 km2 consisting of smoothtopography can have maximum probable rainfalls ofabout 460 mm within an hour (point A in Figure 6.6).If it rains for three hours, this value can rise to 660 mm(point B). However, if the catchment is rough then themaximum probable rainfall for the three-hour periodcan amount to 840 mm – a 27 per cent increase (PointC). Because tropical air holds more moisture pervolume than air at the poles, the values derived from

Figure 6.6 must be corrected for latitude. Typically thiscorrection is 0.65 at mid-latitude locations such asSydney, Australia or Los Angeles, California. Thislatitudinal correction reduces the values mentionedabove to approximately 300 mm, 430 mm, and550 mm, respectively, at these two latter locations.

Figure 6.6 illustrates several factors about flashflooding. First, beware of small catchments. Thatbabbling brook meandering intermittently throughone’s backyard can turn into a raging torrent underinnocuous conditions. In Wollongong, Australia, flashflooding with the attendant movement of boulders hasoccurred in small creeks within 100 m of a similar-sized


1 10 100 1000Drainage Basin Area (km2)

1200

1000

800

600

400

200

6 hr (R)

5 hr (R)

4 hr (R)

3 hr (R)

6 hr (S)5 hr (S)

2 hr (S)

1 hr (R & S)

0.75 hr (R & S)

0.5 hr (R & S)

0.25 hr (R & S)

2 hr (R)

1.5hr (R)

1.5 hr (S)R

ainf

all (

mm

)

R = rough topographyS = smooth topography

4 hr (S)3 hr (S)

X

X

X

A

B

C

Fig. 6.6 Maximum probable rainfalls per unit time by drainage basinarea (Australian Bureau of Meteorology, 1984).

stream in which no rain has fallen. Second, after aboutone hour of continuously heavy rainfall, rough catch-ments receive 10–25 per cent more rainfall thansmooth basins. That babbling brook running through aproperty with hilly topography is incredibly susceptibleto flash floods. Finally, the volumes of rain that can fallwithin small catchments can be awesome. Figure 6.1 isa conservative depiction of the amounts of catastrophicrain that have been measured by our inadequatenetwork of rain gauges. Within pristine catchmentstypical of the hilly topography of Sydney or LosAngeles, over 500 mm of rain can fall within the spaceof one hour. Thankfully rainfalls generating flash floodsrarely continue for longer periods. When they do, theeffects are catastrophic.

Flash flood events (Bolt et al., 1975; Cornell, 1976; Maddox et al., 1980)

Flash flooding in the United States tends to be associ-ated with summer storms and landfall of tropical stormsystems. The Rapid City flood in the Black Hills ofSouth Dakota, 9–10 June 1972, exemplifies the formeraspect. One metre of rain fell in the space of six hours,an amount that had a recurrence interval of 1:2000years. Flash floods swept down all major streamsincluding Rapid Creek, where an old dam – built fortyyears earlier as a Depression relief project – collapsed,sending a wall of water through the downtown sectionof Rapid City. Over 230 people lost their lives and 2900people were injured. Property damage exceeded$US90 million and included over 750 homes and 2000cars. Texas is affected by the intrusion of tropicalcyclones and easterly waves embedded in trade winds.Strong orographic uplift of these westerly movingsystems can occur along the Balcones Escarpment. Asa result, Texas has the reputation for being the mostflood-prone region in the United States, and hasrecorded some of the highest rainfall intensities in theworld (Figure 6.1). For example, during the PecosRiver flood of June 1954, Hurricane Alice moved upthe Rio Grande Valley and stalled over the BalconesEscarpment. The hurricane interacted with an upperair disturbance and turned into an intense, extra-tropical, cold-core cyclone. Over 1 m of rain wasdumped in the lower drainage basin, resulting in flood-waters 20 m deep, with a recurrence interval of 1:2000years. To date, this is the largest recorded flood eventin Texas.

Urban flash floods(Nanson & Hean, 1984; Australian Bureau of Meteorology,1985; Riley et al., 1986a, b)

Urban areas exacerbate flash flooding. The breaking ofthe long drought in eastern Australia following the1982–1983 ENSO event heralded the onset ofextraordinary flash flooding in the Sydney–Wollongongregion, as Walker circulation ‘turned on’ again over thefollowing two years. The flooding was very localizedand can be separated into two components: the Daptoflood of 17–18 February 1984 and the Sydney thun-derstorms of 5–9 November 1984. In each case, slowmoving or stationary convective cells developed inassociation with east-coast lows. Neither the localizedconvection nor the east-coast lows was unusual.Initially, the events represented the rare concurrenceof two commonly experienced meteorologicalprocesses. Unfortunately, the pattern has occurredwith alarming frequency throughout the remainder ofthe twentieth century.

The Dapto flood event began in the late evening of17 February 1984 as a cold front pushed through thearea in front of a high pressure cell centered south-eastof Tasmania. As a result, moist onshore air flowed intothe area at the same time as a small east-coast lowdeveloped south of Newcastle. This low tracked downthe coast and then stabilized over the 500 m highIllawarra escarpment, north of the town of Dapto, at7 am on 18 February. This was followed by develop-ment, during the remainder of the morning, of acomplex, upper-level trough over the Wollongongregion. These lows produced extreme instability andcaused marked convergence of moist north-eastand south-east airflow into the Wollongong area.Orographic uplift along the high escarpment dumpedcopious amounts of rain over the Lake Illawarradrainage basin over 24 hours (Figure 6.7). In a one-hour period on 18 February, 123 mm fell west ofDapto. Over a nine-hour period, 640 mm and 840 mmwere recorded, respectively, at the base and crest ofthe escarpment. The heavier rainfall had a 48-hourrecurrence interval of 1:250 years and is the greatestnine-hour rainfall recorded in Australia.

The resulting flooding of Lake Illawarra, however,had a 1:10-year frequency of occurrence. This casesupports the United States evidence that flash floodingwith severe consequences can occur in parts of adrainage basin with areas less than 50 km2. Figure 6.7


shows the highly localized nature of the event. Within5 km of the escarpment, rainfall amounts had droppedto less than 400 mm; within 10 km, amounts were lessthan 200 mm. Because any area on this map can besubject to this type of flooding – albeit rarely – it hasbeen estimated that the Wollongong urban area couldexperience a Dapto-type event every 25 years. In the20 years since 1984, five flash flood events of similarmagnitude have occurred. Emergency organizationshere should be prepared to handle high-magnitude,localized but rare floods in this region, with a muchshorter expectancy rate than shown by individualevents.

The Sydney metropolitan flooding that followed wasnot a single event, but rather a series of intensethunderstorms, which struck various parts of the metro-politan area on 5–9 November 1984. Accompanyingthe synoptic weather situation was a series of tornadoesand waterspouts that caused over $A1 million damage.The floods themselves cost the insurance industry

$A40 million. One death resulted from lightning. Thisdeath toll is small considering the fact that mostobservers would class the lightning activity on the nightof 8 November as the worst in living memory. Overthe five-day period, rainfall totalled 550 mm south-south-east of the Sydney central business district.Most of the eastern and northern suburbs, as well asthe Royal National Park south of Sydney, receivedtotals in excess of 300 mm.

The floods originated from a low-pressure cellcentered south-east of Brisbane on 5 November. Thislow moved slowly down the coast, rotating counter-clockwise around a strong, blocking high-pressure cell(maximum pressure 1032 hPa) that drifted fromAdelaide to New Zealand over the period (Figure 6.8).The high-pressure cell directed consistent, moist,zonal flow onto the coastline. Concomitant with themovement of the high, a low-pressure trough devel-oped parallel to the coastline. This trough spawned low-pressure cells that forced airflow over Sydney,


Wollongong

Mullet Cr.

Lake Illawarra

Dapto

Tasman Sea

0 2 4 6 8 10 km

200

30040

0500

600

700

800

50040030

0

200

Reservoirs

Illaw

arra

Escarp

men

t

N

Fig. 6.7 Forty-eight-hour isohyets for the Dapto flood, Wollongong, Australia, 18 February 1984 (based on Nanson & Hean, 1984).

resulting in intense atmospheric instability. Theflooding occurred as three separate events during thisperiod.

The first event on Monday 5 November began asthunderstorm activity at Cronulla (Figure 6.9a) andslowly moved northward over the city center. Thisthunderstorm activity originated as a series of intensecells associated with a low-pressure trough parallel tothe coast and a separate low-pressure cell 50–100 kmsouth of Sydney. A second trough existed in the interiorof the continent (west of the Dividing Range), anddrifted over Sydney in the early evening. Both of thesetroughs were responsible for separate rainfall events.Over 220 mm of rainfall fell within 24 hours overRandwick. Rainfall intensities for time spans undernine hours exceeded the 1:100-year event. The mostserious flooding was caused by a storm drain atRandwick racecourse that was unable to cope with therunoff. Streets in adjacent Kensington bore the brunt ofsubstantial flooding, which entered houses to depths of0.5–1.0 m. Evening rush hour was thrown into chaos.

In the second event, on 8 November, a low-pressuretrough was again present, this time directly overSydney. Airflow converged into this trough over thenorthern suburbs. Seventeen convective cells wereresponsible for rainfall during the morning with the firsttwo producing the heaviest falls. Over 230 mm of rainfell at Turramurra, in the northern suburbs, with fallsof 125 mm between 7:15 and 8:15 am (Figure 6.9b).This latter intensity exceeded the 1:100-year event. The

20-minute rainfall intensity record at Ryde had arecurrence interval of between 50 and 500 years.Flooding during the second event began within20 minutes of the commencement of heavy rainfall.Peak discharges occurred within five minutes offlooding. Many of the catchments that flooded were nomore than 1 km long; however, flood depths up to 1.5 mbecame common, both in channels and in shallowdepressions forming part of the drainage network. Manyresidents were unaware that these depressions were partof the flood stream network. Flooding occurred in themiddle reaches of catchments as stormwater drainsreached capacity and surcharged through access-holecovers. This surcharging capacity is deliberately builtinto Sydney’s stormwater system, as it is prohibitivelyexpensive to build an underground drainage networklarge enough to cope with rare events. Some of theflooding was due to storm inlet blockage by trees, loosedebris or, in some cases, cars transported in floodwaters.It was even found that backyard paling fences could sig-nificantly block, divert, or concentrate flows, increasingflood levels locally by as much as 40 cm.

The third event occurred on the night of8 November. Synoptically, it was very similar to thesecond event, with a trough cutting through Sydneyand converging air into the city region. Rainfalls of249 mm were measured at Sydney’s Botanic Gardens,and values near 300 mm in and around the RoyalNational Park to the south (Figure 6.9c). Again, severalcells were involved, and again peak intensities for time


Nov 1984

L

L

L

1008

1024

H

Sydney

H

5

8

6

9

7

L

H

1024

1008

L

L

H

L

L

H L

1008

L

1028H

L 1004

1028

H

H

L

1004

1024

HH

Fig. 6.8 Synoptic pattern for Sydney, Australia, flash floods, 5–9 November 1984.

spans under nine hours exceeded the 1:100-year event.Many of the suburbs affected by this event had experi-enced their previous 1:100 year rainfall only four dayspreviously. During this event, the road network sup-planted natural drainage and took floodwaters awayfrom storm drains normally expected to carry runoff.Because roadways have reduced frictional coefficients,flood peaks were almost instantaneous throughout thedrainage system. Most of the damage during the thirdflash flood was caused by floodwaters reachingshopping centers at the base of steep slopes, sendingwalls of water and debris – in some instances includingexpensive cars – cascading through shops. In addition,the Tank Stream, which runs beneath the lower centralbusiness district of Sydney, was reactivated, floodingthe Sydney Stock Exchange and causing several milliondollars worth of damage. The State Library was alsoflooded, damaging rare exhibits.

There are three points about flash flooding in urbanareas to be drawn from the events described above.Firstly, the probability of occurrence of a high-magnitude, localized rainfall event in a region can bean order of magnitude greater than the probability ofthe event itself. For example, the Dapto Flood was a1:250-year event. The fact that any part of the Illawarracan experience such a localized storm indicates thatthe probability of a similar flood occurrence in theWollongong region – along a 40 km stretch of coast – is

about 1:25 years. Sydney experienced at least five1:100-year events in the space of five days. In somecases, emergency services had to respond to all fiveevents. Since then, Sydney has received the heaviest48-hour rainfall on record. On 5–6 August 1986, aneast-coast low dropped over 430 mm of rain in threedays, causing rivers to flood from Bathurst, in the BlueMountains, to the Georges River, in south Sydney. Atleast six people died, from drowning or being electro-cuted by fallen power lines. Damage exceeded $A100million, including over 3000 motor vehicles swampedby floodwaters. In this case, rainfall with a recurrencefrequency of 1:100 to 1:200 years was experienced overa large area, much of which had been affected bysporadic 1:100-year events two years previously.

Secondly, urban flooding in most cases peakedwithin half-an-hour of the onset of intense rain. Muchof the flash flooding was exacerbated by the structureof the urban drainage system. In the less densely builtnorth-western suburbs of Sydney, surcharging of thestorm drains along existing natural drainage routescaused flooding to homes and shops built in theseareas. In the more densely built up eastern suburbs,the road network often took the place of the naturaldrainage system, causing damage to areas that wouldnot normally receive this type of flooding.

Finally, it is very difficult to comprehend what theincreased frequency in flooding means. While some


5-6 Nov. 1984 7-8 Nov. 1984 8-9 Nov. 1984

MacquarieUniversity Turramurra

Ryde

SydneyRandwick

Cronulla

Royal NationalPark A B C

N

< 50 mm 50–99 mm 100–149 mm 150–199mm > 200 mm

Fig. 6.9 Isohyets for three flash flood events in Sydney, Australia, 5–9 November 1984 (modified from Australian Bureau of Meteorology, 1985).

might see it as a signal of Greenhouse warming, itappears to represent a shift towards extreme rainfallevents in this region. Not only did the climatic patternsgiving rise to one catastrophic flood reappear days laterin the same location, but the patterns also tended torepeat themselves at other locations along the coast,and recur several times over the next few years. Since1984, the 80 km stretch of coast encompassingWollongong and Sydney area has experienced at leasttwelve similar events. A local coroner’s inquiry into onerecent event has shown that local authorities are almostunaware of this change in the rainfall regime. In NewSouth Wales, the state government has realized thaturban flooding has become more prevalent, and hasadded a $A25 surcharge to ratepayers’ assessments tocover clean-up expenses and recover relief costs. Eventhis is a misconception of the nature of the climaticshift, because the surcharge is being applied to thewestern suburbs of Sydney, which in fact have notexperienced the greatest number of flash floods. Thegreatest rainfall has occurred during thunderstormsover the central business district, which has a highersurface roughness coefficient because of tall buildingsand parallel streets.

These events represent a significant increase in theoccurrence of a natural hazard that has economic andpersonal ramifications for most of the four millionresidents in the region. Neither the significance ofpresent flooding, nor its consequences, has been fullyappreciated.

HIGH-MAGNITUDE,REGIONAL FLOODS

High-magnitude, regional floods usually present adisaster of national or international importance.Usually, large-scale drainage basins or many smallerrivers in the same region are flooded. Flooding appearsto have increased in intensity and extent in recent years.As higher precipitation is one of the consequences ofglobal warming, flooding has recently received theattention of media and relief organizations. Figure 6.10plots the locations of major floods for fifteen yearsbetween 1985 and 2003. Flooding has occupied anextraordinary proportion of the world’s land surfaceconcentrated in two main regions: the tropics and themid-latitude storm belts. This section describes histori-cal flooding for three areas – the Mississippi River,eastern Australia, and China – to illustrate flooding’spervasiveness and high economic cost.

Mississippi River floods

(Bolt et al., 1975; Lott, 1993; Larson, 1996; Trotter et al.,1998; Public Broadcasting Service, 2000)

The Mississippi River drainage basin is the thirdlargest in the world, and the largest in North America.The basin covers 3 224 000 km2 or 41 per cent of theconterminous United States (Figure 6.11). Its head-waters can be divided into two areas, one originating atthe continental divide in the Rocky Mountains, and the


Fig. 6.10 Flood locations 1985–2003, excluding 1989, 1992, 1996, and 1997 (based on Dartmouth Flood Observatory, 2003).

other in the Great Lakes lowlands. The latter networkhas been developed only since the last glaciation, andincludes an area that can accumulate large, wintersnowfalls. Floods have been a constant feature of theMississippi River since historical records began(Figure 6.12). Flooding can be divided into threeperiods: before 1927, between 1927 and 1972, andafter 1972. Before 1927, flood control was considereda local responsibility. The flood of 1927 brought therealization that the Mississippi, which crosses stateboundaries, could only be controlled at the nationallevel. There were 246 fatalities, 137 000 buildingsflooded, and 700 000 people made homeless as a resultof this flood. Subsequently, the Flood Control Act of

1928 was passed and levee and reservoir maintenanceand management were put under the auspices of stateand federal authorities. Over 3000 km of levees andfloodwalls, averaging 9 m in height in the lower part ofthe basin, were built to control flooding. Four floodwayswere also built to divert excess floodwaters into largestorage areas, or into the Gulf of Mexico. Some parts ofthe channel susceptible to erosion were also stabilizedwith concrete mats, which cover the whole channelboundary. Finally, reservoirs were constructed onmany of the tributaries to delay passage of floodwatersinto the Mississippi. The period between 1927 and 1972represents one where floods on the Mississippi wereless severe and the mitigation works appeared to besuccessful.

Since 1972, the Mississippi Basin has been subjectedto two periods of flooding that have tested these engi-neering works to the limit. The 1973 flood was thegreatest to hit the basin to that point in time – 200 yearssince records began. Total rainfall for the six-monthperiod exceeded the norm by 45 per cent in all partsof the Basin. On 28 April, the river peaked at a 200-year historical high at St Louis. Damage exceeded$US750 million, and 69 000 people were madehomeless for periods up to three months. Withoutthe flood mitigation works built since 1927, damagewould have been as high as $US10 000 million, and12.5 million hectares of land would have been flooded.


-120-120∞

-120-120∞

-100∞

-100∞

-80∞

-80∞

25∞ 25∞

30∞ 30∞

35∞ 35∞

40∞ 40∞

45∞ 45∞

50∞ 50∞

km

Ohi o

Te

nnessee

Gulf of Mexico

Misso

uri

Red

St. Louis

New Orleans

0 1000 km

1927

1993

Fig. 6.11 The Mississippi River drainage basin and the extent of the 1927 and 1993 floods (based on Public Broadcasting Service, 2000).

5

10

15

20

25

30

35

40

1700 1750 1800 1850 1900 1950 2000Year

Inte

rval

bet

wee

n flo

ods

(yr)

Fig. 6.12 Occurrence of major floods on the Mississippi River since1700 versus interval between floods (based on Trotter et al., 1998).

Because of localized decreases in river depth andhigher flood heights, over 1300 km of main levees hadto be raised after the 1973 flood.

As of 1993, the total damage caused by Mississippiflooding was equivalent to $US4.4 billion. The GreatFlood of 1993 was worse (Figure 6.11). Mobile polarhighs were stronger than normal in the spring of 1993,dragging cold air south into contact with warm air fromthe Gulf of Mexico. By April, the biggest floods onrecord were sequentially swamping the Missouri–Missi-ssippi floodplain from north to south. The pressurepatterns continued into summer with rainfalls exceeding200–350 per cent of the norm. Flood levels reached75–300-year recurrence intervals along major portionsof the upper Mississippi and lower Missouri Rivers. On1 August, the Mississippi crested at 15.1 m at St. Louis,2.0 m above the previous record. Transport across theMississippi for a distance of 1000 km was interrupted,1000 levees collapsed, 55 people lost their lives, 48 000homes in 75 towns were flooded, and 74 000 peoplewere left homeless. The total damage bill across 12states was $US18 billion. Fortunately, modellingforecast the disaster four months in advance, givingplenty of time for preparations and evacuation.

Great Australian floods(McKay, 1979; Shields, 1979; Holthouse, 1986; Yeo,2002; Australian Bureau of Meteorology 2003a, b)

The major rivers of eastern Australia originate withinthe Great Dividing Range, which parallels the eastcoast (Figure 6.13). Two systems flow westward.The Murray–Darling system and its tributaries – theWarrego, Condamine, Macintyre, Macquarie, Lachlanand Murrumbidgee – forms the largest river systemin Australia, entering the southern ocean south ofAdelaide. The other system consists of the Diaman-tina–Coopers Creek rivers, which flow into landlockedLake Eyre. On the eastern side of the Dividing Rangeare a number of much shorter, but just as impressive,systems. These rivers consist of the Hunter, Macleay,Clarence, Brisbane, Burnett, Fitzroy, and Burdekin.Successive tropical cyclones or monsoonal troughshave in the past severely flooded all of these rivers, butnever at the same time. Aboriginal legends from theDreamtime imply that many river valleys have beenflooded over widths of 50 km or more. Three floodevents: the wet of 1973–1974, the Charleville–Nynganfloods of 1990 and the Katherine floods of 1998 – all


Katherine

Daly Gulf of Carpentaria Coral Sea

Queensland

NSW

VictoriaSouthern Ocean

Leichhardt

Mitchell

Gilbert

Flinders

Burdekin

Fitzroy

Brisbane R.

Burnett

Condamine

Clarence

Macintyre

Macleay

Hunter

Murrumbidgee

Coopers Cr.

Warrego

Darling

Brisbane

Lake Eyre

Wentworth

Murray

Lachlan

Macquarie

Diamantina

Alice Springs

Finke

1000 km0

Gippsland

Northern Territory

South Australia

Sydney

Charleville

Nyngan

Tasman Sea

Fig. 6.13 Major rivers of eastern Australia.

related to ENSO – illustrate the broad catastrophicnature of flooding in eastern Australia that substantiatethose Aboriginal legends.

Flooding in the summer of 1973–1974 coincidedwith one of Australia’s wettest La Niña events inthe twentieth century. Rainfall was torrential andcontinuous throughout most of January 1974, as theintertropical convergence settled over tropicalAustralia. On 25 January, Cyclone Wanda moved intothe interior of Queensland and New South Wales,dumping in excess of 300 mm of rain in 24 hours overa wide area, and triggering massive flooding of all riversystems. Rain in the catchment of the Brisbane Riverproduced one of the worst urban floods in Australianhistory. These January floods represented the largestnatural disaster to occur in Australia to that point intime. Flooding covered an area of 3 800 000 km2,larger than the drainage basin of the Mississippi River.From Alice Springs to the Pacific Ocean, and fromthe Gulf of Carpentaria to the Murray River, militaryairlifts had to be arranged to supply isolated towns, cutoff by floodwaters, with emergency food and stockfodder. Around the Gulf of Carpentaria, the tributariesof the Flinders River amalgamated to form a river150 km wide. In northern New South Wales, torrentialrainfall continued week after week, raising river levelsin excess of 20 m. Floodwaters slowly flowed downthe Diamantina and Coopers Creek drainage into theinterior, filling Lake Eyre for only the fourth time thiscentury. At Wentworth in New South Wales, farmlandremained flooded for two years as successive floodscame down the Darling and Murray in 1974–1975.Stock losses to sheep alone in New South Walestotalled 500 000. The scale and magnitude of floodingwas unprecedented.

The Brisbane flood hit an expanding city that had notwitnessed a major flood since 1893. Few suspected theflooding potential of the Brisbane River. Synoptically,the flooding of the city was due to the persistent recur-rence of cyclones tracking over eastern Queensland inthe space of four weeks, culminating in Cyclone Wanda.However, many of the flood’s effects were exacerbatedby human factors. Somerset Dam, built after the 1893flood to contain similar events, was totally inadequatefor this purpose; yet, many people believed that thedam protected them from flooding. The sprawlingnature of growth in the Brisbane River Basin had alsoled to large-scale clearing of forest in rugged terrain,which enhanced runoff during the high-intensity

rainfall periods of the 1974 flood. The character of thedrainage basin was also altered. Urban developmenthad seen some creeks filled in, while others had beenlined with concrete. These latter channels weredesigned to contain only the 1:10-year flood. Roads,driveways, and houses had sealed large sections of thelandscape. These modifications resulted in calculatedflood discharges being twice that of the 1:100-year floodfor adjacent vegetated catchments. Some urbancatchments experienced the 1:1000-year flood event.Additionally, there was very little delay between time ofpeak rainfall intensity and peak flooding. Flash floodingoccurred in urban catchments with flood peaks crestingwithin one hour of peak rainfall intensities. Many areasalso had inadequate rain gauging stations, so thatthe exact amount of rain falling in some parts of thecatchment was unknown, and the forecasting of floodlevels impossible to determine.

The Brisbane flood also saw the collapse oforganized disaster warnings. There was no central dataprocessing authority, so that local flooding in key areaswent unnoticed. Over 70 per cent of residents ques-tioned afterwards about the flood had received noofficial warning. The media, in their efforts to report amajor story from the field, sparked rumors and cloudedthe true picture of flooding. The rapid series of flashfloods, in isolated parts of the catchment 24 hours priorto the main flood, led to public confusion and commit-ted the Disaster Relief Organisation to what afterwardswere evaluated as only minor disasters. The majorflood peak was disastrous. All bridges across theBrisbane River were damaged or destroyed and35 people drowned. At its height, the river broke itsbank and ran through the central business district ofBrisbane. In Ipswich, 1200 homes were destroyed.Overall, 20 000 people were made homeless. Only11 per cent of residents who experienced the mainflood received evacuation assistance from emergencyorganizations, and only 30 per cent acknowledgedthe clean-up help from any relief organization. Mostrelief came from friends and community groups.Over 40 per cent of victims received help fromchurch contacts, and over 30 per cent of volunteerssurveyed stated that they belonged to no organizedgroup. Generally, most people applied themselves toevacuation, alternative accommodation, clean-up andrehabilitation with little reliance on government orsocial organizations for help. This raised seriousquestions about the efficiency of such organizations in


disaster relief in Australia, and led to significantchanges in the response to disasters by both privaterelief organizations and federal government agencies.Eleven months later these revisions were tested to theutmost when Cyclone Tracy destroyed the city ofDarwin.

The floods of 1990 hit a region from centralQueensland in the north to Gippsland in the south. On18 April 1990, a strong upper-level low developed overthe interior as a high-pressure cell stalled over the eastof the continent. Up to 350 mm fell initially overcentral Queensland, triggering major floods of everycreek, stream, and river. On 21 April, floodwatersconverged suddenly on the Warrego River and sweptthrough Charleville. Residents had to be rescued fromrooftops by helicopter. Over the next few days boththe rains and the floods slowly tracked southward,inundating one riverside town after the other. AtNyngan on the Bogan River, residents decided to ‘fightit out’ and began sandbagging the levees around thetown. Television crews moved in to report nightly ontheir heroic but futile efforts. Over 200 000 sandbagswere laid in four days.

In Australia, a national emergency can be declaredonly if someone reports it to the state government whothen requests federal assistance. In anticipation ofthat call, Emergency Management Australia called in500 troops who stationed themselves in the nearesttown waiting to help. In the confusion and excitementof the flood, no one called for help. Finally, as the floodbroke through the levee, television crews filmedpeople flinging themselves into the gaps in a last-ditcheffort to stay the inevitable. Within an hour the townwas flooded and helicopters evacuated the 2500residents to Dubbo where the army had been waiting.The disaster continued southward as rivers sequen-tially broke previous flood heights by a metre or more.The low, which was blocked eastward by the GreatDividing Range, finally exited the continent throughGippsland, Victoria. Here, up to 350 mm of rain fellwithin 48 hours, causing major floods on all riversflowing from the Snowy Mountains. In the space offour days, this single low-pressure cell had flooded alandmass the size of western Europe in one of theworst mass floods in Australian history.

The floods of 1998 illustrate the extent of floodingthat can occur in Australia during a La Niña event.Five flood events occurred from Katherine in the northto the Gippsland in the south. The interior of the

Australian continent receives most of its rainfall fromtropical cyclones that can move inland long distanceswithout the impediment of mountains. In January1998, tropical Cyclone Les, which had developedover the Gulf of Carpentaria, moved westward andweakened into a rain depression. However, it stalledover Katherine in the Northern Territory and between25 and 27 January, it dropped 400–500 mm of rain onthe Katherine, Roper, and Daly River catchments.Floodwaters on the Katherine River rose to a recordlevel 20 metres above normal. This was sufficient tosend two metres of muddy water through the town ofKatherine. The town’s 2000 residents left evacuationtoo late, fearing looters or believing that the floodthreat was highly exaggerated. Over 1100 people hadto be treated for injuries sustained while coping withthe disaster.

The shift to a La Niña event is often marked by anincrease in the frequency and intensity of east-coastlows. One such low caused heavy rains that floodedGippsland on 23–24 June, while another causedflooding at Bathurst, west of Sydney, on 7–8 August.Finally, on 17 August, a small low developed along the500 m escarpment backing Wollongong on the NewSouth Wales coast. In the early evening, extremerainfalls caused flash flooding that destroyed 700 carsand caused over $A125 million damage. In all areas,the flooding was not unique. Historic and palaeofloodevidence indicates that floods with up to three timesthe volume of the 1998 floods have occurred in allthree regions in the last 7000 years. In Wollongong, the1998 floods were simply part of a spectrum consistingof 80 or more floods that have occurred since the early1800s. Much of the perception of increased flooding inAustralia reflects media hype, increased urban growthand a stubborn resistance to learn from the lessons ofprevious floods.

Flooding in China(Bolt et al., 1975; Czaya, 1981; Milne, 1986)

Flooding in China is mainly related to the vagaries ofthe Hwang Ho (Yellow) River channel. The Hwang HoRiver – China’s River of Sorrow – drains 1 250 000 km2

and flows over a distance of 4200 km (Figure 6.14).Throughout much of its upper course, it erodes yellowloess, and for 800 km of its lower length, the HwangHo flows without any tributaries on a raised bed of silt.Much of this material has entered the river because ofpoor land-use practices in the upper drainage basin in


the last 2000 years. Today, the suspended load reaches40 per cent by weight, making the Hwang Ho one ofthe muddiest rivers in the world. Where the riverbreaks onto the flatter coastal plain, vertical aggrada-tion has built up a large alluvial fan-like delta fromBeijing in the north, to Shanghai in the south. Theriver thus flows on the crest of a cone-shaped delta andbuilds up its bed very rapidly during flood events. Asearly as 2356 BC, the river channel was dredged toprevent silting. In 602 BC, the first series of levees wasbuilt to contain its channel. Despite these efforts, inthe last 2500 years the Hwang Ho has broken its banksten times, often switching its exit to the Yellow Seaover a distance of 1100 km on each side of theShandong Peninsula (Figure 6.14). Associated withthese breakouts has been massive flooding – with theresult that China historically has suffered the greatestloss of life in the world from flood hazards.

Between 2300 BC and 602 BC the Hwang Hoflowed through the extreme northern part of its deltain the vicinity of Beijing. In 602 BC, the river movedslightly south to the region of Tianjin, and stayedthere until 361 BC, whereupon it underwent acatastrophic shift south of the Shandong Peninsula.For the next 150 years, the Chinese expended con-siderable effort shifting it north again. In 132 BC, theHwang Ho switched to a course near the presententrance, where dike building tended to keep itstabilized, with minor channel switching, on the northernside of the Shandong Peninsula until 1289 AD. Forthis period, the northern part of the alluvial plainaggraded – a process that steepened the topographicgradient southward. With the fall of the Sung dynasty,dike maintenance lapsed, and in 1289 AD the HwangHo again switched south of the peninsula, whereit was allowed to remain for the next 660 years.


1048–1194

11–70

Kaifeng

Beijing Tianjin

Shandong Peninsula

2300–602 BC

602–361 BC

Hwang Ho R.

1289–1494

1194–1289

1494–1853

1887–1889

Shanghai

East China Sea

Gulf of Chihli

132 BC–1170–1048

1853

Yangtze R.

361–132 BC

N

0 100 200 300 km

Land above 200 m

Fig. 6.14 Location of the Hwang Ho River channel: 2300 BC to the present (after Czaya, 1983). Dates shown are AD unless otherwise indicated.

The most destructive flooding occurred during thisperiod, in 1332 AD, with seven million peopledrowning and over ten million people dying fromsubsequent famine and disease. By 1851, the HwangHo had aggraded its bed in the south to such an extentthat it threatened to engulf the town of Kaifeng.Measured rates of aggradation exceeded 2 m yr-1, anddykes more than 7 m high were built to contain itsflow. Between 1851 and 1853, the Hwang Ho brokeout northward and took on its present course. In1887, a temporary breach saw the river move thefurthest south it has ever been. It temporarily linkedup with the Chang (Yangtze) River and, for a brieftime, both rivers flowed out to sea through the samechannel. The 1887 flood breached 22 m high embank-ments and flooded 22 000 km2 of land to depths of8 m. Over a million people lost their lives. Not onlywas the death toll severe, but the river also buriedflooded land in meters of silt. Farmers relying uponthe river in the northern part of the delta suddenlyfound themselves without irrigation water and facingstarvation. Twentieth century floods have been com-paratively less severe; however 200 000 and 300 000people were killed in the floods of 1911 and 1931,respectively.

The natural tendency of the Hwang Ho to shiftcourse has been greatly exacerbated during wartime.In 1938, General Chiang Kai-shek, in an attempt toprevent advancing Japanese armies taking the city ofChengchow, ordered the levee dikes to be dyna-mited. The diversion failed to stop the Japaneseadvance, but the ensuing floods killed a millionunsuspecting Chinese, and possibly caused the deathof 11 million people through ensuing famines anddisease. It was not until 1947 that extensive engineer-ing works forced the Hwang Ho permanently backinto its prewar channel. Today, the Hwang Ho is sta-bilized for much of its length, and is kept to the northof the Shandong Peninsula. Because the river trans-ports such large quantities of silt, its bed is continuingto aggrade and now sits 20 meters above its flood-plain, the latter protected by an inner and outer set ofdikes spaced 10 km apart. The Chinese governmenthas embarked on a program of dam and silt basinconstruction to minimize the flooding effects of theHwang Ho; however, the river is still not controlled,and will probably breach its dikes in the future duringa large flood.

CONCLUDING COMMENTS(Tropeano et al., 2000; Kunkel, 2003; Sheffer et al., 2003)

The wide scale nature of recent flood events is notunique in time. Rather it represents the continuationof a longstanding global hazard that has takenmillions of lives. For example, in Canada, there hasbeen no change in the frequency and intensity ofextreme precipitation events during the twentiethcentury. Similarly, in the United States (where therehas been an increase of 20–40 per cent in extremeprecipitation over the long-term average of the lasttwo decades of the twentieth century), flooding todayis just as prominent as it was at the end of the nine-teenth century. These changes are due to warmer seasurface temperatures in the tropical Pacific Oceancausing subsequent flooding in the United States.Likewise, what might appear to be the largest floodson record can be shown to be part of a continuum.For example, on 8–9 September 2002, the GardonRiver in southern France reached a flood crest 14 mabove normal following 680 mm of rain over20 hours. However, this flood did not reach a cavelying 3 m higher that contained evidence of fiveseparate flood events dating between 1400 and 1800.Similarly, on 14–15 October 2000, major floodingaccompanied by landslides affected the Italian Alps.The floods were triggered by 400–600 mm of rainthat fell within two days. However, many of the land-slides represented the partial reactivation of muchlarger historical slides. In Australia, at least elevencomparable floods of similar magnitude haveoccurred at 20–40 year intervals since 1810. There isthus substantial evidence that flooding is ubiquitous.Recent floods across the globe have not beenabnormal, but reflect the occurrence of events aspart of an ongoing continuum.


Australian Bureau of Meteorology, Australia 1984. The Estimationof Probable Maximum Precipitation in Australia for ShortDurations and Small Areas. Australian Bureau of MeteorologyBulletin No. 51.

Australian Bureau of Meteorology, Australia 1985. A Report on theFlash Floods in the Sydney Metropolitan Area Over the Period


5 to 9 November 1984. Special Report, Australian Bureau ofMeteorology, Melbourne.

Australian Bureau of Meteorology, Australia 2003a. Nyngan andCharleville, April 1990. <http://www.bom.gov.au/lam/climate/levelthree/c20thc/flood8.htm>

Australian Bureau of Meteorology, Australia 2003b. Katherinefloods, January 1998. <http://www.bom.gov.au/lam/climate/levelthree/c20thc/flood6.htm>

Baker, V.R. and Costa, J.E. 1987. Flood power. In Mayer, L. andNash, D. (eds) Catastrophic Flooding. Allen and Unwin,London, pp. 1–21.

Bolt, B.A., Horn, W.L., MacDonald, G.A., and Scott, R.F. 1975.Geological Hazards. Springer-Verlag, Berlin.


Czaya, E. 1983. Rivers of the World. Cambridge University Press,Cambridge.

Dartmouth Flood Observatory 2003. Global archive map of extremeflood events since 1985. <http://www.dartmouth.edu/~floods/Archives/GlobalArchiveMap.html>

Griffiths, J.F. 1976. Climate and the Environment: The AtmosphericImpact on Man. Paul Elek, London.

Hirschboeck, K.K. 1987. Catastrophic flooding and atmosphericcirculation anomalies. In Mayer, L. and Nash, D. (eds)Catastrophic Flooding. Allen and Unwin, London, pp. 23–56.

Holthouse, H. 1986. Cyclone: A Century of Cyclonic Destruction.Angus and Robertson, Sydney.

Kunkel, K. 2003. North American trends in extreme precipitation.Natural Hazards 29: 291–305.

Larson, L.W. 1996. The great USA flood of 1993. IAHS Conferenceon Destructive Water: Water-Caused Natural Disasters – TheirAbatement and Control, Anaheim, California, June 24–28,<http://www.nwrfc.noaa.gov/floods/papers/oh_2/great.htm>

Lott, N. 1993. The summer of 1993: Flooding in the Midwest anddrought in the southeast. National Climatic Data Center TechnicalReport 93-04.

Maddox, R.A., Canova, F., and Hoxit, L.R. 1980. Meteorologicalcharacteristics of flash flood events over the western UnitedStates. Monthly Weather Review 108: 1866–1877.

McKay, G.R. 1979. Brisbane floods: the paradox of urban drainage.In Heathcote, R.L. and Thom, B.G. (eds) Natural Hazards inAustralia. Australian Academy of Science, Canberra, pp. 460–470.

Milne, A. 1986. Floodshock: The Drowning of Planet Earth. Sutton,Gloucester.

Nanson, G.C. and Hean D.S. 1984. The West Dapto Flood ofFebruary 1984: rainfall characteristics and channel changes.Department of Geography, University of Wollongong, OccasionalPaper No. 3.

Public Broadcasting Service 2000. Fatal flood. <http://www.pbs.org/wgbh/amex/flood/maps/>

Riley, S.J., Luscombe, G.B., and Williams, A.J. (eds) 1986a.Proceedings Urban Flooding Conference: a conference on thestorms of 8 November 1984. Geographical Society N.S.W.Conference Papers No. 5.

Riley, S.J., Luscombe, G.B., and Williams, A.J. 1986b. Urbanstormwater design: lessons from the 8 November 1984 Sydneystorm. Australian Geographer 17: 40–50.

Sheffer, N.A., Enzel, Y., Waldmann, N., Grodek, T., and Benito, G.2003. Claim of largest flood on record proves false. EOS,Transactions American Geophysical Union 84(12): 109.

Shields, A.J. 1979. The Brisbane floods of January 1974, In Heath-cote, R.L. and Thom, B.G. (eds) Natural Hazards in Australia.Australian Academy of Science, Canberra, pp. 439–447.

Tropeano, D., Fabio Luino, F., and Turconi, L. 2000. Flooding inwestern Italian Alps, 14–15 October 2000. <http://www.irpi.to.cnr.it/English/Events%2014-15%20october/Events%20.htm>

Trotter, P.S., Johnson, G.A., Ricks, R., Smith, D.R., and Woods, D.1998. Floods on the lower Mississippi: An historical economic over-view. <http://www.srh.noaa.gov/topics/attach/html/ssd98-9.htm>

Yeo, S. 2002. Flooding in Australia: A review of events in 1998.Natural Hazards 25: 177–191.


INTRODUCTION

Of all natural hazards, the most insidious is drought.However, for some countries such as Australia andUnited States, droughts have not led to starvation, butto spectacular fires as tinder-dry forests ignite, grass-lands burn and eucalyptus bushland erupts in flame. Ofall single natural hazard events in Australia, bushfiresare the most feared. Here, the litany of disasters overthe past 150 years reads like the membership list ofsome satanic cult: ‘Black Thursday’, ‘Black Friday’, and‘Ash Wednesday’ (Figure 7.1). For firefighters, stateemergency personnel and victims, each name willinvoke stories of an inferno unlike any other. North

America has suffered just as badly in the past from fires,as have Europe and the former Soviet Union. Approxi-mately 143 � 106 km2 of the Earth’s surface is coveredby vegetation, of which 0.17 per cent burns on averageeach year. Even tropical rainforests can dry out andburn. For example, the U Minh forest in Vietnam (seeFigure 7.2 for the location of major placenames men-tioned in this chapter) ignited naturally in 1968. Largefires have also burnt through tropical vegetation in theBrazilian highlands and the Amazon Basin, especially inassociation with land clearing. Many urban dwellerswould consider that, following wide-scale deforestation,there is no forest fire threat near cities. Fires are per-ceived to be a hazard only in virgin forest or, perhaps in

Fires in NatureC H A P T E R 7

Fig. 7.1 William Strutt’s painting Black Thursday (LaTrobe Collection, © and with permission of the State Library of Victoria). The title refers to thefires of 6 February 1851, which covered a quarter of the colony of Victoria, Australia.

southern California or southern France, a hazard whereurban expansion has encroached upon vegetated moun-tains. In fact, any country where climate is influencedby aridity or by periodic drought is susceptible tonatural fires. This is particularly true for countriesexperiencing a Mediterranean climate, especially thosewhere highly flammable eucalypts have supplantedmore fire-resistant vegetation. More importantly, largeareas of the northern hemisphere have undergonereforestation this century, with the abandonment ofeconomically marginal farmland. Here, forest fires mayagain pose the severe hazard they were under initialsettlement more than a century ago.

This chapter will examine the hazard of drought-induced natural fires. Those weather patterns favoringintense fires will be described first, followed by adiscussion of the main causes of fire ignition. Majorbushfire disasters will then be described, with particularemphasis upon the American and Australian environ-ments, where some of the most intense conflagrationshave been witnessed.

CONDITIONS FAVORINGINTENSE BUSHFIRES

(Vines, 1974; Luke & McArthur, 1978; Powell, 1983; Voice& Gauntlett, 1984; Yool et al., 1985)

The potential for a major fire hazard depends upon thevegetation type, fuel characteristics, climate, and fire

behavior. Eucalyptus is one of the most fire-pronevegetation types, having leaves with a high oil contentthat fosters burning. Such trees can rapidly renewfoliage through epicormic shoots on the main trunk, orthrough shoots on buried and protected roots. Afteronly a few years post-fire, they can become sites ofconflagrations as severe as the original fire. Eucalyptushas been exported successfully from Australia andnow covers significant areas of southern California,northern Africa, the Middle East, and India. Otherfire-prone vegetation dominated by scrub oak,chamise, and manzanita exists in Mediterraneanclimates. Such vegetation communities are labelled‘chaparral’ in California, ‘garrigue’ or ‘maquis’ inFrance, ‘phrygana’ in the Balkans, ‘matarral’ in SouthAmerica, ‘macchin fynbosch’ in South Africa, and‘mallee’ or ‘mulga’ scrub in Australia. In regionsaffected by severe drought, fuel characteristics are veryimportant to the spread and control of fires. Fuelcharacteristics can be separated into two categories:grass, and bush (mainly Eucalyptus). Grasslandsproduce a fine fuel both on and above the groundsurface. Such material easily ignites and burns. Inforests, the fine fuel consists of leaves, twigs, bark, andstems less than 6 mm in diameter. Apart from Euca-lyptus leaf litter with its high oil content, most forestlitter is difficult to ignite.

The arrangement of fuel is also important. Com-pacted litter is more difficult to burn and to extinguish


Los Angeles

MaineLong Island

Great Lakes

SydneyCanberra

Balkans Amur R.

New South Wales

VictoriaMelbourne

Hobart

South AustraliaWestern Australia

Adelaide

Kalimantan

Straits of Malacca

U Minh Forest

Angara - Tunguska R.

Ural MtnsMiramichi

PhiladelphiaNorth CarolinaGeorgia

Michigan TimminsWisconsin

Montana

IdahoTillamook

Yellowstone Park

Boulder

Minnesota

San Diego Okefenokee Swamp

Oakland

Amazon Basin


than litter with a lot of air mass. Forests with denseundergrowth (especially pine plantations) will burnmore readily than forests where the undergrowth hasbeen removed. The moisture content of living anddead material also dictates the possibility of ignitionand fire spread. Dead grasses readily absorb moisture,their moisture content fluctuating with the diurnalhumidity cycle. Humidity increases at night, andgrasses have their highest moisture content in the earlymorning. Monitoring of the proportion of cured grassand its moisture content can give a good index of thesusceptibility of that fuel to ignition and burning. InEucalyptus forest, the types of tree species conditionfire behavior. Stringy-barks permit fire to reachtreetops and throw off flaming bark, thus favoring spotfires, whereas smooth barks do not generate flyingembers.

The climate before a fire season also conditions thefire threat. If fuel has built up, then climatic condition-ing is less important; however, it still takes dry weatherto cure and dehydrate grass and bush litter. Thisdesiccation process is exacerbated by prolongeddrought in the summer season. Typically, El Niño–Southern Oscillation (ENSO) events occurring atChristmas coincide with Australia’s summer season. InCalifornia, La Niña events produce fire weather.During these types of events, dry conditions areguaranteed for several months, ensuring that fuels dryout more completely. Because such events can now bepredicted months in advance, it is possible in many

countries to warn of severe fire seasons before summerarrives.

The prediction of bush or forest fires may evenextend beyond this time span. As pointed out inChapter 2, the 11-year sunspot cycle appears in rainfallrecords for many countries, while the 18.6-year lunarcycle dominates the coincidence of worldwide droughtor rainfall. Because natural conflagrations are mostlikely during major droughts, it is to be expected thatforest fires evince one of these astronomical cycles.Vines (1974) has performed detailed work on thecyclicity of bushfires in Australia and Canada, andbelieves that cyclicities of 6–7 years and 10–11 yearsappearing in these countries directly reflect sunspotcycles. The 10–11-year periodicity also appears inthe fire records of the southern United States since1930, although at sunspot minima. In the Great Lakesregion, major conflagrations between 1870 and 1920coincided very well with peaks in the 11-year sunspotcycle. Logically, major fire seasons should also corre-late with the 18.6-year lunar cycle; however, littleresearch has been carried out to verify this conjecture.

Figure 7.3 illustrates the typical daily weatherpattern leading to extreme bushfires in south-easternAustralia. A high-pressure system moves slowly south-east of Australia and ridges back over the continent.This directs strong, desiccating, north-west winds fromthe interior of the continent across its south-eastcorner. These winds are followed by a strong southerlychange that may drop humidity but maintain high wind

141Fires in Nature

0 1000 km

L L

L

1008

1012

1016

1020

1024

1008

1004

1000

SubtropicalJet

1000

1004

1008

1012

1016

1020

1024

1028H

7 Feb 1967

Fig. 7.3 Typical synoptic weather pattern conducive to bushfire conflagrations in south-eastern Australia.

speeds. Winds blowing from the interior of the conti-nent warm adiabatically as they drop in elevation fromthe highland areas of Victoria to the coast, producingwhat is commonly known as a föhn wind. This adiabaticwarming can push temperatures to 40°C or above atthe coast. The process also reduces relative humidity tolevels below 20 per cent. Similar winds exist elsewhere,especially in mountainous Mediterranean climatesdominated by Hadley cells. Local variants of föhnwinds are termed ‘mono’ in central California, ‘Diablo’around San Francisco, ‘Santa Ana’ in southern Califor-nia, ‘sirocco’ in southern Europe, and ‘bora’ in theBalkans. In Australia, strengthening of föhn winds isaided by convergence of isobars, and reinforced by afast-moving subtropical jet stream in the upper tropo-sphere on the western side of the high-pressure cell.Because this sequence of winds moves eastward acrossthe continent over several days, there is a gradual shiftin the bushfire threat from South Australia to Victoria,and then to New South Wales. In Western Australia,fire weather is characterized by positioning of high-pressure cells with north-easterly winds out of theinterior.

Fire behavior and local topography ultimatelydetermine the severity of the fire hazard. Often firesaccelerate as the day progresses, because fuel moisturecontent drops and wind speed and atmosphericinstability increase owing to diurnal heating. Addition-ally, as the fire intensifies during the day, convection isgenerated by the extra heat supplied to the atmos-phere. Radiation from the fire dries out fuel in front ofthe fire, and spot fires can prepare forward areas forthe passage of the main fire. Spotting in front of fires isquite common in Australia because of the presence ofstringy-barks. Spotting ensures that convection occursin front of the main fire, thus drawing the main fire-front towards the spot fires.

As wind speed increases, the rate of fire movementincreases exponentially. A wind speed of 10 km hr-1

will move a fire – in Australian bushland of average fuelcontent – at the rate of 0.5 km hr-1. At 20 km hr-1, thisrate increases to 0.8 km hr-1 and, at 40 km hr-1, to1.8 km hr-1. If the fire actually reaches treetops andburns through the crowns of trees (Figure 7.4), it canaccelerate and travel at speeds of 20 km hr-1. Withspotting, fires can move forward or ‘propagate’ fasterthan this, creating a firestorm that can travel at60 km hr-1 or more. The rate at which a bushfire movesdoes not taper off with increased wind velocities.


Fig. 7.4 Crown fire in A) Eucalyptus forest, Australia (photograph byN.P. Cheney, National Bushfire Research Unit, CSIRO,Canberra)

Fig. 7.4 and B) a boreal forest in northern Ontario, Canada(photograph by Jim Bryant, Ontario Department of theEnvironment). Note the similarities between the fires.Eucalyptus is perceived as more flammable, but borealforests produce three to four times more energy per unittime because of their greater biomass.

However, in grassland, fire movement decelerates atwind speeds above 50 km hr-1 because of fragmenta-tion of the fire head. This is not significant when oneconsiders that, under the same wind speed, grass firescan move about eight times faster than a bushfire. Iffires move upslope, their rate of advance will increasebecause rising heat from combustion dries out fuel.Fires will also move faster upslope because the windvelocity profile steepens without the influence ofground friction. In rugged terrain, spot fires are morelikely because embers can be caught by winds blowingat higher speeds at the tops of hills or ridges.

Atmospheric stability also controls fire behavior. Ifthe atmosphere is unstable, then convective instabilitycan be established by the fire’s heating of theatmosphere. Once initiated, convective instability willcontinue even after the heat source is removed. Underconditions of extreme instability, fire whirlwinds aregenerated and can reach speeds exceeding 250 km hr-1.These whirlwinds are of three types. The first is relatedto high combustion rates and generates a rotationalupdraft, which can move embers a couple of hundredmeters beyond the fire-front. The second type repre-sents a thermally induced tornado. This type originatesin the atmosphere on the downwind side of theconvective column or the lee side of topographicrelief. The winds are capable of lifting large logs andproducing gaseous explosions in the atmosphere.These vortices are particularly dangerous becausewind is sucked towards the fire tornado from alldirections. The third type of whirlwind is the post-burnvortex (Figure 7.5). It forms because of the heatemanating from a burnt area up to a day after the fire.These vortices are usually less than 20 m in diameter,but they can carry embers from burnt to unburnt areas.

CAUSES OF FIRES

(Luke & McArthur, 1978; Pyne, 1982; van Nao, 1982)

No matter how dry a forest or grassland becomes,there can be no fire unless it is ignited. The recordedcause of a fire depends upon how rigorously fire statis-tics are collected in individual countries. For instance,in European countries, 45 per cent of all fires are ofunknown cause; however, in Canada, only 5 per cent offires have unknown causes. In Europe, only 2 per centof the known causes of bushfires can be attributed tonatural causes such as lightning. Over half are theresult of arson and 40 per cent, of human carelessness.

The latter group includes burning that gets out ofcontrol, smoking, children playing, fires due to trains,and camp fires. In Canada, lightning is responsible for32 per cent of all forest fires, whereas in the UnitedStates it accounts for fewer than 8 per cent of fires. Thelatter figure is misleading because lightning has playeda major role in igniting fires in isolated areas of theUnited States. Between 1940 and 1975, over 200 000fires were started by lightning in the western UnitedStates alone. This represents an average of 15.6 firesper day. Single storm events or pressure patterns canalso produce a large number of lightning fires in a shortspace of time. Between 1960 and 1971, the westernUnited States experienced six events that produced500–800 lightning fires each. The largest such event, inJune 1940, started 1488 fires.

Humans are the greatest cause of fires. Arson bymentally disturbed people accounts for 32 per cent ofall fires in the United States and 7 per cent in Canada.

143Fires in Nature

Fig. 7.5 Whirlwind associated with a clearing after burning (photographby A.G. Edward, National Bushfire Research Unit, CSIRO,Canberra).

In southern California, nearly a third of all fires arestarted by children, most of whom are boys playingwith matches. Throughout the twentieth century inthe United States, economic downturns have broughtincreased incendiarism, because firefighting affordstemporary job opportunities for the unemployed. Themajor difference between arson and lightning as acause of fire is the fact that lightning-induced fires tendto become conflagrations because they occur inisolated areas, while arson-induced fires tend to occurin accessible areas that are relatively easy to monitorand control. In the United States 2000 fire season,lightning caused 7659 fires burning out 4930 km2 ofvegetation. In contrast, humans caused 96 390 firesthat burnt across 10 050 km2. Of this number,30 per cent were due to arson and consumed34 per cent of the area burnt.

In Australia, the causes of fire are more numerousbecause of the variety in vegetation, climate, and land-use. Additionally, prescribed burning as a fire preven-tion procedure is practiced more in this country thanany other part of the world. This is despite the fact thatprescribed burning was used in the south-east UnitedStates for clearing undergrowth and minimizing theoutbreak of major fires. Between 1966 and 1971 inWestern Australia, 36 per cent of all fires resultedfrom burning-off; only 8 per cent were due to light-ning. Vehicles, farm machinery, and trains ignited15 per cent of all fires, with carelessness accounting for21 per cent. In the Northern Territory, most fires arecaused by either lightning or burning-off. Here,Aborigines traditionally light fires to clear the land-scape so they can catch game. In eastern Australia,where the population is greater, arson has become amajor cause of fires in recent years. Single individualshave been found to light more than one fire on a highfire-risk day. Because the spring burning-off season canencroach upon early warm weather, notably duringENSO years, the incidence of fires caused by burning-off has also increased in recent years.

BUSHFIRE DISASTERS: WORLDPERSPECTIVE

(Cornell, 1976; van Nao, 1982; Seitz, 1986; Couper-Johnston, 2000)

Bushfires as a hazard are usually evaluated in terms ofloss of life, property damage (burning of buildings orentire settlements), and the loss of timber. While loss

of life can still occur in northern hemisphere temper-ate forests, most destruction is evaluated in terms ofthe loss of marketable timber. This statement needsqualification because timber has no value unless it canbe accessed, and there is no imperative to access itunless it is close to centers of consumption. In NorthAmerica, it is mainly the forests lying within 1000 kmof dense urban development that are used to supplybuilding timber, unless there is a need for timber withspecific qualities – such as the Californian Redwood.Beyond this distance, forests are mainly used to supplypulp and paper in the form of newsprint. There arelarge areas in the Canadian and Alaskan boreal forestwhere no harvesting of trees takes place because it isuneconomic. These areas generally are uninhabitedand no attempt is made to suppress forest fires.Fires are allowed to burn uncontrolled, and may takeweeks or the complete summer season to burn out. Innorthern Canada, it is not uncommon for such fires tocover thousands of square kilometers.

In the Middle Ages, large areas of Europe weresubjected to forest fires. Damage and loss of life hasgone virtually undocumented, although it is estimatedthat today’s conflagrations cover only 20 per cent of thearea they did centuries ago. Today, large tracts ofdeciduous, mixed and chaparral forests remain inwhich fires pose a serious threat. In Europe as recentlyas 1978, over 43 000 fires consumed 440 000 hectaresof forest and as much again of grasslands and crops.These figures are not remarkable. No Europeancountry has had more than 2 per cent of its total forestcover burnt in a single year. Individual fires on a muchlarger scale have occurred throughout the northernhemisphere. In October 1825, over 1 200 000 hectaresof North American forest burned in the Miramichiregion of New Brunswick, together with 320 000hectares in Piscataquis County, Maine. The Wisconsinand Michigan forest fires of 1871 consumed 1 700 000hectares and killed 2200 people. Areas of similarmagnitude were burnt in Wisconsin in 1894 and inIdaho and north-western Montana in 1910. The Por-cupine Fire of 1911 raced unchecked from Timmins,Canada, north through virgin forest for over 200 km,killing scores of settlers on isolated farms. Duringthe 1980 Canadian fire season, virtually the wholemixed forest belt from the Rocky Mountains to LakeSuperior was threatened. Over 48 000 km2 of forestwere destroyed, with a potential loss equivalent to thetimber required to supply 340 newsprint mills for one


year. The destruction only got worse – with 73 880 km2

and 64 040 km2 of forest going up in smoke in 1989and 1994, respectively. The magnitude of this burningwas matched along the Amur River (separating Chinaand Siberia) by the Great Black Dragon Fire of 1987,which burnt through 73 000 km2, killed 220 people andleft 34 000 homeless. The areas of these last three firesare equivalent to the area of Scotland, and are 13 timeslarger than the Yellowstone Fire of 1988.

All of these fires pale in significance when comparedto the Great Siberian fires of July–August 1915. In thatyear, up to 1 million km2 of Siberia, from the UralMountains to the Central Siberian Highlands, burned(an area 2–20 times greater than the extent of the 1980fires in Canada). An area the size of Germany, covering250 000 km2, was completely devastated between theAngara and Lower Tunguska Rivers. These giganticfires were brought about by one of the worst droughtsever recorded in Siberia. Most of the taiga and larchforests were desiccated to a combustible state, produc-ing crown fires. Over 500 000 km2 of peat dried outand burnt to a depth of 2 m. The amount of smokegenerated was 20–180 � 106 tonnes, a figure equivalentto the amount of smoke estimated to be produced in alimited-to-extreme nuclear war leading to a nuclearwinter. Thick smoke was lifted more than 12 km intothe sky. In the immediate area, visibility fell to 4–20 mand to less than 100 m as far as 1500 km away. As aresult, the solar radiation flux was significantly attenu-ated at the ground, dropping average temperatures by10°C. At the same time, long wave emission at nightwas suppressed such that the diurnal temperaturerange varied by less than 2°C over a large area. Thesebelow-average temperatures persisted for severalweeks; however, harvests in the area were delayed byno more than two weeks. Little evidence of the fireswas detectable outside Siberia. Because of the isolationof the area and the low population density, reportedloss of life was minimal. The above documentation oflarge fires implies that conflagrations of this magnitudeare frequent occurrences in the boreal forests of thenorthern hemisphere. Forest fires presently posethe most serious natural hazard in the two largestcountries of the world, Canada and Russia.

Nor are fires restricted to temperate latitudes. Since1980, large fires have swept through the tropics ofIndonesia and Amazonia, exacerbated by ENSOevents and human malpractice. During the 1982–1988,1994, and 1997–1998 ENSO events, 32 000 km2,

51 000 km2, and 50 000 km2, respectively, of rainforestburnt across Indonesia, mainly in Kalimantan. Duringthe 1997–1998 ENSO event, smoke due to slash-and-burn agriculture, forest clearing, and firing ofpeatlands enveloped over 100 million km2 of South-East Asia in an acrid, pea-soup haze that reducedvisibilities to 2–100 m. At one point the pollutionreading at Pontianak in west Kalimantan reached1890 mg m-3, 40 times the level considered safe by theWorld Health Organization. In Indonesia, about onemillion people suffered health problems and 40 000had to be hospitalized. At one school, visibility was solow that school children had to be roped togetherto keep them from getting lost. About 2.6 � 109 tonnesof carbon – 1000 times more than produced by theSiberian fires of 1915 – were released into the atmo-sphere, an amount equivalent to 40 per cent of globalemissions that year from fossil fuels. Because ofreduced visibility, two ships collided in the Straitsof Malacca, killing twenty-nine people. The haze wasalso implicated in the crash of a Garuda Airbus intoa mountain at Medan with the loss of 234 passengersand crew. Ultimately, 7000 professionals and 31 000volunteers fought the fires. As Indonesia’s resourcesbecame overwhelmed, the international communityresponded with cash donations, medical assistance,and firefighting expertise at a level usually reservedfor a major earthquake or drought. Indonesia’s grossnational product dropped 2.5 per cent due to thedecline in its tourist and forest industries.

United States(Pyne, 1982; Romme & Despain, 1989; Firewise, 1992;ThinkQuest Team, 2001)

Fire history

In North America at least 13 fires have burned morethan 400 000 hectares. In the twentieth century, almostall such mega-fires have begun as controlled burns, oras wildfires that at some point could be consideredcontrollable. And almost all severe fire seasons werepreceded by drought. The intensity of any fire is notnecessarily dependent upon the growth of biomassduring wet seasons, but on the availability of driedlitter or debris generated by disease or insect infes-tation, on windstorms, previous fires, or land clearing.This point is well-recognized. For example, the 1938New England hurricane mentioned in Chapter 2 left atrail of devastation in the form of uprooted trees and

145Fires in Nature

felled branches. Considerable effort was spent cleaningup this debris before it could contribute to major forestfires. Fire-scarred areas take decades to regenerate,and regeneration in these areas is particularly sus-ceptible to future burning. In North America, largeburnouts such as the Tillamook in Oregon, burnt origi-nally in 1933, spawned numerous fires for the next30 years. Indians who cleared or fired forests for agri-culture, firewood, hunting, and defense conditionedmuch of the early fire history of North America. Thus,the forests that met white settlers were park-like andopen. While European settlement adopted many nativeAmerican fire practices, the demise of such practicesmade possible the reforestation of much of the north-east and began the period of large-scale conflagrations.A similar process followed the decline of prescribedburning – first by the turpentine operators and then bythe loggers in the pine forests of the south-east. Themost dramatic effects are seen on the Prairies, whereforests have reclaimed a grassland habitat as land wasconverted to cultivation or rangeland. In southernCalifornia, urban development has seen grasslandstreed simply for aesthetic reasons. In each case, Euro-peans have attempted to supplant natural fire cycleswith total suppression to protect habitation.

The onset of forest fires in reforested regions of theUnited States was dramatic. Disastrous fires sweptthrough the north-east in 1880, and it became a policyto restrict settlement in forest reserves. That policy ledto reduced burning-off and a build-up of forest fuelthat resulted in 400 000 hectares burning from NewYork to Maine. The fires were perceived as a threat totimber reserves and watershed flood mitigation. Theyresulted in a concerted effort to set up fire lookouts inthe north-east, and prompted the passage of the WeeksAct in 1911 to protect the watersheds of navigablestreams and to subsidize state fire protection efforts.These efforts – along with the formation of the CivilianConservation Corps (CCC) in the 1930s – saw thedemise of large wildland fires; however, urbanizationhad expanded into forest areas. The nation’s first urbanforest fire occurred in October 1947 at Bar Harbor,Maine. Over 200 structures were burnt and 16 liveslost. This scene was repeated over a wide area in 1963when fires threatened suburbs in New York andPhiladelphia, and destroyed over 600 structures onLong Island and in New Jersey. These fires led toincorporation of municipal fire units into the stateprotection system, and the signing of interstate

agreements for fire suppression assistance. By the1970s, the latter had spread nationally and internation-ally. Today bushfires also plague southern California’surban areas, which since the 1950s have encroachedupon fire-prone scrubland. Little thought has beengiven to building design and layout, with a jumble oforganizations responsible at various levels for firesuppression. The costs of firefighting operations insouthern California as a result have reached highlevels. In 1979, the Hollywood Hills and Santa Monicamountain fires involved 7000 firefighters at a cost of amillion dollars per day for a month.

In the south-east of the United States, reforestationand regeneration of undergrowth presented, by 1930,large reserves of biomass that fuelled major fires duringdroughts. In 1930–1932, Kentucky, Virginia, Georgia,and Florida suffered devastating fires. Further firesbetween 1941 and 1943 saw the reintroduction of prescribed burning. In 1952, 800 000 hectares offorest burned across Kentucky and West Virginia;in 1954–1955, 200 000 hectares burned in theOkefenokee Swamp and, in 1955, 240 000 hectaresburned in North Carolina. These occurrences led to theestablishment, in North Carolina, of one of the mostefficient fire suppression organizations in the countryand, in Georgia, of the first fire research station. Thisstation produced the nation’s leading methods inprescribed burning.

In the virgin forests of the Lake states, similar-sizedfires have occurred historically. However, the deathtolls have been horrendous. Here, disastrous firesoccurred as settlers first entered the region, togetherwith logging companies and railways. The loggingproduced enormous quantities of fuel in the form ofslash on the forest floor; the railways provided sourcesof ignition from smokestacks; and the settlers providedbodies for the resulting disaster. Between 1870 and1930, one large fire after another swept through theregion in an identical pattern. Most of the conflagra-tions occurred in autumn following a summer drought.The worst fires occurred in 1871, 1881, 1894, 1908,1910, 1911 and 1918. The Peshtigo and Humboldtfires of 1871 took, respectively, 1500 and 750 lives inWisconsin; the Michigan fires of 1881 killed severalhundred; the Hinckley, Minnesota, fire in 1894 took418 lives; and finally the Cloquet fire in Minnesota in1918 killed 551 people. The fires were so common,that most residents treated fire warnings nonchalantly,to their own detriment.


The great fires of 1871 were the first, and by farthe worst, to sweep the Lake states. Ironically theyoccurred in October on the same day as the GreatChicago fire. In the weeks preceding the fires, much ofthe countryside was continually ablaze. Almost all able-bodied men were employed in fire suppression, andmany inhabitants had inaugurated procedures toprotect buildings. On 8–9 October the fires broke intofirestorms only ever witnessed on a few other occa-sions. Flames leapt 60 m into the air and were drivenahead at a steady rate of 10–16 km hr-1 by spot fires,intense radiation, and updrafts being sucked into themaelstrom. The approach of the flames was heraldedby dense smoke, a rain of firebrands and ash, andby combusting fireballs of exploding gases. Winds of100–130 km hr-1 picked up trees, wagons and flamingcorpses. The fire marched with an overpowering roarlike continual thunder or artillery barrages. Hundredswere asphyxiated hiding in cellars, burned to deathas they sheltered in former lakebeds converted toflammable marshes by the drought, or trampled bylivestock competing for the same refuges. The Lakefires were certainly the worst in documented historyand show how vulnerable any forest can be under theright fuel and climatic conditions.

Recent disasters

Despite these efforts to identify the fire regime andcome to terms with urban encroachment on forest,fires in North America continue to be a major hazard.Three events in recent years illustrate the range ofproblems: the Yellowstone fire of 1988 because itpitted advocates for natural burns against those fortotal fire suppression; the Oakland fire of 20 October1991 because it characterizes the urban threat; and the2000 fire season because it shows how widespreadthe problem can be during regional drought.

The Yellowstone fire was impressive for its intensity.More area was burnt in the region on 20 August 1988than during any decade since 1872. The fires wereignited in June by lightning following six years ofabove-normal rainfall. For the first month, the fireswere allowed to burn unchecked; but in mid-July, after35 km2 had been burnt, a decision was made to containthem. A week later this area had doubled and a monthlater 1600 km2 had been incinerated despite massiveefforts at suppression. The dramatic increase inburning was due to drought conditions that were theworst since the 1930s. Winds as high as 160 km hr-1

stoked the fires despite the 9000 firefighters brought into contain them. By mid-August, at least eight separatefires were burning at rates as high as 20 km day-1 withtemperatures exceeding 32°C. By the time the fireswere finally extinguished, 5666 km2 of forest had beenburnt. Some of the best natural scenery in the UnitedStates had been scarred for decades and a debate wasignited pitting those who favored a laissez faire attitudeto fires against those who favored complete suppres-sion. Supporting the former view was the fact that inthe previous twelve years, 235 fires had been ignited bylightning, allowed to blaze, and had died out afterburning an average of 0.4 km2 of forest. The largest firedestroyed only 30 km2 and had threatened no lives.Slowly people came to realize that the 1988 fireswere an indication that a ‘natural order’ was being re-established – one that had been upset for a century.Suppression of fires for over a century had allowed anecosystem to develop that would otherwise have beenreturned, by fires every ten to 20 years, to a morenatural state. The forests in Yellowstone generallybecome more flammable as they age. This is becauseold-growth forests accumulate greater amounts ofdead biomass at ground level and are susceptible tocrown fires because the trees are taller with unprunedlower limbs. The Yellowstone fires had reached thislimit in the 1930s, but benign climate and the intro-duction of fire suppression had extended the fire-pronestate. The suppression consisted of ‘Smokey the Bear’technology and heroics. Aircraft and helicoptersscoured the landscape for the first sign of fire, waterbombers attacked even the most remote flame, andsmoke jumpers were parachuted into sites withinhours of ignition. All fires were to be suppressed by10 o’clock in the morning. The fires of 1988 were partof a natural 200–300-year cycle of fire that was delayedby 30–40 years of efficient fire suppression.

Opposing fire suppression was the view thatnature should be aided by prescribed burning. Hadthe Yellowstone forests been culled artificially bysanctioned arson, under weather conditions disadvan-tageous to fire, then the smoke that hid the scenery,affected health and deterred the tourists would nothave been so extreme. The Yellowstone fires fuelledthe view that forested landscapes, burnt or unburnt,have different values to diverse interest groups.

The Oakland, California, fire of October 1991 wasnot the first fire to sweep the area. From 1920 to 1995over 3500 structures were lost to fire in the East Bay

147Fires in Nature

Hills – the worst event being the loss of 584 homes in1923. This urban–wildfire conflict was exacerbated bythe dominance of fire-prone vegetation such asEucalyptus. Monterey pine, chaparral, grass, and orna-mental species such as junipers and cedars were alsopresent. All are highly flammable, generate intenseradiation, and produce easily blown embers. The fireswere aided by the weather. The region had experi-enced five years of drought while frosts the previouswinter had increased the amount of dead vegetation inthe forest. On the day the firestorm occurred, relativehumidity dropped to 16 per cent and temperaturesrose to a record 33°C. The local föhn winds enhancedthe firestorm. These winds became turbulent over thelocal terrain and reached gusts of 90 km hr-1. Finally,an inversion layer developed below 1000 m, containingwithin the lower atmosphere the heat generated by thefires and preheating vegetation in front of the flames.

The disaster began as a small brush fire on19 October 1991 in the hills above Oakland. This firewas easily controlled, but it continued to smolderbeneath a 20 cm thick carpet of pine needles. At10:45 am on 20 October, flames reappeared and devel-oped into a wildfire that lasted three days and becamethe worst urban wildfire disaster in United Stateshistory. Within 15 minutes of reignition, the firebecame a firestorm that generated its own weather andproceeded along several fire fronts. Within 30 minutes,eight pumping stations and ten water reservoirs werelost due to power failures. Within an hour, the fire –spreading at a rate of 1.67 m s-1 – destroyed 790buildings. Local firefighting resources were over-whelmed and a Bay-wide call for assistance was issued.Eighty-eight fire trucks, six airplanes, 16 helicoptersand over 1400 service personnel responded to the fires.However, fire units from outside the area foundthat their hose connections were incompatible withOakland’s hydrant system. Communications amongstemergency services failed within the first twelve hoursas call lines became congested. The rugged terraininterfered with radio and cellular (mobile) phonetransmissions. Abandoned vehicles, narrow roads,downed power lines and the thick smoke trapped inthe inversion layer hampered firefighting efforts. Thesame restrictions hindered evacuation of residents.Nineteen people were killed trying to flee the fires;eleven of these died trapped in a traffic jam. By the endof the first day, ten key reservoirs were drained andfirefighters had to rely upon tankers for water supplies.

The fires were extinguished only because they began toburn more slowly as they reached flatter terrain, andbecause the winds diminished. After burning for threedays, the firestorms had burnt through 6.4 km2 ofaffluent neighborhoods, destroyed 3469 structuresworth an estimated $US1.6 billion, injured 150 peopleand taken 25 lives.

Following the 1998–1999 La Niña event in theUnited States, the 2000 fire season was conditionedby widespread drought from Florida in the south-eastto Washington in the north-west. Across the country,122 827 fires burnt across 34 000 km2 – more thandouble the annual average – destroying 861 buildings,killing 11 people, and costing $US1.6 billion tosuppress. At the peak of the fire season in Florida,more than 500 wildfires broke out each day. Eight ofthe ten worst fires were concentrated in the north-west. This situation stretched the country’s firefight-ing capacity. At the season’s peak on 29 August, 1249fire engines, 226 helicopters, 42 air tankers and28 462 people were fighting the fires. Eventually per-sonnel and equipment had to be brought in fromCanada, Mexico, Australia, and New Zealand. Andstill the carnage intensified. In October 2003, fifteenfires in southern California burnt out 3000 km2,destroying 3640 homes and 1174 other structures(Figure 7.6).


Fig. 7.6 An aerial view of total devastation in a housing estate follow-ing the southern Californian fires on 29 October 2003. Notethe lone house that survived unscathed (photo 031029-N-4441P-013 taken by Photographer’s Mate 2nd Class MichaelJ. Pusnik, Jr., of the U.S. Navy assigned to the ‘Golden Gators’of Reserve Helicopter Combat Support Squadron Eighty Five(HC-85). Used with permission of Director, Naval VisualNews Service, U.S. Navy Office of Information.<http://www.news.navy.mil/list_single.asp?id=10314>).

These recent fires have led to a shift in the wayfires are dealt with in the United States. Prior to theYellowstone fire of 1988, fire suppression took prece-dence. This led to the accumulation of abundant fuelstores and more intense firestorms when fire inevitablybroke out. Since then, protection of human life and ofproperty have become the first and second priorities,respectively, in fire prevention and management. Anemphasis has also been placed upon urban wildfires,including fuel reduction programs and education abouthazard minimization. While government bodies andagencies are mainly responsible for fire management,private property owners are also being encouraged andeducated in fire prevention and protection in urbanvegetated areas. Importantly, wildfires are now recog-nized as a critical natural process for the health ofecosystems. The Forest Service has instituted a newfire management policy called Fire 21. Unless lives andproperty are threatened, fires will be allowed to burnas part of their natural ecological role.

Australia(Luke & McArthur, 1978; Powell, 1983; Oliver et al., 1984;Voice & Gauntlett, 1984; Webster, 1986; Pyne, 1991)

Conditions

Australian bushfires are unique, because they representthe present-day occurrence of deadly fires previously

witnessed only in the United States and Canada beforethe 1940s. Australian bushfires are also unique becauseof the highly flammable nature of Australian forestvegetation. Globally, Australian bushfires compare insize to the largest recorded in North America. Whileloss of life is now minimal in North America because ofefficient evacuation procedures and the nature of set-tlement, in Australia it appears to be increasing becauseof expansion of urban populations into rugged bushcountry. Mention has been made of the free-burningnature of forest fires in northern Canada. Whileattempts are made in Australia to put out fires, manyin inaccessible hill country are left to burn. These firespose a threat directly if they change direction understrong winds, and indirectly because they provide asource for fire spotting beyond the limits of the mainfire-front. Australian bushfires are also impossible tocontrol once they become wildfires, because of the highflammability of the bush. Some of the most intense fireswere the conflagrations in Hobart in 1967 and south-eastern Australia in 1983.

The map in Figure 7.7 shows that the spatial occur-rence of fires in Australia is seasonally controlled.The tropical zone has a winter–spring fire season,because the summer is usually too wet and grasses willnot cure and dry out before spring. The fire seasonprogresses southward and across the continent alongthe coasts through the spring. By late spring–summer,

149Fires in Nature

QueenslandQueensland

New SouthWales

VictoriaTasmania

Hobart

Melbourne

SydneySCanberraSouth

Australia

Adelaide

n

WesternAustraliaAustralia

NorthernTerritory

Winter & Spring

Spring

Spring & Summer

Summer

0 1000 km

Fig. 7.7 Pattern of seasonal bushfires in Australia (after Luke & McArthur, 1978).

the fire season has peaked in a line stretching fromGeraldton to Canberra and Sydney. By mid-summer,the zone has moved to the southern part of the conti-nent. At this time, hot, desiccating winds from theinterior can blow towards the coast as mobile polarhighs stall over the Tasman Sea (Figure 7.3). Histori-cally, this is the time of severest bushfires and greatestloss of life in Australia. Finally, the southern extremi-ties of the continent experience bushfires as vegetationdries out at the end of summer.

Bushfires are not necessarily restricted to the seasonsshown in Figure 7.7. In the central spring and summerzone, severe bushfires have been known to occur in theautumn. For example, in 1986 the worst bushfires ofthe season broke out in early April following a drysummer and record-breaking temperatures for themonth.

Historic disasters

The worst fires in Australian history are well-documented and have occurred in the fire-pronebelt of southern South Australia and Victoria. On14 February 1926, fires ignited under temperaturesin excess of 38°C and relative humidity of lessthan 15 per cent. Sixty lives were lost and an untoldnumber of farms, houses and sawmills burnt out. Inthe 1931–1932 fire season, 20 lives were lost. The13 January 1939 fires occurred under record-breakingweather conditions. Fires raced through extremely drybush as temperatures rose to 46°C and humiditydropped as low as 8 per cent. Over 70 lives were lost innumerous fires in the south-eastern part of the country.The 1943–1944 season was a repeat. Over 49 peopledied in grass and bushfires under oppressively hot anddry weather. Fires in the Dandenongs in 1962 killed14 people and burnt 450 houses. On 8 January 1969,23 people died in bushfires as temperatures exceeded40°C and winds topped 100 km hr-1.

Since the Second World War, there have been fourfire events that stand out as exceptional: the Hobartfires of 7 February 1967, the Ash Wednesday firesof 1983, the Sydney fires of January 1994, and theCanberra maelstrom of 18 January 2003. The last threehave occurred at decadal intervals, reflecting one ofthe major climate cycles outlined in Chapter 2.

The Hobart, Tasmania, fires of 1967 were precededby a season of above-average rainfall and heavy growthof grasses. As drought set in and this vegetation driedout, numerous controlled burns were lit to diminish

the fire risk; however, by 7 February, 81 fires had beenleft burning around Hobart under the misconceptionthat rain would extinguish them (Figure 7.8). On thatday, temperatures reached 39°C with humidity as lowas 18 per cent, conditions not witnessed in a century.The fires were driven by north-west winds in excess of100 km hr-1 generated by the pressure pattern shown inFigure 7.3. By noon, fires had raced through the outerdistricts of Hobart and were entering the suburbs.By the time the fires were controlled, 67 people weredead, 1400 homes and 1000 farm buildings destroyed,and 50 000 sheep killed. Insurance claims totalled$A15 million. The fires in Hobart clearly represent anatural hazard exacerbated by human developmentand attitudes. They also signify a major shift – towardsurban conflagrations – in the nature of bushfires in thiscountry. Urban settlements in Hobart had expandedinto hilly, picturesque bushland under the misconcep-tion that metropolitan firefighting and water serviceswould be available to counteract any increased fire risk.The unattended spot fires burning around Hobartbefore 7 February stretched these resources to thelimit, scattering equipment outside the city and


N

0 10 20 30 40 50 km

Hobart

Source fires

Burnt areas

Wind direction

Fig. 7.8 Extent of bushfires and location of spot fires prior toHobart, Tasmania, bushfires, 7 February 1967 (adaptedfrom Cheney, 1979).

reducing the efficacy of communication. On the dayof the disaster, many communities had to rely upontheir own resources to cope with the blazes becauseprofessional assistance was unavailable or non-contactable.

The Ash Wednesday fires of 16 February 1983caused substantial death and injury. The 1982–1983 ElNiño–Southern Oscillation event produced the worstdrought in Australia to that point in time. Bushlandand grasslands were desiccated. While the weatherconditions of 16 February had occurred often through-out the summer, temperature on the day rose above40° C accompanied by winds gusting over 60 km hr-l.Weather forecasts accurately predicted one of thehottest and driest days of the summer. Hot, dry airflowed strongly from the interior of the continent,reinforced by the subtropical jet stream in the upperatmosphere. A strong cool change was forecast tocross the south of the continent in the afternoon andearly evening, bringing some relief to the desiccatingconditions. For weeks, total fire bans had beenenforced and broadcast hourly over much of south-eastern Australia. Bushfire organizations and volunteerfirefighters in South Australia, Victoria, and New SouthWales were put on alert. Every human resource wasavailable and ready.

There was little anyone could do to stop the fires. Bynoon, much of Adelaide was ringed by fires and at3:30 pm the cold change swept through. The changedropped temperatures 10°C, but the winds increasedand drove the fires to the north-east. In Victoria, firesbroke out around Mt Macedon to the north ofMelbourne and in the Dandenong Ranges to the east,with hot, dry north-westerly winds gusting up to70 km hr-1. Other fires developed in rural scrubland inwestern Victoria and in the coastal resort strip west ofGeelong. Firefighters speedily brought under controlall but eight of the 93 fires that broke out. The coolchange, rather than bringing relief as expected, turnedthe fires into raging infernos. Fire fronts were shifted60–90 degrees and pushed forward by winds averaging70 km hr-1 and gusting at times to 100–170 km hr-1. Theconflagrations that swept the Adelaide and Melbourneareas were unprecedented. Only three to four timesper century could such a combination of conditions beexpected anywhere in Australia. Even under idealfuel conditions, it is physically impossible to generatefire intensities in the Australian bush greater than 60 000–100 000 kW m-1 (because of their greater

biomass, boreal forests can have values up to2 500 000 kW m-1). On Ash Wednesday, thousands ofhouses, farms, and towns were subjected to themaximum limit of these intensities as fires bore downon them at velocities close to 20 km hr-1. Aluminumtyre rims and house sidings melted into puddles as airtemperatures reached 660°C. The intense heat gener-ated thermal tornadoes that snapped, at ground level,the trunks of whole stands of trees. Trees in front of thefire exploded first into flames because of the intenseradiation and were then consumed by the maelstrom.Such conditions had not been witnessed since thefirestorms that swept through the cities of Hamburgand Tokyo as the result of fire bombing in the SecondWorld War. People fleeing in cars were outrun andincinerated. Hundreds of people survived only byquick thinking and luck. At McMahons Creek east ofMelbourne, 85 people escaped the inferno by shelter-ing 30 m underground in the 1 m spacing betweenwater pipes servicing a dam. At the town of Cockatoo,in the Dandenongs, over 150 townspeople survived inthe local kindergarten, in the middle of a playground astheir town burnt down around them. Along the Otwaycoast, people fled to the safety of the local beaches, andthen into the ocean as the sand became too hot to bearor wind-blasted them.

Seventy-six people died; 3500 were injured, manywith burns that would cripple them for the rest of theirlives; over 300 000 sheep and 18 000 cattle were killed;over 500 000 hectares of urban forest and pasture burntin a single day; 300 000 km of fencing was destroyed;and 1700 homes and buildings were consumed(Figure 7.9). In South Australia alone, 40 per cent of thecommercial pine forest was burnt. Insurance claimsreached $A200 million, with total property lossesexceeding $A500 million. Channel 7’s national televisionnews simply ended its evening program by silentlylisting the 20 or more towns that had ceased to existthat day.

The Ash Wednesday bushfires were a rareoccurrence in Australia. No modern technology, noprescribed burning, no preparation by home-owners ormunicipal councils, no improvement in prediction, nonumber of firefighters or amount of equipment couldhave stopped those fires from occurring nor negatedtheir effects. The weather conditions leading up to AshWednesday were matched only by those preceding the1939 bushfires. However, the fires exemplify the factthat there is a weather pattern that tends to occur prior

151Fires in Nature

to extreme bushfires in south-eastern Australia – andthis pattern should be researched and predicted.Warnings of this particular hazard and research intothe design of buildings and the structuring of commu-nities to prevent fire intrusion must be undertaken tominimize the threat. Issues relating to prescribedburning practices, and pitting conservationists againstfire authorities in attempts to prevent another AshWednesday, must be resolved. The Ash Wednesdaybushfires of 1983 were the worst natural disasterin Australia up to that time. Research, planning, andpreparation for major bushfires must be seen in thiscountry as the prime challenge for natural hazardmitigation. These activities, coupled with an effectiveeducation program on bushfire survival techniques(staying inside a house or car until the fire passes overis the safest behavior), offer hope of reducing the deathtoll due to this hazard in Australia.

The Sydney 1994 and Canberra 2003 fires weremainly urban events. The lessons learnt during theHobart fires of 1967 were practiced during the Sydneyfires but ignored in the Canberra fire of 18 January2003. The Sydney fires were part of a larger firedisaster affecting 600 km of the eastern seaboard ofNew South Wales. Drought was a major factor aseastern Australia slipped into the 1990–1995 El Niño-Southern Oscillation event. With the advent of firesin early summer, fire authorities knew that climateconditions were deteriorating rapidly and that worsewas to come. For the first time ever, the fires wouldnot stop at the edge of Sydney. Rather, they would

penetrate to the heart of the city along green beltsand through beautified suburbs planted during the1960s and ’70s with highly flammable native trees andshrubs. Following Bureau of Meteorology forecastsfor desiccating conditions in early January, and as thenumber of fires around the city grew from 40 to148 between 3 and 8 January, authorities plannedaccordingly, setting in place a military-style operationto save lives and property. Convoys of urban andrural firefighting equipment totaling 1700 unitswere summoned from as far away as Melbourne andAdelaide, distances of over 1000 km. Firefighters wereflown in from across the continent and from NewZealand. At the peak of the threat, on ‘Hell Fire’Saturday, 8 January, 70 000 firefighters – three-quartersof whom were volunteers – were battling blazes thatwould consume 800 000 hectares of bush. On that day,air temperatures climbed to 40°C and humiditydropped to 4 per cent accompanied by winds of70 km hr-1 blowing out of the continent’s hot interior.The firefighters were supported by 26 000 people pro-viding food, accommodation and logistical support; andassisted by 16 fixed-wing flying water bombers and80 helicopters including three Erickson aircranes –Elvis, The Incredible Hulk and Georgia Peach. Sewagetankers, petrol tankers, and concrete mixers assisted incarting water. Taxis, ambulances, chartered buses, andferries were used to evacuate some of the 20 000people fleeing the fires. The scale of the responsesurpassed anything witnessed during urban wildfires inthe United States – including the 1991 Oakland fires.The effort paid off. Over a three-week period, onlyfour people died and just 170 homes, valued at$US50 million, were lost. A grateful Sydney held aticker tape parade in honor of everyone who had pre-vented a much larger disaster – that at the beginning ofthe fires was predicted to destroy 5000 homes.

In contrast to the Sydney fires, the Canberra fire of18 January 2003 was not well-managed. Again droughtwas a prominent factor, but this time its severity wasthe worst in recorded history. Fires in the previous yearhad burnt more area in the surrounding state ofNew South Wales – 6000 km2 – than in any other yearincluding the 1993–1994 fire season. Lightning in theweek preceding 18 January had sparked numerousfires in the rugged Brindabella Ranges 50 km west ofthe city. As the high-pressure pattern responsible formost catastrophic fires in eastern Australia developed(Figure 7.3), hot air from the arid interior of the


Fig. 7.9 Devastation to the community of Fairhaven along theOtway coastline of Victoria following the Ash Wednesdaybushfires, 16 February 1983. Note similarity of devastationto that in Figure 7.6 (photograph © and reproducedcourtesy of the The Age newspaper, Melbourne).

continent was drawn over the city and humidityplunged due to adiabatic heating. By 2:00 pm temper-atures had reached 40°C, humidity had dropped toless than 4 per cent – reaching 0 per cent at one station– and, ominously, wind speeds began to increase to130 km hr-1. Within half-an-hour, fires had raced out ofthe mountains, through plantations of Pinus radiata(these trees have higher biomass than eucalypts) andinto the outer suburbs of Canberra. As in Hobart35 years before, firefighters had been distracted byfires burning on the outskirts of the city. When the firesreached Canberra, 1000 firefighters, 14 aircraft, 119fire trucks, and 54 water tankers were fighting threeblazes outside the city. Only 12 pieces of equipmentwere available within the urban area to cope with thefires. The unprecedented firestorm that developedsnapped off, like a lawnmower cutting grass, trees thesize of telephone poles 1 m above the ground, flungvehicles through the air, ripped doors off fire trucks,deroofed houses, hurled bits of roofing tile through carwindows like bullets, and carried pieces of wood aslarge as compact discs more than 12 km away. Theground trembled and the sky was filled with a soundsimilar to a fleet of 747s taking off. Fireballs slammedinto houses, palls of dense smoke spontaneously burstinto flame and, as if by magic, houses suddenly ignitedone by one like matches being lit against a reddenedsky. Daylight was supplanted by darkness, eerily lit bythe feeble headlights of fleeing cars and sprays ofracing embers. The helicopters, including an Ericksonaircrane, which had so successfully contained urbanfires in eastern Australia over the previous decade,were helplessly grounded by the lack of visibility.Communications between the fire command centerand field units broke down in chaos. Fire hydrantsfailed, gas meters ignited torching adjacent houses,and exasperated firefighters broke down in tears. Thefires raged for 12 hours, killing four people, destroying431 homes, and causing $250 million damage includingdestruction of the internationally renowned MtStromlo astronomical observatory. It was an urban firethat should never have been allowed to happen andshattered any confidence that Australia had learnedfrom the lessons of previous fires.

CONCLUDING COMMENTS

Of all natural hazards occurring in Australia, bushfiresnow have the potential to be the major cause ofproperty damage and loss of life. The Ash Wednesday

fires of 1983 took more lives and destroyed moreproperty in scattered semi-rural communities thanCyclone Tracy did in moving through the center ofDarwin. Dense settlement, which is now occurring inrugged bushland on the outskirts of major Australiancities, is increasing the probability of another majorbushfire disaster like Ash Wednesday or the 2003Canberra fire. All too often, images of horrendousconflagration fade within a few years from people’sminds. A drive through the outskirts of Melbourne orthe Adelaide Hills will show that houses, destroyed in1983 because they were not fire-resistant, have beenreconstructed exactly as they were before the fire. InNew South Wales, at least 10–20 homes each yearover the last 20 years have been destroyed by bush-fires around urban centers. In the northern suburbsof Sydney, homes with wooden decks and exteriorshave been built up the inaccessible slopes of steepgullies.

Throughout this book, the 1:100-year event isemphasized as an event that has not yet occurred inmost of Australia. Overseas, in countries settled forlonger, structures designed to withstand that sameevent have been shown to be inadequate overperiods longer than 100 years. If 1:100-year bushfiremaps were constructed for settled areas in southernAustralia, many newer suburbs, as well as some ofthe older inner suburbs, would obviously lie withinbushland that has undergone recurrent burning.

Oddly, just when America is abandoning pre-scribed burning in national parks, Australia isadopting it. This is the reverse of the pattern twentyyears ago. The debate whether to burn or not will, inthe end, be resolved by defining which policy bestmitigates loss of life. However, prescribed burningcan also kill. In the middle of winter on 8 June2000, four volunteer firefighters died in a Sydneycontrolled burn that got out of control. Similarly,another controlled burn too close to the summerfire season, in the Goobang National Park, burnt 14 000 hectares of farmland killing 5000 sheep anddestroying 140 km of fencing. One of the reasonsthat prescribed burning has been reintroduced inAustralia is that, despite the world’s best-developedvolunteer firefighting organization, urban conflagra-tions are increasing. This volunteer service will beseverely tested and ultimately fail within the next20 years as the urban bushfire threat continues toescalate in south-eastern Australia.

153Fires in Nature

Nor is this threatening situation confined toAustralia. The same is occurring in southernCalifornia, which has similar fire weather and vege-tation but more rapid urban expansion. What is noteasily recognized is the degree to which urbansettlement has extended into forest areas in all partsof the United States. Since the first major urbanforest fire in 1947, urbanization has continued at arapid pace, and forests have completely reclaimedlarge tracts of abandoned farmland near settlementsand cities. Each urban wildfire takes on a distinctidentity – for example the Black Tiger Fire of July1989 near Boulder, Colorado, or the Stephan BridgeRoad Fire of May 1990 in Crawford County,Michigan. Since the mid-1950s, this urbanization hastaken on a different aspect as millions of urbandwellers have chosen to leave major cities duringsummer for cottages in nearby woodlands. Thisoccurs in the north-eastern Appalachians, northernMichigan, and Wisconsin. In Canada, up to 50 percent of the population of Toronto and Montrealescape the summer’s heat each weekend by travelinginto the Canadian Shield forest. Historically, theseforests have been subjected to infernos. Given thepropensity of humans to play with fire, the regen-eration of forests and the tendency for warmersummers, these woodlands will undoubtedly witnessthe recurrence of large catastrophic fires similar tothose that affected the Great Lakes region in the latenineteenth century.


Cheney, N.P. 1979. Bushfire disasters in Australia 1945–1975. InHeathcote, R.L. and Thom, B.G. (eds) Natural Hazards inAustralia. Australian Academy of Science, Canberra, pp. 72–93.



Firewise 1992. The Oakland/Berkeley Hills fire. <http://www.firewise.org/pubs/theOaklandBerkeleyHillsFire/>

Luke, R.H. and McArthur, A.G. 1978. Bushfires in Australia. AGPS,Canberra.

Oliver, J., Britton, N.R. and James, M.K. 1984. The Ash Wednesdaybushfires in Victoria, 16 February 1983. James Cook Universityof North Queensland Centre for Disaster Studies, DisasterInvestigation Report No.7.

Powell, F.A. 1983. Bushfire weather. Weatherwise 36(3): 126.Pyne, S.J. 1982. Fire in America: A Cultural History of Wildland

and Rural Fire. Princeton University Press, Princeton.Pyne, S.J. 1991. Burning Bush: A Fire History of Australia. Allen &

Unwin, Sydney.Romme, W.H. and Despain, D.G. 1989. The Yellowstone fires.

Scientific American 261(5): 21–29.Seitz, R. 1986. Siberian fire as ‘nuclear winter’ guide. Nature 323:

116–117.ThinkQuest Team 2001. 2000 fire season. <http://library.thinkquest.

org/C0119184/english_text/historical_fires_2000_fire_season.shtml>

van Nao, T. (ed.) 1982. Forest Fire Prevention and Control.Martinus Nijhoff, The Hague.

Vines, R.G. 1974. Weather Patterns and Bushfire Cycles inSouthern Australia. CSIR0 Division of Chemical Technology,Technical Paper No. 2.

Voice, M.E. and Gauntlett, F.J. 1984. The 1983 Ash Wednesdayfires in Australia. Monthly Weather Review 112(3): 584–590.

Webster, J.K. 1986. The Complete Australian Bushfire Book.Nelson, Melbourne.

Yool, S.R., Eckhardt, D.W., Estes, J.E. and Cosentino, M.J. 1985.Describing the bushfire hazard in southern California. Annals ofthe Association of American Geographers 75(3): 417–430.


INTRODUCTION

In Chapter 3, waves were often mentioned as one ofthe main agents of erosion and destruction by storms.In fact, destructive waves can occur without stormevents, and in association with other climatic andoceanographic factors such as heavy rainfall and highsea levels, to produce coastal erosion. In this sense,waves form part of a group of interlinked hazards asso-ciated with oceans. This chapter examines these andother oceanographic hazards. Wave mechanics and theprocess of wave generation will be outlined firstfollowed by a description of wave height distributionworldwide. This section will conclude with an appraisalof the hazards posed by waves in the open ocean.

Wind is the prime mechanism for generating thedestructive energy in waves. In cold oceans, seasor lakes, strong winds can produce another hazard –the beaching of drifting sea-ice. Sea-ice is generallyadvantageous along a coastline that experiences winterstorms, because broken sea-ice can completelydampen the height of all but the highest waves.When frozen to the shoreline as shorefast ice, sea-icecan completely protect the shoreline from any stormerosion. However, floating ice is easily moved by windsof very low velocity. These winds might not generateappreciable waves, but they can drive sea-ice ashoreand hundreds of metres inland. The force exerted bythis ice can destroy almost all structures typically found

adjacent to a shoreline. For this reason, sea-ice mustbe considered a significant hazard in any water bodyundergoing periodic freezing and melting. Thischapter will describe the problem of drifting sea-ice,its worldwide distribution, and the measures taken tonegate its effects.

It was pointed out in Chapter 3 that waves might notbe important as a hazard if the full intensity of a stormreaches landfall during low tide. Such was the caseduring Cyclone Tracy in Darwin, Australia, in 1974(see Figure 8.1 for the location of all major placenamesmentioned in this chapter). Even with a 4 m stormsurge, waves did not reach the high tide mark, becausethe storm made landfall during low tide in an environ-ment where the tidal range is about 7 m. More insidi-ously, low waves can cause considerable damage alonga shoreline if sea levels are either briefly elevatedabove normal levels, or rising in the long term. It isgenerally believed that sea level is presently risingworldwide (eustatic rise) at a rate of 1.0–1.5 mm yr-1.The most likely cause of this rise is the Greenhouseeffect. Human-produced increases in CH4, CO2 andother gases are responsible for a general warming ofthe Earth’s atmosphere. This causes sea level to risebecause of thermal expansion of ocean waters. Notethat this cause is more significant than melting of theworld’s icecaps. Even if recent warming is not due tothe Greenhouse effect, the implications are the same.The recent and threatened rise in sea level poses a

Oceanic HazardsC H A P T E R 8

natural hazard to many coastal areas, especially wheremajor world cities exist close to the limits of presentsea level. The rise in sea level is also believed to beresponsible for the accelerating trend in sandy beacherosion worldwide. Sea level behaviour and beacherosion are more complex than outlined in the abovescenarios. This chapter will describe worldwide sealevel trends, and examine the causes of sea levelfluctuations. The chapter will conclude with an evalua-tion of the factors responsible for beach or generalcoastal erosion. The processes whereby rising sea levelcan induce beach erosion will be described, followedby an examination of other causes, such as increasingrainfall and storminess on the east coasts of NorthAmerica and Australia. The relative importance ofthese causes will be evaluated using as an example partof a data set compiled for Stanwell Park Beach,Australia.

WAVES AS A HAZARD

Theory(Wiegel, 1964; Beer, 1983; Dalrymple, 2000)

The basic form and terminology of a linear wave isdescribed schematically in Figure 8.2. This diagramhas been used to characterize simply general wavephenomena, including ocean waves, tides, storm surges,seiching, and tsunami. The waveform in Figure 8.2 hasthe following simple relationships in deep water:

L = CT (8.1)C = gT(2�)-1 (8.2)

then L = 1.56T2 in meters (8.3)where L = wavelength

C = wave velocityT = wave periodg = acceleration due to gravity

� = 3.1416

Wave motion is transmitted through the water columnwith water particles circumscribing elliptical paths thatdecrease exponentially in height and width with depth.If there is no measurable oscillation of water particlesbefore the seabed is reached (point A in Figure 8.2),then the wave is considered to be in deep water. Ifoscillations are still present when the seabed is reached(point B in Figure 8.2), then the wave is in shallowwater and affected by the seabed. The height andenergy of a shallow-water wave are determined, notsolely by waveform, but also by the nature of inshoretopography. Under these conditions wave height canbe increased by shoaling, by wave refraction – whereoffshore bathymetry has focused wave energy onto aparticular section of coastline such as a headland – orby wave diffraction around headlands, tidal inlets, andharbor entrances. Figure 8.3 exemplifies each of theseconditions, which are controlled in magnitude by waveheight, period, and water depth. Generally, the rate atwhich energy moves with the wave – the energy flux –remains virtually constant as a wave travels through


Venice

Crescent CitySanta Barbara

Southern California

Lake ErieNew York

Florida

Chesapeake Bay

Gulf of MexicoTexas

New Jersey

London

New South Wales

Stanwell Park BeachSydney

Moruya Beach

Cornwall

NorthSea

Scolt Head Is.

DungenessOrford Ness

Stockholm

Caspian Sea

WajimaTokyo

Darwin

Tweed R.

South Australia

Scarborough Beach

Agulhas Plateau

Dead Sea

Assam Dam ArabianSea

QattaraDepression

South CarolinaCape Hatteras

Gulf of St. Lawrence

Denmark


increasingly shallow water towards shore. However,the speed at which the waveform is traveling decreaseswith decreasing water depth. To maintain a constantenergy flux as a wave shoals into shallower water, waveheight must increase. The fact that wave velocity slowsas water depth decreases also determines wave re-fraction. If one part of a wave crest reaches shallowerwater before another part, then the wave crest bendstowards the shallower water segment. The process isanalogous to driving one wheel of a vehicle off a pavedor sealed road onto a gravel verge or shoulder. Thewheel that hits the gravel first slows down dramatically,while the wheel still on the road continues to travel atthe original road speed. The vehicle as a result veerstowards the slower rotating wheel and careers off theroad. For waves, the effect produces a bending ofthe wave crest towards shallow water, an effect that isparticularly prominent around headlands or reefs thatprotrude out into the ocean.

Diffraction is the process whereby energy moves lat-erally along a wave crest. The effect is noticeable whena wave enters a harbor between narrow breakwalls.Once the wave is inside the wider harbor, the crestwidth is no longer restricted by the distance betweenthe breakwalls, but gradually increases as energy freelymoves sideways along the wave crest. In some circum-stances, wave height can increase sideways at theexpense of the original wave height. Such an effectoften accounts for the fact that beaches tucked insideharbor entrances and unaffected by wave refractionexperience considerable wave heights and destructionduring some storms.

The destructiveness of a wave is a function of thewave height and period. Height determines waveenergy, the temporary elevation of sea level, the degreeof overwashing of structures, the height of water

oscillations in the surf zone, and the distance of waverun-up on a beach. Because most waves affecting acoastline vary in period and height over a short periodof time, oceanographers talk about a range of waveheights included in wave spectra. For engineeringpurposes, the highest one-third or one-tenth of waves,called the ‘significant wave height’, is used. Waveperiod determines wave energy, power, and shape inshallow water. The latter aspect determines whetherwave-generated currents at the bed move seaward orshoreward. If seaward, then the beach erodes. Thefollowing equations define wave energy and power fora single wave in deep ocean water:

E = 0.125�gH2L (8.4)P = CgE (CT)-1 (8.5)

where E = wave energy P = wave powerH = wave height L = wavelength

Cg = the group velocity� = density of water

Group velocity is the speed at which a group of wavesmoves forward. In deep water, it is exactly 0.5 times thewave velocity such that individual waves tend to moveforward through a set of waves and disappear at thefront. In very shallow water, the group velocity is equalto the wave velocity. There is a significant differencebetween wave energy and power. Wave energy is thecapacity of a wave to do work. Power is the amount ofwork per unit time. The difference can be explainedsimply by comparing a greyhound to an elephant. Agreyhound is light and when it rubs up against you, youhardly feel it. An elephant is bulky and when it rubs upagainst you, you are pushed aside. The elephant,

157Oceanic Hazards

B ACrest

Wavelength, L

Trough

WaterDepth, dDeep

waterShallow water

Seabed

Still waterlevel Trough

Wave height, H

Fig. 8.2 Schematic representation of wave characteristics.

because of its larger mass, exerts more energy againstyou than the greyhound. A greyhound can move fastand, when it hits you, you are bowled over. An elephantmoves slowly and, when it hits you, you move easily outof its way. Here, the greyhound has more power thanan elephant (unless of course the elephant decides tocharge). Power as defined here applies to an infinitetrain of waves.

Wavelength is much greater than wave height, sothat most of a wave’s energy and power is controlled bywavelength, which is related to the wave period.However, a change in wave height has a much moredramatic affect on changing wave energy and powerthan a change in wavelength. Table 8.1 illustrates thiseffect for the most common wave periods in swellenvironments. (Note that these values would be


10 sec

ocean

wav

e cr

est

harbor

C) Wave diffraction

B) Wave refraction

A) Shoaling

0 sec wave crest20 sec

30 sec

40 sec

50 sec

60 sec

70 sec

60 m

50 m

40 m

30 m

20 m

10 m

0 m

meansea level

continental shelf

C1 > C2 > C3

d1 >> d2 >> d3

wave height increases shorewardand energy flux is therefore conserved

d1 d2 d3

C1 C2 C3speed

depth

Fig. 8.3 Schematic diagrams of wave behavior in shallow water, A) shoaling, B) refraction, and C) diffraction.

reduced 2.4 per cent in fresh water because of its lowerdensity.) A doubling in wavelength leads to a doublingin wave energy and power, whereas a doubling in waveheight leads to a quadrupling in wave energy andpower. In the coastal zone under strong winds, waveheight can change more rapidly than wavelength. Theamount of wave energy in a storm wave can becomeabnormally large (Figure 8.4). In swell environments,wave heights of 1.0 m and periods of 10 seconds arecommon; however, if a storm wave increases in heightto 7 m – a value typical of the storm waves off theeastern coasts of the United States or Australia – thenwave energy will increase almost fiftyfold. This is thereason why storm waves can be so damaging in such ashort period of time.

World distribution of high waves(PO.DAAC, 2003)

The generation of waves represents transference ofenergy from wind to the water body. The height towhich a wave will grow depends upon four factors:wind speed, the duration of the wind, the length ofwater over which wind blows – termed fetch – and theprocesses leading to decay. Generally, wave height isdependent first upon fetch, second upon wind speedand last upon duration. Before the advent of wave-rider buoys to measure waves along a coastline, oraltimeters in satellites to measure wave period andheights over large areas, data on wave characteristicsdepended upon ship observations and hindcastingprocedures. Hindcasting theoretically estimates waveheight and period based on wind duration, speed, andfetch derived from synoptic pressure maps. Excepton small bodies of water or near shore where the

resolution of imaging is too low, satellites haveeffectively eliminated the necessity for hindcasting todetermine wave height. Altimeters in satellites nowmeasure wave heights with an accuracy of 2.5–4.0 cm.Instruments were first used on SEASAT in 1978, thenon GEOSAT from 1985 to 1988, and ERS-1 and 2 from1991, TOPEX/POSEIDON from 1992, and Jason-1from 2001. CNES, the French space agency, andNASA, the United States space agency, jointly operatethe latter two satellites. Waves measured using satellitealtimeters range from 1 to 5 m at daily and monthlylevels. Figure 8.4 illustrates these patterns for Januaryand July 1995, which are typical of these months.Maximum wave heights correspond to belts of maxi-mum winds generated over oceans by mid-latitudelow-pressure cells (Figure 2.3). Wave climate in theextra-tropical north Pacific and north Atlantic isespecially seasonal, with much rougher conditions inthe northern (boreal) winter than in the summer. TheSouthern Ocean has high waves throughout the year,but is roughest in the southern (austral) winter. Nearthe equator, larger waves are produced either by swellpropagating from higher latitudes, or by seasonal winds(for example, monsoons and tropical storms). TheArabian Sea is particularly rough from June to August,coinciding with the south-west monsoon.

Waves as a hazard at sea(Beer, 1983; Bruun, 1985; Kushnir et al., 1997; Zebrowski,1997; Lawson, 2001)

The effect of waves as a hazard really depends uponwhere the wave is experienced. In the open ocean,freak, giant, or rogue waves pose a hazard to naviga-tion. This does not involve a single wave. Instead, it

159Oceanic Hazards

Energy m-1 of wave Power m-1

crest in one wave of wave crest Power relative Wave period (s) Length (m) Height (m) (103 joules) (103 watts) to 1 m high wave

10.0 156 1.0 0.196 9.79 1.01.4 0.384 19.20 2.02.0 0.784 39.18 4.03.0 1.763 88.15 9.07.0 9.598 479.92 49.0

14.0 306 1.0 0.384 13.71 1.02.0 1.536 54.85 4.03.0 3.455 123.41 9.0

Table 8.1 Wave energy and power for typical ocean waves (based on Wiegel, 1964; Dalrymple, 2000).

involves the meeting of several waves of differentheight, period, and direction that exist in the spectrumof waves generated by winds over the ocean. Forexample, the energy per meter of wave crest in a 4 mwave with a period of 10 seconds in deep water is31.4 � 105 joules (Equation 8.4). The energy of a 3 mwave of the same period is 17.6 � 105 joules. If the twowaves intersect then the summation of energy is49 � 105 joules. Using Equation 8.4, the resulting waveheight is 5 m. This wave height is unique in time andspace. If a ship happens to be at the point of confluenceat that time, then it is under threat of capsizing. Suchwaves have always plagued shipping, but today they alsopose a potentially serious hazard to oil platforms, withpotential ecological damage and investment loss.

Sometimes waves will increase in height when theymeet an opposing current. The region along Africa’ssouth-eastern coast is notorious for high waves dueto current interaction. Here, waves generated in theroaring forties travel against the Agulhas current movingsouth, trapped between the submerged Agulhas Plateauand the continental shelf. Waves steepen in this regionas a result. One confluence of wave crests steepenedby this current swamped and sank a supertanker, theGigantic, in the late 1970s. There are other regions ofthe world – including the east coast of New South Waleswhere storm waves often move into the East AustralianCurrent – that are as dangerous.

Algebraic summation of interacting waves does notexplain the incidence of rogue waves currently being


Ice-coveredIce-covered

Ice-covered

A January

1 5Significant wave height

2 3 40 6 m

B July

Ice-coveredIce-covered

Ice-covered

Fig. 8.4 Worldwide distribution of altimeter wave heights measured by the TOPEX/POSEIDON satellite A) January 1995, B) July 1995 (Jet PropulsionLaboratory, 1995a, b).

measured. In the North Sea, the overall height ofwaves is increasing, with freak waves in excess of 17 mbeing recorded around many platforms. In the Gormfield, five waves in excess of 8 m were recorded in a 13-hour period on 6 September 1983. All of thesewaves are the result of chaos described best by thenon-linear Schrödinger equation. This formula canpredict waves four times the average wave height.Non-linear interaction can also produce waves of thesame size, often in sets of three to five monstrouswaves. Perhaps the largest recorded wave occurred on7 February 1933 when the US Ramapo struck a stormwith winds as high as 11 on the Beaufort scale. After aweek of constant storm conditions, the ship leaned intothe trough of a wave and was met by a wall of water34.1 m high. Other waves of significance were the 26 mwave that hit Statoil’s Draupner gas platform in theNorth Sea on New Year’s Day 1995, the 29 m wave thatrocked the QE2 crossing the North Atlantic in 1995,and the 30 m wave that almost downed a helicopterduring the Sydney Hobart Yacht Race on 27 December1998. Such waves have even toppled oil platforms, aswas the case with the Ocean Ranger drilling rig on theGrand Banks of Newfoundland on 15 February 1982.Since 1969, monstrous waves as high as 36 m haveprobably been responsible for the disappearance ofmore than 200 supertankers.

Regional wave heights can also increase substan-tially over time. For instance, waves in the northAtlantic Ocean have increased in height by 20 per centsince 1980. Largest wave heights offshore fromCornwall have risen from 11.9 m in the 1960s to 17.4 min the late 1980s, while mean wave heights haveincreased from 2.2 m to 2.7 m over this period. Theseheights represent a 32 per cent increase in waveenergy, an increase that may not have been accountedfor in the engineering design of shoreline or offshorestructures. Intensification of the north Atlantic windfield has amplified these wave heights.

Waves as a hazard on rocky coasts

Simple wave theory can also explain how rough seascan sweep people unexpectedly off rocks. Presentlyone person dies each week in Australia as the result ofbeing swept off rock platforms. This is currently themost common natural cause of death in Australia.While tourists – especially those with little knowledgeof a coastline – appear at risk, the majority ofdeaths involve anglers. Fishing is the most popular

recreational activity in Australia and many anglers viewrock ledges and platforms as the best spots to fish.

Four factors exacerbate the hazard. First, if peoplehave walked along a beach to get to a headland, theycan be lulled into a false sense of security because,due to wave refraction, wave heights can be up to50 per cent less along the beach than off the headland(Figure 8.3). Second, wave heights are diminished5–20 per cent at the coast because of frictional interac-tion with the seabed. Along rocky coasts, especially inNew South Wales where the continental shelf has awidth as narrow as 12 km, the steep nature of theoffshore contours minimizes this frictional loss to lessthan 5 per cent. Third, the term ‘freak wave’ typicallyrelates to the interaction of wave spectra rather than toa single large wave. Waves rarely arrive along a coast-line with a constant wave period, height, or direction.It is quite common, for example, for wave sets alongthe New South Wales coast to come from two differentdirections. In embayments, any directional variation inwave approach to the shoreline tends to be minimizedbecause of wave refraction. On headlands, wave refrac-tion is not efficient because headlands protrudeseaward from the coast and offshore topography issteep. Here, the directional variation in wave approachis greatest and the potential for interaction betweenwave crests is at a maximum. Thus, wave crests fromtwo different directions can often intersect, resulting inexaggerated wave heights due to the summation of theenergy in the two waves. This combined wave will havea much larger run-up height along rocks than thatproduced by waves individually. Finally, waves along abeach will often break at some distance from shore anddissipate their energy across a surf zone. While set-upin the surf zone and run-up onto the beach can signif-icantly and temporarily raise water levels at shore,most of the energy has been lost in the surf zone.Additionally, wave reflection back from the shoreline isminimal because beaches tend to be flat. Waves onheadlands, however, break by surging at the shoreline.Energy loss occurs over the rocks rather than offshorein a wide surf zone. As well, the steep nature ofheadlands results in considerable reflection of energyseaward, a process that can temporarily raise sealevels several metres from shore. It is possible forthe next incoming wave to be superimposed upon thissupra-elevated water. This process can generate freakwaves that can override rock platforms even when thetide is low (Figure 8.5).

161Oceanic Hazards

The crucial parameters in the generation of freakwaves are high seas and variable wave conditions.These conditions may be associated with periods ofstormier or windier weather. If such conditions havenot occurred for some period, then people’s awarenessof wave behavior – especially along rocky coasts underhigh seas – may be poor. This ignorance is the mainreason for a significant loss of life due to people beingswept off rocks by waves in Australia.

SEA-ICE AS A HAZARD

Ice in the ocean(Defant, 1961; Gross, 1972; Vinje, 2001; Anisimov &Fitzharris, 2001; Long et al., 2002)

Sea-ice has two origins: through the freezing ofseawater and by the discharge of ice from glaciers intothe ocean. The former mechanism is the morecommon method of forming sea-ice. Because theocean has a salinity of 31–35 ‰ (parts per thousand),it freezes at temperatures around –1.9°C. New sea-icedoes not contain pure water but has a salinity of5–15 ‰. The faster ice forms, the more saline it islikely to become. As ice ages, the brine drains away,and the ice becomes salt-free and clear. These charac-teristics can be used to identify the age and origin ofsea-ice. At –30°C, sea-ice can form at the rate of10 cm day-1. Typically in the Canadian Arctic, a winter’sfreezing cycle will produce ice 2–3 m thick. At thispoint, ice insulates the underlying water and freezing issubstantially reduced. The process is similar for thefreezing of ice in lakes and rivers; but because of lowsalinities, ice forms faster in these latter environments.The density of seawater is 1027 kg m-3. With an air

content of 1 per cent, 89 per cent of floating ice issubmerged. If the air content rises to 10 per cent, as itcan do with some icebergs and shelf-ice, then thesubmerged part is reduced to 81 per cent by volume.

Sea-ice can be divided into four categories. The firstconsists of the polar shelf- and sea-ice forming a per-manent cover, 3–4 m thick, over 70 per cent of theArctic Ocean and surrounding the Antarctic continent(Figure 8.6). This ice is very hummocky but may crackopen in summer to form leads. In the following winter,leads will be squeezed shut by the shifting ice formingpressure ridges tens of meters high. Sea-ice may alsomelt patchily, especially in the Antarctic, forming openwater areas called polynyas. Since the mid-1970s, themelting of Arctic sea-ice has increased. The areacovered by sea ice has diminished 21 per cent between1955 and 2000. This varies from a high of 6 per centper decade since 1970 in the Atlantic, to 1.5 percent per decade in the Canadian Arctic and Bering Sea(see Figure 8.6 for placenames mentioned in thissection). While the trend has been attributed to globalwarming, melting has been a longer term process.Figure 8.7 illustrates this trend in the Arctic over thepast two centuries using data from Davis Straitbetween Canada and Greenland, the Greenland Seaeast of Greenland, and the Nordic Sea in the northAtlantic. In the latter region, sea-ice extent in April hasdecreased by 33 per cent (0.79 � 106 km2) with a sub-stantial proportion of this reduction occurring beforeindustrial activity accelerated in the middle of thetwentieth century. The dotted lines in Figure 8.7indicate that changes in sea-ice extent are synchronousacross the Arctic. The limit of sea-ice in the northAtlantic retreats when the North Atlantic Oscillation(NAO) is positive, due to an increased northward fluxof warm air. However, off Labrador, the opposite effectis evident. This correlation fits the pattern, associatedwith the NAO, of warmer temperatures in the north-east Atlantic and cooler ones in eastern NorthAmerica. Concomitant with this retreat, sea-icethinned by 0.13 m between 1970 and 1992. In theeastern Siberian Sea, ice thickness decreased from3.1 m in the 1960s to 1.8 m in the 1990s. In somelocations, especially between Greenland and Norway,ice cover has thinned by 40 per cent over the past threedecades. More dramatic has been the disappearance oficeshelves around the Antarctic Peninsula. Fiveiceshelves, amounting to 5000 km2 of iceshelf, havecollapsed due to regional warming of 2.5°C in the last


Fig. 8.5 Deep-water swell waves of 4 m breaking over the breakwallto Wollongong harbor, Australia (photograph courtesyJohn Telford, Fairymeadow, New South Wales).

50 years. The largest collapse was the Larsen A shelf in1995 (3000 km2). This was matched in areal extentagain in 1999. Because this ice is already floating, it hashad no impact on sea level.

The second type of sea-ice is the annual pack-icethat mainly forms within the seas of Japan andOkhotsk, the Bering and Baltic Seas, the Gulf of StLawrence and along the south-east coast of Canada.

163Oceanic Hazards

Limit of Icebergs

Limit of seasonal pack-ice

Permanent shelf-and sea-ice

Ellesmere Is

Beaufort Sea

Bering Sea

Baffin Is

Gulf of St. LawrenceNewfoundland

Davis StraitIceland

NorwaySpitsbergen

Baltic Sea

Novaya ZemlyaSevernaya Zemlya

Sea of Okhotsk

Sea of Japan

AustraliaSouth Africa

Ross Ice Shelf

AntarcticaFilchner Ice Shelf

Topic of Capricorn

Arctic Ocean Zemlya Frantsa-Iosifa

Bermuda

Greenland

Labrador

Greenland Sea

NordicSea

Siberian Sea

Larsen Ice Shelf

Fig. 8.6 Worldwide distribution of permanent polar shelf- and sea-ice, seasonal pack-ice, and icebergs (after Gross, 1972).

1860 1880 1900 1920 1940 1960

0

2

4

-2

-4Rel

ativ

e Ic

e E

xten

t

1980 2000

Nordic Sea (middle)

1820 1840 1860 1880 1900 1920

0

2

4

-2

-4Rel

ativ

e Ic

e E

xten

t Davis Strait

0

2

4

-2

-4Rel

ativ

e Ic

e E

xten

t

1900 1920 1940 1960

Greenland Sea

Fig. 8.7 Composite April sea-ice extent, 1818–1998, compiled from records in Davis Strait (Defant, 1961), Greenland Sea (Skov, 1970), and middleNordic Sea (Vinje, 2001).

Pack-ice also forms in inland lakes where winterfreezing takes place. Pack-ice generally extends moresouthward in the northern hemisphere along thewestern sides of oceans because of prevailing southward-moving cold currents at these locations. Inparticularly severe winters, pack-ice can extendtowards the equator beyond 50° latitude in both thesouthern and northern hemispheres. If the followingsummers are cold, or the ice build-up has been toosevere, then it is possible for large areas of pack-ice notto melt seasonally.

Excessive pack-ice between Greenland and Icelandterminated the Greenland Viking colony because itprevented contact with Europe after the middle of thefourteenth century. Except for the early 1970s, pack-ice limits in the Arctic have retreated poleward in thetwentieth century, concomitant with global warming.Pack-ice becomes a hazard only when it drifts intoshore, or when it becomes so thick that navigable leadsthrough it have a serious chance of closing and formingpressure ridges. Whaling and exploration ships havebeen crushed in this way. In the summer of 1985–1986and 2001–2002, the Australian Antarctic researchprogram was severely hampered by thick pack-ice thatclosed in, stranding supply vessels and crushing andsinking a ship. In the Canadian Arctic, thick pack-icehas historically posed a threat to shipping and increasedthe cost of re-supplying northern settlements. Pack-iceterminated numerous journeys to the Canadian Arcticin the nineteenth century, the most famous of whichwas the loss of the Franklin expedition.

The third type of sea-ice consists of shore-fast ice.This ice grows from shore and anchors to the seabed.It effectively protects a shoreline from wave attackduring winter storms but, upon melting, can removesignificant beach sediment if it floats out to sea. Thistype of ice poses a unique hazard at the shoreline andwill be discussed later.

The fourth category of sea-ice consists of icebergs.Icebergs form a menace to shipping as exemplified bythat most notorious disaster, the sinking in April 1912of the Titanic with the loss of 1503 passengers andcrew. Icebergs in the northern hemisphere originate inthe north Atlantic, as ice calves from glaciers enteringthe ocean along the west coast of Greenland and theeast coast of Baffin Island. Icebergs coming fromthe east coast of Greenland usually drift south,round the southern tip of Greenland, and join this con-centration of ice in the Davis Strait. Along the southern

half of the west coast of Greenland, icebergs collect infjords and discharge into the Strait at fortnightly inter-vals on high tides. Along the northern half of theGreenland coast, glaciers protrude for several kilo-meters into the sea and icebergs calve directly into DavisStrait. About 90 per cent of all icebergs in the northernhemisphere collect as swarms in Davis Strait duringsevere periods of pack-ice, and move south into theLabrador current as the pack-ice breaks up in spring.These icebergs then pass into the Atlantic Ocean, offNewfoundland, with widely varying numbers from yearto year. Fewer than 10 per cent of icebergs originate inthe Barents Sea from glaciers on the islands of Spitsber-gen, Zemlya Frantsa-Iosifa (Franz Josef Land), NovayaZemlya, and Severnaya Zemlya.

Whereas winds determine the direction that pack-ice moves, ocean currents control the movement oficebergs. Because of Coriolis force, ice drifts to theright of the wind in the northern hemisphere and tothe left in the southern hemisphere. Icebergs in thenorth Atlantic usually have heights of tens of meters,although some are up to 80 m in height and over 500 min length. The most southerly recorded sighting in thenorthern hemisphere was near Bermuda at 30.33°N.In the Antarctic, icebergs usually originate as calvedice from the Ross and Filchner iceshelves. Here,glaciers feed into a shallow sea forming an ice bank 35and 90 m above and below sea level, respectively. Theproduction of icebergs in the Antarctic is prolific,probably numbering 50 000 at any one time. Icebergsbreaking off the Ross Ice Shelf may reach lengths of100 km. The largest yet measured was 334 km by96 km with a height above water of 30 m. This onetabular ice block had a volume that exceeded twicethe annual ice discharge from the entire Antarcticcontinent. In 2000, an iceberg – named B15 andmeasuring 295 km by 37 km – broke off the Ross IceShelf. The number of observed icebergs has increasedfivefold; however, technological advances in icebergobservation account for most of this increase. Inaddition, the shelves are undergoing the end of a majorcalving cycle that recurs every 50–100 years. Tabularicebergs have been found as far north as 40°S, with themost northerly recorded sighting being 26.05°S,350 km from the Tropic of Capricorn. This is unusualbecause the Antarctic circumpolar current between50° and 65°S forms a warm barrier to northward icemovement. About 100 smaller tabular icebergs, up to30 km2 in size, also exist in the Arctic Ocean, drifting


mainly in a clockwise direction with the currents in theBeaufort Sea. These ice rafts have calved from fivesmall iceshelves on Ellesmere Island. Because theyhave little chance of escaping south to warmer waters,individual blocks can persist for 50 years or more.

In the southern hemisphere, icebergs pose littlethreat to navigation because they do not frequentlyenter major shipping lanes. However, it is not unknownfor shipping between South America and Australia,and between Australia and South Africa, to be divertednorth because of numerous sightings of icebergsaround 40–50°S. North Atlantic icebergs do enter themajor shipping lanes between North America andEurope, especially off the Grand Banks, east of theNewfoundland coast. In severe years, between 30 and50 large icebergs may drift into this region. Histori-cally, they have resulted in an appallingly high deathtoll. Between 1882 and 1890, icebergs around theGrand Banks sank 14 passenger liners and damaged 40others. The iceberg hazard climaxed on 14 April 1912– a year of a strong El Niño – when the ‘indestructible’,46 000 tonne Titanic scraped an iceberg five times itssize. The encounter sheared the rivets off plates belowthe waterline over two-fifths the length of the ship.So apparently innocuous was the scrape that mostpassengers and crew members did not realize what hadhappened. Within three hours, the Titanic sank dra-matically, bow first. Only a third of the passengerssurvived in lifeboats, while over 1500 people died asthe ship sank. Immediately, the United States Navybegan patrols for icebergs in the area. In the followingyear, 13 nations met in London to establish and fundthe International Ice Patrol, administered by theUnited States Coast Guard. The Coast Guard issuesadvisories twice daily and all ships in the north Atlanticare obliged to notify authorities of the location of anyicebergs sighted. Since the Second World War, regularair patrols have also been flown. With the advent ofsatellites, large icebergs can be spotted on satelliteimages, while the more threatening ones can bemonitored using radio transmitters implanted on theirsurface. Individual ships can also spot icebergs usingvery sophisticated radar systems.

While the iceberg hazard to shipping has all butdisappeared, the hazard to oil exploration and drillingfacilities has increased in recent years, especially in theBeaufort Sea and along the Labrador coastline.Because over 80 per cent of their mass lies below sealevel, large icebergs can gouge the seabed to a depth of

0.5–1.0 m, forming long, linear grooves on the conti-nental shelf. The icebergs could therefore gouge opena pipeline carrying oil or gas across the seabed. Driftingicebergs also have enough mass and momentum tocrumple an oil rig. Not only could there be substantialfinancial loss with the possibility of deaths, but theresulting oil spill could also pose a significant ecologicaldisaster. Attempts have been made to divert icebergsby towing them away using tugboats; however, the largemass of icebergs gives them enormous inertia, and thefact that they are mostly submerged means that theyare easily moved by any ocean current. At present,waves pose the most serious threat to oil and gasactivities in ice-prone waters, but it is only a matter oftime before an iceberg collides with a drilling-rig orplatform. Present disaster mitigation procedures forthis hazard consist simply of evacuating personnel fromthreatened facilities. Such procedures have been usedon several occasions over the last few years.

Ice at shore(Taylor, 1977; Kovacs & Sodhi, 1978; Bruun, 1985)

Nowhere is the destructive effect of ice displayed morethan at the shoreline. During winter, the icefoot, whichis frozen to shore at the waterline, protects the beachfrom storm wave attack. Pack-ice, which develops inthe open ocean or lake, also attenuates the fetch lengthfor wave development. The middle of winter sees thewhole shoreline protected by ice. In spring, the pack-ice melts and breaks up, whereupon it may driftoffshore under the effect of wind. At this time of year,shorelines are most vulnerable to ice damage if thewind veers onshore and drives the pack-ice shoreward.At shore, pack-ice has little problem overriding thecoastline, especially where the latter is flat. Thisstranding ice can pile up to heights of 15 m at shore,overriding wharfs, harbor buildings, and communica-tion links. Ice can actively drive inland distances up to250 m, moving sediment as big as boulders 3–4 m indiameter. On low-lying islands in the Canadian Arctic,shallow pack-ice, 1–2 m thick, can push over the shore-line along tens of kilometers of coastline (Figure 8.8).Large ridges of gravel or boulders – between 0.5 m and7 m in height, 1–30 m in width and 15–200 m in length– can be plowed up at the shoreline. The grounding ofice and subsequent stacking does not require abnor-mally strong winds. Wind velocities measured duringice-push events have rarely exceeded 15 m s-1

(54 km hr-1). Nor is time important: ice-push events

165Oceanic Hazards

have frequently occurred in as little as 15 minutes. Allthat is required is for loosely packed floating ice to bepresent near shore, and for a slight breeze to drive itashore.

Ice pushing can also occur because of thermalexpansion, especially in lakes. With a rapid fall in tem-perature, lake ice can contract and crack dramatically.If lake water enters the cracks, or subsequent thawinginfills the crack, warming of the whole ice surface canbegin to expand the ice and push it slowly shoreward.If intense freezing alternates with thawing, then over-riding of the shore can be significant for lakes as smallas 5 km2. Larger lakes tend not to suffer from iceexpansion because ice will simply override itself,closing cracks that have opened up in the lake.

The main effect of grounded sea-ice or ice push isthe destruction of shore structures. In lake resorts inthe north-eastern United States and southern Canada,docks, breakwalls, and shoreline buildings are almostimpossible to maintain on small lakes because ofthermal ice expansion. On larger lakes, and along theArctic Ocean coastline in Canada, northern Europe,and Russia, wind-driven ice push has destroyed coastalstructures. In the Beaufort Sea area of North America,oil exploration and drilling programs have beenseverely curtailed by drifting pack-ice. Sites have hadto be surrounded by steep breakwalls to prevent iceoverriding. In the Canadian High Arctic, oil removalwas delayed for years while the government tried todetermine the cheapest and safest means of exportingoil. Pipelines laid across the seabed between islandsare exposed to drifting icebergs, or to ice-push damagenear the shoreline. Alternatively, harbor facilities foroil tankers are threatened by ice push. Even where icepush is not a major problem, drifting sea-ice can push

into a harbor and prevent its being used for most of theshort (two to three months), ice-free navigation season.

SEA LEVEL RISE AS AHAZARD

Introduction

Sea level rise also poses a significant long-term hazard.Over the past century, there has been a measurable risein global (eustatic) sea level of 15.1 1.5 cm. In theearly 1980s, this rate was viewed as accelerating to3 mm yr-1, fuelling speculation that the Earth was aboutto enter a period whereby sea levels could rise50–100 cm eustatically in the next 100 years. A sea levelrise of this magnitude has destructive implications forthe world’s coastline. Beach erosion would accelerate,low-lying areas would be permanently flooded orsubject to more frequent inundation during storms, andthe baseline for watertables raised. Much of the rise insea level in the twentieth century has been attributedto the melting of icecaps, concomitant with increasingtemperatures because of global warming. Whetherglobal warming is occurring and whether it is natural orhuman-induced are moot points. Global warming andthermal expansion of surface ocean waters are forecastto cause sea level rises of 20–40 cm by 2100. Thissection will examine the evidence for sea level rise, itscauses, and the reasons for sea level fluctuations.

Current rates of change in sea levelworldwide(Cazenave et al., 2002; Nerem et al., 2002; ColoradoCenter for Astrodynamics Research, 2003)

Figure 8.9 presents sea level curves measured on tidegauges, 1930–1980, for five major cities generally per-ceived as being threatened by a rise in ocean levels:London, Venice, New York, Tokyo and Sydney. Thecities are not representative necessarily of eustaticchange in sea level because local (isostatic) effectsoperate in some areas. London is in an area where theNorth Sea Basin is tectonically sinking, Venice suffersfrom subsidence due to groundwater extraction, andTokyo has had tectonic subsidence due to earthquakeactivity. However, the diagrams illustrate that some ofthe world’s major cities would require extensive engi-neering works to prevent flooding given predictedrates of sea level rise. In addition, Figure 8.9 presentsthe changes in sea level for three cities: Stockholm;Wajima, Japan; and Crescent City on the west coast of


Fig. 8.8 Pack-ice 1 m thick pushing up gravel and muds along thesouth shore of Cornwall Island, Canadian Arctic. Scars inupper part of the photograph represent ice push at slightlyhigher sea levels.

the United States where uplift very much affects sealevel. In the case of Stockholm, land is still rising as theresult of glacial deloading over 6000 years ago.Tectonic uplift affects the other two cities. All recordsindicate that sea level can at times fall even at placessuch as London and Venice where it is rising in thelong term. More importantly, Figure 8.9 indicates thatsea level records are ‘noisy’. Globally, the averageannual variability in sea level – as measured by tidegauges – is 35 mm. This is 20–35 times the hypothe-sized annual rate of global sea level rise.

Figure 8.10 presents the trends in sea level between1960 and 1979 at shoreline tide gauges having at leastten years of data. Note that tide gauges are not theoptimum means for measuring sea level changes. Theirprimary purpose is to measure tides in harbors fornavigation – not sea level due to global warming. Theyare also unevenly spread across the globe. For thesereasons, only the overall trend in sea level along ashoreline is plotted in Figure 8.10. Large areas of

consistent sea level change are difficult to delineatebecause rates can change from positive to negativeover distances of a few hundred kilometers or less. Thisis the case in northern Europe and eastern Asia. Forexample, rates range from 24 mm yr-1 of submergencealong the south-east coast of Japan, to 6.8 mm yr-1

of emergence along the north-west coast. Sea levelhas fallen since 1960 along significant stretches ofcoastline, mainly in western Europe, western NorthAmerica, and eastern Asia. Increasing sea levels occurin the Gulf of Mexico; along the eastern coastlines ofNorth America, South America, and Japan; and theshorelines of southern Europe.

Satellite altimeters described above that measurewave heights provide better information about sealevel than tide gauges when results are averaged overtime. The patterns of sea level change measured bythese altimeters over the open ocean between January1993 and June 2002 are also plotted in Figure 8.10.There is very little correspondence between the

167Oceanic Hazards

Rising Sea Level

Falling Sea Level

London(Southend)

100

80

60

40

20

0

mm

100

80

60

40

20

0

mm

Venice(Trieste)

New York

Tokyo(Aburatsubo)

Sydney

Stockholm

Wajima, Japan

1940 1960 1980 1940 1960 1980

Crescent CityWest Coast, USA

Fig. 8.9 Annual changes in sea level, 1930–1980, at select tide gauges where sea level is either rising or falling.

measurement of sea level in the open ocean and thatmeasured at shore using tide gauges. More than likely,the two types of instruments are measuring separatephenomena. Tide gauges are measuring ocean basinedge effects along irregular coastlines. These levels areaffected by a myriad of processes that can be groupedunder four headings: 1) changes due to daily weatherfluctuations affecting temperature, wind stress, andatmospheric pressure; 2) seasonal variation due toheating of water or runoff from rivers; 3) inter-annualeffects due to major oscillations in pressure andheating in individual oceans; and 4) tectonics. Satellitesmeasure changes in the overall volume of water in theoceans. Their results indicate that daily, monthly, andinter-annual changes are dominated by steric factorsrelated to heating of the ocean surface and attributableprimarily to changes in the Southern, North Atlantic,and North Pacific Oscillations. For example, sea levelis falling in the central Pacific at a rate that exceeds

15 mm yr-1; however, it is rising at similar rates alongthe adjacent coastline. This is the signature of repeti-tively strong ENSO and La Niña events. There are alsopositive and negative changes at the boundary of theAntarctic circumpolar current in the south PacificOcean. Overall, global sea level has been risingrecently (Figure 8.10b). This is borne out in Figure 8.11,which plots the average of all altimeter readingsbetween 1992 and 2002. Sea level is increasing at therate of 1.85 ± 0.42 mm yr-1, a value that is remarkablysimilar to that determined using tide gauge data.

Factors causing sea level rise(Bryant, 1987; Sahagian et al., 1994; Gornitz, 2000)

Thermal expansion of oceans is not the sole cause ofrises in sea level during the twentieth century. Otherfactors may be responsible, including deceleration oflarge-scale ocean gyres or melting of alpine glaciers.For instance, the rise in sea level during the twentieth


-20 -10 0 mm yr -1

A Falling sea level Ice-covered or out-of-range

Ice-covered or out-of-range

0 10 20 mm yr -1

B Rising sea level Ice-covered or out-of-range

Ice-covered or out-of-range

Tide gauge record

Tide gauge record

Fig. 8.10 Changes in global sea level measured at coastal tide gauges between 1960 and 1979 (from Bryant, 1988) and from altimeters on theTOPEX/POSEIDON satellite between January 1993 and December 1999: A) falls, B) rises (from Cazenave et al., 2002).

century required only a 20 per cent reduction in alpineice volume. Anthropogenic effects in fact dominateglobal sea level changes. The two main mechanismsare through damming and groundwater extraction.These effects work in opposition to each other. Dams,built mainly since 1950, presently impound between530 km3 and 740 km3 of water. This is equivalent to adecrease in sea level of 1.3–1.8 mm yr-1. On the otherhand, groundwater extraction is raising sea level. TheUnited States obtains 20 per cent of its annual waterconsumption from groundwater. This amounts to123 km3, equivalent to a rise in sea level of 0.3 mm yr-1.If the source of the whole world’s annual waterconsumption of 2405 km3 were proportional to that inthe United States, then it would equate to a sea levelrise of 1.3 mm yr-1. Other activities that alter the globalhydrological balance include deforestation, draining ofwetlands, lake reclamation, and irrigation. In total,human activity could be reducing sea level by0.9 mm yr-1. This means that the natural rate of sealevel rise could be as high as 2.75 mm yr-1. Continueddam construction will maintain the human buffer onthis natural rise, but there is a finite limit to theamount of water that can be impounded. By the year2020, this limit should be reached and the rate ofmeasured sea level rise could double. Fortunately,there is an economical solution for negating future sealevel rises. Seawater could be diverted, with minimaleffort, into depressions such as the Dead Sea, theQattara Depression in Egypt, Lake Eyre in Australiaand the Caspian Sea. These basins have very large

areas below sea level that could easily hold increasedocean volumes.

Global sea level and the hydrologicalcycle(Bryant, 1993)

The cyclicity present in Figure 8.11 is problematic.Certainly, a high frequency component of about fourmonths exists with amplitudes of 15–20 mm. A longerterm fluctuation of about two years’ duration alsoexists. The latter is probably the quasi-biennial oscilla-tion, which has a periodicity of 2.2 years. Thisoscillation is a stratospheric phenomenon that appearsin various countries’ rainfall records. This implies thatthe fluctuations are climatically induced, reflectingseasonal and inter-annual variations in the globalhydrological cycle. The tide gauge records presented inFigure 8.9 demonstrate this aspect, despite the domi-nance of local factors. Changes in sea level recorded ontide gauges are remarkably coherent across the globe.For example, sea levels in southern California rise atthe same time as those in southern Australia decrease;sea level at Sydney, Australia, tends to rise one yearbefore it does at London and Venice; and sea level atLondon or Venice tends to rise one year before it doesat New York or Tokyo. This coincidence accounts forabout 62 per cent of the annual variation in sea levelfluctuations amongst these five cities. As Figure 8.9illustrates, sea levels oscillate 40–150 mm every 3–5years. This periodicity is characteristic of the SouthernOscillation, which has the greatest effect on rainfalls,

169Oceanic Hazards

19941992 1996 1998 2000 2002 20040

10

20

30

40

50

60

Cha

nge

in m

ean

sea

leve

l (m

m)

Year

Fig. 8.11 Global sea level change, 1992–2002 (Colorado Center for Astrodynamics Research, 2003).

especially in the southern hemisphere. During ENSOevents, rainfall decreases over Australia, South-EastAsia, India, and Africa. At these times, water drainsfrom continents, accumulates in the oceans, and causesglobal sea level to rise. There is a lag in time as waterspreads out from these source regions across the globe.During La Niña events, increased monsoonal rain fallsover Australia, India, South-East Asia, and Africa andaccumulates in rivers, lakes, and the ground. As aresult, global sea level falls. The land area involved inthese alternating drought and flood cycles is approxi-mately 16 � 106 km2, equivalent to 10–11 per cent ofthe ocean’s surface area. The rainfall differencebetween dry ENSO and wet La Niña phases is typically200–400 mm spread over a period of 2–3 years. Thevolume of rainfall involved, when averaged overthe oceans, equates with a change in global sea levelof 22–45 mm, a value that approximates the averageannual variability of 35 mm present in tide gaugerecords and the satellite altimeter data. As will beshown subsequently, these fluctuations have an impacton coastal erosion that may be just as important as anylong-term rise in sea level.

BEACH EROSION HAZARD

Introduction

There is little argument that 70 per cent of the world’ssandy beaches are eroding at average rates of0.5–1.0 m yr-1. Some of the reasons for this erosioninclude increased storminess, coastal submergence,decreased sediment movement shoreward from thecontinental shelf (associated with leakage out of beachcompartments), shifts in global pressure belts and,hence, changes in the directional components of waveclimates. As well, humans this century have had a largeimpact on coastal erosion. No single explanation hasworldwide applicability because all factors vary inregional importance. A lack of accurate, continuous,long-term erosion data complicates evaluation of thecauses. Historical map evidence spanning 100–1000years has been used in a few isolated areas, for instanceon Scolt Head Island in the United Kingdom, Chesa-peake Bay in the United States, and on the Danishcoast; however, temporal resolution has not beensufficient to evaluate the effect of climatic variables.Aerial photographs exist only in number since the1930s and often suffer from insufficient ground controlfor accurate mapping over time. Ground surveying of

beaches was rarely carried out before 1960 and is oftendiscontinuous in time and space. Examples of longtime series beach change data include: the 16-yearrecord for Scarborough Beach, Western Australia (thisalso goes back discontinuously into the 1930s); the fort-nightly record for Moruya Beach, New South Wales,from 1973 to the present; and the periodic profiling ofthe Florida coastline at 100 m intervals since 1972.These examples are isolated occurrences. Coastalgeomorphologists are still not exactly certain why somuch of the world’s coastline is eroding. This sectionwill examine some of the more popular means ofexplaining the causes of this erosion.

Sea level rise and the Bruun Rule(Schwartz, 1967; Bryant, 1985)

The general emphasis on rising sea level – anassumption made without firm evidence – has led tothis factor being perceived as the main mechanismcausing eroding beaches. Beach retreat operates via theBruun Rule, first conceived by Per Bruun, a Danishresearcher, and formulated by Maurice Schwartz of theUnited States. The Bruun Rule states that, on beacheswhere offshore profiles are in equilibrium and net long-shore transport is minimal, a rise in sea level must leadto retreat of the shoreline. The amount of retreat isnot simply a geometrical effect. Bruun postulated thatexcess sediment must move offshore from the beach, tobe spread over a long distance to raise the seabed andmaintain the same depth of water as before (Figure8.12). The Bruun Rule implies that beaches do notsteepen over time as they erode, but maintain their con-figuration, be it barred, planar or whatever. Generally, a1 cm rise in sea level will cause the shoreline to retreat0.5 m. Numerous studies – ranging from field measure-ments at various coastal locations such as high-energyexposed tidal environments, open ocean beaches andconfined lakes, to experiments in wave tanks – havesubstantiated this rate. The Bruun Rule operates byelevating the zone of wave attack higher up the beach.This is exacerbated by the fact that the baseline of thebeach watertable is also raised. Water draining from thelower beach-face decreases sediment shear resistance,and enhances the possibility of sediment liquefaction,causing seaward erosion of sand and shoreline retreat.The liquefaction process is described in more detail inthe chapters on earthquakes and land instability.

Unfortunately, while the Bruun Rule has beenproven universally, there are limitations to its use. The


actual volume of sediment needed to raise the seabedas sea level rises depends upon the distance from shoreand the depth from which waves can move thatmaterial. The Bruun Rule was formulated when it wasbelieved that this depth was less than 20 m. Individualwaves can move sediment offshore to depths exceeding40 m, and well-defined groups or sets can initiatesediment movement out to the edge of the continentalshelf. Coastlines, where Bruun’s hypothesis was mostsuccessfully applied, include the east coasts of theUnited States and Japan, where sea level is risingrapidly. There are many areas of the world where sealevel is rising at slower rates, or declining, and coastalerosion is still occurring. It would appear that sea levelchange as a major, long-term cause of beach erosionhas evolved from evidence collected on, and appliedwith certainty to, only a small proportion of the world’scoastline. A dichotomy also exists between measuredrates of beach retreat and the present postulatedeustatic rise in sea level. Common magnitudes ofbeach retreat in the United States, and elsewhere inthe world, range from 0.5–1.0 m yr-1. According to theBruun Rule, shoreline retreat at this rate wouldrequire sea level rises three to seven times larger thanthe maximum rise of 3 mm yr-1 that may presently beoccurring globally. Obviously, factors other than sealevel rise must be involved in beach erosion.

The actual rise in sea level may be irrelevant to theapplication of the Bruun Rule. As shown above, sealevel oscillations at all timescales are one of themost widely occurring aspects of sea level behavior.More sections of the world’s coastline have significant

oscillations than simply have rising sea level. Theemphasis upon the Bruun Rule has been upon risinglevels, but the rule can also predict accretion where sealevel is falling. If the Bruun Rule is to account for long-term beach erosion in these locations, then what has tobe taken into account is the imbalance between theamount of erosion during high sea level fluctuationsand the amount of accretion when sea level drops. Ittakes a shorter time to remove material seaward duringerosional, rising sea level phases – storm events – thanit does to return sediment landward during accretional,falling sea level periods. If eroded sediment that hasbeen carried seaward cannot return to the beachbefore the next higher sea level phase, then thebeach will undergo permanent retreat. Bruun’shypothesis only provides the mechanism by whichthese shifts in sea level operate. The more frequent thechange in sea level, the greater the rate of erosion.

Other causes of erosion(Dolan & Hayden, 1983; Bryant, 1985, 1988; Dolan et al.,1987; Bird, 1996; Gibeaut et al., 2002)

Clearly, more than one factor can account for erosionof the world’s coastlines. Table 8.2 lists these additionalfactors. Human impact on beach erosion is becomingmore important, especially with this century’s techno-logical development in coastal engineering. In somecountries such as Japan, breakwalls have been used asa ubiquitous solution to coastal erosion, to the extentthat their use has become a major reason for continuederosion. Wave energy reflects off breakwalls, leading tooffshore or longshore drift of sediment away from

171Oceanic Hazards

Watertable before rise

Watertable after rise

Amount of shoreline retreat

Sea level after rise

Sea level before rise

Amount ofsea level rise

Amount of seabed accretion

Seabed after rise

Seabed before rise

Shorelineerosion

Nearshoredeposition

Zone of enhanced water discharge

Fig. 8.12 The Bruun Rule for the retreat of a shoreline with sea level rise.

structures. Classic examples of breakwall obstructionor longshore sediment drift, leading to erosion down-drift, include the Santa Barbara harbor breakwall inCalifornia, and the Tweed River retaining walls southof the Queensland Gold Coast in eastern Australia.Offshore mining may affect refraction patterns at shoreor provide a sediment sink to which inshore sedimentdrifts over time. Erosion of Point Pelée, Ontario, inLake Erie, can be traced to gravel extraction somedistance offshore. Many cases also exist, along thesouth coast of England, of initiation of beach erosionafter offshore mining. Dam construction is veryprevalent on rivers supplying sediment to the coastline.Californian beaches have eroded in many locationsbecause dams, built to maintain constant watersupplies, have also trapped sediment that maintains astable, coastal sand budget. The best example isthe case of the Aswan Dam in Egypt. Since its con-struction, the Nile Delta shoreline has retreated by1 km or more. Even the simple act of revegetating

dunes can cause shoreline retreat if the sand budget ofa beach is fixed. Sand blown into vegetated dunescannot return to the beach on offshore winds, so thebeach slowly narrows over time.

Two little-known factors may be more importantthan any of the above. Each time someone walks acrossa beach, that person becomes a mini-excavator. Aperson who does not disrobe and endeavor to shake allclothing free of sand after swimming can remove 5–10 gof sand on each visit to the beach. On popular bathingbeaches, this can amount to several tonnes per kilome-ter of beach per year. Secondly, since the Second WorldWar rapid urban expansion near the coastline has led tomodification of urban drainage channels and water-tables. Water pumped from coastal aquifers to supplyurban water may not only cause local subsidence – as isthe case for most cities along the United States eastcoast – but may also lead to accelerated beach erosionbecause that water may raise local watertables if it ispassed through septic systems or used to water lawns.


Exclusively Human-induced• Reduction in longshore sand supply because of construction of breakwalls. • Increased longshore drift because of wave reflection off breakwalls behind, or updrift of, the beach. • Quarrying of beach sediment. • Offshore mining. • Intense recreational use.• Revegetation of dunes.

Exclusively Natural• Sea level rise. • Increased storminess. • Reduction in sand moving shoreward from the shelf because it has been exhausted or because the beach profile is

too steep. • Increased loss of sand shoreward from the beach by increased wind drifting. • Shift in the angle of wave incidence due to shifts in the location of average pressure cells. • Reduction in sediment volume because of sediment attrition, weathering, or solution. • Longshore migration of large beach lobes or forelands. • Climatic warming in cold climates, leading to melting of permanently frozen ground and ice lens at the coastline. • Reduction in sea-ice season leading to increased exposure to wave attack.

Either Human-induced or Natural• Decrease in sand supply from rivers because of reduced runoff from decreased rainfall or dam construction.• Reduction in sand supply from eroding cliffs, either naturally, or because they are protected by seawalls. • Reduction in sand supply from dunes because of migration inland or stabilization. • Increase in the beach watertable because of increased precipitation, use of septic tanks or lawn watering.Source: Based on Bird, 1988

Table 8.2 Factors recognized as controlling beach erosion (based on Bird, 1988).

Natural factors are perhaps still the major causes ofbeach erosion. Long-term changes have been bestdocumented on beach forelands along the Channeland east coast of England at Dungeness, Orford Ness,and Scolt Head Island. These forelands are tens ofkilometers wide and appear to have developed sincethe Holocene rise in sea level, following the last glacia-tion. However, some are slowly migrating along shore.Material is eroded from the updrift ends and depositedas a beach ridge on the downdrift end. In addition,shorter term beach erosion has manifested itself in thetwentieth century, mainly because of the change inglobal climate that has occurred since 1948. Thischange has not been one of magnitude as much as oneof variation: the storms follow longer periods of quies-cence, record rain and floods break severe droughts,and the coldest winter on record is followed by thewarmest summer. Unfortunately, investigations of thisclimate shift have focused on temperature change,whereas the shift also involves dramatic changes incoastal rainfall regimes, and changes in pressure celllocation leading to an upsurge in storminess.

While sea level rise enhances beach erosion by ele-vating beach watertables, logically any factor affectingwatertables can also cause beach erosion. For instance,a sandy backshore absorbing rainfall acts as a time-delaying filter for subsequent discharge of this water tothe sea through the lower foreshore. The process canraise the watertable substantially for periods up to twomonths after heavy rainfalls. Many coastlines, espe-cially those where seasonal orographic rainfall is pro-nounced, can receive over 1000 mm of precipitationwithin a month. Under these intense rainfalls, evensand suspension on the shore face by groundwaterevulsion is possible. Rainfall may be better related tobeach change than sea level. This aspect can be shownfor Stanwell Park Beach (40 km south of Sydney, Aus-tralia): a compartmentalized, exposed, ocean beachwith no permanent longshore leakage of sediment. Thebeach over time has oscillated between extremeerosion and accretion, but with no long-term trend ineither state (Figures 2.11 and 8.13). Figure 8.14 plotsthe relationship defined between average deviationsfrom the mean high tide position and (a) monthly sealevel and (b) monthly rainfall for this beach between1895 and 1980. Rainfall explains more of the change inshoreline position over time than does sea level(8.7 per cent versus 6.6 per cent). Despite all of theabove factors, the most logical cause of beach erosion

173Oceanic Hazards

Fig. 8.13 Difference in beach volume, Stanwell Park, Australia. Above, the beach after sustained accretion in February 1982.Below, the same beach when eroded severely by a storm inAugust 1986 (photograph courtesy of Dr Ann Young,Department of Geography, University of Wollongong).

is increased storminess. Figure 8.14(c) plots, between1943 and 1980, the relationship between the high tideposition and accumulated heights of storm waves forthe previous year on Stanwell Park Beach. Each 1 mincrease in storm wave height led to 0.47 m of beachretreat – a value equivalent to the effect generated bya 1 cm rise in mean sea level. The relationship is twiceas strong as that produced by rainfall and/or a rise insea level. In fact, storms are often accompanied byhigher sea levels and heavy rainfall. Storms concurrentwith these phenomena can explain 40 per cent ofthe variation in shoreline position at Stanwell ParkBeach. In contrast, periods without storms favorbeach accretion.

This effect is also recognizable elsewhere. Evidencefrom the barrier island coastline of eastern NorthAmerica, between New Jersey and South Carolina, notonly indicates large spatial variation in long-term ratesof erosion of barrier islands induced by storms, but alsoshows that particularly large storms tend to erode oroverwash the shoreline more in the places having the

highest historical rates of erosion produced by lesserevents. Beaches along this section of coastline areeroding on average 1.5 m yr-1. Islands with moresoutherly exposure have slower rates of erosion, a factimplying that the direction of storm tracking and waveapproach are important. Along this coastline between1945 and 1985, waves exceeded 3.4 m in height anaverage of 64 hours annually. East-coast lows and strongonshore winds generated by anticyclones accounted for13 and 24 per cent, respectively, of this total. Moststorm waves were produced by storms tracking alongthe polar front eastward across the continent, especiallysouth of Cape Hatteras. The Great Ash Wednesday

Storm of 1962 had little effect upon long-term erosionover this 42-year period, with its impact embedded inthe overall storm wave climatology.

In contrast, normally low-energy coastlines affectedperiodically by big storms respond differently. Here,rare but big storms generate waves with enough energyto exceed a threshold for complete beach erosion.Beaches undergoing this type of change occur in theGulf of St Lawrence and the Gulf of Mexico wheretropical storms are the main cause of shoreline retreatof barrier islands. Along the Texas coast of the Gulf ofMexico, tropical storms frequently affect the coastline.For example, between 1994 and 1998, eight tropicalstorms struck the north-western Gulf. However, onlytropical storm Josephine in October 1996 generatedwaves that exceeded the threshold for significanterosion by overwashing vegetated dunes. In this low-energy environment, this threshold is exceeded by anystorm generating a storm surge of 0.9 m and waves inexcess of 3 m lasting more than 12 hours. Smallerstorms, even though they can persist five times longer,do not cross this threshold. In the Gulf of St Lawrence,episodic events that change the barrier coastline areeven rarer because this region lies at the northernboundary of hurricanes, which here occur as infre-quently as once every twenty-five years. Barriershorelines remain spatially stable for years only toundergo landward retreat by tens of meters duringthese events. More dramatic is the puncturing ofbarriers to form temporary tidal inlets that then infillslowly over the following decade.

CONCLUDING COMMENTS

While sea level rise may be responsible for recentsandy beach erosion (the Bruun Rule), this factor maybe subservient to the role played by rainfall andstorms. Rainfall controls beach position through itsinfluence on the watertable – as does rising sea level.Storms are a continuous process operating over thetime span of decades or longer. Individual stormsare not necessarily responsible for beach erosion.However, if storms occur in clusters, then beacheserode because there is less time for beach recoverybefore the next storm. Three variables: rainfall, sealevel and storms, form a suite of interrelated variables.Analysis of data for Stanwell Park Beach illustrates thisaspect. Groups of environmental parameters thatindividually can be linked to erosion, collectively andlogically, operate together. Furthermore, changes in


Fig. 8.14 Linear regression lines (with 95 per cent confidencelimits) between average deviation from the mean hightide position on Stanwell Park Beach, Australia andA) monthly sea level, B) monthly rainfall, and C) heightof previous year’s storm waves (Bryant 1985, 1988).Graphs A) and B) are for the period 1895–1980, whilegraph C) is for the period 1943–1980.

y = -46.8x + 38.1

R2 = 0.07

-60

-40

-20

0

20

40

0.6 0.7 0.8 0.9 1.0Monthly sea level (m)

Dev

iatio

n H

T P

ositi

on (

m)

A

C-60

-40

-20

0

20

40

0 20 40 60 80

y = -0.33x + 7.82

R2 = 0.12

Previous year's wave height (m)

Dev

iatio

n H

T p

ositi

on (

m)

B

R2 = 0.09

0 200 400 600 800 1000 1200-60

-40

-20

0

20

40y = -0.016x + 6.68

Monthly rainfall (mm)

Dev

iatio

n H

T P

ositi

on (

m)

these environmental parameters reflect the large-scaleclimatic change that the Earth is presently under-going. The more dramatic and measurable increasesin temperature have unfortunately detracted from themore significant nature of that change, namelythe increasing variability of climate since 1948. Thisincrease accelerated in the 1970s and shows no sign ofabating. The world’s beaches will continue to respondto this change; unfortunately, it will be in an erosivemanner.

Finally, a note of caution: this chapter has castdoubt on the relevance of emphasizing a uniform,worldwide rising sea level, and shown that this factoris not solely responsible for eroding beaches. Erosionof the world’s beaches may not be so common, orso permanent, as generally believed. For example,when averaged over the long term, the high tiderecord for Stanwell Park Beach (Figure 2.11) isstable even though sea level off this coast is currentlythought to be rising by 1.6 mm yr-1; rainfall hasincreased about 15 per cent since 1948; and there areno presently active sediment sources. In Florida,detailed surveys at 100 m intervals around the coasthave shown that beaches did not erode between1972 and 1986, even though sea level rose andrainfall increased and sand supplies remainedmeager. Here, the absence of storm activity over thistime span may be the reason for stability.


Anisimov, O. and Fitzharris, B. 2001. Polar regions (Arctic andAntarctic). In McCarthy, J.J., Canziani, O.F., Leary, N.A.,Dokken, D.J. and White, K.S. (eds.) Climate Change 2001:Impacts, Adaptation and Vulnerability – Contribution ofWorking Group II to the Third Assessment Report of IPCC.Cambridge University Press, Cambridge, pp. 801–842.

Beer, T. 1983. Environmental Oceanography. Pergamon, Oxford.Bird, E. C. F. 1996. Beach Management (Coastal Morphology and

Research). Wiley, Chichester.Bruun, P. (ed.) 1985. Design and Construction of Mounds for

Breakwalls and Coastal Protection. Elsevier, Amsterdam.Bryant, E.A. 1985. Rainfall and beach erosion relationships,

Stanwell Park, Australia, 1895–1980: worldwide implications forcoastal erosion. Zeitschrift für Geomorphologie Supplementband57: 51–66.

Bryant, E.A. 1987. CO2-warming, rising sea level and retreatingcoasts: review and critique. Australian Geographer 18(2): 101–113.

Bryant, E.A. 1988. The effect of storms on Stanwell Park, NSWbeach position, 1943–1980. Marine Geology 79: 171–187.

Bryant, E.A. 1993. The magnitude and nature of ‘noise’ in world sealevel records. In Chowdhury, R.N. and Sivakumar, M. (eds)Environmental Management, Geo-Water & EngineeringAspects. Balkema, Rotterdam, pp. 747–751.

Cazenave, A., Do Minh, D., Cretaux, J.F., Cabanes, C., andMangiarotti S. 2002. Interannual sea level change at global andregional scales using Jason-1 altimetry. <http://topex-www.jpl.nasa.gov/science/invest-cazenave.html>

Colorado Center for Astrodynamics Research 2003. Long term sealevel change: global. <http://ccar.colorado.edu/research/gmsl/PDF/msl_global_noib.pdf> (URL defunct as of 2004)

Dalrymple, R. 2000. University of Delaware Wave Calculator.<http://www.coastal.udel.edu/faculty/rad/wavetheory.html>

Defant, A. 1961. Physical Oceanography v. 1. Pergamon, New York.Dolan, R. and Hayden, B. 1983. Patterns and prediction of

shoreline change. In Komar, P.D. (ed.) CRC Handbook ofCoastal Processes and Erosion. CRC Press, Boca Raton,pp. 123–150.

Dolan, R., Hayden, B., Bosserman, K. and Isle, L. 1987. Frequencyand magnitude data on coastal storms. Journal of CoastalResearch 3(2): 245–247.

Gibeaut, J.C., Gutierrez, R., and Hepner, T.L. 2002. Thresholdconditions for episodic beach erosion along the Southeast TexasCoast. Transactions of the Gulf Coast Association of GeologicalSocieties 52: 323–335.

Gornitz, V. 2000. Impoundment, groundwater mining, and otherhydrologic transformations: Impacts on global sea level rise. InDouglas, B.C., Kearney, M.S., and Leatherman, S.P. (eds) SeaLevel Rise: History and Consequences. Academic Press, SanDiego, pp. 97–119.

Gross, M.C. 1972. Oceanography: A View of the Earth. Prentice-Hall, Englewood Cliffs, New Jersey.

Jet Propulsion Laboratory 1995a. ‘TOPEX/POSEDON: Waveheight, January’. <http://education.gsfc.nasa.gov/experimental/all98invProject.Site/Pages/trl/inv47WAVE_HEIGHT_JAN.html>

Jet Propulsion Laboratory 1995b. ‘TOPEX/POSEDON: Waveheight, July’. <http://education.gsfc.nasa.gov/experimental/all98invProject.Site/Pages/trl/inv4-7WAVE_HEIGHT_JUL.html>

Kovacs, A. and Sodhi, D.S. 1978. Shore Ice Pile-up and Tide-up:Field Observations, Models, Theoretical Analyses. United StatesArmy Corps Eng. Cold Regions Research and EngineeringLaboratory Report.

Kushnir, Y., Cardone, V.J., Greenwood, J.G., and Cane, M. 1997. Onthe recent increase in North Atlantic wave heights. Journal ofClimate 10: 2107–2113.

Lawson, G. 2001. Monsters of the deep. New Scientist 30 June: 28–32.Long, D.G., Ballantyne, J., and Bertoia, C. 2002. Is the number of

Antarctic icebergs really increasing?. EOS, Transactions –American Geophysical Union 83: 469, 474.

Nerem, R. S., Mitchum, G.T., Giese, B.S., Leuliette, E.W., andChambers, P. 2002. An investigation of very low frequency sea levelchange using satellite altimeter data. <http://topex-www.jpl.nasa.gov/science/invest-nerem.html> (URL defunct as of 2004)

PO.DAAC (Physical Oceanography Distributed Active ArchiveCenter) 2003. NEREIDS ocean topography overview. JetPropulsion Laboratory, Californian Institute of Technology<http://nereids.jpl.nasa.gov/cgi-bin/ssh.cgi?show=overview>

175Oceanic Hazards

Sahagian, D.L., Schwartz, F.W., and Jacobs, D.K. 1994. Directanthropogenic contributions to sea level rise in the twentiethcentury. Nature 367: 54–57.

Schwartz, M.L. 1967. The Bruun theory of sea-level rise as a causeof shore erosion. Journal of Geology 75: 76–92.

Skov, N.A. 1970. The ice cover of the Greenland Sea. Meddelelserom Grønland 188(2).

Taylor, R.B. 1977. The occurrence of grounded ice ridges and shoreice piling along the northern coast of Somerset Island, N.W.T.Arctic 31(2): 133–149.

Vinje, T. 2001. Anomalies and trends of sea ice extent andatmosphere circulation in the Nordic Seas during the period1864–1998. Journal of Climate 14: 255–267.

Wiegel, R.L. 1964 Oceanographic Engineering. Prentice-Hall,Englewood Cliffs, New Jersey.

Zebrowski, E. 1997. Perils of a Restless Planet: Scientific Perspectiveson Natural Disasters. Cambridge University Press, Cambridge.


GEOLOGICALHAZARDS

P A R T 2

INTRODUCTION

Of all natural hazards, earthquakes and volcanoesrelease the most energy in the shortest time. In the past40 years, scientists have realized that the distribution ofearthquakes and volcanoes is not random across theEarth’s surface, but tends to follow crustal plate bound-aries. In the past 20 years, research has been dedicatedto monitoring these regions of crustal activity with theintention of predicting – several days or months inadvance – major and possibly destructive events. At thesame time, planetary studies have led to speculation thatthe clustering of earthquake or volcanic events over timeis not random, but tends to be cyclic. This knowledgecould lead to prediction of these hazards decades inadvance. Before examining these aspects, it is essentialto define how earthquake intensity is measured, becauseearthquakes are always characterized by their magni-tude. This aspect will be examined first, followed by adescription of the distribution of earthquakes and volca-noes over the Earth’s surface and some of the commoncauses of these natural disasters. The chapter concludeswith a discussion on the long- and short-term methodsfor forecasting earthquake and volcano occurrence.

SCALES FOR MEASURINGEARTHQUAKE INTENSITY

(Holmes, 1965; Wood, 1986; Bolt, 1993; NationalEarthquake Information Center, 2002)

Seismic studies were first undertaken as early as132 AD in China, where crude instruments were madeto detect the occurrence and location of earthquakes.It was not until the end of the nineteenth century thatthese instruments became accurate enough to measurethe passage of individual seismic waves as they traveledthrough, and along, the surface of the Earth. Thecharacteristics of these waves will be described in moredetail in the following chapter. The magnitude orintensity of these waves is commonly measured usingscales based on either measurements or qualitativeassessment. The first measured seismic scale was theMs scale established by Charles Richter in 1935.Richter defined the magnitude of a local earthquake,ML, as the logarithm to base ten of the maximumseismic wave amplitude (in thousandths of a milli-meter) recorded on a standard seismograph at adistance of 100 km from the earthquake epicenter.The scale applied to Californian earthquakes and tothe maximum amplitude of any type of seismic wave.Each unit increase on this scale represented a tenfoldincrease in the amplitude of ground motion. The scalewas later standardized for any earthquake using

Causes and Predictionof Earthquakes and Volcanoes

C H A P T E R 9

seismometers at any distance from the epicenter andsurface (Rayleigh) waves at a period of 18–22 secondsand a wavelength of 60 km. The characteristics ofseismic waves will be described in more detail in thefollowing chapter. This magnitude scale is calledsurface magnitude or Ms. Alternatively, the amplitudeof the initial seismic wave (P-wave), mb, having aperiod of 1–10 seconds can be used, because it is notaffected by the focal depth of the source of the earth-quake. This measure is called the body-wave magni-tude. The body-wave magnitude can be related to theMs scale as follows:

mb = 2.9 + 0.56 Ms (9.1)

Overall the mb value is less representative than the Msone of earthquake magnitude. The energy released byan earthquake can be related to these scales as follows:

log10 E = 4.8 + 1.5Ms (9.2)

log10 E = 5.0 + 2.6Mb (9.3)

The energy values are defined in ergs. For each unitincrease in surface magnitude, Ms, energy increases by31.6 times. None of these scales is appropriate for largeseismic events. Here, the seismic moment, Mw, is nowused. This scale is based upon the surface area of thefault displacement, the average length of movementand the rigidity of the rocks fractured.

Table 9.1 lists the annual frequency of earthquakesover the twentieth century on the Ms scale, while Table9.2 lists the largest events over the same period. Nohistorical earthquake has exceeded a magnitude of 9.0on the Ms scale. An earthquake larger than Ms= 8.0occurs about once a year. Its impact is catastrophic.About 18 earthquakes occur each year in the 7.0–7.9range. These earthquakes are deadly near populationcenters. About 120 and 800 earthquakes occurannually in the 6.0–6.9 and 5.0–5.9 ranges, respec-tively. The former earthquakes are deadly near popula-tion centers with inferior construction standards, whilethe latter can still cause significant property losses. Thelargest earthquakes appear to occur along the westernedges of the North and South American Plates, and inChina (Figure 9.1).

The Ms scale requires instrumentation. It cannot beused to assess the magnitude of prehistoric earthquakesor ones occurring in countries without a network of

seismic stations. Instead, the Mercalli scale (or itsmodified form) is used. This scale, summarized inTable 9.3, can be related approximately to the Ms scale.The Mercalli scale qualitatively estimates the strengthof an earthquake using descriptions of the type ofdamage that has occurred close to the origin or epi-center of the earthquake in built-up areas. Determining

180 Geological Hazards

Magnitude Frequency>8.0 17.0–7.9 186.0–6.9 1205.0–5.9 8004.0–4.9 62003.0–3.9 490002.0–2.9 365000

Table 9.1 Annual frequency of earthquakes by magnitude on theMs scale.

Magnitude, Death-Date Location Ms toll1 November Lisbon, Portugal 8.7 70 0001755

31 January Andes (Columbia) 8.6 5001906

16 August Valparaiso, Chile 8.6 20 0001906

18 April 1906 San Francisco, US 8.3 3 0003 January 1911 Tienshan,

Kazakhstan 8.4 45216 December Kansu, China 8.6 200 0001920

1 September Tokyo, Japan 8.3 143 0001923

3 March 1933 Japanese Trench 8.1 3 06415 August Assam, 8.6 1 5261950 India/China

22 May 1960 Chile 8.5 2 23127 March Alaska 8.6 1001964

26 January Gujarat, India 7.6 20 0002001

Table 9.2 Some of the largest earthquakes, by magnitude, on theMs scale.

181Causes and Prediction of Earthquakes and Volcanoes

Eurasian Plate

African Plate

Indo-AustralianPlate

Antarctic Plate

PacificPlate

North American Plate

Cocos Plate

Caribbean Plate

Nazca Plate

SouthAmerican Plate

Areas of most frequentearthquake activity

Volcanic eruptions onland historically

Plate boundary

Direction of movement

Fig. 9.1 Distribution of plate boundaries, intense earthquakes and historical land-based volcanic eruptions (based on Press & Siever, 1986; Bolt, 1993).

Maximum Acceleration Corresponding Scale Intensity Description of effect in mm s2 Richter ScaleI Instrumental detected only on seismographs <10II Feeble some people feel it <25III Slight felt by people resting; like a large truck <50 <4.2

rumbling by.IV Moderate felt by people walking; loose objects <100

rattle on shelves.V Slightly Strong sleepers awake; church bells ring. <250 <4.8VI Strong trees sway; suspended objects swing; <500 <5.4

objects fall off shelves.VII Very Strong mild alarm; walls crack; plaster falls. <1000 <6.1VIII Destructive moving cars uncontrollable; <2500

chimneys fall and masonry fractures;poorly constructed buildings damaged.

IX Ruinous some houses collapse; ground cracks; <5000 <6.9pipes break open.

X Disastrous ground cracks profusely; <7500 <7.3many buildings destroyed; liquefaction and landslides widespread.

XI Very Disastrous most buildings and bridges collapse; <9800 <8.1roads, railways, pipes, and cablesdestroyed; general triggering of otherhazards.

XII Catastrophic total destruction; trees driven from >9800 >8.1ground; ground rises and falls in waves.

Table 9.3 The Mercalli scale of earthquake intensity.

the magnitude of an earthquake on this scale does notdepend upon seismographs, but instead upon actualfield reports and pictures of damage to culturalfeatures, mainly buildings. The scale has some quanti-tative basis in that the categories on the scale increasewith increasing acceleration of the shock wave throughthe crust. When the maximum acceleration exceeds9.8 m s-2, which is the acceleration due to gravity, theshock wave of the earthquake hits the Earth’s surface sohard that objects are tossed up into the air and treescan be physically hammered from the ground. Theseaccelerations can be used to construct maps of seismicrisks, which will be described in more detail in thefollowing chapter.

DISTRIBUTION OF EARTHQUAKES AND VOLCANOES

(Hodgson, 1964; Doyle et al., 1968; Bolt et al., 1975;Blong, 1984)

Figure 9.1 shows the global distribution of areas of mostfrequent earthquake activity together with historicallyactive volcanoes on land. Superimposed on this map arethe locations of active plate boundaries. The moststriking feature is the correspondence between theoccurrence of these two hazards and the boundaries ofcrustal plates. Active volcanoes occur in three locations:near convergent plate margins, at divergent platemargins and over mantle hot spots. Briefly, convergent

plate boundaries are either subduction zones, wheretwo sections of the Earth’s crust are colliding with oneplate being consumed or subducting beneath the other;or mountain-building zones, where two or more platesare colliding with one overriding the other. The oceantrenches and the Pacific island arcs occur above sub-duction zones, while the Himalayas and European Alpsoccur within mountain-building zones (see Figure 9.2for the location of all major placenames mentioned inthis chapter). Although both these regions give rise toearthquakes, subduction zones are far more importantin accounting for the majority of earthquakes andalmost all explosive volcanism. While volcanoes at theselocations produce only 10–13 per cent of the magmareaching the Earth’s surface, they are responsible for84 per cent of known eruptions and 88 per cent of eruptions with fatalities. Most of these eruptionsoccur around the Pacific Ocean, along what is termedthe Pacific ‘ring of fire’. Plate boundaries throughIndonesia, Italy, and New Zealand have the longestrecord of activity. Between 1600 and 1982, 67 per centof all known deaths resulting from volcanoes occurredin Indonesia, on the convergent boundary of theIndo–Australian and Eurasian Plates. Mantle hot spotsare the next most important region of volcanic activity.The most notable hot spot includes the HawaiianIslands, which lie in the middle of the Pacific Plate.Most divergent plate margin volcanoes occur alongmid-ocean ridges where the crust is separating. Thelatter is the cause of volcanism in Iceland. Volcanoes on


Alaska

Timiskaming

New Madrid

Aleutian IslandsMt St HelensYellowstone

Salt Lake CityDenver

San FranciscoLos Angeles

KilaueaEl ChichonSanta Maria

Cosequina

Nevado del Ruiz

Hoover DamLanders

Chile

New Zealand

NewcastleMeckering

KrakatauGalunggung

Mt AgungTambora

Kariba

Koyna

Gujarat

DeccanPlateau

Yasur

Tennant Cr.

Tangshan

Mt FujiTokyo

KobeHaichengMt Pinatubo

Shensi

Himalayas

Italy

IstanbulIzmit

Mt EtnaVesuvius

Lisbon

Finland

Iceland Mt Hekla

European Alps

CanaryIslands

Charleston

La Soufrière

SoufrièreMt Pelée

France

New England


plate boundaries or over hot spots are usually associatedwith earthquakes.

About 75 per cent of earthquake energy is releasedin the upper 60 km of the crust along plate boundaries.These are termed shallow-focus events compared toearthquakes that occur up to 700 km below the crustbeneath subduction zones. Little if any seismic activityoriginates at depths greater than 700 km. There arehowever a significant number of earthquakes thatare not associated with either volcanic activity or plateboundaries, but in fact have an intra-plate origin. Manyintra-plate earthquakes are not associated with anyknown faults or historical earthquake activity. Forexample, in China, the earthquake record goes back to780 BC, and shows that some of the largest and mostdestructive earthquakes have occurred in a belt fromTibet to Korea, as much as 1000 km from the nearestplate boundary. The most destructive earthquake interms of loss of life occurred on 23 January 1556 inShensi province on the Hwang Ho River. These intra-plate earthquakes result from crustal stresses due tocontinued uplift in the Himalayas, to sinking in theShensi geosyncline or movement along the north-western edge of the north China plain. The Tangshanearthquake of 28 July 1976, which took 250 000 lives,occurred in this latter region. In North America, suchintra-plate earthquakes occured at Timiskaming in 1935in the middle of the Canadian Shield, at New Madridalong the Mississippi River between 16 December 1811

and 7 February 1812, and at Charleston, SouthCarolina, on 31 August 1886. The most recent intra-plate earthquake occurred on 29 January 2001 atGujarat, India. It killed 50 000 people.

Of all locations that should be aseismic, theAustralian continent continues to register the mostintra-plate earthquakes. The Tennant Creek, NorthernTerritory, earthquakes of 22–23 January 1988 are thelargest recorded in Australia, registering slightly morethan 7.2 on the Richter scale. Before these, the Meck-ering earthquake of 1968 in Western Australia was thelargest, registering 6.8. This was one of the few earth-quakes in Australia to be associated with observedfaulting. Until December 1989, no one had been killedby these earthquakes in Australia and propertydamage was minimal. The Newcastle earthquake of28 December 1989 changed this dramatically. In tenseconds, an earthquake of magnitude 5.5 on the Richterscale shook the center of Newcastle, New South Wales.It killed 12 people and left a damage bill of close to$A1000 million. Almost all of these earthquakesoccurred within the upper crust. Shallow earthquakesundoubtedly represent strain release of energy by frac-turing. The source of stress probably originates fromthe thermal imbalance in the upper mantle and may berelated to convective currents under Australia.

Figure 9.3 compares the return interval in years forearthquakes of various magnitudes between California,a seismically active region, and the supposedly aseismic


California 1932–1962

Australia 1920–1973

8

7

6

5

4

3

2

Mag

nitu

de o

n R

icht

er s

cale

Return interval in years1.5 2 3 4 5 10 20 50 1000.5

Fig. 9.3 Return intervals of earthquakes of various magnitudes in California and Australia (after Denham, 1979).

Australian continent. The data for Australia do not fita straight line because the most seismic part of thiscontinent is so isolated, with some events goingunrecorded. In any 100-year period, Australia canexpect an earthquake of magnitude 6.5 or larger. TheMeckering earthquake of 1968 and the Tennant Creekearthquake of 1988 in fact exceeded this value. In thelast 75 years, there have been 18 earthquakes thathave exceeded 6 on the Richter scale. The 1 in 5-yearevent has a magnitude of 5.8 on this scale. In contrast,California can expect an earthquake in excess of 8 onthe Richter scale at least once in 100 years. Severalhistorical earthquakes, including the San Franciscoearthquake of 1906, have exceeded this value. InCalifornia, the 1 in 5-year event exceeds 6.4 on theRichter scale. The mean annual maximum earthquakein California is about 0.5 orders of magnitude largerthan that in Australia. This difference becomes moresubstantial when it is realized that California is aboutone-twentieth the size of Australia.

This review on the distribution of earthquakesemphasizes three facts. Firstly, while earthquakes aremost likely to occur along plate boundaries, the largestearthquakes in terms of death toll and destruction havenot been associated with plate margins. Secondly, nocontinent or region can be considered aseismic. Eventhose that are perceived as being aseismic have aremarkably high incidence of earthquakes above 6 onthe Richter scale. And finally, while a major damagingearthquake is imminently expected along the SanAndreas Fault, major earthquakes have occurred in theeastern part of North America, which if they occurred

near populated centers today would be just as destruc-tive as expected ones in California.

CAUSES OF EARTHQUAKESAND VOLCANOES

(Holmes, 1965; Bolt et al., 1975; Whittow, 1980; Wood,1986; Ritchie & Gates, 2001; Matthews et al., 2002; Heki,2003; Jones, 2003)

Plate boundaries

Most volcanism occurs along the plate boundaries insubduction zones, and along lines of crustal spreadingsuch as the mid-Atlantic ridge. The theory of conti-nental drift states that the crust in the center of oceansspreads because of diverging convection cells in themantle. These cells rise towards the Earth’s crust andspread out at the surface, resulting in a ridge with a riftin the middle (Figure 9.4). This rift is infilled withmolten mantle material, which at weak points or frac-tures can form volcanoes. As the crust continues tospread apart, volcanoes migrate away from the centerof the ridging. If conduits joining the volcanoes to themagma pool can remain open, then the volcanoes willremain active. However, if the crust spreads faster thanlava can be fed to the volcano, the conduits block,the volcano becomes extinct, and newer ones developcloser to the rift.

Where plate boundaries are colliding, a differentprocess occurs. Part of one plate may be drivenbeneath another (Figure 9.5). This plate drags downwith it crustal material and forms a trench. The crustal


Mean sea levelActive volcano

Rift zoneExtinct volcanoes

Oceanic crust

Mantle convectioncell

Fig. 9.4 Origin of volcanism associated with mid-ocean plate spreading.

material is melted and incorporated into the mantle. Alarge zone of thrusting, termed the Benioff zone, formsto depths of 700 km with intense heat generation. Thiszone builds up heat because of magma extrusion,friction generated by earthquake activity and radioac-tive decay of crustal sediments. Beneath the overridingcrust, less dense, heated magma (mainly andesite andrhyolite) rises from the Benioff zone into the upperlayers of the mantle, before breaking sporadicallythrough the crust to build volcanic cones. This magmacontains gases derived from contact with groundwatersor from chemical reactions in dehydrated parts of thecrust that are being subducted. The western PacificOcean ring of fire represents the consumption of thePacific Plate by the Eurasian and Indo–AustralianPlates. A chain of volcanoes – an island arc – formsparallel to the subduction zone. The Japanese, Indo-nesian, and Aleutian Islands are classic examples ofvolcanic island arcs built in this manner.

The Benioff zone is also a major zone of earthquakeactivity because it represents thrust faulting of twoplates over a wide area. Deep-focus earthquakes at70–700 km depth occur mainly beneath the continen-tal side of the islands surrounding the Pacific andIndian Oceans. Over 75 per cent of all earthquakes

occur in the plate rim of the Pacific Ocean. The AndesMountains can be included in this category as they rep-resent an uplifted version of an island arc. As the depthof focus increases in the Benioff zone, the epicentersof deep earthquakes move further inland towardsthe centers of volcanic activity. At times of volcaniceruptions, earthquakes may occur because of volcanicexplosions, shallow magma movements within thecrust, or sympathetic tectonic earth movements.The first two factors result from release of pressuresin the crust, or changes due to expansion or deflationof the magma volume.

Plate collisions can also result in one continentalcrust overriding another, as is the case with theIndo–Australian Plate overriding the Eurasian Plate.Few if any volcanoes are generated in this situationbecause the crust of the Earth generally thickens atthese locations, and fracturing down to the mantle israre. Plates can also slide past each other as in the caseof the Pacific and North American Plates along theSan Andreas Fault. Not only earthquakes but alsovolcanoes can develop because fractures, called‘transcurrent faults’, open up at right angles to the lineof activity. Most of the volcanic activity in the RockyMountains and the Andes is of this nature.


Andesitic volcano

Oceantrench

Ocean crust

Beniof f

Zone

Cont inental crust

Subduct ion

Fig. 9.5 Origin of volcanism over a subduction zone (after Holmes, 1965).

Hot spots

A similar situation to volcano generation at separatingplates can be produced if a hot spot in the mantleremains nearly stationary over time. A hot spot repre-sents a long-lived plume of magma 10–100 km indiameter rising from the edge of the Earth’s core2900 km beneath the Earth’s surface. Volcanoes willdevelop above the hot spot as crustal material is slowlymelted from below. If a crustal plate drifts over thepseudo-stationary hot spot, then volcanoes driftingaway from the spot become extinct, and newer vol-canoes develop in their place. The Hawaiian Islandsowe their origin to drifting of the Pacific Platewestward over a hot spot (Figure 9.6). This process hasbeen occurring for over 40 million years. The oldestactivity occurs at Midway Island and a series ofseamounts northwest of the Hawaiian Islands. On thelatter islands, the active eastern island is about3–4 million years younger than the furthermostwestern and dormant Kauai Island. Rifting in theHawaiian area ensures that fractures penetrate to themagma source below. The melting of crustal materialabove the hot spot, and subsequent fracturing andsinking of the weakened crust, produces this rifting.

The Hawaiian Island chain indicates that the PacificPlate has drifted away from the East Pacific Rise north-west towards subduction zone trenches. Otherisland chains in the Pacific Ocean support this

hypothesis. Figure 9.7 plots the location of main islandstrings in the Pacific Ocean associated with drifting ofthe Pacific Plate over different hot spots. These hotspots probably represent the location of upwardflowing convection currents in the mantle that areresponsible for the spreading of plates away from theEast Pacific Rise. Note that not all volcanoes appearabove the ocean surface. Many are submerged. TheCaroline, Tuamotu, Society, and Austral volcanic islandchains all parallel movement of the Hawaiian Islands,and thus indicate a north-west movement of the PacificPlate away from the East Pacific Rise. The Sala-y-Gomez Island and Galapagos Island chains on the eastside of the Rise indicate migration of plates eastward.At least 25 primary hot spots have been identified, twoof which lie beneath continental crust under theeastern seaboard of Australia and Yellowstone in theUnited States

The theory of hot spots has fundamental flaws. Forexample, the crust under Iceland and the island ofHawaii, which are viewed as classic examples, does nothave anomalous heat flow suggestive of rising magma.The site of volcanism in the Canary Islands on the eastside of the mid-Atlantic ridge also does not appear tobe moving with the plate. Instead, eruptions haveoccurred randomly beneath the islands. Finally, thecomposition of volcanic material discharged abovehypothesized hot spots is not always indicative of deepmantle material. An alternative hypothesis suggests


Kauai

Direction of

plate movement

Maui

Hawaii

Molakai

Oahu4.5–5.6

3.0–4.0

2.7–3.32.2–2.5

1.81.2–1.3

0.4–0.8

0.2–0.7

0 100 200 km

N

Fig. 9.6 Age and direction of movement of Hawaiian Islands drifting over a hot spot (after Muller & Oberlander, 1984).

that lines of volcanic activity represent tearing of theEarth’s crust perpendicular to mid-ocean ridges.The Hawaiian and Tuamotu Island chains certainlyfit the locations where stress fractures should formvia this mechanism. It is now viewed that intra-platevolcanism is more complex than originally theorized inthe 1970s using the concept of hot spots.

Other faulting and dilatancy

The above factors account for the location of mostvolcanic activity and a significant number of earth-quakes. Most damaging earthquakes simply representthe rapid release of strain energy stored within elasticrocks. Such earthquakes are termed tectonic earth-quakes. The Earth’s crust is being continually stretchedand pulled in different directions as plates moverelative to each other, and as forces act within theplates themselves. Even minor crustal movement setsup elastic strain within the surface of the Earth, whichmay be greater than the internal strength of the layersof bedrock. This strain builds up a reservoir of energymuch like a coiled spring in a child’s toy. Where this

strain is excessive, ruptures or faults occur to relievethe pressures that are being built up. Fracturing mayoccur gradually, or as a sudden series of shocks thatradiate outwards from the strain zone in an unevenfashion, dependent upon the spatial variation in rockstrength.

The type of faulting or rupturing that occursdepends upon the characteristics of the fault. Figure9.8 shows the typical range of faults that occur withearthquakes. The zone of earthquake influence isnarrowest for the strike-slip fault, whereas normal orthrust faulting produces a wider zone of influence.However, the difference is really dependent upon theangle of the fault. For example, as the dip of the SanAndreas Fault is virtually vertical, earthquake damagerapidly diminishes within 20 km of the fault line. Thisis why moderate earthquakes this century in southernCalifornia have not caused widespread damage. On theother hand, the high magnitude earthquakes in Chilein 1960, and Alaska in 1964, occurred on low-angledreverse faults (20º or less in the case of Alaska) associ-ated with subduction zones, a feature that extended


Hawaiian Is.

Marquesas Is.Tuamotu Is.

Society Is.

Austral Is.

Cook Is.Sala-y-Gomez

GalapagosCaroline Is.

EastPacif

ic

Pla

teboundary

Plate boundaries

Active or extinct volcanoes

Seamounts or atolls

Fig. 9.7 Location of other hot spots and associated island strings on the Pacific Plate (based on Holmes, 1965).

damage over a much larger area. Weathered materialabove bedrock can absorb considerable amounts of thefault displacement, such that faulting occurring inunderlying bedrock may not be present at the Earth’ssurface. Elastic strain can be relieved slowly over timeas tectonic creep. Such creep can be monitored andused as a tool for earthquake prediction, a facet thatwill be discussed later.

Earthquakes can be triggered by sudden loading ordeloading of the Earth’s crust along most faults. Thecoincidence of tropical cyclones and earthquakes hasalready been noted in Chapter 3. Here the mechanismwas linked to the rapid onset of storm surges on conti-nental shelves. The loading threat may be even moresubtle. In Japan, records over the last 1500 yearsindicate that intra-plate earthquakes with magnitudesgreater than 7 are three times more likely to occur inspring. The cause is related to snow accumulationduring winter that puts stress on reverse and strike-slip faults. Rapid melting in spring then unloads thecrust suddenly, triggering earthquakes. As little as 1 mequivalent depth of water may be involved in theprocess.

Faulting and earthquakes can also operate viadilatancy in crustal rocks. At depths greater than

5 km, the pressure due to the weight of overlying rockis equal to the strength of unfractured rock. Theshearing forces bringing about sudden brittle failureand frictional slip can never be obtained, because therock deforms as a plastic. However, the presence ofwater provides a mechanism for sudden rupture byreduction of the effective friction along any crackboundary. If crustal rocks in the upper crust strainwithout undergoing plastic deformation, they maycrack locally and expand in volume. The process isknown as dilation. Cracking may occur too quickly forimmediate groundwater penetration, but water willeventually penetrate the cracks, providing lubricationfor any remaining stresses to be released. This modelhas prospects for predicting earthquakes because itinvokes measurable changes in ground levels, electri-cal conductivity and other physical factors precedingtremors.

Added water (dams and rain)

The dilatancy theory can also explain the presence ofearthquakes around reservoirs. Reservoir earthquakesoccur in both seismic and non-seismic zones, and areunrelated to rock type or local faulting patterns.Interest in this phenomenon started in 1935 after the


+ +

Normalfault

Lateralnormal

fault

Lateralreverse

fault

Reversefault

Strike-slip orlateral fault

Fig. 9.8 Types of faults associated with earthquake activity.

building of the Hoover Dam, which impounded LakeMead on the Colorado River. The area was not notedfor earthquake activity before the dam was constructed.Since that time, however, over 1000 earthquakeshave been generated with enough strength to be felt bylocal inhabitants. Since then, this effect has been notedfollowing the construction of 10–15 other reservoirsworldwide including the Kariba Dam on the ZambeziRiver in southern Africa and the Koyna Dam nearMumbai, India. It should be noted that not all artificialreservoirs have generated earthquakes, even whenthey have been located in active seismic zones.However, enough moderate earthquakes registering5–6.5 on the Richter scale have been produced atKoyna, India, to cause the loss of 177 lives. It appearsthat reservoir water penetrates the underlyingbedrock, reducing effective frictional resistance alongfractures, and permitting slippage significant enoughto generate earthquakes. Mining, water and oil extrac-tion, and waste-fluid disposal underground have alsoresulted in local seismic activity. For example, waste-water from chemical warfare manufacturing waspumped underground at Denver, Colorado, between1962 and 1965, resulting in measurable tremors.However, the effects of these latter operations appearminor compared to those produced so far by reservoirconstruction. Finally, there is a statistical correlationbetween the wettest time of year and eruptions ofYasur, Mt Etna, Mt St Helens, and Soufrière on theisland of Montserrat. Rain immediately precededrecent eruptions of the latter volcano on three differ-ent occasions. Rain seeps into the volcano, and uponcontact with the magma chamber, turns to steamcausing a blast.

PREDICTION OF EARTH-QUAKES AND VOLCANOES

Clustering of volcanic and seismicevents

Volcanoes(Gribbin, 1978; Lamb, 1972, 1982; Blong, 1984; Pandey& Negi, 1987)

The prediction of earthquakes and volcanoes can bebroken down into two time spans: long-term predictionbeyond a period of several years, and short-term pre-diction from a couple of years to several days or hoursbefore the event. Long-term prediction of volcano andearthquake activity involves the delineation of cycles in

activity over decades or centuries, and clustering ofevents.

Over geological time, periods of volcanism have notoccurred as random events. Over the past 250 millionyears, volcanic activity appears to be enhanced every33 million years with a minor peak occurring every16.5 million years. The greatest peak in activityoccurred 60–65 million years ago, corresponding tothe formation of the Deccan Plateau in India, andthe Cretaceous mass extinction that wiped out thedinosaurs. Periods of volcanic activity correlate withother mass extinctions, major reversals in the Earth’smagnetic field, meteor or comet impact-cratering andother terrestrial processes. The cause of these cata-strophic events is related to the passage of the solarsystem through the galactic disc. Our galaxy – theMilky Way – is a flattened spiral of stars that, side-on,looks like a plate. The solar system rotates relative tothe center of the galaxy, passing from one side of thedisc to the other. A complete circuit takes 33 millionyears, such that the solar system passes through theaxis of the galaxy every 16.5 million years. At present,the solar system is passing through the galactic disc andvolcanism has been increasing over the last 2 millionyears.

Extended periods of volcanism represent a majorchange in the isothermal, circulatory, and convectivebehavior of the mantle. As a result, major plate move-ments take place, leading to faster crustal spreadingrates, to broadening of oceanic ridges and to sea levelrises. More importantly, increased volcanic dust duringmajor periods of volcanic activity screens out enoughsolar radiation, reducing photosynthesis and disturbingthe food web sufficiently, to cause the disappearanceof many land- and ocean-based species. Astronomicalperiodicities at the galactic level are sufficient to accountfor most periods of volcanism on the Earth, concomitantwith increased comet impacts, magnetic reversals,climatic change and, ultimately, mass extinctions. Vol-canism must be one of the most far-reaching andprofound processes affecting the Earth geologically.

Blong (1984) has summarized some interesting dataon the frequency of volcanic eruptions over the past10 000 years. Over the last 500 years, individual volca-noes have erupted at the median rate of once every220 years. The number of eruptions per century, whenplotted on log probability paper (Figure 9.9), forms astraight-line relationship similar to the frequency ofoccurrence of most natural hazard events. About


20 per cent of volcanoes erupt less than once every100 years, and 2 per cent less than once every10 000 years. Both the El Chichon eruption in Mexicoin March 1982, and the Pinatubo eruption in the Philip-pines in June 1991, exemplify this latter aspect. Neitherhad any historic record of eruption and both appearedextinct. The 25 most violent volcanic eruptions occurredwith a median frequency of 865 years. Of the morethan 5500 eruptions by 1340 volcanoes since the lastglaciation, only 40 per cent are known to have eruptedin the historic past. Thus, most of the world’s extinctvolcanoes could become active in the future. Onaverage, one extinct volcano erupts every five years.

While the above analysis shows the randomness oferuptions, the historical record of volcanic activitybased on dust that volcanoes have thrown into theatmosphere illustrates that eruptions are clusteredover time. The amount of dust can be evaluated accu-rately by examining the attenuation of solar radiationby volcanic dust and measuring the widths of tree-rings(Figure 9.10). Professor Lamb in the United Kingdomcompiled indices of volcanic activity based on solarradiation attenuation going back to the sixteenthcentury. His index is referenced to the amount of dustproduced by the Krakatau eruption of 1883 (base valueof 1000) and is known as the Dust Veil Index. TheSantorini eruption in the Aegean in 1470 BC produced

an index value of 3000–10 000, while Mt Vesuvius in79 AD generated a value of 1000–2000. Solar radiationattenuation, lower temperatures, and increased frostsreduce tree-ring growth. These effects are enhancedfollowing major volcanic eruptions, so that tree-ringsshould have increased density and diminished width. Arecord of the 20 lowest tree-ring density measure-ments for the last 400 years, from Europe and NorthAmerica, is also plotted in Figure 9.10. These recordsshow that the eruption of Huaynaputina, in Peru in1601, had the greatest impact on tree-rings, while theeruption of Tambora in 1815 produced the greatestamount of dust in the atmosphere.

There have been some truly cataclysmic eruptionsover the past 300 years. The Tambora eruption on theisland of Sumbawa in Indonesia, during its maineruption in 1815, and the Cosequina eruption inNicaragua in 1835, both released four times as muchdust as Krakatau. These eruptions occurred simultane-ously with many smaller ones, producing a period ofsustained volcanic activity beginning in 1750 andterminating with the eruptions of Mt Pelée inMartinique, Soufrière on St Vincent Island, and SantaMaria in Guatemala in the Caribbean Sea area in 1902.The tree-ring record indicates that the period between1600 and 1710 was also one of persistent volcanism.After the volcanic eruption of Katmai, Alaska, in 1912,


0.01 0.1 1 10 100

Eruptions per century

9998

95

90

80

70

60504030

20

10

5

21

Cum

ulat

ive

perc

enta

ge

Fig. 9.9 Probability diagram of eruptions of individual volcanoes per century (Blong, ©1984, with permission Harcourt Brace Jovanovich Group,Australia).

there was not one significant dust-producing volcaniceruption that affected the northern hemisphere climateuntil Mt Agung in Bali in 1961. Mt Hekla, Iceland, in1947 produced 100 000 m3 s-1 of ash, which reached asfar as Finland; however, the eruption was short-livedand did not inject significant debris into the stratos-phere. Mt Agung represented a mild reawakening ofactivity worldwide, which did not become intense untilthe Mt St Helens eruption. Besides Mt St Helens, therehave been major eruptions of Galunggung, westernJava, Indonesia, in April 1982; El Chichon in March1982; Nevado del Ruiz, Colombia, in November 1985;Augustine, Alaska, in 1986; and Pinatubo in June 1991.Because of these eruptions, the Dust Veil Indexclimbed above 2000 during the last two decades of thetwentieth century.

Also plotted in Figure 9.10 is the average tempera-ture of the northern hemisphere, referenced to 1880.Dust emissions have a significant impact on globaltemperature 2–3 years after a major eruption. Forexample, dust from the Krakatau, Indonesia, eruptionof 27 August 1883, reduced solar radiation in Franceby 10–20 per cent over a three-year period. The corre-

spondence between the Dust Veil Index and globaltemperature is strong (regression coefficient of –0.65,which is significant at the 0.01 level). Periods ofincreased volcanism decrease surface air temperaturesand have had a marked impact on society. For instance,volcanism in the 1780s produced extremely coldwinters and drought throughout western Europe andJapan. In France, the adverse climatic events exacer-bated the social conditions leading to the French Rev-olution in 1789. In 1816, the dust and sulfate emissionsfrom the Tambora eruption, in the previous year,produced temperature drops of 1°C in New England,western South America, Europe, and the south-westPacific. In New England, the event was known as the‘year without a summer’. Frost occurred in everymonth of the year in eastern North America, cropsfailed in Wales and central Europe leading to famine,and the South-East Asian monsoon was very intense.In contrast, the summer was hot in the MississippiValley, the Yukon, and north-western Russia. Thiseruption, and a series of others, was conducive toclimatic conditions that favored the first global epi-demics of typhus and cholera between 1816 and 1819.


Dus

t Vei

l Ind

ex

1550 1600 1650 1700 1750 1800 1850 1900 1950 2000

Year

Inde

x of

min

imum

tree

-rin

g de

nsity

Tem

pera

ture

dev

iatio

nfr

om m

ean

(˚C)

N. hemisphere temperature

0

1000

2000

3000

4000

5000

0.0

0.6

-0.6

2.5

2.0

1.5

1.0

0.5

Dust

Krakatau

Tambora, IndonesiaHuaynaputina, Peru

Tree-rings

Fig. 9.10 Recent volcanism (1600–2000) as determined by Lamb’s Dust Veil Index and tree-ring density measurements. The top panel plotsnorthern hemisphere air temperature over the same time span (based on Gribbin, 1978; Lamb, 1985; and Jones et al., 1995).

The heightened volcanic activity of the last fourcenturies indicates that the Earth may be witnessing aperiod of increased volcanism not seen for 10 000 years.

Earthquakes (Whittow, 1980)

There is some evidence to suggest that regional earth-quake activity is clustered. Figure 9.11 plots the mag-nitude of earthquakes based on the Richter scale inJapan since 684, the Mediterranean for the periodsince 1900, and the United States since 1857. Japan,which is seismically very active, has experienced aseries of high-magnitude earthquakes centered on theeighth, sixteenth and twentieth centuries. In the nine-teenth century, earthquakes appeared to cluster every50 years. The most seismic period in the recordoccurred after the Second World War, but the mostdestructive earthquake in Japan in recent timesoccurred in 1923 at Tokyo. The Kobe earthquake of17 January 1995 may simply have been part of thisenhanced twentieth century activity. In the Mediter-ranean, earthquakes occur more regularly (Figure9.11b). Since 1920, there have been four clusters ofearthquakes, centered around 1930, 1941, 1953–1956,and 1977. Except for Turkey, the Mediterraneanregion has been quiescent in the latter part of thetwentieth century. In North America, clustering ofearthquakes has also occurred along the westerncontinental margin. The most intense phase of seismicevents has occurred since 1980. Other periods ofclustering occurred around 1906 – just before theSan Francisco earthquake – 1930, 1951, and 1964.However, the clustering is not nearly so pronounced asin the Mediterranean region. If clustering of earth-quakes is real, then volcanic eruptions and earthquakeactivity should also be coherent because they have thesame underlying geophysical causes.

Interaction between earthquakes and volcanoes

(Hill et al., 2002)

There is a high occurrence of volcano activity immedi-ately after earthquakes greater than 6.5 in magnitude.For example on 29 November 1975, Kilauea volcanoerupted 30 minutes after a 7.5 magnitude earthquakeon its southern flank. Cordon Caulle volcano in centralChile erupted two days after the great 9.5 Mw magni-tude Chilean earthquake of 22 May 1960. While theseeruptions occur within days of an earthquake, theeffect may extend to hundreds of years afterwards up

to 1000 km away from earthquake epicenters. Thecauses are due to static stress changes near the epicen-ter of an earthquake, stress changes associated with theslow relaxation of the lower crust, or dynamic stressestriggered by the earthquake. Static stress changes arecaused by changes in pressure in a magma body lyingclose to an earthquake. The change may squeeze themagma upward like toothpaste from a toothpaste tubeor enhance melting adjacent to the magma. Morelikely, the stress change may trigger the formation ofbubbles in the magma and the freeing of conduitslinking the magma chamber to the surface. Theseprocesses probably caused Mt Fuji in Japan to erupt in1707, one month after an 8.2 magnitude earthquake;and Mt Pinatubo to erupt in 1991 following a 7.7 mag-nitude earthquake 100 km away. It is also possiblethat inflation of the magma chamber may triggerearthquakes that positively feed back on the inflationprocess. This pairing of events within ten years of eachother seems to be characteristic of major eruptionsof Vesuvius. Earthquake swarms over periods of7–50 years are also related to viscous relaxation of thelower crust. After an earthquake, relaxation of shearstresses in the upper 15–20 km of crust triggers aconcomitant relaxation in the underlying plastic crust.This effect induces volcanic eruptions either immedi-ately or 30–35 years afterwards up to 1000 km fromthe earthquake epicenter.

The most likely process causing volcanic activityafter earthquakes takes place in the magma. Seismicwaves can dislodge bubbles held by surface tension onthe floor and walls of a magma chamber. The sameseismic waves can also increase bubble size. Duringthe dilating phase of a seismic wave, bubbles grow asgas escapes under lower pressure from the magma.During compression, the gases dissolve back into themagma. However, periods of dilation are longer thanthose of compression. Hence, as the seismic wavespass through magma chambers, the bubbles grow insize. Within a few hours, the 28 June 1992 Landers,California, earthquake triggered 14 other earthquakesup to 1200 km away. Many of these earthquakesoccurred in the southern part of the Long ValleyCaldera where it is estimated that the Landers earth-quake increased pressure by 7–10 MPa. Finally,earthquakes may dislodge crystals that have precipi-tated out of the magma body. Because these crystalsare denser than the magma, they settle downward,forcing gas-rich magma upward. This latter magma is


subject to a pressure reduction, releasing gas thatincreases buoyancy and leads to further upwardmovement. This molten material can continue to riseupward in a runaway process at the rate of 10 cm s-1,eventually rupturing the surface within a week of thetriggering earthquake.

Prediction of seismic activity

Earthquake cycles (Wood, 1986)

In 1975, research into earthquake movements in swampdeposits in the Los Angeles area indicated that there hadbeen eight major earthquakes since 565 AD, spaced at


A

B

C

Japan

Mediterranean

North AmericaSan Francisco Alaska

1850 1900 1950 2000

1850 1900 1950 2000

1900 1920 1940 1960 1980

500 1000 1500 2000

Tokyo

9

8

6

7

9

8

7

9

8

6

7

9

8

6

7

Fig. 9.11 Plot of magnitude on the Richter scale of major earthquakes A) in Japan, B) in the Mediterranean and C) in the United States (based onWhittow, 1980). Arrows indicate clustering. Updated from <http://earthquake.usgs.gov/activity/past.html>

intervals of between 55 and 275 years. The averagereturn period was 160 years, with the last major earth-quake occurring in 1857. Based upon this periodicity,the next major earthquake above 8 on the Richter scaleis imminent in the Los Angeles area within the next20–30 years. This forecast is not illogical given the factthat most earthquakes occur along plate boundaries,which are slowly moving past each other at rates averag-ing 10 cm yr-1. If the frictional drag between two platesmoving relative to each other is not released continually,then forces will build up over time, culminating in asingle major jump of the plates along a fault line.Because the plates are moving at a continual rate, thehistoric record of fault displacements should provide thereturn period or frequency of earthquakes in a region.For instance, if the average displacement along a fault is5 m for each earthquake, then the fault should be activeevery 50 years, given the above rate of plate movement,10 cm yr-1. Around the Pacific Ocean, the return periodfor earthquakes is between 75 and 300 years. More sig-nificantly, once a plate has moved along a fault line in anarea, the probability of a similar sized event occurring inthe next few decades after the earthquake is perceivedas small. In the area of the 1960 Chilean earthquake, the

return period of large earthquakes for the past 300 yearsaverages 80 years. The next major earthquake here is notexpected until 2040 ± 20 years.

Around the Pacific region, plates are moving soconsistently that anomalous regions that are notmoving can be easily detected. For instance, in theAlaskan region, plate movements have generatedcontinual earthquake activity over the past 150 years asstresses build up to crucial limits and are periodicallyreleased at various points along the plate margin.However, at some locations, the stresses may not bereleased easily; these points appear in the historicalrecord as abnormally aseismic, while adjacent regionsare seismically active. These locations, called seismicgaps, are prime sites for future earthquake activity,especially if earthquakes have not occurred in theregion in the previous 30 years. The Alaskan earth-quake of 1964 filled in one of these gaps as did theMexican earthquake of 19 September 1985. Majorseismic gaps as of 2003 are shown in Figure 9.12. TheLos Angeles area of California is located at a seismicgap, as is much of the Caribbean Sea.

Away from plate boundaries, the overall rate ofmovement along faults is an order of magnitude slower


Earthquake zone

Seismic gap

Predicted earthquake

Fig. 9.12 Location of major seismic gaps in the Pacific ‘ring of fire’ (Coates, 1985).

than around the Pacific Rim. Here, the repeat timeof an earthquake becomes much larger. This hassignificance for areas in the middle of plates that haveexperienced isolated earthquakes in historical times.For instance, the New Madrid earthquakes in theMississippi Valley in 1811–1812, and the Charleston,South Carolina, earthquake in 1886 appear to beanomalous (Figure 9.2). The concept of earthquakerepeat-cycles would indicate that these regions arehighly unlikely to experience a similar intensityearthquake for hundreds, if not thousands, of years.There are also active fault lines where little or no majorearthquake activity has been recorded historically.For instance, the Wasatch fault, which passes throughSalt Lake City, appears to be more active than others inthe region. Yet, historically it has never experienceda major earthquake. For these types of faults, majorearthquakes may occur at 500–1000-year frequencies.

Short-term prediction of seismic activity

(Rikitake, 1976; Ward, 1976; Simpson & Richards, 1981;Kisslinger & Rikitake, 1985; Lighthill, 1996)

Of all natural hazards, earthquakes generate the largestand longest range of associated phenomena that can beused to foreshadow the timing of the impending event.Research into earthquake prediction has been carriedout since the late nineteenth century. In 1891, theJapanese government set up an earthquake investiga-tion committee, which collected historical records ofearthquakes in Japan as far back as 416 AD. In 1925,the Earthquake Research Institute was established atTokyo University to give earthquake research a scien-tific impetus. Following Japan’s economic recoveryin the 1950s, earthquake prediction was intensified. InChina, where historical records go back 3000 years, amajor effort at identifying earthquake precursors wasinitiated after 1950. On 5 February 1975, the city ofHaicheng became the first major inhabited area tobe evacuated before the occurrence of a significantlylarge earthquake. In the months leading up to theearthquake, ground tilting was noticed together withunusual animal behavior and water spurting from theground. The city was evacuated 12 hours beforethe earthquake struck. The earthquake measured 7.2on the Richter scale and, despite the destruction of90 per cent of the city, only a few lives were lost.

Precursors of earthquakes group into the followingfive categories: land deformation, seismic activity, geo-magnetic and geo-electric activity, ground water, and

natural phenomena. Land deformation studies arebased on the concept that the strain energy building upin the Earth’s crust will evince itself in minor lateralor vertical distortions at the Earth’s surface. Much ofthe Earth’s surface has been surveyed with bench-marks or triangulation stations. In some countries,these stations have been accurately surveyed to forman extensive triangulation network. By resurveyingstations within this network every decade, it is possibleto detect gross movement, ranging from centimeters tometers, in the Earth’s crust. It is also possible to relatethese stations to tide gauges, to look for land move-ments relative to sea level in coastal areas. On a smallerscale, it is now possible with infra-red and laser surveyequipment to mount survey stations on opposite sidesof active fault zones and to measure earth movementsas small as 1 mm over distances of several kilometers.This has been carried out along key sections of theSan Andreas Fault. Points showing rapid changes inearth movement along the fault may signal futureearthquake activity nearby. Crustal deformations canalso be measured using tiltmeters and strain gauges.Increases in strain signal earthquake activity withinmonths, while rapid increases in tilting of the Earth’scrust (called tilt steps) give warnings of activity withinhours.

On a smaller timescale, foreshocks or microseismsalso indicate an impending earthquake. However, thenature of the warning is paradoxical and depends uponthe area. Some earthquake zones give foreshocks daysto hours before the main shock, while others evince adecrease in microseismic activity before the shock.Changes in seismic wave velocities through rock canalso be used to foreshadow an earthquake. Strainchanges the velocity at which a seismic wave will travelthrough the ground. By measuring the velocities ofhuman-induced shocks across known earthquakezones, it is possible to forecast major earthquakes yearsin advance. Decreases in strain of up to 20 per centhave been measured for compression-type wavesbefore earthquakes.

There are also geomagnetic and geo-electric pre-cursors. It has already been pointed out that theEarth’s geomagnetic activity varies with the sunspotcycle, a relationship that may be linked to earthquakeactivity. Geomagnetic activity also varies spatially andtemporally over the surface of the Earth. Anomalouschanges in geomagnetic activity in the order of 4–20gammas have been measured up to ten years before


local earthquake activity. The Earth’s ability totransmit currents, or its electric resistivity, can fore-shadow earthquake activity. Changes in resistivity aremonitored by sending electrical pulses through theground and measuring their strength some distanceaway. Decreases in resistivity of 10–15 per cent havebeen detected several months before major earth-quakes, while stepped increases have been detectedseveral hours in advance, even for earthquakes somedistance from the monitoring site. Electrical currents,termed telluric currents (or Earth currents), are alsocontinually flowing through the Earth’s crust. Changesin the order of 2 mV have been observed hours beforesmall earthquakes.

Many minerals such as quartz have piezoelectricproperties. When stress is applied to rocks containingthese minerals an electrical current is generated.Changes in the electrical potential between two rodsbored into the crust can be used to measure the mag-nitude of stress building up along a fault line before anearthquake. The best example of this technique is thecontroversial VAN method – named after the initials ofProfessors P. Varotsos, K. Alexopoulos and K. Nomicoswho developed it to forecast earthquakes in Greece.The method allegedly predicts the magnitude andepicenter of earthquakes, greater than Ms 5, within anerror of 0.7 units and 100 km respectively, severalhours to days in advance.

Groundwater fluctuations in wells can foreshadowearthquakes with magnitudes greater than 5 on theRichter scale, 0.5–10.0 days before the event. Thistechnique is used extensively now as a forecasting toolin China. The longer the existence of anomalous waterlevels, and the wider the anomaly, the larger theimpending earthquake. Changes in groundwatermay be related to tectonic strain induced by crustalmovements, or to the dilatancy phenomenon. Afeature associated with changes in groundwater level isthe change in the water’s radon content. Radon, espe-cially in mineral waters, will increase exponentiallyyears before an earthquake, and then decrease rapidlyto previous levels after the event. The increase inradon may reflect increased water movement throughcracks opened by dilatancy, permitting more of thisisotope to be incorporated into groundwaters.

There are also natural indicators of impendingearthquakes. Catfish in Japan become very activebefore earthquakes and even jump out of the water.In addition, local fish catches around Japan increase

significantly just before earthquake activity. Groundanimals will leave tunnels, stabled animals will becomerestless, and dogs will start barking hours before anearthquake. Snakes, weasels, and worms deserted theancient city of Helice, Greece, days before it was com-pletely destroyed by an earthquake in 373 BC. Beforethe Lisbon earthquake of 1755, worms crawled fromthe ground and covered the surface. The evacuation ofthe city of Haicheng, China, in February 1975 – justbefore it was destroyed by an earthquake – followedobservations of strange animal behavior: geese flewinto trees, pigs became aggressive, chickens refused toenter coops, and rats appeared drunk. In the San Fran-cisco zoo, animals have been found to group togetherby species about half-an-hour before an earthquake.Animal behavior is now being monitored daily in zoosand marine lands around San Francisco to incorporateanimal behavior into an earthquake prediction system.

It is not exactly clear why animal behavior isprognostic of earthquakes. A response to long waveelectromagnetic radiation has been given as one reasonfor animal panic. Animals may also be supersensitiveto vibrations and ultrasound generated by smallearthquakes preceding the main event. However,reptiles and birds that exhibit unusual behavior do nothear ultrasound. In addition, ultrasound from deepearthquakes is absorbed by rock, leaving only very lowfrequency sound to reach the surface. In this case,humans as well as animals should hear the sounds.Another reason for odd animal behavior may be ananimal’s extreme sensitivity to the smell of methane,known to leak from the ground during tremors.However, many burrowing animals that continuallytolerate natural methane as they dig also panic beforeearthquakes. Finally, a barrage of electrostatic particlesemanating from the ground may precede tremors.Animals with fur or feathers are very sensitive toelectrostatic charges, and the sudden increase inquantity of such particles may simply irritate them tothe point of panic or flight. Whatever the cause, scien-tists in China and other parts of the world now view themonitoring of animal behavior as the best means offorecasting the occurrence of major earthquakes.

Unusual weather changes may also be a harbingerof impending disaster. Mists close to the ground, aswell as glowing skies, have been reported beforeearthquakes. In some cases, the light appears toemanate from the ground as a flash of flame. While thisluminosity has often been dismissed, there is a new


theory that suggests charge separation occurs deep inthe Earth’s crust under the pressures that build upbefore earthquakes. Igneous rocks contain mineralswith paired oxygen atoms. Under stress, these break,with a negative ion remaining trapped in the crystallattice of the mineral and the positive charge flowing tothe Earth’s surface. The positive charge can combinewith an electron to give off infra-red light. Prior to thedevastating 26 January 2001 earthquake in Gujarat,India, increased infra-red emissions were detected bysatellite around the quake’s epicenter. Otherwise, thecharge builds up to 400 kV over a short distance atwhich point it ionizes air causing a luminous plasma –known as earthquake lights – to form. Alternatively,laboratory studies have shown that rock crushed underpressure gives off luminescence.

The prediction of earthquakes is still not an exactscience. China, because it faces such large death tollsfrom earthquakes, has developed the most advancedpredictive techniques. Five major earthquakes struckChina in the early- to mid-1970s. All but one was pre-dicted early enough to permit orderly evacuation ofpeople from buildings with minimal death or injury.The exception, the Tangshan earthquake of 28 July1976, took the lives of 250 000 people – the highestnumber of lives lost in an earthquake in two centuries.While the early warning signs of that earthquake wereobserved, they were too faint to arouse concern.

Randomness versus clustering

(Lighthill, 1996; Bak, 1997; Geller et al., 1997; Stein, 2003)

Foolproof prediction of large earthquakes is still notpossible. The consensus today is that earthquakesoccur randomly. If the cumulative frequencies ofearthquakes are plotted against magnitude for variousdepths through the Earth’s crust, the resulting curveshave a similar shape following a power law. This ischaracteristic of systems that are self-organized andcontain critical thresholds. For example, if you buildup a pile of dry sand, it will reach a critical slope anglewhere failure will occur. The scale of this failure israndom, from individual grains to massive sections ofthe sand pile. Massive failure occurs much less fre-quently than does the movement of individual grains.If the cumulative frequency of failure is plotted againstvolume, the resulting curve fits a power curve. Moreimportantly, small events often trigger much largerones, but there is no direct relationship between the

two. The movement of a single grain may triggeranother grain to move or the collapse of half the sandpile – or anything in between. This is characteristic ofearthquakes, which are often preceded by swarms ofsmaller earthquakes, but you can never predict whichone will trigger the ‘big’ one.

Most of the short-term precursors have only beenidentified as such following an earthquake. Thesealleged signals vary greatly from earthquake toearthquake and rarely can be detected as a regionalphenomenon surrounding the epicenter of an earth-quake. Crucially, the track record at predictingearthquakes has failed dismally. Even the success ofpredictions for the 1975 Haicheng earthquake can bechallenged. Official records now indicate that 1328were killed and 16 980 injured. For every precursorevent linked to an earthquake, there are hundreds thathave produced false alarms. For example, the crustaround Palmdale in Los Angeles swelled rapidly by asmuch as 20 cm in 3 years in the 1970s. After the expen-diture of millions of dollars on monitoring, the bulgewas found to be an artefact of measurement error. Thearea is still awaiting an earthquake. The VAN methodusing geo-electric activity has also failed at predictingearthquakes in Greece, because the signals beingmeasured are indiscernible from background noise.Each false prediction is just another statistical ‘nail inthe coffin’ of earthquake prediction.

The seismic gap concept is flawed. Many earth-quakes occur in swarms with the leading earthquakenot necessarily being the largest one. Earthquakesshould be viewed as chaotic geophysical phenomena,generated by a spectrum of seismic waves with varyingamplitudes and periods. If this is the case, thenearthquakes behave as white noise. One of the aspectsof such systems is that earthquakes recur at the samelocation rather than in areas that are more quiescent.For example, earthquake activity along the westernNorth American Plate boundary increased in the yearsleading up to the San Francisco earthquake of 1906.A predominance of magnitude 5 earthquakes in aregion usually indicates, several months in advance,the imminent occurrence of a much larger earthquake.These smaller earthquakes often ring the futureepicenter, forming what is known as a Mogi doughnut,named after the discoverer of the effect.

Some of these fine details about the spatial structureof earthquake swarms are being teased out. Earth-quakes are produced by changes in stress at an


epicenter. This stress is made up of two components,one parallel to the plane of a fault, and the other per-pendicular to it. If shear stress along the fault exceedsfrictional resistance, or the stress pressing against thefault from the sides is relaxed, the crust on both sidesof the fault will slip, producing an earthquake. Thetotal stress field is known as Coulomb stress. During amajor earthquake, Coulomb stress can alter by lessthan 3 bars. The stress that appears at the epicenter ofan earthquake on a fault line simply shifts elsewherealong the fault because stress must be conserved. Thebest example of this phenomenon occurred in Turkey.Here, the 17 August 1999 earthquake at Izmit thattook 25 000 lives was the twelfth to occur alongthe North Anatolian fault since 1939. Each eventsimply shifted the stress along the fault, triggering anearthquake at another location a few years later. TheIzmit earthquake reduced stress along 50 km of faultline, shifting it to Düzce 100 km to the east, andtowards Istanbul in the west. A magnitude 7.1 earth-quake subsequently affected Düzce in November1999. Fortunately, this stress shift was monitored andpublicized. Engineers pressured authorities in Düzceto close schools two months beforehand. Many of theseschools were destroyed in the November earthquake.The stress build-up towards Istanbul remains andexperts predict that the annual probability of anearthquake in the capital has now risen from1.9 per cent to 4.2 per cent. The probability of anearthquake occurring in this city before 2030 has risenfrom 48 per cent to 62 per cent.

Stress mapping has also been used to explain thepattern of earthquakes that followed the 28 June 1992Landers earthquake. Records show that there is a67 per cent chance of another large earthquake alongthe San Andreas Fault within a day of a 7.3 magnitudeearthquake. The Landers earthquake was of thismagnitude, and within three hours, a 6.5 magnitudeearthquake occurred at Big Bear, 40 km to the south-west. Coulomb stress had not only shifted alongthe San Andreas Fault but also laterally onto a parallelfault line. Seven years later, in 1999, a 7.1 magnitudeearthquake occurred at Hector Mine, 40 km to thenorth of Landers. Swarms of minor earthquakes thatfollowed the Landers earthquake also occupied areasof higher stress. Worldwide, 61 per cent of aftershocksgreater than magnitude 5 – that have occurred within250 km of the epicenter of a magnitude 7 or greaterearthquake – correspond to areas where Coulomb

stress has increased. Seismicity never completely shutsdown around the center of an earthquake. Earthquakeactivity simply increases away from the epicenter. Thismay explain why the number of earthquakes in theSan Francisco area decreased after 1906 relative to thepreceding 60 years. Earthquake activity has simplyshifted elsewhere along the San Andreas Fault. TheLoma Prieta earthquake of 1989 may have beenthe beginning of a return to the previous rate of activityin the San Francisco Bay area.

Prediction of volcanoes(Decker, 1976; Tazieff & Sabroux 1983; Smith, 1985)

While the first outburst of volcanic activity can now bepredicted, it is at present almost impossible to predictthe direction or intensity of activity that follows. Todate, only a handful of eruptions have been forecast,the earliest being the renewed activity of Kilauea,Hawaii, in November 1959. Most of these predictionsso far have been for eruptions that involved fluidmagma and did not threaten life. Volcanic eruptionsthat consist of viscous magmas, or that become explo-sive, still cannot be predicted. For example, while therenewed activity of Nevado del Ruiz, Colombia, inNovember 1985, was noted, the timing of the finaleruption could not be predicted, and 20 000 peoplelost their lives in the ensuing heated mudflows. Therehave also been some notable and costly false alarms forthese types of volcanoes. On 12 April 1976, the resi-dents of Guadeloupe in the West Indies were toldto evacuate because of an imminent eruption ofLa Soufrière volcano. Over 75 000 people heeded thewarning and were evacuated. They waited for 15 weeksuntil the volcano finally produced a small, veryharmless eruption on 8 July. It cost $US500 million tomaintain the evacuation and brought economic ruin tomany. Such predictions, while soundly based, only tendto weaken the credibility of subsequent warnings.

The techniques for predicting volcanic eruptions oractivity are just as varied as, but more technical than,those for predicting earthquakes. Precursors for volca-noes group into the following four categories: landdeformation, seismic activity, geomagnetic and geo-electric effects, and gases. Ground deformations aroundvolcanoes are due to the subterranean movements ofmolten magma. These movements can be vertical,lateral, or oblique. They manifest themselves at thesurface by tilting of the Earth’s surface, which can be


measured using tiltmeters. The movements can be fastor slow, positive or negative. Differences in movementcan occur over short distances. Large, sudden tiltsgenerally herald a violent eruption while slight tilts ofincreasing frequency foreshadow the movement ofmagma closer to the Earth’s surface. Tiltmeters on thenorthern side of Mt St Helens measured dramatic infla-tion at the rate of 0.5–1.5 m day-1 preceding theeventual eruption. Tiltmeters have also been used inHawaii to accurately forecast eruptions of Kilauea.

Studies of seismology have dominated volcanologysince the early 1900s. As magmas flow through sub-terranean channels, they apply stress to rocks, whichcan fracture and set off seismic activity. This appliesmostly to volcanoes characterized by fluid magma.Earthquakes set off by volcanoes differ from tectonicones in that they occur at depths of less than 10 km, andare low in magnitude. Volcanic tremors can be dividedinto two groups. The first group consists of prolonged,continuous volcanic vibrations due to resonance of fluidmagma flowing through cracks. Some explosive volca-noes, notably Mt St Helens, were beset with this type oftremor. These long period vibrations have proved highlysuccessful at forecasting impeding eruptions. The mostnotable success was the prediction twenty-four hours inadvance of the eruption of Mt Redoubt in Alaska on2 January 1990. The second group consists of spasmodicor regular vibrations, the frequency of which indicatesthe origin and nature of the magma. All forecast erup-tions of Kilauea, Hawaii, have been based upon tiltingand earthquake precursors. However, not all seismicactivity associated with volcanism can be used withcertainty to predict subsequent activity. While mosteruptions are usually, but not always, preceded byswarms of earthquakes, the presence of tremors canalso represent collapse of rock into emptying magmachannels, a process that may indicate cessation ofvolcanic activity. In a few cases, the subsidence of earthquake activity may also signal an eruption.

Active volcanoes are dominated by temporally andspatially variable geomagnetic fields. Volcanoes containa high content of ferromagnetic minerals, which can setup changes in the local magnetic field. Magnetization,however, is reduced by increasing temperature, vanish-ing completely above 600°C. The magnetic field of avolcano is thus reliant upon the proximity and temper-ature of molten magma near the surface. As hot magmabetween 200 and 600°C approaches the surface,the geomagnetic field should therefore decrease.

Magnetization can also be enhanced by increasingpressure and stress exerted by flowing magma as itapproaches the surface. This process is termedpiezomagnetism, and at present is being researchedextensively as a new technique for prediction. Geo-electric measurements involving the resistivity of thesubsurface layers of a volcano and the change in thetelluric currents can give an image of the behavior ofmagma at depth. Resistivity depends upon the nature ofthe rock, its water content, salinity, and temperature.Telluric currents depend upon geological structure,lithology, moisture content, and temperature. Atpresent, these methods are used to define the structureof the volcano, including the presence of naturalconduits, which may become the preferred pathwaysfor continued magma movement.

The analysis of gaseous constituents exhaled froma volcano constitutes one of the best techniquesfor understanding and forecasting eruptive activity.Unfortunately, while the technique is so informative, itis restricted by the need to chemically analyze the gasconstituents immediately they are vented from thevolcano. Many gas studies are still in their infancy,mainly because techniques for sampling and analyzinggases from venting volcanoes are still being developed.The most common gases vented by a volcano are H2O,CO2, SO2, H2, CO, CH4, COS, CS2, HCl, H2S, S2,HF, N2 and the rare gases helium, argon, xenon, neonand krypton. The chemical nature of these gasesdepends upon the maturity of the magma melt,because not all gas components have the same solubil-ity at a given pressure. The type of gas also dependsupon the amount of crustal material relative to originalmagma that is incorporated into the melt. If themagma is chemically stable, the partial pressure ratiosbetween various gases can give an indication of thepressure and temperature of the magma en route tothe surface. For example, the partial pressure ratio ofCO:CO2, HF:HCl and H2O:CO2 can all be usedas qualitative geothermometers. The first two ratiosincrease, and the last one decreases, as temperatureincreases. The ratios of CO:CO2, H2:H2O, andH2S:SO2 are sensitive to changes in the conditions thatcontrol the thermodynamic equilibrium of the gasphase, and can be used to distinguish betweenhydrothermal and magma flows. The ratios SO2:CO2and S:Cl, plus the absolute amount of HCl, increaseimmediately prior to eruptions, while the ratio He:CO2decreases. These changes result from the different


solubilities of individual gases and from the ascent offresh magma to the surface. Isotope studies of variousgases can also be used to delineate the different sourcesof material reaching the magma pool from below.

CONCLUDING COMMENTS

This chapter has shown that earthquakes and volca-noes are no longer poorly understood phenomenathat invoke fear, superstition, and pagan attempts atcontrol because of a lack of knowledge about theirorigin and behavior. While earthquakes and volca-noes can still kill a large number of people, and canoccur with extreme suddenness, our knowledge isreaching the stage where a significant number ofthese events can be predicted in space and time topermit measures to be taken to minimize the loss oflife. Since the widespread acceptance of continentaldrift theory, and the accurate measurement ofthousands of earthquake epicenters, the most likelylocations of future earthquake activity have beenmapped for the globe. Sound scientific observationsand instrumental measurements are defining thebest precursors of imminent tremors or eruptions.The success of such techniques in China in the 1970sis exemplary; the failure of authorities in Californiato install sensitive strain gauges along the SanAndreas Fault, which could have forecast the 1989San Francisco earthquake, is alarming. Complacencyor disbelief will always plague our attempts to predictboth earthquakes and volcanic eruptions. Both traitscan be justified because there is still a large randomcomponent to the occurrence of these hazards. TheTangshan earthquake of 1976 and the eruptions ofMt St Helens in 1980 and El Chichon in 1982 areonly a few of the recent examples supporting thisstatement.

The next two chapters will describe in more detailthe exact nature of, and phenomena associated with,earthquakes and volcanoes. As well, some of themore significant events in recorded history will bedescribed to reinforce the view that these twohazards are still amongst the most destructive naturalevents to afflict humanity.


Bak, P. 1997. How Nature Works: The Science of Self-OrganizedCriticality. Oxford University Press, Oxford.

Blong, R.J. 1984. Volcanic Hazards: A Sourcebook on the Effects ofEruptions. Academic Press, Sydney.

Bolt, B.A. 1993. Earthquakes. W.H. Freeman and Co., New York.Bolt, B.A., Horn, W.L., MacDonald, G.A., and Scott, R.F. 1975.

Geological Hazards. Springer-Verlag, Berlin.Coates, D.R. 1985. Geology and Society. Chapman and Hall, New

York.Decker, R.W. 1976. State of the art in volcano forecasting. In

Geophysical Predictions. National Academy of Science,Washington, pp. 47–57.

Denham, D. 1979. Earthquake hazard in Australia. In Heathcote,R.L. and Thom, B.G. (eds) Natural Hazards in Australia.Australian Academy of Science, Canberra, pp. 94–118.

Doyle, H.A., Everingham, I.B., and Sutton, D.J. 1968. Seismicityof the Australian continent. Journal of the Geological Society ofAustralia 15(2): 295–312.

Geller, R.J., Jackson, D.D., Kagan, Y.Y., and Mulargia, F. 1997.Earthquakes cannot be predicted. Science 275: 1616–1617.

Gribbin, J. 1978. The Climatic Threat. Fontana, Glasgow.Heki, K. 2003. Snow load and seasonal variation of earthquake

occurrence in Japan. Earth and Planetary Science Letters 207:159–164.

Hill, D.P., Pollitz, F., and Newhall, C. 2002. Earthquake-volcanointeractions. Physics Today 55(11): 41–47.

Hodgson, J.H. 1964. Earthquakes and Earth Structure. Prentice-Hall, Englewood Cliffs, New Jersey.

Holmes, A. 1965. Principles of Physical Geology. Nelson, London.Jones, N. 2003. Volcanic bombshell. New Scientist, 8 March: 32–37.Jones, P.D., Briffa, K.R., and Schweingruber, F.H. 1995. Tree-ring

evidence of the widespread effects of explosive volcaniceruptions. Geophysical Research Letters 22: 1333–1336.

Kisslinger, C. and Rikitake, T. (eds). 1985. Practical Approaches toEarthquake Prediction and Warning. Reidel, Berlin.

Lamb, H.H. 1972. Climate: Present, Past and Future. v. 1 & 2,Methuen, London.


Lamb, H.H. 1985. Volcanic loading: The dust veil index. CarbonDioxide Information Analysis Center Numeric Data PackageCollection, Dataset No. NDP013.DAT, Oak Ridge NationalLaboratory, Oak Ridge, Tennessee, USA.

Lighthill, J. (ed.) 1996. Critical Review of VAN: Earthquake Predictionfrom Seismic Electrical Signals. World Scientific Publishing,London.

Matthews, A.J., Barclay, J., Carn, S., Thompson, G., Alexander, J.,Herd, R., and Williams, C. 2002. Rainfall-induced volcanicactivity on Montserrat. Geophysical Research Letters 29(13)doi:10.1029/2002GL014863.

Muller, R.A. and Oberlander, T.M. 1984. Physical GeographyToday (3rd edn). Random House, New York.

National Earthquake Information Center 2001. Frequency ofoccurrence of earthquakes based on observations since 1900.United States Geological Survey, <http://neic.usgs.gov/neis/general/magnitude_intensity.html>

National Earthquake Information Center 2002. Measuring thesize of an earthquake. <http://neic.usgs.gov/neis/general/measure.html>


Pandey, O.P. and Negi, J.G. 1987. Global volcanism, biological massextinctions and the galactic vertical motion of the solar system.Geophysical Journal 89(3): 857–868.

Press, F. and Siever, R. 1986. Earth (4th edn). Freeman, New York. Ritchie, D. and Gates, A. E. 2001. Encyclopedia of Earthquakes and

Volcanoes. Facts on File, New York.Rikitake, T. 1976. Earthquake Prediction. Elsevier, Amsterdam. Simpson, D. W. and Richards, P. G. (eds) 1981. Earthquake

Prediction. American Geophysical Union, Washington. Smith, J. V. 1985. Protection of the human race against natural

hazards (asteroids, comets, volcanoes, earthquakes). Geology 13:675–678.

Stein, R.S. 2003. Earthquake conversations. Scientific American288: 60–67.

Tazieff, H. and Sabroux, J.C. (eds) 1983. Forecasting VolcanicEvents. Lange and Springer, Berlin.

Ward, P.L. 1976. Earthquake prediction. In Geophysical Predictions.National Academy of Science, Washington, pp. 37–46.


Wood, R. M. 1986. Earthquakes and Volcanoes. Mitchell Beazley,London.


C H A P T E R 1 0

TYPES OF SHOCK WAVES

(Hodgson, 1964; Holmes, 1965; Whittow, 1980)

Earthquakes are shock waves that are transmitted froman epicenter, which can extend from the surface to700 km beneath the Earth’s crust. Earthquakesgenerate a number of types of waves, illustrated inFigure 10.1. A primary or P-wave is a compressionalwave that spreads out from the center of the earth-quake. It consists of alternating compression anddilation, similar to waves produced by sound travelingthrough air. These waves can pass through gases,liquids and solids, and undergo refraction effects atboundaries between fluids and solids. P-waves can thustravel through the center of the earth; however, at thecore–mantle boundary they are refracted producingtwo shadow zones, each 3000 km wide, without anydetectable P-waves on the opposite side of the globe.

The second type of wave is a shear or S-wave, whichbehaves very much like the propagation of a wavedown a skipping rope that has been shaken up anddown. These waves travel 0.6 times slower thanprimary waves. While the velocity of a primary wavethrough Earth depends upon rock density and com-pressibility, the rate of travel of a shear wave dependsupon rock density and rigidity. Shear waves will travelthrough the mantle, but not through the Earth’s rigidcore. Thus, there is a shadow zone on the opposite sideof the Earth that does not record S-waves. This zone

overlaps completely the two shadow zones producedby refraction of P-waves at the core–mantle boundary.The spatial distribution and time separation betweenthe arrivals of P- and S-waves at a seismograph stationcan be used to determine the location and intensity ofan earthquake. Three stations receiving both P- and S-waves are necessary to determine the exact positionof any epicenter.

Earthquakes also generate several different long orL-waves trapped between the surface of the Earth andthe crustal layers lower down. These waves do nottransmit through the mantle or core, but spread slowlyoutwards from the epicenter along the surface of theglobe. Their energy is dissipated progressively fromthe focal point of the earthquake. One type of L-wave,the Rayleigh wave, behaves similarly to an ocean wave,while another type, the Love wave, literally slithersback and forth through the crust. This latter type ofwave is responsible for much of the damage witnessedduring earthquakes, and produces the shaking thatmakes it impossible to stand up. Long waves are theslowest of all seismic waves, taking about 20 minutes totravel a distance of 5000 km along the Earth’s surface.All three types of waves, because they travel differentlythrough dissimilar rock densities and states of matter,have been used to delineate the structure of the Earth’score, mantle, and crust.

Earthquakes andTsunami as Hazards

203Earthquakes and Tsunami as Hazards

P-waves only

Shadow zonefor S-waves

Shadow zonefor P-waves

Earth's crust

Only P- & S-wavestransmitted

L-, P- & S-waves transmittedEpicenter

L-wavesattenuated

S-waves nottransmitted

P-wavesrefracted

Mantle

Core

Compression Dilation

Crustalcross-section

P-wave

S-wave


amplitude

wavelength

Aerialview

lateralmovement

L-wave (I) Love wave

L-wave (II) Rayleigh wave

CompressionDilation


Fig. 10.1 Description of seismic wave types and their travel behavior through the Earth (adapted from Whittow, 1980).


SEISMIC RISK MAPS

(Keller, 1982; Giardini et al., 1999)

Maps can be constructed, based upon the historicalrecord of seismic waves, to give probability estimatesof the peak ground wave acceleration possible for anyregion over a fixed period. These accelerations areusually expressed as the 10 per cent probability over afifty-year period. In terms of recurrence interval, theyrepresent a 1:475 year event. Engineers can use suchmaps to assess the lateral forces acting on objects, suchas rigid buildings, to determine the risk of theircollapse during earthquakes. Figure 10.2 presents theseismic risk for the world’s landmasses based uponhistorical data. Obviously, the estimated risk is afunction of the historical record and spatial detectionof earthquakes. For example, much of Australia (seeFigure 10.3 for the location of all major placenamesmentioned in this chapter), which lies in the middle ofa continental plate, has a higher risk than Africa, whichcontains the Great Rift Valley and a historically seismicregion in Algeria. In Australia, seismic detection isbetter developed than in most of Africa. In NorthAmerica, publicity is overexaggerated regarding theseismic risk in southern California around the SanAndreas Fault zone. The northern Rocky Mountainsand the lower St Lawrence estuary are equally at risk.Large urban centers such as Salt Lake City, Seattle,Montreal, and Quebec City are all seismic. These latter

regions have a 20 per cent probability in 100 years of adestructive earthquake corresponding to an intensityof 8 or greater on the Mercalli scale, and 6.5 or greateron the Ms scale. The seismic risk in Turkey, Iran, andthe region bounding the Tibetan Plateau exceeds thatin North America. Seismic risk, however, does notequate to damage or loss of life. These latter factorsdepend upon the establishment of building codes byplanners and engineers, and their enforcement, espe-cially in urban areas, by government. For example, the17 August 1999 Izmit earthquake (Ms = 7.4) in Turkey,where codes are lax, killed 25 000 people compared tothe similarly sized, 17 October 1989, earthquake(Ms = 7.1) in San Francisco – where codes are strictlyenforced – which killed only 62 people.

EARTHQUAKE DISASTERS

General(Holmes, 1965; Cornell 1976; Whittow, 1980; Ritchie &Gates, 2001)

Each year on average, earthquakes kill 10 000 peopleand cause $US400 million property damage. In theperiod 1964–1978, over 445 000 people lost theirlives from earthquakes or earthquake-related hazards.Table 10.1 lists significant earthquakes in terms of loss oflife since 800 AD. As with tropical cyclones, there havebeen some notable events. In 1290, in China, two earth-quakes – in the Gulf of Chihli, and at Beijing – took

0.0 5.6

Peak ground acceleration m s-2

Fig. 10.2 Global seismic risk map. Shading defines peak horizontal accelerations with a probability of exceedence of 10 per cent in 50 years or arecurrence interval of 475 years (based on Giardini et al., 1999).


100 000 lives. The Shensi event in China on23 January 1556 claimed 830 000 lives. That earth-quake struck loess country, and caused the collapse ofthousands of human-made caves dug into cliffs andused as homes. Another earthquake at Kansu, China,

in 1920 killed 200 000 people. The Calcutta earth-quake of 11 October 1737 wiped out half the city,killing 300 000 people. In 1923, the Tokyo earthquake– which is discussed below in more detail – killed143 000 inhabitants. Italy has been struck many times

Gulf of Chihli

Alaska

MontrealAleutian IslandsGreat LakesSeattle

Salt Lake CityCrescent City

San FranciscoLos Angeles

HiloMexico City

Panama

San Diego

Chile

New Zealand

NewcastlePerth

Krakatau

Tambora

Rann of Kachchh

Tangshan

Tokyo

Sanriku

KobeOkinawa

Shensi

Messina

Izmit

CataniaNaples

Lisbon

North Sea

AntiguaPort Royal

Boston

Lituya Bay

Honolulu

Tahiti

ValparaisoPitcairn Is.

Quebec CityNewfoundland

St. Lawrence R.

Cadiz

Barbados

Chicxulub

Arica

Balleny Islands

Cape of Good Hope GisborneLake Taupo

Tasmania

Papua New Guinea

MindanaoPhilippines

Niigata

KamchatkaPeninsula

BejingKansu

Calcutta

Iran

English Channel

Santorini


Date Location Death toll Ms Magnitude (if known)23 January 1556 Shensi, China 830 00011 October 1737 Calcutta, India 300 00027 July 1976 Tangshan, China 255 000 89 August 1138 Aleppo, Syria 230 00022 May 1927 Xining, China 200 000 8.322 December 856 Damghan, Iran 200 00016 December 1920 Kansu, China 200 000 8.61 September 1923 Tokyo, Japan 143 000 8.3September 1290 Chihli, China 100 000November 1667 Shemakha, Caucasia 80 00018 November 1727 Tabriz, Iran 77 00028 December 1908 Messina, Italy 70 000 7.51 November 1755 Lisbon, Portugal 70 000 8.725 December 1932 Kansu, China 70 000 7.631 May 1970 Peru 66 000 7.821 June 1903 Silicia, Asia Minor 60 00011 January 1693 Sicily, Italy 60 00023 March 893 Ardabil, Iran 50 0004 February 1783 Calabria, Italy 50 00020 June 1990 Iran 50 000 7.730 May 1935 Quetta, Pakistan 30 000 7.5

Table 10.1 Major historical earthquakes, by death toll (United States Geological Survey, 2002)

by earthquakes, with ones near Naples and Catania,Sicily, in 1693, and at Messina on 29 December 1908,taking between 60 000 and 150 000 lives each. The SanFrancisco earthquake on 18 April 1906 was the worstAmerican earthquake, killing 498 people, with much ofthe city being destroyed by ensuing fires. Mostrecently, the 1976 Tangshan earthquake claimedapproximately 250 000 people, making it one of theworst earthquakes on record.

While many earthquakes stand out historically, verylittle research has been performed on the causes ofsome of the largest such as the Calcutta earthquake. Inmodern times, four to five events stand out becauseof their magnitude and the nature of the disaster.The 1906 San Francisco earthquake occurred in thevulnerable San Andreas Fault area of California andwiped out a major United States city in a country thatbelieved itself to be immune from such disasters. TheTokyo earthquake of 1923 totally destroyed the largestcity in Japan. The Chilean earthquake of 1960 markeda revival in large earthquake activity after a decade ofsmall events, and produced one of the largest wide-spread tsunami events in the Pacific Ocean. TheAlaskan earthquake of 1964 took few lives, but literallyshook the globe like a bell. The 1976 Tangshan earth-quake, notable for one of the most massive death tollsin the twentieth century, struck an area not renownedfor seismic activity at a time when the Chinesebelieved that they could predict major earthquakes.The latter event is also notable for the cloak of secrecysurrounding its aftermath. Not only did China refuseinternational aid, but it also released very few details ofthe event until five years afterwards. Two of theseevents, the 1964 Alaskan earthquake and the 1923Tokyo earthquake, are discussed in more detail below.

Alaskan earthquake of 27 March1964(Hansen, 1965; Whittow, 1980; Hays, 1981; Bolt, 1993)

The Alaskan earthquake struck on Good Friday whenschools and businesses were closed. It occurred in aseismically active zone running westward along thecoast of Alaska (Figure 10.4), parallel to the AleutianTrench, and south of the Aleutian Islands arc. Theearthquake was only one in a series of earthquakes inthe area beginning 10 million years ago in the latePliocene. In 1912 and 1934, earthquakes of magnitude

7.2 on the Ms scale occurred nearby. Other largeearthquakes had occurred in neighboring areas in1937, 1943, 1947 and 1958. These earthquakes weregenerated by the southward movement of Alaska overthe Pacific Plate at a shallow angle of 20 degrees.

The 1964 earthquake was one of the strongest to berecorded since seismograph records were taken. Longwave vibrations lasted 4–7 minutes. The shock waveswere so high in amplitude that seismographs ran off-scale all over the world, thus preventing accuratemeasurement of the earthquake intensity (reaching8.6 and 9.2 on the Ms and MW scales, respectively).The earthquake rang the earth like a bell, and set upseiching in the Great Lakes of America, 5000 km away.Water levels fluctuated in wells in South Africa on theother side of the globe. The ground motion was sosevere that the tops of trees were snapped off. Earthdisplacements covered a distance of 800 km along theDaniel fault system parallel to the Alaskan coastline.Over 1200 aftershocks were recorded along this fault.Tsunami exceeded 10 m in height along the Alaskancoast and swept across the Pacific Ocean at intervals ofone hour. The west coast of North America as far southas San Diego was damaged. In Crescent City, Cali-fornia, over 30 city blocks were flooded, resulting in$US7 million damage.

Maximum uplift and downthrow amounted to 14.4and 3 m, respectively, the greatest deformation from anearthquake yet measured. In some locations, individualfault scarps measured 6 m in relief. Coastal and inlandcliffs in south Alaska were badly affected by landslides.Land subsidence occurred in many places underlain byclays, and in some areas liquefaction was noted (Fig-ure 10.5). Liquefaction is a process whereby firm butwater-saturated material can be shaken and turned intoa liquid that then flows downslope under the effectof gravity. About 60 per cent of the $US500 milliondamage was caused by ground failures. Five landslidescaused $US50 million damage in Anchorage alone.Most of the ground failure in this city occurred in theBootlegger Cove Clay, a glacial estuarine–marinedeposit with low shear strength, high water contentand high sensitivity to vibration. In the ports ofSeward, Whittier, and Valdez, the docks and ware-houses sank into the sea because of flow failures inmarine sediments. Over the next nine hours, each ofthese towns was overwhelmed by tsunami 7–10 m inheight.



The Californian earthquake hazard(Keller, 1982; Oakeshott, 1983; Wood, 1986; Palm &Hodgson, 1992)

Of all seismic areas in the world, the San Andreas Faultrunning through southern California is one of the mostpotentially hazardous zones for earthquakes. Its manysubsidiary branches pass by, or through, a region cur-rently inhabited by an estimated 23 000 000 people(Figure 10.6). The most devastating earthquake in thisregion, in terms of property loss, was the San Franciscoearthquake of 18 April 1906 that killed 498 – andmaybe as many as 3000 – people. This quake was esti-mated at 8.25 on the Ms scale, and had a maximumvertical and lateral displacement of 2.5 m and 6 m,respectively. Fires, which broke out following the

-180˚ -160˚ -140˚ 65˚

60˚

55˚

50˚

45˚

A

24

0-2-4

1 2 3 4 5 6

1965

Latouche Is.

CValdez

Whittier

SewardChenega

Montague Is.

Prince William Sound

0 100 200 km

B

Gulf of Alaska

Trinity Islands

Subsidence

Cape Yakataga

Uplift

KenaiPeninsula

1964 epicentre

Seismic activity since 1938

Bering Sea

ValdezAnchorage

Kodiak Island

Mw = 8.7Gulf of Alaska

Hours after earthquake

Cape Yakataga

Hei

ght (

m)

1938Mw = 8.21946

Mw = 7.4

1964Mw = 9.2

1957Mw = 9.1

1986Mw = 7.7

Aleutian Islands

Continental Shelfedge

Tsunami wavecr

est

Source

ofts

unam

i

25 min

50m

in

Fig. 10.4 Earthquake and tsunami characteristics of the Alaskan earthquake of 27 March 1964: A) Location of seismic activity since 1938; B) Gulfof Alaska land deformation caused by the earthquake and theorized open-Pacific tsunami wave front; C) Detail of Prince William Sound(based on Van Dorn, 1964; Pararas-Carayannis, 1998; and Johnson, 1999).

Fig. 10.5 Head scarp of one of numerous landslides that developedin Anchorage, Alaska, during the 27 March 1964 earth-quake. Sand and silt lenses in the Bootlegger CoveFormation liquefied causing loss of soil strength (photo-graph courtesy of the United States Geological Survey).


earthquake, were responsible for most of the damage.This earthquake was not unique. There have beenearthquakes of similar size in recent geological times,with the earthquake at Fort Tejon on 9 January 1857being the largest yet recorded in southern California.The Owen Valley tremor on 26 March 1872 is consid-ered the third greatest earthquake to strike Californiain recent times. It produced surface faulting extendingup to 150 km, with a maximum vertical displacement of4 m. Other noteworthy earthquakes occurred in 1769south-east of Los Angeles; on 10 June 1836 in the EastBay area of San Francisco; in June 1838 at Santa Clara;on 8 October 1865 in the Santa Cruz Mountains;on 21 October 1868 south of San Francisco; and on24 April 1890 at Chittenden Pass.

The San Andreas Fault has been active since theJurassic, 135 million years ago, and has accumulated550 km of displacement. During the Pleistocene (last2 million years) horizontal displacement has amountedto 16 km. Dating of charcoal deposited over the past fewthousand years over various fault lines indicates thatmajor earthquakes occur at intervals of 50–300 years.The Pacific Plate is moving slowly north-west relative to

the North American Plate at the rate of 5 cm yr-1. Mostof the movement is sideways; but in a few places, thereis a vertical component on the North American Plateside. In most places, this movement is continual;however, in some places, the plates become lockedtogether and the movement is released in one suddenjerk. The San Francisco earthquake gave rise to thegreatest lateral displacement – 6 m – yet measured alongthe San Andreas Fault. Even where slow creep takesplace, earthquakes can occur.

In southern California around Los Angeles, therehas been no major movement since the 1857 earth-quake. The most recent earthquake, at Northbridgeon 17 January 1994, measured only 6.7 on the Msscale. Based upon geological evidence, this areatogether with the San Francisco area is overdue for amajor earthquake greater than 8 on the Ms scale.Both cities exist in seismic gap areas, surrounded byzones that have experienced recent tremors in thepast 30 years. At present, the probability of an earth-quake occurring in these gaps is 50 per cent. One ofthe more overwhelming responses to the earthquakehazard in California is neglect. In the Parkfield area

Los Angeles Faults

San Andreas

San Gabriel

Big PIne

Santa Ynez

San Cayelano

Red Mt.Oak Ridge

Simi

SantaSusana San

Fernando

MalibuCoast Newport

Inglewood

Santa Cruz Island

0 10 20 30 km

Hayward

CalaverasPilarcitos

San Francisco

San Andreas

Garlock

San Gabriel

Big Pine

Los Angeles

BanningElsinore

San Diego

Imperial

San Jacinto

N

0 50 100 150 200 250 km

Ms Magnitude

> 8.07.0–7.96.0-6.9

19801922

18921906

18681838

1989

1952

19831857

1927

1812

1971

18121992

19681987

1979

1940

1994

2003

1873

18361865

1872

19251918

17691933

1890

1892

Mission Creek

Fig. 10.6 Historically active fault systems of southern California and noteworthy earthquake epicenters for the last two centuries (adapted fromOakeshott, 1983 and Robinson, 2002).

of Los Angeles, where an earthquake with a90 per cent probability of occurrence was forecastbetween 1985 and 1993, less than 20 per cent of thepopulation purchased earthquake insurance andfewer than 17 per cent made their homes more earth-quake resistant, despite widespread publicity aboutthe likelihood of the event. Following the LomaPrieta earthquake of 1989, the increase in insurancein the most affected area was only 11 per cent, stillleaving over 50 per cent of the population uninsured.The major populated segments of California lie in oneof the world’s most unstable zones for earthquakesand are poorly prepared for the consequences ofthese events.

A major earthquake in southern California would becatastrophic. Many past earthquakes in southernCalifornia produced little damage because theyoccurred in unpopulated areas. Many of these areasare now densely settled. The San Fernando earthquakeof 1971, with an Ms magnitude of only 6.6, caused thecollapse of public buildings, including two hospitalsbuilt to earthquake building code specifications.Sections of freeway overpasses built to the samestandards also failed to withstand the shock waves.Subsequent to that earthquake, many public buildingsand roadways were reinforced or rebuilt. Buildingcodes have been tightened to account for the type ofdamage produced by that relatively small earthquake.Two dams in the area almost failed. One, the LowerVan Norman earth-filled dam, came within inches ofcollapsing and flooding 80 000 people in the lowerSan Fernando Valley. There are 226 dams in the SanFrancisco region alone with over 500 000 people livingdownstream. Many of these dams, together with othersin southern California, have subsequently beenstrengthened. Some of these remedial measures havenot worked. The 17 October 1989 Loma Prieta earth-quake in San Francisco, measuring 7.1 on the Ms scale,collapsed a large section of the upper deck of theNimitz Freeway in Oakland, even though the supportsholding up this expressway were reinforced in the1970s (Figure 10.7). Much of the housing constructionin the San Francisco area since 1906 has occurred onunconsolidated landfill that could liquefy or amplifyshock waves up to ten times, causing total destructionof buildings. This additional hazard was also broughthome in the October 1989 earthquake, which causedthe complete collapse of many houses in the Marinadistrict for this reason.


Not all of the branches of the San Andreas Faulthave been mapped. For instance, the Diablo Canyonnuclear plant north of Los Angeles was built to with-stand an earthquake of 8.5 on the Ms scale originatingfrom a distant fault. However, during construction amajor and active fault was discovered offshore within5 km of the plant. While the plant’s design was altered,there is still doubt whether or not it could withstand amajor earthquake along the nearby fault. It isinevitable that the next major earthquake in southernCalifornia above 7 on the Ms scale will bring untolddestruction. It has the potential to destroy largesections of major cities and to start numerous fires,which could turn into urban conflagrations similar tothe 1906 earthquake. Numerous dams could failcausing major flooding and loss of life in downstreamvalleys, and the process of ground liquefaction, espe-cially in the San Francisco area, could flatten suburbsin most of that city.

Fig. 10.7 Aerial view of collapsed sections of the Cypress viaduct ofInterstate Highway 880 (H.G. Wilshire, U.S. GeologicalSurvey, <http://wrgis.wr.usgs.gov/dds/dds-29/>).


The Japanese earthquake hazard(Hodgson, 1964; Holmes, 1965; Hadfield, 1992; EQE,1995; Bardet et al., 1997)

The most notable earthquake in Japan in the lastcentury occurred on 1 September 1923 in Sagami Bay,50 km south-west of Tokyo (Figure 10.8). Known asthe Great Kanto Earthquake, the crustal movementstwisted the mainland in a clockwise direction, witha maximum horizontal displacement of 4.5 m and avertical drop of 2 m. Until the Alaskan earthquake,these were some of the greatest crustal shiftsrecorded. Within the bay itself, changes in elevationwere much larger owing to faulting, compaction, andsubmarine landslides. Measurements after the earth-quake indicated that parts of the bay had deepened by100–200 m with a maximum displacement of 400 m.The Tokyo earthquake immediately collapsed over ahalf million buildings and threw up a tsunami 11 m inheight around the sides of Sagami Bay. However,these events were not responsible for the ultimatelylarge death toll, because many of the collapsed build-ings consisted of lightweight materials. The deathsresulted from the fires that immediately broke out inthe cities of Tokyo and Yokohama, and raged for threedays, destroying over 50 per cent of both cities. Thefires rate as one of the major urban conflagrations inhistory – on the same scale as fires that destroyedLondon in 1666, Chicago in 1871, Moscow during the

Napoleonic Wars in 1812, San Francisco during the1906 earthquake, and Dresden and Hamburg duringthe Second World War. The outbreaks of fire wereminor to begin with, but they occurred throughoutboth cities in large numbers, mainly because theearthquake occurred at lunchtime when many openfires were being used for cooking. Within half-an-hour, over 200 small fires were burning in Yokohamaand 136 were raging in Tokyo. The rubble of collapsedbuildings in the streets and cracked water mainscompounded the difficulty in dealing with the fires.Firefighters could neither get to the fires norobtain a reliable water source to extinguish them.Once the fires raged, the cluttered streets hamperedevacuation. In Yokohama, 40 000 people fled to anopen area around the Military Clothing Depot, whichwas subsequently engulfed in flames. All these peopledied from the searing heat or suffocation caused bythe withdrawal of oxygen.

The spread of fire was aided initially by strongtropical cyclone winds. In the chapter on tropicalstorms, it was pointed out that the earthquake wasprobably triggered by the arrival of a tropical cyclonethe previous day. The calculated change in load on theEarth’s crust because of the pressure drop, andthe increased weight of water from the storm surge,was 10 million tonnes km-2. The fires were exacerbatedby the development of strong winds generated by thecyclone in the lee of surrounding mountains. These

0 10 20 30 40 50 km

Fault displacementDirection of horizontalmovement (m)

Direction ofverticalmovement

N

BozoPeninsula

Yokohama

Tokyo

Sagami Bay

1.5

2.7

2.0

Epicenter

2.7IdzuPeninsula

2.7

2.0

3.6

2.0

Fig. 10.8 Distribution of faulting for the Tokyo earthquake, 1 September 1923 (from Holmes, 1965).

föhn winds were desiccating. All descriptions of thefires allude to the strong winds, which swept flames outof control.

Since 1923, the Japanese economy has grown to bea crucial component of the global economy. The worldhas been waiting for the next large Tokyo earthquake.The death toll will be as high as 200 000 with up to3.5 million people left homeless. The 1923 earth-quake cost Japan 37.5 per cent of its gross nationalproduct (GNP). In 1987, a similar sized event wouldhave cost 23 per cent of GNP or $US500–850 billion.This is more than Japan’s entire overseas capitalinvestment. The global insurance industry could coveronly a small percentage of this loss and Japan wouldbegin retrieving international investments to pay forreconstruction. Such a move would trigger a globalrecession that could last over five years. On 17 January1995, a strong earthquake, measuring 7.2 on the Msscale, struck the Kobe–Osaka region, the second mostpopulated area in Japan. The earthquake – nowknown as the Great Hanshin Earthquake – killed 5500people, injured 35 000 others and destroyed or badlydamaged 180 000 buildings. Liquefaction caused thedockyards to sink below sea level and displacedsupports for bridges. Shearing of inadequatelydesigned or poorly constructed columns collapsedexpressways, railway lines, and commercial premises.Building codes were designed to prevent structurescrumbling during such an earthquake; yet, 4 per centof buildings constructed to those codes sufferedextensive damage and 2.5 per cent collapsed. Over25 per cent of buildings higher than eight storiessuffered severe damage. Fires consumed thecrowded, older residential section of the city. Fewerthan 5 per cent of homes were insured. Damagetotalled $US100–150 billion or 13 per cent ofthe national budget, a staggering amount given thelimited size of the earthquake. More significantly,the Nikkei index dropped 5.6 per cent, wipingthe same amount off the share market. The fall of theshare market spread globally. The Japanese economy,already in recession, stagnated further. Bureaucraticrigidity and archaic chains of command paralyzedrelief efforts for several days. The response by theSelf-Defense Forces was tardy, hampered by politics,and ineffectual. Everything about the Kobe earth-quake indicates that a bigger event hitting Tokyo, as itinevitably must in the near future, will cause an evenworse disaster scenario.

LIQUEFACTION ORTHIXOTROPY

(Hansen, 1965; Scheidegger, 1975; Hays, 1981; Lomnitz,1988; Bolt, 1993; Pinter et al., 2003)

Liquefaction is a process whereby relatively firm clay-free sands and silts can become liquefied and flow as afluid. This can lead to the sinking of objects that theyhave been supporting. The mechanics of liquefactionwill be discussed in more detail in Chapter 13. Fornow, liquefaction will be discussed qualitatively as anearthquake-induced hazard. The weight of a soil issupported by the grains, which touch each other.Below the watertable, spaces or voids in the soil fillwith water. As long as the soil weight is borne by grain-to-grain contacts, the soil behaves as a rigid solid.However, if the pressure on water in the voidsincreases to the point that it equals or exceeds theweight of the soil, then individual grains may no longerbe in contact with each other. At this point, the soil par-ticles are suspended in a dense slurry that behaves as afluid. It is this process, whereby a rigid soil becomesa liquid, that is termed liquefaction or thixotropy.

Liquefaction can be induced by earthquake-generated shear or compressional waves. Shear wavesdistort the granular structure of soil causing some voidspaces to collapse. These collapses suddenly transfer theground-bearing load of the sediment from grain-to-grain contacts to pore water. Compressional wavesincrease the pore water pressure with each passage of ashock wave. If pore water pressure does not have timeto return to normal before the arrival of the next shockwave, then the pore water pressure increases incremen-tally with the passage of each wave (Figure 10.9). Theprocess can be illustrated quite easily by going to thewettest part of the beach foreshore at low tide. On flatslopes, one can usually run easily along the sand. It iseven possible to drive vehicles along this zone on somefine sand beaches. However, if you tap the sand in thiszone slowly with your foot, it will suddenly become softand mushy, and spread outwards. If the tapping is lightand continual, the bearing strength of the sand willdecrease and your foot will sink into the sand. All of thedamaging earthquakes towards the end of the twentiethcentury – Loma Prieta, Northbridge, Kobe, and Izmit –were characterized by liquefaction.

Pore water pressure can return to normal only ifthere is free movement of water through the soil.Smaller particles inhibit water movement because



capillary water tension between particles is greater.This cohesion is so large with very fine silt and claysthat the pore water pressure cannot reach the point ofliquefaction. Medium- to fine-grained sands satisfyboth criteria for liquefaction to occur. They are fineenough to inhibit rapid internal water movement, yetcoarse enough that capillary cohesion is no longerrelevant. Since these sand sizes are common inrecently deposited river or marine sediments (in thelast 10 000 years), liquefaction is an almost universalfeature of earthquakes. The younger and looser thesediment deposit and the higher the watertable, themore susceptible the deposit is to liquefaction.

The build-up of pore water pressure can cause waterto vent from fissures in the ground resulting in surfacesand boils or mud fountains. Following the 7.6 Msearthquake on 26 January 2001 in the Rann ofKachchh, India, the magnitude of groundwaterevulsion was so great that it reactivated ancient riverchannels and formed shallow lakes over a 60 000 km2

area. If pore water pressure approaches the weight ofthe overlying soil, then the soil at depth starts to behavelike a fluid. Objects of any density will easily sink intothis quicksand because the subsurface has virtually nobearing strength. Thus, liquefaction causes three typesof failure: lateral spreads, flow failures, and loss ofbearing strength. Lateral spreads involve the lateralmovement of large blocks of soil as the result of lique-faction in subsurface layers, due to ground shaking

during earthquakes. It occurs on slope angles rangingfrom 0.3–3.0 degrees, with horizontal movements of3–5 m. The longer the duration of ground shaking, thegreater the lateral movement and the more the surfacelayers will tend to fracture, forming fissures and scarps.During the 1964 Alaskan earthquake, floodplain sedi-ments underwent this process, and lateral spreadingdestroyed virtually every bridge in the affected zone.Flow failures involve fluidized soils moving downslope,either as a slurry or as surface blocks overtop slurry atdepth. These flows take place on slopes greater than3 degrees. Some of these flows can move tens of kilo-metres at speeds in excess of 15 km hr-1. Morecommonly, these flows can occur in marine sedimentunder water. Loss of bearing strength simply representsthe transference of the ground load from grain-to-graincontacts to the pore water. Any object that has beensitting on this material and using it for support is thenliable to collapse or sink. At present, the exact condi-tions responsible for earthquake-induced liquefactionhave not been determined. Evidence indicates that theacceleration and number of shock waves control theprocess; however, prior history of the sediment cannotbe ignored. Only mapping of subsurface material inearthquake-prone areas will delineate the area of riskfrom this process during earthquakes.

The effects of earthquake-induced liquefaction canbe spectacular. For example, the Japanese earthquakeat Niigata of 16 June 1964 registered 7.5 in magnitude

Pre-earthquake pressure

Point ofliquefaction

Earthquake-induced shear stress (P-wave)

Compression

Dilation

Pore water pressure

Time

Liquid limit

Fig. 10.9 Cumulative effect of compressional P-waves upon pore water pressure leading to liquefaction (after Bolt et al., 1975).

faction will be the major cause of property damage.The intensity of seismic waves does not have to begreat to induce liquefaction – as illustrated in Figure10.10, where the bearing capacity of soils underlying thebuildings failed but the buildings remained virtuallyundamaged. However, the intensity of compressionalshock waves triggering liquefaction may be consider-ably lower than the values experienced at Niigata.For instance, the Meckering, Western Australia,earthquake of 1968 induced liquefaction of sandsunderlying many parts of the expressway system inPerth, 100 km away from the epicenter. At Perth, theearthquake registered only 3 on the Mercalli scale. Notonly did the edges of roadways sink 4 cm within hoursof this seismic event, but the problem also persisted forweeks afterwards, eventually resulting in widespreaddamage to road verges.

The liquefaction process usually involves watermovement in clay-free sand and silt deposits. Inthe loess deposits of China, it can involve air as thesuspending medium. The Kansu earthquake of16 December 1920 broke down the shearing resistancebetween soil particles in loess hills. The low perme-ability of the material prevented air from escapingfrom the loess, which subsequently liquefied; thisresulted in large landslides that swept as flow failuresover numerous towns and villages, killing over 200 000people. The 1556 earthquake in Shensi Province mayhave generated similar loess failure, giving rise to adeath toll in excess of 800 000.

Waterlogged muds can also amplify the longperiod components of shock waves. Earthquakes atNewcastle, Australia, in 1989 and in Mexico in 1985provided clear evidence of this process at work. AtNewcastle, the seismic wave from a weak tremorregistering only 5.5 on the Ms scale amplified in riverclays beneath the city. As a result, what was really onlya small tremor still damaged over 10 000 buildings,creating a damage bill of $A1000 million. With theMexican earthquake, damage was even more exten-sive. The Aztecs founded Mexico City in 1325 ADupon the deep, weathered ash deposits of a formerlake bed, Lake Texcoco, later drained by theSpaniards. These deposits have weathered to formmontmorillonitic clays with the ability to absorb waterinto their internal structure, thus increasing theirwater content by 350–400 per cent. From a geo-physical point of view, these clays can be effectivelymodelled as water, even though they have a solid


on the Ms scale. The earthquake was located 60 kmnorth of the city and, at Niigata, produced maximumground accelerations of 0.16 per cent g – values thatcannot be considered excessive. The city itself hadexpanded since the Second World War onto reclaimedland on the Shinano River floodplain. The earthquakeshock waves did not destroy buildings. Rather, they liq-uefied unconsolidated, floodplain sediments, suddenlyreducing their bearing strength. Large apartmentblocks toppled over or sank undamaged into the groundat all angles (Figure 10.10). Afterwards, many buildingswere jacked back upright, underpinned with supports,and reused.

The San Francisco earthquake of 1906 producedsome evidence of liquefaction around the city. Thebreaking of water mains, which hampered fire sup-pression, was generated by liquefaction-inducedlateral spreads, mainly near the harbor. Much of thedevelopment of the city since has occurred on top ofestuarine muds with high water content. In addition,much of the harborside construction in southernCalifornia occurs on landfill, which can easily undergoliquefaction if water-saturated. The San Fernandoearthquake in 1971 produced liquefaction in manysoils and caused landslides on unstable slopes. Similarliquefaction problems were experienced during theOctober 1989 San Francisco earthquake. When a largeearthquake strikes this city, as it inevitably will, lique-

Fig. 10.10 Liquefaction following the 16 June 1964, Niigata,Japan, earthquake. These workers’ apartments sufferedlittle damage from the earthquake, but toppled overbecause of liquefaction of unevenly distributed sand andsilt lenses on the Shinano floodplain. Horizontalaccelerations were as low as 0.16 per cent g. Thebuildings subsequently were jacked up and re-occupied(photograph courtesy of the United States GeologicalSurvey, Catalogue of Disasters #B64F16-003).

appearance. Today, Mexico City has grown to apopulation in excess of 12 million. Two earthquakesstruck Mexico 400 km south-west of the city, offshorefrom Rio Balsas, on 19–20 September 1985. The firstearthquake measured 8.1 on the Ms scale, while thesecond aftershock, 36 hours later, measured 7.5.Jalisco, Michoacan, and Guerrero, the three coastalstates nearest the epicenter, suffered extensivedamage, with the loss of about 10 000 people. Damagedecreased sharply away from the epicenter. In MexicoCity, peak accelerations corresponded to a calculatedreturn period of 1 in 50 years. However, when theshock waves traversed the lake beds underlyingMexico City, they were amplified sixfold by resonance,such that the underlying sediments shook like a bowlof jelly under complex seiching. Surface gravity waves,akin to waves on the surface of an ocean, crisscrossedthe basin for up to three minutes. These waves hadwavelengths of about 50 m, the same as the basedimensions of many buildings, and caused thedestruction of 10 per cent of the buildings in the citycenter, covering an area of 30 km2. Of the 400multistorey buildings wrecked, all but 2 per cent werebetween 6 and 18 stories high. Many had becomepliable with age and had developed a fundamentalresonance period of two seconds, the same as thatobtained for peak accelerations of surface seismicwaves. The buildings folded like a stack of cards. Closeto 20 000 people were killed in Mexico City and thedamage bill amounted to $US4.1 billion. The Mexicanearthquake illustrates a fundamental lack in ourknowledge regarding the nature of ground motion insedimentary basins containing high water content, oralternating dry and wet sediment layers.

TSUNAMI

Description(Wiegel, 1964; Shepard, 1977; Myles, 1985;Bryant, 2001)

Tsunami (both the singular and plural forms of theword are the same) are water wave phenomena gener-ated by the shock waves associated with seismicactivity, explosive volcanism, or submarine landslides.These shock waves can be transmitted through oceans,lakes, or reservoirs. The term tsunami is Japanese andmeans harbor (tsu) wave (nami), because such wavesoften develop as resonant phenomena in harbors afteroffshore earthquakes. Because some tsunami are

preceded by a down-drawing of water followed by arapid surge over the space of 30–120 minutes, thename tidal waves has been used to describe suchphenomena. However tsunami have nothing to dowith tides and this usage is now discouraged. Tsunamihave a wavelength, a period, and a deep-water height;and can undergo shoaling, refraction and diffraction.Most tsunami originate from submarine seismicdisturbances. The displacement of the Earth’s crust byseveral metres during underwater earthquakes maycover tens of thousands of square kilometres, andimpart tremendous potential energy to the overlyingwater. Tsunami are rare events, in that not all sub-marine earthquakes generate them. Between 1861 and1948, only 124 tsunami were recorded from 15 000earthquakes. Along the west coast of South America,1098 offshore earthquakes have generated only 19tsunami. This low frequency of occurrence maysimply reflect the fact that most tsunami are small inamplitude and go unnoticed, or the fact that mostearthquake-induced tsunami require a shallow focusseismic event greater than 6.5 on the Ms scale.

Submarine earthquakes have the potential togenerate landslides along the steep continental slopethat flanks most coastlines. In addition, steep slopesexist on the sides of ocean trenches and around thethousands of ocean volcanoes, seamounts, atolls, andguyots on the seabed. Because such events are difficultto detect, submarine landslides are considered a minorcause of tsunami. However, small earthquakes havegenerated tsunami and one mechanism includessubsequent tsunami landslides. A large submarinelandslide or even the coalescence of many smallerslides has the potential to displace a large volume ofwater. Geologically, submarine slides involving up totwenty thousand cubic kilometres of material havebeen mapped. Tsunami arising from these eventswould be much larger than earthquake-induced waves.Only in the last thirty years has coastal evidence forthese mega-tsunami been uncovered.

Tsunami can also have a volcanic origin. Of 92 doc-umented cases of tsunami generated by volcanoes,16.5 per cent resulted from tectonic earthquakes asso-ciated with the eruption, 20 per cent from pyroclastic(ash) flows or surges hitting the ocean, and 14 per centfrom submarine eruptions. Only 7 per cent resultedfrom the collapse of the volcano and subsequentcaldera formation. Landslides or avalanches of coldrock accounted for 5 per cent; avalanches of hot


material, 4.5 per cent; lahars (mud flows), 3 per cent;atmospheric shock waves, 3 per cent; and lavaavalanching into the sea, 1 per cent. About 25 per centhad no discernible origin, but probably were producedby submerged volcanic eruptions. The eruptions ofKrakatau in 1883 and Santorini in 1470 BC wereresponsible for the largest and most significantvolcano-induced tsunami.

There has been no historical occurrence of tsunamiproduced by a meteorite impact with the ocean.However, this does not mean that they are an inconse-quential threat. Stony meteorites as small as 300 m indiameter can generate tsunami over 2 m in height thatcan devastate coastlines within a 1000 km radius of theimpact site. The probability of such an event occurringin the next fifty years is just under 1 per cent. Oneof the largest impact-induced tsunami occurred atChicxulub, Mexico, sixty-five million years ago at theCretaceous–Tertiary boundary. While the impact wasresponsible for the extinction of the dinosaurs, theresulting tsunami swept hundreds of kilometers inlandaround the shore of the early Gulf of Mexico. Impactevents are ongoing. Astronomers have compiledevidence that a large comet encroached upon the innersolar system and broke up within the last fourteenthousand years. The Earth has repetitively intersecteddebris and fragments from this comet. However, theseencounters have been clustered in time. Earlier civil-izations in the Middle East were possibly destroyed byone such impact around 2350 BC. The last rendezvousoccurred as recently as 1500 AD; however, it happenedin the southern hemisphere where historical recordsdid not exist at the time. Only in the last decade hasevidence become available to show that the Australiancoastline preserves the signature of mega-tsunamifrom this latest impact event.

A partial geographical distribution of tsunami isprovided in Table 10.2, while the location of tsunami inthe Pacific since 47 BC are plotted in Figure 10.11.The worst regions for tsunami occur along the stretchof islands from Indonesia through to Japan and thecoast of the Russia. Most tsunami in this region origi-nate locally. Of 104 damaging tsunami in the pastcentury, only nine have caused damage beyond theirsource regions. In the Pacific, these originate betweenJapan and the Kamchatka Peninsula inclusive, alongthe Aleutian Islands and south Alaskan coast, and alongthe coast of South America. For example, the Chileanearthquake in 1960 affected most of the coastline


of the Pacific Ocean. Here, earthquakes greater inmagnitude than 8.2 affect the entire Pacific Oceanand have a probability of occurrence of once every

Location PercentAtlantic east coast 1.6Mediterranean 10.1Bay of Bengal 0.8East Indies 20.3Pacific Ocean 25.4Japan-Russia 18.6Pacific east coast 8.9Caribbean 13.8Atlantic west coast 0.4

Table 10.2 Percentage of tsunami in the world’s oceans and seas.

90˚ 150˚ -150˚ -90˚Longitude

60˚

40˚

20˚

0˚

-20˚

-40˚

-60˚

Latit

ude

A

B

90˚ 150˚ -150˚ -90˚Longitude

60˚

40˚

20˚

0˚

-20˚

-40˚

-60˚

Latit

ude

Fig. 10.11 A) Spatial distribution of tsunami since 47 BC;B) source areas for Pacific-wide events (adapted fromNational Geophysical Data Center and World DataCenter A for Solid Earth Geophysics, 1984).

25 years. The high number of tsunami in Oceaniaincludes those at Hawaii, which receives about85 per cent of all tsunami affecting the Pacific. In fact,Japan and Hawaii are the two most tsunami-proneregions in the world, each accounting for 19 per centof all tsunami measured. For this reason, Hawaiimonitors closely all tsunami source regions in thePacific.

Peru and Chile are most affected by local earth-quakes. This coastline is also the only source area toaffect the south-west Pacific region. Japan is mostlyaffected by locally generated tsunami, and directs itsattention to predicting their occurrence locally. Since684 AD, in the Japan region, 73 tsunami have resultedin over 200 000 deaths. Indonesia and the Philippinesare also affected by locally originating tsunami. In thePhilippines, recorded tsunami have killed 50 000people, mainly in two single events in 1863 and 1976.Indonesia has experienced a similar death toll inrecorded times; however, half this total has beencaused by tsunami associated with local volcanism. TheAtlantic coastline is virtually devoid of tsunami;however, the Lisbon earthquake of 1755 produced a3–4 m wave that was felt on all sides of the Atlantic.The continental slope off Newfoundland, Canada, isseismically active and has produced tsunami that haveswept onto that coastline. One recorded tsunami fromthis site reached Boston, with a height of 0.4 m, on18 November 1929. One of the longest records oftsunami occurrence exists in the eastern Mediter-ranean. Between 479 BC and 1981 AD, 7 per cent ofthe 249 known earthquakes produced damaging ordisastrous tsunami. Here, about 30 per cent of allearthquakes produce a measurable seismic wave.

Tsunami generated close to shore are not so large asthose generated in deeper water. Tsunami have waveperiods in the open ocean of minutes rather thanseconds, and heights up to 0.5 m. Waves with thisperiod will travel at speeds of 600–900 km hr-1

(166–250 m s-1) in the deepest part of the ocean and100–300 km hr-1 (28–83 m s-1) across the continentalshelf. In some cases in the Pacific Ocean, initial waveperiods of 60 minutes have been recorded. Waveperiods of this length in the open ocean will oftendegrade to less than 2.5 minutes in shallow water.Tsunami resonating in harbors typically have periods of8–27 minutes. If the wave period is some harmonic ofthe natural frequency of the harbor or bay, then theamplitude of the tsunami wave can greatly increase

over time. This resonance may cause the period of thewave to shift to higher wave periods (lower frequen-cies). Figure 10.12 plots typical tide gauge records ormarigrams of tsunami at various locations in the PacificOcean. The concept that a tsunami consists of only oneor two waves is not borne out by the records. Somewave periods range between 8 and 30 minutes andpersist for over six hours. Wave characteristics arehighly variable. In some cases, the waves consist of aninitial peak that then tapers off in height exponentiallyover four to six hours. In other cases, the tsunami wavetrain consists of a maximum peak well back in the wavesequence.

The wavelength of an average tsunami of eightminutes is about 360 km (refer to Equation 8.3). Asexplained in Chapter 8, with this wavelength it ispossible for the tsunami to feel the ocean bottom at anydepth, and to undergo refraction effects. Refractiondiagrams of tsunami give accurate prediction of arrivaltimes across the Pacific, but not of run-up heights nearshore. It has been shown that the spread of tsunamiclose to shore and around islands follows classic dif-fraction theory (Figure 8.3c). Figure 10.13 illustratesthe movement of the Chilean tsunami wave of 23 May1960, which wreaked havoc throughout the Pacificregion. It clearly shows that the wave radiated sym-metrically away from the epicenter, and was unaffectedby bottom topography until islands or seamounts wereapproached. The island chains in the west Pacificclearly broke up the wavefront; however, much ofJapan received the wave unaffected by refraction.

It is a very noticeable phenomenon that ships,anchored several kilometers out to sea, hardly noticethe arrival of a tsunami wave. However, once the waveapproaches shore, the wave very rapidly shoals andreaches its extreme height. This run-up height controlsthe extent of damage. Because most shorelines arerelatively steep compared to the wavelength of thetsunami, the wave will not break or form a bore but willsurge up over the foreshore (Figure 10.14). Forexample, the earthquake offshore from Gisborne,New Zealand, on 26 March 1947, generated a tsunamirun-up 10 m high along a 13 km stretch of coast, whilethe eruption of Krakatau in 1883 generated a wavethat reached elevations up to 40 m high along thesurrounding coastline. The largest recorded earth-quake-generated tsunami wave occurred in 1737 onthe Kamchatka Peninsula when a 64 m high wavewashed across the southern tip of the peninsula. By far



Midway Island 9–10 March 1957 Antofagasta, Chile 10 April 1957

Unalaska 9 March 1957 Hilo 9 March 1957

GMT

50

40

30

20

10

1.0

0.5

0.0

-0.5

-1.0

3.5

3.0

2.5

2.0

1.5

1.5

1.0

0.5

2.0

1.5

1.0

250 300 350 400

18 21 24 3 6

15 18 21 24

15 18 21 24 6 9 12 15

GMT

GMTGMT

minutes

Cha

rt u

nits

met

res

met

res

met

res

met

res

Wake Island 9 March 1957

Fig. 10.12 Plots or marigrams of tsunami wave trains at various tidal gauges in the Pacific region (after Salsman, 1959; Van Dorn, 1959; Wiegel, 1964).

Epicenter19:11 GMT

23hr

12hr

60˚

30˚

0˚

-30˚

-60 ̊

1hr2hr3 hr4hr5 hr6hr7hr8hr9hr

10hr11hr12hr

14hr15hr16hr17hr

18hr

19hr

20 hr

21hr

22hr

24hr 13hr

16 hr

15 hr

14hr

7hr

Hanas

aki

Japa

n

Hilo

Tahi

ti

Nuku–Hiva

SouthAmerica

ValparaisoAustralia

Passage of tsunami crest,hours after earthquake

NorthAmerica

Santa

Monica

120˚ 150˚ 180˚ -150˚ -120˚ -90˚ -60˚

Fig. 10.13 Tsunami wave refraction pattern for 22 May 1960 Chilean earthquake in the Pacific Ocean (based on Wiegel, 1964 and Pickering et al., 1991).


the bay. Figure 10.15 dramatically illustrates thepotential extent of flooding inland that such tsunamiare able to generate. The American gunship Watereeand the Peruvian ship America were standing offshoreduring the Arica, Chile, tsunami of 13 August 1868.They were both swept 5 km up the coast, and 3 kminland over the top of sand dunes by two successivetsunami waves. The ships came to rest 60 m from avertical cliff at the foot of the coastal range, againstwhich the tsunami had surged to a height 14 m abovesea level. Unfortunately, 25 000 deaths also resultedfrom this event along the coasts of Chile and Peru.

Run-up heights also depend upon the configurationof the shore, diffraction, characteristics of the wave,and resonance. Within some embayments, it takesseveral waves to build up peak tsunami wave heights. Ifthe first wave in the tsunami wave train is the highest,then it can be dampened out. Figure 10.16 maps theheights of tsunami run-up around Hawaii for the 1946Aleutian earthquake. The northern coastline facing thetsunami advance received the highest run-up.However, there was also a tendency for waves to wraparound the islands and strike hardest at supposedlyprotected sides, especially on the islands of Kauai andHawaii. Because of refraction effects, almost everypromontory also experienced high run-ups, often

the largest run-up height recorded is that produced byan earthquake-triggered rockfall into Lituya Bay,Alaska, on 9 July 1958. The resulting wall of watersurged 524 m up the shoreline on the opposite side of

Fig. 10.14 Sequential photographs of the 9 March 1957 tsunamioverriding the backshore at Laie Point, Oahu, Hawaii. Thetsunami was generated by an earthquake with a surfacemagnitude of 8.3 in the Aleutian Islands, 3600 km away.Fifty-four people died in this tsunami (photograph credit:Henry Helbush. Source: United States Geological Survey,Catalogue of Disasters #B57C09-002).

Fig. 10.15 The American gunship Wateree in the foreground andthe Peruvian warship America in the background. Bothships were carried inland 3 km by a 21 m high tsunamiduring the Arica, Peru (now Chile) event of 13 August1868. Retreat of the sea from the coast preceded thewave, bottoming both boats. The Wateree, being flat-hulled, bottomed upright and then surfed the crestof the tsunami wave. The America, being keel-shaped,was rolled repeatedly by the tsunami (photographcourtesy of the United States Geological SurveyCatalogue of Disasters #A68H08-002).


exceeding 5 m. Coastlines where offshore depthsdropped off steeply were hardest hit, because thetsunami waves could approach shore with minimalenergy dissipation. For these reasons, run-up heightswere spatially very variable. In some places, forexample on the north shore of Molokai, heightsexceeded 10 m, while several kilometres away heightsdid not reach 2.5 m.

Tsunami magnitude scales (Abe, 1979, 1983; Horikawa & Shuto, 1983; Hatori,1986; Shuto, 1993; Geist, 1997)

Earthquake-generated tsunami are associated withevents registering more than 6.5 on the Ms scale. Two-thirds of damaging tsunami in the Pacific region havebeen associated with earthquakes of magnitude 7.5 orgreater. Tsunami can be scaled according to the sizeof their run-up. The following scale, known as theImamura–Iida scale, was defined using approximatelyone hundred Japanese tsunami between 1700 and1960:

mII = log2 Hrmax (10.1)where mII = Imamura–Iida’s tsunami

magnitude scale (dimensionless)

Hrmax = maximum tsunami run-up height

On the Imamura–Iida scale, the biggest tsunami inJapan – the Great Meiji Sanriku tsunami of 1896,

which had a run-up height of 38.2 m – had a magnitudeof 4.0. The other great Sanriku tsunami of 1933, whichranks as the second largest recorded tsunami in Japan,had a magnitude of 3.0. Japanese tsunami havebetween 1 and 10 per cent of the total energy of thesource earthquake. This can be related to a tsunami’smagnitude using the Richter scale (Table 10.3). Onlyearthquakes of magnitude 7.0 or greater are responsi-ble for significant tsunami waves in Japan, with run-upheights in excess of 1 m. However, as an earthquake’smagnitude rises above 8.0, the run-up height anddestructive energy of the wave dramatically increase. Amagnitude 8.0 earthquake can produce tsunami run-up of between 4 and 6 m in height. Intensities of 8.75generated the 38.2 m wave run-up of the Great MeijiSanriku tsunami.

The Imamura–Iida magnitude scale has acquiredworldwide usage. However, because the maximum run-up height of a tsunami can be so variable along a coast,Soloviev proposed a more general scale as follows:

is = log2 (1.4 H––

r) (10.2)where is = Soloviev’s tsunami

magnitude (dimensionless)H––

r = Mean tsunami run-up height along a stretch of coast (m)

This scale and its relationship to both mean andmaximum tsunami wave run-up heights are sum-marized in Table 10.4. Neither the Imamura–Iida northe Soloviev scales relate transparently to earthquake

0 50 100 150km6.1

16.4

200 0 m isobath

Run-up heightof tsunami

Direction of tsunami approach

Spot height (m) 16.4

(= 10 m)

Kauai

Oahu

Molokai

Maui

13.7

11.5

10.78.2

9.4

11.210.9

9.1

16.711.5

Hawaii

Fig. 10.16 Run-up heights of the 1 April 1946 tsunami around theHawaiian Islands following an Alaskan earthquake (basedon Shepard, 1977).

EarthquakeMagnitude Tsunami

Richter Scale Magnitude Maximum run-up (m)6 –2 <0.36.5 –1 0.5–0.757 0 1.0–1.57.5 1 2.0–3.08 2 4.0–6.08.25 3 8.0–12.08.5 4 16.0–24.08.75 5 >32.0

Source: Based on Iida, 1963.

Table 10.3 Earthquake magnitude, tsunami magnitude and tsunamirun-up heights in Japan (based on Iida, 1963).

The above scales imply that the size of a tsunamishould increase as the magnitude of the earthquakeincreases. This is true for most teleseismic tsunami inthe Pacific Ocean; however, it is now known that manysmaller earthquakes can produce large, devastatingtsunami. The Great Meiji Sanriku earthquake of 1896and the Alaskan earthquake of 1 April 1946 were ofthis type. The Sanriku earthquake was not felt widelyalong the adjacent coastline. Yet, the tsunami thatarrived thirty minutes afterwards produced run-upsthat exceeded 30 m in places and killed 22 000 people.These types of events are known as tsunami earth-quakes. Submarine landslides are thought to be one ofthe reasons why some small earthquakes can generatelarge tsunami; but this explanation has not beenproven conclusively. Presently, it is believed thatslow rupturing along fault lines causes tsunamiearthquakes.


magnitude. For example, both tsunami scales containnegative numbers and peak around a value of 4.0.Tsunami waves clearly carry quantitative informationabout the details of earthquake-induced deformation ofthe seabed in the source region. Knowing the tsunamimagnitude, Mt, it is possible to calculate the amount ofseabed involved in its generation using the followingformula:

Mt = log10 St + 3.9 (10.3)where St = area of seabed generating

a tsunami (m2)

Disaster descriptions(Cornell, 1976; Myles, 1985; Bryant, 2001)

Over the past two thousand years there have been 462 597 deaths attributed to tsunami in the Pacificregion. Of these deaths, 95.4 per cent occurred inevents that killed more than one thousand people each.The number of deaths is recorded in Table 10.5 foreach of the main causes of tsunami, while the eventswith the largest death tolls are presented in Table 10.6.

Tsunami Mean run-up Maximum run-up magnitude height (m) height (m)

–3.0 0.1 0.1–2.0 0.2 0.2–1.0 0.4 0.40.0 0.7 0.91.0 1.5 2.12.0 2.8 4.82.5 4.0 7.93.0 5.7 13.43.5 8.0 22.94.0 11.3 40.34.5 16.0 73.9

Table 10.4 Soloviev’s scale of tsunami magnitude (source: fromHorikawa and Shuto, 1983).

Date Fatalities Location22 May 1782 50,000 Taiwan27 August 1883 36,417 Krakatau, Indonesia28 October 1707 30,000 Nankaido, Japan15 June 1896 27,122 Sanriku, Japan20 September 1498 26,000 Nankaido, Japan12 August 1868 25,674 Arica, Chile27 May 1293 23,024 Sagami Bay, Japan4 February 1976 22,778 Guatemala29 October 1746 18,000 Lima, Peru21 January 1917 15,000 Bali, Indonesia21 May 1792 14,524 Unzen, Ariake Sea,

Japan24 April 1771 13,486 Ryukyu Archipelago22 November 1815 10,253 Bali, IndonesiaMay 1765 10,000 Guanzhou, South

China Sea16 August 1976 8,000 Moro Gulf, PhilippinesSource: National Geophysical Data Center and WorldData Center A for Solid Earth Geophysics (1998) andIntergovernmental Oceanographic Commission, (1999).

Table 10.6 Large death tolls from tsunami in the Pacific Oceanregion over the last 2000 years.

Number % of Number % of Cause of events events of Deaths deathsLandslides 65 4.6 14,661 3.2Earthquakes 1,171 82.3 390,929 84.5Volcanic 65 4.6 51,643 11.2Unknown 121 8.5 5,364 1.2Total 1,422 100.0 462,597 100.0Source: National Geophysical Data Center and WorldData Center A for Solid Earth Geophysics (1998) andIntergovernmental Oceanographic Commission (1999).

Table 10.5 Causes of tsunami in the Pacific Ocean region over thelast 2000 years.

Tectonically generated tsunami account for the greatestdeath toll, 84.5 per cent, with volcanic eruptionsaccounting for 11.2 per cent – mainly during twoevents, the Krakatau eruption of 26–27 August 1883 (36 417 deaths) and the Unzen, Japan, eruption of21 May 1792 (14 524 deaths). The number of fatalitieshas decreased over time and is slightly concentrated inSouth-East Asia, including Japan. The biggest tsunamiof the twentieth century occurred in Moro Gulf,Philippines, on 16 August 1976 where 8000 peopledied. The largest total death toll is concentrated in theJapanese Islands where 211 300 fatalities haveoccurred. Two events affected the Nankaido region ofJapan on 28 October 1707 and 20 September 1498,killing 30 000 and 26 000 people, respectively. TheSanriku Coast of Japan has the misfortune of beingthe heaviest populated, tsunami-prone, coast in theworld. About once per century, killer tsunami haveswept this coastline, with two events striking within aforty-year time span. On 15 June 1896, a small earth-quake on the ocean floor 120 km south east of the cityof Kamaishi sent a 30 m wall of water crashing into thecoastline and killing 27 122 people. The same tsunamievent was measured 10.5 hours later in San Franciscoon the other side of the Pacific Ocean. In 1933, disasterstruck again when a similarly positioned earthquakesent ashore a wave that killed three thousand inhabi-tants. Deadly tsunami have also affected Indonesia andthe South China Sea. In the South China Sea, recordedtsunami have killed 77 105 people, mainly in two eventsin 1762 and 1782. Indonesia has experienced a compa-rable death toll (69 420 deaths) over this period withthe largest following the eruption of Krakatau in 1883.

While earthquakes are responsible for mostdestructive tsunami, the Santorini and Krakatauvolcanic eruptions were probably the most devastat-ing volcanic events. The Krakatau eruption of27 August 1883, which will be discussed in moredetail in the next chapter, sent out a tsunami wavemeasured around the world. The wave rounded theCape of Good Hope in South Africa 6000 km away,and was recorded along the English Channel 37 hourslater. On the other side of the Pacific Ocean, waterlevels were affected along the west coast of Panamaand in San Francisco Bay. The waves that roundedthe world were probably associated with a substantialatmospheric shock wave, as some tsunami weredetected in bodies of water that were not connectedto each other. For instance, it is difficult to see how

the tsunami wave could have got through the islandarchipelagos of the west Pacific to register at SanFrancisco. In addition, a 0.5 m oscillation was measuredin Lake Taupo, in the center of New Zealand, coinci-dentally with the passage of the atmospheric shockwave. The Santorini eruption of 1470 BC generated atsunami that must have destroyed all coastal towns inthe eastern Mediterranean. The Santorini crater is fivetimes larger in volume than that of Krakatau, and twiceas deep. On adjacent islands, there is evidence ofpumice stranded at elevations up to 250 m above sealevel. The initial tsunami waves may have been 60 m inheight as they spread out from Santorini.

The Lisbon earthquake of 1 November 1755, whichis possibly the largest earthquake known (9.0 on the Msscale), resulted in the lower town being submergedunder a tsunami 15 m in height. The wave raised tidelevels 3–4 m above normal in Barbados and Antigua,West Indies, on the other side of the Atlantic. Tsunamialso affected the west coast of Europe and the Atlanticcoast of Morocco. The Spanish port of Cadiz, as well asMadeira in the Azores, was also hit by waves 15 m high,while a 3–4 m high wave sank ships along the EnglishChannel. Water level oscillations were also noticed inthe North Sea. The Caribbean has also witnessed itsshare of devastating tsunami. Noted events occurredin 1842, 1907, 1918, and 1946. However, the worsttsunami recorded destroyed Port Royal, Jamaica, inJune 1692. The earthquake that triggered the tsunamicollapsed the city, sending much of it sliding into thesea because of liquefaction. The resulting tsunamiflung ships standing in the harbor inland over two-storey buildings. Because of the small size of the town,the death toll was fortunately only 2000.

The most active area seismically producing tsunamiis situated along the eastern edge of the Nazca Plate,along the coastlines of Chile and Peru. This regionhas been inundated by destructive tsunami at roughly30-year intervals in recorded history – in 1562, 1570,1575, 1604, 1657, 1730, 1751, 1819, 1835, 1868, 1877,1906, 1922 and, most recently, 1960. The tsunamievents of 21–22 May 1960 were produced in Chile byover four dozen earthquakes with magnitudes up to8.5 and 9.6 on the Ms and MW scales, respectively.A series of tsunami waves, spread over a period of18 hours, took over 2500 lives across the Pacific, andproduced property damage in such diverse places asHawaii, Pitcairn Island, New Guinea, New Zealand,Japan, Okinawa, and the Philippines (see Figure 10.13).


The tsunami resulted from sudden land displacementof 2–4 m along the fault during the earthquakes.Within 15 minutes of the faulting, three large wavesdevastated coastal towns in Chile, where over 1700people lost their lives. The arrival of the first wave waspredicted at Hilo, Hawaii, five hours in advance to anaccuracy of one minute. Unfortunately, 61 people,mainly sightseers, were killed as the wave hammeredthe Hawaiian Islands. When the wave reached Japan24 hours later, it still had enough energy to reach aheight of 3.5–6 m along the eastern coastline. Fivethousand homes were washed away, hundreds of shipssank and 190 people lost their lives. In Hawaii, thisevent illustrated the folly of some humans in the faceof disaster. Despite plenty of warning, only 33 per centof the residents evacuated the affected area in Hilo.Over 50 per cent evacuated only after the first wavearrived, and 15 per cent stayed behind even after largewaves had beached. Many of those killed were spec-tators who went back to see the action of a tsunamihitting the coast.

Prediction in the Pacific region(Sokolowski, 1999; Bryant, 2001; International TsunamiInformation Center, 2004)

The most devastating ocean-wide tsunami of the pasttwo centuries have occurred in the Pacific Ocean.Following the Alaskan tsunami of 1946, the UnitedStates government established a tsunami warningsystem in the Pacific Ocean under the auspices ofthe Seismic Sea Wave Warning System. In 1948, thissystem evolved into the Pacific Tsunami WarningCenter (PTWC). Warnings were initially issued for theUnited States and Hawaiian areas; but, followingthe 1960 Chilean earthquake, the scheme was extendedto all countries bordering the Pacific Ocean. Until1960, Japan had its own warning network, believing atthe time that all tsunami affecting Japan originatedlocally. The 1960 Chilean tsunami proved that largesubmarine earthquakes in the Pacific Ocean regioncould spread tsunami ocean-wide. The Pacific TsunamiWarning System was significantly tested following theAlaskan earthquake of 1964. Within forty-six minutesof that earthquake, a Pacific-wide tsunami warning wasissued. This earthquake also precipitated the need forfurther international cooperation on tsunami warningin the Pacific. The United States National WeatherService currently maintains the Center. Currently,twenty-five countries cooperate in the Pacific Tsunami

Warning System, in one of the most successful disastermitigation programs in existence. The countriesinvolved include Canada, the United States and itsdependencies, Mexico, Guatemala, Nicaragua,Colombia, Ecuador, Peru, Chile, Tahiti, Cook Islands,Western Samoa, Fiji, New Caledonia, New Zealand,Australia, Indonesia, Philippines, Hong Kong, People’sRepublic of China, Taiwan, Democratic People’sRepublic of Korea, Republic of Korea, Japan and theRussian Federation. An additional ten countries ordependencies receive PTWC warnings. Many of theprimary countries also operate national tsunamiwarning centers, providing warning services for theirlocal area.

The Pacific Tsunami Warning System relies onany earthquake 6.5 or greater on the Richter scale reg-istering on one of thirty-one seismographs outside theshadow zones of any P- or S-waves originating inthe Pacific region (Figure 10.17). These stations areoperated by the Center itself, the West Coast andAlaskan Tsunami Warning Center, the United StatesGeological Survey’s National Earthquake InformationCenter, and various international agencies. Once asuspect earthquake has been detected, information isrelayed to Honolulu where requests for fluctuations insea level on tide gauges are issued to member countriesoperating sixty tide gauges scattered throughout thePacific. These gauges can be polled in real time so thatwarnings can be distributed to one hundred dissemina-tion points with three hours’ advanced notice of thearrival of a tsunami. For any earthquake with a surfacemagnitude of 7 or larger, the warnings are distributedto local, state, national, and international centers.Administrators, in turn, disseminate this informationto the public, generally over commercial radio andtelevision channels. The National Oceanic and Atmos-pheric Administration (NOAA) Weather Radio systemprovides direct broadcast of tsunami information to thepublic via very high frequency (VHF) transmission.The US Coast Guard also broadcasts urgent marinewarnings on medium frequency (MF) and VHFmarine radios. Once a significant tsunami has beendetected, its path is then monitored to obtain infor-mation on wave periods and heights. These data are thenused to define travel paths using refraction–diffractiondiagrams calculated beforehand for any possible tsunamioriginating in any part of the Pacific region.

The United States supplements these warningsusing six seabed transducers that have been installed



and linked to satellites for rapid communication(Figure 10.17). The project is known as the Deep-Ocean Assessment and Reporting of Tsunami (DART).The bottom transducers can detect a tsunami withheights of only 1 cm in water depths of 6000 m. TheUnited States also warns its west coast separately.The West Coast/Alaskan Tsunami Warning Center(WC/ATWC) in Palmer, Alaska, was established afterthe 1964 Alaskan earthquake to provide tsunamiwarnings to coastal areas of Alaska following localearthquakes. In 1982, the Center’s mandate wasextended to include the coasts of California, Oregon,Washington, and British Columbia. Alarms are trigger-ing by any sustained, large earthquake monitored ateight seismometers positioned along the west coast ofNorth America. Regional warnings can be issuedwithin fifteen minutes of the occurrence of any earth-quake having a surface magnitude greater than 7.

Localized earthquakes pose dilemmas in manycountries. Separate warning systems exist for Hawaii,Russia, French Polynesia, Japan and Chile. The Russianwarning system was developed for the Kuril–Kamchatka region following the devastating Kamchatka

tsunami of 1952. This system is geared towards therapid detection of the epicenter of coastal tsunamigenicearthquakes, because some tsunami here take only20–30 minutes to reach shore. In French Polynesia in1987, a system was developed by the PolynesianTsunami Warning Center at Papeete, Tahiti, to detectboth near- and far-field tsunami. The system usesthe automated algorithm TREMORS and calculates theexpected tsunami height from any epicenter in thePacific. Near-field tsunami are a threat in Japan, espe-cially along the Sanriku coastline of north-easternHonshu where only 25–30 minutes lead time existsbetween the beginning of an earthquake and the arrivalat shore of the resulting tsunami. The P-wave for anylocal earthquake can be detected within seconds usingan extensive network of high-frequency and low-magni-fication seismometers. Within the next two minutes, theheight of the tsunami along the adjacent coastline canbe predicted using graphic forecasting models basedupon the size of prior earthquakes and the distanceto their epicenters. Finally, Chile, being the sourceof many teleseismic tsunami, does not rely upon thePacific Tsunami Warning System, because this does not

120˚ 150˚ 180˚ -150˚ -120˚ -90˚ -60˚

60˚

30˚

0˚

-30˚

-60˚

Seismic station

Tide gauge

DART buoy

THRUST coverage

North America

Palmer

Pacific TsunamiWarning Center

HawaiiWake Island

Wellington

Papeete

Easter Island

Tokyo

Guam

Australia

SouthAmerica

Valparaiso

Fig. 10.17 Location of seismic stations and tide gauges making up the Pacific Tsunami Warning System, buoys forming NOAA’s DART network, and area ofpotential coverage of the THRUST satellite warning system (sources: Bernard, 1991; González, 1999; <http://vishnu.glg.nau.edu/wsspc/tsunami/HI/Waves/waves00.html> [cited 2000, now defunct]

give a long enough lead time in which to evacuate thecoast. Project THRUST (Tsunami Hazards ReductionUtilizing Systems Technology) was established offshorefrom Valparaiso, Chile, in 1986 to provide advancedwarning (within two minutes) of locally generatedtsunami along this coastline. When a sensor placed onthe seabed detects a seismic wave above a certainthreshold, it transmits a signal to the GEOS geostation-ary satellite, which then relays a message to groundstations. The signal is processed and another signalis transmitted via satellite to a low-cost receiver andantenna operating twenty-four hours a day, locatedalong a threatened coastline. This designated stationcan be pre-programmed to activate lights and acousticalarms, and to dial telephones and other emergencyresponse apparatus when it receives a signal. For a costof $US15 000, a life-saving tsunami warning can beissued to a remote location within two minutes of atsunamigenic earthquake. The THRUST system hasthe potential to be used over a wide area of the Pacific(Figure 10.16).

The Pacific Tsunami Warning System is not flawless.The risk still exists in many island archipelagos alongthe western rim of the Pacific, that local earthquakescould generate tsunami too close to shore to permitadvanced warning. The 7.8 magnitude (Ms scale)Philippines earthquake of 16 August 1976, in the MoroGulf on the south-west part of the island of Mindanao,generated a local tsunami 3.0–4.5 m high that report-edly killed 8000 people. The event was virtually un-predictable because the earthquake occurred within20 km of a populated coastline. Similarly, the PNGtsunami of 17 July 1978, which killed 2200 people, alsoescaped the notice of the Pacific Tsunami WarningSystem. Finally, our knowledge of tsunami is rudimen-tary for many countries and regions in the PacificOcean. While travel-time maps have been drawn upfor source regions historically generating tsunami – forexample, the coasts of South America, Alaska andKamchatka Peninsula – the entire Pacific Rim coast-line has not been studied. This was made apparent on25 March 1998 when an earthquake with a surfacemagnitude, Ms, of 8.8 occurred in the Balleny Islandsregion of the Antarctic directly south of Tasmania,Australia. Because of the size of the earthquake, atsunami warning was issued; but no one knew whatthe consequences would be. The closest tide gaugeswere located on the south coasts of New Zealand andAustralia. The Pacific Tsunami Warning Center in

Hawaii had to wait to see if any of these gaugesreported a tsunami before they issued warnings furtherafield. While that may have helped residents in theUnited States or Japan, it certainly was little comfort toresidents living along coastlines facing the Antarctic inthe antipodes. In cities such as Adelaide, Melbourne,Hobart and Sydney, emergency hazard personnelknew they were the ‘mine canaries’ in the warningsystem. Fortunately, the Antarctic earthquake was notconducive to tsunami and no major wave propagatedinto the Pacific Ocean.

CONCLUDING COMMENTS

The seismic risk for many locations in the world cannow be assessed to the point that many buildings andstructures can be designed to withstand damagefrom tremors. However, the seismic risk in manylocations, even in developed countries or ones whereearthquakes have been recorded for centuries,cannot be assessed completely. For example, theseismic risk for Australia is very much a product ofthat country’s distribution of cities (Figure 10.2). InNorth America, the assessment of seismic risk in theeastern part of the continent is still deficient becausemajor earthquakes occur so infrequently that, evenafter centuries of settlement, the full extent of thehazard is not known (Figure 10.2). This chapter hasnot highlighted recent earthquakes but concentratedupon two large events, the Alaskan earthquake of1964 and the Tokyo earthquake of 1923, to illustratethe force that can be released in a tremor and thedestruction that can result. The Alaskan earthquakewas one of the largest events in the twentiethcentury. Even though it occurred in a relativelyisolated part of the world, its magnitude and sub-sequent tsunami attracted world attention. The Tokyoearthquake is noteworthy because of the appallingdeath toll and the manner in which those deathsoccurred. The descriptions of these events couldeasily have been set within southern California, aregion where earthquakes with the same magnitudeas the Alaskan one can easily occur, and a regionwhere the effects upon dense human settlementmight be as dramatic. In discussing the Californianearthquake hazard, it was hoped to highlight theextent of the hazard and some of the inadequacies inpeople’s attempts to minimize future damage andloss of life. Despite decades of retro-fitting ofstructures to withstand earthquakes, engineering


Bernard, E.N. 1991. Assessment of Project THRUST: past, present,future. Proceedings of the Second UJNR Tsunami Workshop,Honolulu, Hawaii 5–6 November 1990, National GeophysicalData Center, Boulder, pp. 247–255.

Bolt, B.A. 1993. Earthquakes. W.H. Freeman and Co., New York.Bolt, B.A., Horn, W.L., MacDonald, G.A., and Scott, R.F. 1975.

Geological Hazards. Springer-Verlag, Berlin.Bryant, E. 2001. Tsunami: The Underrated Hazard. Cambridge

University Press, Cambridge.Cornell, J. 1976. The Great International Disaster Book. Scribner’s,

New York.EQE 1995. The January 17, 1995 Kobe earthquake. <http://www.

eqe.com/publications/kobe/kobe.htm> (URL defunct as of 2004)Geist, E.L. 1997. Local tsunamis and earthquake source parame-

ters. Advances in Geophysics 39: 117–209.Giardini, D., Grünthal, G., Shedlock, K. and Zhang, P. 1999. Global

Hazard Seismic Map. Global Seismic Hazard AssessmentProgram, UN/International Decade of Natural Hazard Reduction.

González, F.I. 1999. Tsunami! Scientific American, May: 44–55.Hadfield, P. 1992. Sixty Seconds That Will Change the World: The

Coming Tokyo Earthquake. Tuttle, Boston.Hansen, W.R. 1965. Effects of the Earthquake of March 27, 1964 at

Anchorage, Alaska. United States Geological Survey ProfessionalPaper No. 542-A.

Hatori, T. 1986. Classification of tsunami magnitude scale. BulletinEarthquake Institute 61: 503–515 (in Japanese).

Hays, W.W. 1981. Facing geologic and hydrologic hazards: Earth-science considerations. United States Geological SurveyProfessional Paper No. 1 240-B: 54–85.

Hodgson, J.H. 1964. Earthquakes and Earth Structure. Prentice-Hall, Englewood Cliffs, New Jersey.

Holmes, A. 1965. Principles of Physical Geology. Nelson, London. Horikawa, K. and Shuto, N. 1983. Tsunami disasters and protection

measures in Japan. In Iida, K. and Iwasaki, T. (eds) Tsunamis:Their Science and Engineering. Reidel, Dordrecht, pp. 9–22.

Iida, K. 1963. Magnitude, energy and generation of tsunamis, andcatalogue of earthquakes associated with tsunamis. Proceedings ofthe Tsunami meetings associated with the Tenth Pacific ScienceCongress. International Union of Geodesy and GeophysicsMonograph 24: 7–18.

Intergovernmental Oceanographic Commission 1999. Historicaltsunami database for the Pacific, 47 BC–1998 AD. TsunamiLaboratory, Institute of Computational Mathematics andMathematical Geophysics, Siberian Division, Russian Academyof Sciences, Novosibirsk, Russia, <http://tsun.sscc.ru/HTDBPac1/>(URL defunct as of 2004)

International Tsunami Information Center 2004. <http://www.prh.noaa.gov/itic/more_about/itic/itic.html>

Johnson, J.M. 1999. Heterogeneous coupling along Alaska-Aleutiansas inferred from tsunami, seismic, and geodetic inversions.Advances in Geophysics 39: 1–106.

Keller, E.A. 1982. Environmental Geology (3rd edn). Merrill,Columbus, Ohio, pp. 133–167.

Lomnitz, C. 1988. The 1985 Mexico earthquake. In El-Sabh, M.I.and Murty, T.S. (eds) Natural and Man-Made Hazards, Reidel,Dordrecht, pp. 63–79.

Myles, D. 1985. The Great Waves. McGraw-Hill, New York.


expertise is only recently coming to terms with thefact that the effects of seismic waves upon structureshas been underestimated by a factor of 2–4. The SanFrancisco and Kobe earthquakes of 1989 and 1995,respectively, triggered this reassessment. There areextensive regions prone to large earthquakes wherepeople must live and adjust to this threat.

Finally, very little discussion was given to thethreat of tsunami along coastlines where historicalrecords are minimal. The coastline of Australiaillustrates this point. This country has been scruti-nized by written observations for only the past230 years. Historically, the largest tsunami waveheight measured on a tide gauge – 1.07 m – occurredat Sydney in May 1877. Observations exist on thenorth-west coast of Australia for run-up heights of4 m following the Krakatau eruption in 1883 andof 6 m in August 1977 following an Indonesianearthquake. Yet, Aboriginal legends originating fromcoastal regions frequently note the occurrence oflarge tsunami. Detailed fieldwork examining a rangeof signatures left by tsunami in coastal deposits andthe erosion of bedrock across Australia indicate thattsunami have run up to 35 km inland on the north-west coast and topped headlands 130 m high south ofSydney. The extent and magnitude of this evidencecould be caused only by a comet or meteorite impactwith the ocean. These events are not ancient.Radiocarbon dating indicates that these mega-tsunami occurred within the last 300–500 years. BothAboriginal and Maori legends also substantiate acosmogenic event within this time frame. Ourknowledge of tsunami as a hazard is deficient interms of both cause and extent.


Abe, K. 1979. Size of great earthquakes of 1837–1974 inferred fromtsunami data. Journal Geophysical Research 84: 1561–1568.

Abe, K. 1983. A new scale of tsunami magnitude, Mt. In Iida, K. andIwasaki, T. (eds) Tsunamis: Their Science and Engineering.Reidel, Dordrecht, pp. 91–101.

Bardet, J.P., Idriss, I.M., O’Rourke, T.D., Adachi, N., Hamada, M.,Ishihara, K. 1997. North America–Japan Workshop on theGeotechnical Aspects of the Kobe, Loma Prieta, and NorthridgeEarthquakes Quakes. Report to National Science Foundation, AirForce Office of Scientific Research and Japanese GeotechnicalSociety.

Robinson, A. 2002. Earthshock. Thames & Hudson, London.Salsman, G.G. 1959. The tsunami of March 9, 1957, as recorded at

tide stations. U.S. Coast and Geodetic Survey Technical BulletinNo. 6.

Scheidegger, A.E. 1975. Physical Aspects of Natural Catastrophes.Elsevier, Amsterdam.

Shepard, F.P. 1977. Geological Oceanography. University ofQueensland Press, St Lucia.

Shuto, N. 1993. Tsunami intensity and disasters. In Tinti, S. (ed.)Tsunamis in the World. Kluwer, Dordrecht, pp. 197–216.

Sokolowski, T.J. 1999. The U.S. West Coast and Alaska TsunamiWarning Center. Science of Tsunami Hazards 17: 49–56.

United States Geological Survey 2002. Most destructive knownearthquakes on record in the world. <http://neic.usgs.gov/neis/eqlists/eqsmosde.html>

Van Dorn, W.G. 1964. Tsunamis. Advances in Hydrosciences2:51–46.


Wiegel, R.L. 1964. Oceanographical Engineering. Prentice Hall,Englewood Cliffs, New Jersey.

Wood, R.M. 1986. Earthquakes and Volcanoes. Mitchell Beazley,London.


National Geophysical Data Center and World Data Center A forSolid Earth Geophysics 1984. Tsunamis in the Pacific Basin1900–1983. United States National Oceanic and AtmosphericAdministration, Boulder, Colorado, map 1:17 000 000.

Oakeshott, G.G. 1983. San Andreas Fault: Geologic and EarthquakeHistory. In Tank, R.W. (ed.) Environmental Geology. OxfordUniversity Press, Oxford, pp. 100–107.

Palm, R. and Hodgson, M.E. 1992. After a California Earthquake:Attitude and Behavior Change. Geography Research PaperNo. 233, University of Chicago.

Pararas-Carayannis, G. 1998. The March 27, 1964 Great AlaskaTsunami. <http://www.geocities.com/CapeCanaveral/Lab/1029/Tsunami1964GreatGulf.html>

Pickering, K.T., Soh, W., and Taira, A. 1991. Scale of tsunami-generated sedimentary structures in deep water. Journal of theGeological Society London 148: 211–214.

Pinter, B., Gobron, N., Verstraete, M.M., Mélin, F., Widlowski, J-L.,Govaerts, Y., Diner, D.J., Fielding, E., Nelson, D.L., Madariaga,R., Tuttle, M.P. 2003. Observing earthquake-related dewateringusing MISR/Terra satellite data. EOS, Transactions, AmericanGeophysical Union 84: 37, 43.

Ritchie, D. and Gates, A.E. 2001. Encyclopedia of Earthquakes andVolcanoes. Facts on File, New York.

C H A P T E R 1 1

INTRODUCTION(Bolt et al., 1975; Tazieff & Sabroux, 1983)

Of all natural hazards, volcanoes are the most complex.Whereas a tropical cyclone has a predictable structure,or a drought can generate a predictable sequence ofevents in rural communities, such predictions cannotbe made in respect of volcanoes. There is a multitudeof volcanic forms, and each event appears unique inthe way that it behaves, and the physical and humanconsequences it produces. This chapter will examinethe different types of volcanoes and the secondaryphenomena associated with their occurrence. It willconclude with a detailed description of some of themore spectacular volcanic disasters that have occurredin recorded history.

Volcanoes are conduits in the Earth’s crust throughwhich gas-enriched, molten silicate rock magmareaches the surface from beneath the crust. The originof magma is still debated, but it is generally believedfrom seismic evidence that the mantle is partiallyliquefied 75–300 km below the Earth’s surface. Thereare two types of magma. The first type consists ofsilica-poor material from the mantle, and formsbasaltic volcanoes. The second type consists of silica-rich material originating from either the melting of thecrust in subduction zones or the partial differentiationof liquefied mantle material. This second type forms‘andesitic’ volcanoes, the largest group of which rings

the western Pacific Ocean, where the Pacific Plate issubducting beneath the Eurasian Plate (Figure 9.1).Basaltic lavas have temperatures of between 1050 and1200°C, while andesitic ones are 100–150°C cooler.For each 500°C decrease in temperature, fluiditydecreases tenfold. For this reason, and because of theirhigher silica content, andesitic lavas are 200–2000times more viscous than basaltic lavas. Andesitic lavasare also gas-rich, with 50–90 per cent of the gasconsisting of water assimilated from dissolved crustalmaterial. This high water content and high viscositymakes andesitic volcanoes, located over supposedsubduction zones, much more explosive than basalticvolcanoes sitting over hot spots in the center of crustalplates.

Magma is not necessarily extruded at the Earth’ssurface by pressure originating in the mantle below.Instead, lava penetrates through tens of kilometres ofweakened, fractured crust because of gas pressure. Allmagmas, be they andesitic or basaltic, contain variousamounts of dissolved gas. These magmas can becomeoversaturated in gas due to crystallization of themagma, its cooling down, decreases in hydrostaticpressure, or enrichment of volatile molecules fromoutside the magma. The gas in oversaturated lavasbegins to vesiculate or form bubbles, a process thatimmediately makes the magma lighter or morebuoyant. If the magma is forced quickly upward bypressure from below, then the hydrostatic pressure

Volcanoes as a Hazard

acting on the magma to dissolve these bubbles candecrease rapidly, causing a chain reaction of newbubble formation. If the magma is gas-rich, asandesitic lavas are, and the chain reaction is rapidenough, an explosive eruption can occur. Once explo-sive magma reaches the atmosphere, it is fragmentedinto small particles by the gas and by the force of theeruption. These pieces are called pyroclastic ejecta ortephra and are classified according to size as follows:bombs (liquid, boulder-sized blobs), blocks (angular,solid boulders), scoriae (cinder material of any size),lapilli (fragments 2–60 mm in diameter), and sandsand ash (< 0.1 mm). Depending upon the size anddistance hurled, much of this material falls back to theground around the vent to build up a cone.

The study of the gas phase of magma is crucial indetermining whether the volcano will become explosive.If the magma is rising slowly, a small quantity of gas willcontinuously escape through vents to the surfacethrough fumaroles. This movement tends to be slowbecause the viscosity of the magma inhibits bubblerelease. Bubbles, once formed, tend to stick toliquid–solid interfaces no matter how buoyant theybecome. These gases also have sufficient time to dissolvein groundwater. The monitoring of groundwater orescaping gas can give an accurate indication of thematurity of the magma, and its closeness to the surface.Some of these geochemical techniques for predictionwere discussed in Chapter 10.

In some volcanic eruptions, only this degassingprocess initially affects the upper magma. As thismagma is forced out, the magma immediately beneathdegasses explosively. In this way, there may not be onebig explosion, but a series of continual blasts, eachlasting a few seconds to minutes, over several days.Degassed magma may also pour out of the volcano aslava, which (depending upon viscosity, output and theslope of the ground) can flow from a few meters toseveral hundreds of kilometers from the vent. All lavasoon solidifies to form hard rock, although extremelythick lava beds may take years to cool down completelyto surrounding environmental temperatures. There arethree types of basaltic lava that can be ejected from avolcano: pahoehoe, aa, and block lava. Pahoehoe isbasaltic lava that solidifies at the top, but which is stillfed from below by pipe vesicles. Such lava has asmooth, wrinkled surface, is less than 15 m thick andflows at rates of a few metres per minute. Aa forms asa thin river of lava, less than 10 m thick, with a spiky,

cinder-like topping. Block lava consists solely ofandesite or rhyolite up to 300 m thick, and has asurface coating consisting of smooth-sided fragments.It has the slowest traveling speed of all lava flows,moving only a few metres per day. Both aa and blocklava behave in all respects like a river and can bebordered by levee banks.

The total energy of a volcanic event can be brokendown into four processes as follows:• energy released by volcanic earthquakes,• energy necessary to fracture the overburden,• energy expended in ejecting material, and• energy expended in producing atmospheric shock

waves and, occasionally, tsunami.Volcanic tremor and earthquake energy can be

assessed in similar ways to any other earthquake event.These methods of measuring energy have been dis-cussed under earthquake magnitude in Chapter 10.Earthquakes of magnitude 7 or larger on the Richterscale have been produced by major explosive erup-tions, and contain more than 4 � 1015 joules of energy.The energy required to break up overburden istypically greater than 1 � 1011 joules. Energy spent inejecting material consists of kinetic energy, potentialenergy, and thermal energy. These terms can becalculated for most volcanic eruptions, and indicatethat volcanic explosions release as much energy aslarge nuclear devices – typically 1 � 1015 joules. Theenergy used in generating the atmospheric shock wavecan be expressed as follows:

Es = 1.25 � 1013 sin� 0.5 At2 (11.1)where Es = energy to generate atmospheric

shock wave sin � = distance from source of the

explosion in degrees on the Earth

A = amplitude of shock wave in millibars

t = wave period in seconds

A similar type of equation as 11.1 can be used tomeasure the energy contained in any tsunami wavegenerated by an eruption. These shock waves requireenergy expenditure greater than 1 � 1015 joules. Intotal, volcanic eruptions contain energy in excess of1 � 1016 joules. Some of the larger eruptions, such asKrakatau in 1883, or Tambora in 1815, containedenergy in excess of 1 � 1018 joules (see Figure 11.1 for


229Volcanoes as a Hazard

all major placenames mentioned in this chapter andnot located on individual site maps). By calculating theamount of energy partitioned among these variousprocesses, it is possible to classify the type of volcanoand its magnitude. This is of little help in predictingvolcanic events, but it can give information on the typeof eruption that has occurred. For instance, eruptivevolcanoes expend considerable energy in easilymeasured atmospheric shock waves and tsunami.Knowledge of these values can be used to evaluate theoverall energy of the volcano. In addition, the contin-ual monitoring of energy expenditure, and the way it ispartitioned, can be used to predict whether a volcanois dying down or increasing in activity.

TYPES OF VOLCANICERUPTIONS

(Fielder & Wilson, 1975; Blong, 1984; Wood, 1986)

While most classification schemes refer to a specifichistorical event characterizing an eruption sequence, itshould be realized that each volcanic eruption isunique, and may over time take on the characteristicsof more than one type. There are two types of non-explosive volcanoes. Volcanoes producing flood lavasare the least hazardous of all eruptions, because theyoccur very slowly, permitting plenty of time for evacu-ation. Flood lavas are basaltic in composition. Atpresent, such eruptions are rare, occurring only inIceland, where they happen with sufficient warning to

permit safe evacuation. In the geological past, somevolcanoes have had enormous magma chambers thatemptied and flooded landscapes over vast distances ina short period. Some flood lavas have laid downdeposits 500 m thick over distances of 300 km. Thelargest such deposits on Earth (flood lavas also occuron other planets and their moons) make up the DeccanPlateau in India, the island of Iceland, the Great RiftValley of Africa, and the Columbia River Plateau inthe United States. ‘Hawaiian’-type eruptions are verysimilar to flood lavas. However, at times, they canproduce tephra and faster moving, thin lava flows.Hawaiian eruptions tend to be aperiodic, building upsuccessive deposits over time. These can form largecones that, in the case of the Hawaiian Islands, can betens of kilometres high and hundreds of kilometreswide at their base.

Explosive eruptions can be more complex. Thesimplest form begins with the tossing out of moderateamounts of molten debris. ‘Strombolian’ volcanoes,named after Mt Stromboli in Italy, throw up fluid lavamaterial of all sizes. Most eruptions of this type toss outbombs a few hundreds of metres into the air, rhythmi-cally, every few seconds in events that can go on foryears. They can also produce moderate lava flows.Strombolian volcanoes are characterized by a verysymmetrical cinder cone that grows in elevationaround the vent. Strombolian volcanoes also representan intermediate stage between basaltic and acidicmagma extrusions. ‘Vulcanian’-type eruptions are more

Vatnajökull

Mt St Helens

Kilauea El ChichonSanta Maria

Nevado del Ruiz

Lake Taupo

KrakatauGalunggung

Mt. AgungTambora

Rift Valley

Kolkata

Elsey Cr.

Bandai

Mt. Pinatubo

Mt. StromboliVesuvius

Laki

La Soufrière

SoufrièreMt. Pelée

France

Columbia River Plateau

SurtseyIceland

Katmai

Parícutín

Mauna LoaHilo

Mt. Hekla

Mt. SpurrRedoubt

Augustine

Mt. Erebus

Vancouver

Dieng PlateauKelat

Mt Rainier Usu

Mt. Ruapehu

Mt. EtnaSantorini

Rodriguez Is.Northwest Cape

Aden

Seychelles Islands

Mauritius

Port Elizabeth

English Channel

San Francisco

Kodiak Is


acidic, and give off large blocks of very viscous, hotmagma with few if any flows. This very viscous materialmay be tossed up to heights of 10 km (40 km inextreme cases). Activity is explosive for periods rangingup to several months. These volcanoes build up tephraand block cones. Usually the lava freezes in ventsbetween eruptions, so that the strength of the plug ulti-mately determines the amount of pressure that canbuild up before the next eruption. ‘Surtseyan’ erup-tions are violently explosive because hot, fluid magmaencounters seawater. These types of eruptions aretermed ‘phreatomagmatic’ and make up 8.8 per centof all volcanic events. Because of the interaction ofmagma and water, large clouds of very fine dust areproduced that can drift several kilometres upward intothe atmosphere. Coarser particles tend to form ringsaround the vent, rather than cones or large ash sheets.The rubbing together of fine dust particles builds upstatic electricity, leading to numerous and spectacularlightning displays around the rising column of ejecta.

‘Plinian’ events are more explosive, with widedispersal of tephra. Over 1 km3 of magma may bedriven straight up, at velocities of 600–700 m s-1, toheights of 25 km by a continuous gas jet and thermalexpansion over the course of several hours. As themagma chamber slowly empties, the character of themagma changes and the eruption becomes starved formaterial. Usually, the volcano then collapses internallyunder its own weight to form a caldera. The MtVesuvius eruption of 79 AD was of this type, and willbe described later in more detail. While such eventshave been dramatic on Earth, on other planets withlower gravity and thinner atmospheres, Plinianeruptions have been truly cataclysmic.

If the eruptive ash column from an explosive volcanocollapses under the effect of gravity, it can produce adebris avalanche of ash and hot gas that can racedownslope at speeds reaching 60 km hr-1. This type oferuption is termed a ‘Peléean’ eruption after the explo-sion of Mt Pelée in 1902. The debris clouds are calledpyroclastic flows or nuées ardentes, and the resultingdebris deposits are termed ignimbrites. The large deathtoll of Vesuvius in 79 AD can be attributed to anassociated pyroclastic flow event. Pyroclastic flows willbe examined later in this chapter when secondaryvolcanic hazards are examined in detail. The mostdestructive and largest Peléean eruptions are termedthe ‘Katmaian’ type, after the Katmai eruption at theeastern end of the Alaska Peninsula on 6 June 1912.

This eruption ejected ten times the material of the 1902Mt Pelée event. As the magma chamber emptied,a 2500 m high mountain collapsed into a caldera 5 km across and 1 km deep. The eruption was heard1200 km away, with dust being deposited over a1500 km distance. The pyroclastic flow from thiseruption flooded into a valley 20 km long, leaving anestimated 11 km3 of hot, fused ash spread to a depth of250 m.

Peléean eruptions should not be used to refer to allvolcanic events that generate hot ash flows. Manydestructive pyroclastic flows have not originated as ashcolumns collapsing under the effect of gravity, but havebeen preceded by a ground or basal surge caused bythe lateral eruption of the volcano. This type ofexplosion is termed a ‘Bandaian’ eruption. The 1980eruption of Mt St Helens in the United States was ofthis type, which is more properly known as a basesurge. Because the force of the eruption shoots outparticles at velocities in excess of 150 km hr-1, theblasts have been likened to cyclones of ash. Objects inthe paths of such hot and violent blasts are destroyedby sandblasting. Base surges often precede a large andslower moving pyroclastic flow, and represent the mostdestructive of all types of volcanic eruptions.

VOLCANIC HAZARDS

(Bolt et al., 1975; Whittow, 1980; Hays, 1981; Tazieff &Sabroux, 1983; Blong, 1984; Wood, 1986; Ritchie &Gates, 2001)

There are many hazardous phenomena produceddirectly, or as secondary effects, by volcanic eruptions.The association of earthquakes and tsunami withvolcanoes has already been covered in Chapter 10, andwill be discussed in detail for particular events at theend of this chapter. Excluding these two phenomena, itis possible to group volcanic hazards into six categoriesas follows:• lava flows, • ballistics and tephra clouds,• pyroclastic flows and base surges,• gases and acid rains,• lahars (mudflows), and• glacier bursts (Jökulhlaups).

Lava flows

Since the sixteenth century, 20 km2 of land per year hasbeen buried by lava flows. There are about 60 lava flow


events per century that result in severe social andeconomic disruption. The main reason for thisdisruption is the fact that lava flows in volcanic areasare very fertile, and have attracted intense agriculturalusage and dense settlement. Re-eruption of volcanoesin these areas has, thus, led to the destruction ofthis agricultural land and to loss of life. Much of thisdestruction can be prevented simply by mapping andavoiding those areas most frequently inundated by lavaflows. Techniques also exist to modify flow behavior,especially for pahoehoe and aa flows.

Based upon the more than 1000 flows that haveoccurred historically, low-viscosity flows have a medianlength of 4.1 km and an average thickness of 10 m, whilehigh-viscosity flows have a median length of 1.3 km andan average thickness of 100 m. There is a 1 per centprobability that high- and low-viscosity lava flows willexceed, respectively, 11 km and 45 km in length. Thefastest moving, hottest and most mobile flows are thelow-viscosity Hawaiian- and Icelandic-type eruptions.These flows would appear to be dangerous; however,they are thin and often consist of pahoehoe-type lava,which cools very rapidly. Pahoehoe lavas may developlarge diameter lava tubes extending several kilometresunder the flow. These can provide conduits for subse-quent lava venting once the original flow has cooled atthe surface. Such flows can travel great distances, butare the easiest to divert. Aa flows are thicker and moreviscous. They tend to channelize to depths up to 30 m.Both aa and block lava flows are more difficult to divert.

The characteristics of flows can vary dependingupon temperature, viscosity, yield strength, and expul-sion rate. Because of the high temperatures of flows(880–1130°C), all carbon materials such as cloth,paper and wood are easily ignited upon contact withlava. Once temperatures drop below 800°C, a skinforms on the magma that effectively inhibits furtherheat loss. Lavas have been known to have high internaltemperatures 5–6 years after they have stoppedflowing. As temperature decreases, viscosity increases,and it is this factor that stops most flows. While thevelocity of a flow depends upon gravity or the slope ofthe terrain, the rate at which lava is expelled alsocontrols the speed of movement. Most expulsion ratesrange between 15 and 45 m s-1, but values above1000 m3 s-1 have been measured. The distancereached by lava is a direct function of this expulsionrate, such that the distance of travel in kilometers isabout three times the expulsion rate measured in

m3 s-1. The actual velocity of flows is in a range of9–100 km hr-1, with average values around 30 km hr-1.

While most documented flows rarely exceed 4 km3

in volume, some have caused impressive damage. TheLaki fissure eruption of 1783–1784 in Iceland covered550 km2, had a length of 88 km and a total volume of12.3 km3. This eruption caused melting of glaciers andflooding of valuable farmland. The sulfurous vaporsfrom the lava hung over the countryside for weeks. The‘haze famine’ that resulted killed 20 per cent ofIceland’s population, and wiped out 50 per cent ofits cattle and 75 per cent of its sheep and horses.Extensive flows sweeping down the slopes of MtVesuvius engulfed the town of Bosco Trecase at its basein 1906 and the town of San Sebastiano in 1944. Lavafrom Mt Etna in 1669 reached the wall of the feudalcity of Catania 25 km away, eventually overtopping thewalls and moving slowly through the city. In 1928,the town of Mascali was also destroyed by lava fromMt Etna. Parícutín in Mexico, in 1946, buried 2400hectares of forest and agricultural land before com-pletely covering the town of San Juan Parangaricutiro,5 km away from the main vent. Eruptions from MaunaLoa and Kilauea in Hawaii have consistently floodedagricultural land, destroyed small villages surroundingthe cones, and threatened the major city of Hiloon several occasions. In January 2002, lava from MtNyiragongo in the Democratic Republic of the Congoburied the town of Goma under 1.5 m of lava andcaused over 500 000 people to flee for their lives.

The easiest way to stop lava movement is to enhancethe processes that slow it down. Flow depends uponyield strength, which can be increased by lowering thetemperature of the melt, increasing the rate of gasescape from the flow, stirring, and seeding the lavawith foreign nuclei. In some cases, increasing the yieldstrength of the flow may only turn a fluid flow into amore viscous one, making it thicker and more difficultto stop. Because of the insulating properties of solidlava, cooling by water is not always effective. It hasbeen observed that lava flows can continue to moveeven when submerged under the sea. One notableexception appears to be the attempts in 1973 to stopa block lava flow, which threatened the town ofVestmannaeyjar in Iceland. Water was intensivelypumped for months onto, and just behind, the movinglava front, thus increasing the viscosity of the magmaand slowing its rate of progress. Agitation of the flowcan be difficult to achieve except by bombing. This was


tried on one tube feeding a pahoehoe flow that threat-ened Hilo, Hawaii, in 1935. The flow was disruptedenough that it turned into a more viscous aa flow, andthe tube became blocked. The bombing might nothave been the reason for the flow termination, becauseit was already in the final stages of expulsion. However,there are still contingency plans for bombing lava flowsif they threaten Hilo in the future. The sides of a flowcan also be breached, thus diverting lava to a saferpath. However, what one person considers safe,another may consider a threat. This problem beset theunfortunate inhabitants of Catania in the 1669eruption of Mt Etna. The citizens of the walled citymade the first recorded attempt to alter a lava flow.They clad themselves in wetted animal skins, and duginto one of the lava levees controlling the route of theflow towards their city. Unfortunately, the diversionsent the flow towards the town of Paterno. Fivehundred enraged residents of this latter city raced upthe slopes to attack the citizens of Catania, and termi-nated their efforts after winning a pitched battle. Thelava flow clogged the breach, proceeded to Catania,overflowed its 20 m high walls, and inundated largesections of the town. A similar attempt in Hawaii in1942, rather than diverting the flow, only sent it on aparallel path.

Barrier construction has also been used to divert aaflows, or give more time for effective evacuation. Mostlava flows will pond behind the flimsiest obstacle, andthen simply overtop it, as eventually occurred inCatania. Barriers should not be used to stop a flow, butsimply to divert it. Because lava flows can become veryviscous as they cool down (106 times more viscous thanwater), they may only be flowing over very low gradesand exerting very little force. They can thus be divertedsuccessfully by putting guiding barriers diagonally intheir paths, as long as the diversion route is steep.However, the rapid movement of some flows rules outtime-consuming barrier construction. Even construc-tion of barriers long distances in front of the advancinglava may be impractical, because the flow could easilystop naturally before reaching the barrier. Barriersmust also consist of denser material than the lava, beattached to the ground, or be three times wider at theirbase than the flow thickness itself. Less dense materialwill simply be buoyed up by the lava and incorporatedinto the flow. If barriers only divert a flow, there isalways the problem of finding a safe diversion path thatdoes not affect someone else’s property or life. Only

additional study in the mechanics and behavior oflava flows will permit barrier construction to be usedeffectively.

Ballistics and tephra clouds (Prabaharan, 2002)

The eruption of Vesuvius in 1944 illustrates anothermajor hazard of volcanic eruptions. At that time theAllied war effort in the area was severely hamperedby the bombing of airfields, not by the Germans,but by liquid lava blobs tossed out by the volcano.Strombolian-, Vulcanian-, Surtseyan-, and Plinian-type eruptions all shoot out ash to heights greater than30 km. The larger particles consist of boulder-sizedblobs of fluid magma and remnant blocks of thevolcano walls. Measured velocities are in the range of75–200 m s-1, and maximum distances are obtainedwith trajectory angles of 45°. Measured distances forboulder material rarely exceed 5 km, althoughmaterial weighing 8–30 tonnes has been projectedover distances of 1 km or more. This type of debris canbe voluminous and hot, and can fall over a small area.It can also be extremely destructive. The density ofprojectiles varies considerably so that the kineticenergy of impact is wide-ranging. Larger bombs canbehave exactly as human-made projectiles in theirexplosive effect when hitting the ground. Panoramicphotographs of vegetated landscapes subject tosustained projectile bombardment look exactly likewar bombing scenes.

The production of tephra or ash can also be destruc-tive (Figure 11.2). The eruption of Mt Hekla, Iceland,in 1947 ejected 100 000 m3 s-1 of material, dropping ashas far away as Finland two days later. Tephra can rise atvelocities of 8–30 m s-1, with drift rates downwind of20–100 km hr-1. While much of this dust falls out locally,fine material < 0.01 mm in size can be thrown up toheights in excess of 27 000 m into the stratosphere.Here, residence times may be two or more years. Theeruption of Tambora in Indonesia on 5–10 April 1815 isthe largest recorded tephra eruption. It blasted 151 km3

of material into the atmosphere as fine ash, which wasresponsible for a cooling of the Earth’s surface temper-ature by 0.5–1.0°C. As already mentioned in Chapter 9,volcanic dust exerts a dramatic control on the Earth’stemperature, with many of the changes in temperatureover the last three centuries correlating well withfluctuations in volcanic dust production.



The area covered by tephra deposits can be consid-erable because most tephra falls from atmosphericsuspension within a short distance of the volcano. Fineash settling out from Krakatau in 1883 covered anarea of 800 000 km2. This volcanic material can beextremely fertile once incorporated into the soil as afine dust; however, it can have severe environmentalconsequences. Only a small proportion of volcano-related deaths can be attributed directly to ash fallout,nevertheless, ash fallout 70–80 km from Krakatau wasstill hot enough to burn holes in clothing and vegeta-tion. Tephra can also contain toxic fluoride compounds,which are poisonous to animals attempting to eatcontaminated fodder. The ‘haze famine’ of 1784 inIceland, and the death by famine of 80 000 peoplefollowing the eruption of Tambora in 1815, can beattributed to the destruction of vegetation by tephra.The glassy tephra from Mt St Helens in 1980 had asevere impact on road conditions and on motorvehicles. Tephra from the 1982 eruption of Galung-gung volcano in western Java almost brought down, inseparate incidents, two 747 passenger jets bound forAustralia. The initial volcanic eruption had not beennoticed nor reported to aviation authorities. The firstplane to succumb was a British Airways 747 flyingbetween Singapore and Australia, when it entered theash cloud at 11 300 m. All four engines cut out, oneafter the other, sending the plane into a 13-minutesilent descent, through 7300 m, before the pilot could

restart three of the engines. Impact with the ash pittedthe cockpit windscreen so badly that the pilot had toland the plane by peering out a side window. Twoweeks later, a Singapore Airlines 747 flying the sameroute strayed into another ash cloud and dropped2500 m before the pilot regained control. Only afterthis second incident was a warning issued for planes toavoid the area. Since then more than 80 commercialairplanes have flown into clouds of volcanic ash.Mt Pinatubo in 1991 damaged 20 commercial planes intransit at a cost of $US100 million. Explosions onRedoubt, Galunggung, Mt Spurr, and Mt Pinatubosent clouds of ash into the atmosphere that damagedplanes 150, 200, 1200, and 1740 km, respectively, fromthe site of the eruptions.

Pyroclastic flows and base surges(Branney & Zalasiewicz, 1999)

Mention has already been made of the collapsing ashcolumns – termed ‘nuées ardentes’ or ‘pyroclasticflows’ (Figure 11.3) – and the basal surges produced bythe lateral explosion of volcanoes. Base surges are akinto the lateral blast waves accompanying a nuclearexplosion, and can pick up considerable dust inaddition to carrying the ash originating from thevolcano itself. The Bandai eruption of 1888 saw a300 m high volcano pulverized and converted to adebris avalanche in this manner. The blast wave canreach speeds in excess of 150 km hr-1, and the dust cansandblast objects. The Taal eruption in the Philippinesin 1965 ablated 15 cm of wood from the sides of treeswithin 1 km of the eruption. The basal surge may befollowed by a pyroclastic flow. The Mt St Helenseruption followed this sequence. The classic descrip-tion of a pyroclastic flow is ‘an avalanche of an exceed-ingly dense mass of hot, highly charged gas andconstantly gas-emitting fragmental lava, much of itfinely divided, extraordinarily mobile, and practicallyfrictionless, because each particle is separated from itsneighbors by a cushion of compressed gas’ (Perrett,1935). In pyroclastic flow, fine particles are suspendedand transported by turbulent whirlwinds of gas. Whileappearing as choking clouds of ash, the particles are sowidely spaced that they rarely collide. As these turbu-lent eddies pass over a spot, they can deposit alternat-ing layers of coarse and fine sediment as the velocity ofthe current waxes and wanes in a similar fashion togusts of wind. The destructive force of the flow can

Fig. 11.2 Tephra ejecta from Mt St Helens, 22 July 1980. Pyroclasticflows generated the lower ash clouds to the right. Notethe spreading of the ash cloud at the tropopause and thefallout of coarse particles downwind (photograph courtesyof the United States Geological Survey).

the flow then a process called fluidization may allowsediment particles in dense slurries to move as a fluidand remain unsorted. Different processes probablyoperate at different levels in pyroclastic flows.Turbulence lifts finer particles into the current higherup, while at ground level it enhances fluidization.Similarly, the falling-out of particles from slurries nearthe ground entraps smaller particles into a depositwhile expelling water. The latter process also enhancesfluidization. From time to time, turbulent vorticespenetrate to the bed allowing the deposition of alter-nating layers of fine and coarse material. The resultingproduct is a disorganized deposit containing a widerange of particle sizes with evidence of layering.

Not only can the blast from the flow be destructive,but the flow can also be extremely hot. Pyroclasticflows are also called ‘glowing avalanches’ because ofthe presence of heated debris within the flow. TheMt St Helens flow produced temperatures around 3500°C near the source and 50–200°C near themargins of the flow. These temperatures weresustained for less than two minutes. Temperatures inthe 1902 Mt Pelée flow were as high as 1075°C, asshown by the partial melting of gold coins and bottles.Temperatures can vary as much as velocities do. Forexample, the Mt Vesuvius flow of 79 AD was hotenough to carbonize wood in places, while food inother locations was left uncooked. The deposits frompyroclastic flows can remain hot for considerableperiods of time. Temperatures in excess of 400°C werefound within one deposit from the Augustine, Alaska,eruption of 1975, several days after the event.

Pyroclastic flows can also travel at high speedsbecause they behave like a density flow under theeffects of gravity. Velocity measurements of small ashflows range between 10 and 30 m s-1, while larger flowscan obtain speeds of 200 m s-1. At these velocities,the flows can override high topographic barriers. TheMt Pelée flow of 1902 refracted around topographicobstacles, and appeared to increase in velocity seaward.Both basal surges and pyroclastic flows are exceedinglydeadly because of their speed, and the fact that they aremixed with large amounts of hot, toxic gases.

Historically, pyroclastic flows have covered substan-tial areas. The Katmai eruption in Alaska in 1912 – oneof the biggest flows measured – covered an area of126 km2, while the Mt St Helens pyroclastic flowsextended over 600 km2. The Rabaul eruption,1400 years ago, produced a flow 2 m deep over


thus vary over short distances. The Mt Vesuvius pyro-clastic flow of 79 AD knocked over walls and statues insome places, while short distances away furnitureinside buildings was left undisturbed. Where the cloudmeets the ground, conditions are different. Sedimentconcentrations increase and sediment particles rangingin size from silts to boulders undergo billions ofcollisions. The momentum of the flow is equalizedbetween the particles and the flowing current of air. Insome cases, the grains may flow independent of anyfluid under a phenomenon known as granular flow.Granular flow tends to expel coarse particles to thesurface; however, if fluid moves upwards through

Fig. 11.3 Beginning of a pyroclastic flow or nuée ardente onMt Ngauruhoe, New Zealand, in January 1974. Tephra hasbeen blasted vertically only to collapse under its ownweight. The pyroclastic flow consists of ash suspended inhot carbon dioxide, and can be seen as the rapidly movingcloud closely hugging the ground to the left in thephotograph (photograph courtesy of the United StatesGeological Survey, Catalogue of Disasters #0401-09j).

1200 km2. Eruptions in the Taupo volcanic zone ofNew Zealand are the largest to have occurred onEarth. There is geological evidence that Taupo hasdischarged pyroclastic material at the rate of1 km3 min-1. The 186 AD eruption of Taupo produced60 100 km3 of tephra, an amount five times greaterthan that produced by Krakatau in 1883. The tephracovered all of the North Island of New Zealand with atleast 10 cm of ash, while intense pyroclastic flowsoverran 1000 m high ridges around the crater andflooded an area of 15 000 km2.

From ignimbrites – deposits resulting from pyro-clastic flows – there is some evidence that the flowshave climbed ridges 500–1000 m high up to 50 kmfrom the source of eruptions. At these high velocities,the flows can also be erosive, gouging out channels inless resistant sediments. Ignimbrites produced byPeléean volcanoes are distinct from ash fallout depositsresulting from Plinian eruptions. Because particlessettle from suspension in air, tephra deposits from thelatter tend to be well-sorted and spread evenly overthe landscape. Ignimbrites on the other hand consistof particles of all sizes that have been transported withair as the interstitial medium. They are made up ofvolatile and gas-rich magmas of a rhyolitic or andesiticnature, are chaotically sorted, and tend to pool indepressions because their flow is controlled by gravity.The heat in a pyroclastic flow can weld ignimbritematerial together, whereas Plinian deposits remainunconsolidated. Some ignimbrites tend to have aninverted size grading in which finer material is overlainby coarser blocks. For this to happen, the cloud mustbe very dense, travel at high velocities and be thin.Such a process is called shear sorting, and is analogousto large lumps in a sugar bag coming to the surfaceas the bag is shaken. Alternatively, these well-sorteddeposits may represent evidence of base surges, whichusually produce thin, well-sorted beds less than 10 mthick. Ground surges, thus, appear to precede alarge and slower moving pyroclastic flow, resulting inchaotically sorted ignimbrites being deposited on topof well-sorted ground-surge deposits. Because well-sorted ignimbrites arc plentiful, basal surge events maybe more common than previously thought.

Gases and acid rains (Symons et al., 1988)

Many pyroclastic flows were believed at first to consistsolely of hot gases. While this has been found to be

untrue, gases cannot be ruled out as a hazard. Onaverage, explosive volcanoes eject 4 � 106 tonnes ofsulfur dioxide into the atmosphere annually. This ishighly variable from year to year. For example, the MtPinatubo eruption of 15 June 1991 released 20 � 106

tonnes of sulfur dioxide in the space of a few weeks.The more explosive the eruption, the greater the trans-port of gas on adsorbed particles. Passively, degassingvolcanoes annually release another 9 � 106 tonnes ofsulfur dioxide. While appearing voluminous, theseamounts represent only 5–10 per cent of the annualanthropogenic flux to the atmosphere. However,emissions of sulfur dioxide by volcanoes can havesignificant local and regional effects, mainly on crops.Carbon monoxide emissions are more dangerous,being toxic to mammals in low quantities, and inhibit-ing leaf respiration in plants. Even CO2 can proveharmful. Volcanoes can produce large quantities of thisgas (20–95 per cent of the gas discharge) which, beingdenser than air, can pool in depressions and suffocatelivestock. On 21 August 1986, these gases, plus sulfurdioxide and cyanide, were released by a landslide fromthe bottom sediments of Lake Nios, a dormantvolcanic crater in Cameroon, Africa. Clouds of deadlygases, reaching concentrations of 20–30 per cent CO2,hugged the ground and flowed under gravity intotopographic lows killing all animal life in their path.Over 1700 people died within a matter of minutes,while 10 000 survivors suffered skin burns. A gasdischarge in neighboring Lake Monoun, Cameroon,killed 37 people in 1984, while a similar disaster on theDieng Plateau in Java in 1978 killed 180 people.

Many gases react with water vapor under the hightemperatures of volcanic eruptions to form hydro-chloric, sulfuric, carbonic, and hydrofluoric acids.Passively degassing volcanoes release annually intothe atmosphere 0.05–4.7 million tonnes of fluorine and0.3–10 million tonnes of chlorine as hydrofluoricand hydrochloric acid, respectively. Icelandic eruptionsare notoriously high in fluorides. The hydrofluoric acidcontent in the Hekla, Iceland, eruption of 1970reached 1700 ppm. Notable eruptions containinghigh amounts of HCl have been Agung, Bali(1.5 million tonnes), in 1963; Augustine, Alaska (525 000 tonnes) in 1976; Soufrière, Guadeloupe(1 million tonnes) in 1979; and Mt Erebus in theAntarctic (370 000 tonnes) in 1983. In the troposphere,acid can be precipitated out of the atmospherethrough condensation, scavenged by ash, or dissolved



in water-rich eruption plumes. Above 6000°C, hydro-fluoric acid forms calcium fluorosilicate that adheres tothe glassy surfaces of tephra particles. Despite thisculling, acids can still pose a hazard over a large area.The Katmai eruption of 1912 dispersed a sulfuricacid rain that damaged clothes hung out to dry inVancouver, 2000 km to the south. The Laki, Iceland,eruption in 1783–1784 filled the skies over most of thenorth Atlantic, western Eurasia, and the Arctic with adry sulfuric fog that led to a cold winter. Explosiveeruptions can also inject chlorine in the form of HCldirectly into the stratosphere where it can react withozone. Tambora in 1815 and Krakatau in 1883 wereexplosive eruptions, ejecting a minimum of 2.1 and3.6 million tonnes of chlorine as HCl, respectively. Theeruption of El Chichon, Mexico, in 1982 released40 000 tonnes of HCl into the stratosphere between20 and 40°N latitudes.

Lahars(Newhall & Punongbayan, 1996)

One of the more unusual, though still hazardous, phe-nomena produced by volcanoes consists of lahars, ormudflows, that can occur at the time of the eruption(primary lahars) or several years afterwards (secondarylahars). About 5.6 per cent of volcanic eruptions in thelast 10 000 years have produced mudflows at sometime. Primary lahars can be generated by pyroclasticflows or by eruptions of crater lakes. In the former

case, a pyroclastic flow can easily entrain water fromstreams and rivers as it moves down topographic lows.In the process, the gas-rich flow is slowly convertedto a fast moving, heated mudflow as more water isentrained in the mix. The Toutle River lahars (seeFigure 11.4) from Mt St Helens in 1980 had this origin.Volcanoes with crater lakes can produce mudflows atthe time of any eruption, if the crater lake is ruptured.The size of the mudflow is then related directly to thevolume of water in the lake. The 1919 eruption ofMt Kelat on Java expelled water from a crater lake,covering 200 km2 of farmland and killing over 5000people. Attempts have been made to control the Kelatsituation by drilling tunnels through the walls of thecrater to lower the level of the lake. The first attemptwas in 1929, when Dutch engineers reduced the lakevolume from 21 to 1 million m3. A subsequenteruption in 1951 blocked the tunnels and, even afterrepairs, an eruption in 1966 generated lahars that tookhundreds of lives. Indonesian engineers have sincereplaced the tunnels and drained the lake completelyto negate the threat of future lahars.

Secondary lahars are caused by rain falling onfreshly deposited, uncompacted tephra. Such water-soaked material is very unstable, and can move down-slope as a mudflow that entrains all loose debris in itspath. If acids have not been leached out of the tephra,then lahars can become acidic enough to cause seriousburns. Such flows have covered enormous areas. One

British Columbia

Mt Olympus

Mt St Helens

Mt Rainier

Mt Baker

Yakima

Mt Hood

Mt Adams

Idaho

Washington

NOregon

Mt Shasta

Lassen Peak

Cas

cad

es

N. Toutle R.

S. Toutle R.

Lahar

Muddy R.

Kelso

Columbia R.

Pyroclasticflow

0 10 20 30 km

Mt St Helens

Fig. 11.4 Location map of Mt St Helens (adapted from Blong, 1984).

4–8 m s-1 with peak discharges of 200–1200 m3 s-1. Afew lahars reached speeds of 11 m s-1, with peakdischarges of 5000 m3 s-1. Lahars were triggered by aslittle as 6 mm of rainfall in half-an-hour. By the end ofthe monsoon season, these lahars had moved about0.9 km3 of sediment and buried 300 km2 of lowland.The 1992 monsoon season yielded over 2000 mm ofrain that generated lahars that moved 0.5–0.6 km3 ofsediment. Typically, 1000 m3 of ash was removed fromeach square kilometer of upland watershed by eachmillimeter of rainfall. Each subsequent monsoonseason reiterated the process but with diminishingvolumes of sediment being transported. Because ofthe installation of warning systems, the lahars after themain eruption killed few people.

Because lahars are restricted to topographic lows,they can be predicted in advance. The build-up oftephra on a volcano’s slopes can be easily measured, sothat steps can be taken to evacuate people in threat-ened areas. After the Usu eruption in 1977 in Japan, itwas calculated that tephra deposits thicker than 0.5 m,on slopes longer than 300 m and steeper than 17–18°,would require only 20 mm of rain at a rate of10 mm hr-1 to initiate lahars. The 86 lahars thatoccurred afterwards closely followed these predictions.In New Zealand, lahars threaten the Whakapapa ski-field situated on the north-west slopes of Mt Ruapehu.Here, seismometers have been installed to detect smallearth movements indicating the start of a lahar. Thesetrigger alarms on the ski-fields giving skiers 5–10minutes to ski out of valleys. While lahars are verycommon and can reach velocities in excess of20 km hr-1, they can be prevented in some cases.Unfortunately, lahars can also be initiated by erosion ofold tephra deposits years, or decades, after the originaleruption has taken place. For instance, Mt Rainier wasaffected by a lahar in 1947 even though no volcaniceruption has been recorded historically.

Glacier bursts or Jökulhlaups(Halldórsson & Brandsdóttir, 1998)

Of particular danger during a volcanic eruption is themelting of glaciers or snowfields by hot lava or gases.The water can be subsequently heated and mixed withmud debris to form a steaming hot lahar. These eventsare termed jökulhlaups, an Icelandic word meaning‘glacier bursts’. Jökulhlaups are restricted mainly toIceland and the Andes; however, they can occurelsewhere. For example, the Mt St Helens lahars were


on Mt Rainier in Washington State 5000 years ago hada volume of 1.9 � 109 m3 and flowed 60 km from thesummit. Other lahars preserved in the geologicalrecord in the United States have extended over5000–31 000 km2. Secondary lahars following thePlinian eruption of Mt Pinatubo in the Philippines inJune 1991 left more than 50 000 persons homeless, anddisturbed the livelihoods of 1 400 000 people in 39towns and four large cities over an area of 1000 km2

(Figure 11.5). The eruption was one of the largest ofthe twentieth century and culminated on 15 June in anexplosive eruption that caused the summit to collapseinto a 2.5 km diameter caldera. Pyroclastic flowsexceeded 5–6 km3 and spread up to 16 km from thevolcano over an area of 400 km2. The final eruptiondischarged 3.4–4.4 km3 of ash. On the day of theclimactic eruption, Typhoon Yunya passed close tothe volcano. First its winds scattered tephra to a thick-ness of 10–33 cm over an area of 2000 km2, second itsrain soaked into the ash and caused many buildings tocollapse and, third, runoff turned the pyroclastic flowsinto enormous lahars. The following monsoon rainscontinued the threat, repeatedly generating hot laharsthat buried towns and agricultural land under 5–30 mof ash, destroyed bridges, and cut roads. A typical laharwas 2–3 m deep and 20–50 m wide. It consistedof 50 per cent ash moving as slurry at velocities of

Fig. 11.5 Lahar at Bamban following the eruption of Mt Pinatubo inJune 1991. The lahar has cut the main road linking majorcities in the north. As well, it has caused vertical accretionof the river channel. To prevent sediment-laden watersspilling across the floodplain and into the town, leveeshave been built to contain the lahar (photo by Cees J. vanWestern, International Institute for Aerospace Survey andEarth Sciences, The Netherlands. Taken from NOAA’sNational Geophysical Data Center hazard photos<http://www.ngdc.noaa.gov/seg/hazard/slideset/36/36_728_slide.shtml>).

probably exacerbated by melting of snow and ice byhot gases. In November 1985, over 20 000 people werekilled by a glacier burst mudflow that swept off theNevado del Ruiz volcano in Colombia. This wasthe worst such event in the twentieth century.Jökulhlaups in Iceland are caused by continuousmelting of the Myrdalsjökull or Vatnajökull icecaps,which lie overtop active volcanoes. The Vatnajökullglacier caps the 40 km2 Grimsvötn caldera, whichcontinually emits hot gases. The thickness of theseglaciers prevents heat escape, while their weightprevents water from flowing out of the caldera.Meltwater builds up slowly beneath the ice untilwater depths are sufficiently deep to float the glaciers.Jökulhlaups have dominated Iceland since firstreported in the twelfth century. Until the twentiethcentury, they occurred every 5–15 years. During thetwentieth century, large jökulhlaups took place in 1903,1913, 1922, 1934, 1938 and 1996. Smaller events nowoccur 2–3 times per decade. Discharges during ajökulhlaup create some of the largest floods on Earth.In 1918, the discharge of a jökulhlaup from theMyrdalsjökull icecap exceeded three times the dis-charge of the Amazon, the largest flowing river inthe world. The Katla volcano in Iceland has hadglacier bursts from the Vatnajökull glacier exceeding92 000 m3 s-1, with a total volume in excess of 6 km3.The latter floods over time have formed an outwashplain 1000 km2 in area. The last major jökulhlaupoccurred at Grimsvötn on 5 November 1996. Flowpeaked at 45 000 m3 s-1 as 3 km3 of heated waterdischarged over three days across the southernoutwash plain of Iceland. Blocks of ice 15 m indiameter and weighing up to 1000 tonnes knocked outbridges and destroyed the main highway.

Water discharge can also accompany volcaniceruptions in areas without snow or ice. The 1902 erup-tions of Mt Pelée and Soufrière in the West Indies wereboth accompanied by massive flooding of dry rivers.The 1947 eruption of Mt Hekla, Iceland, led to a dis-charge of 3 000 000 m3 of water, which could not beattributed solely to melting of snow or ice. These latterevents suggest that water may be expelled from thewatertable by escaping gases during some volcaniceruptions. Nor do these discharges have to accompanythe eruption. In 1945, the Ruapehu volcano in NewZealand erupted, emptying its crater lake. This lakeslowly refilled inside a newly formed crater that becamedammed by glacial ice. In 1953, an ice cave formed that

began lowering the crater. On Christmas Eve of thatyear the crater wall suddenly collapsed near the cavecausing the lake to drop 6 m in two hours at the rate of900 m3 s-1. The floodwater swept down the WhangaehuRiver picking up mud and boulders and forming a lahar.Two hours later, at 10:00 pm, the debris-laden waterswept away a section of the Tangiwai Rail Bridge, threeminutes before the Wellington–Auckland night trainarrived. The train plunged through the gap, killing151 people.

VOLCANIC DISASTERS

Volcanoes are visually one of the most spectacularnatural hazards to occur, and probably one of the mostdevastating in terms of loss of human life. Even theSahelian droughts, or earthquakes that completelyflatten cities, leave more survivors than dead.Volcanoes have wiped out entire populations. Therewere only two survivors out of a population of 30 000in the town of St Pierre, Martinique, following theMt Pelée eruption of 1902. Four volcanic events standout in the historic record. Three of these, Mt Vesuvius,Krakatau, and Mt Pelée, are significant because ofeither the enormity of the eruption, or the resultingdeath toll. The other, Mt St Helens, stands out becauseit represents a major eruption beginning a period ofvolcanic activity at the end of the twentieth century.

Santorini, around 1470 BC (Yokoyama, 1978; Pichler & Friedrich, 1980; Kastens & Cita,1981; Cita et al., 1996; Pararas-Carayannis, 1998)

The prehistoric eruption of Santorini, also known asStronghyli – the round island – occurred about1470 BC off the island of Thera, north of Crete inthe southern Aegean Sea (Figure 11.6). Probably thebiggest volcanic explosion witnessed by humans, it isalso one of the most controversial because legend,myth, and archaeological fact are frequently inter-twined and distorted in the interpretation of events.The eruption has been linked to the lost city of Atlantis(described by Plato in his Critias), to destruction ofMinoan civilization on the island of Crete 120 km tothe south (Figure 11.6), and to the exodus described inthe Bible of the Israelites from Egypt. Certainly, Greekflood myths refer to this or similar events that gener-ated tsunami in the Aegean. Plato’s story of Atlantis isbased on an Egyptian story that has similarities withCarthaginian and Phoenician legends. There is no



doubt that the great Bronze Age Minoan Empiredeclined around the time of the eruption.

Santorini is part of a volcanic island chain extendingparallel to the coast of Asia Minor. The volcano haserupted explosively at least twelve times during the lasttwo hundred thousand years. Its height has beenreduced over this time from a single mountain 1500 mhigh to a submerged caldera ringed by three islandsless than 500 m in height. The timing of the last majoreruption around 1470 BC is disputed. It has beenlinked to the plagues of Egypt (Exodus 6:28–14:31)and the exodus of the Israelites from that country. In1 Kings 6:1, the exodus is dated as occurring 476 yearsbefore the rule of King Solomon, which began in960 BC, putting the Exodus around 1436 BC.Other evidence indicates that the exodus occurred in1477 BC. Both dates encompass the reign in Egypt ofeither Hatshepsut or her son Tuthmosis III of theeighteenth dynasty. The transition between the tworulers is known as a time of catastrophes. In the biblicalaccount, the river of blood may refer to pink ash fromthe Santorini eruption preceding the explosion. Thisash, after it was deposited, would easily have mixedwith rainwater and flowed into any stream or river,coloring it red. The three days of darkness possiblyrefer to tephra clouds blowing south across Egypt atthe beginning of the eruption. The darkness wasdescribed as a ‘darkness, which could be felt’. Egyptian

documents around 1470 BC refer to a time ofprolonged darkness and noise; to a period of nine daysthat ‘were in violence and tempest: none ... could seethe face of his fellow’, and to the destruction of townsand wasting of Upper Egypt. Acidity in Greenland icecores indicates that the major eruption occurred in1390 ± 50 BC, although radiocarbon dating onland suggests an age around 1450 BC or 1470 BC.Dendrochronology based on Irish Bog Oak and Cali-fornian Bristlecone Pine puts the age of the event asearly as 1628 BC. At this latter time, Chinese recordsreport a dim sun and failure of cereal crops because offrost. Large volcanic events cool temperatures globallyby as much as 1°C over the space of several years. Therange of dates may not be contradictory because thereis evidence that Santorini may have erupted severaltimes over a time span of two hundred years.

The eruption around 1470 BC had four distinctphases. The first was a Plinian phase with massivepumice falls. This was followed by a series of basalsurges producing profuse quantities of pumice up to30 m thick on Santorini. The third phase was associ-ated with the collapse of the caldera and production ofpyroclastic flows. About 4.5 km3 of dense magmawas ejected from the volcano producing 10 km3 of ash.The volume of ejecta is similar in magnitude to thatproduced by the Krakatau eruption in 1883. The ashdrifted to the east-south-east, but did not exceed 5 mm

30˚

40

20 30 40

3

12

4

4

KnossosCrete

Anapi

SantoriniTurkey

Rhodes

Sites of homogenites

Tsunami wave energydispersed by refraction

Tsunami wave energyconcentrated by refracton

Direction of pumice drift

Black Sea

Africa

Cyprus

Jaffa–Tel AvivGreece

Italy

Mediterranean Sea

Fig. 11.6 Eastern Mediterranean region affected by the Santorini eruption around 1470 BC. The numbers refer to sites where homogenites exist: 1)Calabrian Ridge, 2) Ionian Abyssal Plain, 3) Bannock Basin, and 4) Mediterranean Ridge. Refraction patterns based on Kastens and Cita (1981).

thickness in deposits on any of the adjacent islands,including Crete. The largest thickness of ash measuredin marine cores appears to originate from pumice thatfloated into the eastern Mediterranean. It is possibleat this stage that ocean water made contact with themagma chamber and produced large explosions thatgenerated tsunami. The final phase of the eruption wasassociated with the collapse of the caldera in its south-west corner. The volcano sank over an area of 83 km2

to a depth of 600–800 m. This final collapse producedthe largest tsunami, directed westward (Figure 11.6).It is estimated that the original height of the tsunamiwas 46–68 m, and maybe as high as 90 m. The averageperiod between the dozen or more peaks in the wavetrain was fifteen minutes.

Evidence of the tsunami is found in deposits close toSantorini. On the island of Anapi to the east, sea-bornepumice was deposited to an altitude of 40–50 m abovepresent sea level. Considering that sea levels at thetime of the eruption may have been 10 m lower, thisrepresents run-up heights greater than those producedby Krakatau in the Sunda Strait. On the island ofCrete, the wave arrived within thirty minutes with aheight of approximately 11 m. Refraction focused waveenergy on the north-east corner of Crete, where run-up heights reached 40 m above sea level. In the regionof Knossos, the tsunami swept across a 3 km widecoastal plain reaching the mountains behind. Thebackwash concentrated in valleys and watercourses,and was highly erosive. Evidence for the tsunami is alsofound in the eastern Mediterranean on the westernside of Cyprus, and further away at Jaffa–Tel Aviv inIsrael. At the latter location, pumice has been found ona terrace lying 7 m above sea level at the time of theeruption. However, the tsunami wave here had alreadyundergone substantial defocusing because of waverefraction as it passed between the islands of Crete andRhodes. The greatest tsunami wave heights occurredwest of Santorini. The wave in the central Mediter-ranean Sea was 17 m high; while closer to Italy over thesubmarine Calabrian Ridge it was 7 m high. Bottomcurrent velocities under the wave crest in these regionsranged from 20 to 50 cm s-1 – great enough to entrainclay- to gravel-sized particles. The maximum pressurepulse produced on the seabed by the passage ofthe wave ranged from 350 to 850 kdynes cm-2.Spontaneous liquefaction and flow of water-saturatedmuds is known to occur under pressure pulses of280 kdynes cm-2 and greater.

Vesuvius (25 August 79 AD)(Bolt et al., 1975; Whittow, 1980; Blong, 1984; Sigurdssonet al., 1985)

Mt Vesuvius lies on the south-east corner of the bay ofNaples on the west coast of Italy (Figure 11.7). TheRoman towns of Pompeii and Herculaneum weresituated at the base of the south-western slopes, 5 kmfrom the mountaintop. Both towns were prosperousregional centers serving as summer retreats for wealthyRomans. Mt Vesuvius, prior to its eruption, consistedof a single, flat-topped cone with a small crater. TheRomans recognized it as a volcano, but there is noRoman or Etruscan record of its ever being active. In63 AD, a major earthquake struck the region with itsepicenter at Pompeii. Both the towns of Pompeii andHerculaneum were severely damaged, and by the timeof the eruption in 79 AD, only minor rebuilding hadbeen completed in Pompeii.

The description of the eruption comes from the eye-witness account of Pliny the Younger, who recounts thedeath of his father during the eruption. The writings ofCassius provide descriptions of the destruction of thetwo cities. Pliny the Elder was the commander ofthe Roman fleet at Misenum, 30 km to the west at theentrance to the Bay of Naples. On 24 August, Vesuviusbegan to erupt, accompanied by violent earthquakes.Pliny sailed to the scene to rescue inhabitants at thebase of the mountain. He could not get near the shorebecause of a sudden retreat of the shoreline, and hisboats were covered in a continuous ash fall. He sailedthat night to Stabiae to stay with a friend. Unfortu-nately, Stabiae was downwind from Pompeii and,during the night, the ash fall became so deep thatpeople realized they would have to flee for their ownsafety. Pliny the Elder reached the coast during thedaylight hours of 25 August, but the ash fall was soheavy that it still appeared to be night. As the magmachamber emptied, the top of the volcano collapsedinward, consuming 50 per cent of its volume and pro-ducing a caldera 3 km in diameter. In the early hoursof 25 August, a series of six surges and pyroclastic flowsswept the area. The first surge overwhelmed Hercula-neum. All remaining inhabitants met a grisly deathfrom asphyxiation as they sheltered on the beach. Thefourth surge reached Pompeii at about 7:00 am. Thelast two surges were the largest, and reached the townsof Stabiae and Misenum. At Stabiae, Pliny the Elderdied on the beach, apparently from a heart attack. It



took three days before it was light enough to recoverhis body. The pyroclastic ash falls buried Stabiae andPompeii to depths of 2 and 3 m, respectively. Most ofthe residents of Pompeii appear to have fled, but10 per cent were literally caught dead in their tracks bythe pyroclastic flows that climaxed the eruption. Her-culaneum was also affected by the tephra fallout but, inthe latter stages of the eruption, was engulfed by amudflow 20 m thick. This lahar appears to have beencaused by groundwater expulsion as the calderacollapsed.

The eruption marks one of the first volcanic disas-ters ever documented. The sequence of events atVesuvius has been labelled a Plinian-type eruptionafter Pliny’s descriptions. Many of the buildings inPompeii and Herculaneum protruded above the ash.The buildings were ransacked of valuables, dismantledfor building materials, and subsequently buried bycontinued eruptions. The ash reverted to rich agricul-tural land and Pompeii was forgotten. It was not until1699 that Pompeii was rediscovered and excavated.Vesuvius has continued to erupt aperiodically until thepresent day. Subsequent major eruptions occurred in203, 472, 512, 685, 787 and five times between 968 and1037 AD. It then remained dormant for 600 years until1630, when it produced extensive lava flows that killed700 people. The last major eruption in 1906 producedtephra up to 7 m thick on the north-east side of themountain. About 300 people were killed, mainly under

collapsing roofs. If the 79 AD eruption recurred today,over 1 500 000 people would be affected.

Krakatau (26–27 August 1883)(Verbeek, 1884; Latter, 1981; Self & Rampino, 1981;Nomanbhoy & Satake, 1995; Winchester, 2003)

The eruptions of Tambora and Krakatau in thenineteenth century make Indonesia, with its densepopulation, one of the most hazardous zones ofvolcanic activity in the world. Krakatau lies in theSunda Strait between Sumatra and Java, Indonesia(Figure 11.8). The Javanese Book of Kings describes anearlier eruption that generated a sea wave thatinundated the land and killed many people in thenorthern part of Sunda Strait. Krakatau had last beenactive in 1681 and, during the 1870s, the volcanounderwent increased earthquake activity. In May of1883, one vent became active, throwing ash 10 kminto the air. By the beginning of August, a dozenVesuvian-type eruptions had occurred across the island.On 26 August, loud explosions recurred at intervals often minutes, and a dense tephra cloud rose 25 kmabove the island. The dust particles formed nuclei forwater condensation and muddy rain fell on adjacentislands. The explosions could be heard throughout theislands of Java and Sumatra. In the morning and laterthat evening, small tsunami waves 1–2 m in heightswept the strait striking the towns of Telok Betong onSumatra’s Lampong Bay, Tjaringin on the Java coast

Naples

HerculaneumPompeii

Sorrento

Stabiae

200

100502010

Pyroclastic flow

0 10 20 30 km

Agropoli

N

5020

10

Fig. 11.7 Location map of Mt Vesuvius (Blong, ©1984, with permission Harcourt Brace Jovanovich Group, Australia).


north of Pepper Bay and Merak. On the morning of27 August, four horrific explosions occurred. The firstexplosion, at 5:28 am, destroyed the 130 m peak of Per-boewatan, forming a caldera that immediately infilledwith seawater and generated a tsunami. At 6:44 am, the500 m high peak of Danan exploded and collapsed,sending more seawater into the molten magmachamber of the eruption and producing anothertsunami. A third explosion occurred at 8:20 am. Thefourth blast, at 9:58 am, tore apart the remaining islandof Rakata. Including ejecta, 9–10 km3 of solid rock wasblown out of the volcano. About 18–21 km3 of pyro-clastic deposits spread out over 300 km2 to an averagedepth of 40 m. Fine ash spread over an area of2.8 � 106 km2 and thick pumice rafts impeded naviga-tion in the region up to five months afterwards. Acaldera formed, 6 km in diameter and 270 m deep,where the central island had once stood. This final blastwas the largest sound ever heard by modern humans,and was recorded 4800 km away on the island ofRodriguez in the Indian Ocean, and 3200 km away atElsey Creek in the Northern Territory of Australia.Windows were shattered 150 km away. The atmos-pheric shock wave traveled around the world seventimes. Barometers in Europe and the United Statesmeasured significant oscillations in pressure over ninedays following the blast. The total energy released bythe third eruption was equivalent to 840 � 1015 joules.

The blasts were accompanied by a cloud of ash thatrose to a height of 30 km. The debris from this cloudformed deposits with a volume of 13 km3. On theisland remnants, 60 m of ignimbrite were deposited ontop of 15 m of tephra. Large rafts of pumice ejected bythe volcano blocked the Sunda Straits. Tephra fell overan area of 300 000 km2 and turned day into night200 km downwind for forty-eight hours. Dustencircled the globe within two weeks of the eruption,mainly within the zone of tropical easterlies. Withinthree months, dust had spread into the northernhemisphere, covering most of the United States andEurope. In France, solar radiation dropped 20 per centbelow normal and, three years afterwards, was still10 per cent below normal at the ground surface. Spec-tacular, prolonged sunsets were observed everywhere,caused by high altitude dust scattering incoming solarradiation as it entered the Earth’s atmosphere at lowangles. The first occurrence of such sunsets sparkedfalse alarms to fire departments in the eastern UnitedStates as residents reported what appeared to bedistant, but horrific, fires.

The two pre-dawn blasts each generated tsunamithat drowned thousands in the Sunda Strait. Thefourth blast-induced wave was cataclysmic and devas-tated the adjacent coastlines of Java and Sumatrawithin 30 to 60 minutes. The coastline north of theeruption was struck by waves with a maximum run-up

0 20 40 km

42

15

22

3630

24

1523

24

11

3

15

Borneo

Java

Sumatra

Malaysia

5˚

0˚

-5˚

-10˚

Sunda

StraitKelimbang

Merak

Anjer Lor

TjaringinBataviaKrakatau

Vlakken Hoek

Welcome Bay

Telok Betong

Coastline inundated by tsunami

Height of run-up (m)

105˚ 110˚ 115˚

Fig. 11.8 Coastline in the Sunda Strait affected by tsunami following the eruption of Krakatau on 26–27 August 1883 (based on Verbeek, 1884; Blong,1984 and Myles, 1985).

height of 42 m (Figure 11.8). The tsunami penetrated5 km inland over low-lying areas. The largest wavestruck the town of Merak. Here, the 15 m tsunami roseto 40 m because of the funnel-shaped nature of thebay. The town of Anjer Lor was swamped by an 11 mhigh wave, the town of Tjaringin by one 23 m in height,and the towns of Kelimbang and Telok Betong wereeach struck by a wave 22–24 m high. In the latter town,the Dutch warship Berouw was carried 2 km inlandand left stranded 10 m above sea level. The highestrun-up reached 42 m above sea level at Merak in thenorth-east corner of the Strait. Coral blocks weighingup to 600 tonnes were moved onshore. Within theStrait, eleven waves rolled in over the next fifteenhours, while at Batavia (now Jakarta) fourteen consis-tently spaced waves arrived over a period of thirty-sixhours. Between 5000 and 6000 boats were sunk in theStrait. In total, 36 417 people died in major townsand three hundred villages were destroyed because ofthe tsunami.

Within four hours of the final eruption, a 4 m hightsunami arrived at Northwest Cape, Western Australia,2100 km away. The wave swept through gaps in theNingaloo Reef and penetrated 1 km inland over sanddunes. Nine hours after the blast, three hundred river-boats were swamped and sunk at Kolkata – formerlyCalcutta – on the Ganges River, 3800 km away. Thewave was measured around the Indian Ocean at Adenon the tip of the Arabian Peninsula, Sri Lanka, Mahe inthe Seychelles Islands, and on the island of Mauritius.The furthest this tsunami wave was observed was8300 km away at Port Elizabeth, South Africa. Tsunamiwaves were measured over the next thirty-seven hourson tide gauges in the English Channel, the PacificOcean, and in Lake Taupo in the center of the NorthIsland of New Zealand – where a 0.5 m oscillation inlake level was observed. Around the Pacific Ocean, tidegauges in Australia, Japan, San Francisco, and KodiakIsland measured changes of 0.1 m up to twentyhours after the eruption. Honolulu recorded higheroscillations of 0.24 m with a periodicity of thirtyminutes.

The tsunami in the Pacific have been attributed tothe atmospheric pressure wave, because many islandseffectively obscure the passage of any tsunami from theSunda Strait eastward. The atmospheric pressure wavealso accounts for seiching that occurred in Lake Taupo,which is not connected to the ocean. Finally, it explainsthe long waves observed along the coasts of France and

England when the main tsunami had effectively dissi-pated its energy in the Indian Ocean. The generationof tsunami in Sunda Strait and the Indian Ocean hasbeen attributed to four causes: lateral blast, collapse ofthe caldera that formed on the north side of KrakatauIsland, pyroclastic flows and a submarine explosion.Lateral blasting may have occurred to a small degreeon Krakatau during the fourth explosion; however, itseffect on tsunami generation is not known. During thefinal explosion, Krakatau collapsed in on itself forminga caldera about 270 m deep and with a volume of11.5 km3. However, modelling indicates that thismechanism underestimates tsunami wave heights by afactor of three within Sunda Strait. Krakatau generatedmassive pyroclastic flows. These flows probably gener-ated the tsunami that preceded the final explosion.At the time of the final eruption, ash was ejected intothe atmosphere towards the north-east. Theoretically,a pyroclastic flow in this direction could have gener-ated tsunami up to 10 m in height throughout thestrait; however, the mechanism does not account fortsunami run-ups of more than 15 m in height in thenorthern part of Sunda Strait. The pyroclastic flow nowappears to have sunk to the bottom of the ocean andtraveled 10–15 km along the seabed before depositingtwo large islands of ash. The 40 m high run-upmeasured near Merak to the north-east supports thishypothesis. The tsunami’s wave height correspondswith the depth of water around Krakatau in thisdirection. Water was simply expelled from the seabedby the pyroclastic flows. As well, the fourth explosionof Krakatau, at 9:58 am, more than likely produced asubmarine explosion as ocean water came in contactwith the magma chamber. A submarine explosioncould have generated tsunami 15 m high throughoutthe Strait. If the explosion had a lateral componentnorthward, as indicated by the final configuration ofKrakatau Island, then this blast, in conjunction withthe pyroclastic flow, would account for the increase intsunami wave heights towards the northern entrance ofSunda Strait.

Mt Pelée (8 May 1902)(Bolt et al., 1975; Whittow, 1980; Blong, 1984;Scarth, 2002)

The year 1902 was not a good year for the residentsof the West Indies or the surrounding Caribbean(Figure 11.9). The Pacific coast of Guatemala wasstruck by a strong earthquake on 18 January and again



on 17 April. On 20 April, Mt Pelée began to erupt onthe island of Martinique and on 7 May Soufrière, onthe nearby island of St Vincent, erupted killing 2000people. Izalco in El Salvador erupted on 10 May,Masaya in Nicaragua erupted in July, and Santa Mariain Guatemala exploded on 24 October. The latter erup-tions mainly caused property damage. The worsteruption in terms of loss of life was the explosion ofMt Pelée on 8 May. It ranks as one of the mostcatastrophic natural eruptions witnessed.

The eruption of Soufrière on St Vincent, 160 kmsouth of Martinique, was a prelude to the Mt Peléedisaster. Soufrière became active in 1901 and, in Aprilof 1902, earthquake activity was severe enough thatmost residents were evacuated to the southern part ofthe island. On 6 May, the volcano started emittingsteam, and the Rabaka Dry and Wallibu Rivers becametorrents of muddy water. At 2:00 pm on 7 May,Soufrière generated a pyroclastic flow that destroyedmost of the northern part of the island and killed 1565people. Fortunately, the evacuations avoided a higherdeath toll.

Mt Pelée began to erupt on 20 April 1902. Themountain was conical in shape, with a notch on thesouth-west side that led into a topographic low calledthe Rivière Blanche (Figure 11.10). This valleydescends steeply for about 1.5 km and then bendssharply westward. The town of St Pierre lies on thecoast 3 km south of this bend. St Pierre was one of the

most prosperous cities in the Caribbean. Its businesscentered on the export of rum made from local sugarcane. The eruption was preceded by a long series ofnatural warnings foreboding imminent disaster. Lavawas found in the crater shortly after the initial eruptionand, in the last week of April, the volcano belched ashand sulfurous fumes so intense that birds dropped deadfrom the sky. Reports were issued that the volcano wasstill safe, and that St Pierre was protected from any lavaflows by the ridge between it and the Rivière Blanche.On the night of 3 May, Mt Pelée entered a new stage oferuption. The Roxelane River, running through StPierre, turned into a torrent of mud, knocking out thepower supply in the city. Fissures opened up on theslopes, and sent out steam and boiling mud that killed160 people in the town of Ajoupa-Bouillon. About10 000 people then fled the slopes and crowded intoSt Pierre. Biblical-type plagues then afflicted the cityand countryside. Snakes and insects driven from themountain by the heat and gas emissions invaded thecity’s suburbs. More than 100 deadly pit vipers enteredthe north of the city. The army was called out to shootthem, but not before over 50 people and 200 animalshad been killed. At the sugar plantation at the mouth ofRivière Blanche, ants and deadly centipedes (the latter0.3 m long) invaded the mill buildings, swarmed up thelegs of horses and the workmen, and bit them.

By 5 May, the crater at the top of the mountain hadfilled with muddy water. On 5 May, this was blasted

24 Oct. Santa Maria

18 Jan.17 Apr.

Earthquakes

July Masaya

N

10 May Izalco

8 May Mt Pelée

7 May Soufrière

0 200 400 600 km

Fig. 11.9 Location map of volcanoes and earthquakes in the Caribbean region in 1902.


out, sending a large mudflow down the RivièreBlanche to the coast. The lahar killed 25 people. Whenthe 35 m high wall of mud hit the ocean, it generated atsunami that crashed into the lower part of St Pierre,killing 100 people there. On 6 May, Mt Pelée issuedtephra that was so voluminous that it crushed roofs andclogged streets. All the rivers around the mountainthen flooded for the next two days as the mountainbegan roaring and blowing out pumice. Hot gasesrose to the surface, expelling groundwater from thewatertable. On 7 May, Mt Soufrière on the adjacentisland of St Vincent erupted and reports sooncirculated that this eruption had released the pressuresbuilding up under Mt Pelée. Then Mt Pelée quietened,supporting this rumor.

During these events, the government, headed byGovernor Mouttet, struggled to make sense of theeruptions. Scientific advice was sought, but there wasno expert living on the island. Rumors were dispelledby press releases stating that there was no threat.Comparisons were made with Krakatau, which haderupted nineteen years earlier. Because Mt Pelée wasnot near the ocean, the risk of tsunami was dismissed.Earthquakes were viewed as the main hazard. No onerealized that the nuées ardentes that had begun on7 May heralded an unfamiliar phenomenon that posedfar more of a threat. Without any adequate knowledgeof what to expect next, Mouttet – in what must be ratedas one of the most foolhardy acts in history – took his

wife and most of the senior members of his adminis-tration to St Pierre on the evening of 7 May toallay fears of impending doom amongst the citizens.Troops were ordered to begin patrolling the streets thefollowing morning to prevent panic. Those citizenswho wanted to leave St Pierre had already done so.

At 8:00 am on 8 May, a series of four violent explo-sions shot dust to heights of 15 km. A second cloud wasblasted as a basal surge, horizontally to the south-west.This cloud immediately flowed down the RivièreBlanche. When it came to the bend in the river valley,it overrode the bank, and part of the flow, travelingat velocities in excess of 160 km hr-1, swept throughthe city of St Pierre at 8:02 am. Eyewitness accounts,from the few people who had fled the city early thatmorning, describe the pyroclastic flow as a hurricane offlame. Temperatures within the gas cloud were as highas 1075°C. Walls of buildings were blown down, athree-ton statue of the Virgin Mary was tossed 12 m, 18of the 20 ships in the harbor were sunk, and most ofthe standing city set ablaze. The hot blast exploded arum distillery and ignited rum, which then flowedthrough the streets completing the incineration of thecity (Figure 11.11). The dust cover averaged only30 cm in thickness throughout the city, but it engulfedthe area in complete darkness. Towards the mouth ofthe Rivière Blanche, pyroclastic deposits thickened to4 m. Up to 30 000 people, including GovernorMouttet, died within two minutes. Many had clothing

N

Nueés ardentes

May 8

AfterMay 8

Grand Rivière

Rivière BlancheSt Pierre

Le Prêcheur

MacoubaBasse Pointe

Fort de France

0 10 20 km

Fig. 11.10 Location map of Mt Pelée.

minor volcano. However, research in the mid-1970sbegan to indicate that ash found in the region origi-nated mostly from Mt St Helens. On 20 March 1980, aswarm of micro-earthquakes around Mt St Helenssignaled renewed activity, which culminated in a smalleruption on 27 March sending clouds of ash 6 kmskyward. Harmonic tremors began on 3 April, signalingthe movement of magma into the magma chamberbelow the volcano. By this time, over 70 million m3 oftephra had been ejected. Prior to the eruption, thenorth side of the volcano bulged more than 150 m at arate of 1.5 m per day.

At 8:30 am on 18 May, an earthquake of magnitude5.1 on the Richter scale started landslides in the bulgeregion. Immediately, the volcano erupted and sent outa lateral explosion, which was felt 425 km away. A darkcloud of ash, containing 2.7 km3 of material weighing520 million tonnes, rose to a height of 23 km over thenext nine hours. The height of the mountain wasreduced by 500 m (Figure 11.12). Over 500 km2 offorest were devastated by the resulting base surge thatreached temperatures of 2600°C. The blast overroderidges over 10 km away. In some places, debrisdeposits accumulated to depths of 150 m. Snow and iceon the mountain were melted by the blast and formedlahars that rushed into Spirit Lake, filling the lake todepths of 60 m. The lahars then moved over 50 kmdown the north and south forks of the Toutle River,reaching Kelso on the Columbia River. Lahars alsoflowed down into the Muddy River to the east. Inall, over 300 km of roads and 48 road bridges wereextensively damaged. The lahars and fine ash thatsettled in the surrounding area were soon carried intothe Columbia River, where the 180 m wide shippingcanal was reduced from a depth of 12 m to one of4.3 m. Only 60 people lost their lives, mainly becausethe government had ordered evacuations as the inten-sity of the eruption increased. However, some of thisdeath toll included people who were permitted backinto the area only days before the eruption. Tephrafallout became the worst nuisance after the initialeruption. While most of the dust was blasted out theside (Figure 11.12), ash fell to the ground in easternNorth America and over the Atlantic. Within 17 days,ash had encircled the globe at a height of 9–12 km,at the top of the troposphere. Some ash moved intothe stratosphere, but this amount was minor. Temper-atures were estimated to have decreased by 0.5°C fora few weeks directly downwind because of the reduced


stripped from them by the force of the blast. Otherswere grotesquely disfigured, either as their body fluidsboiled and burst through their skin, or as their musclescontracted in spasms as they fought for air. There wereonly two survivors, one of whom, Auguste Ciparis, wasa condemned prisoner in the local jail. He sufferedsevere burns and later had his sentence commuted.Subsequent blasts on 19 May and 20 August sweptacross most of the northern and western slopes,causing further destruction. The last explosion killed2000 people in five mountain villages. Of all historicalvolcanic disasters, the residents of the city of St Pierreand the island of Martinique appear to have sufferedone of the most devastating eruptions known.

Mt St Helens (18 May 1980)(Hays, 1981; Lipman & Mullineaux, 1981; Keller, 1982;Blong, 1984; Coates, 1985)

The Mt St Helens eruption is notable for severalreasons. Firstly, it was the first large explosive eruptionto occur in the world for several decades. Secondly, itheralded a period of substantial volcanic activity at thebeginning of the 1980s. Thirdly, it had associated withit several hazard phenomena. Finally, it was the largesteruption to occur in the conterminous United States inrecorded history. Mt St Helens is situated in theCascade Mountains of Washington State (Figure 11.4).It had not had a major eruption for 123 years and,compared to Mt Rainier, was considered a relatively

Fig. 11.11 The city of St Pierre, Martinique, after the 8 May 1902eruption of Mt Pelée. The pyroclastic flow, whichdestroyed the city, swept over the ridge at the top ofthe photograph. The waterfront had been wrecked by atsunami caused by a lahar several days previously(photograph from a book by A. Lacroix, courtesy of theGeological Museum, London).

ranges. Mt St Helens also highlighted the fact that theprevious 60 years worldwide were generally quiescentvolcanically.

CONCLUDING COMMENTS

There is no doubt that the Earth is experiencing oneof the most intense periods of volcanism in the last10 000 years. This period began at the beginning ofthe seventeenth century, concomitant with globalcooling that peaked in the Little Ice Age. The volcanicevents of the latter half of the twentieth centurymust be viewed in this context, rather than as freakeruptions of supposedly dormant volcanoes. Theeruptions of El Chichon, Mt St. Helens, and MtPinatubo are the most dramatic examples of thisreawakening. While Mt Pinatubo stands out as adominant event, there is little realization that itwas but one of three major volcanic eruptions in1991–1992. The other two were Mt Hudson in Chileand Mt Spurr in Alaska. Since then, volcanic activityhas become quiescent. Whether or not this representsa resumption of inactivity similar to that between theeruption of Caribbean volcanoes in 1902 and MtAgung in 1961 is academic. However, since theeruption of Mt St Helens in 1980, renewed volcanicactivity and associated phenomena have taken morethan 24 000 lives. In our present era, volcaniceruptions are pervasive, unpredictable, and deadly.


Blong, R.J. 1984. Volcanic Hazards: A Sourcebook on the Effects ofEruptions. Academic Press, Sydney.


Branney, M and Zalasiewicz, J. 1999. Burning clouds. New Scientist17 July: 36–41.

Cita, M.B., Camerlenghi, A., and Rimoldi, B. 1996. Deep-seatsunami deposits in the eastern Mediterranean: new evidenceand depositional models. Sedimentary Geology 104: 155–173.

Coates, D.R. 1985. Geology and Society. Chapman and Hall, NewYork.

Fielder, G. and Wilson, I. 1975. Volcanoes of the Earth, Moon andMars. Elek, London.

Halldórsson, M.M. and Brandsdóttir, B. 1998. Subglacial volcaniceruption in Gjálp, Vatnajökull, 1996: The jökulhlaup. <http://www.hi.is/~mmh/gos/vat-update.html>

Hays, W.W. 1981. Facing geologic and hydrologic hazards: Earth-science considerations. United States Geological SurveyProfessional Paper 1240-B: 86–109.


incoming solar radiation, but the worldwide effect onclimate was virtually irrelevant. The greatest effectoccurred within 700 km of the eruption. Ash accumu-lated to depths of 10–15 mm, 150 km downwind of theeruption. It also made driving conditions in and aroundthe area treacherous for five days because of lowvisibility and slippery roads. Speed limits had to bereduced to 8–10 km hr-1 to prevent accidents. Thedeposition of ash proved a major clean-up problem inurban centers. The small town of Yakima, Washington,with a population of 50 000, took ten weeks toremove half-a-million tonnes of ash, a task that cost$US2 million. The fine dust clogged air filters on cars,got into brakes and wrecked motors. Dust fallout onelectrical equipment caused equipment failure, whilecrop yields noticeably decreased because the ashcoated leaves and lowered the efficiency of photosyn-thesis.

Given the magnitude of the event, the media hype,and the reaction by the public and government, thetrue economic effects of the disaster tended to be over-estimated. Washington State estimated the damage at$US2700 million, and Federal Congress appropriatedjust under $US1000 million for disaster relief. Ineffect, the eruption caused $US844 million damage.Clean-up cost $US270 million; agricultural lossesamounted to $US39 million; property damage, mainlyto roads and bridges, cost $US85 million; and com-mercial timber losses amounted to $US450 million.The eruption sparked renewed research interest intoother volcanoes along the Cascade and Sierra Nevada

Fig. 11.12 The north side of Mt St Helens after the 18 May 1980eruption. The crater has not formed by collapse, but by alateral blast. The material in front represents thepyroclastic debris from that blast (photograph courtesy ofJames Ruhle and Associates, Fullerton, California).

Prabaharan, D.J. 2002. Fear of flying: Assessing the risks of volcanicash clouds for aviation. GIS User 50: 22–23.

Ritchie, D. and Gates, A.E. 2001. Encyclopedia of Earthquakes andVolcanoes. Facts on File, New York.

Scarth, A. 2002. La Catastrophe: Mount Pelée and the Destructionof Saint-Pierre, Martinique. Terra Publishing, Harpenden.

Self, S. and Rampino, M.R. 1981. The 1883 eruption of Krakatau.Nature 294: 699–704.

Sigurdsson, H., Carey, S., Cornell, W., and Pescatore, T. 1985. Theeruption of Vesuvius in 79 AD. National Geographic Research 1:332–387.

Symons, R.B., Rose, W.I., and Reed, M.H. 1988. Contribution ofCl- and Fl-bearing gases to the atmosphere by volcanoes. Nature334: 415–418.

Tazieff, H. and Sabroux, J.C. (eds). 1983. Forecasting VolcanicEvents. Lange and Springer, Berlin.

Verbeek, R. D. M. 1884. The Krakatoa eruption. Nature 30: 10–15.Whittow, J. 1980. Disasters: The Anatomy of Environmental

Hazards. Pelican, Harmondsworth.Winchester, S. 2003. Krakatoa: The Day the World Exploded:

27 August 1883. Viking, New York.Wood, R.M. 1986. Earthquakes and Volcanoes. Mitchell Beazley,

London.Yokoyama, I. 1978. The tsunami caused by the prehistoric eruption

of Thera. In Thera and the Aegean World: 1. Proceedings ofthe Second International Scientific Congress, Santorini, Greece,August 1978, pp. 277–283.


Kastens, K.A. and Cita, M.B. 1981. Tsunami-induced sedimenttransport in the abyssal Mediterranean Sea. Geological Society ofAmerica Bulletin 92: 845–857.

Keller, E.A. 1982. Environmental Geology (3rd edn.) Merrill,Columbus, Ohio.

Latter, J.H. 1981. Tsunamis of volcanic origin: summary of causes, withparticular reference to Krakatau, 1883. Bulletin Volcanologique 44:467–490.

Lipman, P.W. and Mullineaux, D.R. (eds). 1981. The 1980 Eruptionsof Mount St Helens, Washington. United States Geological SurveyProfessional Paper No. 1250.

Myles, D. 1985. The Great Waves. McGraw-Hill, New York.Newhall, C.G. and Punongbayan, R.S. (eds) 1996. Fire and Mud:

Eruptions and Lahars of Mount Pinatubo, Philippines. PhilippineInstitute of Volcanology and Seismology, Quezon City andUniversity of Washington Press, Seattle.

Nomanbhoy, N. and Satake, K. 1995. Numerical computations oftsunamis from the 1883 Krakatau eruption. GeophysicalResearch Letters 22: 509–512.

Pararas-Carayannis, G. 1998. The waves that destroyed the MinoanEmpire (Atlantis). <http://www.geocities.com/CapeCanaveral/Lab/1029/.html> (URL defunct as of 2004)

Perrett, F.A. 1935. Eruption of Mt Pelée 1929–32. Carnegie Instituteof Washington Publication No. 458.

Pichler, H. and Friedrich, W.L. 1980. Mechanism of the Minoaneruption of Santorini. In Thera and the Aegean World: 2.Proceedings of the Second International Scientific Congress,Santorini, Greece, August 1978, pp. 15–30.

INTRODUCTION

One of the most widespread natural hazards is theunexpected and sometimes unpredictable movementof unconsolidated weathered material (regolith) orweathered rock layers near the Earth’s surface. Land-slides and avalanches, while historically not renownedfor causing as large a death toll as other natural disas-ters such as tropical cyclones or earthquakes, have hadjust as dramatic an impact on property and lives. Thesudden movement of slope material is as instantaneousas any earthquake event but it is a more widespreadproblem. In any moderate- to high-relief regionsubject to periods of high rainfall, slippage of part or allof the regolith downslope is probably the mostcommon hazard. Nowhere is this problem more preva-lent than in cold regions underlain by permafrost orground ice. Of a slower nature, and just as widespreada hazard, is land subsidence. While much of a landsurface may be stable or even flat, there is a wide rangeof natural processes that can generate ground collapse.Another important aspect of land instability is themultitude of factors that can trigger groundmovement. Almost all of the hazards presented in thisbook can generate secondary land instability problems.In many cases, associated landslides have contributedsignificantly to the large death toll from earthquakesand cyclones. Even droughts can exacerbate groundinstability through the process of repeated drying and

wetting of expansive clays. This can lead to surfacedeformation and eventual destruction of structureswith insufficient foundations.

In this chapter, a broad overview of the wide rangeof land instability hazards will be presented. Firstly, thebasic principles of soil mechanics will be described toshow how surface material becomes unstable, andto point out what factors exacerbate ground failure.Secondly, each of the main types of land instability willbe examined in turn, together with a description ofsome of the major disasters. To limit the coverageof such a wide topic as land instability, direct, human-induced and ice-related (cryogenic) factors will bediscussed only at a cursory level.

SOIL MECHANICS(Young, 1972; Chowdhury, 1978; Finlayson & Statham,1980; Goudie, 1981; Aune, 1983; Bowles, 1984)

Stress and strain

Consider a body of soil with mass � sitting on a slopeof angle . This mass is affected by gravity and tendsto move downslope. The effect of gravity is directlyrelated to the angle of the slope as follows:

effect of gravity on a slope = � sin (12.1)where � = mass

= the slope angle

Land Instability as a Hazard

C H A P T E R 1 2

The effect of stress upon a soil or a regolith is calledstrain. Strain is not directly equated with stress butincorporates the additional factor of slope angle. Strainmay not uniformly occur in the soil body, but may berestricted to joints where fracturing will eventuate.Strain can affect inter-particle movements, or act onthe overall soil column. In combination, these strainsresult from what is labelled ‘net shear stresses’.

Friction, cohesion and coherence

One of the forces of resistance against the movementof material downslope is friction. Friction is the forcethat tends to resist the sliding of one object over, oragainst, another. For regolith material sitting on a firmbase, friction is due to irregularities present at thecontact between these two materials no matter howsmooth this contact may appear to be. The irregulari-ties tend to interlock with each other and preventmovement. Generally, the greater the weight of anobject pressing down upon the contact, the greater the

amount of friction. While friction is dependent on thedegree of roughness of a surface, it is independent ofthe area of contact between a body of regolith and theunderlying substrata. Small areas of soil material tendto fail at the same angles as larger areas. Figure 12.1illustrates the application of stress to an object on aslope. There is a resisting force against movement dueto friction. Sliding will commence when the appliedstress exceeds maximum frictional resistance.

Friction is expressed as a coefficient µ and thiscoefficient is equal to the tangent of the slope angle atwhich sliding just begins:

tan � = (� sin�) (� cos�)-1 (12.2)where = angle of plane sliding friction

� = coefficient of friction

The equations of movement of objects can beexpressed in terms of critical frictional resistance(Rfcrit) and critical applied force (Fcrit). Failure com-mences when the critical applied force exceeds thecritical frictional resistance as follows:

Fcrit > Rfcritwhere Rfcrit ~ � tan� (12.3)

This equation implies that as the weight of an objectapproaches zero the force required to move it down-slope approaches zero. For non-rigid objects (uncon-solidated material) this is not so, because as the weightof the regolith approaches zero, there is still an addi-tional force resisting downslope movement. This forceis termed cohesion, defined as the bonding that existsbetween particles making up the soil body. The aboverelationship, now including cohesion, can be expressedas follows:

Fcrit > Rf crit (12.4)where Rf crit ~ � tan� + c

c = cohesion

Strictly, the term cohesion refers to chemical orphysical forces between clay particles. The term coher-ence is used to describe the binding of soil particles ofall sizes, as a group or mass, due to capillary cohesionby water at the interface between individual grains, bychemical bonds or by cementation. In cementation,the chemical bonds are considered primary and, thus,are very strong. Common cementing agents include


Any force (gravity is by far the most important) thattends to move material downslope is termed a stress(Figure 12.1). If a building is built on the slope, theweight of the building adds more stress to the soil.Additional weights on a slope are termed ‘stressincrements’. Stress increments can consist of rain, soilmoved from upslope, buildings, or any other mass.While gravitational stress is most obvious, there areother stresses that can exist in the regolith. Molecularstress arises with the movement of soil particles oreven individual molecules. It is associated with suchphenomena as swelling and shrinking of colloids (claysoil particles) on wetting and drying, thermal expan-sion and contraction, and the growth of ice crystals.Biological stress refers to the stress exerted on a soilbody by the growth of plant roots or the activities ofanimals.

ωsinα ωα

Fig. 12.1 Effect of gravity on an object with mass � on a slope withangle .

carbonates, silica, alumina, iron oxide, and organiccompounds. Cementation can also occur because ofcompaction, especially if material comes underpressure from above. The compaction process givesstability to materials on slopes. With certain grainshapes, especially with clays, it is possible to realigngrains so that they interlock effectively, increasingcoherence and the resisting force to movement.Chemical bonds consist mainly of oppositely chargedelectrical fields that develop on the surfaces of largemolecules, especially clay minerals. These attractingcharges are called Van der Waals’ bonds, which for clayminerals remain active even when the clay particles orcolloids are moved relative to each other. For clays, thisgives rise to plasticity, which will be described later.

Figure 12.2 presents the type of bonds that can existdepending upon particle size and the relative strength ofbonding. Van der Waals’ bonds are restricted to materialless than 0.03 mm or 30 microns in diameter. As thematerial gets smaller, the bonding strength increasesconsiderably until it reaches values of 1 kg cm-2 for sizesless than 0.001 microns. As grain size increases, capillarycohesion (mainly due to water) becomes dominant.Capillary cohesion is also very important in bonding clayminerals, creating forces three orders of magnitudestronger than Van der Waals’ bonds. The only way thatcoherence of material coarser than 1 cm can be achievedis by compaction and cementation.

The effect of capillary cohesion can be demon-strated easily. Damp sand can be shaped into almostvertical walls without any sign of failure. As the sanddries out and the cohesive tension of water at sandgrain interfaces is removed, the sand pile begins tocrumple, eventually reaching the angle of repose forloose sand, which is only 33°. Cohesion in damp sandis produced totally by the effect of water tensionbetween sand grains, and gives wet sand a remarkabledegree of stability. Note, however, that the sand cannotbe too wet, otherwise liquefaction occurs. In otherwords, if more water (or any other material for thatmatter) is added to the sand, such that the pore spacesin the material are filled and the capillary thicknessof water at the grain interfaces is exceeded, thenthe cohesion of the material approaches zero and thematerial begins to behave as a fluid. This process willbe discussed in more detail later.

SHEAR STRENGTH OF SOILS:MOHR–COULOMB EQUATION

The way that soil particles behave as a group or mass(coherence) depends not only upon the inner cohesionof the soil particles but also upon the friction generatedbetween individual soil grains. The latter characteristicis internal friction or shearing resistance. How muchshear stress a soil or regolith can withstand is givenby the following equation, termed the Mohr–Coulombequation:

�s = c + �tan� (12.5)where �s = the shear strength of the soil

c = soil cohesion� = the normal stress (at right angles

to the slope)� = the angle of internal friction or

shearing resistance

The Mohr–Coulomb equation is represented diagram-matically in Figure 12.3. Note that this equation is ina form similar to Equation 12.4 except for twodifferences. Firstly, in the Mohr–Coulomb equationthe critical force for movement is determined by thestress normal to the ground surface, rather than bythe weight piled on a slope.

Secondly, the angle of the slope has been replacedby the angle of shearing resistance, which representsthe angle of contact between particles making up

251Land Instability as a Hazard

Cementation withcompaction

Cementation bydeposition

Particle diameter (cm)

Ten

sile

str

engt

h (k

g cm

-2)

Capillary cohesion

Vander W

aals'-type bonding

104

10-6

103

102

101

100

10-1

10-2

10-3

10-7 10-5 10-4 10-3 10-2 10-1 100

Fig. 12.2 The strength of bonding on particles of different sizes(Finlayson & Statham, ©1980 with permissionButterworths, Sydney).

unconsolidated material. Loosely compacted materialtends to have a lower angle of shearing resistance thanvery compressed or compacted material. All unconsoli-dated material tends to fail at internal angles less thanthe slope angle upon which it is resting. The Mohr–Coulomb equation can also be used to define the shearstrength of a unit of rock resting on a failure plane andthe susceptibility of that material to landslides. In thiscase, the formula must be modified to include thecharacteristics of bedding planes. If the stress appliedto the soil exceeds the shear strength, then the materialwill fail and begin to move downslope.

Pore-water pressure

Almost all material, no matter what its state of consoli-dation, contains pores or voids, which may be filled withair or, more often, water. If the water does not adheresolely to individual particles as capillary water, butcompletely fills voids forming a watertable, then it willflow freely under a pressure head. If the soil is incom-pressible, a rise in the watertable will cause the pressureof water in voids at depth to increase. Furthermore,some of the added pressure may be taken up by thegrain-to-grain contacts in the soil. If an external load isapplied to a soil mass on a slope in the form of additionalwater, or overburden, then the pore-water pressure willbuild up in that mass and water will be expelled at weakpoints. These weak points occur along fracture lines,bedding planes, or at the base of the regolith, wherepore-water pressure is highest. As water drains from asoil body, the pore-water pressure will decrease.

The increase in pore-water pressure in theMohr–Coulomb equation reduces the effective resis-tance of the soil body or regolith as shown by thefollowing equation:

�s = c + (� – ) tan � (12.6)where = pore-water pressure

In this case, normal stress � is reduced by the pore-water pressure Í. If overburden is added to a soil, itwill immediately increase pore-water pressure. Unlessthe excess pore-water pressure is reduced throughdrainage of water, the critical stress on that soil willexceed the critical resistance of the soil, causing slopefailure. If, on the other hand, wet material is addedupslope, there is no change in the stress being appliedto the soil. However, the water in this material candrain into the soil downslope, increasing progressivelyover time its pore-water pressure. Failure of the slopemay happen sometime after the overburden wasemplaced upslope. A slope that has been stable underexisting load and watertable conditions can alsobecome unstable if drainage patterns are changed inthe surrounding area. Dams raise watertables locally –a fact that may cause subsequent slope failure inadjacent areas some time after the dam is filled.Construction of buildings on slopes with septic systemscan also increase the pore-water pressure, leading tosubsequent failure. The source of water need not comesolely from septic tanks. It could also come from lawnwatering and other domestic discharges.

Earthquakes can increase the pore-water pressureof a soil with each passage of a compression shockwave. As the pore-water pressure increases, it maynot be reduced fast enough by discharge of waterfrom the ground before the next compression wavesweeps past. The effective stress in the material thusdecreases with each shock wave until the pore-waterpressure is equal to the normal stress in the soil(Figure 10.9). At this point liquefaction will result. Inunconsolidated sediment, the smaller the particle size,the more water movement is inhibited because ofcapillary cohesion. Very fine silt and clays thus shouldbe susceptible to liquefaction during earthquakes.However, depending upon the degree of consoli-dation, weathering, or other factors, cohesion in thesematerials may exceed the increases in pore-waterpressure during the passage of earthquake shockwaves. Thus, liquefaction tends to occur best inmedium- to fine-grained sands that have not com-pletely compacted. Because these types of sedimentsare widespread, especially in marine environments oron river floodplains, liquefaction is an almost universalfeature of moderate-to-large earthquakes.


Shear strength of soil

Slopeangle

Possible angle of internal friction

τ s

��

ασσ

τs = c + σ tan�

Normal stress

Fig. 12.3 Schematic representation of the Mohr–Coulomb equation.Note that the angle � is not necessarily the slope angle,but the angle of internal friction within the slope mass.

There is an exception to the above occurrence ofliquefaction. Some sodium cation-rich clays – termedquick clays – will liquefy if salt is leached by freshwater. This scenario often occurs with glaciallydeposited marine clays that are subsequently elevatedabove sea level and flushed with fresh water. The saltor sodium chloride deposited with the clays acts aselectrolytic glue, adhering to the clay particles and pro-viding cohesiveness and structure to the clay matrix. Ifthe sodium is leached out or replaced by calcium – aprocedure that can be performed even by the mobi-lization of free calcium in cement foundations into thesurrounding clay subsoil – the clay dramatically losesits cohesiveness. These clays then become very respon-sive to shock waves and liquefy during moderatelyintensive earthquakes. Such a process occurred in theBootlegger Cove clay underlying Anchorage duringthe Alaskan earthquake of 1964 (see Figure 12.4for the location of major placenames mentioned in thischapter). The process can also be instigated by theflushing of evaporite lake clays or highly weatheredclays such as montmorillonite, and is certainlyexacerbated by irrigation, pipe leakage, or simple lawnwatering. The latter circumstances exist today in manyparts of southern California including the San JoaquinValley, southern coast ranges, Mojave Desert, and LosAngeles Basin – all regions that are seismically active.

Rigid and elastic solids

Depending upon how stress and strain are related, soilbodies may behave in four ways: as rigid solids, as

elastic solids, as plastics, or as fluids. The process ofliquefaction determines the point at which a soil bodybehaves as a fluid. The concept of a rigid solid is alsoeasily explained. No matter how much stress is appliedto a solid, that solid is considered rigid until the stressexceeds the strength of the material. At this point, thesolid either deforms or fractures. The rate at whichthe stress is applied determines how the solid willbehave. For instance, toffee candy, when given asharp jolt, will fracture. In this case, the toffee can beconsidered a rigid solid. If, however, gentle pressure isapplied to the toffee, it may simply deform. In thiscase, the toffee is not rigid under this type of stress. Asolid may be considered elastic if a stress is applied thatresults in slow, continuous deformation proportional tothe applied stress before fracturing. The deformationat this stage is reversible, hence the description of thebody as an elastic. Earth materials behave elastically aslong as the stress changes applied are small enough.

Plastic solids

The relationship between the moisture content of asoil and its volume permits three factors to be deter-mined: shrinkage, plasticity, and liquid limits. Thesethree terms are labelled Atterberg limits. They refermainly to properties of clays and form a continuum,illustrated schematically in Figure 12.5. The shrinkagelimit is defined as the moisture content of a soil atwhich point the soil stays at a constant volume upondrying. Above the shrinkage limit, soil material willbehave as a rigid or elastic solid because it consists of


Australia

San Francisco

Caracas

SolomonIslands

Australia

Anchorage

Los AngelesMojave Desert

San Joaquin Valley

St. Lawrence Valley

Norway

West Virginia

ShensiKansu

Rio de Janeiro

Hong Kong

KobeTokyo

Kure

New BritainPanama

Fiji

ColoradoWyoming

Mississippi ValleyGreat Plains

Mt. Huascazrán

Chiavenna Valley

KhaitUsoyVaiont Valley

SwitzerlandWellingtonRogers Pass

Burning Mt.Wollongong

St. Jean- Vianney

AberfanNorth Sea

Alps

Russian Steppes


interlocked assemblages of grains and particles thatpossess a finite strength, termed the yield limit.However, if a force is applied to that material such thatthe rate of deformation is proportional to the amountof applied stress above this yield limit, then thematerial is termed a ‘plastic solid’. The differencebetween an elastic solid and a plastic solid dependsupon how the material reacts after the stress isreleased. In an elastic solid, yield strength has not beenexceeded, and the solid will tend to return to itsoriginal configuration after deformation. In a plasticsolid, the deformation is irreversible because the yieldlimit has been exceeded. The yield or plastic limit isdefined by the minimum moisture content at whichpoint that material, usually clay, can be molded. It isalso the point where the angle of shearing resistance inclays approaches zero as moisture content increases.This limit defines plasticity and, for clays, this state canbe maintained over a wide range of moisture contents.As long as clay has a moisture content below the plasticlimit, it will support objects. However, when the plasticlimit is reached, then the bearing strength of the clay isgreatly reduced and it will begin to deform. The liquidlimit defines the moisture content at which the clayflows under its own weight. When the liquid limit isreached, clay behaves like a fluid and easily flowsdownslope. This point also defines when liquefactionoccurs, because the material has no shear strength.

Plastic and shrinkage limits are very much a functionof the type of clay material. There are three main typesof clays, dependent upon the degree of weathering offeldspars or other easily weatherable minerals.Feldspars progressively weather to montmorillonite,illite, and finally kaolinite. Montmorillonite [(Mg,Ca)O.Al2O35SiO2.nH2O] is composed of one alumina

layer sandwiched between two silica layers. The layersare bound together by Van der Waals’ bonds. The clayminerals have a large residual negative charge becauseof a charge imbalance in their structure. This leads tothe extensive addition of water molecules to thestructure of the clay. As a result, the cation exchange,plasticity, and swelling capacity of such clays are high,making them especially susceptible to deformation andswelling. As the clays hydrate through progressivechemical weathering, they begin to lose thesecharacteristics. Illite [KAl2(OH)2(AlSi3(O,OH)10)],a daughter product of montmorillonite, contains ahigher proportion of silicon atoms, inducing a high, netnegative charge between layers. Potassium ions areattracted to these sites in the crystal lattice and bindthe silica layers together, thus reducing their swellingcapacity and plasticity. Kaolinite [Al2O3.2SiO2.2H2O]is a highly weathered clay mineral in which all potas-sium ions have been stripped from the lattice. Theseclay minerals consist of a layer of silica and aluminabonded together by Van der Waals’ bonds or hydrogenions. Both bonds are strong. There is little substitutionof other atoms or molecules possible in the mineral;hence, it has low plasticity and low swelling capacity.

CLASSIFICATION OF LANDINSTABILITY

Introduction(Sharpe, 1968; Varnes, 1978; Finlayson & Statham, 1980;Crozier, 1986)

Land instability can best be described by setting thevarious types of failure into a general classification. Atpresent, land subsidence can be considered as a specialcase of land instability because it is more related tothe behavior of the substratum than it is to the stressapplied to material on a slope. Most classifications ofland instability are based upon the type of materialand the type of movement. There are five types ofmovement: falls, topples, slides, lateral spreads andflows. Material can consist of bedrock, consolidated soiland regolith, loose debris, various mixtures of sedimentand water, and pure water in the form of snow or ice.Unfortunately, many of the classifications vary in theiremphases. Some are based on geotechnical aspects,reflecting an engineering orientation; while otherscenter on processes and morphology, reflecting ageomorphological perspective. Each classification has


Solid Elastic Plastic Liquid

Sam

ple

volu

me

SLshrinkage

limit

PLplasticlimit

LLliquidlimit

Moisturecontent (%)

Fig. 12.5 Schematic representation of Atterberg limits (after Goudie,1981).

tended to bring in different terminology that at times isconfusing and contradictory.

A useful classification in terms of material andmovement is shown in Table 12.1. This classificationhas since been modified to include topples and lateralspreads. The scheme emphasizes the composition ofthe material being moved, but it and its modificationclearly do not include time. From the hazard pointof view – in terms of warning, human response, andprevention – these classifications can be limited. A

temporal classification that gives some concept of thespeed of movement of instability can be more attrac-tive. In Table 1.4, which ranks hazard characteristicsand impacts mentioned in this text, some emphasis isplaced on the suddenness of the event. One suchtemporal classification is shown in Figure 12.6. It isimmediately obvious that this classification includesexpansive soils, which are ignored in most engineer-ing and morphometric classifications. Figure 12.6is used in this chapter, not because it is better than


10-610-7 10-5 10-4 10-3 10-2 10-1 100 101 102 103 10410-9 10-8

1 mm yr-1 1 cm yr-1 1 mm day-1 1 cm day-1 1 cm hr-1 1 km yr-1 1 cm min-1 1 km hr-1 1 m s-1 100 m s-1

creepsolifluction

expansive soils

mud & debris flows

landslides & slumps

rockfalls

debris avalanches

air supported flows

Velocity (cm s-1)

Fig. 12.6 Classification of land instability based upon rate of movement (derived from Finlayson & Statham, 1980).

Table 12.1 United States Highway Research Board Landslide Committee classification of mass movements (from Leopold et al., 1964).

Type of movement

Type of material

Bedrock Soil

Falls Rockfall Soil fall

Slides

No. of UnitsFew

Many

Rotational RotationalPlanar

Slump Block glide Block slump

Rockslide Debris slide Lateral spread

Flows

Moisture content

All unconsolidated

Rockfragments

Sandor silt

Mixed Plastic

Dry

Wet

Rockfragmentflow

Sandrun

Loessflow

Rapidearthflow

Debris flow:sand or silt

Debrisavalanche

Rubble flow:dirty snowavalanche

Slowearthflow

Solifluction

Mud flow

Creep

the other classifications, but because it emphasizestime and includes expansive soils – the latter in thelong term causing the most expensive type of landinstability.

Expansive soils(Hays, 1981)

Expansive soils annually cause more than $US3000million damage (1989 dollar value) to roads andbuildings in the United States. This cost exceeds thecombined, annual damage bill from all climatic hazardsin this country. Because the process works so slowly,damage is not always obvious. The actual cost to societymay in fact exceed $US10 000 million annually.Fifty per cent of the damage occurs to highways andstreets, while 14 per cent occurs to family dwellings orcommercial buildings. Of the 250 000 homes built onexpansive clays in the United States each year, about10 per cent will undergo significant damage and60 per cent will undergo minor damage in their lifespan. Figure 12.7 maps the extent of expansive soils inthe conterminous United States. Over a third of thecountry is affected. The extent of the problem is notunique to the United States: it occurs on a similar scalein Australia and on the Russian steppes, where soiland geological conditions are similar. In Australia,38 per cent of homes in the southern half of thecontinent show evidence of cracking, with again

10 per cent developing serious damage over their lifespan. During the drought of 2002–2003, the damagebill reached $US600 million.

Expansive soils are produced mainly by claysderived from two major groups of rocks. The firstgroup consists of aluminum silicate minerals involcanic material that decomposes to form montmoril-lonite. The second group consists of shales containingthis clay mineral. In Australia, both these rock typesare very prevalent because large areas of the easternhalf of the continent were subject geologically to vol-canism. Additionally, slow evolution of the Australianlandscape has permitted the widespread accumulationof large quantities of montmorillonite clay as crackingclays on inland river systems.

Expansion usually takes place when water penetratesthe lattice structure of clay minerals. Two conditionsmust be met before swelling can take place. First, itmust be possible for a volume change to occur and,second, water must be present. An increase in soilmoisture content of only 1–2 per cent is sufficient tocause expansion. For the volume change to be made,sufficient clay must be present. A thick layer ofexpansive clay has greater potential volume change thana thin layer. If the load on the clay is high, then com-pressive forces may exceed the expansive force exertedby clay upon wetting. Because this compressive loaddecreases towards the ground surface, the presence of


Great Plains

Great LakesAppalachian

Coastal Plain

Mountain &Basin

0 1000 km

Fig. 12.7 The extent of expansive soils in the United States (adapted from Hays, 1981).

expansive clays in the upper soil layers increases theeffect of swelling and shrinking. Large buildings posi-tioned on expansive clays will not usually be affectedbecause the weight of the structure can prevent expan-sion. Extensive surficial cracking due to desiccation isalso indicative of expanding clays. In Australia, thepresence of gilgai (undulations in the soil surface withwavelengths of 1–2 m) is indicative of the problem.

People can enhance the effect of expanding clay bydisrupting ground drainage and modifying vegetation.Drainage from a house gutter or septic tank into anarea of expansive clays can cause that material toexpand, while the drier material under the house doesnot. This differential expansion can affect foundationsclosest to the wetted soil. The removal of vegetationcan also result in increased soil moisture becauseevapotranspiration from trees or shrubs is negated. Inbuilt-up areas where rainfall variation is high, expand-ing clays can undergo repetitive cycles of swelling and

shrinking, hence putting unequal stress on foundationsover time. The process is quite common in Australiawhere drought and flooding rains alternate betweenEl Niño and La Niña events. This is illustratedschematically for my house in Figure 12.8. Duringdroughts, the soil underneath my house will dry outmore slowly than that in the open. The exposed soilshrinks and the outer foundations of the house tendto settle before the inner ones. After rainfall, theexposed soil is wetted and expands. A wetting frontthen proceeds under the house causing expansion overtime towards the centre of the house. Trees around ahouse can accelerate the drying effect during droughtbecause of their higher evapotranspiration, while lawnwatering can accelerate the wetting effect. Crackingof interior walls, especially during a change fromdrought to wet or vice versa, is very common. Floorscan become creaky as pilings sink differentially due tovariations in clay or moisture content.


NormalConditions

Drought

highevaporation

cracksappear

foundationssettle

soi l st i l l wet soil shrinkssoil dries

Wet

water infiltratesgi lgai surfacesoil swells

more cracks!

soi l st i l l dry

wett ingfront

Fig. 12.8 Schematic representation of the effect on a house of alternate wetting and drying of expansive soils.

The most efficient way to reduce damage caused byexpansive soils is to avoid them. In Australia, however,this is almost impossible. Other methods to minimizedamage include removing the expanding soil (especiallyat the surface); applying heavy loads to the surface;preventing water access to the site; pre-wetting the soiland preventing its drying out forever afterwards(although damp soil attracts termites); or ensuring thatfoundations are sunk deep enough to minimize theeffects of near-surface swelling and shrinking. It is alsopossible to change the ionic character of the soil byadding hydrated lime, Ca(OH)2, to the surface of anexpanding clay. The lime lowers the exchange capacityof the clay. Water, thus, cannot be substituted into theinternal lattice structure of this clay mineral as effi-ciently, and expansion is reduced. Note that the additionof lime to a soil subject to cracking is not recommended,if that soil is also quick clay. Liming dramaticallydecreases this latter soil’s cohesiveness.

Creep and solifluction(Leopold et al., 1964; Sharpe, 1968; Young, 1972;Finlayson & Statham, 1980)

Under sustained shearing stresses, all soil and rockmaterials on slopes exhibit viscous behavior, which istermed creep. Where the melting of ground ice exacer-bates the movement, the process is termed solifluction.Rates of creep are not substantial and rarely exceed1–2 mm yr-1, although velocities over 100 mm yr-1 havebeen measured on slopes as steep as 40°. Because creepis more active towards the surface of a soil profile, therate of movement and the resulting displacementdecrease with depth. This is partly due to compactionand increased loading at depth. If material has somelong-term instability, shearing planes or surfaces willdevelop in the region of maximum shear stress. Thispoint usually occurs at the base of the soil profilebetween the B and C horizons, or along fractures(bedding planes) in the underlying weathered bedrock.A stone line, above which banding in the soil occurs,and below which imperceptible movement takes place,often shows this. Material in this zone will shear until itsstrength is reached. If further shearing eventuatesupward in the profile, the displacement becomescumulative and, hence, greatest towards the surface.

Creeping of a regolith downslope can also be accom-plished by alternating expansion and shrinking of a soil.The most obvious mechanism for this process is thepresence of expanding clays under the influence of

seasonal inequalities in rainfall. The process can also beaccomplished by freezing and thawing of water in thesoil, and by salt crystallization. Each time expansionhappens, the soil tends to be pushed upward at rightangles to the slope (Figure 12.9). This process weakensthe coherence of the soil mass. When shrinkage orthawing eventuates, the soil settles back to its originalposition; however, gravity will tend to move the materialslightly downslope. Creep rates should be proportionalto the tangent of the angle of the slope, and all materialshould move downslope. In practice, it has been foundthat the rates of movement on a constant slope canvary greatly over small distances, to the point of beingrandom. There are also measurements indicating thatcreep movement can occur upslope. This result impliesthat expansion and contraction do not operate normalto the slope. In this regard, a slope must be consideredthree-dimensional, with movement possible in anydirection including upslope and laterally. This aspecthas already been recognized in periglacial environ-ments, where lateral sorting of sediment dominatesthe micro-morphological evolution of the landscape. Inenvironments where a distinct thawing season existsafter the soil has been frozen, or where rainfall isseasonal, creep rates will vary seasonally. The process isfastest in spring, when temperatures are still coolenough for the ground to re-freeze at night, and duringthe wet season, when distinct seasonal inequalities inrainfall exist.

Creep in clay does not always depend upon expansionand contraction. If clay is wet enough to become plastic,it will deform under load in the downslope direction. Asthe rate of creep depends upon the weight of over-burden, deformation may occur only at depth in theprofile and not at the surface. Creep, in clays at depth


Settlement under the effects of gravity

Distance AB = Expansion distance = x tanα

α

α

BA

Expansion normal to slope

Soil particle

x

Fig. 12.9 Mechanism for the downhill movement caused by soilcreep of material on a slope.

in a particular zone, can cause the clay minerals torealign themselves, reducing shear strength in this layer,which subsequently becomes the location for develop-ment of a shear plane. The cohesiveness of claydecreases with increasing moisture content; so, contin-ual creep in clays has its highest rate where the clay ismoist for the longest duration. Creep processes are notrestricted solely to clays or to locations with expansivesoils. Because creep operates under the effect of gravity,any disturbance to a soil on a slope will result in downhillmovement of material. Animal burrowing, penetrationby plant roots, tree collapse, and simple trampling byanimals or people can be the cause of this disturbance.While the short-term disturbance seems minor, aggre-gated over time these effects can have significantinfluence on near-surface creep rates.

The process of solifluction involves additionalfactors. Solifluction is restricted to environments whereground freezing takes place. The expansive forceinitiating creep is due to moisture freezing in the soil,because frozen water occupies a greater volume thanthe equivalent weight in liquid form. Solifluction canproduce accelerated rates of downslope movementof material on slopes as gentle as 1°. Freezing in soilcan also concentrate water at a freezing front. Largevolumes of water can be drawn to this front by capillaryaction, forming ice lenses that can constitute up to80 per cent by volume of the soil mass. The melting ofall or some of this ice can increase pore-water pressuresand form shear planes for slope failure. In areas under-lain by permafrost, where seasonal melting prevails inan active surface zone, solifluction can totally dominatea landscape such that material on all slopes appears tobe in a continual state of rapid movement downslope.The discussion of solifluction and its geomorphicramifications are beyond the scope of this text; but itshould be realized that ground ice and solifluctionprocesses represent a serious hazard to constructionand transport in Arctic regions of Canada, Alaska, andRussia. Most mountain regions are also cool enough,because of their high elevation, to be affected bysolifluction. About 20 per cent of the world’s landmassis susceptible to these deleterious agents.

Mud and debris flows (Sharpe, 1968; Bolt et al., 1975; Cornell, 1976; Whittow,1980; Innes, 1983; Wieczorek et al., 2001)

A flow is any slope failure where water constitutesa major component of the material and a major

controlling factor in its behavior. Flows can be sub-divided into two categories, based upon the type ofmaterial. If clays dominate, the flow is termed a‘mudflow’ or ‘earth flow’ in the United States anda ‘mudslide’ in the United Kingdom. Note that thisterm incorporates lahars mentioned in the previouschapter. The term ‘lahar’ is more restrictive in that it isa mudflow originating in volcanic ash. If the range inparticle size is highly variable, then the flow is termeda debris flow. Generally, the material in a flow is water-saturated and unconsolidated. Not only can waterpermeate through the material, but it can also beabsorbed by the sediment. For this reason, most flowsrepresent the further downslope movement of materialthat has already undergone some sort of failure. Forinstance, a rockfall that disintegrates into smallerpieces while falling will produce an unconsolidateddebris mantle when it comes to rest. Over time, thisdebris can chemically weather, breaking down intosmaller particles. When wetted, this material mayabsorb water, increase in weight, and begin to fail againat lower slope angles.

Flows occur where high water content increasespore-water pressure in the deposit, a process thatdecreases shearing strength. Flows usually beginmoving along a basal shear plane. Rates of movementcan be as low as 1–2 m yr-1, or as high as 600 m or moreper year. Fastest movement occurs during the wettestmonths. Mudflows do not necessarily move at con-sistent rates; they commonly surge. In southernCalifornia, surge rates of 3–4 m s-1 have been observedin large mudflows. Typically, even with lahars,velocities average less than 20 km hr-1. Slow-movingearth flows have caused property damage, but minimalloss of life, in Czechoslovakia, England, Switzerland,and Colorado. In periglacial environments dominatedby solifluction, mudflows are a common occurrence.

Coarse debris flows are much more difficult tomove. Fine material must be present to aid waterretention. The debris flow may consist of very lowvolumes of sediment moving downslope at rates ofseveral kilometers per hour. Coarser debris tends toget extruded to the surface and to the sides duringmovement, such that coarse-sized levee banks are builtto the sides and front of the moving deposit. Typically,debris flows require slopes of 30–40° for movementto start, but flow continues over slopes of 12–20°(Figure 12.10). Note that this type of flow does notinclude large, catastrophic events moving coarse debris


at higher velocities. Such an event is termed a debrisavalanche and will be discussed in detail below. Whenmudflows and debris flows stop, excess water tends todrain from the deposit, forming either lobate or planarfans. In some cases, where subsequent filtering out ofsediment has happened, flow deposits can be mistakenfor alluvial fan deposits. The thickness of the deposit isinversely proportional to the fluidity of the deposit, andto the slope angle at rest.

Under extreme wetness, both mudflows and debrisflows can liquefy, whereupon the weight of material isborne by pore-water pressure and not grain-to-graincontacts. This problem is particularly severe inmarine clays laid down following the retreat of thelast major glaciation. Subsequent desalinization formsquick clays, which are very susceptible to lique-faction. The St Lawrence Valley in eastern Canadawas effectively flooded by the ocean during thewaning phases of continental glaciation, with the sub-sequent deposition of extensive, thick clay bedssusceptible to earth flow and liquefaction. Mudflowswith volumes in excess of 20 million m3 have beenquite common in this region. The St Jean-Vianney,Quebec, failure in 1971 took 35 lives. Similar types ofdeposits have been recorded in Scandinavia andRussia. In 1893, at Vaerdael, Norway, 112 peoplewere killed by one such failure.

Unstable mine tailings can also generate debrisflows. The Aberfan, Wales, disaster of 21 October 1966

was one of the more tragic disasters of this nature.A 250 m high, fine-grained, rain-soaked coal tailingspile failed and turned into a debris flow, which sweptthrough a school killing 116 children, the entirejuvenile population of the town. A similar flow in a coalmining waste dump in West Virginia killed 118 peoplein 1972. The Aberfan disaster illustrated the unsafenature of coal tailings in the United Kingdom – over25 per cent were subsequently found to be in a similarcondition of instability.

Nothing in recent history matches the Caracas,Venezuela, floods of 15–17 December 1999. Cen-turies ago, Caracas was deliberately built inland,behind the 2700 m high El Ávila coastal range, to hidethe city from invaders. It has grown to a city of5 million people plagued by the worst characteristicsof Third World urban centers: slums with highdensities of 25 000 km-2 built in hazard-prone areas,haphazard construction, lack of planning, and unen-forceable building codes. Hillsides with 80 per centslopes have been built upon. The disaster beganduring a month of rain, culminating in over 900 mmfalling in a 72-hour period between 15 and 17December. The rain triggered landslides, mudflowsand, worst of all, debris flows, killing an estimated30 000 people. While the rains were excessive, prob-ability analysis has shown that 300 mm or more of rainin 24 hours could be experienced once in 25 years.Similar floods had occurred in 1798, 1912, 1914,1938, 1944, 1948, and 1951. Geomorphic evidenceindicated debris flows were common. The rains of1999 reactivated nine large alluvial fans on the coastalside of the El Ávila range and created 28 slidescontaining as much as 58 500 m3 of material. Unfor-tunately, the fans had been urbanized (Figure 12.11).The rains also stripped steep slopes of their regolithand moved boulders with dimensions as large as11.3 � 5.0 � 3.5 m, and volumes of 1310 m3, down-stream. Debris flow velocities reached 14.5 m s-1.Under these conditions, buildings up to eight storeyshigh were destroyed. The extent of the disaster wasawesome. The volume of sediment moved in individ-ual debris flows reached 1.9 million m3, amongst thelargest ever recorded. However, these volumes are atleast an order of magnitude smaller than the largesttriggered by volcanic explosions, eruptions, orearthquakes. Over 8000 homes and 700 apartmentbuildings were destroyed or damaged. The damagebill totalled $US1.8 billion.


Fig. 12.10 Widespread debris flows initiated on slopes of 12–30°in the Kiwi Valley area, Wairoa, New Zealand, byintense rainfall in February 1977. Rainfall intensities of127 mm in 2 hours, and 252 mm in 5.5 hours, wererecorded (photograph courtesy of the Hawke’s BayCatchment Board, Napier, New Zealand).

Landslides and slumps

Mechanics

(Leopold et al., 1964; Chowdhury, 1978; Whittow, 1980)

Slope failure causing landslides can be modelled usingthe Mohr–Coulomb equation. Because many slidesoccur along a distinct shear plane, this equation can bemodified to include the effects of shear planes withinan unconsolidated deposit, bedding planes and jointsin weathered or unweathered bedrock, and thepresence of a watertable. Landslides, therefore, tendto happen in two different types of material, consistingof either bedrock or unconsolidated sediment, usuallyclay. In the simplest case, where the Mohr–Coulombequation is easily applied, landslides move downslopeparallel to the line of failure. However, landslides canundergo rotation as well. In this case, the failuredevelops slowly as a slump. Figure 12.12 illustrates thisprocess of rotational sliding or slumping. While themechanics of rotation are the same as a simple planarslide, the movement of the material is determined bythe arc of rotation centered at a point outside the slipmaterial. Rotational slides can be discussed in terms ofa head scarp, where the material has separated fromthe main slope; a slip surface, upon which the materialrotates; and a toe, which consists of the debris from thefailure. Rotational slides may also generate transverse

cracks across the body of the slipped mass and ten-sional cracks above the head scarp. These tensionalcracks can be lines of failure for future slides. The toeof the slide controls the terminal point of the failure.Interference with the toe of the slide can also initiatefurther failure.

Causes

(Sharpe, 1968; Finlayson & Statham, 1980; Crozier, 1986)

With landslides and slumps, not only is the stressapplied to the area of potential failure important butthe behavior of the toe also becomes crucial to thetiming of the slide. One of the major causes of land-slides is undercutting of the basal toe in some manner.The causes of landslides in any form are thus multi-tudinous. Geological or topographical factors conditionthe location of a potential slide. If weathered orunweathered bedrock contains inherent lithological orstructural weakness, then these areas will favor slidefailure. Lithologically favorable material includesleached, hydrated, decomposed, chloritic, or micaceousrocks, shales, poorly cemented sediments, or uncon-solidated material. Additionally, if this material is well-rounded or contains clay, then it will flow morefreely under pressure. Stratigraphically favorableconditions for landslides include massive beds overlyingweaker ones, alternating permeable and impermeablebeds, or clay layers. Structurally favorable conditionsinclude steeply or moderately dipping foliations,cleavage, joint, fault, or bedding planes. Rock thatis strongly fractured, jointed, or contains parallelalignment of grains because of crushing, folding,faulting, earthquake shock, columnar cooling or desic-cation is also likely to fail. Internal deformationstructures resulting from tectonic movement, solution


Fig. 12.11 Aerial view of erosion and debris flow deposition duringthe Caracas, Venezuela, urban flood of December 1999.The deposit is up to 6 m in thickness and contains1.9 million m3 of sediment. Gaps denote areas wherebuildings were removed by the floods (photo by Lawson Smith, United States Geological Survey,<http://pubs.usgs.gov/of/2001/ofr-01-0144/Venezuela/tnhires/pages/image030.htm>).

head scarp

central axis ofrotation

toe

slip surface

radius ofrotation

Fig. 12.12 The process of rotational sliding (Whittow, 1980, ©and reproduced with the permission of John Whittow,Department of Geography, University of Reading).

or underground excavation will aid slope failure,especially if such features do not reach the surface. Anywater-saturated, porous lens within a regolith will alsotend to form a zone of preferred shearing. Topographi-cally, any cliffed or steep slope is also susceptible to landsliding, as are areas of block faulting, previous landslidesor subsidence, and artificial excavation. Finally, any areathat has been denuded of its soil-retaining vegetationbecause of deforestation, overgrazing, cultivation, fires,or climatic change has the potential for sliding.

Initiating causes of landslides are just as plentiful.The removal of basal support provides the easiest wayto start a landslide. Natural agents such as undercut-ting by running water, waves, wind, and glacial ice areprime candidates. Extrusion of material from the baseof a slope is also effective. This can take the form ofoutflow of plastic material within the deposit, washoutof fines, and melting of ice. The material at the basecould also change its characteristics through waterabsorption due to flooding, solution of soluble materialsuch as limestone or salt, or chemical alteration. Indeveloped areas, people have now become the majorcause of toe instability on slopes through mining, exca-vation, quarrying, basal construction, and road, rail andcanal works. Overloading of the slope material is also acommon means of initiating failure. Overloading canoccur by saturating the slope with rainwater or waterfrom upslope streams or springs, or by loading theslope with snow or debris from upslope instability.Humans can also initiate overloading by dumping spoilor by building foundations and structures on the slope.

Slope failure can also be induced by reducing internalcoherence in the slope material. Increased lubrication

due to heavy rainfall or runoff filtering through aregolith is a common mechanism for triggering failure,either immediately or some months afterwards.Cracking of the slope material through partial failure,desiccation, earthquakes, or internal movement can alsoallow water to penetrate more easily into the slopematerial. Blockage of drainage, usually by raising thebase of the watertable, can reduce shear resistanceinternally. As already mentioned, the activities ofhumans through drainage modification, deforestation,overgrazing, or water discharge from septic systems anddomestic usage also reduce internal coherence throughlubrication. Earthquakes and volcanoes can eithercrack slope material, or reduce coherence through theprocess of liquefaction. Even thunderstorms can triggerrockfalls and landslides because of the vibrations thatare transmitted through slopes. Similarly, humans nowperform activities that vibrate the ground and couldtrigger landslides. These activities include vehicle move-ments, blasting, pile-driving, drilling, and seismic work.

A number of processes can cause slope failuredue to prying or wedging. Water freezing in cracks,increased pore-water pressure after heavy rain,expansion of soil material because of hydration, oxida-tion, carbonation or the presence of swelling clays, treeroot growth, and temperature changes are the mainmechanisms. Finally, strains in the earth due to suddenchanges in temperature, atmospheric pressure, or thepassage of earth tides are also factors that could triggerlandslides. The above conditions favoring landslidesmay act in concert to trigger instability. As a result,slope failure can happen instantaneously, irregularlybut cumulatively over time, or with considerable


Short-term oscillations in soil strength, for example, changes in pore-water pressure.

A point where shearstrength equals shear stress.Therefore failure occurs.

Long-term change in soil strength, for example, weathering.

Critical shear stress.If soil strength falls belowthis level, failure occurs.

Progressive increase inshear stress due to loading, for example,increasing rainfall.

Short-term increasein shear stress, for example, due to an earthquake

Soi

l str

engt

h, s

hear

str

ess

Time

Fig. 12.13 The complexity of landslide initiation over time (Finlayson & Statham, 1980 © with permission Butterworths, Sydney).

temporal lag after the occurrence of a triggering event.Figure 12.13 illustrates this complexity in landslideinitiation. Ultimately, failure will eventuate when theshear strength of the soil is exceeded by a critical shearstress. The critical shear stress can fluctuate slowly ordramatically over time, while the shear strength of thematerial can oscillate because of seasonal factors, ordecrease slowly because of any of the above conditions.Slope failure may happen at a number of points intime. If stress dramatically increases, for instancebecause of heavy rainfall, and exceeds soil strength,then the cause of the landslide is obvious. However,soil strength could be decreasing, or the stress on aslope increasing slowly and imperceptibly over time, tothe point that the cause of the landslide is not defin-able. Thus, with any landslide, there is randomness asto the timing and cause of the failure.

Landslide disasters(Bolt et al., 1975; Cornell, 1976; Whittow, 1980; Coates,1985; Blong & Johnson, 1986)

Landslide disasters are difficult to separate from othertypes of land instability, and from some of the largerdisasters that trigger the event. For instance, tropicalcyclones, apart from drowning people, usually bringvery heavy rainfalls; this not only increases pore-waterpressure in potentially unstable material but, throughadded weight, also increases stress in slope deposits.Some of the worst natural disasters have occurredbecause of landslides triggered by earthquakes. TheChinese earthquakes in Shensi province in 1556, andat Kansu in 1920, caused failure in loess deposits,collapsing homes dug into the cliffs. Urban areas seemto be most affected by landslides. In Rio de Janeiro in1966, record-breaking rainfall in January and Marchcaused catastrophic landslides that struck shantytownsaround the mountains in the city. Over 500 peoplewere killed in the slides and another 4 000 000were affected by disrupted transportation andcommunications. In the following two-year period,over 2700 people were killed in the Rio de Janeiro areaby landslides and other slope instability events thatafflicted over 170 km2. In Hong Kong, a tropicalcyclone in 1976 dropped 500 mm of rain in two days,triggering landslides on steep slopes and killing 22people. Similar slides there in 1966 killed 64 people.Many of the slides occurred where dense urban sprawlencroached upon steep slopes and undermined thetoe of unstable sediments. Japan is also particularly

vulnerable to typhoon-generated slides near urbanareas. Kobe was hit in 1939 by rain-induced landslidesthat killed 461 people and damaged 100 000 homes.In 1945, comparable landslides killed 1154 people inKure. In 1958, Tokyo was struck by a typhoon thatgenerated over 1000 landslides and killed 61 people.

In underdeveloped, heavily vegetated regions,especially in the tropics, mega-landslides play a majorrole in the downslope movement of material. In 1935,New Guinean landslides cleared 130 km2 of vegetatedslopes. The Bialla earthquake of 10 May 1985, on theisland of New Britain in Papua New Guinea, triggereda mega-landslide in the Nakanai Range that dramati-cally infilled the Bairaman River (Figure 12.14). About12 per cent of slopes in New Guinea are subject tolandslide denudation every century. Elsewhere in thetropics, an area of 54 km2 was cleared on slopes inPanama in 1976. Tropical Cyclone Wally, which struckFiji in April 1980, and Cyclone Namu, which passedover the Solomon Islands in 1986, both generatedhundreds of landslides with an average volume ofmaterial moved in excess of 500 m3 per hectare. Suchlandslides had not been witnessed in these areas in theprevious 50 years. The cyclones effectively destabilizedthe landscape enough to ensure continued massmovements triggered by less severe rainfalls for yearsto come.


Fig. 12.14 Mega-landslide on the northern slope of the NakanaiRange, New Britain, Papua New Guinea. This slide wastriggered by the magnitude 7.0 Bialla earthquake on10 May 1985. The slide occurred in Miocene limestoneand extensively infilled the Bairaman River Valley(photograph by Dr Peter Lowenstein, courtesy of C.O. McKee, Principal Government Volcanologist, Rabaul Volcanology Observatory, Department of Minerals andEnergy, Papua New Guinea Geological Survey).

The ubiquitous nature of landslides is illustratedwith reference to the conterminous United States inFigure 12.15. By far the most prevalent zone forlandslides occurs in the Appalachian Mountains in theeastern part of the continent. Second in importanceare the Rocky Mountains, particularly in the states ofColorado and Wyoming. The coastal ranges along thePacific Ocean form the third most hazardous zone. InJanuary 1982, heavy rains triggered more than 18 000landslides in the San Francisco Bay region, causing25 deaths and more than $US100 million in damage.It was the third biggest disaster in this region afterthe 1906 San Francisco and 1989 Loma Prieta earth-quakes. The threat is even greater in the Los Angelesregion. Landslides in the Los Angeles area killed 200people on 2 March 1938. In all, landslides in someform affect about one-seventh of the United States.While most of the failures happen in mountainousareas, they are not restricted to these zones. TheMississippi Valley has a low, but significant, risk, as doplateaus on the southern Great Plains where weakshales are found. The United States is not unique inthis respect. Landslides are a common geomorphicprocess in most countries with steep or raised terrain.

Rockfalls(Scheidegger, 1975; Voight, 1978)

Rockfalls are one of most rapid land-instability events.They are generally restricted to unvegetated, vertical,

rock cliff-faces but can occur in certain unconsolidatedsediments that permit vertical slopes to develop.Generally, the vertical faces must be continually main-tained by erosion of talus and screes that form at thebase of cliffs. If this debris is permitted to accumulate,then the vertical face becomes buried. The debris isremoved through chemical breakdown, washing out offines and slippage. There is, thus, a balance betweenthe rate of cliff retreat controlled by the occurrenceof rockfalls, and the rate of removal of debris at thebase. The production of talus does not have to occurcatastrophically, but instead debris can simply accumu-late through the continual addition of material as it isloosened from the cliff-face by weathering, rainfall,small shock waves, or frost action.

The preservation of a vertical face is dependentupon the characteristics of the bedrock. Generally,less resistant rock such as shales will not be able todevelop cliff-faces, because the rock is susceptibleto weathering and slippage. Situations where there aremassive beds of sandstone or limestone, or where theseresistant beds overlie weaker beds, are ideal for theformation of cliffs. However, massive beds withoutmuch jointing do not erode easily, and hence do notproduce rockfalls. For example, the escarpments alongthe east Australian coastline have faces up to 100 m ormore high in places, and appear to be retreating atrates as low as 1 mm yr-1 without significant rockfalls.Significant joints or vertical failure planes should also


0 1000 km

Severest landslide hazard

Localized landslide of varying severity

Moderate landslide hazard

Fig. 12.15 Extent of landslides in the coterminous United States (after Hayes, 1981).

be present for the occurrence of large rockfalls. Evenjointing is not a guarantee that rockfalls will eventuate.Blocks separating from cliffs along vertical lines ofweakness must also undergo disintegration for rockfallsto take place. Otherwise, they simply glide downslopeintact.

Globally, rockfall is the major process for theremoval of debris from cliffs in scarped or steepmountainous regions affected by periglacial activity.While many of the factors that trigger landslidesinitiate rockfalls, the majority occur because of earth-quakes. Many of the types of landslides discussedabove, especially those occurring in the EuropeanAlps, began as rockfalls. In deeply snow-coveredmountains, rockfalls are a major triggering mechanismfor avalanches.

Debris avalanches(Varnes, 1978; Voight, 1978)

Debris flows and mudflows can take on catastrophicproportions several orders of magnitude greaterin size and speed than described above. The term‘debris avalanche’ is more appropriate to these typesof flows. Many of these events begin as landslides orrockfalls. All appear eventually to entrain a variety ofparticle sizes and behave as a fluid flow. The highvelocities appear to be achieved by lubricationprovided by a cushion of air entrapped beneath thedebris. The flow literally travels like a hovercraft. Insome instances, there is little evidence of waterinvolved in the avalanche and, in other cases, massesof debris appear to have traveled with minimal defor-mation or re-alteration. By far the most destructivedebris avalanches in recent times occurred aroundMt Huascarán, Peru, in 1962, and again in 1970. The1962 event started as an estimated 2 million m3 of iceavalanched from mountain slopes. This ice mixed withmud and water, and turned into a much largermudflow with a volume of 10 million m3. It sweptdown the Rio Shacsha Valley killing 4000 people,mainly in the town of Ranrahirca. Material, some ofwhich consisted of boulders over 2000 tonnes inweight, traveled down the valley at 100 km hr-1 and100 m up valley slopes. Having survived that event –and believing that another, similar, catastrophe wasunlikely – residents of the area then experienced anearthquake on 31 May 1970, which caused the icecapof Mt Huascarán, together with thousands of tonnesof rock, to cascade down the valley. This debris flow

entrained boulders the size of houses. As it traveleddown the valley at speeds of 320 km hr-1, it picked upmore debris each time it crossed a glacial moraine. Atthe bottom of the valley, the flow had incorporatedenough fine sediment and water to become a mudflowwith a 1 km wide front. Eyewitness accounts describethe flow as an enormous, 80 m high wave with a curllike a huge breaker as it approached the town ofYungay. Ridges over 140 m in height were overridden,and boulders weighing several tonnes were tossed upto 1 km beyond the rims of the flow. Within fourminutes, 50–100 million m3 of debris completelyobliterated Ranrahirca and the much larger town ofYungay. The flow then continued down the Rio Santareaching the Pacific Ocean 160 km downstream. Over25 000 lives were lost. Only those who had raced tothe tops of ridges survived. The disaster ranks withthe Mt Pelée eruption of 1902 for the completenessof its destruction.

Many large landslides described as such in the liter-ature should in fact be classified as debris avalanches.Almost all have originated in steep mountain areas. Forinstance, the Bialla slide referred to above as a mega-slide, and shown in Figure 12.14, is technically a debrisavalanche. Historically, many large landslides fit intothis category. At Elm, Switzerland, on 11 September1881, about 10 million m3 of mountain slipped ontothe town, burying it to a depth of 7 m and killing 115people within 55 seconds. A debris avalanche of similarsize at Goldau in Switzerland, in 1806, buried fourvillages and killed 457 people. On 4 September 1618,debris avalanches in the Chiavenna Valley in Italykilled over 2400 people. Larger disasters have alsotaken place. Reportedly, the total population of 12 000in the town of Khait, Tajikistan, was killed in 1949 byconverging avalanches triggered by earthquakes in thePamir mountains. The largest amount of debris everinvolved in a recorded failure occurred in the samemountains at Usoy in 1911. Approximately 2.5 � 109 m3

of rock failed, damming the Murgab River and forminga lake 284 m deep and 53 km long. The debrisavalanche was triggered by an earthquake registering 7on the Ms scale that destroyed the town of Usoy, killing54 inhabitants. A failure one-tenth this size, containing0.25 � 109 m3 of soil and rock, fell into a reservoiracross the Vaiont Valley of Italy on 9 October 1963.The impact sent a wave of water 100 m high overthe top of the dam and down the valley, killing 2000people.


Air-supported flows (avalanches)(Bolt et al., 1975; Scheidegger, 1975; Whittow, 1980;Wolman, 2003–2004)

Avalanches are commonly thought of in terms of ice orsnow cascading down slopes; however, they are notrestricted to frozen water. Snow avalanches are part ofa phenomenon of debris movement involving theinterstitial entrapment of air. Any material may beinvolved, including hot gases entrained within volcanictephra (as described for pyroclastic flows in Chapter11), mixtures of sediment and water in debrisavalanches (as described in the previous section), andcollapsing loess (as occurred during the Shensi, China,earthquake of 1556). The material in an avalancheflow is supported by air rather than by grain-to-graincontacts. Thus, avalanches are analogous to the lique-faction process whereby air, rather than water, fillsvoids and sustains the weight of the flow. Even if theavalanche makes contact with the ground, or containsa high proportion of liquid water, the internal aircontent ensures that coherence is minimal, and frictionso low that the avalanche can obtain high ground veloc-ities. Without turbulence, it is possible for velocities toreach in excess of 1200 m s-1. However, avalanchesrarely exceed 300 m s-1 because of the disruption tomovement by turbulence. Unfortunately, turbulencepermits most avalanches to entrain loose debris up tothe size of boulders, a process that can cause sub-

stantial erosion along part of its path. The swiftmovement ensures that such flows are very powerful,and that they are preceded by a shock wave, which forsnow avalanches may exert forces greater than0.5 tonnes m-2, as shown in Figure 12.16. The internalturbulence within a snow avalanche tends to generateswirls of debris with velocities in excess of 300 km hr-1,greater than the movement of the avalanche itself.Forces of 5–50 tonnes m-2 can be generated internally,and are of sufficient magnitude to uproot trees andmove large structures.

If the debris in the flow has liquefied, and if veloci-ties are high and the flow thin, it is possible for shearsorting to occur internally in the flow. Shear sortingtends to produce internal sorting of debris such thatlarger objects are expelled to the top and sides of theflow. Basal surges in volcanoes behave in this fashion,and produce ash deposits that are well-sorted with thecoarsest material towards the top of the deposit.The debris avalanche that wiped out Yungay in Peru in1970 preferentially moved boulders towards the topand sides of the flow as it underwent shear sorting;however, it is not certain whether water or air wasinvolved in the sorting process. The results of shearsorting enhance the prospect of survival for peoplecaught in snow avalanches. Because human bodies arerelatively large compared to the rest of the debris inthe flow, they tend to move to the top of the avalancheunder shear sorting. Rescuers simply probing the


air blast0.5 tonnes m-2

windows break,doors pushed in

uproots trees

destroys most buildings

can move reinforced concrete objects

vertical thrust

1-25 tonnes m -2

5-50 tonnes m -2

horizontal thrust

wood-framed buildings

destroyed

snow

Fig. 12.16 Process and impact forces generated by a snow avalanche (Whittow, 1980, © and reproduced with the permission of John Whittow,Department of Geography, University of Reading).

upper layers of the avalanche deposit find and rescuemany people buried by snow avalanches.

The word ‘avalanche’ has always conjured up imagesof snow cascading down mountains at high speed andwiping out unsuspecting alpine villages. This image isbased upon accumulative disasters in the EuropeanAlps. Snow avalanches are also a common hazard in theHimalayas, Andes, and Rocky Mountains, especially inspring. One of the more recent disasters happenedin 1978 at Col des Mosses, Switzerland, when anavalanche overwhelmed a ski-lift and killed 60 people.Even the Prince of Wales was almost killed by anavalanche that overtook his skiing party in 1987. Two ofthe largest disasters have occurred during wars.Hannibal’s 218 BC invasion of Rome across the ItalianAlps in winter was thwarted by snow avalanches thatkilled 15 000–18 000 of his men. During the FirstWorld War, snow avalanches – many of which weredeliberately triggered by shelling – killed around 40 000 Italian and Austrian troops fighting in theItalian Alps. Other disasters have been less costly, butjust as spectacular. Between 1718 and 1720 in Swit-zerland, four separate snow avalanches took over 250 lives. The largest Swiss avalanche, in 1584, took300 lives. In 1910, an avalanche at Wellington,Washington, in the Rocky Mountains, swept throughtwo trains killing 96 people. In the same year at RogersPass, British Columbia, avalanches killed 62 people. Asrecently as 2000–2001, avalanches killed 176 peopleworldwide.

In recent years, snow avalanche disasters havedecreased with the advent of monitoring techniques andremedial measures designed to negate the hazard. Forinstance, most zones have been mapped in the popu-lated mountainous areas where snow avalanches arelikely to occur. The fact that avalanches tend to recur inthe same places has aided this mapping. Sharp increasesin temperature, after a winter of heavy snowfall orduring a spring accompanied by rainfall, produce idealconditions for avalanches. Studies of the dynamics ofsnow and ice on slopes have allowed these conditions tobe identified to a high degree of accuracy. Sensitivemicropenetrometers driven into snow at 2 cm s-1 canproduce detailed profiles of snow hardness and identifylayers of loose snow underlain by snow that has meta-morphosed into ice called depth hoar that forms primeconditions for avalanching. In milder environments, re-freezing of melted slush – called ‘zarame-yuki’ in Japan– can produce similar conditions. In the North

American Rockies and the European Alps, army fieldcannon or explosives are used to trigger snow avalanchesbefore they can build up to catastrophic dimensions.These techniques have proved effective in keepingpasses free of large disruptive flows, and in protectingski slopes in resort areas. Where road and train routesmust be kept open, avalanche chutes are used to divertfrequently recurring flows. These chutes simply consistof roofs that permit the avalanche to pass harmlesslyover the road or railway. Near settlements, where thesetechniques are impractical, other techniques forbraking, deflecting or disrupting the flow have beenestablished. Upward-deflecting barriers can be con-structed along preferred avalanche pathways to converta ground-based avalanche into a slower, less forceful,airborne one. Concrete blocks can be arranged onslopes to divert avalanches away from settlements. Thejudicious alignment and reinforcement of buildingsfacing avalanche-prone slopes can also negate damageand prevent loss of life.

SUBSIDENCE

(Sharpe, 1968; Bolt et al., 1975; Coates, 1979, 1985;Whittow, 1980; Hays, 1981)

Land subsidence is a major type of land instability thatdoes not depend upon many of the basic mechanismsoutlined in this chapter so far. Rather, it is due tofactors that include the effects of humans throughwater and oil extraction, and mining activities. Theproper discussion of these latter factors is beyondthe scope of this text. In this section, only the naturalcauses and mechanisms of land subsidence will bediscussed.

Land subsidence is important at several scales ofmagnitude, spatially and temporally. For instance, asmall sinkhole caused by limestone solution is notsignificant except to a local landowner; however,uniform subsidence of several square kilometres canbecome severely disruptive to a much wider commu-nity. In a few minutes, the Alaskan earthquake of 1964produced regional, tectonically induced subsidencealong the south coast of Alaska that had a profoundeffect upon the landscape. This sudden change in therelative location of land and sea resulted in severeeconomic problems in harbors. At the same time, thereare locations – such as the south coast of England –where long-term subsidence of the crust has beenoccurring since the demise of the last continental


icesheets, and as far back as the Pliocene. While coastalerosion and flooding can result from this regional sub-sidence, it is possible for humans to adjust to thesechanges.

Natural mechanisms causing land subsidence fallinto three groupings: chemical, mechanical andtectonic. Chemical agencies of land subsidence mainlyinvolve solution by groundwater of limestone, halite,gypsum, potash, or other soluble minerals. Becauselimestone is very extensive, the development of karsttopography can lead to sudden ground subsidence overlarge areas. This is a particular problem in the south-eastern and mid-western parts of the United States,areas that are underlain by extensive limestonedeposits. In Australia, subsidence features are wide-spread across the Nullarbor Plain; however, the regionis neither densely settled nor intensely farmed, so thatkarst subsidence has little effect on people. A rarelyrecognized chemical cause of subsidence is the naturalignition (or, more recently, industrial vandalism) of coalseams such as has occurred at Burning Mountain,north of Scone, New South Wales, Australia.

Mechanical factors involve the removal, by extrusion,of material at depth in the soil. This can include theejection of quick clays. However, by far the most effec-tive process is the melting of ice lenses in permafrostareas. Because living and decayed vegetation providesan insulating blanket against the penetration of heatinto permanently frozen ground, its removal will oftenresult in catastrophic and irreversible melting of groundice and subsequent surface collapse. The emplacementof heat conduits, such as foundations and telephonepoles, will also permit the transfer of heat from thesurface to depth where melting will occur. Even heatfrom a building can conduct into permafrost to causemelting and subsidence unless it is prevented fromdoing so by an intervening layer of insulation. Theeffect of subsidence in permafrost regions has majorimplications for landform development and the survivalof fragile, cold-climate ecosystems.

Tectonic factors involve warping of the Earth’s crust.The Earth is an elastic body that can be deformed bythe weight of glaciation, and that will return to itsoriginal shape after the ice load is removed. Glaciationcauses forebulging some distance in front of the icefront. In these regions following glacial retreat, thecrust settles instead of rebounding. Large sections ofthe southern coastline of England, including Londonand northern Europe, are affected by this process, and

are presently undergoing subsidence at rates in excessof 2–3 mm yr-1. This is being exacerbated by the long-term downwarping of the southern North Sea Basin. Ifthere is a threat of sea level rise in the next century,then it is possible for this rise to be exacerbated bytectonic subsidence in these regions. Subsidence canalso be caused by ground loading. This is an obviousfactor where major rivers are presently depositing largevolumes of sediment on shallow continental shelves.Much of the coastline surrounding the MississippiDelta is subsiding because of continual, long-termdeformation of the crust due to the deposition ofsediment debouched from the Mississippi River.

CONCLUDING COMMENTS

Land instability hazards encompass the most diverserange of processes and phenomena of any hazard dis-cussed in this text. They also occur over the greatesttime span, from soil creeping downslope at ratesof millimetres per year, to the various types ofavalanches reaching speeds in excess of 300 km hr-1.In many respects, land instability is the mostcommonly occurring hazard. For instance, solifluc-tion affects 20 per cent of the Earth’s landmass, whileexpansive soils and landslides impact on 33 per centand 14 per cent, respectively, of the conterminousUnited States. However, the topic of land instabilityhazards is not an easy one to cover. The fact thatthese hazards have attracted equal attention fromengineering- and geology-related disciplines has ledto a confusion of terms and a discipline-specificemphasis. This has meant that some importantphenomena such as expansive soils are relegated intreatment to disciplines, in this case soil science, thatdo not always attract the attention of natural hazardresearchers. It has also made exchange betweendisciplines difficult, not only because of the clashover terminology, but also because engineers tendto centre discussion around mathematics whilegeologists and geomorphologists are more interestedin processes and morphology. Despite attractingthe attention of a wide range of researchers, manyaspects of land instability remain unclear. Forinstance, little is known about the mechanics ofdebris avalanche movement. This has meant that ourability to predict the occurrence of many landinstability hazards is far from effective. For example,research in the Wollongong region of Australiawould indicate that landslides should be prevalent


whenever the monthly rainfall total exceeds 600 mm.This condition was met for the month of April in both1988 and 1989. While numerous artificial embank-ments failed, the number of landslides to occur onnatural slopes did not reach these predictions. Finally,land instability hazards are frequently associatedwith other natural hazards. Flows, landslides, andavalanches can all occur as secondary phenomenafollowing tropical cyclones and other large-scalestorms, earthquakes, and volcanic eruptions. Evendrought affects soil stability because of its control onthe frequency of soil contraction and expansion. Forthese reasons, land instability hazards are secondaryphenomena and cannot be ranked as important asfloods, storms, earthquakes, or volcanoes.


Aune, Q.A. 1983. Quick clays and California’s clays: no quicksolutions. In Tank, R.W. (ed.) Environmental Geology. OxfordUniversity Press, Oxford, pp. 145–150.

Blong, R.J. and Johnson, R.W. 1986. Geological hazards in thesouthwest Pacific and southeast Asian region; identification,assessment, and impact. BMR Journal of Australian Geology andGeophysics 10: 1–15.


Bowles, J.E. 1984. Physical and Geotechnical Properties of Soils(2nd edn). McGraw-Hill, Sydney.

Chowdhury, R.N. 1978. Slope Analysis. Elsevier, Amsterdam.Coates, D.R. 1979. Subsurface influences. In Gregory, K.J. and

Walling, D.E. (eds) Man and Environmental Processes. Dawson,Folkestone, pp. 163–188.

Coates, D.R. 1985. Geology and Society. Chapman and Hall, NewYork.

Cornell, J. 1976. The Great International Disaster Book. Scribner’s,New York.

Crozier, M.J. 1986. Landslides: Causes, Consequences and Environ-ment. Croom Helm, London.

Finlayson, B. and Statham, I. 1980. Hillslope Analysis. Butterworths,London.

Goudie, A. (ed.). 1981. Geomorphological Techniques. Allen andUnwin, London.

Hays, W.W. 1981. Facing geologic and hydrologic hazards: Earth-science considerations. United States Geological SurveyProfessional Paper 1240-B: 54–85.

Innes, J. L. 1983. Debris flows. Progress in Physical Geography 7:469–501.

Leopold, L.B., Wolman, M.G., and Miller, J.P. 1964. FluvialProcesses in Geomorphology. Freeman, San Francisco.

Scheidegger, A.E. 1975. Physical Aspects of Natural Catastrophes.Elsevier, Amsterdam.

Sharpe, C.R.S. 1968. Landslides and Related Phenomena: A Studyof Mass-Movements of Soil and Rock. Cooper Square, New York.


Wolman, D. 2003–2004. Charge of the ice brigade. New Scientist20–27 December 2003–3 January 2004: 44–46.

Varnes, D.J. 1978. Slope movement types and processes. InSchuster, R.L. and Krizek, R.J. (eds) Landslides: Analysis andControl. National Research Council Transportation ResearchBoard Special Report No. 176: 11–33.

Voight, B. (ed.). 1978. Rock Slides and Avalanches I: NaturalPhenomena. Elsevier, Amsterdam.

Young, A. 1972. Slopes. Oliver and Boyd, Edinburgh.Wieczorek, G.F., Larsen, M.C., Eaton, L.S., Morgan, B.A. and Blair,

J.L. 2001. Debris-flow and flooding hazards associated with theDecember 1999 storm in coastal Venezuela and strategies formitigation. U.S. Geological Survey Open File Report 01-144,<http://pubs.usgs.gov/of/2001/ofr-01-0144/>


SOCIALIMPACT

P A R T 3

INTRODUCTION

The mitigation and survival of any natural hazardultimately depends upon the individual, family, orcommunity. An individual has the choice to heedwarnings, to prepare for an impending disaster, and torespond to that event. In the end, the individual, familyunit or local community endures the most of a disasterin the form of property loss, injury, loss of friends orrelatives, or personal death. If these small social unitscould recognize the potential for hazards in theirenvironment, and respond to their occurrence beforethey happened, then there would be minimal loss oflife and property. Unfortunately, not everyone heedsadvice or recognizes warning signs, not necessarilybecause of stupidity, but for very important personaland socio-economic reasons. The reaction of individuals,families, and community groups after a disaster alsodetermines their ability to survive physically andmentally. Some individuals and families can overcomeall bureaucratic obstacles in rebuilding. Others end upwith shattered lives despite all the support available tothem. This chapter will describe and account for someof these personal and group reactions.

BEFORE THE EVENT

(Burton et al., 1978)

Warnings and evacuation

People react differently to their perception of a hazard.While some may not know that an earthquake is aboutto occur until they hear the rumbling and feel theshaking of the ground, most will know that they live inan earthquake-prone area. For example, there are23 000 000 people living along the San Andreas Faultin southern California. The bulk of this populationknows that a major earthquake is overdue, but contin-ues to live there despite the risk. People are faced withthe same decision in Wellington, New Zealand, wherethe occurrence of a major earthquake could destroymost of the city at any time. Wellington is expandingalong its fault zone. In both cases, it is the ‘sense ofplace’ or ‘home’ that overrides all common sense aboutthe threat of the hazard. People born in an area tendto want to stay there. Home is familiar, it has peopleone knows and associates with at a personal level.There is some sort of historic identity with the area,which is difficult to give up, and which is defendedagainst change. There is a need to maintain links withthe past or with ancestors, to sustain one’s roots. Peoplerequire an association with place and history. No threatof a hazard will make them leave, and they will ratio-nalize the threat to minimize its occurrence. Residents

Personal and Group Response

to HazardsC H A P T E R 1 3

of San Francisco refer to the ‘Great Fire of 1906’rather than the ‘Great Earthquake of 1906’. In thesepeople’s minds, great fires are rarer than great earth-quakes and, hence, a similar disaster would be lesslikely to recur. If a place is attractive climatically, has analluring lifestyle, or promises better economic oppor-tunities, then it will continue to grow despite anycalamity that can be associated with it. That is whypeople continue to move to California, that is whyMexico City has grown so rapidly since 1960, and thatis why Wellington, New Zealand, continues to grow.

In the short term, even in the above locations, awarning of imminent disaster can also invoke a casualreaction bordering on disregard. For example, duringthe 1983 Ash Wednesday bushfires in Australia, somepeople nonchalantly sat around pubs escaping the heatof the day, while they watched one of the worst con-flagrations witnessed by humanity bearing down onthem (Figure 13.1). A non-evacuation reaction may bemacho-based bravado, a public display of fearlessness,a normal psychological response of a low-anxiety indi-vidual, a religious taboo, or even the sensible thing todo. An example of a religious taboo occurred inBangladesh (East Pakistan) during the 1970 cyclonesurge. At the time the cyclone struck many womenwere prohibited, by established religious practice,from going outside for the period of a month. Even ifresidents had been given warning of the storm surge,few women would have evacuated and broken thattaboo. Non-reaction to imminent disaster can occur for

other reasons. Warnings of disaster imply evacuation,and that may appear too radical a departure fromeveryday life to be acceptable. Peer-group pressurealso may be at work. If you are the only one in yourneighborhood to evacuate, then you may be ridiculedor made to look silly. This has always been a humanresponse to warnings of doom. For example, in 1549, acadi (judge) in eastern Persia predicted an earthquakefor his city. After trying to persuade his friends to leavetheir homes for the open, and after spending timealone in the cold outside, he eventually returned homeonly to be killed along with 3000 others in the ensuingearthquake (Michaelis, 1985).

Impending disasters are also times of worry, fear,and confusion. Most people at Hilo, Hawaii – reactingto sirens warning of the approach of a tsunami follow-ing the Chilean earthquake of 23 May 1960 – wereconfused (Lachman et al., 1961). Only 41 per cent ofpeople evacuated to safer locations following thesirens. Twenty-seven per cent of the population,including some who evacuated, did not know what thesirens meant, while 46 per cent thought they were onlya preliminary warning. More worrying was the fact thatno mechanism existed to inform people it was safe toreturn. Of the sixty-one people killed by the tsunami,many believed it was safe to return after the first fewwaves had arrived at shore. The overt decision to ‘stayput’ may represent a subliminal mechanism for copingwith the unknown and the unexpected. Generally,those of low socio-economic status, ethnic minorities,and women respond least to warnings. Older peoplefind such change difficult to handle physically andmentally. For example, one old man died in the MtSt Helens eruption because he refused to move despiteall the warnings. He had lived there all his life, andnothing was going to make him move. During tropicalcyclones in the United States, the militia has evacuatedpeople, at gunpoint, from coastal areas. Old couplesdied in the 1983 Australian bushfires along the coastsouth of Melbourne, because they had retired to thosecommunities and there was nowhere else for them togo. If their retirement home were to be destroyed byfire, they would stay behind to save it, or ‘go down withthe ship’.

This reaction to imminent disaster can be explainedin psychological terms. High-anxiety people perceiveevents occurring around them as confirming theiranxiety. They tend to take action and in fact be psy-chologically conditioned for survival. Similarly, people

274 Social Impact

Fig. 13.1 Drinkers at a pub in Lorne, Victoria, casually watching abushfire bearing down on them on Ash Wednesday,16 February 1983 (photograph © and reproduced courtesyof The Age, Melbourne). This fire and others turned into oneof the greatest conflagrations ever witnessed.

who believe that they are in control of their life willalso take action to survive. In contrast, people whoare low-anxiety or fatalistic believe their lives arecontrolled externally. This difference also relates tovariation in socio-cultural groups. For example, asmentioned in Chapter 3, Sims and Baumann (1972)found that there was a higher incidence of death due totornadoes in the south of the United States than in thenorth. People in Alabama distrust the Weather Bureauand its warnings. They await their fate and, as a result,there is a greater incidence of death there because oftornadoes. On the other hand, people in Illinoisbelieve they are in control of their own lives, are tech-nologically orientated, and take precautions against thethreat of tornadoes. They better survive the risk asa result.

For some, the decision to evacuate is an economicone. Any evacuation costs money, no matter who orga-nizes it or carries it out. An evacuation involves loss ofincome for most of the community. For example,fishermen in the Caribbean do not evacuate duringcyclones. Staying behind with their boat is moreacceptable than evacuating, knowing that it is going tocost them money to move, and loss of income for thetime they have moved. Loss of income may be unac-ceptable if your income is paying for an expensivemortgage, or keeping you from the moneylender. Inthat case, evacuation will be contemplated only if thehazard is a certainty.

The ‘cry wolf’ syndrome lulls others. For example,even though the residents of Darwin in 1974 had up tothree days’ warning of the arrival of Cyclone Tracy,only 30 per cent took precautions. This was because acyclone had approached Darwin in the previous week,and had veered away. False cyclone threats are alwayshappening in Darwin. Tracy also had the misfortune ofoccurring just before Christmas. Preparing for Tracymeant interrupting Christmas celebrations for acyclone that might never come. Evacuation meantgoing inland, where there were no facilities, and thepossibility of being trapped by heavy rains. Similarly,in the United States, tornado warnings are ignoredbecause they are issued for very large areas, when theirprobability of occurrence in part of that area is small.The Weather Bureau assumes that the total populationwill respond to a warning. However, so few peoplehave actually experienced a tornado that the majorityof the population fails to take any precautionarymeasures. Civil defense groups in the United States

now make more dramatic pleas for evacuation. Forinstance, after a newscaster has broadcast a tropicalcyclone warning in the weather report, the head of thecivil defense organization may interrupt further broad-casting, identify who they are and, in an authoritativevoice, state that the cyclone threat is now real andpeople should take immediate action to evacuate fortheir own personal safety. Of course, this did not occurbefore Hurricane Andrew made landfall in 1992.

There are also those who are quite willing toevacuate, even though the signs of impending disasterare not obvious to the public. Many people in touchwith nature, such as North American Indians, have leftareas of disaster weeks in advance, responding to subtlesigns shown by insects and animals. For example, someIndians in the Mt St Helens area packed up and leftbefore the eruption, because they noticed animalsdoing the same thing. Indian tribes in southern Califor-nia have left flood-prone areas in the midst of drought,because they noticed insects and ground burrowinganimals seeking higher ground. Several weeks laterflash flooding occurred when the drought broke.

Evacuation can also be dangerous. It may seemadvantageous to flee the landfall of an approachingcyclone and, certainly – if warning is given of an earth-quake – it is wise to leave buildings and head for openspaces. The same type of evacuation might also seemappropriate if a wildfire is swooping towards yourhome at 20 km hr-1, as happened in Australia’s 1983Ash Wednesday bushfires. However, a study by Wilsonand Ferguson (1985) of houses occupied at MtMacedon showed that it was better not to flee thisbushfire. Only 44 per cent of unattended housessurvived compared to 82 per cent of occupied houses.This evidence was ignored during the devastatingbushfire that swept through Canberra 20 years later.Six people lost their lives at Mt Macedon; however,this could not be attributed to the fact that peopleremained at home to fight the fires. Evacuation duringthe fire was downright deadly. Other evacuees cloggedroads, smoke made driving impossible, and some carsended up trapped in exposed areas, where death fromheat radiation was a risk. Similarly, it is unwise to try toflee approaching tornadoes in cars. Studies in theUnited States have shown that 50–60 per cent ofdeaths and injuries from tornadoes are caused by flyingglass from broken windscreens.

In some cases, where the disaster occurs suddenly,evacuation may not be a viable option. During most

275Personal and Group Response to Hazards

earthquakes or landslides, individuals have only afew seconds to react. Personal decisions must be delib-erately made against a background of seeminglyorganized flight by tens or hundreds of people. Thedecision to ‘buck the trend’ becomes mentally moredifficult and, in certain cases, physically impossible. Asan expert, I know that standing in the open during alightning storm is dangerous. Yet I have stoodimmobile watching a junior soccer match under suchcircumstances, too shy to march onto the field andorder everyone off because of an imminent threat.Most individuals, be they sane or not, demonstrateremarkable composure at the threat of a nature hazard.

Preparedness, if warned

In many of the above cases, people who stay put anddo not prepare for disaster are the exception. Forexample, in the United States, millions of people haveevacuated from the path of a hurricane in a timely andorderly fashion. The threat of personal loss of life, orloss of close family, outweighs the risk of staying. Formost inhabitants of cyclone-threatened areas in Aus-tralia, the approach of the cyclone means tying downloose objects in the home and yard, stocking threedays’ supply of food and water, and taping windows.Such precautions are taken seriously and initiatedindividually.

Preparation for imminent disaster by individualscenters around four areas: self, family, property, andcommunity – in that order. Organizing can rangefrom packing one’s clothes and toothbrush to mentalpreparation. The latter usually includes some sort ofcatharsis such as conveying apprehension or fear tooneself, family or friends, or taking a sedative or alco-holic drink. In some cases, it may take the form ofprayer for divine aid or intervention. Mental prepara-tion appears to be a first priority in meeting any hazard.For those with immediate family, the second priorityis ensuring their safety, preparedness, and mentalstability. Attempts may be made to check the safetyof close family living in the same threatened area, orcommunicating assurances of safety to distant relatives.

Next, there is a direct concern for property. Usually,this involves the immediate home either well before, orimmediately preceding, any disaster. For example,McKay (1983) found following the 1983 AshWednesday fires that the degree of preparation forbushfires in the Adelaide Hills varied throughout thearea. Fewer than 20 per cent of people did nothing.

Fifty-four per cent were involved in either temporary orpermanent measures. Temporary measures involvedsuch preventative tasks as clearing gutters of leaves, andcutting away undergrowth around buildings. Permanentmeasures involved structural changes to the house,landscape modifications, or the construction of asprinkler system. These remedies required little main-tenance, but were more expensive. There was no socio-economic relationship between the numberand types of measures used. However, people whorented their accommodation tended to perform fewermeasures. About 60 per cent of people carrying outpreparations did so after seeking advice from neighbors,friends, or relatives. Only 9 per cent obtained informa-tion from public bodies and 11 per cent from theCountry Fire Brigade. Drabek (1986) also found asimilar trend for many other hazards worldwide.Obviously, people trust neighbors or relatives more foradvice than they trust government bodies with exper-tise. Significantly, people who had neighbors who hadexperienced bushfires, or people who themselves wereaffected in the previous few years by fires, took greaterprecautions. In Australia, people’s enthusiasm forimplementing preventative measures against bushfiresmay wane with the memory of the last fire. As the threatincreases, large segments of the community may beunprepared for bushfires. Overcoming this may requirean annual advertising program, making people awareof the continuing threat. A similar pattern has beenfound in the United States with people’s degree ofpreparedness to cyclones.

During any tragedy, victims usually attempt to savetheir home. If there is little hope that the house can besaved, then personal belongings or mementos such asphotographs, trophies, or presents are rescued first.If there is time, then more expensive, transportableproperty is saved. Reaction to pets can also be includedin this behavior. Sometimes considerable risk can beexpended with people dying in attempts to save petsor livestock. For example, floods in Australia haveseen farmers sheltering dairy herds in their house.Evacuation of 40 000 families in Mississauga, Ontario,Canada, following a gas leak in 1979, also involved30 000 pets. If preparation for an imminent hazard canbe completely carried out, and the victim perceivesthat such preparation has enhanced their chance ofsurvival, or preserved personal property, the shockof the event is lessened and recovery after the eventis more successful.

276 Social Impact

DEALING WITH THE EVENTAND ITS AFTERMATH

Response during the event

Response during the event centers on the same fourareas focused on by people preparing for a disaster. Itis doubtful if anyone going through a tropical cyclonewill be worrying about the state of the American orAustralian economy. Almost all descriptions of humanbehavior during a natural disaster refer to a lack ofpanic. People may be scared but, in the face of adver-sity, most remain outwardly calm. Hysteria is rarely dis-played. In an extensive search of the literature, Drabek(1986) could find no evidence that disasters generatedmass fear or panic. Generally, the media overtlyreinforced a panic myth by exaggerating incidents ofnon-rational behavior. If panic exists, it usually repre-sents a rapidly reinforced, collective response to fleesome very real threat; or panic occurs simply becausepeople in a rapidly developing disaster situation, suchas a fire, spend critical time confirming the threat, thusleaving little time to flee.

It is a human instinct to try to protect one’s familyduring disaster. This instinct tends to override all fear.One mechanism ensuring maximum protection is forfamily members to remain close or huddled togetherduring life-threatening situations. Tropical cyclonevictims in Australia usually find themselves huddlingtogether for refuge in the toilet, because in new housesthis room is best designed to withstand high winds.During Cyclone Tracy in 1974, huddling environmentswere not always conducive to survival. One family whohad huddled together in bed in the main bedroomfound that they were exposed to the full force of thestorm as the house disintegrated around them. Thebed began to drift downwind across the floor towardsthe edge of the house and came to rest only when oneof the legs fell through the floor. Earthquake victimsare often found huddled together in the crush of fallenbuildings. In miraculous situations, such closenessoften permits an adult to save a child. For instance, afather, fleeing the debris flow of the Mt Huascaránearthquake in Yungay, Peru, in 1970, managed tothrow his two children up a slope to safety just as theflow overtook and killed him. Following the Nevadodel Ruiz lahar of 1985 in Columbia, a child was foundtrapped in a building in mud up to her neck, proppedup by her father who had suffocated to death.

The above examples illustrate another basicresponse during disaster, and that is the innate instinctfor survival. The ‘miracle babies’ of the 1985 Mexicanearthquake survived in the maternity ward of acollapsed hospital for up to two weeks after the earth-quake because their bodies went into a natural state ofhibernation that conserved fluids and energy. Victims –some severely wounded – have been dug out ofcollapsed mines, earthquake rubble, bombed buildings,and avalanches after similar periods of incarceration.Almost all survived because of a determination to live,a feeling that it was too early to die, a feeling that therewas still more to be done in life. The Tarawera eruptionof 10 June 1886, in New Zealand, buried Maori villagesunder more than 2 m of ash and mud, killing 156people. A 100-year-old survivor, Tuhoto of the Arawatribe, was found alive and in good health after beingburied in his house for four days. He, in fact, told hisrescuers to go away because the gods were taking careof him. He died ten days later under hospital care.The girl trapped in the lahar mud in Columbia inNovember 1985 met a similar fate. She remainedcheerful and survived for several days stuck inthe mud, only to die just before being pulled free. Thephysical devastation greeting the first outsidersto arrive in Darwin after Cyclone Tracy led many toexpect more than the 64 deaths that actually occurred.Over 90 per cent of homes were totally destroyed andmany families endured Tracy completely exposed tothe elements. The small number of casualties under-scores the ability of people to endure more hardship,deprivation, undernourishment, and shock than isnormally believed possible. It is only after the eventthat survivors must cope with the psychological traumathat inevitably sets in.

Humans may simply be curious to see what is goingon during a disaster. This behavior is not to beconfused with sightseeing, which involves people whohave not experienced the disaster wanting to see whathas happened. For example, during the passage of theeye of a hurricane, almost everyone wants to go outsideand look around. You and your family appear to beuninjured, but you wonder, ‘What does the house looklike, where’s the dog, how are the neighbors doing?’When a flood crest affects the Mississippi River systemin America, people who could be flooded are alwayswandering down to the river to look at the river’sheight. During eruptions, Mt Vesuvius has alwaysattracted local sightseers curious to find out what is


going on. Even residents of St Pierre journeyed up toMt Pelée in 1902 to see what was happening a few daysprior to the eruption. Ten thousand people flocked toSan Francisco beaches after the 1964 Alaskan earth-quake to see the tsunami coming in. The death toll inAmerica during that tsunami event, and the one thathit Hawaii in 1946, resulted from mainly curious (andprobably uninformed) residents returning home toevaluate the damage too early.

Maybe there is some euphoria to living throughnatural disasters or record-breaking events. The newsis reporting the heaviest one-day rainfall on record andyou wander outside to stand in it. I can still rememberas a child awaiting for days the approach of HurricaneHazel in southern Ontario, because no one in the areahad ever experienced a tropical cyclone. When Hazelfinally struck, there was a sense of relief that it hadarrived at last. The same thrill of expectation goesthrough anyone who is witnessing for the first time theinitial stages of a ground tremor. Then reality sinks in.The tremor becomes a major earthquake; the non-chalant attitude to a cyclone becomes a nightmare asthe house disintegrates around you; the flood hitshome as rumors of friends being wiped out becomefacts. The disaster has struck, and new emotions andneeds arise. Mechanisms for cleaning up after thedisaster must be carried out, and reconstruction andrecovery instigated. At this stage, the emphasis shiftsslowly away from the individual towards social groups.

Death and grief

One of the immediate aspects of a disaster is copingwith the injured and the dead. Often the search forsurvivors and rescuing the injured become majorcomponents of large-scale natural disasters, requiringinternational cooperation. While rescue operationstend to take precedence over coping with the dead,dead bodies pose special problems, both hygienicallyand legally. The health aspect of dead, but unburied,bodies is obvious. Decaying corpses, human or animal,breed disease and must be cleared away within hoursor days if epidemics are to be avoided. The more chal-lenging problem occurs in trying to define people aslegally dead. The problem is minor in the case ofunidentifiable bodies. Forensic science can be used tomatch bodies against reports of missing people. Aftertime, it will simply be classified as a ‘John Doe’ andburied as such. The more difficult legal problem occurswhen a person is reported as missing, and a body

cannot be found. Legally, western society presumesthat such a person is still alive. They could be wander-ing around with amnesia; they may have taken theopportunity to flee an unbearable family or worksituation in the confusion of the disaster; or they mightbe responsible for the disaster and are fleeingconviction. Unless there is a body, the person isconsidered – for up to seven years – to be alive. A willcannot be read, nor an estate settled, within this statu-tory period. In extreme cases, where a will has notbeen written, it is possible for a person’s estate to besealed off. A dead person’s car cannot be moved froma driveway; a house is locked up; and business assetsfrozen. Not only can such action be financially tryingfor the surviving relatives, but the presence of the deadperson’s possessions in the immediate vicinity is also aconstant reminder that they are missing.

In all societies, the trauma of death is resolved by aburial or memorial service. It allows family membersand close friends to publicly show feelings of grief.Mentally, it permits an individual to accept the deathand get on with life. Psychologically and sociologically,funeral services and death rites are essential toovercome issues associated with disaster fatalities. Ifthe disaster is large-scale and the death toll significant,then society organizes mass burials and memorialservices. An official period of national mourning mayalso be instigated following large disasters. Totalstrangers who have never known any of the dead orhad any association with the affected communitiesmay find that they want to attend some sort of funeral.Even in Australia, where less than 10 per cent of thepopulation regularly attends church services, specialservices had to be arranged to fulfill this need followingboth Cyclone Tracy and the Ash Wednesday bushfires.

Possessions and homes

While the death of close family may seem to be theworst outcome of a disaster, the destruction of propertymay leave longer-lasting scars. No funeral service isheld for a wrecked home, or a burnt photograph.Following the Ash Wednesday bushfires, people statedthat the worst aspect was losing, firstly, personalphotographs; secondly, prized trophies or awards;thirdly, property that had some sort of link to deadparents, children, or friends. Some even vocalized thatthe loss of possessions was worse than losing closefamily. Following a bushfire, nothing can replace oldfamily photographs or a melted-down trophy. Almost

278 Social Impact

everyone has seen pictures of people returning homeafter a natural disaster to pick through the ruins to findsome memento. The pictures often show people sittingafterwards in complete despair or shock when nothingcan be found. The personal belongings give some senseof continuity to life. They also provide a sense ofhistory – a connection with past events or ancestors. Asa survivor of Hurricane Andrew in 1992 said, ‘You workall your life, and for what? Insurance took good care ofme, but I lost my home. Not just my house and every-thing in it, but my home’ (Santana, 2002). Survivorshave gone to extreme lengths to try to recover some ofthese possessions. Relatives are contacted to providecopies of significant photographs, and antique shopsmay be searched to obtain replicas of prized objects.

Many people immediately commit themselves torebuilding in exactly the same location. Often a dupli-cate of the original house is built. This was the case formany people rebuilding after the Ash Wednesdaybushfires. Homes were re-established in their entiretyto designs that offered little protection against repeatbushfires. The need to rebuild exact replicas can betaken to extremes. For example, following the SanFrancisco earthquake of 1906, the city rejected amodern design in favor of a street layout and a buildingprogram that would, as quickly as possible, recreatethe original city. Sometimes investment in assetsconditions reconstruction. For instance, following theHawke’s Bay, New Zealand, earthquake of 3 February1931, the town of Hastings was rebuilt to its original,narrow street plan, even though total destructionpermitted an opportunity to rebuild wider streets. Thedecision to rebuild in this manner was made at theadamant insistence of one councilor, who ownedthe only shop with any part of its structure stillstanding. The councilor wanted to save money by nothaving to rebuild a new shop, which he would have todo if the streets were widened.

Feelings about property can be quite intense. Forexample, following the autumn coastal storms of1974 around Sydney, Australia, I was walking withcolleagues along remnants of the beach in front ofa series of houses propped up by makeshift supports.As we dodged waves surging up amongst the rocks,one resident furiously yelled that we were on privateproperty and that he would ‘get the police after us’ fortrespassing. This incident highlights a second stage ofresponse following a natural disaster. After the initialimpact, and after all one’s resources have been utilized

to ensure safety of the family, the anger sets in. Forsome people, the response manifests itself in long-termpsychological shock that alters life forever.

Anti-social behavior

No group of people is so disdained as the sightseer orthe looter. Looting, while rare, does occur. However, italso tends to be exaggerated by the media. In theUnited States, the prospect of looting traditionally hasbeen so great following disasters, that the NationalGuard is often mobilized to block off affected areasand ordered to shoot looters on sight. Residents haveto show evidence that they live in an area, and mustrestrict their activities to the street in which they live.In Australia, looting is a reality after most disasters.People were arrested following both Cyclone Tracyand the Ash Wednesday bushfires, but no one was everjailed. The Lismore flood of 1954 saw looters takingwhatever they could carry from the business district,including a rescue boat that was used openly totransport stolen goods.

A real problem for rescue and relief organizers afterlarge disasters is coping with the sightseers. Whilepeople may just want to see what a natural disastersuch as a cyclone or bushfire can do, the victims oftenperceive sightseeing as an intrusion on their privacy bythe nosy, or as an opportunity by those who survivedunscathed to gloat over their luck and the victims’misfortune. In Australia, sightseeing can take onmammoth proportions, not only by the public but alsoby the politicians and bureaucrats. Almost everycabinet minister in the Whitlam government made afact-finding journey to Darwin after Cyclone Tracy.The same planes that evacuated survivors returnednearly as full with government sightseers. Followingthe Ash Wednesday bushfires, which occurred in themiddle of the 1983 election campaign, almost all partyleaders made the journey to affected areas – not, asthey took pains to state in front of the cameras, forpolitical reasons, but to see if they could offer genuinehelp.

People’s misfortune can also breed extraordinarilyghoulish behavior. For example, an old lady wasrobbed in busy downtown Toronto, Canada, in broaddaylight, after being blown to the ground in January1978 by wind gusts in excess of 160 km hr-1. As she laypinned to the sidewalk, a more agile motorist, in whatat first appeared to be an act of chivalry, stopped his carand ran to assist her. Before stunned onlookers could


react, he snatched her purse, hopped back into his carand sped away. Michaelis (1985) reports that after theNaples earthquake of November 1980, smartly dressedpeople in flashy cars were seen leaving the scene withorphaned babies and young children, abducted forthemselves or to be sold for adoption. Relief goods inNaples were also pilfered and resold on the blackmarket. Holthouse (1986) also reports that during the1918 Mackay, Queensland, cyclone and subsequentmassive flood, boat-owners were seen offering torescue people stranded on rooftops and in trees onpayment of a substantial fee: no fee, no rescue. Follow-ing Cyclone Zoe, which struck the impoverishedSolomon Islands of Tikopia, Anuta and Fatutaka on28 December 2002 with winds of 300 km hr-1, thepolicemen crewing a rescue boat demanded anallowance of $A1250 before they would leave thecapital Honiara to travel 1000 km with urgently neededrelief supplies. In the longer term, disasters attractstrangers and hangers-on, the type of people you werewarned about as a child. As Hauptman (1984) so aptlyputs it, ‘the sort of trashy people who follow disasters . . . people who could not hold a job any place else’. Justlike vultures flocking to encircle a dying body, thesepeople flock to victimize survivors, who often let theirguard down following a disaster. They have to becausethey are now reliant upon total strangers for assistance.The worse their financial plight, the more reliant theybecome. Moreover, for every honest stranger dispens-ing cash and materials for relief, there is a dishonestone disguised as a contractor or household merchantattempting to steal it back.

Perhaps the most gruesome aspects of humanbehavior are those that arise because of severe faminefollowing drought. Societies have ritualized humansacrifice on a grand scale in attempts to appease godsbelieved responsible for a drought. Couper-Johnston(2000) describes some of the more elaborate ritualsestablished by the Aztecs in Mexico in response toENSO and La Niña events. In the drought El Niñoyears around 1450, the Aztecs were sacrificing 250 000victims annually to curry favor with the rain god Tlaloc.Men were beheaded at the tops of pyramids builtespecially for such sacrifices, woman danced in ritualsbefore being slaughtered, and children were drownedafter having their nails extracted so that priests couldinterpret the signs of forthcoming weather from theircries. Another response to impeding death fromfamine is to sell your children and yourself into slavery.

Captains of ships recruiting labor for cotton and sugarcane plantations in Australia, Hawaii, and Fiji in thenineteenth century would scurry to South Pacificislands on the news of drought, because they couldeither steal a weakened population or coerce them intoslavery with offers of food and water. Indenture issimilar to slavery. In Mexico, successive droughtsfound 60 per cent of the rural population locked per-manently into forced labor by the beginning of thetwentieth century.

Finally, when all else fails, people resort to cannibal-ism to survive. The behavior is ubiquitous. Followingcrop failure in 1315–1317 in western Europe, peoplein Ireland waited at the fringes of funerals, then dugup the bodies from churchyards and ate them. InGermany, criminals were snatched from the gallows tobe eaten, often before they had died. Couper-Johnston(2000) describes other examples of cannibalism. In thedrought of 1200–1201 in Egypt, cannibalism becameso rampant that people walking in the streets were atrisk of being snatched by hooks lowered from thewindows above. Doctors ate their patients and guestswere invited to dinner by friends – never to be seenagain. When two parents were caught with a smallroasted child and burnt publicly for their crime, thepopulace ate one of the bodies the next day. Cannibal-ism has often become entrenched in societies experi-encing repeated drought. For example, during thedrought of 1640–1641 in China, the Ming dynastytolerated markets in human flesh.

Resettlement

In almost all cases of a major natural disaster, peoplewant to go back and resettle in the same place theywere living beforehand. This may seem odd if the placehappens to lie in a major earthquake zone such asthe San Andreas Fault, or a coastal area subject torecurring tsunami. There are a number of reasons whypeople return home, even when that environment hasbeen shown to be hazardous.

There may simply be no alternative place to live oroccupation to undertake. People cannot go elsewhere ifthey are employed in agriculture or fishing and theydo not know any other career. Additionally, they areexperts in farming or fishing in that particular area, andcannot afford to lose that expertise. For instance,people farming the rich soils surrounding volcanoesmay want to return after an eruption because they knowthat the tephra will increase crop yields. The residents

280 Social Impact

of the town of Sanriku on the east coast of Japan, whichhas been struck repeatedly by large tsunami, havealways returned because they know the fishing condi-tions along that section of coast. The latter example alsoillustrates that, in some cases, people return to the sameplace because there may be nowhere less hazardousfor them to go. The citizens of Managua, Nicaragua,also found themselves in that situation after the 1972earthquake. The whole country is seismic and no otherlocation is necessarily safer. In Nicaragua or Japan,movement from one area of the country to another doesnot necessarily increase a person’s sense of security.

Evacuation can also mean loss of assets. Byreturning, people can at least salvage something.Residents at Terrigal, north of Sydney, were caught inthat predicament following the Australian coastalstorms of 1974. The value of houses, perched precipi-tously at the edge of dunes eroded by storm waves,collapsed. Residents returned to these homes to try toprotect them from further erosion. Fortunately, forsome, housing prices climbed back to normal and theywere able to sell out with minimal loss. Unfortunately,for others, storms in 1978 encroached further uponproperties, and land had to be resumed for token sumsby the local council. Change after a disaster also raiseslegal complications, especially if governments attemptto restructure cities and cut across existing private orcommercial property boundaries. For traditionalpeoples, a high value is placed upon traditional landrights. All of these examples show that humans areterritorial, and may not want to surrender property nomatter what has happened to it.

There are also psychological benefits in reconstruct-ing at the site of a disaster. By rebuilding as fast aspossible, people can forget the actual event. They alsokeep busy at something they are familiar with, and canmaster. This keeps their minds off the disaster, buildsup self-esteem, and overcomes the feelings of hope-lessness and despair. It has already been pointed outthat the city of San Francisco was rebuilt almost exactlyas it had been before the 1906 earthquake and, afterthe 1983 bushfires in Victoria, some home-ownersrebuilt exact replicas of their original home. In the caseof San Francisco, the decision to rebuild in a hazardousarea leads to outright denial of the hazard. Residents ofSan Francisco will point out that San Francisco is assafe as any other area in southern California.

A loyalty to local history, and a commitment to alocality, also draws people back to a disaster scene.

Rebuilding signals a loyalty to friends or relatives whomight have been killed. Their lives were not lost invain, and they will be remembered. For example,following the Colombian lahar tragedy, the wholeaffected area was declared a national shrine by thegovernment in memory of the 20 000 people buriedbeneath the mud. Within months, survivors wereresettling the area. This sense of loyalty also includesthe concept of ancestor attachments. Many peoplewant to return to their homeland. In some cases, thisrelates to ethnicity, in other cases, to respect forancestor links. Indian tribes in the Americas and theAboriginal groups in Australia have equally strongattachments to the land. To be removed and then notbe allowed to return to an ancestral homeland makeslife meaningless.

Religious beliefs or morals are sometimes used todismiss the hazardousness of an area. Many peoplebelieve that disasters are due to the wrath of a vengefulor angry god. Disasters do not occur because areas arehazardous, but because the people living there didsomething wrong. For instance, the Lisbon earthquakeof 1755 was perceived at the time to be punishment forthe wickedness of its citizens, to be God’s answer tothe cruelty of the Portuguese Inquisition. Accounts ofthe destruction of the town of St Pierre, Martinique,after Mt Pelée erupted, mention the fact that the townhad a reputation for being permissive. Prophets ofdoom have predicted the imminent destruction ofSan Francisco by an earthquake because of its largehomosexual population. In Papua New Guinea, somemembers of the Orokaivan group rationalized the 1951Mt Lamington eruption as payback for the killingof missionaries. Hymns, gifts, offerings, and prayerscan placate an angry god who can cause disaster.Hazardous environments in Indonesia and thePhilippines are often viewed as being imbued with adivine spirit that can be quietened in this fashion.

Similarly, such beliefs can also dictate againstresettlement. For example, Maoris in New Zealandattributed the 1886 eruption of Mt Tarawera to an actof revenge by an alleged sorcerer. This supposedsorcerer was, in fact, Tuhoto, the same man referred toearlier who was found alive after being buried for fourdays. Tuhoto had visited some friends close to the siteof the eruption earlier in the year and, following hisvisit, the child of a chief had died. The child’s grand-mother accused Tuhoto of bewitching the child. WhenTuhoto heard about the accusation, he was so angered


that he called upon the god of volcanoes to wipe outthe whole village. Following the eruption, no Maoriwould have anything to do with Tuhoto. The site of theeruption and the surrounding devastated area wereconsidered taboo, and to this day are avoided byMaoris.

Resettlement may also be rationalized by a beliefthat modern-day technology can prevent destruction infuture disasters. In southern California, great faith isput into reinforcing older buildings and expresswayoverpasses against collapse. Minor earthquakes in 1971and 1989 have shown that not all this ‘technologicalstrengthening’ works. New Zealand is pursuing theidea of building large buildings on rubber bungs thatcan absorb earthquake shock waves and minimizedestruction. Certainly, in areas where this has beendone, earthquakes are not so destructive as they are inregions where buildings are not built to earthquakestandards. The large death tolls of the Agadir,Morocco, earthquake of 1960, the Armenian earth-quake of 1988, and the Bam, Iran earthquake ofDecember 2003, attest to this latter fact – even thoughthese events were of lesser magnitude than recentCalifornian ones with low death tolls.

Moving elsewhere after a disaster may also betoo disruptive to a wider segment of society. Evacua-tion of large numbers of people represents a major taskthat can put pressure on housing, government, andconsumer supplies in other parts of a country. By con-taining the survivors, efforts can be efficiently directedto a single area. For example, the evacuation of Darwinfollowing Cyclone Tracy (described in Chapter 2) wasnot effective. It caused disruption to the people whowere evacuated. Had the numbers been larger, itwould have severely disrupted the economy in thesouth. There is an inertia effect resisting the relocationof so many people. Large communities attract acommunications and business infrastructure that rep-resents a significant capital investment that cannot beeasily abandoned or relocated. For instance, Messina,Italy, after the earthquake of 1908, rather than beingabandoned, was rebuilt in the same area because it wasa significant port. Tangshan, China, provides an impor-tant link between Beijing and Tienshen on the coast.For this reason, it had to be quickly rebuilt followingthe 1976 earthquake. The only concession to thedisaster was the rebuilding of the city as a series ofsatellite cities spread out to minimize damage in anyfuture earthquake. Some countries such as Japan and

China have more centralized economies, where thegovernment decides the movement and location ofpeople. There is no personal choice to leaving orresettlement. The government makes the decision, andeither enforces or encourages it.

A final reason why people may choose not to resettleelsewhere is that the disaster may appear to be a one-off or exceedingly rare event. In this case, alternativesto resettlement in the same area do not arise. Forexample, tropical cyclones affecting southern Ontario,Canada, are relatively rare events. Hurricane Hazel in1954 represented a 1:200-year event with majorflooding in Toronto. While areas prone to such infre-quent flooding have been set aside in Toronto as openspace, and hazard planning takes account of this typeof flooding, no plans for massive evacuation have beenestablished in case another cyclone strikes. The eventis just too rare to be of significant concern. The sameis true of some earthquakes and most volcanoes.Historically, some of the largest earthquakes in theUnited States have occurred at locations such asNew Madrid, Missouri, and Charleston, South Carolina,which are not considered seismic. An earthquakedisaster movie set in South Carolina would not bebelievable. Nor does the disaster have to be infrequentto be dismissed. It just has to be perceived as infre-quent. The slopes of Mt Vesuvius in Italy, and the townof St Pierre, Martinique, have been resettled despitethe fact that the local volcanoes are still considereddangerously active. This perception of infrequencycan form very quickly, because people have shortmemories of disaster events. For example, within sixmonths of the 1974 Brisbane River floods in Australia,property values in flooded areas had returned to theirpre-flood levels.

Myths and heroes

The myth that people panic during disasters hasalready been dispelled in this chapter. There are anumber of other myths about human response duringdisasters that should also be dismissed. The first mythis that people experiencing the calamity appearconfused, are stunned, and helpless. If anything, disas-ters bring out the strength in people. Survivors, farfrom being helpless, are almost immediately back intothe rubble trying to save people and property. At thisstage in disaster relief, the victims are more thancapable of organizing themselves and performingthe rescue work. The only things they may lack are the

282 Social Impact

tools to carry out the tasks efficiently. Secondly, it isgenerally believed that disasters bring out anti-socialbehavior, stories of mothers dropping their babies to goafter a few seeds of grain spilled from a passing truck,of men pushing children from lifeboats on sinkingships. Such behavior is rare, or else it is often theproduct of media distortions and bears little resem-blance to reality. While there may be anti-socialbehavior – for example the ghoulish conduct of baby-snatchers in Naples mentioned above – survivors aregenerally very sociable. Disasters kindle communityspirit and altruistic acts in people and businesses. InAustralia, there are examples of total strangers in newcars driving up to flood and bushfire victims, tossingthem the keys and walking away. In the 1994 Sydneyfires, a major supermarket chain opened up its ware-houses and trucked away its soft drinks to be off-loadedfree at the fire-fronts.

Another aspect about disasters, which may or maynot be mythical, is the role of heroes. People know thatif they ever have the misfortune of being caught in adisaster and require help, there will be someone tryingto save them. In the United States, forest fires makeheroes. For example, the actions of the crew on theSt Paul and Duluth train to Hinckley, Minnesota,during the fire of 1894 were certainly gallant. Despitedecreasing visibility due to smoke, the train wascasually driven into the town – only to be met by 150fleeing townspeople, closely pursued by a wall of flame.The engineer decided to evacuate people to a marsh7 km back up the line, and throttled in reverse at fullspeed. However, the fire overtook the train, consumingone passenger car after another. As flames lickedthrough cracks and windows, the conductor calmlyused a fire extinguisher to put out fires on women’sdresses. In the cab, the situation was worse. Both theengineer’s and fireman’s clothes caught fire, andthe cab began to burn. The glass in the cab windowburst from the heat, cutting the engineer’s jugular vein;the controls caught fire; the seats began to smolder;and the cabin lamp melted. Several times both mencollapsed in the heat, but managed to keep the trainmoving. As the train reached the marsh and eruptedinto flame, both passengers and crew flung themselvesinto the mud where they sheltered for four hours.The fireman, however, after carrying the bleedingengineer to safety, returned to the train, uncoupled theengine and moved it to safety, while the conductorstruggled back to the nearest station to warn the next

train coming through. The crew’s actions saved over300 people.

In Australia, no one would doubt that Major-General Stretton’s work in the first weeks afterCyclone Tracy was heroic. His organizational skillsstand out as superhuman. He was able to mobilizeevacuations and ensure that the city was cleaned up,laying the foundations for the reconstruction that wasto follow. He made decisions that appeared innovativeand effective. He was given an impossible task tomake order out of chaos; and he succeeded withoutany additional casualties. Stretton rose to the occasion,but that is what was expected of him. He had amilitary background, and was in charge of theNational Disasters Organisation in Canberra. Histraining, experience, and position aided him inperforming his assignment.

Yet, there have been situations where commonpeople with few of these attributes have alsospontaneously risen to the occasion. When rescuersarrived at the scene two days after the 4 February 1976earthquake in Guatemala, they found an old man incharge. He had organized rescue, reassured survivors,and when the aid came in, he took charge of thedistribution of food and supplies. The other survivorsdeferred to his wisdom whenever a decision or actionwas required. In a way, hero stories reinforce society’ssense of security. Nowhere was this altruism moreevident than in the terrorist attacks on the World TradeCenter on 11 September 2001. One gentleman pusheda wheelchair-bound employee down 65 flights ofstairs to safety. Another chose to stay behind with ahandicapped person who could not flee. When thetranscripts of telephone calls made from the doomedbuilding were released two years later, they showedemployees carrying out their duties professionally. As aNew York Port Authority spokesperson said, ‘thesepeople were heroes until the end’.

ADDITIONAL IMPACTS

(Dalitz, 1979; Reser, 1980)

Emotional problems

‘Why did it have to happen to me? I’ve worked hard allmy life and the house is gone. There stands George’shouse next door. He built it after winning the lotto. Theshrubs are stolen, and he was at the pub all through thefire. Look, his fence isn’t even scorched.’


This scenario describes well the reaction that begins toset in after a disaster. While humans are quite resilientduring disaster, a state of shock or numbness thatborders on self-pity sets in afterwards. These feelingsinevitably turn to anger within 24–48 hours: maybeanger at oneself for not preparing better for the event,maybe anger at a neighbor who better survived thedisaster, maybe anger at some larger organization suchas the Weather Bureau for not giving enough warningabout the approach of a cyclone, maybe anger at someunidentified group such as the conservationists whoprevented prescribed burning. The object may be realor imaginary. It makes no difference. One just becomesangry. The outlet may be physical or verbal abuse. Acolleague of mine walked into a landslide disaster area,within the anger time frame, to take pictures, and leftwith a bleeding nose after being punched in the face.Commonly, people must talk about the event to anyonewho will listen. The press preys upon this characteris-tic to get pictures and material on the disaster, andoften records this anger stage. In some cases, the pressbecomes the object of the resentment. Relief andsocial workers are aware of the fact that if they remainaround the scene of a disaster long enough, they mustexpect to be on the receiving end of some of the verbalattacks.

For some, a disaster elicits a victim–helpersyndrome. Large natural disasters such as earthquakescan completely destroy the physical fabric of a com-munity. Survivors become very dependent uponoutside aid for their daily requirements such as shelter,clothes, food, and services. Often the victims have noresources left for reconstruction. They may be suffer-ing from injury or shock after losing family members.These people become victims of the disaster. Withoutanything that is familiar, without any ability to copeor power to direct their own lives, they succumb tolethargy and establish a complete dependence uponthe rescuer. This disaster syndrome is not thatcommon, being most pronounced following traumaticcatastrophes for only a minority of people. In somecases, unless aid is directed to re-establishing a normalexistence, the victims can become permanently lockedinto a dependency syndrome. They become disasterrefugees. For example, the Ethiopian government atthe end of the Sahel drought of 1984 forcibly repatri-ated refugees to relief camps in their northern home-lands, because they had become so dependent uponemergency aid that they were hesitant to return to

what they still perceived as drought-ravaged farms. Asignificant number of evacuees of Cyclone Tracy, aftera year of living in the southern states of Australia wherethe standard of living was better, also found it difficultdeciding to return home to Darwin.

The evacuation of survivors causes other psycholog-ical problems. For instance, after Cyclone Tracy, mostresidents were initially evacuated. However, husbandswere permitted back to rebuild. Families were oftenseparated for up to a year. When finally reunited,parents had drifted apart socially, and a gulf had arisenbetween fathers and children who had not seen eachother for months. In this case, the resulting tragedy ofDarwin was not the initial death toll from the cyclone;it was the fact that over 50 per cent of families who hadexperienced the disaster separated within a decade.Nor was this phenomenon restricted to evacuees.Families in the south, who had volunteered theirhomes as temporary billets, were faced with increasedstress and – for some individuals – tempted into anextra-marital affair with younger guests. Evacuees,especially those people who were self-employed, alsoexperienced loss of income. This was not necessarily aproblem for government workers or employees ofbenevolent national companies, who were relocated tosouthern offices. Those who were not so lucky facedunemployment or found temporary work. For some,this alternative employment became so attractive thatthey were reluctant to return to Darwin and face anuncertain future. Single-parent families left in thesouth during the reconstruction also faced difficultieswith children. Because of the trauma and changed sur-roundings, younger children underwent regressivebehavior. Without a father-figure, rejected by theirpeers as odd, and considered transient by the commu-nity, school-aged children found it difficult adjusting toa strange environment and fitting into the community’ssocial activities. Many developed antisocial behavioraltraits, became delinquent, or took up drugs and alcoholto cope. The problems were especially acute foryounger children (see Table 13.1). Behavior of thesenon-returnee children significantly differed from simi-larly aged children who either had returned to Darwinor, for some reason, had not left.

Within Australia, this situation is not unique toCyclone Tracy. The drought of 1982–1983 drove manyfamilies off the land into rural towns and the capitalcities, where society was more permissive than gener-ally experienced on isolated farms. Many children

284 Social Impact

found the availability of drugs, which at the time werebeginning their dramatic spread beyond the capitalcities, irresistible. At the time of the Live Aid Concertfor Ethiopia, Molly Meldrum (the Australian organizerof that relief campaign) observed that we were moreconcerned in Australia with people’s plight in the Sahelthan we were about the unrecognized or deliberatelyignored drug problems of children driven into citiesby the Australian drought. Significant adolescentproblems still existed in country towns in 1989, sixyears after that drought, with many young peopledrifting to capital cities or resort towns and contribut-ing to delinquency rates in those places.

As time passes, the shock of a natural hazard and thetrauma of death also lead to psychiatric problems. TheAberfan, Wales, disaster, while not strictly a naturaldisaster, illustrates this phenomenon well. Failure of a250-metre high, coal tailings pile killed 116 children,the entire juvenile population, in a manner of seconds.The whole population was personally affected andunderwent psychological trauma. Over 158 residentssubsequently sought psychiatric treatment. A similar,but less severe, situation arose in communities badlyaffected by Australia’s 1983 Ash Wednesday bushfires.The number of people requiring psychiatric help to

cope with the disaster and its after-effects increasedsignificantly afterwards. The number of suicides rose inthe year following a particularly bad tornado in WichitaFalls, Kansas, in 1979.

Just as work stress over time can lead to increasedillness and absenteeism in workers, so the culminationof life changes, such as those resulting from anatural disaster, can lead to psychosomatic illness.Psychosomatic illness is any physical disease, such asulcers, depression, or arthritis that owes its origin to apsychological problem. Scales have been constructed toforeshadow the likelihood of an individual undergoing apsychosomatic illness. Death in the immediate family, ahousehold move, the birth of a child, a change in job, ora divorce all contribute equally to the chance of physicalillness in the following year. Again, non-returnees fromthe evacuation of Darwin in 1974 suffered a significantincrease in stress-related illnesses. Further study hasshown that it was not necessarily non-returnees thatsuffered more; it was the people who wanted to returnbut could not (Kearney & Britton, 1985). In this lattergroup, there was a 50 per cent increase in the numberof people who were nervous, depressed, or worriedabout the future, and a 100 per cent increase in thenumber of people complaining about bowel trouble orwho were using sedatives.

The prolonged prominence of psychosomatic illnessmay not only destabilize a family unit in the short termbut, if not resolved, have profound long-term ramifica-tions for society. Where one or more family memberssuffer from a psychological disorder, the chance ofthat disorder appearing in offspring is significantlyincreased. Natural disasters can also breed a cycle ofsociological and psychological disorder, affecting afuture generation who has never actually experiencedthe event. Ten years after Cyclone Tracy, social workersin Brisbane were able to recognize cyclone-trauma-tized victims, who had been evacuated to this city, as adistinct subgroup.

The impression should not be given that all naturaldisasters produce copious psychological problems. Anatural disaster may simply apply stress to an already-stressed person or family. The disaster may simply be‘the straw that broke the camel’s back’. The evacueesfrom Darwin who did not return, and who developedpsychological problems, may have been those peoplein Darwin already suffering problems, who would haveleft the city permanently some day even if CycloneTracy had not struck. Even if the families were stable


Stayers Returnees Non-Returnees Trait % % %fear of rain and wind 19.8 24.2 29.3

fear of dark 10.8 13.2 11.5fear of jet noise 4.5 7.9 15.5

clingingto mother 6.3 5.3 12.6

bed-wetting 2.7 6.8 7.6thumb-sucking 0.0 1.6 2.0tempertantrums 2.7 1.1 7.2

fighting 0.9 3.2 4.9deliberatevandalism 2.7 1.1 3.4

injuries 0.0 2.6 5.5diseases 9.9 5.8 10.9Source: Western and Milne, 1979.

Table 13.1 Frequency of psychological problems amongst youngsurvivors of Cyclone Tracy, Darwin, 1974.

units, the long-term isolation in unfamiliar surround-ings and crowded accommodation was an abnormalsocial pressure. The study by Luketina (1986) of the1984 Southland floods in New Zealand indicated thatpeople evacuated for periods similar to the Darwinevacuees also underwent exceptional stress. However,after life returned to normal, very few admitted to,or sought help for, psychological problems attributedto the floods. The exceptions to this occurred withthe elderly – some of whom became confused by theevents – with those already going through the stages ofmarriage breakdown, and with children. Immediatelyafter the floods, 7 per cent of children from floodedhomes were exhibiting behavioral changes of concernto parents and teachers. However, the majority of theseproblems disappeared as life normalized after severalmonths.

There is a sequence of psychological adjustmentfollowing a disaster. Foremost, the majority of disastervictims appear to be resilient and adaptive to thedisaster. For example, in the Brisbane floods of 1974the community shunned disaster relief. Fewer than2 per cent of the victims subsequently sought helpfrom a counseling service. Of course, in many of theseevents there was no large loss of life nor substantialinjury. A significant criterion for disaster-inducedpsychosis appears to be the occurrence within a shortperiod of major life changes. If the disaster generatesthese changes, or adds to ones that have alreadyoccurred, then an individual or family will be morelikely to be affected psychologically by that disaster.

Critical incidence stress syndrome

A more worrying problem with disasters is the fact thatanyone who encounters a tragedy spawning death orinjury can be affected by that tragedy for the rest oftheir lives. The experience may be as innocuous asbeing a bystander to the tragedy. For instance, duringthe Waterloo, London, subway station fire of 1987,trains coming into the station stopped and the doorsremained closed. Then, the trains pulled swiftly away,despite people on the platform desperately bangingon windows and doors trying to get in to escape theinferno. For the stunned passengers on the trains,the stop lasted only 30 seconds, but the experience willbe with them for the rest of their lives. The closer oneis to the disaster, the more permanent and damagingthe effect. Sadly, the closest people are the heroes.Whereas a hero may have been awarded society’s

everlasting gratitude and should be basking in glory,many in fact suffer from extreme depression and ekeout shattered lives. The pattern is now known asCritical Incidence Stress Syndrome, and is recognizedby rescue workers and heroes as a severe, long-termpsychological response to disaster. The pattern is char-acterized by extreme depression and unfounded guiltfeelings. For example, the policeman who repeatedlyrisked his life to save commuters in the King’s Crosssubway fire, in London in 1987, cannot ascertain whyhe has survived while others died; he now lives a timidlife, afraid to go outdoors. The passenger who becamea human bridge permitting fellow travelers to crawl tosafety after the Herald of Free Enterprise ferry sankat Zeebrugge, Belgium, in 1987, lives a life withoutdirection, even though he has been given everybravery award his country can offer. Another police-man, who dragged a burning man from the Bradfordstadium fire in 1985, finds himself so emotionallyweakened that he cannot help crying, years after theincident, over trivial matters. Even Major-GeneralStretton was affected by guilt and depression becausehe believed his evacuation orders after Cyclone Tracyhad shattered survivors’ lives. As with the victims ofmany disasters who receive psychiatric counselingwithin minutes of being rescued, professionalcounselors are now debriefing rescue workers follow-ing their rescue work. In New South Wales, Australia,these debriefing sessions take the form of teamcounseling and, depending upon the magnitude of thedisaster and the response, can continue for monthsafterwards. The policy is not to let rescue workersdiscuss the disaster clannishly in an unstructured envi-ronment such as the local pub. Instead, counseling isdirected towards relieving initial feelings of anger andguilt that may not only destroy that person’s life buttheir abilities as a rescue worker. Where death isinvolved, there is no rosy conclusion to a disaster.Anyone who has witnessed the tragedy, be they victim,bystander, rescue worker or hero, can potentially havetheir lives irreparably harmed by the experience.


Burton, I., Kates R.W., and White, G.F. 1978. The Environment asHazard. Oxford University Press, Oxford.


286 Social Impact

Dalitz, E.R. 1979. Personal reactions to natural disasters. InHeathcote, R.L. and Thom, B.G. (eds) Natural Hazards inAustralia. Australian Academy of Science, Canberra, pp. 340–351.

Drabek, T.E. 1986. Human System Responses to Disaster: AnInventory of Sociological Findings. Springer-Verlag, New York.

Hauptman, W. 1984. On the dryline. Atlantic Monthly, May: 76–87.Holthouse, H. 1986. Cyclone: A Century of Cyclonic Destruction.

Angus and Robertson, Sydney.Kearney, G.E. and Britton, N.R. 1985. Insurance response in disaster.

Report of Proceedings of Research Workshop on Human Behaviourin Disaster in Australia. Natural Disasters Organisation, MtMacedon, pp. 300–324.

Lachman, R., Tatsouka, M., and Bonk, W.J. 1961. Human behaviourduring the tsunami of May, 1960. Science 133: 1405–1409.

Luketina, F. 1986. The psychological effects of floods. Soil andWater 22(1): 20–24.

McKay, J.M. 1983. Community adoption of bushfire mitigationmeasures in the Adelaide Hills. In Healey, D.T., Jarrett, F.G.and McKay, J.M. (eds) The Economics of Bushfires: The SouthAustralian Experience. Oxford University Press, Melbourne,pp. 116–131.

Michaelis, A.R. 1985. Interdisciplinary disaster research. Report ofProceedings of Research Workshop on Human Behaviour inDisaster in Australia. Natural Disasters Organisation, MtMacedon, pp. 325–346.

Reser, J.P. 1980. The psychological reality of natural disasters. InOllier, J. (ed.) Response to Disaster. Centre for Disaster Studies,James Cook University, Townsville, pp. 29–44.

Santana, S. 2002. Remembering Andrew. Weatherwise 55 July/August: 14–20.

Sims, J.H. and Baumann, D.D. 1972. The tornado threat: copingstyles of the north and south. Science 176: 1386–1392.

Western, J.S. and Milne, G. 1979. Some social effects of a naturalhazard: Darwin residents and cyclone ‘Tracy’. In Heathcote,R.L. and Thom, B.G. (eds) Natural Hazards in Australia.Australian Academy of Science, Canberra, pp. 488–502.

Wilson, A.A.G. and Ferguson, I.S. 1985. Fight or flee? A casestudy of the Mt. Macedon bushfire Pt 2. Bushfire Bulletin8(1 & 2): 7–10.


C H A P T E R 1 4

CHANGING HAZARDREGIMES(Houghton et al., 1995, 1996; Bryant, 2001; Nott, 2003)

Throughout this book the commonness of naturalhazards has been continually emphasized. For example,the eruption of Krakatau in 1883 was not an unusualevent. It was matched by earlier eruptions of similarmagnitude and will be witnessed again. This argumentsmacks of uniformitarianism, but it is not meant tosupport this concept. Rather, it acknowledges that ourexisting realms of natural hazards fit a magnitude–frequency distribution that can be described, and fromwhich the probability of occurrence of future extremescan be predicted. This is not necessarily how naturalhazards behave over time. Existing hazard regimes arenot immutable. They can change. My present researchinto mega-tsunami illustrates this point. In the historicrecord, tsunami in Australia have not exceeded 1.07 mon tide gauges, did not have run-ups of 4 m above sealevel nor penetrate more than 1 km inland. Over thepast fifteen years, evidence has been found alongAustralia’s New South Wales coastline for largetsunami that have gone inland up to 10 km, transportedboulders the size of boxcars up 30 m high cliffs (Figure14.1) and swept over headlands 130 m high. Nor is theevidence restricted to the New South Wales coastline.In north-western Australia, one event penetrated 35 kminland in the Great Sandy Desert. These mega-tsunami

have occurred at periodic intervals over the last8000 years. They were generated by comet or meteoriteimpacts with the surrounding oceans – because noknown earthquake or volcanic eruption has producedsimilar magnitude evidence. Astronomers believe that alarge comet entered the inner solar system about 15 000 years ago. It broke up, and the Earth periodicallysweeps through its debris trail in late June or mid-November each year. This is the Taurids meteorite

Epilogue

Fig. 14.1 Imbricated boulders stacked against a 30 m high cliff faceon the south side of Gumgetters Inlet, New South Wales,Australia. Some of these boulders are the size of a boxcar.Note the person circled for scale.

stream. The Earth passes through the concentratedhead of this stream every 2520 years with anotherminor concurrence every 1100 years. The Earth’s lastencounter occurred around 1500 AD. This is when thelast mega-tsunami affected the New South Wales coast-line. Australian Aboriginal and Maori legends describea comet shower that can be dated to this time. We havealso found a recent impact crater, generated by a cometabout 1 km in diameter, off the south coast of StewartIsland, New Zealand. Currently, the Earth is notaffected by large or plentiful objects in the Taurids. Asa result, Australia – together with much of the globe– exists presently within a quiescent epoch wheretsunami are generated mainly by earthquakes orvolcanoes. In Australia, the dichotomy is that fieldevidence for mega-tsunami does not match a limitedhistorical record devoid of such events. The dichotomycan be resolved knowing that tsunami are very muchcharacteristic of a changing hazard regime.

Other climatic and geological hazard regimes havealso changed. For example, since 1700 the Earth hasexperienced more and larger volcanic eruptions than atany time during the last 10 000 years. We are alsowitnessing the termination of the Little Ice Age,which appears to follow a 1500-year climate cycle.During cold phases, storms are more frequent andintense in northern Europe (Figure 3.12), and temper-atures more extreme. For example, in the 1420s, theBurgundy region of France – during one of the periodsof the Middle Ages when temperatures were coolingrapidly – witnessed six of the coldest years in1000 years. However, embedded within this coolingwas the warmest summer experienced for a millen-nium. This summer was not just slightly above normal,but exceptionally so. The grape harvest in that yearoccurred at the beginning of August instead of duringthe second or third week of September.

Finally, the extreme 1 in 100 year rainfalls that manyregions of the world have been experiencing maysimply represent the occurrence in time of part of amuch more intense rainfall regime. Wollongong,Australia, mentioned in this text, is a region whereextreme rainfalls of 200–400 mm in 24 hours arecommon. Embedded in rainfall records for this regionare rainfalls of 840 mm in 9 hours, and 1000 mm withinone month. These rainfalls do not plot on the straightline in probability of exceedence diagrams. Theyrepresent a different climate regime conducive toextreme floods. This aspect is not unique to Wollongong.

Across Australia, there is geomorphic evidence forfloods that are greater than any that could be producedby maximum probable rainfalls. Nor is Australiaunique in the world. The increase in flooding globally(Figure 6.10) is evidence that rainfall (or drought)regimes are changeable.

The effects of global warming, whether caused bythe termination of the Little Ice Age or by enhancedGreenhouse gas concentrations, can also be accommo-dated within the concept of a changing hazard regime.Climatologists acknowledge that the Earth’s climate isnot consistent over time. At the height of the LastIce Age 22 000 years ago, temperatures fluctuateddramatically, rising and falling 4–6°C over the space ofa few years. These ‘flickerings’ are probably subduedunder our present climate regime, but they appearwhen climate gets colder. The fluctuations in the 1420sin France were probably evidence of this fundamentalfeature of Pleistocene climate.

The warming that is occurring presently and ishypothesized to increase represents a change in hazardregime. The potential effect of changes in carbondioxide induced by human activity has been the subjectof scientific enquiry for the past 155 years. In 1861,John Tyndall suggested that carbon dioxide changesin the atmosphere might be responsible for changes inclimate. In 1896, Svante Arrhenius proposed that adoubling of carbon dioxide might warm the atmosphereby 5°C. In 1938, Callender showed how carbon dioxideand water vapor absorb long wave radiation at differentwavelengths. He suggested that anthropogenic produc-tion of carbon dioxide would alter the natural balancebetween incoming solar radiation and outgoing longwave radiation, and cause global warming that wouldexceed natural variations in historical times. Theseminal conference that catalyzed world attention onthe significant ramifications of enhanced ‘Greenhouse’gas concentrations in the atmosphere was the WMO-sponsored conference of 9–15 October 1985 in Villach,Austria. A policy statement released by this conferencepointed out that most current planning and policydecisions assumed a constant climate – an immutablehazard regime – when, in reality, increases in ‘Green-house’ gases by the 2030s would warm the globebetween 1.5 and 4.5°C, leading to a sea level rise of20–140 cm. The only thing that has changed since thenis the realization that numerous gases – CO2, CH4,N2O, CO, O3 and CFCs – are implicated in thesepredictions. While present temperatures have been

289Epilogue

matched during the Mediaeval Warm Period (centeredon the thirteenth century), there is no match in thegeological record to the current rapid increase inGreenhouse gases. Thus, the climatic consequencesof increased Greenhouse gases are speculative anduncertain.

MODERN CONSEQUENCES OFNATURAL HAZARDS

This book has been concerned with natural hazardsthat are within the present realm of possibility: thatwe have lived with, will continue to experience, andhopefully can survive. The basic purpose of the text isto present the reader with a clear description ofthe mechanisms producing climatic and geologicalhazards. An attempt has been made to enliven thatpresentation with historical descriptions and support-ive diagrams. There are four themes that are indirectlyaddressed throughout the text. It is worthwhile herebriefly summarizing these.

Firstly, natural hazards are predictable. This isespecially applicable to climatic hazards. With satellitemonitoring, short-range computer forecasting modelsand the wealth of experience of trained personnel andscientists, most climatically hazardous events can beforecast early enough to give the general populationtime to respond to any threat. Whether or not thesewarnings are believed is most relevant in today’s world.The fact that the British Meteorological Office ignoredan extra-tropical storm warning issued by theircounterparts in France; that residents in Darwin weremore interested in partying for Christmas whenCyclone Tracy was barreling down on them; or that theduty officers in Sydney ignored their own amateurstorm-watch observers as the worst hailstorm onrecord overtook the city, does not negate the fact thatmost climatically hazardous events are being pre-dicted. It only illustrates that a society, its organizedbureaucracy, or its individuals have a major influencein determining how effectively warnings of, andpreparations for, a natural disaster are accepted.

Also pointed out in this text is the fact that thedegree of prediction is strongly influenced by regularastronomical cycles, such as the 11- and 22-yearsunspot and the 18.6-year lunar cycles. Sunspotnumbers have tended to obscure the more significantprognostic variable, namely solar geomagnetic activity.These avenues of forecasting have been strongly

criticized. Yet, frequencies equating with the returnperiod of sunspots appear in enough climatic andgeophysical data that solar geomagnetic cycles cannotbe ignored as a precursor for natural hazard events.Similarly, the correlation of rainfall and drought cycleswith the 18.6-year lunar cycle should not be ignored.The coincidence of drought, across most of the grainproducing regions of the world, with the MN lunartidal component now demands that this cycle be takenseriously.

Secondly, the occurrence of natural hazards isincreasing in frequency: if not in terms of the numberof events, then in terms of economic cost. The MunichReinsurance Company, especially since 1960, can sub-stantiate this trend through payout claims. This point iswell-illustrated for volcanic activity this century. Afterthe Caribbean eruptions of 1902, volcanism leading tolarge death tolls virtually ceased for half-a-century. Asimilar quiescence is evident in climatic records. The1930s and 1940s were exceedingly passive as far asstorm events were concerned. Air temperaturesglobally tended to be warm, and weather patterns didnot fluctuate strongly. This does not imply that therewere no climatic disasters. After all, this periodwitnessed the dust bowl years on the Great Plains ofthe United States, and Hitler’s armies experienced aRussian winter that was only matched in severity bythat which afflicted Napoleon in 1812. Beginningabout 1948, climate globally began to become moreextreme. This variability has increased substantiallysince 1970. For example, since 1980 Australia has with-stood two of the largest regional flood events on record– sandwiched between three of the largest 1 in200 year droughts recorded. In the United Kingdom,in the late 1980s, temperatures seasonally tended tooscillate from the coldest to the warmest on record.This enhanced variability is prominent in both climaticand geological phenomena. In terms of natural hazardevents, we presently live in a period that may beexperiencing a shift in hazard regime.

Thirdly, natural hazards are pervasive in time andspace. A simple plot of the location of the hazardsmentioned in this book will illustrate this fact.Normally, we do not expect droughts to strike England,tornadoes to sweep through Edmonton, Canada, orsnow to fall in Los Angeles. However, such events havehappened in the latter half of the twentieth century.Part of the reason so-called unusual events are occur-ring in places where they are not expected may lie in

290 Epilogue

the fact that we have lived through a passive period forhazards, as mentioned above. Another explanation isthe fact that, as we develop longer geophysical records,extreme events are more likely to appear in thoserecords. Additionally, technological advances in com-munications have dramatically shrunk our spatialawareness to the point that today’s unusual weatheror geological event is everyone’s experience. Suchdevelopments have heightened people’s awareness ofnatural hazards. I believe that the long-term prognosis,at any location, is for the occurrence of a wide range ofhazards that have not necessarily been detected in thehistorical record. We are experiencing a shift in regimefor a number of hazards. It is not unusual for anearthquake of 7.2 on the Richter scale to occur in thesupposedly aseismic area of Tennant Creek, Australia(September 1988), nor for tornadoes to sweep throughNew York City (July 1989). These extreme events,while not common in these areas, are certainly withinthe realm of possibility. There is a pressing need for theserious assessment of extreme hazards in many parts ofthe world. For example, volcanic activity is possible inAustralia even though there are no active volcanoes,and tropical cyclones could affect an area like southernCalifornia. In the former case, volcanoes have eruptedin Australia in the past 10 000 years. In the latter case,southern California lies close to an area generatingtropical cyclones. Natural hazards, and in some casesvery unexpected hazards, are simply more commonthan perceived.

Finally, the pervasiveness of so many hazards, plusones that we have not discussed, should not be viewedpessimistically. When a natural disaster occurs, and issplattered across our television screens, we tend tobelieve that few survive. In fact, it is surprising that sofew people die or are injured. There have been only afew disasters in the past 150 years that have wiped outmost of a population. Mt Pelée in 1902 left two

survivors out of 30 000; the Mt Huascarán debris flowof May 1970 killed most of the population of Yungay,Peru, while the November 1985 eruption of theNevado del Ruiz volcano in Colombia killed most ofthe people in the path of its subsequent lahar. Considerthe odds of survival for other large natural disastersmentioned in this text: Cyclone Tracy killed 64 peopleout of a population of 25 000 in Darwin on25 December 1974; only one out of every four peopledied in the horrific Tangshan earthquake in China in1976; the death toll in the Tokyo earthquake of 1923was 140 000 people out of a population well in excessof 1 million. Human beings are very resilient and apt atsurviving and, in particular, are gifted at coping withdisasters. These aspects were emphasized in Chapter13. Any text on natural hazards tends to emphasize, asdoes this one, the sensational, negative, or gruesomeaspects of disasters. However, it is only appropriate toend this book by noting people’s remarkable abilityduring calamities to rescue, survive, and recover fromsuch events.


Bryant, E. 2001. Tsunami: The Underrated Hazard. CambridgeUniversity Press, Cambridge.

Houghton, J.T., Meira Filho, L.G., Bruce, J., Hoesung Lee,Callander, B.A., Haites, E., Harris, N. and Maskell, K. (eds)1995. Climate Change 1994: Radiative Forcing of ClimateChange and an Evaluation of the IPCC IS92 Emission Scenarios.Cambridge University Press, Cambridge.

Houghton, J.T., Meira Filho, L.G., Callander, B.A., Harris, N.,Kattenberg, A. and Maskell, K. (eds) 1996. Climate Change1995: The Science of Climate Change. Cambridge UniversityPress, New York.

Nott, J. 2003. The importance of prehistoric data and variability ofhazard regimes in natural hazard risk assessment: example fromAustralia. Natural Hazards 30: 43–58.

291Epilogue

AAdiabatic Lapse Rate. The rate at which air temperature

decreases with elevation because of the decreasingeffect of gravity. The result is an increase in air volumewithout any change in the total amount of heat. Thenormal adiabatic lapse rate is –1 °C for each 100 metresof elevation: but the saturated adiabatic lapse rate,when air cools more slowly, is 0.4–0.9°C per100 metres. If a parcel of air is moved towards theground the reverse process occurs and the air begins towarm.

Agglutinate. The tendency for high-pressure cellsoriginating from the poles to stagnate and stack up at aparticular location over mid-latitude oceans. TheBermuda or Azores High represents one of theselocations at which stagnation tends to occur. SeeMobile Polar High.

Albedo. The degree to which short wave solar radiation isreflected from any surface or object.

Andesite. Volcanic rock containing the mineralsandesine, pyroxene, hornblende, or biotite. In a meltedstate, such material contains a high amount ofdissolved gas.

Aperiodically. Recurring but not with a fixed regularity.For instance, the first day of each month occursaperiodically because there are not the same numberof days in every month.

Aseismic. Characterizing a region that historically has notexperienced earthquakes.

Atterberg Limits. These limits define the moisturecontent of a soil at the points where it becomes elastic,plastic, and liquid.

BBarrier Island. A sandy island formed along a coastline

where relative sea level is rising or has risen sub-stantially in the past. Shoreward movement ofsediment cannot shift the island landward fast enough,so the island becomes separated from the mainland bya lagoon or marsh.

Basaltic. Referring to the most common liquid rockproduced by a volcano, containing mainly magnesium,iron, calcium, aluminum, and some silicon. It forms ahard, but easily weathered, fine-grained, dark-grey rock.Volcanoes producing magma with these constituents areclassified as basaltic.

Base Surge. The sudden eruption of a volcano laterallyrather than vertically.

Bathymetry. Depths below sea level.Bearing Strength. The amount of weight that a soil can

support, including its own weight, before the soildeforms irrevocably.

Benioff Zone. The zone at a depth of 700 km associatedwith earthquakes along a subduction zone.

Biomass. The total weight of all living organisms.Bistable Phasing. Over two to three centuries in many

locations, droughts and wet periods tend to alternateevery 9.3 years. In some cases, a drought is suddenly

Select Glossary of Terms

followed, nine years, later by another drought beforethe alternating cycle is re-established. This is bistablephasing.

Blizzard. Any wind event with velocities exceeding60 km hr-1, with temperatures below –6 °C, and oftenwith snow being blown within tens of metres of groundlevel.

Blocking High. A high-pressure system whose normaleastward movement stalls, blocking the passage of sub-sequent low- or high-pressure systems and deflectingthem towards the pole or equator. Mobile polar highsare high-pressure cells that have either run out ofmomentum in their movement away from the poles orhave not been displaced by another high.

Blocks. A type of lava that consists of angular, boulder-sized pieces of solid rock blown out of an eruptingvolcano.

Bombs. (1) See East-Coast Lows. (2) Boulder-sizedpieces of liquid lava blown out of an erupting volcano.

Boreal. (1) Pertaining to the northern hemisphere. (2)Northern hemisphere forest, consisting mainly ofconifers and birch.

Bruun Rule. A relationship stating that a sandy beachretreats under rising sea level by moving sand fromthe shoreline and spreading it offshore to the samethickness that sea level has risen.

CCaldera. The round depression formed at the summit of a

volcano caused by the collapse of the underlying magmachamber that has been emptied by an eruption.

Capillary Cohesion. The binding together of soilparticles by the tension that exists between watermolecules in a thin film around particles.

Catastrophists. A group of people, little challengedbefore 1760, who believed that the Earth’s surface wasshaped by cataclysmic events that were acts of God.

Cation. An ion or group of ions having a positive charge.Cavitation. A process, highly corrosive to solids, whereby

water velocities over a rigid surface are so high that thevapor pressure of water is exceeded and bubbles beginto form.

Cementation. The process whereby the spaces or voidsbetween particles of sediment become filled withchemical precipitates such as iron oxides and calciumcarbonate, which effectively weld the particlestogether into rock.

Chaparral. The name given to dense, often thorny,thickets of low brushy vegetation, frequently including

small evergreen oaks This vegetation is found in theMediterranean climate of the United States south-west.

Chernozem. A very black soil rich in humus and carbon-ates forming under cool, temperate, semi-arid climates.

Clustering. The tendency for events of a similarmagnitude to be located together in space or time.

Coherence. (1) Of soils: the binding together of soilparticles due to chemical bonding, cementation ortension in thin films of water adhering to the surfaceof particles. (2) Of time series: the tendency for aphenomenon to be mirrored at another location overtime. For instance, wet periods in eastern Australiaoften occur simultaneously with a heavy monsoon inIndia.

Cohesion. The capacity of clay soil particles to stick oradhere together because of physical or chemical forcesindependent of inter-particle friction.

Colloids. Fine clay-sized particles 1–10 microns in size,usually formed in fluid suspensions.

Conflagration. A particularly intense fire with a heatoutput of 60 000–250 000 kW m-1 along the fire front.

Continental Shelf. That part of a continental platesubmerged under the ocean, tapering off seaward at anaverage slope of less than 0.1 degrees, and terminatingat a depth of 100–150 m in a steep drop to the oceanbottom.

Convection. The transfer of heat upward in a fluid or airmass by the movement of heated particles of air orwater.

Convective Instability. A parcel of air continuing to risewhen lifted if its temperature is always greater thanthe surrounding air – a process that is guaranteed ifcondensation occurs releasing heat into the parcel.

Coriolis Force. The apparent force caused by the Earth’srotation, through which a moving body on theEarth’s surface deflects to the right in the northernhemisphere and to the left in the southern hemi-sphere.

Correlation. The statistical measure of the strength of arelationship between two variables.

Cracking Clays. Soils consisting of clays that have a highability to expand when wetted, and to shrink and,hence, crack when dried.

Creep. The imperceptibly slow movement of surface soilmaterial downslope under the effect of gravity. See alsoTectonic Creep.

Cretaceous. The geological time period 135–65 millionyears BP, which was characterized by sudden, largeextinctions of many plant and animal species.

293Glossary

Critical Incidence Stress Syndrome. The conditionaffecting rescuers who perform superhuman, selflessfeats of heroism or who are involved with death duringa disaster. Such people often suffer severe guilt,become reclusive, and generally broken in spirit formonths or years after the event.

Crust. The outermost shell of the Earth, 20–70 km thickunder continents, but only 5 km thick under theoceans. It consists of lighter, silica-rich rocks, whichflow on denser liquid rock making up the mantle.

Cryogenic. Any process involving the freezing of waterwithin soil or rock.

Cumulonimbus. A dense cloud consisting of a lower darklayer that is potentially rain-bearing, and a multiplewhite fluffy crown reaching the top of the troposphere.Such cloud forms by thermal convection and ischaracteristic of thunderstorms.

Cycles, cyclic or cyclicity. Recurring at regular intervalsover time: for instance, sunspot numbers, which recurat 10–11 year intervals.

Cyclogenesis. The process forming an extra-tropical,mid-latitude cell or depression of low air pressure.

DDebris Avalanche. A flow of debris of different particle

sizes that takes on catastrophic proportions. Volumesof 50–100 million cubic metres traveling at velocities of300–400 km hr-1 have been measured.

Debris Flow. Any flow of sediment moving at any velocitydownslope in which a range of particle sizes is involved.

Deflation. The removal by wind, over a long time period,of fine soil particles from the ground surface.

Degassing. The sudden or slow release into theatmosphere of gas dissolved in melted rock.

Degradation. The slow loss from a soil of humus,minerals, and fine particles essential to plant growth; orthe contamination of a soil by salts or heavy mineralsthat are toxic to plant growth.

Desertification. The process whereby semi-arid landcapable of supporting enough plant life for grazingor extensive agriculture becomes infertile because ofincreased aridity, degradation, or overuse.

Dilation. Rock in the Earth’s crust will tend to crack andexpand in volume when subjected to forces capable offracturing it.

Dust Bowl. Part of the American southern Great Plains,centered on Oklahoma, which experienced extensivedust storms during the drought years of 1935–1936 atthe height of the Great Depression.

Dust Veil Index. A subjectively defined index thatattempts to measure the relative amounts of dustinjected into the stratosphere by volcanic eruptions.The index is reference to a value of 1000 for theeruption of Krakatau in 1883.

EEast-Coast Lows. Very intense storms that develop in mid-

latitudes on the seaward, eastern sides of continents.Such lows develop over warm ocean water in the lee ofmountains and are dominated by intense convection,storm waves, and heavy rainfall. Events where windvelocities reach in excess of 100 km hr-1 within a fewhours are called ‘bombs’.

Easterly Wave. In the tropics, pressure tends to increaseparallel to the equator with easterly winds blowing instraight lines over large distances. However, suchairflow tends to wobble, naturally or because of someoutside forcing. As a result, isobars become wavy andthe easterly winds develop a slight degree of rotationthat can eventually intensify over warm bodies of waterinto tropical cyclones.

El Niño. The name given to the summer warming ofwaters off the coastline of Peru at Christmas time(from Spanish, meaning ‘The Child’).

Elastic. A property of any solid that deforms under a loadand then tends to return to its original shape after theload is released.

Electric Resistivity. The degree to which a solid body canconduct an electric current.

Electrosphere. The zone of positively charged air at50 km elevation, the strength of which determines thefrequency of thunderstorms.

Electrostatic. The generation of electrical charges onstationary particles.

Endemic. A term used to describe a disease that existsnaturally within a specific region because of itsadaptation to climate and the presence of vectors andreservoirs. If the vectors and reservoir species returnor controls are relaxed, then the disease can easilybecome re-established. See also Vectors, Reservoirs.

Energy. The capacity of any system to do work. Energycan be described as ‘kinetic’, which is the energy anysystem possesses by virtue of its motion; as ‘potential’,which is the energy any system possesses because of itsposition; and ‘thermal’, which is the energy any systempossesses because of its heat.

ENSO. Abbreviation for the words ‘El Niño–SouthernOscillation’, referring to times when waters off the

294 Glossary

coast of Peru become abnormally warm and theeasterly trade winds blowing in the tropics acrossthe Pacific Ocean weaken or reverse.

Epicenter. The point at the surface immediately above thelocation of a sudden movement in the Earth’s crustgenerating an earthquake.

Episodic. The tendency for natural events to group intotime periods characterized by common or looselyconnected features. Weak cyclicity or regularity maybe implied in this term.

Equinox. The time occurring twice yearly, around21 March and 21 September, when the apparentmotion of the sun crosses the Earth’s equator. Thisresults in all parts of the Earth’s surface receiving equalamounts of daylight.

Eucalyptus. A genus of tree, which is mainly evergreen, isrich in leaf oil, flammable, drought-resistant, andcapable of growing in nutrient-deficient soil.

Eustatic. Worldwide sea level change.Evaporite. A sediment deposit that has formed from the

residue left after the evaporation of salt water.Evapotranspiration. The amount of moisture lost by

plants by evaporation to the atmosphere, mainlythrough leaf openings or stomata. This amount is alwaysgreater than that lost from equivalent, unvegetatedsurfaces.

Expansive Soils. Soils consisting of clay particles – mainlymontmorillonite – that absorb water when wetted andthus expand in volume.

Extra-Tropical Low or Depression. A mid-latitude, low-pressure cell with inwardly and upwardly spiralingwinds. Unlike tropical cyclones, the cell can developover land as well as water, usually in relation to thepolar front and with a core of cold air.

Eye. The center of a tropical cyclone (or tornado), whereair is descending, surrounded by a wall structure createdby a sharp decline in pressure and characterized byintensely inwardly spiraling winds.

FFault. A planar zone where one part of the Earth’s crust

moves relative to another. A fault is categorizedaccording to the direction of movement in the hori-zontal or vertical plane (see Figure 9.8). Transcurrentfaults are fault lines approximately at right angles tomajor lines of spreading crust in the oceans.

Feedback. The situation where one process reinforces(positive) or cancels (negative) the effect of another. Forinstance, a microphone placed close to an amplifier

catches its own sound waves and rebroadcasts themlouder. This is positive feedback.

Fetch. The length of water over which wind blows togenerate waves. The longer the fetch, the bigger thewave.

Firestorm. Under intense burning, air can undergo rapiduplift and draw in surrounding air along the ground toreplace it. This movement can generate destructivewinds at ground level in excess of 100 km hr-1.

Flood. A large discharge of water flowing down – if notoutside – any watercourse in a relatively short periodcompared to normal flows.

Flood Lava. Lava flows occurring over a very large area.Floodplain. Flattish land adjacent to any watercourse

built up by sediment deposition from water flowingoutside the channel during a flood.

Föhn Wind. Air rising over a mountain and undergoingcondensation releases latent heat of evaporation and,thus, cools adiabatically at a slow rate of about 0.5°Cper 100 m. When this air then descends downslope onthe leeward side, it tends to warm at the faster, dryadiabatic lapse rate of 1°C per 100 m. The resultingwind is much warmer and drier than on the windwardside.

Foreshocks. The small seismic shocks preceding a majorearthquake or volcanic event.

Freezing Rain. Light rain falling through a layer of coldair or upon objects with a temperature below freezingwill tend to freeze or adhere to those objects. Overtime the accumulated weight breaks power lines andtree branches.

Friction. The force due to the resistance of one particlesliding over another because of a weight pressingirregularities on the surface of both particles together.

Fumarole. A hole or vent from which volcanic fumes orvapors issue.

GGeneral Air Circulation. Equatorial regions have an

excess of heat energy relative to the poles becauseincoming solar radiation exceeds outgoing long waveradiation. To maintain a heat balance over the Earth’ssurface, this excess heat flows mainly in air currentsfrom the equator to the pole.

Geochemical. Involving chemical reactions on, or inside,the Earth.

Geo-Electric Activity. Processes involving the transmissionor induction of electrical currents along the surface of,or through, the Earth.

295Glossary

Geomagnetic Activity. Fluctuations in the Earth’smagnetic field over space and time. The strength ofthis field and its variation can be changed by solaractivity.

Geophysical. Dealing with the structure, composition,and development of the Earth, including the atmos-phere and oceans.

Geosyncline. An area of sediment accumulation so greatthat the Earth’s crust is deformed downward.

Gilgai. The tendency for soil, affected by alternatingwetting and drying, to overturn near the surface. Theprocess can result in local relief up to 2 m.

Glacier Burst. The sudden release of huge volumes ofwater melted by volcanism under a glacier and held inplace by the weight of ice until the glacier eventuallyfloats.

Graupel. Aggregated ice or hail coated in supercooledwater.

Greenhouse Effect. The process whereby certain gases inthe atmosphere, such as water vapor, carbon dioxideand methane, transmit incoming short wave solarradiation, but absorb outgoing long wave radiation,thus raising surface air temperature.

Gumbel Distribution. The plot of the magnitude of anevent on a logarithmic scale against the average lengthof time it takes for that event to occur again.

Gyres. Large rotating cells of ocean water that often com-pletely occupy an ocean basin. The Gulf Stream formsthe western arm of one such gyre in the north Atlantic.

HHaboob. The sand or dust storm resulting from a Har-

mattan wind. The term is used in Arabia, North Africaand India. See Harmattan.

Hadley Cell. Heated air at the equator rises and movespoleward in the upper atmosphere. It cools throughthe loss of long wave radiation and begins to sink atabout 20–30° north and south of the equator, where-upon part of the air returns to the equator. This formsa large, semi-permanent vertical cell on each side ofthe equator. Postulated by George Hadley in 1735.

Hale Sunspot Cycle. Sunspot numbers peak every10–11 years. Every alternate peak is larger, forminga 22-year cycle, a phenomenon first noted by theAmerican astronomer George Hale. The Hale cycle is,in fact, the full magnetic sunspot cycle.

Harmattan. A dry, dusty trade wind caused by mobilepolar highs moving from the east over North Africa.These are the source of dust storms.

Hectare. Standard international unit measuring area.1 hectare = 1000 m2 = 2.471 acres.

Hectopascal. Standard international unit, abbreviated ashPa, measuring air pressure: 1 hPa = 1 millibar.

Hindcasting. The procedure whereby wave characteristicscan be calculated from the wind direction, strength andduration derived from weather charts.

Holocene. The most recent part of the Quaternary,beginning 15 000 years BP, when the last majorglaciation terminated.

Howling Terrors. An Australian term for small, destructive,tropical cyclones with an eye diameter of less than 20 km– also called ‘kooinar’ by Aborigines.

Hydrostatic. Describing pressure in groundwater exertedby the weight of water at levels higher than a particularlocation on a slope.

Hydrothermal. Describes the processes associated withheated, or hot melted, rock rich in water.

IIce-Dammed Lakes. Lakes formed by glaciers or icesheets

completely infilling a drainage course or flowing againsthigher topography, thus blocking the natural escape ofmeltwater.

Ignimbrites. Very hot ash can be blasted laterally acrossthe landscape. When deposited, the ash particles fusetogether to produce a hard, welded rock.

Illite. A clay mineral slightly more weathered thanmontmorillonite, containing hydroxyl radicals andhaving less iron and silica. Chemical formula[KAl2(OH)2(AlSi3(O,OH)10)].

Interfluves. The hill slopes between streams that flow inthe same direction.

Intertropical Convergence. Air in the tropics risesbecause of heating beneath the seasonal position ofthe sun. Winds flow from the north and south, andconverge on this zone of heating, effectively producinga barrier to the exchange of air between hemispheres.

Intra-Plate. Locations lying near the center of a plate or,at least, away from tectonic activity associated with themargins.

Inverted Barometer Effect. The height of sea levelinversely relates to the pressure of the atmosphereabove, such that a decrease of 1 hectopascal (hPa) inair pressure results in a rise in sea level locally of 1 cm.

Island Arc. A series of islands located on the continentalplate-side of a deep ocean trench, under which gas-rich crust is being subducted to produce andesiticmagma.

296 Glossary

Isobars. Lines plotted on a map joining points of equalpressure, usually at a fixed spacing of 4 hectopascals(hPa).

Isostatic. Describing local changes in sea level induced byregional tectonic variation.

Isothermal. Having equal degrees of heat.

JJet Stream. A fast-moving flow of air at the top of the

troposphere, usually linked to large-scale uplift orsubsidence of air. The two most common jet streamsare the polar jet stream, attached to uplift along thepolar front, and the subtropical jet centered overthe poleward and trailing side of Hadley cells.

Jökulhlaups. See Glacier Burst.Jurassic. The geological period 180–135 million years BP,

dominated by the dinosaurs.

KKaolinite. A common, clay mineral, which is resistant to

further weathering, containing only metals of aluminumand silica. Chemical formula [Al2O3.2SiO2.2H2O].

Karst. The process in which solution is a major mechanismof erosion, or landscape developed by that process.Karst most commonly occurs in limestone, but in someinstances can involve sandstone and volcanic rocks.

Katabatic Winds. Winds formed by the flow undergravity down topographic lines of denser air cooledby long-wave emission under clear skies and calmconditions. The resultant wind can reach speeds inexcess of 100 km hr-1 and is prevalent in winter,mountainous terrain or adjacent to icecaps.

Kelvin Wave. A water wave trapped along the boundaryof a coast, channel, embayment, or two bodies of waterdiffering in some physical characteristic. The wavemoves slowly parallel to the boundary and has a heightthat decreases rapidly away from the boundary. Themovement of water across the Pacific Ocean during anENSO event occurs as a 500–1000 km long, 10–50 cmhigh Kelvin wave, trapped within 5° of the equator andmoving eastward at a speed of 1–3 m s-1.

LLa Niña. The opposite of an ENSO event, during which

waters in the west Pacific are warmer than normal;trade winds or Walker circulation is stronger; andconsequently rainfalls heavier in South-East Asia.

Lahar. A mudslide induced by volcanic eruption either atthe time of the eruption – by the mixing of hot gases,

melted ice or water, and ash – or years later by the failureof volcanic ash deposits in the presence of heavy rain.

Landslide. The general term given to movement ofmaterial downslope in a mass.

Lapilli. Volcanic fragments about 2–60 mm in diameter.Latent Heat. (1) of evaporation: each kilogram of water

converted to water vapor at 26°C requires 2425 kilo-joules of heat energy. Because this heat energy is notfelt, it is termed ‘latent’. (2) Of condensation: if this airis cooled and moisture condenses, the latent heatenergy is released back to the atmosphere, warmingthe air.

Lava. Molten rock. Different terms are used to describethe nature of the lava, mainly as determined by viscosity.

Levee. When a river overflows its banks, there is animmediate decrease in velocity. This results in deposi-tion of suspended mud, and eventually the build-up ofan embankment that can contain the river above theelevation of its adjacent floodplain.

Liquefaction. If the weight of a sediment deposit or flowis not supported by the touching of individual grainsagainst each other, but by the pressure of water orair entrapped in voids between particles, then suchsediment will liquefy – that is, behave as a fluid.When liquefaction occurs, the sediment loses itsability to support objects and is able to travel at highvelocities.

Lithology. Refers to the physical composition of any rock.Little Ice Age. The period between 1650 and 1850 AD,

when sunspot activity was minimal, global tempera-tures were 1°C cooler, and temperate mountainglaciers advanced.

Loess. Dust picked up from barren ground in front oficecaps by dense, cold winds blowing off the icecapand deposited hundreds of kilometers downwind.

Longitudinal Or Meridional. Any movement that occursin a north–south direction.

Longshore Drift. The movement of water and sedimentwithin surf zones due to waves approaching shore at anangle.

Lunar Cycle. See Mn Lunar Tide.

MMagma. Liquid rock derived from the mantle.Magnetic Reversal. The Earth has a magnetic north–

south polarity centered within 1000 km of its presentaxis of rotation. The intensity of this magnetic field canvary and, at intervals of hundreds of thousands of years,weaken and reverse polarity very rapidly.

297Glossary

Magnitude–Frequency. The intensity or magnitude of allgeophysical events such as floods occur at discrete timeintervals in such a way that the lower the magnitude ofthe event, the more frequent its occurrence. Themagnitude of an event can be plotted against how oftenthat event occurs, to produce a magnitude–frequencydiagram.

Mantle. That part of the Earth from the crust to depths of3000 km, which consists of molten, dense rock made upof silicates of magnesium, iron, calcium, and aluminum.

Marigrams. Records of tsunami wave height on tidegauges.

Mass Extinctions. The dying-out or disappearance oflarge numbers of plant and animal species in shortperiods of time, supposedly because of a catastrophicevent that affected the climate of the Earth in theprehistoric past.

Maunder Minimum. Refers to the lack of sunspots between1650 and 1700 AD at the height of the Little Ice Age.

Maximum Probable Rainfalls. The amount of rain thatcan theoretically fall within a given time interval overa catchment under the most favorable conditions ofhumidity, convection, and condensation.

Meandering. The process whereby a flow of air or watertends to become unstable and travel in a winding path.

Mediterranean Climate. A seasonally dry climate havingrainfall in winter months. Its average temperature forat least four months exceeds 10°C, with no monthlyaverage falling below –3°C. Rainfall of the driestsummer month is less than 40 mm and less than one-third that of the wettest winter month.

Meridional Air Flow. Movement of air in a north–southdirection.

Microseisms. Very small earthquakes that can bedetected only by sensitive instruments.

Mn Lunar Tide. A periodicity of 18.6 years that exists inthe orbit of the moon about the Earth because of theorientation of the moon’s orbit to the equatorial planeof the sun.

Mobile Polar High. Pools of cold, shallow air that moveout from particular locations around the poles into the mid-latitudes. They tend to lose momentum such thatsuccessive highs stack up or agglutinate in a singlelocation over the oceans. When their pressure isaveraged over time, these locations appear as areas ofconstant high pressure, which have been labelledHadley cells.

Mogi Doughnut. In some seismic areas, the epicenter ofa large earthquake is ringed in the prior decade by

moderate earthquakes forming a doughnut-likepattern. Named after its discoverer.

Mohr–Coulomb Equation. An equation in which theshear strength of a soil is related to soil cohesion, aforce applied to the soil at right angles to the slope, andthe angle at which particles move relative to each otherwithin the soil.

Monsoon. Defines any region characterized by a distinct180° change in wind direction between summer andwinter, resulting in seasonal alternation of copious rainand aridity.

Montmorillonite. A group of clay minerals with analuminum layer sandwiched between two silica layersbonded with magnesium or calcium cations. Such soilsexpand easily because they can absorb water into thelayered structure when wetted. Chemical formula[(Mg,Ca)O.Al2O35SiO2.nH2O].

Mulching. Any process that causes soils not to compact,but to become more friable towards the surface.

NNormal Stress. Any force applied at right angles to a soil

surface (see Figure 12.3).Normalize. A statistical process whereby the magnitude

of values in a time series are made independent oftheir unit of measurement. This is usually done by sub-tracting the mean of the data set from each value, anddividing the result by the standard deviation. In mostclimate time series, such as the Southern Oscillationindex, the resulting value is multiplied by 10.

North Atlantic Oscillation. The tendency for air pressureto oscillate in intensity between the Icelandic Low andthe Azores High. The oscillation reflects inter-annualvariability in the strength of Rossby waves in thenorthern hemisphere. See also Southern Oscillation.

North Pacific Oscillation. A statistical measure of thestrength and position of the Aleutian low-pressuresystem, mainly in winter. The oscillation is linked toshort-term climate change in North America. See alsoSouthern Oscillation.

Nuée Ardente. A cloud of hot ash that, blown upwardsduring an eruption, collapses under the weight ofgravity and begins to flow downslope at great velocitysuspended by the gases expelled in the eruption(see Figure 11.3). Also called pyroclastic flow.

OOrographic. Any aspect of physical geography dealing

with mountains.

298 Glossary

PParent Cloud. A large cloud characterized by turbulent

airflow, and subdividing into separate areas ofconvection that can give rise to tornadoes, intenserainfall, and hail.

Partial Pressure. The contribution made by a gas to thetotal pressure of the atmosphere.

Permafrost. Permanently frozen ground.Permeability. The degree to which water can flow through

the voids in sediment deposits or micro-fractures inrock.

Persistence. The tendency for an abnormal climaticsituation to continue over time.

Photosynthesis. The process whereby plants convertcarbon dioxide and water in the presence of sunlight,and aided by chlorophyll, into carbohydrates or biomass.

Piezoelectric. The ability of a mineral to generatean electric current when pressure or stress is appliedto it.

Piezomagnetism. The generation of magnetism in iron-rich magma by increases in pressure or stress.

Plastic. A property of any solid that permanentlydeforms without breaking up or rupturing whensubject to a load.

Plate. The Earth’s crust is subdivided into six or sevenlarge segments or plates that move independently ofeach other at rates of 4–10 cm per year over the under-lying mantle. These plates can collide with (converge),or separate from (diverge) each other.

Pleistocene. The geological age in the Quaternarybeginning approximately 2–3 million years BP. It wascharacterized by periods of worldwide glaciation.

Pliocene. The last period in the Tertiary, 12–2 millionyears BP. The Pliocene terminated when large-scaleglaciation began to dominate the Earth.

Pluviometers. Instruments used to measure the rate atwhich rain falls over very short time spans.

Polar Front. Dense air, cooled by long wave emissionof heat in the polar regions and flowing towardsthe equator, differs substantially in temperatureand pressure from the air it displaces. This differ-ence occurs over a short distance termed the ‘polarfront’.

Pore Water Pressure. The pressure that builds upbecause of gravity in water filling the spaces or voidsbetween sediment particles.

Power. The rate at which energy is expended.Precursors. Variables that give an indication of the

occurrence of some future event.

Prescribed Burning. The deliberate and controlledburning of vegetation growing close to or on the groundto minimize the fuel supply for future bush or forestfires. Also called (in Australia especially) ‘burning-off’.

Probability of Exceedence. The probability or likelihoodof an event – with a certain magnitude or size –occurring, expressed in the range of 0–100 per cent.

Pumice. Volcanic ash filled with gas particles and henceable to float on water.

Pyroclastic. Referring to the fine, hot ash or sedimentproduced from the fragmentation of magma in theatmosphere by the force of a volcanic eruption, andthe gases contained within.

Pyroclastic Ejecta. Pyroclastic material blown out into theatmosphere. See also Tephra.

Pyroclastic Flow. See Nuée Ardente.

QQuaternary. The most recent geological period, beginning

approximately 2–3 million years BP and consisting of thePleistocene and the Holocene.

Quick Clays. Clays that have been deposited with a highdegree of sodium cations that tend to glue particlestogether, giving the deposit a high degree of stability.However, sodium cations are easily leached by ground-water, with the result that the clay deposit becomesprogressively unstable over time.

RRecurrence Interval. The average time between repeat

occurrences of an event of given magnitude.Reg Desert. The residual, stony desert formed after clay

and sand-sized particles are blown away.Regolith. The layer of weathered or unweathered, loose,

incoherent rock material of whatever origin formingthe surface of the Earth.

Regression Coefficient. A term used in statistics to definethe relationship between two variables. The coefficientcan vary from –1.0, a perfect inverse relationship, to1.0, a perfect direct relationship.

Reservoir. (1) A structure deliberately constructed toimpound water. (2) A pocket of resistance that exists inthe natural environment in which an organism orpathogen can exist with impunity and can spreadoutwards under favorable conditions.

Resonance. A process involving the development of airor water waves in which the wavelength is equal to, orsome harmonic of, the physical dimensions of a basinor embayment.

299Glossary

Rift Valley. A zone of near-vertical fracturing of theEarth’s surface that causes one segment to dropsubstantially below another.

Rossby Waves. Any wave motion that develops with along wavelength of hundreds or thousands of kilo-meters in a fluid moving parallel to the Earth’s surfaceand controlled by Coriolis force. Generally restrictedto the upper westerlies attached to the polar jet streamat mid-latitudes.

Run-Up. When any ocean wave reaches the shoreline, itsmomentum tends to carry a mass of water landward.The distance landward of the shoreline is the run-updistance, while the elevation above sea level is the run-up elevation.

Rhyolite. A fine-grained, extruded, volcanic rock with awavy, banded texture. Mineralogically it is similar togranite, consisting of quartz, feldspar, and mica.

SSahel. The seven African countries – Gambia, Senegal,

Mauritania, Mali, Burkina Faso, Niger, and Chad – atthe southern edge of the Sahara Desert.

Saltation. The tendency for sediment particles to movefrom the bed into air- or water-flow and back to the bedin zigzag paths, where the distance beneath upwardmovement is much shorter than the distance beneathdownward movement.

Scoriae. Fused volcanic slag or ash that results from apyroclastic flow and contains a high degree of gas.

Seamounts. Volcanoes that have formed in the ocean,planed off at the top because of close proximity to theocean’s surface, and have then been lowered below thesea as the Earth’s crust moves away from the center ofvolcanic activity.

Seiching. Excitation of a regular oscillation of waterwaves within a basin or embayment – caused by anearthquake, the passage of an atmospheric pressurewave, or the occurrence of strong winds.

Seismic. Related to sudden and usually large movementof the Earth’s crust.

Seismic Gap. That part of an active fault zone that hasnot experienced moderate or major earthquake activityfor at least three years.

Seismic Risk. The probability of earthquakes of givenmagnitude occurring in a region.

Set-Back Line. A line established shoreward of a coastline,incorporating 50–100 years of postulated shorelineretreat, and in front of which no development shouldproceed.

Set-Up. The enhanced elevation of sea level at a coastlinedue mainly to the process of wave-breaking acrossa surf zone, but also aided by wind, shelf waves,Coriolis force, upwelling, and current impinging alonga coast.

Shallow Focus. Earthquakes occurring within 60 km ofthe surface.

Shear Sorting. Any process that causes coarse particles tobe transported more efficiently in a liquid or air flowthat is thin, full of sediment, and traveling at highvelocity.

Shear Strength. The internal resistance of a body ofmaterial to any stress applied with a horizontalcomponent.

Shear Stress. The force applied to a body of material,which tends to move it parallel to the contact withanother solid or fluid.

Shearing. A stress caused by two adjacent moving objectstending to slide past each other parallel to the plane ofcontact.

Slump. The tendency for some landslides to undergorotation, with the greatest failure of material at thedownslope end.

Solar Flares. Large ejections of ionized hydrogenfrom the sun’s surface accompanied by a pulse ofelectromagnetic radiation.

Solifluction. The slow movement of water-saturatedmaterial downslope, usually in environments wherefreezing and thawing are the dominant processesaffecting the surface.

Southern Oscillation. The usual movement of air in thetropical Pacific is strong easterly trades reinforced bylow pressure over Indonesia–Australia and highpressure over the south-east Pacific. Every two toseven years, this low and high pattern weakens or evenreverses, causing the trades to fail or become wester-lies. It is this switching from one pattern to anotherthat is the Southern Oscillation. See also North AtlanticOscillation and North Pacific Oscillation.

Spot Fires. Small fires ignited by flying embers carried onupdrafts generated by heated air from the main core ofa forest or bushfire.

Steric. Refers to the change in volume of seawater due toheating or cooling.

Storm Surge. The elevation of sea or lake levels abovepredicted or normal levels resulting mainly from adrop in air pressure, or from the physical piling upof water by wind associated with a low-pressuredisturbance or storm.

300 Glossary

Strain. The deformation or movement caused by applyinga force to material on a slope.

Stratosphere. That part of the atmosphere lying12–60 km above the Earth, and characterized bystability and an increase in temperature with altitude.

Stratospheric Fountains. Little air is exchanged betweenthe troposphere and stratosphere. However, intenseconvection – generated by thunderstorms in thetropics at the west side of the Pacific – breaches thisbarrier and, like a fountain, injects air originating fromthe ground into the stratosphere.

Strip Farming. An agricultural practice used to conservesoil moisture whereby crops are alternated in stripswith fallow land.

Subcritical Flow. Tranquil river flow in which velocitiesrelative to the channel depth are low enough thatsediment transport is not excessive.

Subduction Zones. The zone down which an oceanicplate passes beneath a continental plate to beconsumed into the mantle.

Subsidence. The downward failure of the Earth’s surfacebrought about mainly by removal of material from below.

Sunspot. An area 20 000 km in diameter on the sun’ssurface featuring strong magnetic disturbance so strongthat convective movement of heat to the surface iscurtailed. Hence, the spot is darker because it is cooler.

Supercell Thunderstorm. A particularly large, convectivethunderstorm cell with a diameter of 50–100 km.Such a storm has a greater chance of producing hail,tornadoes, strong wind, or copious rainfall.

Supercritical Flow. River flow in which velocities relativeto the channel depth are so high that the flow begins toshoot or jet. Such flow is very erosive.

Surcharging. The escape of stormwater through access-hole covers that are designed to lift off underexcess pressures caused by flooding. This preventsstormwater pipes from bursting underground, whichwould necessitate costly repairs.

Suspension Load. That portion of sediment being carriedin air or water flow by turbulence.

TTaiga. Swampy, coniferous, evergreen forest of sub-Arctic

Siberia.Tectonic. The process characterizing the deformation of

the Earth’s crust because of earthquakes.Tectonic Creep. Deformation of the Earth’s crust that

occurs in such small increments that it is virtuallyundetectable, except over time.

Teleconnection. The linkage of a measured climatic timeseries to another one at some distance across the globe.For instance, rainfall in Florida can be related, throughthe Southern Oscillation, to sea-surface temperature innorthern Australia.

Teleseismic Tsunami. A tsunami generated by an earth-quake from a distant source. The Chilean earthquake of1960 generated a teleseismic tsunami that impacted onmost coasts surrounding the Pacific Ocean.

Telluric Currents. Electric currents moving through theEarth.

Tephra. Volcanic ash that is disaggregated and blown bythe force of an eruption, usually vertically, into theatmosphere.

Tertiary. The geological period from 65 to 2–3 millionyears BP.

Thermal Expansion. Water is unusual in that it has itsgreatest density at 4°C. Thus, it expands readily whenheated. Ice, however, has its greatest volume aroundthe freezing point. It shrinks and cracks around –10°C.If water can enter the cracks and refreeze, such ice willexpand and buckle as it warms.

Thermocline. The abrupt change in water temperaturewith depth within a large body of water.

Thixotropy. The property of becoming fluid when stirredor agitated, especially by compressive shock waves. Seealso Liquefaction.

Threshold. The point at which the application of a forcetriggers a dramatic response that did not exist before-hand. For example, increasing stress applied to rockmay produce no observable effect until a threshold isreached and the rock fractures.

Tidal Inlet. The opening that forms across a barrier islandin which there is a free and substantial exchange ofsaline water between the ocean and a landward lagoonduring each tide cycle.

Tidal Waves. A misnomer applied to tsunami and stormsurges because these phenomena sometimes produce aslow drop in sea level followed by a rapid rise, analogousto tides.

Tilt Steps. Tilting of the Earth’s crust that occurs in aseries of discrete steps rather than smoothly.

Topographic Control. Any airflow whose path is deflectedor altered by terrain that is higher than the surround-ing landscape. For example, a moving mass of air maybe blocked by high mountains and go around theobstruction.

Tornado. An extremely violent vortex or whirlwind of airmeasuring only hundreds of meters in diameter andwith wind speeds up to 1300 km hr-1.

301Glossary

Trade Winds. Winds on the equatorial side of mobile polarcells, blowing from east to west within 20–30° of theequator. Their steadiness for three to four months ofthe year greatly benefited trading ships in the days of sail.

Trench. An area of oceanic crust that has been draggeddownward by the movement of oceanic crust beneatha continental plate.

Tropical Cyclone. A large-scale vortex of rising airhundreds of kilometers in diameter that forms over thetropical oceans. It is characterized by copious rain anda central area of calm surrounded by rotating windsblowing at speeds in excess of 200–250 km hr-1.

Tropopause. The band separating the troposphere –where temperature decreases with altitude – from thestratosphere – where temperature increases withaltitude. It forms a barrier to the ready exchange of airbetween these two zones.

Troposphere. The lower part of the atmosphere,10–12 km in altitude, characterized by cloud formationand a drop in temperature with elevation.

Tsunami. A Japanese word for ‘harbor’ (tsu) ‘wave’(nami). It is used to define a water wave generated bya sudden change in the seabed resulting from anearthquake, volcanic eruption, or landslide.

Tsunami Earthquake. Some earthquakes that arenormally too small to produce any sea waves locallygenerate devastating and deadly tsunami.

Turbulence. Flow – also called turbulent flow – in whichindividual particles or molecules travel unpredictablepaths apparently unaffected by gravity.

UUltrasound. Sound with a frequency above 20 000

vibrations per second and generally not perceivable bythe human ear.

Uniformitarianism. A geological explanation developed inthe eighteenth century to counteract the Catastrophists.It involves two concepts: the first implies that geologicalprocesses follow natural laws applicable to science(methodological uniformitarianism); the second impliesthat the type and rate of processes operative todayhave remained constant throughout geological time(inductive reasoning).

VVan Der Waals’ Bonds. The weak attraction exerted by all

molecules to each other, caused by the electrostaticattraction of the nuclei of one molecule for the electronsof another.

Vertical Aggradation. The build-up of a floodplain in avertical direction because of the deposition of sediment.

VHF. Abbreviation for ‘very high frequency’. That partof the electromagnetic spectrum with a wavelengthbetween 30 and 300 megahertz, characteristic of radioand television transmission.

Victim–Helper Relationship. In disasters where survivorsare at the mercy of outside support for survival for theirdaily requirements, a strong dependence can developbetween a victim and those who are providing the aid.This dependency may not be easily terminated whenthe necessity for relief has ended.

Viscosity. The resistance of a fluid to the motion of itsown molecules because they tend to stick together.

Voids. The spaces or gaps that exist between particles ina sedimentary deposit, filled with air or water.

Vortex. A spinning or swirling mass of air or water thatcan reach considerable velocity.

WWalker Circulation. The normal movement of air across

the tropical Pacific is from the east, between a stationaryhigh-pressure cell off the coast of South America, and alow-pressure zone over Indonesia–India. In the 1920s,Gilbert Walker discovered that the strength of this airmovement determined the intensity of the Indianmonsoon.

Wave Diffraction. The tendency for wave energy to spreadout along the crest of a wave. Hence, when a wavepasses through a narrow harbor entrance, the wave willspread out to affect all the coastline inside a harbor.

Wave Refraction. The tendency for the velocity at whicha wave travels to reduce over shallow bathymetry.Hence, a wave crest tends to wrap around headlandsand to spread out into bays.

Wave-rider Buoy. An instrument, consisting of sensitiveaccelerometers encased inside a buoy and floating onthe ocean surface, used for measuring the height andenergy of waves.

Wind-Chill Factor. The factor by which the temperatureof a surface is reduced because of the ability of windto remove heat from that surface more efficiently byconvection (see Figure 3.23).

Wind Shear. The tendency for winds to move at differentvelocities with elevation because of strong temperaturedifferences. The wind, as a result, may accelerate inspeed over short altitudes or even change direction – aparticularly hazardous situation for airplanes landing ortaking off.

302 Glossary

ZZonal. Restricted to an area bounded by parallels of

latitude on the Earth’s surface.

303Glossary

YYield Limit. The limit beyond which a soil can no longer

resist a force without permanently deforming. Yield Strength. The maximum resistance of a solid to a

force before it is permanently deformed.

aa lava 228, 232acid rain 235–6‘acts of God’ 1, 4Adelaide, Ash Wednesday fires 151adiabatic lapse rate 85Africa

drought, Sahel region 105, 106, 109–11dust storms 79pre-colonial response to drought 103–5

agriculture, drought impacts 105, 109, 110air circulation model 18–19air movement 17–18air-supported flows (avalanches) 266–7Alaskan earthquake (1964) 194, 206–7,

220, 253and land subsidence 267and subsequent tsunami 206–7, 220, 278

Aleutian Lows 22, 23, 35, 37Amazonia, fires 145andesite 185, 228andesitic lavas 227animal behavior

and impending earthquakes 196and impending volcanic eruption 275

Antarcticaiceberg movement 165iceberg production 164pack-ice 164

anti-social behavior following disasters279–80, 283

Arcticiceberg movement 164–5iceberg production 164sea-ice, melting of 162

arson, as cause of fires 143, 144aseismic locations 183ash clouds 233, 240–1, 242, 246Ash Wednesday bushfires (Australia, 1983)

32, 151–2firestorms 95people’s attitude to 274preparedness for 276rebuilding following 279trauma following 285

Ash Wednesday storm, 7 March 1962(USA) 64–5

Asia, dust storms 81astronomical cycles 37–41

see also MN lunar tidal cycleAtlantis 238atmospheric circulation models 18–19atmospheric shock waves 221, 242Atterberg limits 253Australia

1982–93 drought 112–13, 284–5bushfire preparedness 276bushfires 32, 139, 149–53, 274cyclone preparedness 54drought policy 112–13drought survival 105droughts 32, 111dust storms 32, 79, 81earthquakes 183–4, 213east-coast lows 63, 64east-coast storms 65–7eastern, river systems 133expansive soils 256farmers’ response to drought 112

Index

fire weather 141flash floods 127–31great floods 133–5hailstorms 91, 92–3land subsidence 268SOI impacts 32–4thunderstorms 86, 87tornadoes 99tropical cyclones 5, 34, 51, 54–6, 97, 275

avalanches 249, 266–7debris 230, 260, 265preventing/deflecting 267snow 266, 267

Azores High 35

Balcones Escarpment, Texas, flash floods120, 126, 127

ballistics and tephra clouds 232–3Bam, Iran earthquake 282Band Aid 116, 117, 118Bandai eruption 233Bandaian eruptions 230Bangladesh

cyclone shelters 58floods 30hailstorms 92storm surges 51, 57, 73, 76, 275tornadoes 99tropical cyclones 30, 51, 57–8

barrier shorelines 173–4basaltic lavas 227

types of 228base surges 233, 247Bathurst Bay Cyclone 51

bathymetry 156Bay of Bengal

storm surges 76, 77tropical cyclones 57, 58

Bay of Fundy, Canada 76beach erosion 170–4

human-induced factors 171–2natural factors 173–4sea level rise and Bruun Rule 170–1

Benioff zone 185biblical accounts

of Exodus of Israelites 238–9of floods 2

blizzards 71–2and wind-chill factor 71definition 71United States 68, 70–1

block lava 228, 231blocking highs 21, 22, 23‘bombs’ see east-coast lowsboreal forests, fires in 142Botswana, response to drought 111–12Brazil

fires 139floods 27landslides 263

breakwalls, impact of 171–2Brisbane River flood, 1974 134–5Bruun Rule, and sea level rise 170–1burning off, as cause of fires 144bushfires

Australia 32, 139, 149–53, 275conditions favouring intense 140–3, 149–60evacuation warnings 275fuel arrangement in 140–1historic disasters, Australia 32, 150–3prediction 141–2seasonal patterns, Australia 149world perspective 144–53

caldera formation 140, 214, 230, 242, 243California

coastal storms 28earthquake hazard 207–9earthquake risk 273earthquakes 183, 184, 193–4, 198, 206fire weather 141fires 140, 144, 146, 147–8

Canadaforest fires 144–5ice storms 73icebergs 164pack-ice 164snow avalanches 267snowmelt flooding 71snowstorms 67, 68SOI and precipitation 34

tornadoes 98–9tropical cyclones 48–9, 51

Canberra bushfires (2003) 152–3cannibalism 280Caracas, Venezuela, debris flows 260Caribbean

ENSO and tropical cyclones 34tsunami 221

catastrophism versus uniformitarianism 3–5catastrophists 4cavitation flow 124cementation 250–1Central America, tropical cyclones 53Charleville–Nyngan (Qld) flood (1990) 135Cherrapunji, India, rainfall 120–1Chile

tsunami 218, 221–2tsunami warning system 223–4

Chilean earthquake 192, 194, 206tsunami wave pattern 216, 217, 221–2

Chinadust storms 81earthquake prediction 196, 197earthquakes 183, 197, 204–5, 206, 213flooding 51, 135–7forest fires 145landslides 263rebuilding following earthquakes 282storm surges 73, 76tropical cyclones 51

chronic hazards 1clays

cracking 256creep 258–9plasticity 254sodium cation-rich, liquefaction 253types of 254

climate changeand ENSO events 28–9and jet stream paths 21–2and mobile polar highs 22–4

climate variability mechanisms 17–42cloud seeding, to reduce hail damage 91–2coastal erosion 65coherence (soils) 251cohesion (soils) 250–1convection 19, 24convective instability 24, 85–6Cordon Caulle eruption, Chile 192Coriolis force 17, 18, 19, 59Coulomb stress 198cracking clays 256, 257creep 258–9Crete

destruction of Minoan civilisation 238tsunami following Santorini eruption 240

critical incidence stress syndrome 286

305Index

crop lossesfrom hail 91–2from tropical cyclones 54

crop management practices, and drought105

cyclogenesis 22Cyclone Mahina 51Cyclone Tracy 5, 51, 97

disaster coordination centre role 55economic impact 54evacuation of residents 55–6, 282, 284failure to respond to warnings 275heroes 283, 286lack of preparation for 55psychological problems following 284,

285–6rebuilding following 56, 284response during the event 277

Dapto (NSW), flood 127, 130Darwin

Cyclone Tracy 5, 51, 54–6, 275, 282, 284rebuilding following Cyclone Tracy 56,

284death and grief, following disasters 278–9deaths from natural hazards 8–9, 291debris avalanches 230, 260, 265debris flows 259–61, 265Deep-Ocean Assessment and Reporting of

Tsunami (DART) 223desertification 80, 110dilation, and earthquakes 188downbursts 94drought 22, 25, 103–19

Africa’s Sahel region 105, 106, 109–11and dust storms 82and ENSO events 29, 30–1, 32, 79, 152and gruesome human behavior 280and migration 110and MN lunar tidal cycle 39, 41, 107Australian policy 112–13Bob Geldorf’s private response 115–18Botswana’s response to 111–12coping with 5environmental impacts of 110–11exacerbated by modern societies 106frequency 6, 7international aid ‘flops’ 114–15international response to 111, 113–15modern response to 106–13post-colonial response 105pre-colonial response, Africa 103–5social impact, Australia 284–5technological response to 105United Kingdom 108–9United States response to expected 106–8

drought relief 107, 110, 111, 113–15

drought-reduction strategies, United States107–8

dustfrom volcanic eruptions 190–1global signature 80, 81in upper troposphere 106role of 80–1

dust devils 94dust storms 32, 78–9, 110

and degradation of arable land 81, 82formation 79–80frequency 79, 81–2from cold fronts 79human activities affect on 81–2materials carried in 81notable events 82–3particle size 79–80, 82trajectories 81transport of pathogens 81

Dust Veil Index 190, 191Dvorak scale 50

earthquake cycles 193–5earthquake intensity scales 179–82earthquake lights 197earthquakes

and landslides 263annual frequency by magnitude on Ms

scale 180at plate boundaries 180, 181, 182,

184–5causes of 185–9clustering of 192, 193, 197–8distribution 182–4economic cost 7frequency 6, 7in myths and legends 3interaction with volcanoes 192–3largest by magnitude on Ms scale 180liquefaction or thixotropy 211–14, 252location of 180, 181major events 205near dams and reservoirs 188–9notable events 204–14occurrence during cyclones 45precursors 195prediction 193–8randomness versus clustering 197–8rebuilding following 279resettlement following 282seismic wave generation 202–3submarine 214through faulting and dilation 187–8tsunami following 206–7, 210, 216–19,

221see also specific localities

East Pakistan see Bangladesh

east-coast lows 58formation 63–4topographic control effects 64

east-coast stormsAustralia 65–7United States 64–5

eastern Australiabig wet of 1973–74 134major rivers 133rainfall 30

economic costs, of natural hazards 7–8, 290Ecuador, floods 27Egypt, and Santorini eruption 238–9El Chichon eruption, Mexico 29, 190, 191El Niño event 25–6El Niño–Southern Oscillation events 25–9

1982–83 27, 28, 291990–95 291997–98 28and droughts 30–1and fires 141, 145, 152and floods 30–1, 134and Kelvin waves 28and sea level changes 168, 170and sea surface temperatures 25–6, 29and short-term climate change 28–9and tropical cyclones 28, 49–50and volcanic eruptions 29causes 27evolution and effects 25

elastic deformation 253elastic solids 253, 254electrosphere 86emotional problems following disasters 56,

283–6English Channel storms 60, 62ENSO events see El Niño–Southern

Oscillation eventsepicenter 181episodic hazards 1Ethiopia

drought 107, 110, 111, 113drought relief debacle 114–15ENSO and historical development 32individual response to drought in 116

Eucalytpus, as fire-prone vegetation type140

Eucalyptus forests, crown fires 142Eurasian Plate 227evacuation (following the disaster) 55–6,

282, 284evacuation (prior to the disaster) see

warnings and evacuationevapotranspiration 106exceedance probability 77–9expansive soils 256–8

impact on houses 257

306 Index

lime effects 258extra-tropical depressions 23, 59

east-coast lows 63–7historical events 60–3polar-front lows, formation 59–60

faminedeaths from 9following drought 280

faulting, and earthquakes 187–8Federal Emergency Management Agency

(FEMA) 52, 57, 100–1feedback 24fire behaviour 142–3fire prevention/suppression 144, 146, 147,

148fire tornadoes 95fire weather 141–2fireballs 153fires 139–54

causes of 143–4following Tokyo earthquake 210–11see also bushfires; forest fires

firestorms 95, 142, 153flash floods

Brisbane area 134flood power 122–4from anomalous large-scale atmospheric

circulation 124–5magnitude and frequency of heavy

rainfall 120–2maximum probable rainfall 125–7notable events 127Sydney–Wollongong area 127–31, 135urban 127–31

flood lavas 229flood mitigation works 132–3, 136, 137flood power 122–4floods 120–37

and ENSO events 27, 29, 30–1and La Niña events 30and MN lunar tidal cycle 39–40, 41economic cost 7frequency 6, 7high-magnitude, regional floods 131–7in myths and legends 2–3see also flash floods; specific localities

Floridafires 148tropical cyclones 29

flow failure 212föhn wind 143Food and Agriculture Organization (FAO)

113, 114foreshocks 195forest fires 140, 141

Canada 144–5

fire behavior 142prediction 141–2United States 141, 144, 145–9

Francefires 140hailstorms 92

freak waves 159–62freezing rain 34, 72–3friction (soils) 250frictional coefficient of ice 19Fujita Tornado Intensity Scale 97fumaroles 228

galactic disc 189Galungung eruption, Java 233Geldorf, Bob 11, 111, 115–18

Band Aid Christmas aid album 116critics of Geldorf’s fundraising 118Live Aid concert 116–17

geo-electric precursors of earthquakes 195,196, 197

geomagnetic activity 37, 38and thunderstorms 86–7as earthquake precursor 195–6

geosynclines 183geothermometers 199–200Germany

hailstorms 92storms and floods 60, 61

ghoulish behavior following disasters 279–80giant waves 159–61gilgai 257Gisborne earthquake, New Zealand 216glacier bursts 237–8global cooling, and dust 80, 81global sea level and the hydrological cycle

169–70global seismic risk maps 204global warming

and changing hazard regime 289–90and melting of sea-ice 162–3, 164and sea level rise 166

Gobi Desertdust from 78dust storms 79, 81

governmentsAfrica’s Sahel, inaction over drought

109–11UK, unprepared for drought 108–9USA, preparedness for drought 106–8

granular flow 234grass fires 140

fire behavior 143graupel 88, 91Great Black Dragon Fire 145Great Britain

drought 22, 108, 109

land subsidence 267–8storms and floods 61–2tornadoes 99water supplies 108–9

Great Drowning Disaster storm 60–1Great Lakes region

fires 141ice storms 72–3snowfalls 70storm surges 74

Great Meiji Sanriku earthquake 220Great Plains (USA), drought 39, 107–8Great World Drought, 1982–83 25Greenhouse effect 155Greenhouse gases 155, 289–90Greenland

icebergs 164pack-ice 164

grounded sea-ice 165, 166groundwater fluctuations, as earthquake

precursors 196group response to hazards see personal and

group response to hazardsGujarat, India earthquake 183, 197

haboob 79Hadley cells 19, 22, 23, 142hail 90–3

building damage 92–3crop damage 91, 92deaths from 92locations and frequency 91size of 91

hail suppression programs 92–3Hale Cycle 37Halloween Storm, October 1991 (USA)

65harmattan winds 79Hawaiian Islands

tropical cyclones 29tsunami 216, 275volcanoes 182, 186, 198, 199, 229, 231,

232Hawke’s Bay earthquake, New Zealand

279hazard regimes, changing 288–90hazard response 5–6hazard statistics 6–9headlands, waves on 161hectopascals 17Herculaneum 240, 241heroes, in disasters 283high waves, world distribution 159, 160high-magnitude, regional floods 131–7

Australia 133–5China 135–7Mississippi River floods 131–3

307Index

Hobart bushfires (1967) 150–1Honduras, tropical cyclones 5Hong Kong, landslides 263hot spots 186–7houses

expansive soils effect on 257loss following disasters 279

Huaynaputina eruption, Peru 190human activity, and atmospheric dust 80humans

and natural hazards 5–6anti-social behavior following disasters

279–80, 283as cause of fires 143–4concerns about possessions and homes

278–9death and grief following disasters 278emotional problems following disasters

283–6myths about response during disasters

282–3preparedness, if warned 276reactions to warnings and evacuation

273–5resettlement following disasters 280–2response during the disaster 277–8role of heroes 283

Hurricane Andrew (Florida and Louisiana)50, 51–3, 54, 57, 279

wind strength and level of destruction 52Hurricane Carol (east coast USA) 74, 75Hurricane Grace (east coast USA) 65Hurricane Hazel (Ontario) 48–9, 51, 278,

282Hurricane Mitch (Central America) 53hurricanes see tropical cyclonesHwang Ho (Yellow) River, China, flooding

135–7

ice-dammed lakes 123ice-expansion, in lakes 166ice-push 165–6ice storms 72–3icebergs 164–5

and ENSO events 34hazards from 165monitoring 165movement 164–5

Icelandjökulhlaups 238lava flows 231volcanoes 182, 232, 236, 238

Icelandic Lows 22, 23, 35iceshelves 163–4ignimbrites 230, 235illite 254Imamura–Iida scale 219

imminent disasterpreparedness for 276reaction to 274–5

Indiaearthquakes 183, 197, 205floods and droughts, and SOI 31

Indigenous peoples, attachment to the land281

Indonesiafires 145monsoon droughts and ENSO 31, 32tsunami 216volcanic eruptions 26, 182, 190, 191,

215, 221, 228, 232, 233, 236, 241–3

infra-red emissions, and earthquakes 197

insects 27, 244, 275instability, convective 24, 85–6international aid ‘flops’ 114–15International Ice Patrol 165International Red Cross 114international relief organizations 113–14international relief programs, organized by

Bob Geldorf 115–18intertropical convergence 22intra-plate earthquakes 183introduced animal species, impact of 106inverted barometer effect 74ionosphere 86Iran, earthquakes 282irrigation projects, for drought-reduction

107island arcs 182, 185isobars 17

and wind movements 18Italy

debris avalanches 265earthquakes 205–6snow avalanches 267

Izmit earthquake, Turkey 198, 204

Japanearthquake hazard 210–11earthquake prediction 195earthquakes 45, 192, 205, 206, 212–13east-coast lows 63landslides 263storm surges 73tropical cyclones 51tsunami 216, 219, 221tsunami warning systems 222, 223

jet streamsand hail 90changes in 21–2looping effects 21, 67

jökulhlaups 237–8

Kamchatka tsunami 223kaolinite 254katabatic winds 22, 79Katherine (NT), flood of 1998 135Katmai eruption, Alaska 230, 234, 236‘Katmaian’ eruptions 230Kelvin wave 28, 61Kilauea eruptions, Hawaii 19, 192, 198,

199lava flows 231Kobe earthquake, Japan (1995) 7, 192, 211Krakatau eruption, Indonesia 26, 190, 191,

215, 228, 233, 236, 241–3ash and dust 242explosions 242tsunami following 216, 221, 241–3volume of ejecta and deposits 242

L-waves 202, 203La Niña events 29–30

and fires 141, 148and flooding 134, 135and sea level changes 168, 170

lahars 215, 236–7, 246, 277prediction 237primary 236secondary 236–7

Lake Illawarra flood 127–8, 130lakes, ice-expansion in 166land clearing, impacts of 107land instability 249

air-supported flows (avalanches) 266–7classification 254–6creep and solifluction 258–9debris avalanches 265expansive soils 256–8landslides and slumps 261–4mud and debris flow 259–61rockfalls 264–5

land management practices, and drought105

land subsidence 249, 267–8natural mechanisms 268

landslides 45, 249causes 261–2eastern Australia 30initiating failure 262mechanics 261notable disasters 263–4submarine 214

latent heat 18lateral spreads 212, 213lava 227, 228, 229lava flows 230–2

characteristics 231stopping/diverting 231–2

lava tubes 231

308 Index

lightning 88–90as cause of fire 143, 144damage from 90deaths from 90, 128electromagnetic energy 90frequency 89sunspot activity effects 37

lightning discharge 88–9liquefaction 211

and types of failure 212earthquake-induced 211–14, 252sodium cation-rich clays 253

Lisbon earthquake, Portugal 3–4, 221, 281Little Ice Age 37, 289Live Aid concert 116–17, 118, 285longshore drift 171–2looting 279Los Angeles

earthquake likelihood 208–9earthquake prediction 194

loss of bearing strength 212Lothar (windstorm) 62Love wave 202luminescence 197

magma 182, 186, 192degassed 228extrusion 227–8gas phase 227, 228types of 227

magnetic reversals 189magnetization 199malnutrition 110mantle 185, 189Martin (windstorm) 62Marxist view of natural hazards 5, 6mass extinctions 189Maunder Minimum 37maximum probable rainfall 125–7maximum rainfall equation 121–2Meckering, WA, earthquake 213Mediterranean climate 140, 142mega-ENSO events 29mega-tsunami theory 5, 288–9Mercalli scale of earthquake intensity

180–2meridional airflow 24

and flash floods 124–5mesohigh floods 125mesoscale synoptic storms 126meteorite impact, and tsunami 215meteorite streams, and mega-tsunami 288–9Mexico

earthquakes 213–14volcanic eruptions 29, 190, 191

microseisms 195mid-latitude cyclonic depressions 67–8

mid-ocean ridges 182migration, as a result of drought 110mine tailings

and debris flows 260disasters, trauma following 285

Minoan civilization, destruction of 238Mississippi River floods 131–3

1927 flood 1321973 flood 132Great Flood of 1993 133levee and floodwall development 132

MN lunar tidal cycle (18.6 years) 8, 17,39–41, 290

bistable phasing 40impact on drought 39, 41, 82, 107, 111impact on flood 39–40, 41lunar orbit relative to Earth and Sun

39, 40mobile polar highs 22–4, 25, 35, 36

dynamic structure 24Mogi doughnut 197Mohr–Coulomb equation 251–2, 261montmorillonite 254, 256Moon, orbit relative to Earth and Sun

39, 40Mt Etna eruption, Sicily 231, 232Mt Fuji eruption, Japan 192Mt Hekla eruption, Iceland 232, 238Mt Huascarán avalanche, Peru 265, 277Mt Hudson eruption, Chile 29, 233Mt Kelat eruption, Java 236Mt Ngauruhoe eruption, New Zealand

234Mt Nyiragongo eruption, Democratic

Republic of Congo 231Mt Pelée eruption, Martinique 230, 234,

243–6natural warnings prior to 244pyroclastic flow 245–6tsunami from 245violent explosions 245–6

Mt Pinatubo eruption, Philippines 29, 190,191, 192, 233, 237

Mt Rainier, Washington State 237, 246Mt Redoubt eruption, Alaska 199, 233Mt Ruapehu, New Zealand 237, 238Mt St Helens eruption, Washington State

191, 199, 233, 236ash accumulation 247ash falls 246lahars 246pyroclastic flow 233, 234

Mt Soufrière eruption, St Vincent 244, 245Mt Spurr eruption, Alaska 29Mt Stromboli eruption, Italy 229Mt Tarawera eruption, New Zealand

281–2

Mt Vesuvius eruption, Italy 190, 192, 230,240–1

ash falls 240–1lava flows 231pyroclastic flows 234

mountainadoes 95mudflows 215, 236–7, 259–61myths and legends 2–3, 238

natural hazardsdeaths from 8–9, 291economic costs 7–8, 290impacts on humans 5–6location of 290–1Marxist view 5, 6modern consequences 290–1occurrence 290prediction of 290

Netherlandsstorm surges, recurrence intervals 76–7storms and floods 61, 62tornadoes 99

New South Waleseast-coast storms, May–June 1974 65–6flash floods 127–31storm surges 73, 74waves, rocky coasts 161see also Sydney

New Zealandearthquake risk 273earthquakes 216, 279floods 286volcanic eruptions 234, 235, 281–2

Newcastle earthquake (1989) 183, 213Nigeria, Hausan people response to

drought 104–5Niigata earthquake, Japan 45, 212–13North American Plate 185, 208North Atlantic Oscillation (NAO) 34–5

and dust storms 79and geomagnetic activity 38and sea-ice 162atmospheric circulation patterns and

climatic impacts 35, 36decadal phases 35indices 26, 35

North Pacific Oscillation (NPO) 35–7indices 26, 37

North Seastorm surges 73, 74storms 60, 61, 62

northern hemisphere, mobile polar highs22, 23, 24

nuées ardentes 230, 233–5, 245

Oakland fires, California 147, 148ocean trenches 182

309Index

oceanic hazards 155–75beach erosion 170–4sea-ice 25, 155, 162–6sea level rise 166–70waves 155, 156–62

P-waves 202, 203, 223Pacific Ocean

and ENSO events 24–5, 27–9tsunami 215–16, 221–2tsunami prediction 222–4

Pacific Ocean ‘ring of fire’ 182, 185Pacific Plate 185, 186, 208, 227Pacific Tsunami Warning System 222, 224pack-ice 163–4

on shore 165–6pahoehoe lavas 228, 231Palmén–Newton model of global

circulation 18–19, 24and areas of forcing 22and monsoonal circulation 22limitations 19terminology problems 22

panic during disasters, as myth 277, 282Papua New Guinea

landslides 263tsunami 224

parent cloud, and tornado formation 94,95–6

Peléean eruptions 230, 235periodic hazards 1permafrost 249personal and group response to hazards 273

anti-social behavior following disasters279–80

critical incidence stress syndrome 286death and grief 278–9emotional problems 283–6myths and heroes 282–3possessions and homes 278–9preparedness, if warned 276resettlement 280–2response during the event 277–8warnings and evacuation 273–6

Perudebris avalanches 265, 267ENSO and historical development 32floods 27, 32volcanic eruptions 190

Philippinestsunami 216, 224volcanic eruptions 29, 190, 191, 192, 233,

237phreatomagmatic eruptions 230piezoelectric properties 196piezomagnetism 199plastic deformation 188

plastic limits 254plastic solids 253–4plate boundaries

and earthquakes 180, 181, 182, 184–5and volcanoes 182, 184, 185

‘Plinian’ eruptions 230, 240–1pluviometers 121–2polar front lows

formation 59–60historical events 60–3

polar fronts 19polar jet stream 21, 35polar shelf- and sea-ice 162–3Pompeii 240–1pore-water pressure 211–12, 252–3Portugal, earthquakes 3–4, 221, 281possessions, loss of following disasters 278–9power (waves) 157–8preparedness for imminent disaster 276prescribed burning 144, 146pressure gradients, and wind intensity 18probability of occurrence (storm surges) 76

exceedance probability 77–8recurrence intervals 76–7

psychosomatic illness following disasters283–6

pyroclastic ejecta 228pyroclastic flows 214, 230, 233–5, 245–6

and lahars 236, 237speed of 234temperatures in 234

quick clays 253, 260

Rabaul eruptions 234rainfall

and beach erosion 173and La Niña events 30and MN lunar tidal cycle 39–40and urban flash floods 127–31from tropical cyclones 45heavy, magnitude and frequency 120–2,

289maximum probable 125–7

Rayleigh wave 202, 203rebuilding at site of the disaster 56, 279,

281–2regional floods, high-magnitude 131–7regoliths 249, 250, 258religious beliefs, impact of 281–2reporting incidence of natural hazards 7–8reservoir earthquakes 188–9resettlement following disasters 280–2Reunion Island, rainfall 45rhyolite 185, 228Richter Scale 179–80rigid solids 253

ringbarking of trees 106ritualized human sacrifice 280rockfalls 264–5rocky coasts, waves on 161–2Rocky Mountains 21, 267rogue waves 159–61Rossby waves 21, 28, 35, 36, 59, 124rotational sliding 261Russia

snowmelt flooding 71storm surges 74, 75

S-waves 202, 203Saffir–Simpson scale 50Sahara region

dust from 78dust storms 81rainfall 22

Sahelian droughts 105, 106, 109–11St Petersburg, storm surges 74salinity 106saltation 79San Andreas Fault 185, 187, 195, 198, 207

activity 208risks to people living along 273

San Fernando earthquake 213San Francisco

earthquake (1906) 206, 207–8, 213earthquake likelihood 208Great Fire of 1906 274rebuilding following earthquake 281

Santorini eruption1470 BC 215, 221, 238–40tsunami following 240

sea, giant waves at 159–61sea-ice

categories 162–5formation 162grounded 165, 166

sea-ice as a hazard 35, 155, 162–6ice at shore 165–6ice in the ocean 163–5

sea level rise as a hazard 155–6, 166–70current rate of change in sea level

worldwide 166–8sea level rises

and Bruun Rule 170–1cities threatened by 166–7factors affecting 168factors causing 168–9global sea level and the hydrological

cycle 169–70satellite altimeter data 167–8tide gauge data 167, 168, 169

sea surface temperaturesand ENSO 25–6, 29and NAO 35

310 Index

seamounts 186sediment transport, in floods 123, 136, 137seismic activity

prediction of 193–9short-term prediction 195–7

seismic gap concept 197seismic risk maps 204seismic scales 179–82

Mercalli scale 180–2Richter scale 179–80

Seismic Sea Wave Warning System 222seismic waves 179, 192

types of 202–3setback line 65shallow-focus events 183shear sorting 235shear strength of soils 251–4

plastic solids 253–4pore-water pressure 252–3rigid and elastic solids 253

shear stressand landslides 263from flash floods 122

shore-fast ice 164shoreline retreat, and Bruun Rule 170–1shrinkage limits 253–4Siberia, forest fires 145sightseers to disaster sites 279silver iodide 91slope failure 261, 262–3

causes 262–3slumps 261snakes 244snow avalanches 266, 267snowmelt flooding 71snowstorms

economic impact of 69–71formation 67–8impact on cities 70–1United States 67, 68–9

social impacts of drought 106, 109–10, 284–5sociological perspective of natural hazards

5–6soil mechanics 249–51soils

expansive 256–8friction, cohesion and coherence 250–1shear strength 251–4stress and strain 249–50

solar cycles 37–8solar flares 37solar radiation, and dust 80solar system 189solifluction 258–9Soloviev scale 219South Africa, east-coast lows 63South Pacific Convergence Zone 27

southern hemisphere, mobile polar highs22–3, 24–5

Southern Oscillation 24–34global long-term links to droughts and

floods 30–1indices 26links to other hazards 31–4time series, Australia 32–4see also El Niño–Southern Oscillation

events; La Niña eventsSoviet Union, floods 7storm surges 44–5, 51, 57, 64, 73

causes 73–6deaths from 73inundation from 54, 73potential 50probability of occurrence 76–8recurrence intervals 76–7

storm-surge waves 74and beach erosion 173–4shoaling effects 74, 75

strain (soils) 250stratosphere 24stratospheric fountains 24stream power, in floods 122–4stress and strain (soils) 249–50stress mapping 198‘Strombolian’ volcanoes 229subcritical flow 124subduction zones 182, 183and volcanism 184, 185submarine earthquakes 214submarine landslides 214, 220Sudan

drought 110, 113, 115dust storms 79floods 30

Sunda Strait, affected by tsunami fromKrakatau eruption 242–3

sunken lands, in myths and legends 2–3sunspot cycles 37

and fires 141and thunderstorms 86–7impact on lightning and tropical cyclone

activity 37–8sunspot–global temperature association 37supercell thunderstorms 24, 94supercell tornado formation 94supercritical flow 123surcharging 129Switzerland

debris avalanches 265snow avalanches 267

Sydneybushfires (1994) 152east-coast storms (May–June 1974) 65–6flash flooding 127, 128–31, 135

hailstorms 92–3thunderstorms 88

Sydney Hobart Yacht Race 1998, storm 66–7Sydney Hobart Yacht Race 2001,

waterspout 95synoptic patterns

favoring bushfires 141favoring flash flooding 124–5

Tahiti, tropical cyclones 28Tajikstan, debris avalanche 265Tambora eruption, Indonesia 190, 191,

228, 232, 236, 241Taupo eruptions, New Zealand 235tectonic creep 188teleseismic tsunami 220telluric currents 196tephra 228, 230, 235tephra clouds, and ballistics 232–3Thailand, floods 30Thar Desert, dust storms 79, 81thermoclines 27–8Third World countries 5–6thixotropy 211–14thunderclouds, and lightning 88, 89thunderstorms 85–8

attracted to urban heat islands 87–8damage from 88electricity generation 86link to tornadoes 87stationary 126

Tibetan Plateau 21, 27tidal waves 214tilt steps 195tiltmeters 199Tokyo earthquake (1923) 45, 192, 206,

210–11Tonga, tropical cyclones 28topographic control 64tornadoes 19, 93–101

attitudes towards 101deaths from 99destruction 99forecasts and warnings 100, 275form and formation 93, 94–7frequency 6, 93, 97funnels 95–6intensity measurement 97link to thunderstorms 87multiple-vortex formation 97occurrence 97–9path width and length 93response to 100–1speed of 94structure 95–7supercell tornado formation 94topographical influence 101

311Index

wind speeds 96–7trade winds 19, 20transcurrent faults 185transport networks, to supply aid 114–15transportation, snowstorm effects 69tropical cyclones 19, 30, 44–83

and earthquakes 45and ENSO events 28, 29, 34, 49–50and wind-spread fires 45conditions for development 47–8early warning system, USA 56energy sources 45–6evacuation procedures, USA 56–7frequency 6, 7hazards related to 44–5human and economic impact 54–8intensity measures 50magnitude and frequency 48–50notable disasters 50–8origins 46preparedness for 276response during the event 277sunspot activity effects 37–8see also named events, eg Cyclone

Tracy; specific localitiestropical storms, economic cost 7tropopause 21, 24troposphere 18, 80, 235tsunami 214–24

deaths from 220–1disaster descriptions 220–2flooding from 218following Alaskan earthquake 206–7,

222, 278following Chilean earthquake 216, 217following earthquakes 206–7, 210,

216–19, 221following Krakatau eruption 216, 221,

241–3following Lisbon earthquake 221following Santorini eruption 240following Tokyo earthquake 210frequency 6, 7, 214geographical distribution 215–16in myths and legends 3marigrams 216, 217meteorite impact origin 215origins 214–15prediction in the Pacific Ocean 222–4run-up height 216, 218–19submarine landslide origins 214teleseismic 220volcanic origins 214–15warning systems 222–4wave periods 216, 217wavelengths 216see also mega-tsunami theory

tsunami earthquakes 220Tsunami Hazards Reduction Utilizing

System Technology (THRUST) 224tsunami magnitude scales 219–20Tuamota Archipelago, tropical cyclones 28Tully (Qld), rainfall 121Turkey, earthquakes 198, 204Turkmenistan, dust storms 81typhoons see tropical cyclones

Ukraine, dust storms 82uniformitarianism 4United Kingdom see Great BritainUnited Nations Children’s Fund

(UNICEF) 113United Nations Disaster Relief Office

(UNDRO) 113United Nations Office of Emergency

Operations in Africa (UNOEOA)113–14

United States of Americablizzards 70–1drought 39, 106–8dust storms 79, 81, 82–3earthquakes 183, 192, 193–5, 198, 206–9east-coast lows 63east-coast storms 64–5expansive soils 256flash floods 122–3, 124, 125, 126, 127forest fires 139, 144, 145–9freezing rain 72hailstorms 91, 92lahars 236landslides 264lightning 89preparedness for disaster 275, 276snow avalanches 267snowstorms 67, 68–9SOI and precipitation 34storm surges 51, 73, 74, 75stream power during floods 122–3thunderstorms 86, 87tornadoes 87, 97–8, 99, 275tropical cyclones 49, 50–3, 56–7, 276tsunami 206–7, 222tsunami warning system 222–3volcanic eruptions 191, 199, 233, 234,

236, 246–7Unzen eruption, Japan 221urban flash floods 127–31urban heat islands, and thunderstorms

87–8

Van der Waals’ bonds 251, 254VAN method 196, 197vegetation destruction, impact of 107vertical aggradation 136, 137victim–helper syndrome 56, 284

Victoria, Ash Wednesday fires see AshWednesday bushfires

volcanic disasters 238Krakatau (1883) 26, 190, 191, 215, 221,

228, 233, 236, 241–3Mt Pelée (1902) 243–6Mt St Helens (1980) 191, 199, 233, 234,

236, 246–7Mt Vesuvius (79 AD) 190, 192, 230, 231,

234, 240–1Santorini (1470 BC) 215, 221, 238–40

volcanic eruptionsand ENSO events 29and projectile behavior 232ash clouds 233, 240–1, 242dust from 190–1, 242energy from 228–9, 242explosive 229–30frequency of 189–90, 289interaction with earthquakes 192–3magma extrusion 227–8non-explosive 229prediction 198–200tephra ejecta 232–3types of 229–30see also specific eruptions

volcanic events, total energy 228volcanic gases 199–200, 235–6volcanic hazards 230

ballistics and tephra clouds 232–3gases and acid rain 235–6glacier bursts or jökulhlaups 237–8lahars 236–7lava flows 230–2pyroclastic flows and base surges 233–5

volcanic island arcs 182, 185volcanic tremors 199volcanism

and decreased surface air temperatures191

at hot spots 186–7at plate boundaries 182, 184, 185in myths and legends 2–3prediction 189–92

volcanoes 227–47causes of 185–9distribution 182–4gas emission analysis 199–200magnetic fields 199

voltage gradient 86vortices 18, 94–5‘Vulcanian’-type eruptions 229–30

Walker circulation 25, 27, 29–30, 31and SOI 33–4

warning systemstornadoes 100, 275tropical cyclones 56

312 Index

tsunami 222–4warnings and evacuation 100, 273–6

bushfires 275concern for property 276confusion over warnings 274cost of evacuation 275non-evacuation reaction 274people’s perceptions of hazards 273–4preparedness if warned 276reaction to imminent disaster 274–5

Washington scablands 4water management, United States 107water supplies, United Kingdom 108–9waterlogged muds, shock wave effects

213–14waterspouts 95, 96watertables 106wave behavior, in shallow water 156, 158wave characteristics 156–9wave destructiveness 157wave diffraction 156, 157, 158, 216wave heights 156, 157, 158

giant waves 159–61rocky coasts 161–2

wave period 157, 158–9wave reflection 161wave refraction 156, 158, 161wave velocity 157wavelength 158waves as a hazard 155, 156–62

at sea 159–61on rocky coasts 161–2theory 156–9world distribution of high waves 159

weather maps, pressure gradients 17West Coast/Alaskan Tsunami Warning

Centre 223willy-willys 94wind-chill factor 71wind chill index 71–2wind movements, relative to isobars 18wind set-up 74wind shear 89, 94wind speeds, tornadoes 96–7wind storms, economic cost 7winds

in bushfires 142in tropical cyclones 45

windstorms, Europe 62–3Wollongong (NSW), flash flooding 126–7,

128, 130, 131, 135, 289World Food Programme (WFP) 113World Health Organization (WHO) 113World Meteorological Organization

(WMO) 113

Yellowstone fires (1988) 147yield limit 254

Documents

NATURAL HAZARDS by : EDWARD BRYANT