You are on page 1of 162

This page intentionally left blank

Volcanoes and the Environment


Volcanoes and the Environment incorporates contribu- in 1986. His Ph.D. research focused on the ancient
tions from some of the foremost authorities in explosive volcanism of the Pyrenees. Since the begin-
volcanology from around the world to form a com- ning of his scientific career, he has conducted
prehensive and accessible text. This book is an indis- much research relating to the structure of calderas
pensable guide for all those interested in how vol- (giant collapse craters), and the dynamics of caldera-
canism has affected our planet’s environment in the forming eruptions and associated hazards. Dr. Martı́
past and will continue to do so in the future. Span- has studied explosive volcanism in different modern
ning a wide variety of topics from geology to clima- and ancient volcanic environments. In particular, he
tology and ecology, it also considers the economic has contributed many studies on explosive volcan-
and social impacts of volcanic activity on humans. ism in Tenerife (Canary Islands), where he continues
Chapters cover the role of volcanoes in shap- to coordinate several research projects.
ing our planet’s environment through the eons, and
their effect on the geological cycle; the impacts on g e r a l d e r n s t was awarded a B.Sc. in geol-
atmosphere and climate; impacts on the health of ogy and mineralogy from the University of Liège,
those living on active volcanoes; the role of vol- Belgium, in 1990 and a Ph.D. in Earth sciences
canism on early life; effects of eruptions on mod- from the University of Bristol, UK, in 1997. Follow-
ern plant and animal life and implications from ing appointments as a postdoctoral fellow and lec-
these studies; links between large eruptions and turer in volcanology and geological fluid mechanics
mass extinctions; relationships between contrasting at the University of Bristol he joined the Belgian
human societies and volcanic disasters; how vol- National Science Foundation (FWO-Vlaanderen) and
canoes can provide heat energy and supply pre- the University of Ghent. as a researcher in 2003. He
cious base metals, as well as raw material, for our is currently pursuing new research focused on how
industries; and impacts of volcanic disasters on the volcanoes work and how they affect people, their
economy. activities, and the wider environment. Dr. Ernst
This book is intended for students and is working to establish the Mercator and Ortelius
researchers interested in environmental change Research Centre for Eruption Dynamics, and the
from the fields of earth and environmental science, Joseph Plateau Geological Fluid Dynamic Labora-
geography, ecology, and social science. It will also tory at the University of Ghent. He is a Honorary
interest and inform policy-makers and professionals Research Fellow of the University of Bristol, a
working in natural hazard planning and mitigation. Scientific Collaborator of the University of Brussels,
and an Adjunct Assistant Professor of Volcanology
j o a n m a r t í is a Research Professor at the Institute at Michigan Tech University (USA). He has been the
of Earth Sciences of the Spanish National Research recipient of four awards for his work including the
Council (CSIC) in Barcelona. He studied geology prestigious 2002 Golden Clover Prize of the Fonda-
and obtained his Ph.D. at the University of Barcelona tion Belge de la Vocation.
Volcanoes and the Environment
Edited by
Joan Martı́
Institute of Earth Sciences ‘‘ Jaume Almera”, Consejo Superior de
Investigaciones Cientificas

and

Gerald Ernst
Mercaton and Ortelius Research Centre for Eruption Dynamics,
and Joseph Plateau Geological Fluid Dynamics Laboratory,
University of Ghent
  
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo

Cambridge University Press


The Edinburgh Building, Cambridge  , UK
Published in the United States of America by Cambridge University Press, New York
www.cambridge.org
Information on this title: www.cambridge.org/9780521592543

© Cambridge University Press 2005

This publication is in copyright. Subject to statutory exception and to the provision of


relevant collective licensing agreements, no reproduction of any part may take place
without the written permission of Cambridge University Press.

First published in print format 2005

- ---- eBook (EBL)


- --- eBook (EBL)

- ---- hardback


- --- hardback

Cambridge University Press has no responsibility for the persistence or accuracy of s
for external or third-party internet websites referred to in this publication, and does not
guarantee that any content on such websites is, or will remain, accurate or appropriate.
Contents

List of contributors page vii


Preface ix
Acknowledgments xv

Chapter 1 Understanding the physical behavior


of volcanoes 1
Steven N. Carey

Chapter 2 Volcano hazards 55


Robert I. Tilling

Chapter 3 Anticipating volcanic eruptions 90


Joan Mart´
ı and Arnau Folch

Chapter 4 Volcanoes and the geological cycle 121


Ray A. F. Cas

Chapter 5 Effects of volcanic eruptions on the


atmosphere and climate 152
Stephen Self

Chapter 6 Volcanoes, hydrothermal venting, and the


origin of life 175
Karl O. Stetter

Chapter 7 Volcanism and mass extinctions 207


Paul B. Wignall

Chapter 8 Effects of modern volcanic eruptions


on vegetation 227
Virginia H. Dale, Johanna
Delgado-Acevedo, and James MacMahon

Chapter 9 Animals and volcanoes: survival and revival 250


John S. Edwards

Chapter 10 Human impacts of volcanoes 273


Peter J. Baxter

Chapter 11 Volcanoes, geothermal energy, and the


environment 304
Wendell A. Duffield
vi CONTENTS

Chapter 12 Volcano-hosted ore deposits 333


Harold L. Gibson

Chapter 13 Industrial uses of volcanic materials 387


Grant Heiken

Chapter 14 Volcanoes, society, and culture 404


David K. Chester

Chapter 15 Volcanoes and the economy 440


Charlotte Benson

Index 468
Contributors

Peter J. Baxter Wendell A. Duffield


Department of Community Medicine Geology Department
University of Cambridge Northern Arizona University
Addenbrooke’s Hospital Flagstaff
Cambridge AZ 86011
CB2 2QQ USA
UK
John S. Edwards
Charlotte Benson Department of Zoology
209 Jalan Ara University of Washington
Bangsar Baru Seattle
59100 Kuala Lumpur WA 98195
Malaysia USA
Steven N. Carey
Graduate School of Oceanography Gerald G. J. Ernst
University of Rhode Island Mercaton and Ontelius Research Centre
Narragansett for Eruption Dynamics and
RI 02882 Joseph Plateau Geological Fluid
USA Dynamics Laboratory
Geological Institute
Ray A. F. Cas University of Ghent,
School of Geosciences Krijgsloan 281/S8
Monash University B-9000 Ghent
P.O. Box 28E Belgium
Victoria, 3800
Australia Arnau Folch
Institute of Earth Sciences ‘‘Jaume Almera”
David K. Chester Consejo Superior de Investigaciones Cientificas
Department of Geography Lluis Solé Sabaris s/n
University of Liverpool 08028 Barcelona
Liverpool Spain
L69 3BX
UK Harold L. Gibson
Virginia H. Dale Department of Earth Sciences
Environmental Sciences Division Laurentian University
Oak Ridge National Laboratory Sudbury
Oak Ridge Ontario
TN 37831 P3E 2C6
USA Canada

Johanna Delgado-Acevedo Grant Heiken


Universidad de Puerto Rico Earth and Environmental Sciences Division
Departamento de Biologia Los Alamos National Laboratory
Apdo. 23360 Los Alamos
00931 San Juan NM 87545
Puerto Rico USA
viii LIST OF CONTRIBUTORS

James MacMahon Karl O. Stetter


College of Science Lehrstuhl für Mikrobiologie
Utah State University Universität Regensburg
5305 University Blvd Universitätsstrasse 31
Logan D-93053 Regensburg
UT 84322 Germany
USA
Robert I. Tilling
Joan Martı́
Volcano Hazards Team
Institute of Earth Sciences ‘‘Jaume Almera”
US Geological Survey
Consejo Superior de Investigaciones Cientificas
345 Middsfield Road, MS-910
Lluis Solé Sabaris s/n
Menlo Park
08028 Barcelona
CA 94025
Spain
USA
Stephen Self
Department of Earth Sciences Paul B. Wignall
The Open University School of Earth Sciences
Walton Hall University of Leeds
Milton Keynes Leeds
MK7 6AA LS2 9JT
UK UK
Preface

Volcanic eruptions are among the most fascinat- populations are the most vulnerable to natural
ing natural phenomena and can have significant hazards, remain largely unstudied. In this age
impacts upon the environment. One only has to of supercomputers, nanotechnology and space
think of the 1883 eruption of Krakatau or of the exploration, we know very little about these vol-
1980 eruption of Mount St. Helens to get a sense canoes of our own planet, most of which are not
of the awesome power and wide-ranging impacts monitored to any extent, or about the impacts
of eruptions. Some readers will remember hear- they could have in future. Changing this situa-
ing or reading about the loss of life and devas- tion needs to be one of the highest priorities of
tation around those volcanoes, while others will volcanological efforts in coming decades.
remember how even larger eruptions than these Volcanoes also contribute positively to the
cooled the Earth’s climate and affected the ozone environment. They have brought much to ‘‘life”
layer. Most will recall controversial discussions in the past, and represent, at present, an impor-
about how volcanoes may have eradicated the tant source of benefits for humanity. For example:
dinosaurs some 65 million years ago or about
how super-eruptions may have nearly wiped out r Volcanoes have decisively contributed to the
our human ancestors some 75 000 years ago. We origin of life and Earth’s atmosphere, and are
like to think it is fortunate that they did not, but often regarded as directly responsible for the
what will happen when the next super-eruption existence of highly fertile soils in many parts
strikes? Most recently, eruptions at ocean island of our planet. Uplift related to volcanism in
volcanoes have even been proposed as triggers easternmost Africa, starting millions of years
for catastrophic, massive volcano failure and the ago, and associated climatic and vegetation
generation of tsunami waves of unimaginable changes, have been related to the emergence
proportion. The 26 December 2004 Banda Aceh of our direct ancestors in the Rift Valley some
tsunami; off Sumatra, which devastated coastal 2–3 million years ago.
areas in SE Asia, killing 300000 and precipitat- r The volcano-derived heat that may have fueled
ing several million people into a state of absolute the emergence of life in the oceans is now being
poverty, is a small event in comparison. Volcano- mined as geothermal energy on several con-
related mega-tsunamis represent a very great risk tinents. Volcanoes are also the source of ore
to many coastal cities around the world and to deposits and an important source of material
their populations, and no doubt to the world for industry.
economy. Moreover, it has now been demon- r By studying the dispersal of giant clouds pro-
strated that periods of severe cooling lasting 1000 duced by explosive volcanic eruptions, atmo-
years are unambiguously correlated with erup- spheric scientists have made great advances in
tions. This will no doubt fuel heated debates as to understanding how the atmosphere works, and
whether eruptions can indeed trigger glaciations particularly in understanding its energy bal-
on that timescale. ance and the complexity of atmospheric circu-
Eruptions arguably have the potential to have lation. Each time a large eruption happens, it is
all these impacts and many more upon their sur- a natural experiment on the scale of our planet
roundings and, in some cases, upon the global as a whole, enabling atmospheric scientists to
environment. This is a sobering thought, espe- put to challenging tests their models predicting
cially when considering that there are still a our current and future climate.
large number of potentially hazardous volcanoes r Thick volcanic ash layers from powerful erup-
of which we know next to nothing. For exam- tions have also buried and preserved the
ple, about 80% of active volcanoes, most of which records of ancient civilizations worldwide. In
are located in developing countries where local Europe, this is illustrated by the late Bronze
x PREFACE

Age towns on the Island of Thera (Santorini, eruptions can be anticipated. These topics are
Greece; arguably the legendary Atlantis or part fascinating aspects in their own right and set the
of it), which have provided profound insights scene for the discussions that follow.
into the early development of European art, This wide coverage of topics related to the
from the study of its beautiful paintings on impacts of volcanism and volcanic eruptions
walls and pottery. Two thousand years later, upon the many interrelated aspects of the envi-
another large eruption buried the Roman town ronment considered in the widest sense is what
of Pompeii (near Naples, Italy), again providing makes this book unique. A few years ago we
invaluable clues into what made a great civ- came to realize that many of the aspects now
ilization. Of course, these buried civilizations discussed in this new text were covered only in
are also a stark reminder of the extraordinary part and only in scattered texts. There was not
destructive power of the largest of eruptions, a single text attempting to present an integrated
for which we have either no or only limited treatment. Many of the discussions appeared in
historical records. specialized journal articles and books, making
r Studying the origin of life at hot vents on them inaccessible to amateur environmentalists
deep submarine volcanoes can also shed light and to most undergraduates and postgraduates.
on whether life may exist on other planets We felt that what was needed was a text that
such as Europa, a satellite of Jupiter; and on would be not only comprehensive in its treat-
whether life may have thrived on Mars in the ment of the subject but also accessible to a wide
past. audience of naturalist amateurs, undergraduate
and postgraduate students in the environmen-
The list of benefits derived from volcanoes and tal, geographical and earth sciences as well as an
their study extends much beyond this short list of easy-to-carry reference text for all our research
examples as will become clear during the reading and teaching colleagues across many scientific
of this book. disciplines.
In this textbook we have taken the broad- Seventeen of our colleagues among the lead-
est possible view of the environment. We have ing authorities on the subject enthusiastically
considered not just the impacts of eruptions on shared our vision and together we started prepar-
the atmosphere and climate, and on the flora, ing what was to become this book, the very first
fauna and humans around volcanoes but also the textbook extensively discussing most aspects of
impacts on human health, human societies, and the impacts of volcanic eruptions upon the envi-
the local and national economies. We have also ronment.
considered the role of volcanism in generating We hope that like all the contributors to this
precious base metal resources, the use of volcanic book the readers will be irresistibly enthused by
materials in industry and the recovery of volcano- this exciting subject. The following paragraphs
derived thermal energy, the contribution of vol- introduce the successive chapters in more detail,
canic eruptions to past mass extinctions, and the outlining some of the basic questions, which are
role of volcanism and volcano-derived hydrother- discussed by the contributors.
mal venting on the emergence of the earliest
forms of life on Earth and on the development Chapters 1–3 are foundation chapters needed
of the primitive atmosphere and ocean. We have before the diverse effects of volcanism can be
also included a treatment of the role of volcanism studied. Chapter 1, by Steven Carey, a senior
in the geological cycle. physical volcanologist at the University of Rhode
In order to enable discussions of all these Island, sets the scene by reviewing our under-
interrelated impacts to be easily followed it has standing of volcanoes and of the physical pro-
been necessary to introduce a basic treatment cesses associated with volcanic eruptions. Some
on the physical understanding of volcanic erup- of the questions considered include: where is
tions, as well as on volcanic hazards and on how volcanic activity concentrated and why? What
PREFACE xi

are the main contrasting types of volcanic set- their activities or the environment at large.
tings and how do they differ in the styles of the Chapters 4–5 explore in turn the relationships
eruptions which occur there? How variable is the between volcanism and a specific aspect of the
composition of magma and how does this come ‘‘environment.” Chapters 4–5 are related, respec-
about? How are magmas generated? In what way tively, to the relationship of volcanism with
do volcanic eruptions contribute to the devel- geological time and space, and with a specific
opment of the landscape (or the ‘‘seascape”) – aspect of the physical environment of our planet,
mountains and topographic depressions? namely its atmosphere. Some of the key questions
Volcanic hazards can affect humans and the discussed include: what has been the contribu-
environment. But what are the main types of tion of volcanoes in shaping our planet and its
hazards and how can humans cope with them? environment through the history of the Earth?
How can we assess them? Can we predict the Are eruptions becoming more or less frequent
onset of eruptions? What are the techniques now than millions of years ago? What is the
used? Is it sufficient to anticipate where and role of plate tectonics and when did plate tec-
when eruptions will occur to anticipate poten- tonics start to act as a major driving force in the
tial disasters, either environmental or human in evolution of our planet? What was, and what is
character? These are some of the crucial ques- now the contribution of volcanic degassing and
tions considered, in Chapter 2, by Robert Tilling, volcanic eruptions to the atmosphere? – to the
a veteran of volcano monitoring and volcano dis- oceans? – to life? These questions are discussed
aster response with the US Geological Survey. in Chapter 4 by Ray Cas, a veteran volcanologist
Tilling discusses some of these questions by com- at Monash University who has studied volcanoes
paring the cases and responses to two classic first-hand in many parts of the world.
eruptions, the eruptions in 1985 at Nevado del In recent years there has been a growing real-
Ruiz, Colombia and in 1991 at Mt. Pinatubo, ization that volcanic degassing and eruptions of
Philippines. He concludes by summarizing some a range of styles can impact upon both the chem-
of the key lessons that have been learnt by those ical composition of the atmosphere and upon its
who have lived through volcanic crises and vol- radiative energy balance. This, in turn, changes
canic disasters such as these. the temperature distribution on the Earth and
Chapter 3, by Joan Martı́ and Arnau Folch, thus the weather and climate that we experience.
both experienced volcanologists and modelers In Chapter 5 Stephen Self, a leading volcanologist
at CSIC (Consejo Superior de Investigaciones at the Open University and pioneer of research
Cientı́ficas, i.e., the Spanish National Research in this area, introduces the main mechanisms by
Council) in Barcelona, builds up on Chapters 1 which contrasting styles of eruptions can affect
and 2. It focuses specifically on how we can antic- atmospheric composition and climate. Through
ipate volcanic eruptions. It discusses what tech- a few key examples that include super-eruptions
niques are used in the wide variety of volcanic and flood basalt volcanism, he illustrates that
settings and corresponding volcano types. It also we are increasingly able to document how large
discusses how theory and laboratory work can eruptions affected the atmosphere and climate in
help by providing invaluable insights and under- the past. This effort is crucial to help us foresee
standing. The chapter shows along its different the impacts that future eruptions will have.
sections that the best way to anticipate a vol- Chapters 6–9 explore in turn several key
canic eruption and its effects is by combining a aspects of the relationship between volcanism
good knowledge of the volcano’s eruptive behav- and life, both in the past and at present. One of
ior (physical volcanology) and of the level of its the most fascinating aspects of our relationship
current activity (volcano monitoring). to volcanism is in how volcanic activity may have
These three chapters are fundamental to played a major role in our ultimate origin. That is
understanding the character of volcanic per- the origin of ancestral forms of life that evolved
turbations, which can impact on humans and into increasingly complex forms and eventually
xii PREFACE

some 4 billion years later or so, to us Homo sapiens. succession after each specific type of volcanic
In Chapter 6, Karl Stetter, a senior researcher at disturbance.
the University of California–Los Angeles and the In Chapter 9, John Edwards, a leading volcano
University of Regensburg and a leading authority zoologist and ecologist at the University of Wash-
on the subject, considers some of the following ington, discusses how animals may or may not
questions: what are the most primitive forms of survive volcanic eruptions and in the case where
life still in existence today? How do they relate the eruptions are so strong that they wipe out
to volcanism? How can we describe their rela- all life forms, how and in what order the animals
tionships with other species in an evolutionary are observed to recolonize a barren volcanic area.
tree? How diverse are these primitive forms of The chapter not only emphasizes survival and
life and how have they survived the competition revival of animal communities but also stresses
for almost 4 billion years? Professor Stetter con- that eruptions are a major process on evolution-
siders these fascinating questions and more, and ary timescales of millions of years. The discussion
places them in the framework of the history of includes the effects of ash dispersal on insects,
life on our planet. the issue of animal survival after eruptions, the
An area of heated debate has been whether or processes of recolonization by animals in six case
not the dinosaurs were wiped out by an asteroid studies, the recolonization in the zone destroyed
that fell over the Yucatan Peninsula in Mexico, by the ‘‘ash hurricane” at Mount St. Helens and
or whether longer-term volcanic activity covering later events in animal recolonization. One impor-
the largest part of India with layer upon layer of tant finding is that arthropods are key players, as
lava has anything to do with it. In Chapter 7, Paul primary colonizers. Another key finding is that
Wignall, a geologist and paleontologist leading recolonization appears extremely rapid compared
work in this area at the University of Leeds, seeks to evolutionary times. A largely unresolved ques-
evidence to resolve this issue. Wignall does not tion which deserves more attention is the ‘‘refu-
limit the discussion to the great extinction of the gia question” – whether or not some animals
dinosaurs and other species at the K–T boundary, may be able to survive extreme volcanic disas-
but considers a variety of well-documented mass ters and how this could come about? The chapter
extinction events, which may be related to mas- concludes by highlighting three key areas where
sive outpourings of lava on a continental scale. research is much needed in order to make further
The chapter closes by balancing the evidence in advances in understanding.
favor and against a major role of volcanism in the Chapter 10 considers the effect of recent erup-
mass extinction scenarios. tions on human health. Over half a billion people
In Chapter 8, Virginia Dale, Johanna Delgado- now live in the immediate vicinity of active vol-
Acevedo, and James MacMahon, all volcano canoes, the majority of them in developing coun-
botanists and ecologists, from the Oak Ridge tries and highly vulnerable to volcanic emissions
National Laboratory, the University of Puerto and eruptions. Therefore, a very important con-
Rico, and Utah State University, respectively, dis- cern is to discover how volcanic degassing and
cuss how plant life is affected by volcanic erup- eruptions can affect the health of a proportion
tions. These scientists, also at the forefront of of those people and to explore what could be
their field, consider recent eruptions of con- done to monitor and mitigate any adverse effects
trasting styles ranging from those emitting lava on human health. What are the health effects of
flows to those involving pyroclastic flows. What being exposed to volcanic gases? What are the
determines plant life survival? How does plant effects of eruptions of a range of styles? Peter
life recover after a major volcanic eruption? The Baxter, a medical scientist at the University of
chapter also summarizes the physical impacts Cambridge and the leading authority in the field
of eruptions and their impacts on vegetation of volcano medicine, has studied first-hand the
around volcanoes worldwide. They also com- effects of volcanic activity and eruptions world-
pare the patterns of the surviving floral com- wide and draws from case studies at many vol-
position, vegetation re-establishment, and plant canoes to discuss these fundamental questions.
PREFACE xiii

Chapter 10 also makes recommendations for good The exploitation of these volcanic materials in
practices during volcanic crises and raises con- turn has a direct impact on the environment.
cerns for the future where appropriate. What are the different types of materials that
Chapters 11–13 discuss in turn three valuable can be used and to what purposes? Chapter 13, by
by-products of volcanism which are extensively Grant Heiken, a veteran volcanologist at the Los
used by humans: volcano-derived heat, precious Alamos National Laboratory, describes these dif-
metals, and raw materials for industry. The pres- ferent types of materials and goes on to analyze
ence of hot magmas, below the surface in associ- their past and current uses. In particular, pumice
ation with volcanoes, offers the prospect of har- and scoria, which are both produced by a vari-
nessing a huge amount of thermal energy that ety of types of explosive eruptions, have many
can be used in our homes, both for heat or to pro- uses ranging from raw material in the prepara-
vide electricity. This so-called geothermal energy tion of abrasives, building-blocks for cathedrals,
is an important source of renewable energy cat litter, cement, and concrete, etc. But where
and the implications for the environment are sig- can we find these materials? What problems are
nificant if we can successfully recover it. Where we faced with as we try to identify the most eco-
does geothermal energy originate? What does it nomical varieties of rock types, and how is this
take for a geothermal deposit to be economical? related to the processes that deposited the mate-
How do we estimate reserves? Are there any rials in the first place? The chapter is not limited
adverse effects for the environment as geother- to pumice and scoria but encompasses the whole
mal energy is tapped and recovered, in compari- range of volcanic materials.
son with other types of energy? These are some Chapters 14–15 discuss additional aspects
of the key questions considered in Chapter 11 by of the relationship between eruptions and
Wendell Duffield, a veteran US Geological Survey humans – namely the relationship of volcanism
scientist and volcanologist. Geothermal energy is with human societies and the impact of erup-
already important and is bound to become even tions on the human economies. Human soci-
more so, as the reserves of fossil fuels run out and eties have had to respond to the many types of
as the cost of cleaning up adverse environmental volcanic disasters that have occurred and con-
impacts can no longer be avoided in energy cost tinue to occur worldwide. In Chapter 14, David
calculations. Chester, a geographer and volcano sociologist at
One of the many benefits of volcanism comes the University of Liverpool and researcher leading
from the exploitation of valuable ore deposits, this field, considers the relationship between con-
which are related to the presence of volcanoes trasting human societies and volcanic disasters.
in the Earth’s crust. But what do we mean when The chapter starts by discussing the relationship
we speak of ore deposits? How do we determine between volcanoes and society in time and space
whether an ore body is exploitable? What are from a historical perspective. It then moves on to
the different types of ore bodies associated with review the interface between social theory and
volcanism? How can we use our understanding eruptions. How have societies responded to erup-
of volcanoes and ore deposit formation to dis- tions in the past? What is the modern response
cover the new resources that will be needed in of contrasting societies to volcanic disasters? The
the future. In Chapter 12, Harold Gibson, an chapter challenges the dominant approach that
economic geologist and volcanologist expert on has been adopted almost up to the present
this subject at Laurentian University, considers and proposes more radical alternative societal
these questions and describes the volcanic envi- responses. It concludes by highlighting areas of
ronments and processes which hold the key to research that will have to be developed in the
the formation and location of volcano-derived ore future.
deposits. It is clear that eruptions have direct and indi-
A key area of the impact of eruptions on rect effects on the economy. Do they impact only
our environment and activities is through pro- local economies, or can there be effects on a
viding materials that can be used in industry. national or even global level? How big are these
xiv PREFACE

effects, and how is the cost of a volcanic disaster We hope that you, the readers, will enjoy read-
estimated? What are the key determinants of ing this book as much as we enjoyed putting
whether an eruption will or will not impact an it together. We will have succeeded in our
economy? In Chapter 15, Charlotte Benson, a vol- endeavor if reading this textbook inspires some of
cano disaster economist who has been pioneering you to pursue unanswered or poorly understood
research on this topic at the UK Overseas Devel- aspects of the fascinating, ‘‘volcanoes and envi-
opment Institute, draws from documentation on ronment” problems. More importantly, we hope
the economic impacts of volcanic activity. She that this book will enhance public awareness of
clearly demonstrates that this is an increasingly our rapidly changing and evolving environment
important and fascinating new area of research and of how volcanic eruptions contribute to the
and one where much more research is needed. changes.
Acknowledgments

In preparing this book, a challenging task for support from the respective home institutions of
all involved, and extending over several years, the authors, to which we extend our gratitude,
we received tremendous help and encouragement grants from public and private institutions have
from our families and colleagues. We are most also provided support for the preparation of some
grateful for their support and patience with us chapters. These generous institutions include the
while we were preparing and editing the book. Deutsche Forschungsgemeinschaft; the Bundes-
We are indebted to many friends and scientific ministerium für Bildung, Wissenschaft, Forsch-
colleagues. Their current and past research forms ung und Technologie; the European Union; the
the backbone around which the chapters of this Fonds der Chemischen Industrie; the Minority
book have in part been built up. Their thought- Biomedical Research Support program; the Uni-
ful, incisive comments and suggestions greatly versity of Puerto Rico; the Oak Ridge Institute of
influenced the book’s preparation. Particular Science and Education and Oak Ridge National
thanks are given to Russell J. Blong, William Laboratory; The National Geographic Society; the
E. Scott, Steven R. Brantley, Edward W. Wolfe, National Science Foundation and the Mazamas;
Harald Huber, Reinhard Rachel, Wendy Adams, the Fondation Belge de la Vocation; the Belgian
Robert O’Neill, Patricia Parr, Roger del Moral, National Science Foundation (FWO-Vlaanderen)
Frederick M. O’Hara, Linda O’Hara, Rick Sugg, the UK Natural Environment Research Council
Eldon Ball, Gordon Orians, Ian Thornton, Robert and the UK Nuffield Foundation (NAL award to
Fournier, Arthur Lachenbruch, Patrick Muffler, GGJE).
Herbert Shaw, Robert Smith, Steve Sparks, Chris We wish to thank Matt Lloyd, Susan Francis,
German, Martin Palmer, Mike Carroll, Bill Rose, Sally Thomas, Anna Hodson and Jo Bottrill of
Donald White, John Sass, Michael Sorey, John Cambridge University Press for their expert guid-
Lund, Sue Priest, Kevin Rafferty, J. M. Franklin, ance during the preparation of this book. With-
M. D. Hannington, J. Hedenquist, C. M. Lesher, out their expertise and dedication to this effort,
A. J. Naldrett, S. D. Scott, R. H. Sillitoe, D. H. the preparation of this book might not have been
Watkinson, R. E. S. Whitehead, R. R. Keays, and possible. Special thanks are given to Silvia Zafrilla
G. J. Ablay. for her contribution to the editorial tasks and for
Permission granted to use published figures the preparation of many of the figures in final
is gratefully acknowledged. In addition to the form.
Chapter 1

Understanding the physical behavior


of volcanoes
Steven N. Carey

Introduction Each eruption represents a unique culmination


of this complex series of steps, making the predic-
Volcanism is a spectacular display of the com- tion of volcanic eruptions one of the most chal-
plex way in which energy and materials are lenging tasks in the geosciences.
exchanged between three major components of The science of volcanology draws from many
our planet: the solid Earth, oceans, and atmo- different fields, including petrology, geochem-
sphere. Mankind has long been both fascinated istry, seismology, and sedimentology. It began
and terrified by erupting volcanoes. Yet through- with mainly qualitative observations about the
out history people have been drawn to their fer- distribution of volcanoes, their eruptive behavior,
tile slopes and have developed a unique sym- and the types of products they create. In AD 79,
biosis. In many cultures, volcanoes symbolize a Pliny the Younger made the first written descrip-
source of tremendous power that must be pla- tion of a volcanic eruption in two letters describ-
cated by worship or sacrifice. Volcanologists, on ing the death of his uncle, Pliny the Elder, during
the other hand, strive to understand how volca- the eruption of Vesuvius volcano in Italy. During
noes work in order to better predict their behav- the past two decades the field of volcanology has
iour and reduce the hazards to people who live undergone revolutionary changes in the under-
near them. But volcanoes are not merely destruc- standing of volcanic activity. This has resulted
tive and need to be viewed as an integral part from a more quantitative approach to the anal-
of the dynamic Earth system. They create new ysis of volcanic processes through the applica-
land, replenish soil, and provide essential water tion of physics, chemistry, and fluid dynamics. In
and other gases to our oceans and atmosphere. many instances, strides in the field were spurred
Much of this book will focus on the relationship on by observations of major eruptions such as
of volcanoes to the environment and to mankind. Mount St. Helens, USA, in 1980, El Chichón,
However, before these topics are addressed we Mexico, in 1982, Nevado del Ruiz, Colombia,
need to begin with some fundamental concepts in 1985, and Mt. Pinatubo, Philippines, in 1991
about the causes and processes of volcanism. In (described further in Chapter 2). These eruptions
this chapter we explore how volcanoes work by often provided the opportunity to observe vol-
examining the complex path that must be taken canic processes that had not previously been doc-
before an eruption takes place at the Earth’s sur- umented quantitatively and led to the ability to
face. This includes the generation of magma at interpret such processes from deposits in the geo-
depth, its rise, storage and evolution within the logic record (this will be discussed further in
Earth’s crust, and finally the factors that deter- Chapter 3). With the advent of Earth-observing
mine the nature of the eruption at the surface. satellites, the remote sensing of volcanoes from

Volcanoes and the Environment, eds. J. Martı́ and G. G. J. Ernst. Published by


Cambridge University Press.  C Cambridge University Press 2005.
2 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

space has led to a new appreciation for the global the line continues southward all the way to New
impact of some large-scale eruptions and pro- Zealand. This remarkable clustering of volcanoes
vided volcanologists with new tools to assess the is known as the ‘‘Ring of Fire.” Large earthquakes
relationship of volcanoes to the environment (e.g. are also concentrated along this belt adding to
Rose et al., 2000). A particularly exciting break- the hazards associated with the volcanoes. Other
through has enabled the mapping of the ocean’s significant belts of subaerial volcanoes occur in
bathymetry from satellites by measuring subtle the Indonesian Archipelago, the Mediterranean
changes in the elevation of the seasurface, allow- region and in the Lesser Antilles islands of the
ing for the recognition and mapping of volcanic West Indies.
features on the seafloor with high resolution. As It would take one of the major revolutions
a result of land-based, marine, and space obser- in the earth sciences, that of plate tectonics, to
vations there is now an excellent documenta- provide the unifying explanation for the distri-
tion of the distribution of volcanic activity on bution of volcanoes on the Earth. It has now
Earth. been demonstrated that the surface of the Earth
consists of numerous large plates that are con-
stantly in motion relative to one another and rel-
Distribution of volcanic activity: ative to the deep interior of the Earth (Fig. 1.1).
the plate tectonic framework The theory of plate tectonics provides the fun-
damental linkage between our understanding
The distribution of volcanoes on the Earth’s sur- of the Earth’s surface and its interior (Keary
face provides important information about the and Vine, 1990; Canon-Tapia and Walker, 2003). A
underlying causes of volcanism, yet making a cross-section through the planet reveals a stron-
global inventory of volcanoes is not a simple gly layered internal structure (Fig. 1.2). Much of
task. Simkin (1993) notes that it is difficult to our knowledge of the Earth’s interior has been
define precisely the number of active volcanoes derived from the study of how earthquakes are
on Earth. During historic times there have been transmitted in different types of materials. By
some 538 with documented eruptions and over studying the arrival of different types of seismic
1300 have erupted during the last 10 000 years. waves, such as P- and S-types, at various positions
However, two-thirds of the planet’s surface is cov- around the globe, it is possible to infer the nature
ered by water and thus observations of volcanic of the deep Earth. This remote-sensing approach
activity are strongly biased towards the conti- is necessary because direct sampling of the deep
nents and islands. We now know that virtually all Earth by drilling is usually limited to depths of
of the seafloor is composed of volcanic rock and about 5 km. Some deeper samples are brought to
that most volcanic activity occurs far removed the surface naturally by volcanic eruptions. These
from sight in the depths of the ocean. xenoliths have been critical for determining the
Early workers who documented the distribu- chemical composition and physical nature of the
tion of volcanoes realized that they did not occur source areas of volcanism.
randomly but tended to run in belts or linear From the center of the Earth to a distance
segments. This is particularly evident when one of 1228 km lies the inner core, a dense solid
looks at the present location of volcanism on consisting mostly of iron with lesser amounts of
the Earth’s surface (Fig. 1.1). A large majority of nickel and sulfur. Surrounding the solid inner
the Earth’s active and conspicuous volcanoes can core out to a distance of 3500 km is a liquid
be found along a belt that rims the Pacific Ocean outer core, also consisting of iron and sulfur.
beginning at the southern tip of South America, The presence of a liquid region deep in the
proceeding north along its western coast across Earth is inferred from the blocking of earthquake
Mexico into the western United States (Fig. 1.1). shear waves (S-waves), which are unable to prop-
From there it winds into Alaska, across the agate through a liquid. Within this layer, convec-
Aleutian islands to the western Pacific, where tive movement of material is responsible for the
Fig. 1.1. Distribution of active volcanic centers (solid circles) and the boundaries between major lithospheric plates. Arrows indicate directions of
relative plate motion. Modified from Sparks et al. (1997).
4 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

Fig. 1.2. Cross-section of the Earth’s


interior showing the major concentric
regions. Arrows indicate the motions
of the major lithospheric plates.
Thickness of the lithosphere is
exaggerated for viewing purposes.
Modified from Frankel (1996).

generation of the Earth’s magnetic field. At a dis- the lithosphere, a combination of crustal rocks
tance of 3500 km from the Earth’s center there and upper mantle material (Fig. 1.2). They are of
is an abrupt transition to the largest volumet- the order of 100 km thick and are considered
ric component of the solid Earth, the mantle. to behave in a generally rigid form. Lithospheric
Extending to less than 100 km from the sur- plates ride over an underlying asthenosphere
face, the mantle consists of silicate minerals and that consists of mantle rocks at higher temper-
is the source of most magma for volcanism on atures and pressures. These conditions allow the
the planet. As a result of gradients in pressure asthenosphere material to deform under stress in
and temperature throughout the mantle, the a ductile fashion and to accommodate the move-
assemblage of minerals changes as a function of ment of the overriding plates.
depth. In the lower mantle, dense phases such as The boundaries between plates can be subdi-
perovskite, magnesiowustite, and stishovite pre- vided into three main types: divergent, conver-
dominate, whereas above 670 km (upper man- gent, and transform (Fig. 1.2). Each one repre-
tle), minerals such as pyroxene, olivine, and an sents a different sense of motion between plates.
aluminum-bearing phase are stable. The mantle At divergent boundaries plates are moving away
represents 80% of the solid Earth by volume and from one another and magma wells up from the
67% by weight (Ringwood, 1975). mantle along fissures, so creating new oceanic
Overlying the mantle is a thin veneer, or crust, crust and lithosphere. An example of a divergent
on which we live. In simple terms, the crust can boundary is the one between the North Ameri-
divided into two principal types: oceanic and con- can and the Eurasian plates centered roughly in
tinental. Oceanic crust is relatively thin (up to the middle of the Atlantic Ocean (Fig. 1.1). Con-
15 km), dense (3.0 g/cm3 ), consists of rocks rich vergent boundaries represent the opposite sense
in iron and magnesium called basalt, dolerite, of motion where plates are coming together. In
and gabbro, and underlies most of the ocean these areas one plate moves under another in a
basins. In contrast, continental crust is thicker process called subduction (Fig. 1.2). The descend-
(up to 50 km), less dense (∼2.8 g/cm3 ), consists ing plate is typically bent downwards as it is
of rocks that are richer in silica, sodium, and returned to the mantle, producing a deep area,
potassium, such as granite, and forms most of the or trench, near the boundary. Earthquakes occur
continental masses. The plates that are in con- along the descending plate as it returns to the
stant motion over the Earth’s surface are termed mantle and their location can be used to define
DISTRIBUTION OF VOLCANIC ACTIVITY: THE PLATE TECTONIC FRAMEWORK 5

the slope of the downgoing slab, referred to Spreading rates are generally low by human
as the Benioff Zone. In the Pacific Ocean, the standards, averaging only a few centimeters per
deepest oceanic depth of 13,000 m is associated year. However, over long periods of geologic time
with the trench of the Mariana subduction zone. movements of plates can produce or consume
Convergent margins can be produced by subduc- entire ocean basins.
tion of oceanic lithosphere beneath other oceanic There is considerable variation in the rates of
lithosphere (island arc) or oceanic lithosphere spreading along different segments of the mid-
beneath continental lithosphere (active continen- ocean ridge system. In the Atlantic Ocean the
tal margin). Finally, transform boundaries are spreading along the mid-Atlantic ridge is only
areas where plates slide past one another, such as about 2 cm/year, whereas along the East Pacific
along the San Andreas fault system of the west- Rise the rate is up to 18 cm/year. It should be
ern United States, where the Pacific plate is mov- emphasized that the spreading rates are based
ing northward relative to the North American on calculations that average the production of
plate (Fig. 1.1). Significant earthquake hazards new seafloor over hundreds of thousands or mil-
are associated with this type of boundary when lions of years. There is not necessarily an addi-
plate motion occurs by rock failure. At transform tion of a few centimeters of new oceanic crust
boundaries, there is not, however, significant vol- all along the mid-ocean ridge every year. Erup-
canic activity. tions are likely to be episodic, occurring on the
Volcanoes and earthquakes are largely con- timescale of hundreds to thousands of years, and
fined to the boundaries between lithospheric affecting only parts of the ridge at any given
plates and are a direct consequence of plate time. Only recently has there been an actual erup-
motion. The style of volcanism, the composition tion detected along the mid-ocean ridge system.
of the erupted products, and the nature of vol- It occurred in 1993 when an event was recorded
canic features that are produced vary consid- from a 6-km-long section of the Juan de Fuca
erably between different boundaries and reflect spreading center off the northwest coast of the
the fundamental influence of tectonic forces on United States (Chadwick et al., 1995).
volcanism. The crest of the ridge is generally located
at a depth of several thousand meters. Topo-
Mid-ocean ridge volcanism graphic profiles across ridges exhibit distinctive
(divergent boundary) morphologies that relate to spreading rate. The
Volcanism at divergent boundaries has produced slow-spreading mid-Atlantic ridge consists of a
a globally encircling mountain range known as well-defined axial valley surrounded on each side
the mid-ocean ridge system. It extends through- by steep, fault-bounded ridges, whereas the fast-
out the major ocean basins of the world for a spreading East Pacific Rise is defined by a broad
total length of some 70 000 km. Discovery of elevated ridge with only a small-scale axial valley
this undersea mountain chain and its recogni- (Fig. 1.3).
tion as a primary volcanic feature was one of the Although they are primarily confined to the
most important advances in the field of marine seafloor there are areas, such as Iceland and the
geology. Dredging of rocks from the ridge and African Rift Valley, where divergent boundaries
direct observations by submersibles confirmed are exposed on land, providing a unique opportu-
that fresh volcanic rock is being emplaced along nity to directly observe volcanic processes. Iceland
this boundary, forming new seafloor that pro- is an example where an anomalous rate of
gressively moves away from the ridge axis. The magma discharge along the mid-Atlantic ridge
rate of production has been determined by dat- has elevated the spreading center above sea level.
ing magnetic anomalies in the oceanic crust that The island is being enlarged through volcanic
are arranged in symmetrical fashion on each side activity occurring along its neovolcanic zone that
of the ridge (Vine and Matthews, 1963). These extends from the southwest to the northeast
anomalies record times when the Earth’s mag- (Fig. 1.4). Along this zone is a set of prominent
netic field changed polarity direction by 180◦ . central volcanoes that rise to elevations up to
6 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

Fig. 1.3. Bathymetric cross-sections across


the East Pacific Rise at 3◦ N and 21◦ N, and
the mid-Atlantic ridge at 37◦ N showing the
differences in morphology between fast,
intermediate, and slow-spreading ridges.
Modified from Brown et al. (1989).

Krafla

ICELAND

ull
jok
tna
Va
Reykjavik
0 40 km Hekla

Katla Volcanic centers


Fissures
Surtsey
Neovolcanic
e

zone
dg
Ri

Atlantic Ocean Glaciers


ic
nt
tla
-A
id
M

Fig. 1.4. Location of the neovolcanic zone (<7 Ma


volcanism) in the central part of Iceland. Hotspot volcanism The global importance of divergent boundary
centered along the mid-Atlantic ridge has led to the volcanism can be appreciated by considering the
emergence of Iceland above sea level. Modified from annual budget of volcanic activity on Earth. Esti-
Gudmunsson (1996). mates of the total annual volcanic output from
all volcanoes is about 4 km3 of magma. Of that,
2000 m and are often glaciated. Between these
3 km3 , or about 75%, can be attributed to the
centers are zones of fissures (rift zones) where
mid-ocean system (Crisp, 1984). Thus, a large frac-
magma can travel laterally from storage areas
tion of the global volcanic activity occurs at great
beneath the central volcanoes. In eastern Africa
depth in the ocean, out of view.
a nascent divergent boundary has created a 4000-
km-long rift zone that extends from the intersec-
tion of the Red Sea and the Gulf of Aden, south Subduction zone volcanism
to the latitude of Madagascar. This rift valley is Most of the documented volcanic eruptions occur
home to some of Africa’s largest volcanoes such along convergent plate boundaries, or subduction
Mt. Kilimanjaro and Mt. Kenya. zones, such as the ones that make up the Pacific
DISTRIBUTION OF VOLCANIC ACTIVITY: THE PLATE TECTONIC FRAMEWORK 7

the western Pacific and the Andean active conti-


nental margin of South America (Fig. 1.6). A cross-
section through the Marianas arc shows that sub-
duction of oceanic crust is occurring along a
steeply dipping Benioff Zone. The lack of great
thrust earthquakes during historical times sug-
gests that the mechanical coupling between the
downgoing and the overriding plate is relatively
weak. Volcanic activity in the Marianas occurs
as a narrow, curved array of individual volcanic
islands positioned about 120 km above the top
of the descending plate. The curvature of the
Fig. 1.5. Mount Rainier stratovolcano in Washington state,
island chain is merely a reflection of the geo-
elevation 4392 m. The volcano is dominanted by lava flows
and breccias which are mostly andesitic in composition.
metrical constraints of subducting a rigid plate
Explosive eruptions have also produced pumice and ash into the surface of a sphere. In detail, however,
deposits on the slopes of the volcano. Photograph by S. Carey. some island arcs show distinct linear segmen-
tation of volcanic centers along the subduction
trend (e.g., Marsh, 1979; Carr, 1984). Behind the
‘‘Ring of Fire” (Fig. 1.1). These eruptions are often line of volcanic islands in the Marianas is a series
explosive in nature, result in immediate threats of small oceanic basins called back-arc basins.
to human populations, and consequently attract These basins are inferred to form by spreading,
our attention. Many of the world’s most famous in a manner similar to mid-ocean ridges, but on
volcanoes such as Krakatau, Mt. Pelée, Mount St. a smaller scale (Karig, 1971). The Marianas arc
Helens, and Mt. Fuji are related to the subduc- is thus considered as an example of a typical
tion process. Their structure is often what we con- island arc involving the subduction of oceanic
sider as a ‘‘typical” volcano; a high, steep-sided crust beneath other oceanic crust.
mountain with a crater at the summit (Fig. 1.5). A cross-section through the west coast of
But as we’ve seen, the dominant and thus more South America shows a very different type of sub-
‘‘typical” type of volcanic landform is that occur- duction process (Fig. 1.6). Here, oceanic crust is
ring beneath the sea in the form of mid-ocean descending beneath thick continental crust at a
ridges. shallow angle. The frequent occurrence of strong
A striking feature of the global distribution thrust earthquakes suggests that, unlike in the
of subduction-related volcanism is that it’s highly Mariana arc, there is strong coupling, or resis-
asymmetrical, with most of it occurring around tance, to the subduction of the plate beneath
the rim of the Pacific Ocean (Fig. 1.1). The Atlantic South America. Volcanism occurs as a line of
Ocean has only two short segments of subduc- individual volcanoes that typically is located at
tion volcanism; the Lesser Antilles island arc at a position about 120 km above the top of the
the eastern boundary with the Caribbean Sea, descending plate. However, because the slope of
and the South Sandwich island arc off the south- the Benioff Zone is much shallower than in the
ern tip of South America. Other major subduc- Marianas, the location of the volcanic front in
tion zones occur in the Indian Ocean (Indonesian the Andes occurs further back from the trench
island arc) and the Mediterranean Sea (Aeolian (Fig. 1.6). Another important difference is that
and Hellenic island arcs). the Andean subduction zone exhibits compres-
Just as there were significant differences in sion in the area behind the volcanic front and
the morphology of divergent plate boundaries consequently this type of margin lacks the exten-
(Fig. 1.3) there are also marked differences in sional back-arc basins that are characteristic of
the types of convergent boundaries. Uyeda (1982) the western Pacific.
has suggested that there are at least two distinct The differences in these two modes of subduc-
types represented by the Mariana island arc in tion have been related to factors such as the age
8 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

volcanic front Fig. 1.6. Summary of the


Mariana-type
subduction differences between the
Mariana-type and Chilean-type

deep trench
old dense plate subduction modes based on the
back-arc basin model of Uyeda (1982).
retreating Mariana-type subduction involves
continent
steeply descending plates with few
tension great thrust earthquakes, whereas
plate subduction in the Chilean-type
zones is much shallower and

Zone
typically more seismically active.

Benioff
dipping
steeply

volcanic front

Chilean-type subduction

compression
shallow filled trench
advancing
continent

younger buoyant plate great thrust


shall
earthquakes ow d
ippin
g Be
nioff
Zon
e

of the subducting slab and the absolute motion of output of the Earth (Fisher and Schmincke, 1984).
the overriding plate relative to the trench (Molnar The volcanism associated with subduction is only
and Atwater, 1978; Uyeda and Kanamori, 1978). 15%, and thus, even though subduction volcan-
Subduction of the Mariana type is thought to take ism poses the largest threats to the human popu-
place when old dense seafloor descends beneath lation and is the most conspicuous type of activ-
oceanic crust and the overriding plate is moving ity, it still represents a relatively small proportion
away from the active trench. In the Andean type of the annual volcanic production.
of subduction the descent of the plate is more
difficult because it is younger and more buoy-
ant. The absolute motion of the overriding plate Intraplate volcanism
is also towards the trench, resulting in compres- There are many areas of the Earth’s surface
sion behind the volcanic front. where volcanism is occurring that is unrelated
Subduction is a process of recycling where to a major plate boundary. These can be found
oceanic crust is returned to the mantle. The size both in the ocean basins and on the continents.
of the Earth has remained relatively constant Focused, persistent areas of volcanic activity such
and therefore the production of new seafloor at as these that remain active for millions of years
the mid-ocean ridges must be balanced by the are known as hotspots, or mantle plume vol-
subduction of crust at convergent margins. As canism (Wilson, 1973). They are areas of anoma-
discussed earlier, volcanism at mid-ocean ridges lously high volcanic production that are related
accounts for about 75% of the annual volcanic to the rise of mantle plumes from deep within
DISTRIBUTION OF VOLCANIC ACTIVITY: THE PLATE TECTONIC FRAMEWORK 9

Fig. 1.7. Location of the major volcanic hotspots (solid Formation of this large-scale feature is attributed
circles) in the ocean basins and on the continents. Modified to hot spot volcanism currently located beneath
from Wilson (1989). the southeast flank of Hawaii. During the past
70 million years, volcanic activity from this
hotspot has built a series of islands that are con-
the Earth. There are currently between 50 and tinually moved to the northwest by the motion
100 active hotspots, depending on the factors of the Pacific plate. This is confirmed by the
that are used to define them (Fig. 1.7). Common systematic increase in island age from the south-
features include voluminous outpourings of east to the northwest along the chain (Clague and
basaltic magma, higher than normal heat flow, Dalrymple, 1987). As the islands age, they sub-
thinning of the overlying crust, and development side and are eroded below sea level. A pronounced
of a broad topographic high. shift in the orientation of the Hawaiian–Emperor
Because the source of magma for hotspots seamount chain occurred about 44 Ma ago when
is derived from a long-lived source beneath the the motion of the Pacific plate changed to a
lithosphere, plate movement over millions of more westerly direction of motion (Fig. 1.8). The
years results in the formation of linear volcanic most recent island associated with the Hawaiian
features. Hotspots are thus extremely useful for hotspot is Loihi, now forming under water on
inferring the absolute motions of plates relative the southeast flank of the big island of Hawaii.
to the deep interior of the Earth. An excellent Other examples of linear chains of islands and
example of this can be found in the middle of seamounts includes the Tuamotu and Austral
the Pacific plate where the Hawaiian islands form groups of the South Pacific. These chains show
a distinctive linear group that trends from the a similar kink associated with the major change
southeast to the northwest. The Hawaiian islands in Pacific plate motion.
are actually a small part of a much larger sub- In the ocean basins some hotspots coincide
marine linear feature that includes the Hawai- with divergent plate boundaries. Iceland is an
ian ridge and Emperor seamount chain (Fig. 1.8). example of a hotspot that is centered along the
10 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

160o 180o 160o 140o 120o

Aleutian Islands North


America
50o 50o
Empe
ror Se

Suiko (64 Ma)


PACIFIC OCEAN
40o 40o
amou

Yuryaku (44 Ma)


nts

30o Midway (28 Ma)

Haw Nihoa (8 Ma)


aiia
nR
idge
20o o 20o
160 180o 160o Hawaii 140o 120o

Fig. 1.8. Orientation of the Hawaiian ridge and Emperor which there are no modern analogues. Many
seamount chain in the Pacific Ocean. Ages of various islands continents and ocean basins contain widespread
are shown in parentheses. Note the change in alignment of
accumulations of basaltic lava flows that repre-
the two seamount chains at Yuryaku, corresponding to an age
sent massive outpourings of magma. These are
of 44 Ma. Modified from Clague and Dalrymple (1987).
collectively referred to as large igneous provinces,
or LIPs (Coffin and Eldholm, 1994). Included in
mid-Atlantic ridge axis. The excess volcanic out- this category are continental flood basalts and
put of the hotspot results in the elevation of submarine basalt plateaus (Fig. 1.9). The Deccan
the ridge above sea level and the formation of Traps of India is an example of one of the largest
elevated submarine ridges on each side of the continental flood basalt provinces on Earth.
Iceland Plateau. Other hotspots are located When it was formed 65 millions years ago it may
beneath the continents, especially Africa (Fig. 1.7). have covered an area of 1.5 million km2 with an
Plate motion of continental masses can also pro- average thickness of at least 1 km. Such provinces
duce linear traces of past volcanic activity like are noteworthy because of the large areas affected
those in the ocean basins. For example, volcanic by volcanism and the relatively short times of
activity beneath Yellowstone National Park in formation. An estimate for the emplacement of
the western United States has been attributed to the Deccan Traps basalts is of the order of 1 Ma
the location of an active hotspot. A belt of vol- (Duncan and Richards, 1991; see also Chapter 7).
canic rocks ranging from Recent to 15 Ma old Such durations imply volumetric discharge rates
(Eastern Snake River Plain) can be traced from that are up to an order of magnitude larger than
Yellowstone back across Wyoming into Idaho. those of modern volcanic provinces, such as mid-
This is likely to represent the record of volcan- ocean ridges. As a result of the high eruption
ism as the North American plate migrated to the rates this type of activity does not tend to pro-
southwest (Greeley, 1982). duce high individual volcanoes, but instead, erup-
tions from many vents leads to the accumulation
Large igneous provinces (LIPs) of laterally extensive flows into a thick plateau.
Much of the previous discussion has centered The Ontong–Java plateau, located in the western
on describing the current distribution of vol- Pacific (Fig. 1.9) is an example of a large subma-
canic activity on the Earth’s surface. However, rine flood basalt province that was formed 124 Ma
there are examples of spectacular volcanic activ- ago. It is three times the area of the Deccan Traps
ity that have occurred in the geologic past for and was produced in about 3 Ma.
COMPOSITION AND PHYSICAL PROPERTIES OF MAGMA 11

Fig. 1.9. Location of some of the major large igneous on Earth are silicate liquids whose composition,
provinces (LIPs) NAVP, North Atlantic Volcanic Province; including volatile content, viscosity, and tem-
Dec, Deccan Traps; Kerg, Kerguelen; Elan, Elan Bank; Brok, perature can vary considerably depending upon
Broken Ridge; Piga, Pigafetta Basin; Naur, Nauru Basin; Onto, the nature of the eruption and the tectonic set-
Ontong–Java plateau; Mani, Manihiki; Hawaii, Hawaiian islands;
ting of the volcano. Magmas are complex mix-
Col Riv, Columbia River. From Coffin and Eldholm (1994).
tures of melt, suspended crystals, and gas bub-
bles on approaching the surface. Melt is usually
The origin of LIPs remains controversial and the major component and magma consequently
involves models based on deep mantle plumes, behaves as a fluid. Most magmas, with the excep-
shallow mantle plumes, and lithospheric rifting. tion of some rare carbonatite magmas, have SiO2
Deep mantle plumes may initiate massive out- contents in the range of about 45 to 77 wt.%.
pourings of lavas by actively impinging upon the Other major elements include aluminum, cal-
base of the lithosphere, causing thinning, heat- cium, iron, magnesium, potassium, sodium, and
ing, and melt production (Richards et al., 1989). titanium. The compositional diversity observed
The initial phase of volcanism (LIP formation) is in magmas is caused primarily by variations in
the most voluminous because the arriving plume the composition of source rocks being melted,
consists of an enlarged head followed by a long, differences in the amount of source rock melt-
thin conduit. Such plumes may originate near ing, variations in the amount of volatiles present
the core–mantle boundary. Alternatively, passive in the source regions, crystallization of magma
thermal plumes from layered mantle convection during its ascent and storage in the crust, mix-
may generate large melting events when litho- ing with other magmas, and contamination by
spheric extension allows for adiabatic decompres- country rocks (Wilson, 1989).
sion of hot mantle (White and McKenzie, 1989). Unraveling the details of these processes falls
in the realm of igneous petrology and exten-
sive treatment of these processes is beyond the
Composition and physical properties scope of this chapter. Instead we will focus on
of magma some basic details of magma composition as they
relate to the behavior of volcanic systems. As a
Composition of magmas starting point we will use a simple classification
Any discussion of how volcanoes work needs scheme for magmas based on SiO2 , Na2 O, and
first to consider the essential component of vol- K2 O (Fig. 1.10). This classification allows for the
canism, magma. Virtually all magmas erupted assignment of general names to different magma
12 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

Table 1.1 Major element composition of some common volcanic rocks and nodules

Mid-ocean ridge Peridotite


Oxidea basalt (MORB) Andesite Dacite Rhyolite (garnet lherzolite)

SiO2 50.58 56.86 66.36 74.00 45.89


TiO2 1.49 0.88 0.58 0.27 0.09
Al2 O3 15.60 17.22 16.12 13.53 1.57
FeO 9.85 4.26 2.41 1.16 6.91
Fe2 O3 – 3.29 2.39 1.47 –
MgO 7.69 3.40 1.74 0.41 43.46
CaO 11.44 6.87 4.29 1.16 1.16
Na2 O 2.66 3.54 3.89 3.62 0.16
K2 O 0.17 1.67 2.22 4.38 0.12
a
Oxides are expressed as weight percentages.
Source: Data from Wilson (1989) and Le Maitre (1976).

Ultrabasic Basic Intermediate Acid Fig. 1.10. Classification of volcanic rocks based
16 on Na2 O + K2 O versus SiO2 content. Modified
from Wilson (1989).
14
Phonolites
eli litic
es
Na2O + K2O (wt.%)

12
nit
ph no

Trachytes
Ne Pho

10 Benmoreites

tes
lts s
ite
tes

sa Rhyolites
s

ari hyba es
ite

8 ge
hri

d
u c
an
lin

a
ep

M r
T
hy
e

c
dt

Tra
ph

Hawaiites
an

6
Ne

Dacites
es
nit
sa

andesites

Andesites
Basaltic
Ba

4 Basalts

40 50 60 70
SiO2 (wt.%)

compositions, although it should be emphasized Magmas typically occur as distinct compo-


that there is often a complete gradation of sitional suites that suggest a genetic relation-
magma compositions. Broad categories include ship between the members. The suites consist
silicic (>63% SiO2 ), intermediate (52–63% SiO2 ), of compositions that exhibit relatively smooth
basic (45–52% SiO2 ), and ultrabasic (<45% SiO2 ). trends in the various oxide components such as
The most common type of magma is basalt SiO2 , FeO, MgO, CaO, and K2 O. Primitive mem-
erupted along ‘‘normal” segments of the mid- bers of a suite are closer to the primary magma
ocean ridge system where mantle plumes are not composition derived from the source regions
influencing the composition of the melts. This so whereas evolved members have undergone dif-
called mid-ocean ridge basalt, or MORB, makes ferentiation, usually as a result of crystalliza-
up a large fraction of the seafloor (Table 1.1). tion. Evolved members of a suite typically have
COMPOSITION AND PHYSICAL PROPERTIES OF MAGMA 13

higher contents of SiO2 , K2 O, and Na2 O. Different


suites are associated with specific geological envi-
ronments and styles of eruption. At mid-ocean 8.0
ridge spreading centers, magmas commonly form
part of the tholeiitic suite, a group of composi-

lite
yo
tions that exhibit a strong increase in FeO at the 6.0

Rh
Pressure (kbar)
more primitive end with a subsequent decrease

site
towards more evolved compositions. Although

de
lt
sa
An
the suite can include compositions that range up

Ba
to >70% SiO2 , these solicic magma types are vol- 4.0
umetrically minor. In contrast, subduction zone
volcanism is dominated by the calcalkaline suite.
This suite spans the compositional range basalt,
andesite, dacite, and rhyolite, or a SiO2 content of 2.0
about 50–77% (Table 1.1) and is characterized by
a lack of strong iron enrichment in moving from
the primitive to evolved members. Unlike mid-
ocean ridges, subduction zones exhibit a broader 0 4.0 8.0 12.0
range of magma compositions with more evolved
Dissolved H2O (wt.%)
magmas, such as andesites, being volumetrically
dominant. Fig. 1.11. Variation in dissolved water content as a function
Hotspot volcanism is dominated by basaltic of pressure for basalt, andesite, and rhyolite magma. Modified
magma but it differs in many respects from the from Burnham (1979).
voluminous magma erupted along the normal
mid-ocean ridge spreading centers. In particu-
lar, the trace element and isotopic signature of Volatile contents in magmas vary consider-
plume magmas suggests derivation from source ably from one tectonic environment to another.
regions that are different from that feeding At divergent boundaries the basalts of mid-
the normal mid-ocean ridge segments (Wilson, ocean ridge systems are generally volatile-poor
1989). (H2 O <1%, CO2 <0.10%) and erupt quiescently.
Magmas typically contain small amounts of Degassing in this environment is also inhibited
volatile components such as water, CO2 , sulfur, by the pressure of the overlying water column.
and halogens, such as Cl and F. These volatiles At depths of 2000 m the pressure is 200 bar, or
are dissolved in the magma and are typically enough to retain 0.1% H2 O and 0.1% CO2 dis-
more soluble at higher pressure. Their solubility solved in the magma. In contrast, subduction
is also a function of temperature, melt composi- zone magmas are noted for their higher volatile
tion and oxygen fugacity. Figure 1.11 shows the contents, especially magmas with SiO2 contents
solubility of water in magmas of different com- in excess of 60%. Water is the most abundant
position as a function of pressure. At a similar volatile in these types of magmas and provides
pressure, more evolved magmas such as rhyo- the driving force for the explosive volcanism
lites can dissolve larger amounts of water than that is so characteristic of the subduction zone
less evolved magmas like basalts. As magmas environment. The dacite magma erupted dur-
rise towards the surface and are erupted, the ing the May 18, 1980 Mount St. Helens eruption
dissolved gases can become supersaturated and in Washington, USA contained about 4.5% H2 O
come out of solution as bubbles. This degassing (Rutherford et al., 1985; Rutherford and Devine,
process has a fundamental influence on the style 1988) and the Bishop Tuff magma of California
of volcanic eruptions and plays a major role in showed H2 O contents as high as 7.0% (Anderson
the production of explosive volcanism. et al., 1989). Even some basic magmas in back-arc
14 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

environments have been inferred to contain up


to 4% H2 O (Sisson and Grove, 1993; Stolper and Bingham
σo
Newman, 1994).
Sulfur is an important volatile component in

tic
las
magmas because of its potential climatic effects y)
sit

-p
o
isc

do
once it is released into the atmosphere (Rampino,

Shear stress
v

eu
1988; Sigurdsson, 1990; see also Chapter 5). Like high

Ps
ia n(
H2 O and CO2 its solubility is dependent upon ton
pressure but it is also strongly a function of the New )
osity
w visc
FeO content of the magma and oxygen fugacity n (lo
tonia
(Carroll and Rutherford, 1988). It is dissolved as New
sulfide under relatively reducing conditions, and
as sulfate as magmas become more oxidized. In
general there is an inverse correlation between Strain rate
the SiO2 content of magma and the amount of Fig. 1.12. Relationship between shear stress and strain rate
sulfur that can be dissolved. Basalts, for example, for Bingham, pseudo-plastic, and Newtonian fluids. Modified
can contain up to to 1000 ppm (parts per mil- from Wolff and Wright (1981).
lion), whereas rhyolites are generally poor in sul-
fur, with only about 20 ppm. Some magmas, such
as those erupted during the 1982 eruption of El of shear stress to strain rate (Fig. 1.12). Some flu-
Chichón volcano in Mexico and the 1991 erup- ids exhibit a linear relationship between shear
tion of Mt. Pinatubo, have sufficient sulfur and stress and strain rate that passes through the
are oxidized enough to allow the mineral anhy- origin on Figure 1.12. These are referred to as
drite (CaSO4 ) to crystallize as a phenocryst. Newtonian fluids and they will flow even with
only an infinitesimally low shear stress. In gen-
Physical properties of magmas eral, most magmas do not behave as Newtonian
The physical properties of magma play a funda- fluids unless they are at high temperatures or
mental role in determining the style of a volcanic are relatively crystal-free. Fluids which show a
eruption. Composition is an important factor but non-linear relationship between stress and strain
other parameters such as temperature, propor- rate are described as non-Newtonian (pseudo-
tion of crystals, amount of dissolved volatiles, plastic behavior). Many magmas actually behave
and the abundance of gas bubbles all contribute as Bingham fluids (Fig. 1.12). Such fluids exhibit
to determining the rheology of the erupting an intercept with the stress axis at zero strain
magma. Temperature is a relatively simple param- rate and a linear variation of stress and strain
eter to measure and significant amounts of data rate. The intercept represents a minimum shear
are available for different magma types. There stress that must be exceeded before a fluid will
is typically an inverse correlation between erup- begin to flow. This parameter is called the yield
tion temperature and SiO2 content. Basaltic mag- strength and in magmas results from the pres-
mas (50% SiO2 ) erupt at temperatures of about ence of crystals, bubbles, and changes in viscos-
1200 ◦ C, whereas silicic magmas such as rhyolites ity caused by cooling during eruption. The slope
(75% SiO2 ) are cooler and erupt in the tempera- of a line connecting the origin with a point on a
ture range of 700–900 ◦ C. rheological curve is called the apparent viscosity.
One of the most important magmatic proper- The viscosity of silicate melts varies as a
ties that influence the nature of volcanic erup- function of temperature and composition (e.g.,
tions is magma viscosity. In simple terms it is Bottinga and Weill, 1972; Ryan and Blevins, 1987).
the amount of internal resistance to flow that Silicate melts show hyperexponential depen-
a fluid exerts when a force is applied to it. The dence of viscosity on temperature, with increas-
lower the viscosity, the more easily a fluid can ing temperatures resulting in significant viscosity
flow. Specifically, viscosity is the slope of the ratio reduction. The effect of composition is complex
MELTING MECHANISMS FOR MAGMA GENERATION 15

but in general viscosity is inversely related to 11 4 a


Basalt

log10 viscosity (Pa s)


SiO2 content at normal eruption temperatures.
3
Basaltic melts have relatively low viscosities (102 80

to 103 Pa s) compared to silicic magmas such as 9 100
2 0°
rhyolites (106 to 1012 Pa s). The presence of crys- 120

tals and gas bubbles have important effects on 1
1400°

log10 viscosity (Pa s)


magma rheology. As crystal abundance increases,
7
viscosity increases according to a power law
0 5 10
(Pinkerton and Stevenson, 1992). The transition
H2O content (wt.%)
from Newtonian to non-Newtonian properties
can develop when the volume of suspended crys- 5
tals reaches between 20% and 30%. In contrast,

80

the effect of bubbles is more complicated and

10
00
12
can result in either viscosity increase or decrease

o
00
o
depending on factors such as bubble size, surface 3

14
00
o
tension of the melt, and strain rate (Dingwell and
Webb, 1989).
Dissolved volatiles in magmas, such as water 1
Rhyolite
and CO2 , can significantly impact magma vis-
cosity even though their mass fraction is rela- 0 4 8 12
tively small. The effects are due to the ways in H2O content (wt.%)
which some of these components are held in solu-
tion. For example, water is dissolved in magma Fig. 1.13. Variation in magma viscosity as a function of
through the breaking of strong Si O bonds and dissolved water content for rhyolite and basalt magma at
temperatures from 800 to 1400 ◦ C. Modified from Williams
the formation of OH− . This causes a reduction
and McBirney (1979).
in the internal resistance to flow and thus a
reduction in viscosity (Fig. 1.13). If 4% H2 O is
added to an initially dry rhyolite the viscosity
by volume. In contrast, high-viscosity magmas
will decrease by five orders of magnitude. Addi-
react to cooling by forming supercooled melts
tion of H2 O to basalt also decreases the viscosity,
and glasses. The glass transition temperature rep-
but the magnitude of change is considerably less
resents the boundary between ductile and brit-
(Fig. 1.13). As magmas approach the surface dur-
tle behavior. It is dependent on melt viscosity
ing an eruption they typically lose volatiles as
and shear rate (Dingwell and Webb, 1990). Thus,
a result of degassing induced by pressure reduc-
siliceous magma in a slow lava flow may be
tion. Figure 1.13 shows that a consequence of
able to behave as ductile viscous fluids at much
the degassing process is a dramatic increase in
lower temperatures than in an explosive eruption
the viscosity of magma, especially one with high
where strain rates are much higher.
silica content.
During eruptions, magma cools and devel-
ops mechanical strength. How magmas react to
cooling plays an important role in the way they Melting mechanisms for
respond to stresses as the eruption proceeds. Two magma generation
end member responses to cooling can be consid-
ered. Low-viscosity magmas, such as basalts, gen- The generation of magma occurs deep within
erally react to cooling by crystallization. If crystal the Earth and thus out of the range of direct
content increases to an extent that the crystals observations. This process involves the melting of
form a touching framework, then it becomes a solid material (source rocks) to yield a melt phase
partially molten rock rather than a magma. This that segregates and rises to the surface buoyantly.
transition typically occurs at around 50% crystals Our understanding of the melting process comes
16 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

primarily from the study of the thermodynamic of low velocity in the upper mantle may repre-
properties of magma and potential source rocks, sent a small proportion of interstitial melt, but
together with the experimental determination this is unlikely to represent a major source of
of the melting behavior of geological materials magma for volcanism at the surface. The lack of
under conditions appropriate for the Earth’s inte- large melted zones in the mantle indicates that
rior. The important questions about magma gen- the interior temperature must be below the melt-
eration relate to the nature of the source rocks ing temperature, or solidus, of mantle material.
and the conditions that are necessary to cause Magma generation must therefore involve the
melting. Laboratory experiments have demon- melting of essentially solid source rocks at depth.
strated that the melting of rocks consisting of The temperature within the Earth increases with
more than one mineral occurs over a range of depth and is the result of energy accumulated
temperatures. As temperature is increased, first during initial accretion of the planet, formation
melting begins at the solidus and complete melt- of the core by iron segregation, and radioac-
ing takes place above the liquidus. Within this tive heating resulting from the decay of radioac-
temperature range both the amount of melt and tive elements (Frankel, 1996). Energy is dissipated
its composition vary. For the most part, magmas from the Earth’s interior to the surface by both
erupted at the surface are generated within the conduction and convection. In the mantle, the
Earth’s mantle, although some are produced by temperature increases to close to 3000 ◦ C near the
melting of the crust. boundary with the core and the lack of extensive
The composition of the mantle has been melting reflects the increase in melting tempera-
inferred primarily from observed seismic veloc- ture of mantle material at high pressures.
ities and the composition of nodules, thought There are essentially three ways to initiate
to represent direct samples of the mantle that melting of source rocks: (1) increase the temper-
are brought to the surface during volcanic erup- ature, (2) decrease the pressure, or (3) change
tions. Detailed geochemical and isotopic stud- the composition. It would seem most logical to
ies of magmas erupted from divergent bound- suppose that melting is generally caused by an
aries and intraplate volcanism indicate that the increase in temperature within a region of the
source rocks in the mantle are heterogeneous Earth’s interior. In fact, this is not likely to be an
on a variety of scales resulting from the recy- important way in which the majority of magmas
cling of material back into the mantle at sub- are formed. The reason for this is the difficulty in
duction zones and previous melting episodes in developing enough heat locally to induce melting
Earth history that have depleted (i.e. impover- of source rocks. Focusing of most volcanic activity
ished) certain areas in melt. In general, the com- at plate boundaries indicates that the dynamics
position of the mantle is ultramafic (rocks high of plate motion plays a fundamental role in the
in magnesium and iron; see Table 1.1) and a melting mechanism.
likely source rock is known as peridotite (Yoder,
1976). The mineralogy of peridotites varies as a Magma generation at mid-ocean ridges
function of pressure (depth) as different mineral Considering that mid-ocean ridge volcanism
phases attain equilibrium. At the depths likely accounts for 75% of the annual volcanic pro-
to correspond to melting zones, the peridotites duction, the generation of magma beneath the
consists of olivine ((Mg,Fe)2 SiO4 ), orthopyroxene ridges represents the most important mechanism
((Mg,Fe)SiO3 ), clinopyroxene ((Ca,Mg)Si2 O6 ), and for melt generation on the Earth. The geomet-
garnet (MgAl2 Si9 O12 ). At shallower levels, the ric configuration of a divergent boundary con-
aluminum-bearing phase garnet is replaced by sists of two lithospheric plates moving apart with
spinel (MgAl2 O4 ) and then plagioclase (NaAlSi3 O8– new seafloor created in the middle. This pat-
CaAl2 Si2 O8 ). tern of motion requires upwelling of material
Seismic studies indicate that except for the from depth in the mantle to replace material
molten outer core, the greater part of the Earth’s moving laterally away from the ridge. Hot man-
interior, i.e. mantle, is in the solid state. A zone tle is brought into a lower pressure regime and
MELTING MECHANISMS FOR MAGMA GENERATION 17

48–70 km depth, when the temperature is equal


to the solidus. Continued rise of this material
results in a mixture of melt and residual solid,
where the fraction of melt is indicated by the
dashed lines. Material initially starting at 1300 ◦ C
arrives at the surface with a temperature of about
1200 ◦ C and the melt fraction would be about
20%. The rapid cooling of the material once it
intersects the solidus (∼4 ◦ C/km) is caused by the
loss of heat of fusion necessary to continue the
melting process as the material decompresses.
The extent of melting and the temperature of
the melt at the surface is a function of the start-
ing temperature when the mantle material inter-
sects the solidus. Thus, material derived from
deeper depths and higher temperatures has the
potential for generating more melt at higher
temperatures by the time it reaches the sur-
face. This decompression melting mechanism is
shown schematically in the context of a divergent
boundary (Fig. 1.14b). Below the ridge axis, ris-
ing mantle material experiences decompression
melting as it rises, with the formation of a par-
tial melting zone. The degree of melting is largest
for material in the center of the upwelling, as it
has the potential for the maximum decompres-
sion. Upwelling material follows flow lines that
Fig. 1.14. (a) Paths of decompression melting for rising become parallel with the surface and track the
mantle peridotite (lines with arrowheads) at temperatures lateral motion of the plates. Segregation of melt
from 1300 to 1500 ◦ C. Melting begins at the solidus in the occurs within the partial melting zone, and the
depth range of ∼50–75 km. Dashed lines indicate the fraction melt makes its way to the surface where it forms
of melt generated at a given pressure and temperature new ocean crust (Fig. 1.14b).
between the solidus and liquidus. (b) Schematic
Experimental melting of mantle peridotites
representation of the partial melting zone generated beneath
indicate that it is possible to generate a magma
a mid-ocean spreading center by decompression melting of
rising mantle peridotite. Modified from Brown et al. (1992). that resembles typical mid-ocean ridge basalt by
partial melting of the order of 20%. A 6–8-km-
thick oceanic crust of basaltic composition could
can undergo melting, referred to as decompres- thus be produced by upwelling and decompres-
sion melting (Fig. 1.14a). The solidus curve of sion melting of mantle material initially at a
mantle material is seen to increase in temper- temperature of about 1350 ◦ C (Sparks, 1992).
ature with increasing depth or pressure. At all
depths it is above the local temperature and melt- Magma generation at subduction zones
ing is not predicted to take place. However, if The association of volcanism with plate subduc-
mantle material from depth is physically trans- tion presents something of a paradox when con-
ported towards the surface without losing sub- sidering the generation of magma. In these areas,
stantial heat, i.e. adiabatically, it can intersect the cold, dense seafloor is being returned to the man-
solidus and begin to melt. Ascent paths are shown tle by gravitational sinking and the production
in Figure 1.14a for mantle material initially at of magma appears counter-intuitive. Modeling
1300, 1400, and 1500 ◦ C. Melting occurs at about of the temperature distribution in subduction
18 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

Fig. 1.15. Temperature distribution induced by the


subduction of a lithospheric plate into the mantle. Modified Fig. 1.16. Solidus curves for the melting of mantle
from Wilson (1989). peridotite under wet (H2 O-present) and dry conditions.
Heavy dashed lines show the geothermal gradient in the
mantle wedge beneath a subduction zone. Melting of
zones, indeed, shows that the mantle is cooled by peridotite is possible at subduction zones under wet
the descent of the slab (Fig. 1.15). A clue to solv- conditions at depths of approximately 75–120 km (gray
ing this apparent paradox lies in the types of mag- shaded zone). Modified from Wilson (1989).
mas erupted. As pointed out earlier, an important
difference between the magmas erupted at diver-
gent and convergent boundaries is that magmas observation that most subduction zone volcanoes
in subduction zones are much richer in volatile are located between 100 and 150 km above the top
components, such as water and carbon dioxide. of the descending slab.
Thus it is necessary to evaluate the role of these If water is of fundamental importance to sub-
components in the melting process. duction zone magma genesis, what is its origin?
Figure 1.16 shows the solidus of mantle mater- Two possibilities are that it is present in the
ial with no volatiles (dry) in the context of the upper mantle in general, or it is introduced into
inferred geothermal gradient that exists beneath upper mantle through the process of subduc-
a subduction zone. At all depths the tempera- tion. If the former were true then the influence
ture along the geothermal gradient is less than of water would be evident at mid-ocean ridge
the dry solidus for mantle material and melting spreading centers as well. However, the volatile-
is not predicted. Decompression melting could poor nature of MORBs suggests that high volatile
occur if material is convected upwards, but the contents in the mantle are a direct result of
plate motions are not conducive to the sustained the subduction process. As new seafloor is cre-
large-scale upward movement of mantle material. ated at mid-ocean ridges it immediately begins
Also shown on Figure 1.16 are the solidus curves to cool and subside. Fracturing of the crust
for peridotite in the presence of water and car- allows for the penetration of seawater to deep
bon dioxide. The principal effect of these compo- levels where chemical exchanges occur between
nents is to dramatically reduce the melting tem- hot rock and water. This hydrothermal circula-
perature at all pressures and depths. If water is tion is an effective mechanism for enhancing
available, then the geothermal gradient beneath heat loss from mid-ocean ridge spreading cen-
subduction zones can intersect the solidus and ters and sustains an exotic biological commu-
melting can take place at depths in the range nity that is based on chemosynthesis (Humphris
80–120 km. This prediction is in accord with the et al., 1995). One of the principal effects of this
RISE AND STORAGE OF MAGMA 19

process is the hydration of the oceanic crust by


the conversion of anhydrous minerals to hydrous
minerals such as serpentine (Mg3 Si2 O5 (OH)4 ) and
amphibole ((Na,Ca)2 (Mg,Fe,Al)5 (Si,Al)8 O22 (OH)2 ) at
different depths. In addition, low temperature
alteration of the ocean crust takes place for mil-
lions of years as the plates move away from the
spreading centers. The result is that when the
plates are recycled into the mantle at subduc-
tion zones they are carrying with them water
and other components that are bound in certain
mineral phases. However, during the subduction Fig. 1.17. Schematic representation of the partial melting
process increases in pressure and temperature of rising mantle plume material beneath a hotspot volcano.
as the slab descends into the mantle cause the Melting occurs at greater depth in the middle of the rising
hydrous phases to become unstable and release plume because of the higher temperature. Modified from
their water. These dehydration reactions occur Brown et al. (1992).
over a range of depths from about 60 to 130 km.
Once released, the volatiles migrate upwards as a volcanism must also involve partial melting of
fluid phase into the overlying mantle wedge. The mantle peridotite. The higher production rate
influx of water and other volatiles into the man- can be accounted for by applying the decompres-
tle wedge may have two effects. First, the temper- sion melting model, used previously in the con-
ature regime in the area of volatile influx may text of normal mid-ocean ridges. In order to yield
be such that melting is possible by intersection larger amounts of melt by this mechanism it is
with the ‘‘wet” solidus of peridotite (Fig. 1.16). necessary to begin with hotter mantle material
Second, if temperature conditions are not con- (Fig. 1.14a).
ducive for ‘‘wet” melting, then the introduction Deep in the mantle, density differences
of volatiles may reduce the density of the mantle caused by compositional heterogeneity or other
wedge sufficiently to cause diapiric uprise of factors trigger diapiric uprise of hot material.
material. Melting could then occur by decom- Based on the observation that hotspot volcan-
pression when the wet solidus is intersected ism is not uniform over time, but instead occurs
during the ascent path. Whatever the ultimate as a series of pulses, it is likely that the uprise
mechanism, the solution to the subduction zone of material occurs as a stream of discrete blobs.
paradox appears to lie with the recycling of water Figure 1.17 shows the development of a par-
and other components back into the mantle and tial melt zone beneath a hotspot caused by the
their effect on the melting behavior of mantle upwelling hot mantle material. Melting is max-
peridotite. imized in the center of the upwelling plume
because this is the hottest region and conse-
Magma generation at intraplate hotspots quently it begins to melt first.
Hotspot, or intraplate, volcanism appears to be
largely decoupled from the motion of the litho-
spheric plates, implying that magma generation Rise and storage of magma
is not dependent upon the upper-level dynamics
of plate interactions. When a hotspot is coinci- Ascent of magma
dent with a mid-ocean ridge spreading center, Virtually all magma is generated by the partial
such as is the case with Iceland (Fig. 1.4), the melting of source rocks, either in the upper man-
net volcanic production is high and an elevated tle or crust. In order for magma to be erupted it
portion of the ridge is built, yet the composi- must separate from the residual source material
tion of magmas produced is broadly similar to and make its way towards the surface. The main
sections of the adjacent ridge. Therefore, hotspot driving force for the rise of magma is buoyancy
20 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

associated with the difference in density between the equivalency in the effective, large-scale in situ
melt (2300–2500 kg/m3 ) and source rock (∼2700– magma density and the density of the surround-
3200 kg/m3 ). During its rise to the surface, ing country rock. Under these conditions there is
magma will lose heat to the surrounding rocks no longer any driving force causing the magma
and begin to crystallize. If the ascent rate is too to rise and it will stagnate at some level beneath
slow, the amount of crystallization may be suf- the surface.
ficient to freeze the material en route, forming For neutral buoyancy to be the controlling fac-
an intrusive body. Observations of erupted mag- tor in magma chamber formation there must be
mas suggest that more than 55% crystallization some level beneath a volcano where the density
renders a melt ‘‘uneruptible,” owing to the very of the country rock is equal to the density of the
high bulk viscosity as crystal content is increased ascending magma, and additionally, that above
(Marsh, 1981). One way for magmas to reach the this level the density of the country rock is lower
surface is for ascent to take place along previous than the magma density. At first this appears
magma paths in order to take advantage of the to be counter-intuitive because if a volcano is
elevated temperatures of the surrounding rocks. constructed out of solidified magma, shouldn’t
Thus, volcanic centers tend to focus their activ- it be everywhere denser than magma? In fact,
ity by establishing paths of minimum thermal there are large variations in the density of vol-
resistance. canic rocks that make up the upper portions of
Ascent of magma to the surface is invariably volcanoes. This results from variations in rock
a complex path that involves storage at differ- composition and from the extent of fracturing.
ent levels. Accumulation of magma within the Density generally increases with depth, but in
lithosphere takes place in magma chambers, or many cases exhibits a distinctive trend. From the
reservoirs. Deep reservoirs (tens of kilometers in surface to a depth corresponding to a pressure
depth) may be established in the zone of partial of about 200 MPa, density increases in a strongly
melting where upwelling mantle material under- non-linear fashion, whereas at greater depths the
goes decompression (Head and Wilson, 1992). The increase is broadly linear (Ryan, 1987). The non-
position of such reservoirs will be controlled by linear portion of the trend is attributed to reduc-
the thickness and movement of the lithosphere, tion in the macroscopic and megascopic pore
the transition in rheological properties of the space as pressure is increased and fractures are
mantle (i.e., brittle versus ductile), and existence progressively closed and sealed. At higher pres-
of phase changes encountered by rising mantle sures, the increase in density is accommodated by
material. Further ascent of magma to shallower adjustments in the structure of the constituent
levels may be instigated by the development of minerals.
smaller diapirs, as long as the magma density As magma rises into the upper levels of a vol-
is still less than the surrounding rocks and the cano and attains neutral buoyancy, it will stag-
rheological/thermal properties of the diapir allow nate and begin to spread out laterally within a
it to rise. neutral buoyancy horizon. This horizon is ex-
Direct eruption of magma from deep reser- pected to have limited vertical extent but a more
voirs may take place occasionally but it is more extensive lateral expanse. Over time magma will
common for magma to be stored in a shallow accumulate in this zone to form a chamber from
reservoir. Much more is known about the geom- which surface eruptions may be fed. In Hawaii,
etry and development of these storage areas and in situ density measurements and knowledge of
significant progress has been made during the melt density allow predictions of the location of
last decade in understanding the nature of shal- the neutral buoyancy horizon (Fig. 1.18). There is
low chambers in different types of volcanic envi- an excellent agreement between the predicted
ronments. Ryan (1987) has proposed that the cre- level of neutral buoyancy (2–4 km) and the
ation of shallow magma chambers is controlled inferred level of magma storage based on a mul-
by the principle of neutral buoyancy, defined as titude of geophysical observations (Ryan, 1987).
RISE AND STORAGE OF MAGMA 21

Fig. 1.19. Variation in the level of neutral buoyancy


between Loihi, Kilauea, and Mokuaweoweo volcanic centers
in the Hawaiian islands. Modified from Ryan (1987).

keep pace with the density adjustments of the


edifice. In essence, a shallow magma chamber
moves upwards to follow the absolute elevation
of the edifice. This is portrayed in Figure 1.19
where the neutral buoyancy horizons are
followed across volcanic centers of different ele-
vation in the Hawaiian islands. The net effect is
to keep the level of neutral buoyancy at a rela-
tively constant depth beneath the top of volcano.
Fig. 1.18. Variation of in situ density beneath Kilauea
This level will be a function of the dominant com-
volcano, Hawaii (shaded curve). Magma density of an olivine position of the erupted products, the nature of
tholeiite composition as a function of pressure shown as the eruptions, and the morphological structure
stippled bar. Neutral buoyancy region occurs where magma of the volcano.
and in situ density are equivalent. Modified from Ryan (1987). The principle of neutral buoyancy is expected
to operate in any volcanic system where magma
is rising from depth and encountering a decreas-
An important aspect of the neutral buoyancy ing density profile towards the surface. How-
horizon is that it is not a static level during ever, because the composition of magmas and
the growth of a volcano. As eruptions occur and stress regimes differ considerably between tec-
volcanic centers grow in elevation the density tonic environments at different plate boundaries,
profiles will adjust to the new overlying load. The it is expected that the geometry and location of
neutral buoyancy level will rise in an attempt to magma chambers should reflect these factors.
22 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

Fig. 1.20. Cross-axis and along-axis sections of the Magma chambers


compound magma chamber model for a fast-spreading Mid-ocean ridges
mid-ocean ridge system. Magma that is available for eruption For many years there was considerable debate
is stored as a thin melt lens (black area) that is segmented about the nature and existence of magma cham-
along the spreading axis at axial discontinuities. From Sinton
bers beneath mid-ocean ridge spreading centers.
and Detrick (1992).
Seismic experiments on the East Pacific Rise
and the mid-Atlantic ridge have now led to a
Magma chambers play a fundamental role in new model for reservoirs along divergent bound-
the way in which volcanoes erupt because they aries that takes into account differences in
act as the source reservoirs for eruptions at the spreading rates (Sinton and Detrick, 1992). At
surface. For example, the size of an eruption is fast-spreading ridges, such as the East Pacific
limited by how much eruptible magma is con- Rise, magma derived from decompression melt-
tained in a magma chamber and the style of ing of upwelling mantle material accumulates
the eruption can be related to the physicochemi- in an axial magma chamber that is located at
cal changes that have taken place in the magma between 1 and 2 km below the seafloor (Fig. 1.20).
since its arrival in the chamber. The existence of Geophysical evidence suggests that eruptible
magma chambers has been suspected for some melt is concentrated in a relatively thin (∼50 m to
time but it has only been recently that our view <1 km) lens that is about 1–2 km in width. This
has progressed significantly from the simple ‘‘bal- melt lens is located directly beneath the spread-
loon and straw” model often portrayed in intro- ing axis and overlies a more extensive zone of
ductory geology textbooks. Recent advances in crystal mush (small amounts of interstitial melt
the use of geophysical techniques to image sub- and crystals). The mush has a vertical extent of
surface magma chambers has led to new insights several kilometers and merges with a transition
into their shape, location, and evolution. zone consisting of hot, but essentially solid rock.
RISE AND STORAGE OF MAGMA 23

geophysical evidence for the existence of a steady-


state magma lens, like that imaged on the East
Pacific Rise. Instead, Sinton and Detrick (1992)
propose that a mush and hot rock zone is located
several kilometers below the axial rift valley
(Fig. 1.22). The supply rate of magma from depth
is insufficient to produce accumulations of melt
that can be drawn upon to feed surface erup-
tions. It is more likely that surface eruptions co-
incide with injection events of new magma into
the mush zone.

Fig. 1.21. Model for the development of axial magma Subduction zones
chambers by Rayleigh–Taylor instabilities in the zone of partial In contrast to the elongated nature of mid-ocean
melting beneath a mid-ocean ridge spreading center. ridges, volcanism at subduction zones is focused
Instabilities within the partial melt zone feed overlying magma at individual volcanoes and thus the geometry
chambers at regularly spaced intervals. Modified from
of the underlying magma chambers is expected
Schouten et al. (1985).
to be more analogous to the traditional view of
a spherical reservoir connected to the surface by
a conduit. Considerable effort has been made to
At fast-spreading centers, the melt lens and infer the nature of magma chambers beneath
mush zone are considered to be in a quasi- subduction zone volcanoes by a variety of geo-
steady state, and can easily feed new eruptions physical techniques such as seismic, gravity, mag-
to form ocean crust. Along the axis there are netic, and electrical techniques (e.g., Ryan, 1988;
variations in the relative volumes of melt mush Iyer et al., 1990; Barker and Malone, 1991). A
that appear to be coincident with morpholog- particularly useful technique is seismic tomogra-
ical discontinuities of the spreading boundary phy which relies on the attenuation of seismic
(Fig. 1.20). These differences may be related to waves (natural or artificial) by the presence of
along-axis variations in the supply of upwelling magma. The technique enables the generation of
magma diapirs beneath spreading centers (e.g., a three-dimensional image of magma chambers,
Whitehead et al., 1984; Schouten et al., 1985). although it is not possible to quantitatively infer
As upwelling mantle encounters overlying man- the amount of melt, or its composition. In the
tle of greater density and viscosity, gravitational Cascade subduction zone of the western United
instabilities with regular spacing are predicted to States three types of situation have been found
form (Fig. 1.21). These are analogous to Rayleigh– using this technique (Iyer et al., 1990): (1) no dis-
Taylor instabilities produced whenever a fluid of cernable magma chambers, (2) magma chambers
low density is placed beneath a fluid of higher with volumes between 200 and 1000 km3 , and
density. The spacing of the resulting diapirs is a (3) small chambers (few cubic kilometers) embed-
function of the thickness and rheological proper- ded in intrusions within the upper 5 km of the
ties of the two fluids. This process can therefore crust.
lead to the regular focusing of magma supply A good example of a well-studied subduction
to the shallow chamber beneath fast-spreading zone magma reservoir is the one beneath Mount
ridges and have a fundamental influence on St. Helens volcano in the western United States
the morphological signature of the accretion (Pallister et al., 1992). As a result of the 1980 explo-
process. sive eruption a large variety of geophysical, petro-
Geophysical observations at slow-spreading logical, and experimental data has contributed to
ridges, such as the mid-Atlantic ridge, indicate the definition of the system’s location and geom-
a major difference in the nature of the sub- etry. Its shape has been largely inferred from the
surface reservoir system. There is currently no location of an earthquake-free zone in the depth
24 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

Fig. 1.22. Model for the magma


chamber configuration beneath a
slow-spreading mid-ocean ridge center.
In contrast to fast-spreading ridges
there is no evidence for a long-lived lens
of melt beneath the axial valley. From
Sinton and Detrick (1992).

range of 7–15 km, where the presence of magma


precludes brittle failure (Fig. 1.23). The depth to
the top of the chamber based on seismic evi-
dence coincides well with petrologic determina-
tions of the pressure/temperature conditions of
the Mount St. Helens magma just prior to the
1980 eruption (Rutherford et al., 1985). Ascent
of magma during the 1980 eruption occurred
through a conduit that was of the order of 50 m
diameter (Carey and Sigurdsson, 1985; Scandone
and Malone, 1985).
An important point about subduction zone
magma chambers is their size relative to those
beneath the mid-ocean ridge system. At spread-
ing centers the chambers are relatively small and
the amount of eruptible magma per unit length
of ridge is only of the order of <0.5 km3 . In con-
trast, the chambers beneath subduction zone vol-
canoes can vary in size by several orders of mag-
nitude. At Mount St. Helens the inferred volume
of the reservoir system is about 10 km3 . During
the 1980 eruption only 0.5 km3 was erupted from
the chamber and thus the majority remained at
depth. Other eruptions of subduction zone vol-
canoes have discharged hundreds or even thou- Fig. 1.23. Inferred configuration of the magma chamber
sands of cubic kilometers of magma during sin- and conduit system beneath Mount St. Helens prior to the
gle eruptions so the volume of their chambers 1980 eruptions, based on geophysical and petrologic data.
Modified from Pallister et al. (1992).
may be 103 –104 km3 in size.
The establishment of shallow magma cham-
bers beneath subduction zone volcanoes is also centers may favor stabilization at deeper levels
likely to be governed by the principle of neutral beneath the surface.
buoyancy (Ryan, 1987). However, the dominance
of more evolved (higher silica), lower density Intraplate hotspots
rocks in the crust of subductions zone and the Perhaps in no other environment has the com-
subaerial expression of many of the volcanic plexity of magma chambers and their plumbing
STYLES OF VOLCANIC ACTIVITY 25

systems been revealed with more detail than at content, structure of the volcano, rate of magma
Hawaii, a prime example of an intraplate hotspot. discharge, and the environment of eruption, i.e.,
Decades of observations on eruptive activity and subaerial versus submarine.
seismic events at Kilauea volcano have been syn-
thesized to reveal the internal three-dimensional Effusive volcanism
system of magma storage (Ryan, 1988). In particu- One of the most common forms of volcanic activ-
lar, specific types of earthquake have been used to ity on Earth is the eruption of basaltic magma
infer the location and direction of magma move- as lava flows. Lava flows are produced where
ment beneath the volcano. Magma rises beneath magma issues from a vent quiescently or from
Kilauea through a primary conduit that extends high magma discharge lava-fountains and flows
from the upper mantle to a shallow reservoir. away from the source. These vents may be located
This conduit is an integrated zone of magma at the summit of a volcano, on the flanks, or adja-
ascent that is about 3 km in diameter. cent to the base. The geometry of the vent can
Magma is stored within a complex summit take two different forms: a central vent or a linear
reservoir that extends from 2 to 7 km depth and fissure. Lava flow behavior can be quite complex
whose formation is governed by neutral buoy- and is related to factors such as magma compo-
ancy. Based on inflation of the summit area, the sition, volatile content, crystal content, viscosity,
reservoir is envisioned to be a network of magma local slope of the terrain, and eruption rate.
veins separated by a boxwork of country-rock The principal driving force for the movement
blocks. Eruptions at Kilauea occur in the Kilauea of lava flows is gravity and thus movement will
caldera and along the East Rift zone. These erup- be constrained by the local topography. In order
tions are fed from the summit reservoir crown for flow to occur the internal viscous forces of
and from the summit reservoir base via the upper the magma must be overcome. If lavas were truly
East Rift zone pipe. It should be emphasized that Newtonian fluids they would flow on any slope,
this picture of the Kilauea system is an integrated even though it might be very slow. However, most
one built up from data collected during many lavas have a finite yield strength (Fig. 1.12) and
periods of activity and thus may not represent behave as Bingham fluids. Consequently, there is
the configuration of the system at any one time. a minimum amount of shear stress that needs
Nevertheless, it provides important insights into to be exceeded before flow will proceed. Yield
the complexities of magma ascent, storage, and strength is a result of the partial solidification
eruption in a large hotspot volcano. of magma as it migrates to the surface and
is erupted. It is a function of magma composi-
tion and the extent to which crystallization has
Styles of volcanic activity occurred.
For a given slope, the lava must be a certain
When magma erupts at the surface the discharge thickness in order for the weight to overcome the
can take two different forms. If the magma is finite yield strength. The thickness of the flow, t,
volatile-poor then it will remain intact and flow is given by
as a mixture of melt and crystals, eventually
τ
solidifying as it cools to ambient temperature. t= (1.1)
ρg tan α
This type of activity is referred to as effusive, or
quiescent. However, if the magma contains signif- where τ is the yield strength, ρ is the lava den-
icant amounts of dissolved volatiles, or if it comes sity, g the acceleration of gravity, and α is the
into contact with water near the surface, it can slope in degrees (Hulme, 1974). It can be seen
be catastrophically disrupted into small pieces, from this relationship that the higher the yield
or pyroclasts, by the rapid expansion of gases. strength, the thicker the lava flow must be in
Such events are termed explosive eruptions. The order to move.
style of an eruption depends on many factors Once a flow is moving the velocity of the flow
including the composition of the magma, gas will be affected by many factors. In general, the
26 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

Fig. 1.24. Levee development in a


basaltic lava flow. During movement
solidified pieces of the lava flow
build up along the sides to form a
confined channel within which lava
can continue to flow. Modified from
Francis (1993).

speed of the flow, V, can be approximated by this of over 1000 m. Fallback of material from lava-
relationship: fountaining may form quenched scoria frag-
ments or, if the fragments are still very hot, spat-
ρ gt 2
V = sin α (1.2) ter develops around the vent, fuses together, and

may begin to move away as a lava flow.
where η is the lava viscosity, ρ is its density, g is
the acceleration of gravity, t is flow thickness, B Subaerial conditions
is a constant, and α is the slope. An important As magma is discharged from a vent under sub-
result of this equation is that the speed is criti- aerial conditions cooling takes place by conduc-
cally dependent upon the lava viscosity. As viscos- tion, convection and radiation. At the high tem-
ity increases, the speed will decrease. Hawaiian perature of eruption, radiation is initially the
lava flows have been measured at speeds of up most important mechanism of heat loss and can
to 60 km/hr on slopes of 10–28◦ . This represents be approximated by
the upper range of lava flow speeds, with most
moving considerably more slowly. Q ≈ σT 4 (1.4)
The flow regime of moving lava can be inter-
where Q is the rate of heat loss, σ is the Stefan–
preted using the dimensionless Reynolds number
Boltzman constant (5.67 × 10−12 J s−1 cm2 ◦ C−4 ),
given by
and T is the absolute temperature. Because Q is
R e = 2ρ H V /η (1.3) proportional to the fourth power of temperature
there is a dramatic decrease in the rate of heat
where ρ is the magma density, H is the depth of loss associated with only slight decreases in tem-
the flow, V is the velocity, and η is the magma perature. Loss of heat by radiation immediately
viscosity. A Reynolds number value of less than lowers the surface temperature and changes the
2000 corresponds to laminar flow where move- color of a flow. This results in an increase in vis-
ment is smooth and flow lines do no cross. If the cosity and a decrease in flow speed as pieces of
Re is greater than 2000, the flow becomes turbu- chilled lava appear on the surface. These are car-
lent with eddies and crossing flow lines. For the ried along on the lava surface and only parts
majority of basaltic eruptions the viscosity of the of the incandescent interior remain visible. As
magmas is such that laminar flow conditions pre- pieces of quenched lava fall off the sides of a lava
vail. Only in the case of extremely low viscosities flow they build up a natural levee that confines
or high velocities, such as might occur over very the flow and insulates it from cooling (Fig. 1.24).
steep topography, would flow conditions become This process results in lava being funneled down
turbulent. the slope of a volcano along a similar path for
Discharge of magma at the vent may be pas- long periods of time. The width of the levees is
sive, in the form of slow outpourings of magma, a function of the yield strength of the magma
or it may be more spectacular with fountains of and of the local slope. Hulme (1974) has devel-
magma (lava-fountaining) that can attain heights oped a relationship for quantitatively estimating
STYLES OF VOLCANIC ACTIVITY 27

yield strength, τ , from levee dimensions as


follows
τ = 2w b gρα 2 (1.5)

where w b is levee width, g is local gravity, ρ is


magma density, and α is the slope.
Continued development of a quenched crust
on the lava flow surface may eventually lead to
isolation of the flowing interior from the atmo-
sphere as a roof of lava crust forms over an inte-
rior lava tube. Because lava and quenched magma
are poor conductors of heat, but have high heat
capacities, lavas may continue to flow for consid-
erable distances from source without suffering
significant heat loss. The confinement of lava to
these subsurface tubes is thus an effective mecha-
nism for flows to increase their maximum runout Fig. 1.25. Distribution of the 14 Ma Pomona basalt lava
from source. When an eruption ceases, downs- flow in Washington state. Modified from Francis (1993).
lope drainage of the lava may result in the for-
mation of lava tunnels that can extend for many
kilometers away from the vent. In the distal parts
of lava flows, the pressure build-up within the
interior of the flow may causing upwelling of
the surface and occasional breakout of magma.
The uplifted breaks in the surface are referred to
as tumuli.
An important question about lava flows is:
how far will they travel? Most historic lava flows
have traveled several to tens of kilometers from
the source vent. However, there is evidence in
the geologic record of basaltic lava flows that
have traveled up to 300 km from source. The
14 Ma Pomona flow in the western United States Fig. 1.26. Surface morphologies of an aa lava flow
was erupted in Idaho and can be traced almost (background) and a pahoehoe lava flow (foreground) from
to the Pacific coast in Oregon (Fig. 1.25). Walker Kilauea volcano, Hawaii. Differences in the texture are
(1974) proposed that the discharge rate is the attributed to variations in shear strain and magma viscosity.
most important factor in determining the ulti- Photograph by S. Carey.
mate length of lava flows. Discharge rates for his-
toric lava flow have varied by several orders of
magnitude (0.5 to 5 × 103 m3 /s), and there is longer runouts because heat loss, and thus vis-
evidence for truly enormous rates of discharge cosity increase, is reduced for flows in subsurface
(∼1 × 104 m3 /s) during some flood basalt episodes tubes.
(Swanson et al., 1975). In addition, flow volume Quiescent subaerial discharge of basaltic
and cross-sectional area appear to correlate with magma generally leads to the production of two
flow length (Malin, 1980), although this is to be distinct type of lava flows: aa and pahoehoe. Aa
expected if discharge rate plays a major role in is the most common type and consists of flows
flow runout (Woods, 1988). Another factor that that have a rubbly top with sharp angular blocks
influences how far lava flows travel is whether a and clasts (Fig. 1.26). These flows tend to form
flow is channel or tube-fed. The latter will favor units 10–100 m thick that cover areas of between
28 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

inversely related to viscosity. Thus, high-viscosity


magmas require lower rates of shear strain to
make the transition to aa behavior. Several stud-
ies have proposed that discharge rate plays a
major role in determining the production of
aa versus pahoehoe (e.g., Pinkerton and Sparks,
1976; Rowland and Walker, 1990). Field observa-
tions suggest that the formation of aa flows is
favored by higher discharge rates (>5–10 m3 /s).
Magmas of more evolved composition are also
erupted as lava flows but their resulting mor-
phologies reflect fundamental differences in rhe-
ology. In general as silica content increases, vis-
cosity increases (Fig. 1.13) and thus flow mobil-
ity is limited. Flows of andesitic or dacitic com-
Fig. 1.27. Configuration of the proposed transition position tend to be thicker and less widespread
threshold zone (TTZ) defining the behavior of basaltic lava
compared to basaltic flow. Aspect ratio is a con-
flows. An increase in rate of shear strain or magma viscosity
venient parameter to characterize the general
favors the development of aa type flows over pahoehoe.
Modified from Peterson and Tilling (1980).
shapes of lava flows. In this case it is defined
as the ratio of the thickness of a flow to the
area that it covers. Because of their low vis-
1 and 100 km2 . In cross-section the flow typically cosity, basaltic magma generally form thin and
consists of an upper rubble zone overlying a mas- widespread flows, whereas more viscous rhyolite
sive interior. The interior often exhibits abundant flows are fat and compact (Fig. 1.28).
vesicles formed by the exsolution of dissolved Some evolved magmas have such high viscosi-
gases. At the base of the flow there is also a thin ties that they are unable to flow any substantial
rubble zone separating the massive flow from distance away from the vent, and instead build
the ground surface. In contrast, pahoehoe flows up domes. Extrusion of viscous magma can lead
are characterized by a much smoother surface to four principal types of domes: Peléean type,
with ropey and entrail-like morphology (Fig. 1.26). low lava dome, upheaved plug, or coulee (Blake,
These flows are generally thinner (<15 m thick) 1989). Peléean domes are named after the famous
and can cover areas of 1–1000 km2 . structure that grew in the crater of Mt. Pelée fol-
In many cases the composition of aa and lowing the explosive eruption of 1902 (Lacroix,
pahoehoe flows is identical and thus composition 1904). These structures are characterized by very
alone can not be called upon to explain the ori- steep sides and a sharp, craggy spine at the top
gin of the two different flow morphologies. Most (Fig. 1.29). Surrounding their base is an apron
Hawaiian basaltic eruptions begin as pahoehoe of debris generated by repeated collapse of the
but some undergo a transition to aa. The reverse, dome’s sides. A low lava dome has a more
however, has never been observed. Peterson and subdued profile and symmetrical distribution
Tilling (1980) suggest that the transition from around the source vent (Fig. 1.29). It grows
pahoehoe to aa is determined by the relation- by outward displacement of previously erupted
ship between magma viscosity and rate of shear material as new magma is introduced inter-
strain. The latter is considered to be a measure nally. Coulees represent something of a transi-
of how fast differential motion occurs between tion between lava domes and lava flows. They are
adjacent parts of the flowing lava. They proposed generated when viscous magma is erupted on a
the existence of a transition threshold zone (TTZ) steep slope and there is sufficient shear stress
that separates lava flow behavior into pahoehoe for some downslope movement. This flow results
and aa regimes (Fig. 1.27). The rate of shear strain in an asymmetrical distribution about the vent,
necessary to initiate transition to aa behavior is but near source they may still be as thick as low
STYLES OF VOLCANIC ACTIVITY 29

Fig. 1.28. Fields of thickness


versus area covered for lava flows of
different composition. Fractions
indicate the ratio of thickness to
area. Modified from Walker (1973).

and bottom photography have revealed two prin-


cipal types of basaltic submarine lava flows along
mid-ocean ridge spreading centers: pillows and
sheet flows. Pillow lavas consist of masses of inter-
connected flow lobes with bulbous, spherical, or
elongated morphologies (Fig. 1.30). Individual pil-
lows range from about 10 cm to 1 m in diameter.
The surface of pillows consists of a thin glassy
crust (<1 cm) formed by the rapid quenching of
magma by seawater. Inside, the pillows exhibit
radial fractures, vesicles, and a more crystalline
groundmass resulting from slower cooling. The
outside of some elongated pillows is decorated
with corrugations and contraction joints formed
as pillows grow and subsequently cool. On the
Fig. 1.29. Different types of domes produced by the Juan de Fuca ridge two types of pillow morpholo-
extrusion of viscous magma. Variations in dome morphology
gies have been observed (Chadwick and Embley,
are related to differences in magma viscosity and extrusion
1994). One type was characterized by large elon-
rate. Modified from Blake (1989).
gated pillows with striated surfaces, while the
other consisted of smoothed surface pillow lobes
lava domes (Fig. 1.29). Finally, upheaved plugs are that resembled subaerial pahoehoe toes.
pushed up sections of viscous magma that rise Sheet flows, as their name imply, are flows
almost vertically and lift sections of the surface with a large horizontal extent relative to their
stratigraphy (Fig. 1.29). thickness (Ballard et al., 1979). The surfaces of the
flows can be relatively smooth or exhibit differ-
Submarine conditions ent types of deformation features, such as ropey,
The majority of effusive volcanism on Earth is folded, or coiled morphology. Some, referred to
basaltic in composition and occurs underwater as hackly flows, have irregular surfaces that are
either along the mid-ocean ridge system or at made up of jagged blocks about 1 m in size. These
submarine hotspots. Discharge of magma into bear some resemblance to the rough surfaces of
water produces distinctive types of flows that some subaerial aa flows. The glassy rinds of sheet
reflect the more efficient cooling compared to the flows are commonly thicker than those of pillow
subaerial environment. Submersible observations basalt, measuring about 5 to 15 cm (Hekinian
30 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

As magma is erupted slowly it is rapidly chilled to


form a glassy rind that inflates due to the inter-
nal magma pressure. The flow advances intermit-
tently in a series of pulses, usually moving a few
meters at a time. Formation of the quenched
crust temporarily stops the flow advancement
until the internal pressure is sufficient to rup-
ture the skin and a new pillow is budded out.
This new growth may occur from the front of the
pre-existing pillow or, more commonly, through
the side. The directional changes caused by later-
ally advancement leads to the complex intercon-
nected appearance of pillow lava fields. Changes
in the rate of magma discharge appears to con-
trol the fine-scale structure of pillows (Ballard
and Moore, 1977). Low magma discharge pro-
duces elongated pillows with cracks and stri-
ations because a thick brittle crust is devel-
oped and budding to form new pillows occurs
through irregular fractures. At higher discharge
rates the quenched skin is thinner and thus can
stretch without fracturing, producing a smoother
surface.
The formation of submarine sheet flows has
never been directly observed but their morpho-
logical features provide important clues to their
Fig. 1.30. Formation of pillow basalt in the submarine
environment. Rapid cooling of lava against seawater generates origin. Using analogies with subaerial lava flows
a solid crust. Continued growth occurs by budding and Ballard et al. (1979) proposed a model for the
elongation of pillow lobes as magmatic pressure exceeds the formation of pillow and sheet flows. Sheet flows
strength of the pillow walls. Modified from Brown et al. are thought to form from high-discharge events
(1989). of fluid lava from fissures. They are thus analo-
gous to high-discharge-rate subaerial surface-fed
pahoehoe in which the flow is not tightly con-
et al., 1989). Sheet flows are often found in associ- strained by channels. The rapid discharge of lava
ation with collapse pits in summit calderas along can produce ponding in topographic depressions.
spreading ridges. These pits are up to several As breakout of magma occurs at the margins of
hundred meters in diameter and can be 25 m the pond, or drainback to the source vents takes
deep (Francheteau et al., 1979). Inside there is place, collapse of the quenched surface produces
commonly basaltic rubble scattered across the pits. If the drainage is discontinuous, a record of
floor, horizontal ledges at various levels along the the changes will be preserved in the form of hor-
sides, and unusual circular pillars with distinc- izontal terraces along the pit margin. The pillars
tive ribbed surfaces. that are often found in association with sheet
In many submarine volcanic fields the pil- flows most likely represent places where water
low and sheet lavas have the same composition that was trapped below the flow has been ejected
and thus the origin of the contrasting morpholo- upwards through the flow creating a cylindrical
gies cannot be dependent on magma composi- conduit from quenched lava.
tion alone. The formation of pillow lavas has The important role of discharge rate on flow
been well documented from observations of lava morphology of submarine lava flows has been
flows entering the sea off Hawaii (Moore, 1975). demonstrated by laboratory experiments using
STYLES OF VOLCANIC ACTIVITY 31

polyethylene glycol wax (Griffiths and Fink, 1992; volatile components such as water and carbon
Gregg and Fink, 1995). Different flow morpholo- dioxide (e.g., Verhoogen, 1951; McBirney, 1973).
gies can be characterized by a dimensionless At depth, magma may be undersaturated, satu-
parameter, ψ, defined by rated, or supersaturated with these components.
The degassing process has been largely inferred
ψ = ts /ta (1.6) from experimental and theoretical work on bub-
ble growth in gas-saturated liquids, coupled with
where ts is the time required for the formation observations of erupted products (e.g. Sparks,
of a solid crust on the surface of the flow and 1978; Cashman and Mangan, 1994; Sparks et al.,
ta is some characteristic timescale for horizontal 1994). In order for bubbles to begin to grow,
advection of the flow. Pillow lavas form when the supersaturation of only a few tens of bars is
value of ψ is less than 3, whereas various types required for heterogeneous nucleation (Hurwitz
of sheet flows are produced if ψ is between 3 and Navon, 1994). During ascent, bubbles increase
and 25. in size by a combination of diffusion (mass trans-
The distribution of different submarine lava fer) and depressurization (Sparks, 1978). Diffu-
flow types has been studied in a number of dif- sional growth dominates at depth while decom-
ferent mid-ocean spreading areas such as the pression is more important near the surface. The
mid-Atlantic ridge (slow spreading rate), the Juan ultimate size of the bubbles is a function of
de Fuca ridge and Galapagos Rift (intermediate magma ascent rate, initial volatile content, and
spreading rate) and the East Pacific Rise (fast- diffusion coefficient of the volatile component.
spreading rate). At slow-spreading ridges, pillow Bubbles in basaltic magma tend to be larger than
basalts are dominant, whereas sheet flows are in silicic magma because the diffusion coeffi-
most common along intermediate and fast- cients of most volatile components in basalt are
spreading ridges. This suggests that magma dis- high and the viscous forces that impede bubble
charge rates are likely to vary between these growth are smaller.
different spreading environments. In addition, When bubbles form in ascending magma they
there are variations in the relative abundances are less dense than the surrounding liquid and
of different flow types along the axis of ridge will also begin to rise. The rate at which they rise
segments. Francheteau and Ballard (1983) found will be determined by their size, density contrast
that sheet flows were more abundant at the topo- with the magma, and magma viscosity. If the rise
graphic highs of first order ridge segments on rate is similar to or much less than the rise rate
the East Pacific Rise. These areas correspond to of the magma, then the bubbles are essentially
zones of high magma supply and thus eruptions locked into the volume of magma from which
can tap into the subaxial magma chambers for they grew. This situation typically occurs for sili-
voluminous discharges of fluid magma (Fig. 1.20). cic magmas whose viscosity is so high that bub-
With increasing distance from the topographic bles cannot rise relative to the magma even for
high, the magma supply diminishes and erup- low rates of magma rise. In contrast, the viscosi-
tions of lower discharge rate produce increasing ties of basaltic magmas can be low enough such
amounts of pillowed flows. In constrast, the lack that bubbles can rise faster than the magma. This
of a well-developed melt lens at slow-spreading can result in the accumulation of gases in parts
ridges (Fig. 1.22) limits the amount of available of the magma chamber or conduit system.
magma and leads to common eruptions of low As magma approaches the surface the vol-
discharge rate and production of large amounts ume fraction of gas continues to increase as
of pillowed lavas on the rift axis floor. bubbles grow larger. An important aspect of the
degassing process is the rheological changes asso-
Explosive volcanism ciated with the decrease in dissolved volatiles
Fragmentation of magma by volatile degassing during ascent. Figure 1.13 shows that for a rhyo-
One of the most important mechanisms of lite magma, the loss of only 1.0% water causes
magma fragmentation is degassing of dissolved an increase in viscosity of about an order of
32 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

magnitude. Consequently, the magma is essen- winds. The heights of the fountains are a function
tially transformed into a foam that becomes of magma volume flux and volatile content (Head
increasingly rigid. Crowding of bubbles and the and Wilson, 1987).
increase in strength of the foam makes it diffi- In contrast to the continuous fountain of
cult for further bubble growth and excess pres- basaltic magma produced by lava-fountaining,
sure builds within bubbles. Eventually the inter- Strombolian eruptions are characterized by a
nal pressure in the bubbles is high enough to series of discrete explosions separated by <0.1 s
cause bubble bursting and fragmentation of the to several hours. Each explosion represents the
magma into a mixture of exsolved gases and hot, bursting of one or more very large bubbles (up
liquid particles. This transition from a state in to meter-size diameter) near the surface, usu-
which the magma was the continuous phase to ally within a standing lava lake (Blackburn et al.,
one where the gas is the continuous phase is 1976). Fragmentation is also relatively poor dur-
referred to as the fragmentation level. The posi- ing these types of events and incandescent clasts
tion of this level will be a function of the initial are typically ejected along well-defined ballistic
gas content of the magma, the magma ascent trajectories through the atmosphere (e.g., Chouet
rate, and the magma composition. In general, et al., 1974). Accumulation of coarse ejecta from
fragmentation during explosive eruptions occurs Strombolian eruptions can construct steep-sided
at depths of less than 1 km below the vent, cones such as the one produced during the 1943–
although in some cases it may migrate to deeper 52 eruption of Parı́cutin volcano in Mexico (e.g.,
levels. Above this level the mixture is rapidly Riedel et al., 2003).
accelerated due the expansion of the gases under The mechanism of lava-fountaining and Strom-
high pressure. It is this expansion that is the driv- bolian eruptions involve fragmentation of low-
ing force for the high-speed ejection of material viscosity basaltic magma by volatile degassing.
from a vent during an explosive eruption. The Parfitt and Wilson (1995) have suggested that
degassing process can lead to a variety of erup- a fundamental factor that differentiates lava-
tion styles depending on the amount of volatiles fountaining from Strombolian eruptions is the
in the magma, the composition of the magma, extent to which bubble coalescence occurs dur-
and the rate at which it is supplied from depth. ing magma ascent. In basaltic magma the viscos-
ity is sufficiently low lava that bubble rise due to
lava-fountaining and strombolian buoyancy can approach or exceed the rise rate of
eruptions the magma. When bubbles are able to rise rela-
Lava-fountaining is the disruption of basaltic tive to the magma the probability of coalescence
magma into a spray that is ejected up to over increases dramatically as portions of the magma
1000 m above a vent, although typical heights are become enriched in bubbles. As magma rise rate
usually tens to hundreds of meters. An important increases bubbles can no longer move relative to
aspect of this style of eruption is that it produces the magma and are thus fixed to the portion
activity that may be sustained for several hours or of melt from which they exsolved, reducing the
days. It is a common type of activity for Hawaiian potential for coalescence.
volcanoes and can occur from both central vents Lava-fountaining is favored by high rates
or fissures. The degree of fragmentation is typ- of magma rise. Under these conditions bubble
ically poor and thus relatively large clots up to growth occurs more homogeneously throughout
several tens of centimeters are formed. Because the rising magma and fragmentation can be
the clasts can be large and the ejection heights attained at deep levels. This leads to a contin-
modest, the pyroclasts from lava-fountaining may uous discharge of fragmented spray above the
remain molten during their transport and coa- vent. Modeling by Parfitt and Wilson (1995) sug-
lesce after fallout to form a lava flow. Finer pyro- gests that for volatile contents from 0.1% to 1.0%,
clasts may cool sufficiently during fallout to pro- lava-fountaining will occur if magma rise speeds
duce a deposit of solid tephra particles that is are in excess of 0.1 m/s (Fig. 1.31). As magma rise
controlled by the direction and speed of the local rate is decreased, bubbles are able to move and
STYLES OF VOLCANIC ACTIVITY 33

Hawaiian (fire-fountaining)
Magma rise speed (m/s)

0.1

nal
sitio
0.01 Tran

Strombolian
0.001

0.0001
0 0.2 0.4 0.6 0.8 1.0
Gas content (wt.%)

Fig. 1.31. Predicted conditions for the production of


lava-fountaining versus Strombolian eruptions during the
discharge of basaltic magma. Based on the modeling of Parfitt
and Wilson (1995).

concentrate relative to the rising magma. Some


parts of the magma become volatile-depleted and
can thus rise to the surface without fragmenta-
tion. Other portions, where volatiles accumulate
and bubble coalescence takes place, will contain
large bubbles that burst intermittently as they
arrive at the surface. Low magma rise speeds thus
favor Strombolian eruptions (Fig. 1.31) and transi-
tions in eruption style are more likely to be deter-
mined by changes in ascent rate as opposed to
variations in volatile content.
Fig. 1.32. Cross-section showing the development of a
Vulcanian eruption (bottom to top). Modified from Wilson
vulcanian eruptions (1980).
Another type of explosive eruption that involves
a series of discrete events, but which discharges
more evolved magma, such as basaltic andesite ween these types of explosive eruptions is likely
or andesite, is called Vulcanian. These eruptions to be transitional and complicated by the differ-
occur at intervals of a few minutes to several ent mechanism by which they are generated.
hours and eject pyroclastic material at velocities Vulcanian eruptions are generally attributed
up to 400 m/s (Self et al., 1979). Fragmentation of to the sudden release of pressure within a con-
magma and country rock is more efficient than in duit or dome as a result of the failure of a
Strombolian eruptions and the ejected mixture partially cooled overlying carapace (Fig. 1.32). As
typically forms a turbulent plume of particles magma rises into the edifice of a volcano, crys-
and gases that rises by thermal convection above tallization and volatile exsolution can build up
the vent to altitudes up to several thousand high pressures. If the magma is blocked by a
meters. The plumes generated by such eruption plug of solidified magma or country rock, the
may reach as high as 20 km into the atmosphere pressure may continue to build until it exceeds
and result in much wider dispersal of material the strength of the cap. Sudden pressure release
compared with Strombolian eruptions. It should allows the gases to rapidly expand and fragment
be emphasized, however, that the division bet- both magma and the solidified cap. Observed
34 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

Fig. 1.33. Cross-section through a volcano


during a Plinian-style eruption. Hb is altitude
where the plume density is equal to the
density of surrounding atmosphere and Ht is
maximum altitude attained by excess
momentum.

ejection velocities of up to 200 m/s are consis- During these eruptions magma is withdrawn
tent with the shallow level degassing of several from a crustal magma chamber and undergoes
weight percent of dissolved volatile components gas exsolution as it moves toward the surface. Ini-
from magmas (Wilson, 1980). In some cases, Vul- tially, bubbles grow and move with the magma
canian eruptions may be driven by the interac- because of its high viscosity (Fig. 1.33). Recent
tion of magma with groundwater, whereby ejec- experimental and theoretical work suggests that
tion velocities up to 400 m/s may be developed. a large part of the degassing may actually occur
over a limited region of the conduit (Mader et al.,
1994; Sparks et al., 1994). Large degrees of super-
plinian eruptions saturation caused by magma ascent may result
The most intense type of explosive eruption in homogeneous nucleation of bubbles and catas-
involving fragmentation of magma by volatile trophic gas release. Rapid acceleration of a stiff
degassing occurs when large volumes of silicic foam and breakage of bubble membranes results
magma are erupted in a quasi-steady-state fash- in disintegration into a gas/pyroclastic mixture
ion for periods of hours to days. These events within a fragmentation zone. Above this zone the
are referred to as Plinian eruptions in honor of mixture moves in a turbulent, high-speed fashion
Pliny the Younger who described the classic AD 79 before being ejected from the vent.
explosive eruption of Vesuvius which buried A key feature of Plinian eruptions is the
the cities of Pompeii and Herculaneum in Italy. sustained discharge of a highly fragmented
STYLES OF VOLCANIC ACTIVITY 35

constitutes a large part of the rising plume that


may eventually reach several tens of kilometers
in height. However, because the atmosphere is
stratified, the plume will eventually reach an alti-
tude, Hb , where its bulk density is equal to that of
the surrounding air and there is no longer suf-
ficient thermal energy to heat entrained air. At
that point the plume will begin to spread out
laterally into an umbrella region (Fig. 1.33). The
top of the umbrella region, Ht , represents the
maximum height attained by the rising mixture
caused by its momentum when it arrives at the
level of neutral bouyancy. Away from the convec-
tive portion of the plume the umbrella region
spreads like a giant gravity current in the atmo-
sphere (Bursik et al., 1992a, 1992b). The lateral
velocities can be very large near source and over-
whelm the strength of the local winds allowing
an umbrella current to spread radially for great
distances (e.g., Sparks et al., 1986). Four hours
after the beginning of the 1991 explosive erup-
tion of Mt. Pinatubo the umbrella region had
grown to a diameter in excess of 400 km.
The height of a Plinian plume is mainly deter-
Fig. 1.34. Plinian-style plume from an explosive eruption mined by the thermal flux at the vent and to
of Mount St. Helens in 1980. Photograph by M. Doukas a first approximation behaves in a manner simi-
(US Geological Survey).
lar to turbulent convective plumes in a stratified
environment where the height, h, is given by
gas/pyroclast mixture for long periods of time.
h = b Q 0.25 (1.7)
Exsolution of only 1% to 5% water is sufficient
to produce exit velocities of 100 to 500 m/s at and Q is the thermal flux (directly related to
the vent (Wilson et al., 1980). The hallmark of magma discharge rate), and b is a parameter
Plinian eruptions is the production of a spectac- related to the density stratification of the atmo-
ular mushroom-shaped plume (Figs. 1.33, 1.34), sphere (Sparks, 1986). Most Plinian eruptions gen-
such as described by Pliny the Younger during the erate sufficient thermal output to form plumes
AD 79 eruption of Vesuvius. As the gas/pyroclast that extend well into the Earth’s stratosphere.
mixture is ejected from the vent its bulk density For example, the 1991 eruption of Mt. Pinatubo
is greater than the atmosphere density and it will in the Philippines produced a giant mushroom-
initially rise to several hundreds or thousands of shaped plume that reached between 35 and 40
meters owing to its momentum (Sparks, 1986). km altitude (Koyaguchi and Tokuno, 1993). More
During this jet phase the density of the mix- sophisticated models of Plinian plumes allow for
ture must decrease in order for continued rise the specific calculation of plume density, tem-
to take place above the volcano. This is accom- perature, and velocity as a function of altitude
plished by the entrainment and heating of ambi- (e.g., Woods, 1988).
ent air at the sides of the high-velocity jet. A tran- Fallout of pumice and ash occurs from the
sition occurs where the motion of the material sides of Plinian eruption columns and the base
is now determined by buoyant thermal convec- of the umbrella region (Carey and Sparks, 1986;
tion (Fig. 1.33). Essentially the plume rises like a Bursik et al., 1992a; Bonadonna et al., 1998).
hot-air balloon. This convective region typically Because of the great height of the plumes and
36 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

Fig. 1.35. Isopachs of the AD 79 Plinian pumice fall deposit


from Vesuvius volcano. Thicknesses are in centimeters.
Modified from Sigurdsson et al. (1985).

Fig. 1.36. Pyroclastic flow descending the slopes of Mount


continuous nature of Plinian eruptions, large St. Helens during the August 7, 1980 eruption. Photograph by
areas can be impacted by thick accumulation of P. Lipman (US Geological Survey).
tephra fall. The AD 79 eruption of Vesuvius lasted
for approximately 19 hours and resulted in the
burial of Pompeii under about 2 m of pumice
highly turbulent fashion and can overrun many
fallout (Fig. 1.35).
topographic barriers (Fisher et al., 1980; Carey,
1991). They tend to produce a thinner, stratified
Generation of pyroclastic flows and surges deposit of fine-grained pumice and ash.
One of the deadliest types of behavior during An important mechanism for the generation
explosive eruptions is the generation of mixtures of flows involves the behavior of the high-speed
of hot gases and particles that descend down the jet of gas and pyroclasts after it leaves the vent.
slopes of a volcano at high speeds (Fig. 1.36). They The gas/pyroclast mixture that is ejected from
can travel at velocities in excess of 100 km/hr and vents during explosive eruptions is in most cases
maintain temperatures of >400 ◦ C at great dis- denser than the surrounding atmosphere and
tances from source. Their speed and high tem- can only continue to rise if it entrains and heats
perature make them particularly lethal to human enough ambient air to reduce its bulk density
populations around volcanoes and have resulted to less than that of the atmosphere (Sparks and
in tens of thousands of fatalities during historic Wilson, 1976). When that occurs, a high-altitude
times (Tilling, 1989). Pyroclastic flows are high- convective plume is formed, as described in
particle concentration types of flows that tend to the previous section. However, if the mixture is
be strongly controlled by the local topography. unable to become buoyant, then it will collapse
They can produce widespread deposits of coarse to form pyroclastic flows or surges. The collapse
pumice and ash known as ignimbrites. Pyroclas- may affect the entire jet, in which case a low
tic surges are more dilute flows that move in a collapsing fountain may occur over the vent, or
STYLES OF VOLCANIC ACTIVITY 37

1011
Taupo

Magma discharge rate (kg/s)


Crater Lake

Bandelier Tuff
1010

Katmai
109

108
10 100
Runout distance (km)

Fig. 1.38. Inferred magma discharge rates versus runout


distance for some large-scale pyroclastic-flow-generating
Fig. 1.37. Predicted conditions for the generation of
eruptions. Modified from Bursik and Woods (1996).
convecting columns versus collapsing fountains (pyroclastic
flow generation). Values next to the boundary curves indicate
the volatile mass fraction in the magma. Modified from Sparks accord with theoretical predictions of eruption
et al. (1997). column behavior (Sigurdsson et al., 1985; Carey
and Sigurdsson, 1987).
Pyroclastic flows and surges may also be gen-
it can take place only along the column margin, erated by the collapse of growing lava domes by
with some parts of the column continuing to rise explosive or gravitational failure of the sides or
convectively (Carey et al., 1988). explosive eruption. A lateral blast from a growing
The main factors that determine whether an dome on the summit of Mt. Pelée in Martinique
erupting mixture will form a rising plume or on May 8, 1902 generated a devastating pyro-
a collapsing fountain are the magmatic volatile clastic surge that killed more than 28 000 peo-
content, exit velocity, vent radius, and magma ple in the city of St. Pierre (Fisher et al., 1980).
discharge rate. Figure 1.37 shows the configu- More recently, pyroclastic flows and surges have
ration of the two different regimes based on been generated frequently during dome growth
theoretical modeling of eruption column behav- at Unzen volcano in Japan and the Soufrière Hills
ior (Sparks et al., 1997). Exit velocity is strongly of Montserrat (Young et al., 1997).
dependent on magmatic volatile content (Wilson, Once generated, pyroclastic flows and surges
1980) and magma discharge rate is controlled in travel downslope under the influence of grav-
part by the vent radius (Wilson et al., 1980). The ity and entrain air as they move. Deposition of
modeling results indicate that the generation of material from the base and heating of entrained
pyroclastic flows and surges is favored by high air results in the generation of a buoyant co-
magma discharge rates and low exit velocities. ignimbrite plume that rises off the top of the flow
A common feature of many Plinian eruptions is (Fig. 1.36) and may match or exceed the heights of
for the event to begin with a high-altitude con- some Plinian plumes (Woods and Wohletz, 1991).
vective plume that forms widespread fallout fol- An important aspect of pyroclastic flows, as with
lowed by a transition to the generation of pyro- lava flows, is how far they travel from source.
clastic flows. This transition may be the result Bursik and Woods (1996) recently developed a
of decreasing volatile content of the magma or model for the dynamics and thermodynamics of
increases in the magma discharge rate during the large pyroclastic flows. They found that model-
course of the eruption. The AD 79 Plinian erup- derived calculations of magma discharge rate for a
tion of Vesuvius exhibited this type of evolution number of large pyroclastic-flow-generating erup-
and calculation of eruption parameters, such as tions were well correlated with the runout dista-
magma discharge rate and exit velocity, are in nce (Fig. 1.38). Furthermore, the inferred magma
38 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

Strombolian Surtseyan Submarine Fig. 1.39. Variation in the scaled


1.00
energy (efficiency of conversion of
Scaled energy

0.10 thermal energy to explosive magma


disruption) versus the mass ratio of
0.01 water to magma. Maximum
efficiency occurs at a mass ratio of
0.001 pillow lavas
about 0.3. Modified from Wohletz
and McQueen (1984).
0.01 0.1 1.0 10.0 100.0 1000
Mass ratio (H2O/magma)

discharge rates of the associated Plinian phases of nisms involving external water. The first, contact-
these eruptions were always less than the pyro- steam explosivity, involves the generation of a
clastic flow-phase rates, in support of predicted thin film of vapor at the interface between hot
eruption column behavior shown in Fig. 1.37. magma and water. The film subsequently col-
Generation of pyroclastic flows during large- lapses as a result of cooling by the remain-
volume silicic eruptions represents one of the ing reservoir of water and disrupts the surface
most spectacular forms of volcanism known on into small droplets. This process exposes new
Earth. Enormous volumes of magma can be hot magma below and the process repeats itself.
expelled during single events to form widespread Each cycle occurs on a timescale of microseconds
deposits known as ignimbrites, or ash flow tuffs and evolves into an explosive process by high-
(Ross and Smith, 1961). Such deposits consist of frequency repetition (Wohletz, 1986). However,
a poorly sorted mixture of pumice, ash, crys- explosive disruption of magma does not always
tals, and lithics. Individual flow units may reach occur when it comes in contact with water. A
thicknesses of tens of meters and can extend major factor in determining the nature of the
in excess of 100 km from source. In some cases interaction is the mass ratio of water to magma.
the deposits are emplaced at such high temper- Explosive fragmentation is favored by a ratio of
atures and with sufficient thickness that indi- about 0.3 (Fig. 1.39). Contact-steam explosivity can
vidual glassy particles fuse together to form a occur during underwater eruptions although the
dense welded tuff. Historic eruptions have pro- maximum depth at which the process can take
duced pyroclastic flow deposits with volumes up place is poorly known.
to several tens of cubic kilometers, but there are The second mechanism for magma fragmen-
many examples in the geological record of much tation by external water, called bulk interaction
larger deposits (Smith, 1979). For example, the steam explosivity, involves the local trapping of
volume of the Fish Canyon Tuff on the La Garita water and conversion to vapor. For example, a
caldera in the San Juan Mountains of Colorado, lava flow erupted underwater may trap water-
USA, has been estimated as >3000 km3 (Steven rich sediments beneath the flow. Heat from the
and Lipman, 1976). flow leads to superheating of the trapped water
and the build-up of high pressure as vapor is
produced. If the pressure exceeds the strength
Fragmentation of magma by interaction of the overlying flow, then fragmentation occurs
with external water as the cap fails catastrophically. This process is
The second major mechanism for the fragmen- likely to occur in water depths up to 2 to 3 km
tation of magma to generate explosive volcan- (Kokelaar, 1986), being limited by the pressure
ism is the interaction with some external source at which the volume change of water from liq-
of water. This may include groundwater, rivers, uid to vapor is no longer significant enough
lakes, seawater, or ice and snow (e.g., Colgate to cause high internal pressure. As mentioned
and Sigurgeirsson, 1973). Magma is fragmented previously, some Vulcanian-style eruptions may
by the rapid conversion of water to steam or by be triggered by the heating of groundwater by
rapid quenching. Kokelaar (1986) has suggested magma that is blocked by a solidified plug.
three principal types of fragmentation mecha- As pressure increases the plug eventually fails
STYLES OF VOLCANIC ACTIVITY 39

Fig. 1.40. Structure of an


underwater lava fountain that
generates hyaloclastite gravity
currents by collapse. Modified from
Smith and Batiza (1989).

allowing the vapor to expand and the magma to rapid discharge of magma generates a fountain of
undergo further degassing of dissolved volatile magma above the submarine vent (Fig. 1.40). Frag-
components. mentation occurs by cooling contraction granu-
A third mechanism of magma fragmentation lation and steam explosivity as seawater mixes
is significantly less explosive than the other two, with magma to produce a slurry of basaltic glass
but nevertheless can result in significant pro- shards, hot water, and possibly steam. Collapse of
duction of fragmental material during volcanic the slurry from the top of the fountain produces
eruptions. As magma comes into contact with gravity currents that move away from the vent
water, rapid quenching transforms the melt to and eventually deposit the bedded hyaloclastites
a glass. Significant contraction of the material (Fig. 1.40).
takes place as a result, leading to the fracturing
of glass into small pieces, or hyaloclastites. This surtseyan eruptions
cooling contraction granulation is an important The construction of many oceanic volcanoes
process associated with the emplacement of sub- begins in deep water but progresses towards
marine lava flows. Unlike the other fragmenta- the surface as new volcanic material is added.
tion mechanisms, there is no depth limit to con- A Surtseyan eruption is an explosive event
traction granulation, as steam expansion is not a associated with discharge of magma as a vol-
consideration. cano approaches the surface and becomes
emergent. Vigorous explosions send steam-rich
submarine lava-fountaining cocks-tail style plumes along parabolic trajec-
Lava-fountaining is a common style of volcanic tories (Fig. 1.41). The tephra-laden plumes are
activity for many subaerial basaltic eruptions and produced as a series of discrete events that may
there is now evidence that an analogous process take place on a variety of timescales ranging
may occur in relatively deep water as a result of from minutes to hours. In 1963, the island of
magma–water interactions. Submersible work on Surtsey emerged from the sea south of Iceland
seamounts near the East Pacific Rise has discov- in a series of explosive eruptions that are
ered the common occurrence of bedded hyalo- considered the type example of this activity. A
clastite deposits at depths from 1240 to 2500 m cone was built up from the accumulation of
(Smith and Batiza, 1989). The deposits are closely basaltic ash and scoria ejected by the eruptions.
associated with pahoehoe-like flows suggesting Kokelaar (1983) has proposed that in the vent
that high eruption rates are necessary for their area of a Surtseyan eruption a slurry of hot
formation. Smith and Batiza (1989) propose that water and tephra is mixed with new magma
40 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

Fig. 1.41. Eruption of Surtsey


volcano, Iceland in 1963. Photograph
by I. Einarsson.

Fig. 1.42. Internal structure of a Surtseyan eruption dissolved in the magma. Explosive activity at Surt-
column. Rising, vesiculating magma is mixed with a slurry of sey ceased once an island was formed with suffi-
water and volcaniclastic particles near the surface and then cient relief to isolate interaction of magma with
rapidly ejected as the liquid is converted to steam. Modified seawater, indicating that the dominant fragmen-
from Kokelaar (1983).
tation mechanism was related to interaction of
magma with seawater.

(Fig. 1.42). Upward migration of the mixture and


heat transfer cause the mixture to flash to steam
and be rapidly accelerated out of the vent as a jet. Construction of volcanic edifices
The primary mechanisms of magma fragmenta-
tion are thus contact-surface explosivity coupled The great variety in magma compositions, styles
with degassing of juvenile volatile components of eruption, and rates of discharge combine to
CONSTRUCTION OF VOLCANIC EDIFICES 41

0 layer 1 sediments
Fig. 1.43. Idealized stratigraphic section
through normal oceanic crust showing
layer 2a basaltic
1 layer 2b lavas lithologic subdivisions and their associated
seismic velocities. Modified from Brown
layer 2c basaltic
Depth below seafloor (km)

et al. (1989).
2 dykes

3
Oceanic
4 crust

layer 3
5 gabbros

6
Moho
7
Mantle
8 peridotite

2 4 6 8
Seismic wave velocity (km/s)

produce the myriad of volcanic structures that across-axis profile is typically smoother, with a
have been recognized throughout the world. central axial high. The height of the mid-ocean
Understanding the physical mechanisms of vol- ridges is the result of the elevated temperatures
canic eruptions and the types of volcanic deposits of the upwelling mantle material. As new ocean
that they produce has proved critical to recon- crust is formed and spreads away from the ridge
structing the evolution of volcanic landforms. it cools and becomes denser. The increase in den-
In many cases the growth and development of sity results in subsidence of the plate accord-
volcanoes has been directly witnessed but in ing to a systematic relationship where the
other cases, such as in the marine environ- depth is related to the square root of the
ment, the development of models for the evolu- age.
tion of volcanoes has had to rely largely on the Volcanism at the ridge is dominated by the
interpretation of the erupted products and indi- quiescent extrusion of basaltic magma as pil-
rect measurements of their structure. low lavas and sheet flows. The magma for these
flows is derived from a subaxial melt lens in the
Submarine and subaerial rift structures case of fast-spreading ridges and from isolated
Most volcanism on Earth occurs along linear rift melt bodies beneath slow-spreading ridges
systems where the surface plates are in exten- (Figs. 1.20 and 1.22). This style of volcanism pro-
sion and volcanism produces new ocean crust. duces a distinctive layered structure character-
The mid-ocean ridge system is a globally encir- istic of oceanic crust as inferred from geophys-
cling, elevated volcanic feature that extends for ical studies and the examination of uplifted
a total length of about 70 000 km (Fig. 1.1). Its submarine sequences known as ophiolites (Cas,
structure varies considerably and reflects the bal- Chapter 4, this volume) (Fig. 1.43). Layer 1 con-
ance between tectonism and magmatism. Where sists of a sequence of deep-sea sediments that
the mid-ocean ridge is spreading slowly a cen- accumulates on top of the oceanic crust as it
tral rift valley bounded on each side by faulted migrates away from the ridge. Layer 2 is subdi-
ridges is usually developed (Fig. 1.3). The cen- vided into three units based on seismic velocity
tral valley is less than 3 km wide and approxi- and morphology of erupted products. Layers 2a
mately 400 m deep. At high rates of spreading the and 2b are composed of a sequence of overlapping
42 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

pillow basalts and sheet flows formed by the bathymetric highs leading to flanking fissure
extrusion of magma onto the seafloor. Layer 2c eruptions. Little is known about the frequency
lies beneath these extrusives and is made up of of eruptions that take place along the mid-ocean
vertically oriented dykes of similar composition ridge system although morphological constraints
to the pillow basalts and sheet flows. These dykes suggest that individual axial volcanoes may have
represent the feeder system whereby magma is a lifetime of the order of 102 to 103 years.
transferred from the subaxial magma chamber Rift-dominated volcanism is also developed on
to the surface. Magma that remains in these some of the continents and although some rep-
conduits solidifies to form the dykes. An abrupt resent extensions of mid-ocean spreading ridges,
change in lithology occurs between the base of others are more complex. In general, many rifts
layers 2 and 3 where gabbros and metagabbros occur within a broad topographic elevation that
are encountered (Fig. 1.43). These coarse-grained is attributed to hot, buoyant mantle rising from
rocks are compositionally equivalent to basalt but below. A fundamental question for many rifts is
have crystallized over much longer periods of the relationship between the rifting and volcan-
time. They represent magma that was unable to ism. Two alternatives are that (1) rifting and vol-
be erupted at the surface and has frozen in place canism are the product of the active rise of man-
as the lithospheric plate migrated away from the tle material from depth, such as a hotspot, or (2)
ridge. Below the gabbros is another sharp contact volcanism is a passive response to rifting of con-
that marks the transition to periodotites of ultra- tinental crust that is driven by a wider regional
mafic composition (Fig. 1.43). These mark the base tectonism. The East African rift is often cited as a
of the oceanic crust sequence and represent the classic continental rift where extension is accom-
residual mantle material after magma has been panied by active volcanism. This rift is an exten-
extracted by pressure-release melting. sion of a triple junction occurring at the inter-
The layered structured of the oceanic crust is section of the Red Sea and the Gulf of Aden.
a general feature that persists in many types of It extends to the southwest for about 4000 km
spreading environments. Other important struc- with a width of 100 km. Within the rift are
tural features of the ridge system include along- some of Africa’s largest volcanoes, such as Kili-
axis segmentation (e.g., Macdonald et al., 1991). manjaro, Nyiragongo, and Longonot. Formation
The nature of segmentation varies considerably of the East African rift is attributed to the forma-
in scale along the ridge and again reflects compet- tion of a hotspot beneath Africa that is progres-
ing processes of tectonism and volcanism. First- sively forming a series of new small ocean basins
order segments are long sections of the ridge (up by thinning and extension of the continental
to hundreds of kilometers) that are bounded on crust.
each end by a transform fault. Their production The association of some continental flood
is tied to the tectonics and rate of plate spread- basalt provinces with the location of ancient rift
ing. Prime segments are sections of the ridge that zones has led to the suggestion that hotspot
are bounded by smaller-scale ridge axis discon- development beneath continents will necessar-
tinuities such as overlapping spreading centers ily lead to rifting. However, areas such as the
or axial offsets of >0.5 km and 20 m elevation Columbia River flood basalts and the Siberian
(Tighe, 1997). They are of the order of 100 km in province do not show evidence of rifting (Francis,
length along the East Pacific Rise and coincide 1993). Clearly the relationship between hotspots,
closely with magma that can be related to a sin- flood basalt provinces, and continental rifting
gle parental magma by fractional crystallization. has yet to be precisely understood.
Within each of the prime segments there is a sin-
gle or series of axial volcanoes defined by <0.5 km Shield volcanoes
axial offsets and single bathymetric highs. These Large-volume discharge of basaltic magma as lava
are interpreted to represent the locus of melt flows is responsible for the construction of the
uprise and eruption. Lateral flow of magma along world’s largest type of individual volcano, known
the ridge is likely to take place away from the as shields. They are characterized by gentle slopes
CONSTRUCTION OF VOLCANIC EDIFICES 43

submarine to subaerial conditions. The internal


structure of such volcanoes reflects the influence
Shield volcano of these different environments on the nature of
the eruptive style and morphology of the
resulting deposits. A well-studied example is the
Caldera Stratovolcano island of Gran Canaria in the Canary Islands of
the Atlantic. The evolution of the island has
Cinder Tuff been reconstructed by a combination of land-
cone ring
based and marine geological studies (Schmincke,
Fig. 1.44. Comparison of different volcano sizes and 1994; Schmincke et al., 1995). Submarine growth
morphologies. Modified from Decker and Decker (1997). of the island produced a series of distinct over-
lapping facies (Fig. 1.45). The core of the vol-
cano is built up of intrusions, pillow lavas, pillow
of usually less than 10◦ and broad extent giving breccias, debris flows, and hyaloclastites. With
the overall appearance of an overturned shield increasing height above the seafloor the pillow
(Fig. 1.44). Mauna Loa on the island of Hawaii lavas in the core become more vesicular as gases
is the world’s largest shield with an elevation of are able to exsolve under lower pressure and the
10 km as measured from the seafloor to the sum- abundance of hyaloclastites is higher. Surround-
mit and basal diameter of 190 km. The volume ing the core facies is a seismically chaotic flank
of Mauna Loa has been estimated as at least facies composed of pillow breccias, hyaloclastites,
40,000 km3 , or an order of magnitude higher and debris flow deposits (Fig. 1.45). This facies
than most other volcanoes. It has been built is constructed by shallow submarine eruptions
by the repeated eruption of low-viscosity tholei- and the eventual emergence of the island above
itic magma beginning at water depths of sev- sea level. Overlying the flank facies are the slope
eral thousand meters and eventually building up and basin facies that define the submarine mor-
above sea level for 4 km. The morphology of phology of the volcano. The slope facies is more
shield volcanoes reflects the ability of lava flows proximal to the source and represents a series
to travel far from the vent and produce gentle of slumps, debris flows, and massive units that
slopes. In many instances the enhanced transport were produced by subaerial eruptions discharg-
of flows is facilitated by movement through lava ing clastic material into the sea and the redepo-
tubes which insulate the flows from cooling and sition of epiclastic materials. At Gran Canaria this
greatly extend their runout. At the summit of facies extends from the shoreline to about 45 km
many shield volcanoes are calderas where lava offshore. With increasing distance away from
lakes may be present during eruptions. The for- the volcano the slope facies interfingers with
mation of these structures is due to the draining a more well-bedded basin facies. This consists
of magma from the summit regions into fissures of interbedded pelagic sediment, fine-grained
feeding flank eruptions or lateral intrusions. turbidites, and tephra fall layers. Its extent is
There are differences in the morphology and up to hundreds of kilometers from source and
size of shields in different geologic settings. For marks the edge of the island’s volcanic apron
example, some shield volcanoes of the Galapa- (Fig. 1.45).
gos Islands exhibit a distinctive profile with a An important discovery related to the growth
relatively flat top, moderately steep upper slopes of oceanic shield volcanoes is that they are
(>10◦ ), and a relatively abrupt change to more susceptible to major mass-wasting processes
gentle distal slopes (Francis, 1993). In addition, that can drastically modify their flanks. In the
many of the Galapagos shields have much deeper Hawaiian islands seafloor mapping has revealed
summit calderas (up to 800 m deep) than their the presence of enormous slumps and debris
Hawaiian counterparts. avalanches on the flanks and adjacent to many
Many of the large shield volcanoes form of the islands (e.g., Moore and Normark, 1994).
oceanic islands and thus have evolved from Slumps are slow displacements of large portions
44 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

Fig. 1.45. Cross-section through Gran


Canaria island based on seismic
stratigraphy and inferences about
submarine volcanic process. Modified
from Schmincke et al. (1995).

from the volcanic rift zone seaward to the base


of the volcanic pile. The slumps can be as much
as 110 km wide and 10 km thick, with gradients
of 3◦ . The Hilina slump on the south coast of the
island of Hawaii serves as a good example of an
active slump on the flank of an oceanic island
(Fig. 1.46). On land the slump shows up as a series
of normal faults that traverse the south coast of
the island. In 1975 during a 7.5-magnitude earth-
quake, a 60-km-long section of Kilauea’s coast sub-
sided by as much as 3.5 m and moved seaward by
8 m.
Debris avalanches tend to be longer, thinner,
and less steep than the slumps, and often can be
traced back upslope to a well-defined amphithe-
ater. A characteristic feature of debris avalanches
that allows for their recognition in the deep sea
is the presence of abundant hummocks, similar
to debris avalanche deposits on land. On sonar
returns they show up as a speckled pattern. The
size distribution of hummocks can vary signifi-
cantly from one slide to another. Some are 1 km
Fig. 1.46. Distribution of major submarine slumps and in diameter whereas others are up to 10 km. High-
debris avalanches (stippled patterns) around the island of resolution bottom photography indicates that the
Hawaii. Modified from Moore et al. (1995). areas between large hummocks are filled with
much smaller fragments that do not show up
of an island’s flank that occur over an extended on the large-scale mapping images. The Alika-
period of time, keeping pace with the production 2 debris avalanche exhibits hummock sizes and
of new volcanic material. They are thought to be spacing that are similar to the 1980 Mount St.
deeply rooted in the volcanic edifice, extending Helens debris avalanche deposit. This suggests
CONSTRUCTION OF VOLCANIC EDIFICES 45

Fig. 1.47. Mount St. Helens volcano prior to the The scoria cone of Parı́cutin in Mexico was con-
catastrophic eruption of May 18, 1980. Spirit Lake is in the structed during an eruptive episode from 1943
foreground. to 1952 and is just one of many such cones
on the flank of Pico de Tancitaro in the Trans-
that the debris avalanches are rapidly moving Mexican volcanic belt. In some cases the erup-
events, in contrast to the slow-moving slumps. tion of basalt during monogenetic activity will
There are two major hazards associated interact with water and become explosive. Maars
with the generation of major submarine debris are low craters formed by explosive interaction of
avalanches on the flanks of oceanic shield volca- groundwater and magma. The explosions exca-
noes. First, if the debris avalanche begins on land vate the subsurface rocks to form a crater sur-
then significant portions of an island’s coast may rounded by a rim of ejected debris. If the explo-
rapidly slide into the sea, destroying property and sive discharge leads predominantly to accumu-
lives. The second principal hazard is related to lation of material over the subsurface rocks by
the displacement of water during the generation Surtseyan-type explosions then the feature is gen-
of a debris avalanche and the formation of large erally referred to as a tuff ring.
tsunamis. Although this has not been witnessed Most volcanoes have a more complex evolu-
directly there is evidence in the geological record tion and are built up from both effusive and
that such events have occurred and may be com- explosive volcanism. Their internal structures
mon in the evolution of island volcanoes (Moore thus consist of a combination of lava flows and
et al., 1994). pyroclastic deposits. These composite volcanoes,
or stratovolcanoes, differ significantly in mor-
Volcanic cones and composite volcanoes phology from shield volcanoes, having much
On a small scale, localized volcanism often builds steeper slopes and flank profiles that are concave
volcanic cones that represent a single eruptive downward. They are most common in subduc-
episode (Fig. 1.44) (Riedel et al., 2003). The dura- tion zone environments (island arcs and conti-
tion of such episodes may vary from a few days to nental margins) where the magma compositions
several years. These volcanoes are known as are more evolved and generally carry more dissol-
monogenetic because once the eruptive episode ved volatiles. Mount St. Helens in the northwest
has ceased there is no more activity at the site. United States is a good example of a stratovolcano
Accumulation of scoria and lava flows typically (Fig. 1.47).
builds a steep-sided cone (∼33◦ ) several hun- Composite volcanoes are built up over
dred meters in height, with a relatively flat top. extended periods of time by a large number of
46 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

Fig. 1.48. Cross-section through a composite volcano Mullineaux, 1981). It is now apparent that catas-
showing the changes in structure as a function of distance trophic sector collapse of volcanoes is a relatively
from the peak. Modified from Williams and McBirney (1979). common phenomenon and a natural process in
the evolution of composite volcanoes (Siebert,
eruptive events and are referred to as polygenetic. 1984). Volcanic debris avalanches are quite dif-
The location of volcanism remains relatively ferent from normal dry rock avalanches in the
fixed, but the nature of the eruptions and the amount of material that is involved. Non-volcanic
development of the peak may change dramati- dry rock avalanches tend to form along bedding
cally. The core of the volcano consists of intru- or joint planes and are shallow relative to their
sive bodies and dykes that represent the path- length. In contrast, volcanic debris avalanches are
ways for magmas to the surface and locations much deeper relative to their length and have a
where magma was stored without eruption. Indi- steeply sloping back wall. The scar that remains
vidual eruptive vents may migrate around a cen- on the flank of the volcano often appears as a
tral location but the integrated products still pro- horseshoe-shaped crater (Fig. 1.49).
duce a structure with conical symmetry. At many One of the most distinctive features of debris
composite volcanoes, the summit area consists avalanche deposits is the occurrence of numer-
of a complex of domes formed by effusive activ- ous hills and depressions on their surfaces, giving
ity. Many have some type of summit crater that an overall hummocky topography. In addition,
formed as a result of the ejection of material by there are also longitudinal and transverse ridges.
explosive eruptions or collapse associated with The deposits usually consist of a very poorly
magma withdrawal. The flanks are built up of sorted mixture of brecciated debris. Most of the
multiple layers of lava flows, primary pyroclastic material is lithic clasts and blocks from the volca-
deposits, and the reworked products of the vol- nic edifice. However, there can be some juvenile
cano (Fig. 1.48). Slopes on the flanks may vary material. The majority of debris avalanches travel
from about 10◦ on the lower parts to up to 35◦ in excess of 10 km from the volcano, with several
near the summit. events reaching 50–60 km. These are thus similar
The accumulation of alternating layers of lava to the runout of some pyroclastic flows. Velocities
flows and pyroclastic deposits at relatively steep can also be high, with many in excess of 150 km/hr.
angles on the slopes of composite volcanoes often An important aspect of debris avalanches is
leads to unstable conditions. Catastrophic col- the sudden release of pressure on the volcano as a
lapse of the volcano’s flank may lead to the gener- result of slope failure. This can lead to triggering
ation of landslides, debris avalanches, and debris of explosive activity. There are two types of erup-
flows, such as occurred at the onset of the May 18, tions that generally occur (Siebert et al., 1987). The
1980 eruption of Mount St. Helens (Lipman and Bezimianny-type involves a magmatic component
CONSTRUCTION OF VOLCANIC EDIFICES 47

Fig. 1.49. Summit area of Mount


St. Helens showing the
horseshoe-shaped crater that
formed as a result of the May 18,
1980 explosive eruption.
Photograph from Lipman and
Mullineaux (1981).

significantly, with the largest attaining dimen-


and is driven by degassing. It may be preceded
sions of tens of kilometers in diameter. Irrespec-
by several days to months of precursory seismic
tive of size, the fundamental mechanism for their
and volcanic activity. In contrast, the Bandai-type
origin is similar. As magma is withdrawn from
is phreatic with no juvenile magma discharge
shallow magma chambers during eruption, the
and is driven by expansion of the hydrother-
overlying volcanic center must adjust its struc-
mal system beneath the volcano. These types of
ture in order to accommodate the reduction in
events may occur with or without preceding seis-
volume. If the amount and rate of withdrawal
mic activity. The horseshoe-like craters formed by
are low, the reduction in volume may be adjusted
flank collapse tend to ‘‘heal” by the growth of lava
by general subsidence without significant surface
domes, although the rates can vary widely. At
deformation. However, large volume discharges at
Bezimianny the 1956 amphitheater has been
high rates may reduce the pressure of the magma
almost completely filled in. At Mount St. Helens a
chamber to such an extent that the overlying
new lava dome has been building since the 1980
rocks are left unsupported and collapse of the
eruption and the crater is about one-third filled.
roof occurs (Fig. 1.50). Such collapse forms calde-
The healing process makes it difficult to deter-
ras that can vary in size from a few kilometers to
mine whether a volcano has experienced a col-
a hundred kilometers in diameter. Foundering of
lapse in the recent past.
the magma chamber roof can produce a dynamic
pressure on the remaining magma causing high
Calderas rates of magma discharge through radial fissures
The largest volcanic structures that are formed and favoring the production of additional pyro-
during the course of single eruptions are the clastic flows (Druitt and Sparks, 1984).
great circular depressions known as calderas. Smaller calderas, a few kilometers in diam-
These spectacular features are typically associ- eter, typically form within a pre-existing edi-
ated with voluminous silicic volcanism that gen- fice and thus mostly involve the collapse of vol-
erates thick pyroclastic flow deposits, or ignim- canic rocks. A recent example is the formation
brites (Smith, 1979). The size of calderas can vary of a ∼6-km-diameter caldera on Tambora volcano
48 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

Fig. 1.51. View of the caldera of Tambora volcano in


Indonesia. The caldera was formed during the great 1815
explosive eruption and is approximately 6 km in diameter and
1.2 km deep. Photograph by S. Carey.

is Lake Toba on the island of Sumatra in Indone-


sia (Fig. 1.52). It was produced by a series of
very large explosive eruptions, the last of which
occurred about 74,000 years ago (Chesner and
Fig. 1.50. Schematic representation of the formation of a
Rose, 1991). Approximately 2000 km3 of silicic
caldera during a large explosive eruption of a stratovolcano.
An intitial phase of pumice fallout (a) and (b) is followed by
magma was ejected during the eruption, the
the generation of pyroclastic flows (c). Rapid evacuation of greater part by pyroclastic flows that formed
the underlying magma chamber eventually results in the extensive ignimbrite deposits inside and outside
collapse of the volcanic edifice to form a caldera (d). Modified the caldera. The last Toba eruption occurred
from Williams and McBirney (1979). at about the same time that the Earth’s cli-
mate entered an ice age and it has been spec-
ulated that the injection of ash and aerosols
in Indonesia (Fig. 1.51). Prior to 1815, Tambora into the atmosphere might have been a trigger-
was a broad stratovolcano with an elevation of ing mechanism for catastrophic climatic change
about 4000 m. In 1815 the volcano produced the (Rampino and Self, 1992). Other examples of giant
largest explosive eruption of historic times, eject- calderas include many of the well-studied centers
ing about 50 km3 of magma during a 2-day span. in the western United States such as Yellowstone
Rapid evacuation of a shallow magma cham- in Wyoming, Valles in New Mexico, and Long
ber resulted in the generation of voluminous Valley in California.
pyroclastic flows and eventual collapse of the The size of the largest calderas, such as the
upper 1200 m of the summit. The resulting ones in the western United States, is so great
caldera is about 1250 m deep from its present that the collapse that formed them includes not
rim and about 6 km across. Studies of the erupted only pre-existing volcanic rocks but large pieces
products indicate that the caldera volume is sim- of the upper crust. It is difficult to imagine the
ilar to that of the widely dispersed pyroclastic scale of an eruption that is associated with this
material. type of structure. Observations of historic erup-
Tambora caldera is an impressive sight from tions are limited to the formation of relatively
the top of the volcano (Fig. 1.51) but its size is small calderas on existing volcanoes where only
dwarfed by the largest of calderas found on Earth. a few tens of cubic kilometers of magma were
These giant structures can be close to 100 km erupted. In contrast, the formation of the large
in diameter and in some cases are filled with calderas likely involves discharge of thousands of
water. One of the largest calderas in the world cubic kilometers of magma (Smith, 1979). Massive
REFERENCES 49

the existence of a topographic high in the cen-


tral part of the caldera. This center is attributed
to uplift after the formation of the caldera by
a process known as resurgence (Smith and Bai-
ley, 1968). In most cases this uplift is not associ-
ated with the eruption of magma. The ultimate
cause of resurgence is still being debated but
may be the result of upward pressure exerted by
magma remaining in the chamber as regional
subsidence occurs over an area larger than the
caldera or expansion of magma remaining the
chamber as a result of exsolution of dissolved
volatile components.
Calderas are also formed on basaltic shield
volcanoes, although their sizes are usually
smaller than those formed by explosive eruption
of evolved magmas. The caldera on the summit of
Kilauea volcano in Hawaii is about 4 km in diam-
eter with steep, almost vertical sides. Caldera
formation in basaltic systems typically involves
lateral transfer of magma away from a summit
magma chamber through a system of dykes and
Fig. 1.52. Lake Toba caldera on the island of Sumatra in fissures. For example, magma supplied to the
Indonesia. The caldera has been produced by multiple Kilauea summit is distributed through eruptions
large-volume silicic explosive eruptions. Modified from in the caldera and along the upper and middle
Francis (1993). east rift zones that extend more than 20 km from
the summit.

deposits of pyroclastic flows (ignimbrites) are gen-


erated during this type of event and may extend References
up to a hundred kilometers from source. In addi-
tion, because caldera collapse occurs over such a Anderson, A. T., Jr., Newman, S., Williams, S., et al.
large area magma can be discharged along many 1989. H2 O, CO2 , Cl, and gas in Plinian and
of the ring faults that define the subsiding block. ash-flow Bishop rhyolite. Geology, 17, 221–225.
Despite, or perhaps more accurately, because Ballard, R. and Moore, J. 1977. Photographic Atlas of the
of their great size, some calderas have eluded Mid-Atlantic Ridge Rift Valley. Berlin, Springer-
detection. A good example is the Cerro Galán Verlag.
caldera in northwest Argentina which was only Ballard, R., Holcomb, R., and van Andel, T. 1979. The
recognized as a caldera from satellite images Galapagos Rift at 86◦ W. III. Sheet flows, collapse
(Francis et al., 1978). Its structure exhibits many pits, and lava lakes of the rift valley. Journal of
Geophysical Research, 84, 5407–5422.
features of large calderas from around the world.
Barker, S. and Malone, S. 1991. Magmatic system
The boundaries are defined by a series of ring
geometry at Mount St. Helens modeled from the
fractures that have acted as passageways for
stress field associated with posteruptive
magma to escape from a large shallow cham- earthquakes. Journal of Geophysical Research, 96,
ber. Exposures of Paleozoic basement occur along 11883–11894.
the caldera margin indicating the involvement of Blackburn, E., Wilson, L., and Sparks, R. S. J. 1976.
large crustal blocks during the collapse. Another Mechanisms and dynamics of strombolian
prominent feature of the Cerro Galán caldera and activity. Journal of the Geological Society of London,
many other large calderas around the world is 132, 429–440.
50 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

Blake, S. 1989. Viscoplastic models of lava domes. plumes. Journal of Geophysical Research, 93,
IAVCEI Proceedings in Volcanology, Lava Flows and 15314–15328.
Domes, 2, 88–126. Caroll, M. and Rutherford, M. 1988. Sulfur speciation
Bonadonna, C., Ernst, G. G. J., and Sparks, R. S. J. 1998. in hydrous experimental glasses of varying
Thickness variations and volume estimates of oxidation state: results from measured
tephra fall deposits: the importance of particle wavelength shifts of sulfur x-rays. American
Reynolds number. Journal of Volcanology and Mineralogist, 73, 845–849.
Geothermal Research, 81, 173–187. Carr, M. J. 1984. Symmetrical and segmented
Bottinga, Y. and Weill, D. 1972. The viscosity of variation of physical and geochemical
magmatic silicate liquids: a model for calculation. characteristics of the Central American volcanic
American Journal of Science, 272, 438–475. front. Journal of Volcanology and Geothermal Research,
Brown, G. C., Hawkesworth, C., and Wilson, M. 1992. 20, 231–252.
Understanding the Earth. Cambridge, UK, Cashman, K. and Mangan, M. 1994. Physical
Cambridge University Press. aspects of magma degassing. II. Constraints on
Brown, J., Colling, A., Park, D., et al. 1989. The Ocean vesiculation processes from textural studies of
Basins: Their Structure and Evolution. New York, eruptive processes. Reviews in Mineralogy, 30,
Pergamon Press. 447–474.
Burnham, C. W. 1979. The importance of volatile Chadwick, W. and Embley, R. 1994. Lava flows from a
constituents. In H. Yoder (ed.) The Evolution of mid-1980s submarine eruption on the Cleft
Igneous Rocks. Princeton, NJ, Princeton University segment, Juan de Fuca Ridge. Journal of Geophysical
Press, pp. 439–478. Research, 99, 4761–4776.
Bursik, M. and Woods, A. 1996. The dynamics and Chadwick, W., Embley, R., and Fox, C. 1995. SeaBeam
thermodynamics of large ash flows. Bulletin of depth changes associated with recent lava flows,
Volcanology, 58, 175–193. CoAxial segment, Juan de Fuca Ridge: evidence
Bursik, M. I., Sparks, R. S. J., Gilbert, J. S., et al. for multiple eruptions between 1981–1993.
1992a. Sedimentation of tephra by volcanic Geophysical Research Letters, 22, 167–170.
plumes. I. Theory and its comparison with a Chesner, C. and Rose, W. I. 1991. Stratigraphy of the
study of the Fogo A plinian deposit, São Miguel Toba Tuffs and evolution of the Toba caldera
(Azores). Bulletin of Volcanology, 54, 329–344. complex, Sumatra, Indonesia. Bulletin of
Bursik, M., Carey, S., and Sparks, R. S. J. 1992b. A Volcanology, 53, 343–356.
gravity current model for the May 18, 1980 Chouet, B., Hamisevicz, N., and McGuetchin, T. R.
Mount St. Helens plume. Geophysical Research 1974. Photoballistics of volcanic jet activity at
Letters, 19, 1663–1666. Stromboli, Italy. Journal of Geophysical Research, 79,
Carey, S. N. 1991. Transport and deposition of tephra 4961–4976.
by pyroclastic flows and surges. In R. V. Fisher and Clague, D. and Dalrymple, G. B. 1987. The
G. Smith (eds.) Sedimentation in Volcanic Settings, Hawaiian–Emperor volcanic chain. I. Geologic
Special Publication no. 45. Tulsa, OK, Society for evolution. In Volcanism in Hawaii, US Geological
Sedimentary Geology, pp. 39–57. Survey Professional Paper no. 1350. Washington,
Carey, S. N. and Sigurdsson, H. 1985. The May 18, 1980 DC, US Government Printing Office, pp. 5–54.
eruption of Mount St. Helens. II. Modeling of Coffin, M. and Eldholm, O. 1994. Large igneous
dynamics of the plinian phase. Journal of provinces: crustal structure, dimensions, and
Geophysical Research, 90, 2948–2958. environmental consequences. Reviews of Geophysics,
1987. Temporal variations in column height and 32, 1–36.
magma discharge rate during the 79 AD eruption Colgate, S. and Sigurgeirsson, T. 1973. Dynamic
of Vesuvius. Geological Society of America Bulletin, 99, mixing of water and lava. Nature, 244, 552–555.
303–314. Crisp, J. A. 1984. Rates of magma emplacement and
Carey, S. N. and Sparks, R. S. J. 1986. Quantitative volcanic output. Journal of Volcanology and
models of the fallout and dispersal of tephra Geothermal Research, 20, 177–211.
from volcanic eruption columns. Bulletin of Decker, R. and Decker, B. 1997. Volcanoes. New York,
Volcanology, 48, 109–125. W. H. Freeman.
Carey, S. N., Sigurdsson, H., and Sparks, R. S. J. Dingwell, D. B. and Webb, S. L. 1989. Structural
1988. Experimental studies of particle-laden relaxation in silicate melts and non-Newtonian
REFERENCES 51

melt rheology in geologic processes. Physics and the East Pacific Rise at 12◦ 35 N–12◦ 51 N and
Chemistry of Minerals, 16, 508–516. 11◦ 26 N–11◦ 30 N. Journal of Geophysical Research,
1990. Relaxation in silicate melts. European Journal of 94, 17437–17463.
Mineralogy, 2, 427–449. Hulme, G. 1974. Interpretation of lava flow
Druitt, T. and Sparks, R. S. J. 1984. On the formation morphology. Royal Astronomical Society Geophysical
of calderas during ignimbrite eruptions. Nature, Journal, 39, 361–383.
310, 679–681. Humphris, S., Zierenberg, R., Mullineaux, L., et al.
Duncan, R. and Richards, M. 1991. Hotspots, mantle 1995. Seafloor Hydrothermal Systems: Physical,
plumes, flood basalts and true polar wanderings. Chemical, Biological and Geological Interactions,
Reviews of Geophysics, 29, 31–50. Geophysical Monograph no. 91. Washington, DC,
Fisher, R. V. and Schmincke, H. U. 1984. Pyroclastic American Geophysical Union.
Rocks. New York, Springer-Verlag. Hurwitz, S. and Navon, O. 1994. Bubble nucleation in
Fisher, R. V., Smith, A. L., and Roobol, M. J. 1980. rhyolitic melts: experiments at high pressure,
Destruction of St. Pierre, Martinique, by ash temperature and water content. Earth and
cloud surges. Geology, 8, 472–476. Planetary Science Letters, 122, 267–280.
Francheteau, J. and Ballard, R. 1983. The East Pacific Iyer, H., Evans, J., Dawson, P., et al. 1990. Differences
Rise near 21◦ N, 13◦ N, and 20◦ N: inferences for in magma storage in different volcanic
along-strike variability of axial processes of the environments as revealed by seismic tomography:
mid-ocean ridge. Earth and Planetary Science Letters, silicic volcanic centers. In M. Ryan (ed.) Magma
64, 93–116. Transport and Storage. Chichester, UK, John Wiley,
Francheteau, J., Juteau, T., and Rangin, C. 1979. pp. 293–316.
Basaltic pillars in collapsed lava-pools on the Karig, D. 1971. Origin and development of marginal
deep sea floor. Nature, 281, 209–211. basins in the western Pacific. Journal of Geophysical
Francis, P. 1993. Volcanoes: A Planetary Perspective. Research, 76, 2542–2561.
Oxford, UK, Oxford University Press. Keary, P. and Vine, F. 1990. Global Tectonics. Oxford, UK,
Francis, P. W., Hammill, M., Kretzschmar, G., et al. Blackwell Scientific Publications.
1978. The Cerro Galán Caldera, Northwest Kokelaar, P. 1983. The mechanism of Surtseyan
Argentina. Nature, 274, 749–751. volcanism. Journal of the Geological Society of London,
Frankel, C. 1996. Volcanoes of the Solar System. 140, 939–944.
Cambridge, UK, Cambridge University Press. 1986. Magma–water interactions in subaqueous and
Greeley, R. 1982. The Snake River Plain, Idaho: emergent basaltic volcanism. Bulletin of
representative of a new category of volcanism. Volcanology, 48, 275–289.
Journal of Geophysical Research, 87, 2705–2712. Koyaguchi, T. and Tokuno, M. 1993. Origin of the
Gregg, T. and Fink, J. 1995. Quantification of giant eruption cloud of Pinatubo, June 15, 1991.
submarine lava-flow morphology through analog Journal of Volcanology and Geothermal Research, 55,
experiments. Geology, 23, 73–76. 85–96.
Griffiths, R. and Fink, J. 1992. Solidification and Lacroix, A. 1904. La Montagne Pelée et ses eruptions. Paris,
morphology of submarine lavas: a dependence in Masson.
extrusion rate. Journal of Geophysical Research, 97, Le Maitre, R. 1976. The chemical variability of some
19729–19737. common igneous rocks. Journal of Petrology, 17,
Gudmunsson, A. 1996. Volcanoes in Iceland. Reykjavik, 589–637.
Vaka-Helgafell. Lipman, P. and Mullineaux, D. 1981. The 1980
Head, J. and Wilson, L. 1987. Lava fountain heights at Eruptions of Mount St. Helens, Washington, US
Pu’u ‘O’o, Kilauea, Hawaii: indicators of amount Geological Survey Professional Paper no. 1250.
and variations of exsolved magma volatiles. Washington, DC, US Government Printing
Journal of Geophysical Research, 92, 13715–13719. Office.
1992. Magma reservoirs and neutral buoyancy Macdonald, K. C., Scheirer, C., and Carbotte, S. 1991.
zones on Venus: implications for the Mid-ocean ridges: discontinuities, segments, and
formation and evoution of volcanic landforms. giant cracks. Science, 253, 986–994.
Journal of Geophysical Research, 97, 3877–3903. Mader, H., Zhang, Y., Phillips, J., et al. 1994.
Hekinian, R., Thompson, G., and Bidcau, D. 1989. Axial Experimental simulations of explosive degassing
and off-axial heterogeneity of basaltic rocks from of magma. Nature, 372, 85–88.
52 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

Malin, M. 1980. The lengths of Hawaiian lava flows. Rampino, M. 1988. Introduction: the volcano/climate
Geology, 8, 306–308. connection. Annual Reviews of Earth and Planetary
Marsh, B. 1979. Island-arc volcanism. American Scientist, Sciences, 16, 73–99.
67, 161–172. Rampino, M. and Self, S. 1992. Volcanic winters and
1981. On the crystallinity, probability of occurrence, accelerated glaciations following the Toba
and rheology of lava and magma. Contributions to super-eruption. Nature, 359, 50–52.
Mineralogy and Petrology, 78, 85–98. Richards, M., Duncan, R., and Courtillot, V. 1989.
McBirney, A. R. 1973. Factors governing the intensity Flood basalts and hot spot tracks: plume heads
of explosive andesitic eruptions. Bulletin of and tails. Science, 246, 103–107.
Volcanology, 36, 443–453. Riedel, C., Ernst G. G. J., and Riley, M. 2003. Controls
Molnar, P. and Atwater, T. 1978. Interarc spreading on the growth and geometry of pyroclastic
and Cordilleran tectonics as alternates related to constructs. Journal of Volcanology and Geothermal
the age of subducted oceanic lithosphere. Earth Research, 127, 121–152.
and Planetary Science Letters, 41, 330–340. Ringwood, A. E. 1975. Composition and Petrology of the
Moore, J. 1975. Mechanism of formation of pillow Earth’s Mantle. New York, McGraw-Hill.
lava. American Journal of Science, 63, 269–277. Ross, C. and Smith, R. L. 1961. Ash-Flow Tuffs: Their
Moore, J. G. and Normark, W. 1994. Giant Hawaiian Origin, Geological Reactions and Identification. US
landslides. Annual Review of Earth and Planetary Geological Survey Professional Paper no. 366.
Sciences, 22, 119–144. Washington, DC, US Government Printing Office.
Moore, J. G., Bryan, W. B., and Ludwig, K. 1994. Chaotic Rowland, S. and Walker, G. P. L. 1990. Pahoehoe and
deposition by a giant wave, Molokai, Hawaii. aa in Hawaii: volumetric flow rate controls the
Geological Society of America Bulletin, 106, 962–967. lava structure. Bulletin of Volcanology, 52, 615–628.
Moore, J. G., Bryan, W., Beeson, M., et al. 1995. Giant Rutherford, M. J. and Devine, J. 1988. The May 18,
blocks in the South Kona landslide, Hawaii. 1980 eruption of Mount St. Helens. III. Stability
Geology, 23, 125–128. and chemistry amphibole in the magma
Pallister, J., Hoblitt, R., Crandell, D., et al. 1992. Mount chamber. Journal of Geophysical Research, 93, 1310,
St. Helens a decade after the 1980 eruptions: 11949–11959.
magmatic models, chemical cycles, and a revisted Rutherford, M., Sigurdsson, H., Carey, S., et al. 1985.
hazards assessment. Bulletin of Volcanology, 54, The May 18, 1980 eruption of Mount St. Helens. I.
126–146. Melt composition and experimental phase
Parfitt, E. and Wilson, L. 1995. Explosive volcanic equilibria. Journal of Geophysical Research, 90,
eruptions. IX. The transition between 2929–2947.
Hawaiian-style lava fountaining and strombolian Ryan, M. P. 1987. Neutral buoyancy and the
explosive activity. Geophysical Journal International, mechanical evolution of magmatic systems. In
121, 226–232. B. O. Mysen (ed.) Magmatic Processes: Physiochemical
Peacock, S. M. 1996. Thermal and petrologic structure Principles. Special Publication no. 1. Chichester,
of subduction zones. In G. Bebout, D. Scholl, S. UK, Geochemical Society, pp. 259–287.
Kirby, et al. (eds.) Subduction Top to Bottom. 1988. The mechanics and 3-d internal structure of
Geophysical Monograph no. 96. Washington, DC, active magma systems: Kilauea volcano, Hawaii.
American Geophysical Union, pp. 119–134. Journal of Geophysical Research, 93, 4213–4248.
Peterson, D. and Tilling, R. 1980. Transition of basaltic Ryan, M. P. and Blevins, J. Y. K. 1987. The viscosity of
lava from pahoehoe to aa, Kilauea volcano, synthetic and natural silicate melts and glasses at
Hawaii: field observations and key factors. Journal high temperature and 1 bar (105 Pascals) pressure
of Volcanology and Geothermal Research, 7, 271–293. and at higher pressure. US Geological Survey
Pinkerton, H. and Sparks, R. S. J. 1976. The 1975 Bulletin, 1764, 1–563.
subterminal lavas, Mount Etna: a case history of Scandone, R. and Malone, S. 1985. Magma supply,
the formation of a compound lava field. Journal of magma discharge and readjustment of the
Volcanology and Geothermal Research, 1, 167–182. feeding system of Mount St. Helens during 1980.
Pinkerton, H. and Stevenson, R. 1992. Methods of Journal of Volcanology and Geothermal Research, 23,
determining the rheological properties of 239–262.
magmas at sub-liquidus temperatures. Journal of Schmincke, H. U. 1994. Geological Field Guide of Gran
Volcanology and Geothermal Research, 53, 47–66. Canaria, 6th edn. Kiel, Pluto Press.
REFERENCES 53

Schmincke, H. U., Weaver, P. P. E., Firth, J. V., et al. Sparks, R. S. J. and Wilson, L. 1976. Model for the
1995. Proceeding of the ODP, Initial reports no. 157. formation of ignimbrite by gravitational column
College Station, TX, Ocean Drilling Program. collapse. Journal of the Geological Society of London,
Schouten, H., Klitgord, K., and Whitehead, J. 1985. 132, 441–452.
Segmentation of mid-ocean ridges. Nature, 317, Sparks, R. S. J., Moore, J. G., and Rice, C. J. 1986. The
225–229. initial giant umbrella cloud of the May 18, 1980
Self, S., Wilson, L., and Nairn, L. 1979. Vulcanian explosive eruption of Mount St. Helens. Journal of
eruption mechanisms. Nature, 277, 440–443. Volcanology and Geothermal Research, 28, 257–274.
Siebert, L. 1984. Large volcanic debris avalanches: Sparks, R. S. J., Barclay, J., Jaupart, C., et al. 1994.
characteristics of source areas, deposits and Physical aspects of magma degassing. I.
associated eruptions. Journal of Volcanology and Experimental and theoretical constraints on
Geothermal Research, 22, 163–197. vesiculation. Reviews in Mineralogy, 30, 413–443.
Siebert, L., Glicken, H. X., and Ui, T. 1987. Volcanic Sparks, R. S. J., Bursik, M., Carey, S., et al. 1997. Volcanic
hazards from Bezymianny- and Bandai-type Plumes. Chichester, UK, John Wiley.
eruptions. Bulletin of Volcanology, 49, 435–459. Steven, T. A. and Lipman, P. W. 1976. Calderas of the San
Sigurdsson, H. 1990. Evidence of volcanic loading of Juan Volcanic Field, Southwestern Colorado, US
the atmosphere and climate response. Geological Survey Professional Paper no. 958.
Paleogeography, Paleoclimatology, Paleoecology, 89, Washington, DC, US Government Printing Office.
277–289. Stolper, E. and Newman, S. 1994. The roll of water in
Sigurdsson, H., Carey, S., Cornell, W., et al. 1985. The the petrogenesis of Mariana Trough magmas.
eruption of Vesuvius in AD 79. National Geographic Earth and Planetary Science Letters, 121, 293–325.
Research, 1(3), 332–387. Swanson, D., Wright, T., and Helz, R. 1975. Linear vent
Simpkin, T. 1993. Terrestrial volcanism in space and systems and estimated rates of eruption for the
time. Annual Reviews of Earth and Planetary Sciences, Yakima basalt on the Columbia Plateau. American
21, 427–452. Journal of Science, 275, 877–905.
Sinton, J. M. and Detrick, R. S. 1992. Mid-ocean ridge Tighe, S. 1997. The morphological characteristics and
magma chambers. Journal of Geophysical Research, geological implications of prime segments and
97, 197–216. axial volcanoes along the East Pacific Rise.
Sisson, T. and Grove, T. 1993. Experimental Unpublished Ph.D. thesis, University of Rhode
investigations of the role of H2 O in calc–alkaline Island.
differentiation and subduction zone magmatism. Tilling, R. I. 1989. Introduction and overview. In R. I.
Contributions to Mineralogy and Petrology, 113,143–166. Tilling (ed.) Short Course in Geology, vol. 1, Volcanic
Smith, R. L. 1979. Ash flow magmatism. Special Paper Hazards. Washington, DC, American Geophysical
no. 180. Boulder, CO, Geological Society of Union, pp. 1–8.
America, pp. 5–27. Uyeda, S. 1982. Subduction zones: an introduction to
Smith, R. L. and Bailey, R. A. 1968. Resurgent comparative subductology. Tectonophysics, 81,
cauldrons. In R. L. Hay and C. A. Anderson (eds.) 133–159.
Studies in Volcanology, Geological Society of Uyeda, S. and Kanamori, H. 1978. Back-arc opening
America Memoir no. 116. Boulder, CO, Geological and the mode of subduction. Journal of Geophysical
Society of America, pp. 153–210. Research, 84, 1049–1061.
Smith, T. and Batiza, R. 1989. New field and laboratory Verhoogen, J. 1951. Mechanics of ash formation.
evidence for the origin of hyaloclastite flows on American Journal of Science, 249, 239–246.
seamount summits. Bulletin of Volcanology, 51, Vine, F. J. and Matthews, D. H. 1963. Magnetic
96–114. anomalies over oceanic ridges. Nature, 199,
Sparks, R. S. J. 1978. The dynamics of bubble 947–949.
formation and growth in magmas. Journal of Walker, G. P. L. 1974. The lengths of lava flows.
Volcanology and Geothermal Research, 3, 1–37. Philosophical Transactions of the Royal Society of
1986. The dimensions and dynamics of volcanic London, Series A, 274, 107–118.
eruption columns. Bulletin of Volcanology, 48, 3–15. White, R. and McKenzie, D. 1989. Magmatism at rift
1992. Magma generation in the Earth. In zones: the generation of volcanic continental
Understanding the Earth. Cambridge, UK, margins and flood basalts. Journal of Geophysical
Cambridge University Press, pp. 91–114. Research, 94, 7685–7729.
54 UNDERSTANDING THE PHYSICAL BEHAVIOR OF VOLCANOES

Whitehead, J. A., Dick, H. J. B., and Schouten, H. 1984. mechanisms, and field studies. Bulletin of
A mechanism for magmatic accretion under Volcanology, 48, 245–264.
spreading centers. Nature, 312, 146–148. Wohletz, K. and McQueen, R. 1984. Experimental
Williams, H. and McBirney, A. R. 1979. Volcanology. San studies of hydromagmatic volcanism. In Explosive
Francisco, CA, Freeman, Cooper and Co. Volcanism: Inception, Evolution, and Hazards.
Wilson, J. T. 1973. Mantle plumes and plate motions. Washington, DC, National Academy Press,
Tectonophysics, 19, 149–164. pp. 158–169.
Wilson, L. 1980. Relationships between pressure, Wolff, J. and Wright, J. 1981. Rheomorphism of
volatile content and ejecta velocity in three types welded tuffs. Journal of Volcanology and Geothermal
of volcanic explosion. Journal of Volcanology and Research, 10, 13–34.
Geothermal Research, 8, 297–313. Woods, A. W. 1988. The dynamics and
Wilson, L., Sparks, R. S. J., and Walker, G. P. L. 1980. thermodynamics of eruption columns. Bulletin of
Explosive volcanic eruptions. IV. The control Volcanology, 50, 169–191.
of magma properties and conduit geometry Woods, A. and Wohletz, K. 1991. Dimensions and
on eruption column behaviour. Geophysical dynamics of co-ignimbrite eruption columns.
Journal of the Royal Astronomy Society, 63, Nature, 350, 225–227.
117–148. Yoder, H. S. 1976. Generation of Basaltic Magma.
Wilson, M. 1989. Igneous Petrogenesis: A Global Tectonic Washington, DC, National Academy of Sciences.
Approach. London, Unwin Hyman. Young, S., Sparks, R. S. J., Robertson, R., et al. 1997.
Wohletz, K. 1986. Explosive magma–water Eruption of Soufrière Hills volcano in Montserrat
interactions: thermodynamics, explosion continues. Eos, 78, 401–409.
Chapter 2

Volcano hazards
Robert I. Tilling

Introduction and convergent boundaries between the tectonic


plates and above well-documented intraplate
Not only do ‘‘Volcanoes assail the senses . . .” ‘‘hot spots,” such as Hawaii and the Galapagos
(Decker and Decker, 1997, p. vii), but they also (Simkin et al., 1994). Even though convergent-
assail the environment when they erupt, terrify- plate (subduction-zone) volcanism accounts for
ing and fascinating humankind for countless mil- only about 15% of the averaged global volcanic
lennia. Volcanic processes and products – benefi- output (Simkin, 1993), it produces more than
cial and hazardous – have profoundly impacted 80% of documented historical eruptions (Simkin
and continue to impact society (Chester, 1993, and Siebert, 1994; Tilling, 1996) (Fig. 2.1a). This
Chapter 14, this volume). reflects the fact that the overwhelming majority
It is estimated that about 10% of the world’s of the Earth’s eruptive activity takes places on the
population live within proximity of active and deep ocean floor along the global oceanic ridge
potentially active volcanoes. With projected pop- systems – sight unseen and posing no volcanic
ulation growth, by the twenty-first century more hazards.
than 500 million people could be at risk from vol- In a previous review of the topic of vol-
cano hazards (Peterson, 1986; Tilling and Lipman, canic hazards, British volcanologist George P. L.
1993). Of these half billion people at risk, roughly Walker commented (wryly?): ‘‘In principle, vol-
equal to the estimated entire world population at canic risk could be eliminated by the total aban-
the beginning of the seventeenth century, about donment of all volcanic areas.” (Walker, 1982,
90% live in the circum-Pacific region. Also in p. 156). However, he also stated clearly that such
the twenty-first century, with continued increas- an option ‘‘is not realistic” because these areas
ing urbanization there will be more than 100 are already inhabited – some for many centuries
cities with greater than 2 million population, or millennia. Indeed, the combined pressures of
and many of these will be located within 200 km population rise and urbanization, human set-
of volcanically active subduction zones (McGuire, tlement, agricultural cultivation, and industrial
1995a, Fig. 15.1). development increasingly encroach upon volca-
Over 1500 subaerial volcanoes have been noes heretofore considered ‘‘remote” and, hence,
active during the Holocene (i.e., the past 10 000 presumed to pose minimal risk to people and
years), and more than a third of these have property. Moreover, explosive eruptions of still-
erupted one or more times during recorded his- remote volcanoes (e.g., many along the Aleutian,
tory (Simkin and Siebert, 1994). Active or poten- Kamchatka, and Kurile volcanic arcs) constitute
tially active volcanoes occur in narrow belts a substantial and growing risk to aircraft flying
that collectively comprise less than 1% of the over them along the heavily traveled North Pacific
Earth’s total surface area, dotting the divergent air routes (Casadevall, 1994a, 1994b; Casadevall
Volcanoes and the Environment, eds. J. Martı́ and G. G. J. Ernst. Published by
Cambridge University Press.  C Cambridge University Press 2005.
56 VOLCANO HAZARDS

Fig. 2.1. (a) A powerful explosive eruption of Mt. Pinatubo, the more severe will be the losses once an eruption
Philippines, on June 12, 1991; the climactic eruption on June occurs.
15 was more voluminous (∼5 km3 ), perhaps the second
largest eruption in the twentieth century. About 250 times Thus, a realistic approach to reducing volcano
larger than the 1985 Ruiz eruption, the Pinatubo eruption risk must necessarily embody the main theme of
had, and continues to have, tremendous impact on the the 1988 Kagoshima International Conference on
environment, yet it caused only a few hundred deaths (for Volcanoes: ‘‘Towards better coexistence between
reasons, see discussion on the case histories of two human beings and volcanoes” (Kagoshima Con-
eruptions). Photograph by David H. Harlow (US Geological ference, 1988). We must become more resourceful
Survey). (b) Aerial view of the remains of the city of Armero,
in enjoying the long-term benefits of volcanoes
which was devastated by lahars triggered by a small-volume
during their repose, while planning for, and cop-
(0.02 km3 ) magmatic eruption of Nevado del Ruiz, Colombia,
on November 13, 1985; the lahars poured from the canyon
ing with, the short-term hazardous consequences
mouth of Rı́o Lagunillas (upper right corner), killing more when they erupt.
than 22 000 people. The Ruiz catastrophe is the worst This chapter is not intended as a comprehen-
volcanic disaster in the world since the 1902 eruption of sive treatment of the wide-ranging topic of vol-
Mt. Pelée (Martinique). (Photograph by Darrell G. Herd, US cano hazards, but instead it highlights some suc-
Geological Survey.) cesses and failures in volcano-hazards mitigation
as well as some challenging issues confronting
volcanologists in this century. In this selective
et al., 1999). David Chester (1993, p. 232) cogently
review, I draw liberally, in places verbatim, from
summarizes the crux of the problem:
several relatively recent summary papers and
Without people there can be no natural hazards; they books on the topic: Blong (1984); Crandell et al.
are an artifice of the interaction between people and (1984); Latter (1988); Tilling (1989a, 1989b); Scott
nature. In general, the greater the population density (1989a, 1989b); Ewert and Swanson (1992); Chester
and the higher the level of economic development, (1993); Peterson and Tilling (1993); Tilling and
VOLCANO HAZARDS 57

Fig. 2.1. (cont.) Ironically, earlier in the nineteenth century,


the 1815 eruption of Tambora (Sumbawa Island,
Indonesia), the largest and deadliest eruption
Lipman (1993); Simkin and Siebert (1994); Mc-
(92 000 deaths) in recorded history, received scant
Guire et al. (1995); and Scarpa and Tilling (1996).
notice worldwide because of the volcano’s remote
For more detailed discussion of any particular
location and the poor global communications
aspect of volcano hazards, the interested reader
and meager scientific understanding of volcanic
is referred to these works and the references con-
phenomena prevailing at the time. It was not
tained therein.
until the 1902 eruption of Mt. Pelée (island of
Martinique, West Indies), whose pyroclastic flows
Volcano hazards obliterated the city of St. Pierre and swiftly
killed all but two of its 30 000 inhabitants, that
Hazardous processes associated with volcanic the devastating impacts of volcano hazards cap-
eruptions largely were little appreciated perhaps tured widespread public and scientific attention,
until the 1883 eruption of Krakatau (Sunda Strait, spurring the establishment of instrumented vol-
between Java and Sumatra, Indonesia), when cano observatories around the world and foster-
more than 36 000 people were killed, mostly ing the development of the modern science of
by volcanogenic tsunamis (Simkin and Fiske, volcanology.
1983, 1984). The news of this destructive erup- The powerful impact of hazardous volcanic
tion quickly became public knowledge – via the processes on the environment and society has
newly invented telegraph communications sys- been vividly demonstrated by several eruptions
tem – following observations by the Dutch and in recent decades, especially the voluminous
British scientific expeditions dispatched quickly explosive eruptions of Mt. Pinatubo (Luzon,
to the scene (Verbeek, 1885; Symons, 1888). Philippines) in 1991 (Newhall and Punongbayan,
58 VOLCANO HAZARDS

Fig. 2.2. Looking eastward from the site of the former the Caribbean region and has resulted in substan-
capital city of Plymouth, near Sagar Bay on the southwestern tial socio-economic impact and human misery for
side of Montserrat. Structures have been damaged and the approximately 12 000 inhabitants of this tiny
destroyed by a combination of pyroclastic flows and lahars. island (<160 km2 in area). For a detailed account
Soufrière Hills Volcano is visible in the background; the
of the eruptive activity through 1997 and related
pyroclastic flows originated from collapse of the unstable lava
some located between the two prominent peaks.
topical studies, see Aspinall et al. (1998), Young
(Photograph by Richard P. Hoblitt, US Geological Survey.) et al. (1998b), and Druitt and Kokelaar (2002), but
a brief summary is given below.
The initial phase of the Soufrière Hills erup-
1996b) (Fig. 2.1a). However, eruptions need not tion (July to November 1995) was characterized
be large to be deadly and (or) cause massive by phreatic explosions producing steam and ash
socio-economic disruption. For example, the 1985 columns, the most energetic of which gener-
eruption of Nevado del Ruiz (Colombia) was ated cold base-surge ash clouds and light ash-
quite small, erupting only ∼0.02 km3 of magma falls of non-juvenile materials blasted from the
compared to the ∼5 km3 for the climactic existing domes (Young et al., 1998a). The erup-
eruption of Pinatubo, but it resulted in more tive activity became magmatic in mid November
than 22 000 fatalities (Williams, 1990a, 1990b) 1995 and has involved pyroclastic flows, surges
(Fig. 2.1b). Moreover, the hazardous juxtaposition and co-ignimbrite ash clouds, generated by a
of society and active volcanoes also is emphasized series of collapses of actively growing, gravita-
by the recent eruption of Soufrière Hills Volcano, tionally unstable lava domes. Tephra falls and
on the island of Montserrat (British West Indies), pyroclastic flows from an explosive eruption in
located approximately 250 km north-northwest of mid September 1996 devastated all flanks of the
Mt. Pelée along the Caribbean volcanic arc. This volcano (Fig. 2.2), which make up the entire
eruption, which began in mid July 1995, was the southern half of the island (Young et al., 1998a).
first on Montserrat since European settlement in Pyroclastic flows and associated ash-cloud surges
VOLCANO HAZARDS 59

triggered by a large dome collapse on June 25, half of island’s pre-crisis population, are still liv-
1997 destroyed hundreds of houses and killed ing elsewhere (Rozdilsky, 2001). The Montserrat
19 people (Aspinall et al., 1998). Rapid magma and British governments have developed long-
extrusion immediately ensued and, by Decem- range plans (e.g., Government of Montserrat and
ber 25 (Christmas Day) 1997, the lava dome had Her Majesty’s Government, 1998) for rehabilita-
attained a volume of about 115 × 106 m3 . Then tion of the island and restoration of its infras-
on December 26 (Boxing Day) 1997, a massive col- tructure, to allow the eventual return of the
lapse of the dome occurred, involving about 40% people evacuated. Aspinall et al. (1998, p. 3387)
of its volume and producing debris avalanches, remarked: ‘‘Montserrat is a small island, and the
pyroclastic flows and surges, co-ignimbrite ash hazards there are hard to avoid . . .” According
clouds, lateral blasts, and a small tsunami. The to media reports and observers, toward the end
Boxing Day processes and deposits, which per- of 2001 infrastructure rebuilding is under way
haps constituted the volcano’s most intense out- and the general quality of life is beginning to
burst to date, produced no human fatalities but improve, but housing remains in short supply.
severely damaged or completely buried several However, entering the year 2004, Soufrière Hills
(already evacuated) settlements 2–3 km downs- Volcano remains intermittently active with dome
lope from the summit. After the collapse, magma growth and collapses, rockfalls, and pyroclastic
extrusion and dome growth quickly resumed and flows. Because of the continuing activity and
continued at varying rates through mid March associated volcanic hazards, the southern two-
1998, accompanied by small collapses, explosions, thirds of the island, with exception of a small
and weak to moderate ash venting. From April daytime entry zone on the western coast, remains
1998 through mid November 1999, even though an Exclusion Zone (Montserrat Volcano Observa-
no new magma was added to the dome com- tory, 2001), which is off-limits to ordinary human
plex, numerous so-called ‘‘passive” gravitational activities and only accessible for purposes of sci-
collapses took place. These events typically only entific monitoring and national security. Thus,
produced small to moderate rockfalls and local- the volcanic crisis that began at Montserrat in
ized associated pyroclastic flows and ash clouds, 1995 is persisting into the twenty-first century.
but some resulted in substantial changes in dome Hazardous volcanic processes (i.e., volcano
morphology. On November 29, however, another hazards) have been well described elsewhere (e.g.,
new dome emerged, and the first new magma Macdonald, 1972; Blong, 1984; Crandell et al.,
erupted since March 1998. Dome growth then 1984a; Scott, 1989a; Myers et al., 1997; Sigurdsson
continued episodically through 2001, and the et al., 2000; US Geological Survey, 2005). There-
dome reached its largest size (>120 × 106 m3 ) fore, only a short, selective summary is given
and highest elevation (>1000 m) since the erup- herein, drawing heavily from Blong (1984) and,
tion began in 1995. especially, Scott (1989a) and emphasizing the haz-
Because of the intensive volcano monitoring ards that have occurred frequently in histori-
and effective communications of hazard warn- cal time. Volcano hazards can be grouped con-
ings to government officials by the newly and veniently into two broad categories – direct and
quickly established Montserrat Volcano Obser- indirect (Table 2.1) – although the distinction
vatory, eruption-caused fatalities were relatively between them can be arbitrary and subjective in
few. However, nearly all of the island’s means some instances, depending on the time elapsed
of livelihood and infrastructure, including the between the eruptive activity and the occurrence
principal city of Plymouth and the airport, were of the hazardous event(s).
lost during the eruption, necessitating the evac-
uation of most of the population to the north-
ern (less hazardous) end of the island and to Direct volcano hazards
other locales, including Great Britain. In late Hazardous events that are produced during
1997, only about 3300 people remained on the or shortly following the eruption (i.e., within
island; at present, about 6500 people, or about minutes to several days) are considered to be
60 VOLCANO HAZARDS

Table 2.1 Principal types of volcanic hazards and selected examples

Type Selected examples References

Direct volcano hazards


Fall processes
Tephra falls Rabaul, 1994 Blong and McKee (1995)
Ballistic projectiles Soufrière (St. Vincent), 1812 Anderson and Flett (1903)
Flowage processes
Pyroclastic flows, surges Mt. Pelée, 1902 Fisher et al. (1980)
Laterally directed blasts Bezimianny, 1956 Gorshkov (1959); Belousov (1996)
Mount St. Helens, 1980 Hoblitt et al. (1981); Kieffer (1981)
Debris avalanches Mount St. Helens, 1980 Voight et al. (1981); Glicken (1998)
Primary debris flows Nevado del Ruiz, 1985 Pierson and Janda (1990)
(lahars)
Floods (jökulhlaups) Katla, 1918 Thorarinsson (1957)
Grı́msvötn (Vatnajökull), 1996 Gudmundsson et al. (1997);
Jónsson et al. (1998)
Lava flows Kilauea, 1959–60 Macdonald (1962)
Kilauea, 1983–present Wolfe et al. (1988); Heliker et al.
(1998)
Other processes
Phreatic explosions Soufrière (Guadeloupe), 1976 Feuillard et al. (1983)
Volcanic gases and acid Dieng Plateau (Indonesia), 1979 Le Guern et al. (1982)
rains Lake Nyos (Cameroon), 1986 Kling et al. (1987)
Kilauea, 1983–present Sutton et al. (1997)
Long Valley Caldera, Farrar et al. (1995); Sorey et al.
1989–present (1998)
Indirect volcano hazards
Earthquakes and ground Sakurajima, 1914 Shimozuru (1972)
movements
Tsunami (seismic seawave) Krakatau, 1883 Simkin and Fiske (1983)
Secondary debris flows Mt. Pinatubo, 1991–2 Rodolfo et al. (1996)
(lahars)
Secondary pyroclastic flows Mt. Pinatubo, 1991–3 Torres et al. (1996)
Post-eruption erosion and Mt. Pinatubo, 1991–4 Punongbayan et al. (1996a)
sedimentation
Atmospheric effects Mayon, 1814 COMVOL (1975)
Climate change Tambora, 1815–16 Stommel and Stommel (1983)
Mt. Pinatubo, 1991–3 Self et al. (1996)
Post-eruption famine and Lakagı́gar (Laki), 1783 Thorarinsson (1979)
disease
Aircraft encounters with Redoubt, 1989–90 Casadevall (1994a)
volcanic ash Mt. Pinatubo, 1991 Casadevall et al. (1996)

Source: Updated and expanded from Tilling (1989a, Table 2).


VOLCANO HAZARDS 61

direct volcano hazards. Direct hazards include around the volcano. For example, the 1.1 km3
(see Table 2.1): fall processes, involving the fall (bulk volume) of tephra ejected during the
and accumulation of air-borne volcanic ejecta; 9-hour-long climactic eruption (May 18, 1980) of
flowage processes, involving eruption-triggered, Mount St. Helens covered more than 57 000 km2 ,
ground-hugging movement and deposition of pri- with accumulated thickness of many meters in
mary volcanic materials and (or) of mixtures of the proximal areas and measurable thicknesses
volcanic debris with eruption-induced meltwa- (≥0.5 mm) at distances of more than 700 km
ter or other runoff (e.g., co-eruption rainfall or from the vent (Sarna-Wojcicki et al., 1981). While
evacuation of crater lakes); and other processes, it had a devastating impact on surrounding
involving phreatic explosions, emission of vol- areas, this eruption was a relatively modest-
canic gases, and eruption-related acid rains or size event, when compared to some voluminous
volcanic air pollution. Some common fall and explosive eruptions in the geologic past, such as
flow processes are briefly discussed below to pro- those forming Yellowstone Caldera (northwestern
vide context for the discussion to follow. However, Wyoming) and Long Valley Caldera (east-central
because the hazards associated with phreatic California) in the United States. These huge erup-
explosions, volcanic gases, and acid rains are com- tions erupted several orders of magnitude more
paratively minor, they will not be considered in magma (102 –103 km3 ) (Christiansen, 1984) than
this chapter, and the interested reader is referred did Mount St. Helens in 1980, and the impact
to the excellent summaries in Blong (1984) and of the tephra falls from such eruptions covered
Scott (1989a). areas on a continental scale.
The high-velocity ejection and fall of large bal-
listic projectiles (volcanic bombs and blocks) during
Fall processes
energetic explosive eruptions constitute another
The ejection and deposition of air-borne fragmen-
common, but more localized, fall hazard. Because
tal volcanic materials from explosive eruptions
such projectiles exit the vent at speeds of tens to
perhaps constitute the most common, if not most
hundreds of meters per second on trajectories,
severe, volcano hazard. The origin of the frag-
they are only minimally affected by eruption col-
mental ejecta may be juvenile (formed of magma
umn dynamics or the wind; their areas of impact
involved in the eruption), accidental (derived from
are typically restricted to within 5 km of vents
pre-existing rocks), or, most commonly, a mixture
(Blong, 1984).
of both.
Tephra fall and ballistic projectiles affect
the environment by: (1) the force of impact of
tephra falls and ballistic projectiles falling fragments, (2) burial, (3) production of a
Falls of tephra – air-borne fragments of rock and suspension of fine-grained particles in air and
lava of any size or shape expelled during explo- water, and (4) venting of noxious gases, acids,
sive eruptions – constitute the commonest and salts, and, close to the vent, heat. The most-
most hazardous fall process. The temperature damaging impacts, however, are the collapse of
of the erupted material and mass eruption rate roofs of buildings, interruption of power, disrup-
determine the height of an eruption column, tion of societal infrastructures (e.g., water, waste-
which, along with wind strength and direction, treatment, power, transportation, and communi-
exert the principal controls on the long-distance cations systems), and the damage or killing of
transport and, ultimately, deposition of tephra. vegetation, including agricultural crops.
Tephra typically becomes finer-grained and forms
thinner deposits, with increasing distance down-
wind from the eruptive vent. Flowage processes
Depending on the volume and duration of Flowage processes are among the most deadly
the eruption, tephra falls can blanket many of the direct volcano hazards, even though
tens of thousands of square kilometers of areas their impacts are much more restricted to the
62 VOLCANO HAZARDS

immediate areas around the volcano than are low-angle component of explosive energy release
those of fall processes. that can fan out and affect large sectors (up
to 180◦ ) of the volcano out to distances of tens
pyroclastic flows, pyroclastic surges, of kilometers. A laterally directed blast is little
and laterally directed blasts affected by topographic features. Large directed
These phenomena all involve rapidly moving, blasts result from the sudden depressurization of
ground-hugging mixtures of rock fragments and a magmatic and (or) hydrothermal system within
gases. Pyroclastic flows and surges are masses a volcano, commonly by flank failure and associ-
of hot (300 – >800 ◦ C), dry, pyroclastic debris ated landsliding, as occurred at Mount St. Helens
and gases that sweep along the ground surface in 1980 (Christiansen and Peterson, 1981), affect-
at extremely high velocities, ranging from ten ing an area of about 600 km2 . Well-documented
to several hundred meters per second. Such pro- notable catastrophic directed blasts during histor-
cesses are initiated by discontinuous or contin- ical time took place at Bezimianny Volcano, Kam-
uous collapse of an eruption column, buoyant chatka, in 1956 (Gorshkov, 1959; Belousov, 1996).
upwelling at the vent, and gravitational or explo- Owing to their mass, high temperature and
sive collapse of a growing lava dome (Scott, 1989a) gas content, high velocity, and great mobility,
(Fig. 2.2). pyroclastic flowage processes are among the most
Pyroclastic surges are distinguished from deadly of volcano hazards, causing some or all of
pyroclastic flows by having a relatively low ratio the following consequences: asphyxiation, burial,
of solid materials to gases and, hence, lower den- incineration, and physical impact. In addition
sity, even though a continuum exists between to these direct effects, pyroclastic flows can mix
them. Because of the density difference, pyroclas- with surface water or water melted from snow
tic flows tend to be more controlled by topog- and ice to form destructive lahars and floods that
raphy, mostly restricted to valley floors, whereas can affect valleys farther downstream. Volcanic
less dense, more mobile surges can affect areas catastrophes (each causing 2000 or more fatali-
high on valley walls and even overtop ridges ties) involving pyroclastic flows and (or) surges
to enter adjacent valleys. Pyroclastic surges can include: Vesuvius, Italy, in AD 79; Mt. Pelée,
either be hot or cold. Hot pyroclastic surges are Martinique, in 1902; Mt. Lamington, Papua New
closely associated with pyroclastic flows, both Guinea, in 1951; and El Chichón, Mexico, in 1982.
generated by the same processes; cold pyroclas-
tic surges, however, are generated by hydromag- debris avalanches
matic or hydrothermal explosions (Scott, 1989a). Active and potentially active volcanoes are inher-
Convecting clouds of finer ash commonly accom- ently unstable structures, because they are built
pany pyroclastic flows and surges and form one of weak or unconsolidated rock, have steep
type of tephra-fall deposit. slopes, are highly fractured and faulted, and may
Pyroclastic flows and surges are common at be undergoing deformation related to magma
many andesitic and dacitic composite volcanoes movement and (or) hydrothermal pressurization.
and at silicic calderas. Pyroclastic flows contain- Thus, structural collapse at restless volcanoes
ing abundant dense to slightly vesicular lithic can precede, accompany, or follow eruptive activ-
fragments entrained in an ash matrix (‘‘block- ity, triggering rockfalls, rockslides, and debris
and-ash” flows) are generally of smaller volume avalanches, which can move rapidly downs-
and typically restricted to within a few tens of lope and may pose significant hazards. Most
kilometers of vents. In contrast, large pumiceous volcanic debris avalanches have followed days
pyroclastic flows, composed mostly of lapilli and to months of precursory activity (e.g., seismic-
ash, can extend up to 200 km from vents and can ity, ground deformation, phreatic explosions),
cover tens of thousands of square kilometers. but some apparently have occurred with little
Laterally directed blasts (also called lateral detectable precursors. In addition to the well-
blasts) share characteristics of pyroclastic flows documented 1980 event at Mount St. Helens
and surges but have in addition an initial (Voight et al., 1981, 1983; Glicken, 1998), debris
VOLCANO HAZARDS 63

avalanches have occurred at numerous compos- tinuum of processes ranging from dense lahars
ite volcanoes in historical time (e.g., Schuster and (with consistency of wet concrete) dominated by
Crandell, 1984; Siebert, 1984, 1996). laminar flow to turbulent watery floods. The tran-
Volcanic debris avalanches apparently are sition downstream from a lahar, to a lahar-runout
more mobile than their non-volcanic counter- flow, and then to normal stream flow (i.e., flood)
parts, and, for a given volume and vertical drop, is primarily a function of decreasing sediment
volcanic debris avalanches travel farther (Scott, concentration; see Scott (1988, 1989) for details of
1989a, Fig. 2.5). Large-volume avalanches can the transition and the controlling factors (grain
extend as far as 85 km beyond their sources and size, water content, yield strength, bulk density,
can deposit volcanic debris tens of meters thick etc.). As mentioned earlier, water-saturated debris
over areas 102 –103 km2 (Siebert, 1996). With suffi- avalanches can transform into lahars.
cient momentum, debris avalanches can run up Measured velocities of historical lahars have
slopes and cross topographic barriers up to sev- varied greatly depending on volume, grain-size
eral hundred meters high. The Mount St. Helens distribution, channel dimensions and configura-
debris avalanche (2.5 km3 volume) was the largest tion, and slope gradient, ranging from about 1
such event, volcanic or non-volcanic in origin, in to 40 m/s, with mean values on the order of
historical time (Glicken, 1998). However, a pre- 10–20 m/s (Macdonald, 1972; Janda et al., 1981;
historic debris avalanche (∼300 ka) at Mt. Shasta Blong, 1984). High-velocity lahars have sufficient
Volcano, California covered nearly 700 km2 and momentum to rise on the outside of bends of
involved a volume of 45 km3 (Crandell et al., channels and to surmount topographic barriers.
1984b; Crandell, 1989). Large-volume lahars, rich in clay and confined to
Debris avalanches are highly destructive, narrow channels, can travel hundreds of kilome-
burying and destroying everything in their paths. ters down valleys.
Moreover, the ‘‘dewatering” of a debris avalanche Because of their high bulk density and veloc-
can generate lahars and floods downvalley, as ity, lahars pose a major hazard along valleys
observed at Mount St. Helens in 1980 (Janda draining the volcano, even for communities at
et al., 1981). Lahars and floods can also be trig- great distances from the volcano. Lahars are
gered by the catastrophic failure and draining highly destructive, destroying and burying every-
of lakes formed by damming of streams by thing in their paths. They can also fill and mod-
avalanches (Costa and Schuster, 1988; Scott, 1988). ify stream channels, thus decreasing the chan-
Avalanches may also cause indirect hazards, when nels’ normal carrying capacity and increasing the
the volcanic debris enters lakes or bays and sud- potential for water floods; also, increased sedi-
denly displaces large volumes of water to produce mentation in channels by lahars can affect their
high waves, or enters the sea to generate vol- navigation.
canogenic tsunamis (Scott 1989a; Siebert, 1996). Lahars and pyroclastic flows and surges have
been the deadliest volcano hazards in recorded
primary debris flows (lahars) history (Blong, 1984; Tilling, 1989a, 1989b). Com-
and floods pared to pyroclastic flow and surge hazards,
A lahar (an Indonesian term for volcanic mud- lahars and flood hazards – in theory at least –
flow) is a slurry of rock debris and water that can be more easily mitigated, because they have
originates on the slopes of volcanoes. Such flows relatively well-defined upper limits of potential
are called primary if they occur during eruptive impact along valleys. With early detection of a
activity, and secondary if they are post-eruption. lahar close to its source, and if such detection
The source(s) of the water to mobilize lahars can is communicated effectively to the settlements
be one or more of the following: melting of ice downvalley, people at risk can quickly climb to
and snow by hot volcanic ejecta; crater lakes and safety if high-ground safe areas are identified
other surface waters; water in the groundwa- beforehand. Had such a detection and communi-
ter and geothermal systems; and torrential rains. cation system been established at Volcán Nevado
Lahars and floods are end members of a con- del Ruiz (Colombia) in 1985, many lives probably
64 VOLCANO HAZARDS

could have been saved (see p. 74). Also, ‘‘sabo” and Because the movement of lava flows is pri-
other engineering works along valleys (e.g., check marily controlled by gravity, once potential or
dams, sediment-retention and flood-control struc- actual vents of lava flows are identified, their
tures, levees) can mitigate the hazards of small- paths can be predicted based on the surround-
volume lahars but not large ones (Scott, 1989a). ing topography. With sufficient data, accurate
hazard-zonation maps for lava flows (e.g., Wright
lava flows and others, 1992) provide a basis for emergency-
Lava flows constitute the most common volcano response plans. The most cost-effective mitigation
hazards from non-explosive eruptions, especially of lava-flow hazards is prudent land-use planning,
at basaltic systems. How far a lava flow can but several attempts have been made in recent
travel from the vent depends on its viscosity decades to divert or control the paths of lava
(itself determined by temperature, chemical com- flows after the start of eruption (for examples, see
position, gas content, and crystallinity), effusion Macdonald, 1972; Scott, 1989a; Williams, 1997).
rate, and ground slope. At low effusion rates
(<10 m3 /s), basaltic lava tends to produce many Indirect volcano hazards
small flows that puddle and pile up near the vent, As the term implies, indirect volcano hazards
whereas at higher rates (101 –103 m3 /s), the flows are destructive processes that are incidental to
produced can travel tens of kilometers and can the eruptive phenomena themselves and include
cover hundreds of square kilometers. For exam- (see Table 2.1): ground-shaking and movements
ple, the 1783 Lakagı́gar (Laki) fissure eruption in caused by volcanogenic earthquakes; tsunamis
Iceland, with an effusion rate of 5000 m3 /s, gen- generated by eruption-induced collapse, debris
erated lava flows that covered more than 500 km2 avalanche or slump of a volcanic edifice; ‘‘sec-
(Thorarinsson, 1969). During episodes of flood- ondary” debris flows and floods triggered by
basalt volcanism in the geologic past, such as heavy rainfall and other post-eruption causes;
those producing the Columbia River Basalts in post-eruption erosion and sedimentation; atmo-
the United States, single flows discharged at esti- spheric effects (electrical discharges, shock
mated rates of about 1 × 106 m3 /s covered tens of waves); climate change; post-eruption famine and
thousands of square kilometers (Swanson et al., disease; and, only within recent decades, dam-
1975). age to aircraft encountering volcanic ash clouds.
Unlike the sheet-like flows formed of basaltic Only the most severe of these indirect hazards
and other mafic lavas, more viscous lavas will be reviewed briefly below; for the others,
(e.g., andesite, dacite, and rhyolite) are typically the interested reader is referred to Blong (1984),
erupted at low rates, forming short, stubby lava Scott (1989a), and the relevant references given in
flows or steep-sided domes that cover only a few Table 2.1.
square kilometers. Rates of movement of lava
flows vary considerably, from a few to hundreds Volcanogenic tsunamis
of meters per hour for silicic lava flows to sev- A tsunami (a Japanese word meaning ‘‘harbor
eral kilometers per hour for basaltic lava flows on wave”) is a long-period seawave that is gener-
steep slopes. Because of their slow rates of move- ated by the sudden displacement of water, most
ment, lava flows seldom threaten human life, but commonly by fault displacements of the seafloor
they can be highly destructive, burying, crush- by tectonic earthquakes (e.g., along subduction
ing, or burning everything in their paths. In addi- zones). Volcanogenic tsunamis, much less com-
tion, fires started by lava flows can affect areas far mon, can be produced by one or more of the fol-
beyond their borders. Lava flows also melt snow lowing mechanisms: volcanic or volcano-tectonic
and ice that can produce small debris flows and earthquakes; explosions; collapse or subsidence
floods; however, catastrophic floods can occur if of volcanic edifices; debris avalanches, lahars,
the meltwater can be stored and later released or pyroclastic flows entering water bodies; and,
suddenly in large quantities, as during Icelandic atmospheric shock waves that couple with the
jökulhlaups. sea (Latter, 1981).
VOLCANO HAZARDS 65

Tsunamis travel at high speeds (>800 km/hr) of water impounded by natural dams – formed by
as imperceptible low broad waves in the open lava flows, lahars, debris avalanches, pyroclastic
ocean but build to great heights as they approach flows, or crater rims – when such dams fail or are
shore and ‘‘touch bottom.” Because of the tremen- overtopped (e.g., Houghton et al., 1987; Costa and
dous release of wave energy upon slamming Schuster, 1988; Lockwood et al., 1988).
onto the shore, tsunamis are highly hazardous That hazards associated with secondary lahars
for life and property on low-lying shore areas are not inconsequential is well illustrated at Mt.
of lakes and oceans, even at great distances Pinatubo (Luzon, Philippines). Since Pinatubo’s
(103 km) from the tsunami source. For example, climactic eruption in 1991, monsoon rains have
the 1883 Krakatau eruption produced tsunamis caused massive redistribution of the 1991 erup-
that ran up coasts to heights of 35 m, devastat- tive and lahar deposits, generating numerous sec-
ing nearly 300 coastal villages and claiming more ondary lahars that have buried villages and valu-
than 30 000 victims (see numerous references able agricultural lands and, to date, have killed
in Simkin and Fiske, 1983). Another deadly vol- about an additional 100 people since the erup-
canogenic tsunami occurred in 1792, when a mas- tion, during which 250–300 people perished. Sec-
sive debris avalanche produced by the collapse ondary lahars, involving progressively declining
of the Mayuyama dome (Unzen Volcano, Japan) volumes, are expected to continue well into the
entered the Ariake Sea, generating a tsunami that next decade, as Pinatubo’s drainages gradually
killed about 5000 people (Tanguy et al., 1998). become re-established (Wolfe and Hoblitt, 1996).
The international Pacific Tsunami Warning
Center (Honolulu, Hawaii), created in 1965, Post-eruption famine and disease
employs a circum-Pacific network of seismic and In addition to the direct threat to human safety,
tide-gauge stations to provide timely warnings volcano hazards also can adversely affect the
of approaching tsunamis to areas hundreds to daily lives of people by destruction of dwellings
thousands of kilometers from sources. Such early and societal infrastructure but, more impor-
warnings are very useful for tsunamis generated tantly, by the disruption of food supply by the
by distant sources, allowing several hours for offi- immediate loss of livestock and crops, and by
cials and people to take safety measures. How- the longer-term (years to decades) loss of agricul-
ever, because of the high velocity of tsunamis, it tural productivity of farm lands buried by erup-
is impossible to provide similar early warnings to tive materials. Before the twentieth century, post-
people living close to a locally generated tsunami. eruption famine and epidemic disease – in the
aftermath of the 1783 eruption of Laki (Iceland)
Secondary debris flows (lahars) and the 1815 Tambora eruption – constituted the
Lahars and floods generated after or between deadliest volcano hazard (Fig. 2.3). Previous esti-
eruptions occur commonly at volcanoes. This mates of the deaths caused by post-eruption star-
comes as no surprise because most active or vation after the Tambora eruption were on the
geologically young volcanoes are constructed of order of 80 000 (e.g., Blong, 1984; Tilling, 1989a),
poorly consolidated volcanic products and have but a recent estimate is ∼49 000 (Tanguy et al.,
steep slopes. In addition, valleys draining volca- 1998). The estimate of the number of starvation
noes typically contain abundant unconsolidated deaths (∼10 000) caused by the Laki eruption is
volcanic debris from previous eruptive activity, more reliable; the deaths represented about a
and such deposits act to disrupt the established quarter of Iceland’s entire population at the time.
drainages. Given such conditions, all that is In the twentieth century, fatalities from post-
needed to produce secondary lahars is a sud- eruption famine and epidemic disease have been
den massive infusion of water, which is most fre- greatly reduced (Fig. 2.3).
quently supplied by torrential rainfall, quite com-
mon in many volcanic regions during the rainy or Aircraft encounters with volcanic ash
monsoon season. A less common mechanism for A significant and growing indirect volcano
generating secondary lahars is the sudden release hazard is the encounter between aircraft and
66 VOLCANO HAZARDS

December 1989. A new Boeing 747–400 encoun-


tered Redoubt’s volcanic ash cloud and temporar-
ily lost all four engines, which (luckily) were able
to be restarted in ash-free atmosphere after a pow-
erless descent of nearly 4000 meters, and the
plane landed safely at Anchorage airport. For-
tunately, no passengers were injured, but the
costs to replace all four engines and repair other
damage exceeded US$80 million (Casadevall,
1992).

Spatial and temporal scales of


volcano hazards
From the previous discussion, it is obvious that
Fig. 2.3. Eruption-related fatalities according to principal
the various volcano hazards differ markedly, in
cause for the periods 1600–1899 and 1900–1986. Modified
from Tilling (1996, Fig. 2.2); data sources: Blong (1984); Tilling
spatial and temporal scale, in their impact on
(1989a). the physical environment and on people. Of
the direct hazards, flowage processes (pyroclastic
flows and surges, lateral blasts, debris avalanches
volcanic ash clouds, which are not detectable and flows, and lava flows) have an immediate
by the aircraft’s onboard radar instrumentation. impact on the environment. With the exception
This problem only has emerged in recent decades, of basaltic lava flows, most other direct flowage
with the advent of high-performance jet engines. hazards typically occur over a timescale of
The hazard stems from the ingestion of silicate seconds to days. As exemplified by the current
ash into the aircraft’s jet engines when operating eruption at Kilauea Volcano (Hawaii, USA), which
in volcanic ash clouds (for details, see Casade- began in January 1983 and has continued to date
vall, 1992; Dunn and Wade, 1994). The ash par- (as of February 2000) virtually non-stop in a quasi-
ticles can abrade, and (or) accumulate in, noz- steady-state manner (Wolfe et al., 1988; Mattox
zle guide vane, ducts, compressor and turbine et al., 1993; Heliker et al., 1998), basaltic lava flows,
blades, and other engine parts; also, the high on the surface and below (in lava tubes), can
operating engine temperatures are sufficient to occur on a timescale of hours to years. In general,
partially melt the ash, which then adheres to the environmental impacts of all flowage hazards
or clogs critical engine parts and openings. Ash are restricted to areas around the volcano and its
ingestion degrades engine performance and, in principal drainages out to distances on the order
the worst case, causes engine flame-out and loss of several hundred kilometers maximum.
of power. The ash can also cause exterior damage, Tephra fall poses the widest-ranging direct
such as abrasion of the windshield and windows, hazard from volcanic eruptions. For instance,
erosion of leading edges, etc. areas of 103 to 104 km2 may be covered with
Since the early 1970s, there have been more >10 cm of tephra during some large eruptions,
than 60 volcanic ash–aircraft encounters, with and fine ash can be carried over areas of conti-
several of them experiencing total power loss, nental size or larger. For this reason, tephra lay-
necessitating emergency landings. Most encoun- ers produced by eruptions of known ages serve
ters occur within about 250 km of the erupting as important regional time markers in strati-
volcano, but many have happened at greater dis- graphic and geologic studies over extensive areas.
tances (>900 km) from the volcano. Drifting vol- For example, the Bishop tuff deposit (∼600 km3
canic ash clouds, following stratospheric winds, volume) produced by a cataclysmic eruption
can be tracked for thousands of kilometers before that formed Long Valley Caldera (eastern
dissipation. A very costly encounter was during California, USA) 760 000 years ago can be traced
the eruption of Redoubt Volcano (Alaska, USA) in as a stratigraphic marker in the mid-continent
VOLCANO HAZARDS 67

of the United States, thousands of kilometers of the atmosphere by volcanic aerosols and asso-
distant from the caldera (Izett et al., 1988). ciated changes in global climate are short-lived,
Two indirect volcano hazards also cause wide- lasting only for a few years.
ranging impact on the environment, but on dif- It also should be emphasized that more than
ferent temporal scales: volcanogenic tsunamis one hazardous process can occur during the same
and eruption-influenced global climate change. eruption, each process with its own temporal
The series of powerful tsunamis generated by the and spatial scales. For example, climactic erup-
1883 Krakatau eruption not only wreaked havoc tion of Mount St. Helens on May 18, 1980, within
locally, but the larger waves were recorded within the span of a few hours, involved flank col-
about 12 hours by tide gauges as far away as lapse, laterally directed blast, debris avalanches
the Arabian Peninsula, more than 7000 km from and flows, pyroclastic flows, and areally exten-
the eruption site. Recent studies demonstrate sive tephra fall (Fig. 2.4; see also Lipman and
that the largest-scale, but lowest-frequency, debris Mullineaux, 1981; Tilling et al., 1990). The lat-
avalanches are of submarine rather than sub- eral blast initiated within 2 s after the flank
aerial origin, from collapses of flanks of volcanic collapse and onset of the debris avalanche (esti-
islands in the Pacific, Indian, and Atlantic Oceans mated at 08:32.21.0 local time). The blast itself
(e.g., Moore et al., 1989, 1994; Holcomb and Searle, probably lasted no more than about 30 s at the
1991; Barsczus et al., 1992). Some studies have vent, but the associated expanding and radiating
suggested that these prodigious avalanches have blast cloud persisted for about another minute;
produced tsunamis orders of magnitude greater during this short time, the blast affected a large
than any observed in historical time, leaving sector (600 km2 ) north of the volcano. A few sec-
deposits of their occurrence >300 m above cur- onds following the blast, the vertically directed
rent sea level (Moore and Moore, 1984, 1988; tephra column reached an altitude of 26 km in
Young and Bryant, 1992). less than 15 min. Eruption of tephra continued
Perhaps the widest-ranging volcano hazard is for the next 9 hr, and the fallout area covered
global climate change caused by ejection of vol- more than 57 000 km2 downwind. The finer ash
canic gases (principally SO2 ) into the stratosphere and volcanic gases remained air-borne, spread-
during powerful explosive eruptions. The SO2 ing across the USA in 3 days and circling the
forms an aerosol layer of sulfuric acid droplets, world in 15 days. Several small pyroclastic flows,
which tends to cool the troposphere by reflect- though first directly observed shortly after noon,
ing solar radiation, and to warm the stratosphere probably were erupted shortly after the lateral
by absorbing radiated Earth heat. Such global blast, pyroclastic flow activity continued inter-
climate change is best illustrated by the 1815 mittently for about 5 hr. Small lahars in the
Tambora eruption. The stratospheric volcanic South Fork of the Toutle River were observed as
cloud associated with this huge eruption – the early as 08:50, and the larger and more destruc-
largest in recorded history – resulted in the well- tive lahars in the North Fork of the Toutle
documented ‘‘Year without Summer” the follow- River developed several hours later (Fig. 2.4).
ing year (1816), when usually cold weather pre-
vailed across the northern hemisphere, marked Global and historical perspective
by severe frosts in July and crop failures (Stom- Compared to other hazards, natural or man-
mel and Stommel, 1983). For example, the annual made, volcano hazards have caused far
summer temperature for 1816 for New Haven fewer fatalities, largely because they occur
(Connecticut, USA) dropped 3 ◦ C below the 145- infrequently; floods, storms, earthquakes, and
year average. The more recent eruptions of El droughts are the most frequently occurring
Chichón Volcano in 1982 and of Mt. Pinatubo natural hazards and affect far more people
in 1991 also produced measurable tempera- (Tilling, 1989b, Figs. 1.1 and 1.2). Since AD 1000,
ture decreases for the northern hemisphere, direct and indirect volcano hazards have killed
by 0.2 to 0.5 ◦ C respectively (Simarski, 1992; about 300 000 people, which is much less than
Self et al., 1996). However, such perturbations the fatalities for a single worst earthquake or
68 VOLCANO HAZARDS

122 30' 122 00'

Ri
n
Gree River
G re
en ver

or
N

th

Fo Spirit
rk
To Lake
u tl e

Rive
r
46 46
15' 15'
So
u th
Fork
To u tl e Rive
r

Crater outline
Pyroclastic flow deposits

Mud d y
Mudflow deposits
Lateral blast zone

Riv e r
P in
Debris avalanche deposit 0 5 10

e
C
re e
km

k
122 30' 122 00'
Fig. 2.4. Generalized geologic map showing the impacts tsunamis and post-eruption famine and disease;
and proximal deposits of the climactic eruption of Mount St. in contrast, the principal causes of eruption-
Helens Volcano on May 18, 1980; not shown are the related deaths in the twentieth century have been
tephra-fall deposits. The hazardous processes that produced the direct volcano hazards – pyroclastic flows
these deposits all occurred within about one hour (see text).
(including surges) and lahars (Fig. 2.3). As noted
Modified from Tilling et al. (1990, p. 22); original data source:
previously by Tilling (1989a, p. 260), the reduc-
Kieffer (1981, Fig. 219).
tion in the incidence of deaths resulting from
post-eruption starvation in the twentieth century
hurricane disaster, such as the 1556 Huah- ‘‘is real and reflects the existence of modern,
sien (Shensi, China) earthquake which killed rapid communications and disaster relief-delivery
more than 820 000 or the 1970 Ganges Delta systems,” which were lacking in centuries past.
(Bangladesh) hurricane in which about 500 000 However, the decrease in the number of volcano-
perished. From a global and historical stand- related tsunami deaths in the twentieth century
point, the adverse consequences of volcano (Fig. 2.3), while real, cannot simply be cred-
hazards on people are comparatively quite small, ited to any technological or societal advance.
but volcanic disasters still can have signifi- The observed decrease could merely reflect the
cant short-term human and economic impact fortunate circumstance that no significant vol-
(Chester, 1993; Chapter 14, this volume). canogenic tsunami has occurred since 1883.
During the seventeenth to nineteenth cen- With the international Pacific Tsunami Warn-
turies, most eruption-related casualties were ing Center (Honolulu, Hawaii) in operation since
caused by indirect volcano hazards, most notably the mid-1960s, it is now possible to provide
VOLCANO HAZARDS 69

early warnings for future, remote volcanogenic to provide timely warnings of eruptions from vol-
tsunamis that are expected to arrive onshore an canoes along air routes and to predict and track
hour or more after the triggering event (Lander the possible paths of the volcanic ash clouds once
and Lockridge, 1989). However, such expectation formed (Casadevall, 1994b).
can be realized only if the volcanogenic source of The average number of eruption-caused
the tsunami is centered at least 1000 km distant deaths per year for the twentieth century (880)
from the threatened coastal areas. Given the high is greater than that for the seventeenth to nine-
velocity at which tsunamis travel in the open teenth centuries (620) (Fig. 2.3). Inasmuch as erup-
ocean (∼800 km/hr), it would be nearly impos- tion frequency and severity have not increased
sible for timely warnings to be given to people in recent centuries (Simkin and Siebert, 1994),
at risk from locally generated tsunamis, with the higher toll of casualties in the twentieth
triggering sources within 1000 km or less. century could be interpreted as reflecting world
Most of the countries in the developing population growth. Such interpretation, however,
regions facing severe risk from volcano hazards is inconsistent with the approximately ten-fold
lack the economic and technical resources to increase in world population during the sev-
adequately study their dangerous volcanoes and enteenth to twentieth centuries; thus the fre-
to develop effective volcanic-emergency manage- quency of eruption-caused deaths per capita
ment systems (UNDRO/UNESCO, 1985). Not sur- worldwide actually decreased in the twentieth
prisingly, more than 99% of the eruption-related century. Lacking an analysis of relevant popula-
deaths in the twentieth century have occurred tion and eruption fatality data specific to the vol-
in the developing regions, reflecting one or more canic regions, a plausible inference would be that
of the following factors: (1) high population den- progress in volcano-hazards mitigation has more
sity and growth, (2) insufficient scientific and eco- than kept pace with population growth.
nomic resources, and (3) the fact that no major Perhaps the data for eruption-caused fatali-
eruptions took place in any densely populated ties in the twentieth century (Fig. 2.5) may sup-
volcanic area in an industrialized region (e.g., port this inference, if the 1902 Mt. Pelée and
Japan, Italy) during the twentieth century. 1985 Nevado del Ruiz disasters are not taken into
The last two decades of the twentieth cen- account. The rationale for such exclusion is that,
tury saw the emergence of a new indirect vol- for the 1902 disaster, the inhabitants of city of
cano hazard: volcanic ash in eruption clouds, St. Pierre were not ordered to evacuate and all but
which severely affects jet aircraft flying into two perished during the eruption. Even though
them (Casadevall, 1994b). To date, while no political motivation was a factor (Heilprin, 1903),
fatal accidents have happened from in-flight the primary reason for non-evacuation of the
encounters between aircraft and volcanic ash, city was that its inhabitants and officials had
many hundreds of millions of dollars have been no scientific understanding of eruptive phenom-
incurred from damage to aircraft (including total ena and were totally ignorant of the developing
power loss), unscheduled emergency landings, unrest of Mt. Pelée and its ominous portent of
and increased flight time and fuel consumption the calamity to come. Awareness of the destruc-
in rerouting the planes from known or suspected tive power of pyroclastic flows and surges and
volcanic ash plumes. With continued rapidly the emergence of volcanology as a modern sci-
escalating air travel across the North Pacific and ence came after the obliteration of the city. As
through other air routes over active volcanic belts discussed in detail later (p. 74), in the case of the
(Casadevall et al., 1999), the risk posed by air- Ruiz catastrophe, the high death tolls resulted
borne ash will increase proportionately and could from a failure in communication and human
well become the prime indirect volcano hazard error, despite the availability of monitoring data
in this century. Volcanologists are now working and a hazards assessment; government officials
closely with aircraft controllers, the International simply failed to order evacuations even though
Civil Aviation Organization, the aviation indus- the volcano and the scientists had given sufficient
try, and pilots to develop more effective systems warning.
70 VOLCANO HAZARDS

40,000

Nevado del Ruiz, November 13, 1985


Number of fatalities

30,000

Mont Pelée, May 8,1902

Mt. Pinatubo, June 15, 1991


20,000

10,000

0
1900- 1910- 1920- 1930- 1940- 1950- 1960- 1970- 1980- 1990-
09 19 29 39 49 59 69 79 89 93

Time period

Fig. 2.5. Eruption-related fatalities by decade for the


twentieth century through 1993. The shaded bars show the
Coping with volcano hazards
fatalities for the two deadliest eruptions of the century: the
1902 Mt. Pelée eruption (deaths from pyroclastic flows), and Given the projected growth in world population
the 1985 Nevado del Ruiz eruption (deaths from mudflows). and the observation that on average 50–60 vol-
Dashed arrow hints a possible decline in fatalities with time canoes are active each year (Simkin and Siebert,
if the fatalities from these two events are not considered 1994), the conclusion seems inescapable that the
(see text). Modified from Tilling and Lipman (1993, number of people and the quantity of economic
Figure 1). infrastructure at risk from volcano hazards will
increase (Fig. 2.6). With this daunting demo-
graphic scenario, which is unlikely to change
much during the next several decades, what can
A comparison of the 1902 Mt. Pelée erup-
be done to reduce risk from volcanoes? For active
tion with the 1991 Mt. Pinatubo eruption also
or potentially active volcanoes, the three most
bears on the notion that a possible decreasing
important elements of an effective approach to
trend of eruption-related deaths for the twenti-
volcano-hazards mitigation are:
eth century (Fig. 2.5) reflects advances in scien-
tific understanding of eruptive phenomena and (1) Acquisition of a good understanding of a vol-
in volcano-hazards mitigation. ‘‘If Pinatubo (with cano’s past behavior by deciphering its eruptive
its current population) had erupted in 1902, the history from basic geologic, geochronologic,
deaths would have been at least comparable in and other geoscience studies. Such an under-
number to those in St. Pierre. In these two cases – standing is prerequisite for the preparation of
in both of which devastating pyroclastic flows a hazards assessment and associated hazards-
impacted some densely populated areas – the zonation map, as well as for long-term fore-
decrease in casualties truly reflects a century of casts of potential future activity.
advance in hazards assessment, monitoring, and (2) Initiation or expansion of volcano monitor-
communication.” (Edward W. Wolfe, 1998, per- ing to determine a volcano’s current behavior,
sonal communication.) to characterize its ‘‘normal” baseline level of
COPING WITH VOLCANO HAZARDS 71

6 Tephra volume (km3)

World population (billions)


Active volcanoes per year
100

IC
0.001 0.01 0.1 1 10 100 1,000

AFF
5

R
100 000

RT

Krakatau 1883, Mt. Pinatubo 1991


Ruíz 1985
4

Galunggung 1982
AI

Mount St. Helens 1980


NG
60

SI
3

A
50 RE 10 000
I NC

ON
I 2
AT

Eruptions per thousand years

Yellowstone Caldera (2 million years ago)


UL
POP 1000
NG
EASI 1
IN CR
0

Tambora 1815
1600 1700 1800 1900 2000 100
YEAR
Fig. 2.6. Schematic diagram showing that, given a fixed 10
global eruption frequency of 50–60 volcanoes active each
year on average (shaded bar with erupting volcano symbols)
and an increasing world population (heavy dashed curve), 1
more and more people inevitably will be at risk from
volcanoes into the foreseeable future.
0.1

activity, and to detect, reliably and quickly, 0.01


2 3 4 5 6 7 8
any significant departures from it. Volcano-
monitoring data provide the scientific basis Volcanic explosivity index (VEI)
for short-term forecasts of possible future Fig. 2.7. Frequency and eruption size expressed in terms of
behavior. the volcanic explosivity index (VEI), with some representative
(3) Establishment or maintenance of credible, examples; extrapolation to the huge Yellowstone
effective communications with civil authori- caldera-forming eruption 2 million years ago is unconstrained
ties, the media, and the general public, while and only shown for relative comparison. The VEI is a
semiquantitative estimate of the “magnitude” of explosive
striving to accomplish the other two ele-
eruptions in historical time; a non-explosive eruption is
ments. Equally important is the creation of
assigned a VEI = 0 (see Newhall and Self, 1982). As observed
emergency-response or ‘‘preparedness” plans; also for earthquakes, small events (VEI ≤ 2) are much more
effective communications and mutual trust frequent than larger events (VEI ≥ 5). Eruptions of VEI 6 (e.g.,
will promote the development and testing of Krakatau 1883 and Mt. Pinatubo 1991) are very infrequent,
such plans. Non-achievement of element (3) renders occurring on average about once a century, and the only
elements (1) and (2) useless. VEI 7 is given to the world’s largest historical eruption
(Tambora 1815). Modified from Simkin and Siebert (1994,
Fig. 10).
Hazards assessment, hazards-zonation
maps, and long-term forecasts
An assessment of a volcano’s natural hazards is eruptions (volcanic explosivity index (VEI) ≥ 2) or
only as good as the abundance and quality of rare huge events (VEI ≥ 7) (Fig. 2.7)? Which sectors
geologic, geochronologic, petrologic, geochemi- of the volcano have been repeatedly affected by
cal and other pertinent data collected to recon- hazardous processes? Definitive answers to these
struct its eruptive history in the most comprehen- and related questions provide the key constraints
sive manner. The key questions to be answered on the types, scales, duration, recurrence, and
are straightforward but crucial: what is the typi- other characteristics of hazardous processes that
cal or characteristic style and frequency of its past have dominated the volcano’s eruptive history.
activity? Explosive or non-explosive eruptions? An in-depth understanding of eruptive history
Eruption sizes always similar? Frequent small and characteristic behavior in turn provides the
72 VOLCANO HAZARDS

best basis for long-term forecasts of future eruptive areas, escape routes, resettlement shelters, etc.,
activity and associated hazards. should evacuation or other public-safety mea-
A long-term forecast is generally taken to mean sures be required. Also, citizens should have suf-
a forecast made one year or longer in advance of ficient information about the hazards to make
the anticipated activity. An excellent example of knowledgeable decisions about where to live and
a successful long-term forecast was that made by work within the context of level of acceptable
US Geological Survey scientists in February 1975 risk.
(Crandell et al., 1975) that Mount St. Helens in the
Cascade Range was the one volcano in the con- Volcano monitoring and short-term
terminous United States most likely to reawaken forecasts
and to erupt ‘‘perhaps before the end of this Volcano monitoring refers to the systematic
century” (i.e., the twentieth century). Five years visual observation and instrumental measure-
later (March 1980), Mount St. Helens erupted. ment of structural, geochemical, and seismic and
For more information on the related broad top- other geophysical changes in the state of a rest-
ics of hazards assessment, hazard-zonation maps, less volcanic system. Although the regular moni-
and long-term forecasts, the interested reader is toring of active volcanoes – during and between
referred to the review papers of Crandell et al. eruptions – was the raison d’être for establish-
(1984a), Scott (1989b), and Tilling (1989a). Also, ing modern volcano observatories early in the
the volumes of Latter (1988) and Scarpa and Till- twentieth century, it also was the catalyst that
ing (1996) each contain several papers that treat quickened the transformation of volcanology into
more specialized aspects (including probabilistic the multidisciplinary science that it is today.
analysis) of hazards assessment. Only volcano-monitoring data can provide a diag-
For volcanically active areas that still are nosis of a volcano’s current behavior, which in
sparsely populated and relatively undeveloped, we turn constitutes the only scientific basis for short-
must accelerate basic geoscience studies of volca- term (hours to months in advance) forecasts or
noes to make the best possible hazards-zonation ‘‘predictions” of impending eruption, or of mid-
maps to guide long-term land-use planning by course changes of an ongoing eruption.
government authorities, developers, and the insu- Since the 1972 UNESCO review of volcano-
rance industry. High-density development in high- surveillance techniques (UNESCO, 1972), there
risk zones should be avoided or minimized have been many hundreds (thousands?) of papers
to the extent practical economically. In theory on the general theme of volcano monitoring
at least, scientifically based land-use planning and its applications at diverse volcanic systems
would allow land managers to follow George around the world. This remarkable surge in
Walker’s (1982) ideal dictum to mitigate volcano volcano-monitoring studies clearly reflects the
hazards: simply do not live or work near them. greatly enhanced general awareness of eruptions
However, for the much more common situation and their impacts that began with the May 18,
where the volcanic area is already densely pop- 1980 eruption of Mount St. Helens, which caused
ulated and developed and (or) where land-use the worst volcanic disaster in the history of the
patterns have long been fixed by tradition and United States. The increased awareness whetted
culture, the option of prudent land-use planning the public’s appetite for information on volca-
to mitigate potential hazards cannot be exer- noes and eruptions, encouraged (justified?) pub-
cised. Nonetheless, even in these situations, accu- lic officials to increase funding for volcanologic
rately prepared hazards-zonation maps provide investigations, and spurred scientists to pursue
essential information needed by civil authorities an improved understanding of ‘‘how volcanoes
to develop contingency plans for an acute vol- work.”
canic crisis that might culminate in eruption. It is beyond the scope of this chapter to
Specifically, the hazards-zonation maps should treat volcano monitoring in detail; the interested
help the officials to determine – ideally, before reader is referred to the following works and
a crisis strikes – the least-risk zones for staging the references contained therein: Civetta et al.
COPING WITH VOLCANO HAZARDS 73

(1974), Fournier d’Albe (1979), Lipman et al. (1981), Both the RSAM and SSAM systems – together
Swanson et al. (1983, 1985), Tazieff and Sabroux with associated software for collecting, process-
(1983), Hill (1984), Latter (1988), Banks et al. ing, and displaying in near-time their data along
(1989), Ewert and Swanson (1992), Swanson with non-seismic monitoring data – form the
(1992), Andres and Rose (1995), McGuire (1995a, core of an integrated ‘‘mobile volcano obser-
1995b), McGuire et al. (1995), Murray et al. vatory” successfully used at Mount St. Helens,
(1995, 1996), Tedesco (1995), Tilling (1995), Chouet Redoubt Volcano, and Mt. Pinatubo (Murray et al.,
(1996a, 1996b), Francis et al. (1996b), Giggenbach 1996).
(1996), McNutt (1996), Rymer (1996), Scarpa and The increasing deployment of networks of
Gasparini (1996), Scarpa and Tilling (1996), Dvo- broadband seismometers (e.g., Dawson et al.,
rak and Dzurisin (1997), and Sigurdsson et al. 1998) at some volcanoes is now enabling the
(2000). Instead, the discussion below is selective, complete recording and sophisticated analysis
focusing on some notable recent advances and of LP and VLP (very long-period, >10 s) seismic
developments in volcano monitoring. signals. Such research is expected to quantify
Seismic monitoring and geodetic monitoring our fundamental knowledge of magma transport,
remain the primary means of volcano surveil- source mechanism of volcanic tremor, magma or
lance (Scarpa and Gasparini, 1996). Singly or com- fluid pathways, and other attributes of magmatic
bined, ‘‘these two techniques measure the direct systems. A refined understanding of the source
responses of the volcanic system – brittle frac- mechanisms producing long-period volcanic seis-
ture, inflation, and deflation – in adjusting to micity, which almost invariably precedes eruptive
subsurface magma movement and accompany- activity, is critical to improved short-term fore-
ing stresses and (or) hydrothermal-pressurization casts (Chouet et al., 1994; Chouet, 1996a, 1996b;
effects” (Tilling, 1995, p. 371). Thus, they reliably McNutt, 1996).
detect the earliest signs of departure from a It has long been recognized that optimum
volcano’s ‘‘normal” (i.e., baseline) behavior. In volcano monitoring is achieved by a combina-
recent decades, with improvements in instrumen- tion of techniques rather than sole reliance on
tation, electronics, digital-telemetry technology, seismic monitoring, or some other single tech-
and computerized data collection and process- nique. In this regard, in recent decades geode-
ing, modern seismic networks can monitor the tic monitoring methods and networks have also
seismicity of a volcanic system in ‘‘real time” or improved significantly (Murray et al., 1995), with
‘‘near real time” in different ways. Thus, an over- increasing use of satellite-based technology such
all improvement in seismic monitoring networks as the global positioning system (GPS) and syn-
has been achieved at many, perhaps most, per- thetic aperture radar interferometry (InSAR) (Mas-
manent volcano observatories. In addition, the sonnet et al., 1995; Nunnari and Puglisi, 1995,
recent development of robust, inexpensive, and Sigmundsson et al., 1995; Francis et al., 1996b;
easily portable personal computer (PC)-based sys- Lanari et al., 1998; Wicks et al., 1998). While GPS
tems for various ‘‘real-time” seismic-data analy- and InSAR monitoring of ground deformation
sis software, including ‘‘real-time seismic ampli- has several major advantages over most ground-
tude measurement” (RSAM), real-time ‘‘seismic based systems (e.g., immense ground coverage,
spectral amplitude measurement” (SSAM), and an areal vs. point perspective, ‘‘all weather”),
automatic earthquake locations (Lee, 1989), per- at present these techniques are still not cost-
mits rapid deployment at restless volcanoes not effective, require much computer data manip-
monitored by any permanent observatories. The ulation, and generally cannot acquire and pro-
RSAM system (Endo and Murray, 1991) enables the cess data continuously or quickly enough to be
continuous monitoring of total seismic energy useful during a rapidly evolving volcanic crisis.
release even when conventional analogue sys- However, relatively low-cost, prototype continu-
tems are saturated, and the SSAM (Stephens ous GPS monitoring methods are being devel-
et al., 1994) aids in the rapid identification of oped and tested in the United States, Italy, and
precursory long-period (LP) volcanic seismicity. elsewhere.
74 VOLCANO HAZARDS

Monitoring of volcanic gases has also to alert officials of possible activity – consti-
advanced significantly, especially with improve- tutes the only practical strategy to reduce volcano
ments in the remote techniques for measurement risk. In such cases, the primary focus should be
of sulfur dioxide (SO2 ) and carbon dioxide (CO2 ) – to save human lives, by the timely issuance of
the two most abundant gases after water vapor – monitoring-based warnings.
that can be applied in the field rather than
in the laboratory. The correlation spectrome- Effective communication and
ter (COSPEC) is being increasingly used, in a emergency-preparedness plans
ground-based or air-borne mode, at volcanoes The need for effective communications for the
for measurement of SO2 emission (Andres and successful mitigation of volcano hazards has
Rose, 1995). Another recently developed remote been emphasized in many previous studies (e.g.,
technique showing promise involves Fourier Fiske, 1984, 1988; Peterson, 1986, 1988, 1996;
transform infrared (FTIR) spectroscopy for the Tilling, 1989a, 1989b; Voight, 1990, 1996; Peter-
measurement of SO2 , SO2 /HCl ratio, and SiF4 in son and Tilling, 1993; Tilling, 1995). The critical
volcanic plumes and has been successfully used importance of effective communications, or lack
at Kilauea, Etna, Vulcano, Asama, and Unzen thereof, will become apparent in the two case
volcanoes (see Francis et al., 1996a; McGee and histories discussed below, but the essence is:
Gerlach, 1998; and references therein). A highly
Monitoring data or eruption forecasts, no matter how
useful satellite-based method for the measure-
timely or precise, have little practical consequence
ment of SO2 release of recent eruptions utilizes
unless they can be communicated effectively to, and
the total ozone mapping spectrometer (TOMS), acted upon in a timely manner by, emergency-
an instrument originally designed for ozone management authorities.
measurement but subsequently discovered to (Tilling, 1995, p. 396a)
be able also to measure SO2 in volcanic plumes
that reach the upper atmosphere (Krueger et al.,
1990; Bluth et al., 1993). Particularly important
have been the advances in the measurement Two instructive recent case histories
of, and acquisition of time-series data on, SO2
and CO2 flux (i.e., volcanic degassing) at many The case histories selected for review here were
volcanoes in repose or activity (e.g., Mt. Etna, chosen because they illustrate two very different
Strómboli, Vulcano, Popocatépetl, Mount St. outcomes of recent scientific and public response
Helens, Kilauea, Long Valley Caldera, Grimsvötn), to a volcanic crisis: the disastrous outcome at
using the COSPEC, TOMS, and infrared CO2 Nevado del Ruiz represents a tragic failure in
analyzers (e.g., LI-COR) (Gerlach et al., 1997). Such hazards mitigation, whereas the outcome at
studies furnish additional insights into and con- Mt. Pinatubo in 1991 is widely hailed as a genuine
straints on intrusion of magma to shallow levels volcanologic success story, even if luck played a
and other dynamic processes within volcanic key role in it (Punongbayan and Newhall, 1995;
systems. Measurement of helium emission from Newhall and Punongbayan, 1996a). The discus-
fumaroles also provide a means of determining sion below draws liberally, in places verbatim,
the magmatic component in volcanic gas (e.g., from a review by Tilling (1995), amended and
Sorey et al., 1998). updated as appropriate in light of the recent com-
As mentioned previously, socio-economic and prehensive studies of Voight (1996) for Ruiz, and
demographic realities preclude the utilization of of Newhall and Punongbayan (1996a, 1996b) for
prudent land-use planning and zoning to miti- Pinatubo and works cited therein.
gate volcano hazards as a viable strategy for most
active volcanic areas of the world. Thus, the reg- Nevado del Ruiz, Colombia, 1985: A tragic
ular monitoring of high-risk volcanoes in already failure in hazards mitigation
densely populated regions – to enable early detec- At 15:06 h of November 13, 1985, a brief phreatic
tion of volcanic unrest and, if the data warrant, burst, lasting about a quarter hour, occurred
TWO INSTRUCTIVE RECENT CASE HISTORIES 75

at the summit of Volcán Nevado del Ruiz, a Armero to be located in a high hazard zone for
5389-m-high, glacier-capped mountain that is mudflows (lahars). A version of this map contain-
the northernmost active volcano in the Andes ing notable errors was published on October 9 in
(Williams, 1990a, 1990b). Volcanic tremor then the Bogotá newspaper El Espectador (Voight, 1996,
persisted through 21:08 h, when a small-volume Fig. 5a, c).
(∼0.02 km3 ) magmatic eruption began and sev- Analysis of fumarolic gases in late October
eral pyroclastic flows were emplaced in rapid suc- indicated substantial magmatic contributions of
cession onto the glacier surface. The hot ejecta CO2 and SO2 and the need ‘‘to seriously con-
melted and mixed with snow and ice to form sider the possibility of an impending magmatic
several mudflows that swept down the steep, nar- eruption.” (Barberi et al., 1990, p. 5). Although
row valleys draining the volcano. These mud- expanded in the month prior to the November
flows destroyed or buried everything in their 13, 1985 eruption and catastrophe, monitoring
paths, killing about 23 000 people, the vast studies still were ‘‘too little–too late” to allow any
majority in Armero, an agricultural community precise forecast of impending eruption. As Hall
in the valley of Rı́o Lagunillas. The mudflows (1990, p. 114) concluded: ‘‘Limited scientific data
also injured another 5000 people, left about from a marginal monitoring program, no baseline
7700 homeless, and caused over US$7.7 billion data (italics added), and greatly delayed process-
in economic loss, representing about 20% of ing precluded a realistic attempt to understand
Colombia’s GNP in 1985 (Voight, 1996) (Table 2.1). or predict an eruptive event.”
The 1985 Ruiz catastrophe was the worst vol- In hindsight, however, it is arguable that,
canic disaster in the recorded history of Colombia even if it had been possible to make a pre-
and the most deadly in the world since the 1902 cise short-term forecast based on more extensive
eruption of Mt. Pelée on the Caribbean island of monitoring, the Ruiz disaster could have been
Martinique. averted. Well-documented post-mortems of the
The 1985 eruption and associated mudflows at Ruiz tragedy (e.g., CERESIS, 1990; Hall, 1990, 1992;
Ruiz should not have come as a surprise, because Voight, 1990, 1996) have clearly demonstrated
similarly destructive events have occurred in the that, whatever might have been the inadequa-
past (AD 1595, AD 1845) and inundated the same cies of the scientific information available prior
area upon which Armero later developed (Voight, to the November 13 eruption, the root cause of
1990, 1996, Fig. 13). In addition, efforts were made the disaster was the lack of effective and timely
to conduct monitoring studies in response to warnings, which in turn stemmed from the lack
the precursory signals at Ruiz (felt earthquakes, of effective communications between the scien-
increased fumarolic activity, phreatic explosions, tists, the responsible authorities, the media, and
bursts of volcanic tremor) that were noticed in the populations affected. As first emphasized by
late November 1984. Due to administrative iner- Voight (1990) and later reiterated (Voight, 1996,
tia and logistical problems, it was not until the p. 764):
summer of 1985 that rudimentary seismic mon-
itoring began. However, when a strong phreatic The Armero catastrophe was not produced by
eruption on September 11 produced measur- technological ineffectiveness, nor by an overwhelming
eruption, nor by an improbable run of bad luck. The
able ash fall at Manizales, the capital of Cal-
disaster happened because of cumulative human
das Province (population 230 000), and several
error – by misjudgment, indecision and bureaucratic
sizeable debris flows, governmental concern was shortsightedness . . . The national and provincial
heightened and monitoring increased. By mid authorities were unwilling to bear the economic or
October, a monitoring network of five smoke- political costs of early evacuation or a false alarm,
drum seismometers and four precise levelling and they delayed action to the last possible minute.
(‘‘dry tilt”) arrays was established; two electronic
tiltmeters operated briefly in early November Tragically, the authorities’ delaying ‘‘action to
(Banks et al., 1990). A preliminary volcanic haz- the last possible minute” was too late for 23 000
ard map was issued on October 7, clearly showing people.
76 VOLCANO HAZARDS

Mount Pinatubo, Philippines, 1991: a based on qualitative criteria, was not intended
volcanologic success story to ‘‘make predictions,” but rather only to reflect
Mount Pinatubo is a dacitic volcanic complex ‘‘increasing levels of unrest” and ‘‘what might
located about 100 km northwest of Manila, on occur,” to ‘‘provide a simple set of steps” for
the island of Luzon in the Philippines. Recon- emergency-management officials to ‘‘design cor-
naissance geologic and radiocarbon dating stud- responding response plans” (Punongbayan et al.,
ies in the early 1980s, in part to assess geother- 1996b, p. 73). On May 23, a hazard-zonation map,
mal potential, indicated that its summit dome prepared from rapid geologic reconnaissance by
complex was flanked by extensive fans of young PHIVOLCS–USGS scientists, was issued to provin-
(600–8000 BP (years before present)) pyroclastic cial authorities and to the officials of the US Air
and lahar deposits (Pinatubo Volcano Observatory Force at Clark Air Base, located on the eastern
Team, 1991). Because of three radiocarbon ages foot of Mt. Pinatubo. During the course of the
available in the 1970s (635 ± 80 BP dating the eruption, the map was used to determine hazard
most recent explosive eruption), Pinatubo was zones, expressed as increasing radii from the vol-
reclassified in 1987 from ‘‘inactive” to ‘‘active,” cano, recommended by PHIVOLCS for evacuation
even though it had not erupted in the recorded (Punongbayan et al., 1996b).
history of the Philippines (Punongbayan, 1987; SO2 flux increased ten-fold during the latter
PHIVOLCS, 1988). part of May, from 500 tons/day to 5000 tons/day
On April 2, 1991, phreatic explosions began and then decreased sharply to only 280 tons/day
about 1.5 km northwest of the summit, accom- by June 4 (Pinatubo Volcano Observatory Team,
panied by earthquake activity and increased 1991; Daag et al., 1996). This sudden decrease
steaming at the geothermal vents. A few days in SO2 emission, together with accompanying
later, scientists from the Philippine Institute of increased shallow seismicity, occurrence of vol-
Volcanology and Seismology (PHIVOLCS) began canic tremor, and recording of deep long-period
to monitor the continuing high seismicity events beginning in late May (Harlow et al., 1996;
with portable seismographs. Continuing activity White, 1996), was interpreted to indicate upward
prompted on April 7 the evacuation of more movement of magma and, ultimately, blockage
than 5000 people living within a 10-km radius of escaping gas within the conduit. This devel-
of the volcano. A team of scientists from the US opment led PHIVOLCS to declare on June 5 an
Geological Survey’s (USGS) Volcano Disaster Assis- Alert Level 3, which states: ‘‘If trend of increas-
tance Program (VDAP) arrived in late April in ing unrest continues, eruption possible within 2
response to a request from the Philippine gov- weeks.” (Table 2.2). In essence, an Alert Level 3
ernment for assistance; the VDAP is sponsored constituted a short-term forecast. Intense shallow
jointly by the USGS and the Office of Foreign Dis- seismicity and a sharp inflationary tilt during the
aster Assistance of the US Department of State. following two days culminated in the extrusion
With increasing concern about the possibility of a lava dome on June 7, marking the beginning
of a large magmatic eruption, monitoring was of magmatic activity. The emergence and growth
intensified and included the use of a correlation of the dome, combined with a shift of seismic
spectrometer (COSPEC) for measuring SO2 emis- foci to directly beneath the dome and an irregu-
sion and two electronic tiltmeters for measur- lar increase in the amplitude and frequency of
ing ground deformation. A radio-telemetered net- occurrence of volcanic tremor, collectively pro-
work of seven seismic stations was also installed vided the basis for raising the alert level to 4
by a PHIVOLCS–USGS team, making possible the on June 7, and then to 5 on June 9 (Table 2.2).
near real-time monitoring of earthquake loca- The increasing activity and attendant elevations
tion, total seismic energy, RSAM, and SSAM. of alert level led to a 20-km radius of evacu-
At the same time, PHIVOLCS implemented ation, involving an additional 20 000 evacuees
a five-level alert scheme (Table 2.2) that could from Zambales, Tarlac, and Pampanga Provinces;
be easily communicated to, and understood by, all aircraft and supporting facilities at Clark Air
emergency-management officials. This scheme, Base were moved to safer locations elsewhere. On
TWO INSTRUCTIVE RECENT CASE HISTORIES 77

Table 2.2 The five-level alert scheme for the 1991 eruption of Mt. Pinatubo, Luzon, Philippines, implemented
May 13, 1991

Alert
levela Criteria Interpretation Date declared

No alert Background, quiet No eruption in foreseeable n/a


future
1 Low-level seismic, fumarolic, Magmatic, tectonic, or n/a
other unrest hydrothermal disturbance;
no eruption imminent
2 Moderate level of seismic, other Probable magmatic intrusion; May 13, 1991
unrest, with positive evidence could eventually lead to an
for involvement of magma eruption
3 Relatively high and increasing If trend of increasing unrest June 5, 1991
unrest including numerous continues, eruption possible
b-type earthquakes, within 2 weeks
accelerating ground
deformation, increased vigor
of fumaroles, gas emission
4 Intense unrest, including Eruption possible within June 7, 1991 17:00 h
harmonic tremor and/or 24 hours
many “long-period” (=
low-frequency) earthquakes
5 Eruption in progress Eruption in progress June 9, 1991 17:15 h
a
Stand-down procedures: in order to protect against “lull before the storm” phenomena, alert levels
will be maintained for the following periods after activity decreases to the next lower level. From level 4
to level 3, wait 1 week; from level 3 to level 2, wait 72 hours.
Source: Pinatubo Volcano Observatory Team (1991, Table 1).

June 10, more than 14 500 military personnel and the stratosphere (Gerlach et al., 1996). A new 2.5-
dependents were evacuated. km-diameter caldera was formed, centered about
At 08:51 h on the morning of June 12, the 1 km north of the former summit, which was
first of several vigorous Plinian eruptions began, destroyed; the new summit is about 200 m lower.
producing a 19-km-high ash column. This escala- Pyroclastic flows and tephra deposits blanketed
tion in activity prompted PHIVOLCS to declare a an area of more than 100 km2 around the vol-
40-km radius ‘‘danger zone” for the next three cano and greatly altered local drainage patterns;
days (Punongbayan and Newhall, 1995). A series the loose debris making up these new deposits
of strong explosions ensued the next several days, provided ample additional solid materials for
culminating in the climactic eruption that began subsequent lahars. Heavy rains associated with
at 13:42 h (local time) on 15 June and lasted about Typhoon Yunya, which passed only 50 km north
9 hours (Wolfe and Hoblitt, 1996). The climac- of Pinatubo on the same day as the climactic
tic event was one of largest volcanic eruptions eruption, triggered destructive debris flows and
in the twentieth century (Dartevelle et al., 2002). also wetted the ash that accumulated on roofs.
The combined dense-rock-equivalent volume of The thick accumulation of wet ash, together with
pyroclastic-flow and tephra deposits was about intense seismicity accompanying the climactic
3.7–5.3 km3 (Scott et al., 1996; Wolfe and Hoblitt, eruption, resulted in extensive roof collapse, the
1996), and about 17 Mt of SO2 were released into principal cause of eruption-related deaths (∼250).
78 VOLCANO HAZARDS

After the climactic eruption, seismicity and critical shifts in activity, thereby partly com-
tephra emission gradually diminished over the pensating for the lack of any previous baseline
next several months, and the alert level was monitoring data, and made possible reliable
lowered from 5 to 2 by December 1991. A 10- forecasts during the course of the eruption.
km radius danger zone was maintained, how- (2) Preparation of a hazards map – based on
ever. Following the emplacement of a lava reconnaissance fieldwork in a short time –
dome within the new summit caldera during accurately identified the zones around the vol-
July–October 1992, eruptive and seismic activ- cano most vulnerable to pyroclastic flow and
ity at Pinatubo continued to declined grad- lahar hazards. Geologic studies of prehistoric
ually through mid 1994 (Punongbayan et al., volcanic deposits also indicated that Pinatubo
1996b; Wolfe and Hoblitt, 1996). However, during had produced numerous large-volume mag-
the 1991 and 1992 rainy seasons (typically June– matic eruptions in its history, providing
November), the long-term, post-eruption hazards diagnostic evidence in assessing volcano
involving secondary lahars became increasingly hazards.
apparent. Numerous secondary lahars, triggered (3) PHIVOLCS’ very effective use of a simple
by rainfall as well as by breakouts of temporary alert-level scheme (Table 2.2), together with
lakes impounded by natural dams of volcanic near-real-time analysis and interpretation of
debris, resulted in widespread sedimentation in geologic and monitoring data, was critical in
towns and covered valuable farmlands. Lahar- prompting civil authorities to order timely
related erosion of still-hot volcanic debris caused evacuations of areas at risk and in serving
numerous rootless, secondary explosions and as an understandable framework for commu-
large secondary pyroclastic flows that traveled nicating volcano information to the officials,
several kilometers from source. From June 1991 media, and the general public.
through November 1992, over 8000 dwellings (4) The doggedness and aggressive campaign-
were totally destroyed and about 73 000 were ing of the PHIVOLCS–USGS scientists to edu-
partially damaged, affecting 329 000 families cate the local authorities and the people at
(∼2.1 million people), or about one-third of the risk, and to persist in the face of indiffer-
region’s population; the overriding cause of such ence, denial, skepticism, and (or) rebuffs (see
massive damage and human suffering was the Punongbayan and Newhall (1995) and Newhall
destructive lahar activity during the 1991 and and Punongbayan (1996a) for discussion). A
subsequent rainy seasons (Mercado et al., 1996; major tool in this aggressive campaign was
Newhall and Punongbayan, 1996b). the use, during the weeks building up to the
In retrospect, considering the huge size of the climactic eruption, of a still-unfinished ver-
June 15, 1991 Pinatubo eruption and the large sion of Understanding Volcanic Hazards, a 30-
number of people affected, the death toll directly minute videotape (IAVCEI, 1995) that vividly
related to the climactic eruption was compara- illustrates the devastating impacts of volcanic
tively small, only between 250 and 300 (Punong- hazards. It was produced for the International
bayan and Newhall, 1995). The effectiveness of the Association of Volcanology and Chemistry of
scientific and emergency-management response the Earth’s Interior (IAVCEI) by the late volca-
to the reawakening of Pinatubo after a 600-year nologist Maurice Krafft. Sadly, both Maurice
dormancy can be attributed in large part to: and his wife (Katia Krafft) were killed while
filming the eruption at Unzen Volcano, Japan,
(1) Provision of international assistance to sup- on June 3, 1991, while his work was being
plement the scientific resources of PHIVOLCS, used so effectively at Pinatubo. The video was
especially the rapid deployment of a USGS– instrumental in convincing the government
VDAP mobile observatory (Murray et al., 1996) authorities and the populace that Pinatubo
to initiate near-real-time volcano monitoring. posed a very real danger, and, consequently,
The monitoring data, plus geologic observa- official evacuation orders were heeded with
tions, allowed the detection and evaluation of little or no reluctance.
LESSONS AND TRENDS 79

Finally, as cautioned by Newhall and Punong- information constitutes the basis for hazards
bayan (1996a, p. 807), despite the positive assessments, hazards-zonation maps, and long-
outcome at Pinatubo, ‘‘a suite of problems – term forecasts of potential activity we must
including skepticism, bureaucratic and logistic promote and accelerate such basic geoscience
difficulties, and short warning from the volcano – studies.
practically eliminated the margin of safety. Ulti- r ‘‘Too few active and potentially active volca-
mately, luck and last-minute efforts played key noes of the world are being monitored ade-
roles, but warnings and response could easily have quately, and many are not monitored at all
been too late.” (Italics added). or only minimally.” (Tilling, 1995, p. 395; see
also Scarpa and Gasparini, 1996, Table 1.) As
discussed previously, because the option of
Lessons and trends hazards-assessment-based land-use planning to
reduce volcano risk is impractical for many
From this brief review and numerous other publi- volcanic areas, volcano monitoring and the
cations addressing the broad topic of monitoring capability to make short-term forecasts provide
and mitigation of volcano hazards, many specific the only means to mitigate hazards by pro-
‘‘lessons,” observations, and recommendations viding the essential data to officials to make
have been emphasized. Not surprisingly, many of well-informed decisions about evacuations and
these statements are variations on several similar other countermeasures.
r Therefore, we must monitor more of the world’s
but paramount themes. As we enter the twenty-
first century, it seems appropriate to reiterate high-risk volcanoes in densely populated areas,
some of the main lessons learned from recent vol- to begin acquiring monitoring data, ideally
canic crises and disasters of the previous century, before a volcanic crisis strikes. A long-term
in the hope that the volcanologic community and approach to increase monitoring worldwide
civil authorities can better apply these lessons would be to establish well-equipped and well-
in responding to volcanic emergencies. Outlined staffed permanent observatories at many more
below are some prime points of general agree- high-risk volcanoes, but such an endeavor
ment (not in any order of priority) that emerge would be costly and is unlikely to be suppor-
from the collective experience in confronting ted any time soon by national or international
volcano hazards to date: funding agencies, given the current world econ-
omy. However, as an interim solution and as
r Because the eruption frequency (50 to 60 vol- demonstrated by the Pinatubo case history,
canoes active each year on average) is unlikely under favorable circumstances, the rapid dep-
to decrease, the global problem of reducing vol- loyment of a mobile observatory (Murray et al.,
cano risk can only worsen with time, with con- 1996) can be highly effective and achieve suc-
tinued growth in world population, commerce cessful mitigation.
and economic development, urbanization, and r The longer the pre-crisis monitoring period, the
air traffic. While the industrialized countries earlier the reliable detection of departure of a
with active volcanoes (e.g., Japan, the United dormant volcanic system from its baseline or
States, Italy, France, Iceland, and New Zealand) ‘‘normal” behavior. Early detection allows more
must also confront this reality, this problem lead time for augmented monitoring and for
is more acute for developing countries, which scientists to work with – and to establish cred-
host most of the world’s high-risk volcanoes ibility and rapport with – public officials and
but possess inadequate resources for their study the public in the development of emergency-
and monitoring. response plans, hopefully under non-crisis
r Fundamental geologic and geochronologic data conditions.
for reconstruction of eruptive histories are r Despite recent advances in volcano monitor-
lacking for most of the world’s active and ing, successful forecasting of explosive erup-
potentially active volcanoes. Because such tions, with rare exception (e.g., Mt. Pinatubo in
80 VOLCANO HAZARDS

1991), still frustrate and elude scientists. With used. Continuous monitoring of SO2 , CO2 , and
present state-of-the-art in volcanology, ‘‘Mon- possibly other volcanic gases will almost cer-
itoring data, no matter how good or com- tainly also become routine. However, the con-
plete, do not guarantee ‘successful’ eruption cept of ‘‘keeping monitoring as simple as prac-
forecasts” (Tilling, 1995, p. 396). There is an tical” (Swanson, 1992) should not be dismissed,
urgent need in this century to conceptual- because in some situations simple methods
ize and develop new approaches to eruption might be most cost-effective and applicable.
prediction, apart from the current empirical Certainly, simple monitoring approaches are
approach largely based on pattern recognition better than no monitoring at all.
in monitoring data. In this regard, the recent r Scientists must redouble efforts to establish
research (e.g., Chouet, 1996a, 1996b) on long- and maintain effective communications with
period seismicity produced by magma move- the emergency-management authorities, the
ment or hydrothermal-pressurization phenom- media, and the populations at risk from
ena, coupled with a refined petrological and volcano hazards. A prerequisite of effective
geochemical understanding of the behavior of communications, credibility, and mutual trust
magmatic and hydrothermal fluids before and between all involved parties is absolutely
during eruption, may hold promise. critical in the development of emergency-
r As emphasized and recommended in numerous preparedness plans, based on the data obtained
works (e.g., Tilling, 1989a; Tilling and Lipman, from hazards assessments and a volcano-
1993; Punongbayan and Newhall, 1995; Newhall monitoring program as well as the consid-
and Punongbayan, 1996a, 1996b; Voight, 1996), eration of logistical, socio-economic, political,
there needs to be more effective international cultural, and other human factors of the juris-
cooperation in responding to volcanic crises. diction in which the volcano(es) is (are) situ-
The Pinatubo case history clearly demonstrates ated. In working with emergency-response offi-
the effectiveness of such collaboration. There cials, the media, and the public, scientists must
are simply too many high-risk volcanoes and learn patience and be diligent and persistent
too few trained scientists with first-hand experi- in overcoming indifference and skepticism of
ence or knowledge of eruptions in each nation. many community leaders and citizens.
Thus, it is vital to pool resources and to share r As was done successfully at Pinatubo, scientists
and exchange data and knowledge; this should need to work actively to launch broadly based
become increasingly more convenient with public-education campaigns focused on people
advances in telecommunications (e.g., e-mail, at risk in hazardous areas, the civil authori-
Internet). In particular, scientists in the devel- ties, and the emergency-response communities.
oped countries must become more involved The themes of ‘‘Prevention begins with infor-
in bilateral or international programs to work mation” and ‘‘Building a culture of prevention”
with their counterparts in the developing coun- are resoundingly sounded in the press kit for
tries, many of which lack self-sufficiency in the 1998 World Disaster Reduction Campaign of
volcanology and (or) the necessary economic the United Nations’ Secretariat for the Inter-
resources to conduct monitoring and other national Decade for Natural Disaster Reduction
hazards-mitigation studies. (1990–2000).
r Continuing the trend already in progress,
volcano-monitoring systems in this century will We must never forget the many victims of the
increasingly employ remote techniques that 1985 Ruiz catastrophe, who perished needlessly
will acquire, process, and display the data in because of failed communications and the lack
real-time or near real-time. Satellite-based sys- of an emergency-preparedness plan. By the same
tems for monitoring ground deformation (e.g., token, we also must remember that many thou-
GPS, InSAR) are likely to increasingly comple- sands of lives at Pinatubo were saved because the
ment and, in some cases, supplant the con- scientists effectively communicated the findings
ventional ground-based techniques currently and implications of their studies and convinced
REFERENCES 81

the officials to order timely evacuations, guided of this paper was critically reviewed by Steven
by a simple but effective alert-level scheme. There R. Brantley (US Geological Survey, Hawaiian Vol-
is no assurance that the successful outcome at cano Observatory), Edward W. Wolfe (US Geolog-
Pinatubo in 1991 can be routinely duplicated for ical Survey, Cascades Volcano Observatory), and
future volcanic crises, but it certainly defines a co-editor Joan Martı́. Their thoughtful, incisive
worthy goal. comments and suggestions greatly influenced my
preparation of the final manuscript. I owe them
hearty thanks for their help, but I remain solely
Note added in proof responsible for any errors, illogic, or misplaced
emphasis in the paper.
By May 2003, the growing lava dome at Soufrière
Hills Volcano, Montserrat, attained its maximum
size. On July 12–13, the dome failed catastroph- References
ically and resulted in the largest dome-collapse
event (>12 million m3 ) since the onset of erup- Anderson, T. and Flett, J. S. 1903. Report on the
tive activity in July 1995. Moreover, large explo- eruptions of the Soufrière in St. Vincent, and on
sions accompanied the collapse, producing volu- a visit to Montagne Pelée in Martinique.
minous pyroclastic flows and ash clouds as high Philosophical Transactions of the Royal Society of
as 15 km. Yet, despite the huge eruption size London, Series A, 200, 353–553.
and associated hazardous processes, there were Andres, R. J. and Rose, W. I. 1995. Remote sensing
no deaths or reported injuries. This is because sci- spectroscopy of volcanic plumes and clouds. In
entists at the Montserrat Volcano Observatory – W. McGuire, C. Kilburn, and J. Murray (eds.)
anticipating that such a large collapse was the Monitoring Active Volcanoes: Strategies, Procedures and
most likely scenario based on volcano-monitoring Techniques. London, UCL Press, pp. 301–314.
data – weeks before had advised government Aspinall, W. P., Lynch, L. L., Robertson, R. E. A., et al.
authorities to enforce rigorously the Exclusion (eds.) 1998. Special section on the Soufrière Hills
eruption, Montserrat, British West Indies, Part 1.
Zone around the volcano. Intermittent activity
Geophysical Research Letters, 25, 3387–3440.
was still continuing at the volcano as this book
Banks, N. G., Carvajal, C., Mora, H., et al. 1990.
went to press.
Deformation monitoring at Nevado del Ruiz,
Colombia, October 1985–March 1988. Journal of
Volcanology and Geothermal Research, 41, 269–295.
Acknowledgments Banks, N. G., Tilling, R. I., Harlow, D. H., et al. 1989.
Volcano monitoring and short-term forecasts. In
This brief article builds on the large and grow- R. I. Tilling (ed.) A Short Course in Geology, vol. 1,
ing scientific literature on volcanic phenomena Volcanic Hazards. Washington, DC, American
and, more recently, on volcano hazards and their Geophysical Union, pp. 51–80.
mitigation. Many of the papers cited herein are Barberi, F., Martini, M., and Rosi, M. 1990. Nevado
works published by many good friends and col- del Ruiz volcano (Colombia): pre-eruption
leagues, within and outside the United States, observations and the November 13, 1985
with whom I have worked or discussed topics catastrophe. Journal of Volcanology and Geothermal
of mutual interest. Because my paper draws so Research, 42, 1–12.
Barsczus, H. G., Filmer, P. M., and Desonie, D. 1992.
heavily on their comprehensive reviews on vol-
Cataclysmic collapse and mass wasting processes
cano hazards, I would to express my apprecia-
in the Marquesas. Eos, Transactions of the American
tion to Russell J. Blong (Natural Hazards Research
Geophysical Union, 73 (14), 313.
Centre, Macquarie University, North Ryde, NSW, Belousov, A. 1996. Deposits of the 30 March 1956
Australia) and to William E. Scott (US Geological directed blast at Bezymianny volcano, Kamchatka,
Survey, Cascades Volcano Observatory, Vancouver, Russia. Bulletin of Volcanology, 57, 649–662.
Washington) for allowing me to freely excerpt, Blong, R. J. 1984. Volcanic Hazards: A Sourcebook on the
paraphrase, or distill their texts. An earlier draft Effects of Eruptions. San Diego, CA, Academic Press.
82 VOLCANO HAZARDS

Blong, R. J. and McKee, C. 1995. The Rabaul Eruption Christiansen, R. L. 1984. Yellowstone magmatic
1994: Destruction of a Town. North Ryde, NSW, evolution: its bearing on understanding
Natural Hazards Research Centre, Macquarie large-volume explosive volcanism. In Geophysics
University. Study Committee (National Research Council)
Bluth, G. J. S., Schnetzler, C. C., Kreuger, A. J., et al. Explosive Volcanism: Inception, Evolution and Hazards.
1993. The contribution of explosive volcanism to Washington, DC, National Academic Press,
global atmospheric sulfur dioxide concentrations. pp. 84–95.
Nature, 366, 327–329. Christiansen, R. L. and Peterson, D. W. 1981.
Casadevall, T. J. 1992. Volcanic hazards and aviation Chronology of the 1980 eruptive activity. In P. W.
safety: lessons of the past decade. Federal Aviation Lipman and D. R. Mullineaux (eds.) The 1980
Administration Aviation Safety Journal, 2 (3), 9–17. Eruptions of Mount St. Helens, Washington, US
1994a. The 1989–1990 eruptions of Redoubt Volcano, Geological Survey Professional Paper no. 1250.
Alaska: impacts on aircraft operations. Journal of Washington, DC, US Government Printing Office,
Volcanology and Geothermal Research, 62, 301–316. pp. 17–67.
(ed.) 1994b. Volcanic Ash and Aviation Safety, Civetta, L., Gasparini, P., Luongo, G., et al. (eds.) 1974.
Proceedings of the 1st International Symposium Developments in Solid Earth Geophysics, vol. 6, Physical
on Volcanic Ash and Aviation Safety, US Volcanology. Amsterdam, The Netherlands, Elsevier.
Geological Survey Bulletin no. 2047. Washington, COMVOL 1975. Mayon Volcano. Manila, Philippines,
DC, US Government Printing Office. Commission on Volcanology.
Casadevall, T. J., Delos Reyes, P. J., and Schneider, D. J. Costa, J. E. and Schuster, R. L. 1988. The formation
1996. The 1991 Pinatubo eruptions and their and failure of natural dams. Geological Society of
effects on aircraft operations. In C. G. Newhall America Bulletin, 100, 1054–1068.
and R. S. Punongbayan (eds.) Fire and Mud: Crandell, D. R. 1989. Gigantic Debris Avalanche of
Eruption and Lahars of Mount Pinatubo, Philippines. Pleistocene Age from Ancestral Mount Shasta Volcano,
Quezon City, Philippines, Philippine Institute of California, and Debris-Avalanche Hazard Zonation, US
Volcanology and Seismology, pp. 1071–1088. Geological Survey Bulletin no. 1861. Washington,
Casadevall, T. J., Thompson, T. B., and Fox, T. 1999. DC, US Government Printing Office.
World Map of Volcanoes and Principal Aeronautical Crandell, D. R., Mullineaux, D. R., and Rubin, M. 1975.
Features, US Geological Survey Investigations Mount St. Helens: recent and future behavior.
Series Map no. I-2700. Washington, DC, US Science, 187 (4175), 438–441.
Government Printing Office. Crandell, D. R., Booth, B., Kusumadinata, K., et al.
CERESIS 1990. Riesgo volcánico: evaluación y 1984a. Source-Book for Volcanic-Hazard Zonation.
mitigación en América Latina: aspectos sociales, Paris, France, UNESCO.
institucionales y cientı́ficos. Lima, Perú, Centro Crandell, D. R., Miller, C. D., Glicken, H. X., et al.
Regional de Sismologı́a para América del Sur 1984b. Catastrophic debris avalanche from
(CERESIS). ancestral Mount Shasta volcano, California.
Chester, D. 1993. Volcanoes and Society. London, Edward Geology, 12, 143–146.
Arnold. Daag, A. S., Tubianosa, B. S., Newhall, C. G., et al. 1996.
Chouet, B. A. 1996a. Long-period volcano seismicity: Monitoring sulfur dioxide emission at Mount
its source and use in eruption forecasting. Nature, Pinatubo. In C. G. Newhall and R. S. Punongbayan
380, 309–316. (eds.) Fire and Mud: Eruption and Lahars of Mount
1996b. New methods and future trends in Pinatubo, Philippines. Quezon City, Philippines,
seismological volcano monitoring. In R. Scarpa Philippine Institute of Volcanology and
and R. I. Tilling (eds.) Monitoring and Mitigation of Seismology, pp. 409–414.
Volcano Hazards. Heidelberg, Germany, Dartevelle, S., Ernst, G. G. J., Stix, J., et al. 2002. Origin
Springer-Verlag, pp. 23–97. of the Mount Pinatubo climactic eruption cloud:
Chouet, B. A., Page, R. A., Stephens, C. D., et al. 1994. implications for volcanic hazards and
Precursory swarms of long-period events at atmospheric impacts. Geology, 30, 663–666.
Redoubt Volcano (1989–1990), Alaska: their origin Dawson, P. B., Dietel, C., Chouet, B. A., et al. 1998. A
and use as a forecasting tool. Journal of Volcanology Digitally Telemetered Broadband Seismic Network at
and Geothermal Research, 62, 95–135. Kilauea Volcano, Hawaii, US Geological Survey
REFERENCES 83

Open-File Report no. 98–108. Washington, DC, US Committee (National Research Council) Explosive
Government Printing Office. Volcanism: Inception, Evolution, and Hazards.
Decker, R. and Decker, B. 1997. Volcanoes, 3rd edn. New Washington, DC, National Academy Press,
York, W. H. Freeman. pp. 170–176.
Druitt, T. H. and Kokelaar, B. P. (eds.) 2002. The Eruption 1988. Volcanoes and society: challenges of
of Soufrière Hills Volcano Montserrat from 1995 to 1999. coexistence. Proceedings of the Kagoshima
Geological Society Memoir no. 21. International Conference on Volcanoes, Kagoshima
Dunn, M. G. and Wade, D. P. 1994. Influence of Prefectural Government and the National
volcanic ash clouds on gas turbine engines. In Institute for Research Advancement (NIRA), Japan,
T. J. Casadevall (ed.) Volcanic Ash and Aviation Safety, pp. 14–21.
Proceedings of the 1st International Symposium Fournier d’Albe, E. M. 1979. Objectives of volcanic
on Volcanic Ash and Aviation Safety, US monitoring and prediction, Journal of Geological
Geological Survey Bulletin no. 2047. Washington, Society of London, 136, 321–326.
DC, US Government Printing Office, pp. 107–117. Francis, P. W., Chaffin, C., Maciejewski, A., et al. 1996a.
Dvorak, J. J. and Dzurisin, D. 1997. Volcano geodesy: Remote determination of SiF4 in volcanic plumes:
the search for magma reservoirs and the a new tool for volcano monitoring. Geophysical
formation of eruptive vents. Reviews of Geophysics, Research Letters, 23, 249–252.
35, 343–384. Francis, P. W., Wadge, G., and Mouginis-Mark, P. J.
Endo, E. T. and Murray, T. L. 1991. Real-time seismic 1996b. Satellite monitoring of volcanoes. In
amplitude measurement (RSAM): a volcano R. Scarpa and R. I. Tilling (eds.) Monitoring and
monitoring and prediction tool. Bulletin of Mitigation of Volcano Hazards. Heidelberg, Germany,
Volcanology, 53, 533–545. Springer-Verlag, pp. 257–298.
Ewert, J. W. and Swanson, D. A. (eds.) 1992. Monitoring Gerlach, T. J., Westrich, H. R., and Symonds, R. B. 1996.
Volcanoes: Techniques and Strategies Used by the Staff Preeruption vapor in magma of the climactic
of the Cascades Volcano Observatory, 1980--90, US Mount Pinatubo eruption: source of the giant
Geological Survey Bulletin no. 1966. Washington, stratospheric sulfur dioxide cloud. In C. G.
DC, US Government Printing Office. [A Spanish Newhall and R. S. Punongbayan (eds.) Fire and
translation of this volume, Vigilando volcanes: Mud: Eruption and Lahars of Mount Pinatubo,
Técnicas y estrategias empleadas por el personal del Philippines. Quezon City, Philippines, Philippine
Observatorio Vulcanológico Cascades, 1980--90, was Institute of Volcanology and Seismology,
published in 1993 in cooperation with the Office pp. 415–433.
of Foreign Disaster Assistance (OFDA), US Agency Gerlach, T. J., Delgado, H., McGee, K. A., et al. 1997.
for International Development; it is available Application of the LI-COR CO2 analyzer to
upon request from the Scientist-in-Charge, volcanic plumes: a case study, volcán
Cascades Volcano Observatory, US Geological Popocatépetl, Mexico, June 7 and 10, 1995. Journal
Survey, 5400 MacArthur Blvd., Vancouver, WA of Geophysical Research, 102, 8005–8019.
98661, USA.] Giggenbach, W. F. 1996. Chemical composition of
Farrar, C. D., Sorey, M. L., Evans, W. C., et al. 1995. volcanic gases. In R. Scarpa and R. I. Tilling (eds.)
Forest-killing diffuse CO2 emissions at Mammoth Monitoring and Mitigation of Volcano Hazards.
Mountain as a sign of magmatic unrest. Nature, Heidelberg, Germany, Springer-Verlag, pp. 221–256.
376, 675–678. Glicken, H. 1998. Rockslide-debris avalanche of May
Feuillard, M., Allegre, C. J., Brandeis, G., et al. 1983. The 18, 1980, Mount St. Helens Volcano, Washington.
1975–1977 crisis of la Soufrière de Guadeloupe Bulletin of Geological Society of Japan, 49, 55–106.
(F.W.I.): a still-born magmatic eruption. Journal of Gorshkov, G. S. 1959. Giant eruption of the volcano
Volcanology and Geothermal Research, 16, 317–334. Bezymianny. Bulletin Volcanologique, 20, 76–109.
Fisher, R. V., Smith, A. L., and Roobol, M. J. 1980. Government of Montserrat and Her Majesty’s
Destruction of St. Pierre, Martinique by ash cloud Government 1998. Sustainable Development Plan:
surges, May 8 and 20, 1902. Geology, 8, 472–476. Montserrat Social and Economic Recovery Programme --
Fiske, R. S. 1984. Volcanologists, journalists, and the A Path to Sustainable Development. HGM Office of
concerned local public: a tale of two crises in the the Governor, Olveston, Montserrat, Government
eastern Caribbean. In Geophysics Study of Montserrat and Her Majesty’s Government.
84 VOLCANO HAZARDS

Gudmundsson, M. T., Sigmundsson, R., and Björnsson, Maurice Krafft. Available from: Northwest
B. 1997. Ice–volcano interaction of the 1996 Gjálp Interpretive Association, 3029 Spirit Lake
subglacial eruption, Vatnajökull, Iceland. Nature, Highway, Castle Rock, Washington 98611, USA.
389, 954–957. Izett, G. A., Obradovich, J. D., and Mehnert, H. H.
Hall, M. L. 1990. Chronology of the principal scientific 1988. The Bishop Ash Bed (middle Pleistocene) and
and governmental actions leading up to the some Older (Pliocene and Pleistocene) Chemically and
November 13, 1985 eruption of Nevado del Ruiz, Mineralogically Similar Ash Beds in California, Nevada,
Colombia. Journal of Volcanology and Geothermal and Utah, US Geological Survey Bulletin no. 1675.
Research, 42, 101–115. Washington, DC, US Government Printing Office.
1992. The 1985 Nevado del Ruiz eruption: scientific, Jackson, D. B., Kauahikaua, J., and Zablocki, C. J. 1985.
social, and governmental responses and Resistivity monitoring of an active volcano using
interaction before the event. In G. J. H. McCall, the controlled-source electromagnetic technique,
D. J. C. Laming, and S. C. Scott, Geohazards: Natural Kilauea, Hawaii. Journal of Volcanology and
and Man-Made. London, Chapman and Hall, Geothermal Research, 90, 545–555.
pp. 43–52. Janda, R. J., Scott, K. M., Nolan, K. M., et al. 1981. Lahar
Harlow, D. H., Power, J. A., Laguerta, E. P., et al. 1996. movement, effects, and deposits. In P. W. Lipman
Precursory seismicity and forecasting of the June and D. R. Mullineaux (eds.) The 1980 Eruptions of
15, 1991, eruption of Mount Pinatubo. In C. G. Mount St. Helens, Washington, US Geological Survey
Newhall and R. S. Punongbayan (eds.) Fire and Professional Paper no. 1250. Washington, DC, US
Mud: Eruption and Lahars of Mount Pinatubo, Government Printing Office, pp. 461–478.
Philippines. Quezon City, Philippines, Philippine Jónsson, P., Sigurdsson, O., Snorrason, A., et al. 1998.
Institute of Volcanology and Seismology, Course of events of the jökulhlaup on
pp. 285–305. Skeidarársandur, Iceland, in November 1996.
Heilprin, A. 1903. Mount Pelée and the Tragedy of Abstracts Volume, 15th International
Martinique. Philadelphia, PA, J. B. Lippincott. Sedimentological Congress, Alicante, Spain, 12–17
Heliker, C. C., Mangan, M. T., Mattox, T. N., et al. 1998. April, pp. 456–457.
The character of long-term eruptions: inferences Kagoshima Conference 1988. Proceedings of the
from episodes 50–53 of the Pu’o ‘O’o – Kagoshima International Conference on Volcanoes,
Kupaianaha eruption of Kilauea Volcano. Bulletin 19--23 July 1988. Kagoshima, Japan, Kagoshima
of Volcanology, 59, 381–393. Prefectural Government and the National
Hill, D. P. 1984. Monitoring unrest in a large silicic Institute for Research Advancement
caldera, the Long Valley–Inyo Craters Volcanic (NIRA).
Complex in east-central California. Bulletin Kieffer, S. W. 1981. Fluid dynamics of the May 18 blast
Volcanologique, 47, 371–396. at Mount St. Helens. In P. W. Lipman and D. R.
Hoblitt, R. P., Miller, C. D., and Vallance, J. W., 1981. Mullineaux (eds.) The 1980 Eruptions of Mount St.
Origin and stratigraphy of the deposit produced Helens, Washington, US Geological Survey
by the May 18 directed blast. In P. W. Lipman and Professional Paper no. 1250. Washington, DC, US
D. R. Mullineaux (eds.) The 1980 Eruptions of Mount Government Printing Office, pp. 379–400.
St. Helens, Washington, US Geological Survey Kling, G. W., Clark, M. A., Compton, H. R., et al. 1987.
Professional Paper no. 1250. Washington, DC, US The 1986 Lake Nyos gas disaster in Cameroon,
Government Printing Office, pp. 401–419. west Africa. Science, 236, 169–175.
Holcomb, R. T. and Searle, R. C. 1991. Large landslides Krueger, A. J., Walter, L. S., Schnetzler, C. C., et al.
from oceanic volcanoes. Marine Geotechnology, 10, 1990. TOMS measurement of the sulphur dioxide
19–32. emitted during the 1985 Nevado del Ruiz
Houghton, B. F., Latter, J. H., and Hackett, W. R. 1987. eruptions. Journal of Volcanology and Geothermal
Volcanic hazard assessment for Ruapehu Research, 41, 7–15.
composite volcano, Taupo volcanic zone, New Lanari, R., Lundgren, P., and Sansosti, E. 1998.
Zealand. Bulletin of Volcanology, 49, 737–751. Dynamic deformation of Etna volcano observed
IAVCEI 1995. Understanding Volcanic Hazards: a by satellite radar interferometry. Geophysical
30-minute videotape (copyrighted), International Research Letters, 25, 1541–1548.
Association of Volcanology and Chemistry of the Lander, J. F. and Lockridge, P. A. 1989. United States
Earth’s Interior (IAVCEI), produced by the (late) Tsunamis (including United States Possessions)
REFERENCES 85

1690--1988, National Oceanic and Atmospheric McGuire, W. J. 1995a. Monitoring active volcanoes: an
Administration Publication no. 41–2. Washington, introduction. In W. J. McGuire, C. Kilburn, and
DC, US Government Printing Office. J. Murray (eds.) Monitoring Active Volcanoes:
Latter, J. H. 1981. Tsunamis of volcanic origin. Bulletin Strategies, Procedures and Techniques. London, UCL
Volcanologique, 44, 467–490. Press, pp. 1–31.
(ed.) 1988. Volcanic Hazards: Assessment and Monitoring, 1995b. Prospects for volcano surveillance. In W. J.
IAVCEI Proceedings in Volcanology vol. 1. McGuire, C. Kilburn, and J. Murray (eds.)
Heidelberg, Germany, Springer-Verlag. Monitoring Active Volcanoes: Strategies, Procedures and
Lee, W. H. K. (ed.) 1989. Toolbox for Seismic Data Techniques. London, UCL Press, pp. 403–410.
Acquisition, Processing, and Analysis, International McGuire, W. J., Kilburn, C., and Murray, J. (eds.) 1995.
Association of Seismology and Physics of the Monitoring Active Volcanoes: Strategies, Procedures and
Earth’s Interior (IASPEI) Software Library. El Techniques. London, UCL Press.
Cerrito, CA, Seismological Society of America, 1, McNutt, S. R. 1996. Seismic monitoring and eruption
21–46. forecasting of volcanoes: a review of the
Le Guern, F., Tazieff, H., and Faivre-Pierret, R. 1982. An state-of-the-art and case histories. In R. Scarpa and
example of health hazards, people killed by gas R. I. Tilling (eds.) Monitoring and Mitigation of
during a phreatic eruption: Dieng Plateau (Java, Volcano Hazards. Heidelberg, Germany,
Indonesia), February 19, 1979. Bulletin Springer-Verlag, pp. 99–146.
Volcanologique, 45, 153–156. Mercado, R. A., Lacsamana, J. B. T., and Pineda, G. L.
Lipman, P. W. and Mullineaux, D. R. (eds.) 1981. The 1996. Socioeconomic impacts of the Mount
1980 Eruptions of Mount St. Helens, Washington, US Pinatubo eruption. In C. G. Newhall and R. S.
Geological Survey Professional Paper no. 1250. Punongbayan (eds.) Fire and Mud: Eruption and
Washington, DC, US Government Printing Office. Lahars of Mount Pinatubo, Philippines. Quezon City,
Lipman, P. W., Moore, J. G., and Swanson, D. A. 1981. Philippines, Philippine Institute of Volcanology
Bulging of the north flank before the May 18 and Seismology, pp. 1063–1069.
eruption: geodetic data. In P. W. Lipman and D. R. Montserrat Volcano Observatory 2001. Scientific and
Mullineaux (eds.) The 1980 Eruptions of Mount St. hazards assessment of the Soufrière Hills
Helens, Washington, US Geological Survey Volcano, Montserrat: The Montserrat Volcano
Professional Paper no. 1250. Washington, DC, US Observatory (MVO), 6–8 September 2001. Available
Government Printing Office, pp. 143–155. online at: http://www.mvo.ms/shvha.htm
Lockwood, J. P., Costa, J. E., Tuttle, M. L., et al. 1988. Moore, J. G. and Moore, G. W. 1984. Deposit from a
The potential for catastrophic dam failure at Lake giant wave on the Island of Lanai, Hawaii. Science,
Nyos maar, Cameroon. Bulletin of Volcanology, 50, 226, 1312–1315.
340–349. 1988. Large-scale bedforms in boulder gravel
Macdonald, G. A. 1962. The 1959 and 1960 eruptions produced by giant waves in Hawaii. Geological
of Kilauea Volcano, Hawaii, and the construction Society of America Special Paper, 229, 101–109.
of walls to restrict the spread of the lava flows. Moore, J. G., Clague, D. A., Holcomb, R. T., et al. 1989.
Bulletin Volcanologique, 24, 249–294. Prodigious submarine landslides on the Hawaiian
1972. Volcanoes. Englewood Cliffs, NJ, Prentice- Ridge. Journal of Geophysical Research, 94,
Hall. 17465–17484.
Massonnet, D., Briole, P., and Arnaud, A. 1995. Moore, J. G., Normark, W. R., and Holcomb, R. T. 1994.
Deflation of Mount Etna monitored by Giant Hawaiian landslides. Annual Reviews of Earth
spaceborne radar interferometry. Nature, 75, and Planetary Science, 22, 119–144.
567–570. Murray, J. B., Pullen, A. D., and Saunders, S. 1995.
Mattox, T. N., Heliker, C., Kauahikaua, J., et al. 1993. Ground deformation surveying of active
Development of the 1990 Kalapana Flow Field, volcanoes. In W. J. McGuire, C. Kilburn, and J.
Kilauea Volcano, Hawaii. Bulletin of Volcanology, 55, Murray (eds.) Monitoring Active Volcanoes: Strategies,
407–413. Procedures and Techniques. London, UCL Press,
McGee, K. A. and Gerlach, T. M. 1998. Airborne pp. 113–150.
volcanic plume measurements using a FTIR Murray, T. L., Ewert, J. W., Lockhart, A. B., et al. 1996.
spectrometer, Kilauea volcano, Hawaii. Geophysical The integrated mobile volcano-monitoring system
Research Letters, 25, 615–618. used by the Volcano Disaster Assistance Program
86 VOLCANO HAZARDS

(VDAP). In R. Scarpa and R. I. Tilling (eds.) and lahars. Journal of Volcanology and Geothermal
Monitoring and Mitigation of Volcano Hazards. Research, 41, 17–66.
Heidelberg, Germany, Springer-Verlag, Pinatubo Volcano Observatory Team 1991. Lessons
pp. 315–362. from a major eruption: Mt. Pinatubo, Philippines.
Myers, B., Brantley, S. R., Stauffer, P., et al. 1997. What Eos, Transactions of the American Geophysical Union,
Are Volcano hazards? US Geological Survey Fact 72, 545, 552–553, 555.
Sheet no. 002–97. Washington, DC, US Punongbayan, R. S. 1987. Disaster preparedness
Government Printing Office. systems for natural hazards in the Philippines: an
Newhall, C. G. and Punongbayan, R. S. 1996a. The assessment. In Geologic Hazards and Disaster
narrow margin of successful volcanic-risk Preparedness Systems. Quezon City, Philippines,
mitigation. In R. Scarpa and R. I. Tilling (eds.) Philippine Institute of Volcanology and
Monitoring and Mitigation of Volcano Hazards. Seismology, pp. 77–101.
Heidelberg, Germany, Springer-Verlag, Punongbayan, R. S. and Newhall, C. G. 1995. Warning
pp. 807–838. the public about Mount Pinatubo: a story worth
(eds.) 1996b. Fire and Mud: Eruption and Lahars of repeating. Stop Disasters, 25(III), 11–14.
Mount Pinatubo, Philippines. Quezon City, Punongbayan, R. S., Newhall, C. G., and Hoblitt, R. P.
Philippines, Philippine Institute of Volcanology 1996a. Photographic record of rapid geomorphic
and Seismology. change at Mount Pinatubo, 1991–94. In C. G.
Newhall, C. G. and Self, S. 1982. The volcanic Newhall and R. S. Punongbayan (eds.) Fire and
explosivity index (VEI): an estimate of explosive Mud: Eruption and Lahars of Mount Pinatubo,
magnitude for historical volcanism. Journal of Philippines. Quezon City, Philippines, Philippine
Geophysical Research (Oceans and Atmospheres), 87, Institute of Volcanology and Seismology,
1231–1238. pp. 21–66.
Nunnari, G. and Puglisi, G. 1995. GPS: monitoring Punongbayan, R. S., Newhall, C. G., Bautista, M. A.,
volcanic deformation from space. In W. J. et al. 1996b. Eruption hazard assessments and
McGuire, C. Kilburn, and J. Murray (eds.) warnings. In C. G. Newhall and R. S. Punongbayan
Monitoring Active Volcanoes: Strategies, Procedures and (eds.) Fire and Mud: Eruption and Lahars of Mount
Techniques. London, UCL Press, pp. 151–183. Pinatubo, Philippines. Quezon City, Philippines,
Peterson, D. W. 1986. Volcanoes: tectonic setting and Philippine Institute of Volcanology and
impact on society. In Geophysics Study Seismology, pp. 67–85.
Committee (National Research Council) Active Rodolfo, K. S., Umbal, J. V., Alonso, R. A., et al. 1996.
Tectonics. Washington, DC, National Academy Two years of lahars on the western flank of
Press, pp. 231–246. Mount Pinatubo: initiation, flow processes,
1988. Volcanic hazards and public response. Journal deposits, and attendant geomorphic and
of Geophysical Research, 93, 4161–4170. hydraulic changes. In C. G. Newhall and R. S.
1996. Mitigation measures and preparedness plans Punongbayan (eds.) Fire and Mud: Eruption and
for volcanic emergencies. In R. Scarpa and R. I. Lahars of Mount Pinatubo, Philippines. Quezon City,
Tilling (eds.) Monitoring and Mitigation of Volcano Philippines, Philippine Institute of Volcanology
Hazards. Heidelberg, Germany, Springer-Verlag, and Seismology, pp. 989–1014.
pp. 701–718. Rozdilsky, J. L. 2001. Second hazards assessment
Peterson, D. W. and Tilling, R. I. 1993. Interactions and sustainable hazards mitigation: disaster
between scientists, civil authorities and the recovery on Montserrat. Natural Hazards Review,
public at hazardous volcanoes. In C. R. J. Kilburn 2, 64–71.
and G. Luongo (eds.), Active Lavas: Monitoring and Rymer, H. 1996. Microgravity monitoring. In R. Scarpa
Modelling. London, UCL Press, pp. 339–365. and R. I. Tilling (eds.) Monitoring and Mitigation of
PHIVOLCS 1988. Distribution of active and inactive Volcano Hazards. Heidelberg, Germany,
volcanoes in the Philippines (educational poster, 1 Springer-Verlag, pp. 169–197.
sheet). Manila, Philippines, Philippine Institute of Sarna-Wojcicki, A. M., Shipley, S., Waitt, R. B., Jr., et al.
Volcanology and Seismology (PHIVOLCS). 1981. Areal distribution, thickness, mass, volume,
Pierson, T. C. and Janda, R. J. 1990. Perturbation and and grain size of air-fall ash from the six major
melting of snow and ice by the 13 November 1985 eruptions of 1980. In P. W. Lipman and D. R.
eruption of Nevado del Ruiz, Colombia, and Mullineaux (eds.) The 1980 Eruptions of Mount St.
consequent mobilization, flow and deposition Helens, Washington, US Geological Survey
REFERENCES 87

Professional Paper no. 1250. Washington, DC, US 1996. Hazards of large volcanic debris avalanches
Government Printing Office, pp. 577–600. and associated eruptive phenomena. In R. Scarpa
Scarpa, R. and Gasparini, P. 1996. A review of volcano and R. I. Tilling (eds.) Monitoring and Mitigation of
geophysics and volcano-monitoring methods. In Volcano Hazards. Heidelberg, Germany,
R. Scarpa and R. I. Tilling (eds.) Monitoring and Springer-Verlag, pp. 541–572.
Mitigation of Volcano Hazards. Heidelberg, Germany, Sigmundsson, F., Vadon, H., and Massonnet, D. 1995.
Springer-Verlag, pp. 3–22. Readjustment of the Krafla spreading segment to
Scarpa, R. and Tilling, R. I. (eds.) 1996. Monitoring and crustal rifting measured by Satellite Radar
Mitigation of Volcano Hazards. Heidelberg, Germany, Interferometry. Geophysical Research Letters, 24,
Springer-Verlag. 1843–1846.
Schuster, R. L. and Crandell, D. R. 1984. Catastrophic Sigurdsson, H., Houghton, B., McNutt, S. R., et al. (eds.)
debris avalanches from volcanoes. Proceedings of 2000. Encyclopedia of Volcanoes. San Diego, CA,
the 4th International Symposium on Landslides, Academic Press.
Toronto, Canada, vol. 1, pp. 567–572. Simarski, L. T. 1992. Volcanism and Climate Changes:
Scott, K. M. 1988. Origins, Behavior, and Sedimentology of Special Report. Washington, DC, American
Lahars and Lahar-Runout Flows in the Toutle--Cowlitz Geophysical Union.
River System, US Geological Survey Professional Simkin, T. 1993. Terrestrial volcanism in space and
Paper no. 1447-A. Washington, DC, US time. Annual Reviews Earth and Planetary Science, 21,
Government Printing Office. 427–452.
1989. Magnitude and Frequency of Lahars and Lahar- Simkin, T. and Fiske R. S. 1983. Krakatau 1883: The
Runout Flows in the Toutle-Cowlitz River System, US Volcanic Eruption and its Effects. Washington, DC,
Geological Survey Professional Paper no. 1447-B. Smithsonian Institution Press.
Washington, DC, US Government Printing Office. 1984. Krakatau 1883: a classic geophysical event. In
Scott, W. E. 1989a. Volcanic and related hazards. In C. S. Gillmor (ed.) History of Geophysics, vol. 1.
R. I. Tilling (ed.) A Short Course in Geology, vol. 1, Washington, DC, American Geophysical Union,
Volcanic Hazards. Washington, DC, American pp. 46–48.
Geophysical Union, pp. 9–23. Simkin, T. and Siebert, L. 1994. Volcanoes of the World: A
1989b. Volcanic-hazards zonation and long-term Regional Directory, Gazetteer, and Chronology of
forecasts. In R. I. Tilling (ed.) A Short Course in Volcanism during the Last 10,000 Years. Tucson, AZ,
Geology, vol. 1, Volcanic Hazards. Washington, DC, Geoscience Press.
American Geophysical Union, pp. 25–49. Simkin, T., Unger, J. D., Tilling, R. I. et al. (compilers)
Scott, W. E., Hoblitt, R. P., Torres, R. C., et al. 1996. 1994. This Dynamic Planet: World Map of Volcanoes,
Pyroclastic flows of the June 15, 1991, climactic Earthquakes, Impact Craters, and Plate Tectonics, US
eruption of Mount Pinatubo. In C. G. Newhall and Geological Survey Special Map, 2nd edn.
R. S. Punongbayan (eds.) Fire and Mud: Eruption and Washington, DC, US Government Printing
Lahars of Mount Pinatubo, Philippines. Quezon City, Office.
Philippines, Philippine Institute of Volcanology Sorey, M. L., Evans, W. C., Kennedy, B. M., et al. 1998.
and Seismology, pp. 545–570. Carbon dioxide and helium emissions from a
Self, S., Zhao, J.-X., Holasek, R. E., et al. 1996. The reservoir of magmatic gas beneath Mammoth
atmospheric impact of the 1991 Mount Pinatubo Mountain, California. Journal of Geophysical
eruption. In C. G. Newhall and R. S. Punongbayan Research, 103, 303–315, 323.
(eds.) Fire and Mud: Eruption and Lahars of Mount Stephens, C. D., Chouet, B. A., Page, R. A., et al. 1994.
Pinatubo, Philippines. Quezon City, Philippines, Seismological aspects of the 1989–1990 eruptions
Philippine Institute of Volcanology and at Redoubt Volcano, Alaska: the SSAM perspective.
Seismology, pp. 1089–1115. Journal of Volcanology and Geothermal Research, 62,
Shimozuru, D. 1972. A seismological approach to the 153–182.
prediction of volcanic eruptions. In The Stommel, H. and Stommel, E. 1983. Volcano Weather:
Surveillance and Prediction of Volcanic Activity. Paris, The Story of 1816, The Year without a Summer.
France, UNESCO, pp. 19–45. Newport, RI, Seven Seas Press.
Siebert, L. 1984. Large volcanic debris avalanches: Sutton, J., Elias, T., Hendley, J. W., II, et al. 1997. Volcanic
characteristics and source areas, deposits, and Air Pollution: A Hazard in Hawaii, US Geological
associated eruptions. Journal of Volcanology and Survey Fact Sheet no. 169–97. Washington, DC, US
Geothermal Research, 22, 163–197. Government Printing Office.
88 VOLCANO HAZARDS

Swanson, D. A. 1992. The importance of field Peligros Volcánicos, was published in 1993 by the
observations for monitoring volcanoes, and the World Organization of Volcano Observatories
approach of ‘‘keeping monitoring as simple as (WOVO), a Commission of the International
practical.” In J. W. Ewert and D. A. Swanson (eds.) Association of Volcanology and Chemistry of the
Monitoring Volcanoes: Techniques and Strategies used Earth’s Interior (IAVCEI); it is available through
by the Staff of the Cascades Volcano Observatory, the editor.]
1980--90, US Geological Survey Bulletin no. 1966. 1995. The role of monitoring in forecasting volcanic
Washington, DC, US Government Printing Office, events. In W. J. McGuire, C. Kilburn, and J. Murray
pp. 219–223. (eds.) Monitoring Active Volcanoes: Strategies,
Swanson, D. A., Wright, T. L., and Helz, R. T. 1975. Procedures and Techniques. London, UCL Press,
Linear vent systems and estimated rates of pp. 369–402.
magma production and eruption for the Yakima 1996. Hazards and climatic impact of
Basalt on the Columbia Plateau. American Journal subduction-zone volcanism: a global and
of Science, 275, 877–905. historical perspective. In G. E. Bebout, D. W.
Swanson, D. A., Casadevall, T. J., Dzurisin, D., et al. Scholl, S. H. Kirby, et al. (eds.) Subduction: Top to
1983. Predicting eruptions at Mount St. Helens, Bottom, Geophysical Monograph no. 96.
June 1980 through December 1982. Science, 221, Washington, DC, American Geophysical Union,
1369–1376. pp. 331–335.
1985. Forecasts and predictions of eruptive activity Tilling, R. I. and Lipman, P. W. 1993. Lessons in
at Mount St. Helens, USA: 1975–1984. Journal of reducing volcano risk. Nature, 364, 277–280.
Geodynamics, 3, 397–423. Tilling, R. I., Topinka, L., and Swanson, D. A. 1990.
Symons, G. J. (ed.) 1888. The Eruption of Krakatoa, and Eruptions of Mount St. Helens: Past, Present, and
Subsequent Phenomena, Report of the Krakatoa Future, US Geological Survey General-Interest
Committee of the Royal Society of London. Publications Series. Washington, DC, US
London, Trübner & Co. Government Printing Office.
Tanguy, J.-C., Ribière, C., Scarth, A., et al. 1998. Victims Torres, R. C., Self, S., and Martinez, M. M. L. 1996.
from volcanic eruptions: a revised database. Secondary pyroclastic flows from the June 15,
Bulletin of Volcanology, 60, 137–144. 1991, ignimbrite of Mount Pinatubo. In C. G.
Tazieff, H. and Sabroux, J.-C. 1983. Developments in Newhall and R. S. Punongbayan (eds.) Fire and
Volcanology, vol. 1, Forecasting Volcanic Events. Mud: Eruption and Lahars of Mount Pinatubo,
Amsterdam, The Netherlands, Elsevier. Philippines. Quezon City, Philippines, Philippine
Tedesco, D. 1995. Monitoring fluids and gases at active Institute of Volcanology and Seismology,
volcanoes. In W. J. McGuire, C. Kilburn, and J. pp. 665–678.
Murray (eds.) Monitoring Active Volcanoes: Strategies, UNDRO/UNESCO 1985. Volcanic Emergency Management.
Procedures and Techniques. London, UCL Press, New York, Office of the United Nations Disaster
pp. 315–345. Relief Co-ordinator (UNDRO), United Nations
Thorarinsson, S. 1957. The jökulhlaup from the Katla Educational, Scientific, and Cultural Organization
area in 1955 compared with other jökulhlaups in (UNESCO).
Iceland. Jökull, 7, 21–25. UNESCO 1972. The Surveillance and Prediction of Volcanic
1969. The Lakagigar eruption of 1783. Bulletin Activity: A Review of Methods and Techniques. Paris,
Volcanologique, 33, 919–929. France, United Nations Educational, Scientific,
1979. On the damage caused by volcanic eruptions and Cultural Organization (UNESCO).
with special reference to tephra and gases. In US Geological Survey 2005. Photoglossary. Available
P. D. Sheets and D. K. Grayson (eds.) Volcanic online at: http://volcanoes.vsgs.gov
Activity and Human Ecology. San Diego, CA, Verbeek, R. D. M. 1885. Krakatau. Batavia, Indonesia,
Academic Press, pp. 125–159. Imprimerie de l’Etat.
Tilling, R. I. 1989a. Volcanic hazards and their Voight, B. 1990. The 1985 Nevado del Ruiz volcano
mitigation: progress and problems. Reviews of catastrophe: anatomy and retrospection. Journal of
Geophysics, 27, 237–269. Volcanology and Geothermal Research, 42,
(ed.) 1989b. A Short Course in Geology, vol. 1, Volcanic 151–188.
Hazards. Washington, DC, American Geophysical 1996. The management of volcano emergencies:
Union. [A Spanish translation of this volume, Los Nevado del Ruiz. In R. Scarpa and R. I. Tilling
REFERENCES 89

(eds.) Monitoring and Mitigation of Volcano Hazards. (ed.) 1990b. Special Issue on Nevado del Ruiz,
Heidelberg, Germany, Springer-Verlag, pp. 719–769. Colombia, II. Journal of Volcanology and Geothermal
Voight, B., Glicken, H., Janda, R. J., et al. 1981. Research, 42, 1–224.
Catastrophic rockslide avalanche of May 18. In Wolfe, E. W. and Hoblitt, R. P. 1996. Overview of the
P. W. Lipman and D. R. Mullineaux (eds.) The 1980 eruptions. In C. G. Newhall and R. S.
Eruptions of Mount St. Helens, Washington, US Punongbayan (eds.) Fire and Mud: Eruption and
Geological Survey Professional Paper no. 1250. Lahars of Mount Pinatubo, Philippines. Quezon City,
Washington, DC, US Government Printing Office, Philippines, Philippine Institute of Volcanology
pp. 347–377. and Seismology, pp. 3–20.
Voight, B., Janda, R. J., Glicken, H., et al. 1983. Nature Wolfe, E. W., Neal, C. A., Banks, N. G., et al. 1988.
and mechanism of the Mount St. Helens rockslide Geologic observations and chronology of eruptive
avalanche. Geotechnique, 33, 224–273. events. In E. W. Wolfe (ed.) The Puu Oo eruption of
Walker, G. P. L. 1982. Volcanic risk. Interdisciplinary Kilauea Volcano, Hawaii: Episodes 1 through 20,
Science Reviews, 7, 148–157. January 3, 1983, through June 8, 1984, US Geological
White, R. A. 1996. Precursory deep long-period Survey Professional Paper no. 1463. Washington,
earthquakes at Mount Pinatubo: spatio-temporal DC, US Government Printing Office, pp. 1–97.
link to a basalt trigger. In C. G. Newhall and R. S. Wright, T. L., Chun, J. Y. F., Esposo, J., et al. 1992. Map
Punongbayan (eds.) Fire and Mud: Eruption and Showing Lava Flow Hazard Zones, Island of Hawaii, US
Lahars of Mount Pinatubo, Philippines. Quezon City, Geological Survey Miscellaneous Field Studies
Philippines, Philippine Institute of Volcanology Map no. MF-2193. Washington, DC, US
and Seismology, pp. 307–327. Government Printing Office.
Wicks, C., Jr., Thatcher, W., and Dzurisin, D. 1998. Young, R. W. and Bryant, E. A. 1992. Catastrophic wave
Migration of fluids beneath Yellowstone Caldera erosion on the southeastern coast of Australia:
inferred from satellite radar interferometry. impact of the Lanai tsunami ca. 105 ka? Geology,
Science, 282 (5388), 458–462. 20, 199–202.
Williams, R. S., Jr. (ed.) 1997. Lava-Cooling Operations Young, S. R., Sparks, R. S. J., Aspinall, W. P., et al.
during the 1973 Eruption of Eldfell Volcano, Heimaey, 1998a. Overview of the eruption of Soufrière Hills
Vestmannaeyjar, Iceland, US Geological Survey volcano, Montserrat, 18 July 1995 to December
Open-File Report no. 97–724. Washington, DC, US 1997. Geophysical Research Letters, 25, 3389–3392.
Government Printing Office. Young, S. R., Voight, B., Sparks, R. S. J., et al. (eds.)
Williams, S. N. (ed.) 1990a. Special Issue on Nevado del 1998b. Special section on the Soufrière Hills
Ruiz, Colombia, I. Journal of Volcanology and eruption, Montserrat, British West Indies, Part 2.
Geothermal Research, 41, 1–377. Geophysical Research Letters, 25, 3651–3700.
Chapter 3

Anticipating volcanic eruptions


Joan Martı́ and Arnau Folch

Introduction During the lifetime of the volcano the corre-


sponding magma chamber may change in size
and shape. A chamber’s shape and size strongly
Volcanic activity involves movements of fluids, influence the stress field associated with the
magma, and vapor, inside the volcano and its magma-filled reservoir, which controls the geom-
feeding systems. These movements will cause etry and frequency of dyke injection and extru-
external signals, eruption precursors, which may sion from the volcano. Eruptions are typically
alert us to the proximity of a volcanic event. triggered by the injection of new magma into the
Such precursory signals can be detected by an chamber from deeper levels, leading to chamber
adequate volcano monitoring program. Moni- overpressure either directly through the added
toring techniques include a range of geophysi- magma volume, or indirectly by sudden cooling
cal and geochemical techniques, encompassing and crystallization of the new magma leading
seismic, ground-deformation, gravity, and mag- to wholesale volatile exsolution. Improved under-
netic observations, gas monitoring, and remote standing of magma chamber evolution, host-rock
sensing. Successful forecasting of volcanic events mechanics, volatile solubilities, and degassing is
depends on the precision of the surveillance net- needed if eruptive triggers are to be understood,
work in detecting any changes in the volcano’s and quantitative models of eruption probability
current behavior. To interpret the geochemical developed.
and geophysical precursors correctly, however, Few large volcanic eruptions have been
it is also important to understand the physics observed directly, which imposes important
of the volcanic processes involved in volcanic restrictions on the study of eruptive phenomena.
eruptions. Detailed knowledge of the volcano, its Our knowledge of eruption dynamics is based
internal structure and style, and potential trig- mostly on studies of the solid products of past
gering mechanisms of past eruptions must be eruptions, and the development of appropriate
combined with adequate monitoring if future vol- theoretical and experimental models. Significant
canic eruptions are to be anticipated and their advances have been made in these fields over
effects mitigated (Fig. 3.1). In summary, predic- recent years, though there remains much to learn
tion of volcanic activity has the aims of determin- from this combined field, experimental, and the-
ing when and where a future eruption will occur, oretical approach. Causes, dynamics, and effects
and how it will proceed. However, it is also impor- of volcanic eruptions have traditionally been
tant to understand why the next eruption will studied separately, due to the complexity of the
occur. problems involved and due to difficulties inher-
Volcanic eruptions are caused mainly by pro- ent in a multidisciplinary approach. However,
cesses occurring in magma chambers at depth. that a volcanic eruption, its causes and effects,
Volcanoes and the Environment, eds. J. Martı́ and G. G. J. Ernst. Published by
Cambridge University Press.  C Cambridge University Press 2005.
MAGMAS AND MAGMA CHAMBERS 91

has been treated in considerable detail elsewhere


VOLCANO (e.g., McGuire et al., 1995; Scarpa and Tilling,
1996 and references therein) and this chapter also
gives a brief review of the main geophysical and
geochemical monitoring methods, outlining the
fundamental concepts and current state of the
PHYSICAL VOLCANO art in eruption forecasting.
VOLCANOLOGY MONITORING
- field and laboratory - ground-based and
studies of volcanic remote geophysical Magmas and magma chambers
products and geochemical
- experimental and monitoring The causes of volcanic activity are ultimately
theoretical modeling explained within the framework of the plate tec-
tonics theory (see Chapters 1 and 4, this volume)
relating global tectonics and magmatism. Most
volcanic activity is driven by basaltic magma
that forms in the mantle and rises to litho-
VOLCANO
spheric levels when significant volume is avail-
FORECASTING
able. Magma ascent is controlled by density dif-
ferences with the host rocks and by the rheology
Fig. 3.1. Integration of studies of physical volcanology and
volcano monitoring is necessary to understand a volcano’s of the magma and host rocks. At deeper levels
past and present activity and to predict its future behavior. magma generated by fusion of the mantle will
rise diapirically while surrounding host rocks
deform plastically, while at shallower levels in
should be regarded as a continuum of related the brittle lithosphere it will ascend though dyke-
phenomena is increasingly being accepted by like fractures. Although eruptions directly from
volcanologists. the source region are thought to occur some-
Understanding the physical principles that times, magma will generally accumulate at lev-
control pre-eruptive and eruptive processes is els above the source region, forming reservoirs or
not only crucial in reducing the risk to human magma chambers where cooling will cause phys-
lives and properties in volcanic areas but also in ical and chemical changes. The intrusion of hot,
assessing the effects of volcanic eruptions on the mantle-derived basaltic magma into the crust
environment. Though the precise prediction of can also cause extensive melting of shallow host
future volcanic activity remains elusive, signifi- rocks, generating crustal-derived, silicic magmas
cant progress has been made in understanding (Huppert and Sparks, 1988).
the processes involved, aiding the assessment of Many factors will influence whether or not
precursors of eruptions. In this chapter, we will a magma will erupt at the Earth’s surface, and
review the processes that accompany volcanic how and when this will occur. These include tec-
eruptions and the methods typically used in their tonic setting, source rock composition and melt-
characterization. We will describe in some detail ing process, magma chamber characteristics, and
the theoretical modeling approach for a general magmatic evolution. The chemistry and physical
audience in recognition of its increasing suc- properties of magmas, as well as their ascent
cess in anticipating volcanic eruptions and their and magma chamber formation, are covered in
effects. We refer mainly to explosive eruptions Chapter 1. Here, we use these concepts to explain
of large silicic volcanoes, as these represent per- how magmatic evolution can cause a volcanic
haps the most important of all volcanic hazards. eruption.
However, the concepts covered in this chapter are Volcanic eruptions range from quiet lava out-
also applicable to quieter, less dangerous erup- pourings to explosive events that can inject mater-
tions of basaltic volcanoes. Volcano monitoring ial into the stratosphere and cause damage over
92 ANTICIPATING VOLCANIC ERUPTIONS

setting, gives rise to primary magmas generi-


cally termed basalts, characterized by low volatile
contents, high density, high temperature, and
low viscosity. In contrast, when the crust melts
the resulting magma is more silicic, cooler
and lower in density, and more viscous and
richer in volatile species, particularly water. The
differences between mantle and crustal-derived
primary magmas are mainly due to different
source rocks compositions. Primary basaltic mag-
mas also vary in composition depending on the
nature of the source mantle which varies with
tectonic setting. The conditions of pressure and
temperature at which partial melting occurs, also
Fig. 3.2. The influence of source rock composition and dependent on tectonic setting, are different. They
degree of partial melting on the composition of primary
also influence the composition of the resulting
(first-generated) magmas.
primary melts, as the composition of the source
rocks and the conditions of melting are different
thousands of square kilometers. The reasons for in each case.
these differences are likely to be related to stor- The wide diversity of primary magma com-
age and differentiation of magma, its movement positions shows consequently significant varia-
in dykes connected to the surface, and the near- tions in rheology and other physical properties.
surface degassing of volatiles. The chemical com- As magma is of lower density than its solid host
position of a magma, including the presence of rocks, it will tend to move away from the source
volatiles, will strongly influence its physical prop- region by ascending to shallower levels. As it rises
erties, specifically its rheology, and the manner it will cool and the primary magma will begin
in which it will erupt. Useful additional infor- to crystallize and experience magmatic differen-
mation on the genesis and evolution of magmas tiation. It may accumulate in shallower reser-
may be found in several petrological textbooks voirs, interact with wall rocks, and mix with
(e.g., Carmichael et al., 1974; McBirney, 1984; other magmas (Fig. 3.3). All of these influences
Wilson, 1989; Hall, 1996). on evolving magma composition will cause vari-
Magma is a silicate liquid melt that may con- ations in rheology, eventually influencing the
tain solid (mineral) and volatile (dissolved and characteristics of any eruption. Any combination
exsolved fluid) components. It originates by par- of magmatic differentiation, mixing, and con-
tial melting of mantle or crustal rocks. During tamination might occur during the evolution of
partial melting, under given conditions of pres- a magma from source to eruption and solidifi-
sure and temperature, minerals with low melting cation at the Earth’s surface. The complexity of
points will melt, leaving part of the crustal or magmatic evolution has been revealed by stud-
mantle source rock unmelted. Mantle and crustal ies of magmatic rocks solidified at the Earth’s
source rocks are composed mainly of silicate min- surface (volcanic rocks) or beneath the surface
erals, whose chemical components, and the pro- (plutonic rocks and hypabyssal rocks). Studies of
portion in which these are added to the melt, the chemistry and mineralogy of volcanic rocks
define the resulting primary magma composition are a key component in determining the magma’s
(Fig. 3.2). evolution beneath the surface. Petrological stud-
Mantle-derived magmas erupt in all tec- ies are particularly important in determining the
tonic environments, notably at mid-ocean ridges, pre-eruptive conditions of the magma and in
oceanic islands, continental and oceanic arcs, predicting a volcano’s future behavior. Eruption
and continental rifts (see Chapters 1 and 4). Par- style is a direct consequence of a magma’s chem-
tial melting of mantle, whatever the tectonic ical and physical properties, so that petrological
MAGMAS AND MAGMA CHAMBERS 93

PRIMARY
magma chambers have been inferred convinc-
MAGMA ingly from geophysical studies able to detect the
presence of molten rock within the solid litho-
LIQUID
A sphere, but also following from the existence
ROCK of plutons, igneous bodies representing ancient
A
magma chambers at the Earth’s surface. The
LIQUID chemical and mineralogical features of volcanic
B
ROCK CRYSTAL FRACTIONATION rocks also typically indicate the temporary arrest
B and differentiation of magma at a certain depth,
typically between 2 and 8 km, before eruption.
ROCK Magma chambers apparently range in size from
LIQUID D
C less than 1 km3 to more than 1000 km3 . The
ROCK
C size of chamber attained, and the volume of
magma that may be accumulated locally within
the lithosphere, will depend mainly on the stress
field that dominates at the site, on the com-
MAGMA position and rheology of the magma, and on
A the rate of injection of new magma into the
MAGMA
+ C MAGMA MIXING chamber. Over its lifetime, a magma chamber
MAGMA may undergo episodes of inflation, when new
B magma is injected into it, and episodes of defla-
tion when magma leaves the chamber and is
intruded into the host rocks or erupted at the sur-
face. Crystallization and subsequent differentia-
tion of magma will tend to lead to a progressive
ROCK
HOST

MAGMA MAGMA ASSIMILATION


A B change in composition and physical properties.
This may eventually cause the magma to become
oversaturated in volatiles, which will vaporize
Fig. 3.3. The effect of fractional crystallization, magma and cause the chamber to expand until magma
mixing and crustal assimilation on magma composition. or gas is voided from the chamber.
Magmatic differentiation describes mechanisms by which a Chapter 1 describes how magma chambers
magma changes its physicochemical (due to phase changes)
form at neutral buoyancy levels, where the bulk
composition during cooling. Fractional crystallization, the
density of magma is equal to that of the host
most important mechanism of magmatic differentiation,
involves the separation of crystals from the magmatic liquid. rocks. Levels of neutral buoyancy are associated
Fractional crystallization of magma by gravitational separation with mechanical discontinuities in the Earth,
of crystals that progressively form during cooling may occur such as the asthenosphere–lithosphere boundary,
over an extended period and by the accumulation of magma the mantle–crust or Moho discontinuity, the base
in reservoirs or magma chambers. Removal of crystals of a volcano edifice, or stratigraphic discontinu-
induces the composition of the remaining liquid magma to ities in the interior of a volcanic pile. Depending
change. In addition to changing its composition, a magma may on its location, the regional stress field influenc-
experience mixing with other magmas and host rock
ing the chamber’s development will vary.
contamination. In both cases the physicochemical
The magma chamber feeding a volcano will
composition of the magma will also change.
change shape during the volcano’s lifetime. A
chamber’s shape and size will strongly influence
studies of products of past eruptions are crucial the local stress field, associated with the pres-
in explaining eruptive phenomena. ence of the chamber, controlling magma intru-
Magma chambers are reservoirs of molten sion and eruption (Gudmundsson, 1988). Under-
silicate liquid plus any crystals it may contain standing the geometric evolution of a magma
(Marsh, 1989). The existence and dimensions of chamber is thus a prerequisite for understanding
94 ANTICIPATING VOLCANIC ERUPTIONS

Fig. 3.4. Triggering mechanisms of


explosive eruptions. Pressure inside
Pr the magma chamber (Pm ) may
increase by an amount (Pmg ) due
Pl Pl to volatile oversaturation or to
intrusion of fresh magma (Pmi ). If
∆ Pm
the chamber pressure exceeds
g
lithostatic pressure (Pl ) plus the
Pm tensile strength of the surrounding
∆ Pm rocks (Pr ), tensile vertical fractures
i

may open allowing magma to escape


from the chamber, sometimes
eruption : P m >P l + P r reaching the surface to cause a
volcanic eruption.

the eruption patterns of volcanoes. Moreover, the The pressure increase can result from two main
behavior of volcanic systems depends greatly on causes discussed above: volatile oversaturation
how magma accumulates at shallow levels in the driven by differentiation or the injection of fresh
crust, and at what depth and how the result- magma into the chamber (Blake, 1981, 1984; Tait
ing magma chamber connects to the surface. The et al., 1989; Folch and Martı́, 1998) (Fig. 3.4). Other,
distribution of volcanic vents at the surface will less common, mechanisms such as seismic excita-
depend on the combination of regional tectonic tion of entrapped bubbles (Sturtevant et al., 1996)
structure and that of the active magmatic sys- and tectonic triggers (Linde and Sacks, 1998) have
tem, including the strength of host rocks, the also been recognized. Whatever the case, the
intensive parameters of the magma chamber, and magma chamber responds to restore mechanical
its deformation history of inflation and deflation equilibrium with its surroundings by injecting
episodes, and by the stress distribution within dykes or deforming the wall rocks. If the excess
the volcanic edifice. Understanding of vent dis- pressure cannot be released, a magma-filled frac-
tribution and vent development during explosive ture may propagate from the chamber to the sur-
and effusive volcanic eruptions constitutes one of face and trigger an eruption. A volcanic eruption
the most important elements of volcanic hazard typically ends once mechanical equilibrium has
assessment. been restored by the withdrawal of a few percent
of the stored magma. When the eruption ceases,
the remaining magma can continue to cool and
Eruptive processes evolve, and a new eruptive event can occur if con-
ditions for overpressure are again reached. The
As we have seen, a volcanic cycle typically starts period between consecutive eruptions is known
with the intrusion and accumulation of deeply as the repose period, and is typically of the order
sourced magma into the crust and its accu- of a few years to a few thousands of years. A sili-
mulation to a magma chamber. Once magma cic magma chamber will commonly produce a
is accumulated in a shallow chamber, heat is sequence of eruptions rather than a single event;
slowly lost through the chamber walls and the the products extruded contributing to the con-
magma gradually changes its composition and struction of a volcanic edifice.
physical properties as it differentiates. If the A correlation between the length of volcanic
conditions for magma to exit the chamber are repose periods and the volatile content and explo-
never achieved, the chamber will eventually com- sivity of the subsequent eruption suggested to
pletely crystallize to form plutonic rocks. How- early workers that the exsolution of magmatic
ever, during its lifetime the chamber may become volatiles within the chamber could produce over-
overpressurized occasionally leading to eruption. pressures sufficiently high to trigger an explosive
ERUPTIVE PROCESSES 95

eruptive event (Morey, 1922; Smith and Bailey,


1968; Smith, 1979; Blake, 1984; Tait et al., 1989).
This hypothesis was supported by petrological evi-
dence that magmatic water content progressively
increased during the repose periods. A further
section describes mathematical models that have
allowed quantification of the effect of volatile
oversaturation on the increase of magmatic
pressure.
The second, and probably the most common,
Dispersion of
mechanism triggering volcanic eruptions is the pyroclasts and gas Fragmentation
injection of new magma into the chamber (Blake, level
1981). Essentially the volume of material able to
Magma with
be stored within the chamber is limited by the exsolved volatiles
(gas bubbles)
compressibility of the resident magma and the
expansivity of the chamber walls. This critical
Exsolution
volume cannot be exceeded, and excess magma level
is forced to exit the chamber through dykes.
Evidence that magma chambers are open sys- Magma with
dissolved volatiles
tems periodically experiencing inputs of magma
is compelling. For example, volcano inflation and
ground deformation (see p. 113), often accompa- Fig. 3.5. Volcano undergoing an explosive eruption.
nying seismic activity, can be explained by an Exsolution of volatiles may occur by decompression when
influx of fresh magma into the chamber. Active magma approaches the surface, or by oversaturation in the
eruptive periods separated by repose periods of residual liquid as magma cools down and crystallizes.
a few years are suggested to be related to the Whichever the case, volatile exsolution causes small bubbles
uprise of magma at rates of several cubic meters to form above the exsolution level, where magma becomes a
per second (Blake, 1981). A further section gives a two-phase mixture of liquid and gas. Once nucleated, gas
bubbles grow by a combination of mass transfer of volatile
detailed description of theoretical models devel-
components from magma to bubbles and by depressurization
oped to quantify the effect of magma injection (Sparks, 1978). This occurs until magma disrupts at the
on triggering volcanic eruptions. fragmentation level due to the bursting of bubbles. The
The physics of explosive volcanism began to magma thus becomes a gas continuum with dispersed
be understood during the late 1970s (e.g., Walker, magmatic particles called pyroclasts, which are ejected from
1973; Wilson, 1976; Wilson et al., 1980). Processes the vent at high velocity. After Fisher and Schmincke (1984).
related to explosive volcanism have become a sub-
ject of major interest to scientists in the last two
decades, due to the potential hazards for human within the conduit leading to the surface. Above
life and environment. the exsolution level magma becomes a two-phase
Explosive volcanic activity is strongly related mixture composed of a gas-saturated liquid with
to the presence of volatiles originally dissolved gas bubbles dispersed in it. Gas bubbles, once
in the magma. The solubilities of the main nucleated, grow by several mechanisms. These
volatile species (H2 O, CO2 , SO2 ) are highly depen- include diffusion as more exsolved gas migrates
dent on pressure. Therefore, during the ascent of into the bubbles, decompression as ascent to shal-
magma, a level is reached where magma pressure low levels lowers ambient pressure and expands
equals the exsolution pressure of the dissolved the bubbles, and coalescence when small gas bub-
gas species and vapor is exsolved to form small bles join to form larger ones (Sparks, 1978; Prou-
bubbles. The level at which this occurs is known sevitch et al., 1993; Toramaru, 1995). Bubble size,
as the exsolution level (Fig. 3.5). The exsolution however, cannot be arbitrarily large and a level is
level can be located either in the chamber, or reached where magma fragments become a gas
96 ANTICIPATING VOLCANIC ERUPTIONS

continuum containing liquid magma fragments, a basis for reconstructing previous eruptions.
called pyroclasts, dispersed within. This level is Such studies might also reveal to us the most
known as the fragmentation level, above which probable cause of an eruption. A detailed field
the gas–pyroclast mixture is accelerated to super- study of products of previous eruptions and
sonic velocities. their comprehensive petrological and geochemi-
Magma fragmentation is a defining feature cal analysis can, therefore, provide information
of explosive volcanism but is still not well necessary to constrain the pre-eruptive condi-
understood and controversial (Dingwell, 1998). tions of magma and indicate conditions likely
The debate over fragmentation encompasses two to trigger future eruptions. Field and labora-
major models. The first to be developed (Sparks, tory data also include the measurements of the
1978) concentrates on strain criteria, where physicochemical properties of volcanic materials
magma disrupts and fragments when its vesic- necessary to constrain physical and numerical
ularity attains a critical value. During ascent the models of volcano behavior and eruption dynam-
exsolution of volatiles from the magma causes ics used to anticipate future volcanic activity.
viscosity to increase by several orders of magni- The reconstruction of past eruptions and their
tude. The viscosity increase of the bubbly mix- comparison with observed examples constitutes a
ture prevents further growth of bubbles, allow- key to understanding the eruptive behavior of vol-
ing them to generate sufficient stress to disrupt canoes (e.g., Bonadonna et al., 1998). While direct
the surrounding magma. This model is supported observations of volcanic eruptions are restricted
by the observation that pumice clasts produced in number, improved observational techniques
in explosive eruptions typically exhibit vesiculari- have permitted detailed records of the different
ties in the range 70–75%, the value for maximum phases of these eruptions to be compiled allow-
spherical bubble-packing. The second model con- ing correlation between volcanic products and
centrates on strain rate criteria. The vigorous their corresponding phases of the eruption. Such
kinematics of the mixture of gas and pyroclasts observations also improve our knowledge of the
in the uppermost parts of the conduit may pro- emplacement dynamics of the eruption products.
duce strain rates sufficient to induce a viscoelas- The direct observation of recent eruptions, such
tic response of the liquid melt and subsequent as that of Mt. Pinatubo in the Philippines in 1991,
fragmentation when stresses exceed the magma has demonstrated how complex the interaction
tensile strength (Papale, 1998; Martı́ et al., 1999). may be between the deposition of volcanic prod-
In the foregoing paragraphs of this section ucts and later erosion and secondary processes
we have briefly summarized the main features that may rapidly transform their original appear-
of explosive volcanism. Characteristics of explo- ance. There are many volcanoes, however, from
sive eruptions and their products have already which no eruption has been recorded in historic
been described in Chapter 1. The reader will find times, but which may become active in the near
additional information on the subject in several future. Again Pinatubo is a good example of such
recent textbooks and specialized volumes and a volcano. In such cases, the only way to antic-
reviews (Willams and McBirney, 1979; Fisher and ipate the volcano’s future behavior is the study
Schmincke, 1984; Cas and Wright, 1987; Wohletz of the products of past eruptions and compari-
and Heiken, 1992; Martı́ and Araña, 1993; Papale, son with observed eruptions, and inferences from
1996; Sparks et al., 1997; Freund and Rosi, 1998; theoretical and experimental models.
Gilbert and Sparks, 1998; Sigurdsson, 2000).
Field studies
Field studies of the products of an eruption are
performed to characterize the sequence of vol-
Field and laboratory studies: the canic deposits and to reconstruct the sequence
reconstruction of past eruptions of eruptive and depositional events (Fig. 3.6). Vol-
canological field studies aim to determine the rel-
Products derived from a volcanic eruption are ative stratigraphy and distribution of the differ-
incorporated into the geological record and form ent units forming a particular eruption sequence
FIELD AND LABORATORY STUDIES: THE RECONSTRUCTION OF PAST ERUPTIONS 97

Fig. 3.6. Interpretation of the dynamics of


past eruptions requires detailed field studies
Ignimbrite
of the stratigraphy, distribution and nature of
Plinian
pumice their products. The photograph shows a
fallout detail of the stratigraphic sequence of
Eruption C youngest (0.3–0.2 Ma) eruptive episodes of
the Las Cañadas volcano in Tenerife (Canary
Islands). The products of three different
eruptions are visible clearly separated by
Phreatoplinian Phreatoplinian
ash paleosoils. The first two eruptions are
ash
represented by ignimbrite deposits, while the
Paleosoil youngest shows a more complex sequence of
Ignimbrite
Eruption B products including two phreatoplinian ash
deposits separated by a Plinian pumice fallout
deposit at the base, and an ignimbrite on top.
Paleosoil

Eruption A Ignimbrite

(Rosi, 1996). A volcanic eruption may involve sev- to discriminate them from others related to dif-
eral phases, each giving rise to different products. ferent eruptions or to non-volcanic processes.
For example, the eruption may include the forma- Geological mapping to determine the distribu-
tion of a Plinian column that generates units of tion of volcanic units, may be carried out at scales
fallout deposits, after which several collapses of from regional to local, depending on the size
the Plinian column may produce pyroclastic flows of the eruption. This will determine what com-
and surges. Each of the deposits produced by bination of photogeological or remote sensing
such an eruption will exhibit different litholog- and direct field reconnaissance is required. Strati-
ical, sedimentological, and stratigraphic charac- graphic correlation is necessary in order to con-
teristics, as well as having different distributions firm the distribution of deposits but also to char-
around the volcano. Plinian fallout deposits will acterize lateral variations in thickness, geometry,
be widely distributed and will tend to mantle the and lithology, which may occur in each unit or
surrounding topography. Deposits from pyroclas- facies. Stratigraphic studies are important also in
tic flows and surges tend to accumulate in low- distinguishing groups of deposits from different
lying areas (see Chapter 1). Following deposition eruptions or volcanic centers. Field lithological
eruptive products may be affected by other geo- studies of volcanic deposits are needed to identify
logical processes producing secondary volcanic an appropriate sampling policy for accompanying
deposits. Primary and secondary deposits from a mineralogical and geochemical studies. More-
particular eruption appearing in the geological over, identifying the sedimentological character-
record show complex stratigraphic relationships istics, such as grain-size distribution and sed-
depending on the characteristics of the eruption, imentary structures, is crucial in determining
the topography of the surrounding, and the envi- emplacement mechanisms of the deposits.
ronment in which the eruption occurs. Each volcanic unit should be described in
Field studies should provide the informa- detail, its stratigraphic relationships noted, and
tion necessary to identify the different prod- outcrops spatially correlated. In a sequence of
ucts and phases of a particular eruption, and volcanic deposits the different units should be
98 ANTICIPATING VOLCANIC ERUPTIONS

widespread paleosoils or erosion and alteration


surfaces allowing unequivocal recognition of dif-
ferent eruption sequences, are thus desirable.
Such criteria can also be useful in reconstruct-
ing the long-term evolution of volcanic systems,
identifying volcanic cyclicity and making future
volcanic activity more predictable. Stratigraphic
correlations between outcrops are also necessary
to identify the provenance and thickness varia-
tions of the different volcanic deposits. Here, the
use of isopach and isoplet maps (Fig. 3.8) will con-
strain the source vent for each deposit. However,
the exact location of eruptive vents is not always
possible to identify, particularly in complex vol-
canic fields where the activity of several volcanoes
may coincide in time and space.
While stratigraphic studies are vital, determi-
Fig. 3.7. Interpreting the products of volcanic eruptions
nations of the absolute age of a volcano’s deposits
exposed in the geological record becomes more difficult will help greatly in reconstructing its eruptive
when products of different vents are interbedded in the same history. Absolute age determinations, typically
stratigraphic sequence. In this example from Tenerife, a distal estimated from isotopic compositions, will allow
Plinian fallout deposit, light in color, is interbedded with the different eruptions and possible cycles of erup-
dark-colored proximal products of several small basaltic tions to be recognized. However, recovering an
cones. absolute age of a deposit is not always possi-
ble, depending on the dating technique and the
type and alteration of the sample. The establish-
grouped according to a hierarchy allowing tem- ment of the relative age scheme for a group
poral succession of volcanic events to be recog- of volcanic products using stratigraphic methods
nized (Fisher and Schmincke, 1984; Martı́, 1993). should therefore always be a step before deter-
Each eruption may last days to months, or sev- mining the chronology of a sequence using radio-
eral years, and may include several phases with metric methods.
durations of hours to days. Moreover, each phase
may include several pulses of activity of seconds Laboratory studies of volcanic rocks
to minutes duration. Different deposits may be Studies of the samples of volcanic units collected
generated by each eruptive phase or pulse, so that from the field continue in the laboratory with
the resulting sequence of deposits will be as com- mineralogical and geochemical analysis. This will
plex as the eruption. Where direct observation of enable the magma composition and possibly also
an eruption is lacking, a good interpretation of its pre-eruptive conditions to be determined.
the lithological nature and stratigraphic position When a magma erupts at the Earth’s surface it
of each volcanic deposit will provide a basis for cools rapidly and its liquid part quenches to a
understanding a particular eruption and predict- glass. The crystals observed in the lavas and pyro-
ing a volcano’s potential future behavior. clasts forming the magmatic products comprise
Understanding a sequence of deposits from a the mineral assemblage stable during cooling of
volcanic eruption becomes more difficult as the the magma in the chamber (Fig. 3.9). Some reac-
age of the deposits increases. This is because of tions of the crystals with the liquid may have
post-eruptive processes that may remove or trans- occurred during the eruptive processes. However,
form the deposits and the incorporation into the it is generally assumed that the observed com-
stratigraphic sequence of products from other positions and assemblage of large crystals (phe-
eruptions (Fig. 3.7). Stratigraphic criteria, such as nocrysts), and the quenched glassy groundmass,
FIELD AND LABORATORY STUDIES: THE RECONSTRUCTION OF PAST ERUPTIONS 99

Las Cañadas caldera


Total
Guajara Peak thickness
3120

1
2 1.2
0.33
0.28
4 1.2 0.25
>0.3
3115 6 1.37 1.6 0.34
>4.13 8 1.89
4 6 8 4.62 0.74
4.8 1.5
>5.6 5.64 4.89 1.84
5.5 2.27 1
8.87 6.63 5.5 3.8
2 2
1 >5.05 5.76 >3 2 1.9 1.05
>6.73 7.67 4.36 1.84
3.2 2.5 2.2 2
3110 >6 7.62 2.5
8.35 3.3 2.16
>2.06 7.4
>5 3.1 2.74
4.81 >7 2.75
>2.46
3.61 >3.6 3.8
>1.8 8.29 3.2
2.13 >2.5 >3 3.5
4.6
3.78
1.67
3105 3.2 6.36 >3.08
5.83
5.94 0 5
4.5 Km
>3.9

3100

330 335 340 345 350 355 360

Fig. 3.8. Isopach maps of the Granadilla pumice fallout composition. If minerals and liquid were in equi-
deposit, Tenerife. Isopach maps show thickness variations of librium with local conditions then thermody-
the deposit. Each contour joins points where the deposit has namic techniques such as geothermometry and
a particular thickness value. Isopach maps, together with geobarometry can be used to determine the tem-
isoplet maps showing grain-size variations within the deposit,
perature and pressure at which a mineral assem-
are useful for discriminating fallout deposits from other kinds
blage formed (Wood and Fraser, 1976). Experimen-
of pyroclastic deposits formed by flow processes, and help to
identify the vent area and distribution of the deposit. tal petrology also provides tools to determine
Simplified from Bryan et al. (2000). the pre-eruptive conditions of magmas. Natural
volcanic samples can be melted and recrystal-
lized under a variety of physical conditions until
will be representative of the magma composition the mineral assemblage contained in the nat-
immediately before the eruption. Petrological ural sample is reproduced and natural condi-
investigation of the mineralogy and geochem- tions deduced. Of particular interest in experi-
istry of the minerals and glass can also constrain mental petrological studies of volcanic rocks are
physicochemical conditions within the magma those focused on the stability relations of hydrous
chamber prior to eruption. minerals, i.e., minerals that contain water or
A magma’s mineralogy and chemical composi- other volatiles in their structures. The stability of
tion reflect its thermodynamic history. For exam- hydrous minerals is closely linked to the activity
ple, mineral assemblages and compositions vary of water in the magma, which provides a major
in response to pressure, temperature, and liquid driving force for explosive eruptions.
100 ANTICIPATING VOLCANIC ERUPTIONS

Fig. 3.9. Thin section of a phonolitic lava flow from also include the analysis of samples of melt
Tenerife showing large phenocrysts of alkali feldspar, grown in included in crystals. Such melt inclusions repre-
the magma chamber before the eruption, embedded in a sent trapped samples of magmatic liquid preserv-
glassy groundmass formed by chilling of magmatic liquid at the ing the original volatile content of the magma,
Earth’s surface.
unmodified by degassing prior to or during the
eruption. Melt inclusion data from a sequence of
One of the key parameters controlling frag- eruptions from the same volcano may show tem-
mentation and eruption dynamics is pre-eruptive poral variations in the content and composition
magmatic gas content. Estimating the pre- of volatiles. Such variations can be interpreted in
eruptive gas content of a magma from its erup- light of eruptive behavior, making future trends
tive products is thus desirable in order to char- and activity more predictable.
acterize the hazard potential of related future Concentrations of many stable and radioac-
volcanic activity. Analysis of hydrous mineral sta- tive isotopes, such as U, Th, Ra, Sr, Nd, O, H,
bility relations is one possible way to estimate and S, are measured in the laboratory, in order
pre-eruptive gas content, but this will give only to characterize the origin and evolution of mag-
the gas content at the time prior to that at mas. The source region of a particular magma has
which the minerals equilibrated. The gas con- an isotopic signature that is transferred to the
tent at the time of fragmentation might be esti- magma it generates, allowing us to distinguish
mated from either analysis of bubble size distri- between different melting zones and sources. If
butions or water contents of pumice produced by magma assimilates host rocks in its way to the
an eruption. However, this would require a firm surface the isotopic signature of the host rock
understanding of post-eruptive degassing pro- will be imported to the magma, allowing isotopes
cesses, perhaps achievable through theoretical to detect contamination processes during ascent.
calculations. Laboratory studies aimed at deter- In addition, isotopes are important in constrain-
mining the pre-eruptive gas content of magmas ing processes occurring in the shallow magmatic
SCALE AND ANALOGUE MODELING 101

plumbing systems. From isotopic petrogenetic the tapping of a volatile-rich magma chamber.
studies it is possible to identify the presence of Such models provide no real quantification of
a shallow reservoir and constrain its volume, the the factors critical in nature, as their scales and
average residence time of the magma, and the proportions are different. Scale experiments, on
rates of magma replenishment, fractional crys- the other hand, allow quantitative comparison
tallization, and magmatic degassing. between the experimental model and the natural
Radioactive disequilibrium between short- system, by reproducing the geometry and relative
lived nuclides of the U and Th decay series proportions of forces pertaining in the natural
can provide abundant information on magmatic case. Scale experiments require not only dimen-
fractionation processes and their timescales and sional or scale equivalence (geometrical similar-
has become a subject of increasing interest in ity) with the natural model, but also the equiva-
modern volcanology (Lambert et al., 1986; Con- lence between body and surface forces acting on
domines et al., 1988, 1995). Increasingly precise the system (dynamical similarity). The major dif-
measurements and theoretical modeling of short- ficulty with such experiments is the scaling of
lived daughter nuclides of the U and Th decay gravity forces, which requires a centrifuge, pro-
series can be used to constrain timescales of mag- viding for unworkable complexity in experimen-
matic processes ranging from days to thousands tal design. This restriction has significantly lim-
of years. Investigation and modeling of the sys- ited the application of true scale experiments in
tematics of the daughter isotopes of radon (210 Pb, volcanology and has favored analogue and dimen-
210
Po, 210 Bi), which can occur in volatile S and sionally scaled approaches.
Cl compounds, also contribute to understanding Analogue experiments have become useful
degassing timescales and mechanisms and open in visualizing and understanding hidden pro-
up new possibilities for the use of volcanic gas cesses occurring in magma chambers, which can
monitoring in eruption prediction (Lambert et al., be identified from the study of volcanic prod-
1986; Le Cloarec and Marty, 1991; Le Cloarec et al., ucts but cannot be observed directly. These pro-
1992; Pennisi and Le Cloarec, 1998). cesses, including fractional crystallization, mix-
ing, convection, and zonation, help to deter-
mine the characteristics of the eruptive processes
Scale and analogue modeling and should be considered a tool for forecast-
ing volcano behavior. Certain models involve the
Scale and analogue modeling has, over the last use of saline solutions with contrasting physi-
few decades, become a useful tool in the study cal properties, to simulate interactions between
of volcanic processes, particularly applied to the magmas, such as mixing or density stratifica-
dynamics of shallow magma chambers, erup- tion. In some experiments a hot, dense saline
tion mechanisms, volcanic plumes, pyroclastic solution is injected into the bottom of a labo-
flows, and volcano stability. Scale experiments in ratory tank containing a cooler, lighter solution,
the laboratory can simulate volcanic processes to simulate replenishment of a shallow magma
under conditions analogous to those occurring chamber by fresh magma (Huppert and Sparks,
in nature, allowing visualization of phenomena 1980; Huppert et al., 1983). Convection in homo-
that cannot normally be directly observed. Exper- geneous magma chambers has been simulated by
imental work, together with field and laboratory an experimental tank containing a homogeneous
data, is also important in validating numerical saline solution heated from below (Turner and
models. Campbell, 1986). Analogue experiments with flu-
What are the differences between analogue ids of different physical properties have also been
and scale models? An analogue model is one that used to investigate magma mixing during flow
reproduces some fundamental aspect of the nat- in volcanic conduits during explosive eruptions
ural phenomenon without direct physical scal- (Freund and Tait, 1986; Blake and Fink, 1987).
ing between the two. For example, opening a Experimental models have been specifically
bottle of champagne is a simple analogue of developed to understand eruption dynamics,
102 ANTICIPATING VOLCANIC ERUPTIONS

from bubble growth to the emplacement of pyro- provide good analogues of natural process and
clastic flows. Bubble nucleation and growth in assist understanding of the physics of magma
magmas is a crucial aspect of explosive volcan- fragmentation during explosive eruptions. How-
ism mainly understood through theoretical mod- ever, experiments involving analogue magmas
eling. Experiments performed in order to improve and shock-tubes cannot be properly scaled to
our understanding of magma vesiculation (Hur- nature restricting the quantitative interpretation
witz and Navon, 1994; Lyakhovsky et al., 1996) of experimental results derived from them.
mainly use natural rhyolites heated and hydrated Phreatomagmatic eruptions, involving the
at high pressures to form volatile saturated melts, interaction of magma and surface or ground-
and then decompressed instantaneously allowing water, have been successfully investigated using
bubbles to nucleate and grow. Quenching of sam- experimental models (Zimanowski, 1998). In such
ples after different times at the new pressure per- eruptions, the thermal energy of magma is
mits examination of the number, density, and transformed into mechanical energy when water
spatial distribution of the bubbles formed. In is heated and suddenly vaporized. Experiments
other experiments, natural samples are heated designed specifically to reproduce this partic-
and the growth of individual, pre-existing bub- ular type of explosive interaction have been
bles observed and compared with the viscosity performed by several researchers (Wohletz and
and temperature variation of the sample (Bagdas- Sheridan, 1983; Wohletz, 1986; Zimanowski et al.,
sarov et al., 1996). 1991, 1997). In pioneering work, Wohletz and
An important group of experimental models Sheridan (1983) used metallic melt, in some cases
of eruption dynamics, which has received great mixed with molten quartz sand, to approximate
attention over the last few years, refers to magma the silicate magma, to provide crucial infor-
fragmentation during explosive eruptions. Frag- mation concerning the control of water/melt
mentation experiments are basically designed to ratios on the resulting pyroclasts formation.
produce controlled explosions in shock tubes, Further experimental work by Zimanowski and
and include different techniques (Mader, 1998). co-workers used remelted volcanic rocks to pro-
Some experiments are an extension of the vesic- vide new insights into the process.
ulation experiments described above and involve The dynamics of eruption columns and pyro-
explosive vaporization of a volatile saturated liq- clastic flows has been also investigated through
uid by sudden decompression, producing explo- experimental models. Experiments on eruption
sive boiling. Other fragmentation experiments columns (Carey et al., 1988; Sparks et al., 1991;
suddenly decompress a sample of highly viscous Ernst et al., 1994, 1996) simulate the physical pro-
vesicular magma, producing experimental pyro- cesses that control column behavior and particle
clasts very similar to natural samples from Peléan sedimentation, allowing one to establish predic-
and Vulcanian eruptions (Alidibirov and Ding- tive models of ejecta dispersal. The experiments
well, 1996). These experiments show that brittle include a constant source of fresh, particle-laden
fragmentation of magma may occur by expansion water injected at the base of a tank filled with
of gases in nearly cooled magmatic materials. aqueous saline solution (Fig. 3.10). The exper-
To simulate the triggering effect of exsolution imental design allows variation of the injec-
of volatiles from silicic magmas, experiments tion rate and the particle loading, two of the
have used chemical reactions to generate large most important variables controlling eruption
volumes of CO2 within non-magmatic liquids, columns.
which are then subjected to sudden decompres- In a similar way, the emplacement of pyro-
sion at room temperature (Mader et al., 1994, clastic flows has been investigated through ana-
1996). These experiments cause vesiculation and logue models. Pyroclastic flows represent one of
disruption of the magma analogue and suggest the most important volcanic hazards and a good
that magma fragmentation results from acceler- knowledge of their physics is crucial for elab-
ation of the gas–liquid mixture by gas expan- orating accurate predictive models. Unable to
sion during decompression. Such experiments observe the interior of a moving pyroclastic flow,
SCALE AND ANALOGUE MODELING 103

structures that are observed in natural deposits,


helping in the interpretation of their transport
physics. Other experiments have investigated the
effects of topography on the emplacement of
pyroclastic flows (Woods and Bursik, 1994; Woods
et al., 1998). In this case, the experiments use a
laboratory tank, 1–2 m long, which contains an
ambient salt-water medium. An inclined plane
simulating a ridged topography is placed in the
tank and a finite volume of dense fluid consist-
ing of a mixture of fresh water and solid particles
is released into the tank over the inclined plane.
The experiments contribute to the understand-
ing of the relative importance of sedimentation
and entrainment in generating buoyancy in the
pyroclastic flow and formation of co-ignimbrite
clouds (see Chapter 1).
Scale and analogue models have been devel-
oped to investigate both the geometry of rel-
atively small volcanic cones (e.g., Riedel et al.,
2003) on the internal structure and stability of
large volcanoes. Formation of collapse calderas
and failure of volcano slopes have been success-
fully reproduced with simple analogue experi-
ments, which allow the reproduction of perma-
nent deformation structures such as fractures
and faults. The experiments simulate the rigid
crust using cohesive, dry, powder mixtures (sand,
fused alumina, flour, etc.), while silicone is used
Fig. 3.10. Plan-view photograph of an experiment to study
to simulate magma or a mantle plastic layer. For
the dynamics of eruption plumes. The experiment was example, three different experimental models
designed to study plume bifurcation (Ernst et al., 1994), a have been designed to study the formation of
phenomenon observed in laboratory experiments and natural collapse calderas, volcanic depressions originated
eruptions, including that of Rabaul (Papua New Guinea) in by the collapse of the magma chamber roof dur-
1994. The experiment consists of a low-speed water channel ing large volume eruptions. In all cases, a labo-
where dyed (cold or warm) water can be injected vertically ratory tank was filled with powder material and
through a nozzle on the channel floor into cold flowing water,
the main difference between experiments is the
generating turbulent buoyant plumes in a cross-current.
nature of the experimental magma chamber: an
Photograph by G. G. J. Ernst.
ice ball (Komuro et al., 1984), a inflatable balloon
(Martı́ et al., 1994), and silicone (Roche et al., 2000),
volcanologists have designed experimental mod- respectively. The results obtained are similar and
els addressed to investigate their emplacement allow one to reproduce the formation of tectonic
mechanism. Some experiments have focused on structures (inward dipping faults, ring faults,
the effect of fluidization on the emplacement of radial faults) commonly associated with the for-
pyroclastic flows (Wilson, 1980, 1984). This has mation of collapse calderas (Fig. 3.11). Analogue
been achieved in the laboratory by passing gas experiments using a cohesive mixture of powder
upwards through a natural pyroclastic deposit, in materials simulating a volcanic cone and silicone
a similar way to what is used in some industrial simulating intrusion of magma or a plastic man-
applications. These experiments have reproduced tle layer that deforms due to the load of the
104 ANTICIPATING VOLCANIC ERUPTIONS

Fig. 3.11. Analogue experiments on the formation of a the complexity of the mathematics needed to
collapse caldera. The objective was to investigate the describe theoretical models has led to the appli-
influence of the geometry and location of the magma cation of computer-based numerical simulations
chamber on the geometry and location of the resulting to volcanic processes.
caldera. An elastic-skinned balloon was inflated and emplaced
A theoretical model is a simplified abstrac-
within a cohesive powdered material (fused alumina), and
tion of a given observed phenomenon. Such a
deflated to simulate eruption and decompression of a shallow
silicic magma chamber (Martı́ et al., 1994). Scaling of material model is acceptable when it can not only repro-
properties ensures geometrical similarity of the model with duce experimental data but can also make pre-
nature. The experimental model provides insights into the dictions confirmable by further measurements.
geometry and motion of faults that control the processes of Any such theoretical model is characterized by
caldera collapse. a set of equations, termed the governing equa-
tions of the model, which describe the physics of
the problem in mathematical notation. Typically,
volcanic edifice have been performed to inves- solution of the governing equations cannot be
tigate the effect of cryptodomes (Donadieu and achieved analytically with the use of simple alge-
Merle, 1998) and spreading (Merle and Borgia, bra, and numerical techniques must be employed.
1996), respectively, on the stability of volcanic A numerical simulation is simply the solution of
edifices. a particular theoretical model under a given (or
assumed) set of conditions. Such conditions might
include the boundary and initial conditions for
Theoretical modeling in volcanology the commonest type of governing equations,
which are time-dependent differential equations
Theoretical models based on the physical dis- that explain and predict the phenomena. A
ciplines of thermodynamics, solid and fluid numerical simulation performed with a certain
mechanics, etc., have been applied to many differ- technique can be checked against other numeri-
ent volcanic phenomena and have progressively cal procedures to see whether they yield similar
become an indispensable tool. In several cases, results when applied to the same problem.
THEORETICAL MODELING IN VOLCANOLOGY 105

Pre-eruptive models addition of new magma into the chamber. Two


Modeling pre-eruptive phenomena is important types of theoretical model have been developed
in understanding processes that potentially lead to determine the main physical processes that
to volcanic eruptions. Correct comprehension of occur in each case. The first group may be called
the problem is a crucial aspect in interpret- ‘‘closed system models,” because they investigate
ing the meaning of some of the geophysical the effect of volatile oversaturation on magma
and geochemical precursors signaling an erup- pressure, assuming that no new magma is added
tive event. For example, seismic signals may to or removed from the chamber. The first model
be recorded months or years before an erup- of this type to be developed was that of Blake
tion begins. On the other hand, eruptive prod- (1984). Although the model was limited to the
ucts sometimes show clear evidences that fresh effect of water in rhyolitic magmas, its general
magma was intruded into the magma cham- approach can be applied to other magma-volatile
ber shortly, perhaps weeks, months, or a few systems. This model allowed quantification of the
years, prior to eruption. Based on previous sec- water content required to generate a critical over-
tions, this fact may be interpreted to indicate pressure equal to the tensile strength of the coun-
that an eruption was triggered by a pressure try rock, in terms of parameters such as cham-
increase inside the magma chamber related to ber depth, magma compressibility, and strength
the intrusion. Consequently, preceding seismic of the chamber walls. According to Blake’s (1984)
signals may be interpreted to record such intru- model, water contents in the range of 3–6 wt.%
sions of new magma into the chamber that have are in most cases sufficient to rupture the cham-
caused a critical pressure increase and a subse- ber. Another important conclusion of the study
quent rupture of the chamber walls. The erup- is that fractionated magmas evolving at shallow
tion of Mt. Pinatubo, Philippines, in 1991 is a depth cannot attain water contents much above
good example of this sequence of events (Newhall 6 or 7 wt.%, at which point the chamber becomes
and Punongbayan, 1996). Clearly, a good knowl- critically overpressurized and a volcanic eruption
edge of the processes occurring during replenish- is triggered. More elaborate models aim to calcu-
ment of a shallow magma chamber might help late chamber overpressure as a function of the
to predict the time lag between the seismic pre- amount of crystallization and the solubility law
cursors and the eventual eruption. As shown pre- of the volatile species present, assuming some
viously, a first approach to the understanding of initial mass of gas (Tait et al., 1989; Bower and
magma chamber processes may be obtained from Woods, 1997; Folch et al., 1998) (Fig. 3.12). These
experimental modeling and petrological studies. models have shown that the more soluble the
However, if the magnitude and duration of these volatile species, the more significant is the devel-
thermodynamic and mechanical changes affect- opment of chamber overpressure.
ing the magma chamber are to be quantified, More complex models consider a magma
a theoretical model is required. This is just an chamber as an open system, where pressure
example to illustrate how theoretical modeling of increase is caused by addition of new magma
pre-eruptive processes can help us to anticipate into the chamber. The first quantitative model
volcanic eruptions. explaining magma chamber replenishment as a
Several groups of theoretical models have first-order mechanism to trigger volcanic erup-
been developed to understand relevant pre- tions was that of Blake (1981). According to his
eruptive processes occurring in shallow magma model a volume of injected magma of about
chambers. Important aspects include cooling, 0.1% of the chamber volume or approximately 1%
differentiation, and mixing of magma, magma when a gas phase is present (Bower and Woods,
chamber degassing, and the achievement of crit- 1997) is able to produce sufficient overpressure to
ical conditions leading to rupture of the magma trigger an eruption.
chamber walls. Magma chamber pressure may The period of repose occurring between injec-
increase owing to oversaturation of volatiles tion of new magma and any subsequent erup-
when magma cools and differentiates, or by the tion suggests that the injection of a volume of
106 ANTICIPATING VOLCANIC ERUPTIONS

Fig. 3.12. Theoretically calculated


10 000
magma chamber pressure increases
5000 due to volatiles oversaturation
-2 Andesite

V (km3)
during crystallization (a) or
LOG (ηg)

replenishment (b). (a) The


Basalt
1000 relationship between the volume
-3 500
fraction of gas (ηg ) and the
100 overpressure as crystallization
-4 proceeds, assuming that no gas is
0
present at t = 0. Simplified from Tait
0 4 8 12 16 10-2 10-1 1 10
et al. (1989). (b) The volume of
OVERPRESSURE (MPa) ∆V (km3) magma, V, that must be introduced
into a chamber of constant volume,
(a) (b) Vc , to trigger eruption, for different
magma compositions. Simplified
from Blake (1981).

new magma is not, in itself, sufficient to pro- and Martı́ (1998) give a quantification of the rel-
voke an eruption immediately, and that some sec- ative importance of these processes in causing
ondary effect(s) of the injection causes the critical chamber overpressure.
overpressure. This observation has led to an Volcanic eruptions are mostly supplied with
important group of pre-eruptive models focused magma through dykes, or other types of sheet
on the effects of magma mixing as a volcanic intrusions. However, commonly, the eruption fre-
eruption trigger. This idea was first introduced quency of a volcano is only a fraction of the
to explain the 1875 Plinian eruption of Askja inferred sheet-injection frequency of its source
Volcano, Iceland (Sparks et al., 1977) and has since magma chamber, meaning that many sheets
then been invoked to explain other eruptions must stop within the crust, never reaching the
including the 1991 eruption of Mt. Pinatubo, surface to erupt. Thus in order to be able to pre-
Philippines (Pallister et al., 1992). When mafic dict volcanic eruptions and to estimate volcanic
magma is injected into the base of a chamber con- risk, not only must the condition for injection of
taining lighter and cooler silicic magma, a com- sheets from the chamber be known, but also the
plex sequence of processes occurs (Sparks et al., condition for sheet arrest (Gudmundsson et al.,
1977; Huppert et al., 1982a, 1982b; Campbell and 1999 and references therein). The types of models
Turner, 1986; Sparks and Marshall, 1986; Snyder mentioned above provide valuable insights into
and Tait, 1995, 1996). These include cooling the causes of magma overpressure, but do not
and crystallization of the mafic magma, heat- treat the resistance limits and rupture mecha-
ing and convection of the silicic magma, volatile- nisms of the country rock. Another set of theo-
oversaturation and subsequent gas release from retical models based mostly on principles of rock
both magmas, convective overturning, and large- mechanics have been developed to explain rock
scale mixing between the two magmas. Erup- failure and dyke injection under conditions of
tion may occur at any time during this sequence magmatic overpressure. These models are signif-
of mixing events if the critical overpressure is icant, allowing interpretation of the seismic sig-
attained. To summarize, exsolution of volatiles nals that precede volcanic eruptions, as well as
from the felsic magma due to a increase in tem- geodetic and gravimetric data from volcano mon-
perature or upward convection, or exsolution of itoring. They treat the physical principles control-
volatiles from the mafic magma during cooling ling rupture of dykes and dyke injection from a
and crystallization, have both been cited as mech- magma chamber and the general mechanics of
anisms by which the injection of new magma dyke propagation and arrest, as well as the pro-
and subsequent mixing events could contribute portion of dykes that reach the surface to erupt
to overpressurizing the magma chamber. Folch (Roberts, 1970; Gudmundsson, 1988; Rubin, 1995).
THEORETICAL MODELING IN VOLCANOLOGY 107

The conditions leading to sheet arrest form a


subject that has received little attention and is
poorly understood. One possibility is that sheet
propagation is arrested when magma flow into
the sheet tip is blocked by solidification when
the propagation velocity decreases below a criti-
cal level. Alternatively, arrest may occur when the
sheet enters crustal layers where sheet-normal
compressive stresses exceed the magmatic over-
pressure driving the sheet. These and others pos-
sibilities might be tested in detail by combining
extensive field studies of exposed sheet tips with
Fig. 3.13. Contours of the maximum tensile stress, in MPa, theoretical modeling (Gudmundsson et al., 1999).
around a sill-like magma chamber at the base of the volcanic Magma degassing has also been modeled theo-
area subjected to 10 MPa magmatic excess pressure. The retically. Differences in eruptive behavior broadly
results allow deduction of the position of ring faults allowing a correlate with differences in magma type, with
caldera collapse event to occur, and show that the depression gas-poor basaltic magma more likely to erupt as
would be very similar in dimension to the cross-sectional area lava, whereas gas-rich evolved magma types typi-
of the source chamber. In these boundary-element cally erupt explosively. These generalizations are
calculations the static Young’s modulus of the crustal plate is
not always correct, however, as some mafic mag-
40 GPa and Poisson’s ratio is 0.25. From Gudmundsson et al.
(1997).
mas may erupt with great energy and destruc-
tive power, while felsic magmas may erupt qui-
etly to form thick lava flows or domes with
The stress field associated with the source little impact on the surrounding environment.
magma chamber is the main parameter control- Magma degassing in shallow chambers and con-
ling whether a sheet is intruded into surround- duits seems to play a significant role in control-
ing rocks from the chamber. This stress field ling such variations in eruptive behavior. The-
requires detailed analysis, particularly regarding oretical modeling of magma degassing consists
the three-dimensional shape of the chamber and mainly of thermodynamic calculations, based on
the location of stress concentrations where sheet mineral composition data combined with experi-
injection is most likely. Traditional models of mental determinations of volatiles solubility and
rupture of magma chambers and dyke propaga- phase equilibria, and has mainly been applied
tion, however, make drastic simplifying assump- to determining degassing kinetics and magma
tions, not taking into account the complexities chamber evolution (Carroll and Holloway, 1994
of magma dynamics and the geometric evolution and references therein).
of dykes during emplacement. Such complexities
must be analyzed numerically in combination Eruptive models
with analytical work to advance our understand- Models of volcanic eruptions aim to describe
ing. Numerical modeling is now widely used physical aspects of the eruptive process using con-
to simulate the mechanics of magma chambers cepts from thermodynamic and fluid mechani-
and to study the conditions of stress distribu- cal theory. Ideally, an eruptive model should con-
tion for magma chamber rupture for different sider simultaneously the physical processes that
chamber depths, shapes, volumes, rock strengths, occur within the magma chamber, within the vol-
and loading conditions (Gudmundsson, 1988, canic conduit during the ascent of magma, and
1998; Gudmundsson et al., 1997). These numerical within the atmosphere after eruption, since the
results provide, for example, good insights on the processes that occur in each domain can affect
most favorable stress conditions that in each case the others. Unfortunately, a successful solution of
can determine the chamber rupture and eventu- such a large, coupled problem remains unachiev-
ally a volcanic eruption (Fig. 3.13). able as yet. A major difficulty is that each region
108 ANTICIPATING VOLCANIC ERUPTIONS

parameters (volatile content, eruption rate, size


Eruption of the vent, etc.) of the magma at the vent.
column SUBAERIAL Future eruptive models should address the
MODELS coupling of these partial models that to date
Pyroclastic Lava have considered only parts of the general prob-
flow Pyroclastic
flow
fall
lem. Efforts should focus on the creation of a gen-
eral model able to describe simultaneously these
Fragmentation three spatial domains. Recently, some attempts
have been made to partially couple conduit and
Volcanic CONDUIT subaerial models (Dobran et al., 1993; Neri et al.,
conduit MODELS 1998). The idea behind this has been to use as
Exsolution input boundary conditions for the subaerial mod-
of volatiles
els the conditions at the vent computed inde-
Magma pendently from a conduit model. The same idea
chamber CHAMBER could be also extended to couple the chamber
MODELS and the conduit models.

Magma chamber models


Fig. 3.14. Domains of theoretical modeling in physical A first group of syn-eruptive magma cham-
volcanology. Magma chamber models deal with the dynamics
ber mathematical models comprises analyti-
of withdrawal during the course of an eruption. Conduit
cal approaches that consider only differences
models focus on the physics of the conduit and aim to
describe processes such as exsolution of volatiles, growth of between initial and final stages and do not
bubbles, or magma fragmentation. Subaerial models focus on contemplate temporal variations of the physi-
the characteristics, dynamics, and emplacement of erupted cal parameters during the course of the erup-
products and include, among other aspects, models for the tion. Although there are limitations inherent to
eruptive column, pyroclastic and lava flows, or pyroclastic falls. these models, they provide a first-order approxi-
mation to what can be expected from an eruptive
event. Thus, for instance, an important question
is considerably different physically and the gov- in terms of volcanic hazard concerns the amount
erning equations for each have different math- of material that can potentially be extruded dur-
ematical requirements. Moreover, many of the ing a volcanic eruption. Several analytical mod-
processes involved are not well understood. els aim to answer this question calculating the
To simplify the problem, modeling of erup- differences of mass in the chamber between an
tive processes has traditionally been consid- initial overpressurized stage, associated with the
ered in the three spatial domains separately: beginning of the eruption, and a final equilib-
magma chamber, volcanic conduit, and the rium stage associated with the end of it (Blake,
Earth’s surface (Fig. 3.14). Magma chamber models 1981; Bower and Woods, 1997; Folch et al., 1998)
aim to describe the dynamics of magma with- (Fig. 3.15). According to these models, the mass
drawal over the course of a volcanic eruption. erupted depends critically on whether magma
Conduit models focus on processes occurring dur- is undersaturated or oversaturated in volatiles
ing the ascent of magma in the volcanic con- prior to the eruption. Undersaturated magma is
duit, including exsolution of volatiles, nucleation essentially incompressible, so that the extrusion
and growth of gas bubbles, and, in the case of of small fractions (between 0.1% and 1%) of the
explosive volcanism, magma fragmentation. Con- chamber contents, is sufficient to relieve the over-
duit models also provide information on physi- pressure, bringing the chamber to the lithostatic
cal conditions at the vent. Finally, subaerial models equilibrium and ending the eruption. The small
refer to the characteristics, dynamics, and sub- expansion coefficient of the resident magma and
sequent emplacement of the erupted materials contraction of the chamber walls in response
generally assuming given values for the physical to the pressure decrease therefore restricts the
THEORETICAL MODELING IN VOLCANOLOGY 109

110 importance in constraining the inflow conditions


of conduit models, described in the following sec-
100 Lithostatic tion, but have received surprisingly little atten-
pressure
tion. The first approaches assumed magma to
90 be an incompressible fluid of constant viscosity,
Pressure at the top of
the chamber (MPa)

and considered only simple chamber geometries


80 (Spera, 1984). Despite these oversimplifications,
70 these models allow the study of the evacua-
tion process and how it depends on the aspect
60 ratio of the magmatic reservoir, by considering
evacuation isochrons. An evacuation isochron is
50 wt = 5% wt = 6%
a locus of points within the magma chamber
wt = 3.5% wt = 4%
marking parcels of magma which will arrive at
40 the conduit entrance at the same time. Succes-
0 5 10 15 20 25 sive models have included compositional gradi-
Erupted mass fraction (in %) ents within the chamber (Spera et al., 1986; Trial
et al., 1992), and when combined with geologi-
Fig. 3.15. Theoretically calculated variations in magma cal and geochemical studies of volcanic deposits
pressure at the chamber roof during a Plinian eruptions as a can lead to the reconstruction of the pattern of
function of erupted mass fraction for different water pre-eruptive compositional zonation within the
contents. The magma chamber is located at 4 km below the chamber.
Earth surface, equivalent to a lithostatic pressure of 94 Mpa,
The assumption of magmatic incompressibil-
and the overpressure necessary to trigger the eruption is
ity has restricted the applicability of numeri-
assumed to be 10 Mpa. The diagram shows the decrease of
magma chamber pressure during the eruption. Note that the cal models to eruptions triggered by the injec-
pressure inside the chamber can fall well below the lithostatic, tion of fresh magma into the chamber. Recent
eventually causing the collapse of the chamber roof if required approaches have tried to overcome this by consid-
mechanical conditions are also met. After Martı́ et al. (2000). ering some parts of the chamber filled with vesic-
ulated, i.e., compressible, magma. Such models
have allowed simulations of eruptions driven
amount of material that can be extruded from by volatile oversaturation (Folch et al., 1998).
the chamber under these conditions. However, Results from these new mathematical simula-
if magma is initially volatile oversaturated and tions allow one to deduce the temporal evolution
contains gas bubbles, the magmatic mixture has of the most relevant physical parameters, includ-
a much higher compressibility, and the amount ing pressure, position of the exsolution level,
of magma that can be extruded from the cham- velocity field, eruption rate, and the amount
ber is much greater (between 1% and 10%). Since of erupted material, during a Plinian eruption
such models predict the mass fraction of erupted from a shallow, voltile-rich, magma chamber
magma, they can be used a posteriori to con- (Fig. 3.16).
strain chamber size once the volume of mater-
ial produced by a particular eruptive event is Conduit models
known. During the course of a volcanic eruption, the
A second group of chamber models looks at ascent of magma up the volcanic conduit and its
temporal variations of the physical parameters interaction with the conduit walls exert a strong
and the dynamics of the magma withdrawal pro- influence. For this reason, theoretical conduit
cess. The introduction of time-dependence neces- models are important in anticipating the proper-
sitates the use of numerical techniques to solve ties and the potential hazards of the eruption. For
the governing equations. The results and predic- example, depending on its composition and con-
tions of these models for the dynamics of magma tent of volatiles, ascending magma can fragment
withdrawal from crustal reservoirs are of crucial explosively in the conduit. If these parameters
110 ANTICIPATING VOLCANIC ERUPTIONS

Fig. 3.16. Example of a numerical calculation to determine or gas volume fraction inside the volcanic
the velocity field for magma inside a chamber of known conduit. Ideally, magma ascent along conduits
geometry and volume, after 22 hours of simulated Plinian should be modeled by solving a complex three-
eruption. The model allows prediction of the movement of a dimensional, unsteady (time-dependent), three-
particular parcel of magma inside the chamber at any instant
phase (gas–liquid mixture plus solid crystals), and
of time. This has implications for the interpretation of
eruption products and helps in the understanding of the
non-homogeneous flow problem. Indeed, other
dynamics of such eruptions. Simplified from Folch et al. effects such as crack formation and propagation,
(1998). erosion of conduit walls, and the existence of a
multicomponent gas phase should also be taken
into account. Clearly, however, the complete solu-
could be known a priori, conduit models could tion of such problem is difficult while knowl-
allow prediction of when and where magma will edge of the variables involved and their inter-
fragment, and what will be the destructive poten- relations are only partially known. Therefore,
tial of the eruption. several simplifications including the neglect of
Conduit models involve the numerical solu- certain terms of the governing equations, or
tion of the equations describing the mechanics the use of a simplified state law and constitu-
of fluids. These include the continuity, momen- tive equations, have typically been made. The
tum and thermal equations, derived respectively great majority of conduit models neglect time-
from the general principles of mass, momentum, dependent terms of the governing equations and
and energy conservation. In addition, a state law restrict themselves to steady conditions. Steady
and several constitutive relationships character- models do not consider coupling with either
izing the physical and rheological properties of chamber dynamics or subaerial processes and
the ascending magma require definition. The aim are best applied to phases of an eruption dur-
of the models is to find the values of variables ing which no significant variation of flow param-
such as pressure, density, velocity, temperature, eters occurs. The different components (i.e., gas
THEORETICAL MODELING IN VOLCANOLOGY 111

bubbles and liquid) of a magmatic mixture


have different dynamic behavior and require
separate solutions of the governing equations.
Homogeneous models simplify the simulation
by assuming thermal and mechanical equilib-
rium between liquid and gas phases, while
non-homogeneous models take such mechani-
cal disequilibrium into account. The homoge-
neous approach is justified if magma viscosity
is high enough to prevent relative movement
between gas bubbles and liquid, and is a gen-
erally accepted approach for chemically evolved
magmas in the bubbly flow regime. However, the
assumption of homogeneity is not justified above
the fragmentation level where the viscosity of
the mixture is much lower, allowing for relative
motion between liquid and gas phases.
The theoretical basis for conduit models Fig. 3.17. General model relating vent radius, magmatic gas
comes from the important contribution of Sparks content, eruption rate, and gas velocity to convecting and
(1978), who introduced a series of fundamen- collapsing states of eruption columns. After Wilson et al.
tal concepts. These include the concepts of sat- (1980).
uration, nucleation, and exsolution levels, the
growth of gas bubbles by diffusion and decom-
pression during magma ascent, and the con-
dition of magma fragmentation. These ideas of the ascending magma in terms of composition,
became the basis for a first generation of com- water content, etc.
prehensive conduit models which, despite their A more recent and realistic generation of
simplicity, gave insights into magma ascent conduit models introduces non-homogeneity
(Wilson et al., 1980; Wilson and Head, 1981; (Dobran, 1987, 1992). In this approach separate
Gilberti and Wilson, 1990). These early models are equations are developed for the liquid and gas
one-dimensional, steady, and homogeneous, and phases describing their mechanical relationships.
ignore thermal effects. Their numerical solution Non-homogeneous models give results that can
gives a first approach to understanding the roles differ appreciably from homogeneous types, and
of magma density, velocity, pressure, and conduit results depend greatly on the choice of free
radius, and permits effects of volatile content, parameters. A detailed discussion between the
magma viscosity, and conduit shape on the out- differences between these types of conduit mod-
flow conditions. Since vent conditions are closely els is given by Papale (1996).
related to eruptive style (Sparks and Wilson, 1976; Contemporaneous non-homogeneous con-
Wilson 1976), these models were able to explain duit models consider conduit erosion, through
different explosive regimes in terms of the prop- motion of particles against the conduit walls
erties of the ascending magma. For example, a (Macedonio et al., 1994), and incorporate other
decrease in magma volatile content associated refinements mainly concerning magma proper-
with the extrusion of deeper layers of a stratified ties (Papale and Dobran, 1993, 1994; Papale et al.,
magma chamber, or an increase in vent radius 1998). The recent versions (Papale et al., 1998)
due to conduit erosion, produce lower exit veloc- are the sharpest tools available for predicting
ities, explaining the collapse of Plinian columns flow in volcanic conduits and assessing the roles
and the formation of pyroclastic flows (Fig. 3.17). of magmatic composition on the dynamics of
From a hazard perspective eruptive style can be eruptions. These new conduit models represent
predicted providing knowledge of the properties a significant advance on the development of
112 ANTICIPATING VOLCANIC ERUPTIONS

numerical simulations aimed to predict eruptive consequences of these types of volcanic pro-
behavior. cesses. Theoretical results can be input into a geo-
graphical information system (GIS) package devel-
oped specifically to predict the effects of future
Subaerial models eruptions, allowing areas likely to be affected
Modeling of subaerial volcanic processes has been by a particular volcanic hazard to be assessed
one of the most active areas of volcanology over (Gomez-Fernandez, 1998; Gomez-Fernandez and
the last two decades, owing to its great impor- Macedonio, 1998) (Fig. 3.18).
tance for the assessment of volcanic hazards.
Reviews of the subject are given by Neri and Mace-
donio, (1996), Sparks et al. (1997), Freund and Volcano monitoring
Rosi (1998), and Gilbert and Sparks (1998), who
give detailed information on this group of the- This chapter has discussed the importance of
oretical models. In the context of explosive vol- understanding volcanic phenomena in order to
canism, such models solve two-phase flow prob- accurately predict the occurrence and effects
lems involving gas plus ash particles, for jets, of future eruptions. How volcanic processes are
plumes, and gravity currents to find the dynam- studied in the field and laboratory, and mod-
ics of the transported volcanic products. Efforts eled by experiments and theory have also been
have focused mainly on modeling the dynamics covered. However, as mentioned in the introduc-
of eruptive columns including their collapse to tion, this comprises only part of the information
produce pyroclastic flows, and the dispersal of needed to correctly anticipate volcanic eruptions.
pyroclastic material into the atmosphere. Given Information is also required about the present
certain physical conditions at the vent, the mod- state of a volcano if any changes in its behav-
els aim to assess pyroclastic transport processes ior are to be noted. Changes such as deforma-
and related volcanic hazards. tion of the volcano, seismicity, temperature varia-
The results of this group of theoretical mod- tion, changes in gas composition, etc. will provide
els are related directly to the short- and long- information about hidden movements of magma
term predictions of the effects of volcanic erup- and related fluids, and a correct interpretation
tions. For example, the long-term environmental will most likely come from all the precursory phe-
impact of explosive eruptions is related mainly to nomena together.
climate change, such as temperature decreases, Experience gained from reconstructing his-
which is of global scale. Climate changes torical volcanic eruptions has shown that large
have occurred following the recent eruption of volcanoes may remain inactive for hundreds to
Mt. Pinatubo in 1991; and larger eruptions, thousands of years before erupting once more.
such as that of Toba on the island of Sumatra Reactivation, however, may occur over a very
in Indonesia about 74 000 years ago, or more short period of a few months to a few years,
recently that of Tambora Volcano, also in Indone- showing the importance of a well-equipped vol-
sia, in 1815, are believed to have produced much cano observatory to alert attention promptly to
more dramatic variations (Rampino et al., 1988). signs of renewed activity. Several different types
Such climate effects occur due to the injection of of monitoring equipment are required to detect
volcanic aerosols and ash particles into the high the various precursory signals of a forthcoming
atmosphere, carried by the eruption column and eruption. Traditionally, volcano monitoring has
associated ash clouds (see Chapter 5). Models of focused on ground-based geophysical and geoche-
subaerial processes help to predict the transport mical techniques. However, technological im-
of volcanic particles in the atmosphere, help- provements over the last 20 years have brought
ing to anticipate potential climate effects of cur- the addition of sophisticated remote-sensing tech-
rent and future eruptions. The development of niques based on radar or satellite systems to vol-
theoretical models of pyroclastic flows emplace- cano monitoring. Monitoring is one of the main
ment and the movement of lava flows and aspects of volcanology and has received conti-
lahars are also important given the short-term nuous attention from the scientific community.
VOLCANO MONITORING 113

Fig. 3.18. Example of volcanic hazard scenario obtained are ground deformation, seismic activity, temper-
from a computer-based geographical information system ature variations, gravity and stress-field changes,
(GIS) tool, specifically aimed to volcanic risk assessment. This and electromagnetic effects.
example shows the simulation of a lava eruption from the Deformation of the ground surface is one
Dorsal Ridge volcano in Tenerife, Canary Islands. It allows
of the main precursory signals in active vol-
zones to be defined according to their potential hazard
(Gomez-Fernandez, 1998; Gomez-Fernandez and Macedonio,
canic areas. Ground deformation is commonly
1998). attributed to an increase in magma chamber
pressure (Mogi, 1958) or the pressurization of
a shallow aquifer by a magmatic heat source
Comprehensive reviews containing detailed ex- (Bonafede, 1991). Whichever the case, a pres-
planations of the techniques of volcano mon- sure increase at depth will cause the surround-
itoring have recently been given by McGuire ing media to deform, which may be measured
et al. (1995), Scarpa and Tilling (1996), Sigurdsson at the surface in terms of vertical and/or hori-
(2000), and Sparks (2003) to which the readers zontal displacements of previously surveyed ref-
are referred for more information. Early detec- erence points. Since pressurization is a major
tion and tracking of volcanic pleemes by radar potential eruptive trigger, monitoring and inter-
or satellite remote sensing is not discussed here preting surface deformation patterns is a key
(e.g., see Rose et al., 2000; Lacasse et al., 2004). aspect in volcanic eruption forecasting, and has
been an active research topic over the last few
Geophysical and geodetic monitoring decades (Van Der Laat, 1996; Dvorak and Dzurisin,
Geophysical and geodetic monitoring techniques 1997 and references therein). Ground deforma-
include ground-based and remote methods tion has traditionally been measured by precise
designed to allow changes in the main physi- optical leveling, using electronic distance mea-
cal properties of the volcano and its surround- surement (EDM) equipment, and tiltmeters, and
ings to be detected. The main physical signals significant progress is now being made with the
related to mass movement in the interior of use of satellite based systems. Global positioning
the volcano and that can be observed at surface system (GPS) instruments, using a group of more
114 ANTICIPATING VOLCANIC ERUPTIONS

Fig. 3.19. Seismic monitoring and


ground deformation are the main
methods used in volcano
surveillance. Magma movement
inside the volcano may eventually
cause volcanic activity, generally
accompanied by an increases in
seismic activity, surface deformation,
changes in gravity and magnetic
fields, temperature variations, and
changes in the composition of
escaping gases. Volcano
observatories, using appropriate
techniques, record these precursory
signals, allowing future eruptions to
be anticipated.

than 20 fixed for reference satellites, allow topo- combined with measurements from stress meters
graphic coordinates of latitude, longitude, and are occasionally used to determine changes in
elevation to be measured at any point of the the accompanying stress-field associated with the
Earth’s surface to a precision of a few centime- seismic events, and provide a method to distin-
ters, irrespectively of topography. Compared to guish volcano-tectonic events from other purely
traditional methods GPS allows the monitoring tectonic events unrelated to volcanism.
of larger areas and is more efficient and easy Gravimetric and magnetic methods of vol-
to use. Another satellite-based technique being cano monitoring are also used. The Earth’s grav-
increasingly used in ground deformation moni- itational and magnetic fields can be disturbed
toring is synthetic aperture radar (SAR) interfer- by volcanic activity, owing to the movement of
ometry. This was introduced in 1974 to generate magma beneath the surface. Gravity measure-
topographic maps, and has since been used to ments are usually carried out in combination
detect changes in the Earth’s surface. Differences with observations of ground deformation. This
are detected by superimposing radar images of is in order to distinguish small variations in
large areas, taken at different times. The cur- the gravity field due to subsurface mass or den-
rent maximum frequency of sampling is monthly. sity changes from elevation changes caused by
Observable ground deformation is of the order of magma intrusion or withdrawal (McGuire, 1995).
a few centimeters. Displacements of fresh magma inside the vol-
Seismic activity related to earthquakes or cano can also cause small variations of the inten-
tremors is one of the main indicators of vol- sity of the magnetic field, which can be iden-
canic unrest and seismic monitoring is one of the tified as precursory signals of eruptive activity
main tools of geophysical volcano surveillance (Zlotnicki, 1995). Finally, geoelectrical methods,
(McGuire et al., 1995; Scarpa and Tilling, 1996) including mainly the development of resistivity
(Fig. 3.19). Pressure increases and movements maps of the volcano’s interior, allow any variation
of magma and related fluids inside the vol- attributable to magma intrusion to be detected
cano are normally associated with the open- (Lenat, 1995).
ing of fractures, which causes seismic signals
usually recorded by short-period seismometers.
Data interpretation allows determination of the Geochemical methods
energy and location of earthquake hypocenters When combined with geophysical methods, the
and their evolution with time. Seismic data analysis of volcanic gases constitutes a powerful
FINAL REMARKS 115

tool for volcano monitoring. The intrusion of short-lived nuclides released in gases is becom-
fresh magma into the volcanic system, and the ing one of the effective methods of continuous
ascent of magma toward the Earth’s surface, geochemical monitoring at active volcanoes.
can cause profound changes in the composi-
tion of gases escaping as fumaroles or diffu-
sively from the volcano. Continuous monitor- Final remarks
ing equipment is increasingly being installed at
active volcanoes in order to detect any change This chapter has briefly summarized the main
in state by measuring the concentration of dif- methods used in contemporaneous volcanology
ferent gas species. The gases (H2 O, CO2 , Cl, S, to understand how volcanoes work, and to mon-
SO2 , etc.) are released in significant amounts itor their current activity. Previous sections have
through near-vent, high-temperature fumaroles covered the causes and main features of the erup-
and through newly opened fractures, when fresh tion process, and have described how field and
magma reaches shallow depths. These geochem- laboratory studies combined with experimental
ical indicators, occurring with other precur- and theoretical modeling constitute the main
sory signals such as thermal anomalies, ground tools for discovering a particular volcano’s char-
deformation, and seismic activity, would confirm acteristics and likely future behavior. We have
the proximity of a forthcoming eruption. Dif- emphasized mainly the application of mathemat-
fuse degassing, through subsurface fractures, of ical modeling to the study of volcano dynamics,
species such as CO2 , helium, and radon, which as this is an emerging and powerful way of antic-
can be measured in soils or springs, provides ipating volcano behavior. However, all theoretical
information about the overall permeability of the models need validation by observation, which is
volcanic edifice, the potential for degassing from provided by field, laboratory, and experimental
areas other than the active crater, and the abil- data. The examples of techniques included here
ity of a volcano to release large quantities of vol- comprise only a brief survey of the wide spec-
canic gases diffusively (Stix and Gaonac’H, 2000). trum available, but cover the main problems of
Direct measurement of volcanic gas compositions physical volcanology.
is difficult work, owing to the obvious logistical A good knowledge of volcanic processes is
difficulties and risk involved, and may be comple- required to ensure the correct interpretation of
mented by remote sensing of the gas plume. This eruptive precursors. Geophysical and geochemi-
has the advantage of distance from the volcano, cal monitoring of volcanoes is able to detect pre-
and the greater sampling frequencies that may cursory signals of eruptive activity. However, vol-
be obtained than when sampling volcanic gases cano forecasting is required not only to detect
manually. whether a particular volcano has become active,
Measurements of radionuclides released from but also whether renewed activity will culmi-
volcanoes provide constraints on the rate of deep nate in an eruption, and how the eruption will
magma input to shallow depth and the rate proceed. Mitigation of volcanic hazards will be
of magma degassing. An increase in the first effective only if monitoring systems are available
parameter provides evidence of increasing activ- working currently and if the effects of a forth-
ity. Knowledge of the second parameter helps coming eruption can be anticipated correctly.
predict the volume of magma expelled during Clearly, volcanic activity cannot be stopped but
eruptions, which is of particular interest for we can avoid its effects and reduce its poten-
studies of explosive activity. Besides providing a tial damage if we can predict it well in advance.
geochemical tracer of shallow magma evolution Physical volcanology and volcano monitoring
through crystallization and degassing, radionu- need, therefore, to work together. Although there
clide studies offer a unique opportunity to is still a long way to go, present knowledge of vol-
estimate the timescales of magmatic processes, canic processes is adequate for the task of antic-
essential for any attempt to predict its future evo- ipating future activity. Modern GIS have consid-
lution and associated volcanic risk. Analyses of erably improved the management and mitigation
116 ANTICIPATING VOLCANIC ERUPTIONS

of volcanic risks, particularly when a good knowl- Carmichael, I. S. E., Turner, F. J., and Verhoogen, J.
edge of the volcano and its past activity has been 1974. Igneous Petrology. New York, McGraw-Hill.
compiled. Integration of data from real-time mon- Carroll, M. C. and Holloway, J. R. (eds.) 1994. Volatiles in
itoring and the results of numerical modeling Magmas, Reviews in Mineralogy no. 30.
into GIS frameworks is one of modern volcanol- Washington, DC, Mineralogical Society of America.
ogy’s major challenges for the near future. Cas, R. A. F. and Wright, J. V. 1987. Volcanic Successions
Modern and Ancient. London, Allen and Unwin.
Condomines, M., Hémond, C. H., and Allègre, C. J.
1988. U–Th–Ra radioactive disequilibria and
References magmatic processes. Earth and Planetary Science
Letters, 90, 243–262.
Alidibirov, M. and Dingwell, D. B. 1996. Magma Condomines, M., Tanguy, J. C., and Michaud, V. 1995.
fragmentation by rapid decompression. Nature, Magma dynamics at Mt. Etna: constraints from
380, 146–148. U–Th–Ra–Pb radioactive disequilibria and Sr
Bagdassarov, N., Dingwell, D., and Wilding, M. 1996. isotopes in historical lavas. Earth and Planetary
Rhyolite magma degassing: an experimental Science Letters, 132, 25–41.
study of melt vesiculation. Bulletin of Volcanology, Dingwell, D. 1998. Magma degassing and
57, 587–601. fragmentation. In A. Freund and M. Resi (eds.)
Blake, S. 1981. Volcanism and the dynamics of open Modelling Physical Processes of Exposive Volcanic
magma chambers. Nature, 289, 783–785. Eruptions. Amsterdam, The Netherlands, Elsevier,
1984. Volatile oversaturation during the evolution pp. 1–25.
of silicic magma chambers as an eruption trigger. Dobran, F. 1987. Nonequilibrium modeling of
Journal of Geophysical Research, 89, 8237–8244. two-phase critical flows in tubes. Journal of Heat
Blake, S. and Fink, J. H. 1987. The dynamics of magma Transfer, 109, 731–738.
withdrawal from a density stratified dyke. Earth 1992. Nonequilibrium flow in volcanic conduits and
and Planetary Sciences Letters, 85, 516–524. application to the eruptions of Mt. St. Helens on
Bonadonna, C., Ernst, G. G. J., and Sparks, R. S. J. 1998. May 18, 1980 and Vesuvius in AD 79. Journal of
Thickness variations and volume estimates of Volcanology and Geothermal Research, 49, 285–311.
tephra fall deposits: the importance of particle Dobran, F., Neri, A., and Macedonio, G. 1993.
Reynolds number. Journal of Volcanology and Numerical simulation of collapsing volcanic
Geothermal Research, 81, 173–187. columns. Journal of Geophysical Research, 98,
Bonafede, M. 1991. Hot fluid migration: an efficient 4231–4259.
source of ground deformation: application to the Donadieu, F. and Merle, O. 1998. Experiments on the
1982–1984 crisis at Plegrean Fields, Italy. Journal of indentation process during cryptodome
Volcanology and Geothermal Research, 48, intrusions: new insights into Mount St. Helens
187–198. deformation. Geology, 26, 79–82.
Bower, S. M. and Woods, A. W. 1997. Control of magma Dvorak, J. J. and Dzurisin, D. 1997. Volcano geodesy:
volatile content and chamber depth on the mass the search for magma reservoirs and the
erupted during explosive volcanic eruptions. formation of eruptive vents. Reviews of Geophysics,
Journal of Geophysical Research, 102, 10273–10290. 35, 343–384.
Bryan, S. E., Cas, R. A. F., and Martı́, J. 2000. The 0.57 Ernst, G. G. J., Davis, J. P., and Sparks, R. S. J. 1994.
Ma plinian eruption of the Granadilla Member, Bifurcation of volcanic plumes in a crosswind.
Tenerife (Canary Islands): an example of Bulletin of Volcanology, 56, 159–169.
complexity in eruption dynamics and evolution. Ernst, G. G. J., Sparks, R. S. S., Carey, S. N., et al. 1996.
Journal of Volcanology and Geothermal Research, 103, Sedimentation from turbulent jets and plumes.
209–238. Journal of Geophysical Research, 101, 5575–5590.
Campbell, I. H. and Turner, J. S. 1986. The influence of Fisher, R. V. and Schmincke, H. U. 1984. Pyroclastic
viscosity on fountains in magma chambers. Rocks. Berlin, Germany, Springer-Verlag.
Journal of Petrology, 27, 1–30. Folch, A. and Martı́, J. 1998. The generation of
Carey, S. N., Sigurdsson, H., and Sparks, R. S. J. overpressure in felsic magma chambers by
1988. Experimental studies of particle-laden replenishment. Earth and Planetary Science Letters,
plumes. Journal of Geophysical Research, 93, 314–328. 163, 301–314.
REFERENCES 117

Folch, A., Martı́, J., Codina, R., et al. 1998. A numerical magma. Contribitions to Mineralogy and Petrology,
model for temporal variations during explosive 75, 279–289.
central vent eruptions. Journal of Geophysical 1988. The fluid dynamics of crustal melting by
Research, 103, 20883–20899. injection of basaltic sills. Transactions of the Royal
Freund, A. and Rosi, M. (eds.) 1998. Developments in Society of Edinburgh, 79, 237–243.
Volcanology, vol. 4, From Magma to Tephra. Huppert, H. E., Turner, J. S., and Sparks, R. S. J. 1982a.
Amsterdam, The Netherlands, Elsevier. Replenished magma chambers: effects of
Freund, A. and Tait, S. R. 1986. The entrainment of compositional zonation and input rates. Earth and
high-viscosity magma into low-viscosity magma in Planetary Science Letters, 57, 345–357.
eruption conduits. Bulletin of Volcanology, 48, 1982b. Effects of volatiles on mixing in calc-alkaline
325–339. magma systems. Nature, 297, 554–557.
Gilbert, G. J. and Sparks, R. S. J. (eds.) 1998. The Physics 1983. Laboratory investigations of viscous effects in
of Explosive Volcanic Eruptions, Geological Society replenished magma chambers. Earth and Planetary
Special Publication no. 145. London, The Science Letters, 65, 377–381.
Geological Society. Hurwitz, S. and Navon, O. 1994. Bubble nucleation in
Gilberti, G. and Wilson, L. 1990. The influence of rhyolitic melts: experiments at high pressure,
geometry on the ascent of magma in open temperature and water content. Earth and
fissures. Bulletin of Volcanology, 52, 515–521. Planetary Science Letters, 122, 267–280.
Gomez-Fernandez, F., 1998. Development of a volcanic Komuro, H., Fujita, Y., and Kodama, K. 1984.
risk assessment information system for the Numerical and experimental models on the
prevention and management of volcanic crisis: formation mechanism of collapse basins during
stating the fundamentals. In C. A. Brebbia and P. the Green Tuff orogenesis of Japan. Bulletin of
Pascolo (eds.) GIS Technologies and their Volcanology, 47, 649–666.
Environmental Applications. Computational Lacasse, S., Karlsdóttir, S., Larsen, G., et al. 2004.
Mechanics Publications, pp. 111–120. Weather radar observations of the Hekla 2000
Gomez-Fernandez, F. and Macedonio, G. 1998. eruption cloud, Iceland. Bulletin of Volcanology, 66,
Integration of physical simulation models in the 457–473.
frame of a GIS for the development of a volcanic Lambert, G., Le Cloarec, M. F., Ardouin, B., et al. 1986.
assessment information system. In C. A. Brebbia, Volcanic emission of radionuclides and magma
J. L. Rubio, and J. L. Uso (eds.) Risk Analysis. dynamics. Earth and Planetary Science Letters, 76,
Computational Mechanics Publications, 185–192.
pp. 265–274. Le Cloarec, M. F. and Marty, B. 1991. Volcanic fluxes
Gudmundsson, A. 1988. Effect of tensile stress from volcanoes. Terra Nova, 3, 17–27.
concentration around magma chambers on Le Cloarec, M. F., Allard, P., Ardouin, B., et al. 1992.
intrusion and extrusion frequencies. Journal of Radioactive isotopes and trace elements in
Volcanology and Geothermal Research, 35, 179–194. gaseous emissions from White Island, New
1998. Magma chambers modelled as cavities explain Zeland. Earth and Planetary Science Letters, 108, 19–28.
the formation of rift zone central volcanoes and Lenat, J. F. 1995. Geoelectrical methods in volcano
the eruption and intrusion statistics. Journal of monitoring. In W. J. McGuire, C. Kilburn and J.
Geophysical Research, 103, 7401–7412 Murray (eds.) Monitoring Active Volcanoes: Strategies,
Gudmundsson, A., Martı́, J., and Turon, E. 1997. Stress Procedures and Techniques. London, UCL Press,
fields generating ring faults in volcanoes. pp. 248–274.
Geophysical Research Letters, 24, 1559–1562. Linde, A. T. and Sacks, I. S. 1998. Triggering of
Gudmundsson, A., Marinoni, L., and Martı́, J. 1999. volcanic eruptions. Nature, 395, 888–890.
Dyke injection and arrest: implications for Lyakhovsky, V., Hurwitz, S., and Navon, O. 1996.
volcanic hazards. Journal of Volcanology and Bubble growth in rhyolitic melts, experimental
Geothermal Research, 88, 1–13. and numerical investigation. Bulletin of
Hall, A. 1996. Igneous Petrology. Chichester, UK, John Volcanology, 58, 19–32.
Wiley. Macedonio, G., Dobran, F., and Neri, A. 1994. Erosion
Huppert, H. E. and Sparks, R. S. J. 1980. The fluid processes in volcanic conduits and application to
dynamics of basaltic magma chambers the AD 79 eruption of Vesuvius. Earth and
replenished by influx of hot, dense ultrabasic Planetary Science Letters, 121, 137–152.
118 ANTICIPATING VOLCANIC ERUPTIONS

Mader, H. M. 1998. Conduit flow and fragmentation. Neri, A. and Macedonio, G. 1996. Physical modelling
In J. S. Gilbert and R. S. J. Sparks (eds.) The physics of collapsing volcanic columns and pyroclastic
of Explosive Volcanic Eruptions, Geological Society flows. In R. Scarpa and R. I. Tilling (eds.)
Special Publication no. 145. London, The Monitoring and Mitigation of Volcano Hazards.
Geological Society, pp. 51–71. Berlin, Germany, Springer-Verlag, pp. 389–427.
Mader, H. M., Phillips, J. C., and Sparks, R. S. J. 1996. Neri, A., Papale, P., and Macedonio, G. 1998. The role
Dynamics of explosive degassing of magma: of magma composition and water content in
observations of fragmenting two phase flows. explosive eruptions. II. Pyroclastic dispersion
Journal of Geophysical Research, 101, dynamics. Journal of Volcanology and Geothermal
5547–5560. Research, 87, 95–115.
Mader, H. M., Zhang, Y., Phillips, J., et al. 1994. Newhall, C. G. and Punongbayan, R. S. 1996. The
Experimental simulations of explosive degassing narrow margin of successful volcanic-risk
of magma. Nature, 372, 85–88. mitigation. In R. Scarpa and R. I. Tilling (eds.)
Marsh, B. D. 1989. Magma chambers. Annual Reviews of Monitoring and Mitigation of Volcano Hazards. Berlin,
Earth and Planetary Sciences, 17, 439–474. Germany, Springer-Verlag, pp. 807–838.
Martı́, J. 1993. Paleovolcanismo. In J. Martı́ and V. Pallister, J. S., Hoblitt, R. P., and Reyes, A. G. 1992. A
Araña (eds.) La Vulcanolog´ ıa Actual: Nuevas basalt trigger for the 1991 eruptions of Pinatubo
Tendencias. Madrid, Spain, Consejc Superier de volcano. Nature, 356, 426–428.
Investigaciones Cientificas, pp. 531–578. Papale, P. 1996. Modelling of magma ascent along
Martı́, J. and Araña, V. (eds.), 1993. La Vulcanolog´ ıa conduits: a review. In F. Barberi and R. Casale
Actual: Nuevas Tendencias. Madrid, Spain, CSIC. (eds.) The Mitigation of Volcanic Hazards.
Martı́, J., Ablay, G., Redshaw, L. T., et al. 1994. Luxembourg, Office for Official Publications of
Experimental studies of collapse calderas. Journal the European Communities, pp. 3–40.
of the Geological Society of London, 151, 919–929. 1998. Strain-induced magma fragmentation in
Martı́, J., Soriano, C., and Dingwell, D. 1999. Tube explosive eruptions. Nature, 397, 425–428.
pumices: strain markers of a ductile–brittle Papale, P. and Dobran, F. 1993. Modeling of the ascent
transition in explosive eruptions. Nature, 402, of magma during the plinian eruption of
650–653. Vesuvius in AD 79. Journal of Volcanology and
Martı́, J., Folch, A., Neri, A., et al. 2000. Pressure Geothermal Research, 58, 101–132.
evolution during explosive caldera-forming 1994. Magma flow along the volcanic conduit
eruptions. Earth and Planetary Science Letters, 175, during the Plinian and pyroclastic flow
275–287. phases of the May 18, 1980, Mount St. Helens
McBirney, A. R. 1984. Igneous Petrology. San Francisco, eruption. Journal of Geophysical Research, 99,
CA, Freeman, Cooper and Co. 4355–4374.
McGuire, W. J., 1995. Monitoring active volcanoes: an Papale, P., Neri, A., and Macedonio, G., 1998. The role
introduction. In W. J. McGuire, C. Kilburn, and J. of magma composition and water content in
Murray (eds.) Monitoring Active Volcanoes: Strategies, explosive eruptions. I. Conduit ascent dynamics.
Procedures and Techniques. London, UCL Press, Journal of Volcanology and Geothermal Research, 87,
pp. 1–31. 75–93.
McGuire, W. J., Kilburn, C., and Murray, J. (eds.) 1995. Pennisi, M. and Le Cloarec, M. F. 1998. Variations of
Monitoring Active Volcanoes: Strategies, Procedures and Cl, F, and S in Mount Etna’s plume, Italy, between
Techniques. London, UCL Press. 1992 and 1995. Journal of Geophysical Research, 103,
Merle, O. and Borgia, A. 1996. Scaled experiments of 5061–5066.
volcanic spreading. Journal of Geophysical Research, Prousevitch, A. A., Shagian, D. L., and Anderson, A. T.
101, 13805–13818. 1993. Dynamics of diffusive bubble growth in
Mogi, K. 1958. Relations of the eruptions of various magmas: isothermal case. Journal of Geophysical
volcanoes and the deformations of the ground Research, 98, 22283–22307.
surface around them. Bulletin of the Earthquake Rampino, M. R., Self, S., and Stothers, R. B. 1988.
Research Institute, 36, 99–134. Volcanic winters. Annual Reviews of Earth and
Morey, G. W. 1922. The development of pressure in Planetary Sciences, 16, 73–99.
magmas as a result of crystallization. Journal of Riedel, C., Ernst, G. G. J., and Riley, M. 2003. Controls
the Washington Academy of Science, 12, 219–230. on the growth and geometry of pyroclastic
REFERENCES 119

constructs. Journal of Volcanology and Geothermal collapse. Journal of the Geological Society of London,
Research, 127, 121–152 132, 441–451.
Roberts, J. L. 1970. The intrusion of magma into Sparks, R. S. J., Sigurdsson, H., and Wilson, L. 1977.
brittle rocks. In G. Newall and N. Rast (eds.) Magma mixing: a mechanism for triggering acid
Mechanism of Igneous Intrusion. Liverpool, UK, explosive eruptions. Nature, 267, 315–318.
Gallery Press, pp. 287–338. Sparks, R. S. J., Carey, S. N., and Sigurdsson, H., 1991.
Roche, O., Druitt, T. M., and Merle, O. 2000. Sedimentation from gravity currents
Experimental study of caldera formation. Journal generated by turbulent plumes. Sedimentology, 38,
of Geophysical Research, 105, 395–416. 839–856.
Rose, W. I., Bluth, G. J. S., and Ernst, G. G. J. 2000. Sparks, R. S. J., Bursik, M. I., Carey, S. N., et al. 1997.
Integrating retrievals of volcanic cloud Volcanic Plumes. New York, John Wiley.
characteristics from satellite remote sensors: Spera, F. 1984. Some numerical experiments
a summary. Philosophical Transactions of the on the withdrawal of magma from crustal
Royal Society of London, Series A, 358, reservoirs. Journal of Geophysical Research, 89,
1585–1606. 8222–8236.
Rosi, M. 1996. Quantitative reconstruction of Spera, F., Yuen, D. A., Greer, J., et al. 1986. Dynamics of
recent volcanic activity: a contribution to magma withdrawal from stratified magma
forecasting future eruptions. In R. Scarpa and R. I. chambers. Geology, 14, 723–726.
Tilling (eds.) Monitoring and Mitigation of Volcano Stix, J. and Gaonac’H, H. 2000. Gas, plume and
Hazards. Berlin, Germany, Springer-Verlag, thermal monitoring. In H. Sigurdsson (ed.)
pp. 631–674. Encyclopedia of Volcanoes. San Diego, CA, Academic
Rubin, A. 1995. Propagation of magma filled cracks. Press, pp. 1141–1163.
Annual Reviews of Earth and Planetary Sciences, 23, Sturtevant, B., Kanamori, B., and Brodsky, E. E. 1996.
287–336. Seismic triggering by rectified diffusion in
Scarpa, R. and Tilling, R. I. (eds.) 1996. Monitoring and geothermal systems. Journal of Geophysical Research,
Mitigation of Volcano Hazards. Berlin, Germany, 101, 25269–25282.
Springer-Verlag. Tait, S., Jaupart, C., and Vergniolle, S. 1989. Pressure,
Sigurdsson, H. (ed.) 2000. Encyclopedia of Volcanoes. San gas content and eruption periodicity of a shallow
Diego, CA, Academic Press. crystallizing magma chamber. Earth and Planetary
Smith, R. L. 1979. Ash flow magmatism. Geological Science Letters, 92, 107–123.
Society of America Special Paper, 180, 5–27. Toramaru, A. 1995. Numerical study of nucleation and
Smith, R. L. and Bailey, R. A. 1968. Resurgent growth of bubbles in viscous magmas. Journal of
caulderons. Geological Society of American Memoirs, Geophysical Research, 100, 1913–1931.
116, 613–622. Trial, A. F., Spera, F., Greer, J., et al. 1992. Simulations
Snyder, D. and Tait, S. 1995. Replenishment of magma of magma withdrawal from compositionally
chambers: comparison of fluid-mechanic zoned bodies. Journal of Geophysical Research, 97,
experiments with field relations. Contributions to 6713–6733.
Mineralogy and Petrology, 122, 230–240. Turner, J. S. and Campbell, I. H. 1986. Convection and
1996. Magma mixing by convective entrainment. mixing in magma chambers. Earth Sciences Reviews,
Nature, 379, 529–531. 23, 255–352.
Sparks, R. S. J. 1978. The dynamics of bubble Van Der Laat, R. 1996. Ground-deformation
formation and growth in magmas, a review and methods and results. In R. Scarpa and R. I.
analysis. Journal of Volcanology and Geothermal Tilling (eds.) Monitoring and Mitigation of Volcano
Research, 3, 1–37. Hazards. Berlin, Germany, Springer-Verlag,
2003. Forecasting volcanic eruptions. Earth and pp. 147–168.
Planetary Science Letters, 210, 1–15. Walker, G. P. L. 1973. Explosive volcanic eruptions: a
Sparks, R. S. J. and Marshall, L. A. 1986. Thermal and new classification scheme. Geologisctic Rundschall,
mechanical constraints on mixing between mafic 62, 431–446.
and silicic magmas. Journal of Volcanology and Willams, H. and McBirney, A. R. 1979. Volcanology. San
Geothermal Research, 29, 99–124. Francisco, CA, W. H. Freeman.
Sparks, R. S. J. and Wilson, L. 1976. A model for the Wilson, C. J. N. 1980. The role of fluidization in the
formation of ignimbrite by gravitational column emplacement of pyroclastic flows: an
120 ANTICIPATING VOLCANIC ERUPTIONS

experimental approach. Journal of Volcanology and Wohletz, K. H. and Sheridan, M. F. 1983.


Geothermal Research, 8, 231–249. Hydrovolcanic explosions II. Evolution of basaltic
1984. The role of fluidization in the emplacements tuff rings and tuff cones. American Journal of
of pyroclasic flows. II. Experimental results and Science, 283, 385–413.
their interpretation. Journal of Volcanology and Wood, W. J. and Fraser, D. G. 1976. Elementary
Geothermal Research, 20, 55–84. Thermodynamics for Geologists. Oxford, UK, Oxford
Wilson, L. 1976. Explosive volcanic eruptions. III. University Press.
Plinian eruption columns. Geophysical Woods, A. W. and Bursik, M. I. 1994. A laboratory
Journal of the Royal Astronomy Society, 45, study of ash flows. Journal of Geophysical Research,
543–556. 99, 4375–4394.
Wilson, L. and Head, J. W. 1981. Ascent and Woods, A. W., Bursik, M. I., and Kurvatov, A. V. 1998.
eruption of basaltic magma on the earth and The interaction of ash flows with ridges. Bulletin of
moon. Journal of Geophysical Research, 86, Volcanology, 60, 38–51.
2971–3001. Zimanowski, B. 1998. Phreatomagmatic explosions. In
Wilson, L., Sparks, R. S. J., and Walker, G. P. L. A. Freund and M. Rosi (eds.) Developments in
1980. Explosive volcanic eruptions. IV. Volcanology, vol. 4, From Magma to Tephra.
The control of magma properties and conduit Amsterdam, The Netherlands, Elsevier, pp. 25–53.
geometry on eruption column behaviour. Zimanowski, B., Büttner, R., Lorenz, V., et al. 1997.
Geophysical Journal of the Royal Astronomy Society, 63, Fragmentation of basaltic melt in the course of
117–148. explosive volcanism. Journal of Geophysical Research,
Wilson, M. 1989. Igneous Petrogenesis: A Global Tectonic 102, 803–814.
Approach. London, Unwin Hyman. Zimanowski, B., Fröhlich, G., and Lorenz, V. 1991.
Wohletz, K. H. 1986. Explosive magma–water Quantitative experiments on phreatomagmatic
interactions: thermodynamics, explosion explosions. Journal of Volcanology and Geothermal
mechanisms, and field studies. Bulletin of Research, 48: 341–358.
Volcanology, 48, 245–264. Zlotnicki, J. 1995. Geomagnetic surveying methods. In
Wohletz, K. H. and Heiken, G., 1992. Volcanology and W. J. McGuire, C. Kilburn, and J. Murray (eds.)
Geothermal Energy. Berkeley, CA, University of Monitoring Active Volcanoes. London, UCL Press,
California Press. pp. 275–300.
Chapter 4

Volcanoes and the geological cycle


Ray A. F. Cas

Introduction as the driving force for the geological cycle and


associated volcanism. We will also consider how
volcanism has contributed to the development of
Volcanoes are one of the most exciting land- the atmosphere, hydrosphere and even life, dur-
forms on Earth today, and frequent volcanic erup- ing the geological cycle.
tions around the world every year are a constant
reminder of the dynamic nature of the Earth
as a planet. With many well-documented major The geological cycle
volcanic eruptions in the last two decades (e.g.,
the eruptions of Mount St. Helens in 1980 in the ‘‘The geological or rock cycle” is the term that
northwestern United States, El Chichón in 1982 in is used for the recycling of rock materials, min-
Mexico, Galungung in 1982 in Indonesia, Hawaii erals, molecules, and ions by dynamic natural
from 1992 to present, Mt. Pinatubo in 1991 in processes throughout Earth history, involving the
the Philippines, Mt. Unzen from 1991 to 1994 in exchange of matter between the Earth’s inte-
Japan, and Soufriere Hills, Montserrat from 1996 rior, its outer shells (lithosphere and crust), the
to present), it may appear that the frequency of atmosphere, hydrosphere, and even the biosphere
eruptions is increasing and that volcanic erup- (Fig. 4.1). It has been recognized for a long
tions are a relatively recent or increasingly signifi- time that the major rock groups are the prod-
cant phenomenon in the Earth’s history. However, ucts of these dynamic recycling processes. The
the remains of ancient volcanic rock successions geological or rock cycle works something like
around the world indicate that volcanic activity this. When magmas are erupted as volcanics on
has been a fundamental aspect of the evolution the Earth’s surface, they are subjected to chem-
of our planet throughout its history. The appar- ical and physical weathering and erosional pro-
ent increase in the frequency of eruptive activity cesses, which strip ionic and solid mineral mat-
in recent times is probably largely an artefact of ter off the volcanics and transport them away
improved surveillance opportunities, technology, to be incorporated into sedimentary deposits
and media coverage. In this chapter the signifi- such as sands and clays, much of which are
cance of volcanic activity in the history and evolu- ultimately deposited on the ocean floor. There,
tion of the Earth will be examined. In particular, those sediments, and the volcanic rocks mak-
we will explore how volcanism plays a fundamen- ing up the ocean crust, are rafted along on
tal role in the geological cycle and the dynamic lithospheric plates and may eventually be sub-
mechanism that drives the geological cycle, plate ducted at the oceanic trenches (Carey, Chapter 1,
tectonics. We will also explore how far back in this volume). They then become deeply buried
time modern plate tectonics can be recognized during continued subduction, and are subjected
Volcanoes and the Environment, eds. J. Martı́ and G. G. J. Ernst. Published by
Cambridge University Press.  C Cambridge University Press 2005.
122 VOLCANOES AND THE GEOLOGICAL CYCLE

Here
H2 , He I
come!
Atmospheric recycling
Asteroid
Cosmic recycling
H2O, CO2, N2, Ar2

Surface process
recycling CO2 + H2O O2
Weathering Biospheric recycling
Erosion
Erosion CO2 + H2O
v CH2O

v v v Coal CaCO3
v
Metamorphism v v
v + Partial melting and crystallization
Volcanic +
v v v
recycling
Underplating
v
v
Partial
v melting
Mantle convection and
recycling
Subduction v
of volcanics,
sediments, v
water, v
skeletal
Magma CaCO3 v

Fig. 4.1. A schematic representation of the geological cycle, density compared with mantle rocks, and it
depicting the interaction between the Earth’s four spheres therefore resists being subducted. It is under-
(lithosphere, atmosphere, hydrosphere, and biosphere) as thrust under the overriding continental crust,
well as the interaction between the Earth and the cosmos. where it is buckled, folded, faulted, and meta-
morphosed. The combined crust is thus enor-
mously thickened forming huge, high collision
to metamorphism as pressure and temperature mountain ranges such as the Himalayas. The
increase. Some of this deeply subducted rock lower part of the thickened crust, which may
material may then begin to partially melt, and be up to 80 or more kilometers thick, is sub-
rise towards the Earth’s surface as magma. Some jected to heat and pressure and may also begin
of this magma crystallizes in the subsurface as to melt forming granitic magmas. These magmas
plutonic bodies of rock, often causing the crust to may also begin to rise because of their buoyancy.
thicken from below by addition of magma to the Meanwhile, the surface of the uplifted, thickened
base of the crust in a process called underplat- mountain range is being subjected to very high
ing, but some may be erupted again as volcanic rates of physical erosion, with huge volumes of
magmas on the Earth’s surface. sediment being shed into nearby rivers. These
In addition, sediments and seafloor volcanics rivers will continue to downcut and will begin to
may be scraped off onto the leading edge of erode deeper levels of the crust, including high-
the overriding lithospheric plate in the trenches, grade metamorphic and granitic rocks. The rivers
where they added to the highly deformed and eventually deliver much of the sediment to the
metamorphosed accretionary prisms there. At deep ocean floor where it again begins its cycle of
some stage a continental mass may be carried transport on top of oceanic lithosphere towards
into the subduction zone, leading to major colli- a subduction zone.
sion between continents or between continents At the same time atmospheric gases are react-
and island arcs. The continental crust on the ing with the lithosphere in processes such as
subducting plate is buoyant because of its low chemical weathering, rainwater is absorbed into
VOLCANIC ACTIVITY IN THE PLANETARY SYSTEM: CAUSES AND IMPLICATIONS FOR EARTH 123

the soils and bedrock, organisms and plants are many orbiting moons. Volcanic activity is there-
extracting compounds such as nitrogen, oxygen fore not just confined to the Earth, and although
and carbon dioxide from the atmosphere, hydro- the plate tectonic model (Carey, Chapter 1, this
sphere, and lithosphere, and fixing them dur- volume) provides a dynamic framework within
ing growth. When these organisms and flora which the location of many volcanoes on Earth
die, their remains become incorporated into the can be explained, this is clearly not the case for
lithosphere (Fig. 4.1), and are thus recycled back all the planetary bodies, most of which show
through the geological cycle. little or no evidence for plate tectonic activity
Volcanic activity therefore plays a fundamen- (Frankel, 1996; Thomas et al., 1997).
tal role in the geological cycle, transferring mat- In some cases, such as the Moon, some large-
ter from the mantle and the lower crust to the volume volcanic events associated with major
Earth’s surface, atmosphere, and hydrosphere. impact structures, especially of the older lunar
Subsequently this volcanic material, as well as highlands, appear to have been related to and
volcanically derived elements fixed by the bio- probably triggered by the late stages of the major
sphere and lithosphere, is recycled when the meteorite impact phase of the Moon’s early his-
products of erosion are deposited in the sea tory. Analysis of the history of the Moon (Head,
and may eventually be subducted back into the 1976) suggests that the Moon underwent a pro-
Earth’s interior. We will now see how volcanic longed phase of major impact activity from its
activity has been fundamental in shaping the origins to about 3800 Ma, and this is thought to
Earth we live on, as well as other planetary bodies have produced the rugged impact generated ter-
of the solar system. rains of the lunar highlands. This was followed
In this discussion, common time terms will be by a declining phase of impacts between 3800
used, and the reader is referred to Harland et al. and 3300 Ma. Smaller isolated impact craters
(1990) for reference. In particular, the Priscoan which pockmark the lunar surface testify to ongo-
eon is used for the time interval from the origin ing isolated events throughout the Moon’s his-
of the Earth, 4.57 billion years ago to 4.0 billion tory to the present day. The large-volume lunar
years ago, the Archean eon extends from 4.0 to mare flood basalt fields have been dated to have
2.5 billion years ago (some people combine these formed between 3900 and 3100 Ma. Their erup-
two eons into one, the Archean), the Proterozoic tion therefore coincides with the late stages of
eon extends from 2.5 billion to 570 million years major impacts, and continues through the wan-
ago (Ma), and the Phanerozoic eon extends from ing phase of activity. Originally the eruptions
570 Ma to the present. of the mare lavas were thought to have been
triggered by impacts, and this could be argued
for those whose ages coincide with the period
of heavy bombardment. However, it is difficult
Volcanic activity in the planetary
to argue this for those mare lavas erupted hun-
system: causes and implications dreds of millions of years after major bombard-
for Earth ments ceased. It also appears that the mare
lavas were derived from great depth within the
The Earth and the planetary solar system formed Moon’s mantle (Frankel, 1996). Although this
about 4570 Ma through progressive accretion of does not exclude a major impact origin, it also
solid and gaseous matter into planetary bod- permits a deep-seated hotspot or plume origin
ies in fixed orbits in the solar nebula. Exam- sourced by radioactivity-generated heat (Frankel,
ination of the surfaces of the planets during 1996).
landings by manned and unmanned spacecraft, Mercury appears to have had a similar history
from the images from orbiting satellite radar and to the Moon. Both have a very thick lithosphere,
photographic surveys, and various other remote- and neither shows any evidence of ever hav-
sensing techniques indicate that volcanic activity ing experienced plate tectonic processes (Thomas
has affected at least all of the rocky planets and et al., 1997). Mercury, like the Moon, shows no
124 VOLCANOES AND THE GEOLOGICAL CYCLE

evidence of tectonic activity other than that asso- associated with internal convection and thermal
ciated with meteorite impacts. instability.
Mars and Venus have experienced more recent Thomas et al. (1997) suggest that the criti-
volcanic activity than the Moon and Mercury. The cal factor determining whether or not plate tec-
volcanism on Venus and Mars is also more diverse tonics operates is the viscosity contrast between
in character than on the Moon and Mercury, and the lithosphere and asthenosphere, and not
includes silicate igneous rocks with compositions the thickness of the lithosphere. For example,
unlike any known on Earth. The surface of Venus although Venus has a lithosphere about the same
has relatively few impact craters and many vol- thickness as that of the Earth, and has about
canic landforms, indicating that volcanic activ- the same planetary diameter, it does not expe-
ity has been active to relatively recent times. rience plate tectonics. The Earth’s lithosphere is
In particular, it appears to have experienced a also more rigid than that of Mars, which has a
major phase of flood-like lava volcanism around surface temperature of about 450 ◦ C, and yet Mars
500 Ma, that smoothed the Venutian landscape also does not experience plate tectonics.
and buried earlier impact generated landscapes On Io, one of the rocky moons of Jupiter,
(Frankel, 1996). and the most volcanically active planetary body
On Mars, impact craters occur and it appears in our solar system, some of the volcanoes are
that volcanism stopped perhaps as recently as sev- sulfur-erupting volcanoes (Sagan, 1979) and spec-
eral tens of millions of years ago as suggested for tacular modern explosive eruption plumes have
the giant volcano, Olympus Mons (Frankel, 1996). been imaged from these (Frankel, 1996). The vol-
Mars and Venus show evidence of fault scarps, canic activity on Io is thought to be triggered by
and rifts, and there has been considerable specu- extremely strong tidal forces, generated by the
lation on the likelihood of plate tectonic systems multiple gravitational forces acting on Io from
during their history (e.g., Sleep, 1994). For Mars, Jupiter and two of the other moons, Ganymede
plate tectonics could only have occurred in its and Europa (Frankel, 1996). These forces are
earliest history when it was still hot, and per- thought to deform and generate frictional melt-
haps subjected to widespread mantle convection. ing of the interior of Io, leading to virtually
Because Mars is substantially smaller than Earth, continuous volcanism.
its interior would have cooled quickly during its It is thus apparent that volcanism in our solar
early history, leading to a thick, cold rigid crust system cannot be related only to plate tecton-
and lithosphere (Thomas et al., 1997). ics. There are diverse causes. It also follows from
Venus is only slightly smaller than Earth, this that the volcanism on Earth need not be
and there have been numerous suggestions that completely related just to plate tectonic factors,
plate tectonic processes may have operated, although much of it is. Impacts, especially dur-
based on linear alignments of volcanic edifices ing Earth’s early history, could also be a factor,
(arcs?), extensive fracture zones, and ridge-like internal thermal instabilities and plumes could
features (Frankel, 1996). However, although colli- be a factor, and at some times, even tidal forces
sion mountain belts exist, subduction zones, mid- between Earth and the Moon could be a factor. To
ocean spreading ridge-like features, and other illustrate that the Earth is not a completely pre-
plate features do not appear to be present dictable volcanic entity, although the majority of
(Thomas et al., 1997). Some of the widespread magmas erupted on Earth have a silicate chem-
early volcanism on Mars and Venus may thus be ical and mineral composition, eruptions of car-
related to the same bombardment events that bonatite magmas are known, and even iron oxide,
appear to have caused volcanism on the Moon magnetite lavas have been found in the Andes, in
and Mercury; some may be related to early plate Chile (e.g., Henriquez and Martin, 1978).
tectonic-like effects, but more recent volcanic The Earth shows clear evidence of extrater-
and tectonic activity cannot be ascribed to plate restrial impact events in the geological past.
tectonics. Relatively recent volcanism can only Impact craters ranging from Precambrian to Pleis-
be attributed to hotspot or plume volcanism, tocene are known. Some of these also appear to
WHEN DID VOLCANIC ACTIVITY BEGIN IN EARTH HISTORY? 125

coincide approximately with episodes of region- was young, its surface was hot, internal temper-
ally extensive and voluminous volcanic activity. ature gradients high, and its outer margins less
For example, both major impact events (Alvarez rigid than today.
et al., 1980) and major phases of volcanic activity
such as flood basalt eruption phases (Rampino,
1987; Rampino and Stothers, 1988; Courtillot
et al., 1990; Rice, 1990) have been proposed as When did volcanic activity begin in
being responsible for mass extinctions, includ- Earth history?
ing that of the dinosaurs, at the end of the Cre-
taceous time period. If major impact events or The oldest known volcanic rocks on Earth occur
phases were capable of triggering volcanic activ- in the Isua greenstone belt of Greenland (Appel
ity on other planetary bodies, then they should et al., 1998). These rocks are between 3700 and
also have created such volcanic responses dur- 3800 Ma old and consist of metamorphosed
ing the Earth’s history, as proposed by Rampino submarine basalt pillow lavas, massive basalt
(1987). This cause–effect relationship may have lavas, subaqueous hyaloclastite (lava-associated
been especially significant during Earth’s early breccias), and interbedded fine-grained sedimen-
history when the major impact phases of activ- tary rocks. They indicate that at least as long ago
ity affected the solar system (Lowman, 1976). as 3800 Ma, volcanic activity was significant on
Although the present oxygen-rich nature of the the Earth’s surface. In addition, the oldest dated
Earth’s atmosphere leads to combustion of much rocks on Earth, the Acasta gneiss of the Slave
incoming extraterrestrial matter due to fric- Province, Canada, which have an emplacement
tional heating, large meteoritic objects could age of 3962 Ma, consist of metamorphosed gran-
still impact on the Earth’s surface. For example, ite and tonalite (Bowring et al., 1989), which indi-
Meteor Crater in Arizona, USA is only 25 000 years cates that subsurface crustal plutonic magmatic
old. In the geological past, when atmospheric oxy- activity was well established then, and probably
gen levels were significantly less than at present, surface volcanism as well. However, the age of
the combustion effect would have been minimal, these oldest known rocks does not reflect the age
so enhancing the number of impact events and of the Earth. The reason why older rocks have not
their effects. Such events are random events not been discovered, even though the age of the Earth
easily accommodated by the steady-state geologi- is known to be 4570 Ma, lies in the processes of
cal cycle, but their effects on Earth history could the ‘‘geological or rock cycle,” as outlined above.
be profound. Because of the Earth’s dynamic nature, differ-
In summary, therefore, although the plate tec- ent types of rocks are constantly being created,
tonic paradigm provides a logical explanation modified, and recycled through the processes of
for the occurrence of many of the volcanic pat- volcanism, subduction, uplift, erosion, sedimen-
terns on Earth today and in the past, it does tation, burial of sediments, lithification, tectonic
not explain them all. Even within the plate tec- deformation, and metamorphism. It thus seems
tonic framework volcanism can occur in random likely that rocks formed during the earliest his-
localities within plates, and at random times tory of the Earth would have been subject to the
(hotspot activity) (Carey, Chapter 1, this volume). processes of the rock cycle and evidence of their
There has also been some debate as to how existence removed.
long plate tectonics has been part of the Earth’s Do we have any indication when a solid Earth
structural and tectonic make-up, and its geolog- first formed? Although the oldest dated rock suc-
ical cycle (see below). In addition, random vol- cession occurs in Canada, there are older dated
canic events could also have been associated with minerals. Detrital crystal fragments of the min-
extraterrestrial impact events throughout geo- eral zircon (ZrSiO4 ), with a lead isotopic age
logical time, externally triggered or indigenous of up to 4400 Ma, have been collected from
deep-level plume events, or even gravitationally a quartzite and conglomerate sedimentary rock
or ‘‘tidally” induced, especially when the Earth sequence from the Archean of Western Australia
126 VOLCANOES AND THE GEOLOGICAL CYCLE

(Maas et al., 1992; Wilde et al., 2001). The depo- r it originated endogenously from within the
sitional age of the metasedimentary host rocks Earth during its long term evolution since orig-
is about 3000 Ma. Zircon is a mineral that crys- inal accretion.
tallizes in granitic magmas formed by the melt-
ing of continental-type crust. It appears that the The arguments against the present atmosphere
dated zircon grains were eroded from 4270 Ma forming at the time of initial planetary accre-
granites exposed on continental crust about 3000 tion are compelling. Accretion would have pro-
Ma ago. More importantly they indicate that a duced enormous heat from the friction caused
solid continental crust existed on Earth at least by collision of solids. The Earth was probably a
4270 Ma ago, and that igneous processes, and largely molten mass, with its surface temperature
therefore probably volcanic processes, were oper- extremely high. Evidence lies in the abnormally
ating then. This is also consistent with the ages of
low concentrations in the Earth’s atmosphere of
the oldest dated samples from the Moon, includ- the heavy rare gas xenon, which has an atomic
ing lava fragments (4200 Ma), lava dust (4350 Ma), mass of 131. Increasing the temperature of gases
and mantle rock, a peridotite with an age of 4500 causes them to become more energetic and the
Ma. These suggest that the Moon had developed molecules to circulate with increased velocities.
a solid volcanic surface by at least 4350 Ma, and It is envisaged that the temperature of the Earth’s
perhaps as early as 4500 Ma. The peridotite rep- surface and atmosphere was so high as to cause
resents one of the earliest crystallized rocks in atmospheric xenon to escape the Earth’s gravity
the Moon’s (and solar system’s) history (Frankel, field during and in the period after accretion. It
1996). then follows that if the heavy gas xenon escaped,
Because the Moon is significantly smaller than then all lighter gases, including all the common
the Earth, it would have cooled more quickly gases making up our present atmosphere, and
than Earth, and it is likely that it formed a crustperhaps even heavier ones as well, would have
earlier than Earth. We also know that the Earth’s escaped.
mantle must have been at least 200 ◦ C average The Earth’s atmosphere therefore originated
hotter than today (Frankel, 1996; Ranalli, 1997), after accretion. Although some theories suggest
causing more vigorous convection and there- that ‘‘capture” of the atmosphere from space
fore volcanism than today. It is therefore likely or from ‘‘cosmic showers” may have occurred
that volcanism was a fundamental, and probably after the Earth cooled sufficiently, and it is likely
the most dominant, process affecting the Earth’s that there is some contribution from such a
earliest crust and surface. source, the most reasonable source for most of
the Earth’s atmosphere is from the release of
volcanic gases from the Earth’s interior. As we
have already seen, volcanic activity has probably
The role of volcanic activity in occurred since the Earth formed and so there has
creating the atmosphere, been an extremely long time-span of Earth his-
hydrosphere, and life tory for accumulation of released gases.
Although modern volcanoes are known
The origin of the Earth’s atmosphere and hydro- mostly for the large volumes of lavas and pyro-
sphere has been a matter of scientific discussion clastic deposits they erupt (see Cas and Wright
for a long time. Possible origins include: (1987) for a discussion of the processes), they
also release huge volumes of gases. For exam-
r it originated at the time of the accretion of ple, during the eruption of Mt. Pinatubo in the
the Earth 4570 Ma ago from gases condensing Philippines on June 15, 1991, 20 Mt of SO2 alone
above the hot surface of the Earth, or were released, this then converting to 30 Mt of
r it and its gaseous component were captured climate-modifying H2 SO4 /H2 O aerosol gas com-
from space during the orbit of the Earth, or plexes (McCormick et al., 1995). Although the gas
it was subjected to periodic ‘‘cosmic showers,” types and their abundances vary from volcano to
or volcano depending on the magma composition, a
THE ROLE OF VOLCANIC ACTIVITY IN CREATING THE ATMOSPHERE, HYDROSPHERE, AND LIFE 127

typical gas composition by volume would include forming an insulating layer of steam (Mills, 1984).
70–80% H2 O, 8–12% CO2 , 3–5% N2 , 5–8% SO2 , and This allows the lava surface to cool sufficiently
minor proportions of H2 , CO, S2 , Cl2 , and Ar. slowly to form a plastic skin. As more magma is
Whereas the abundance of most gases in injected under the skin from the vent it breaks
the atmosphere can be explained by progressive through the skin as a lobe of lava on average
long-term accumulation under the influence of about 0.5 to 1 m in diameter. This in turn also
nearly continuous volcanic activity through geo- forms a skin, preserving a pillow-like form. Many
logic time, the abundance of particularly oxy- such lobate breakouts lead to a lava whose inter-
gen, carbon dioxide, and nitrogen cannot be nal structure is very complex and consists of mul-
thus explained. Volcanic gases are almost totally tiple lobes or pillows stacked on top of each other.
devoid of free oxygen, but conversely they con- Sometimes, however, while the magma is still
tain very large volumes (10% or more) of car- liquid, the insulating steam layer collapses, in
bon dioxide. By contrast the atmosphere con- part because of natural instabilities at the steam–
tains some 21% oxygen and only 0.031% carbon water boundary, in part because of the effects
dioxide. The inconsistencies in oxygen and car- of strong convection currents in the surrounding
bon dioxide abundances in volcanic gases and water mass. Water then comes directly in contact
the Earth’s atmosphere can only be explained by with the liquid lava, which chills instantaneously
the influence of photosynthetic and respiratory to solid glass, shrinks in volume, and by this shat-
organisms through geologic time (Cloud, 1988). ters to an aggregate of glassy debris called hyalo-
However, herein lies a conundrum. Modern respi- clastite (Cas and Wright, 1987).
ratory organisms require oxygen, yet it is almost Indirect data even suggest that surface water
certain that the early Earth’s atmosphere con- may have existed as long ago as 4400 Ma (Wilde
tained none. Furthermore, the early atmosphere et al., 2001). Zircon crystals of this age from
would have been ‘‘thin” and easily penetrated by the Jack Hills of Western Australia have elevated
18
ultraviolet (UV) radiation from the Sun that is O values suggesting interaction with surface
harmful to life. water.
Life could not have evolved until it could be We therefore know that surface bodies of
protected from UV radiation. The early atmo- water, resulting from condensation, atmospheric
sphere must have been carbon dioxide- and water- precipitation, and surface runoff existed by
vapor-rich (a super greenhouse), based on their 3800 Ma and possibly as early as 4400 Ma. This in
high abundances in volcanic gases, but neither turn tells us that the surface temperature of the
carbon dioxide nor water vapor are as effective in Earth and its atmosphere must have fallen below
absorbing and screening the Earth’s surface from 100 ◦ C by then, otherwise liquid water could not
UV as oxygen and ozone. In the absence of the have accumulated.
shielding effects of oxygen and ozone, it is likely At just what stage life first evolved is not clear.
the initiation of life, by whatever means, had to However, the oldest known organic molecules
wait until liquid water bodies had accumulated that represent the first known ‘‘life forms” occur
on the Earth’s surface, and that it occurred under in the Pilbara region of Western Australia, in
anaerobic or reducing conditions. When did this rocks at least 3500 Ma old (Schopf, 1993). The
happen? fossil life forms consist of anaerobic cyanobac-
The oldest supracrustal rocks on Earth, the teria, non-cellular procaryotic organisms some
3800–3700 Ma volcanic rocks of the Isua green- of which were stromatolitic, and some filamen-
stone belt of Greenland, provide the answer. tous (Schopf, 1983, 1993). Stromatolitic fossils of
These volcanic rocks contain two particular erup- algae of 3500 Ma age are known from both the
tive forms of subaqueous lavas, pillow lava and Pilbara region of Western Australia (Walter
hyaloclastite (Appel et al., 1998). Pillow lavas form et al., 1980) and the Barberton Mountain Land,
when fluidal magma such as basalt is erupted South Africa (Byerly et al., 1986). Nisbet (1995)
under water or flows into water at a slow rate. As argues that the earliest organisms may have
the surface of the lava comes in contact with the been chemotrophic bacteria that evolved around
water it causes the water at the interface to boil, subaqueous volcanic hydrothermal vents under
128 VOLCANOES AND THE GEOLOGICAL CYCLE

Fig. 4.2. Cartoon showing the


possible setting for the origin of
life around submarine volcanic
rain volcanic hydrothermal systems. From Nisbet
CO2 - rich atmosphere vent (1995).

light
photosynthesis optional H2S CH4

water
chemotrophy water
obligatory hot lava

anaerobic conditions, perhaps as long as 3800 Ma crucial to the survival of the earliest life forms.
ago (Fig. 4.2), consistent with evidence for both Being non-cellular, those organisms had no cell
aqueous environments and the activity of a car- walls or membranes to protect them from direct
bon cycle at that time (see also Stetter, Chapter 6, contact with oxygen, which is poisonous to the
this volume). Davies (1998) overcomes the prob- organic compounds that constituted the first life
lem of ultraviolet effects on non-cellular organ- forms. The source or origin of the iron and sil-
isms by suggesting that life could have evolved ica in banded iron formations is still unclear,
underground, within the influence of subter- but there have been suggestions that they orig-
ranean, volcanically related, hydrothermal fluid inated from exhalative release of hydrothermal
systems. This is based on the discovery that prim- fluids or volcanic volatiles, perhaps from sub-
itive bacteria-like organisms occur even today marine eruptions or volcanic hydrothermal vent
several kilometers below the Earth’s surface. systems (Derry and Jacobsen, 1990; Eriksson,
However, the push for life forming earlier and 1995).
earlier continues, with proposals that evidence After this time, oxygen levels appear to
for biogenic carbon occurs in the 3800–3700 Ma have progressively increased, as indicated by the
Isua belt rocks in Greenland (Schidlowski, 1988; appearance of pervasively oxidized continental
Mojzsis et al., 1996; Rosing, 1999). ‘‘red bed” sedimentary rocks at about 2000 Ma
Some of the earliest organisms may have been (Eriksson, 1995). Eucaryotic cells, with protective
photosynthetic organisms (Schopf, 1993), because cell membranes or walls and oxygen-mediating
about this time in Earth history the geological enzymes, developed at about 1300 Ma, respi-
record begins to contain ‘‘banded iron forma- ratory metazoans appeared about 700 Ma ago,
tions” (Cloud, 1988). These are fine-grained, lat- and organisms with skeletons of CaCO3 appeared
erally very continuous sequences of red siliceous about 600 Ma ago. Land plants appeared about
or cherty layers and white cherty layers. The red 420 Ma ago, indicating high enough oxygen and
layers are rich in haematite and magnetite, which ozone levels in the atmosphere to shield out suf-
are iron oxide minerals rich in Fe3+ , the oxidized ficient UV to make the land surfaces habitable
form of iron, whereas the white layers consist (Cloud, 1988).
of iron-free chert. Banded iron formations indi- The development of photosynthetic organisms
cate that there was abundant soluble Fe2+ in the was a very important stage in terms of changing
oceans, indicating reducing conditions, but peri- global climate. Up to this time, atmospheric con-
odically this Fe2+ was oxidized to produce the ditions would have been influenced very much
Fe3+ -rich haematite and magnetite layers, indicat- by the high carbon dioxide levels created by ongo-
ing flushes of oxygen in the environment. The ing volcanic activity, creating a super-greenhouse
Fe2+ in the environment acted as a very impor- effect much more severe than the current
tant oxygen receptor or sink, and was probably climatic effects of industrial greenhouse gases. In
THE MODERN PLATE TECTONIC SYSTEM 129

the process of photosynthesis, plants take in car- major mantle convection systems. Mantle-derived
bon dioxide and water from the environment and basaltic magma rises into the fissures along the
convert these into carbohydrate, which becomes MORs from the asthenosphere as the lithospheric
fixed as plant tissue. Oxygen is also produced plates on either side are rafted in opposite direc-
and is released to the atmosphere where it then tions on the diverging limbs of convection cells.
becomes available to help shield out UV. The This leads to the nearly continuous eruption of
removal of carbon dioxide from the atmosphere submarine basalt lavas, of both the pillowed and
when photosynthetic organisms developed would massive sheet types, and also associated hyalo-
thus have made climate more amenable. This clastite. MOR fissure vents are usually so deep
effect would have been reversed slightly by respi- that the hydrostatic pressure due to water depth
ratory release of carbon dioxide by newly evolved prevents explosive eruptions from them. MOR vol-
respiratory organisms in the Late Precambrian canism is the most voluminous on Earth, and the
and through the Phanerozoic. basaltic volcanic oceanic crust produced covers
It is thus clear that volcanism created most almost 70% of the Earth’s surface and has a
of the atmospheric components, and it was relatively uniform chemical composition. It is
thus also responsible for the hydrosphere. It called oceanic tholeiite basalt or MORB (Wilson,
can also be argued that without volcanism 1989).
there would be no life, because the atmospheric Volcanic activity is also common along the vol-
and hydrospheric environment in which life canic arcs associated with the oceanic trenches,
evolved resulted from volcanism, and perhaps which are the sites of the subduction or ‘‘recyc-
the earliest life forms, chemotrophic bacteria, ling” of the oceanic lithosphere back into the
originated in the immediate environment of Earth’s interior (Fig. 4.1). As the downgoing
submarine volcanic hydrothermal vents once sur- oceanic plates plunge back into the mantle, com-
face waters had developed or even within subter- monly at angles of 30–70◦ , they are progressively
ranean hydrothermal systems. Even the chemical heated as they go deeper into the mantle. Eventu-
building-blocks of higher life forms, amino acids ally hydrous fluids are released from the downgo-
and carbohydrates, consist of the chemical com- ing slab and these rise into the overlying wedge
ponents in gases released by volcanoes. of mantle. This causes the melting temperature
in the mantle wedge to decrease, causing par-
tial melting, of perhaps as much as 5% of the
The modern plate tectonic system as mantle wedge. The resulting melts and further
hydrous fluids rise buoyantly towards the Earth’s
a guide for recognizing the tectonic
surface through the crust, and often cause fur-
setting of ancient volcanic ther partial melting of mantle and crust through
successions which they pass, especially if the uprise velocity is
low.
As outined in Chapter 1 (Carey, this volume), the Because magma is rising from many localities
settings in which volcanic activity occurs on the along the slab in a band parallel to the trench,
present-day Earth’s surface can most readily be a distinct line of volcanoes or a volcanic arc
related to the modern plate tectonic framework. forms parallel to the trench. Where the trench
In this section, the essentials of the modern plate and subduction zone are wholly located within
tectonic framework that can be used to identify a large ocean basin away from continental land-
diagnostic elements in ancient counterparts will masses, the volcanoes of the arc begin forming
be briefly reviewed. as submarine volcanoes, and may rise above sea
Volcanic activity is semicontinuous along the level to form lines of volcanic islands called vol-
mid-oceanic ridge (MOR) systems which encircle canic island arcs. Where trenches and subduction
the globe in a connected network of divergent zones occur along the margins of major conti-
plate margins. At these ridges, new oceanic crust nents, the arcs form on land and are called con-
is being created nearly continuously because tinental margin volcanic arcs. Island arcs erupt
MORs are the sites of upwelling limbs of largely tholeiitic island arc basalts (IAB), and
130 VOLCANOES AND THE GEOLOGICAL CYCLE

lavas predominate over pyroclastic deposits and Oceanic crust produced by back arc spreading
eruptions. Continental arc volcanoes also erupt ridges is dominated by basalts known as back-arc
some mantle-derived basalts, but mostly derive basin basalts (BAB), which may be composition-
voluminous magmas from melting of continen- ally distinguishable from MOR basalts, especially
tal crust, resulting in intermediate to high SiO2 in terms of some trace elements and isotopes
magmas such as andesites, dacites, and rhyolites. (Wilson, 1989). Where back arc spreading begins
Arc magmas include a spectrum of geochem- behind continental volcanic arcs, significant vol-
ical compositional suites such as calc-alkaline, umes of silicic magmas such as andesites, dacites,
tholeiitic, and alkaline (Wilson, 1989). Although and rhyolites are produced. Often significant vol-
lava-producing volcanic centers such as stratovol- umes of basalt may also be erupted, producing
canoes and dome complexes are common, very bimodal (i.e., low SiO2 and high SiO2 ) associations
large explosive volcanic centers, such as calderas, of magmas.
and voluminous, widespread pyroclastic deposits, Such bimodal volcanic associations may also
including huge ignimbrite sheets (the deposits be characteristic of areas of the Earth’s conti-
of dynamic pyroclastic flows), are more common nental crust where spreading systems intersect
than in island arcs (see Cas and Wright (1987) or propogate through continental crust (Wilson,
and Sigurdsson (2000) for a discussion of erup- 1989). For example, the Basin and Range Province
tion styles and deposit characteristics of strato- of the western United States may be the result of
volcanoes and calderas). the subduction of a segment of the East Pacific
As lithospheric plates move away from the MOR under the western United States continen-
MORs, they begin to cool. In addition man- tal crust (Eaton, 1982). There, magmatism is very
tle magmas accrete to the bases of the sliding diverse in its chemical composition, including
plates causing them to thicken by the addition alkaline, tholeiitic, and calc-alkaline suites (Chris-
of dense peridotite from below. The lithospheric tiansen and Lipman, 1972). The East Africa rift
slab becomes denser and heavier. As it plunges zone appears to be a limb of the Indian Ocean
into the mantle the density and weight of the system to which it joins in the Afar region of
plate may cause its distal margin in front of the the Red Sea. Although the subsided crust in
trench to begin to sink, not just slide laterally. As the East African rift zone is continental, if fur-
a result, the position of the trench migrates for- ther extension, or continued subcrustal spread-
ward, and extensional stresses begin to develop ing occurs, then oceanic crust may begin to
in the leading margin of the overriding plate. form. Again there, magma compositions are very
Often, the area of the plate behind the volcanic diverse, including peralkaline suites of volcanic
arc begins to rift, or extend in a direction nor- rocks and early flood basalts (Wilson, 1989).
mal to the plate margin (Karig, 1974). This per- In addition, volcanism may occur within the
turbs the configuration of the major mantle con- interior of lithospheric plates (intraplate) as a
vection cell under the overriding plate, and may result of apparently randomly located mantle
cause smaller subsidiary convection cells to form ‘‘hotspots.” These are regions of the mantle which
beneath the arc, particularly under the influence are anomalously hot, and generate large vol-
of uprising magmas feeding the volcanic arc. umes of mantle magmas which rise buoyantly
As a result, divergent spreading ridges may to the Earth’s surface as ‘‘plumes” of magmas.
form behind such volcanic arcs (Karig, 1974) lead- Such hotspots may remain active for millions or
ing to the opening of relatively small back arc even tens of millions of years. For example, the
basins (e.g., Taylor, 1995). In volcanic island arc Hawaiian chain of volcanic islands, and to their
systems, such back arc basins develop oceanic north, the Emperor island chain, are thought
crust; in continental margin arc systems such to have resulted from a reasonably stationary
basins are initially continental, but may develop hotspot located under the currently active ‘‘big
into oceanic basins if extension and rifting of island” of Hawaii, which occurs at the south-
the continental crust proceeds to the spreading ern end of the chain (Carey, Chapter 1, this vol-
stage, producing oceanic crust and lithosphere. ume). The islands become progressively older to
THE MODERN PLATE TECTONIC SYSTEM 131

the north, a trend that is consistent with the due to cooling as it drifts away from MORs. When
drift of the Pacific lithospheric plate northwest- continental masses are rafted into subduction
wards over the deep hotspot, which has succes- zones as passengers on subducting lithospheric
sively released batches of oceanic island basalt plates, they resist being subducted into the man-
magma (OIB) (Wilson, 1989) through the plate, tle because of their relatively low density and
producing the succession of volcanic islands over buoyancy relative to the mantle. Continental
tens of millions of years. Other areas of intraplate masses may thus collide with each other at
hotspot activity may also occur within continen- subduction zones. The downgoing continental
tal masses, and the Tertiary basaltic province of mass may be underthrust under the leading edge
eastern Australia (Johnson, 1989) is a good exam- of the overriding continental mass, and resists
ple. Eruptive products are diverse, ranging from being subducted any further, causing an anoma-
undersaturated basic alkaline to highly silicic per- lously thickened zone of continental crust. This
alkaline magmas. Although dominated by mafic appears to have happened in the Himalaya moun-
lavas, which may be localized to the confines of tain range of Asia as a result of the attempted
volcanic edifices or be far-flowing, confined by subduction of India under Asia.
topography and flowing up to tens of kilometers Curiously, such continental collisions and
from the vent, the products also include more attempted subduction do not produce large vol-
evolved lavas and pyroclastic deposits. Pyroclastic umes of volcanic rocks or volcanic arcs. It seems
deposits may be localized to small monogenetic likely, however, that in the lower crust of such
volcanic centers such as scoria cones, maars, and collision belts, partial melting and generation of
tuff cones, but may also be well developed in crustal magmas must eventually occur because
large continental shield volcanoes (see Cas (1989) of the high pressure and temperature conditions
for a summary of eruption styles, volcano types, at the base of such thickened crust. If such mag-
and nature of the eruption products). mas form, they do not appear to migrate up suf-
Large-volume basalt volcanic provinces, called ficiently to erupt, but must presumably reside or
flood basalt or plateau basalt provinces, and large solidify in the crust, forming granite batholiths.
igneous provinces, are also likely to be the prod- Why is volcanic activity not prevalent in conti-
ucts of major mantle hotspots or plumes. These nental collision belts?
are known from both continental and oceanic Examination of settings in which volcanic
crustal settings. They are characterized by multi- activity occurs in the modern plate tectonic
ple, huge-volume basalt lavas that are sheet-like, framework suggests that there is a relationship
cover areas of 104−105 km2 , and flow up to 300 km, between settings where voluminous volcanism
or more. Huge oceanic plateaus such as the occurs, and regional tectonic stress fields. Clearly,
Ontong Java Plateau of the western Pacific Ocean divergent plate boundaries, where voluminous
are probably long-lived submarine flood basalt mid-oceanic spreading ridge volcanism occurs,
provinces or large igneous provinces. Well-known and in young incipient divergent plate margin
examples of continental flood basalt provinces settings such as the Red Sea and East Africa
include the Deccan Traps, India, the Karoo and rift system, are areas where the regional tectonic
Etendecka Basalts, South Africa and Namibia, the stress field is extensional. Under these conditions
Ethiopian Province, and the Parana Basalts of σ 1 , the maximum principal stress component
South America. Intriguingly, these flood basalt is (sub)vertical, and σ 3 , the minimum principal
provinces are associated with the rifted conti- stress component is horizontal, leading to nor-
nental margins resulting from the break-up of mal and lystric faulting and graben formation.
the Mesozoic supercontinent of Gondwana. The These represent the ideal conditions for large vol-
Columbia River Province of the western United umes of buoyant magma to reach the Earth’s sur-
States is associated with the northern margin of face, utilising the numerous (sub)vertical fault
the extensional Basin and Range rift province. and fracture systems as magma conduits.
Subduction of oceanic lithosphere appears to It used to be considered that convergent sub-
occur readily because of its increasing density duction plate settings coincided with widespread
132 VOLCANOES AND THE GEOLOGICAL CYCLE

compressional stress fields. This is certainly the sediments are being scraped off the downgoing
case in areas of collision of continental masses plate and where there is some frictional coupling
with each other, or where significant island sys- between the two plates.
tems, oceanic plateaus or ridges collide with Even in continental margin subduction set-
trench walls or continental masses. In these set- tings such as the Andes and the Cascades arc of
tings, interestingly, volcanic activity ceases. For northwestern America, there appears to be evi-
example, the Himalayas, the greatest collision dence of extension where the volcanic arc volca-
mountain belt on Earth, is not marked by a vol- noes occur. For example, in the Andes the loca-
canic arc. The simple explanation might be that tion of the arc volcanoes coincides closely with
because subduction of oceanic lithosphere has major regional grabens such as the Puna and Alti-
ceased, partial melting of that lithosphere is no plano, indicating extension. Lamb et al. (1997),
longer taking place. And yet, because of the enor- however, argue that the grabens are bounded
mous crustal thickness (>80 km), it is likely that by differentially uplifted ‘‘pop-up” fault blocks
melting of the lower crust is occurring, produc- in a subduction system that is in widespread
ing granitic magmas. Why do such magmas not regional compression for up to 1000 km from
reach the Earth’s surface and produce volcanism? the trench. During the Mesozoic, the Andean sub-
Similarly in areas where the Louisiade Ridge is duction margin experienced limited rifting and
colliding with the Vanuatu Trench, there is an extension, forming small back arc basins which
absence of arc volcanoes, whereas away from developed oceanic crust (Dalziel et al., 1974). The
the collision zone there are numerous arc vol- alternation between compressional tectonics and
canoes. Similarly along the Peru–Chile subduc- extensional tectonics that the Andes has expe-
tion margin of the Andean mountain range of rienced can be explained as a function of the
South America, where oceanic ridges such as the relative rates at which the leading edge of the
Carnegie, Nazca, and Juan Fernandez ridges col- South American plate is advancing westwards,
lide with the trench, there is a break in the pres- which is controlled by the spreading rate at the
ence of volcanoes, which elsewhere form a nearly Atlantic mid-oceanic spreading ridge, and the
continuous line of arc volcanoes parallel to the rate at which the hinge zone of the subduct-
trench. ing Nazca plate in the Pacific is rolling back
It has now been recognized, however, that (Sebrier and Soler, 1991). If the rate of rollback
many subduction settings, especially where vol- is greater than the spreading rate at the mid-
canic systems are well developed, show signif- Atlantic ridge, then extension at the Andean mar-
icant evidence of extensional processes within gin will occur, accompanied by voluminous vol-
and behind the arc. This is most clearly obvi- canism. If the spreading rate exceeds the rollback
ous in oceanic arc systems, such as the Mari- rate, then the Andean margin will undergo com-
anas and Tonga–Kermadec arcs, which have expe- pression, with the likely effect of reducing the
rienced back arc basin formation and extension. level of volcanism. Even under mild compression
Such basins have well-defined spreading ridge some volcanism may occur if the magma fluid
systems in their center and the arc volcanoes pressure exceeds the regional comressional stress
are usually located along the trenchward margin field or where local transtensional basins form
of the extensional back arc basin. In these con- along transverse fault zones.
vergent subduction systems, the volcanic activ- Interestingly, many of the the relatively
ity is entirely located in the extensional stress young, large-scale continental flood basalt
field domain of the arc system. This extensional provinces of the world occur at, or very close to,
stress field has developed due to passive, gravita- the rifted margins of the major continents.
tionally induced sinking of the subducting plate, It thus seems that voluminous volcanic
allowing the hinge zone to retreat or ‘‘roll back” episodes and provinces are commonly related to
oceanwards, and the overriding plate to extend. regions of the Earth’s crust experiencing exten-
A compressional stress field domain occurs only sion or transtension in either divergent oceanic,
at the leading edge of the overriding plate where continental rift, or arc–back arc settings. This
RECONSTRUCTING THE TECTONIC SETTING OF ANCIENT VOLCANIC SUCCESSIONS 133

guide can then be used to assess the paleotectonic tinental lithosphere under oceanic lithosphere.
settings for ancient rock successions containing As argued above, because of the buoyancy of
widespread, voluminous volcanic successions. continental crust, the subduction system gets
jammed up, the continent buoys up, and a seg-
ment of oceanic crust and lithosphere ends up
perched on land.
Reconstructing the tectonic setting
By contrast, the existence of ancient continen-
of ancient volcanic successions: the tal crust is characterized by abundant, thick, vari-
Phanerozoic record ably folded and metamorphosed continental and
continental shelf and slope sedimentary rock suc-
Identifying the original setting of ancient vol- cessions, especially those with abundant quartz
canic successions is a challenging task. In the first sand and silt grains. In addition, such metased-
instance, the task is to identify criteria for recog- imentary rocks will be intruded by regionally
nizing the type of crust the volcanism occurred extensive granite batholiths, indicating a source
on. Where volcanism is thought to have occurred of melted silica and aluminum-rich continental
on oceanic crust, the subjacent rock succession crust, and the rock successions may be signifi-
must preserve the rock succession that makes up cantly metamorphosed, producing high temper-
oceanic crust. Deep-sea drilling through present- ature, low to moderate pressure metamorphic
day oceanic crust, as well as onland exposures mineral assemblages indicative of the middle to
through ancient oceanic crust which has been lower crust.
uplifted and tectonically emplaced onto land, The influence of a subjacent continental-type
reveals a consistent stratigraphy and succession crust may also be indicated by the composition
of rock units. The uppermost unit, called layer 1, of the volcanics. If voluminous andesites, dacites,
consists of typical open ocean fine-grained sed- and rhyolites occur, they also indicate a magma
iments, often with abundant pelagic microfos- source of partially molten subjacent continen-
sils preserved. This lies above layer 2, consisting tal crust. The corollary, however, does not apply:
of a thick succession of basalt lavas, including abundant basaltic volcanic rocks do not neces-
both classical submarine pillow lavas, and mas- sarily indicate a subjacent oceanic crust, unless
sive sheet lava (layers 2a, 2b). This passes down the previously defined succession of rock units of
into layer 2c, which consists of (sub)vertically ori- oceanic crust also occur. Basalts originate from
entated sheet-like dykes of basalt and dolerite partial melting of mantle, whether it be under
which may pass down into coarse plutonic rocks oceans, or continents. However of all volcanic
such as gabbros of layer 3 (Brown et al., 1989). rocks, basalts from different settings may have
The base of the oceanic crust, called the oceanic some distinctive geochemical characteristics that
Moho, separates the crust from the coarse olivine may help to narrow the possible settings (Pearce
rich peridotites of the upper mantle. Many cases and Cann, 1973; Wilson, 1989).
of fault emplaced successions of ancient oceanic Beyond establishing the nature of the crust,
crust, known as ophiolites are known from con- assessing the tectonic setting in which volcanism
tinental and/or arc collision belts of Phanerozoic occurred is the next step in understanding the
age (i.e. 0–570 Ma) (Coleman, 1977, 1984; Moores, role of volcanism in the geological cycle. ‘‘Tecton-
1982), clearly indicating that oceanic and/or back ics” refers to the large-scale dynamic lithospheric
arc basin oceanic crust existed, that processes of and crustal processes which affect the Earth’s
subduction were active during the Phanerozoic, surface and cause crustal deformation.
and that periodically fragments of oceanic crust Of the various settings outlined above, mid-
became incorporated into subduction complexes, oceanic and back arc basin spreading ridge vol-
or were overthrust tectonically onto continental canism are the easiest to recognize. These settings
land masses in a process called obduction. Obduc- should be easily discernible in the preserved
tion is the process whereby a subduction zone stratigraphy of oceanic crust, including the layer
forms involving the attempted subduction of con- 1 oceanic sediment assemblage, layer 2 massive
134 VOLCANOES AND THE GEOLOGICAL CYCLE

and pillowed basalt lavas, and sheeted dolerite 1989) is one of the best-documented accretionary
dyke complex, and layer 3 gabbroic plutonic prism systems of Phanerozoic age preserved in
complex, overlying mantle peridotites. We have the geological record.
clear evidence of such settings occurring in the In addition, ancient subduction settings
Phanerozoic, e.g., the Cretaceous Samail ophio- should be marked by assemblages of volcanic arc
lite of Oman and the Jurassic Vourinos ophiolite rock assemblages, including for young oceanic
of Cyprus (Coleman, 1977) and the Early Ordovi- arcs, island arc tholeiite basalts and basaltic
cian ophiolites of Newfoundland, Canada (Dewey andesites, and subsurface plutonic equivalents,
and Bird, 1971). open ocean pelagic and hemipelagic sedimentary
By contrast, subduction settings are more dif- rocks, deposited by deep water suspension set-
ficult to recognize in the geological record. This tling, together with associated volcaniclastic sed-
is because subduction settings commonly show imentary rocks which represent aprons of debris
signs of both compression and extension. Com- derived from the arcs and transported into deep
pression is evident at the trench and in the lead- water by mass-flow sedimentary processes. Such a
ing edge of the overriding lithospheric plate, succession should be preserved on a ‘‘basement”
where sediment rafted into the trench on the of old oceanic crust.
downgoing plate is ‘‘scraped off,” and added to For continental margin arcs, although some
the leading edge of the overriding plate, form- basaltic rock successions will occur, there
ing a highly folded and faulted metasedimentary would also be a significant association of more
wedge, which may also include fault slices of ‘‘evolved,” more felsic volcanic and plutonic rock
dismembered oceanic crust and even seamounts associations, including andesites and diorites,
rafted into the subduction zone. dacites and granodiorites, and rhyolites and
These highly deformed wedges of ‘‘off- granites which comonly show calc-alkaline geo-
scraped” sedimentary and volcanic rocks are chemical affinities. Sedimentary rock associa-
called accretionary prisms or subduction com- tions will include abundant coarse volcanic con-
plexes (Karig and Sharman, 1975). They represent glomerates and sandstones, as well as coarse
the only physical evidence of the subduction pro- older continental crust derived bedrock sedimen-
cess. Sometimes, the rocks are so deformed that tary debris derived from sources such as older
they lose all coherence of the original stratigra- metasedimentary and metaigneous source rocks.
phy. They become so dismembered by the very Paleoenvironments represented by these sedi-
high shear stresses in the subduction zone that mentary associations include continental alluvial
they break into highly distorted or strained sliv- fans and braided rivers, coarse shoreline deltas,
ers centimeters to many kilometers in dimen- and submarine fan and apron settings.
sion, which become chaotically mixed and highly The geochemical compositions of continental
sheared, and are called ‘‘tectonic melange,” or arc volcanic belts vary from being alkaline to
‘‘broken formation.” Accretionary complexes may tholeiitic to calc-alkaline. Such associations are
also contain domains of rock which have been considered to be diagnostic of subduction arc
subducted tens of kilometers below the seafloor settings by some people. However, similar asso-
in the downgoing plate and have been subjected ciations may occur in major continental rift set-
to very high pressure, relatively low tempera- tings, such as the Cainozoic Basin and Range Pro-
ture metamorphism, before being jostled back vince of the western United States, which include
as faulted slices to the surface. The distinctive examples of almost every chemical association
high pressure, low temperature metamorphism known (e.g., Christiansen and Lipman, 1972).
produces distinctive assemblages of metamorphic Behind the arc, there may be a complex of
minerals called blueschists, which include high extensional graben basins. In fact, the arc is
pressure minerals such as glaucophone, and rock almost certainly located within intra-arc or back
types such as eclogite. The famous Mesozoic Fran- arc extensional basins, in an area of the subduc-
ciscan accretionary prism association of Califor- tion system subjected to neutral, or extensional
nia (Bachman, 1982; Underwood, 1984; Aalto, regional stresses. For example, volcanoes of the
RECONSTRUCTING THE TECTONIC SETTING OF ANCIENT VOLCANIC SUCCESSIONS 135

Cascades arc of the northwestern United States extremely wide (hundreds of kilometers or more)
are associated with intra-arc grabens. The volca- relative to the regional tectonic trend, then it
noes of the Andes are also closely associated with is unlikely that the setting was a subduction-
major grabens. At times extension may even lead related one, especially where there is no clear
to arc rifting, the onset of spreading, and the for- accretionary complex identifiable. It has not been
mation of new oceanic crust and basins, which demonstrated that very shallow dipping subduc-
appears to have occurred along the Andean conti- tion zones will produce very wide volcanic arc
nental margin during the Jurassic and Cretaceous belts hundreds or more kilometers wide. Compli-
(Dalziel et al., 1974). The sedimentary associations cations occur in reconstructing ancient tectonic
that may be preserved in these back arc exten- settings when late stage strike–slip faulting on
sional basins could include a complex associa- a regional scale dismembers an original tectonic
tion of alluvial fans, braided rivers, lakes, deltas, system and displaces particular elements along
fan deltas, and shallow to deep marine fans and strike. For example, the displacement of a subduc-
aprons. Such volcanic and sedimentary associa- tion complex away from its arc and back arc sys-
tions are unfortunately similar to volcanic and tem may make such an arc system difficult to dis-
sedimentary associations that might be devel- tinguish from a rift system. Resolving such a sit-
oped in continental rift settings, such as the rela- uation cannot be done by guessing. It is better to
tively narrow East Africa rift zone, or the broader be objective and list the possibilities and pros and
Basin and Range Province of the western United cons rather than inventing, for example, subduc-
States. tion settings, even invoking multiple coexisting
This raises a major problem. Unless well- subduction zones to explain closely spaced linear
defined regionally extensive (hundreds of kilome- volcanic belts, when no real evidence exists.
ters) accretionary complexes are found in associ- As an example of the problems of reconstruct-
ation with relatively narrow belts of volcanic and ing ancient configurations, the Paleozoic Lach-
sedimentary rocks that define a linear or curved lan Fold Belt of southeastern Australia evolved
trace also over hundreds of kilometers long, evi- from a widespread oceanic paleogeography in
dence of subduction settings is poor to say the the Cambro-Ordovician to a continental paleo-
least. To put this in a true scale context, major arc geography in the Late Devonian (Cas, 1983). Cam-
volanoes such as stratovolcanoes or calderas are brian and Ordovician rock systems consist of
tens of kilometers in diameter, and are separated mafic volcanics and intrusions and interbedded
by kilometers to tens of kilometers. Regrettably, deep marine sedimentary rocks indicative of a
some reconstructions of ancient terrains invoke setting marginal to the paleo-Australian land-
the existence of volcanic arcs, when the dimen- mass. The volcanic successions are commonly
sions of belts of volcanic rocks are only on the fault bounded, with the faults commonly hav-
order of tens of kilometers long, and evidence for ing experienced the last phases of movement
accretionary complexes are lacking. Similarly, the in the Devonian and Carboniferous. There has
existence of calc-alkaline volcanic associations is been prolonged debate on the nature of the
considered by some to be sufficiently indicative crust upon which volcanism and sedimentation
of subduction arc settings, when clearly, calc- occurred, some favoring Cambrian oceanic and
alkaline associations also occur in other settings associated arc crust (Crawford and Keays, 1978;
such as rift settings. Gray, 1997), others continental crust (Cas, 1983;
As outlined above, continental rift settings Chappell et al., 1988). There have also been vari-
may show similar features to continental intra- ous suggestions that volcanism occurred at plate
arc and back arc settings, and in the absence margin arc subduction settings or intraplate rift
of accretionary prisms, the two may be impos- settings. However, no obvious Cambro-Ordovician
sible to distinguish, especially if the rift zone is accretionary prism successions have been recog-
narrow, or has several narrow branches such as nized, except for a possible remnant belt of poly-
the East Africa rift zone. However, if the zone deformed metasedimentary and volcanic rocks
of volcanism (and associated plutonic activity) is outcropping locally along the south coast of New
136 VOLCANOES AND THE GEOLOGICAL CYCLE

South Wales, probably related to an Ordovician enced by a large range of variables including the
to earliest Silurian arc system. composition and previous history of the source
For the Silurian to Middle Devonian rock sys- region in the Earth’s interior from which the
tems of the Lachlan Fold Belt, it is clear from magma is derived, the degree of partial melting
the widespread occurrence of granite batholiths the source region undergoes, the pressure and
of this age throughout the fold belt, that the temperature in the source region, the influence
crust was almost everywhere continental in char- of external fluids, the magma ascent rate and the
acter, thickened during an end-Ordovician – Early degree of interaction with country rock through
Silurian compressional orogenic event. The sys- which the magma rises, to mention only just a
tem is marked by numerous extensional basins. few of the variables.
Most of these appear to have been floored by sub- Reconstructions and interpretations based
sided continental crust, but one may have devel- solely on geochemistry should be based on a
oped embryonic oceanic or transitional crust. probability-based likelihood. For example, plots
Despite this, a number of people have proposed of various geochemical parameters on tectonic
that multiple subduction zones and associated affinity diagrams show a spread of data points,
arcs existed within the Lachlan Fold Belt during and not uncommonly data from volcanics in one
the Silurian to Middle Devonian. However, clear tectonic setting overlap with fields for other tec-
accretionary prisms with high pressure metamor- tonic settings. Data for particular tectonic set-
phic assemblages have not been identified where tings should be contoured. Then when data for
the subduction zones are supposed to occur, nor ancient successions are plotted on such con-
have extensive tracts of remnant oceanic crust, toured tectonic affinity diagrams, the likelihood
which is a prerequisite for subduction: subduc- of the ancient succession representing one tec-
tion cannot occur without consuming oceanic tonic setting or another can be stated in sta-
lithosphere. Also, the scale of the proposed indi- tistically realistic probability terms, rather than
vidual ‘‘volcanic arcs” is too small. A more likely being depicted in a single biased manner. The rea-
scenario is a broad magmatically active rift ter- son for suggesting such a statistically meaningful
rain, perhaps similar to the present day Basin approach is that a particular geochemical com-
and Range Province, influenced by major man- positional suite of volcanics does not necessarily
tle activity, perhaps intraplate hotspot or plume uniquely depict a particular setting.
activity, and widespread crustal melting. It seems It is clear that tectonic reconstructions of
difficult to explain magmatic activity in a belt ancient rock systems and their settings cannot be
hundreds of kilometers wide in any other way, done on the basis of one parameter only, such as
especially in the absence of clearly defined sub- structural style, or the geochemical characteris-
duction system(s), and when geochemical charac- tics of igneous suites of rocks. Successful tectonic
teristics of ancient volcanic successions may be reconstructions depend on a holistic approach –
ambiguous in their significance. considering all geological aspects of the entire
Often attempts are made to evaluate the tec- rock belt in question. So these are some of the
tonic setting of ancient volcanic successions by problems of reconstructing the setting of ancient
the process of ‘‘geochemical fingerprinting.” This volcanic successions and associated rocks. The
involves comparing the geochemical character- question then is, how far back in time can we
istics of ancient volcanic successions to those recognize evidence for the existence of plate tec-
of modern volcanic suites from particular tec- tonic settings based on the modern plate tectonic
tonic settings, and by analogy, interpreting the paradigm?
tectonic affinity of the ancient succession based To answer this we must work backwards in
on similarities of geochemistry. This is fraught time. Clearly plate tectonics can be ascribed with
with dangers because there is nothing sacro- certainty as far back as the age of the oldest
sanct about magma geochemistry. The geochem- oceanic crust on the modern ocean floor, which
istry of particular magmatic successions is influ- is Jurassic in age. What evidence is there for
PRECAMBRIAN VOLCANISM AND TECTONIC SETTINGS 137

plate tectonic processes during the Paleozoic? Were Precambrian processes significantly
Cambro-Ordovician ophiolite complexes (slices of alien to modern or even Phanerozoic geolog-
ancient oceanic crust), occurring in complex tec- ical processes? While this is almost certainly
tonic zones consistent with plate subduction pro- true in some cases, basic physical and chemi-
cesses is well documented for the Appalachian cal principles were the same. Thurston (1994)
orogen (Dewey and Bird, 1971) indicating the and Sylvester et al. (1997) find that many vol-
likelihood of plate subduction processes as long canic and associated volcaniclastic sedimentary
ago as the beginning of the Paleozoic era almost successions (and therefore the processes) were
600 Ma ago. the same in the Precambrian as in Phanerozoic
successions. Common modern-day volcanic prod-
ucts such as viscous felsic lava domes, low vis-
cosity mafic sheet lavas, autobreccias, subaque-
Precambrian volcanism and ous hyaloclastites, and ignimbrites can all be
tectonic settings found in the Archean rock record. Exceptions
include ultramafic komatiite lavas which are
There can be extreme difficulties in evaluating almost exclusively Precambrian in age, and pre-
the tectonic setting of Precambrian volcanic and dominantly middle to late Archean, but there are
associated rock successions for several reasons. even some Mesozoic ones reported on Gorgona
First, in the absence of zone macrofossils, dating Island in Mexico.
and correlating rocks is difficult and expensive Eriksson (1995) and Eriksson et al. (1997,
because of the costs of using precision radiomet- 1998a) demonstrate that normal sedimentary
ric or isotopic dating methods. Second, even if it processes and environments also existed during
is possible to date rocks, the experimental errors, the Archean and Proterozoic. However, in the
given the ages involved, may be significant and on absence of vegetation to provide stable flood-
the order of millions to tens of millions of years. plains for single channel meandering rivers
It thus becomes very difficult to establish a com- almost all alluvial plain rivers would have been
plete and exact chronology of events in such a braided because of the lack of stable banks. As a
long, distant interval of time. Too often attempts consequence sediment flux rates would have been
to reconstruct events in the Precambrian lose high, and in aqueous environments, fan systems
sight of this and present perspectives couched would have been common (fan deltas, deltas, and
in modern or recent geological time and rate subaqueous fans). In addition, banded iron for-
terms. Third, because of the range of effects of the mations are peculiar to the late Archean and
geological cycle as outlined above, the geological early Proterozoic, not because weird processes
record is often likely to be incomplete, although occurred, but because particular environmental
in some areas of Archean geology it is remark- geochemical conditions existed then that do not
able how continuous or complete the geological today. If those conditions did occur today, banded
record appears to be. iron formations would also form today. Perhaps
There is an aura about Precambrian rock sys- what was different then during much of the Pre-
tems because of their antiquity, somehow imply- cambrian, were some of the essential conditions
ing that Precambrian processes were abnormal that existed both inside and on the Earth’s sur-
relative to modern Earth processes. However, it face.
is easy to forget that the Priscoan, Archean, and Many works have been written and com-
Proterozoic eons represent 87.5% of geological piled on Precambrian geology, including Windley
time. On a total Earth timescale, therefore, what (1976), Kroner (1981), Condie (1992, 1994, 1997),
happened during the Precambrian was more Coward and Ries (1995), de Wit and Ashwal (1997),
correctly the norm, and the Phanerozoic rock Eriksson et al. (1998b), Percival and Ludden (1998),
systems represent very recent evolutionary stages and Bleeker (2002) and from these it is clear
in the history of the Earth. that there is still a lot of uncertainty and debate
138 VOLCANOES AND THE GEOLOGICAL CYCLE

about the physical, chemical, and as a result, because metamorphism and alteration have given
the geological state of the Earth during the them a greenish color because of the secondary
Precambrian. The peculiarities of Precambrian minerals developed, such as serpentine, chlorite,
atmospheric evolution and the role of volcanism epidote, and actinolite) have been considered in
have already been discussed above. We will now some instances to represent remnants of oceanic
consider the state of the lithosphere at the time, crust, in the absence of ophiolite stratigraphy,
the implications this has for the tectonic regime, this is difficult to argue. Certainly komatiites can-
and the effects of this on the understanding of not be considered to represent a component, or
Precambrian volcanism. product, of normal oceanic crust-forming pro-
Since a fundamental requirement for plate cesses, because they are derived from the deep
tectonics is the existence of, and the subduction mantle, not the asthenosphere (Arndt, 1994).
of, oceanic lithosphere, it is difficult to advo- Another significant problem in terms of consid-
cate plate tectonic settings for Precambrian vol- ering the tectonic significance of mafic and ultra-
canics unless evidence for oceanic crust associ- mafic successions of Archean greenstone belts
ated with accretionary prisms can be found. In is that in at least some greenstone belts (e.g.,
this regard, there is considerable debate in the the Norseman–Wiluna belt, Western Australia)
geological literature on whether or not clearly the mafic–ultramafic volcanics contain signifi-
defined ophiolite successions of Precambrian age cant zircon xenocrysts, indicating that the mag-
exist, whether or not plate tectonics as we know mas passed through granodioritic bodies or conti-
it today operated, or if some modified form of nental crust during uprise to the Earth’s surface
global tectonics existed. Bickle et al. (1994) and (Squire et al., 1998). It would appear that such
Hamilton (1998) have proposed that there is no greenstone belts were erupted onto continental
clear evidence for the existence of remnants of type crust, and therefore do not represent oceanic
oceanic crust (ophiolites) of Archean age. Some crust.
authors such as de Wit (1998) and Choukrone Other uncertainties associated with adopting
et al. (1997) dispute this, arguing that because a modern plate tectonics perspective of Archean
of the long-term effects of the geological cycle, tectonics include the absence of any known
including subduction, uplift, tectonic dismem- blueschist facies metamorphic assemblages of
bering, and erosion of orogenic rock systems of Archean age. As previously outlined, blueschists
Archean age, the chances of preserving ophio- are high pressure metamorphic rocks thought to
lites are low. De Wit (1998) suggests that there result from the unique high pressure, deep meta-
are sufficiently well-preserved mafic volcanic and morphic regime of subduction zones. The old-
plutonic complexes with a number of the strati- est known blueschist assemblages are 1.8 billion
graphic elements of the ophiolite succession pre- years old from China, and although blueschists
served to warrant considering them to be the are known from Paleozoic successions, they
relicts of true Archean oceanic crust. He cites become relatively common from the Mesozoic
several examples in the pan-African orogenic onwards (Choukrone et al., 1997). Does this mean
belts as good examples of preserved Archean that subduction processes did not occur dur-
ophiolites. ing the Archean or that a different subduction
Archean greenstone–granite belts, which rep- regime existed?
resent the most important architectural com- It is commonly accepted that the mantle was
ponent of Archean cratons, commonly consist at least 200 ◦ C hotter in the Archean than at
of associations of basalts, komatiites (which are present, resulting mostly from higher levels of
ultrabasic lavas and intrusions; see below), fel- heat released from radioactive decay than at
sic volcanic rocks, granitic bodies and associated present (Choukrone et al., 1997; Ranalli, 1997),
sedimentary successions. The basalts are com- but presumably also from residual accretion-
monly tholeiitic or hybrids of komatiites, called generated heat. Higher heat flow would have cre-
komatiitic basalts. Although the mafic and ultra- ated more vigorous convection in the mantle
mafic components of greenstone belts (so called than at present and more ductile deformation.
PRECAMBRIAN VOLCANISM AND TECTONIC SETTINGS 139

a 2.8 OPATICA N. ABITIBI


Fig. 4.3. Representation of two
possible Archean tectonic systems,
showing a convergent tectonic model
based on the geology of the Superior
2.75 S. ABITIBI Province of Canada, and a diapiric

CANADIAN STAGES
b tectonic model based on the geology of
the Dharwar Craton of India. Successions
of cross-sections show possible evolution
2.7 of tectonic stages. Numbers above each
c cross-section refer to the time depicted in
gigayears. From Choukrone et al. (1997).

2.65 PONTIAC
d

e 3.3

3.2
f

HEAT INDIAN STAGES


2.5
g
Shortening

Higher rates of convection almost certainly would would have been to make oceanic lithosphere
have led to more rifting and more subduction more buoyant and more difficult to subduct than
on smaller scales and under more ductile con- at present. In addition, it is possible that with a
ditions. Evidence for crustal convergence, com- thicker crust and under more ductile mantle con-
pressional deformation on a regional scale, and ditions, during attempted subduction, delamina-
crustal accretion exists in the nappe and fold tion or flaking of the crust from the upper man-
and thrust belt terranes of the Archean, such tle along the oceanic Moho may have occurred,
as the Superior Province of Canada (Choukrone leading to sinking and subduction of lithospheric
et al., 1997) (Fig. 4.3). Although these show clear mantle but not the crust, which would have
evidence of large-scale crustal convergence and been accreted to the overriding crust. Ranalli
shortening, and achieve high metamorphic (1997) calls this type of subduction ‘‘flake tec-
grades indicating burial to perhaps several tens tonics.” Under a higher temperature, ductile sub-
of kilometers, they lack blueschists. duction regime, high pressure–low temperature
Calculations of the effect of higher heat blueschist metamorphic conditions were proba-
flow on crustal formation processes suggest that bly not achieved. What the effect of such a sub-
oceanic crust would have been thicker, perhaps duction regime may have been on the generation
greater than 25 km, and lithosphere thinner and of subduction generated arc magmas is not clear.
more ductile (Ranalli, 1997). The effect of this Hamilton (1998) argues strongly that there is no
140 VOLCANOES AND THE GEOLOGICAL CYCLE

evidence for plate tectonics in the Archean, and intraplate basaltic continental volcanic fields,
that Archean tectonics was driven by the more especially flood basalts, can be related to hotspot
ductile and higher thermal regimes inherent in plume activity (Richards et al., 1989). In the
the Earth’s early history. Archean, the clearest representation of this
In addition to convergent tectonics (plate was komatiite and associated basaltic volcanism
or flake), it also appears that in some regions (Campbell and Hill, 1988; Campbell et al., 1989).
(e.g., Pilbara Craton, Western Australia, or Dhar- As depicted by Campbell and Hill (1988), the
war Craton, India) the ductile thermal regime final tectonic effects on the crust are similar to
in the crust and lithosphere allowed buoy- those proposed in the diapiric tectonic model
ancy driven granite diapiric tectonics (Choukrone (Fig. 4.4).
et al., 1997). The alternative way of viewing Komatiites are unusual igneous rocks. They
this is as gravitational sinking (‘‘sagduction”) of are marked by very low SiO2 (<45 wt.%), and high
dense, supracrustal flood lava-like greenstones MgO (>18 wt.%) contents. They are thought to
into older, relatively thin and ductile continen- have been erupted as high temperature (1450–
tal type crust which then squeezes up diapir- 1650 ◦ C), and low viscosity (0.1 –? 5 Pa s) lavas
ically (Chardon et al., 1998) (Fig. 4.3). Such tec- capable of high velocity, turbulent flow and of
tonic domains are marked by multiple, relatively thermally eroding their substrate (Huppert et al.,
low density, granitoid or gneissic domes which 1984). Komatiite lavas also preserve spectacular
have intruded, and are surrounded by, concentri- crystallization textures, including needle-like and
cally deformed, denser, volcanic and sedimentary plate-like olivine and clinopyroxene spinifex tex-
greenstone belts or structural basins. No linear- tures, thought to result from rapid chilling of
ity or polarity exists, and the dome and basin magma, for example in the surface crust of a flow-
relationship between granites and greenstones ing lava. They also preserve spectacular cumu-
suggests that gravitationally controlled tectonic late olivine textures, thought to result from the
processes dominated by buoyant granitoid diapirs slower cooling, crystallization, and settling of oli-
and their deformational effects on the green- vine crystals in the interior of a lava, much of
stone belts occurred in these terranes, not plate it probably after the lava had stopped flowing.
tectonic processes (see also Bleeker, 2002). Although komatiites are most commonly preser-
Since modern felsic magmas and basalts can ved in Archean rock successions, occurring in
be found in continental crustal settings, and in two main pulses of activity globally at 3400 and
both subduction and rift settings, and with a 2700 Ma (Arndt, 1994), some are also known in
wide range of geochemical characteristics, their Proterozoic successions, although these are
tectonic significance is also equivocal. Since it mostly komatiitic basalts, and minor occurrences
appears that mantle convection was more vig- are as young as Cretaceous (e.g., Gorgona Island,
orous in the Archean than today, rifting and Mexico). Komatiite lavas are known from all the
rift related volcanism would have been com- major Archean cratons of the Earth. Cratons are
mon. Felsic volcanics cannot therefore be taken areas of continental crust which have been tecto-
as being indicative of subduction settings; they nically stable for a long period of time, usually
could equally be rift related, in the dome and one or more eons long. Archean cratons are the
basin terranes, they could be related to granite nuclei of the large present-day continental land
diapiric intrusions, and they could also be related masses and are thought to have resulted from
to melting caused by plume activity. crustal accretion processes during the Archean
In addition to magmatism associated with and have been largely tectonically stable ever
convergent, extensional and granite diapiric tec- since.
tonics, it also seems that ultramafic and mafic Komatiites represent a unique type of volcan-
plume tectonics was an important process dur- ism in Earth history. Their source is thought
ing the Archean, and probably since then to the to have been in the deep mantle, and probably
present day. Modern hotspot activity which forms involved as much as 20% melting of the source
oceanic islands, basaltic oceanic plateaus, and mantle region (Arndt, 1994). They are generally
PRECAMBRIAN VOLCANISM AND TECTONIC SETTINGS 141

Sea Level uncontaminated by upper mantle and crustal


(a) Primitive Continental Crust rocks, indicating rapid uprise to the Earth’s sur-
face. This has given rise to the proposal that
Lithosphere
komatiite magmas rose as huge mantle diapirs
or plumes (Campbell et al., 1989). The reason for
Hot Asthenosphere their occurrence mostly in the Archean rock sys-
tem is not clear, especially the occurrence in two
distinct pulses at 3400 and 2700 Ma. It presum-
Sea Level
(b) v v v v v v v v v v v v v v v
v
v v v v v v ably relates to the high thermal regime and the
v
enhanced levels of mantle convection during the
Archean. Since their age coincides with the last
stages of the major meteorite–asteroid bombard-
ment phase in the history of the solar system,
their origin is perhaps related to such an event,
(c) Sea Level or several such events. Apart from the outpouring
v v v v v v v v v v v v v v v v v v v v v v v v
v v v v v v v v of regionally extensive komatiite and associated
basalt lava sheets, it is unclear whether komatiite
Zone of Crustal Melting
events and their host plumes caused any signif-
icant crustal deformation, either extensional or
compressional. It is unclear if komatite volcan-
ism coincided with the formation of extensional
basins, which would certainly have enhanced the
(d) uprise and eruption of the magmas, as discussed
Sediments
Gravity Slide
Sea Level previously. Plume tectonics may thus be rep-
v v
v v v
v v
v v
v v
resented only by voluminous volcanic outpour-
v v
v v v v v
v v v
v v ings, but it may also be related to crustal defor-
Melt Zone mation, especially crustal extension, or impact
events.
(e) v v v v
v v v v vv
v v v
v v
v It therefore seems unclear whether or not
v v v v v
v v v v
evidence of normal, modern plate tectonic set-
Melt Zone tings is preserved in rock systems of Archean
age (see also Bleeker, 2002). The debate about
v v Basalt Gneiss the occurrence of ophiolites, and the absence
Sediments and felsic volcanics Gabbro of high pressure blueschist metamorphic facies
Ancient andesitic crust Granite
casts doubt on whether ‘‘normal” subduction
Basaltic melt
occurred, although some form of convergent
tectonics certainly did. There is clear evidence
Fig. 4.4. Sequence of cartoons depicting the uprise of an (voluminous magmatism) for extensional tecton-
Archean mantle plume and its effects on the crust and the ics, implying extensional stress fields and signifi-
structure of greenstone belts. (a) Generation and uprise of a
cant rifting activity. The abundance of granites
mantle plume. (b) Plume spread laterally in upper mantle
suggests extensive areas of continental crust,
beneath the crust and generating plume-derived flood
volcanism and extensional tectonics. (c) Partial melting of especially in the late Archean. Unusual granite
lower crust by regionally high heat flow and underplated dome and greenstone basin belts also suggest
plume-derived magmas. (d) Diapiric uprise of lower crustal that gravitationally controlled diapiric or sag-
magmas to high levels in the upper crust and initial duction tectonics occurred in some regions of
supracrustal gravitational deformation effects. (e) Lateral the Earth’s crust. Also, pulses of voluminous
spreading of diapirs in the upper crust and peripheral komatiite and associated basalt volcanism clearly
compressional shortening or deformation of upper crustal indicate dynamic, deep-seated mantle plume
strata. From Campbell and Hill (1988).
influenced tectonics was also important.
142 VOLCANOES AND THE GEOLOGICAL CYCLE

The settings for Proterozoic (2500–600 Ma) vol- there are many Proterozoic rift zones not asso-
canism can also be difficult to establish. How- ciated with subduction zones. There is therefore
ever, there appears to be greater agreement in the a clear case for intracontinental and intraplate
literature that a more clearly defined plate tec- orogeny, indicating that magmatism associated
tonic regime existed during the Proterozoic eon with fold and thrust belts does not necessarily
(Condie, 1992). This is supported by the strong have to occur at plate margin subduction zones,
evidence for Proterozoic ophiolites from 2000 an important lesson for other ancient orogenic
Ma (Helmstaedt and Scott, 1992), and the exis- belts, including younger Phanerozoic ones.
tence of accretionary prism-like fold and thrust It is also important to stress that this principle
belts, as well as high pressure blueschist meta- applies not just to Proterozoic cases. Even in the
morphic assemblages from at least 1800 Ma modern world there are many examples of intra-
onwards (Choukrone et al., 1997), indicating per- continental rift terrains including narrow rift sys-
haps deeper subduction of more rigid litho- tems such as the East Africa rift zone, the Rhine
spheric plates. Volcanic rock successions also Graben and North Sea rift systems of Europe,
appear to be generally similar counterparts, and and the Baikal rift of Russia (Olsen, 1995). A very
numerous studies clearly demonstrate that pale- broad extensional and magmatically active rift
oenvironments were similar to today, although terrain is of course the Basin and Range Province
some exceptions are indicated by the extensive of the United States. In these settings if regional
banded iron formations which extended from the stress fields changed from extensional to com-
Precambrian to about 1800 Ma. pressional or transpressional, deformation would
Such a brief summary gives the impression most easily be taken up by pre-existing normal
that the Proterozoic tectonic and volcanic world faults formed during extension, and so causing
was similar to today’s and was dominated by large basin inversion.
rigid plate interaction as today, with all the usual The concept of intraplate orogeny is not new
plate tectonic settings for volcanism and defor- (Park and Jaroszewski, 1994; Neil and Houseman,
mation. 1999). Neil and Houseman (1999) suggest that if a
However, such a simple assessment has to depression of the base of the lithosphere occurs
be approached with considerable caution. First, into the asthenosphere then even a slight lat-
there are extensive Proterozoic orogenic or fold eral compressive stress transmitted through the
and thrust belts that do not appear to have lithosphere could cause the dense lithosphere to
involved subduction. They originated as intracon- downwell due to Rayleigh–Taylor instability that
tinental rifts, never developed to the break-up, develops at the boundary. Such downwelling will
spreading and oceanic crust stage, and therefore cause shortening of the crust, leading to thick-
were not subjected to the tectonic effects of sub- ening and uplift. If detachment of the down-
duction, but were subjected to major regionally welling mass has occurred, the crust adjusts by
extensive orogenic deformation and metamor- extending. Extension and compression can then
phism. O’Dea et al. (1997) document the evolution be achieved.
of the Mt. Isa Rift and Fold Belt of northern Aus-
tralia, an extensive mid-Proterozoic intraconti-
nental rift and fold belt terrain, that experienced
multiple episodes of rift basin formation, vol- Volcanism, the magma cycle, and the
canism, sedimentation, deformation, and uplift, growth of the continental crust
culminating in the compressional Isan Orogeny
between 1590 and 1500 Ma. The Mt. Isa Rift It is appropriate to return to the relationship
and Fold Belt is not an isolated example in the between volcanism and the geological cycle to
Proterozoic. Etheridge et al. (1987) noted that assess how volcanism has contributed to the
none of the Proterozoic orogenic belts of Aus- growth of the continental crust throughout geo-
tralia show evidence for spreading, oceanic crust, logical time. Although there has been consider-
and subduction. Green (1992) notes that globally able debate on the nature of the Earth’s early
VOLCANISM, THE MAGMA CYCLE, AND THE GROWTH OF THE CONTINENTAL CRUST 143

Orogeny, Continental rift, Mid-ocean ridge, Subduction Fig. 4.5. A schematic


accretion intraplate continental, back-arc basin, representation of the magmatic
continental arc island arc,
volcanism cycle showing where magmas are
oceanic island
volcanism generated and the possible paths
that those magmatic products may
Burial, Partial follow within and on top of the
metamorphism, melting Earth as part of the larger-scale and
anatexis
long-term geological cycle
throughout geological time.

Crustal + Heat Mantle


magmas magmas

crust and how much continental versus oceanic mantle wedge above subduction zones. If the arc
crust there may have been, it seems clear that is on a continental margin, then again there is
throughout geological time there has been a net direct transfer of mantle magmas to continental
transfer of matter from the mantle to the conti- crust. Mantle magmas also contribute directly to
nental crust through the activity of the magma the building of continental crust through volcan-
cycle as part of the geological cycle. ism in continental rift zones and in intraplate
The magma cycle (Fig. 4.5) involves the man- continental hotspot volcanic provinces. Even if
tle as the ultimate source for all magmas. Man- rifting is successul to the spreading stage, some
tle magmas can take several possible paths. First mantle-derived volcanics and intrusions will be
and most obvious is eruption as lavas at mid- incorporated into the crust of the newly formed
oceanic spreading ridges, forming seafloor crust, rifted continental margin. If rifting is aborted,
whereafter that crust is convected to subduction then the volcanics become part of the upper
zones. It then undergoes partial melting, which crustal succession, emplaced there as part of the
contributes to volcanic arc magmas. The remain- rift valley volcanic succession.
der is reassimilated into the mantle and becomes In all of these cases, only a small part of the
part of the long-term mantle magma cycle again. rising mantle magmas reach the Earth’s surface.
Other mantle-derived magmas contribute to Much of the rising magma is trapped at the base
oceanic island and oceanic plateau volcanism of the crust and is added to it from underneath
and edifices, which are also ultimately trans- in a process called underplating, or is emplaced
ported into subduction zones as part of downgo- as intrusions within the crust.
ing recycled crust and lithosphere. Some oceanic Magmas resulting from partial melting or
island/plateau edifices are completely recycled anatexis of continental or evolved arc crust form
back into the mantle, but some may be decapi- after input of heat from the mantle either
tated in or collide with the subduction zone and through direct conduction or through heat trans-
become accreted to the overriding plate, as part fer from the passage of mantle-derived aqueous
of the accretionary prism and crust of the arc fluids or magmas through the crust. These mag-
system of the overriding plate. Similarly slices mas rise, and are either emplaced as intrusions
of oceanic lithosphere (ophiolites) may become higher in the crust or are erupted on the Earth’s
accreted to the arc crust and accretionary prism. surface, leading to the surface build-up of the
In this way, mantle-derived volcanics are directly continental crust.
and physically incorporated into arc crust, which Volcanic systems exposed subaerially on the
ultimately becomes accreted to continental crust Earth’s surface are weathered and eroded, and
through collison, if in fact the arc system is not the products are ultimately deposited in sedimen-
already a part of a continental margin. tary basins, where they are buried, deformed,
Mantle magmas may also contribute directly and incorporated into the crust. Eroded sediment
to arc volcanoes if they are derived from the which is deposited at the continental shoreline as
144 VOLCANOES AND THE GEOLOGICAL CYCLE

part of a growing continental margin shelf–slope the Late Paleozoic onwards would this sediment
complex may simply be buried and become part flux rate have declined or stabilized.
of the continental margin crust, or may at some
later stage become involved in collision in a sub-
duction zone, and then becomes part of the inte- Magmatic cycles
rior collision zone crust of a larger, amalgamated
continental mass. A fundamental task of science is to recognize
Clearly then volcanic products derived from patterns, because patterns represent activity that
the mantle have been contributing to the growth is repeated and suggest there is an underlying
of continental crust throughout geological time scientific cause or mechanism that is systematic
through the magma cycle and its role in the in its course and has natural in-built ‘‘memory”
larger global geological cycle. Various attempts or a feedback process. There have at times been
have been made to evaluate the rate of growth attempts to identify magmatic cycles in Earth
of continental crust through geological time. For history, or regular patterns, and then seek the
example, McLennan and Taylor (1982) used iso- underlying tectonic or tectonomagmatic cause.
topic fractionation trends in shales of various This approach has often been attempted with
ages to evaluate the relative roles of primitive Archean volcanic successions with proposals that
mantle source components relative to recycled magma cycles begin with voluminous mafic and
supracrustally eroded sources. They found that ultramafic magmatism and conclude with felsic
through most of the Archean, primitive sources magmatism (e.g., Hutchinson, 1973).
predominate, but that from the beginning of the Barley et al. (1998) have proposed that there
Proterozoic there is an increasing input from have been specific periods in the Earth’s history
recycled ‘‘mature” continental crustal sources. when there were major increases, or cycles, of
This suggests that areas of continental crust magmatism, tectonic activity, and the formation
developed during the Archean by a combination of ore deposits. They cite the interval from about
of arc terrane accretion at Archean-style subduc- 2.8 to 2.6 billion years ago as one of these cycles
tion or accretion zones, and by rift related and and suggest that it was related to the break-up
intraplate magmatism, leading to the develop- of an Archean supercontinent. They draw ana-
ment of Archean cratons. From the Proterozoic logy with a younger magmatotectonic cycle in the
onwards larger continents existed, shedding more Mesozoic that coincided with the break-up of the
sediment into the surrounding oceans than dur- Gondwana and Laurasian supercontinents.
ing the Archean. In addition, a more structured On a global scale there is currently constant
plate tectonic regime appears to have developed volcanic activity. Seafloor lavas are constantly
leading to arc–arc, arc–continent, and continent– being erupted at mid-oceanic ridges, and if we
continent collisions, and growth of continents by accept that in the past, when the Earth’s interior
accretion and collision. Rift and intraplate hot- was hotter than now, mantle convection must
spot magmatism would have continued to contri- have been more active than today, then it is likely
bute to growth of continents. Additionally, under that there has always been constant seafloor
a regime of increasing atmospheric oxygen, and volcanism throughout geological time. However,
therefore increasing rates of weathering and ero- from time to time, spreading rates and magma
sion, increasing rates of sediment input to conti- production rates may have changed, depending
nental margin shelf and slope systems. The pro- on the total length of spreading ridge axes on
cess would have become increasingly more signif- the Earth’s surface and convection rates.
icant as the surface area of landmasses exposed The most likely settings where cyclicity of
above sea level increased, therefore leading to magmatism would occur would be in plate mar-
increased surface area subject to weathering and gin arc related subduction settings and continen-
erosion and increased sediment flux rates to the tal rift settings. At subduction settings magma
continental margins and oceans. Only when veg- activity may vary depending on events in the
etation became widespread on landmasses from subduction system. For example, during normal

You might also like