Assignment

Can Poor Countries Afford to Prepare for Low-Probability Risks?

Michele McNabb, Freeplay Energy

Kristine Pearson, Freeplay Foundation

Overview This chapter examines how less-developed countries can prepare

for low-probability risks in the face of so many other pressing needs such as health care, education, clean water, and roads. The case of Cyclone Nargis in Myanmar (Burma) is examined, among other recent disasters. Although it would be easy to vilify the globally unpopular government of Myanmar and blame it for the deaths of 130,000 people, the truth is more complex. Cyclone Nargis hit a part of the country with a low probability of cyclones. Can poor countries like Myanmar build comprehensive cyclone warning systems for areas that might not face another cyclone for decades or even cen- turies? What good is a warning system if the people are too poor to evacuate and roads are nearly nonexistent? With global climate change, anomalistic weather is projected to increase, meaning extreme weather events are likely to impact areas previously unaf- fected. How can early-warning systems be created for every potential disaster in every region? The authors argue that developing countries cannot afford to build individual early-warning systems for low- probability disasters, so they must rely on (1) multihazard warning systems, (2) disaster risk-reduction education and training, (3) low- cost/low-technology solutions, and (4) multiuse communication

6

100

From the Library of Daniel Johnson

structures that can serve general development and early-warning purposes.

Can Poor Countries Afford to Prepare for Low-Probability Risks?

Natural disasters don’t discriminate between rich people and poor people. Whether over shacks or mansions, floodwaters rise evenly, and hurricane winds blow with the same intensity. Yet the poor, the disenfranchised, and the weak usually suffer the greatest loss of life and lose a greater proportion of their livelihoods in disas- ters. The reasons disasters have such disproportional effects on the poor are clear: They have less ability to prepare for and mitigate the effects of disasters. They live in flimsy houses. They are likely to hold any wealth in assets like cattle or jewelry instead of in banks. They don’t have insurance. Their jobs often depend on the land. They lack vehicles or money to evacuate quickly. They may lack means of com- munication to learn of an impending disaster or to plan an escape.

In rich countries, governments normally accept the responsibility to provide extra assistance to vulnerable groups when a disaster strikes. When the world’s richest country failed to protect its most vulnerable citizens as Hurricane Katrina hit New Orleans, the entire moral fiber of the country was shaken. Stories of old people aban- doned by their caregivers and the obvious racial and socioeconomic composition of people crowded on roof tops and in sports stadiums caused countless editorials and deep questioning about government’s duty to protect its own most vulnerable citizens. Barack Obama high- lighted the moral responsibility of governments when he accepted the Democratic nomination for president, saying “We are more com- passionate than a government...that sits on its hands while a major American city drowns before our eyes.”1

However, what about the governments of poor countries? Do they not have the same moral responsibility to protect their citizens? Cyclone Nargis killed 130,000 people in 2008 in Myanmar (Burma). Although it would be easy to vilify the globally unpopular military gov- ernment of Myanmar and blame it for the atrocities, the reality is much more complicated. The government clearly impeded relief

6 • CAN POOR COUNTRIES AFFORD TO PREPARE FOR LOW-PROBABILITY RISKS? 101

102 LEARNING FROM CATASTROPHES

operations in the first days after the storm due to a combination of bureaucracy, paranoia, and paralysis; however, many other factors sug- gest that blaming the Myanmar government alone would be simplistic.

First, the government’s meteorological department did issue cyclone warnings, and the excellent international tropical cyclone warning system functioned well, tracking the storm at sea for 6 days and sending warnings from the regional center in India to Myanmar 48 hours in advance. It is unclear how many people living in the Irrawaddy Delta actually received the warnings, but even if they had received them, the poor roads, lack of transport, and extreme poverty in the region would have prevented a mass evacuation. If a govern- ment cannot afford schools, roads, or hospitals, how it is supposed to protect its citizens from the ravages of nature?

Second, Cyclone Nargis was a highly unusual event, hitting a part of the country that had not experienced cyclones for decades.2

Parts of Myanmar regularly experience cyclones, but the wide flat floodplains along the Irrawaddy Delta had not faced a cyclone in more than 40 years—indeed, the director of a U.S.-based weather service called Nargis “one of those once-in-every-500-years kind of things.”3 The storm surge quickly flooded the vast flat plains where most of the population lives. Although there had been investment in early warning for high-probability areas, similar investments were not made in the delta because of the low probability of cyclones. So how can a poor country prepare itself for events that might not recur for decades?

Third, some experts claim the unusual trajectory and intensity of Cyclone Nargis resulted from climate change. The Centre for Science and Environment in India claimed that Nargis was “not just a natural disaster, but a human-made disaster caused by climate change.”4

Although most scientists warn again labeling a single event a “sign” of climate change (extreme, anomalistic events have always occurred), clear consensus exists that the intensity of severe weather will increase and that areas previously unaccustomed to cyclones, floods, heat waves, and so on will face these threats.5 So, does the moral responsibility for preparing a poor country like Myanmar from climate-change-induced hazards lie with the poor country or with the

6 • CAN POOR COUNTRIES AFFORD TO PREPARE FOR LOW-PROBABILITY RISKS? 103

countries that created the greenhouse gases and caused climate change?

Finally, some environmentalists pin Nargis’s high death toll on the destruction of mangrove swamps for rice and shrimp farming; whereas other experts say years of Western sanctions against the mil- itary regime had only exacerbated poverty and driven the environ- mental destruction as millions of desperately poor struggled to survive. Therefore, protecting the most vulnerable from disasters— especially low-probability disasters—is exceedingly complex.

State of Early Warning in Developing Countries: High-Probability Events

Over the past 30 years, significant investment has been made in early warning throughout the world. Advances in technology have greatly increased our ability to predict many types of natural hazards. Scientists have a much greater understanding of the earth’s weather systems thanks to satellites, sensors, radar, and computer modeling. Global cooperation has led to major improvements in short- and long- term forecasting, with rich and poor countries working together to share resources and knowledge. The global tropical cyclone6 warning system described next is perhaps the best example of the benefits of global technical and scientific collaboration in early warning.

Although there is room for improvement in nearly every early- warning system, most countries have at least rudimentary ability to provide warnings for high-probability events—events that have hap- pened regularly in the past and are expected to continue. Many flood- prone countries operate flood early-warning systems; most cyclone-prone countries have strong early-warning systems in place; and drought prone countries in Africa monitor rainfall and crop conditions to sense the onset of drought before conditions lead to food insecurity or famine.

For example, Bangladesh, Cuba, and Mozambique, although some of the world’s poorest countries, have good early-warning systems to cope with cyclones, which hit all three countries nearly

104 LEARNING FROM CATASTROPHES

every year. Each of the three countries has adopted cyclone-warning systems tailored to their specific geographic, cultural, and political situation. In all cases, the strength of their systems is based not only on expensive technologies but also on the involvement of the vulner- able populations themselves. Scientific knowledge that a cyclone is developing in the ocean is meaningless unless there are also systems to get the warnings out to at-risk populations. But even that is not enough; vulnerable groups must have options available to protect themselves when they receive the warning.

One of the reasons Bangladesh’s cyclone-warning system is widely lauded is that it not only provides official warnings from authorities and includes an extensive network of volunteers who com- municate down to the village level, but the country has also built sev- eral thousand cyclone shelters. The raised shelters provide emergency accommodation in safe durable buildings not far from people’s homes. Thousands of the poorest people in the poor nation of Bangladesh now have a place to go when a cyclone hits. When the massive Cyclone Sidr struck in November 2007, 3,400 people still died, but an estimated 1.5 million people sought refuge in 2,168 cyclone shelters.7 The death of more than 3,000 people is a terrible tragedy, the number of shelters is still woefully inadequate, and main- tenance is a perpetual challenge, but without the option offered by the shelters, the death toll would have been much higher.8

The cyclone early-warning systems in many other countries lack this last component—options of last resort, especially for people who do not have the means to evacuate or live in poor communities with no sturdy buildings able to withstand the wind and rain of cyclones. To advise people to evacuate when they have no vehicles, to take shel- ter when they have no permanent buildings, or to seek higher ground on a flat flood plain is of little value.

For high-probability events, successful models exist, and most countries have implemented early-warning systems. Regional and global collaboration assist governments with the information, resources, and tools to build and continuously improve these sys- tems. If this is true, then how could more than 200,000 people per- ish in the Indian Ocean tsunami? The answer is simple: The tsunami was a low-probability event.

6 • CAN POOR COUNTRIES AFFORD TO PREPARE FOR LOW-PROBABILITY RISKS? 105

This chapter argues against spending resources to establish early-warning systems for low-probability events. Highly anomalis- tic events will continue to happen, and it is impossible to be fully prepared for every event that might happen once in 50 or 100 or 1,000 years. It is a poor use of scarce development resources to set up a tsunami early-warning system in Somalia, Kenya, Tanzania, or Madagascar—all of which suffered minor impacts in the Indian Ocean tsunami. It is well documented that early-warning systems that are not activated regularly lose effectiveness and fall into disrepair.9,10

State of Early Warning in Developing Countries: Low-Probability Events

The challenges poor countries face in protecting their poorest citi- zens from high-probability events are magnified in the case of low- probability events. If Mozambique struggles to create and maintain flood and cyclone early-warning systems despite the fact that it faces an average of 3 cyclones every season and it has 11 international rivers flowing through its soil, how can it possibly prepare for a tsunami that might not occur in a lifetime? Should it use scarce resources to educate its citizens about tsunamis or other low-probability events that may not strike for decades or even centuries?

The tsunami early-warning system for the Pacific Ocean is more than 40 years old. One of the drivers of the system was the 1960 Chilean earthquake, the most powerful instrumentally recorded in his- tory. The resulting tsunami affected Chile, Hawaii, the Aleutian Islands in Alaska, California, Samoa, Japan, the Philippines, New Zealand, and Australia. Hilo, Hawaii, was one of the worst effected cities, where waves as high as 35 feet were recorded.

Based outside of Honolulu, Hawaii, the Pacific Tsunami Warning Center has provided dozens warnings for the Pacific Ocean countries. Scientists continuously improve the system, tracking the effects of underwater earthquakes and landslides to model the potential occur- rence of tsunamis. But no system was in place in the Indian Ocean

106 LEARNING FROM CATASTROPHES

because tsunamis are relatively rare in that basin. In fact, as millions of dollars were rushed into creating a tsunami early-warning system for the Indian Ocean after the fact, some experts were suggesting the money would be better spent creating a system for the Mediterranean Sea—statistically more vulnerable to tsunamis than the Indian Ocean, with 140 million people living near the shoreline, plus millions of tourists visiting at any given moment. Some basic work has gone into the Mediterranean tsunami early-warning system, mainly in terms of data collection through ocean buoys, but little public outreach or edu- cation has taken place.11 A large undersea earthquake triggering a tsunami in the Mediterranean today could kill tens of thousands. Based on probabilities and risk, additional investment in public education about tsunamis in the Mediterranean is urgently needed.12

After a mega-disaster, there is a strong desire to “do something.” Immediately after the Indian Ocean disaster, money was put into tsunami early warning in Tanzania and other East African countries that suffered only minor losses during the event and had never expe- rienced a tsunami before. More than $10 million was requested for tsunami-related activities in Somalia, for example. Such a reaction, although understandable, is misguided. Money spent on single haz- ard early-warning systems for low-probability events in poor coun- tries could be better spent on general development activities that can contribute to disaster preparedness: improved communications, roads, education, building codes, and so forth.

High or Low Probability: How Climate Change Is Changing the Nature of Risk

Risk assessment is considered the first step in risk management. These risk calculations are often based on available historical data. The most common type of probability calculations are “return peri- ods” on flooding—calculations of how often a specific size of flood occurs, based on data from the past 100 or more years. A rigorous analysis of data can allow experts to classify floods as a “1 in 100-year event” or a “1 in 10-year event” with some degree of confidence.

However, it is widely agreed that climate change is making fore- casts based on historical data less relevant (although how much less

6 • CAN POOR COUNTRIES AFFORD TO PREPARE FOR LOW-PROBABILITY RISKS? 107

relevant remains unknown). The past is no longer an accurate predic- tor of future events—floods are occurring more frequently and impacting areas previously unaffected. Changes in temperature mean that vector borne diseases such as malaria may spread to areas previously free from the disease, unusual heat waves are affecting Europe, and Australia is suffering severe drought and wild fires, for example.

Climate change means risk-assessment methodologies must fac- tor in a greater degree of uncertainty than simply an analysis of histor- ical data. Increased uncertainty makes it that much more difficult to label an event “high probability” or “low probability.” Some high- probability events will become more extreme meaning an even higher level of preparedness is needed. Some low-probability events may happen more frequently or affect new areas, blurring the line between high- and low-probability events.

Who Should Pay for Early Warning? Just as natural hazards don’t discriminate between rich and poor,

they know no borders. The same hurricanes threaten Cuba, the United States, and Mexico. Ashes from a volcano eruption in the Philippines affect rainfall and weather patterns around the world. Droughts in Africa cause people to cross borders in search of water and food. Heavy rainfall in landlocked countries floods coastal com- munities thousands of miles away. So while the state bears primary role for early warning, responsibility and funding is rightly shared from the global to the household level.

Arguably the best example of integrated early-warning system with collaboration from the international to the household level is the global cyclone early-warning system. The World Meteorological Organization’s global operational network enables continuous obser- vation, data exchange, and regional forecasting. Six regional special- ized meteorological centers around the globe provide forecasts, alerts, and bulletins to national meteorological services to all coun- tries at risk with lead times of 24 to 72 hours. The national services then issue warnings to government, media, and the general public according to national protocols. Historical risk areas are well

108 LEARNING FROM CATASTROPHES

established (although historical patterns are becoming a less reliable predictor, as demonstrated in the case of Myanmar), and five Regional Tropical Cyclone Committees work continuously to enhance forecasting skills of all members. Costs and responsibilities are shared and the system works extremely well.

Global Collaboration

Global collaboration in early warning has been driven by both humanitarian imperatives and self-interest. Often, the two are entan- gled and difficult to separate. More technologically advanced coun- tries may support enhanced flood or tsunami warning systems, but they also sell river gauge or ocean-monitoring equipment, sophisti- cated computer modeling capacity, and so on.

There is nothing inherently wrong with self-interest playing a role in global collaboration on disasters—in fact, a greater recognition of the potential benefits to richer countries can help increase invest- ment in early warning and disaster risk reduction in poor countries. Unmitigated disasters often lead to a downward spiral of poverty, increased social inequities and tensions, and even migration. For example, families lose their homes and assets in a hurricane or earth- quake; they are forced to send family members on difficult and dan- gerous journeys to find work abroad; and unskilled and uneducated migrants may fail to find work and end up relying on social programs in the host country or turning to illegal activities to survive and sup- port family at home. Disaster risk-reduction and -mitigation efforts can reduce this downward spiral.

Modeling of the global climate system has lead to significant advances in understanding how sea surface temperatures (El Niño and La Niña) affect seasonal weather patterns, which, in turn, has important implications for drought, flood, and malaria early warning in developing countries. Satellites launched into orbit for weather- monitoring purposes in developed countries were inadvertently dis- covered to have the capacity to monitor vegetative vigor on the ground—which has become a key indicator for drought early warning in Africa. River systems modeling developed in Japan and the United States has been shared with developing countries where the under- standing of rivers and flooding was nonexistent or rudimentary.

6 • CAN POOR COUNTRIES AFFORD TO PREPARE FOR LOW-PROBABILITY RISKS? 109

Regional Collaboration

Regional collaboration in early warning is generally weak, especially in the developing world. There are a few examples of regional bodies that successfully share information and resources, leading to better early warning, such as CILSS (Comite Permanent Inter-Etats de Lutte Contre la Secheresse) for drought in West Africa and the Mekong Delta River Authority for flooding in Southeast Asia. But there are many more cases where regional cooperation is ineffective.

Flood monitoring in sub-Saharan Africa offers an example of how the weaknesses in regional collaboration negatively impacts early- warning efforts and increases people’s vulnerability. Many of Africa’s river basins are international—meaning that rain falling in one country will eventually find its way into rivers that pass through other coun- tries, potentially causing flooding downstream even if there is no rain locally. It is imperative for downstream communities to know how much rain is falling upstream and how fast the rivers are rising, yet in many cases, the information is not readily shared. Linking upstream information with downstream communities can provide 48 to 72 hours’ warning of an impending flood and enough time to evacuate their assets and move to higher ground. These types of upstream-to- downstream linkages have been created in some places, notably in Central America, but examples of multicountry collaboration in con- tributing to regional early warning are few.

One reason regional collaboration is weak may be a lack of perceived self-interest. Using the flood early-warning example, four Southern African countries share the Limpopo river basin, yet all of the water eventually flows through Mozambique into the Indian Ocean, presenting regular flood risks to communities living near the river’s mouth. Even though the flood risk is only in Mozambique, most of the catchment area is in South Africa, with small parts in Botswana and Zimbabwe.13 For South Africa, heavy rainfall in the high elevation catchment areas has little national impact and only lim- ited local impact. South Africa may accept it has a humanitarian imperative to help its poorer neighbor (its military has dispatched hel- icopters to rescue Mozambicans stranded by floodwaters in 2000), but when South Africa itself has countless internal demands for improved

110 LEARNING FROM CATASTROPHES

housing, health care, and schools, should it prioritize the establish- ment of an extensive network of rainfall gauges for the main benefit of its neighbor?

State’s Responsibility

The World Conference on Disaster Reduction in Kobe, Japan, in January 2005 released the Hyogo Framework for Action 2005-2015: Building the Resilience of Nations and Communities (HFA).14 The HFA placed the primary responsibility for implementation of early warning and disaster risk reduction on national governments. Ensur- ing its citizens’ safety is a primary responsibility of government, and national leadership and ownership are keys to effective early warning and early action.

Even poor countries can mount effective early-warning systems for high-probability events. The example of Cuba’s hurricane early- warning system illustrates this clearly. The country is hit by hurri- canes nearly every year, yet the fatalities in storms are usually far fewer than on neighboring islands. A combination of effective gov- ernment planning, annual simulation exercises, and citizen responsi- bility ensures everyone is aware of a storm’s approach and knows exactly how to respond.15 Yet nearby Haiti suffers immensely from storms, and its government is ill equipped to protect its citizens. The lack of government preparedness coupled with decades of deforesta- tion has resulted in uncontrolled landslides and flooding, further exacerbating the problem and making its population even poorer (in fact, the poorest in the Western Hemisphere).

When a state lacks human or financial resources to protect its cit- izens or lacks the commitment, what can and should the global com- munity do to protect the most vulnerable? This issue is addressed in the final section.

Community Responsibility

In the last decade, significant advances have been made in rec- ognizing a community’s responsibility for protecting themselves from disasters. Even the most vulnerable communities should not be seen as helpless victims of a natural disaster but as the group with the largest vested interest in early warning and early action.

6 • CAN POOR COUNTRIES AFFORD TO PREPARE FOR LOW-PROBABILITY RISKS? 111

Empowering communities to help themselves prevent and mitigate disasters is many times more effective than any other form of risk reduction.

Initial advances in early warning had a strongly scientific bias. Investments were made in technological solutions without much practical thought about how communities would be warned and how they would react to warnings. This mistake is still being made as evidenced by the lack of progress over the past few years in building a tsunami early-warning system for the Mediterranean. Almost no resources have been devoted to educating the millions of people who live near the shores of the Mediterranean about how to recog- nize a tsunami and what actions to take. This is a disaster waiting to happen—the threat exists, and little has been done to mitigate the risks for the vulnerable people.

Community involvement in early warning and early action must incorporate three levels of preparedness:

1. Communities need to understand their risks. All people, not just those in developing countries, misjudge their risks because they don’t have an objective basis for assessing them over a long time frame. When a catastrophic, low-probability hazard occurs, like the Indian Ocean tsunami, people along coastlines worldwide overestimate their risk of another event. Communities living in floodplains generally have a good con- ception of how often small, medium, or large floods occur—but if a large devastating flood occurred recently, they often overes- timate their risks.

It has been believed that bringing scientific data on historical disaster patterns (100-year records of rainfall or cyclone tracks) together with local knowledge about past events gathered from older residents or collective memory was the best way to assess risks. However, with climate change, there is a new challenge: Those historical records or stories passed down over genera- tions may no longer predict future frequency or intensity. There is evidence in the Mozambique floods of 2000, for example, that people were warned on radio the impending floods would be “a major flood,” which community elders interpreted to mean sim- ilar in magnitude to the floods experienced in the early 1900s.16

112 LEARNING FROM CATASTROPHES

However, the floods vastly exceeded floods in anyone’s living memory.

It is essential that communities map their own knowledge of prior events, incorporate any available data on historical events, but also understand that climate change adds a level of uncer- tainty never before experienced. The past is no longer an accu- rate guide to the future of disaster events.

2. Communities need to understand warnings from out- side, and their local knowledge needs to be shared and assessed scientifically. Warnings may be shared from national authorities down to community level, but insufficient analysis has been carried out on how the messages are perceived. In some cases, messages may lack specificity (“floods are likely along the Mekong”) or be overly technical (“a Category 3 cyclone with maximum sustained winds of 150km/hour will hit between 2100 and 2300 hours tomorrow”). Although these messages provide a basic level of alert, they are unlikely to engender any actions from the vulnerable populations—unless they have confirming evidence from traditional warning indicators.

Almost all communities have “traditional” hazard warnings embedded in their culture. In developing countries, many of these warning signs are based on animal behavior including birds singing at unusual times of day, monkeys or small animals fleeing an area, or livestock refusing to approach shorelines.17

In the past, local knowledge about early warning signs has been largely dismissed as unscientific, but it is increasingly clear that local knowledge can complement technical warnings. For example, in Mozambique, downstream communities watch the color of the river water and the size and type of debris floating down to judge the magnitude of a potential flood.

In Simeuleu, an island off the coast of Indonesia only about 100 kilometers from the epicenter of the earthquake that triggered the tsunami, only 7 people of a total population of 83,000 were killed. The island had suffered from a tsunami in 1907 and knowledge of the warning signs—especially the ocean receding after an earthquake—had been passed from generation to generation through songs and poems. Instead of dismissing this

6 • CAN POOR COUNTRIES AFFORD TO PREPARE FOR LOW-PROBABILITY RISKS? 113

local knowledge, it should be studied and integrated into warning systems as appropriate.

3. Communities need to have options if a warning is sounded. Although this step sounds obvious, it has received remarkably little attention by practitioners and is a missing link in many early-warning systems in developing countries. In the case of Cyclone Nargis, for example, even if risk assess- ments had been carried out and warnings had been received by all people living in the Irrawaddy Delta, there would have been little concrete action they could have taken to escape the devastation. The roads were insufficient to handle a mass evacuation, the population lacked means of transport, and people were unwilling to leave their assets behind for fear of theft.

The Bangladesh cyclone early-warning system presents one of the few examples of warning systems that have found ways to increase communities’ options for responding to a warning. Poor populations in Bangladesh face almost the same chal- lenges that affected the residents of the Irrawaddy Delta. With few durable buildings, options to shelter from cyclonic winds were few, and even robustly constructed buildings could be flooded by the heavy rains accompanying the cyclone. But Bangladesh has built a series of more than 2,000 raised cyclone shelters, near the at-risk areas. People no longer have to choose between leaving all of their belongings and livestock and flee- ing long distances.

Understanding and supporting traditional means of self- protection can be an important step. For example, much global attention was paid to the birth of Baby Rosita in a tree in Mozambique during the massive floods of 2000. Press reports created images of a pregnant woman suddenly stranded in a tree. Almost no acknowledgment of the full story took place— Rosita’s family began constructing a platform shelter in a tree several days before the big floods came and storing critical sup- plies there, following traditional practices. In the low-lying floodplains of southern Mozambique, there is nearly no high ground. A few trees are the only thing standing a few meters above the waterline, and so the idea of building shelters in

114 LEARNING FROM CATASTROPHES

trees is perfectly rational. Efforts to provide options for react- ing to a disaster could involve developing improved techniques for building tree shelters, providing durable construction mate- rials, and advising on sanitation during an extended period in the tree shelter.

Options for Early Warning for Devastating Low-Probability Events in Poor Countries

Can poor countries ever be prepared for catastrophic, low- probability events? Should they allocate scarce resources to an event that may not happen for decades or even centuries when they have critical needs for investment in education, health, and infrastructure? Or does the industrialized world have a responsibility, given its contribu- tion to climate change, for funding early warning for low-probability, high-impact events in developing countries?

The solution to these difficult challenges isn’t fatalism or inaction. A combination of four actions can help mitigate the effects of anom- alistic events and also can contribute to general development:

1. Multihazard early-warning systems

2. Disaster risk-reduction education

3. Low-cost/low-technology solutions

4. Multiuse communication structures

Multihazard Early-Warning Systems

The concept of multihazard early-warning systems is gaining favor around the world, especially for places that face numerous, fairly low- probability events. In many cases, high-probability events need their own early-warning systems because every part of the hazard is unique, from data collection to response options. For example, an area might face high risks of both volcanoes and droughts. But the chain of information flow and action is completely different between a volcano early-warning system and a drought early-warning system, for exam- ple, so separate systems make sense.

6 • CAN POOR COUNTRIES AFFORD TO PREPARE FOR LOW-PROBABILITY RISKS? 115

Multihazard early warning does not advocate for the creation of a mega-early-warning system but for grouping low-probability hazards and for sharing data and structures among all systems when it makes sense. One notable effort is the Global Earth Observation System of Systems (GEOSS) initiated by the Group on Earth Observation, which intends to build on existing information systems such as the Global Telecommunications System and add new initiatives to create timely, accurate, and interoperable data on all aspects of the earth, for use in early warning, risk reduction, and other endeavors.

For poor countries facing multiple low-probability hazards, a multihazard system can increase efficiency and reduce costs. Group- ing hazards also increases the likelihood that the system will be trig- gered more frequently—which is absolutely essential for early-warning systems to improve over time.

Shanghai, China, has pioneered efforts to build a multihazard early-warning system that can be a useful model for mega-cities in the developing world. The approach grouped together the many hazards faced by the 17 million residents—typhoons, tornados, strong winds, and floods and also chemical spills, nuclear accidents, public health emergencies, and so on. The Shanghai system has integrated a “top- down” approach with unified policies, data collection systems, and multiagency command structures with a “bottom-up” approach that ensures the community is aware of the risks, understands appropriate responses, and can channel information upward to emergency response authorities and receive information transmitted from authorities.18

Similar principles can be applied to nonurban settings, too, with the focus on disaster risk-reduction education and low-technology solutions, and on multiuse communications systems.

Disaster Risk-Reduction Education

Education plays a fundamental role in reducing disaster risks, whether for high-probability or low-probability events. It could be argued that education is even more important for low-probability events because people will have their own firsthand knowledge and experience with high-probability events, whereas for low-probability events, this personal experience will be missing.

116 LEARNING FROM CATASTROPHES

This was compellingly illustrated by 11-year-old Tilly Smith during the Indian Ocean tsunami. The British schoolgirl was vacationing in Phuket, Thailand, with her family when she saw an event exactly as her geography teacher had described to her in England a few weeks ear- lier. The ocean was bubbling “like the foam on a beer,” said Tilly, and receding—the signs of an impending tsunami. She warned her family and dozens of other hotel guests on the beach, all of whom escaped to safety.19 Initially, her mother did not believe her and nearly refused to leave the beach, because she had never heard the word tsunami and didn’t understand the risks. If all the local and visiting schoolchildren in Phuket had studied the same lesson, the death toll could have been greatly reduced.

Educational programs also teach people about general safety improvements they can make to their homes to become better pre- pared for many types of hazards. Various low-cost/low-technology solu- tions can help prepare people for many types of disasters.

Low-Cost/Technology Solutions

Preparedness for low-probability events does not have to be expensive. It can involve “normal” development interventions that have additional benefits if a low-probability event strikes. Supporting the development of small-scale savings programs, where poor people trust banks and begin to accumulate money in cash rather than assets, can play a big role in recovery after a disaster. Improving construction practices in “nonengineered,” traditionally constructed buildings sim- ilarly can provide better, safer, and more comfortable living condi- tions, but also can enhance resilience to disaster events.

Many governments have tightened building codes to stop the construction of flimsy housing that easily collapses in an earthquake, cyclone, or flood. After the Bam earthquake in 2003 killed more than a quarter of the town’s 100,000 people, Iranian authorities banned traditional mud and adobe houses and prevented the building of dome structures.20

Avalanches are becoming more common in parts of the Alps, per- haps due to climate change. The Swiss government operates sophisti- cated detection and early-warning systems, but local residents near

6 • CAN POOR COUNTRIES AFFORD TO PREPARE FOR LOW-PROBABILITY RISKS? 117

Davos have taken their protection into their own hands with low- technology solutions. Every year, they trek up the mountains to install snow fences to impede avalanches. At the same time, they have planted thousands of trees. When the trees grow to full size in 40 to 50 years, they will replace the fences and break the movement of avalanches.

Although insurance is scarce in developing countries in general, and insurance against low-probability events even more unlikely, households can be encouraged to take actions to protect themselves from future disasters. With climate change, farmers who traditionally counted on one moderate drought every three years and one major drought every decade may suddenly face a major drought every three years. This means that farmers must be encouraged to adapt to the reality that the future may be even harder to predict than the past. Engaging farmers in climate change adaptation discussions is essen- tial. Radio programs, extension messages, and SMS/cell phone text messaging technologies can help inform farmers about how climate change may impact the predictability of the seasons and require changes in farming practices.

Thorough risk assessments at the community level can help iden- tify mechanisms through which the community can become more resilient.

Multiuse Communications Systems

Communication saves lives during a disaster. Communications are also an essential component of a nation’s development. How can authorities in the national or regional capital get information to the community level? And how can receivers of information at a commu- nity level get it out to the most remote members of community—the so-called last mile? And how can communications flow from the remote members of the community upward to national authorities, in the case of a disease outbreak, for example? A communications assessment can reveal gaps in the ability to reach certain groups of people. For example, cell phones are becoming common throughout the developing world. Many communities have cell phone coverage and at least a few residents own cell phones. Cell phone text

118 LEARNING FROM CATASTROPHES

messages about an impending disaster might reach the residents with cell phones. But how do those residents spread the word to others who are less likely to own cell phones such as child-headed households, rural women, elderly, or disabled? Redundancy in com- munications structures is essential, too. What happens if the cell phone network is damaged in an earthquake, for example?

Special attention needs to be paid to the vulnerability of groups such as women, children (especially girls), and the disabled and ill. The groups may fall outside the normal flow of information in a community—HIV/AIDS widows or orphans may be shunned by the community and have no regular or reliable access to information flowing into the area. Vulnerable groups have less money to buy cell phones or radio batteries, so they may live on the margins of the com- munity. Enhancements to communications systems need to focus on these traditionally neglected groups.

Various options for “last mile” communications are being tested around the developing world.21 Cell phone broadcasting of warnings has potential applications, but with the limitations previously described. In some densely populated places, volunteers with megaphones and whistles can pass through the community sounding an alarm. In many places in the developing world, radio broadcasts remain the most effective means to reach large numbers of people, even when they are dispersed widely. Local community radio stations that broadcast in local languages can receive and rebroadcast warn- ings effectively, provided they have not been damaged by the event.

The RANET project22 has created models to disseminate infor- mation via satellite to community radio stations, which then broadcast them over FM frequencies to local populations. In places where radio ownership is limited by poverty, or where people cannot afford to buy disposable batteries every few weeks, windup and solar-powered radios, like the Freeplay Lifeline radio, provide a low-technology, low-cost solution. If community radio stations are damaged, these radios can receive warnings broadcast over regional, national, or shortwave frequencies.

6 • CAN POOR COUNTRIES AFFORD TO PREPARE FOR LOW-PROBABILITY RISKS? 119

Ensuring communications to at-risk populations has many other benefits. When no disaster threatens, radios can broadcast health, agriculture, and educational programming, for example. The RANET stations described above educate farmers about seasonal forecasts as well as planting techniques and market prices. Communications systems put in place for early warning can help medical staff evacuate injured or ill patients from remote areas.

Conclusion The world is confronted by almost limitless natural hazards, some

regular and fairly predictable and others extremely rare and unpredictable. Although it is reasonable to expect all countries to protect their citizens from disasters that occur regularly, it is difficult, even for rich countries, to prepare for low-probability events. Although there are many remaining uncertainties about global climate change, there is consensus that the number, location, and intensity of natural hazards will become less predictable. For poor countries, with critical needs for investment in health, education, and infrastructure, investments in single-hazard early-warning systems for low-probability events are unwise.

Following a mega-disaster like the Indian Ocean tsunami, it is understandable that the world would want to help not only the coun- tries devastated by the tsunami (for example, Indonesia and Sri Lanka), but also countries that suffered only slightly from the disaster (for example, Tanzania and Somalia). This chapter has argued that investments in tsunami early warning in East Africa could be much more wisely made in multihazard early-warning systems, disaster risk-reduction education, or even general development activities that contribute to risk reduction such as improved building codes or bet- ter communications. If emotion were completely removed from the risk-assessment process, post-tsunami investment may have priori- tized the creation of a tsunami early-warning system in the Mediterranean Sea rather than in the Indian Ocean, given the fact that it is statistically more vulnerable.

120 LEARNING FROM CATASTROPHES

Global collaboration, national government commitment, and community involvement have created excellent early-warning systems for high-probability disasters even in some of the world’s poorest countries. Yet for low-probability events, a different model is needed—one that emphasizes not the hypothetical hazard but the underlying vulnerabilities.