The increasing exposure of cities to the effects of volcanic eruptions: a global survey

Environmental Hazards 2 (2001) 89}103

The increasing exposure of cities to the e!ects of volcanic eruptions: a global survey D.K. Chester *, M. Degg
, A.M. Duncan, J.E. Guest Department of Geography, University of Liverpool, Roxby Building, P.O. Box 147, Liverpool L69 3BX, UK
Department of Geography, Chester College of Higher Education, Parkgate Road, Chester CH1 4BJ, UK Centre for Volcanic Studies, University of Luton, Park Square, Luton, UK Department of Geological Sciences, University College London, Gower Street, London, WC1E 6BT, UK

Abstract The most dynamic demographic process of the past 250 years has been the movement of people from rural areas to cities. For most of this period urbanisation has been concentrated in economically more developed parts of the world, but during the last 50 years the focus has shifted to economically less developed regions. Urbanisation, particularly in developing countries, has led to increasing global exposure to a variety of natural hazards, not the least of which are risks posed to large cities by volcanoes. In this paper we monitor these demographic changes and detail the various types of volcanic hazard to which cities are exposed. A major eruption a!ecting a city in a developing country could cause widespread loss of life and regional disruption. E!ective response, however, might minimise casualties in a city within a developed nation a!ected by a major eruption, but the economic impact could have global consequences. We argue that global hazard exposure is often subtle and involves not only the size of a city and the types of volcanic product that may occur, but also the strategic position of the threatened city within the economy of a country and/or region and the fact that volcano-induced tsunami and other consequences of eruptions, such as climatic change, may a!ect cities far removed from a given eruption site. Mitigation measures informed by both speci"c prediction (surveillance) and general prediction (hazard mapping) are providing the potential to reduce hazard exposure. The paper concludes with a consideration of ongoing research, in particular the emphasis currently being placed on con#ating hazard analysis with studies of place, economy, society and culture.  2001 Elsevier Science Ltd. All rights reserved. Keywords: Volcanoes; Hazards; Cities; Urbanisation; Eruption prediction; Developing countries

1. Urbanisation and escalating global exposure to hazards The movement of large numbers of people from rural areas to cities has been one of the most dynamic demographic processes of the last 250 years, heralding massive socio-economic and environmental changes that a!ect the day-to-day living of the vast majority of people on the planet. From 1750 onward, migration was closely linked to industrialisation in Europe and North America, but during the second part of the 20th century the focus of world urbanisation shifted to the developing world where rates of natural increase, rural depopulation, and urban

* Corresponding author. Fax: #44-151-794-2866. E-mail addresses: [email protected] (D.K. Chester), [email protected] (M. Degg), [email protected] (A.M. Duncan), [email protected] (J.E. Guest).

growth during the last 50 years have far outstripped anything seen in more developed countries (Jones and Kandel, 1992). Roughly 50% of the world's population now lives in urban areas, but this is expected to rise to 60% by the year 2025 (Domeisen and Palm, 1996). These "gures may be compared to an urban population in 1950 of just under 30%. The shifting focus of global urbanisation from developed to developing countries is emphasised by the fact that only seven of the world's 20 largest cities were in developing countries in 1950, but this "gure has now risen to 17 out of 20 (Fig. 1A). By 2025, 80% of the world's urban population will reside in developing countries (Domeisen and Palm, 1996). There are many rami"cations of this dramatic change, especially in terms of societies' increasing vulnerability to natural hazards (Mitchell, 1999). Hazards, which are de"ned as volcanic phenomena that pose a potential threat to people or property in a given area within a given

1464-2867/01/$ - see front matter  2001 Elsevier Science Ltd. All rights reserved. PII: S 1 4 6 4 - 2 8 6 7 ( 0 1 ) 0 0 0 0 4 - 3

90

D.K. Chester et al. / Environmental Hazards 2 (2001) 89}103

Fig. 1. (A) Cities of the world with over 1 million inhabitants, showing the 20 largest metropolitan areas in 2000 (compiled from UNFPA, 1993, United Nations 1996 and additional sources). (B) Active volcanoes of the World (information from Chester, 2000 with additions and amendments). A more detailed picture is provided by Simkin et al., 1994a, b.

period of time, continue to exact disproportionately high death tolls on developing countries (Fig. 1B). This is because risk, the probability of losses (in particular to life),

is exacerbated by increasing human vulnerability (Tilling, 1989, p. 241, see also Susman et al., 1983). Annual natural disaster related deaths have averaged nearly 24,000

D.K. Chester et al. / Environmental Hazards 2 (2001) 89}103

91

Fig. 2. Economic and insured losses from natural disasters: 1960}1998 and projected to 2000 (based on Munich Re., 1999, p. 5).

during the 20 years from 1977 to 1997 and it is estimated that around 90% of these have occurred in developing countries (Alexander, 1997). In contrast, huge economic losses have occurred in developed nations, accounting for some 82% of the global total during the period 1977}1997 (Alexander, 1997) (Fig. 2). Although the absolute (i.e. monetary) value of losses in developing countries is much smaller than in developed ones, the relative impact is much greater. The World Bank estimates that when losses are related to a country's gross national product (GNP), and for events of a similar magnitude, costs are often twenty times greater in a developing country than a developed one (Anon, 1994, p. 37). Major disasters in poor countries may also wipe out any development gains for many years (Harriss et al., 1985; Kates, 1987). High rates of natural population increase and wholesale urbanisation are widely considered to be important factors contributing to this trend because of their e!ects on: (1) concentrating people and economic investment at particular points on the Earth's surface, many of which are exposed to a variety of natural hazards; and (2) diminishing public awareness of environmental hazards and the ways in which they might be responded to. Densely packed urban concentrations of population and economic value in hazard zones present greater potential for catastrophes than dispersed rural populations. The most extreme examples of this come from the developing world, where the phenomenal rates of urbanisation and associated industrialisation witnessed in recent decades have often led to the over-concentration of people and investment in one or two major conurbations: the so called primate cities (Table 1). These cities dominate their respective countries, but have catastrophic po-

tential when located in areas exposed to destructive natural events. The risk of human losses is accentuated by overcrowding which forces people*often the poorest of urban dwellers*to live in marginal areas that are not suited to development, such as along drainage channels and #ood plains susceptible to inundation. The pressure of numbers means that city authorities are often unable to provide even the most basic infrastructure and services, forcing people to live in squatter conditions and thus enhancing human vulnerability. Unfortunately, this is often the rule rather than the exception in the largest cities in the developing world; where 30}60% of urban dwellers often live in densely populated squatter settlements (Domeisen and Palm, 1996). Risk of economic losses is further exacerbated by the inherent vulnerability of much of the modern urban infrastructure and by the growing dependency of modern industry and commerce upon a perfectly working infrastructure. Even a relatively minor hazard causing power failure in a city can now lead to severe urban/industrial disruption, and an enormous loss of revenue and pro"ts with considerable socio-political after-e!ects. Major hazards a!ecting regional or national urban centres can cause losses of such proportions that national economies are set o!-course for decades. For example, the 1972 Managua earthquake, which killed 5000 people and caused economic losses of $US2.0 billion, represented around 40% of Nicaragua's GNP (Coburn and Spence, 1992). Managua is also located within 20 km of two active volcanoes: Masaya Nindiri and Apoyeque. Unfortunately, the disaster of 1972 does not appear to have altered the reconstruction of the city. Despite subsequent attempts at decentralisation, Managua remains a classic example of a primate city in a developing country: the national capital; centre of industry; and business and home to 42% of Nicaragua's urban population. Increasingly, there is concern about the possible global economic impact of a serious hazard on a &world city'. For

92

D.K. Chester et al. / Environmental Hazards 2 (2001) 89}103

Table 1 Some primate cities of developing world (based on various sources) Country

Urban population 1995 (millions)

Average annual growth rate (%) of urban population 1995}2000

Argentina Bolivia Brazil Chile Colombia Ecuador

30.5 4.5 126.6 12.0 25.5 6.7

1.5 3.8 2.3 1.6 2.2 3.1

El Salvador French Guiana Guatemala Guyana Mexico Paraguay Peru Surinam Uruguay Venezuela UK USA

2.5 0.1 4.4 0.3 70.2 2.6 17.2 0.2 2.9 20.3 52.1 200.7

#2.0 4.4 &3.0 2.9 &2.0 3.9 2.5 2.5 0.8 2.4 0.4 1.2

example, if the 1995 Kobe earthquake in Japan, had occurred in the Greater Tokyo region, estimated losses would have been around 1}3 trillion $US (Had"eld, 1995; Munich Re., 1997). The e!ects of urbanisation in reducing public perception of natural hazards are largely related to lifestyle. Rural communities are, generally speaking, closer to the natural environment than their urban counterparts because their livelihoods depend upon successful interaction with it. Stimuli from the environment that might warn of potential danger from hazards (e.g. marshy ground de"ning a low-lying area susceptible to inundation, or changing gas emissions and/or felt earthquakes from an active volcano) are readily perceived and there is a strong motivation to respond to them. In contrast, the built environment serves to smother natural stimuli and replace them with human-made ones, such as noise, pollution, and building skylines that obscure the natural horizon. All too often, engineered defence schemes are implemented to protect cities from minor to moderate events (e.g. leveH es to contain the one in 50-year #ood), thereby creating an unwarranted sense of absolute security and further reducing hazard perception and the motivation to respond (see Stoddart, 1987, p. 332, for an example from Bangladesh). Rural dwellers also tend to be more self-reliant and less dependent upon infrastructure than their urban counterparts. Rural communities often demonstrate strong resilience in the face of environmental extremes, based upon indigenous mitigation practices and strategies of coping that have evolved through generations, such as safe site

Largest urban centre(s)

% of total urban population in the largest centre(s), 1995

Buenos Aires La Paz Sa o Paulo Santiago BogotaH Guayaquil Quito San Salvador Cayenne Guatemala City Georgetown Mexico City AsuncioH n Lima Paramaribo Montevideo Caracas London New York

36.1 7.7 13.0 42.3 22.0 25.6 18.6 60.8 36.6 26.1 78.1 21.4 24.1 43.4 46.9 46.1 14.6 14.1 8.1

selection and resistant building practices (Aysan et al., 1995). All too often, this so called pre-industrial or folk response (White, 1973) becomes diluted and lost through the process of urbanisation, as people move to new, unfamiliar areas and old community and family ties are broken. On Mount Etna, in Sicily, well into the present century, rural communities coped with lava inundations using a mixture of indigenous procedures (Duncan et al., 1981, 1996; Chester et al., 2001). These included: farmers holding land in widely separated plots; assistance from members of extended families; and systems of agricultural practice that had evolved over millennia and were geared towards bringing land back into cultivation as quickly as possible. Traditional practices are often viewed as &unfashionable' in the urban context, where individuals prefer to abdicate responsibility for dealing with hazards to municipal authorities and/or national governments. Large cities &constitute extensive areas of highly concentrated vulnerability to catastrophes' (Alexander, 1997, p. 293) and as the Munich Reinsurance Company conclude `there is a direct correlation between catastrophe risk and the number of cities exposed to hazardsa (Munich Re., 1997, p. 10). In the 1970s and 1980s, much of the debate surrounding global urbanisation and natural hazards focused on the so-called megacities (Anon, 1995); cities with populations of over eight million, several of which were projected to reach 40 million by 2030. Data now show that most of these cities grew considerably slower in the 1980s than in earlier decades, so that the world is now less dominated by megacities than many were once predicting.

D.K. Chester et al. / Environmental Hazards 2 (2001) 89}103

93

Fig. 3. Number of cities with 1 million or more inhabitants: 1950}2015. Cities are not necessarily o$cial administrative units, but are de"ned as urban agglomerations of continuous territory inhabited at urban level of residential density (based on information in United Nations, 1997).

Only around 5% of the world's population currently lives in agglomerations of this type, and forecasts of the number of megacities likely to exist in the early part of this century have been revised downwards to 33 by 2015 (United Nations, 1996). What is also interesting is the changing distribution of these cities: six in developed countries (no change from 1970), three in Africa (none in 1970), "ve in Latin America (three in 1970) and 19 in Asia (two in 1970). There has been some reshu%ing in the ranking of the largest cities, with those in Latin America growing slower than anticipated, and those in Asia much faster (Jones and Kandel, 1992, p. 73}75). The world has not been spawning megacities at the rates anticipated, yet the number of cities with over one million inhabitants has risen far more sharply than expected, with particularly explosive growth in developing countries from the early 1970s (see Fig. 3). Again there are large regional di!erences, with 50% of the world's million-plus cities projected to be in Asia by 2015, 14% in Latin America, 13% in Africa, 13% in Europe, 9% in North America and 1% in Oceania (United Nations, 1995). Most major port cities and traditional industrial centres in the developed world were among the world's largest cities during the 1850s}1950s, but experienced declining populations during the latter part of the 20th century. Conversely, Asia experienced phenomenal urban growth linked to a greatly increased role within the world economy during the last three decades.

2. Urban trends and volcanic hazard exposure It has been argued by some that volcanic eruptions do not present a serious hazard to humankind (Wijkman

Table 2 Estimated number of people a!ected, but not killed, by natural disasters 1980}1990 (based on Chester, 2001 and information in UNESCO, 1993, p. 5, with modi"cations and amendments) Disaster type

Approximate numbers a!ected (thousands)

Numbers a!ected (&% of total)

Droughts Floods Windstorms Earthquakes Landslides Volcanic eruptions Wild"res Tsunami Total

952,200 524,600 150,300 28,400 3100 620 610 &1 1,659,831

57 32 9 2 0.2 0.04 0.04 0.0001

and Timberlake, 1984, p. 100), and at "rst sight there appears to be some justi"cation for this position. It is likely, for instance, that the number of people killed by the 1976 Tangshan earthquake in China greatly exceeded the o$cial estimate of 250,000 (Tilling, 1989; Bolt, 1993); yet even if this "gure is accepted, it is still greater than the 220,000*an average of 1000 per annum* who have died directly and indirectly from the e!ects of volcanism since the 1783 Laki "ssure eruption in Iceland (Tanguy et al., 1998). If the "gures for the numbers of people a!ected by, but not killed in, natural disasters for one decade are compared (Table 2), then the impact of eruptions appears to be minimal. There are two reasons why caution is required when interpreting these data. In the "rst place, it is only a matter of chance that casualties have not been higher. Several major eruptions in the 20th century (e.g. Katmai, Alaska 1912; Bezymianny, Kamchatka Russia 1955/56) and earlier have

94

D.K. Chester et al. / Environmental Hazards 2 (2001) 89}103

occurred in regions with low population densities. In reviewing major volcanic disasters, Tilling and Lipman (1993) noted three episodes of &restlessness' at major calderas*Long Valley (USA), Campi Flegrei (Italy) and Rabaul (Papua, New Guinea)*that could eventually culminate in large eruptions. Indeed, Rabaul did erupt in 1994, displaying both strombolian and vulcanian activity. An eruption at any one of these calderas today would have catastrophic implications for the concentrations of people that have grown up around them. The size of a volcanic eruption usually is measured by the volcano explosivity index (VEI). The index is logarithmic and runs from 0 to 8, and is based on the volume of material ejected and the height of the eruption column (Newhall and Self, 1982). Major volcanic hazards are normally associated with explosive eruptions, which have high VEI values. Events with a VEI of 3 and below usually involve lava #ows and/or minor explosive activity and are generally localised in their e!ects. Events of VEI 4-5 often disrupt regional economies, while eruptions of over 6 may a!ect the whole earth because of their impact on global climate. The Mount St. Helens eruption in the USA in 1980 had a VEI of 5, while Pinatubo in the Philippines in 1991 had a VEI value of 6, one of the largest volcanic events of the 20th century. The most recent magnitude 7 event was the 1815 eruption of Tambora in Indonesia, which led to huge loss of life, famine, disease, short-term climatic change and crop failure in North America and Europe (Rampino and Self, 1992). Today an event of this magnitude in the vicinity of a large urban area would be devastating for the city in question, could cause disruption to the economy of an important economic region, and have e!ects throughout the world, including signi"cant global climatic changes. It is estimated that even a modest eruption at Campi Flegrei, adjacent to Naples in Italy, would put between 200,000 and 400,000 people at risk (Barberi and Carapezza, 1996). An eruption at Campi Flegrei, such as that which emplaced the Campanian ignimbrite about 33,000 years ago (VEI 6/7), would devastate a region from Monte Cassino in the north to Salerno in the south. An event of this size, or even a smaller eruption such as the &Minoan eruption' of Santorini around 3600 years ago, would have catastrophic economic consequences not only for the country concerned, but also for the entire European economy. VEI events of magnitude 8 are rare and there is no certain example from the Holocene, the last probably occurring at Toba volcano in Sumatra some 74,000 years ago. The recurrence intervals of some forms of volcanic activity are very long when compared with the history of human occupancy of volcanic regions. The Long Valley and Yellowstone volcanic regions in the USA are, for instance, capable of producing large volume explosive eruptions (Wood and Kienle, 1990) and could devastate areas in excess of 30,000 km. Recently such volcanic

regions have received signi"cant scienti"c and media attention (Connor, 2000). Although repose periods of events from such volcanoes may be as great as 10 or even 10 years, e!ects would not only a!ect the region in question, but could also have signi"cant and deleterious impacts on the world economy and global weather. Similar considerations of long recurrence, when compared with the relatively short span of human occupancy, also apply to potential hazards posed by tsunami generated by: violent explosions; landslides into the sea and/or collapses of volcanic edi"ces (Tables 3 and 4). Indeed the possible e!ects on human a!airs of volcanic events with long recurrence intervals is currently a fashionable research frontier, being evoked to explain both several civilization changing events in history (e.g. Keys, 1999) and the putative impacts of so called mega-tsunami in the future (e.g. McGuire, 1999). The unfortunate history of uncritical environmental determinism in the 20th century must, however, be kept in mind (Peet, 1998, p. 12}15). A second reason why historic data on hazard losses fail to capture the present and future dangers of volcanic eruptions is the process of urbanisation. High rates of urban population growth in developing countries are a cause of great concern because many cities are located astride active tectonic belts, which contain the vast majority of the world's most explosive volcanoes. In fact volcanoes are the major hazard facing a growing number of large and small cities around the world and the contemporary switch in focus of world urbanisation has probably impacted more on volcanic risk than on any other type of natural hazard. More than 500 million people are currently at risk from volcanic hazards (Tilling and Lipman, 1993). It is not only the dangers of death, economic disruption, and climatic perturbation that make volcanoes dangerous places to live alongside, there is also increasing evidence that some active volcanoes present a persistent risk to health due to longterm exposure to carbon dioxide, radon and other pollutants (Baxter, 2000) (Table 3). Fig. 4A shows the location of a selection of the world's most highly exposed conurbations, plotted according to distance and direction from the nearest volcano(es). The "gure includes the megacities of Tokyo (25 million) and Mexico City (15.1 million), as well as a selection of smaller urban settlements, and is an updated and reworked version of a diagram "rst produced in 1984 by the Munich Reinsurance Company (Munich Re., 1984). Fig. 4B indicates the theoretical distances over which volcanic processes and products may adversely a!ect human populations. When comparing these diagrams with the map of million cities (Fig. 1A), it is clear that volcanic hazard exposure*especially for developing countries*is both signi"cant and, on the basis of earlier discussion, increasing. In recent years it is merely a combination of statistical chance and good fortune that urban losses have not been greater.

D.K. Chester et al. / Environmental Hazards 2 (2001) 89}103

95

Table 3 Volcanic hazards and measures of mitigation (based on information in: Crandell et al., 1984; Chester, 1993; McGuire, 1998 and additional sources) Volcanic process

Characteristics and mitigation

Lava #ows

Lava #ows generally move slowly, along paths determined by topography and do not normally threaten life. Lava #ows cause destruction through burial and burning, and this may end all existing landuse, preventing it being re-established for many centuries. Mitigation measures include: general & speci"c prediction; damming and/or diversion and slowing the #ow using water-cooling. Domes are formed when magma is too viscous and immobile to #ow very far. Pyroclastic #ows may be generated by collapse (see below). Pyroclastic falls range in calibre from ash (less than 2 mm), lapilli (2 mm*(69 mm) to blocks/bombs (greater than 64 mm). Falls present a number of hazards. Near to the volcano, failure of roofs, power lines and cables and death and injury through the impact of large calibre blocks and bombs, may occur. Ash can destroy vegetation, crops, block roads, clog drains, watercourses and cause damage to equipment. In large eruptions, the dangers to aircraft may be considerable and particles can be spread over the globe and impact on climate and weather. Mitigation measures include: general and speci"c prediction; pre-eruption evacuation; measures to ensure that the roofs of buildings are of su$cient strength; provision of face masks and uncontaminated water; and warnings to air tra$c. Lateral blasts are highly destructive. Blasts kill by heat, burial and impact. Blasts are caused by decompression of magmatic gases, or explosion of high pressure hydrothermal systems. Blasts involve ballistic action from ashes, blocks and bombs, and may also include pyroclastic surges and #ows (see below). Blasts often cause the partial destruction of the volcanic edi"ce. Examples include: Bezymianny (Russia, 1956) and Mt. St Helens (USA, 1980). For mitigation measures see pyroclastic falls, #ows and surges. Volcanic gases present hazards to the health and may damage vegetation. The e!ects of gases are at their most severe near to a volcano and wind directions are important in determining distribution. In large eruptions, gases may have an impact on global climate. Mitigation measures include: general and speci"c prediction; re-settlement if persistent, and pre-eruption evacuation if transient. Pyroclastic #ows are hot dry masses of particulate volcanic material that move along the ground surface. Maximum temperatures in pyroclastic #ows range from 3303C to more than 5503C. Flows are mobile and travel at speeds of over 100 km h\. On eroded volcanoes, #ows are restricted to valleys if thin, but on lowlands may be deposited as fans. Flows are very hazardous. Flows of modest size were associated with Krakatau (Sunda Straits, 1883) and Katmai (Alaska, 1912) and are relatively low frequency events, but if one did occur in a densely populated area, then it would cause one of the greatest of human disasters. Around many volcanoes there are thin deposits that cover large areas. These are thought to be emplaced by very high velocity #ows, which can surmount high mountain barriers. Even smaller #ows, of which there are many historic examples (e.g. Mont PeleH e and La Soufrie`re*Caribbean 1902), constitute a very serious hazard and would cause death and destruction to all in their paths. Mitigation measures include: general and speci"c prediction; judicious siting of settlements and pre-evacuation. A surge is a turbulent, low density cloud of gases and rock debris, perhaps accompanied by water and steam, that moves at high speed. Two sub-types are recognised: (a) Hot surges are very dangerous and can cause death, injury, destruction of buildings, impact damage and burial. Hot surges move at great speed and may be formed by a number of processes (e.g. explosive disruption of domes, collapse of domes, collapse of eruption columns and lateral blasts); (b) Cold surges are associated with hydrovolcanic eruptions and are produced by vertical explosions and by material falling from eruption columns. Temperatures are usually below 1003C. Normally surges are con"ned to within 10 km of their source. Surges kill and injure, destroy structures through burial and impact. Mitigation measures are similar to those for #ows. Pyroclastic #ows and pyroclastic surges may be regarded as the end members of a relatively continuous range of #ow types.

Domes Pyroclastic falls

Lateral (directed) blasts

Gases

Pyroclastic #ows

Pyroclastic surges

3. Theoretical and actual volcanic hazard exposure Although Figs. 4A and B give a theoretical picture of exposure, in reality global hazardousness is much more subtle. In the "rst place, it is not just the size of a settlement that is important in risk assessment, but its location and &strategic' position within the economy of a country or region. Iceland is a rich country in terms of its GNP per capita (27,000 $ US in 1997), yet because of its small

population (273,000), it has a modest national wealth. The 1973 eruption of Heimaey, which only a!ected the relatively small settlement of Vestmannaeyjar (population 5300, 2% of the national total), &cost' the Icelandic economy just over 2% of its GNP, but this represented an estimated 10% of average family income in extra taxes to fund recovery. By way of comparison the 1980 eruption of Mount St. Helens cost the economy of the United States only a small proportion of its national wealth

96

D.K. Chester et al. / Environmental Hazards 2 (2001) 89}103

Table 4 Volcano-related hazards and measures of mitigation (based on information in: Crandell et al., 1984; Chester, 1993; McGuire, 1998 and additional sources) Process

Characteristics and mitigation

Lahars and #oods

Flooding is often associated with volcanoes with crater lakes, and in situations where eruptions occur beneath snow and ice. When concentrations of sediment increase a Lahar may be generated. Using Lahar to include all sediment/water mixtures, e.g. mud#ows, debris #ows etc., then these may be generated in a number of ways. (a) Rapid melting of snow and ice. (b) By an avalanche, induced by an eruption, depositing material into a stream or lake. (c) By a lava or pyroclastic #ow moving into a stream or lake. (d) By a volcanic earthquake inducing slope failure and initiating mass movement. (e) By heavy rains on a volcano. (f) Sub-glacial eruptions. (g). Breaching of a crater lake. Lahars have in the past caused substantial damage and catastrophic loss of life (e.g. Nevado del Ruiz}Colombia*1985* over 20,000 deaths). Lahars may travel distances of up to 300 km from their sources at speeds of around 100 km h\ and are concentrated in and follow existing valleys and topographical depressions. Mitigation measures include: planning and evacuation based on general and speci"c prediction, judicious siting of settlements and sediment dams. Structural collapse of volcanoes to form calderas and major sector collapses are rare, but examples are known (e.g. San volcano (Japan, 1888) and Tenerife, at least 9 sector collapses and 3 caldera collapses are recognised). Tsunami may also be initiated if collapse takes place near to the sea. Collapses represent a high potential hazard, but one with a long recurrence interval. Debris avalanches are generated by the collapse of part of a volcanic edi"ce, e.g. Unzen (Japan, 1792). Run-out distances of studied avalanches range from less than 10 km to over 50 km. Mitigation measures include: general and speci"c prediction and pre-eruption evacuation. Some volcanoes in#ate before eruptions as magma is stored e.g. before the 1980 eruption of Mount St. Helens an elliptical area moved outwards at 2m/day for nearly six weeks. Magma intrusion at shallow depth without eruption (e.g. Campi Flegrei, Italy 1982}4) also causes ground in#ation. The major hazards are seismic e!ects on buildings and evacuation is often required. Volcano induced tsunami have not been a major cause of death in the twentieth century, but in earlier centuries their e!ects were catastrophic. Examples include: Unzen Volcano (Japan 1792); Krakatau (Sunda Straits between Sumatra and Java 1883) and Tambora (Indonesia 1815). Volcanoes in coastal locations may induce tsunami through a variety of processes which include: (a) violent explosion (e.g. Krakatau); (b) a landslide into the sea (e.g. Unzen) and (c) sector collapse. All volcanoes near to the sea have the capacity to generate tsumanis. Mitigation measures include those listed for collapse and tsunamis warning networks. Include: starvation; epidemic disease; contamination of water and land; drowning; transport accidents; exposure; cardiac arrests and the breakdown of law and order. Most of these are preventable given good civil defence.

Collapse

Deformation and earthquakes

Tsunami

Other hazards

(0.03% of GNP), even though its GNP per capita was at the time, and remains, lower than that of Iceland (Chester, 1993). The costs of an Icelandic eruption a!ecting a larger town or the capital Reykjavik*a classic primate city containing some 37% of the national population*clearly would be devastating. Other developed regions with signi"cant urban volcanic hazard exposure, relatively high GNP "gures per capita, but modest wealth include: the Atlantic Islands of the Azores and Canaries; the Greek islands and, to a lesser extent, New Zealand. In all cases potentially serious economic consequences are mitigated by other factors. Under anything but the most extreme, long-term and, therefore, most unlikely eruption scenario, volcanic hazards are regional problems and only pose a threat to part of a country. The Azores and Canaries are, respectively, autonomous regions of Portugal and Spain, and help would be available both from national governments and from the European Union. Greece is also a member of the European Union. Volcanism in New Zealand is con"ned to North Island.

Although the potential losses are serious for small, rich developed states, potential risks for small developing countries are of a di!erent order and much more serious. The e!ects of the 1994 eruption of Rabaul volcano in Papua New Guinea, which a!ected Rabaul town*an important regional centre*provide a good example. The high cost (6% of GNP) was mostly borne by Rabaul town, which at the time of eruption, only contained 0.4% of the country's population. Rabaul is but one of many settlements in Papua New Guinea of a similar size and exposure to one or more of the 37 volcanoes that have

䉴 Fig. 4. (A) A selection of the world's most highly exposed conurbations, plotted according to relative distance and direction from the nearest volcano(es) (modi"ed from Munich Re., 1984, Fig. 9, p. 13, population data from numerous sources). (B) The theoretical in#uence of distance on the impact of destructive phenomena associated with volcanic eruptions (modi"ed from McGuire, 1998, Fig. 1b, p. 80, plus additional information).

D.K. Chester et al. / Environmental Hazards 2 (2001) 89}103

97

98

D.K. Chester et al. / Environmental Hazards 2 (2001) 89}103

been active during the Holocene (Blong and McKee, 1995). Other examples of poor island states, with small populations and with signi"cant urban risks, include many countries in the Caribbean and the Comoro islands in the Indian Ocean. In some cases, losses would no doubt be mitigated by the colonial or former colonial power, as is the case in Montserrat following the 1995 eruption (Possekel, 1999), but for others the outlook is more uncertain and bleak. Further instances of settlements of strategic importance occur in some of the developing countries, which display marked urban primacy (Table 1). Here, either the primate city is unexposed, or has a signi"cantly lower risk than a smaller, but still important regional urban centre. Several examples occur in the central Andes (de Silva and Francis, 1991) and include the regionally important cities of CamanaH , Peru (pop. 20,000*Coropuna and Sabancaya volcanoes), and Arequipa, Peru (pop. &634,000*Chachani and El Misti volcanoes). Whereas the impact of a major eruption on an urban centre in a developing country is likely to have severe social and economic consequences at a regional level, the in#uence on a global scale is likely to be minimal. A major eruption in a developed country, devastating an individual region, could however, lead to severe global economic repercussions, the prime examples being Tokyo and several smaller Japanese cities (Fig. 4A). A second subtlety brought out in Fig. 4B is that many cities face a threat even though they are not located on or near to volcanoes. As mentioned above, low frequency high magnitude events (i.e. VEI 6 and 7) are rare but not unknown. Further many coastal cities of the circumPaci"c (Fig. 1A) face threats from volcano-induced tsunami, while other more distant cities could be a!ected adversely by the consequences of major weather perturbations produced by an eruption equal to or greater than Tambora in 1815 (Tables 3 and 4). In the case of

tsunami, the vast majority of events are generated by submarine tectonic earthquakes or seismically-triggered submarine landslides, and during the past 50 years few deaths and little devastation have resulted from volcanoinduced events, though the harbour of Rabaul town was badly damaged during the 1994 eruption (Blong and McKee, 1995). There are many historical examples of major damage and many deaths caused by volcanoinduced tsunami (Unzen, Japan*1792 and Krakatau, Indonesia*1883), but these are dwarfed by estimates of current risk that have been made on the basis of past collapses on the Hawaiian Islands and in the Canaries (Moore and Moore, 1984; McGuire, 1998, p. 92; Carracedo, 1999). In the case of the Canaries, modelling indicates run-up heights near to the source of up to 375 m and destruction spreading over thousands of kilometres. Under the most extreme scenarios large sections of the western seaboard of the USA would be threatened. A third subtlety relates to detailed issues of city site and situation. In Tables 3 and 4, mitigation measures are listed for each type of volcanic and volcano-related hazard. These include the damming and/or diversion of lava #ows, the strengthening of roofs so that they can withstand ash loading and the construction of sediment dams to contain #ood waters and lahars. Such mitigation measures rely on costly engineering solutions and are palliatives, in the sense that they attempt to reduce losses once an eruption has started. General and specixc predictions are prominent measures of mitigation for a wide range of volcano and volcano-related hazards (Table 3 and 4), and the characteristics of both approaches are summarised in Table 5. Although early attempts at prediction were carried out several decades ago, it is only in the last 20 years, and especially since 1990, that they have become a signi"cant research focus in applied volcanology. Although there has been abundant research into understanding and in-

Table 5 The characteristics of general and speci"c prediction (based on: Walker, 1974 and additional sources) General prediction (also known as hazard mapping and assessment)

Speci"c prediction

The raison d'etre is a desire to know the products which are to be erupted over di!erent time scales and the e!ects these will have on people living in a volcanic region. De"ned as, study of the past behaviour of a volcano to determine the frequency, magnitude and style of eruptions, and to delineate high risk areas, general prediction uses geological and historical evidence and eruption statistics, to produce a map showing the range of volcanic hazards under di!ering eruption scenarios. Traditionally the output has been cartographic, but increasingly computer-based geographical information systems (GIS) are being employed. In pro-active planning, general prediction is used to steer new economic development away from particularly hazard prone areas, while in reactive mode it may be used in civil defence and evacuation planning. The raison d'etre is to forecast the time and type of eruption. Speci"c prediction is based on surveillance of a volcano, and monitoring of changes (e.g. in seismic activity, ground deformation, thermal characteristics and the geochemistry of gases), to forecast the time, place and magnitude of an eruption. This involves the identi"cation of precursory signs of activity and necessitates careful observation of the volcano for months, and in some cases years, before an eruption occurs. Although frequently described as separate approaches, general and speci"c are complementary, because once an eruption has started both may be used to determine how it will develop and what measures will be necessary to minimise its e!ects.

D.K. Chester et al. / Environmental Hazards 2 (2001) 89}103

terpreting precursory phenomena related to volcanic eruptions (McGuire et al., 1995; Scarpa and Tilling, 1996), it is not yet possible to predict the time of onset of most eruptive events. In the case of a limited number of well-studied volcanoes, for example Kilauea (Hawaii), it has proved possible to predict some eruptive events. Another example is the Soufriere Hills Volcano in Montserrat. In mid-1997 the eruptive andesitic dome of the Soufriere Hills volcano showed a pronounced rhythmic pattern of repetitive ground in#ation and de#ation. Each cycle lasted about 12}18 h, with rock falls and pyroclastic #ows being associated with the onset of de#ation. Once in#ation began to ease, rock falls and pyroclastic #ows could be expected to peak within the next few hours (Voight et al., 1998). Cyclic behaviours in the deformation of the dome and eruptive events were also re#ected in the pattern of earthquake swarms. In the case of the major (June 25) eruption, which caused 19 deaths in the exclusion zone, in#ation started at about 09:00 h and "nished around 12:00 h, the accompanying hybrid seismic swarm began at 10:50 h and merged into felt ground tremor at 12:50 h, when rapid de#ation occurred. Ten minutes later there was a major collapse of the lava dome and pyroclastic #ows swept down the #ank of the volcano (Voight et al., 1999). Similar cyclicity was displayed by volcanian explosions and it proved possible to predict some of these. On the basis of this predictable pattern of eruptive behaviour, it was possible to control the access of "eld parties to the volcano, so making surveillance by scientists much safer. The above example from Montserrat illustrates how under certain circumstances it is possible to make speci"c predictions using detailed instrumental surveillance, but there is clearly a long way to go before scientists are able to provide precise predictions of eruptions on a routine basis. In most cases it is possible, however, to recognise when a volcano is entering a dangerous phase. Volcanoes typically show increased seismicity for up to a year prior to an onset of new eruptive activity (McNutt, 1996). Mount St. Helens in 1980, Pinatubo in 1991 and Soufriere Hills in 1995, all showed phreatic (i.e. steam) explosions in the weeks and months prior to the main eruptive phase. The last 10 years of the 20th century were designated by the United Nations The International Decade for Natural Disaster Reduction (IDNDR), and many countries initiated research programmes to improve understanding of the threat posed to major population centres by natural hazards. In addition the international volcanological research community through its professional body, the International Association for Volcanology and Chemistry of the Earth's Interior (IAVCEI), designated 16 `decade volcanoesa for detailed study, to which the European Union/European Science Foundation added a further 6 `laboratory volcanoesa. The decade volcanoes were selected to represent a wide range of volcanic

99

styles*some explosive others e!usive, located in dissimilar countries*some from developed others from developing countries, the aim being to develop expertise that could be applied more generally (Newhall, 1998). Among the volcanoes in Fig. 4A, those listed as threatening major cities, Rainier, Taal, Sakura-jima, Vesuvius, Unzen and Merapi were decade volcanoes, Etna was a laboratory volcano, and those in the United States were and are being actively studied under the Volcanic Hazards Program of the United States Geological Survey. Similar programmes have been initiated in many other countries. For most of the volcanoes listed in Fig. 4A and many others that threaten cities in developed countries, hazard maps are now complete, but for most*though by no means all*cities at risk in developing countries, progress has been more patchy (Fisher et al., 1997, p. 262}285). From the hazard maps that have been published, it is already clear that distance and direction of a city from a volcano are not the sole, or even the most important, determinants of risk. Lahars from Mount Rainier, for example, constitute a major potential volcanic hazard in the Cascade Range. Reconstruction of past events and the mapping of river valley routes radiating from the volcano, show that during the past 10,000 years 60 lahars have been generated, but only the most severe have a!ected the sites of the cities of Tacoma and Seattle (Hoblitt et al., 1998). While these cities are at risk from high magnitude/low frequency eruptions the greatest need is to devise civil defence measures to protect the populations of the many smaller towns which are in greater danger. Lava #ows move down slope and follow topographic depressions and on Mount Etna in Sicily many villages are at risk. The city of Catania (Fig. 4A), which has been destroyed and damaged by lava #ows on several occasions during its long history, has a complex hazard exposure. Although some of the city is threatened, other areas are elevated and e!ectively `topographically protecteda from lava inundation (Chester et al., 1985). Vesuvius presents a major threat to the Naples conurbation (Fig. 4A). The greatest risk is not to the city itself, but to its sprawling suburbs which have spread over the #anks of Vesuvius in an unplanned and chaotic manner. Here the risk is of mass destruction by pyroclastic #ows and surges, combined with major economic losses from tephra fall deposition. On the basis of general prediction, the routes*mostly topographic depressions and river valleys*which would be followed by these phenomena, have been determined (Dipartimento della Protezione Civile, 1995).

4. Conclusion: present trends and future prospects As far as the volcanological community is concerned the development and application of general and speci"c

100

D.K. Chester et al. / Environmental Hazards 2 (2001) 89}103

prediction are probably the greatest achievements of the IDNDR, but other aspects of research, nascent in the early 1990s, are now having a direct impact on understanding urban hazards. From the early 1980s, social scientists were arguing (Hewitt, 1983, 1997) that hazard research placed too much emphasis on scienti"c and engineering solutions to hazards; and the uncritical transfer of technology and planning expertise from rich to poor countries. During the IDNDR this critique was voiced at a United Nations conference in Yokohama in 1994, charged with reviewing progress during the midpoint of the decade (United Nations, 1995). Accordingly much greater emphasis is now placed on the particularities of the societies living on volcanoes and how they will react to damaging eruptions. In 1994, the hazard analyst Piers Blaikie argued that what is now required is a con#ation of traditional hazard mapping, speci"c prediction and civil defence planning with vulnerability analysis. Whilst hazard mapping is concerned with physical mechanisms, vulnerability analysis concerns `everything else: monitoring changes in root causes . . . , and understanding how these are channelled into unsafe conditions for . . . , the population by social and economic mechanismsa (Blaikie et al., 1994, p. 225). For volcanology this means that the mitigation measures listed in Tables 3 and 4 are but the "rst stage in assessment and these have to be combined with parallel studies of aspects of demography, economy and society which either increase or decrease susceptibility to losses. An evolving international consensus*involving interdisciplinary research with earth scientists working with social scientists and civil defence planners*is throwing new light on the exposure of large and regionally important cities. Research on Furnas European laboratory volcano in the Azores is instructive and has highlighted several aspects of vulnerability that have to be taken into account if civil defence policies are to be successful (Cole et al., 1999; Pomonis et al., 1999). These include high `dependency ratiosa (i.e. the % under 15, plus the % over 65 in the population), and high levels of illiteracy, suggesting that considerable assistance would be required in the event of evacuation; and a strong attachment of people to place, which could discourage some inhabitants from agreeing to leave their homes, land, villages and livestock (Dibben and Chester, 1999). On Teide laboratory volcano in the Canaries, the new &incultured' agenda is manifest in a desire to foster public education in order to give local people ownership of emergency plans (GoH mez, 1996; SansoH n-Cerrato, 1996), while on Santorini (Greece) the achilles heel of civil defence has been identi"ed as widespread mistrust and resentment towards government and all attempts at hazard planning (Fytikas, 1995). Applied volcanologists and hazard analysts view education as a priority and developing good relationships with the media and political leaders is recognised as a vital ingredient in e!ective implementation of policy.

The complexities involved in developing a more incultured agenda for volcanic hazard analysis are far greater for large cities than for the island communities of the Azores, Canaries and Greece. In the case of many cities in developing countries, natural hazards in general and volcanic hazards in particular are but one of many aspects of vulnerability. Puente (1999) using Mexico City as an example, with parallels relevant to many hazard exposed cities in the region, argued that the crux of the problem is the inability of local and national governments to cope with rapid population growth, and the socio-economic polarisation between rich and poor neighbourhoods following in its wake. Outer suburbs in Mexico City and elsewhere are characterised by illegal squatter settlements often located in particularly hazard prone localities (e.g. in valley bottoms and on steep slopes). The sheer pace of growth and lack of land-use planning means that plans for hazard zoning and civil defence are dated before they can be implemented. Large numbers of people are not only economically and socially marginalised, but are also e!ectively untouched by disaster planning. Complexity at the interface between hazard assessment and planning is not con"ned to cities in developing countries. In Japan and as recently as 1991, pyroclastic #ows from Unzen (Fig. 4A) killed 44 people, cost the city of Shimabara and surrounding towns 2 billion $US and necessitated the prolonged evacuation of 11,000 citizens (Shimozuru, 1996). Japan has great strengths in pure and applied volcanology, its cities are well ordered and its population well educated, but even here recent disasters have highlighted problems in disaster management. Following the 1995 Kobe earthquake the authorities were severely criticised for bureaucratic delays, lack of emergency preparedness and an unwillingness to accept overseas help (Dawkins, 1995; Bolt, 1999). At present the authorities are trying to overcome these de"ciencies (Shimozuru, 1996; Kumagai and Nojima, 1999). Planning for an eruption of Vesuvius in Italy is perhaps the most extreme example of the issues raised by volcanic hazard planning in large complex cities (Matthews, 1998). Vesuvius can give rise to explosive eruptions which generate pyroclastic #ows (Dobran et al., 1994). A comprehensive draft hazard evaluation and evacuation plan has been proposed, based on a 1631 AD eruption scenario. Using a combination of the areal distribution of 1631 AD eruption products and computerbased hazard mapping, modelling of likely tephra fall and pyroclastic #ow emplacement has been undertaken. The plan envisages the evacuation of 700,000 people following a 7-day warning (Dipartimento della Protezione Civile, 1995). The rationale is that before the 1631 eruption, earthquakes were felt for at least 15 days. Preparation of an evacuation plan is one important step in disaster management. Clearly there are many issues that need addressing, three being particularly important:

D.K. Chester et al. / Environmental Hazards 2 (2001) 89}103

(a) the validity of the assumptions being made about both the veracity of precursory phenomena and the validity of the 1631 eruption scenario (Masood, 1995); (b) the impact of volcanic processes and products; and (c) the logistic s of evacuation. The plan's success depends critically on the interpretation of precursory phenomena using techniques of speci"c prediction, and the di$culty is in deciding how long a seismic crisis should be allowed to continue before an evacuation is ordered. Compliance with an evacuation order will depend on the hazard awareness of the population and at the present time there is a renewed focus on education and dialogue with local people. A limited trial evacuation on Vesuvius has been declared a success and the draft evacuation plan is being revised by its editor, Professor Lucia Civetta (Director of the Vesuvius Volcano Observatory) based on comments from interested parties. Despite a local milieu, where scientists deal with severe problems of unplanned and often illegal urban growth, a long-standing denial by much of the population that Vesuvius poses any threat at all, organised crime, and allegations of corrupt government, the evacuation plan at least tackles di$cult issues that have long been recognised and are only now being addressed. Recently the European Science Foundation/European Union has sponsored volcanic risk assessment seminars in several countries including Italy. At the present time the United Nations, national governments and the international volcanological community are making plans for what will follow the IDNDR. The United Nations terms its strategy: A Safer World in the Twenty-First Century: Risk and Disaster Reduction (United Nations, 1999); and, as well as focusing on interdisciplinary studies of the type described in this paper, attaches great importance to the integration of disaster planning and policies of sustainable development. According to Robert Hamilton of the National Academy of Sciences in the USA, (Hamilton, 1999, p. 306), two key priorities in the immediate future will be: (1) further study of hazard exposed world cities; and (2) greater integration of conventional scienti"cally based hazard analysis and social science based studies into human vulnerability. In other words the two research themes that have emerged within applied volcanology during the second half of the 1990s, and which are discussed in this paper, are likely to become even more important during the present decade.

References Alexander, D., 1997. The study of natural disasters, 1977}1997: some re#ections on a changing "eld of knowledge. Disasters 21 (4), 284}304. Anon, 1994. A World Safe from Natural Disasters: The Journey of Latin America and the Caribbean. Pan-American Health Organisation, Washington, DC. Anon, 1995. Megacities: Reducing Vulnerability to Natural Disasters. Thomas Telford, London.

101

Aysan, Y., Clayton, A., Cory, A., Davis, I., Sanderson, D. (Eds.), 1995. Developing Building for Safety Programmes: Guidelines for Organizing Safe Building Improvement Programmes in DisasterProne Areas, Intermediate Technology, London. Barberi, F., Carapezza, M.L., 1996. The problem of volcanic unrest: the Campi Flegrei case history. In: Scarpa, R., Tilling, R.I. (Eds.), Monitoring and Mitigation of Volcanic Hazards. Springer, Berlin, Heidelberg, pp. 771}785. Baxter, P.J., 2000. Impacts of eruptions on human health. In: Sigurdsson, H., Houghton, B.F., Mc Nutt, S.R., Rymer, H., Stix, J. (Eds.), Encyclopedia of Volcanoes. Academic Press, San Diego, pp. 1035}1043. Blaikie, P., Cannon, T., Davis, I., Wisner, B., 1994. At Risk: Natural Hazards, People's Vulnerability and Disasters. Routledge, London, New York. Blong, R.A., McKee, C., 1995. The Rabaul Eruption 1994. Natural Hazards Research Centre, Macquarrie University, New South Wales. Bolt, B.A., 1993. Earthquakes, 3rd Edition. W.H. Freeman, New York. Bolt, B.A., 1999. Earthquakes, 4th Edition. W.H. Freeman, New York. Carracedo, J.C., 1999. Growth, structure, instability and collapse of Canarian volcanoes and comparisons with Hawaiian volcanoes. Journal of Volcanology and Geothermal Research 94, 1}19. Chester, D.K., 1993. Volcanoes and Society. Edward Arnold, London. Chester, D.K., 2001. Volcanoes, society and culture. In: Marti, J., Ernst, G.J. (Eds.), Volcanoes and the Environment. Cambridge University Press, Cambridge (forthcoming). Chester, D.K., Duncan, A.M., Guest, J.E., 2001. Loss-bearing on Etna from the Classical Period to 1900 CE. In: Johnston, P. (Ed.), The Cultural Response to Volcanic Landscapes. Archaeological Institute of America (in press). Chester, D.K., Duncan, A.M., Guest, J.E., Kilburn, C.R.J., 1985. Mount Etna: The Anatomy of a Volcano. Chapman & Hall, London. Coburn, A., Spence, R., 1992. Earthquake Protection. Wiley, Chichester. Cole, P.D., Guest, J.E., Queiroz, G., Wallenstein, N., Pacheco, J.M., Gaspar, J.L., Ferreira, T., Duncan, A.M., 1999. Styles of volcanism and volcanic hazards on Furnas Volcano, Sa o Miguel, Azores. Journal of Volcanology and Geothermal Research 92, 39}53. Connor, S., 2000. Scientists warn of supervolcano that could blot out the sun and bring a global winter. The Independent (London), 3 February, p. 9. Crandell, D.W., Booth, B., Kusumadinata, K., Shimozuru, D., Walker, G.P.L., Westerkamp, D. (Eds.), 1984. Source Book for VolcanicHazards Zonation. UNESCO, Paris. Dawkins, W., 1995. Faith in the authority fades. Financial Times (London), 21 January, p. 9. De Silva, S.L., Francis, P., 1991. Volcanoes of the Central Andes. Springer, Berlin. Dibben, C., Chester, D.K., 1999. Human vulnerability in volcanic environments: the case of Furnas, Sa o Miguel, Azores. Journal of Volcanology and Geothermal Research 92, 133}150. Dipartimento della Protezione Civile, 1995. Piani"cazione Nazionale D'Emergenza Dell' Area Vesuviana. Prefettura di Napoli, Napoli and Dipartemento della Protezione Civile, Roma. Dobran, F., Neri, F., Tedesco, M., 1994. Pyroclastic #ow at Vesuvius. Nature 367, 551}554. Domeisen, N., Palm, E., 1996. Urban trends. STOP Disasters 28 (2), 4. Duncan, A.M., Chester, D.K., Guest, J.E., 1981. Mount Etna Volcano: environmental impact and problems of volcanic prediction. Geographical Journal 147, 164}179. Duncan, A.M., Dibben, C., Chester, D.K., Guest, J.E., 1996. The 1928 eruption of Mount Etna Volcano, Sicily, and the destruction of the town of Mascali. Disasters 20, 1}20. Fisher, R.V., Heiken, G., Hulen, J.B., 1997. Volcanoes Crucibles of Change. Princeton University Press, Priceton, NJ.

102

D.K. Chester et al. / Environmental Hazards 2 (2001) 89}103

Fytikas, M., 1995 Santorini volcano laboratory. In: Barberi, F., Casale, R., Fratta, M. (Eds), The European Laboratory Volcanoes. Workshop Proceedings*Aci Castello (Catania), 18}21 June 1994. European Commission, Brussels, pp. 174}183. GoH mez, F., 1996. Los sistemas de informacioH n geograH "ca (GIS) en la evalucioH n del riesgo volcaH nico. In: Ortiz, R. (Ed.) Riesgo VolcaH nico. Servicio de Publicaciones, Cabildo de Lanzarote: Conserjeria de Cultura, Serie de Los Volcanes Vol. 5, pp. 123}153. Had"eld, P., 1995. Sixty Seconds that will Change the World: How the Coming Tokyo Earthquake will Wreak Worldwide Economic Devastation. Pan Books, London. Hamilton, R., 1999. Natural disaster reduction in the 21st Century. In: Ingleton, J. (Ed.), Natural Disaster Management. Tudor Rose, Leicester, pp. 304}307. Harriss, R.W., Hohenemser, C., Kates, R.W., 1985. Human and nonhuman mortality. In: Kates R.W., Hohenemser, C., Kasperson, J.X. (Eds.), Perilous Progress: Managing the Hazards of Technology. Westview Press, Boulder, CO, pp. 129}155. Hewitt, K. (Ed.), 1983. Interpretations of Calamity. Allen & Unwin, London. Hewitt, K., 1997. Regions of Risk: A Geographical Introduction to Disasters. Longman, London. Hoblitt, R.P., Walder, J.S., Driedger, C.L., Scott, K.M., Pringle, P.T., Vallance, J.W., 1998. Volcano hazards from Mount Rainier, Washington, Revised 1998, Open-File Report 98-428. United States Geological Survey. Washington, DC. Jones, B.G., Kandel, W.A., 1992. Population growth, urbanization, and disaster risk and vulnerability in metropolitan areas: a conceptual framework. In: Kreimer, A., Munasinghe, M. (Eds.), Environmental Management and Urban Vulnerability, Vol. 168. World Bank Discussion Paper, Washington, DC, pp. 51}76. Kates, R.W., 1987. Hazard assessment and management. In: McLaren, D.J., Skinner, B.J. (Eds.), Resources and World Development. Wiley, Chichester, pp. 741}753. Keys, D., 1999. Catastrophe. Century, London. Kumagai, Y., Nojima, Y., 1999. Urbanization and disaster mitigation in Tokyo. In: Michell, J.K. (Ed.), Crucibles of Hazard: Mega-cities and Disasters in Transition. United Nations University Press, Tokyo, pp. 56}91. Masood, E., 1995. Row erupts over evacuation plans for Mount Vesuvius. Nature 377, 471. Matthews, R., 1998. The Vesuvius Dilemma. The Sunday Telegraph (London), 12 April, p. 21. McGuire, W.J., 1998. Volcanic hazards and their mitigation. In: Maund, J.G., Eddleston, M. (Eds.), Geohazards in Engineering Geology, Vol. 15. Geological Society Engineering Geology Special Publication, London, pp. 79}95. McGuire, B. (W.J), 1999. Apocalypse. Cassell, London. McGuire, W.J., Kilburn, C., Murray, J. (Eds.), 1995. Monitoring Active Volcanoes. UCL Press, London. McNutt, S.R., 1996. Seismic monitoring and eruption forecasting of volcanoes. In: Scarpa, R., Tilling, R.I. (Eds.), Monitoring and Mitigation of Volcanic Hazards. Springer, Berlin, pp. 99}146. Mitchell, J.K., 1999. Natural disasters in the context of mega-cities. In: Mitchell, J.K. (Ed.), Crucibles of Hazard: Mega-cities and Disasters in Transition. United Nations University Press, Tokyo, pp. 15}55. Moore, J.G., Moore, G.W., 1984. Deposit from a giant wave on the island of Lanai, Hawaii. Science 226, 1312}1315. Munich Re., 1984. Volcanic Eruption*Causes and Risks. MuK chener RuK chversicherungs-Gesellschaft, Munich. Munich Re., 1997. Topics: Annual Review of Natural Catastrophes, Munich Reinsurance Company, Munich. Munich Re., 1999. Topics: Annual Review of Natural Catastrophes, 1998. Munich Reinsurance Company, Munich. Newhall, C., 1998. IAVCEI/International Council of Scienti"c Unions' Decade Volcano Projects*Reducing Volcanic Disaster Status Report. United States Geological Survey, Washington, DC.

Newhall, C.G., Self, S., 1982. The volcanic explosivity index (VEI): an estimate of explosive magnitude for historical volcanism. Journal of Geophysical Research 87 (C), 1231}1238. Peet, R., 1998. Modern Geographical Thought. Blackwell, Oxford. Pomonis, A., Spence, R., Baxter, P., 1999. Risk assessment of residential buildings for an eruption of Furnas Volcano, Sa o Miguel, the Azores. Journal of Volcanology and Geothermal Research 92 (1}2), 107}131. Possekel, A.J., 1999. Living with the Unexpected: Linking Disaster Recovery to Sustainable Development in Montserrat. Springer, Berlin. Puente, S., 1999. Social vulnerability to disasters in Mexico City: an assessment method. In: Michell, J.K. (Ed.), Crucibles of Hazard: Mega-cities and Disasters in Transition. United Nations University Press, Tokyo, pp. 295}334. Rampino, M.R., Self, S., 1992. Volcanic winter and accelerated glaciation following the Toba super-eruption. Nature 359, 50}52. SansoH n-Cerrato, 1996. La protecioH n civil ante en riesgo de erupciones volcaH nicas. In: Ortiz, R. (Ed.), Riesgo VolcaH nico. Servicio de Publicaciones, Cabildo de Lanzarote: Conserjeria de Cultura, Serie de Los Volcanes Vol. 5, pp. 197}216. Scarpa, R., Tilling, R.I. (Eds.), 1996. Monitoring and Mitigation of Volcanic Hazards. Springer, Berlin. Shimozuru, D., 1996. Volcanic emergency management in Japan: case histories of Izu-Oshima and Unzen. In: Scarpa, R., Tilling, R. (Eds.), Monitoring and Mitigation of Volcanic Hazards. Springer, Berlin, pp. 788}806. Simkin, T., Siebert, L., Blong, R., 1994a. Volcanoes of the World. Geoscience Press, Tucson, Arizona. Simkin, T., Unger, J.D., Tilling, R.I., Vogt, P.R., Spall, H., 1994b. Dynamic Planet: World Map of Volcanoes, Earthquakes, Impact Craters and Plate Tectonics. United States Geological Survey, Washington, DC. Stoddart, D.R., 1987. To claim the high ground: geography for the end of the century. Transactions of the Institute of British Geographers NS12, 327}336. Susman, P., O'Keefe, P., Wisner, B., 1983. Global disasters, a radical interpretation. In: Hewitt, K. (Ed.), Interpretations of Calamity. Allen and Unwin, London, pp. 263}283. Tanguy, J.-C., Ribie`re, Ch., Scarth, A., Tjetep, W.S., 1998. Victims from volcanic eruptions: a revised database. Bulletin of Volcanology 60, 137}144. Tilling, R.I., 1989. Volcanic hazards and their mitigation: progress and problems. Reviews of Geophysics 27, 237}267. Tilling, R.I., Lipman, R.W., 1993. Lessons in reducing volcanic risk. Nature 364, 277}280. UNESCO, 1993. Disaster reduction. Environment and Development Briefs 5. Banson, London. UNFPA, 1993. The Individual and the World: Population, Migration and Development in the 1990s. United Nations Population Fund, Geneva. United Nations, 1995. World Urbanization Prospects: The 1994 Revision. The United Nations Organisation, New York. United Nations, 1996. An Urbanizing World: Global Report on Human Settlements 1996. Oxford University Press, Oxford. United Nations, 1997. Urban Agglomerations, 1996 (wallchart). The United Nations Organisation, New York. United Nations, 1999. International Decade for Natural Disaster Reduction: Successor Arrangement. United Nations, New York. Voight, B., Hoblitt, R.P., Clarke, A.B., Lockhart, A.B., Miller, A.D., Lynch, L., 1998. Remarkable cyclic ground deformation monitoring in real-time on Montserrat, and its use in eruption forecasting. Geophysical Research Letters 25 (18), 3405}3408. Voight, B., Sparks, R.S.J., Miller, A.D., Stewart, R.C., Hoblitt, R.P., Clarke, A., Ewart, J., Aspinall, W.P., Baptie, B., Calder, E.S., Cole, P., Druitt, T.H., Hartford, C., Herd, R.A., Jackson, P., Lejeune, A.M., Lockhart, A.B., Loughlin, C., Luckett, R., Lynch, L., Norton, G.E., Robertson, R., Watson, I.M., Watts, R., Young, S.R., 1999. Magma #ow instability and cyclic activity at Soufriere Hills Volcano, Montserrat, British West Indies. Science 283, 1138}1142.

D.K. Chester et al. / Environmental Hazards 2 (2001) 89}103 Walker, G.P.L., 1974. Volcanic hazards and the prediction of volcanic eruptions. In: Funnell, B.M. (Ed.), Prediction of Geological Hazards. Geological Society of London, Miscellaneous Paper 3, 23}41. White, G.F., 1973. Natural hazards research. In: Chorley, R.J. (Ed.), Directions in Geography. Methuen, London, pp. 193}212.

103

Wijkman, A., Timberlake, L., 1984. Natural Disaster: Acts of God or Acts of Man? Earthscan. International Institute for Environment and Development, London, New York. Wood, C.A., Kienle, J., 1990. Volcanoes of North America. Cambridge University Press, Cambridge.