Hazardous environments

Where earthquakes and volcanoes happen, and why

Earthquakes and volcanoes are not scattered at random across the map: they trace the edges of the tectonic plates. Roughly three-quarters of the world's active volcanoes and most of its earthquakes lie in a belt around the Pacific Ocean known as the Ring of Fire, where the Pacific plate and several smaller plates are being pushed beneath their neighbours. A second belt runs from the Mediterranean through the Middle East and the Himalayas to Indonesia, following the collision zone between Africa, Arabia and India and the Eurasian plate. The third lies under the oceans, along the mid-ocean ridges, where plates are pulling apart and new crust is being created.

At a constructive (divergent) boundary two plates move apart and hot mantle rock rises into the gap. As it rises the pressure on it falls, so it partly melts without any extra heat, and it erupts as runny basaltic lava that cools into new crust. The earthquakes here are shallow, because the crust is thin, hot and broken, and they are usually small to moderate: strain is released often rather than stored for centuries. Iceland, sitting astride the Mid-Atlantic Ridge, erupts in this way, and the East African Rift shows the same process tearing a continent apart.

At a destructive (convergent) boundary an oceanic plate, being denser, is forced down beneath the other plate into a deep ocean trench. It carries seawater and wet sediment down with it, and that water lowers the melting point of the mantle above, so magma rises and feeds a line of explosive strato-volcanoes set back from the trench. The earthquakes are the largest on Earth, because a subduction zone is a locked boundary hundreds of kilometres long that finally ruptures along its whole length: they are shallow near the trench and get progressively deeper down the sinking slab, reaching 600 to 700 km. The 2004 Sumatra-Andaman and 2011 Tohoku earthquakes both occurred on subduction zones of this kind.

Where two continental plates meet, at a collision boundary, neither can sink far because continental rock is too light, so the crust is crumpled upwards into fold mountains. That is the Himalayas and the Alps: many earthquakes, but almost no volcanoes, because very little rock is pushed deep enough to melt. At a conservative (transform) boundary the plates slide past one another instead, as along the San Andreas Fault in California and the North Anatolian Fault in Turkey, producing shallow strike-slip earthquakes and no magma at all. Volcanoes therefore map the places where rock melts, which means spreading ridges and subduction zones, and earthquakes map every boundary where plates are locked together.

Earthquake severity: focus, epicentre, waves and magnitude scales

An earthquake happens when rock that has been bent by plate movement suddenly snaps and slips along a fault, releasing stored elastic strain as seismic waves. The point inside the Earth where the rupture starts is the focus; the point on the surface directly above it is the epicentre. Depth matters more than most students expect: a shallow-focus earthquake (less than about 70 km down) releases its energy close to the surface, so the shaking is violent, while a deep-focus earthquake of the same magnitude, 300 to 700 km down, has already spread its energy over a wide area by the time it reaches the surface and does far less damage. Shaking also weakens with distance from the epicentre, and it is amplified by soft, wet sediment and reclaimed ground while being damped by solid rock, so two districts of one city can suffer very differently.

The waves that arrive first are the body waves, which travel through the interior. P-waves (primary) are compressional: the rock is pushed and pulled along the direction the wave is travelling, like a slinky, and they are the fastest seismic waves at roughly 6 to 8 km/s in the crust, so they arrive and are recorded first. S-waves (secondary) are transverse: the rock is shaken from side to side at right angles to the direction of travel, they move at about half the speed of P-waves, and they cannot pass through a liquid. That last fact is how we know the outer core is liquid, and it is also why S-waves shake buildings more strongly than P-waves do.

The surface waves travel along the ground rather than through it. Love waves move the ground horizontally from side to side, parallel to the surface, and shear the foundations of buildings as they pass; they are the faster of the two surface waves. Rayleigh waves roll the ground in an ellipse, like a wave on the sea, and although they are the slowest of all they have the largest amplitude, so they cause most of the structural damage and are the waves that trigger liquefaction. Because surface waves have a long period, they also reach further from the epicentre, which is why a tall building many kilometres away can sway badly while a low one nearby is barely affected.

Two different questions then get asked, and they need two different scales. How much energy was released? That is magnitude, a property of the earthquake itself. How badly did it shake here, and what did it do? That is intensity, a property of a particular place. The Richter scale, devised by Charles Richter in the 1930s for southern California, is logarithmic and based on the amplitude of the largest wave recorded on a standard instrument at a standard distance; it works well for small and moderate local earthquakes but saturates above about magnitude 7, so it understates the greatest events. The moment magnitude scale (Mw) replaced it and is now the standard: it is calculated from the seismic moment, the rigidity of the rock multiplied by the area of fault that slipped and the average distance it slipped, so it measures the work done by the rupture, does not saturate, and gives one value for one earthquake wherever it is recorded. The Modified Mercalli scale measures intensity instead, on a scale from I (not felt) to XII (total destruction), assessed from what people felt and what happened to buildings, roads and the ground.

Because the Mercalli scale depends on local effects, the same earthquake scores high near the epicentre and low hundreds of kilometres away, which makes it ideal for mapping the pattern of damage but useless for comparing the size of two events. Magnitude is logarithmic, and it is worth being able to handle: an increase of one whole number means about ten times the ground motion and about 32 times the energy released, because the energy rises by a factor of about 31.6 per unit. A magnitude 7 earthquake therefore releases roughly 32 times the energy of a magnitude 6, and about 1 000 times the energy of a magnitude 5.

• Richter scale: logarithmic, from wave amplitude on a standard instrument; good for small local earthquakes but saturates above about magnitude 7, so it understates great earthquakes.
• Moment magnitude (Mw): logarithmic, from the seismic moment of the rupture; does not saturate and gives one value worldwide, so it is now the standard measure of magnitude.
• Modified Mercalli: intensity from I to XII, from observed effects on people, buildings and ground; different in different places for the same earthquake, so it maps damage but cannot compare sizes.
• Logarithmic, not linear: plus one magnitude is about ten times the ground motion and about 32 times the energy, so the step from 6 to 7 matters far more than the step from 4 to 5.

Earthquake hazards, vulnerability and impacts

Ground-shaking is the hazard everything else follows from. It damages or collapses buildings, breaks pipes and cables, and throws people and vehicles around, and its severity at a place depends on magnitude, depth, distance and the ground beneath. Shaking is amplified on soft, wet sediment and on reclaimed land, and damped on solid rock, which is why damage in one city is often concentrated in the districts built on old river deposits or on land won from the sea.

Liquefaction happens when shaking strikes saturated, loosely packed sand or silt. The shaking raises the water pressure in the pores between the grains until the grains stop touching each other, and the soil behaves like a liquid rather than a solid. Buildings tilt and sink, roads and airport runways buckle, buried pipes and storage tanks float upwards, and sand and water are forced out of the ground in sand boils. It is why so much earthquake damage is concentrated on waterfronts, river floodplains and areas of artificial fill.

Landslides are triggered where shaking reduces the strength of a slope that was already close to failure, especially on steep hillsides of weathered rock or saturated soil. They bury roads, houses and villages, cut the routes that rescuers need, and can dam a river valley to create a lake that later bursts and floods downstream. Tsunami are generated when the sea floor is displaced vertically by a shallow subduction-zone earthquake, or when a large mass of sediment slides into the sea. In the open ocean a tsunami travels at up to about 800 km/h with a wavelength of hundreds of kilometres and a height of only a metre or so, so ships barely notice it; as it reaches shallow coastal water it slows, the water piles up and it comes ashore as a series of surges that can run hundreds of metres inland and arrive within minutes of local shaking, or hours later on a distant coast.

Aftershocks are the smaller earthquakes that follow the main shock as the surrounding crust adjusts to the new stress. They are usually weaker, but they last for days, weeks or months, they bring down buildings that were already cracked, and they make search and rescue dangerous, which is why rescue teams do not send people into damaged structures without shoring. Earthquake damage is also compounded by secondary hazards: fires from ruptured gas mains and downed power lines, contaminated water supplies and disease in crowded emergency camps, and the failure of dams and levees.

Vulnerability is the product of physical and human factors. Physical: the depth of the focus, the distance from the epicentre, the local ground conditions, the steepness of the land and whether the coast is exposed to a tsunami. Human: population density (how many people are in the buildings that fall), infrastructure (whether buildings, bridges, hospitals and water mains were designed for shaking and whether they are maintained), economic development (whether the country can afford codes, enforcement, monitoring and a trained emergency service), and the time of day, which decides where people are: a shock at 3 a.m. traps families in their homes, while one at 3 p.m. on a school day traps children in schools and workers in offices, and a shock during the evening rush hour kills people in cars and on bridges.

• Short-term impacts: deaths and injuries from collapsing buildings; people trapped; homelessness and displacement; loss of electricity, water and communications; roads and bridges blocked; fires; hospitals overwhelmed; schools and workplaces closed.
• Long-term impacts: rebuilding costs and debt; lost livelihoods and business closure; permanent disability and psychological trauma; migration away from the area; damaged heritage; disruption to education; and, in the environment, landslide scars, sediment in rivers, salinised or contaminated aquifers and lost vegetation.
• Why the death toll differs: the same magnitude can kill a few dozen people in one country and hundreds of thousands in another, and the difference is building quality, density, ground conditions, depth of focus, time of day and the quality of the response - not the size of the earthquake.

Volcanoes: how they form, the five types and the VEI

A volcano is the landform built where magma reaches the surface, and that magma forms in two main ways. At a constructive boundary the rising mantle decompresses as it approaches the surface and melts without gaining heat; at a destructive boundary water carried down by the subducting plate lowers the melting point of the mantle above it, so melting happens where there would otherwise be none. The magma is less dense than the rock around it, so it rises, gathers in a magma chamber, and is forced up a vent to erupt at a crater or along a fissure once the pressure of the dissolved gases inside it exceeds the strength of the rock above.

What the eruption looks like depends on two properties of the magma. Silica content controls viscosity: basalt is low in silica, hot and runny, so gas escapes easily and lava flows far, giving effusive eruptions; andesite and rhyolite are high in silica, cooler and sticky, so gas is trapped until it explodes out, giving explosive eruptions that shatter the magma into ash and tear the cone apart. Gas content controls how violent that explosion is. The same volcano can do both: a volcano can begin with an explosive blast that empties the chamber and then effuse lava afterwards.

The five volcano types in the syllabus are distinguished by their shape, their magma and their eruption style. A strato-volcano (composite cone) is built of alternating layers of lava and ash, steep-sided and concave, from viscous andesitic magma in explosive eruptions, and it is the classic dangerous cone of a subduction zone. A shield volcano is broad and gently sloping, built by countless flows of runny basalt, so it is large in area but low in profile. A cinder cone is a small, steep cone of scoria and ash thrown from a single vent, usually built in one short eruption and often on the flank of a larger volcano. A lava dome forms when thick, sticky lava piles up around the vent instead of flowing away, producing a steep-sided mound that can collapse without warning into a pyroclastic flow. A fissure eruption opens a long crack rather than a central vent, and where the lava is basaltic it floods out in sheets that build lava plains rather than a cone.

• VEI 0-1: effusive, lava fountains and flows, no real explosion - the Hawaiian eruptions of Kilauea.
• VEI 2-3: moderate explosions, an ash column a few kilometres high, local ash fall - many routine eruptions of strato-volcanoes.
• VEI 4-5: violent explosions, ash columns 10-25 km high, pyroclastic flows and widespread ash fall - Mount St Helens in 1980 was a VEI 5.
• VEI 6-7: catastrophic, with ash reaching the stratosphere and affecting climate worldwide - Pinatubo in 1991 was a VEI 6.
• VEI 8: a super-eruption, over 1 000 cubic km of material - the Yellowstone caldera eruptions were of this size.
• The VEI is logarithmic, like magnitude: each step up is a tenfold increase in the volume of material ejected, and it is calculated from that volume and the height of the eruption column.

Volcanic hazards, vulnerability and impacts

Effusive eruptions produce lava flows, which are slow enough for people to walk away from but destroy everything they cross: buildings, roads, pipelines, farmland and, on the coast, harbours. Lava can sometimes be slowed or diverted by earth barriers, water or bombing the flow, but only where the ground allows it and only for a slow-moving hazard. Effusive eruptions also release large volumes of toxic gases, and the quietest hazard of all is often the most underestimated.

Explosive eruptions produce the killers. Tephra is the general term for fragments thrown out of the vent, from fine ash to blocks and bombs, and ash fall carries the finest material hundreds or thousands of kilometres downwind, collapsing roofs under its weight, killing crops and pasture, poisoning livestock, closing airports and causing respiratory illness. Pyroclastic flows (French: nuees ardentes) are mixtures of hot gas, ash and rock at several hundred degrees Celsius that sweep down the slopes at 100 to 700 km/h, and they destroy and incinerate everything in their path: they buried Pompeii and Herculaneum, and they destroyed the town of Saint-Pierre on Martinique in 1902. Lahars are volcanic mudflows, formed when heavy rain or melted snow mixes with loose ash and runs down the valleys like wet concrete, carrying boulders and trees, destroying bridges and burying settlements; they can occur for years after an eruption, long after the volcano has gone quiet, as happened around Pinatubo.

Three more hazards complete the list. Volcanic landslides happen when the flank of a volcano becomes unstable and collapses, sometimes as a huge debris avalanche, as at Mount St Helens in 1980, and sometimes as a tsunami-generating collapse into the sea. Jokulhlaups are glacial outburst floods: an eruption beneath an ice cap melts enormous quantities of ice, the meltwater is trapped and then released suddenly, and a wall of water and icebergs sweeps down the valleys, taking out roads and bridges, as Iceland has experienced repeatedly. Toxic gases - carbon dioxide, sulphur dioxide and hydrogen sulphide - suffocate people and animals, with carbon dioxide being denser than air and collecting in valleys, cellars and hollows, while sulphur dioxide forms acid rain and damages crops and lungs.

• Vulnerability to a volcano is decided by speed, size, frequency and extent. Speed decides whether people can leave at all, which is why pyroclastic flows and lahars kill and lava usually does not; size and extent decide how many people and how much farmland lie in the path; frequency decides how prepared the population and the authorities are.
• Human factors work on top of that: population density around the cone, whether roads and hospitals are in the danger zone, whether the country can afford monitoring and evacuation, and whether people trust the warning enough to leave their animals and crops.
• Short-term impacts: deaths, burns and respiratory injuries; evacuation and displacement; destroyed homes and crops; ash on roofs and runways; contaminated water; lahars cutting roads.
• Long-term impacts: farmland buried or abandoned; resettlement and permanent migration; loss of tourism and geothermal income; damaged infrastructure; but also, over years, extremely fertile volcanic soils and new land, which is why people return.

Tropical cyclones: formation, hazards, impacts and a changing climate

Tropical cyclones are one phenomenon with three names: hurricanes in the North Atlantic and the eastern North Pacific, typhoons in the western North Pacific, and cyclones in the Indian Ocean and around Australia. They form over warm tropical seas between roughly 5 and 20 degrees of latitude in both hemispheres, and they are seasonal, arriving in late summer and autumn when the sea surface is at its warmest: the North Atlantic season runs from June to November, and the southern hemisphere season from November to April. They are absent from the South Atlantic and the eastern South Pacific, where the water is cooler and the winds aloft too strong.

Five conditions are needed, and each one explains part of the distribution. Sea temperature must be at least about 26-27 °C down to a depth of around 50 m, because the storm is powered by the latent heat released when water evaporates from the sea and condenses in the clouds. The Coriolis effect is needed to set the air spinning, which is why cyclones do not form within about 5 degrees of the equator, where the Coriolis effect is too weak. Low vertical wind shear is needed so that the rising column of air is not torn apart, and moist, unstable air must extend through a deep layer of the atmosphere. Finally, there must be a pre-existing disturbance - a cluster of thunderstorms, often an easterly wave - to start the process.

The sequence runs from disturbance to depression to tropical storm to cyclone. Latent heat released by condensation warms the core, the pressure falls, more air flows in at low levels, and the Coriolis effect turns that inflow into a spiral. The result is the familiar structure: a calm, clear eye of sinking air; a ring of the most violent winds and heaviest rain around it, the eye wall; spiral rainbands further out; and air flowing out at the top. A cyclone weakens when it moves over land, because friction and the loss of its warm water supply cut off the energy, or when it meets cold water or strong wind shear.

Intensity is graded on the Saffir-Simpson scale, from 1 to 5, by the sustained wind speed. It is used for hurricanes in the Atlantic and eastern North Pacific, and other ocean basins have their own equivalent scales. Damage does not rise in step with the category: a category 5 storm is far more than five times as destructive as a category 1, because the energy of moving air rises with the square of its speed, and because the surge and the rainfall that come with it also grow.

The hazards come in four linked packages. High winds tear roofs off, bring down power lines and turn loose material into flying debris, and they affect a swathe a hundred kilometres wide. Storm surge is a rise in sea level ahead of and beneath the storm, produced by the low pressure of the centre, by the wind driving water onshore, and by the tide; it is worst where the sea bed shelves gently, where the coast is funnel-shaped and where the surge coincides with a high spring tide, and it is the hazard that drowns most people. Coastal flooding carries salt water into fields, soils and wells, and the saltwater intrusion that follows can ruin farmland and contaminate freshwater supplies for years. Intense rainfall brings river floods and triggers landslides and debris flows inland, often hundreds of kilometres from the coast and long after the wind has dropped.

Vulnerability to a tropical cyclone depends first on the strength and extent of the storm - its category, its size, and how far the damaging winds and the surge reach - and on the location of landfall. A storm that hits a sparsely populated delta is a different event from the same storm hitting a city built on reclaimed coastal land, and a slight change of track can put the surge into a bay rather than against a sea wall. Population density, infrastructure and economic development then decide what happens: whether buildings are engineered for wind and water, whether there is a seawall or a mangrove belt in front of them, whether the roads out are usable, and whether the country can afford warning systems, shelters and a trained emergency service.

The impacts are best sorted by time. Short term: deaths and injuries, mostly from drowning in the surge and from collapsing buildings; homes destroyed; power, water and communications cut; crops flattened; roads and ports closed; people crowded into shelters; and, afterwards, contaminated water and disease. Long term: the cost of rebuilding and the debt it creates; lost harvests and livelihoods; salt-damaged soils that will not grow crops for years; people who never return; damaged reefs, mangroves and forests; and the disruption to schooling and health care that outlasts the storm itself. Poorer countries lose a much larger share of their national wealth to the same storm, which is why one event can set development back by a decade.

• Climate change and magnitude: warmer sea surfaces supply more latent heat, so a greater proportion of cyclones are likely to reach the highest categories, with higher wind speeds and heavier rainfall, since a warmer atmosphere holds more water vapour.
• Climate change and storm surge: sea-level rise raises the baseline that a surge is built on, so the same storm floods further inland and more often.
• Climate change and frequency: the total number of tropical cyclones may stay the same or fall slightly, while the proportion of intense ones rises - so 'fewer but stronger' is the careful answer, not 'more of everything'.
• Climate change and location: the zone in which cyclones can form and intensify may shift polewards, bringing the hazard to coasts that have little experience of it, and rapid intensification close to landfall is expected to become more common.
• Uncertainty: these are projections from models, not observations of the future, so a strong answer says 'likely' and 'may' and explains the mechanism rather than quoting a figure.

Wildfires: conditions, causes, spread and impacts

A wildfire needs three things at once - fuel, oxygen and heat - and the risk is highest where a wet season has built up a dense growth of vegetation and a dry season then dries it out. The world's fire-prone regions follow that pattern: the Mediterranean-type climates of southern Europe, California, Chile, South Africa and southern Australia; the semi-arid interiors of Australia and North America; the boreal forests of Canada, Alaska and Siberia; and the savannas of Africa, where fires are frequent but usually smaller. Fuel matters in three ways: its type (resinous conifers and oil-rich eucalypts burn fiercely; fine dead grass and litter ignite easily), its flammability (how dry and how oily it is) and its density and continuity (a continuous bed of fuel carries fire; a broken one does not). Climatic factors set the background - seasonal drought, high temperatures, low humidity - and recent weather decides the day: a heatwave, strong winds and a long dry spell turn a manageable fire into an uncontrollable one.

The physical causes are the ones students forget. Desiccation of vegetation by prolonged heat and drought lowers the moisture content of fuel until a spark will take; lightning strikes start fires without any human involvement, especially in dry thunderstorms and in remote boreal forest; and sparks from falling rocks can ignite dry grass on a steep slope. The human causes are more numerous and are split into accidental and deliberate. Accidental: unattended campfires and barbecues; equipment use and malfunction, including power lines brought down by wind, machinery striking stones, and vehicle exhausts; and rubbish - a piece of glass or a discarded bottle acting as a lens to focus sunlight, or a cigarette end thrown from a car. Deliberate: arson, whether for insurance, land clearance or other motives, and the careless burning of agricultural waste that escapes control.

Once alight, a fire spreads in three ways: as a surface fire through litter and low vegetation, as a crown fire through the canopy, which is the fastest and most lethal, and as a ground fire smouldering in peat and roots. Wind is the single most important control: it supplies oxygen, it leans the flames forward so that they preheat the fuel ahead, and it carries burning embers downwind to start spot fires hundreds of metres beyond the front. Topography matters just as much. A fire running uphill spreads faster because the slope preheats the fuel above it and because flames and smoke rise along it, and in the northern hemisphere south-facing slopes are drier and burn more readily; ridges, saddles and narrow valleys channel and accelerate a fire, which is why they become death traps.

Vulnerability depends on the speed and extent of spread, on land use (plantations of oil-rich trees, and the wildland-urban interface where housing pushes into flammable vegetation, are the worst cases), on population density and its spread (dispersed rural housing is harder to defend than a compact town), on infrastructure (access roads, water supply, escape routes) and on economic development (whether there is a fire service, aerial support, building standards and insurance). Impacts include deaths and burns, smoke-related illness, homes and livelihoods lost, evacuation and displacement, power and transport disruption, and insurance costs; environmentally, lost vegetation and wildlife, soil erosion and landslides after the roots decay, degraded water quality, and carbon released into the atmosphere. Not all of it is damage: many ecosystems, from eucalypt forest to Mediterranean scrub, depend on fire to release seed and recycle nutrients, which is why total fire exclusion eventually makes the next fire worse.

Sort the impacts by time. Short term: deaths, burns and smoke inhalation; homes, farms and livestock lost; evacuation and displacement; power, water and road closures; and the cost of firefighting itself. Long term: rebuilding and insurance costs; lost timber, grazing and tourism income; soil erosion, landslides and silted reservoirs after the vegetation that held the slope has gone; lost habitat and carbon storage; and, in some places, a permanent shift to a different vegetation type because the interval between fires is now too short for the original forest to return.

• Climate change and location: longer, hotter and drier fire seasons extend the risk into regions that were rarely fire-prone, and push it into the boreal forest and higher latitudes.
• Climate change and extent: larger areas burn where drought and heat coincide, and fires reach higher intensities and are harder to suppress.
• Climate change and frequency: extreme fire weather becomes more frequent, and in some regions the interval between fires shortens so much that vegetation cannot recover, converting forest to scrub or grass.
• The mechanism is fuel moisture: warmer air and more evaporation dry vegetation for longer each year, so the window in which a fire can start and spread widens at both ends of the season.

Managing hazardous environments: prediction, perception and strategy

Management works on the three parts of the disaster equation - the hazard, the exposure of people and property, and their vulnerability - and it can act before, during or after the event. Before: monitoring and prediction, warning systems, hazard mapping and land-use planning, building codes and engineering, environmental modification, education and insurance. During and immediately after: evacuation, search and rescue, emergency shelters, medical treatment, water and sanitation. Afterwards: rebuilding, livelihood recovery, counselling, and reviewing what failed. The strategies that act before the event save the most lives; the ones that act afterwards are the most visible.

Prediction differs enormously between hazards, and the reliability is the point of the comparison. Earthquakes cannot be predicted: there is no precursor that reliably precedes one, so what exists is long-term probabilistic forecasting from fault maps and past recurrence, plus early warning, which detects the fast P-waves a few seconds before the damaging surface waves arrive and automatically shuts down trains, lifts and gas supplies. Volcanic eruptions can often be forecast, because magma moving towards the surface produces precursor events: swarms of small earthquakes, ground deformation measurable by tiltmeters, GPS and satellite radar, changes in the temperature and chemistry of gas emissions, and changes in springs and fumaroles - giving warning times of days to months, though not of the size or exact timing. Tropical cyclones are tracked by satellites, radar, instrumented aircraft and numerical models, with track forecasts usually reliable two to five days ahead and intensity forecasts much less so. Wildfires are forecast through fire danger indices that combine temperature, humidity, wind and fuel moisture, and monitored by satellite hotspots, lightning detectors, lookouts and drones, giving hours to days of warning of the conditions, but very little of where a fire will start.

Perception of risk is how people and governments judge a hazard: how likely they think it is, how serious they think it will be, and whether they are prepared to accept it. It explains almost everything that looks irrational. People who have never experienced an event underestimate it; people who have experienced several become fatalistic or stop responding to warnings; people weigh the certain benefits of fertile volcanic soil, coastal fishing or cheap housing against an uncertain future loss, and stay. Optimism bias ('it will not happen in my lifetime'), warning fatigue from false alarms, mistrust of authorities, the cost of evacuating with livestock, and the simple fact that a decision to leave has an immediate cost while the benefit is only probable, all reduce the take-up of warnings. Risk perception also runs the other way: a spectacular, unfamiliar hazard such as a tsunami frightens people more than a familiar one such as a river flood that kills more people each year.

Environmental and infrastructure design or modification reduces the physical impact of the hazard. For earthquakes: reinforced concrete frames, shear walls, cross-bracing, base isolation and light roofs so that buildings sway rather than collapse; retrofitting of older buildings, bridges and hospitals; and ground improvement or avoiding liquefiable ground altogether. For volcanoes: earth barriers and water to divert lava, diversion channels for lahars, strengthening of roofs against ash, and the relocation of critical infrastructure out of the highest-risk zone. For cyclones: sea walls, storm-surge barriers and levees; soft engineering such as mangrove belts, dune systems and coral reef protection that absorb wave energy; cyclone shelters on raised ground; and buildings tied down, shuttered and roofed to resist wind. For wildfires: fuel reduction by controlled burning, grazing and thinning; firebreaks; clearance of vegetation around settlements; fire-resistant building materials, closed eaves and sprinkler systems.

Planning, education and awareness work on exposure and vulnerability together. Hazard maps show which land floods, shakes, burns or lava flows, and land-use zoning keeps new housing, schools and hospitals off the worst of it; building codes set the engineering standard and, crucially, are enforced; evacuation plans designate routes, shelters and triggers, and are rehearsed in drills at school and at work; education teaches people what a warning means and what to do; insurance spreads the cost of loss and, in some countries, is refused for the most exposed properties, which acts as a planning control. The weakness is always enforcement: zoning and codes are resisted where land is scarce, housing is informal and enforcement is poorly funded.

Short-term emergency response covers the first days: warning dissemination, pre-emptive evacuation, search and rescue, medical treatment, emergency shelters, food, water and sanitation, restoring communications and keeping lifelines running. Long-term recovery covers the years after: rebuilding to higher standards - build back better - rehousing on safer land, restoring livelihoods, repairing infrastructure, psychological support, and revising plans in the light of what went wrong. The temptation after a disaster is to rebuild exactly what was destroyed as fast as possible; the opportunity is to use the political will and the aid money to reduce the next disaster, and that opportunity is usually lost within a year or two.

• Strengths of pre-event strategies: they are permanent, they protect everyone regardless of income, and they cost far less than rebuilding - but they must be funded every year whether or not an event occurs, and their success is invisible.
• Weaknesses of pre-event strategies: monitoring is expensive, codes and zoning are hard to enforce, false alarms erode trust, and defences designed for one size of event are overwhelmed by a larger one.
• Strengths of post-event response: it is immediately visible and politically necessary, and it saves lives in the hours after an event - but it arrives after the deaths and depends on roads, hospitals and coordination that may themselves be damaged.
• The judgement examiners want: engineering and planning reduce loss of life most per dollar over time, but they only work where there is money, enforcement and public trust, so the effective answer in a low-income country is often community-based preparedness and simple, affordable engineering.

A multi-hazard environment: the Philippines

The Philippines is the best example in this syllabus of an environment where hazards overlap. It is an archipelago of more than 7 000 islands sitting between the Philippine Sea plate to the east and the Eurasian plate to the west, with subduction along the Philippine Trench and the Manila Trench, an active strike-slip fault system running through the islands, and at least twenty active volcanoes. On top of that it lies in the western Pacific typhoon belt, so tectonic and climatic hazards strike the same places, sometimes in the same week.

The tectonic hazards come from that setting: shallow and deep earthquakes along the trenches and the Philippine Fault, several of them destructive in the last half century; explosive volcanism from strato-volcanoes such as Mayon, Taal and Pinatubo, whose 1991 eruption was a VEI 6 that cooled global temperatures for a year or two and left lahars choking the river valleys around it for years afterwards; and tsunami generated by offshore subduction earthquakes. The climatic hazards come from around twenty tropical cyclones entering the country's area of responsibility each year, about half of which make landfall, bringing high winds, storm surges, intense rainfall, river floods and landslides. Typhoon Haiyan in November 2013 was one of the strongest storms ever to make landfall, and the surge it pushed into the city of Tacloban killed thousands of people and flattened whole coastal districts.

Management is organised around three national agencies: PAGASA for weather, floods and cyclones, PHIVOLCS for earthquakes, tsunami and volcanoes, and the NDRRMC for coordinating disaster risk reduction and response, with hazard maps for all of them published through the GeoRisk Philippines portal. The Philippine Disaster Risk Reduction and Management Act of 2010 moved the emphasis from relief to preparedness, requiring local government units to hold their own plans and funds. In practice this means Doppler radar and satellite tracking with typhoon warnings issued days ahead and storm surge advisories issued since Haiyan; a volcano alert-level system with pre-emptive evacuation of declared danger zones, which emptied the volcano island and its lakeshore villages before Taal's 2020 eruption; seismic monitoring and tsunami warnings; structural codes for new buildings; designated shelters and evacuation routes; no-build zones and mangrove replanting along exposed coasts; and community drills in schools and barangays.

Evaluating the success of this management means holding two ideas at once. Warning and evacuation are genuinely effective and have improved: death tolls from comparable typhoons have fallen, tens of thousands of people are moved out of the path of eruptions before they happen, and the 2010 Act gave local government a legal duty to plan. But the underlying vulnerability has not been removed. Rapid population growth has pushed informal settlements onto riverbanks, steep slopes and the coast; zoning and building codes are unevenly enforced where land is scarce and people are poor; reconstruction after a major typhoon is slow; and capacity varies enormously between wealthy cities and remote rural municipalities. The hardest problem is the interaction between hazards: ash on the slopes of Pinatubo became lahars in the next rainy season, and an earthquake during a typhoon would defeat a response system built to deal with one hazard at a time. A multi-hazard approach therefore has to plan for combinations, and to treat land-use planning and poverty reduction as part of hazard management.