346 AAssignmentsReading #6

Get_the_Data_and_Make_Them_Usable.pdf

Home >English homework help > 346 AAssignmentsReading #6

[5]

G E T T H E D ATA A N D M A K E

T H E M U S A B L E

In the early days of the Cold War, as the mistrust between the United States and Soviet Union gave way to open confrontation, the US military flew regular spy missions over Soviet territory. Flying above the range of Soviet antiaircraft weapons, the spy planes would collect valuable intelligence. But the military had a problem. Bad weather sometimes prevented the planes from successfully completing the mission. To ensure that its pilots would have good visibility when they arrived in enemy airspace, the US military needed to forecast conditions over the Soviet Union. Collecting data through conven- tional means, such as weather balloons, was not going to work; the Soviets would immediately shoot down any foreign objects within range of their antiaircraft weapons. So the American generals turned to a think tank called the RAND Corporation for advice. In 1951, RAND issued a secret report that would not be declassified for over forty years.1 It proposed a novel approach to forecasting that in- volved “weather observations made by means of a television camera placed in an unmanned vehicle flying above the normal range of de- fense weapons.” In other words, a weather satellite.

Almost a decade later, RAND’s proposal came to life when the US government launched the Television and InfraRed Observation

Building a Resilient Tomorrow: How to Prepare for the Coming Climate Disruption. Alice C. Hill and Leonardo Martinez-Diaz, Oxford University Press (2020). © Oxford University Press. DOI: 10.1093/oso/9780190909345.003.0006

G E T T H E D ATA A N D M A K E T H E M U S A B L E

Satellite, or TIROS- 1— the first weather satellite in history (Figure 5). The satellite had two television cameras that took a dis- torted, black- and- white picture of a small part of the Earth’s surface every ten seconds. Each camera linked to a magnetic tape recorder that could store up to thirty- two photographs when the satellite’s signal was out of the range of stations on the ground. A power failure permanently disabled the cameras after just a few weeks in oper- ation, ending the mission. Still, in its brief life, TIROS- 1 beamed back some 20,000 images, including one of a cyclone above New Zealand, proving that weather observation from space was possible.

Since the days when TIROS- 1 represented the cutting- edge of technology, the world’s capacity to collect and analyze climate and

Figure 5 Artist’s rendering of the instruments aboard TIROS- 1 satellite. Source: National Aeronautics and Space Administration.

B U I L D I N G A R E S I L I E N T T O M O R R O W

100

weather data has exploded. The problem now is that we face a par- adox. At no point in history have these data been more plentiful and predictive models more powerful. Yet many of the people who need them most lack both access to the right data and the means to make them useful for decision- making.

Consider, for example, Perdido Beach, a tiny, climate- vulnerable town on the coast of Alabama in the United States. Given its loca- tion, the community suffers from high exposure to hurricanes and sea- level rise. In 2012, Hurricane Isaac washed away portions of the beachfront. Two years later, floods washed out some roads in Perdido Beach, trapping residents inside their homes. Coastal pro- tection is critical for the town’s survival, but how much infrastruc- ture should it build? What kind? At what cost? And what is the best placement to maximize protection? In order to make these critical decisions, the community needs data and modeling, but even in the United States, small communities lack the resources necessary to plan for resilience. Speaking to federal authorities, Perdido Beach Mayor Patsy Parker explained her predicament: “I’m just a part- time mayor in a small town. I don’t have a big planning staff, grant writers, or any resources. So how can I even know the size of the threats we are facing— and what can I do to protect the people of my town?”2 In most developing countries, the problem is even more dire.

As climate change advances and its impacts become clearer, more communities in the United States will need deeper insight into the future, both immediate and distant. Decision- makers will require information to make high- impact, hard- to- reverse decisions about water, agriculture, and where and how to build infrastructure in a world experiencing climate change. They will demand data to calculate the likelihood of catastrophic events and to figure out how best to plan and pay for defenses. They must model the projected evolution of droughts, heatwaves, and wildfires, so they can help

G E T T H E D ATA A N D M A K E T H E M U S A B L E

101

people get out of harm’s way. And they will need data to make disaster- relief operations more effective. In this chapter, we describe the data paradox and offer a few ideas on how to escape it.

THE GOLDEN AGE OF CLIMATE AND WEATHER DATA

Since the 1960s, technology has made possible what can only be described as a revolution in weather and climate data. Consider sat- ellites. One of the most modern is the GOES- 16 (Geostationary Operational Environmental Satellite), launched by the US gov- ernment in 2016. It provides a “full disk” image of the Earth cen- tered on the Western Hemisphere every 15 minutes and an image of the continental United States every five minutes. It can mon- itor hurricanes, wildfires, and other severe weather events as often as every 30 seconds and take pictures with sixteen spectral bands, meaning that it can “see” dust, haze, smoke, fog, ice, snow, changes in vegetation, and moisture levels in the atmosphere. When Hurricane Maria knocked out radar systems in Puerto Rico in 2017, GOES- 16 enabled meteorologists to monitor the massive storm as it made its way across the Caribbean. Satellites have come a long way since TIROS- 1’s wobbly, black- and- white transmission.

The data revolution no longer depends just on expensive, government- owned satellites. After working at the US National Aeronautics and Space Administration (NASA), British- born phys- icist Will Marshall launched the company Planet Labs (now simply called Planet). His big idea was to piggyback on advances in min- iaturization to build tiny satellites, or “cubesats.” No bigger than a shoebox and costing a small fraction of the amount of traditional satellites, cubesats are launched dozens at a time in “flocks.” Once

B U I L D I N G A R E S I L I E N T T O M O R R O W

102

they are in orbit, they spread out to cover the planet. Although they are not substitutes for GOES- type satellites, each Planet satellite can collect 10,000 images, covering 770,000 square miles (2 mil- lion km2) per day.3 Today, a flock of cubesats that is already in orbit provides daily images of the Earth’s entire landmass.

Despite the notoriety that drones have gained for their con- troversial role in targeted assassinations, when it comes to cli- mate resilience, drones provide a valuable service. Packed with sensors, digital radios, and powerful microprocessors, drones en- able farmers to see their fields at high resolution with multispec- tral bands, which can reveal potential irrigation problems, water stress, or pest or fungal infestations. Drones can detect healthy and stressed plants more accurately than the naked eye, allowing farmers to focus their efforts on trouble spots. Drones can do this cheaply and regularly, providing the farmer with time- series animations that reveal how the field is responding to weather and farmer interventions. Heavy- duty drones, such as those used by NASA’s Hurricane and Severe Storm Sentinel Mission, can get close to hurricanes and collect data on the forces that drive them. After a disaster has struck, drones can survey the damage quickly, allowing for instant identification of areas most in need of assis- tance, which may speed up the recovery process.

Ground and sea- based sensors also play a role in the data revolu- tion by providing superlocalized information. As Hurricane Sandy approached the northeastern United States in 2012, US government scientists scrambled to deploy water- level and barometric pressure sensors in over two hundred locations along the Atlantic coast, from Virginia to Maine.4 Data from these devices proved crucial to emergency managers and insurers, who required the most precise information about the extent and costs of the damage. Startup com- panies have developed ground- based, wireless sensing devices that

G E T T H E D ATA A N D M A K E T H E M U S A B L E

103

are solar- powered and independent of the cellular network, so that they can keep collecting and transmitting data even if electricity and telecoms grids go down.

Mobile phones can turn anyone into an environmental data- collection agent. Thousands of volunteers armed with cell phones now regularly collect information about the impacts of extreme weather events and disseminate it on social media. These efforts are also making it possible to map previously unmapped parts of the world, so that communities can better understand flood and heatwave risks. People carrying cell phones can even generate useful information unwittingly. Call detail records, which time stamp and pinpoint the location every time a person makes or receives a mo- bile phone call, can help emergency- management agencies un- derstand, in near- real time, how a population responds to disaster alerts. They also enable emergency managers to track population movements after a natural disaster.

Thanks to all this technology, the volume of data collected today is impressive. The US National Oceanic and Atmospheric Administration (NOAA) alone collects 20 terabytes of environ- mental data from its satellites and other sources every day. That’s the daily equivalent of about 8,300 ninety- minute Netflix videos.5 As of 2018, the satellite company Planet was collecting about six terabytes a day from its satellites— about 2,500 Netflix videos a day.6 We are awash in data. But how do we make it useful for decision- making and accessible to the people who need it?

MODELING THE FUTURE

To render all the data we collect useful, weather data agencies and companies must clean, process, and package it so it can be digested

B U I L D I N G A R E S I L I E N T T O M O R R O W

104

by decision- makers. Machine learning, and artificial intelligence (AI) more generally, can help us sort through mountains of data rap- idly. Programmers can teach computers to “read” the massive inflow of data and look for relevant patterns. Computers can do this much faster and more accurately than humans can, and they work around the clock, without coffee breaks. Machine- learning programs can also modify themselves as they interact with the data, making au- tomatic adjustments to get better at their task. They can also share what they learn with other machines, multiplying the speed of im- provement across the network.

But even when decision- makers can access and crunch the right data, they need insights about the future, not just about the past and present. For that, decision- makers need models that can inform forecasts. Models are simplified simulations of how some- thing works. The stronger our understanding of the science behind climatic events, and the larger the volume of historical data we can feed into the models, the better the models become at forecasting the future. And the more powerful the models become, the greater the ability of government and business leaders to make informed decisions about climate resilience.

Climate and weather models are getting better all the time. Take weather forecasts. Today’s five- day forecasts are as good as two- day forecasts were twenty- five years ago. Five- day cyclone forecasts are now the global standard.7 Cloud computing has helped make this possible. By buying massive online storage capacity, software, and computing power and renting it out to thousands of users, “cloud” providers enable their customers to store and analyze large quan- tities of data at a fraction of what it would cost if they bought their own equipment. This switch has helped advance the development and use of climate and catastrophe models.

G E T T H E D ATA A N D M A K E T H E M U S A B L E

105

Few groups should be as motivated to model the future of ex- treme weather as insurance and reinsurance companies and catas- trophe bond investors (see chapter 4). These groups have lots of money riding on this information. Depending on how much damage hurricanes, earthquakes, floods, windstorms, and wildfires cause (or don’t cause), insurers and investors can either earn or lose millions or, increasingly, billions. Catastrophe (“cat”) models can deliver an informational edge by helping them understand the likelihood that a disaster will strike a given place, the amount of property that is vul- nerable to the disaster, and the value of the damage that could result. The better the models, the more accurately insurers and cat- bond markets can set the price for protection, and the better the chance that those being insured are not underpaying (or overpaying) for coverage. Of course, cat models have other uses. For instance, urban planners can use them to make better decisions about where to put infrastructure, and emergency managers can employ them to un- derstand which parts of a country or region are at greatest risk.

Blind spots still bedevil cat models, however. The models are good, for example, at forecasting wind damage from a storm. But they have much more trouble computing flood damage, which depends on many more factors, such as elevation and topography, that determine where the water flows after it hits the ground. For this reason, all the major commercial cat models in the past have consistently underestimated the size of insured losses caused by major hurricanes affecting the United States.8 Wildfire models are getting more and more accurate at forecasting which areas are at greatest risk for fire, and once a fire starts, how it might behave. But as with hurricanes, the models break down when it comes to the biggest fires. Fires that get big enough and hot enough to generate their own complex wind patterns, including “firenados,” are, so far, too complex to model accurately.9

B U I L D I N G A R E S I L I E N T T O M O R R O W

106

W hen it comes to modeling climate impacts, one important challenge is “downscaling.” Scientists have generated reams of information regarding climate impacts on global, regional, and national scales. But information to forecast how climate impacts will unfold in a small geographic area or corporate facility re- mains harder and more expensive to obtain. Yet this information is critical to making good resilience decisions. Everyone from local officials to supply- chain risk managers, urban planners, and emergency- management agencies needs to understand how cli- mate change- related events might impact specific factories, farms, ports, buildings, and critical pieces of infrastructure. If the models spit out a “pixel size” that is too large, then the information will be too general to be actionable.

Entrepreneurs are taking on the challenge. Startups in Silicon Valley and other places are hoping to bring super- localized forecasting to government and to the private sector. Highly gran- ular modeling is becoming available, especially in the most lucrative markets of the United States and Europe. One Concern, a Silicon Valley startup, has launched products that promise to forecast flood levels and other perils with “asset level” precision— meaning at the level of individual buildings and industrial facilities.10 In Florida, an environmental lawyer and a climate scientist have teamed up to help clients protect themselves from flooding related to sea- level rise using proprietary climate- modeling technology. The firm, Coastal Risk, provides clients with maps of projected tidal and storm- surge flooding at a resolution of up to one square meter (10.76 square feet). As the business case for climate- predictive analytics becomes more and more compelling, entrepreneurs will compete with growing intensity to give their clients an informational edge to build resilience.

G E T T H E D ATA A N D M A K E T H E M U S A B L E

107

WHO GETS THE DATA, AND AT WHAT PRICE?

While climate and weather data have become more plentiful, and models more powerful, the question is who can access them and at what price? As climate change advances and the need to build resilience becomes a matter of burning urgency, this question will become political and ethical, not just economic. In the United States, policymakers have long debated whether government or pri- vate sector should take the leading role in the provision of weather and climate data, and they have gone back and forth on the issue several times.

Consider the story of Landsat, a series of US government weather satellites that have been gathering data since 1972, pro- viding an unbroken stream of images of the Earth and documenting natural and human- caused changes to the planet in ever- greater detail. During the 1970s and 1980s, the government considered privatizing American weather satellites and transferring Landsat data to private companies for distribution. In 1984, Congress pro- hibited the privatization of the government’s satellites, but it did pass a law allowing the transfer of Landsat data to a private vendor under an exclusive license.11 That vendor could then sell the data for a profit. The company, called the Earth Observation Satellite Company (EOSAT), promptly raised prices for Landsat data and focused on developing sophisticated data products for the high end of the market.12 After the price hike, many customers for basic data products, including academic institutions, could no longer afford it.

Less than a decade later, Congress reversed itself, concluding that “the cost of Landsat data has impeded the use of such data for scientific purposes, such as for global environmental change

B U I L D I N G A R E S I L I E N T T O M O R R O W

108

research, as well as for other public sector applications.”13 Landsat data distribution shifted back to the government, which charged only the cost of fulfilling users’ requests. But the pressures to privatize continued. Another attempt to turn Landsat data over to the private sector came in 2002, unsuccessfully, and in 2012, when the Department of Interior asked the Landsat Advisory Group to consider privatization. The group advised against it. Today, Landsat data is freely and openly available to all, at a cost of essentially zero. But as the economic value of weather and cli- mate data continues to rise, the debate over privatization will likely return.

Limited access to catastrophe models, not just to the underlying data, also poses concerns. Cat models remain expensive. A handful of cat modeling companies has dominated the market for years, and some blame this situation for slowing innovation and keeping costs high.14 The proprietary models of those companies are closed “black boxes.” Insurers and other clients pay handsomely to use these tools under a commercial license, but they only own the outputs of the model, not the model itself. Also, insurers have historically had a hard time comparing the cost and quality of the models produced by the various providers.

Developing countries have an especially hard time getting access to models that will enable them to understand the impacts of climate change. Poorer regions of the world are often not lucrative places for insurers and reinsurers to do business. As a result, modeling com- panies have not spent resources modeling natural disasters there, and so climate risk in those countries remains poorly understood. That, in turn, makes it less likely that people there will want to buy insurance. In an effort to break this vicious cycle, government and academic institutions have developed free, open- source models, but they are still few and far between. The most comprehensive

G E T T H E D ATA A N D M A K E T H E M U S A B L E

109

inventory available currently lists 240 cat models, but only 9 percent are freely available, and some provide only limited functionality.15

The government’s role in providing basic public goods in the form of climate data and information is more important than ever. Public agencies should provide a bedrock of free, taxpayer- funded climate and weather data that all levels of government can use to inform public policy, to protect public assets, and to build the resil- ience of poor and vulnerable people. It should also continue to fund the basic climate science that underlies climate models. Meanwhile, private companies can specialize in customized, proprietary products. Governments typically lack the necessary expertise to do the latter, and the private sector usually innovates more quickly than government agencies. Also, if companies need highly downscaled data and models to safeguard their own facilities and supply chains, it seems fair that they pay for that.

Realizing that government needs to play a strong role in pro- viding climate information, countries around the world have started to strengthen their weather and climate agencies. Uruguay, for ex- ample, transformed its meteorological agency from a unit within the Defense Ministry into a separate entity, giving it greater inde- pendence and flexibility. In 2016, the government of Uruguay also established a national- level platform that serves as a one- stop source of climate data and information. It also provides tools to help farmers make climate- informed decisions.16 Another example is the India Meteorological Department, which replaced an obsolete weather- prediction model with a customized version of a system developed by the US National Centers for Environmental Prediction.17 The new system is designed to enable India’s meteorologists to issue more granular and accurate forecasts in a country where under- standing monsoon rain patterns and extreme heat events is vital to lives and livelihoods.

B U I L D I N G A R E S I L I E N T T O M O R R O W

110

Meanwhile, some US government officials have given serious thought to how best to strengthen the nation’s climate informa- tion system. In 2015, the US Government Accountability Office (GAO), the nonpartisan watchdog agency that serves Congress, published the results of an illuminating investigation.18 Climate information in the United States, the report found, “exists in an uncoordinated confederation of networks and institutions.”19 The federal government’s climate information, the report went on to say, “is fragmented across many individual agencies that use the information in different ways to meet their respective missions . . . [D] ecision makers are vastly underserved by the cur- rent ad hoc collection of federal climate information services.”20 The GAO also studied the national climate information institutions of the Netherlands, the United Kingdom, and Germany, all of which the watchdog regarded as more effective than the American approach.

What to do about the US government’s fragmented system? Creating a new centralized agency to manage climate information may seem tempting. Yet the GAO inquiry rightly concluded that a more decentralized system would best serve the United States be- cause no single agency has the expertise required to provide a full menu of climate information services. At the same time, a strength- ened US approach should give one federal agency the authority to push other government departments to share information with the public in “customer- friendly” formats. Above all, the federal govern- ment should put a system in place that can harness the vast amount of climate and weather information it collects and make it available in accurate, reliable, and easily accessible ways across multiple, com- plementary platforms. Any such effort must remain laser focused on ensuring that those who most need the information can readily put it to immediate use.

G E T T H E D ATA A N D M A K E T H E M U S A B L E

111

Meanwhile, cities are also building data infrastructure to help them manage climate impacts. New Orleans, for instance, built one of the best data hubs in the country. Its Data Center brings together information from multiple sources to track the city’s demographic and economic recovery after shocks such as Hurricane Katrina. New Orleans has also created geospatial maps that enable decision- makers to visualize multiple layers of information about the urban environment simultaneously. These maps enable officials and others to pinpoint city blocks where temperatures may be dangerously high and to identify the most strategic locations for placing green infrastructure. Officials in Norfolk, Virginia have taken notice, and they are studying New Orleans’s achievements. They hope to estab- lish a data center of their own.

LOST WITHOUT TRANSLATION

Noted psychologist Daniel Kahneman has observed, “No one ever made a decision because of a number. They need a story.”21 With climate information, the challenge is how to make the data speak to decision- makers and the public. The experts who generate the data sometimes forget the critical nature of this translation challenge. What good are the best models powered by the best data if the users fail to draw actionable insights from them?

Translating climate and weather data can take different forms. Some approaches harness the power of visuals. One example is Aqueduct, an open- source online tool developed by the World Resources Institute (WRI), a nonprofit research organization.22 Aqueduct takes data and modeled projections of water stress and generates colorized maps. The user can see how his or her commu- nity is expected to fare in terms of water availability as far as 2040.

B U I L D I N G A R E S I L I E N T T O M O R R O W

112

Areas expected to remain rich in water are shown in a soothing cream color, while water- stressed areas become progressively redder. High- risk places appear in bright red, and those at “extremely high risk” show up in a deep, burnt maroon.

Owens Corning, a company based in the Rust Belt town of Toledo, Ohio, put this tool to use. The firm’s 20,000 employees produce insulation, roofing, and fiberglass composites. The work requires lots of water to cool the machinery and materials used in high- temperature manufacturing processes. Owens Corning has facilities in areas of Mexico, Spain, China, and the United States that will struggle to meet water demand. The company layered a map of its global facilities on top of Aqueduct’s colorized maps, making it possible to identify which facilities faced the highest risk of water stress, now and in future decades.23 Guided by this analysis, Owens Corning ranked its facilities worldwide to determine where to focus its resilience budget.24

Sometimes translation is all about tailoring the message to the audience and delivering it through the right messenger. In Colombia, for example, climate experts noticed that farmers there were not thinking enough about how climate change might af- fect their crops. Climate- impact scientist Julian Ramirez- Villegas recalled, “[A] lot of people were aware of the importance of the cli- mate, but didn’t know what to do about it.” “After going to the field, we realized that Colombian farmers were planting based on what happened last year . . . With the amount of climate variability that we have here in Colombia, particularly rainfall, that’s a recipe for disaster.”25 Farmers needed information not just about the weather over the next few days, but also about conditions several months into the future.

A nonprofit organization dedicated to helping farmers in de- veloping countries cope with climate change impacts took on the

G E T T H E D ATA A N D M A K E T H E M U S A B L E

113

challenge. Based in southwestern Colombia, the International Center for Tropical Agriculture (CIAT), employs over three hundred scientists and boasts a depository with seeds for 37,000 varieties of beans. To help Colombian farmers understand future risk, CIAT generated forecasts using crop models and big- data analytics. The forecasts provided the farmers with very localized climate projections extending several months into the future. CIAT aimed to help the farmers answer three practical questions. W here should they plant, given the forecast? W hen is the best time to do so? And which crop variety should they put in the ground?

The next challenge was how to get the information to the people who needed it most. CIAT worked with farmer organizations to build tools for relaying the forecasts to their member farmers. But many small- scale farmers lived in remote areas and didn’t belong to farmers’ organizations. To reach them, CIAT turned to radio, tele- vision, and text messages. CIAT calculates that 300,000 farmers are now receiving climate information thanks to this initiative. The or- ganization wants to take the effort global. “Imagine if we could im- plement similar systems in sub- Saharan Africa or South East Asia,” says Ramirez- Villegas. “There are potentially millions and millions that could benefit.”26

In the United States, the Obama administration launched an initiative called Climate Services for Resilient Development, along with a slew of other efforts aimed at helping make climate infor- mation more useful for developing nations, including Colombia. Another novel initiative launched by the administration was the Climate Resilience Toolkit, a web portal offering tools and infor- mation designed to help Americans inside and outside government understand climate risks and their implications for different regions of the country better. In a report issued in the administration’s final

B U I L D I N G A R E S I L I E N T T O M O R R O W

114

days, the White House recognized that despite these efforts, a lot more needed to be done in this area.27

We live in a golden age of climate and weather data and ana- lytical and modeling power. Yet we experience an acute shortage of insights to help drive resilience. Breaking out of this paradox involves collecting and downscaling data and ensuring that it is accessible. It demands addressing the weaknesses of the current catastrophe models and translating climate information in ways decision- makers can use. And it requires that we put in place effec- tive weather and climate agencies and foster a cadre of translators who can demystify climate information.

PRESCRIPTIONS AND PROVOCATIONS

• The federal government should rethink the organization of the US climate data system, drawing on lessons from other countries, to ensure that climate and weather data from across the government remains free, openly sourced, and dissemi- nated in user- friendly ways.

• Cities and states should consider creating data centers that collect and analyze local demographic and environmental data that are useful for resilience planning.

• Insurers and reinsurers should invest in catastrophe models that better reflect future climate risk and work with regulators to ensure that insurance premiums reflect that risk. The fed- eral government, in partnership with the private and non- profit sectors, should expand the availability of open- source, free catastrophe models.

G E T T H E D ATA A N D M A K E T H E M U S A B L E

115

• The federal government, in partnership with the private sector, should create a cadre of climate- science translators to help decision- makers in state and local governments evaluate climate risks, develop resilience strategies, and access federal funding.